BIOENERGETICS OF T H E C E L L : QUANTITATIVE ASPECTS

Developments in Molecular and Cellular Biochemistry

Series Editor: Naranjan S. Dhalla, Ph.D., M . D . (Hon.), F A C C

1. V.A. Najjar (ed.): Biological Effects of Glutamic Acid and Its Derivatives. 1981 ISBN 90-6193-841-4
2. V A . Najjar (ed.): Immunologically Active Peptides. 1981 ISBN 90-6193-842-2
3. V A . Najjar (ed.): Enzyme Induction and Modulation. 1983 ISBN 0-89838-583-0
4. V A . Najjar and L . Lorand (eds.): Transglutaminase. 1984 ISBN 0-89838-593-8
5. G.J. van der Vusse (ed.): Lipid Metabolism in Normoxic and Ischemic Heart. 1989 ISBN 0-7923-0479-9
6. J.F.C. Glatz and G.J. van der Vusse (eds.): Cellular Fatty Acid-Binding Proteins. 1990 ISBN 0-7923-0896-4
7. H.E. Morgan (ed.): Molecular Mechanisms of Cellular Growth. 1991 ISBN 0-7923-1183-3
8. G.J. van der Vusse and H . Stam (eds.): Lipid Metabolism in the Healthy and Diseased Heart. 1992
I S B N 0-7923-1850-1
9. Y. Yazaki and S. Mochizuki (eds.): Cellular Function and Metabolism. 1993 ISBN 0-7923-2158-8
10. J.F.C. Glatz and G.J. van der Vusse (eds.): Cellular Fatty-Acid-Binding Proteins, IL 1993 ISBN 0-7923-2395-5
11. R.L. Khandelwal and J.H. Wang (eds.): Reversible Protein Phosphorylation in Cell Regulation. 1993
ISBN 0-7923-2637-7
12. J. Moss and P Zahradka (eds.): ADP-Ribosylation: Metabolic Effects and Regulatory Functions. 1994
I S B N 0-7923-2951-1
13. V A . Saks and R. Ventura-Clapier (eds.): Cellular Bioenergetics: Role of Coupled Creatine Kinases. 1994
ISBN 0-7923-2952-X
14. J. Slezk and A . Ziegelhffer (eds.): Cellular Interactions in Cardiac Pathophysiology. 1995 I S B N 0-7923-3573-2
15. J.A. Barnes, H.G. Coore, A . H . Mohammed and R.K. Sharma (eds.): Signal Transduction Mechanisms. 1995
I S B N 0-7923-3663-1
16. A . K . Srivastava and J.-L. Chiasson (eds.): Vanadium Compounds: Biochemical and Therapeutic Applications. 1995
I S B N 0-7923-3763-8
17. J . M J Lamers and P.D. Verdouw (eds.): Biochemistry of Signal Transduction in Myocardium. 1996
ISBN 0-7923-4067-1
18. E.-G. Krause and R. Vetter (eds.): Biochemical Mechanisms in Heart Function. 1996 ISBN 0-7923-4118-X
19. R. Vetter and E.-G. Krause (eds.): Biochemical Regulation of Myocardium. 1996 ISBN 0-7923-4259-3
20. G.N. Pierce and W.C. Claycomb (eds.): Novel Methods in Molecular and Cellular Biochemistry of Muscle. 1997
ISBN 0-7923-4387-5
21. F.N. Gellerich and S. Zierz (eds.): Detection of Mitochondrial Diseases. 1997 ISBN 0-7923-9925-0
22. P.K. Singal, V Panagia and G.N. Pierce (eds.): The Cellular Basis of Cardiovascular Function in Health and
Disease. 1997 ISBN 0-7923-9974-9
23. S. Abdel-aleem and J.E. Lowe (eds.): Cardiac Metabolism in Health and Disease. 1998
ISBN 0-7923-8104-1
24. A . K . Srivastava and B . Posner (eds.): Insulin Action. 1998 ISBN 0-7923-8113-0
25. V . A . Saks, R. Ventura-Clapier, X . Leverve, M . Rigoulet and A . Rossi (eds.): Bioenergetics of the Cell: Quantitative
Aspects. 1998 ISBN 0-7923-8118-1

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Bioenergetics of the CeH:
Quantitative Aspects

Edited by

VALDURA. SAKS RENEE VENTURA-CLAPIER

Laboratories of Bioenergetics Cellular and Molecular Cardiology

Institute of Chemical and Biological Physics U 446INSERM
Tallinn, Estonia and Chatenay-Malabry, France
Joseph Fourier University, Grenoble, France

XAVIER LEVERVE ANDREROSSI

Laboratory of Bioenergetics Laboratory of Bioenergetics

Joseph Fourier University, Grenoble, France Joseph Fourier University, Grenoble, France

MICHEL RIGOULET

Laboratory of Bioenergetics
Institute of Biochemistry and Cellular Genetics
CNRS, Bordeaux, France

Reprinted from Molecular and Cellular Biochemistry, Volume 184 (1998)

Springer-Science+Business Media, B. V.
Library of Congress Cataloging-in-Publication Data

Bioenergetics of the cell: quantitative aspects / edited by Valdur A. Saks . . . [et al.].
p. cm. ~ (Developments in molecular and cellular biochemistry ; v. 24)
ISBN 978-1-4613-7587-6
1. Mitochondria. 2. BioenergeticsMathematical models. 3. Cell
metabolismMathematical models. 4. Energy metabolismMathematical
models. I. Saks, V. A. II. Series
QH603.M5B526 1998 97-42565
571.6'57-dc21 CIP

ISBN 978-1-4613-7587-6 ISBN 978-1-4615-5653-4 (eBook)

DOI 10.1007/978-1-4615-5653-4

Printed on acid-free paper

A l l rights reserved

1998 Springer Science+Business Media Dordrecht

Originally published by Kluwer Academic Publishers in 1998
Softcover reprint of the hardcover 1st edition 1998
No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical,
including photocopying, recording or by any information storage and
retrieval system, without written permission from the copyright owner
Molecular and Cellular Biochemistry:
An International Journal for Chemical Biology in Health and Disease
CONTENTS VOLUME 184, Nos. 1 & 2, July 1998

BIOENERGETICS OF THE CELL: QUANTITATIVE ASPECTS

Valdur A. Saks, Renee Ventura-Clapier, Xavier Leverve, Michel Rigoulet and Andre Rossi, guest editors

Preface
What do we not know of cellular bioenergetics? - A general view on the state of the art

Part I: Bioenergetics of mitochondria: In vitro and in vivo studies

M.D. Brand: Top-down elasticity analysis and its application to energy metabolism in isolated mitochondria and intact cells
O.v. Demin, B.N. Kholodenko and v.P. Skulachev: A model of 0'2- generation in the complex III of the electron transport
chain 21-33
M. Rigoulet, X. Leverve, E. Fontaine, R. Ouhabi and B. Guerin: Quantitative analysis of some mechanisms affecting the yield
of oxidative phosphorylation: Dependence upon both fluxes and forces 35-52
X. Leverve, B. Sibille, A. Devin, M.-A. Piquet, P. Espie and M. Rigoulet: Oxidative phosphorylation in intact hepatocytes:
Quantitative characterization of the mechanisms of change in efficiency and cellular consequences 53--65
N. Averet, V. Fitton, O. Bunoust, M. Rigoulet and B. Guerin: Yeast mitochondrial metabolism: From in vitro to in situ quantitative
study 67-79
V.A. Saks, v.I. Veksler, A.V. Kuznetsov, L. Kay, P. Sikk, T. Tiivel, L. Tranqui, 1 Olivares, K. Winkler, F. Wiedemann and W.S.
Kunz: Permeabilized cell and skinned fiber techniques in studies of mitochondrial function in vivo 81-100
L. Rappaport, P. Oliviero and J.L. Samuel: Cytoskeleton and mitochondrial morphology and function 101-105
A. Devin, P. Espie, B. Guerin and M. Rigoulet: Energetics of swelling in isolated hepatocytes: A comprehensive study 107-121

Part II: Energy transfer networks: Molecular physiology ofkinases, lessons from transgenic mice, mathematical theories
U. Schlattner, M. Forstner, M. Eder, O. Stachowiak, K. Fritz-Wolf and T. Wallimann: Functional aspects ofthe X-ray structure
of mitochondrial creatine kinase: A molecular physiology approach 125-140
O. Stachowiak, U. Schlattner, M. Dolder and T. Wallimann: Oligomeric state and membrane binding behaviour of creatine
kinase isoenzymes: Implications for cellular function and mitochondrial structure 141-151
W. Qin, Z. Khuchua, J. Cheng, 1 Boero, R.M. Payne andA.W. Strauss: Molecular characterization ofthe creatine kinases and
some historical perspectives 153-167
P.P. Dzeja, RJ. Zeleznikar and N.D. Goldberg: Adenylate kinase: Kinetic behavior in intact cells indicates it is integral to multiple
cellular processes 169--182
K. Steeghs, F. Oeriemans,A. de Haan,A. Heerschap, L. Verdoodt, M. de Bie, W. Ruitenbeek,A. Benders, C. Jost, 1 van Deursen,
P. Tullson, R. Terjung, P. Jap, W. Jacob, D. Pette and B. Wieringa: Cytoarchitectural and metabolic adaptations in
muscles with mitochondrial and cytosolic creatine kinase deficiencies 183-194
K. Nicolay, F.A. van Dorsten, T. Reese, M.J. Kruiskamp, J.F. Gellerich and C.J.A.van Echteld: In situ measurements of creatine
kinase flux by NMR. The lessons from bioengineered mice 195-208
M.K. Aliev, F.A. van Dorsten, M.G. Nederhoff, C.lA. van Echteld, V. Veksler, K. Nicolay and V.A. Saks: Mathematical model
of compartmentalized energy transfer: Its use for analysis and interpretation oP 1P-NMR studies of isolated heart of creatine
kinase deficient mice 209--229
R. Ventura-Clapier, A. Kuznetsov, V. Veksler, E. Boehm and K. Anflous: Functional coupling of creatine kinases in muscles:
Species and tissue specificity 231-247
GJ. Kemp, D.N. Manners, IF. Clark, M.E. Bastin and G.K. Radda: Theoretical modelling of some spatial and temporal aspects
of the mitochondrion/creatine kinase/myofibril system in muscle 249--289
V. Saks, P. Dos Santos, F.N. Gellerich and P. Diolez: Quantitative studies of enzyme-substrate compartmentation, functional
coupling and metabolic channelling in muscle cells 291-307
Part III: Metabolic signalling and calcium: Regulation of mitochondrial oxidative phosphorylation in vivo
B.N. Kholodenko, J.M. Rohwer, M. Cascante and H.V Westerhoff: Subtleties in control by metabolic channelling and enzyme
organization 311-320
J.H.G.M. van Beek, X. Tian, C.J. Zuurbier, B. de Groot, C.J.A. van Echteld, M.H.J. Eijgelshoven and lB. Hak: The dynamic
regulation of myocardial oxidative phosphorylation: Analysis of the response time of oxygen consumption 321-344
B. Korzeniewski: Is it possible to predict any properties of oxidative phosphorylation in a theoretical way? 345-358
R.G. Hansford and D. Zorov: Role of mitochondrial calcium transport in the control of substrate oxidation 359-369
L.S. Jouaville, F. Ichas and l-P. Mazat: Modulation of cell calcium signals by mitochondria 371-376

Part IV: Bioenergetics and medicine

F. Di Lisa and P. Bernardi: Mitochondrial function as a determinant of recovery or death in cell response to injury 379-391
I.E. Hassinen, K.H. Vuorinen, K. Ylitalo and A. Ala-Rami: Role of cellular energetics in ischemia-reperfusion and ischemic
preconditioning of myocardium 393-400
A. Rossi, L. Kay and V Saks: Early ischemia-induced alterations ofthe outer mitochondrial membrane and the intermembrane
space: A potential cause for altered energy transfer in cardiac muscle? 401-408
T. Letellier, M. Malgat, R. Rossignol and J.-P. Mazat: Metabolic control analysis and mitochondrial pathologies 409-417
E.K. Seppet, A. Kaasik, A. Minajeva, K. Paju, J.J. Ohisalo, R. Vetter and U. Braun: Mechanisms of thyroid hormone control
over sensitivity and maximal contractile responsiveness to ~-adrenergic agonists in atria 419-426
M.L. Guerrero-Ontiveros and T. Wallimann: Creatine supplementation in health and disease. Effects of chronic creatine ingestion
in vivo: Down-regulation of the expression of creatine transporter isoforms in skeletal muscle 427-437
S. Neubauer, M. Hom, D. Hahn and K. Kochsiek: Clinical cardiac magnetic resonance spectroscopy - present state and future
directions 439-443
B. Beauvoit and B. Chance: Time-Resolved Spectroscopy of mitochondria, cells and tissues under normal and pathological
conditions 445-455

Index to Volume 184 457-460

Molecular and Cellular Biochemistry 184: 1, 1998.
Preface
This volume continues the discussion of the problems of
cellular bioenergetics - the tradition we started in 1994 by
publication of the focused issue of Molecular and Cellular
Biochemistry, volume 133/134 and a book 'Cellular Bio-
energetics: role of coupled creatine kinases' edited by V.Saks
and R. Ventura-Clapier and published by Kluwer Publishers,
Dordrecht - Boston. In the present volume, use of quantitative
methods of studies of organized metabolic systems, such as
mathematical modeling and Metabolic Control Analysis, for
investigation of the problems of bioenergetics of the cell is
described together with presentation of new experimental
results. The following central problems of the cellular bio-
energetics are the focus of the discussions: the mechanisms
of regulation of oxidative phosphorylation in the cells in vivo
in health and disease, the influence of the fine intracellular
structural organization on the mitochondrial function, the
intracellular networks of energy transfer and metabolic
feedback signaling, the mitochondrial role in intracellular
calcium cycling and changes in the mitochondrial membrane
systems and cellular energetics under pathological conditions.
Systematic studies from many laboratories presented in this
volume convincingly show that many mitochondrial prop-
erties, including affinities for substrates and ADP are different
in vivo and in vitro. The recently solved X-ray structure of
the mitochondrial creatine kinase and its molecular biology
are analyzed with respect to its molecular physiology and
functional coupling to the adenine nucleotide translocase, as
well as its participation, together with the adenylate kinase
system, in intracellular energy transfer. The results of the
studies of creatine kinase deficient transgenic mice are
summarized and analyzed by using mathematical models of
the compartmentalized energy transfer, thus combining two
powerful new methods of the research. All these results,
together with the physiological and NMR data on the cardiac
metabolic and mitochondrial responses to work-load changes
concord to the concept of metabolic networks of energy
transfer and feedback regulation. The papers, mostly review
articles, are written by the leading experts in the field who
analyze their experience of research and give good insights
into the future development ofthis increasingly important area
of biological sciences. The editors thank all authors for their
active participation, and hope that this volume will be very
interesting and useful for both students and active researchers
in the field of bioenergetics, cellular physiology, biochemistry,
molecular biology and medicine.
VALDURA. SAKS, Grenoble, France and Tallinn, Estonia;
RENEE VENTURA-CLAPIER, Chatenay - Malabry, France;
XAVIER LEVERVE, Grenoble, France;
MICHEL RIGOULET, Bordeaux, France;
ANDRE ROSSI, Grenoble, France.
Molecular and Cellular Biochemistry 184: 3-9, 1998.
1998 Kluwer Academic Publishers.
What do we not know of cellular bioenergetics? - a
general view on the state of the art
The definition
Our science, bioenergetics of the cell (or cellular bio-
energetics), the most important problems of which are
discussed in this volume, studies integrated biochemical and
biophysical processes of all kind of energy conversion in
living cells, and investigates the mechanisms of their
regulation. This is the definition of bioenergetics of the cell
we give to the students here, at Joseph Fourier University
in Grenoble, France, in the opening lecture. And it may serve
equally well for the purpose of active science, to define
precisely the topics of our discussions. The discussions in
multiple papers in this volume written by authors actively
working in the field are mostly related to the consideration
of quantitative aspects of the experimental research in
bioenergetics which have proven most effective and useful
for identification of complex cellular mechanisms in vivo.
The problems
In most non-photosynthetic and aerobic eucaryotes, bio-
energetics is usually associated with experimental studies
of isolated mitochondria which playa central role in ATP
synthesis through oxidative phosphorylation. These studies
have been performed for half a century and were pioneered
by Lehninger, Chance and many others. A decisive con-
tribution was made by Peter Mitchell who proposed the
chemiosmotic theory which provides the conceptual basis
for understanding a great variety of energy conversion
mechanisms including oxidative phosphorylation. His
laboratory and others (Skulachev, Lieberman and their
co-workers among the first) experimentally proved proton
pumping out of mitochondria during substrate oxidation, the
existence of the difference in membrane electrical potential
and proton-motive force (pm), the dependence of ATP
synthesis on the pmf and recently Boyer and Walker identified
the rotary mechanism of ATP-synthase reaction. Gradually,
it appeared that oxidative phosphorylation is a complex and
highly controlled network, and the adjustment betweenATP
formation and expenditure could be modulated through
Address/or offprints: V.A. Saks, Laboratory of Bioenergetics, Institnte of Chemical and Biological Physics,Akadeemia tee 23, EE 0026 Tallinn, Estonia
different sites by different effectors. From studies on
isolated mitochondria, it was clear that not only ADP was
important for setting the rates of respiration, but also
mitochondrial NADH, calcium, pmf and inorganic phosphate
play their roles in determining the energy fluxes in mito-
chondria. More recently, to explain the short-term control of
mitochondrial respiration and ATP synthesis, studies have
been focused on some enzymatic steps involved in the control
and regulation of these processes i.e. cytochrome oxidase,
some dehydrogenases, mitochondrial carriers and/or ATP
synthase. The application of the Metabolic Control Analysis
(MCA) developed by Kacser and Burns and Heinrich and
Rapoport afforded a powerful method for the quantification
of the control exerted by various enzymes and effectors on
oxidative phosphorylation under different steady states. In
this volume, a particular development ofthis approach called
'top-down elasticity analysis' is presented by M. Brand.
The intensive work done using this tool has led to two main
results: (1) among the number of potential control and/or
regulation sites and effectors, several of these can operate
simultaneously in determining a given steady state of
oxidative phosphorylation i.e. the control is distributed; (2)
these control and/or regulation factors may contribute to a
variable extent in the control of the whole pathway depending
on the cell and the steady state considered, i.e. control
distribution is highly variable. For readers more interested
in these problems we may refer to the recent publication by
D. Fell, Understanding the Control of Metabolism, Portland
Press, London and Miami, 1997.
Another central problem of the bioenergetics in its
classical meaning is the mechanism of the proton transport
involved in the chemiosmotic energy transduction. Indeed,
two different models have been proposed: direct coupling,
initially induced by Mitchell, in which translocated protons
are directly involved in the coupled chemical reaction (redox
and ATP synthesis or hydrolysis), and an indirect coupling,
linked to a conformational change in the enzymatic complex
between chemical reaction and proton transport. Analysis of
the nature of some mechanisms leading to a change in the
yield of oxidative phosphorylation (see chapter by Rigoulet
et al. in this volume) has given some information concerning
4
both the putative mechanism of coupling and the kinetics
control versus regulation of some pumps involved in
mitochondrial oxidative phosphorylation.
However, the relatively new but the most important
question in cellular bioenergetics is how mitochondria
behave in the cell and how they are regulated in vivo?
Without no doubt, all information collected in mitochondrial
studies in vitro is extremely valuable to solve this question.
However, this information appears not to be sufficient. The
already classical and many times described example is
directly related to the central issue of cellular bioenergetics:
very multiple metabolic studies, by using NMR or more
conventional biochemical techniques have shown that in many
cell types energy fluxes can be changed very significantly
while practically no change in the total levels of metabolites
is observed. This is contrary to mitochondria in vitro which
are regulated mostly by ADP concentration in the medium
- this is the classical respiratory control phenomenon.
Therefore, the in vitro data are not a priori sufficient to
explain the regulation of cellular energetics in vivo.
Evidently, it is the nature of intracellular factors of regulation
of mitochondrial activity which we do not know as yet and
do not understand completely. They are, without any doubt,
associated with cellular architecture which is broken and
lost when we try to isolate mitochondria from the cells.
The results of very numerous studies show that the intra-
cellular medium can not be taken as a homogeneous
solution. In living cells, with a very highly organized
ultrastructure, there is practically no free protein in
solution and even soluble macromolecules, such as compo-
nents of glycolytic system etc. are organized into supra-
molecular structures. Not only the macromolecules but
even significant part of the molecules of the water are not
free in the cells. The use of high voltage or embedment-free
electron microscopy has demonstrated the existence of very
highly elaborate and dynamic cytoskeletal structural network
(called 'microtrabecular lattice' by Porter) connecting many
cellular structures and most importantly, the mitochondria
with cell membranes, myofibrils and other cellular structures.
These novel developments and new aspects of cellular
regulation are reviewed in our volume by Rappaport and
Samuel (chapter 7). The highly organized cytoarchitecture
inevitably results in compartmentation, microcompart-
mentation, metabolic channelling, and functional coupling
- all phenomena based on direct interactions between cellular
structures, and which are the basis of cell function. This
changes, in many unexpected ways, the mechanisms of
control and regulation of metabolic processes as it is clearly
shown by a theoretical analysis, given in chapter 19 by
Kholodenko et al. All these factors of control and regulation
are lost in vitro. This is what makes the intracellular medium
different from simple homogeneous solution. And to under-
stand how mitochondrial function is regulated in vivo, all
these cellular factors should be accounted for, and at least
should not be ignored.
This is why the purpose of this book is to consider the
current state of the art of the studies of phenomena of
metabolic regulation in cellular bioenergetics, based on the
macromolecular and structural organization in the cell
interior, compartmentation and channelling of metabolic
intermediates, and to identify the most interesting perspect-
ives for further developments. These complicated systems
are best understood by using methods of quantitative
analysis, and hence the title of the volume: we are most
interested in the quantitative aspects of in vivo studies of
mitochondrial function. Among different powerful methods
of these studies, the application of mathematical modeling
for description of the energy fluxes in the cell seems to be
one of most promising for understanding of the metabolic
organization of the cellular life and regulation of energy
production, transfer and utilization. It is also important to
emphasize that not only the studies of normal cells but the
investigation of cellular pathologies give important in-
formation of the mechanism of cellular regulations. At the
same time, these studies provide the satisfying means of
direct application of the results of fundamental research for
the analysis of the mechanisms of pathogenesis of diseases,
and thus for practical purposes.
Commentaries to the text
All contributors to this volume have a long time and serious
experience in bioenergetics, including in vivo studies and
quantitative analysis of complex intracellular systems. Of
course, this group of authors represent only a small part of
investigators actively working in the field of cellular
bioenergetics and contributing in the increase of our
knowledge of the mechanisms of cell regulation. The picture
given in this book can only be representative. The main
attention was focused mostly on the highly specialized cells
with high and fluctuating energy fluxes like muscle cells.
The book starts with clear and simple description of the
principles of top-down elasticity analysis for the quantitative
studies of mitochondrial regulation by different effectors,
both in vitro and in vivo, given by M. Brand, as already
mentioned above. An interesting and elegant quantitative,
kinetic analysis of electron transport in respiratory chain in
mitochondria is presented from Skulachev's laboratory by
O. Demin et al. (chapter 2). Using a kinetic model of direct
interaction of fixed electron carriers and based on the
investigation of isolated mitochondria, this theoretical study
gives very practical results useful for cellular bioenergetics
in vivo: it seems that it is a good idea to induce some degree
of uncoupling in the inner mitochondrial membrane by
increasing proton leak in order to avoid the excess of oxygen

free radical generation in the cell. The next chapter by
Rigoulet et al. investigates the stoichiometry of ATP
production in oxidative phosphorylation in isolated yeast
mitochondria, and the same problem is studied in in vivo in
hepatocytes by Leverve et al. and Averet et al. in permeabil-
ized yeast cells (chapters 4 and 5, correspondingly). While
decrease in ATP/O ratio with increase of respiration rate is
characteristic for mitochondria independently from their
localization, many important functional differences between
yeast mitochondria in vivo and in vitro have been observed
(chapter 5). Among them, most important seems to be the
new and rather general phenomenon of very low affinity of
the mitochondrial oxidative phosphorylation for ADP
observed in vivo both in permeabilized yeast (chapter 5)
and described earlier for many types of mammalian cells
(chapter 6). In the latter case, the general phenomenon,
nevertheless, appears to be tissue-specific (chapter 6). All
new evidence confirm that this is due to a very significant
limitation of diffusion of ADP (and it appears that also of
other substrates, such as NADH) through porin pores in the
outer mitochondrial membrane. The participation of porin
channels in this regulation is directly confirmed by Averet et
al. by using porin-deficient mutants of yeast. The conclusion
from these works is that in the cells in vivo outer mito-
chondrial membrane porin channels are controlled, in
tissue-specific manner, by some extramitochondrial protein
structures hypothetically associated with cytoskeleton
(chapter 6). Already existing multiple data on the mito-
chondrial-cytoskeleton interaction are reviewed shortly by
Rappaport and Samuel in chapter 7.
In the next paper, chapter 8 by Devin et al., the in vitro
data are used for analysis of the influence of cell swelling
on the cellular energetics of hepatocytes. This is a rather
unusual, osmotic mechanism of regulation of mito-
chondrial function in vivo, and the data regarding this
mechanism of regulation are still very controversial in the
literature.
Central for the book is the next part describing the
networks of the energy transfer and distribution in the cells
in vivo. It is now becoming generally (although not totally)
accepted that the energy transport is an enzymatic process
of facilitated diffusion and vectorial ligand transduction,
mostly carried out by different kinase systems, starting from
mitochondrial intermembrane space and mitochondrial
outer membrane and ending up with functional coupling with
processes of energy utilization. As usually in our publications,
very strong contribution has been made by T. Wallimann's
laboratory. The review by Schlattner et al. presents recent
progress in the studies of the structure of chicken heart
mitochondrial creatine kinase (mi-CK) - the first creatine
kinase for which X-ray structure was established. These new
principal results of analysis of the X-ray structure ofmi-CK
obtained in collaboration with Fritz-Wolff are discussed in
5
relation to the molecular aspects of its functional coupling
with adenine nucleotide translocase (ANT). O. Stachowiak
et al. describe the oligomeric state and membrane binding
of mitochondrial creatine kinase as the basis of functional
coupling and aerobic phosphocreatine production in mito-
chondria. Both these works are important contributions to
our understanding of cellular bioenergetics - the physio-
logical process of aerobic phosphocreatine production is
first time discussed in its molecular dimension on very
firm structural basis. This is also the molecular description
of the most important site of the regulation of mito-
chondrial function in vivo by local ADP produced by
mi-CK. Reading these chapters we may feel ourselves as
Maxwell's demon looking at all those molecules ofATP and
ADP coming in and out of microcompartment between
mi-CK and ANT.
Detailed description of molecular biology of the com-
ponents ofthe creatine kinase systems - different isoenzymes
of the creatine kinase, the historical aspects of discoveries
of their gene structure and further perspectives of research
in this area are given in this volume by Qin et al. from Arnold
Strauss laboratory in St. Louis, USA. One of important results
of these fundamental studies of molecular genetics of creatine
kinases is the description of co-expression of different
creatine kinase isoenzymes in heart cells to build up the
phosphocreatine pathway for energy transfer. In this volume,
these ideas are given quantitative, mathematical description
later in chapter 15, which shows coordinated non-equilibrium
functioning of such coexpressed creatine kinase isoenzymes
within a PCr pathway.
The creatine kinase system is not, however, the only one
in the energy transfer process in the cells. The review by Dzeja
et al. of experimental data from Goldberger's laboratory
given in chapter 12 brings to us convincing evidence that
similar function is carried out by the adenyl ate kinase
system, in close and finely orchestrated interaction with the
creatine kinase system. In the case of failure of the latter,
and also in the cells lacking the creatine kinase system such
as liver cells, the adenyl ate kinase system becomes the main
energy transport system. Similar function can be played by
other kinases, such as mono- and dinucleotide phospho-
kinases, and increasing amount of evidence points also to
the possibly important role of the structurally organized
glycolytic system (associated with the cytoskeleton) in the
phoshoryl and energy transfer. Thus, there is indeed a whole
network for energy transfer in the cells, a network of
interchangeable pathways built up on the principles of
vectorial ligand transduction. And it seems that this network
is perfectly organized, with well functioning 'infrastructure'
which allows to transfer the energy in the best possible,
under given conditions, way. The mechanisms of this kind
offine regulation between different pathways of the network
is unknown (see chapter 12).
6
An important, fascinating approach for the studies of
these systems in the cells in vivo is the use of transgenic
technology. Three following chapters (chapters 1-3, 14
and 15) discuss different aspects of the use of this powerful
approach for the studies of importance of creatine kinase
compartmentation in the cell. Description of cytoarchi-
tectural and metabolic adaptations in muscles of mice with
knocked-out CK genes is given by Steeghs et al. from Be
Wieringa's laboratory from The Netherlands. These
adaptations should be accounted for in interpretation ofthe
results of any kind of experiments with the transgenic
animals. Klaas Nicolay et al. give a detailed account of in
situ measurements of creatine kinase flux by NMR in the
heart and skeletal muscle of the transgenic mice. These
studies revealed several new phenomena, such as depen-
dence of the creatine kinase flux on isoenzyme localization
and degree of its expression, and different workload
dependencies of the flux through different creatine kinase
isoenzymes. None of these observation can be explained
by conventional, textbook equilibrium theories of homo-
genous creatine kinase systems in the cells. On the contrary,
it is the mathematical theory of compartmentalized energy
transfer system which quantitatively explains the different
workload dependencies of distinct creatine kinase iso-
enzymes. This mathematical model was constructed by M.
Aliev and V. Saks on the basis of the experimental data on
creatine kinase compartmentalization and functional coupling,
obtained in studies of both isolated mitochondria and
permeabilized cardiac cells, and in chapter 15 it is used
jointly with Klaas Nicolay group for analysis of the
experimental data on hearts isolated from transgenic mice.
Thus, here we join two new powerful methods of research.
This analysis shows that in the intact heart muscle cells
creatine kinase isoenzymes function in steady state out of
equilibrium in perfectly coordinated manner to transfer
almost all energy between sites of its production and
utilization (phosphocreatine pathway, or circuit). When left
alone by genetic manipulations, mitochondrial creatine
kinase starts to catalyze futile equilibrium reaction of ATP
production and utilization in mitochondria which is not
dependent on the workload, and energy transfer is taken over
by other enzyme systems. Thus, the mathematical model of
compartmentalized energy transfer appears to be effective
and practical tool of research. All these results demonstrate,
in fact, the importance of compartmentation phenomenon
in the cellular bioenergetics. Comparative experimental
study of heart and skeletal muscle tissues in different animal
species shows that functional coupling of different CK
isoenzymes is characteristic for highly organized adult
mammalian muscle cells and was specialized during evolu-
tion, as it is described by Ventura-Clapier et al. in chapter 16.
All these results show high degree of variability and
plasticity of energy transfer networks in the cells.
General analyses of quantitative methods and mathe-
matical models of different degree of complexity for
studies and simulation of energy transfer in heart and
skeletal muscle metabolism are given by Kemp et al. in
chapter 17, and also, with some historical aspects, by V.
Saks in chapter 18.
The next part of this volume deals with the problems of
intracellular metabolic control and regulation. A theoretical
study by Kholodenko et al. from Hans Westerhoff labora-
tory shows how the Metabolic Control Analysis can be
extended to account for metabolic channelling and enzyme
organization, including the effects of macromolecular
crowding, this leading to completely new regulatory
properties of pathways involving protein-protein interactions.
One of very important aspects of the functioning of energy
transfer systems is the feedback metabolic signaling in cells
- in fact, this is a central question in cellular bioenergetics:
how mitochondrial respiration rate is regulated in vivo in
response to the increased workload. This question is
directly addressed by Van Beek et al. by using sophisticated
measurements of pre-steady state kinetics of mitochondrial
respiration (response times) and high time resolution NMR
for determination of high energy phosphates in isolated
heart. The conclusion from this very precise, quantitative
experimental work is that the metabolic signal propagates
with time from one cellular compartment to another with
participation of organized cytoplasmic structures. This
conclusion is in very good concord with many other studies
discussed above in this volume. The authors discuss also
the possible participation of changes in intracellular
calcium concentrations in regulation of mitochondrial
respiration via activation of Krebs cycle dehydrogenases.
This important question is analyzed first theoretically by
Korzeniewsky in chapter 21, and then in chapter 22 one of
pioneers, riow classics of these studies, Richard Hansford,
carefully reviews, in collaboration with Dmitry Zorov, all
recent developments in this rapidly changing area of cellular
bioenergetics. The very reasonable conclusion from this
work is that both metabolic (via local changes in ADP
concentration) and calcium-mediated signaling should be
accounted for in explaining in vivo regulation of mito-
chondrial respiration.
The very important problem of calcium regulation of
mitochondrial function is further investigated by Jouaville,
Ichas and Mazat in chapter 23. They describe the other side
of this story: how mitochondria, by calcium-induced
calcium release, that takes the form of calcium spike may
play an active role in cell calcium signaling and in calcium
wave propagation.
Increasingly important part of cellular bioenergetics is
investigation of mitochondrial function in pathological
conditions. Mitochondria playa central role in many
pathologies. Preservation of all mitochondrial functions is

vitally important for tissue preservation and for avoiding
irreversible damages in ischemia, hypoxia and reperfusion,
but near and after 'no return point' mitochondria may play
a deleterious role in accelerating fatal events. Recent
publications by Yang et al. and Kluck et al. in Science, vol.
275, 1129-1135, 1997 tell us even that by releasing
cytochrome c mitochondria may even participate in initiation
of apoptosis. Detailed account of responses of mito-
chondria to ischemia and of the role of mitochondrial
permeability transition pores in this process is given by
Fabio Di Lisa and Paolo Bernardi in chapter 24. In next
chapter Hassinen et al. review the data on participation of
mitochondrial systems in ischemic preconditioning - a
phenomenon which still remains slightly mysterious
because of the lack of our understanding of its mechanism.
Rossi et al. consider in chapter 26 the very recent informa-
tion on alteration of mitochondrial outer membrane and
creatine kinase systems in the ischemic heart cells and
conclude that the most early ischemic damage is observed
in a systems of feedback metabolic signaling, a conclusion
which is in good concord with those made by Van Beek et
al. in chapter 20. And in chapter 27 Letellier et al. from J.-P.
Mazat laboratory give us good instructions how to use
I MECHANICAL ACTIVITY
(MYOFILAMENTS)
18
myosin ATPase
CHANNELLING
- anabolic reactions
- reserves
IANABOLIC REACTIONS 18
Fig. 1. Distribution of energy from mitochondria into cytoplasm for various processes of energy utilization in myocytes. The processes of energy utilization
are arbitrarily divided into four sectors. For each of them the estimate of the percentage of energy used in a given process is shown. The pathways for energy
transfer making up the energy transfer network are shown as 'Routes'.
7
metabolic control analyses for studies of mitochondrial
pathologies.
Two next chapters, by Seppet et al. (chapter 28) and by
Guerreros-Ontiveros and Wallimann (chapter 29) describe the
experimental data on thyroid hormone control in atria and
creatine supplementation on muscle, respectively. In both
cases cellular energy metabolism is changed. Interestingly,
there are abundant data showing the benefit of creatine
supplementation, such as increase in endurance and force
development but the cellular mechanism of these effects are
not clear. Some of them are considered in the paper by
Guerreros-Ontiveros and Wallimann who report that chronic
supplementation of creatine down regulates the expression of
the creatine transporter. However, this observation is still not
sufficient to explain the effects of exogenous creatine supp-
lementation.
Finally, Neubauer et al. demonstrate in chapter 30 how the
knowledge of bioenergetics of the cell can be effectively used
for analysis of human pathologies in cardiological clinics with
application of magnetic resonance spectroscopy. The volume
is closed by B. Chance, one offounders of quantitative studies
of bioenergetics, who describes, with his young colleague B.
Beauvoit, how a new method of time-resolved infrared
IONIC (Ca++) MOVEMENTS
- SARCOLEMMA
- RETICULUM
- Ca++ ATPases
- others
/
?
- cyciases
- G proteins
SIGNALLING 18
8

1 REGULATORS I MODULATORS I
metabolic signals related to
, energy transfer related to the integrated activity of the cell

, ADP Cr
,
(Pi)
, AMP
various [Ca++j "Second messengers"
, UDP
,,
kinases
, GDP
/ c~ /
NADH
,
I "' /
NAD
, CHANNELS cascades
CHANNELLING ~
cellular ~ \(metaboIiC protein
redox state waves ?) kinases
./ inter-correlation
metabolic signals SHUTILES ADP , with others cells
related to energy ~
metabolism

- substrates sarcolemma
-2 DIFFUSION myofilaments
- Pi
- creatine
- adenylate pools

MECHANICAL
INTEGRATORS

- free radical - excess calcium

- osmolar load - participation in apoptosis
'-1 -PA-T-H-O-P-H-Y-SI-=-O-L-O-G-IC-A-L-F-A-CT-O-R-S~I

Fig. 2. Schematic presentation of the putative factors of regulation of mitochondrial activity in the cells in vivo. The putative factors of regulation of
mitochondrial function are arbitrarily divided into five categories shown. The mechanisms of signaling to mitochondria are indicated in the central part.

spectroscopy can by used for estimation ofthe mitochondrial
contents in the cells in vivo.
Conclusions
So what do we not know?
All the papers presented in this issue show the complexity
of mitochondrial regulation in the cells. It is very clear that
in the cell in vivo mitochondria are not alone with their
substrates and products in homogenous solution, as they are
in the oxygraphic cell. Instead, they are influenced by their
surroundings and multiple intracellular regulatory factors.
We have tried to summarize and illustrate the information
given in this volume on mitochondrial function and its
regulation in the cells in vivo in two following Figs. As
producers of energy, mitochondria have the key role in the
distribution of energy to many intracellular processes to
ATP (GTP, UTP, CTP) consuming sites. Figure 1 shows
schematically the distribution of energy from mitochondria
to the various ATP consuming systems in muscle cells.
Regarding energy production in mitochondria and its
distribution in the cells, we still do not know:
Putative and not as yet described mechanisms of energy
transduction: if the chemiosmotic coupling is mainly due
to proton movements through proton pumps, in many
species of marine aerobic and anaerobic eubacteria and
archaebacteria sodium cycle is responsible for energy
conversion through sodium pumps. It can be expected that
in the nature other kinds of energy transduction may exist

using different ions alone or alternatively with protons
or other cations in response to strong environmental
constraints. Studies of such new chemiosmotic coupling
mechanisms could improve our understanding of the
principles of energy transduction.
The transfer of energy from mitochondria to myofilaments
and subcellular membranes through the PCr and/or
myokinase shuttles through vectorial ligand conduction
are the most studied systems of energy channelling until
now, but the interaction between these two and other
possible systems of energy transfer should be clarified
in further studies. The structural organization of these
energy transfer networks, their relationship to cyto-
skeleton needs further very vigorous studies. The role of
the outer mitochondrial membrane porin pores in energy
distribution between different pathways is still not
precisely described. For understanding metabolic feed-
back-signaling, the nature of factors controlling outer
mitochondrial membrane permeability for ADP should
also be clarified.
Evidences have been recently accumulating for a similar
channelling in the intracellular structures responsible for
calcium homeostasis. Interaction between the systems of
calcium and energy channelling needs further studies.
This comprises bound kinases, functional coupling and
participation of the so-called glycolytic ATP due to
compartmentalized glycolytic systems.
In a similar way we can propose that in the subsarcolemmal
compartments various cyclases, G-proteins, protein
kinase, phosphodiesterases are coupled to different
phosphokinases (CK, AK, NDPK, GK ..... ) in order to
supply energy to the hormonal signaling pathways. And
finally, some part of energy is used for anabolic reactions.
These processes also need further study.
In tum, all processes of free energy utilization influence
mitochondrial function by intracellular feedback mechan-
isms. This gives rise to multiple regulatory factors,
tentatively classified in Fig. 2. Which ofthem represents
a real feedback signal, most important for regulation of
the rate of oxidative phosphorylation, is unknown. One of
plausible mechanism for feedback regulation is metabolic
9
signaling based on the vectorial ligand conduction pro-
posed by Michell. This most probably gives rise to
metabolic wave propagation (analogous to that already
described for calcium). Thus, theoretical and experimental
studies of the implication of vectorial ligand conduction
in cellular energy transfer are necessary.
Also, the next problems of general interest are not solved
as yet: molecular and genetic mechanisms of tissue
specificity of energy transfer pathways and mitochondrial
function; structure-function analysis of enzyme or protein
interactions in multi enzyme complexes, contact sites,
macromolecular arrangement, cytoskeletallmitochondrial
interactions.
Finally, implications of deficiency in energy transfer
systems in pathogenesis and cellular diseases, including
apoptosis, myopathies and contractile failure are becoming
increasingly important.
These are only some of the important problems of bio-
energetics of the cell. They are listed here to reflect the
beliefs and interests of authors and editors of this volume.
But even this list shows that we still cannot go out of the
laboratory, close the door and say that we have done our job.
As usual, all questions which have been solved will mean
more questions to answer in the future.
VALDURA. SAKS
Laboratories of Bioenergetics, Institute of Chemical and
Biological Physics, Tallinn, Estonia and Joseph Fourier
University, Grenoble, France
RENEE VENTURA-CLAPIER
Cellular and Molecular Cardiology, U446 INSERM
Chatenay-Malabry, France
XAVIERLEVERVE and ANDRE ROSSI
Laboratory of Bioenergetics, Joseph Fourier
University, Grenoble, France
MICHELRIGOULET
Laboratory of Bioenergetics, Institute of Biochemistry
and Cellular Genetics, CNRS, Bordeaux, France
 PART I

BIOENERGETICS OF MITOCHONDRIA:
IN VITRO AND IN VIVO STUDIES
Molecular and Cellular Biochemistry 184: 13-20, 1998.
1998 Kluwer Academic Publishers.

Top-down elasticity analysis and its application to

energy metabolism in isolated mitochondria and
intact cells
Martin D. Brand
Department of Biochemistry, University of Cambl'idge, Tennis Court Road, Cambridge, CB2 lQW, UK

Abstract
This paper reviews top-down elasticity analysis, which is a subset of metabolic control analysis. Top-down elasticity analysis
provides a systematic yet simple experimental method to identify all the primary sites of action of an effector in complex
systems and to distinguish them from all the secondary, indirect, sites of action. In the top-down approach, the complex system
(for example, a mitochondrion, cell, organ or organism) is first conceptually divided into a small number of blocks of reactions
interconnected by one or more metabolic intermediates. By changing the concentration of one intermediate when all others
are held constant and measuring the fluxes through each block of reactions, the overall kinetic response of each block to
each intermediate can be established. The concentrations of intermediates can be changed by adding new branches to the
system or by manipulating the activities of blocks of reactions whose kinetics are not under investigation. To determine how
much an effector alters the overall kinetics of a block of reactions, the overall kinetic response of the block to the intermediate
is remeasured in the presence of the effector. Blocks that contain significant primary sites of action will display altered kinetics;
blocks that change rate only because of secondary alterations in the concentrations of other metabolites will not. If desired,
this elasticity analysis can be repeated with the primary target blocks subdivided into simpler blocks so that the primary sites
of action can be defined with more and more precision until, with sufficient subdivision, they are mapped onto individual
kinetic steps. Top-down elasticity analysis has been used to identify the targets of effectors of oxygen consumption in
mitochondria, hepatocytes and thymocytes. Effectors include poisons such as cadmium and hormones such as tri-
iodothyronine. However, the method is more general than this; in principle it can be applied to any metabolic or other
steady-state system. (Mol Cell Biochem 184: 13-20, 1998)

Key words: control analysis, top-down elasticity analysis, enzyme kinetics, energy metabolism, mitochondria, oxidative
phosphorylation

Introduction Traditional methods of discovering the primary targets

Externally added molecules that modify the properties of a
system are referred to as effectors. Examples are inhibitors
or hormones. Physical influences such as temperature or
pressure can also be classed as effectors, as can less tangible
influences such as evolutionary history. Effectors change the
steady states of biochemical systems by interacting directly
with a finite number of primary targets within them. The
more specific the effector, the fewer the number of primary
sites of action. The rest of the system responds to the primary
changes through secondary alterations in the concentrations
of internal metabolites.
of an effector are rather hit-and-miss. Usually, isolated
enzymes are treated with a chemical or physical effector to
see if it changes their activity. Sometimes the choice of
target to test is guided by observations in the whole system,
for example from crossover experiments where a target is
suspected if upstream and downstream metabolite con-
centrations change in opposite ways when the effector is
added. Sometimes a general screen is used in which a
standard range of different possible targets are tested
routinely. Once a primary target is identified, the specificity
of the effector for this target often remains in doubt until
many other potential targets have been screened. When a

Address for offprints: M.D. Brand, Department of Biochemistry, University of Cambridge, Tennis Court Road, Cambridge, CB2 1QW, UK
14
second primary target is discovered, it is usually very
difficult to untangle the two primary effects to see which is
responsible for a given change in system behaviour, so
multi-site inhibitors are generally considered to be less
useful than highly specific ones, and hormones with several
primary effects are more difficult to analyse than those with
a single receptor or signal transduction pathway.
An alternative strategy is to examine the behaviour of a
whole system when an effector is added, and to work down
into the system to uncover primary sites of action. This
'top-down' approach can be very simple to apply but can
identify primary targets and provide specific information
about their relative importance. This review outlines top-down
elasticity analysis [1-11] and discusses some examples ofits
application. Top-down elasticity analysis is related to
top-down control analysis and top-down regulation analysis,
which are reviewed elsewhere [11, 12], and is often carried
out as an integral part of such a control analysis.
Measurement of block kinetics in steady states
Elasticity analysis has been applied to steady states, in which
the rates of reactions and the concentrations of metabolic
intermediates are effectively constant. The properties of the
steady state can be exploited to allow measurement of the
overall kinetics of blocks of reactions. Consider a very
simple metabolic system made up of two reactions that
convert a fixed concentration of external metabolite X to a
fixed concentration of external metabolite Y, through a
common intermediate, metabolite M (Fig. 1). For all
ordinary kinetic schemes, the system will evolve in a few
minutes from any starting condition to a final steady state
in which the rate of production of M by the supply reaction
equals its rate of consumption by the demand reaction. The
system will then have a steady rate of conversion of X to Y
and a stable concentration of M. Figure 1 shows arbitrary
diagrams of the rates of the two reactions as a function of
the concentration of M. The demand reaction will normally
go faster as the concentration of M, its substrate, rises and
the supply reaction will normally go slower as the con-
centration of M, its product, rises. However, the exact form
of the curves is not important for the analysis, and it makes
no difference in principle if the kinetics of the enzymes are
quite different from those shown in Fig. 1. The steady state
occurs when the concentration of M evolves to the value
where the supply rate exactly equals the demand rate; this is
what happens at the intersection ofthe lines in Fig. 1. Nearby
states return to this steady state because any rise in the
concentration of the intermediate tends to inhibit its own
manufacture by product inhibition and increase its own
depletion by substrate stimulation, so opposing the original
perturbation. Conversely, any drop in the concentration of
the intermediate tends to stimulate its manufacture and
decrease its depletion and the system moves back towards
the steady state. In this way the variables in the system (the
concentration of M and the rates of M supply and demand)
adjust to maintain the steady state that is set by the parameters
(the kinetics of the two reactions, the fixed concentrations
of X and Yand the prevailing conditions of pH, temperature
and all other external effectors).
It is possible to exploit this steady state behaviour and use
it to measure the kinetic response of each reaction to the
concentration ofM. Imagine that we make a parameter change
that alters the relationship between the rate of the supply
reaction and the concentration of M. For example, we could
raise or lower the concentration of the supply enzyme, or treat
it with an activator or with a specific inhibitor that lowers its
V max or decreases its affinity for M. If we inhibit, the
system will evolve to a new steady state with a decreased
concentration of M and decreased rates of both supply and
demand (Fig. 2). The new steady state will not lie on the line
describing the original kinetic response of the supply reaction
to M, because we have interfered with the kinetics of this
reaction. It will, however, lie on the line describing the original
kinetic response of the demand reaction to M, because we
have not altered the kinetics of this reaction, but only changed
the concentration of its substrate. By progressively inhibiting
the supply reaction and measuring the flux through the system
and the concentration ofM, we can observe successive steady
states that map out the kinetic response of the demand reaction
toM as shown in Fig. 2. This strategy gives us a simple way to
measure the kinetic response of the demand reaction to the
concentration of M, in a complex system in situ, whatever
form the kinetic response may take. We can use the same
strategy to measure the kinetic response of the supply
reaction. In a separate series of experiments we can
successively alter the kinetics of the demand reaction and map
out the kinetic response of the supply reaction to M as shown
in Fig. 3. An alternative strategy is to add a new reaction that
consumes or produces M, and independently measure the
steady-state rates of the two original reactions as a function
of the concentration of M. Either strategy gives a full kinetic
description of the two reactions in the steady state from
simple measurements of rates and concentrations.
The kinetic responses in schemes with more than two
reactions and more than one metabolic intermediate can be
measured in the same sort of way. If three or more enzymes
interact with a single intermediate, we simply measure their
rates and plot out the kinetics of each reaction at different
concentrations of the intermediate. The intermediate can
be changed by inserting a new branch into the scheme.
Alternatively, it can be changed by altering one reaction
(Reaction i) and determining the kinetics of the others, then
altering a second reaction and determining the kinetics of
Reaction i.
 15

supply demand
reaction reaction
x ----il.~ M y
.9
><
'0
c:::
o
"~
Q)
>
c:::
8
'0
Q) Steady state
~

demand reaction supply reaction

concentration of M

Fig. 1. The steady state in a simple system consisting of two reactions (the supply reaction and the demand reaction) that share a common internal metabolite
M that can vary in concentration. The system is provided with fixed concentrations of external substrate X and external product Y. The graph shows arbitrary
relationships between the rates of the supply and demand reactions as functions of the concentration of M. Where the lines intersect there is a unique stable
steady state where supply equals demand. This steady state can only be varied by altering the parameters of the system.

Progressive inhibition of the supply reaction

I
:>-

>-<
0

- 0
c:::

,
0
"iii
....
Q)
> I
c:::

-
0
0

-
0
Q)

....
CO

supply reaction

concentration of M
Fig. 2. Measurement of the kinetic response of the demand reaction to the concentration ofM. The kinetic parameters of the supply reaction are changed (in
this example, by adding a specific inhibitor). The system evolves to a new steady state that lies at a different point on the original line describing the kinetic
response of the demand reaction to M. By progressively inhibiting supply through successive steady states, a series of values of rate and [M] are obtained
that fully describe the kinetics of the demand reaction in the range investigated. The lines are also an example of an elasticity analysis to show that the inhibitor
has all its primary effects on the supply reaction, with the effects on the rate of the demand reactions entirely due to secondary changes in response to
changes in the concentration of M.
16

Progressive inhibition
of the demand reaction
I

demand reaction supply reaction

concentration of M

Fig. 3. Measurement of the kinetic response of the supply reaction to the concentration ofM. The kinetic parameters of the demand reaction are changed (by
adding a specific inhibitor). The system evolves to a new steady state that lies at a different point on the original line describing the kinetic response of the
supply reaction to M. By progressively inhibiting demand through successive steady states, a series of values of rate and [M] are obtained that fully describe
the kinetics of the supply reaction in the range investigated. The lines are also an example of an elasticity analysis to show that the inhibitor has all its primary
effects on the demand reaction, with the effects on the rate of the supply reactions entirely due to secondary changes in response to changes in the
concentration of M.

If there are two intermediates,Ml andM2, we can clamp
the concentration of M2 in experiments to measure the
kinetic responses to Ml , and vice versa. If we cannot clamp
the concentrations then we can perform double titrations
and analyse the rate of an enzyme at various concentrations
ofMl when, by titration of a different step, we have brought
the concentration of M2 back to its original value. For an
example of this approach used to measure the kinetic
response of different parts of a system to an external
effector at a single concentration of an internal metabolite,
see [13]. As the number of intermediates in the system
increases, the number of such secondary titrations increases
too, so only relatively simple systems can be analysed in
this way.
We can deal with more complex, real pathways by simplify-
ing them until they are experimentally accessible. Because
we empirically measure the overall kinetic response of a
reaction to M with all other effectors kept constant, the
reactions can have any type of kinetics without affecting the
analysis. So we can group several reactions together into a
larger block of reactions and carry out the analysis as if they
were a single (but complicated) enzyme. Grouping reactions
together in this way is fundamental to top-down metabolic
control analysis, which simplifies complex pathways and
makes them more amenable to experimental application of
control analysis [11]. In fact, a block of reactions can be
of any complexity, containing enzymes, transporters,
non-enzymatic steps, many intermediates, compartments,
feedback loops and allosteric interactions. There are some
rules to be observed. Firstly, there must be no intermediates
within one block of reactions that have any direct effects
on another block; any such intermediates must be con-
sidered explicitly [14]. Secondly, when inhibitors or
activators are used to titrate steady states they must have
sites of action (of known or unknown type) that are
confined to blocks whose kinetics are not being measured
in that experiment [15].
Graphs of the sort shown in Figs 1-3 can be used to
measure the elasticities of blocks of reactions to inter-
mediates - hence the name 'top-down elasticity analysis'
[8]. An elasticity is defined as the fractional change in rate
of a reaction in response to an infinitesimal fractional
change in the concentration of an effector (for example,
its substrate, or an allosteric activator), with all other
effectors held constant. The elasticities of the two reactions
to M are given by the normalised slopes of the lines in Fig.
1 at each concentration of M. Elasticities need not be
calculated explicitly from the raw data when carrying out
an elasticity analysis, but they usually are, because they can
be used for the calculation of control coefficients and
partial response coefficients in control and regulation
analysis [11, 12].

Top-down elasticity analysis
The method discussed above provides a simple experimental
way to measure the kinetic responses of blocks of reactions
embedded in a complex system to the concentrations of one
or more intermediates within the system. This method can
be expanded to discover the primary site or sites of action
of an effector that interacts with the system - we have
dubbed this approach 'top-down elasticity analysis' [8, 9],
and it has also been described as 'phenomenological kinetics'
[16]. By comparing the kinetic curves for the different
reaction blocks in the presence and absence of the effector
it is a simple matter to identify these primary sites. Blocks
of reactions that have a changed kinetic response to one
or more of the intermediates are primary targets of the
effector. Blocks of reactions that do not have a changed
kinetic response to one or more of the intermediates
change rate when the effector is added only as a secondary
consequence of the transmission of signals within the system
by changes in the system variables and are not primary
targets. For example, Fig. 2 demonstrates that the inhibitor
that was used acts only on the supply reactions and not on
the demand reactions, because the supply block has altered
kinetics but the demand block does not. Conversely, Fig. 3
demonstrates that the sole primary effect of the inhibitor
used in this Figure is on the demand reactions, even though
both reactions have a new rate in each new steady state.
Elasticity analysis is the method of choice for the identifica-
tion of the sites of action of effectors in complex systems.
As discussed by Kesseler and Brand [17] and Kholodenko
and Brown [18], conventional control analysis is unsuitable
for this purpose: flux control coefficients do not necessarily
decrease when a step is stimulated or increase when it is
inhibited, and cannot be used as a reliable indicator of the
primary target of an effector.
When there is more than one primary site of effector
action in a system, top-down elasticity analysis can be used
to compare the magnitudes of the effects on different blocks.
The relative sizes of the different primary effects can easily
be established from the kinetic plots by comparing the
fractional changes in the rates of the blocks at the original
concentrations of all the internal effectors in the system:
one block may increase twofold whereas another may
increase only 5% when the effector is present. For an
example of this use of elasticity analysis, see [19-22].
It is possible to conduct an elasticity analysis without full
measurements of the kinetic curves, although this has not yet
been done. Measurement of system fluxes and intermediate
concentrations for a set of at least n independent perturba-
tions of the steady state in a system with n intermediates
yields a set of simultaneous equations that can be solved to
give the elasticities of each block to each intermediate [14,
23]. The changes in flux through each block and the new
17
concentrations of each of the intermediates can be measured
when an effector is added. The predicted flux through each
block can be calculated for the new metabolite con-
centrations using the original elasticities. Comparison of
these predicted fluxes with the observed fluxes identifies
blocks that contain a primary site of action of the effector:
blocks with primary sites of action will have rates that do
not fit the prediction but the fluxes through blocks that
change rate as a secondary result of metabolite changes will
be predicted correctly.
Top-down elasticity analysis of energy metabolism
Top-down elasticity analysis has been used extensively to
identify the sites of action of effectors of energy metabolism.
Most common are analyses of the targets of effectors of
oxygen consumption in mitochondria and hepatocytes, but
there are also examples using other cell types such as
thymocytes. Effectors include inhibitors, poisons such as
cadmium and hormones such as tri-iodothyronine. But the
method is more general than this: in principle it can be
applied to any metabolic or other steady-state system that
is of interest.
In the simplest cases, the whole metabolism of the system
(mitochondria or cell, etc.) is divided into two blocks: those
reactions that produce mitochondrial membrane potential
(~"', the major component of the proton motive force),
namely substrate transport and mobilisation, catabolism and
electron transport, and those that consume it, namely
proton leak reactions, ATP synthesis and all ATP consuming
reactions. For any external effector that changes the rate of
oxygen consumption by the cells, measurement of the
kinetics of the ~",-producers and the ~",-consumers shows
whether the effector works upstream or downstream of ~'"
(or both). By progressively re-dividing affected blocks we
can work from the top level down into the system to each
target step where the effector has a primary site of action.
Often, the ~",-consumers are subdivided into those that lead
to ATP turnover and those that do not (proton leak reactions),
to give a three-block system branched around ~"'.
The kinetics of the proton leak reaction can be assayed in
isolated mitochondria by plotting the relationship between
respiration rate and proton motive force when the system is
titrated with an inhibitor of substrate oxidation in the
presence of oligomycin [24, 25]. This corresponds to the
titration shown in Fig. 2. Respiration rate is a measure of
the rate of outward proton pumping by the electron transport
chain, and, in the presence of oligomycin to inhibit ATP
synthesis, all of this proton flux returns to the matrix by leak
reactions, so it is a measure of the proton leak flux. The
pumping and leak reactions are connected through the shared
intermediate, proton motive force. Proton motive force is
18
a thermodynamic quantity describing the electrochemical
potential for protons across the mitochondrial inner
membrane, but it can be treated just like a more conventional
metabolite within control analysis [26]. Top-down elasticity
analysis arose from the realisation by Hafner [1, 2] that if
this assay was repeated in the presence of an effector of the
leak, a different curve would be generated. In this way,
changes in leak flux secondary to changes in proton motive
force could be distinguished from primary changes caused by
direct effects on the proton permeability of the membrane.
Hafner and co-workers [1, 2] showed that liver mito-
chondria isolated from hypothyroid rats had decreased
proton permeability compared to mitochondria isolated
from euthyroid control rats, and liver mitochondria from
hyperthyroid rats had increased proton permeability. This is
an example of the use of top-down elasticity analysis to
identify a primary target of an effector, with thyroid hormone
status of the rats as the external effector.
Subsequent elasticity analysis [3, 8] showed that the
activity of the ATP producing reactions was also decreased
in liver mitochondria from hypothyroid rats, but that any
changes in the substrate oxidation reactions were too small
to measure, and made no significant contribution to the
changed respiration rate. Even though traditional approaches
had shown that there are quite large changes in cytochrome
content and in the activities of the electron transport chain
enzymes in hypothyroidism, the elasticity analysis showed
that these changes could be disregarded as explanations for
the system-level effects of hypothyroidism on respiration
rate under the conditions examined in [3]. This illustrates a
powerful use of top-down elasticity analysis: it can eliminate
a whole group of reactions as significant primary targets of
an effector. In liver mitochondria from hyperthyroid rats
there were no significant effects on the ATP-producing
reactions, but substrate oxidation was stimulated [20].
The relative sizes of the primary effects of thyroid status
on different blocks of reactions can be determined from the
kinetic plots. In liver mitochondria from hypothyroid rats,
proton leak and ATP turnover were both inhibited by about
50% at a given membrane potential whereas substrate
oxidation was unaffected [8]. In mitochondria from hyper-
thyroid rats, proton leak was stimulated about 2.5-fold, but
substrate oxidation was stimulated a maximum of 1.3-fold
and ATP turnover was unaffected [9]. This example illustrates
the use of top-down elasticity analysis to quantify the
relative sizes of the primary effects of an agent.
Elasticity analysis was then used to show that the proton
leak changes caused by thyroid hormone status were not
observed in proteoliposomes containing liver cytochrome
oxidase from animals of different thyroid status [27], so they
were not caused by a leak or slip reaction associated with
this enzyme. However, the effects could be observed in
intact hepatocytes [5, 9,19-22], so they were not an artefact
of mitochondrial isolation. From the kinetic curves obtained
in hepatocytes it was possible to calculate the extent to
which primary plus secondary changes in the rates of each
block of reactions contributed to the overall change in
respiration rate caused by thyroid hormone status. In this
case, unlike the analysis in the previous paragraph, there
was no correction for the change in the activity of the
intermediate, so secondary changes in rate were not
excluded. The results showed that the decreased respiration
rates in hypothyroid cells relative to euthyroid cells were
about 50% due to decreases in the proton leak flux and 50%
due to decreases in non-mitochondrial oxygen consumption,
with no decrease in ATP turnover [19-21]. On the other
hand, the increased respiration rates in hyperthyroid cells
compared to euthyroid controls were about 40-50% due to
increases in proton leak flux and 55-60% due to increases
in ATP turnover, whereas changes in non-mitochondrial
oxygen consumption did not contribute to the increase [9,
19]. These examples show how top-down elasticity analysis
can be used to quantify and compare the extent to which
various system fluxes change in response to an effector.
There are many other examples of the application of
top-down elasticity analysis to identify primary sites of
effector action. Other hormones have been investigated:
vasopressin and extracellular ATP stimulate respiration in
hepatocytes through a primary activation of substrate
oxidation, mostly before NADH is reduced [28], and
glucagon treatment of rats causes increased respiration rates
in isolated liver mitochondria solely through primary effects
on the respiratory chain [7]. The glucocorticoid methyl-
prednisolone inhibits substrate oxidation and stimulates
proton leak in thymocytes [29]. Nobes and co-workers [61
used the method to analyse the mechanisms by which fatty
acids activate respiration rate in hepatocytes, and showed
that fatty acids at physiological concentrations stimulate
substrate oxidation (probably by acting as a substrate) and
ATP turnover, but do not uncouple (i.e. do not change the
mitochondrial proton permeability) in situ, although they do
so quite effectively in isolated liver mitochondria [30].
The analysis has been used to examine evolutionary
effects on the properties of isolated mitochondria and
hepatocytes. Reptile liver mitochondria have a lower proton
permeability than mammal liver mitochondria [3 1 ], and liver
mitochondria from mammals with a small body mass have a
greater proton permeability than those from larger mammals
[32, 33]. Elasticity analysis of hepatocytes from mammals
of different body mass showed that the substrate oxidation
system increases in activity with body mass, but proton
permeability in situ and the activity of the ATP turnover
reactions decrease [34]. Top-down elasticity analysis has
quantified the differences in kinetics of substrate oxidation,
proton leak and phosphorylation reactions in mitochondria
isolated from different tissues of the rat [35].

Top-down elasticity analysis has been used to characterise
the effects of inhibitors that act at more than one site in a
complex system. The effects of salicylhydroxamic acid on
respiration in potato tuber mitochondria are partly caused by
stimulation of proton permeability and partly by inhibition of
the cyanide-insensitive alternative oxidase - the two effects
could readily be distinguished using elasticity analysis [36].
The analysis has also been used to investigate mitochondrial
effects of chloroform [37], butylated hydroxy anisole [38],
malonyl CoA [39], supra-physiological calcium [40] and of
cadmium [10, 13, 41] as well as those of physiological and
pathophysiological changes: ischemia [42, 43], diet [44, 45]
and temperature [46, 47].
Conclusions
Top-down elasticity analysis provides a simple and straight-
forward way to discover and quantify how an effector
changes a complicated metabolic system through its
immediate primary effects and its secondary effects. It has
been used with a variety of different types of effector ranging
from pleiotropic inhibitors to hormones and evolutionary
influences. Although it has been used mainly to investigate
effects on energy metabolism in mitochondria and cells, it
is a general method that could be profitably applied to a great
many different types of system.
One intriguing possibility that has been suggested is its
potential for identifying the functions of proteins coded for
by the open reading frames discovered during genome
sequencing [48]: more than half of all yeast ORFs have no
known function and approximately one third are currently
orphans with neither known function nor sequence homology
to any other ORFs [49]. If it proves to be possible to set up
a suitable conceptual division of the whole function of a
yeast cell (or other organism) into a relatively small
number of blocks of functions connected by a few pools
of intermediates, then application of top-down elasticity
analysis would provide a ready way to identify which blocks
of functions were altered as a primary consequence of
altering the expression of the open reading frame, and which
blocks changed only as a secondary consequence of the
primary changes. Once a block was identified as a primary
target, it could be subdivided into smaller blocks to narrow
down the identity of each primary target until it was finally
assigned to one particular function in the cell.
References
1. Hafner RP, Nobes CD, McGown AD, Brand MD: An altered relationship
between proton motive force and respiration rate in liver mitochondria
from hypothyroid rats shows that hypothyroidism decreases the proton
leak (or slip). EBEC Short Rep 5: 128, 1988
19
2. Hafner RP, Nobes CD, MeGown AD, Brand MD: Altered relationship
between proton motive force and respiration rate in non-phosphorylating
liver mitochondria isolatcd from rats of different thyroid hormone status.
Eur 1 Biochem 178: 511-518, 1988
3. Hafner RP, Brown GC, Brand MD: Thyroid-hormone control of state-3
respiration in isolated rat liver mitochondria. Biochem 1 265: 731-734,
lcm
4. Brand MD: The proton leak across the mitochondrial inner membrane.
Biochim Biophys Acta 1018: 128-133, 1990
5. Nobes CD, Brown GC, Olive PN, Brand MD: Non-ohmic proton
conductance of the mitochondrial inner membrane in hepatocytes. 1
Bioi Chern 265: 12903-12909, 1990
6. Nobes CD, Hay WW, Brand MD: The mechanism of stimulation of
respiration by fatty acid in isolated hepatocytes. 1 Bioi Chern 265:
12910--12915,1990
7. Brand MD, D' Allessandri L, Reis HMGPV, Hafner RP: Stimulation of
the electron transport chain in mitochondria isolated from rats treated
with mannoheptulose or glucagon. Arch Biochem Biophys 283: 278-
284,1990
8. Brand MD: Control of oxidative phosphorylation in liver mitochondria
and cells: Top-down control analysis and top-down elasticity analysis.
In: PW Hochachka, PL Lutz, T Sick, M Rosenthal, G van den Thillart
(eds). Surviving Hypoxia: Mechanisms of Control and Adaptation.
CRC Press, Boca Raton, 1993, pp 295-309
9. Harper M-E, Brand MD: Use of top-down elasticity analysis to identify
sites of thyroid hormone-induced thermogenesis. Proc Soc Exp Bioi
Med 208: 228-237, 1995
10. Brand MD, Kesseler A: Control analysis of energy metabolism in
mitochondria. Biochem Soc Trans 23: 371-376, 1995
11. Brand MD: Top down metabolic control analysis. 1 Theor Bioi 182:
351-360,1996
12. Brand MD: Regulation analysis of energy metabolism. 1 Exp Bioi 200:
193-202,1997
13. Kesseler A, Brand MD: Localisation of the sites of action of cadmium
on oxidative phosphorylation in potato tuber mitochondria using
top-down elasticity analysis. Eur 1 Biochem 225: 897-906, 1994
14. Ainscow EK, Brand MD: Top-down control analysis of systems with
more than one common intermediate. Eur 1 Biochem 231: 579-586,
1995
15. Brown GC, Hafner RP, Brand MD: A 'top-down' approach to the
determination of control coefficients in metabolic control theory. Eur 1
Biochem 188: 321-325,1990
16. Brown GC: Phenomenological kinetics and the top-down approach to
metabolic control analysis. In: S Schuster, M Rigoulet, R Ouhabi, 1.-P.
Mazat (eds). Modern Trcnds in Biothermokineties. Plenum Press, New
York, 1994, pp 225-228
17. Kesseler A, Brand MD: Effects of cadmium on the control and internal
regulation of oxidative phosphorylation in potato tuber mitochondria.
Eur 1 Biochem 225: 907-922, 1994
18. Kholodenko BN, Brown GC: Paradoxical control properties of enzymes
within pathways: Can activation cause an enzyme to have increased
control? Biocheml 314: 753-760, 1996
19. Harper M-E, Brand MD: The quantitative contributions of mitochondrial
proton leak and ATP turnover reactions to the changed respiration
rates of hepatocytes from rats of different thyroid status. 1 Bioi Chern
268:14850--14860,1993
20 Harper M-E, Ballantyne IS, Leach M, Brand MD: Effects of thyroid
hormones on oxidative phosphorylation. Biochem Soc Trans 21: 785-
792,1993
21. Harper M-E, Brand MD: Mechanisms of the effects of hypothyroidism
and hyperthyroidism in rats on respiration rate in isolated hepatocytes.
In: S Schuster, M Rigoulet, R Ouhabi, I-P Mazat (eds). Modern Trends
in Biothermokinetics. Plenum Press, New York, 1994, pp 351-356
20
22. Harper M-E, Brand MD: Hyperthyroidism stimulates mitochondrial
proton leak and ATP turnover in rat hepatocytes but does not change
the overall kinetics of substrate oxidation reactions. Can J Physiol
Pharmacol72: 899-908, 1994
23. Ainscow EK, Brand MD: Top-down control analysis of oxidative
phosphorylation around tl'P and ATP in isolated rat hepatocytes. In:
HV Westerhoff, JL Snoep, JE Wijker, FE Sluse, BN Kholodenko (eds).
Biothermokinetics of the Living Cell. BioThermoKinetics Press,
Amsterdam, 1996, pp 75-79
24. Nicholls DG: The influence of respiration and ATP hydrolysis on the
proton-electrochemical gradient across the inner membrane of rat-liver
mitochondria as determined by ion distribution .. Eur J Biochem 50:
305-315,1974
25. Nicholls DG: The effective proton conductance of the inner membrane
of mitochondria from brown adipose tissue. Dependency on proton
electrochemical gradient. Eur J Biochem 77: 349-356, 1977
26 Westerhoff HV, van Darn K: Thermodynamics and Control of Biological
Free Energy Transduction. Elsevier, Amsterdam, 1987
27 Brand MD, Steverding D, Kadenbach B, Stevenson PM, Hafner RP:
The mechanism of the increase in mitochondrial proton permeability
induced by thyroid hormones. Eur J Biochem 206: 775-781, 1992
28. Brown GC, Lakin-Thomas PL, Brand MD: Control of respiration and
oxidative phosphorylation in isolated rat liver cells. Eur J Biochem 192:
355-362,1990
29. Buttgereit F, Grant A, Mtiller M, Brand MD: The effects of methyl-
prednisolone on oxidative phosphorylation in Concanavalin-A-
stimulated thymocytes. Top down elasticity analysis and control
analysis. Eur J Biochem 223: 513-519, 1994
30. Brown GC, Brand MD: On the nature of the mitochondrial proton leak.
Biochim Biophys Acta 1059: 55-62, 1991
31. Brand MD, Couture P, Else PL, Withers KW, Hulbert AJ: Evolution of
energy metabolism: Proton permeability of the inner membrane of liver
mitochondria is greater in a mammal than in a reptile. Biochem J 275:
81--86,1991
32. Porter RK, Brand MD: The contribution of mitochondrial proton leak
to basal metabolic rate: A comparison of the proton leak kinetics in
mammals of different body mass. EBEC Reports 7: 63, 1992
33. Porter RK, Brand MD: Body mass dependence of H+ leak in mito-
chondria and its relevance to metabolic rate. Nature 362: 628-630, 1993
34. Porter RK, Brand MD: Causes of the differences in respiration rate of
hepatocytes from mammals of different body mass. Am J Physiol269:
RI213-RI224,1995
35. Rolfe DFS, Hulbert AJ, Brand MD: Characteristics of mitochondrial
proton leak and control of oxidative phosphorylation in the major
oxygen consuming tissues of the rat. Biochim Biophys Acta 1188:
405-416,1994
36. Kesseler A, Diolez P, Brinkman K, Brand MD: Characterisation of
the control of respiration in potato tuber mitochondria using the
top-down approach of metabolic control analysis. Eur J Biochem
210:775-784,1992
37. Brand MD, Chien L-F, Rolfe DFS: Control of oxidative phosphorylation
in liver mitochondria and hepatocytes. Biochem Soc Trans 21: 757-
762,1993
38. Fusi F, Sgaragli G, Murphy MP: Interaction ofbutylated hydroxyanisole
with mitochondrial oxidative phosphorylation. Biochem Pharm 43:
1203-1208,1992
39. Makins RA, Drynan LF, Zammit VA, Quant PA: Top-down regulation
analyses of palmitoyl-CoA oxidation and ketogenesis in isolated rat
liver mitochondria. Biochem Soc Trans 23: 288S, 1995
40. Mildaziene V, Marcinkeviciute A, Baniene R, Mazat J-P: Supra-
physiological concentration of calcium inhibits phosphorylation,
respiratory subsystem and uncouples rat heart mitochondria. In: HV
Westerhoff, JL Snoep, JE Wijker, FE Sluse, BN Kholodenko (eds).
Biothermokinetics of the Living Cell. BioThermoKinetics Press,
Amsterdam, 1996, pp 168-173
41. Kesseler A, Brand MD: The mechanism of the stimulation of state 4
respiration by cadmium in potato tuber (Solanum tuberosum) mito-
chondria. Plant Physiol Biochem 33: 519-528, 1995
42. Borutaite V, Mildaziene V, Brown GC, Brand MD: Control and kinetic
analysis of ischemia-damaged heart mitochondria: Which parts of the
oxidative phosphorylation system are affected by ischemia? Biochim
BiophysActa 1272: 154-158, 1995
43. Borutaite V, Morkuniene R, Budriunaite A, Brown GC: Phenomeno-
logical kinetic analysis of ischemic damage to heart mitochondria: Role
of Ca'+, fatty acids and cytochrome c. In: HV Westerhoff, JL Snoep, JE
Wijker, FE Sluse, BN Kholodenko (eds). Biotherrnokinetics ofthe Living
Cell. BioThermoKinetics Press, Amsterdam, 1996, pp 89-93
44. Lionetti L, Iossa S, Brand MD, Liverini G: Relationship between
membrane potential and respiration rate in isolated liver mitochondria
from rats fed an energy dense diet. Mol Cell Biochem 158: 133-138,
1996
45. Lionetti L, Iossa S, Brand MD, Liverini G: The mechanism of stimulation
of respiration in isolated hepatocytes from rats fed an energy dense
diet. JNutr Biochem 7: 571-576, 1996
46. Dufour S, Rousse N, Canioni P, Diolez P: Top-down control analysis of
temperature effect on oxidative phosphorylation. Biochem J 314: 743-
751,1996
47. Dufour S, Rousse N, Canioni P, Diolez P: Study of a complex effector of
oxidative phosphorylation using top down elasticity analysis and
control analysis. In: HV Westerhoff, JL Snoep, JE Wijker, FE Sluse, BN
Kholodenko (eds). Biothermokinetics of the Living Cell. BioThermo-
Kinetics Press, Amsterdam, 1996, pp 26-30
48. Oliver SG: From DNA sequence to biological function. Nature 379:
697-600,1996
49. Dujon B: The yeast genome project: What did we learn? Trends Genet
12:263-270,1996
Molecular and Cellular Biochemistry 184: 21-33, 1998.
1998 Kluwer Academic Publishers.

A model ofO-2-generation in the complex III of the

electron transport chain
O.Y. Demin,l B.N. Kholodenkol, 2 and Y.P. Skulachev l
IDepartment of Bioenergetics, A.N. Belozersky Institute of Physico-Chemical Biology, Moscow State University,
Moscow 119899, Russia; 2Department of Microbial Physiology, Free University, De Boelelaan 1087, NL-1081 HV
Amsterdam, The Netherlands

Abstract
Oxidation of semiquinone by O 2 in the Q cycle is known to be one of the sources of superoxide anion (0. 2-) in aerobic cells.
In this paper, such a phenomenon was analyzed using the chemical kinetics model of electron transfer from succinate to
cytochrome c, including coenzyme Q, the complex III non-heme iron protein FeSm and cytochromes bl' bh and c l . Electron
transfers from QH2 to FeSm and cytochrome b l were assumed to occur according to direct transfer mechanism (dynamic
channelling) involving the formation of FeS'~I~ -Q-and Q--cytochrome bI complexes. For oxidation/reduction reactions
involving cytochromes bh and bl' the dependence ofthe equilibrium and elementary rate constants on the membrane potential
(~'I') was taken into consideration. The rate of 0. 2- generation was found to increase dramatically with increase in ~'I' above
the values found in State 3. On the other hand, the rate of cytochrome c reduction decreased sharply at the same values of the
membrane potential. This explains experimental data that the 0. 2- generation at State 4 appears to be very much faster than at
State 3. A mild uncoupling in State 4 can markedly decrease the superoxide generation due to a decrease in ~'I' below the
above mentioned critical level. ~pH appears to be equally effective as ~'I' in stimulation of superoxide production which
depends, in fact, upon the ~~H+ level. (Mol Cell Biochem 184: 21-33,1998)

Key words: superoxide, Q cycle, mathematical model

Introduction using methods of chemical kinetics and metabolic control

Cellular respiration is accompanied by production of
reactive oxygen species (ROS), which are toxic for the living
cells. The survival of aerobic cells depends on a balance
between the production of ROS and functioning of anti-
oxidant systems.
At steady-state conditions, respiring mitochondria may
have different oxygen consumption rates and different
magnitudes of the membrane potential depending on the work
load. At State 4 (no ATP production), oxygen consumption
rate is low, whereas the membrane potential reaches the
highest values. At State 3 (maximalATPproduction), oxygen
consumption rate is high, and the membrane potential is
lower as compared to State 4. In this paper, we shall analyze
the relationships between the rate of oxygen radicals
production and the mitochondrial membrane potential
analysis [1]. It will be shown that the rate ofO' 2-generation
increases dramatically with increase in ~'I' and ~pH above
some critical level.
The Q cycle as a source of oxygen radicals: a kinetic
scheme
The mitochondrial cytochrome bC I complex catalyzes the
transfer of two electrons from ubiquinol to cytochrome c, a
release of two protons on the positive side of the membrane
and uptake of two protons on the negative membrane side. The
overall reaction consists of several steps, and is known as
the Q cycle (shown schematically in Fig. 1). In the scheme
(Fig. 1), the supply of electrons to the Q cycle is shown as
reduction of ubiquinone to ubiquinol accompanying oxidation

Address/or o.f.!Prints: v.P. Skulachev, Department ofBioenergetics,A.N. Belozersky Institute of Physico-Chemical Biology, Moscow State University, Moscow
119899, Russia
22
Fig. 1. Electron transfer through the Q cycle pathway (see the text for
explanations).
of the respiratory substrate, succinate, to fumarate. This
reduction is catalyzed by succinate dehydrogenase and
takes place at the negative side of mitochondrial membrane
(N-side). This is considered as step 1 of the overall reaction.
At step 2, ubiquionol diffuses from the n-side to the
positive outer side of the mitochondrial membrane (P-
side). This process is shown as the (QH2 )n ~ (QH2 )p
transition. Then (QH2)p is oxidised in such a way that one
electron goes to iron-sulfur protein of complex III (FeSnJ
As a result, ubi semiquinone anion (Q.p-) is formed with two
protons (Hp +) released on the positive side of the membrane
(step 3). At step 4, Q'p-reduces the b l heme and ubiquinone
(Qp ) is formed. At step 5, uniquinone diffuses from the p-
side to the n-nside (the Qp ~ Q n transition). At step 6,
electron removed from Q. p - is transferred across the
membrane from low potential (b l ) to high potential (b h )
heme. Then bh reduces ubiquinone (Q) to ubisemiquinone
anion (Q'n-) at then-side (step 7). The Q cycle is completed
by ubi semiquinone dismutation reaction (step 8), in which
two molecules of ubi semiquinone anion (Q'n-) and two
protons (Hn +) at the n-side combine to yield ubiquinone (Q)
and ubiquinol (QH2 )n . Note, that another possible route of
reduction ofQ'n-to (QH2)n can include the electron transfer
from bh (cf. [2]). It should be stressed that dismutation of
Q'p-' in contrast to that ofQ'n-' is forbidden (probably, due to
tight binding ofQ'p- toFeS'~~so Q'p--FeS'~I~' rather than free
Q'p-' is oxidized by bJ Such a restriction is necessary to
organize the cyclic electron flow in the bC I complex [3].
The electron, which has been received by FeSm from
ubiquinol, is transferred to cytochrome c i (step 9), then to
cytochrome c, and eventually to oxygen in cytochrome oxidize
reaction to form water. As mentioned in the introduction,
oxidation of ubi semiquinone anion by O 2in the Q cycle seems
to be one of the major sources of superoxide anions (0'2-) in
aerobic cells (step 10 in Fig. 1 shows the corresponding
reaction supplementing the traditional scheme of the Q
cycle). It should be also stressed that processes of the 0'2-
generation in the Q cycle are pH dependent. This is due to
that both the ubi semiquinone anion and its metabolic
precursor (QH') p which is an intermediate product of the
quinol oxidation at center P
Q.p- + 0 2 ~ Q + O' 2-
(QH)p + 0 2 ~ Q + O' 2-+ Hp +
In this paper, we shall consider only first way to form
0'2-' that is oxidation of ubi semiquinone anion by 0'2-'
Complete description of 0'2-production is an aim of our
future work.
Kinetic modelling approach
The construction of any model requires knowledge of the
kinetic scheme, including all the intermediates considered
and the stoichiometry ofthe corresponding reaction steps. Let
us begin with a simple ('minimal') model that corresponds
to a minimal kinetic scheme of the Q cycle, as depicted in
Fig. 1. This scheme includes free (non-bound) ubi semiquinone
anion at P-side (Q. p-) as a kinetic intermediate. On the other
hand, experimental data suggest that at the P-side ubi-
semiquinone anion is always bound to the proteins of bc I -
complex. This is why we shall then consider a 'channel'
model, in which FeS m and cytochrome b interact directly,
so Q. p-remains bound to one of the proteins (metabolic
channelling). In the channel model, the bulk phase pool of
Q. p- is insignificant (cf. [4D. Another complication consists
in the mechanism of interheme electron transfer. In a
minimal model, the hemes bh and blare independent electron
carriers, which can be in the reduced or in the oxidized
states. In a channel model, the hemes, covalently bound to
the cytochrome b apoprotein, transfer electrons directly
within a single complex, which can be in four different states.
The rate of the electron transfer is proportional to the
concentration (probability) of states, in which the b h heme
is reduced and b l heme is oxidized, rather than to the
mathematical product of the concentrations of reduced b h
hemes and oxidized bl hemes, as in the minimal model. The
kinetic scheme, corresponding to a channel model includes
substantially more kinetic intermediates and reaction steps
than the minimal scheme in Fig. 1.
 23

Detailed description of kinetic models Four moiety conservation relations (Eqs 3 and 4) reduce the
number of differential equations to be considered in Eq. 2 from
To analyze the Q cycle kinetics, we shall convert the 12 to 8. It is worth mentioning that although the second electron
reaction scheme, shown diagrammatically in Fig. 1, into a from ubiquinol, (QH2 )p' recycles through cytochrome b, no
set of mathematical equations, known as chemical kinetics additional moiety conservation associated with this cycle
equations. For changes with time ofthe concentration of any occurs because the Q cycle is open with respect to electrons
Q cycle intermediate, e.g. Q n , we can write: moving from succinate to cytochrome c.
At steady-states, all metabolic fluxes and concentrations
do not change with time. Mathematically this implies that all
rate of change ofl (total rate Of) (total rate of )
( the time derivatives in Eq. 2 are equal to 0. Then, fluxes through
Qn concentration) - Qn production Q consumption different steps must be related to each other to sum to zero
(1) at all 'node' points of kinetic diagram. For a minimal model
(Eq. 2) the following flux relations have to be fulfilled:
Here the total rate is the sum of the rates that produce or
consume Qn according to the kinetic diagram. For instance, V = V suee
1
the total rate of Q n consumption equals the sum of steps 1 VIO= vO' 2-
and 7, see Fig. 1. V = V = V = V = 2v - vO' 2-
2 3 5 9 suee
Similar balance equations can be written for all inter- V4 = V6 = V 7 = 2(vsucc - vO. 2-)
mediates. In the left column of Table 1, stoichiometries of V
8
= V
suee
- vO-
2
(5)
the reaction steps, corresponding to a minimal model, are
depicted. The set of balance equations, comprising the Here the rates, V I and V 10' i.e. the rates of succinate con-
minimal model (Table 1), is the following: sumption and superoxide (0'2-) production, are designated
by v,,'" and vO 2-. Equation 5 shows that in minimal model
dQ" Idt = Vs - v 7 - VI dbO~ Idt = V6 - v4 all fluxes can be expressed in terms of two independent
dQ'n- Idt = v 7 - 2vs dbre~ Idt = v4 - V6 fluxes, e.g. vsucc and vO 2-. In particular, the rate of electron
d(QH2 )n Idt = Vs + VI - v2 dbo~ Idt = v7 - V6 transfer to cytochrome c (v 9 ) equals 2vsucc - vO 2 This
d(QH2)p Idt = v2 - V3 db redh Idt = V6 -v7 decrease in the rate of reduction of cytochrome c is due to
dQp- Idt = V3 -VIO - v4 dFeSo~n Idt = V9 - V3 bypass of the interheme b electron transfer (and the
dQp Idt = vlO - v4 - Vj dFeS'etn Idt = V3 - V9 (2) electron recycling through cytochrome b to reduce by
direct reduction of oxygen with the rate vO 2-.
Here bred I' box I and bredh' bO\ are reduced and oxidized states As was already mentioned, along with the minimal model
of heme b I and bh, F eS'edm andFeSO xm are reduced and oxidized (Table 1), we have also considered the more realistic channel
states of FeSnJ' Stoichiometric coefficient 2 in the kinetic model ofthe Q cycle (Table 2) . One can see that the number
equation for Q'n- specifies that two molecules of Q'n are of kinetic intermediates increased in a channel model as
consumed in ubi semiquinone dismutase reaction (step 7 in compared to the minimal model. In a channel model, the
Fig. 1). The concentrations of succinate, fumarate, oxygen, complexes with bound Qp-(e.g. FeS'etll- Q'p-) as well as all
superoxide, external and internal pH and the ratio of (cyto- the possible states of hemes bland bh , are to be considered.
chrome credl)/(cytochrome CO\) are assumed to be constant. States, corresponding to different oxidation/reduction states
Equation 2 shows several features of the Q cycle stoichio- ofhemes b l and b h are designated as follows (Table 2): bO~x
metry, as depicted in Fig. l. Let [total Q], [total bl]' [total bh], and br~ed correspond to states with both b l and b h either
[total F eSm] refer to the total concentrations of quinone, hemes oxidized or reduced, b;:d corresponds to state with oxidized
b l and bh, and iron sulfur protein, respectively. Adding together b l and reduced b h, br~~ corresponds to state with reduced b l
first six equations (for different Q forms) in Eq. 2 we find that, and oxidized bh For instance, complexes, Q. p-_box ox and Q.p-
d[total Q]/dt = 0, i.e. [total Q] does not change in the Q cycle, -boxred (see Table 2, steps 4, 5 and 6, 7, respectively),
correspond to states of cytochrome b with bound Q. p- in
Q +Qp +Q' -+Q-+(QH)
1/ P
11 2n
+(QH)
2p
= [totalQ] = const (3) which bI is oxidized and bh oxidized or reduced, respectively.
The transfer of electrons to/from and between the hemes
Similarly, by grouping other equations in Eq. 2, we find that b is described now as the transitions between different states
moieties conserved in the Q cycle: of cytochrome b, corresponding to different oxidationl
reduction states of hemes b l and bh . In channel model, the
box I + bred I = [total b I] complex Fes red m -Q'p- is donor of electrons. Before an
bO\ + bredh = [total bh] electron is transferred to heme b l, the bound Q'p- is transferred
FeSo~11 + Fesre1n = [total Fes lu ] (4) to cytochrome b (steps 4 and 5). These two steps correspond
24

Table 1. Reaction steps, considered in the minimal model, and the At the steady-state conditions, all the fluxes in the channel
corresponding rate laws
model can be expressed in terms of three independent fluxes
Reaction Rate equation which can be chosen as the steady-state rates of succinate
consumption (VI)' superoxide production (V I4 ) and one of the
I. Succ + Qn = Fum + (QH2 )n See the text, Eq. 8 two rates of Q.p- transfer (v5) (cf. similar relationships for
2. (QH2 )n =(QH2 )p kd ' ((QH2), - (QH2 )p)
3. (QH2 )p + FeS;~ = Q;-+ 2H; + FeS:,~d kJ (QH,)p . FeS;;,-k_J . Q;-'
the minimal model, Eq. 5):
FeS~~
4. Q;- + b~' = Qp + b~d k . Q;-' b;' - k-4 . Qp ' b;'d VI = v,ucc
5. Qp = Qn kd ' (Qp -Q) VI4 = vO' 2-
6. b~d + b~' = b~' + b;:d k6 ' b;"d. b~'- k .. . b~' b;:d
V5 = VB
7. Qn +b~'d=Q~-+b~' k7 ' Q, . b~'d_k_7' Q~-' b~'
V2 = V3 = V I3 = = vO' 2-
8. 2Q~-+ 2H; = Qn+ (QH,)n kg . (Q~ -)2 - k_g . Q, . (QH2 )n Vs 2v,ucc -

9. FeS,~d + c~' = FeS;;, + C~d k, . FeS:~d . c~' . k_9 . FeS;~ . V9 = 2(v,ucc - vO' 2-
c~ed v7 = VIO = VB
k ,o ' Q;-' 02 -k_1O Qp 0i- VII = V6 = 2(v,ucc - vO' 2-J - VB
V I2 = V,ucc - vO' 2- (6)
to different states of cytochrome b in which b l is oxidized
In both minimal and channel model, the rate of interheme
and bh is either oxidized (step 4) or reduced (step 5). In steps
electron transfer equals 2(v suce - vO' 2-) i.e. the difference
6 and 7, an electron is transferred within the complexes, Q.p-
between the rate of electron supply to the Q cycle and the
_bOX and Q. p-_box respectively, to reduce b l As a result,
ox red leak of electron to oxygen in the 0'2- - generation step.
cytochrome b switches to states br~~ or br;:d respectIvely,
Moiety conservation relationships for a channel model
and quinone is formed at P side. The transfer of an electron
are similar to those for minimal model. However, they
between hemes band I
bh takes place in step 9, where the
include more intermediates, i.e. the complexes with bound
transition of state bred to box occurs. In steps 10 and 11 Qn
ox red
semiubiquinone, as well as the concentrations of states,
is reduced to relatively stable ubisemiquinone anion (Q'n-)
corresponding to different oxidation/reduction states of
at N-side.
hemes b I and bh :
In the channel model, it is assumed that ubi semiquinone
anion (Q.p-) bound to FeSIII can be oxidized by oxygen to
Q n + Qp + Q. n- + (QH) + (QH) + Q. P- - box + Q. p--
yield O' - (step 14). This step bypasses the transfer of 2 n 2 p ox
box + FeS'ed - Q. - = [total Q];
electrons\etween hemes. For the sake of simplicity, direct red III p
FeS':x + FeS'ed + FeS'eIIdI - Q.P- = [total FeSIII ];
oxidation of Q. - bound to cytochrome b was not taken into III III
bOX + bred + box + bred + Q. - -box + Q. - - box = [total b]
account. Addition of these steps (i.e. direct oxidation ofQp- ox ox red red p ox p red (7)
_box
~
and Q. p-_boX
~
similar to oxidation of FeS'~I~-Q'p-) did not
affect the kinetic behavior of the model.

Table 2. Reaction steps of the channel model, and the corresponding rate expressions

Reaction Rate equation

Designations: bO' and bred correspond to states with both b, and bh either oxidized or reduced; b~;d corresponds to stat~ with oxidi~e~ b, a~d redu~ed ~h'. b~~
o, wit. rehd red uced b I and 0 x idized bh'. Q'--
correspond s to state p bO'
0' and Q'p - - bO'
red correspond to states of cytochrome b with bound Q; , m which b , IS oXidize
and b. is oxidized or reduced, respectively.
 25

Rate laws 5 ,-------------------------------------,

'2
The right column in Tables 1 and 2 gives the rate laws of :
::2
steps depicted in the left column. The reaction supplying 34
electrons to the Q cycle, is succinate oxidation (step 1 for z
0
both minimal and channel models). When succinate is in i=
0
excess, the rate of succinate dehydrogenase (v) depends ;:)
0 3
on the ratio Q/(QH)n and can be described by the follow- 0
ex::
ing equation [5, 6]: Q.
UJ
0
VQnl(Qn + (QH)) X
0 2
VI = -K-+-"-Q-n--"/('--Q-n-+-(Q-=-H"-2-)n~) (8) ex::
UJ
Q.
;:)

where V = 256 mM/min, K = 0.6.

(J)
LL
0
In either model the remaining reaction steps of the Q cycle UJ 2
f-
(Tables 1 and 2, Fig. 1) correspond to unimolecular transitions
ex::
and bimolecular interactions. The rates of these steps are
given by the law of mass action (Tables 1 and 2). For every 0
step obeying the mass action law, only the values of the 0 50 100 150 200 250 300 350
equilibrium and forward rate constant should be specified. t.' (mV)
Given a concrete values of the equilibrium constants, the
reverse rate constant (k) was expressed through the forward Fig. 2. Increase in the steady-state rate of superoxide generation with the
membrane potential (A') for the minimal (line I) and the channel model
rate constant (k) as:
(line 2). The values of parameters are given in Tables 3 and 4.

k.=kKieq (9)
-/ I
overall reactions including steps 3 and 4, 3 and 5, 3 and 14,
were corrected in accordance with pH values at N- and
The equilibrium constants of oxidation-reduction reactions
P-sides of mitochondrial membrane, respectively (see Eq.
were calculated from the experimental data on midpoint
11 ).
potentials:
If a kinetic scheme includes 'true' cycles, in which the
Keq = exp (10) initial and final states are identical, the magnitudes of the
equilibrium constants of the reactions included in the cycle
satisfy so called 'detailed balance' relationships (see e.g.
[8]). Detailed balance conditions require the products ofthe
Here, T is absolute temperature, L1E m h is the difference of
equilibrium constants along a cycle to be equal to 1, as at
the midpoint potentials with respect to standard hydrogen
equilibrium the net flux through any cycle vanishes. For
electrode at pH =h, n is the number of electrons transferred.
channel model moving along steps 5, 7 and lOin the positive
For reactions that include protons as the substrate or
direction and, then, along steps 6, 4 and 11 in the negative
product (e.g. steps 3 and 8 in a minimal model), L1E m,h were
direction completes a cycle without any concomitant trans-
obtained usingL1Em ,7 data and pH values at P-side and N-side
formations and changes in the free energy. This cycle
of mitochondrial membrane as follows (see [7]):
implies the following constraint on the possible magnitudes
of the kinetic constants:
L1Emh = L1E m ,7 + 2.32(RT/F) . (7.0 - pH) (11)

In a channel model, L1E m,h values for steps, in which one of
the complexes , FeS'ed III - Q. p'
- Q. p- - bOX ox and Q. p- - bOX red'
is the substrate or the product, can not be obtained from
experimental data. As data only for free forms are available,
only the equilibrium constant for overall reaction, which
includes both oxidation-reduction step and dissociation
step, is known. Therefore, the choice of either equilibrium
constant was arbitrary, provided that the product of the two
constants is determined by L1E m,h of the overall reaction.
For a channel model, L1E m,h values for step 12 and for
= 1 (12)
Rate constants for the diffusion of QH2 from P- to N-side
and Q from N- to P-side of membrane (kd in Tables 1 and 2,
respectively) were estimated using the data on the diffusion
coefficient (DQ) of ubiquinone. For one-dimensional
diffusion of Q, the steady-state diffusion flux (JQ) can be
expressed in terms of the diffusion coefficient (D Q), the
diffusion distance (1) and the surface area of the inner
mitochondrial membrane per gram protein (crm ) as follows:
26

Some results of application of the above approach

J = DQ(J . (Qp - Q) (13)
Q m I
where Qp and Q n are ubiquinone concentrations at P- and N-
sides of the mitochondrial membrane, respectively. On the
other hand, the flux J Q can be expressed in terms of the
diffusion rate constants (k) as:
J Q = k d . (QP -Q)
,/
(14)
Equating these two expressions for J Q, we express the rate
constant kd as:
(15)
Estimating the diffusion coefficient DQ of ubiquinone as D Q
= 10-8 cm2 /sec [9], the specific surface area of the mito-
chondrial membrane as (J = 2.10 6 cm 2/(g mitochondrial
protein) [10] and the diffusi;n distance (1) as 1 = 4.5lO-7 cm
[I], one arrives at the following magnitude of the diffusion
rate constant, kd = 1.32.10 6 min-I.
Any step including electric charge transfer across the
membrane produces membrane potential (L1\f) which affects
the rate and equilibrium constants of these steps [11-12].
In the models considered here, steps ofthe electron transfer
to, from or between hemes b h and b l are assumed to be
affected by L1 \f:
Using the kinetic models described above we have calculated
the dependencies of the rate of superoxide radical formation
on the membrane potential. Figure 2 shows a significant
increase in the generation of 0'2- with increase in the
membrane potential (L1\f) above some 'critical' magnitudes.
Increase in superoxide production is due to the accumulation
of ubi semiquinone forms which can be directly oxidized by
oxygen (Q.p - and FeS'ed - Q.- in a minimal and channel
III P
model, respectively), with increase in L1\f (Fig. 3).
The membrane potential is a thermodynamic force that
counteracts transmembrane electron transfer between
hemes b l and bh Equation 16 shows that L1\f lowers the
equilibrium constant of the interheme electron transfer,
decreasing the forward and increasing the reverse rate
constant. At high membrane potential, the electron transfer
from b l to b h becomes thermodynamically unfavorable, and
the reaction can proceed only if the ratio of bre~ /bo~ (in a
channel model the ratio of total concentrations of states
corresponding to reduced and oxidized forms of heme b l )
becomes very high (Fig. 4). A deep depletion of boxi and a
significant increase in bred I concentrations causes accumulation
ofthe ubi semiquinone forms at P-side of the membrane to drive
reduction of heme b l An increase in the concentrations of
ubi semiquinone forms (Q. p -, FeS'ed - Q. -) results in almost
II l p
linear increase in the rate of their direct oxidation and
generation of the superoxide radicals (cf. Figs 2 and 3) .

Keq (L1\f) = exp(a'L1\f/(RT/F-Keq

k+ (L1\f) = exp(---{)'a'L1\f/(RT/F'k+
kJL1\f) = exp(o'(1- a)'L1\f/(RT/F'k_, (16)

Here a'L1\f is the membrane potential difference that drops z

along the path of an electron in a particular step, 0 is the o
i=
relative fraction of the membrane potential difference
a::
I-
(a'L1\f), which affects the rate ofthe forward reaction. Note, z
W 2
that the effects of L1 \f on the forward and reverse rate o
z
constant are in line with Eq. 9. Using the data on the locations o
o
ofhemesb h andb l within the membrane [I], we shall assume w
that the difference in L1 \f that drops between these hemes
>
~
accounts for 80% of the potential difference across the -'
w
mitochondrial membrane, i.e. a equals 0.8 for the corres- a::
ponding step.
Tables 3 and 4 summarize the parameter values for the
models of the Q cycle considered. The parameters include
all the rate and equilibrium constants of steps and conserved
moieties. Experimental data on midpoint redox potentials, o 50 100 150 200 250 300 350
required for estimation of the equilibrium constants, are A'P (mV)
given as well. Note that due to lack of experimental data
some forward rate constants were chosen arbitrarily. Fig. 3. Dependencies of the relative concentrations Q.p I[total QJ (line I)
and FeS'edlll-Q'p-/[total FeSIIIJ (line 2), i.e. ubi semiquinone forms directly
oxidized by 02' on the membrane potential for the minimal and the channel
model.
 27

Table 3. Parameter values for the minimal model ofQ cycle

Reaction Midpoint potential Equilibrium constants at Rate constants, Free parameters Other parameters
no. Em 7 = E, (V) [Ref.] 6'1' = 0 (V), pHp = pH" = 7 mMI'min- 1

V = 256 mMimin,
K=0.6

3 E(QH;IQH,) = 0.29 [9] K;q=exp(-39' 0.01) k J = k03 ' K;q exp(39' 0.118' k03 = 200 pHp = 7.2
E(FeS;;,IFeS;,~d) = 0.28 [3] (7 - pH), k3 = kOJ

4 E(QpIQ;) =--0.16 [2] K~q = exp(39' 0.12) k4 = k04 . K~q . exp(--a' '\ 39 6'1') k04 = 200, 0.=0.1
E(b~'lb;'d) =--0.04 [3] k-4 =k04 ' exp(o. (1-8,)' 39 '6'1') 8, =0.5

6 E(b~'lb;"I) =--0.04 [3] K~q = exp(39 . 0.08) k6 = k06 . K~q . exp(-~ . 8, . 39 . 6'1') k06 = 200, ~=0.8
E(b~'lb;'d) = 0.04 [3] k-6 = k 06 ' exp(~' (1-8,)' 39 6'1') 8, = 0.5

7 E(b~'lb;'d) = 0.04 [3] K;q = exp(39 . 0.03) k7 = k07 . K;"' exp(-y 83 . 39 . 6'1') k07 = 1000, y=O.1
E(Q)Q;) = 0.07 [9] k 7 = k07 ' exp(y' (1-8) 39 6'1') 83 =0.5

8 E(Q"IQ;) = 0.07 [9] K~q= exp(39' 0.1) kg = k08 . K~q . exp(39 . 0.118 . (7 - pH) k08 = 200 pH =7.8
"
E(Q;IQH,") = 0.17 [9] k.8 = k08

9 E(FeS;~IFeS;,~d) = 0.28 [3] K~q = exp(-39 . 0.035) k, = k09 . K;q, k09 = 1000
E(c~'lc;'d) = 0.245 k9 = k09

\0 E(O,IO;) =--0.15 [3] K~ci = exp(39 . 0.0 I) k lo = kOlo . K;.r, ko,o = 0.1
E(Q)Qp) =--0.16 [2] k 10 = kOlo

Boundary conditions: 02 = 2f.lM [30], O~' = O.OIf.lM,c~' = 0.2 nmole/mg pr., c;'d = 0.01 nmole/mgpr. (c~' + c;'d = 0.21 nmole/mg pro [28]).
Conserved moieties (nmol/mg mitochondrial protein), [28-29] [total Q] = 4, [total b l ] = [total bh ] = [totaIISF] = 0.325.

Thus, at high membrane potentials, the electron transport It is important to elucidate the activities of which steps
from heme b l to bh is heavily suppressed and the bypassing can be modulated to restore the recycling of electrons
flux of 0'2- production increases dramatically (cf. Fig. 3). through hemes b l and bh and, consequently, to decrease the
rate of superoxide production. At first glance, it seems that
the electron flux should restore if one increases the activity
1600
of the electron transfer between hemes, since this step is
mostly affected by L1' (up to 80% of the membrane
1400
potential difference drops at the distance between hemes).
(f) Simulations show, however, that even a 10 fold increase in
z 1200 the activity does not released the inhibition of electron
0
i= recycling (Fig. 5).
<{
0:: 1000
I-
Z 2 To approach this problem we have used methods of
W
o Metabolic Control Analysis (MCA) [13-16]. MCA quan-
z 800 tifies the control exerted by an enzyme over the steady state
0
0
ll..
flux or concentration as a control coefficient of this enzyme.
0 600
The flux control coefficient is the fractional change in
0
i=
<{
pathway flux divided by the fractional change in the enzyme
400
0:: activity, extrapolated to infinitesimally small change. The
sum of the control coefficients of all the enzymes involved
200 is equal to 1 [l3].
So far MCA was mainly used to analyze the extent to
0 which different enzymes in a pathway limit the pathway flux.
0 50 100 150 200 250 300 However the same principles may be used to analyze control
6'1' (mV) within any molecular process such as: (i) an enzyme or
Fig. 4. Dependence of the ratio of reduced and oxidized hemes b l on the transporter [17], (ii) multiprotein complexes [18], (iii)
membrane potential for the minimal model (b /'dlb, ox, line I) and the channel group-transfer pathways, e.g. mitochondrial respiratory
model (b eed" + b"deed)/(bO'o, + bO\'d + Q.p. - bO'm + Q'/; - bO\,d)' line 2). chain (and Q-cycle) carrying an electron [19-22]. The
28

control exerted by any (elemental) step within Q-cycle on a
steady-state rate of electron transfer is quantified as the
percent change in the flux brought about by 1% change in
the forward (k+) and reverse (kJ rate constants of this step.
A strict definition extrapolates the percent change to an
infinitesimally small change:
CJi = Jk+i (ddkJ+i J I
kjk+i = const
Here J stands for any steady-state flux within Q-cycle
Simultaneous proportional changes in the forward and
reverse rate constants of a step do not compromise micro-
scopic reversibility. Thus, the ability of a step to change
the pathway flux can be quantified and the relative im-
portance of different enzymes to control this flux can be
simply assessed, since the sum of flux control coefficients
of all elemental steps is equal to 1 [18] (see [23] for review).
Table 5 presents the control coefficients of all steps with
respect to the electron flux through hemes for the minimal
model. One can see that the control is shared between steps
1, 3 and 9 at low ~'II values, and step 1 at high ~'II values.
Therefore, when the electron transport between hemes is
inhibited by ~'II, this inhibition can be partly released by
increasing the activities of the following steps: cytochrome
( 17)
c[ reduction (step 9) and succinate dehydrogenase (1), but
not the activity of interheme electron transfer step itself.
This effect of non-sensitivity of the flux to the activity of
the step directly affected by ~', is due to a particular way
of the influence of ~', which changes the equilibrium
constant, rather the activity of the step. If the activity were
inhibited (i.e. the forward and reverse rate constants
decreased simultaneously), the inhibition of the flux through
this step would be accompanied by the shift of the control
towards this step, and the flux would be highly sensitive to
changes in its activity. The distribution of the control for a

Table 4. Parameter values for the channel model of Q cycle

Reaction Midpoint potential Equilibrium constants at Rate constants, Free parameters Other parameters
no. E m.7=E,(V)[Ref.] L1'P = 0 (mV), pHp = pH, = 7 mM . min- J or min- J

V = 256 mMlmin,
K=0.6

3,4,6 E(QH;IQH,p) = 0.29 [9] K;q = R3 k3 = k03 . K;q . exp(39 . 0.118 . (7 - pHp ' kG3 = 200 pHp =7.2
E(FeS;~JIFeS~~d) = 0.28 [3] K;q = exp(-39' 0.01)IR 3, k_3 = k03' R3 = 100 a=O.l
E(Q)Q;) =-0.16 [2] K~q = R3 . exp(39 . 0.13) k4 = k04 . K;q, k4 = k 04' k04 = 1000 a=O.l
E(b:,'lb~~d) =-0.04 [3] k6 = kG6 . K~q . exp(--a' oJ 39 L1'P), k06 = 200
k.(, = k 06 ' exp(a' (I-oJ)' 39 L1'P) oJ =0.5

3,5,7 E(QH;IQH2p ) = 0.29 [9] K;q = R3 k3 = k03 . K;q . exp(39 . 0.118 . (7 -pHp ' k03 = 200 pHp = 7.2
E(FeS;~IFeS;~d) = 0.28 [3] Ks'" = 1IR3' exp(-39' 0.01), k_3 = k03' R3 = 100 a=O.l
E(Q)Q;) =-0.16 [2] K;q = R3 . exp(39 . 0.13) ks = kGS . K;', k, = kos' k05 = 1000
E(b~~dlb;~~) = -0.04 [3] k, = ko,' K;q . exp(--a' 02 . 39 . L1'P), kG, = 200
k_,=ko,' exp(a (1-0,)' 39 'L1'P) oJ =0.5

9 E(b~'lb;'d) =-0.04 [3] K;q = exp(39 . 0.08) k9 = k09 . K;q . exp(-p 03 . 39 . L1'P), k09 = 10' p=0.8
E(b~Xlb~'d) = 0.04 [3] k9 = kG9 exp(p, (1-03) . 39 . L1'P) 03 = 0.5

(1-
10 E(b ;:~Ib~~d) = 0.04 [3] K;,r = exp(39' 0.03) kJO = kOJO . KJ'ri . exp(-y' 04 . 39 . ;1.'P), kOJG = 1000 y= 0.1
E(Q.lQ,-) = 0.07 [9] kJO = kOJo ' exp(y' 4 ) ' 39 . L1'P) 04 =0.5

II E(b~:dlb:,') = 0.04 [3] K;i = exp(39 . 0.03)

k JJ =kOJJ' Kti' exp(-y' 5 ' 39 L1'P), kOJJ = 1000 y= 0.1
E(Q,IQ,-) = 0.07 [9] k_JJ =kOJJ' exp(y' (1-0')' 39 'L1'P) 05 = 0.5

12 E(Q.lQ,-) = 0.07 [9] K;i = exp(39 . 0.1) kJ2 = kGJ2 . K;r . exp(39 . 0.118 . (7 -pH,, kGJ' = 200 PH, = 7.8
E(Q,-IQH2 ) = 0.17 [9] kJ2 = kOJ2

13 E(FeS;~IFeS~d) = 0.28 [3] K;j = exp(-39 . 0.35) k JJ = kOJ3 . K;j, kOJ3 = 1000
E(c~'lc;'d) = 0.245 k_J3 = kOJ3

9, 14 E(QH;IQH,) = 0.29 [9] K;X = exp(39' 0.01) kJ4 = kOJ4 . KJ'X, kOJ4 = 0.1
E(QpIQ;) =-0.16 [2] k_ J4 = kOJ4
E(O,IO,) =-0.15 [3]

Boundary conditions: 02 = 2~M [30], 0, = O.Ol~M,c~x = 0.2 nmole/mgpr., c;'d = 0.01 nmole/mg pro (c~X +c;'d = 0.21 nmole/mg pro [28]).
Conserved moieties (nmollmg mitochondrial protein), [28-29] [total Q] = 4, [totalb] = [total ISF] = 0.325
 29

Table 5. Control exerted by different steps on the flux of electrons from Q n, Q. n-, pHp and the same (fixed) ratio of reduced and
heme b l to heme b. in the minimal model oxidized forms of FeSnJ" Figure 6 shows how the relative
~'(mV) 100 ISO 200 250
steady-state rate of transmembrane transfer of electron by
cytochrome b depends on A'll. One can see that the electron
CI 0.68 0.66 0.54 0.34 transfer is suppressed at lower A'll values for the channel
C2 10-' 10-' 10-' 10-'
than for the minimal model. Suppression of the electron
C3 0.1 0.1 0.09 0.06
C. 0.001 0.002 0.01 0.05 transport through cytochrome b implies an increase in the
C, 10-6 10-" 10-' 10-' bypassing flux of direct reduction of oxygen generating
C. 10-' 10-' 10-" 10-6 O' 2- radicals (see Eqs. 5 and 6 relating the steady-state
C, 1<f4 1<f4 0.04 0.02 fluxes).
Cg 1<f4 1<f4 0.002 0.01
0.23 0.24
Calculated A'll, which is critical for the generation of O' 2-
C. 0.35 0.52
CIO 0 _10-9 _10-9 _10-9 (around 170 mV for a channel model, Fig. 2), depends on the
SUM 1.002 1.000 1.000 1.00001 magnitudes of the rate constants. To analyze this dependence
we again calculated the control coefficients of all steps, but
now with respect to the concentration of the ubi semiquinone
channeled model confirms that transfer of electrons from forms directly oxidized by O 2 , Concentration control
heme bI to bh is not controlled by the corresponding step, coefficient of a step is defined similar to flux control
althoughjust this step is mostly influenced by the membrane coefficient (Eq. 17). The difference, however, is that one
potential (data not shown). should follow the fractional change in the steady-state
At similar magnitudes of parameters a critical increase concentration rather than in the steady-state flux caused by
in 0'2- production was observed at lower A'll for the channel, the fractional change in the step activity. The relative
than for the minimal model (Fig. 2). To understand which contributions of different steps to the control can be assessed,
design features are responsible for this difference in the since the sum of concentration control coefficients of all the
kinetic behavior, we have analyzed how the membrane steps is equal to 0 [13,19].
potential affects the rate of electron recycling through
cytochrome b under the same magnitudes of the driving
forces, i.e. at the same fixed concentrations of (QH2) p , Qp ,

'2 300 - , - - - - - - - - - - - - - - - - - - - - - - ,

~
.sc:: 250 w
w
lL
en
2 ~
z w
~ 200 >
~
I-
Z :5
oc:: w
c::
t-.
U 150
W
..J
W
.Q "
I) 100
I
.Q-
W
I
l- 50 04-----.-----.-----.-----,-----,---~
lL
o o 50 100 150 200 250 300
W

~ O+----.----.------.----.---.--------l
~' (mY)

o 50 100 150 200 250 300

~' (mV)
Fig. 5. Dependence of the steady-state rate of electron transfer from heme
b I to heme bh on the membrane potential (~') for the minimal model. Line 2
corresponds to a ten-fold increase in the rate constants, k, and k_,' as
compared to the magnitudes given in Table 3 (line I).
Fig. 6. Dependencies of the steady-state rates of electron transfer from
heme bI to heme bh on the membrane potential ~'1', calculated at pH. = 7.2
and at fixed concentrations of all ubiquinone forms, (QH2)p =1.6 nmollmg
protein, Qp =I nmol/mg protein, Q. =I nmollmg protein, Q-. =0.2 nmol/mg
protein, and forms of iron sulfur protein FeS'"d m =0.005 nmol/mg protein,
FeS"1II = 0.32 nmollmg protein for the minimal (line 1) and the channel model
(line 2).
30

Table 6 contains the control coefficients of all steps with 5

respect to the concentration of Q.p- for a minimal model. At '2
the membrane potentials below the values, where the 0'2- :
::;!
production rate increases dramatically (Fig. 2), the positive 34 2
control resides on the succinate dehydrogenase reaction (step z
0
I for a minimal model) and the negative control resides on i=
()
steps of electron transfer through hemes bl and bh (steps 4 ::J
0
0 3
and 7 for the minimal model). At critical membrane potentials, a::
all the steps loose their control over Q.p-. Therefore, the high IL
w
concentration of Q.p - appeared to be stabilized as the con- 0
X
centration of b,e1d approaches the maximal possible magnitude 0 2
a::
equal to btott~ and the concentration of bo x1 reduces to O. In w
IL
::J
view of this distribution of the control over Q.p - one may guess (jJ

that for the minimal model, the critical L1'P shifts to the lower LL
0
L1 'P values when the activities of steps 4 and 7 decrease. w
I-
<{
Figure 7 confirms this conjecture, showing a shift in the a::
critical L1 'P towards the lower magnitudes, caused by changes
in the corresponding activities.
o 50 100 150 200 250 300 350
For a channel model, Table 7 summarizes control co-
L'>'P (mV)
efficients of all the steps with respect to the concentration
of FeS'~I~ - Q'p-' At low membrane potentials, the almost
Fig. 7. Shift in critical L'.'P values in the minimal model, caused by 10- fold
entire all positive control is exerted by succinate de-
increase in the activities of steps 3 and 9, and I O-fold decrease in the activities
hydrogenase reaction (step I) whereas the negative control of steps 4 and 7. Line I corresponds to the parameter values given in Table
resides on the branch leading to cytochrome c 1 (step 13) 3; line 2, k03 = 2000, k09 = 10000, k04 = 20, k07 = 100.
and on the pathway of electron transfer through hemes b l
and bh (for the channel model, the sums of control co-
efficients with respect to 'parallel' transitions 4 and 5 are actlVlty, V (see Eq. 8) , over FeS'~~ - Q.p- (the channel
negative). At critical membrane potentials, all the steps model) essentially exceeds that over Q. p - (the minimal
loose their control over FeS'~I~ - Q'p-' Comparing Tables 7 model). This means that decrease in the succinate de-
and 6, one can see the essential difference in the control hydrogenase activity will result in essential shift of the
properties of a channel and a minimal model, which can be critical L1 'P to higher values in case of the channel model
used for testing these models against the experimental data and insignificant changes of a dependence of the superoxide
(Demin et al., in preparation) . In particular, this difference production rate on membrane potential in case of the
manifests itself in ability of succinate dehydrogenase to minimal model (Figs 8a and 8b). The results specify one
control concentration of the ubisemiquinone form directly of the possible ways to decrease superoxide concentration
oxidized by 02 (Q'p-' FeS'~J~ - Q.p- for the minimal and the during respiration in mitochondria. This way consists in
channel model, respectively). Tables 6 and 7 show that control reducing of the succinate dehydrogenase rate to a minimal
coefficients (C I) with respect to succinate dehydrogenase possible value.

Table 7. Control exerted by different steps on the concentration of

Table 6. Control exerted by different steps on the ubi semiquinone ubisemiquinone bound to FeSIIl (FeS;I~d ~ Q;) in the channel model
concentration at side P in the minimal model
L'.'P (mV) 50 80 110 140
L'.'P (mV) 100 ISO 200 2S0
Cl 0.93 2.1 0.6S 0.04
C, 0.78 0.77 0.49 0.03 C, 10-1> 10-5 10-7 10-"
C, ~105 ~lO' ~IO-s -10-1> C3 0.003 --O.OS 0.002 10 4
C] --0.096 --0. IS -0.06 0.01 C, + C, ~1.03 ~1.3 --0.7 --0.072
C4 --0.41 --0.17 --0.11 --0.1 C6 + C) 0.043 0.03 --0.02 ~1O.4

Cs ~IO-s ~IO-s ~10-5 ~1O 5 Cs 10-) ~1O-) ~IO-l> ~107

C, ~1O-5 ~IOs ~10-5 ~10-5 C, ~IO-l> ~IO-" ~10' ~IO-l>

C7 --0.04 --O.OS --O.OS -0.04 C'O+C'1 --0.003 --0.016 --0.036 --0.006

Cs --0.01 --0.02 -0.02 --0.01 C l2 --0.002 --O.OIS --0.01 --0.001
C9 --0.22 --0.38 --O.2S 0.11 C l3 0.08 --0.74 0.1 0.004
C,o 0 ~109 ~10-9 ~109 C l4 10-9 10 9 10-9 10-9
SUM 10-) 10 7 10-' 10- 7 SUM 10-) 10 7 10-1> 10"
 31

500,----------------------------------.

;-(l:
It is instructive to analyze how the dependence of 0'2-
production rate on L\'II is influenced by L\pH. Simulations
(Fig. 9) show that a decrease in pHn i.e. increase in L\pH up
.-
C II
~
1/'
I 3 2 1

~4001
to 1.8 unit results in a significant elevation of superoxide
production at the same L\'II, which means shift of the /

~ ~ /1 jl;1
threshold to lower L\'P values. At first glance, an increase

A 300
5.----------------------------------.
~
1
I
200
,)
z
o

i '~J ==
i=
u
::>
o
o0::: 3
n.
w
o
2 o 50 100 150 200 250
0:::
w 6'1' (mV)
n.
::>
(f)
Fig. 9. Dependencies of the steady-state superoxide generation on L\ 'I' for
the channel model at pHp = 7.2 and different pH. Lines 1, 2 and 3 correspond
to pH. = 7.4, pH. = 7.8 and pH. = 9, respectively.

in pH. and vanishing of L\pH (i.e. the conversion of L\pH into

o 50 100 150 200 250 300 350 L\'P) can be used to drop ROS generation. However, Fig. 10
shows that this apparent expectation is wrong. It is known

- - ,:
L\'I' (mV)
B that

I : -.J-----------~---'
II / \
(18)

Plot of JO' 2- vs. L\JfH (Fig. 10) demonstrates that in practice

there are no difference between the generation of superoxide
5 I
radicals at the same L\J1H when L\pH varied from 0-1.8.

/2 3 I
In the present paper, we consider kinetic models of the
mitochondrial bel complex without taking into account of
a proton leak. Introducing of this important feature of
150 mitochondrial respiration has to result in essential changes
in global kinetic characteristics (steady-state fluxes and

~ ~ concentrations). However, the proton leak can not be

;::
100 described quantitatively in the framework of our models,

~ ~l
since variables influenced with the proton leak, pHn,pHp as
well as L\'P, are parameters, i.e. assume the same single
values for any moment of time. On the other hand, we can
consider a limiting case of an equilibrium proton distribution
o 50 100 150 200 driven by membrane potential. In this case, pHD is determined
L\'I' (mV)
by L\'II as follows:

Fig, 8. Increase in the steady-state rate of superoxide generation with the pHn = pHp + 0.59(FIRT)-L\'P - C (19)
membrane potential (L\'I') for the minimal (a) and the channel (at R3 = I) (b)
model. Curves 1,2 and 3 correspond to normal (V=256mM/min) , half(V= Here C is a constant equal to 3.25 which is chosen in a such
128 mM/min) and eight part (V = 32 mM/min) activity of succinate
way that L\pH would equal 0.6 at L\'P = 140 mY, pHp = 7. 2
dehydrogenase, respectively.
32

500 I (State 3). Figure 11 shows that taking Eq. 19 (and, con-
I

Il
'2
.i= 450
..::
J sequently equilibrium proton distribution) into account
shifts threshold of dependence of superoxide production rate
~
.s on potential difference to the higher ~'I' values .
z In conclusion, it should be emphasized that similar
0
i= m -I relationships may be inherent in the bJ complex of chloro-

1/
0 350
::J
0 I plasts and cyanobacteria as well as in a Q cycle-like
0
mechanism which is probably involved in the NADH-CoQ

:~1
0::
a. reductase ~jIH+ generator [3]. This means that some increase
ill
0
I in the proton leak at State 4 may, in fact, prevent a burst of
X
0 the superoxide production which otherwise occurs at this
0::
ill 200 / 1/ I State [24]. In other words 'mild', uncoupling actuated at State
3 /1
Il.
::J
(/)
I 4 can be a mechanism of preventing the O2 '- production by
u.. 150
0 mitochondria [25-27].
ill
f-
100
0::

50 Acknowledgements
0 50 100 150 200 250
This study was supported by the Moscow State University,
the Netherlands organization for Scientific Research (NWO)
Fig. 10. Dependencies of the steady-state superoxide generation on li~ + and INCO-COPERNICUS grant ERBIC15CT960307.
for the channel model at pHp = 7.2 and different pH,. Lines 1, 2 and 3
correspond to pH" = 7.4, pH, = 7.8 and pH, = 9, respectively.

References
1. Kholodenko BN: Control of molecular transformations in multienzyme

/1
systems: quantitative theory of metabolic regulation. Mol Biologiya

/1
Russia 22: 1238-1256, 1988
2. Trumpower BL: The protonmotive Q cycle. J BioI Chern 265: 11409-

I
11412,1990
3. Skulachev VP: Membrane Bioenergetics. Springer, Berlin, 1988
I 4. Kholodenko BN, Westerhoff HV, Cascante M: Channelling and the

1;1
concentration of bulk phase intermediates as cytosolic proteins get

'/ !
more concentrated. Biochem J 313: 921-926, 1996
5. Grivennikova VG, Vinogradov AD: Kinetics of ubiquinone reduction by

I
the resolved succinate: Ubiquinone reductase. BBA 682: 491--495, 1982
; I 6. Kotlyar AB, Vinogradov AD: Interaction of membrane-bound succinate

1
I
dehydrogenase with substrate and competitive inhibitors. BBA 784:
I 24-34,1984
7. CramerWA, KnaffDB: Energy Transduction in Biological Membranes.
Springer-Verlag, Berlin, 1991
I I 8. WesterhoffHY, Van Dam K: Thermodynamics and Control of Biological
u.. Free-Energy Transduction. Elsevier, Amsterdam, 1987
o ) / I 9. Rich PR: Electron and proton transfers through qui nones and
cytochromebc j complexes. BBA 768: 53-79, 1984
10. Hoffman P, Morgan TV, Der Vartanian DV: Respiratory chain

o -+-----.---=L=--.---=----r--I-------.-------' characteristics mutants of Azotobacter vinelandii negative to

tetramethyl-p-phenylenediamine oxidase. Eur J Biochem 100: 19-27,
o 100 200 300 400 500 1979
11. Boork J, Wennestrom H: The influence of membrane potentials on
li't' (mY)
reaction rates. Control in free-energy transducing systems. BBA 767:
314-320,1984
Fig. 11. Dependencies of the steady-state superoxide generation on li't' 12. Reynolds lA, Johnson EA, Tanford C: Incorporation of membrane
for the minimal model. Line 1 corresponds to the case when pHp and pH, potential into theoretical analysis of electrogenic ion pumps. Proc Natl
can be chosen independently (pHp = 7. 2, pH, =7.8) . Line 2 corresponds Acad Sci USA 82: 6869--6873, 1985
to the model where limiting case of the electric field influence-equilibrium 13. Kacser H, Burns JA: The control of the flux. In: DO Davies (ed). Rate
proton distribution driven by AT in accordance with Eq. 19 - is taken into Control of Biological Processes. Cambridge University Press, London,
account. 1972, pp 65-104

14. Kacser H, Burns 1A: Molecular democracy: Who shares the controls?
Biochem Soc Trans 7: 1149-1160, 1979
15. Heinrich R, Rapoport TA: A linear steady-state treatment of enzymatic 23.
chains. General properties, control and effective strength. Eur 1
Biochem42: 89-95, 1974 24.
16. Heinrich R, Rapoport TA: Mathematical analysis of multienzyme
systems: II. Steady state and transient control. BioSystems 7: 130-
136,1975 25.
17. Kholodenko BN, WesterhoffHV: Control theory of one enzyme. Biochim
Biophys Acta 1208: 294--305, 1994
18. Kholodenko BN, Westerhoff HV: Metabolic channelling and control 26.
of the flux. FEBS Lett 320: 71-74, 1993
19. Van Dam K, Van derVlag J, Kholodenko BN, WesterhoffHV: The sum
of the flux control coefficients of all enzymes on the flux through a 27.
group-transfer pathway can be as high as two. Eur 1 Biochem 212:
791-799,1993
20. Kholodenko BN, Westerhoff HV: Control theory of group-transfer 28.
pathways. Biochim Biophys Acta 1229: 256--274, 1995
21. Kholodenko BN, Westerhoff HV: How to reveal various aspects of
regulation in group-transfer pathways. Biochim Biophys Acta 1229: 29.
275--289,1995 30.
22. Brand MD, Vallis BPS, KesselerA: The sum of flux control coefficients
33
in the electron-transport chain of mitochondria. Eur 1 Biochem 226:
819-829,1994
Kholodenko BN, Westerhoff HV: The macro- versus microworld of
biochemical regulation and control. Trends Biochem Sci 20: 52-54, 1995
Boveris A, Chance B: The mitochondrial generation and hydrogen
peroxide. General properties and effect of hyperbaric oxygen. Biochem
1134:707-716,1973
Skulachev VP: Lowering of intracellular 0, concentration as a special
function of respiratory systems of the cells. Biochemistry (Moscow)
59: 1910-1912, 1994 (Russian)
Skulachev VP: Non-phosphorylating respiration as a mechanism to
minimize formation of reactive oxygen species in the cell. Mol Biologiya
29: 709-715,1995 (Russian)
Skulachev VP: Role of uncoupled and non-coupled oxidations
maintenance of safely low levels of oxygen and its one-electron
reductants. Quart Rev Biophys 29: 169-202, 1996
Green DE, Wharton DC: Stoichiometry of the fixed oxidation-reduction
components of the electron transfer chain of beef heart mitochondria.
Biochem Z 336: 335--346, 1963
Srere PA: TIBS 6: 4--6, 1981
Jones DP: Intracellular diffusion gradients ofO, andATP. Am 1 Physiol
250:C66~675,1986
Molecular and Cellular Biochemistry 184: 35-52, 1998.
1998 Kluwer Academic Publishers.

Quantitative analysis of some mechanisms

affecting the yield of oxidative phosphorylation:
Dependence upon both fluxes and forces

Michel Rigoulet, l Xavier Leverve, 2 Eric Fontaine,2 Rachid Ouhabi 1 and

Bernard Guerin 1
IInstitut de Biochimie et de Genetique Cellulaires du CNRS, Universite Bordeaux II, Bordeaux; 2Laboratoire de
Bioenergetique Fondamentale et Appliquee, Universite Joseph Fourier, Grenoble, France

Abstract
The purpose of this work was to show how the quantitative definition of the different parameters involved in mitochondrial
oxidative phosphorylation makes it possible to characterize the mechanisms by which the yield of ATP synthesis is affected.
Three different factors have to be considered: (i) the size of the different forces involved (free energy of redox reactions and
ATP synthesis, proton electrochemical difference); (ii) the physical properties of the inner mitochondrial membrane in terms
of leaks (H+ and cations); and finally (iii) the properties of the different proton pumps involved in this system (kinetic
properties, regulation, modification of intrinsic stoichiometry).
The data presented different situations where one or more of these parameters are affected, leading to a different yield of
oxidative phosphorylation.
(I) By manipulating the actual flux through each of the respiratory chain units at constant protonmotive force in yeast
mitochondria, we show that theATP/O ratio decreases when the flux increases. Moreover, the highest efficiency was obtained
when the respiratory rate was low and almost entirely controlled by the electron supply. (2) By using almitrine in different
kinds of mitochondria, we show that this drug leads to a decrease in ATP synthesis efficiency by increasing the H+/ATP
stoichiometry of ATP synthase (Rigoulet M et al. Biochim BiophysActa 1018: 91-97, 1990). Since this enzyme is reversible,
it was possible to test the effect of this drug on the reverse reaction of the enzyme i.e. extrusion of protons catalyzed by ATP
hydrolysis. Hence, we are able to prove that, in this case, the decrease in efficiency of oxidative phosphorylation is due to a
change in the mechanistic stoichiometry of this proton pump. To our knowledge, this is the first example of a modification
in oxidative phosphorylation yield by a change in mechanistic stoichiometry of one of the proton pumps involved. (3) In a
model of polyunsaturated fatty acid deficiency in rat, it was found that non-ohmic proton leak was increased, while ohmic
leak was unchanged. Moreover, an increase in redox slipping was also involved, leading to a complex picture. However, the
respective role of these two mechanisms may be deduced from their intrinsic properties. For each steady state condition, the
quantitative effect of these two mechanisms in the decrease of oxidative phosphorylation efficiency depends on the values
of different fluxes or forces involved. (4) Finally the comparison of the thermokinetic data in view of the three dimensional-
structure of some pumps (X-ray diffraction) also gives some information concerning the putative mechanism of coupling
(i.e. redox loop or proton pump) and their kinetic control versus regulation of mitochondrial oxidative phosphorylation. (Mol
Cell Biochem 184: 35-52, 1998)

Key words: oxidative phosphorylation, leak, slip, almitrine mechanistic change in stoichiometry, fatty acid, yeast, rat liver,
mitochondria

Address for offprints: M. Rigoulet, IBGC-CNRS, 1, rue Camille Saint-Saens, 33077 Bordeaux cedex, France
36
Introduction
According to the chemiosmotic hypothesis [1, 2] the
energy transduction between redox free energy and phosphate
potential is permitted by inner mitochondrial membrane
proton pumps structurally independent but functionally
connected by a proton electrochemical difference between
two bulk phases: intermembranal space and mitochondrial
matrix (l1IlH+). By analogy with the functioning of an
electric battery, Mitchell introduced the term proton-
motive force (l1p = l1IlWIF) which is largely used in bio-
energetics. Even if some alternative models have been
proposed (see [3~7] for reviews), the assumption that the
bulk l1p is the only intermediate in proton-dependent
energy conversion remains one of the bases of modern
bioenergetics.
One of the main problems remaining is to understand how
the values of the different coupled fluxes are determined in
an integrated system, such as oxidative phosphorylation.
Obviously, the coupling of fluxes is mediated by forces, but
at first sight the quantitative relationships between fluxes
may depend on all properties of the whole system. In a
complex metabolic network like mitochondrial oxidative
phosphorylation, a very simple quantitative analysis is the
determination of the yield of the overall reaction, as is
currently performed by measuring ATP synthesis over
oxygen consumption (ATP/O). With this approach, the yield
may vary and several mechanisms possibly involved have
been proposed.
The first mechanism decreasing the coupling efficiency,
the proton leak, is a direct consequence of the nature of the
energetic intermediary, the protonmotive force. Indeed,
biological membranes always present some proton con-
ductance (LH), and the resulting proton flux is strictly dependent
on protonmotive force (JH = LH.l1p). This membrane con-
ductance is a specific property of the membrane itself, but
is not entirely independent from the protonmotive force: at
a high value of this force, the proton membrane conductance
increases. This fact determining a non-ohmic relationship
between passive proton flux (proton leak) and proton-
motive force has been observed in many kinds of mito-
chondria [8~13], including yeast mitochondria [14].
Obviously, the size of this proton leak may modulate the
yield of oxidative phosphorylation (ATP/O), and it has been
clearly shown that a large group of agents (protonophores)
uncouples oxidative phosphorylation by increasing proton
membrane conductance [15~21]. In the classical definition
of proton leak, this depends only on two factors: (i) the
nature of the membrane and (ii) the size of the proton-
motive force. In this uncoupling process, the first event is
a dissipation of protonmotive force which leads to an
increase in respiratory rate and a decrease in the ATP
synthesis rate, thus leading to a decrease in ATP/O ratio.
Even if much experimental work has established strong
evidence that some protonophoric action can quantitatively
account for the uncoupling of oxidative phosphorylation
(see [22] for review), no definitive proofthat the uncoupling
is exclusively and quantitatively due to an increase in cation
membranal conductance has been obtained. In fact, from a
growing number of data, it is evident that the question of
the multiplicity of uncoupling mechanisms is still largely
open. Among them, two kinds of experimental evidence
can be noted: it has been shown that (i) some uncoupling
effects are not linked to a significant decrease in proton-
motive force, which indicates that the decrease in oxidative
phosphorylation efficiency is not, in this case, the conse-
quence of an increase in membranal proton conductance
[23~30]; (ii) direct or indirect estimations of the coupled
flow through different proton pumps indicate that their
intrinsic stoichiometry i.e. the H+/2e- stoichiometry of the
respiratory chain and the WIATP stoichiometry oftheATP
synthase, may vary as a function of many physical para-
meters or some drug addition [14, 27, 28, 30-42]. This
leads Azzone et al. [43, 44] to propose another mechanism
causing a loss of oxidative phosphorylation yield. Such a
new possibility called slip is a decrease in the
efficiency of a proton pump due to partial and variable
decoupling of chemical reaction and proton transport i.e.
a decrease in the H+IO stoichiometry of the respiratory
chain or an increase in the H+IATP stoichiometry oftheATP
synthase. A kinetic model for proton pump functioning,
using a Hill diagram, has been proposed by Pietro bon and
Caplan [45] (see also Fig. 1): the whole reaction is divided
into two parts, a triangle representing the catalytic pathway
of the chemical reaction and a rectangle representing the
pathway of the protons. In the absence of slip, chemical
reaction and proton transport are closely coupled (cycle
without dashed line) . Slip is due to the possibility of
chemical reaction without concomitant proton movement
or vice versa. Hence, with this mechanism of decrease in
efficiency, another parameter, proton pump activity may
also control the yield of oxidative phosphorylation. An
alternative explanation has been proposed by Garlid et al.
[46]. In that hypothesis, the passive proton flux does not
depend only on the protonmotive force and on the charact-
eristic of the membrane, but also on the activity of the
different membranal proteins. Although in the slipping
process proton pumps only are concerned, in Garlid's hypo-
thesis any membranal protein could possibly be involved in
the proton flux back down the electrochemical proton
gradient. For our purpose, this kind ofleak can be assimilated
to proton slip.
When considering the mechanism of the proton transport
involved in chemiosmotic energy transduction, two different
models have been proposed: a direct coupling, initially
introduced by Mitchell [47], in which translocated protons

INTERNAL EXTERNAL
-nE _ _ _ _ _ _ _ _ _ _ E-n
H n E*
Fig. 1. Six-state model of a proton pump as proposed by D. Pietrobon and
R. Caplan [45]. E is an enzyme bearing n proton binding sites accessible
either to the internal medium or to the external medium. The chemical reaction
S (substrates) P (products) induces a conformational change in E to E *, this
transition leading to a change in the accessibility of the bound protons
(from the internal to the external media). The transition indicated by the
dashed line generates the possibility of two mechanisms of slipping (intrinsic
uncoupling) as discussed in the text.
participate directly in the coupled chemical reaction (redox and
ATP synthesis or hydrolysis) and an indirect coupling, linked
to a conformational change in the enzymatic complex, between
chemical reaction and the proton transport [48]. Undoubtedly,
an intrinsic uncoupling, such as slipping, is in favor of the
existence of an indirect coupling. In this case, other mechan-
isms for changing the efficiency of proton pumps can be
theoretically proposed. One of these possibilities could be a
mechanistic change in the actual stoichiometry, as is described
and discussed in this paper.
Materials and methods
Isolation of mitochondria
Yeast mitochondria
Cells of diploId wild strain Saccharomyces cerevisiae (Yeast
foam) were grown aerobically at 28C in a complete medium
37
(1 % yeast extract, 0.1 % KHl04 and 0.12% (NH4)2S04' pH =
4.5) with 2% D-L lactate as carbon source. The cells were
harvested in logarithmic growth phase and mitochondria
were isolated from spheroplastes as described by Guerin et
al. [49]. Protein concentration was measured by the biuret
method using BSA as standard [50].
Rat liver mitochondria
Rat liver mitochondria were prepared according to [51] in
the following medium: 250 mM sucrose, 1 mM EGTA, 20 mM
tris-HCI, pH = 7.2. Mitochondria protein content was deter-
mined by the biuret method.
For experiments performed with mitochondria isolated from
polyunsaturated fatty acid deficient rats, we fed young male
weaning Wi star rats (60 g) a semi -synthetic diet for at least 4
weeks: Casein 21, D.L. methionine 0.12, com starch 44.26,
sucrose 23.4, cellulose 1.87, mineral mixture 3.3, vitamin mixture
0.94 (% weight). This diet was supplemented with either
stearic and palmitic acid (2.65 plus 2.65% weight; PUFA-
deficient diet) or soya oil (5.3% weight; control diet). Animals
had access to food and tap water ad libitum.
Oxygen consumption rate was measured polarographically
at 27C using a Clark electrode. Yeast mitochondria were
incubated in basal medium (0.65 M mannitol, 0.36 mM EGTA,
3 mM tris-phosphate, 10 mM tris-maleate pH =6.7, 5 ~M RbCI
and 0.1 ~g/ml valinomycin).
Rat liver mitochondria were incubated in 250 mM sucrose, 1
mM EGTA, 10 mMTris-HCI pH 7.2 and different additions as
described in the legend of the figs or tables.
ATP/O ratios with different substrates were measured in
conditions of saturating ADP concentration (1 mM) and
determination ofATP accumulation in time (4 samples every
15 or 30 sec) either by [32P]Pi incorporation in nucIeotides
[52] or by HPLC using a reverse phase (Spherisorb, ODS
II, 5 ~m) column (0.46 x 25 cm) at 30C. Elution was
performed with a 25 mM sodium pyrophosphate/pyro-
phosphoric acid (pH 5.75) buffer at a flow rate of 1.2 mll
min [53]. In this case, netATP synthase flux was obtained
by subtracting AMP production in order to eliminate
adenylate kinase activity. In some cases P/O ratios were
measured by using hexokinase in the presence of 1 mM
MgCI 2, 20 mM glucose and 125 ~M ATP. In this case, ATP
production was monitored by glucose 6-phosphate formation,
which was measured enzymatically with spectrophotometric
determination according to Bergmeyer [54].
Ap determination was performed in parallel experiments.
Matrix space was determined by using [3H]water and inner
membrane impermeable [14C]mannitol, A", and ApH by
distribution of 86Rb (in the presence of valinomycin) and
pH]acetate (or PH]DMO for some experiments), respectively
[55]. After equilibration of radioactive probes, mitochondria
were separated from the medium by centrifugation (20 s,
12000 g) through a silicone oil layer (Wacker AR200).
38
ATPase activity was assayed under particular conditions
as described in the legends to the figs or tables. At defined
times, 0.25 ml of mitochondrial suspension was quenched
into HCIOifinal concentration 10% mass/vol) and 0.1 mM
EDTA. As indicated above ATP, ADP and AMP were deter-
mined in the protein-free neutralized extract by HPLC and
Pi was measured according to [56].
Mitochondrial matrix energetic swelling is a very simple
system to investigate the efficiency of a proton pump (either
respiratory chain or ATP synthase). The H+ efflux catalyzed
by a given proton pump sustains a A'I'-dependent potassium
accumulation and a ApH-dependent acetate or phosphate
accumulation. Thus, when mitochondria are suspended in basal
medium supplemented with 10 mM potassium phosphate or
potassium acetate in the presence of a potassium ionophore
(valinomycin), energy supply (either redox potential or
ATP) is converted into salt accumulation thus inducing
mitochondrial swelling. The mitochondrial volume is
monitored by using an Eppendorf spectrophotometer
(wavelength: 540nm).
ATP, ADP, Pi, hexokinase, free fatty acid bovine albumin,
NADH were purchased from Boehringer (Meylan, France),
succinic acid, rotenone, EGTA, CCCP, valinomycin, glutamic
acid, oligomycin, malonic acid from Sigma (L'Isle d' Abeau,
France); Tris, HCl, malic acid, MgCl 2 from Merck (Nogent
sur Marne, France) and labelled compounds from Amersham
(Les VIis, France).
Results and discussion
The oxidative phosphorylation yield in yeast
mitochondria depends on both respiratory chain flux
and kinetic constraints
When compared to mitochondria from mammals, yeast
mitochondria isolated from Saccharomyces cerevisiae
present three main differences: (i) lack of phosphorylation
site corresponding to coupling site 1 of mammalian
mitochondria [57, 58], (ii) ability to oxidize exogenous
NADH by NADH dehydrogenase located toward the outer
surface of the inner membrane [59], (iii) the ability to
oxidize lactate by using directly the third site span [58]. These
last two properties can be used as a tool in order to limit the
electron supply to the respiratory chain in a true steady
state. For instance, it is easy in such a system to limit the
respiratory rate by using a NADH regenerating system as
presented in Fig. 2, where glucose-6-phosphate and NAD+
were at saturating concentration. The addition of various
amounts of glucose-6-phosphate dehydrogenase induces
different steady states of respiration. As shown in Fig. 3A,
the non-phosphorylating respiration (also called state 4)
increased as a function of the glucose-6-phosphate de-
hydrogenase concentration to reach a maximal value of about
300 natom oxygen/min/mg protein for 0.5 unit/ml of added
enzyme. This value and the corresponding protonmotive force
are identical to those measured in state 4 when saturating
NADH concentration is used as respiratory substrate instead
ofNADH regenerating system (not shown) . In the presence
of either a protonophore or ADP, the respiratory rate was not
modified for low concentrations of glucose-6-phosphate
dehydrogenase (up to 0.1 unit/ml). For higher concentrations
of this enzyme, respiratory rate was stimulated. This shows
that up to 0.1 unit/ml, glucose-6-phosphate dehydrogenase
exerts a full kinetic control on oxidative flux. Another
possibility for modulating the respiratory rate under coupled
or uncoupled respiration was inhibitor titration at saturating
concentration of the substrate i.e. NADH. In Fig. 3B, such a
titration with myxothiazol (inhibitor at the bel complex) was
performed. Even when respiratory rate was very low in state
4, it was always largely stimulated by protonmotive force
dissipation with protonophore uncoupler (CCCP) or by
phosphorylation (ADP addition).
At various steady states obtained by different amounts of
glucose-6-phosphate dehydrogenase, respiratory and ATP
synthesis rates were measured. As shown in Fig. 4A, theATP/
o ratio decreased as a function of the respiratory rate from
1.8 when J02 was 100 natom oxygen/min/mg protein to 1.0
when J02 was 400 natom oxygen/min/mg protein, theATP/O
ratio remaining constant for higher respiratory rates. A change
inATP/O stoichiometry indicates either a modification in the
stoichiometry of at least one of the proton pumps involved
(respiratory chain or ATP synthase) or a variation in proton
leak. However, it is clear from Fig. 4A that protonmotive
force measured during phosphorylation is constant when the
respiratory rate increases up to 400 natom oxygen/min/mg
protein. Above this value, Ap decreased when respiratory rate
increased. The main observation was that ATP/O decreased
in a flux-dependent manner while protonmotive force was
constant. Therefore, this change in ATP/O ratio cannot be a
consequence of proton leak modification but must be due to
a change in either the W/2e- stoichiometry of the respiratory
chain or the H+/ATP stoichiometry of ATP synthase or both.
In Fig. 4B, it is shown that myxothiazol titration of oxidative
phosphorylation does not significantly changeATP/O ratio
or Ap under a wide range of respiratory rates. This is
classical in respiratory chain inhibitor titration [60, 61]. The
control ofthe respiratory rate by limiting the NADH supply
with a NADH regenerating system is a completely different
situation as compared to the use of inhibitor titration (see
Fig. 5). In fact, the method leads to a modulation of the
electron flux through each respiratory chain unit, while in
the second procedure the number of functional units is
decreased in the presence of a saturating concentration of
NADH: thus, through each functional respiratory chain unit

the electron flux remains maximal. Consequently, the first
method appears to be the only way to investigate the actual
relationship between the change in flux and the coupling
efficiency of the proton pumps.
As ATP/O ratio depends on both respiratory and phos-
phorylation rates, three different fluxes through this pathway
must be considered at each steady state: electron and proton
movements and ATP synthesis. Even if proton flux cannot
be directly measured, it is easy to calculate the control
exerted on electron andATP fluxes by the enzyme producing
NADH. In a metabolic pathway under steady state, the flux
control coefficient (CD of an enzymatic step (E) on the E
whole flux (1) is defined as follows:
[62, 63]
Figure 6 shows the evolution oftheATP/O ratio and the flux
control coefficient of glucose-6-phosphate dehydrogenase
over respiratory and phosphorylation rates as a function of
respiration. When respiratory rate increased from 1OO~
400 natom oxygen/min/mg protein, ATP/O decreased as
previously shown. The flux control coefficient of glucose-
6-phosphate dehydrogenase over respiration decreased
slowly but remained very high (from 0.95--D.86), whereas
Glucose-6-P + NAD+
Glucose-6-P dehydrogenase
from leuconostoc mesenteroiaes
(NAD+ -dependent)
6-P-gluconate + NADH + H+
Fig. 2. Scheme ofNADH-regenerating system used. NAD+ (2 mM) and glucose-6-phosphate (4 mM) were at saturating concentration. Various amounts of
glucose-6-phosphate dehydrogenase led to different steady states of NADH and respiratory rate.
39
over phosphorylation it increased (from 0.43~0.65).
Hence, the kinetic control exerted by glucose-6-phosphate
dehydrogenase on either electron flux or ATP synthesis rate
is different and changes in an opposite manner over this
oxygen consumption rate range. Thus, an increase in
efficiency of oxydative phsophorylation is linked to both a
low rate of respiration and a nearly total kinetic control of
oxidative flux by the electron supply. In contrast, the control
exerted on the phosphorylation rate remains distributed over
different steps. A positive control was exerted by the
electron supply while a proton leak led to a negative value.
This flux-yield dependence of oxidative phosphorylation
in yeast mitochondria is a general characteristic of this
system. Indeed, as already reported, by using different
substrates leading to matricial NADH formation, very
different respiratory rates in state 3 may be obtained [40].
These differences are due to different kinetic limitations of
the various dehydrogenases. However, the same dependence
between ATP/O ratio and respiratory rate was observed in
both permeabilized spheroplasts and isolated mitochondria,
even if the curves are not entirely superposable (not shown
but see Fig. I in [64]).
From previous studies in our laboratory, it is well
established that in yeast mitochondria, the ATP/O ratios
obtained by classical procedures with respiratory chain
inhibitor titration are in agreement with those found in
mammalian mitochondria: 0.5 and 1 respectively for sites 2
out in
Inner
Membrane
NADH
dehydrogenase
40

A
1200

..- 1000
....=
.~

0
I.
Q..
bn
.
800
/0
~o
..--0
---- 0_

/0
]= 600
0 /0
E
....0
~
400
1

= 0

,
'-'

0"-
0 ' ....... - - - - - . - --. --. ----. - .
... 200
0 ~.
"""
0
0 QS 1 I.S
Glucose-6-P dehydrogenase (unit Iml)

1200
B
1000
-0-
0 .......
'0
800

600
\
\0
0"-"-
400

... .-.-.-.-.......
o
"""
200

o
.-.
. . . . . . . -- ----..0
~o

o O.OS 0.1 O.IS

Myxothiazol (pg Iml)

Fig. 3. Dependence of respiratory rates on either glucose-6-phosphate dehydrogenase or myxothiazol concentrations. Mitochondria were suspended in 1.5
ml respiratory medium (see Materials and methods section) containing a NADH-regenerating system (as presented in Fig. 2) and in (A) with indicated amount
of glucose-6-phosphate dehydrogenase. For a given enzyme concentration, respiratory rate was determined at steady state: state 4 (e); state 3 (0) with I
mM ADP. (B) glucose-6-phosphate dehydrogenase was at 1.5 unit/ml and respiratory rate under state 4 (e); state 3 (0) was modified by addition of indicated
amounts of myxothiazol.
 41

A B
Change in [Glucose-6-P dehydrogenase] Myxothiazol Titration

200 0 2000
2 .2

~
o ~
"0
0
3'
=:..
3' Eo< ~
-<
........
~
....
0'" ...
.... 0

. .-.
.... o ~ 0 __ _

---.- .. _-_.-.
., ..... ' 0 oO--o~------O
~ CD 100 1 100

o o o ~i--~--Ti--~--~i--~--~ir-~---ri--~--'i 0
o 200 400 600 800 1000
o 250 500
J02 (natom 0 Imin Img protein)
750 1000
JOz (natom Imin Img protein)
Fig. 4. Dependence of ATP/O andL'.p on respiratory rate. Experimental conditions were as in Fig. 3 in the presence of ADP (1 mM) . For a given concentration
of enzyme (A) or at high glucose-6-phosphate dehydrogenase concentration, for a gi ven amount of myxothiazol (B) ATP10 (.) and L'.p (0) were determined
as described in Materials and methods section.

out in out
in

I
--+--"jf-+---~ H+ Substrate
~--~----~~H+
of ~--~~--~~H+ Limitation

'---+-----f-... H+

Flux titration Flux titration

by respiratory chain inhibition by [substrate] limitation

Fig. 5. Scheme of two different kinds of change in respiratory chain activity. In the first experiment (respiratory chain inhibitor titration) the number of active
chain units was decreased without any alteration in the electron flux through either of them. In the second procedure (substrate limitation), all the respiratory
chain units remained active even if the electron flux catalyzed was decreased (see text).
42
2 below), studies on the ATPase/ATP-synthase complex offer
o the possibility to investigate the exact nature of the increase
\ o
\o
\ of the reaction (i.e. ATP synthesis or ATP hydrolysis). In
~
0,
0 .......
0- _ proton pump, also called slipping, must induce an increase
e-e_ 0 0-"0-0-0-000-00
.- _e--.e---...
.... ..........
+-+-+-+-+- . ,
+........ +.....
+~ +~.
+...
o~----~------~----~------~----~
o 200 400 600 800 1000
J0 2 (natom 0 Imin Img protein)
Fig. 6. Relationships between either the ATP/O ratio or the flux control
coefficients and the respiratory rate. Respiration in state 3 was modulated
by different concentrations of glucose-6-phosphate dehydrogenase in the
presence of2 mM NAD+ and 4 mM glucose-6-phosphate. The ATP/O ratio
(0) was from Fig. 4A. The flux control coefficients of glucose-6-phosphate
dehydrogenase either on respiratory (e) or on phosphorylation (+) rates
were calculated as indicated in the text.
and 3 [65]. However, the data reported in this paper indicate
that when the electron flux through each respiratory chain unit
decreases, the ATP/O ratio increases irrespective of the
protonmotive force. Thus, the value ofATP/O measured under
state 3 is not the highest one. Moreover, the highest value of
ATP/O ratio is linked to a nearly total kinetic control of
oxidative activity upstream of the respiratory chain. The nature
of the mechanism involved here is very likely to be slipping
at the level of the respiratory chain, which would decrease with
the flux. Indeed, many works have shown that respiratory chain
slipping is due to an increase in protonmotive force (see
above), but, as predicted from the redox slipping model [45]
we found here that it may also depend on the flux.
Change in ATPaselATP synthase efficiency: effect of
almUrine
As with the effect of redox slipping reported on the
respiratory chain, some slipping effects have also been
reported on ATP synthase [31, 33, 39]. Although it is not
possible to discriminate between a pure slipping process and
a decrease in H+/2e- stoichiometry in the respiratory chain
because of the irreversible nature of the reaction (see
in H+/ATP ratio. Indeed, a change in proton ATPase/ATP
synthase stoichiometry could be interpreted as an increase
in slipping if coupling efficiency between the proton flux
and chemical reaction is decreased, whatever the direction
other words, the increase in intrinsic uncoupling of this
in H+IATP ratio for ATP synthesis and conversely a decrease
in WIATP ratio during ATP hydrolysis. So whatever the
direction of the reaction, the process always results in a
decrease in coupling.
,+ An experimental proof of an actual mechanistic change
.,+
.+ in stoichiometry would be given by the observation of an
'\,. identical change in stoichiometry both in the forward and
reverse chemical reaction. Hence, in the situation where the
WIATP ratio increases (as is the case in a slipping process),
the same variation ofthe ratio must be observed duringATP
hydrolysis: for each ATP hydrolyzed a higher amount of
proton should be extruded.
Almitrine is a drug which is used for the treatment of
patients suffering from hypoxia due to chronic lung diseases
[66]. The study of the effects of this drug on isolated
hepatocytes have shown that the effect on the energetic
metabolism could be explained by an increased slipping at
the level of ATPase: we have observed a decrease in ATPI
ADP ratio in both cytosol and mitochondria without any
change in either respiratory rate or in mitochondrial i1'1'
[66,67]. This hypothetical increase in WIATP ratio has
been confirmed on isolated mitochondria (Fig. 7). Indeed,
almitrine addition does not change respiratory rate or
protonmotive force but largely decreases the ATP synthesis
rate leading to a lower ATP/O ratio. It is important to
remember that as already explained above, in a situation
where ATP synthase flux results in a net ATP synthesis, it
is not possible from an increase in the H+/ATP ratio to
discriminate between a slipping effect in this proton pump
and a mechanistic change in stoichiometry. As largely
discussed above, the H+/ATP ratio measured in the reverse
reaction (ATP hydrolysis) must solve this question. As
described in the materials and methods section, a very
simple system for testing the efficiency of a mitochondrial
proton pump is energy-linked swelling. Figure 8A shows that
almitrine addition does not change the extent of the swelling
of rat liver mitochondria induced by valinomycin in potassium
phosphate salt when the respiratory chain proton pumps
generate the driving force. In contrast (Fig. 8B), when the
driving force is generated by ATP hydrolysis throughATPase
activity, almitrine induces a dramatic increase in the extent
of mitochondrial swelling. Moreover, this increase in
swelling is associated with a decrease in ATPase activity as
shown in Fig. 9 where the ratio between the extent of swelling
 43

1.5

....
.~
=
r..

--
0 1
~

~
0.5

o~----~------~------~------~----~~----~
o 5 10 15 20 25 30
[Almitrine] (PM)

Fig. 7 Effect of almitrine on the ATP/O ratio. Rat liver mitochondria (3 mg protein) were suspended in the following medium: 225 mM sucrose, I mM EGTA,
10 mM tris-HCI, (PH 7.2), I mM malate, 5 11M rotenone, 10 mM Tris-Pi and 10 mM succinate in the presence of non-limiting amounts of hexokinase, I mM MgCI,
and 10 mM glucose at 28C. Oxygen consumption rate was measured polarographically by using a Clark electrode and ATP production was monitored by
glucose-6-phosphate formation as described in Materials and methods section.

A (in arbitrary units) is divided by the ATPase activity. It is

B clear that almitrine addition increases this ratio in a dose-
dependent manner. As explained in the materials and
methods section, the extent of swelling depends on the salt
accumulation (salt )saltout ratio, see also [68]). Hence, in
.....
, the presence of almitrine, a lower ATP hydrolysis maintains

,,,
/
/
a higher protonmotive force since the swelling extent is

,,,
I
I
, higher. This clearly indicates that almitrine increases the H+/

, ATP ratio also in the reverse direction: i.e. net ATPase

,,
I

,,,
I activity. This is in accordance with the hypothesis of a change

\ rt
I
\
\ at the actual mechanistic stoichiometry of the ATPase/ATP
\ synthase proton pump.
-J \1
t
CCCP
CCCP

O.1AL Polyunsaturated fatty acid (PUFA) deficiency induces

1min
Fig. 8. Effect of almitrine on the energy-linked swelling of mitochondria in
potassium salt. Swelling was monitored at 546 nm by using an Eppendorf
photometer. Rat liver mitochondria (I mg protein) were incubated in 1 ml of
the following medium 225 mM sucrose, I mMEGTA, 2 mMMgCI2 and 10
mM tris-HCI (pH 7.2) As indicated on the curves, either 5 mM succinate
(plus 20 11M oligomycin) (A) or 2 mM ATP (plus 5 11M rotenone) (B) and \0
mM potassium phosphate plus 2 nmol valinomycin and I 11M CCCP in the
absence (-) or in the presence (----) of20 11M almitrine, were added.
an increase in both non-ohmic proton leak and redox
slipping
As previously reported in the literature, dietary PUFA
deficiency is responsible for a large change in membrane
phospholipid composition including that of mitochondria
[69-73]. It was long recognized that PUFA deficiency affects
energy metabolism [74-77]. On isolated mitochondria an
increased respiratory rate in non-phosphorylating mito-
chondria was reported [78] while the P/O ratio was either
unchanged or decreased [72, 79-82]. We have reinvestigated
44

.---
7

6 .-
./
<Il
"!il
~ 5
.ac-...
Q

.==
~"

4
~~
<E
-.c 3
~
.........

~r..
-bi)$
2
:5
'ii
~
00 1

0
0 5 10 15 20 25 30
[Almitrine] (PM)

Fig. 9. Effect of almitrine on the efficiency of oligomycin-sensitive ATPase. The efficiency of oligomycin-sensitive ATPase was estimated by the ratio
of the degree of swelling on oligomycin-sensitive ATPase activity and expressed as absorbance unitlnmol ATP per min per mg protein. Swelling was
monitored as described in the legend of Fig. 8. ATPase activity was measured when swelling was maximum and stable.

the consequences of PUFA deficiency on some oxidative
phosphorylation parameters. Figure 10 shows that state 4
respiratory rate was increased by PUFA deficiency, while
state 3 was not significantly affected. In addition, mito-
chondria from the PUFA-deficient group exhibited a large
increase in matrix volume which was more pronounced in
state 3 than in state 4. In state 4, protonmotive force was
slightly but significantly decreased whereas in state 3 the
decline in protonmotive force was much larger (40 mV). The
increased state 4 respiratory rate associated with a decrease
in protonmotive force may indicate an increase in proton
leak. To further investigate the relationship between
respiratory rate and protonmotive force, non-phosphorylating
mitochondria were titrated with malonate, a competitive
inhibitor of succinodehydrogenase. Figure 11 shows that as
classically reported since the Nicholl's first observation [8],
the relationship between respiratory rate and protonmotive
force is non-linear, thus making it possible to define both an
ohmic and a non-ohmic part ofthe relationship. Surprisingly,
the non-ohmicity of the curve was more pronounced in
PUFA-deficient mitochondria and the curve was shifted to
the left. Indeed, it has been theoretically predicted and
experimentally proven [83, 84] that the non-ohmicity is
strongly reduced by small amounts of protonophoric
uncoupler, which lead to an increase in state 4 respiratory
rate together with a decrease in protonmotive force as
reported here. The protonmotive force of PUFA-deficient
mitochondria was lower for each respiratory rate than that
of controls. Moreover, the non-ohmic part was observed
over a larger respiratory rate range in PUFA-deficient
mitochondria. These data led us to exclude a single proton-
ophoric effect (increase in ohmic leak), whereas a con-
sequence of a change in non-ohmic proton leak might well
be the case. Besides these effects in non-phosphorylating
mitochondria, the main problem is the possible modification
of the efficiency of oxidative phosphorylation since, as
shown in Fig. 10, state 3 respiratory rate was unaffected
while protonmotive force was largely decreased. To avoid
intrinsic uncoupling, it is generally accepted that ATP/O
stoichiometry must be determined by respiratory chain
inhibitor titration [85]; the slope of the linear relationship
between ATP synthesis and respiratory rate gives the actual
stoichiometry. Figure 12A shows that the experimental
points obtained by mitochondria from either PUFA-deficient
or control rats with antimycin titration are distributed on the
same straight line. From these data, we conclude that at
saturating ADP concentration the ATP/O was not affected,
even if the protonmotive forces were different. This clearly
indicates that the losses (proton leak and/or redox slip) are
identical while the protonmotive forces are different. Hence,
for an identical protonmotive force, this loss must be higher
in the PUFA-deficient group. In such experimental conditions
the flux through each active respiratory chain complex is
always maximal, even if the number of active complexes
 45

30 200
,-.., A ,-..,
C
B
C .Ql
.Ql
....
e
Q.
'0
"'"
Q.

-]
CIl CIl
e 20 ~
C C
] 100
0 0
10 e
Q

=
C
'-'
~
C
'-'
N
N
0 0
""" 0 """ 0

1.5 C 1.5 D
,-.., ,-..,
c .5
1a 1.0 .s
a 1.0
Q

CIl CIl
~ ~
'3. '3.
'-' '-'

e 0.5
ell ell
e 0.5
=
Q Q =
;> ;>

0 0

300
E 200
F

200
,-..,
;> ,-..,
e
'-'
;>
el00
'-'
~ ~
100

o o

Fig. 10. Oxygen consumption, matrix volume and protonmotive force on isolated mitochondria of control and PUFA-deficient rats. For oxygen consumption
rate determination, rat liver mitochondria (1 mg proteinlml) were suspended in the isolated medium described in Materials and methods section supplemented
with 5 mM tris-Pi, 5 mM tris-succinate plus 5 J.lM rotenone. State 3 respirations were obtained after addition of 200 J.lM ADP. Matrix volumes and ~p were
measured in the same medium as described in Material and methods section. (A) oxygen consumption at state 4; (B) oxygen consumption at state 3; (C) matrix
volume at state 4 (D) matrix volume at state 3; (E) protonmotive force at state 4 (F) protonmotive force at state 3. Mitochondria isolated from control (white)
and PUFA-deficient (hachured) rat livers.
46
30
-..
.-....cuc
Q
s..
C.
bIl 20
e
......
.-c
.e
0
e 10
....
Q
=
C
'-'
M
0 14::~::::::;;;--
~ phosphorylation. Concerning the exact nature of these
0 "
0 50 100 150 200 250
Ap (mV)
Fig. 11. Relationship between oxygen conswnption rate and protonrnotive
force in isolated liver mitochondria from control and PUFA-deficient rats.
Rat liver mitochondria (2 mg protein/ml) were suspended in the following
mediwn: 250 mM sucrose, I mM EGTA, 10 mM tris-HCI, 5 mM tris-succinate,
5 mM tris-phosphate, 5 I!M rotenone, O. 02 I!g/mg protein valinomycin, 2
I!g/mg protein oligomycin, 10 I!g RbCI; pH = 7.2; 25C. The respiration rate
was modulated by the addition of Tris-malonates. L1p measurements were
performed in parallel experiments, in the same conditions, as described in
Materials and methods section. Mitochondria isolated from control (L1) and
PUFA-deficient (.A.) rat livers.
changes with inhibitor additions (see above). Thus, such
conditions are far from the actual intracellular situation in
which the steady state of oxidative phosphorylation is
between state 3 and 4. To avoid this problem and to measure
ATP/O ratio in a more physiological manner, it is possible
to manipulate oxidative phosphorylation by generating
different steady states of ADP concentration with hexo-
kinase titration. These results are presented in Fig. l2B
which shows two different and parallel relationships, the line
from the PUFA-deficient group being shifted to the right,
thus indicating a lower ATP/O ratio. It is very interesting to
observe such an apparent discrepancy between the unique
relationship in state 3 (identical P/O, see Fig. l2A) and a
double relationship with hexokinase (lower P/O ratio in the
PUFA-deficient group, see Fig. l2B) . In the first case the
greater losses in the PUFA group are compensated by a
lower protonmotive force, while in the second case this loss
is always higher. By using such an approach, Westerhoff and
van Dam proposed to discriminate between ohmic leak and
slipping by comparing the relationships between ATP
synthesis and respiratory rate [86]. When a slipping-like
effect was experimentally induced, they observed parallel
relationships as predicted by the model [87] . Since this is
the case in Fig. 12B it can be proposed that the nature of the
loss during the phosphorylation process is essentially due
to a slipping in the mitochondria of PUFA-deficient rats.
These results in PUFA-deficent mitochondria suggest the
working hypothesis that these mitochondria present an
increase in both non-ohmic proton leak and slipping. Of
course, the respective role of both processes depends on
protonmotive force and on the rate of oxidative phos-
phorylation. In state 4 of isolated mitochondria, the
modification in non-ohmic proton leak appears to be pre-
dominant, while in state 3 the effect on slipping plays the
main role. It is difficult to extrapolate from these in vitro
studies the actual physiological role of both processes. Indeed,
many parameters may modulate the relative contribution of
these two effects in the actual efficiency of oxidative
effects, both the lipid composition of the membrane and
changes in mitochondria volume may be involved. The
effects of the change in the phospholipid composition ofthe
mitochondrial membrane on the oxidative phosphorylation
pathway have been largely investigated, but no relevant
results indicate clear conclusions for the present study.
Besides membrane fatty acid composition, changes in
mitochondrial volume led to another possible explanation
of the effect of PUFA deficiency. Indeed, a large increase
in mitochondrial volume after a dietary lipid change is a
prominent feature in this work which confirms other data
from the literature [88, 89]. The link between mitochondrial
volume and metabolism is of great interest. We have recently
reported that an increase in matricial volume due to hypo-
osmotic exposure in normal rat liver mitochondria involves
some changes in oxidative phosphorylation very close to the
findings presented here. A similar increase in matricial
volume enhances state 4 respiration together with a decrease
in protonmotive force, such changes being accompanied by
a decrease in state 3 protonmotive force without any change
inATP/O [90]. Preliminary experiments performed in order
to discriminate between the effects due either to volume
increase or change in membrane lipid composition in PUFA
deficiency indicate that the change in matrix volume is not
completely responsible for the oxidative phosphorylation
modifications in these mitochondria.
Structure of proton pumps and putative mechanism of
coupling
In conceptual models of proton translocation, the proton-
chemical reaction coupling may be direct or indirect. Direct
models for redox proton pumps initially proposed by
Mitchell [1,2, 91] stipulated that protons were directly
carried by redox elements such as quinones, tlavins and the
dioxygenlwater couple, which are disposed in the membrane
in such a manner that the redox dependence on protonationl
 47

250 A B
140

C
;('
'Cii 200 C 120
....0 '....
Cii

/
r..
Q.
OIl
a
0
100
8 150 OIl
.......
-E
's=
80

.......
'0 100
's=
....... 60
8 '0
=
'-'
5
8
40
c:lo. c:lo.
E-- 50 E--
-<
~
-<
~
20

0 0
0
J0 2 (natom
100
Imin Img protein)
200 0 20 40
J0 2 (natom 60 80
Imin Img protein)
100 120

Fig. 12. Relationship between ATP synthesis and oxygen consumption rate in isolated liver mitochondria from control and PUFA-deficient rats when titrated
by antimycin (A) or by hexokinase (B). Rat liver mitochondria (I mg protein/ml) were suspended in the isolated medium described in Materials and methods
section supplemented with 5 mM tris-phosphate, 5 ).1M rotenone, 20 mM glucose, I mM MgC I,. In Fig. 12A, after addition of I mM ADP, ATP synthesis flux
was modulated by different amounts of antimycin. In Figure 12B, after addition of 125 ).1M ATP, ATP synthesis flux was modulated by different concentrations
of hexokinase. Mitochondria isolated from control (open symbols) and PUFA-deficient (filled symbols) rat livers.

deprotonation results in a translocation from the inside to
the outside of the mitochondria (redox loop) . This model
predicts a fixed stoichiometry Wle- of 1. It is accepted that
the example of such an arrangement is the quinone cycle
mechanism for the cytochrome bel and b6f complexes [47,
92] The consequence of this direct mechanism is that the
protonmotive force value, in the absence of enzymatic
H+-gradient consuming pathways, depends exclusively on
the redox span and the membrane proton permeability
coefficient.
The most recent data concerning ATP synthase or cyto-
chrome oxidase mechanism studies support the indirect
coupling view. For instance, a model for energy coupling by
ATP synthase is based on the following assumptions [93-95]:
(i) proton translocation is needed to promote ATP release
from a high affinity site where it forms spontaneously; (ii)
substrate binding is also associated with the energization
step; (iii) both steps occur simultaneously at separate but
interacting sites; (iv) the binding change is due to the rotation
of the y subunit which modifies the structure of the catalytic
sites. From this model, a mechanistic H+/ATP stoichiometry
change should be related to the energy required to induce a
rotation of the subunit, which involves the number and
nature of the different bonds existing between the subunits
concerned. It is not difficult to admit that this energy can
vary with a modification of the conformational state of the
enzyme or with a change in its surrounding. Moreover, this
indirect conformational coupling betweenATP synthesis and
proton flow can lead to slippage if there is no perfect
concordance between the high and low affinity conform-
ational state transition of the catalytic site and the presence
of ATP (remember that ADP. Pi conversion to ATP at this
site is reversible).
Cytochrome c oxidase or ubiquinol oxidase are now
considered as proton pumps ([96, 97] for reviews). Protons
are required for two different purposes: chemical
protons are consumed upon reduction of dioxygen by 4
electrons to form 2 water molecules; the H+/e- stoichiometry
of 1 is fixed by the chemical reaction and cannot vary. The
pumped protons are translocated from inside to outside and
the stoichiometry H+/e- must depend on the energy made
available by the redox reaction and on the way in which
electron and proton flux are coupled. An important observa-
tion was that the substitution of asparagine for aspartate-l35
in subunit I of ubi qui no I oxidase of E. coli abolished proton
pumping with little change in electron transfer activities [98,
99]. This observation leads to two predictions: (i) two
separate proton pathways may exist, one for chemical
and the other for pumped protons; H+/e- stoichiometry
of pumped protons may vary from 0 to its maximal value. This
prediction was supported by the structural determination of
the cytochrome oxidase in both Para coccus denitrificans
[100] and heart bovine mitochondria [10 I]. The notion of
stoichiometry changes according to the conditions applied
to the system was significantly supported by reports on
reconstituted cytochrome c oxidase, where stoichiometry
48

k;
I
AT

------------~---
f: ~

Fig. 13. Working hypothesis of change in A TPase/ ATP synthase induced by almitrine. As it is generally admitted, the main energy span in the enzyme cycle
corresponds to the binding or to the release of ATP to its binding site. Hence in absence of almitrine, n protons are involved in this process. In our hypothesis
the affinity of ATP toward its binding site is increased by almitrine addition and consequently, the energy span becomes higher. So, a larger number of protons
(n' + n) is involved in this step.

varies according to the electron flux and the L1pH value [41]
and the medium composition in adenylic nucleotides [102].
Combining their results with the structural information
available, Papa et al. proposed that the actual H+Ie- in oxidase
is determined by the relative contribution of two electron
transfer pathways, one coupled to proton pumping, the other
not [41, 97]. The relative contributions of these pathways
should be determined by kinetic and thermodynamic factors.
This view can explain our results obtained with yeast mito-
chondria and particularly those in Fig. 4. Effectively, at
constant protonmotive force, ATP/O decreased when the
electron flux rose (in part due to a decrease in cytochrome c
oxidase H+/e- ratio) to obtain a minimal value, and a further
increase in respiration rate at constantATP/O was accompanied
by a decrease in protonmotive force (see also [65]). Another
model for redox slippage is that the rate of the conformational
transition of the protein residue(s) involved in the move of
the proton binding site from the input to the output states is
not completely synchronized with the electron flux through
the oxidase. Whatever the mechanism, slippage should depend
on both electron rate and protonmotive force, of which the
latter acts as a resistance to proton efflux.
Contrary to a redox loop (see above), the maximal value
of the protonmotive force established by a proton pump, in
the absence of a consuming system, should depend not only
on redox span and membrane H' conductance but also on
slippage. This means that oxidase alone should maintain a
lower L1p than bel complex does. In such an explanation, it
is interesting to note that if optimal H+/e- is lower in the bel
complex than in oxidase, this complex maintains a lower L1p
value under steady state and at the same electron flux.
Conclusion
While the model of chemio-osmotic coupling proposed by
Mitchell is widely accepted, the mechanisms responsible for
the actual efficiency of oxidative phosphorylation (energy
transduction from redox potential to phosphate potential)
are widely debated (see Introduction section). The first
mechanism which was very early identified in affecting the
efficiency of the coupling was the proton and/or cation leak.
In its classical sense, the leak depends on two parameters:
the value of the protonmotive force and the mitochondrial

inner membrane conductance. With protonophores, the
effects of an increased leak have been well characterized in
different systems. Moreover, the observation of some
changes in oxidative phosphorylation efficiency without any
change in protonmotive force or in mitochondrial membrane
composition suggests the existence of alternative mechanisms
of efficiency control. One of these alternative mechanisms has
been described as slipping : intrinsic uncoupling in the
proton pumps [43,44]. Our data on yeast mitochondria show
that when the oxidative phosphorylation flux is modulated
by a steady state supply of NADH at non-saturating con-
centrations, the ATP/O ratio increases while the flux
decreases at constant protonmotive force: the highest ATPI
a efficiency was observed at a nearly total kinetic control
upstream of the respiratory chain. The use of inhibitors is
of no interest for such a purpose since they make it possible
to manipulate the flux through the respiratory chain only by
decreasing the number of active complexes without changing
the flux through the remaining active complexes. Such a
decrease in efficiency while oxidative phosphorylation
increases at constant protonmotive force can be explained
in the framework of the slipping model.
Another explanation for the observation of a change in
efficiency at constant protonmotive force could be an actual
change in the mechanistic stoichiometry of the proton
pumps. As discussed in the present work, the only possibility
to discriminate between slipping and an actual mechanistic
change in stoichiometry at a given proton pump level is to
compare the effects in the forward and reverse reaction.
Concerning the almitrine effect, we found an identical
change in WIATP stoichiometry in both ATP synthesis and
ATP hydrolysis reactions. Hence, we conclude in an actual
change in the mechanistic stoichiometry of ATPasel ATP
synthase due to almitrine. The mechanism whereby almitrine
induces a change in H+IATP ratio remains an open question.
It is generally proposed that ATP synthase works through a
reaction cycle in which conformational transitions of the
complex corresponding to the energy span occur [93-95].
In this view the energy dependent conformational cycle is
coupled to proton flux through the complex. As a working
hypothesis, one may propose that almitrine modifies at least
a given state of the enzyme into the cycle in such a way that
the energy span between this and the following is increased
(Fig. 13). Indeed, in the presence of almitrine more energy
is necessary to synthesize one mol ofATP, at a given ilp, and
reciprocally more energy is delivered by the hydrolysis of
one mol of ATP.
Mitochondria in PUFA deficiency is a complex model for
the study of oxidative phosphorylation efficiency since many
parameters are simultaneously affected. Although the data
in state 4 indicate an increase in the non-ohmic leak, the
ohmic part of the leak is clearly not affected. In phos-
phorylation conditions, the main process able to affect the
49
yield of oxidative phosphorylation is the slipping. In such a
case where both non-ohmic leak and slipping are affected,
the respective role of these two mechanisms of energy loss
may largely vary, according to the values of the protomotive
force and respiratory flux. Owing to the complexity of the
cellular network, it is not possible a priori to link precisely
a given mechanism of modification to a specific pattern of
cellular metabolic change. It is highly probable that in intact
cells a pure mechanism of efficiency change almost never
occurs alone.
Acknowledgements
The authors wish to thank Dr. R. Cooke for his contribution
to the editing of the manuscript. This work was supported
by grants from the Conseil Regional d' Aquitaine and the
Pole Medicament Aquitain.
References
1. Mitchell P: Coupling of phosphorylation to electron and hydrogen
transfer by a chemi-osmotic type of mechanism. Nature 191: 144-148,
1%1
2. Mitchell P: Vectorial chemistry and the molecular mechanisms of
chemiosmotic coupling: Power transmission by proticity. Biochem Soc
Trans 4: 399-430, 1976
3. Williams RJP: The history and the hypotheses concerning ATP-
formation by energised protons. FEBS Lett 85: 9--19,1978
4. Kell DB: On the functional proton current pathway of electron transport
phosphorylation: An electronic view. Biochim BiophysActa 549: 55-
99,1979
5. Westerhoff HV, Melandri BA, Venturoli G, Azzone GF, Kell DB: A
minimal hypothesis for membrane-linked free-energy transduction: The
role of independent, small, coupling units. Biochim Biophys Acta 768:
257-292,1984
6. Slater EC, Berden JA, Herweijer MA: A hypothesis for the mechanism
of respiratory-chain phosphorylation not involving the electrochemical
gradient of protons as obligatory intermediate. Biochim Biophys Acta
811:217-231,1985
7. Rottenberg H: Decoupling of oxidative phosphorylation and photo-
phosphorylation. Biochim Biophys Acta 1018: 1-17, 1990
8. Nicholls DG: The influence of respiration and ATP hydrolysis on the
proton-electrochemical gradient across the inner membrane of rat -liver
mitochondria as determined by ion distribution. Eur J Biochem 50:
305--315,1974
9. Krisnamoorthy G, Hinkle PC: Non-ohmic proton conductance of
mitochondria and liposomes. Biochemistry 23: 1640--1645,1984
10. O'Shea PS, Petrone G, Casey RP,AzziA: The current-voltage relationship
ofliposomes and mitochondria. Biochem J 219: 719--726, 1984
II. Brown GC, Brand MD: Changes in permeability to protons and other
cations at high proton motive force in rat liver mitochondria. Biochem
J 234: 75-81,1986
12. Zoratti M, Favaron M, Pietrobon D, Azzone GF: Intrinsic uncoupling
of mitochondrial proton pumps. 1. Non-ohmic conductance cannot
account for nonlinear dependence of static head respiration on L'l.IlH+
Biochemistry 25: 760--767, 1986
50
13. Brown GC: The relative proton stoichiometries of the mitochondrial
proton pumps are independent ofthe proton motive force. J BioI Chern
25: 14704-14709,1989
14. Ouhabi R, Rigoulet M, Lavie JL, Guerin B: Respiration in non-
phosphorylating yeast mitochondria. Roles of non-ohmic proton
conductance and intrinsic uncoupling. Biochim Biophys Acta 1060:
293-298, 1991
15. Mitchell P, Moyle J: Acid-base titration across the membrane system
ofthe rat liver mitochondria. Catalysis by uncouplers. Biochem J 104:
588-600,1967
16. Mitchell P, Moyle J: Respiration-driven proton translocation in rat
liver mitochondria. BiochemJ 105: 1147-1167, 1967
17. Bielawski J, Thompson TE, Lehninger AL: The effect of2-4 dinitro-
phenol on the electrical resistance of phospholipid bilayer membranes.
Biochem Biophys Res Commun 24: 948--953, 1966
18. Hopfer U, Lehninger AL, Thompson TE: Protonic conductance across
phospholipid bilayer membranes induced by uncoupling agents for
oxidative phosphorylation. Proc Nat! Acad Sci USA 59: 484-490, 1968
19. Liberman YA, Topaly VP: Selective transport of ions through bimolecular
phospholipid membranes. Biochim Biophys Acta 163: 125-136, 1968
20. Leblanc HO: The effect of uncouplers of oxidative phosphorylation on
lipid bilayer membranes carbonylcyanide-m-chlorophenylhydrazone. J
Membr BioI 4: 227-251, 1971
21. McLaughlin S: The mechanism of action ofDNP on phospholipid bilayer
membranes. J Membr Biol9: 361-372, 1972
22. Hanstein WG: Uncoupling of oxidative phosphorylation. Biochim
Biophys Acta 456: 129--148, 1976
23. Rottenberg H: Uncoupling of oxidative phosphorylation in rat liver
mitochondria by general anesthetics. Proc Nat! Acad Sci USA 80: 3313-
3317,1983
24. Rottenberg H, Hashimoto K: Fatty acid uncoupling of oxidative
phosphorylation in rat liver mitochondria. Biochemistry 25: 1747-1755,
1986
26. Pick U, Weiss M, Rottenberg H: Anomalous uncoupling of photo-
phosphorylation by palmitic acid and by gramicidin D. Biochemistry
26:8295-8302,1987
27. Luvisetto S, Pietro bon D, Azzone GF: Uncoupling of oxidative
phosphorylation. I. Protonophoric effects account only partially for
uncoupling. Biochemistry 26: 7332-7338, 1987
28. Luvisetto S, Azzone GF: Nature of proton cycling during gramicidin
uncoupling of oxidative phosphorylation. Biochemistry 28: 1100--1108,
1989
29. Rottenberg H, Koeppe RE: Mechanism of uncoupling of oxidative
phosphorylation by gramicidin. Biochemistry 28: 4355-4360, 1989
30. Rigoulet M, Ouhabi R, Leverve X, Putod-Paramelle F, Guerin B:
Almitrine, a new kind of energy-transduction inhibitor acting on
mitochondrialATP synthase. Biochim BiophysActa 975: 325-329, 1989
31. Krenn BE, Van Walraven HS, Scholts MJC, KraayenhofR: Modulation
of the proton-translocation stoichiometry ofH+ I ATP synthases in two
phototrophic prokaryotes by external pH. Biochem J 294: 705-709,
1993
32. Murphy MP, Brand MD: Variable stoichiometry of proton pumping by
the mitochondrial respiratory chain. Nature 329: 170--172, 1987
33. Van Walraven HS, Scholts MJC, Koppenaal F, Bakels RHA, Krab K:
Dependence of the proton translocation stoichiometry of cyanobacterial
and chloroplast H+-ATP synthase on the membrane composition.
Biochim Biophys Acta 10 15: 425-434, 1990
34. Luvisetto S, Conti E, Buso M, Azzone GF: Flux ratios and pump
stoichiometries at sites II and III in liver mitochondria. Effect of slips
and leaks. J Bioi Chern 266: 1034-1042, 1991
35. Bechmann G, Weiss H: Regulation ofthe proton/electron stoichiometry
of mitochondrial ubiquinol: cytochrome c reductase by the membrane
potential. Eur J Biochem 195: 431-438, 1991
36. Steverding D, Kadenbach B: Influence ofN-ethoxycarbonyl-2ethoxy-
1,2-dihydroquinoline modification on proton translocation and
membrane potential of reconstituted cytochrome-c-oxidase support
proton slippage. J Bioi Chern 266: 8087--8101,1991
37. Capitanio N, Capitanio G, De Nitto E, Villani G, Papa S: Wle- stoichio-
metry of mitochondrial cytochrome complexes reconstituted in
liposomes. Rate-dependent changes of the stoichiometry in the
cytochromec oxidase vesicles. FEBS Lett 288: 179--182, 1991
38. Papa S, Capitanio N, Capitanio G, De Nitto E, Minuto M: The cytochrome
chain of mitochondria exhibits variable Wle- stoichiometry. FEBS Lett
288: 183-186, 1991
39. Groth G, Junge W: Proton slip of the chloroplast ATPase: Its nucleotide
dependence, energetic threshold, and relation to an alternating site
mechanism of catalysis. Biochemistry 32: 8103-8111, 1993
40. Fitton V, Rigoulet M, Ouhabi R, Guerin B: Mechanistic stoichiometry
of yeast mitochondrial oxidative phosphorylation. Biochemistry 33:
9692-9698, 1994
41. Capitanio N, Capitanio G, Demarinis DA, De Nitto E, Massari S, Papa
S: Factors affecting the H+/e-stoichiometry in mitochondrial cytochrome
c oxidase: Influence of the rate of electron flow and transmembrane
LlpH. Biochemistry 35: 10800--10806, 1996
42. Canton M, Gennari F, Luvisetto S,Azzone GF: The nature of uncoupling
by n-hexane, I-hexanethiol and I-hexanol in rat liver mitochondria.
Biochim Biophys Acta 1274: 39-47, 1996
43. Pietrobon D,Azzone G.F, Walz D: Effect offuniculosin and antimycin
A on the redox-driven H+-pumps in mitochondria: on the nature of
leaks. Eur J Biochem 117: 389--394, 1981
44. Pietrobon D, Zoratti M, Azzone GF: Molecular slipping in redox and
ATPase W-pumps. Biochim BiophysActa 723: 317-321,1983
45. Pietrobon D, Caplan R: Flow-force relationships for a six-state proton
pump model: Intrinsic uncoupling, kinetic equivalence of input and
output forces, and domain of approximate linearity. Biochemistry 24:
5764-5776,1985
46. Garlid KD, Beavis AD, Ratkje SK: On the nature of ion leaks in
energy-transducing membranes. Biochim Biophys Acta 976: 109--120, 1989
47. Mitchell P: The protonmotive Q cycle: A general formulation. FEBS
Lett 59: 137-139, 1975
48. Wikstrom MKF, Krab K: Proton-pumping cytochrome c oxidase.
Biochim Biophys Acta 549: 177-222, 1979
49. Guerin B, Labbe P, Somlo M: Preparation of yeast mitochondria
(Saccharomyces cerevisiae) with good P/O and respiratory control
ratios. Meth Enzymol55: 149--159, 1979
50. Gornall AG, Bardawill CJ, David MM: Determination of serum protein
by means ofthe biuret reaction. JBiol Chern 177: 751-766,1949
51. Cooper C, Lehninger AL: Oxidative phosphorylation by an enzyme
complex from extracts of mitochondria. J Bioi Chern 219: 489--506, 1956
52. Rigoulet M, Guerin B: Phosphate transport andATP synthesis in yeast
mitochondria. Effect ofa new inhibitor: Thetribenzylphosphate. FEBS
Lett 102: 18--22,1979
53. Argaud D, Roth H, Wiernsperger N, Leverve XM: Metformine decreases
gluconeogenesis by enhancing pyruvate kinase flux in isolated rat
hepatocytes. Eur J Biochem 213: 1341-1348, 1993
54. Bergmeyer HU: Methods in Enzymatic Analysis. Academic Press, New
York., 1974
55. Rottenberg H: The measurement of membrane potential and LlpH in
cells, organelles and vesicles. Meth Enzymol55: 547-569, 1979
56. Sumner J: Method for colorimetric determination of Pi. Science 100:
413-418,1944
57. Ohnishi T, Kawaguchi K, Hagihara B: Preparation and some properties
of yeast mitochondria. J Bioi Chern 241: 1797-1806, 1966
58. Kovac L, Bednarova H, Greksak M: Oxidative phosphorylation in yeast.
I Isolation and properties of phosphorylating mitochondria from
stationary phase cells. Biochim Biophys Acta 153: 32-42,1968

59. Von Jagow G, Klingenberg M: Pathways of hydrogen in mitochondria
of Saccharomyces carlsbergensis. Eur J Biochem 12: 583-592,1970
60. Sorgato MC, Branca D, Ferguson SJ: Factors affecting the proton
conductance of submitochondrial particles. Biochem J 188: 945-948,
1980
61. Zoratti M, Pietrobon D, Azzone GF: On the relationship between rate
of ATP synthsis and H+ electrochemical gradient in rat-liver mito-
chondria. Eur J Biochem 126: 443-451, 1982
62. Kacser H, Bums JA: The control offlux. Symp Soc Exp Bioi 27: 65-104,
1973
63 Heinrich R, RapoportTA: A linear steady-state treatment of enzymatic
chains. General properties, control and effector strength. Eur J Biochem
42:89-95,1974
64. Fitton V, Ouhabi R, Guerin B, Rigoulet M: Permeabilized spheroplasts
as a tool for studying the mechanistic stoichiometry of yeast oxidative
phosphorylation. In: E Gnaiger, FN Gellerich, M Wyss (eds). What is
Controlling Life? Innsbruck University Press, 1994, pp 145-149
65. Ouhabi R, Rigoulet M, Guerin B: Flux-yield dependence of oxidative
phosphorylation at constant L'l.IlW. FEBS Lett 254: 199-202, 1989
66. Leverve XM, Fontaine E, Putod-Paramelle F, Rigoulet M: Decrease in
cytosolic ATP/ADP ratio and activation of pyruvate kinase after in
vitro addition of almitrine in hepatocytes isolated from fasted rats. Eur
J Biochem 224: 967-974, 1994
67. Leverve X, Fontaine E, Espie P, Keriel C, Guerin B, Rigoulet M: Influence
of the mechanism of change in ATP synthaselATPase stoichiometry
on the control of oxidative phosphorylation in situ. In: E Gnaiger, FN
Gellerich, M Wyss (eds). What is Controlling Life? Innsbruck
University Press, 1994,pp 137-140
68. Rigoulet M, Fraisse L, Ouhabi R. Guerin B, Fontaine E, Leverve X:
Flux-dependent increase in the stoichiometry of charge translocation
by mitochondrialATPaselATPsynthase induced by almitrine. Biochim
BiophysActa 1018: 91-97,1990
69. Johnson RM, Ito T: Effects of nutritional deficiency of unsaturated
fats on the distribution of fatty acids in rat liver mitochondrial
phospholipids. J Lip Res 6: 75-79,1965
70. Biran LA, Bartley W, Carter WC, Renshaw A: Studies on essential
fatty acid deficiency. Biochem J 94: 247-251, 1965
71. Divakaran P, Venkataraman A: Effect of dietary fats on oxidative
phosphorylation and fatty acid profile of rat liver mitochondria. J Nutr
107: 1621-1631, 1977
72. StancliffR, Williams MA, Utsumi K, Packer L: Essential fatty acid
deficiency and mitochondrial function. Arch Biochem Biophys 131:
629-642,1969
73. Fontaine EF, Moussa M, Devin A, Garcia J, Ghisolfi J, Rigoulet M,
Leverve XM: Effect of polyunsaturated fatty acids deficiency on
oxidative phosphorylation in rat liver mitochondria. Biochim Biophys
Acta 1276: 181-187, 1996
74. Wesson LG, BUIT GO: The metabolic rate and respiratory quotients of
rats on fat-deficient diet. J Bioi Chern 91: 525-539, 1931
75. BUIT G, Berber AJ: Metabolism studies with rats suffering from fat
deficiency. JNutr 14: 553-566, 1937
76. Panos TC, Finerty JC: Effects of fat-free diet on growing male rats with
special reference to the endocrine system. J Nutr 54: 315-329, 1954
77. Morris DM, Panos TC, Finerty JC, Wall RL, Klein GF: Relation
of thyroid activity to increased metabolism induced by fat
deficiency. J Nutr 62: 119-128, 1957
78. Goubert M, Yazbeck J, Chapey MF, Diolez P, Moreau F: Variation in
energization parameters and proton conductance induced by cold
adaptation and essential fatty acid deficiency in mitochondria of
brown adipose tissue in the rat. Biochim Biophys Acta 1015: 334-
340,1990
79. Ito T, Jonhson RM: Effects of a nutritional deficiency of unsaturated
fats on rat liver mitochondria. J Bioi Chern 239: 3201-3208, 1964
51
80. Williams MA, Stancliff RC, Packer L, Keith AD: Relation of
unsaturated fatty acid composition of rat liver mitochondria to
oscillation period, spin label motion, permeability and oxidative
phosphorylation. Biochim Biophys Acta 246: 444-456, 1972
81. Christensen K: Influence of different dietary concentrations oflinoleic
acid on the essential fatty acid (EFA) status and functional characteristics
of porcine hepatic and cardiac mitochondria. Comp Biochem Physiol
85B:419-425,1986
82. Deaver OE, Wander RC, McCusker RH, Berdanier CD: Diet effects on
membrane phospholipid fatty acids and mitochondrial function in BHE
rats. JNutr 116: 1148-1155,1986
83. Brand MD, Chien L, Diolez P: Experimental discrimination between
proton leak and redox slip during electron transport. Biochem J 297:
27-29,1994
84. Luvisetto S, Canton M, Schmehl I, Azzone GF: Leaks and slips during
mitochondrial electron transfer. In: Gnaiger E, Gellerich FN, Wyss M
(eds). What is Controlling Life? Innsbruck University Press, 1994, pp
122-124
85. Tsou CS, van Dam K: Dependence of the efficiency of the uncouplers
on the respiratory rate. Biochim Biophys Acta 172: 174-176, 1969
86. WesterhoffHV, van Dam K: Thermodynamics and control of biological
free-energy transduction. Elsevier, Amsterdam, 1987
87. Groen BH, Berden JA, van Dam K: Differentiation between leak and
slips in oxidative phosphorylation. Biochim Biophys Acta 1019: 121-
127,1990
88. Levin E, Jonhson RM, Albert S: Mitochondrial changes associated
with essential fatty acid deficiency in rats. J Bioi Chern 228: 15-21, 1957
89. Smith JA, De Lucas HF: Structural changes in isolated liver mito-
chondria of rats during essential fatty acid deficiency. J Cell Bioi 21:
15-26,1964
90. Devin A, Guerin B, Rigoulet M: Dependence offlux size and efficiency
of oxidative phosphorylation on external osmolarity in isolated rat
liver mitochondria: role of adenine nucleotide carrier. Biochim Biophys
Acta 1273: 13-20, 1996
91. Mitchell P: Possible molecular mechanisms ofthe protonmotive function
of cytochrome systems. J Theor Bioi 62: 327-367, 1976
92. Gennis RB, Barquera B, Hacker B, Van Doren SR,Arnaud S, CroftsAR,
Davidson E, Gray KA, Daldal F: The bel complexes of Rhodobacter
spharoides and Rhodobacter capsulatus. J Bioenerg Biomembr 25:
195-209,1993
93. Boyer PD: The binding change mechanism for ATP synthase. Some
probabilities and possibilities. Biochim Biophys Acta 1140: 215-250, 1993
94 Duncan TM, Bulygin VV, Zhou Y, Hutcheon ML, Cross RL: Rotation of
subunits during catalysis by Escherichia coli FI-ATPase. Proc Nat!
Acad Sci USA 92: 10964-10968, 1995
95. Zhou Y, Duncan TM, Bulygin VV, Hutcheon ML, Cross RL: ATP
hydrolysis by membrane-bound Escherichia coli FoFI-ATPase causes
rotation of the g subunit relative to the y subunits. Biochim Biophys
Acta 1275: 96-100, 1996
96. Morgan JL, Verkhovsky ML, Wikstrom, M: The histidine cycle:Anew
model for proton translocation in the respiratory heme-copper oxidases.
J Bioenerg Biomembr 26: 599-608, 1994
97. Papa S, Lorusso M, Capitanio N: Mechanistic and phenomenological
features of proton pumps in the respiratory chain of mitochondria. J
Bioenerg Biomembr 26: 609-618, 1994
98. Thomas JW, Puustinen A, Alben JO, Gennis RB, Wikstrom M:
Substitution of asparagine for aspartate-135 in subunit I of the
cytochrome bo ubiquinol oxidase of Escherichia coli eliminates
proton-pumping activity. Biochemistry 32: 10923-10928, 1993
99. Garcia-Horsman JA, Puustinen A, Gennis RB, Wikstrom, M: Proton
transfer in cytochrome bo3 ubiquinol oxidase of Escherichia coli:
Second-site mutations in subunit I that restore proton pumping in the
mutant Asp135---Asn. Biochemistry 34: 4428-4433, 1995
52
100. Iwata S, Ostermeir C, Ludwig B, Michel H: Structure at 2.8 A
resolution of cytochrome c oxidase from Paracoccus denitrificans.
Nature 376: 660-669, 1995
101. Tsukihara T, Aoyama H, Yamashita E, Tomisaki T, Yamaguchi H,
Shinzawa-Itoh K, Nakashima R, Yaono R, Yoshikawa, S: The whole
structure of the 13-subunit oxidized cytochrome c oxidase at 2.8 A.
Science 272: 1136-1144, 1996
102. Frank V, Kadenbach B: Regulation of the Wle- stoichiometry of
cytochrome oxidase from bovine heart by intramitochondrialATP/ADP
ratios. FEBS Lett 382: 121-124, 1996
Molecular and Cellular Biochemistry 184: 53--65, 1998.
1998 Kluwer Academic Publishers.

Oxidative phosphorylation in intact hepatocytes:

Quantitative characterization of the mechanisms of
change in efficiency and cellular consequences

Xavier Leverve, l Brigitte Sibille, 1 Anne Devin,2 Marie-Astrid Piquet, 1

Pascal Espie2 and Michel Rigoulet2
ILaboratoire de Bioenergetique Fondamentale et Appliquee, Universite Joseph Fourier, Grenoble; 2Institut de Biochimie et
de Genetique Cellula ires du CNRS, Universite Bordeaux II, Bordeaux, France

Abstract
Two mechanisms may affect the yield of the oxidative phosphorylation pathway in isolated mitochondria: (i) a decrease in the
intrinsic coupling of the proton pumps (H+l2e- or H+/ATP), and (ii) an increase in the inner membrane conductance (proton or
cation leak). Hence three kinds of modifications can occur and each of them have been characterized in isolated rat liver
mitochondria (see preceding chapter by Rigouletet al.). In intact isolated hepatocytes, these modifications are linked to specific
patterns of bioenergetic parameters, i.e. respiratory flux, mitochondrial redox potential, DY, and phosphate potential.
(1) The increase in H+/ATP stoichiometry ofthe mitochondrialATP synthase, as induced by almitrine [20], leads to a decrease
in mitochondrial and cytosolic ATP/ADP ratios without any change in the protonmotive force nor in the respiratory rate or
redox potential. (2) In comparison to carbohydrate, octanoate metabolism by ~-oxidation increases the proportion of electrons
supplied at the second coupling site of the respiratory chain. This mimics a redox slipping. Octanoate addition results in an
increased respiratory rate and mitochondrial NADHINAD ratio while protonmotive force and phosphate potential are almost
unaffected. The respiratory rate increase is associated with a decrease in the overall apparent thermodynamic driving force
(2~E'0 - n~p) which confirms the 'redox-slipping-like' effect. (3) An increase in proton conductance as induced by the
protonophoric uncoupler 2,4-dinitrophenol (DNP) leads to a decrease, as expected, in the mitochondrial NADHINAD andATPI
ADP ratios and in ~'P while respiratory rate is increased.
Thus, each kind of modification (proton leak, respiratory chain redox slipping or increase in WI ATP stoichiometry ofATPase )
is related to a specific set of bioenergetic parameters in intact cells. Moreover, these patterns are in good agreement with the
data found in isolated mitochondria.
From this work, we conclude that quantitative analysis offour bioenergetic parameters (respiration rate, mitochondrial NADHI
NAD ratio, protonmotive force and mitochondrial phosphate potential) gives adequate tools to investigate the mechanism by
which some alterations may affect the yield of the oxidative phosphorylation pathway in intact cells. (Mol Cell Biochem 184:
53-65, 1998)

Key words: hepatocytes, oxidative phosphorylation, uncoupling, leak, slipping, almitrine, octanoate, PUFA

Introduction of the protonmotive force (proton and or cation leak);

secondly, at the coupling site between electron and proton
As described in a preceding chapter (Rigoulet et al.), there fluxes through the respiratory chain and thirdly, at the level
are theoretically three possibilities which affect the yield of of coupling between proton andATP synthesis fluxes through
the oxidative phosphorylation pathway: firstly, at the level mitochondrial ATPase.

Address/or offprints: X.M. Leverve, Laboratoire de Bioenergetique Fondamentale etAppliquee, Universite Joseph Fourier Grenoble, BP 53X, 3804 I-Grenoble
cedex, France
54
Changes in the proton permeability of the inner mito-
chondrial membrane have been widely studied [1-6] and a
large number of compounds known as 'protonophores' are
responsible for such an effect [7~ 14]. From the studies
performed in isolated mitochondria, it is possible to expect
the following effects in intact cells: (1) a decrease in proton-
motive force, (2) an increase in the respiratory rate associated
with a decrease in the NADHINAD ratio and (3) a decrease
in the ATP synthesis flux with a decreased phosphate poten-
tial (Fig. 1). Hence, the absolute values of the three forces
involved in oxidative phosphorylation coupling decreases
when proton leak increases. In isolated mitochondria, the
consequence induced by uncouplers on both redox and
phosphate potentials are strictly dependant on the experi-
mental conditions. In intact cells by contrast, these three
forces are involved in a complex metabolic network and the
resulting effect of uncoupling may depend on a large variety
of cellular parameters.
The change in efficiency at the level of the respiratory
chain may be due either to a redox slipping [3, 15, 16] or to
a change in the proportion of electrons delivered to the first
or to the second coupling site. In this situation, an increase
in respiratory rate with a decrease in the NADHINAD ratio
can be theoretically expected, without any change in both
protonmotive force [17, 18] and phosphate potential. Hence
for such a change of efficiency at the level ofthe respiratory
chain, a pure effect on the redox side may be predicted.
The third manner in which the coupling between oxidation
and phosphorylation would be affected is a change in
efficiency at the level ofATPase/ATP synthase. Whatever the
mechanism involved, protonslip or change in mechanistic
stoichiometry [19~21] the increase in H+/ATP apparent
stoichiometry must lead to a pure change in the phosphate
potential without any effect on both proton motive force and
redox potential [22, 23].
From these considerations it can be concluded that (1) a
leak effect would affect all three forces, (2) an intrinsic
uncoupling of the respiratory chain would affect only the
redox side of the oxidative phosphorylation pathway, whereas
(3) an intrinsic uncoupling of mitochondrial ATPase would
affect only the phosphate potential (see Fig. 1). Even if the
molecular mechanism involved is still a matter of debate,
these different situations are well described in isolated
mitochondria, contrasting with the low amount of data
available for intact liver cells (but see [6,24-26]). Despite
the complexity of the cellular machinery, we have undertaken
experimental work to investigate this problem. Our objective
can be separated into three different questions: (1) Is it
possible or not to find such pure effects in intact cells? (2) Is
it possible to discriminate between the three kinds of mech-
anisms? (3) What are the consequences of these different
effects on cellular metabolism?
Materials and methods
Liver cells were isolated from male Wistar rats (250-280 g)
fasted for 24 h. The deficiency in polyunsaturated fatty acids
was obtained in male weanling Wi star rats (60 g), fed for at
least 4 weeks a semi-synthetic diet containing: casein 21, D.L.
methinonine 0.12, com starch 44.26, sucrose 23.4, cellulose
1.87, mineral mixture 3.3 [39], vitamin mixture 0.94 [39] (%
weight). This diet was supplemented with either stearic and
palmitic acids (2.65 plus 2.65% weight; PUFA-deficient diet)
or soya oil (5.3% weight; control diet). Animals had access
to food and tap water ad libitum.
Hepatocytes were isolated by the method of Berry and
Friend [24] as modified by Groen et al. [27].
Incubations in closed vials
Hepatocytes (approximately 10 mg dry cellsml l) were incu-
bated as described in [22] with different amounts of substrates
as indicated in the legends. At t = 0 and 30 min of incubation,
400 III samples of the cell suspension were taken, quenched in
HCI04(4% mass/vol. final concentration) and neutralized with
KOH (2N) Mops (0.3 moH- l). Glucose, lactate, pyruvate,
3-hydroxybutyrate and acetoacetate were measured enzym-
atically as described in Bergmeyer [28]. ATP, ADP and AMP
were determined by HPLC as described previously [29].
Measurement of respiratory rates
After 20 min of incubation, the cell suspension was trans-
ferred into a thermostatically controlled oxygraph vessel
equipped with a Clark-type electrode for determination of
respiratory rate. Myxothiazol was used at 2 Ilg/ml in order
to abolish mitochondrial respiration activity. Cellular respira-
tion was the oxygen consumption sensitive to myxothiazol.
Measurement of cell and mitochondria volume in situ
Cell and mitochondrial matrix volumes were determined by
subtracting either l4C-carboxymethylinulin or l4C-manitol
spaces respectively from water space determined by 3Hp as
described [30].
Measurement of mitochondrial and cytosolic membrane
potentials in situ
Hepatocytes were incubated in the presence of3H -TPMP+ for
measurement of TPMP+ accumulation or 36Cl- for measure-
 55

A Schematic representation of the mitochondrial oxidative

phosphorylation pathway

respiratory chain ATPsynthase

tlE~ CD ADP+Pi ATP+HP

matrix

membrane

cytosol
I
I

H+
or
cations

B Working hypothesis: predicted consequences of some changes in

the coupling on cellular bioenergetic parameters

J0 2 NADH
---
NAD
-
LlJ.l w LlG p
I

2LlE h- nLlp

" " " "

pro tonophoric
effect ~ ~
redox
slipping ~ , o r. . . ~ ~

"
A Imitrine
~ ~ ~ ~

Fig. 1. Factors affecting the yie ld of oxidative phosphorylation.

(A) Schematic representation of the mitochondrial oxidative phosphorylation pathway. Respiratory chain, inner mitochondrial membrane and ATP
synthase are represented on this scheme with the related forces and fluxes (oxygen consumption and ATP synthesis). I - ~E' h represents the span in redox
potential between the donor electron couple (NADHINAD) and the final electron acceptor couple (oxygen, water) across the respiratory chain. II - ~iIH +
represent the protonmotive force (~p) which is the differen ce in proton electrochemical potential across the membrane divided by the Faraday constant (F).
1lI - ~G p represent the Gibbs- free energy of the ATP hydrolysis reaction (currently called phosphate potential for simplicity). Coupling efficiency may be
affected by two different mechanisms (see text) affecting either the coupling efficiency of proton pumps or the proton or cation inner membrane conductance.
These effects can be located at three different levels: I - at the respiratory chain by a mechanism of redox slipping; 2 - at the membrane by an increase in
proton or cation leak; 3 - at the proton ATPase/ATP synthase by increasing the H+/ ATP stoichiometry.
(8) Working hypothesis: predicted consequences of some changes in the coupling on cellular bioenergetic parameters. From the current knowledge
concerning the mitochondrial effects of these three kinds of modification in oxidative phosphorylation. Some putative cellular consequences concerning the
main bioenergetic parameters accessible in intact cells are proposed in this figure , as aworking hypothesi s.

ment of Cl- distribution according to [31]. The TPMP+ Perifusion of hepatocytes

accumulation ratio and the Cl- distribution were calculated
as described in [26]. For mitochondrial membrane potential
determination we have determined the actual apparent
activity coefficients for TPMP+ in the extracellular medium,
in the cytoplasm and in the mitochondrial matrix [30].
Liver cells (100--120 mg dry cells) were perifused by the
method of van der Meer and Tager 1976 [32] as modified by
Groen et al. [27, 33). Exogenous substrates were infused at
constant rate in order to obtain steady state concentrations as
56

indicated in the legends of the figure. Samples of perifusate
were taken throughout the experiments for subsequent deter-
minations of glucose, lactate, pyruvate, 3-hydroxybutyrate and
acetoacetate. Since ketone bodies and pyruvate are not stable
when frozen, perifusate samples were kept at +4C and
determinations were performed within less than 16 h. Proteins
in the perifusate were denaturated by heating the samples (80C
for 10 min) before centrifugation [34]. During the course of
the perifusions, samples of cell suspension were taken from
the chambers. However, in order to avoid any change in the
steady state due to opening of the chamber and interrupting the
flow when sampling the cell suspension, as initially described
[35], we inserted a needle directly into the chamber for
removing samples at the same rate as the perifusate entered.
This method allows us to take up to 12 samples from each
chamber at a minimum of 2 min intervals when necessary.
Mitochondrial and cytosolic spaces were separated by using
the digitonin fractionation method as described by Zuurendonk
and Tager [36]. The neutralized extracts were kept at -20C
for subsequent determination ofnucleotides by HPLC [29].
Collagenase A and enzymes were purchased from Boeh-
ringer, octanoate from Janssen, dihydroxyacetone from Merck,
DNP from Sigma and aminooxyacetate from Aldrich.
Results and discussion
We have investigated the cellular effect of agents known to
affect isolated mitochondria at each ofthe three possibilities:
i.e. (1) change in the efficiency of mitochondrial ATPase, (2)
intrinsic uncoupling ofthe respiratory chain and (3) increase
in the proton leak.
(1) Effect of a change in H+/ATP stoichiometry of
mitochondrial ATPase
Almitrine is a drug which is used for the treatment of patients
suffering from hypoxia due to chronic lung diseases [37, 38].
The increase in arterial blood oxygen pressure in patients
treated with this drug is mainly explained by a specific effect
on chemoreceptors [39-42] and/or by changes in the venti-
lation/perfusion ratio [43,44]. Moreover some effects on
coupling between oxidation and phosphorylation in brain
mitochondria isolated from senescent rats have been reported:
an increase at low almitrine concentrations and a decrease at
high almitrine concentrations [45].
In previous works [19-21, 46] we have described the
effect of almitrine in mitochondria isolated from yeast,
bovine heart and rat liver mitochondria. Briefly, almitrine
(1) inhibits ATPase activity, (2) decreases the ATP/O ratio
without any change in the magnitude of the protonmotive
force and (3) had no direct effect on respiratory chain
activity. Moreover, the number of protons transferred by
ATPase for each mole of hydrolyzed ATP is increased.
Taken together, these properties show that almitrine acts by
changing the mechanistic stoichiometry of mitochondrial
ATPase/ATPsynthase. Since in intact cells there is a netATP
synthesis flux, the consequence of an increase in H+/ATP
stoichiometry is in fact similar to a slipping ofATP synthase.
Figure 2 shows that almitrine addition led to a large

30 10 0.8
A C
8
-
(II

III iii 0.6

III
III 20 C.
( II
u
E C 6
III
...>- <I:
Q;
0
Gi
u
'C
I-
0.4
~ ...
<I: 4 ~
...
-
'E iii
10 ~
>-
~ .2
Gi 2
:::I 0.2
E ,Q
:1 U >-
>C
N
0
...0
'C
'")
0 0 >-
.s::. 0.0
I

endogenous octanoate endogenous octanoate ~

endogenous octanoate

Fig. 2. Effect ofalmitrine on oxygen uptake, cellular ATPIADP ratio and 3-hydroxybutyrate/acetoacetate ratio in isolated hepatocytes. Hepatocytes (approx
20 mg dry cells) isolated from 24 h starved Wistar rats were incubated in 2.5 ml Krebs-bicarbonate buffer (pH 7.4) with (open column) or without (black
column) 15 mmol/l almitrine. Experiments were performed as indicated with or without 4 mmolll octanoate. After 20 min of incubation, an aliquot of cell
suspension was injected into an oxygraph vessel for oxygen uptake measurement (Fig. 2A) and simultaneously samples were taken out and quenched for
ATP, ADP, (Fig. 2B - ATP/ADP ratio) 3-hydroxybutyrate and acetoacetate (Fig. 2C - 3-hydroxybutyrate/acetoacetate ratio) (see Materials and methods).
Results are means S.E.M.; n =21 for respiration and ATP/ADP ratio, and 8 for ketone body determination (4 different hepatocyte preparations in each
group). *p < 0.0 I.

decrease in the cellular ATP/ADP ratio without any change
in both the mitochondrial redox state (as expressed by the 3-
hydroxybutyrate/aceto acetate ratio, [47] and the respiratory
rate. Moreover as expected from the result obtained in
isolated mitochondria, almitrine addition does not affect the
mitochondrial Ll\f in intact cells (Fig. 3). Hence, the addition
of almitrine to intact cells affects only the phosphate potential,
as speculated above) and therefore this drug is a useful tool
for obtaining different cellular phosphate potentials at similar
respiratory rates and redox states. Figure 4A, B, C and D
shows the metabolic consequences of this effect in peri fused
hepatocytes. Cells were titrated with dihydroxyacetone
(DHA) as exogenous substrate for both gluconeogenesis and
glycolysis (lactate plus pyruvate production). As reported
previously, the control of DHA metabolism appears to be
almost entirely located at the first step of the pathway under
these experimental conditions [22] while the balance between
gluconeogenesis and glycolysis was dependent on the
cytosolic ATPIADP ratio. It is clear from Fig. 4C and D that
both mitochondrial and cytosolic ATPIADP ratios were
almost totally unaffected by the increasing concentrations of
peri fused DHA while they were significantly lowered by
almitrine addition. The cytosolic redox state (expressed as
lactate/pyruvate ratio [47], was not affected by almitrine (Fig.
4B). The main effect was a large enhancement in the glyco-
lytic flux (Fig. 4A) which was due to an activation of the flux
through pyruvate kinase (Fig. 4D). Hence the change in WI
ATP stoichiometry of mitochondrial ATPase affects theATPI
ADP ratio and leads to a redistribution of the balance between
glycolysis and gluconeogenesis. This effect on mitochondrial
ATP synthase appears to be the unique possibility in which
the phosphate potential would be changed without affecting
redox state nor mitochondrial respiration.
180
160
r--.
140
. n is the H+10 stoichiometry of the electron transport chain .
120
> In our experiments we have only measuredLl\f since it is the
E 100
~
<] 80
I
60
40
20
0 sented in Fig. 5 where the respiratory flux was manipulated
0 5 10 15 20
almitrine, jJmol/l
Fig. 3. Ll 'f' measurements by TPMP+ in intact isolated hepatocytes titrated
with almitrine. Isolated liver cells were titrated with increasing concentra-
tions of almitrine as indicated. Incubations were performed in Krebs-
bicarbonate medium for 25 min. Ll'f' was determined from triphenylmethy-
phosphonium (TPMP+) distribution (see Materials and methods).
57
(2) Effect of octanoate respiratory chain response to
thermodynamic forces
The effects of fatty acids on cellular respiration have been
known forlong time (see [48] for review). It has been reported
by Nobes et al. [6] that the increase in respiratory rate due to
fatty acid addition in isolated hepatocytes is accompanied by
a large increase in mitochondrial NADHINAD ratio which
is associated with a significant increase inLl\f. These authors
proposed that fatty acid addition to isolated hepatocytes raises
the respiratory rate by increasing the mitochondrial NADH
supply to the respiratory chain. Although, membrane proton
conductance was not changed, proton leak increased due to
the increase in protonmotive force, thus contributing to the
increase in respiratory rate together with an increase in the
Llp consuming processes. Table 1 shows similar results: large
increases in respiratory rate and in mitochondrial NADHI
NAD ratio with a slight but significant increase in Ll\f which
is not in agreement with an uncoupling effect, in accord with
the conclusion reached by Nobes et al. [6]. Cytosolic and
mitochondrial phosphate potentials were not significantly
modified by fatty acid addition. Since fatty acids are metab-
olized by ~-oxidation, they lead to a higher proportion of
FADH2 reducing equivalents than from substrates metab-
olized through glycolysis and the Krebs cycle. This shift in
electron supply toward a two coupling site system instead of
three coupling site system leads to a decrease in the overall
proton!2 e- stoichiometry of the respiratory chain, which may
be considered phenomenologically equivalent to intrinsic
respiratory chain uncoupling.
This hypothesis of intrinsic respiratory chain uncoupling
by octanoate could be tested by comparing the relationships
between the overall thermodyamic driving force through the
electron transport chain and the respiratory rate in different
conditions of substrate supply. This overall thermodynamic
driving force is expressed as 2LlE' 0 - nLlp, where LlE' 0 is the
difference in redox potentials across the respiratory chain and
main component of the protonmotive force. For substrates
giving their electrons to site I it is generally assumed that n
is equal to 10 [49]. But, it is worth noting that the absolute
value of this stoichiometry does not play any role in the shape
of the relationship between respiratory rate and the overall
thermodynamic driving force. This relationship was repre-
by amino acid additions in presence of glucose or glucose-
octanoate medium. It clear from this figure that two distinct
relationships were observed in the presence or absence of
octanoate: for a similar overall thermodynamic driving force,
the respiratory rate was much higher in the presence of
octanoate. This difference between the two curves is indepen-
dent ofthe value of n chosen. This may indicate either a large
58

8 8

7 7
A B
6 6

5 5
1
4

1 1
o+-~--~~--~~--~~--~~--,
18 2.0
16 1.8
D
CIS
14
.~

"C 1.6
C
12 o
s:. 1.4
10tr.~~~~~-r____~ ~ 1.2
8 E 1.0
6 o0:- 0.8
...
~ 0.6
4 Q.

c:[ 0.4
2
0.2
o +-~--r-~~~~~--~~--~~ O.O+---.----r--r--"""T'"--..-~--.---,--r--__.
o 2 4 6 8 10 o 2 4 6 8 10
dihydroxyacetone, mmolll dihydroxyacetone, mmolll
8
7

6
E
5

4
3

100 200 300 400 500

phosphoenolpyruvate, nmollg dry cells

Fig. 4. Metabolic effect of almitrine on isolated hepatocytes perifused with dihydroxyacetone. Hepatocytes (200 mg dry cells in 15 ml) were peri fused with
dihydroxyacetone in Krebs-Ringer-bicarbonate buffer (pH 7.4) in the absence (D) or presence (.) of 15 Ilmol/l almitrine. Lactate and pyruvate concentrations
were measured in the protein free perifusate. At the same time, samples of cell suspension were removed and separated by the digitonin fractionation
procedure into mitochondrial and cytosolic spaces for determination of adenine nucleotides by HPLC. Results are means S.E.M.; n = 5. (A) Lactate +
pyruvate production = glycolysis. (B) Lactate/pyruvate = cytosolic redox potential. (C) Cytosolic ATP/ADP ratio. (D) Mitochondrial ATP/ADP ratio. (E)
Effect on pyruvate kinase flux.

50
CD
~
iii
"- .-
c:
c:.! 40
o "-
.- 0
aD..
Em
::I E
C
c: ._
III 30
~
o E
~E
CD 0
m-II!
>
>C c: 20
0 It is generally accepted that 2,4-dinitrophenol (DNP) acts like
10+-~-'--~-'~--'-~~~~~
-0.9 -0.8 -0.7 -0.6 -0.5 -0.4
Fig. 5. Relationship between respiratory rate and overall thermodynamic
driving force in isolated hepatocytes. Cells were incubated as described in
the legend to Fig. 3 in either 20 mmolll glucose (e) or 20 mmolll glucose
and 4 mmol/l octanoate (0). After 25 min cellular and mitochondrial
volumes, mitochondrial A', 3-hydroxybutyrate and acetoacetate were
determined as described in the Materials and methods section. In both
conditions (glucose or glucose/octanoate) respiratory rate was manipulated
by addition of several compounds: 10 mmol/l alanine with or without 0.3
mmolll amino-oxyacetate, 10 mmolll proline, 10 mmolll, a-amino isobutyric
acid, 10 mmol/l glutamine with or without 2.5 mmollileucine. AE'h is the
difference in redox potential across the electron transport chain and Ap is
the protonmotive force multiplied by 10 for the WIO stoichiometry of the
electron transport chain. The value of 10 was chosen considering the three
coupling sites when NADH is the electron source (see text for details).
activation by octanoate of one or more enzymes of the
respiratory chain itself or a decrease in the apparent H+/2 e~
stoichiometry (n). The latter possibility could be the con-
sequence of either an intrinsic uncoupling of the respiratory
chain or of a change in the proportion of electrons supplied
site 1 and 2. Although the first explanation, of a direct effect
of octanoate on the chain activity is possible, when the
differences in the oxidative pathways of octanoate and
glucose are considered, the second hypothesis must be
preferred. Nevertheless, as predicted above for this case
(slipping of the respiratory chain), the increase of respiratory
rate was not accompanied by a decrease in A\{' nor in phos-
phate potential. The large increase in mitochondrial NADH/
NAD is due to octanoate metabolism.
From these data, it appears that the best tool to investigate
respiratory chain efficiency is the relationship between the
respiratory rate and the overall thermodynamic driving force.
Of course, the actual thermodynamic driving force must be
calculated from the actual stoichiometry (n). Hence in our
case, if we consider that respiratory rate is maintained by a
similar overall driving force with or without octanoate, this
means that the actual stoichiometry is respectively 8.4 or 10.
This value is in agreement with the change in the ratio of
FADH/NADH in the case of fatty acid or carbohydrate
59
oxidation. Nevertheless for our purpose, the comparison of
the respiratory rate and ofthe overall thermodynamic driving
force under different conditions permits us to investigate the
efficiency of the respiratory chain.
(3) Effect ofprotonophoric uncoupling in intact
hepatocytes
a proton shuttle, increasing the proton permeability of the
mitochondrial membrane leading to a decrease, or when fully
uncoupled, to abolish the protonmotive force [12, 24, 50-52].
Figure 6 shows that DNP uncoupling in cells perifused with
dihydroxyacetone and octanoate led to a sustained and large
increase in respiratory rate (Fig. 6A), and glucose production
(Fig. 6B). Moreover, both cytosolic and mitochondrialATP/
ADP ratios (Fig. 6C and D) as well as NADHINAD ratios
(expressed as lactate/pyruvate and 3-hydroxybutyrate/
acetoacetate ratios, Fig. 6E and F) decreased. These findings
are in good agreement with data from the literature showing
a decrease in protonmotive force andATP synthesis in similar
conditions [12, 24, 50-52]. Hence as mentioned above,
increase in proton leak results in a decrease of the three forces
involved in the oxidative phosphorylation pathway: redox
potential, protonmotive force and phosphate potential. The
cellular consequences of uncoupling are dependent on the
parameters controlling the metabolic pathway. For instance,
with DHA, whose metabolism is mainly controlled by the
phosphate potential [22, 34], the decrease in the cytosolic
ATP/ADP ratio led to a decrease in glucose production (ATP
consuming process) and to an increase in the rate of lactate
plus pyruvate production (ATP producing pathway). In
contrast glycerol metabolism is controlled by the cytosolic
redox potential [24, 34, 35] and DNP-protonophoric un-
coupling by decreasing the NADHINAD ratio (Fig. 7A and
B) led to a clear enhancement of glucose production (Fig. 7D)
In this experiment, the decrease in the ratio of mitochondrial
(NADHINAD)/cytosolic (NADHINAD) (Fig. 7C) is in line
with the expected decrease in mitochondrial protonmotive
force due to uncoupling.
In fact such clear cellular consequences of uncoupling by
protonophoric addition do not depend only on the nature of
the uncoupling agent but also on the metabolic pathway
generating mitochondrial NADH supply to the respiratory
chain. Indeed, as shown in Fig. 8, when DHA is the only
exogenous substrate, uncoupling by DNP led only to a small
and transient increase in respiratory rate (Fig. 8A) while
glucose production was abolished (Fig. 8B). The decrease in
cytosolic and mitochondrial phosphate potentials (Fig. 8C
and D) and in the mitochondrial NADH/NAD ratio (as
indicated by 3-hudroxybutyrate/acetoacetate ratio, Fig. 8E)
contrasted with the increase in lactate/pyruvate ratio in-
60

,.... A 8 B
'. ~
E 7
40 E
~ui
, CD cO~ 6
00-
C\J 0 30 o ~ 5
0 ~ (.) >.
::::J .... 4
""") "0
q> 20 0,"0
q> 3
""")
(5
E 10
0
E
2
~ ~ 1
0 .J

a.. 1. C 0
0 a.. 6
~ 0
a..
~ 1 ~
Ci1
''::
~
"0 .2
c: "0
~ O. 00
0 2
(.)
0
;t=
>-
() 1
~
0 0

- 0.6 E F
Q)
C6 0.8 Q)
Q)
(.) C6 6
-
>
~ ....::::J
Q) >-
~ 4
(.)
a:s
:S0.4
CD
J:
0 0.2
-C6 2
(.)
a:s

0.0 0
0 20 40 60 80 100 0 20 40 60 80 100
time, min time, min

Fig. 6. Effect of uncoupling (DNP) in isolated hepatocytes perifused with octanoate. Hepatocytes were perifused as described in Fig. 4 with dihydroxyacetone
(10 mmolI- I ). When a steady state had been reached, octanoate then 2,4 dinitrophenol were infused at a constant rate to reach final concentrations of 0.4
mmolI- 1 and 250 Ilmol'I- I , respectively as indicated. Oxygen consumption (Fig. 6A) of the hepatocyte suspension was continuously monitored with a Clark
electrode. Glucose (Fig. 6B), lactate, pyruvate, 3-hydroxybutyrate (OHBut) and acetoacetate concentrations were measured in the protein free perifusate. At
the same time, samples of cell suspension were removed and separated by the digitonin fractionation procedure into mitochondrial (Fig. 6C) and cytosolic
(Fig. 6D) spaces for determination of adenine nuc1eotides by HPLC. Figure 6E hydroxybutyrate/acetoacetate, Fig. 6F: lactate/pyruvate. This figure represents
a typical experiment; similar results were obtained in two others.

dicating an increased reducing power in the cytosolic space
despite the uncoupled state of the cells (Fig. 8F). This
particular consequence of uncoupling is the result of the
limitation of the mitochondrial supply of reducing equivalents
to the respiratory chain since under these metabolic con-
ditions, the entry of cytosolic reducing equivalents into the
matrix via the malate-aspartate shuttle is dependant on the
protonmotive force [53].
It must be stressed that a large decrease in the protonmotive
force is a drastic event for the cells metabolism. Indeed all
mitochondrial pathways are related directly or indirectly to
this protonmotive force, and except when a large supply of
 61

0.005
0 0.08
0

~
0

0.004
A
-

Z
:r::
0 0.06

Z
B

Z 0.003 (ij
.!:! 'J::
0UJ
0 0.002 "
J::
0
0.04
>.
u
~
u
~ 0.02
0.001 E

0 g
--
80

-
6 __
0 60
C ';'
.!!J.
iii
u
4
D

Z Q)~
:c Z UJ" 3
O:c 001
u_
0 ::l'
Z 40 -J::
01._
"")E; 2
~~
,,- 0
J::~ E
00
~>. ..:!.
Uu 20
~ .......
.s-
O 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
time (min) time (min)

Fig. 7. Effect of uncoupling (DNP) in isolated hepatocytes perifused with glycerol. Hepatocytes were perifused without exogenous substrate. When a
steady state had been reached, glycerol 20 mmol'l~l was added and cells were titrated with 2,4 dinitrophenol (DNP: 12.5,25 and 50 Ilmol'l ~1). Cytosolic
(Fig. 7A) and mitochondrial (Fig. 7B) NADHINAD were calculated from lactate, pyruvate, 3-hydroxybutyrate and acetoacetate concentrations in the
perifusate by using the equilibrium constant given in [38]. Figure 7C shows the ratio between mitochondrial and cytosolic (NADHINAD). The rate of
gluconeogenesis (Fig. 7D) was calculated from the glucose in the perifusate. This figure represent a typical experiment; similar results were obtained in
another.

NADH is provided directly into the matrix and independently
ofthe protonmotive force (as is the case with octanoate) the
supply of reducing equivalents to the respiratory chain is a
very controlling step. Nevertheless in all conditions, proton
leak is always responsible for a decrease in all three forces
involved in the oxidative phosphorylation pathway.
Hence by using these different tools, we have shown that
it is possible, in intact cells, as was the case with isolated
mitochondria, to deduce from the bioenergetic picture the
nature of the alteration of the oxidative phosphorylation
pathway. In contrast, when analysing the metabolic con-
sequences, the situation is often more confusing since they
depend on the connection between metabolic pathways and
bioenergetic parameters. Then, the question arises as to the
usefulness of such an approach to characterize more com-
plex modifications, particularly when different effects are
combined.
(4) Cellular effect of a change in mitochondrial oxidative
phosphorylation yield due to a deficiency in
polyunsaturated fatty acids (PUPA)
As we have already reported, and described in a preceding
chapter Rigoulet et al. [54], in isolated mitochondria, PUFA
deficiency led to an increase in non ohmic proton leak and
in redox slipping. Moreover, the relative importance of these
two effects on the overall efficiency of the oxidative phos-
phorylation pathway depends on both the rate of oxidative
phosphorylation and the maintained protonmotive force.
62

.....
'C
25 A ..... 8
'C
'E 'E 7
or;-cn 20 6
a)'wl
C\I (]) (J)-
15 o ~ 5
O~ g~ 4
~ "C
01 10 C)"C
01 3
(3 ~
(3
E 2
:::t 5 E
:::t 1
0 ()

0- C D
Q 0- 6
~ Q
0- 5
..... 1 ~
0-
..... 4
ra
.;::
u
"0
c:
3
0
0 (J)
.c: 0 2

u >.
0
+-'

~
() 1
0 0

(]) G H
-
(50.4
+-' (1)
ra 6
(1)
u

-~ 0.3 >
....
::::l
>.
..e-
(1)

--
u 4
ra
0.2

(1)
";:l
::::l ra
!lJ u
ra 2
I
0 0.1
0.0 0
0 20 40 60 80 0 20 40 60 80
time, min time, min
Fig. 8. Effect of uncoupling (DNP) in isolated heptatocytes perifused with dihydroxyacetone. Hepatocytes (100-120 mg dry cells in 15 ml) were peri fused
with dihydroxyacetone (10 mmolI-'). When a steady state had been reached, 2,4 dinitrophenol was infused at a constant rate to reach final concentrations
of250 IlmolI-'. Oxygen consumption of the hepatocyte suspension was continuously monitored with a Clark electrode (Fig. SA). Glucose (Fig. SB), lactate,
pyruvate, 3 hydroxybutyrate (OHBut) and acetoacetate concentrations were measured in the protein free perifusate. At the same time, samples of cell
suspension were removed and separated by the digitonin fractionation procedure in mitochondrial and cytosolic spaces for determination of adenine nucleotides
by HPLC. This figure represents a typical experiment; similar results were obtained in two others.

Hence, the respective role of each of these effects remains
unclear particularly in intact cells. Table 2 shows that cells
from PUFA deficient rats presented an increased respiratory
rate with or without octanoate, and also in the presence of
oligomycin. Conversely, the mitochondrial NADH/NAD
ratio (assessed by the 3-hydroxybutyrate/acetoacetate ratio)
and the mitochondrial ATP/ADP ratio were not affected by
the PUFAdeficiency. Hence, two of the three forces involved
in the oxidative phosphorylation pathway were not affected
by the PUFA deficiency (i.e. mitochondrial redox potential
and ATP/ADP ratio). According to our conclusion above in
which a protonophoric uncoupling effect is accompanied by
a decrease in the three forces, we can conclude that PUFA
deficiency does not behave as a slight 'protonophoric like'
effect. Indeed the mitochondria from PUFA deficient rats do
not behave like mitochondria from control rats in the presence
of a small amount of uncoupler, but behave as mitochondria
in which the non ohmic proton conductance is higher, as was
shown in the experiments conducted on isolated mitochondria
[37,54). In phosphorylating cells where the mitochondrial
protonmotive force is far from the non ohmic range of the
curve, the question of the cause of the increased respiratory
rate remains. As we have shown above, the relationship
between respiratory rate and the overall thermodynamic
driving force will probably help to answer the question.
Experiments are in progress to address this problem.
It must be stressed that this situation in intact cells is very
complex since PUFA deficiency induces a dramatic change
in lipid membrane composition [54-63] and also some
possible changes in the activity of several membranal
enzymes (glycerol-3-phosphate for example [64]). Hence it
is not possible to assume that these mitochondria are modified
solely by one unique and well characterized alteration.
Conclusion
It is now generally accepted that the oxidative phosphory-
lation pathway in eucaryotes is a chimio-osmotic coupling
involving three different forces related by two fluxes (Fig.
1). The three forces are the mitochondrial redox span between
electron donors and acceptors, the protonmotive force and the
mitochondrial phosphate potential. The two fluxes are
electron transfer through the respiratory chain supporting
proton extrusion from the matrix and ATP synthesis at the
level of mitochondrial ATP synthase coupled to the proton
entry into the matrix. Two different processes can in principle
affect the yield of this coupled pathway: (1) a change in the
intrinsic coupling of the respiratory chain (H+l2e-) or ATPasel
ATPsynthase (WIATP) and (2) a membrane conductance for
protons and or cations which results in a consumption of
protonmotive force. Two main mechanisms may account for
63
the change in intrinsic coupling of the mitochondrial proton
pumps: (1) a slipping process which is a decrease in the
efficiency of coupling and which may affect both the res-
piratory chain and ATP synthase and an actual mechanistic
change in the stoichiometry as has been proposed for ATP
synthase. From a phenomenological point of view there is no
difference in cells during active oxidative phosphorylation,
between proton slipping in ATP synthase and increase in
protonl ATP mechanistic stoichiometry: both cases result to
lessATP synthesized for a given proton flux. These different
effects have been characterized on isolated rat liver mito-
chondria and it has been shown that each of them is associated
with specific modifications of one or more of the main
parameters involved, i.e. the three forces and the two fluxes.
In this paper, we have used well characterized modifications
of the oxidative phosphorylation pathway, as a tool to
describe their effects on the bioenergetics in intact cells and
some of the consequences on metabolic pathways. As far as
modifications of the oxidative phosphorylation pathway in
intact cells are concerned we found that (1) each kind of
modification (proton leak, respiratory chain redox slipping
or increase in H+I ATP stoichiometry ofATPase ) corresponds
to a specific pattern of bioenergetic modifications; and (2)
these patterns are in good agreement with what could be
expected from the studies performed in isolated mito-
chondria. Hence it turns out from this work that we have
indeed adequate tools to investigate the mechanism by which
some alterations may affect the yield of the oxidative phos-
phorylation pathway in intact cells.
It appears also from this work that the cellular metabolic
consequences of a given alteration of the oxidative phos-
phorylation pathway are multiple and not specific. This is in
accord with the complex network of the metabolic machinery
which can be controlled at numerous points by each of the
three forces involved in the oxidative phosphorylation
pathway mitochondrial redox state, protonmotive force and
mitochondrial phosphate potential.
References
I. Nicholls DG: The influence of respiration and ATP hydrolysis on the
proton electrochemical gradient across the inner membrane of rat liver
mitochondria as determined by ion distribution. Eur J Biochem 50:
305-315, 1974
2. Nicholls DG: The effective proton conductance of the inner membrane
of mitochondria from brown adipose tissue. Dependency on proton
electrochemical potential gradient. Eur J Biochem 77: 349-356, 1977
3. Azzone GF, PozzanT, Massari S, Bragadin M: Proton electrochemical
gradient and rate of controlled respiration in mitochondria. Biochim
Biophys Acta 501: 296-306, 1978
4. Krishnamoorthy G, Hinkle PC: Non ohmic conductance of mito-
chondria and liposomes. Biochemistry 23: 1640-1645, 1984
5. Brown GC, Brand MD: Changes in penneability to protons and other
64
cations at high protonmotive force in rat liver mitochondria. Biochem
J 234: 75-81,1986
6. Nobes CD, Brown GC, Olive PN, Brand MD: Non-ohmic proton
conductance of the mitochondrial inner membrane in hepatocytes. J
BioI Chem 265: 12903-12909, 1990
7. Mitchell P: Coupling of phosphorylation to electron and hydrogen
transfer by chemi-osmotic type of mechanism. Nature 191: 144-148,
1961
8. Wilson DF, Ting HP, Koppelman MS: Mechanism of action of
uncouplers of oxidative phosphorylation. Biochemistry 10: 2897-2902,
1971
9. Mac Laughlin SGA, Dilger JP: Transport of protons across membranes
by weak acids. Physiol Rev 60: 825-863, 1980
10. Hanstein WG: Uncoupling of oxidative phosphorylation. Biochim
Biophys Acta 456: 129-148, 1976
II. Terada H: The interaction of highly active uncouplers with mitochodria.
Biochim Biophys Acta 639,225-242,1981
12. Terada H: Uncouplers of oxidative phosphorylation. Environ Health
Perspect 87: 213-218,1990
13. Miyoshi H, Nishioka T, Fujita T: Quantitative relationship between
protonophoric and uncoupling activities of analogs of SF6847
(2,6-di-t-butyl-4-(2 ,2 -dieyanovinyl)phenol). Biochim Biophys Acta
891: 194-204, 1987
14. Westerhoff HV, Van Dam K: Thermodynamics and Control of
Biological Free Energy Transduction. Elsevier, Amsterdam, New York,
Oxford, 1987
15. Pietrobon D,Azzone GF, Walz D: Effect offuniculosin and antimycin
A on the redox-driving H+-pumps in mitochondria: on the nature of
'leaks'. Eur J Biochem 117: 389-394,1981
16. Pietrobon, D., Zoratti, M.,Azzone, G.F.: Molecular slipping in redox
and ATPase W pumps. Biochim Biophys Acta 723: 317-321, 1983
17. Lusivetto S, Pietrobon D, Azzone GF: Uncoupling of oxidative
phosphorylation. I. Protonophoric effects account only partially for
uncoupling. Biochemistry 26: 7332-7338, 1987
18. Pietrobon D, Lusivetto S, Azzone GF: Uncoupling of oxidative
phsophorylation. 2. Alternative mechanisms: intrinsic uncoupling or
decoupling? Biochemistry 26: 7339-7347,1987
19. Rigoulet M, Ouhabi R, Leverve X, Putod-Paramelle F, Guerin B:
Almitrine, a new kind of energy transduction inhibitor acting on
mitochondrial ATP synthase. Biochim Biophys Acta 975: 325-329,
1989
20. Rigoulet M, Fraisse L, Ouhabi R, Guerin B, Fontaine E, Leverve X:
Flux-dependent increase in the stoichiometry of charge translocation
by mitochondrialATPaselATPsynthase induced by Almitrine. Biochim
Biophys Acta 1018: 91-97,1990
21. Jumelle-Lac1au M, Rigoulet M, Averet N, Leverve X, Dubourg L,
Carbonneau A, Clerc M, Guerin B: Relationships between age-
dependent changes in the effect of Almitrine on H+ ATPase/ATP
synthase and the pattern of membrane fatty acid composition. Biochim
Biophys Acta 1141: 90--94, 1993
22. Leverve X, Fontaine E, Putod-Paramelle F, Rigoulet M: Decrease in
cytosolic ATPIADP ratio and activation of pyruvate kinase after in
vitro addition of almitrine in hepatocytes isolated from fasted rats.
Eur J Biochem 224: 967-974,1994
23. Leverve X, Fontaine E, Espie P, Keriel C, Guerin B, Rigoulet M:
Influence of the mechanism of change in ATP synthasel ATPase
stoichiometry on the control of oxidative phosphorylation in situ. In:
E Gnaiger, F Gellerich, M Wyss (eds). What is Controlling Life?
Modern Trends in BioThermoKinetics. Innsbruck University Press,
Innsbruck, 1994, pp 137-140
24. Berry MN, Kun E, Werner HV: Regulatory role of reducting-equivalent
transfer from substrate to oxygen in the hepatic metabolism of glycerol
and sorbitol. Eur J Biochem 33: 407-417, 1973
25. Berry MN, Grivell AR, Wallace PG: Energy-dependent regulation of
the steady-state concentrations of the components of the lactate
dehydrogenase reaction in liver. FEBS Lett 119: 317-322, 1980
26. Nobes CD, Hay WW, Brand MD: The mechanism of stimulation of
respiration by fatty acids in isolated hepatocytes. J Bioi Chem 265:
12910--12915,1990
27. Groen AK, Sips HJ, Vervoorn RC, Tager JM: Intracellular compart-
mentation and control of alanine metabolism in rat liver parenchymal
cells. Eur J Biochem 122: 87-93, 1982
28. Bergmeyer HU: In: VC Weinheinn (ed.). Methods of Enzymatic
Analysis. Academic Press, New York, 1974, pp 1-2302
29. Argaud D, Roth H, Wiernsperger N, Leverve X: Metformine decreases
gluconeogenesis by enhancing pyruvate kinase flux in isolated rat
hepatocytes. Eur J Biochem 213: 1341-1348, 1993
30. Espie P, Guerin B, Rigoulet M: On isolated hepatocytes mitochondrial
swelling induced in hypoosmotic medium does not affect the respiration
rate. Biochim Biophys Acta, 1230: 139-146, 1995
31. Nobes CD, Brand MD: A quantitative assessment of the use of 36CI-
distribution to measure plasma membrane potential in isolated
hepatocytes. Biochim BiophysActa 987: 115-123, 1989
32. van der Meer R, Tager JM: Simple method for peri fusion of isolated
liver cells. FEBS Lett 67: 36-40, 1976
33. Groen AI(, Vervoorn RC, van der Meer R, Tager JM: Control of
gluconeogenesis in rat liver cells. J Bioi Chern 258: 14346--14353, 1983
34. Leverve X, Verhoeven AJ, Groen AK, Meijer AJ, Tager JM: The
malate-aspartate shuttle and pyruvate kinase as targets involved in the
stimulation of gluconeogenesis by phenylephrine. Eur J Biochem 155:
551-556,1986
35. Leverve X, Groen AK, Verhoeven AJ, Tager JM: Kinetic analysis of
short-term effects of alpha-agonists on gluconeogenesis in isolated
hepatocytes. FEBS Lett 181: 81-84, 1985
36. Zuurendonk PF, Tager JM: Rapid separation of particulate and soluble
fractions from isolated cell preparations (digitonin and cell cavitation
procedures). Meth Enzymol56: 207-223, 1979
37. Stanley NN, Pieczora JA, Pauly N: Effects ofalmitrine bismesylate
on chemosensitivity in patients with chronic airways obstruction. Eur
J Resp Dis 64(suppI126): 233-237, 1983
38. Voisin C, Howard P, Ansquer JC: Almitrine bismesylate: a long-term
placebo-controlled double-blind study in COAD--Vectarion Inter-
national Multicentre Study Group. Eur J Resp Dis 23(suppl 146): 169-
182 S, 1987
39. Laubie M, Schmitt H: Long-lasting hyperventilation induced by almi-
trine: evidence for a specific effect on carotid thoracic chemoreceptors.
Eur J Pharmacol61: 121-136, 1980
40. Biscard GE: The response of few-fiber carotid chemoreceptor
preparations to almitrine in the dog. Can J Physiol Pharmacol59: 369-
401,1981
41. Roumy M, Leitner LM: Stimulant effect of almitrine (S 2620) on the
rabbit carotid chemoreceptor afferent activity. Bull Eur Physiopath
Resp 17: 255-259, 1981
42. Lahiri S, Mokashi A, Huang W, Sherpa AK, Di Giulio C: Stimulus
interaction between CO, and almitrine in the cat carotid chemo-
receptors. J Appl Physiol 67: 232-238, 1989
43. Bee D, Gill GE, Emery CJ, Salmon GL, Evans TW, Barrer GR: Action
of almitrine on the pulmonary vasculature in ferrets and rats. Bull Eur
Physiopathol Respir 19: 539-545, 1983
44. Russell PC, Emery CJ, Nicholl J, Barer GR, Howard P: Interactions
between hypoxic and almitrine-induced vasoconstriction in the rat lung.
Clin Sci 82: 265-272, 1992
45. Castaing Y, Manier G, Guenard H: Improvement in ventilation-perfusion
relationships by almitrine in patients with chronic obstructive
pulmonary disease during mechanical ventilation. Am Rev Respir Dis
134(5): 910--916,1986

46. Rigoulet M, Leverve X, Fontaine E, Guerin B: Stoichiometry of
oxidative phosphorylation in mitochondria. J Mol Cell Cardiol (this
issue), xxx-xxx
47. Williamson DH, Lund P, Krebs HA: The redox state of free nicotin-
amide-adenine dinucleotide in the cytoplasm and mitochondria of rat
liver. Biochem J 103: 514-527, 1967
48. Wojtczak L, Schonfeld P: Effect of fatty acids on energy coupling
processes in mitochondria. Biochim BiophysActa 1183: 41-57,1993
49. Stoner CD: Determination of P/2e- stoichiometries at the individual
coupling sites in mitochondrial oxidative phosphorylation. Evidence
for maximum values of 1.0, 0.5, and 1.0 at sites 1,2, and 3. J Bioi
Chern 262: 10445-10453, 1987
50. Steen H, Maring JG, Meijer DK: Differential effects of metabolic
inhibitors on cellular and mitochondrial uptake of organic cations in
rat liver. Biochem Pharmacol45: 809-818, 1993
51. Rahn CA, Bomdick DW, Doolittle DJ: Assessment of mitochondrial
membrane potential as an indicator of cytotoxicity. Fund and applied
Toxicol16: 435-448, 1991
52. Fromenty B, Letteron P, Fish C, BersonA, Deschamps D, Pessayre D:
Evaluation of human blood lymphocytes as a model to study the effects
of drugs on human mitochondria. Effects of low concentrations of
amiodarone on fatty acid oxidation, ATP levels and cell survival.
Biochem Pharmacol46: 421--432, 1993
53. Sibille B, Keriel C, Catelloni F, Rigoulet M, Leverve X: Octanoate
affects 2,4-dinitrophenbol uncoupling in intact isolated rat hepatocytes.
Eur J Biochem 231: 498-502, 1995
54. Fontaine EF, Moussa M, Devin A, Garcia J, Ghisolfi J, Rigoulet M,
Leverve X: Effect of polyunsaturated fatty acids deficiency on
oxidative phosphorylation in rat liver mitochondria. Biochem Biophys
Acta 1276: 181-187, 1996
55. Yazbeck J, Goubert M, Senault C, Chapey MF, Portet R: The effects
of essential fatty acid deficiency on brown adipose tissue activity in
65
rats maintained at thermal neutrality. Comp Biochem Physiol 94A,
273-276, 1989
56. Biran LA, Bartley W, Carter WC, Renshaw A: Studies on essential
fatty acid deficiency. Biochem J 94: 247-251, 1965
57. Divakaran P, Venkataraman A: Effect of dietary fats on oxidative
phosphorylation and fatty acid profile of rat liver mitochondria. J Nutr
107: 1621-1631,1977
58. StancliffR, Williams MA, Utsumi K, Packer L: Essential fatty acid
deficiency and mitochondrial function. Arch Biochem Biophys 131:
629-642, 1969
59. Christensen K: Influence of different dietary concentrations oflinoleic
acid on the essential fatty acid (EFA) status and functional charac-
teristics of porcine hepatic and cardiac mitochondria. Comp Biochem
Physiol85B: 419--425, 1986
60. Deaver OE, Wander RC, McCusker RH, Berdanier CD: Diet effects
on membrane phospholipid fatty acids and mitochondrial function in
BHE rats. J Nutr 116: 1148-1155, 1986
61. Thomassen MS, RortveitT, Christiansen EN, Norum KR: Changes in
the content of n-6 fatty acids in liver phospholipids in rats as a
consequence of partially hydrogenated dietary oils. Brit J Nutr 51:
315-322, 1984
62. Senault C, Yazbeck J, Goubern M, Portet R, Vincent M, Gallay J:
Relation between membrane phospholipid composition, fluidity and
function in mitochondria of rat brown adipose tissue. Effect of thermal
adaptation and essential fatty acid deficiency. Biochim Biophys Acta
1023:283-289,1990
63. Johnson RM, Ito T: Effects of nutritional deficiency of unsaturated
fats on the distribution of fatty acids in rat liver mitochondrial
phospholipids. J Lip Res 6: 75-79, 1965
64. Piquet MA, Fontaine E, Sibille B, Keriel C, Leverve X: Effect of
polyunsaturated fatty acid deficiency in isolated rat hepatocytes effect
on glycerol metabolism. Biochem J (in press)
Molecular and Cellular Biochemistry 184: 67-79, 1998.
1998 Kluwer Academic Publishers.

Yeast mitochondrial metabolism: From in vitro to

in situ quantitative study

Nicole Averet, Valerie Fitton, Odile Bunoust, Michel Rigoulet and

Bernard Guerin
Institut de Biochimie et Gent?tique Cellula ires du CNRS, Universite Victor Segalen Bordeaux 2, I rue Camille Saint-
Saens, 33077 Bordeaux-Cedex, France

Abstract
In this work, we first compared yeast mitochondrial oxidative metabolism at different levels of organization: whole cells
(C), spheroplasts (S), permeabilized spheroplasts (PS) or isolated mitochondria (M). At present, S are more suitable for use
than C for biochemical techniques such as fast extraction of metabolites and permeabilization. We show here that respiratory
rates of S with various substrates are similar to C, which demonstrate that they are adapted to yeast bioenergetic studies. It
appeared from ethanol metabolism NAD+ or NADH respiratory rates on PS that ethanol metabolism was largely cytosolic;
moreover, the activity of NADH dehydrogenase was lesser in the case of PS than in S. By comparing PS and M, the biggest
difference concerned the respiratory rates of pyruvate and pyruvate-malate, which were much lower for M. Thus mitochondria
preparation caused an unidentified loss involved directly in pyruvate metabolism. When the respiratory rate was lowered as
a consequence of a high kinetic control of oxidative activity upstream from the respiratory chain, a similar correlation between
the increase in ATP/O and decrease in respiratory rate was observed. So, the intrinsic uncoupling of proton pumps is not a
particularity of M.
Secondly, we demonstrate the existence of a mechanism of retarded diffusion in yeast similar to that already observed in
permeabilized mammalian cells for ADP. Such a mechanism also occurs in yeast for several respiratory substrates: the KO.5
for each substrate toward the respiration rate in PS always exceeds that for M. It is proposed that such a discrepancy is due
to a restriction of metabolite movement across the outer mitochondrial membrane in permeabilized cells, i.e. regulation of the
substrate permeability through porin channels. In the porin-deficient yeast mutant, the K05 for NADH is not significantly different
in either M or PS and is comparable to that ofthe parent strain PS. This result confirms that this retarded diffusion is essentially
due to the opening-closing ofthe porin channel. (Mol Cell Biochem 184: 67-79, 1998)

Key words: Saccharomyces cerevisiae, spheroplast, permeabilization, mitochondria, oxidative phosphorylation, porin

Introduction that this biological system may be modulated through

In most non-photosynthetic and aerobic cells, mitochondrial
oxidative phosphorylation is the main process of ATP
synthesis. This pathway is a complex and highly controlled
network through whichATP synthesis must be continuously
adapted to changes in the cell energy needed to maintain
homeostasis. The long history of efforts to understand the
mechanisms involved in the control and regulation of
oxidative phosphorylation and in process of adjustment
between ATP formation and expenditure has demonstrated
different sites by various effectors (see [\,2] for recent
reviews). Over the last four decades, both kinetic and
thermodynamic models have been proposed to explain the
short-term control of mitochondrial respiration and ATP
synthesis in isolated mitochondria, cells and tissues. They
have involved: (i) a simple substrate limitation for ATP
synthesis by ADP and Pi [3, 4]; (ii) the near-equilibrium
hypothesis in which all oxidative phosphorylation pathways
operate close to thermodynamic equilibrium (from mito-
chondrial NADHINAD+ to cytosolicATPIADP.Pi ratio) apart

Address for offPrints: N . Averet, Institut de Biochimie et Genetique Cellulaires du CNRS, Universite Victor Segalen Bordeaux 2, I rue Camille Saint-Saens, 33077
Bordeaux-Cedex, France
68

from cytochrome-oxidase which is far from equilibrium NADH+W

and is considered to be kinetically regulated by some
effectors [5]; in such a model, oxidative phosphorylation
is controlled by redox span between NADH/NAD+ and
cytochrome c, cytosolic phosphate potential and potentially
all the substrates and effectors of cytochrome-oxidase
(redox state of cytochrome c, ADP, ATP, oxygen, pH, ... );
(iii) a control by calcium, in addition withADP, acting as a
powerful activator of some matrix dehydrogenases ([6, 7]
for reviews). Thus, in addition to numerous energetic or
metabolic intermediates, some enzymatic steps participate
in the control and the regulation of oxidative phosphoryl-
ation. These steps might not only be cytochrome-oxidase
and some dehydrogenases, but also mitochondrial carriers
and/or ATP synthase. In addition, the passive proton flux
(H+-leak) through the inner mitochondrial membrane
acting as a branched proton pathway may be involved in the
control of oxidative phosphorylation. The flux control
coefficient is positive on respiratory rate and negative on
ATP synthesis, and its activity leads to changes in the yield
of oxidative phosphorylation i.e.ATP/O [8, 9]. By modifying
the intrinsic coupling of the proton pump involved in
oxidative phosphorylation i.e. H+/2e- or H+IATP apparent
stoichiometries, other mechanisms can also contribute to
a change inATP/O ratio and to control in both oxygen con-
sumption and ATP synthesis fluxes.
In view of the complexity of this network and the number
of potential control and/or regulation sites and effectors, it
is highly conceivable that several of these factors operate
simultaneously in determining of a particular steady state of
oxidative phosphorylation. Moreover, these control and/or
regulation factors may contribute to a variable extent to the
control of the whole pathway depending on the cell and the
steady state considered. Thus, an important issue remaining Fig. 1. Scheme illustrating the membranal dehydrogenases and the
is the identification of the mechanisms actually acting in the respiratory chain in mitochondria of yeast Saccharomyces cerevisiae.
control of oxidative phosphorylation and the determination Dehydrogenases located on the outer surface of the inner membrane: (I)
External NADH dehydrogenase; (III) Glycerol-3-P dehydrogenase; (V)
of their involvement in a particular organism and under Lactate dehydrogenase. Dehydrogenases located on the inner surface of
various steady state conditions. the inner membrane: (II) Internal NADH dehydrogenase; (rV) Succinate
Mitochondria isolated from Saccharomyces cerevisiae in dehydrogenase.
the exponential growth phase present three main character-
istics (Fig. 1): (i) lack of phosphorylation site corresponding In a previous study using an external NADH -regenerating
to the coupling site 1 of animal mitochondria [10-12], (ii) system and the ability of yeast mitochondria to oxidize
ability to oxidize exogenous NADH by a NADH dehydrogen- exogenous NADH, we showed that the value of ATP/O
ase located towards the outer surface of the inner membrane significantly increased when respiratory flux slowed down
[l3]; (iii) ability to oxidize lactate by using the third site span in response to a limitation of substrate supply [19]. In another
directly [10]. study, we observed that the mechanistic stoichiometry of
By applying the control analysis [14, 15] to isolated yeast yeast mitochondrial oxidative phosphorylation varied when
mitochondria, we have demonstrated the key role of phosphate the electron supply to the respiratory chain was changed
carrier in the control of oxidative phosphorylation [16, 17] according to the different dehydrogenases involved, each
and its dependence on transmembranal~pH [17,18] when the working at a determined rate in a given steady state: the
respiratory substrates were either ethanol or NADH at ATP/O ratio decreased irrespective of protonmotive force
saturating concentrations. when the electron flux through each respiratory chain unit

increased. Moreover, we showed that the highest ATP/O
values were linked to a high kinetic control of the oxidative
activity upstream from the respiratory chain, at substrate
supply or dehydrogenase activity levels [20].
However, these studies were conducted on mitochondria
isolated from yeast and it seems difficult to determine the
role of dehydrogenases in the control of oxidative phos-
phorylation, because of certain interrelations between
mitochondria and the cytosolic (or other) compartments. It
is possible that mitochondria in vivo are subject to kinetic and
thermodynamic constraints other than isolated mitochondria.
So many authors have studied mitochondria in situ in their
cellular environment [21-23]. It has recently been shown,
in permeabilized isolated cardiomyocytes, skinned cardiac
fibers, slow twitch skeletal muscle fibers and permeabilized
hepatocytes, that the apparent Kos for ADP on the stimulation
of respiration is more than one order of magnitude higher
than that measured in the mitochondria isolated from the
same tissues [24--2S]. In permeabilized heart and liver cells,
these authors observed that a hypoosmotic treatment induced
an increase of the apparent ADP affinity correlated to the
extent of the rupture of the outer mitochondrial membrane.
They conclude that both in heart and liver cells in vivo, the
ADP diffusion is limited by a low permeability of the outer
mitochondrial membrane for ADP [27]. In another work,
Fontaine et al. [2S] established that a trypsin treatment in
digitonin-permeabilized hepatocytes decreased the apparent
Ko.s for ADP. From these results, they conclude that an
important site of respiration control in liver cells in vivo is
located at the porin channels of the outer mitochondrial
membrane.
The permeabilization of cells consists in circumventing
the permeability barrier of the plasma membrane to study
enzymatic activities in situ and in gaining free access to
intracellular compartments such as mitochondria without
damaging the cell. This has already been done with yeast cells
by using different treatments such as toluene [29], toluene-
ethanol mixtures [30] dimethylsulfoxide [31], basic proteins
[32], triton X-I 00 [33], freeze-thaw cycles [34] and digitonin
[35, 36]. Unlike all the above techniques which attempt to
study enzymic activities in situ, our purpose was to do
bioenergetic studies in situ. So the cells had to be completely
accessible to exogenous substrates and cofactors without
damaging the mitochondria. Yet an important prerequisite was
to obtain intact mitochondria with efficient respiratory and
phosphorylation abilities.
We chose to use spheroplasts instead of whole cells
because they are more suitable for quenching reactions and
for biochemical techniques such as extraction procedures
[37]. Nystatin, a polyene antibiotic, is known to cause
specific permeability changes in a variety of models and
biological membranes after interacting with cholesterol or
ergosterol (in the case offungi and yeasts) in the membrane.
69
Permeabilization of yeast spheroplasts was performed in
a previous work [3S]. It was shown, on one hand, that
respiration on glucose was restored by glycolytic cofactors,
so mitochondria were not damaged by permeabilization, and
on the other hand, the glycolytic activity remained inside the
cells and the respiratory function was preserved. However,
the coupling capacity was not tested.
In this work, we first characterize the ability of sphero-
plasts and, after permeabilization tests, we compare the
characteristics of oxidative phosphorylation with various
substrates in isolated mitochondria and in mitochondria in
situ. Secondly, by using the specificities of this biological
material, we propose to answer three questions: (i) is this
phenomenon of ADP restricted diffusion observable in
yeast? (ii) what is its degree of generality? (iii) is it possible
to show directly the involvement of porin in this process?
Materials and methods
Yeast aerobic cultures
Cells of the diploid wild strain Saccharomyces cerevisiae
(yeast foam) were grown in a New Brunswick incubator at
28C in a complete medium: 1% yeast extract, 0.1 %
potassium phosphate, 0.12% ammonium sulfate, supple-
mented with 2% lactate as carbon source. The haploid wild
strain Saccharomyces cerevisiae (DBY747, Mata, ura3-52,
leu2-3, leu2-112, his3111, trpl-289) and porin-deficient
mutant B5, derived from this parent strain [39] were grown at
2S0C in the same medium. The cells were harvested in the
logarithmic growth phase.
Preparation and permeabilization of spheroplasts
Cells (1 g dry weight) were washed twice with 20 ml distilled
water and incubated for 10 min with shaking at 32C in 0.5
M ~-mercaptoethanol, 0.1 M Tris-HCI, pH 9.3. Then they
were washed 3-fold with 0.5 M KCI, 10 mM Tris-HCI, pH
7. For 1 g dry weight of cells suspended in 10 ml of 1.3 5 M
sorbitol, 1 mM ethylene glycol bis (~-aminoethylether)-N,
N, N', N'-tetraacetic acid (EGTA), 10 mM citrate-phosphate,
pH 5.S, 0.17 g of cytohelicase (Biosepra) were added and
cells incubated for about 40 min at 32C with gentle shaking.
The spheroplasts were centrifuged at SOO g for 5 min, washed
3-fold with the same solution and suspended in the following
buffer (medium 1): 1 M sorbitol, 0.5 mM EGTA, 2 mM
MgS0 4 , l.7 mM NaCI, 10 mM potassium phosphate, 0.1%
bovine serum albumin (BSA), pH 6.S.
Permeabilization of spheroplasts was performed as
follows: cells were diluted in medium 1 at 1 mg protein per
ml and incubated for 10 min at 2SoC with 20 f!g.ml- ' nystatin
70
(Sigma). It is possible to use this preparation (in the presence
of nystatin) for experiments lasting about 1 h (see Results
section).
Mitochondria were isolated from spheroplasts as previously
described [40]. Mitoplasts were prepared by the swelling-
shrinking procedure as described by Daum et al. [41].
The protein concentration was estimated by the biuret
method using bovine serum albumin as a standard.
Measurement of mannitol-impermeable spaces of intact
or permeabilized spheroplasts
Spheroplasts were incubated in medium 1 with [14C]mannitol
at 0.1IlCi.ml-1and PH]H20 at 2.5 IlCi.ml-l supplemented with
0.5 mM mannitol and 20 Ilg.ml-1 nystatin for permeabilized
ones. After 10 min of incubation, aliquots of cell suspension
(l ml) were centrifuged for 2 min at 1500 g and 0.1 mlofthe
supernatant was counted. After aspirating all the supernatant,
the cell pellet was suspended in 0.3 ml of perchloric acid
(PCA) 10%, then centrifuged at 1500 g for 2 min and the
new supernatant was counted. The mannitol-impermeable
volume was defined as the water volume of the pellet minus
the mannitol volume.
Measurement of spheroplast volume
Spheroplasts were incubated in medium 1 with PH]HP at
2.5 IlCi.ml-l and [14C]carboxymethylinulin at 0.1 IlCi.ml-l
(added 1 min before sampling) supplemented with 0.24 mM
inulin and 20 Ilg.ml-1 nystatin for permeabilized ones. After
25 min of incubation, aliquots of cell suspension (1 ml) were
centrifuged for 2 min at 1500 g and 0.1 ml of the supernatant
was counted. After aspirating all the supernatant, the cell
pellet was suspended in 0.3 rnl ofperchloric acid 10%, then
centrifuged at 1500 g for 2 min and the new supernatant was
counted. The cell volume was defined as the water volume
of the pellet minus the inulin volume.
Oxygen consumption and determination of ATP/O
Respiration measurement
Oxygen consumption was measured at 28C in a 2 ml
thermostatically controlled chamber equipped with a Clark
oxygen electrode (Gilson) connected to a microcomputer
giving an on-line display of rate values. Spheroplasts and
mitochondria were incubated in medium 1.
ATP/O determination
The incubation medium (medium 2) was as follows: 1 M
sorbitol, 0.5 mM EGTA, 2.5 rnM ethylene diamine tetraacetic
acid (EDTA), 3 mM iodoacetate to inhibit glycolysis, 0.1 mM
PI, P5-diadenosine pentaphosphate (Ap5A, Sigma) to inhibit
the adenyl ate kinase activity, 10 mM KHlO 4' 1.7 rnM N aCI,
pH 6.8 [42]. Permeabilization of spheroplasts was obtained
by addition of nystatin, at a concentration of 20 Ilg.ml-1 . ATP
was measured by the bioluminescence technique using an
ATP monitoring kit (Bioorbit); luminescence was monitored
with a Labsystems Luminoskan.
Results and discussion
Spheroplast permeabilization
Figure 2 presents the effect of nystatin on spheroplast
permeabilization. For a given concentration, the maximal
effect of nystatin was reached after 10 min of incubation and
the material was stable for about 1 h, although nystatin
remained in the medium (not shown). Respiration with
succinate as substrate was maximal for a nystatin con-
centration of 15-20 Ilg.mg-1 protein (or ml-1), indicating
that this concentration was sufficient to permeabilize
spheroplasts to metabolites. This conclusion is supported
100 ~/1ij-O-O-.O ________
r
80
I
10
10 40 60 80 100
nystatin (pg I ml)
Fig. 2. Permeabilization of spheroplasts by nystatin: Concentration
dependence. Spheroplasts were aerated 30 min in medium I (I mg protein in
I ml), then nystatin was added at different concentrations and tests described
below were started 10 min later and monitored at 28C. (I) Respiration was
measured after addition of 10 mM succinate (0); (2) Glucose-6P
dehydrogenase was measured fluorimetrically (Kontron SFM 25) by
addition of 0.2 mM NADP+ and 5 mM glucose-6P in the presence of
antimycin (30 J.Ig.ml-'); 'excitation = 340 nm, 'emission = 450 nm (.). In both
types of experiments the results are expressed in % of maximal effect obtained
at high concentration of nystatin.

by the observation that under this condition, the glucose-6P
dehydrogenase activity from externally added glucose-6P
and NADP+ was also maximal.
Given these results, we chose to use a 20 f.lg.ml-1 con-
centration of nystatin for this study. Under these conditions,
the volume permeable to water but impermeable to mannitol
(mannitol is considered as a non-permeant molecule) was
0.8 vs. 3 f.ll.mg- I protein for the total cellular volume of intact
spheroplasts, indicating that the accessibility to some
cellular compartments was not modified by nystatin.
Obviously, mitochondria contributed to this mannitol-
impermeable space. To characterize further the permeabilized
spheroplasts, the permeation of this material to dextrans of
different molecular weight was tested. Figure 3 shows that
the population of spheroplasts was not homogeneous with
regard to their permeability properties, but that more than
70% of them were impermeable to molecules of a molecular
weight equal or higher to 10,000 daltons under conditions
where spheroplasts were completely permeabilized to
metabolites and coenzymes.
To investigate the effect of cellular organization on the
respiratory capacity ofthe yeast cell, we measured respiration
on whole cells (C), spheroplasts (S), permeabilized sphero-
plasts (PS) and mitochondria (M). The digestion of the cell
wall by cytohelicase did not affect cellular respiration, as
100
80
c
~ 60
E
0
Co
....
0
40
.5"
~ appears that 60% of the respiration in ethanol was due to
20
~I
0
0 so 100 150300 400 500 600
FITC dextran (kDa)
Fig. 3. Incorporation of fluorescent derivatives of dextrans in
permeabilized spheroplasts. Spheroplasts were permeabilized with 20
flgml-1 nystatin as described in the legend of Fig. I in the presence of
3.25 flM of a fluorescein isothiocyanate dextran (FITC-D) of a given
molecular weight. After 10 min of incubation the suspension was
centrifuged, the supernatant conserved (S 1), the pellet suspended in a
solution of I % Triton X-IOO and incubated 30 min with frequent
shaking and centrifugated (S2). Fluorescence of supernatant Sl and S2
was measured separately in a Kontron spectrofluorimeter; A . =
489 nm, Aem,,,,o" = 520nm. The volume occupied by FITC-D in ;h~,,~oell
was calculated and expressed as a percent of the permeable mannitol
space defined as the difference between the cellular space of intact
spheroplasts and the mannitol-impermeable space of the nystatin-
treated spheroplasts.
71
shown in Table 1. A relevant parameter was the degree of
stimulation of the cellular respiration, i.e. the percentage of
stimulation between non-phosphorylating mitochondria (in
the presence of triethyl-tin (TEE), an inhibitor of ATP-
synthase) [43] and fully stimulated respiration by carbonyl
cyanide m-chlorophenylhydrazone (CCCP). It can be
observed that the degree of stimulation was slightly but
significantly higher in S than in C on glucose (40 vs. 30%).
Moreover, in both materials the non-phosphorylating
respiration was much higher with lactate than glucose or
ethanol, in accordance with the fact that lactate delivers
electrons to cytochrome-c. The electron flux is therefore
submitted to the control of only one coupling site instead
of two for the other substrates ([44] for a review).
Table 1 also compares the respiration rates between Sand
PS. Maximal respiration was measured either with CCCP or
ADP according to the substrate and the state of the cells
(permeabilized or not). Maximal respirations on glucose or
ethanol (in the presence of externally added NAD+ for PS)
were twice as high in S than in PS. It is worth noting that
respiration rates with NADH or ethanol plus NAD+ were
identical in PS, thus suggesting that the cytosolic alcohol
dehydrogenases did not exert any control. Therefore, the
difference observed with ethanol between Sand PS could
be due to the different capacities of mitochondria in vitro
or in vivo to oxidize external NADH, rather than to a
difference with alcohol dehydrogenase activities. It should
be noted that PS and M oxidized NADH at the same rate,
therefore indicating that there may be an eventual difference
in NADH oxidation capacity between Sand PS but not
between PS and M.
Ethanol is oxidized in both cytoplasmic and mitochondrial
matrix compartments ([45] for review). At high ethanol
concentration, on the basis of the stimulation by NAD+, it
the matricial metabolism in PS (Table 1). However, PS
respiration in the absence of added NAD+ was higher than
the mitochondrial one (see Table 2), suggesting that either
the cytosolic ADH activity was not completely abolished by
the permeabilization (NAD+ always present in PS), or that
the mitochondrial metabolism of ethanol was decreased in M
(e.g. decrease in acetyl acceptor). Moreover, we measured
apparent Km for ethanol oxidation with Sand M, and found
3.10- 5 M for the former and 10-3 for the latter; Km for PS was
in between these values. Our results clearly show that yeast
preferentially oxidize extramitochondrial ethanol, particularly
when the ethanol concentration is low. These results, in
addition to the difference in respiration with ethanol observed
between Sand PS (see above), show that the matricial ADH
is only slightly involved in ethanol oxidation in vivo.
Table 2 presents the respiration rates on different sub-
strates between M and PS. To compare the results, we
measured cytochrome-c oxidase with ascorbate plus TMPD,
72

Table 1. Effect of mitochondrial inhibitors and uncoupler on respiratory rate of different types of yeast cells metabolizing different substrates

Respiratory rate
(natom O.min-1.mg-1protein)
Whole cells Spheroplasts Permeabilized spheroplasts
Substrates
Additions

none TEE CCCP Anti. A none TEE CCCP Anti. A oligo ADP CCCP
glucose 164 13 71 07 391 18 0 21921 7737 40363 0 120 12
ethanol 22724 9809 62227 0 21O21 8922 63068 0 6504 148 13 17725
ethanol + NAD 670.3 20120 298 15
lactate 18707 133 11 22909 8409 20220 152 06 232 10 88 10
NADH 5806 22037 292 15

Respiratory rates were measured as described in Materials and methods and are the mean of 5 determinations S.D. Respiratory rates on whole cells were
expressed as natom O.min-1.mg-' protein by considering the following relationship: 1 mg dry weight =0.45 mg protein. For whole cells and spheroplasts, when
used, 100 !-1M TEE, 50 !-1M CCCP, 30 !-Ig.ml-1antimycin A (anti. A), when indicated 20 mM glucose, 109 mM ethanol or 50 mM lactate. For permeabilized
spheroplasts, when used, 100 !-Ig.ml-1oligomycin (oligo), 10 !-1M CCCP, when indicated 2 mM ADP, 2 mM NAD+, 109 mM ethanol or 10 mM NADH; when
glucose was used, medium 1 was supplemented with 2 mMATP and 2 mM NAD+ and 10 mM NH.CI.

an artificial substrate, in both M and PS. Surprisingly, I mg
protein ofpermeabilized cells contained 32% of the activity
measured in I mg of isolated mitochondria. Although this
value appeared very high, we used it to standardize the results
in mg protein of mitochondria, by assuming the same
specific activity of ascorbate oxidase in PS and M. In accord-
ance with this method of standardization, the respiration on
NADH was similar in M and PS. The respiration rate on
succinate or 2-oxoglutarate was about 2-fold higher with PS
than with M in state 3, 1.7-fold higher on glycerol-3P, and
1.5-fold on isocitrate. It can be argued that either the
metabolism of these substrates is faster in PS than in M, in
accordance with our method of standardization described
above, or that the mitochondrial content in spheroplasts was
overestimated by about 1.7-fold.
Nevertheless, the main result in Table 2 is the high
respiration rate of PS on pyruvate compared to that with
isolated mitochondria. This respiration appeared to be
exclusively intramitochondrial, since it was not stimulated
by NAD+ and completely inhibited by 4a-cyanocinnamate
(not shown), a potent inhibitor of pyruvate transport through
the internal mitochondrial membrane [46]. Pyruvate is
carboxylated into oxaloacetate in the cytoplasm which can
be converted to malate by reduction [47], so the respiration
of PS on pyruvate should be compared to that on pyruvate
with malate in M. It is clear from Table 2 that the presence
of malate cannot alone explain the difference of respiration
between PS and M; it should also be noticed that malate
stimulated the respiration of PS on pyruvate. From these
results, there was no clear evident metabolic explanation for
the difference of respiration, so this problem is under
investigation.
To evaluate the integrity of mitochondria in the nystatin-
treated spheroplasts, we measured phosphorylations and

Table 2. Comparison of respiratory rates between isolated mitochondria and permeabilized spheroplasts incubated with different substrates

Respiratory rate
(natom O.min-1.mg-1mitochondria protein)
Permeabilized spheroplasts Isolated mitochondria

Substrates state 4 state 3 +cCCP state 4 state 3 +cCCP

malate 1906 16 13 25 06 1802 1604 1401

2-oxoglu 128 16 250 22 122 16 8809 122 19 9502
isocitrate 100 16 272 53 222 16 13020 17608 16834
pyruvate 181 38 428 31 516 131 1303 1003 1202
glycerol-3P 17809 519 19 619 63 14925 36330 37947
NADH 181 19 688116 913 47 20639 57855 83260
pyr-malate 197 12 559 31 769 31 65 15 98 10 61 03
succinate 16909 566 16 341 09 122 16 26655 9001
ethanol 203 13 463 41 553 78 15008 305 16 30023

Respiratory rates were measured as described in Materials and methods and are the mean of 5 determinations S.D. Respiratory rates on permeabilized
spheroplasts were expressed as natom O.min-1.mg-1 mitochondria protein by considering the following relationship: 1 mg mitochondria protein = 0.32 mg
spheroplast protein. In both cases, when used, 100 !-Ig.ml-I oligomycin (oligo), 10 !-1M CCCP, 2 mM ADP. All the substrates were used at 10 mM except for
ethanol: 109 mM. For 2-oxoglutarate, respiratory rates in state 4 and with CCCP were determined in presence of 100 !-Ig.ml-I oligomycin and 2 mMADP.

calculated the ATP/O ratio for different substrates. For this
purpose, it is necessary to inhibit all forms of ATP con-
sumption. This was achieved by adding EDTA, sinceATPase
and kinase activities require Mg2+. ATP production by
adenylate kinase or by metabolism of endogenous carbo-
hydrates was also inhibited (see Materials and methods). By
using an external NADH-regenerating system and the ability
of yeast mitochondria to oxidize exogenous NADH, we have
previously shown that the value of ATP/O significantly
increases when respiratory flux slows down in response to
a limitation of substrate supply [48]. This result was extended
to the internal mitochondrial metabolism where at high
concentration of respiratory substrate, the electron supply
to the respiratory chain varied according to the different
dehydrogenases involved [20]. We observed that the ATP/O
ratio decreased irrespective of protonmotive force when
the electron flux increased. Figure 4 shows that the same
relationship between ATP/O ratio and electron flux was
observed in PS.
The main conclusions of the first part of this work are as
follows: (i) we show that the oxidation of ethanol is essentially
cytosolic through cytosolic alcohol dehydrogenase and
NADH .dehydrogenase located on the external side of the
inner membrane. (ii) there is a great difference between PS
and M concerning respiratory rates on pyruvate alone and
pyruvate-malate. These were much greater for PS although
pyruvate metabolism was mitochondrial in both cases. So it
may be postulated that mitochondrial preparation causes the
loss of a function associated with the transport or the
metabolism of pyruvate. (iii) by modulating the electron flux
through the respiratory chain with different substrates, it is
shown in PS and in M that the ATP/O ratio decreases when
the respiratory rate increases. This observation may be
important in the physiology of the cell since with some
substrates the respiratory rate could be controlled by
dehydrogenase activity.
Substrate diffusion
Figure 5 shows the NADH concentration dependence of the
phosphorylating respiratory rate in both isolated yeast mito-
chondria (M) and permeabilized spheroplasts (PS). The
apparent ~.5 for NADH is one order of magnitude higher in
PS than in M. However, in PS, much of the NADH (or NAD+)
could be bound on cytosolic dehydrogenases and this could
participate in this apparent increase ofKO.5 for NADH. Figure
6 rules out this possibility since the KO.5 measured in PS was
independent of the protein concentration in the respiratory
buffer. So we can assume that the high ~.5 for NADH in PS
is not due to a decrease in free NADH in the medium.
Then, various substrates were assayed in this way. The
results are summarized in Table 3. In PS, after substrate
73
3
o.....
~2
-<:
l+-----~----~------------~----~-----,
o 300 600 900
J02 (natom 0 I min I mg protein)
Fig. 4. Relationship between ATP/O and respiratory rate in permeabilized
spheroplasts and isolated mitochondria with various substrates. Permeabil-
ized spheroplasts were incubated in medium 2 and phosphorylation was
induced by adding 3 mM ADP (as described in Materials and methods).
Experimental value for mitochondria were from Fig. I in [20). As the respiratory
rate increased, for mitochondria (0) the substrates used were pyruvate,
malate, pyruvate-malate, isocitrate, 2-oxoglutarate, succinate, ethanol, at a
final concentration of 5 mM, except for ethanol 109 mM; for spheroplasts
(e) the substrates used were malate, 2-oxoglutarate, isocitrate, pyruvate-
malate, ethanol, pyruvate, succinate, at a final concentration of 10 mM,
except for ethanol I 09 mM. The values presented are from at least 5 different
experiments performed with three different mitochondria and spheroplast
preparations.
addition, we found a respiratory rate slightly higher than that
observed at state 4 of respiration (i.e. in the presence of
oligomycin at a concentration which completely blocked
ATPsynthase activity). This seems to indicate that in spite of
metabolite dilution in external medium during permeabil-
ization, a small ADP amount is still present in the cytosol. To
compare the ADP effect on M and PS respiratory rate, we
studied ADP stimulation on substrate level phosphorylation
in the presence of oligomycin and by using 2-oxoglutarate as
substrate. Indeed, it has been shown that the rate ofNADH
formation by 2-oxoglutarate dehydrogenase complex is
controlled by intramitochondrial ADP concentration [49].
External ADP addition leads to an increase in the external ADP/
internal ATP exchange, and in this way maintained a high
intramatricialADP concentration level. In these conditions,ATP
synthase was completely inhibited, and we could observe the
ADP concentration dependence of the respiratory rate on 2-
oxoglutarate. As previously found in mammals, the~.5 for ADP
in PS was one order of magnitude higher than in M.
KO.5 for NADH was measured in three respiratory states:
non-phosphorylating state in the presence of oligomycin and
74

Vm = 669 natoo'l 0 I min I mg mitochondrial protein

A K . 5 = 0.09 mM
600

500

- 400
300

200

100

o~----~------~------~------~----~
o 0.1 0.2 0.3 0.4 0.5
[NADH] (mM)

B Vm = 738 natom 0 I min I mg mitochondrial protein

Ku = 0.75 mM
700

S-
600
~

'i [ 500

-l
0
S
400

SJ
=
'-'Su
g.. s
I
o~----~------~----~------~----~
o 1 2 3 4 5
[NADH] (mM)

Fig. 5. NADH concentration dependence of the phosphorylating respiratory rate in yeast isolated mitochondria (A) and permeabilized spheroplasts (B).
Respiratory fluxes were measured as described in Materials and methods in medium I containing either 1 mM or 2 mM ADP in mitochondria and spheroplasts
respectively, and various concentrations ofNADH. Spheroplasts (l mg.ml- I) were permeabilized with 20 Ilg.ml-I nystatin.

ADP; state 3 with ADP saturating concentrations; uncoupled always higher in PS than in M. However, during non-phos-
state with CCCP concentrations inducing optimal respiratory phorylating respiration, the difference between the PS and
rates (Table 3). In all these conditions, KO.5 for NADH was M KO.5 for NADH was much higher than that between the PS
 75

Table 3. Kinetic parameters of respiratory activity with various substrates in isolated yeast mitochondria and permeabilized spheroplasts

Substrates Isolated mitochondria Permeabilized spheroplasts

and effectors V K o., V K o.,
m" m"
(natom O.min-' (mM) (natom O.min-1 (mM)
mg-' mitochondria mg-' mitochondria
protein) protein)

ADP 145 26 0.026 0.006 231 78 0.25 0.08

NADH
(+oligomycin + ADP) 162 40 0.Ql5 0.003 147 31 0.51 0.09
NADH (+ADP) 542 152 0.055 0.034 734 153 0.72 0.22
NADH(+CCCP) 866257 0.12 0.03 1034 178 4.l82.11
D-Lactate 401 79 0.12 0.02 691 109 0.530.09
2-oxoglutarate 168 23 0.39 0.21 175 34 1.740.38
Glycerol-3-P 670 133 0.91 O.l3 303 16 1.57 0.46

Spheroplasts and mitochondria were prepared as described in Materials and methods. Respiratory rates were measured as described in Materials and methods
with 0.5 mg.ml-' mitochondria and I mg.ml- 1 spheroplasts permeabilized by 20 )lg.ml-' nystatin and expressed as natom O.min-'.mg-' mitochondria protein by
considering the following relationship: I mg mitochondria protein = 0.32 mg spheroplast protein. The kinetic parameters for ADP toward respiratory rate were
measured in medium I containing 5 mM 2-oxoglutarate, 0.1 mMAp5A, 20)lg and 100)lg oligomycin per mg protein in M and PS respectively. For NADH, they
were measured in medium I containing either 20 )lg and 100 )lg oligomycin per mg protein in M and PS respectively, or I mM ADP, or 10 )lM CCCP. For D-
lactate, respiratory rate was determined in the presence of2 )lg.ml-' antimycin; for 2-oxoglutarate, in the presence of either 20 )lg and 100 )lg oligomycin per mg
protein in M and PS, respectively. With D-lactate, 2-oxoglutarate and glycerol-3P, the incubation medium I was supplemented with I mMADP (M) or 2 mM
ADP (PS). In all cases, kinetic parameters were determined after addition of various concentrations of substrates as described in Fig. 4, and were the mean of
at least three determinations S.D.

and M K05 for ADP measured with 2-oxoglutarate as substrate
(34-fold compared to 10-fold). At state 3 of respiration, this
difference was reduced (14-fold). It is worth noting that
in M as in PS, the expected increase in V max when the
respiratory state corresponded to a lower L1p was linked to a
change in K05 for NADH in such a manner that the V ma/K05
ratio was only slightly changed. Such a change in the kinetic
properties ofNADH dehydrogenase has been described by
Panov and Scaduto [50] on rat heart submitochondrial
particles; the affinity of NADH dehydrogenase to NADH
increased with a concomitant decrease in V max' certainly due
to the increase in L1",.
When D-Iactate, 2-oxoglutarate and glycerol-3-phosphate
were used as substrates, we always found a higher K05 in PS
than in M even if the difference was small compared to that
measured with both the previous substrates. In fact, we

0.8

0.2

o~------~--------~------~--------~
o 0.5 1 1.5 2
Protein (mg/ml)

Fig. 6. Dependence of the K05 for NADH on the amount ofpermeabilized spheroplasts. K05 were determined as in Fig. 4B in permeabilized spheroplasts (20
)lg.ml-' nystatin) in medium I containing 2 mM ADP and various concentrations ofNADH.
76

observed a correlation between the ratios of PS K05
Kos (an index of the diffusion kinetic constraints) and the
molecular weight of the respective metabolites (Fig. 7).
Surprisingly, this phenomenon was independent of the
respiratory rate: one could expect that the higher the
respiratory rate, the higher the diffusion kinetic constraints.
At the present time, this result is still unexplained.
From these results, we conclude that the metabolite
diffusion constraints in permeabilized yeast cells is a
general phenomenon. Nevertheless, this control is strongly
dependent on the molecular weight of the metabolites and
also probably on their ionization state.
The question arises as to whether this retarded diffusion
is due to a decrease in the permeability of porin itself.
Several years ago, yeast Saccharomyces cerevisiae
mutants characterized by the phenotypic absence of the
mitochondrial porin were constructed by disruption-deletion
of the cloned porin gene [51, 52]. These mutants exhibited
a respiration, after addition of ethanol, which was coupled
in vivo in a similar way to the parent strain cells [39, 53]. It
was observed that in these porin-deficient mutant mito-
chondria, compared to the wild type mitochondria, much
greater concentrations of ADP and the presence of Mg2+
were necessary to reach the maximal phosphorylating
respiratory rate with ethanol as substrate [53, 54]. These
authors described limitations in diffusion ofADP, CATR and
NADH through the outer membrane of mutant mitochondria
vs. M which disappeared in mitoplasts, and postulated the existence
of a novel pore to explain the viability of these cells [55].
In our work, we have continued their study by comparing
M and PS from one porin-deficient mutant of yeast Sac char-
omyces cerevisiae. Table 4 shows that when porin was
missing, the Kos for NADH in non-phosphorylating state
was lower and comparable to that of mitochondria isolated
from yeast foam (see Table 3). However, it is worth noting
that Kos for NADH measured in M and PS isolated from the
porin-deficient mutant was only four times higher than in
mitoplasts isolated from the same mutant. This was very
different from yeast foam Kos for NADH which was about
thirty times higher in PS than in M. Nevertheless, in the
parent strain, the Kos for NADH was 0.105 0.014 mM in
PS and 0.021 0.003 mM in M; these values were similar
to those of PS and mitoplasts from the porin-deficient
mutant, respectively.
In mitochondria isolated from the parent strain, porin and
unidentified channels in the open state facilitate the diffusion
of NADH through the external membrane (low Kos) (Fig.
8). When spheroplast is permeabilized, porin channel is
closed and the NADH diffusion is restricted through an
unidentified channel (high KoJ In porin-deficient mutant,
porin channel is missing, NADH diffuses through the
unidentified channel (high KoJ In mitoplast, outer mem-
brane is missing, NADH accesses directly to the external
NADH dehydrogenase (low KoJ

15
==
~
Z
=
'C
"CI

=
0
.c
CJ
oS
's 10
=
II)

~
.....
.l!l
=
CJ
.! ~
0
~
0
e
Q.. I

5 Q I

M
~
.c
Q..
CIl =- I

=
II) ~
I
~
~

0
0 200 400 600 800
Molecular weight (dalton)

Fig. 7. Relationship between permeabilized spheroplasts Kos and mitochondria Kos ratio and the molecular weight of the different metabolites. The K05 ratio
was detennined from the data of Table 3. For NADH, kinetic parameters measured in the phosphorylating state were used.
 77

Table 4. Kinetic parameters of respiratory activity with NADH in
isolated mitochondria, mitoplasts and permeabilized spheroplasts from
porin-deficient yeast mutant
V max K O.5
(natom O.min-'.mg-' (mM)
mitochondria protein)
Mitochondria 597 0.10 0.02
Mitoplasts 467 0.025 0.008
Permeabilized 449 0.12 0.02
Spheroplasts
Mitochondria, mitoplasts and spheroplasts were prepared as described in
Materials and methods. Respiratory rates were measured in medium I
containing 8 mM MgCI2 and various concentrations ofNADH on mitochondria
and mitoplasts (0.5 mg.ml-'), and on spheroplasts (1 mg.ml-') permeabilized by
20 Ilg.ml-' nystatin and expressed as natom O.min-'.mg-' mitochondria protein
by considering the following relationship: I mg mitochondria protein = 0.32
mg spheroplast protein. Kinetic parameters were determined as described in
Fig. 4 and were the mean of3 determinations S.D.
Conclusion
In this study, we adapted a permeabilization model previously
described in yeast in order to work on in situ mitochondria
of which structural and functional integrities are preserved.
In this way, we show that in permeabilized spheroplasts,
mitochondria present two main characteristics different
from isolated mitochondria: ethanol metabolism is mainly
cytosolic through cytosolic alcohol dehydrogenase and
mitochondrial external NADH dehydrogenase; in situ,
mitochondria present a respiration rate on pyruvate + malate
as substrates which is lost during the isolation procedure.
However, the decrease ofATP/O when the respiratory rate
increases previously described on isolated mitochondria is
also observed on in situ mitochondria.
We show that the restricted diffusion of ADP previously
observed in permeabilized cells from various mammalian
tissues is also present in permeabilized yeast cells.
Moreover, we have obtained findings on the nature of this
phenomenon: (i) the restricted diffusion of metabolites is
a general feature and the size of the constraint seems to
depend on both their molecular weight and their ionization
state. Thus, the kinetic constraint for diffusion is particularly
high with a high weight molecule like NADH. This may be
important for the physiology of yeast in which cytosolic
NADH could be either reoxidized directly by mitochondria
or by some cytosolic dehydrogenases. (ii) the use of the
porin-deficient mutant gives a direct proof that the
closure of porin in permeabilized cells from wild type is
responsible for the restriction of metabolite diffusion
through mitochondrial outer membrane.
However, permeabilized spheroplasts and isolated
mitochondria seem to be borderline cases in which porin
is either always closed (PS) or always open (M). In a recent
work on glucose repression in yeast [56], it has been shown
that alumine hydroxide, an inhibitor of porin channel
closure, prevents the inhibition of respiration by glucose.
This inhibition never takes place in the porin-deficient
mutant. These results seem to indicate that the opening-

IN r LATED MlTOCHO DKIA I PERM ABILIZ D PHER PLAST

PARENT TRAI

PORIN-DEFICIENT MUTANT

LE END

opeo porto d porto uold nllfled channel

Fig. 8. Scheme illustrating the permeability states of outer membrane in isolated mitochondria and permeabilized spheroplast from porin-deficient mutant and
its parent strain.
78
closing of the porin channels of the outer mitochondrial
membrane could playa role in the regulation of some
mitochondrial functions.
Acknowledgements
The authors wish to thank Professor Guy Lauquin for the
generous gift of mutant and parent strains and for helpful
discussion and Dr. R. Cooke for his contribution to the
editing of the manuscript. This work was supported by grants
from the Conseil Regional d' Aquitaine.
References
1. Balaban RS: Regulation of oxidative phosphorylation in the mammalian
cell. Am J Physiol258 (Cell PhysioI27): C377--C389, 1990
2. Brown GC: Control of respiration and ATP synthesis in mammalian
mitochondria and cells. Biochem J 284: 1-13, 1992
3. Lardy HA, Wellman H: Oxidative phosphorylations: Role of inorganic
phosphate as an acceptor system in control of metabolic rates. J Bioi
Chern 195: 215-224,1952
4. Chance B, Williams CM: The respiratory chain and oxidative phos-
phorylation. Adv Enzymo117: 65-134,1956
5. Erecinska M, Wilson DF: Regulation of cellular energy metabolism. J
Membr Biology 70: 1-14, 1982
6. Hansford RG: Control of mitochondrial substrate oxidation. CurrTop
Bioenerg 10: 217-278, 1980
7. Denton RM, McCormack JG: Ca2+ as a second messenger within
mitochondria of the heart and other tissues. Annu Rev Physiol 52:
451-466,1990
8. GroenAK, Wanders RJA, WesterhoffHV, van der Meer R, Tager JM:
Quantification of the contribution of various steps to control
mitochondrial respiration. J Bioi Chern 257: 2754-2757, 1982
9. Brand MD, Hafner RP, Brown GC: Control of respiration in non-
phosphorylating mitochondria is shared between the proton leak and
the respiratory chain. Biochem J 225: 535-539, 1988
10. Ohnishi T, Kawaguchi K, Hagihara B: Preparation and some properties
of yeast mitochondria. J Bioi Chern 241: 1797-1806, 1966
II. Mattoon JR, Sherman F: Reconstitution of phosphorylating electron
transport in mitochondria from a cytochrome c-deficient mutant. J Bioi
Chem241:4330-4338,1966
12. Kovac L, Bednarova H, Greksak M: Oxidative phosphorylation in yeast.
I. Isolation and properties of phosphorylating mitochondria from
stationary phase cells. Biochim Biophys Acta 153: 32-42, 1968
13. von Jagow G, Klingenberg M: Pathways of hydrogen in mitochondria
ofSaccharomyces cerevisiae. Eur J Biochem 12: 583-592, 1970
14. Kacser H, Burns JA: The control of flux. In: DD Davies (ed). Rate
Control of Biological Processes. Cambridge University Press, London,
1973, pp 65-104
15. Heinrich R, RapoportTA: A linear steady-state treatment of enzymatic
chains. General properties, control and effector strength. Eur J Biochem
42:89-95,1974
16. Mazat JP, Jean-Bart E, Rigoulet M, Guerin B: Control of oxidative
phosphorylations in yeast mitochondria. Role ofthe phosphate carrier.
BiochimBiophys Acta 849: 7-15, 1986
17. Mazat JP, Jean-Bart E, Rigoulet M, Reder C, Guerin B: Control of
oxidative phosphorylation in yeast mitochondria: The role of phosphate
carrier and pH. In: A Azzi et al. (eds). Anion Carriers of
Mitochondrial Membranes. Springer-Verlag, Berlin Heidelberg,
1989, pp 327-336
18. Beauvoit B, Rigoulet M, Guerin B: Thermodynamic and kinetic control
of ATP synthesis in yeast mitochondria: Role of L1pH. FEBS Lett
244: 255-258, 1989
19. Ouhabi R, Rigoulet M, Guerin B: Flux-yield dependence of oxidative
phosphorylation at constant L1J.1-H+ FEBS Lett.254: 199-202, 1989
20. Fitton Y, Rigoulet M, Ouhabi R, Guerin B: Mechanistic stoichiometry of
yeast mitochondrial oxidative phosphorylation. Biochemistry 33:
9692-9698, 1994
21. Robinson BH, Ward J, Goodyer P, BaudetA: Respiratory chain defects
in the mitochondria of cultured skin fibroblasts from three patients
with lacticademia. J Clin Invest 77: 1422-1427, 1986
22. Bogucka K, WroniszewskaA, Bednarek M, Duszynski J, Wojtczak L:
Energetics of Ehrlich ascites mitochondria: Membrane potential of
isolated mitochondria and mitochondria within digitonin-permeabilized
23.
cells. BiochimBiophysActa 1015: 503--509, 1990
Kunz WS, Kuznetsov AY, Schulze W, Eichhorn K, Schild L, Striggow F,
Bohnensack R, NeuhofS, GrasshoffH, Neumann HW, Gellerich FN:
Functional characterization of mitochondrial oxidative phosphorylation
in saponin-skinned human muscle fibers. Biochim Biophys Acta 1144:
46-53,1993
24. Saks VA, Belikova YO, Kuznetsov AV: In vivo regulation of mitochondrial
respiration in cardiomyocytes: specific restrictions for intracellular
diffusion ofADP. Biochim BiophysActa 1074: 302-311, 1991
25. Saks VA, Vasilyeva E, Belikova YO, Kuznetsov AV, Lyapina S, Petrova
L, Perov NA: Retarded diffusion of ADP in cardiomyocytes: Possible
role of mitochondrial outer membrane and creatine kinase in cellular
regulation of oxidative phosphorylation. Biochim Biophys Acta 1144:
134-148,1993
26. Saks VA, Kuznetsov AY, Khuchua ZA, Vasilyeva EY, Belikova JO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis: possible involvement of mitochondrial-
cytoskeleton interactions. J Mol Cell Cardiol27: 625--645, 1995
27. Saks V, Belikova Y, Vasilyeva E, Kuznetsov A, Fontaine E, Keriel C,
Leverve X: Correlation between degree of rupture of outer mitochondrial
membrane and changes of kinetics of regulation of respiration by ADP
in permeabilized heart and liver cells. Biochem Biophys Res Commun
208:919-926,1995
28. Fontaine E, Keriel C, Lantuejoul S, Rigoulet M, Leverve XM, Saks VA:
Cytoplasmic cellular structures control permeability of outer mito-
chondrial membrane for ADP and oxidative phosphorylation in rat
liver cells. Biochem Biophys Res Commun 213: 138-146, 1995
29. Murakami K, Nagura H, Yoshino M: Permeabilization of yeast cells:
Application to study on the regulation of AMP deaminase activity in
situ. Anal Biochem 105: 407-413, 1980
30. Serrano R, Gancedo JM, Gancedo C: Assay of yeast enzymes in situ.
EurJBiochem34:479-482,1973
31. Adams BG: Method for decryptification of a-glucosidase in yeast with
dimethylsulfoxide. Anal Biochem45: 137-146, 1972
32. Jaspers HTA, Christianse K, van Steveninck J: An improved method
for the preparation of yeast enzymes in situ. Biochem Biophys
Res Commun 65: 1434-1439, 1975
33. Miozzari GF, Niederberger P, Hutter R: Permeabilization of micro-
organisms by triton X-IOO. Anal Biochem 90: 220-223, 1978
34. Oliveira DE, Santos Neto ALC, Panek AD: Permeabilization of yeast for
in situ determination ofa-glucosidase. Anal Biochem 113: 188-192, 1981
35. Gowda LR, Joshi MS, Bhat SG: In situ assay of intracellular enzymes of
yeast (Klayvermyces fragilis) by digitonin permeabilization of cell
membrane. Anal Biochem 175: 531-536,1988
36. Cordeiro C, Freire AP: Digitonin permeabilization of Saccharomyces

cerevisiae cells for in situ enzyme assay. Anal Biochem 229:
145-148, 1995
37. Hutchison HT, Hartwell LH: Macromolecule synthesis in yeast
spheroplasts. J Bacteriol 94: 1697-1705, 1967
38. Sutton DD, Arnow PM, Lampen JO: Effect of high concentrations of
nystatin upon glycolysis and cellular permeability in yeast. Proc Soc
Exp Bioi NY 108: 170--175,1961
39. Michejda J, Guo XJ, Lauquin GJM: Bioenergetic consequences ofthe
lack of mitochondrial porin: Identification of a putative new pore. In: A
Azzi et al. (eds). Anion Carriers of Mitochondrial Membranes. Springer-
Verlag, Berlin Heidelberg, 1989, pp 225-235
40. Guerin B, Labbe P, Somlo M: Preparation of yeast mitochondria
(Saccharomyces cerevisiae) with a good PIO and respiratory ratio.
MethEnzymol. 55: 149--159, 1979
41. Daum G, Bohni PC, Schatz G: Import of protein into mitochondria:
cytochrome b, and cytochrome c peroxidase are located in the
intermembrane space of yeast mitochondria. J Bioi Chern 257: 13028-
13033,1982
42. Ouhabi R, Boue-Grabot M, Mazat JP: ATP synthesis in permeabilized
cells. Assessment oftheATPIO ratio in situ. In: E Gnaiger, FN Gellerich,
M Wyss (eds). Modem Trends in BioThermoKinetics. Vol. 3. Innsbruck
University Press, 1994, pp 141-144
43. Cain K, Griffith DE: Studies of energy-linked reactions: Localisation of
the site of action oftrialkyltin in yeast mitochondria. Biochem. J 162:
575--580,1977
44. Guerin B: Mitochondria. In: AH Rose, JS Harrison (eds). The Yeasts.
Academic Press, London, 4,1991, pp 541--600
45. Fraenkel DG: Carbohydrate metabolism. In: IN Strathern, EW Jones,
JR Broach (eds). The Molecular Biology of the Yeast Saccharomyces.
Metabolism and Gene Expression. Cold Spring Harbor Laboratory,
1982, pp 1-37
46. Nalecz MJ, Nalecz KA, Azzi A: Purification and functional character-
isation of the pyruvate (monocarboxylate) carrier from baker's yeast
mitochondria (Saccharomyces cerevisiae). Biochim Biophys Acta
79
1079: 87-95, 1991
47. Lim F, Rohde M, Morris CP, Wallace JC: Pyruvate carboxylase in the
yeast pyc mutant. Arch Biochem Biophys 258: 259--264, 1987
48. Ouhabi R, Rigoulet M, Lavie JL, Guerin B: Respiration in non-
phosphorylating yeast mitochondria. Roles of non-ohmic proton
conductance and intrinsic uncoupling. Biochim Biophys Acta 1060:
293--298, 1991
49. Rigoulet M, Velours J, Guerin B: Substrate-level phosphorylation in
isolated mitochondria. Eur JBiochem 153: 601--607,1985
50. Panov AV, Scaduto RC Jr: Influence of calcium on NADH and succinate
oxidation by heart submitochondrial particles. Arch Biochem Biophys
316:815--820,1995
51. Guo XJ, Lauquin GJM: Mitochondrial porin-deficient mutant of S.
cerevisiae: In vitro construction and preliminary properties. EBEC
Reports 4: 292, 1986
52. Dihanich M, Suda K, Schatz G: A yeast mutant lacking mitochondrial
porin is respiratory deficient, but can recover respiration with
simultaneous accumulation of an 86 kD extramitochondrial protein.
EMBOJ 6: 723--728, 1987
53. Michejda J, Guo XJ, Lauquin GJM: The respiration of cells and
mitochondria of porin deficient yeast mutants is coupled. Biochem
Biophys Res Commun 171: 354--361, 1990
54. Michejda J, Guo XJ, Lauquin GJM: Energy coupling in cells and
mitochondria ofporin yeast mutants. EBEC Reports 5: 132, 1988
55. Michejda J, Kmita H, Stobiana 0, Budzinska M, Lauquin GJM:
Restrictions of metabolite permeation through the outer mito-
chondrial membrane of porin-deficient yeast mutant. In: M Forte,
M Colombini (eds). NATO ASI Series Vol. H 83. Molecular Biology
of Mitochondrial Transport Systems. Springer-Verlag, Berlin
Heidelberg, 1994, pp 341-356
56. Ahmadzadeh M, Horng A, Colombini M: The control of
mitochondrial respiration in yeast: A possible role of the outer
mitochondrial membrane. Cell Biochem Function 14: 201-208,
1996
Molecular and Cellular Biochemistry 184: 81-100, 1998.
1998 Kluwer Academic Publishers.

Permeabilized cell and skinned fiber techniques in

studies of mitochondrial function in vivo

Valdur A. Saks, 1 Vladimir 1. Veksler,2 Andrei V. Kuznetsov, 3Laurence

Kay, 1 Peeter Sikk, 1 Toomas Tiivel, 1 Leone Tranqui, 1 Jose Olivares, 1
Kirstin Winkler,3 Falk Wiedemann3and Wolfram S. Kunz3
lLaboratories of Bioenergetics, Institute of Chemical Physics and Biophysics, Tallinn, Estonia and J. Fourier University,
Grenoble, France; 2INSERM U446, Chatenay-Malabry, France; 3Neurobiochemisches Labor, Klinikfur Neurologie,
Universitatsklinikum Magdeburg, Germany

Abstract
In this chapter we describe in details the permeabilized cell and skinned fiber techniques and their applications for studies of
mitochondrial function in vivo. The experience of more than 10 years of research in four countries is summarized. The use of
saponin in very low concentration (50-100 I-lg/ml) for permeabilisation of the sarcolemma leaves all intracellular structures,
including mitochondria, completely intact. The intactness of mitochondrial function in these skinned muscle fibers is
demonstrated in this work by multiple methods, such as NADH and flavoprotein fluorescence studies, fluorescence imaging,
confocal immunofluorescence microscopy and respiratory analysis. Permeabilized cell and skinned fiber techniques have several
very significant advantages for studies of mitochondrial function, in comparison with the traditional methods of use of isolated
mitochondria: (1) very small tissue samples are required; (2) all cellular population of mitochondria can be investigated; (3)
most important, however, is that mitochondria are studied in their natural surrounding. The results of research by using this
method show the existence of several new phenomenon - tissue dependence of the mechanism of regulation of mitochondrial
respiration, and activation of respiration by selective proteolysis. These phenomena are explained by interaction of mitochondria
with other cellular structures in vivo. The details of experimental studies with use of these techniques and problems of kinetic
analysis ofthe results are discussed. Examples oflarge-scale clinical application of these methods are given. (Mol Cell Biochem
184: 81-100, 1998)

Key words: mitochondrial respiration, skinned fibers, permeabilized cell, heart, skeletal muscle, regulation, cytoskeleton, myopathies

I. Historical background
Investigation of the mitochondrial function in muscular tissue
under various conditions was and still is the aim of very
numerous studies in the field of cellular bioenergetics. Very
often such studies are performed on whole tissues or organs.
Nuclear magnetic resonance (NMR) spectroscopy is used to
evaluate the intracellular concentrations of metabolites
controlling respiratory activity. An important advantage of
this approach is that cellular architecture and interactions
between different cellular compartments are well preserved,
and the mitochondria can be studied in situ. The main
disadvantage of this in vivo method is that the extra-
mitochondrial milieu can not be controlled. It is not possible
to change intracellular medium (for example, to eliminate
undesirable respiratory substrates) in living tissue; moreover,
it is not even possible to precisely determine concentrations
of metabolites which are known to regulate respiration. One
of the main mitochondrial characteristics is the dependence
of the oxygen consumption rate on freeADP concentration.
The latter, however, can only be calculated (not measured
directly) using the NMR data and several assumptions have

Present address: v.A. Saks, Laboratory of Bioenergetics, Institute of Chemical Physics and Biophysics, Tallinn, Estonia
Address for offprints: V.A. Saks, Laboratory of Bioenergetics, Joseph Fourier University, BPS3 X-3804l Grenoble Cedex, France
82
to be made (pennanent temporal and spatial equilibrium of the
creatine kinase reaction, homogeneous cytosolic distribution
of pH, ATP, phosphocreatine, creatine, etc.) which may not
always be valid.
Another approach which has been exploited since more
than 40 years is the using of isolated mitochondria. This
method allows to perfectly control the medium composition
around the organelles, it is reliable, rapid and relatively cheap.
However, isolation of mitochondria by homogenization of
tissue and sedimentation of mitochondria by centrifugation
can seriously affect structural and functional properties ofthe
organelles. Disruption of the cellular architecture is likely to
alter the parameters concerning the interaction between the
mitochondria and the extramitochondrial space and to destroy
the organized structure of cytoplasm. This is particularly
important ifthe mitochondria are being isolated from tissue
already damaged by a pathological process. Furthennore, the
limited mitochondrial yield achieved during the procedure
of isolation does not allow to investigate the total tissue
mitochondrial population.
About 10 years ago, a novel method for studies of mito-
chondria in vivo which combined the advantages of different
approaches was proposed [11]. According to this method,
using muscle fibers penneabilized with saponin, the function
of the total mitochondrial population can be studied in situ,
without isolation of mitochondria from tissue. This technique
gives the possibility to precisely control the extramito-
chondrial medium. Very important, the method can be
applied to extremely small pieces of tissue that made it
possible to study the mitochondrial function with very small
samples which is important if the amount of material is
limited (biopsy samples, tissues of expensive transgenic
animals, cell cultures). Also, by using this technique unique
properties of mitochondria in the intact cells which are lost
during their isolation can be studied (see Section III 2. ofthis
chapter).
The idea to use a detergent for penneabilization of the
plasmalemma to study the functioning of intracellular
membrane organelles actually came from a work of Endo
and Kitazawa [2]. These authors used a low (50 Ilg/ml)
concentration of saponin to penneabilize bundles of cardiac
and skeletal muscle. They found that in these preparations
the intracellular space was easily accessible for solutes such
as Ca 2+, Mg2+, ATP, and EGTA, this suggesting that the
sarcolemma was perforated and lost its integrity. At the
same time, however, sarcoplasmic reticulum of saponin-
skinned cardiac fibers was still essentially similar to that of
mechanically skinned skeletal fibers indicating the integrity
of this intracellular membrane system.
Such a selective action of glycosides of plant-origin like
digitonin and saponin, which contain hydrophobic steroid
core and hydrophilic glycoside tails can be explained by
different lipid composition of cellular membranes. Saponin
and digitonin, due to the hydrophobic steroid core, have high
affinity for cholesterol and preferentially extract cholesterol
from membranes, therefore they specifically 'attack'
cholesterol-rich membranes such as the plasmalemma [3].
Endoplasmic reticulum has much lower cholesterol content;
inner mitochondrial membranes contain even lower amount
of this lipid [4, 5]. As a result, saponin displays much less
deleterious effects on endoplasmic reticulum and mito-
chondria, and for their disrupture much higher concentrations
of these detergents are required.
These well known differences in the cholesterol content
of plasmalemma, mitochondrial outer and inner membranes
are the basis for selecting of appropriate saponin or digitonin
concentrations necessary to selectively dissolve some
membranes but leaving the others intact, as discussed above.
Therefore, to permeabilize the sarcolemma, to prepare
mitoplasts with dissolved mitochondrial outer but intact
inner membranes, and to prepare so called right-side out
submitochondrial particles, one has to use 0.05,6 or higher
than 10 mg/ml of saponin or digitonin, respectively [1, 2,
6, 7]. Very recently and very surprisingly, Kuchmerick's
group failed [8] to account for this concentration difference
when explaining the low mitochondrial affinity for ADP in
penneabilized cells or skinned fibers by results obtained for
submitochondrial particles in 1967 which showed also low
affinity for ADP [7]. However, already in good old times
of Albert Lehninger who started to use these agents, more
than 100 times difference in the concentrations of a detergent
to be used was apparent, with all very clear consequences
for the cellular structures directly and experimentally
demonstrated during subsequent 30 years [2-5, 9-14]. In
fact, when submitochondrial particles are prepared by using
very high concentration of digitonin or by sonication, the
mitochondrial matrix enzymes and cofactors are released,
and the low affinity of these systems for ADP may by related
to kinetic factors such as slow reloading of vesicles by ADP
or ATP, but not denaturation of membrane proteins: even
after isolation from the membrane, adenine nucleotide
translocator preserves high affinity for ADP [15].
The method of investigation of mitochondrial function in
penneabilized preparation was validated using the following
criteria: (1) well-preserved mitochondrial morphology; (2)
complete penneabilization of the plasmalemma; (3) nonnal
mitochondrial functions.
Careful ultrastructural study of penneabilized preparations
was perfonned by several groups [1, 9-14]. Altschuld et at.
[9] found that digitonin-treated cardiomyocytes retained good
overall morphology. Although, as expected, the sarcolemma
was missing, mitochondria remained intact and retained the
appearance of those in control, non-treated cells. The
sarcomers of the digitonin-lysed cells were well-aligned
with only a slight thickening ofthe Z-lines. The sarcoplasmic
reticulum appeared intact in these cells and numerous

mitochondria remained attached along the surface. These
mitochondria were retained following washing and fixation
steps and appeared to be firmly bound to the cell by interme-
diate filaments or other attachments. Scanning micrographs
show these surface mitochondria as spherical particles with
rather thick filamentous attachments.
Electron microscopic investigation of saponin-skinned
ventricular fibers [1, 10] also revealed the intact ultrastructure
of mitochondria and sarcoplasmic reticulum, though the
sarcolemma was completely removed. The mitochondrial outer
membrane is known to have higher sensitivity to saponin and
digitonin as compared to the inner membrane; nevertheless
mitochondrial population in saponin-permeabilized fibers
demonstrated well-preserved outer mitochondrial membranes
[10]. Similarly, isolated cardiomyocytes treated with saponin
retained normal mitochondrial morphology in spite of the
absence of visible sarcolemma [11]. All these data clearly show
that the permeabilizing treatment with low concentrations of
saponin or digitonin does not alter mitochondrial ultrastructure.
Most importantly, permeabilization of fibers does not alter the
permeability of the outer mitochondrial membrane since
addition of exogenous cytochrome c has no effect on respiration
of the fibres [10]. Also, creatine significantly modulates the
respiratory activity of skinned ventricular fibers at sub-
maximal ADP concentrations thus indicating the presence of
active mitochondrial creatine kinase. Also, in skinned cardiac
fibers creatine kinase stayed fixed in the coupled state to the
membrane of sarcoplasmic reticulum [16] that demonstrating
complete intactness of this membrane system.
Direct morphological evidence for intactness of the
mitochondrial population in saponin-permeabilized muscle
cells has been given by Penman group [12, 13]. They used
treatment with 100 Ilg/ml saponin for 30 min to permeabilize
smooth muscle cells, and embedment-free electron micro-
scopy to unmask mitochondrial-cytoskeleton interactions,
and showed that under these conditions not only outer
mitochondrial membrane but also its multiple contacts with
cytoskeleton are perfectly preserved [12, 13]
The next criterion for complete permeabilization of the
sarcolemma is the absence of barriers for solutes between
the intracellular and extracellular spaces. Digitonin-lysed
cardiomyocytes were shown to loose 98% of initial lactic
dehydrogenase as well as myoglobin and other cytosolic
components but not mitochondrial cytochromes [9]. Similar
results were obtained in skinned ventricular [1] and skeletal
muscle [14] fibers. Numerous experiments carried out in
saponin-skinned fibers have demonstrated that intracellular
membrane structures such as myofilaments or sarcoplamic
reticulum respond quickly to changes in concentrations of
ions, adenine nucleotides, phosphocreatine, EGTA, amino
acids etc. This indicates that the composition of the intra-
cellular space in the saponin-permeabilized tissues and cells
can be well controlled by external additions.
83
The third criterion is the functional state of mitochondria in
permeabilized preparations. Cytochrome aa 3 content in
saponin-skinned cardiac fibers was found to be very similar
to that in intact cardiac muscle [1]. Digitonin treatment did not
produce any significant release of the mitochondrial matrix
enzyme, citrate synthase, from cardiomyocytes [17]; release
of the enzyme during the treatment of skeletal muscle with
saponin was very weak [14]. Oxidative activity of the
permeabilized preparations strongly depended on the presence
of respiratory substrates [1]. Mitochondria in the saponin-
skinned fibers are capable to use various substrates [14, 18]
which suggests that all mitochondrial complexes are active. No
difference from isolated mitochondria was detected with
respect to the oxidation of the different substrates applied and
with the sensitivity to different inhibitors [14]. ADP stimulates
respiration of permeabilized preparations manyfold so that the
acceptor control ratio is in the range of 5-7 for cardiac muscle
of various species. Carboxyatractyloside, an inhibitor of the
adenine nucleotide translocator, completely inhibits theADP-
stimulated respiration in the permeabilized cells. These results
demonstrated a high level of coupling between oxidation and
phosphorylation processes. Direct measurement of the ATP
synthesis in digitoninpermeabilized cells [19] showed ATP /0
ratios which were very similar to those obtained with isolated
mitochondria using the same respiratory substrates.
Intactness of mitochondria and good preservation of the
mitochondrial function after permeabilization have been
demonstrated not only in muscular tissues but also in other
preparations, for example, in unicellular eukariotic organisms
[20].
Thus, all the criteria clearly show that the properties of
mitochondria in situ in permeabilized preparations are well
preserved. This is further confirmed and described below
with presentation of new experimental results from our
laboratories. Analysis ofthe mitochondrial function in saponin-
skinned fibers is sensitive enough to reveal alterations in
various forms of cardiac or muscular pathology [14, 21, for
review see ref.22])
II. Materials and methods
Preparation of skinned muscle fibers in experiments
Skinned fibers are prepared from tissue samples oflaboratory
animals or biopsy samples taken from human muscles,
according to the methods described in [1,10, 11,21-27]. In
summary, to prepare the skinned fibers from the heart and
skeletal muscles of small laboratory animals, the latter are
anaesthetised with sodium pentobarbital (50 mg/kg body
weight, i.p.) and treated with heparin (1500 IV/kg body
weight, i.v.), chest opened and hearts when still beating
excised and put into cooled solution A (see below). Cooled
84
hearts are cut into halves and muscle strips (2-4 mm long and
1-1.5 mm in diameter, 5-7 mg of wet weight) cut from
endocardium ofleft ventricles along fiber orientation to avoid
mechanical damage of the cells. Muscle fiber bundles (3- 4
mm long, about 1 mm in diameter) can be taken also from
m. soleus (oxidative, slow twitch) and m. gastrocnemius white
(glycolytic, fast twitch), m. gastrocnemius red (oxidative
glycolytic, fast twitch) and also from m. plantaris (oxidative-
glycolytic, fast twitch white), tibialis anterior (fast twitch,
oxidative-glycolytic) and m. quadriceps (oxidative-glyco-
lytic, mixed) and other muscles. By using sharpended forceps
or needles, the muscle fibers are separated from each other
leaving only small areas of contact. After that, the fibers are
transferred into vessels with cooled (in ice) solution A
containing 50 J.lg of saponin per ml and incubated at mild
stirring for 30 min for complete solubilization of the
sarcolemma. Note that this concentration is lower by a factor
of 200 compared to that used by Lehninger in 1967! [7].
Permeabilized (skinned) fibers are then washed in solution
B for 10 min; this procedure of washing is repeated two more
times to remove completely all metabo1ises, especially trace
amounts offreeADP. Complete removal offreeADPcan be
easily seen from respiration recordings which should show
very reproducible initial State 2 rates (designated as v J not
sensitive to inhibition by atractyloside (see below).
Isolation ofhuman skeletal muscle fibers
This is a slight modification [14] of the method described
above. The human skeletal muscle biopsy sample (about
50--100 mg wet weight m. vastus lateralis) is transferred
immediately into 25 ml ofan ice-could isotonic NaCI solution.
Fiber bundles are isolated by the careful mechanical dissection
of approximately 50 mg of the muscle tissue in a relaxing
solution A into 5-10 mm long and 100 to maximally 250 J.lffi
thick bundles (containing each between 2-5 single fibers).
Saponin-skinned human muscle fibers are prepared by 30 min
of gentle agitation of these bundles in the relaxing solution
supplemented with 50 J.lg/ml saponin. Thereafter, saponin is
removed by 10 min incubation of the fibers (gentle agitation)
in the medium for respiration measurements containing 5 mM
MgCI2,60mMKCl, 100 mM mannitol, IOmMKH2P04,0.5
mM Na2EDTA, 60 mMTris-HCl (PH 7.4) supplemented with
10 mg/ml essential fatty acid free serum albumin. For
subsequent oxygraphic measurements (within 5 h) the fibers
can be stored in this solution at OC. The long term stability of
these preparations can be considerably improved (up to several
days) by storage of isolated fiber bundles in the relaxing
solution. All of these procedures have to be performed on ice.
Isolation and culturing ofadult cardiac myocytes and
phantom cardiomyocytes.
Male Wistarrats weighing 300--350 g are used in experiments.
Calcium-tolerant myocytes are isolated by perfusion with a
collagenase-containing medium in a Langendorff apparatus
under sterile conditions. The hearts are perfused by about 100
ml of a calcium free solution (initial solution) at perfusate
flow rate 12-14 ml/min, pH 7.3, 37C, saturated with a gas
mixture 95% 02 - 5% CO2 and containing, in mM: NaCI 70,
KCI10,KH:P04 4,NaHC03 15,MgCI21,glucose ll,sucrose
120 and pyruvate 2. Then the hearts are perfused during about
30 min at 4 ml/min with the same solution to which 1%
bovine serum albumin (BSA) and 1 mg:ml of collagenase D
(Boehringer) are added. After that ventricles are placed on
Petri dishes ( 100 mm in diameter) with 10 ml of this
solution and 10 ml of initial solution with 2% ofBSA. The
cardiomyocytes are then gently dissociated by rubber-
polisman and pipette suction and reversing action. The
suspension of cardiomyocytes is then centrifuged at 30 x g
for 60 sec and myocytes are washed three times and sedimented
during 10 min in the initial solution with 2% BSA. Cardio-
myocytes are then progressively transferred into calcium-
tolerant medium by three washes and sedimentation with
mixing of initial solution with 2% BSA and Sigma medium
199 in ratios 3/1, 3/2 and 3/3 respectively. About 1-1.5 x 106
calcium tolerant cardiomyocyte in 15 ml of culture medium
o (Sigma 199 added of insuline 10--9 M, sucrose 55 mM,
Hepes 20 mM, NaHC03 10 mM, Penicillin 100 IV/ml and
streptomycin 0.1 mg/ml) are placed into one Petri dish and
incubated at 37C in atmosphere containing 5% CO2 and 95%
air. After 2-3 hours of culture the cardiomyocytes are washed
once with culture medium O. The rod shape calcium- tolerant
cardiomyocytes were then incubated during 18 h until
saponin treatment
To prepare phantom cardiomyocytes without myosin
[28], the cardiomyocytes are sedimented from incubation
medium and suspended in solution A (see below). For
permeabilization of the sarcolemma, saponin is added in
concentration of 50 J.lg/ml and the suspension is incubated
for 30 min with slight stirring, and washed by sedimentation
5 times in solution B (control permeabilized cardiomyocytes)
or in solution C (see below) when phantom cardiomyocytes
are prepared. Washed cardiomyocytes are suspended in
solution D, containing 800 mM KCI to extract myosin, and
incubated for 30 min. After that, phantom cardiomyocytes are
sedimented and washed again 5 times in solution B. The
extraction of myosin is controlled by determination of the
protein concentration in suspension before extraction and in
extracts after sedimentation. Usually, the extracts contain 15-
20% of the total protein.
In experiments with cardiomyocytes, centrifugation is
never used to avoid the damage of the cells.
Respiration studies
Respiration of skinned fibers is usually measured at 25C
using a high resolution Oroboros oxygraph (Anton Paar,
Graz) or by using the Yellow Spring Clark oxygen electrode

in solution B, containing respiratory substrates (see below)
and 2 mg per ml of bovine serum albumin. Solubility of
oxygen at that temperature is taken as 215 nmoles 02 per ml
[1]. Alternatively, iso-osmotic solutions on sucrose or
mannitol basis used in studies of isolated mitochondria can
be used.
Solutions
Composition of the solutions used is based on the information
of the ionic contents in the muscle cells cytoplasm (see ref.
19 for further references). Solution (A) contains, in mM:
CaK2 EGTA 2.77, K 2EGTA 7.23, MgCl2 6.56, dithiothreitol
(DTT) 0.5, K-MES 50, imidazole 20, taurine 20, Na2ATP 5.3,
phosphocreatine 15, pH 7.1 adjusted at 25C. Solution (B)
contains, in mM: CaK 2EGTA 2.77, K 2 EGTA 7.23, MgCl 2
1.38, dithiothreitol (DTT) 0.5, K-MES 100, imidazole 20,
taurine 20, K 2HP0 4 3 and pyruvate 5 (or glutamate 5) +
malate, 2, pH 7.1 adjusted at 25C. Solution (C) contains, in
mM: taurine 20, dithiothreitol 0.5, MgCl2 1O,ATP 10, K-MES
80, HEPES 50, pH 7.1 adjusted with KOH. Solution (D)
contains, in mM: taurin 20, dithiothreitol 0.5, MgCl 2 10,ATP
10, KC1800, HEPES 50, pH 7.1 adjusted with KOH. Solution
KCl for cytochrome c test contains in mM: KC1125, Hepes
20, glutamate 4, malate 2, Mg-acetate 3, KHl04 5, EGTA
004 and DTT 0.3, pH 7.1 adjusted at 25C and 2 mg ofBSA
per ml is added.
Registration offfluorescence changes in mitochondria in
skinned fibers
Important information of the mitochondrial function in
saponin-skinned fibers can be obtained by measuring of
fluorescence ofNAD(P)H and fluorescent flavoproteins [29,
30]. At 325 nm HeCd-laser excitation the fluorescence
changes detectable at 450 nm emission can be attributed
nearly exclusively to mitochondrial NADH. On the other
hand, the flavoprotein fluorescence signal detected at 520 nm
using 454 nm argon-ion laser excitation is caused mainly by
the oxidized form of a-lipoamide dehydrogenase, being in
close redox equilibrium with the mitochondrial NAD-system
[31]. For the detection of these fluorescence signals in
bundles of saponin-skinned muscle fibers the following
experimental setup, described in detail in [29], is used. Between
2 and 5 mg wet weight skinned fibers are immobilized (by the
attachment to glass wool) in a light screened quartz tube and
peri fused at I mllmin with the medium used for the oxygraphic
measurements. The NAD(P)H fluorescence is excited at 325
nm using the 8 mW beam of an Omnichrom OMI-2056 HeCd
dual-wavelength laser. The flavoprotein fluorescence is excited
at 454 nm using a 75 mW Omnichrom OMI-523 AP argon-
ion laser. The excitation light is guided to the immobilized
sample using a quartz fiber. The fluorescence light is
85
monitored by a Shimadzu RF 5001 spectrofluorimeter
connected at 40 (with respect to the excitation beam) with
a quartz light guide (6 mm inner diameter, Schott, Mainz)
to the peri fusion chamber at the emission wavelength 450
nm (NAD(P)H fluorescence) or 520 nm (flavoprotein
fluorescence). This method allows the determination of redox
state changes of mitochondrial NAD-system on substrate,
ADP and inhibitor additions
Immunofluorescence confocal microscopy of
permeabilized and phantom cardiomyocytes
Double labelling of desmin and mitochondria is performed
on intact cardiomyocytes and phantoms in suspension. We
used a modified method of Granger and Lazarides [32]. The
fixation of the cells is achieved by incubating them during
15 min in freshly prepared 3% paraforaldehyde in phosphate
buffered saline (PBS) at 37C. Permeabilization of intact
cardiomyocytes is achieved by incubation in 1% Triton in
PBS for 30 min. Phantom cardiomyocytes do not need to be
treated by Triton since they are already permeabilized. Then
intact and phantom cardiomyocytes are incubated with
polyclonal antidesmin antibodies for 30 min at 37C. Then
all samples are incubated with rodhamin-conjugated F(ab')2
secondary antibodies and then with nonylorange solution (0.1
J.1g/ml) during 15 min. At this time, the cells are washed three
times for 15 min in PBS and then in water. Finally the labelled
cells in suspension are deposited on class coverships and
mounted in a mixture of mowiol and glycerol to which
1,4diazobocyclo-(2,2,2)-octane was added. Samples are
observed by confocal microscopy performed with a LSSM41 0
Zeiss confocal microimaging system equipped with a plan apo
x 63 oil immersion, NA lAO objective lens. The micrographics
shown in this work represent optical sections 0,5 J.1ffi thick.
III. Results
Intactness of mitochondrial function in saponin skinned
muscle fibers: substrate-dependent fluorescence changes
In Fig. 1, the typical responses ofNAD(P)H fluorescence
(upper trace) and of flavoprotein fluorescence (lower
trace) for human vastus lateralis fibers to the addition of
octanoylcamitine+malate and ADP7 glutamate and cyanide
are shown. Substrate addition (1 mM octanoylcamitine and
5 mM malate) led to an increase in NADH fluorescence and
decrease in flavoprotein fluorescence. Subsequent ADP
addition (1 mM) caused a reoxidation of the mitochondrial
NAD-system visible as decrease in NAD(P)H fluorescence
and increase in flavoprotein fluorescence. This is the result of
the stimulation of the mitochondrial oxidative phosphorylation.
86
KCN
f The opposite fluorescence behavior of NAD(P)H and
OC + MAL
Fig. 1. Fluorescence changes of NAD(P)H (upper trace) and fluorescent
flavoproteins (lower trace) in saponin-skinned human skeletal muscle fibers.
Additions: I mM octanoylcamitine (OC) and 5 mM malate (MAL), I mM
ADP, 10 mM glutamate (GLU), 4 mM KeN.
Improvement of the supply of reducing equivalents in the
presence of 10 mM glutamate caused under these conditions
an increase of reduction of the mitochondrial NAD-system,
visible as increase in NAD(P)H fluorescence and decrease in
flavoprotein fluorescence. The maximal reduction of the NAD-
system can be reached by cyanide addition (4 mM). Under
these circumstances the highest NAD(P)H fluorescence and
the lowest flavoprotein fluorescence is reached. The redox
states obtained under conditions of intermediate reduction can
be calibrated using the fluorescence signals in the endogenous
state as the 0% reduction reference value and the state in the
presence of 10 mM glutamate and 4 mM cyanide as the 100%
reduction reference value, respectively. Therefore, this method
allows to detect at a well defiled influx of reducing equivalents
(constant activity of primary dehydrogenases) changes related
to the rates of mitochondrial oxidative phosphorylation: an
increase in rate of respiration leads to a lowered redox state
of the mitochondrial NAD-system (decrease in NAD(P)H
fluorescence and increase in flavoprotein fluorescence). On
the other hand, at a defiled level of oxidative phosphorylation
activity changes of primary dehydrogenases (e.g. activation
by calcium ions) should cause an increased reduction of the
mitochondrial NAD-system (increase in NAD(P)H fluore-
scence and decrease in flavoprotein fluorescence). This
behavior can be visualized in the following flow diagram
of reducing equivalents:
primary dehydrogenases => (NADHlNAD+)mjt => respiratory
chain => 02
t
(FpH/FP*)a_liP DH
In addition, at comparable conditions of fluorescence
measurements it is possible to estimate from the maximal
fluorescence signal changes the relative amount of fluoro-
chroms within a muscle fiber sample [29-31]. This amount
depends on the mitochondrial content and can be correlated
to the citrate synthase activity.
fluorescent flavoproteins on reduction-oxidation of the
mitochondrial NAD-system allows to monitor the quotients
of both signals as very sensitive indicators of changes in
mitochondrial function, previously applied to the surface of
frozen tissue slices [33]. In Fig. 2 microscopic fluorescence
ratio images (flavoprotein image - 436 nm excitation, 525 nm
emission)/(NAD(P)H image 366 nm excitation, 450 nm
emission, of two single human saponin-skinned muscle fibers
are shown. The fiber bundle was fixed at both ends and
incubated in the medium for oxygraphic measurements. The
fluorescence images were acquired using an Olympus IX70
inverse microscope equiped with a Kappa CCD-camera CF
811 DXC. For the calculation of the digital ratio image
(flavoprotein fluorescence I NAD(P)H fluorescence) the
LSM software (Carl Zeiss) was used In the endogenous
oxidized state of the fiber bundle we obtained a high signal
ratio, leading to a bright ratio image (A, compare the
respective levels of fluorescence signals in Fig. 1). Addition
of 1 mM octanoylcarnitine and 5mM malate caused a
decrease in intensity of the ratio image due to a quenching
of flavoprotein fluorescence and increase of NAD(P)H
fluorescence (B, compare the levels of fluorescence signals
in Fig. 1). The uniform distribution of the fluorescence
changes is an indication for the complete accessibility of the
entire mitochondrial population in both single fibers for
substrates. Addition of 1 mM ADP caused thereafter again a
bright ratio image (C, compare the respective levels of
fluorescence signals in Fig. 1). This is again in line with a
complete, rather uniform accessibility of all mitochondria for
external added substances within these two fibers. The
described diffusion limitations for adenine nucleotides in
skinned fibers (see below) originate therefore clearly not from
diffusion problems in the fiber lattice, but are closely related
to the mitochondrial outer membrane [23-25]. The final
addition of 4 mM potassium cyanide and 10 mM glutamate to
the fiber bundle resulted in complete quenching of flavoprotein
fluorescence and maximal NAD(P)H fluorescence causing a
dark ratio image (D, compare the respective fluorescence
levels in Fig. 1). Therefore, ratio imaging of flavoprotein
and NAD(P)H fluorescence allow to obtain a mapping of
 87

A:endogen B: 1 mM OC + 5 mM MAL

c: 1 mM AOP 0:4 mM KCN

Fig. 2. Digital fluorescence ratio (Fp/PN) images of saponin-skinned human skeletal muscle fibers. A bundle consisting of two single fibers was fixed at
both ends in a chamber containing 300 III medium for oxygraphic measurements. Additions as in Fig. 1. The individual digital images were aquired using an
Olympus IX 70 microscope equipped with a Kappa CCD-camera CF 811 DXC. Conditions of fluorescence measurements: NAD(P)H fluorescence (PN) -
366 nm excitation and 450 nm emission, flavoprotein fluorescence (Fp) - 436 nm excitation and 525 nm emission. The digital ratio images were calculated
using the Zeiss LSM software. Magnification: 100 fold.

mitochondrial function on the level of a single muscle fiber
and can be used for the investigation of possible functional
heterogeneity. Summarizing, the obtained data provide
strong evidence for the complete functional intactness and
accessibility for external substrates of the entire mito-
chondrial population in saponin-skinned muscle fibers.
2. Permeabilized cardiomyocytes after myosin extraction -
'phantoms'
One of striking observations we have made is that isolated
cardiomyocytes when extracted for 30 min with 800 mM KCI
to remove myosin, completely retain their structure. This
is seen in Fig. 3 which shows fluorescence and immuno-
fluorescence confocal microscopic images of these 'phantom'
cardiomyocytes in which mitochondria are located in almost
parallel rows, as in intact cells (Fig. 3A). Their position also
shows, very interestingly, some pattern of striation, and when
immunofluorescence images of mitochondria and desmin are
superimposed, striations are identified at Z-line level due to
desmin (Fig. 3C). Indeed, anti-desmin antibodies stain the
cytoskeleton mostly at the level of Z-line (Fig. 3B). Thus,
cytoskeleton represents a rather stable structures ofthe cell-
it perfectly survives not only permeabilization but also rather
agressive treatment with the solution with very high ionic
strength. Figure 3 shows also that the cytoskeletal system
fixes mitochondria in rather exact position in the cells in vivo.
Our earlier electron microscopic studies showed good
preservation of both inner and outer mitochondrial membranes
88
Fig. 3. Double labelling immunofluorescence confocal microscopy of
phantom cardiomyocytes. (A) The labelling of mitochondria by nonyl
orange (see ' Materials and methods'). The rows of mitochondria (green
color) and striated pattern of their position are seen. Bars represent 4pm
(B) The labelling of desmin in phantom cardiomyocytes. Phantom cells
were incubated with anti-desmin antiserum (dilution 100 times), then
incubated with rhodamine-conjugated F( ab ')2 secondary antibodies. Bright
red color corresponds to anti-desmin antibodies. Bars represent 4 ~m. (C)
superposition of fluorescence of nonyl orange bound to mitochondria
(green) and fluorescence of rhodhamine-labelled anti-desmin antibodies
(red). We see that mitochondria are always between the Z-lines whereby
des min labelling is clearly seen.
under those conditions [10). For the convenience of the
readers, the transmission electron microscopic image of
phantom cell in skinned fibers is reproduced in Fig. 4. One
Fig. 4. Transmission electron microscopic image of the cardiac cell in
'phantom' fibers. Magnifiction 54 000 x . R eproduced from. Ref.lO with
pemmission.
can see the perfect preservation of the mitochondrial inner
and outer membranes in these phantom cells. In these cells
mitochondrial respiratory activity is completely intact [10]
and characterized by very high apparent Km for ADP in
regulation of oxidative phosphorylation (see below and ref.
10). Thus, both mitochondrial structure and function is
intact in the skinned fibers; moreover, the mitochondrial
position in the cells is preserved due to intact interaction
with cytoskeletal elements. This is in very good accordance
with multiple earlier studies [1, 2, 9- 14, 19-26] and
completely opposite to the proposal of Kushmerick group
of vesicularization of mitochondria into submitochondrial
particles under these conditions [8] .
Since anti-desmin antibodies stain all Z-lines in these
cardiomyocytes, from which the myosin was already extracted,
there should be no diffusion problem even for big immuno-
globulin molecules. And since the mitochondrial respiratory
activities in these phantom cells were found to be identical
to those in the skinned fibers or permeabilized cells before
myosin extraction [1], they can be considered as the simplest
and most suitable experimental system for studies of
mitochondrial systems in vivo.
Given below are multiple functional tests for mito-
chondrial respiratory activities in skinned fibers or per-
meabilized cells. Application of these methods allows us to
observe new properties of mitochondria which cannot be
observed in vitro.
3. Respiratory characteristics of mitochondria in
permeabilized cells
a) Useful functional characteristics of mitochondria in the
cells in vivo
Our experience has shown that there are very simple tests of
the functional state of the mitochondrial systems in vivo

which may have a very useful application both in fundamental
research and in clinical biochemistry for diagnostic purposes.
These tests are the following.
Cytochrome c test. The cytochrome c test is used to investigate
the state of the outer mitochondrial membrane before and
after any treatment of the tissue, including isolated mito-
chondria. In the KCl medium cytochrome c dissociates from
the outer surface of the inner mitochondrial membrane and
if the outer mitochondrial membrane is damaged, it leaves
mitochondrial intermembrane space, thus decreasing the
rate of respiration [10]. However, addition of the exogenous
cytochrome c completely restores respiration under these
conditions. Illustration of the use of this method is given in
the Fig. 5A which shows the recordings of the respiration
of the skinned cardiac fibers prepared from the intact heart
and from isolated perfused heart after 1 h of normothermic
ischemia.
Since KCI in high concentration dissociates also creatine
kinase from mitochondrial inner membrane surface, it is not
advised to use this solution to study the effects of creatine
on respiration.
Creatine test. This test is to investigate the intactness of
coupling between mitochondrial creatine kinase and adenine
nucleotide translocase [27]. Since this coupling is based on
metabolic channelling of adenine nucleotides, any changes in
membrane morphology, especially swelling of mitochondria
and dissociation of creatine kinase by increased inorganic
phosphate from the membrane may eliminate creatine
stimulation of respiration and decrease the efficiency of
control of respiration by the creatine kinase reaction. Fig. 5B
gives examples of this test. In fibers from the intact heart,
addition of creatine in the presence of submaximal ADP
concentration (0.1 mM) further activates respiration by
activating production of ADP and phosphocreatine from
mitochondrial ATP. In fibers from ischemic heart, when due
to mitochondrial swelling the creatine kinase dissociates from
the membrane, creatine stimulation is lost. This is one ofthe
most sensitive indicators of the ischemic damage ofthe heart
(see the Chapter by Rossi et al. in this volume).
Kinetics of ADP regulation of respiration. Due to specific
properties of mitochondria in vivo, very valuable information
may be obtained if the kinetics of respiration regulation by
ADP in the absence and presence of creatine is recorded, as
shown in Fig. 5C. This Figure shows rapid method of
determination of the affinity of mitochondrial oxidative
phosphorylation for ADP. However, in most important cases
it is better to analyze the initial rates of respiration stimulated
by ADP in different concentrations [11], to avoid the non-
linearities ofrespiration sometimes observed. This investigation
gives simultaneously information of the state of the outer
89
mitochondrial membrane and coupling between adenine
nucleotide translocase and creatine kinase in mitochondria
(see next section).
Activities of the segments of the respiratory chain or ATP
synthase. The use of specific substrates and inhibitors allows
to determine easily the activities of various segments of the
mitochondrial respiratory chain in skinned fibers [14, 18].
Also precise determination of the ATP synthase activity in
vivo [19] is very often of the interest.
b) Properties of mitochondria in the cells in vivo which
are lost during their isolation:
tissue specificity ofregulation ofrespiration by ADP Isolated
mitochondria in vitro, independently from tissues used for
preparation, are considered to be characterised by very high
affinity for ADP simply because of conserved nature of adenine
nucleotide translocase binding site and high permeability of
the outer mitochondrial membranes with respect to molecules
with the mass below 10 kD. The experimentally determined
value of the apparent Km for ADP is in the range of 10-20
~M with isolated mitochondria [23].
However, in the skinned fibers in which mitochondria are
located inside almost intact cellular structures, a new
phenomenon unknown before is observed. It becomes
evident that the affinity of mitochondria for ADP is tissue
specific - in oxidative tissues such as heart, m. soleus (also
in permeabilized hepatocytes and brain homogenates)
apparent Km for ADP in regulation of oxidative phos-
phorylation is surprisingly high, exceeding that for mito-
chondria in vitro by more than order of magnitude [7, 8,
19--24]. This is illustrated in Fig. 6. For skinned fibers from
heart or soleus skeletal muscle the apparent Km value for
ADP is in the range of 300-500 ~M (Fig. 6A,B). It is of the
same order for phantom cardiomyocytes. For fast twitch
skeletal muscles, gastroenemius white, red and mixed (Gw,
Gr, Gm, respectively) its value is 10-15 ~, fitting well
with that for isolated mitochondria (Fig. 6B).
'respiratory control' by proteolytic enzymes. Another very
interesting new phenomenon observed in skinned fibers with
low mitochondrial affinity for ADP is that this affinity can
be changed by short time treatment of fibers with trypsin (Fig.
7). Figure 7A shows that short treatment of skinned fibers with
trypsin strongly changes the dependence of the respiration rate
on the ADP concentration, decreasing the apparent Km to
about 40 ~M (Fig. 7B). If we fix the ADP concentration at
submaximallevel without trypsin treatment, let us say at 0.1
mM, addition of trypsin (or other proteolytic agents) rapidly
increases the rate of respiration due to changes in kinetics
of ADP regulation of respiration. This then leads to a
phenomenon of 'respiratory control' by trypsin shown in Fig.
90

A Test for Intactness B Effect of creatine

of outer membrane
F F

t t

~~~a.. L
1 min

(a) (b) (c) (d)

Intact membrane Damaged membrane Stimulating effect No effect
Control Ischemia Control Ischemia

CP C ATP
ADP ...

C Kinetics

Fibers Fibers
Fibers Fibers
I ADP:

t
ADP:
\0.05 \
t
0.0125
0.025
~
ADP:
,\0.05 \
,,
~ ADP:
0.025
I
"J t 10.05 \
h10.05
\t 0.1
\~
t

25 ng at. O/ml

1 min

(a) (b) (c) (d)

Creatine + Craatlne Creatine + Creatine

Control Ischemia

Fig. 5. (A) Recording of a cytocrome c test for intactness ofthe outer mitochondrial membrane. To the fibers ADP was added to final concentration of I mM
for maximal activation of respiration, then cytochrome c was added to final concentration of 8 11M. The changes in the oxygen concentration are recorded by
oxygraphy. Control- fibers isolated from heart in situ; ischemia - the heart before isolation of fibers was kept for I h under normothermic total ischemia. (B)
Recording of the creatine test for the functional state of coupled mitochondrial creatine kinase reaction in skinned fibers. ADP was added in concentration
of 0.1 mM, then creatine in concentration of 20 mM. (C) Recording of kinetics of respiration regulation by ADP in skinned fibers from heart. Creatine
concentration was 20 mM. For control and ischemia see 5A.

25
~
'e:
'E
20
Heart
'~ 15
o Gastrocnemius Mixed
E
.9
co, 10
OJ
e:
->
5
E......----'--_-----.l_
0,1 0,2 0,3'
L--/~
1,0
,---~~~--,--~~..c---''--Lt~~~~------,--------~-
- 150
[ADP], mM
Fig. 6. (A) The dependence of the respiration rate of skinned fibers from heart, soleus and gastrocnemius white (w), red (r) and mixed (m) skeletal muscles
on the external ADP concentration in the medium. Note that maximal rates of respiration are rapidly achieved for gastrocnemius but not for soleus and heart
muscles. (B) Analysis of the data from Fig. 6A in double reciprocal plots.
7C. In details, this phenomenon is described in our recent
publication [34].
All these phenomena are summarized in the Fig. 7D, which
demonstrates the phenomenological relationship between the
apparent Km for ADP in skinned fibers and the Vmax' the
maximal respiration rate. Very clearly, in the fibers with high
oxidative activity - Vmax exceeding 10 ng-atoms of oxygen
consumed per min per mg of dry weight - the apparent Km
for ADP is very high, around 300-400 11M, and in all fibers
with low oxidative activity the apparent Km for ADP in vivo
is not different from that in vitro, both being between lO-
20 11M. In cardiomyocytes and soleus, the affinity of
mitochondria to ADP can be increased - apparent Km for
ADP significantly decreased by disrupting the mitochondrial
outer membrane by a swelling procedure [l 0, 25]. And since
this affinity is also increased several times by treatment of
fibers by trypsin (see Fig. 7D), the observed high apparent
Km for ADP is obviously related to expression of some
specific cytoplasmic, extramitochondrial protein(s) which
control( s) the permeability of the porin pores in mitochondrial
outer membrane (VDAC) for ADP in the cells of tissues with
high oxidative activity [10, 11,23-27]. Indeed, when ADP
is produced in coupled mitochondrial creatine kinase reaction
in intermembrane space in the presence of creatine, the
affinity toADP is again increased and concomitantly the Km
for ADP is decreased (Fig. 7D) due to the increased turnover
of adenine nucleotides in the coupled reactions [10, 11,27].
91
1,0
0
Heart
Soleus
Gastrocnemius White
0 Gastrocnemius Red
< 0,5
-75 o 75
~-~---~- .. - - ' -
150
1/[ADP]
This most probable explanation of the multiple experimental
data by control ofthe permeability of the porin pores in the
mitochondrial outer membranes in vivo by some still
unknown proteins [23, 24] is schematically illustrated in
Fig. 8A and B. The conclusion that porin pores, or channels,
in the mitochondrial outer membrane limit the permeability
for ADP in the cells in vivo was made on the basis of the
results of hypo-osmotic swelling resulting in rupture of the
outer mitochondrial membrane and drastic decrease in
apparent Km for ADP [10,23, 24]. The validity of this
conclusion is confirmed in the laboratory ofRigoulet (see the
chapter by Averet et al. in this volume) by showing that while
wild-type permeabilized yeast cells also show very high
apparent Km for ADP in regulation of respiration, that
exceeding this parameter value for isolated mitochondria 30
times, in porin-deficient mutant yeast cells there is no
difference between these mitochondrial parameters in vivo
and in vitro. Since the phantom cardiomyocytes containing
only mitochondria and cytoskeleton show the same high
value of Km for ADP as all permeabilized cells, it was
concluded that the protein factor controlling mitochondrial
properties in vivo is associated with cytoskeleton. Multiple
experimental data on mitochondrial-cytoskeleton contacts are
reviewed in this volume by Rappaport and Samuel (see
Chapter 7). The most interesting problem now is to find out
which protein controls the affinity of mitochondria for ADP
in the cells of tissues with high oxidative activity.
92

15 Skinned Fibers + 0,1 mM ADP

A
~
,
c c
E 500
5 jJM Chymotrypsin ( T =0 )
'";"0>
E
10
1
0 E
E o
E 400
.....
0
co
5 o Soleus o
I roOJ
0>
C
Soleus + Trypsin c

->
0
0,5 1,0 1,5 300

[ADP], mM
0 5 10 15
0,8 T, min

400 D
Soleus

Diaphragm I

300 Peroneals
IHeart

~
:J..
+ Cr, or
~
~
E 200 +proteinases

-20 -10 0 10 20 30 40 Soleus + Cr IHeart + EL

100 Gastr. mix TI
1/ [ADP] I
Gastr. ~ Soleus + KT Heart + Cr -Heart + TR
TAr\d) -./ Quadriceps
Gastr. white
Plantaris
0
10 20 30
V m' ng-atom 0 . mg -1. min-1

Fig. 7. The effect of the treatment of skinned fibers from rat soleus muscles with trypsin on kinetics of regulation of respiration by ADP. Trypsin was used
in concentration of 1251lg/ml for 15 min. (A) Changes in respiration rates as function of ADP concentration. (B) Treatment of data from Fig.7A in double
reciprocal plots. (C) Oxygraph recordings of the repiration of soleus skinned fibers at 0.1 mM ADP: activation of the respiration by chymotrypsin ('respiratory
contro!'). (D) Tissue specificity ofthe affinity ofthe mitochondrial respiratory system for ADP. The affinities are low (apparent Km high) in heart, m. soleus,
diaphragm, m. peroneals and increased (apparent Km decreased) by treatment with trypsin (TR), elastase (EL), chymotrypsin (KT) or by adding creatine. In
all fibers from fast twitch skeletal muscles (gastrocnemius white, mixed, red, quadriceps, tibialis anterior and plantaris) the mitochondrial affmity for ADP
as measured by apparent Km for this substrate is very high (apparent Km very low) and not sensiyive to trypsin or creatine additions. The respiration rates
were determined at 21C.

b) Practical aspects of the determination of respiratory
ADP - control parameters
A typical difficulty in the evaluation ofparameters of the respira-
tory control by ADP in skinned fibers arises from the necessity
to decide whether the studied sample contains a single, uniform
population of mitochondria or a mixture of (two) populations
with different Km values. The latter situation could first of all
arise from improper handling of the sample (e.g., mechanical
damage of bundles of fibers), but it also could resemble to actual
heterogeneity of the muscle samples as well. This situation will
be analyzed below from a solely pragmatic standpoint that may
help to avoid pitfalls in the handling of the data.

Cr
ADP
AI
~~~-ADP
Cr
81
Cr
Fig. 8. (A) Hypothetical connection between mitochondria and a
cytoplasmic protein probably associated to cytoskeleton, controlling the
permeability ofthe mitochondrial outer membrane for adenine nucleotides.
ANT - adenine nucleotide translocator, CK mito - mitochondrial isoenzyme
of creatine kinase (see the chapter by Schlattneret al. in this volume). This
protein controls the permeability of porin pores for ADP (and ATP) but
not for creatine or phoshocreatine. (8) Isolated mitochondria - the porin
pores are open for all substrates including adenine nucleotides since during
isolation the unknown protein, factor x, is separated. Reproduced from ref.
23 with permission.
It is convenient to determine Km and Vm values by direct
fitting of the data to the hyperbolic relationship:
Vm [ADP]
v=v lotal - v0 = (1)
Km + [ADP]
by the non-linear least squares' fit. When two populations of
mitochondria are present in the sample, we have:
93
Vml [ADP] Vm2 [ADP]
v= + ------ (2)
Kml + [ADP] Km2 + [ADP]
Unfortunately, very often superposition of two hyperbolae
with different Km values is surprisingly 'smooth' even at
rather different Km values. Fig. 9A shows that deviations of
the data calculated according to Eqn. 2for equal amounts of
two populations of mitochondria with Km values 300 ~M and
30 ~M from the curve determined according to Eqn. I did
not exceed few percents (4%, see inset) of the Vm value. The
obtained 'average' parameters, Km(av) = 68.9 4.2 and
Vm(av) = 0.901 0.016 (10---1000 ~M ADP) or Km(av) 74.0
4.7 ~M and Vm(av) = 0 .926 0.015 (10---2000 ~M ADP)
are determined with good accuracy and apparently showed
a good fit ofthe data to the used equation. However, the value
ofKm(av) remained between Kml and Km2 values that have
been actually introduced, Vm(av) was lower than the expected
value Vm = 1.0 and the double reciprocal plot revealed
systematic deviations of the data from the calculated straight
line. Thus, the presence of two populations of mitochondria
in the sample cannot be discovered from the rate vs. [ADP]
plot and one should attempt to create a more sophisticated
fitting programs that work in conditions of a limited amount
of scattered experimental data or return to the double
reciprocal plot (Fig. 9B) taking care that the used ADP
concentration range and estimated Km values does not fall
into contradiction. There are, however, the cases when
differences in the apparent Km for different mitochondrial
populations are sufficiently significant to observe them even
in the rate versus ADP concentration plot. .
'Average' Km. Nevertheless, Km(av) determined from the
averaged experimental data by using Eqn.1 could be a useful
parameter from practical viewpoint. Fig. 10 shows averaged
Km(av) as well as Vm(av) values plotted against the fraction
of the low-Km population of mitochondria in the sample
described in Fig. 9. It is evident from Fig. 10 that the presence
of even 10% of the low-Km fraction of mitochondria in the
sample results in a pronounced decrease in Km(av), e.g. , to
21 0 ~M for the set of data calculated for Kml = 300 ~M and
Km2 = 30 ~M. Thus, Km(av) is a sensitive parameter for the
quality of the fibre samples and even can be used to calculate
the ratio of different mitochondrial population provided that
we know the extreme values ofKm. The situation in Fig. 10,
for example, mimics changes in Km(av) in rat heart skinned
fibres after hypoosmotic treatment. Besides of the well-
known high-Km population (315 48 ~M) a new population
of mitochondria with the decreased Km value (32.5 5.0
~M) appeared in the sample in this case (10) and Km(av)
decreased to 70---80 ~M. The latter value corresponds to the
presence of 42% of the low-Km population of mitochondria
in good agreement with the results ofthe cytochrome c test on
94

1.00 l' 8,
I
0, 90 ~ -------~--------
e
---<r-
0.80
1i _~___----ii-- 6
!
/--'.
0.701
lIv
o.60 ~

i 1Q.{. 'I'
I 4.00~

0.30
! i
i.
i~
1-;
\,g 0.00
r-t"1""--~---c;.,---------- ~_ _1~_ _~_ _ _ ~ _ _~_ _ _~_ _~

!';;; J9
I..
jiX_4 . 00 j
:..
., -0.05 o 0.05 0.10
~
0.1Q { I
0.00
I
0.40
I
0.30 1.20
I i i
L60 2,(}O 10 3 1
j
lI[ADP], J1M- 1

1 [flDP]. pM
o.00 -,---.:---.------:---.------,----,---.--------.,------,----.---,------.------7- :
1

1),00 0.40 0.80 1.20 1.60 2.00 10 "31

[AvPJ, pH

Fig. 9. (A) Calculated (modeled) dependencies of the respiration rates on the ADP concentration in the range of 10--1000 11M according to Eqn. 2 for Kml
= 300 11M, Vml = 0.5; Km2 = 30 11M, Vm2 = 0.5 and fitted according to Eqn. 1. Inset shows the deviation of the data from the calculated single hyperbola.
(B) Double reciprocal plot of the data from Fig. 7A. Two populations of mitochondria are clearly seen, as expected from input data.

400

0
0 0 0 ~ 0
\;l ~ ~
1 absence ofthe data at low ADP concentrations) may lead to
misleading conclusions, when the second population of
mitochondria may not be detected.
300 t .....
'"
0.75 .....
::E::::l..
0
Vm =
::l
Determination ofVm. It seems valuable to determine the dry-
weight-related Vm values (Vm(DW) by correlating the
Q
~
e 200 0 0.5 b
ro
.....
obtained Vm with the determined dry weight values. This
approach is especially useful for the determination of the
~
0
E
Vm(DW) for fast twitch muscles that contain low amounts
of respiring mitochondria. Fig. 11 shows that an intuitive
/. >
~~~~~
100 0 0.25 attempt to use large fibre samples (more than about 6-7 mg
Km dry weight) in order to increase the rate of oxygen consumption
could result in a pronounced deviation of the data and
underestimation of the Vm(DW) value.
o 20 40 60 80 100
Acceptor control index (A Cl) as the quality control. Two types
% of low Km population
of acceptor control indices of respiration can be used. Most
usual is the acceptor control by ADP, also called respiratory
Fig. 10. Averaged Km (av) and Vm (av) values calculated from single control index, ReI, which was originally defined as the ratio
hyperbola as shown in Fig. 7 and plotted against the fraction oflow - Km of the respiration rates in State 3 and State 4. Due to high
population of mitochondria in the sample. For both populations Vm" was
MgATPase activity offibers, ReI is taken to correspond to the
taken to be equal to unity. Open symbols - the low Km is 40 11M; closed
symbols - the low Km is 30 11M. ratio of(vo + Vm)/vo = 1 + Vrnlvo where Vm is the maximal
respiration rate at saturating ADP concentrations. In the present
context it could be determined from the Vm vs. v0 plot (Fig. 12).
the amount of mitochondria with damaged outer membrane in For rat heart skinned fibres ReI = 6.4 (Fig. 12). The same ReI
the swelled fibres [25]. value seems to hold for a number of slow and fast twitch
It could also be noted that improperly selected range of skeletal muscles regardless of their Vm(DW) values. Generality
ADP concentrations (too small concentration range or the of this conclusion might be the subject for discussions;
 95

In Table 1, the typical values of the maximal respiration

30 1
rates in skinned fibers from different types of muscles are

=
...... I given .

.s
. . . 20 1.
0
The second useful index is the acceptor control exerted by
creatine due to the coupled creatine kinase reaction in
N I
mitochondria (see above), expressed as CrI,
0
a
s
=a 101 I

This is the relative activation of respiration by 20 mM creatine

> I
concentration in the presence of 0.1 mM ADP, see Fig. 5B.
In intact cardiac fibers it is 0.6--D.65.

o 2 4 6 8 10 12
Dry weight, mg 3. Clinical application of the skinned fiber technique

Fig. 11. Detennination ofthe maximal rate of respiration of skinned fibers a) Analysis of a biopsy sample of m interosseus taken from
from rat gastrocnemius white muscle. It is dangerous to use too big fibers the left hand of V Saks, an author, during emergency
because of the linearity problems which may be related to the reaggregation operation
of fibers in the oxygraphic cells.
Rather good illustration of the amount of biochemical
information which can be obtained rapidly by skinned fiber
60 - technique is given in Fig. 13. On February 14, 1997 one of
o Gastrocnemius Red the authors of this chapter V. Saks was operated by Prof. F.
<) Gastrocnemius White Moutet in Grenoble on the left hand to repair the finger
o Gastrocnemius Mixed
Peroneals broken in small car accident. During the operation, biopsy
sample was taken from slow type m. interosseus of his left
I Soleus hand, permeabilized by saponin and analyzed in the Laboratory

s 40 ~ of Bioenergetics of Joseph Fourier University by another

i::

.......
N /*I/ author, T. Tiivel, for determination of the kinetics of respiration
regulation by ADP. Two populations of mitochondria are
0
"0 ./ clearly seen in Fig. 13. This is the case when two kinetic phases
S of respiration regulation (two population of mitochondria) are

i::
clearly seen even in primary analysis of data, in V 02 vs. ADP
E
20 ~.
concentration plot. And very clearly two populations of
> # mitochondria with very different properties are revealed in
double reciprocal plots. One population of mitochondria is
characterized by very high apparent Km for ADP, 748 ~M,
~CO which is decreased by creatine to 400 ~M (this rather small
effect of creatine may be explained by some intraoperational
I / ischemia), and the second population is characterized by very
~ ~-.-- L J
low apparent Km for ADP, equal to 18 ~M. The difference
in the apparent Km for ADP exceeding factor 40 explains
0 2 4 6 8 why in this case two populations of mitochondria are easily
Vo, nmol 02 / min seen. These two populations of mitochondria, most probably,
represent different types of fibers in this biopsy sample. The
Fig. 12. Detennination of the respiratory control index (ReI) for skinned total maximal rate of respiration is characteristic for tissues
fibers from rat.
with high respiratory activity. Thus, this singe biopsy sample
illustrates two important things: in the human muscle tissue,
nevertheless, to our opinion, every deviation from this value regulation of the mitochondrial respiration seems to be
for the muscle not studied before deserves special attention. specific for fiber types, and in muscles with high respiratory
Reduced RCI value typically points to improper prepara- activity the factor 'x' - a protein of still unknown nature [23,
tion of the fibres, damage of the fibres due to the performed 24], strongly controlling the permeability ofthe mitochondrial
additional operations or due to prolonged storing. outer membrane for ADP is present.
96

Table 1. Specific mitochondrial respiratory activities of saponin-skinned muscle fibers compared to isolated mitochondria

Species Muscle type Isolated Skinned fibers

Mitochondria,
Vmw<' nmol 0,1 Vmo<' nmol 0,1 cytochrome aa" Vmo<' nmol 0,1
min/nmol aa, min/mgdwt nmol/gwwt min/nmol aa,

human vastus lateralis 298 55 8.6 1.9 5.6 1.8 260 57

mice quadriceps 343 62* 11.3 35 5.6 0.4 341 105
mice heart n.d. 33.9 0.7 44 9 131 26
rat heart 190 20 - 25 22.5 2 152 - 187
rat soleus 247 32* 18.2 0.7 12.4 0.5 224 10
rat quadriceps 247 32* 79 0.7 5.2 0.6 230 8

*The mitochondria were isolated from total skeletal muscle. All respiratory activities were determined in the presence of 10 mM glutamate, 5 mM malate and
I mM ADP. A wet weight/dry weight ratio of6.2 0.3 (25 determinations) was used for the calculation of the specific rates

b) cardiomyopathy - analysis of biopsy samples taken from

the endocardium ofpatients during angiography
.... One of the most effective use ofthe skinned fiber technique
l:: in clinical medicine has been an analysis of biopsy material
'E taken from the endocardium of the hearts ofthe patients with
o
~c:
1
A dilated cardiomyopathy during angiographic observations
[35,35]. Figure 14 reproduces these results. The parameter
which changes very significantly in these fibers is creatine -
~ stimulated respiration: in comparison with control, it decreases
about 3 times in functional class III due to diminished
o 1000 2000 expression of the mitochondia; isoform of creatine kinase,
at
:g 0,5
[ADP], jJM
and its decrease correlates well with decrease in ejection
-c: fraction [36].

~ 0,3 c) acute ischemic damage and problem of heart

~ preservation
~ 0,1
E In ischemic heart most rapid cellular change is mitochondrial
,g8 -01' swelling, and this may result in disrupter of the outer
mitochondrial membrane or dissociation of creatine kinase
> [ADP], jJM
'0 -0,3 from the inner mitochondrial membrane. This is why creatine
c: stimulated respiration in skinned fibers is a very sensitive
o
til -0,5 2,0
indicator of the extent of acute ischemic damage and

~ correspondingly of the efficiency of the cardioplegic

protection of the heart in cardiac surgery and preservation.
In more details, these results are considered in the chapter by
Rossi et al. in this volume.

d) detection of mitochondrial myopathies

Dysfunction of mitochondria is found in a rapidly increasing
number of neuromuscular diseases. Muscle tissue has
remained to be the tissue of frst choice in the diagnosis of
mitochondrial cytopathies since mitochondrial defects are
-0,06 -0,04 -0,02 0,00 0,02
most frequently expressed in muscle. It has been demonstrated
1 I [ADP], jJM-1 that mtDNA heteroplasmy can occur in man and that these
diseases may be associated with defects of the mitochondrial
Fig. 13. Respiratory analysis of the biopsy sample of the m. interdossseous
taken from the left hand of one of the authors, V. Saks. (A) Respiration
genome due to either deletions or point mutations [30]. In
rate as a function of ADP concentration; (B) Deviations of the experimental functional analysis of oxidative phosphorylation capacity, the
values from the hyperbola; (C) double reciprocal plot of data from 13 A. measurement of maximal rates of mitochondrial respiration
 97

HAMSTER HEART HUMAN HEART

control

myopathy

FC I
Skinned fibers
I
, ,
D.1mM AOP
20mM Creatine

l.OmM AOP

FC ill
I

Skinned fiber~ I
35 }JMCAT
0.1mMAOpl
20mM Creatine I
50 ng-at O2 / ml

1
t-- 5 min ---t 35}JM CAT

Vrrox ngatansO/min mg a.w. N5 Yroox/Yo Yr:r, ~

"r

FC I FC " FC m FC n FCIFcnFcm FCU FC I FC I FC m FC n

Do1ated CX70aTly~y Myocaot OITated CX7diaTlYopalhy Myocadit Dilated CX7~ Myocaot

Fig. 14. Analysis ofthe creatine kinase isoenzyme profiles and respiration in biopsy samples of normal and cardiomyopathic patients taken during angiographic
observation. In the upper part, creatine kinase izoenzyme profiles are shown for both intact (control) and cardiomyopathic hamster and human hearts. FC I
and FC III show the respiration of skinned fibers, taken during angiography from the hearts of patients suffering from dilated cardiomyopthy, functional
class I and III. The respiratory activities are statistically analyzed and shown for different functional classes of heart failure in the lower part of Figure. Most
remarkable changes are seen in creatine - stimulated respiration. Reproduced from ref. 36 with permission.
98

is often used for the determination of the different mito-
chondrial defects. It has been shown that assessment of
respiratory activities and responses of permeabilized fibers
can be especially applicable for the study of human muscle
mitochondria due to the obviation of mitochondrial isolation
and minimization of the amount of tissue (between 20 and
50 mg wet weight) needed for this assessment. Similar to
cardiac muscle fibers (see above), the human skinned skeletal
muscle fibers behave like isolated mitochondria: - the
respiration of mitochondrial substrates can be stimulated by
ADP with high ReI value, inhibited by carboxyatractyloside
[29] and it is possible to detect fluorescence changes of
NAD(P)H and fluorescent flavoproteins on additions of
substrate, ADP and cyanide [30, 31]. As previously reported,
a comparison of rates of respiration per cytochrome aa 3
content of isolated human muscle mitochondria and saponin-
skinned muscle fibers revealed that more than 85% of mito-
chondria in those fibers are accessible for the investigation of
oxidative phosphorylation [29]. We applied this method to
study the functional changes of mitochondria of patients with
mitochondrial myopathies (chronic progressive external
ophthalmoplegia, Morbus Leigh and mtDNA depletion). The
values of maximal rates of respiration ofthese fibers and the
corresponding citrate synthase activities in the skeletal
muscle biopsy are shown in Table 2. As seen from the Table,
in many cases an adaptational increase in the mitochondrial
content (citrate synthase) in response to the mitochondrial
defect can be observed. On the other hand, the rates of
respiration of saponin-skinned fibers (expressed per mg dry
weight) are lower. Therefore, the determination of maximal
rates of respiration of skinned fibers is a useful method for
screening of biopsy samples of patients suspected for mito-
chondrial defects. However, not in all cases the differences in
maximal rates of respiration are statistically significant. In
order to visualize small defects and to localize enzymatic
deficiencies using skinned fiber preparations titration
experiments applying specific inhibitors are especially
useful [37] In Fig. 15 an example of rotenone titration
experiments of the glutamate + malate supported maximal
respiration is shown. Skinned fibers from patients with
complex I deficiencies (D.S., K.G.) show a much higher
sensitivity to the specific inhibitor of this enzyme. The
quantitative description of these titration curves within the
framework of metabolic control analysis indicated elevated
flux control coefficients of the affected enzyme: for D.S. a
flux control coefficient of 0.54 0.02 and for K.G. of 0.46
0.01 were determined (control value- 0.16 0.05, n = 11)
(see also the related paper by Letellier et al. in this volume).
Conclusions
From all the results described in this chapter, the following
conclusions can be made. (1) In permeabilized cells and

Table 2. Maximal mitochondrial respiratory activities of saponin skinned human muscle fbers of patients with mitochondrial myopathies

mtDNA defect Citrate synthose Succ + rot Glu+ mal

controls 8.7 3.8 9.2 1.9 8.6 1.9

(N= 84)
S.B. 67% of 11.9 kb 11.40.1 5.3 6.6 0.1
(CPEO) L'>mtDNA (n = 2)
M.K. missing Apa J 12.0 0.3 6.1 0.2 5.3
(CPEO) restriction site (n = 2)
M.T. 53% of 14.3 kb 125 0.2 8.3 2.1 6.2 0.8
(CPEO) L'>mtDNA (n = 4) (n = 3)
K.G. 84% of 12 kb 34.8 1.7 5.6 1 3.4 0.6
(CPEO) L'>mtDNA (n= 8) (n = 14)
T.w. 65% of 11.6 kb 11.0 0.1 9.3 1.1 7.9 1.7
(CPEO) L'>mtDNA (n = 6) (n = 10)
E.B. none (?) 13.2 0.3 8.9 1.5 6.1 0.8
(Morbus Leigh, (n = 12) (n = 10)
complex I
deficiency)
N.N. 24% of 9.10.3 6.3 0.9 5.2 0.5
(mtDNAD) wtmtDNA (n = 6) (n = 10)
D.S. none (?) 17.4 l.l 9.8 0.9 8.6 1.2
(complex I (n = 4) (n = 10)
deficiency)

The rates of respiration (expressed in nmol O/min/mg drsr weight, at 25C) were determined using the substrates 10 mM succinate + 10 11M rotenone (succ
+ rot) or 10 mM glutamate + 5 mM malate (glu + mal) in the presence of I mM ADP. The activities of citrate synthase in the biopsy are given in U/g wet
weight (at 30C). N - number of orthopedic patients, n - number of independent determinations. CPEO - chronic progressive external ophthalmoplegia,
mtDNA D - mitochondrial DNA depletion.
 99

+> 2. Endo M, Kitazawa T: E-C coupling studies in skinned cardiac fibers.

it 10 In: M Morad (ed). Biophysical Aspects of Cardiac Muscle, Academic,
'0
blJ New York, 1978 pp 307-327
S
..........
B 3. Glauert A, Dingle J, Lucy J: Action of saponin on biological cell
I=i membranes. Nature 196: 952-955, 1962
...... 4. Kom E: Cell membranes: Structure and synthesis. Annul Rev Biochem
S B 38:263-288,1969
..........
t\l 5. Comte J, Maisterrena B, Gautheron DC: Lipid composition and protein
0
...... 4 profiles of outer and inner membranes from pig heart mitochondria com-
0 parison with microsomes. Biochim Biophys Acta 419: 271-284, 1976
S 6. Saks VA, Kuznetsov AV, Kupriyanov VV, Micely MV, Jacobus WE:
I=i 2 Creatine kinase of rat heart mitochondria. The demonstration of
p.. functional coupling to oxidative phosphorylation in an inner membrane
rn - matrix preparation. J Bioi Chern 260: 7757-7764, 1985
Q)

>r-. 10 20 30 40 7. Bygrave FL, Lehninger AL: The affinity of mitochondrial oxidative

[rotenone], f..LM
Fig. 15. Rotenone titrations of ADP stimulated respiration of saponin-
skinned muscle fibers. Circles - orthopedic patient, 1.7 mg dry weight;
squares - patient D.S., 2.0 mg dry weight; triangles - patient K.G., 2.2 mg
dry weight of skinned fibers in the medium for respiration measurements
containing I mM ADP, 10 mM glutamate and 5 mM malate.
skinned fibers prepared with the use of saponin in low
concentration (50-100 Jlg/ml) mitochondria become accessible
for external substrates and maintain their structural and
functional intactness and connection to the intracellular
structures. (2) At the same time, new specifc properties of
mitochondria in vivo such as tissuedependence of the
mechanism of respiration regulation by ADP can be revealed
and studied. (3) Due to very small amount of tissue necessary
for these studies, the skinned fiber technique is a very useful
tool of clinical studies. (4) The limitation of th~method is
that the cytoplasm is lost during permeabilization procedure
and therefore, the precise mechanisms of feedback signal
transduction cannot be studied by this method.
Acknowledgements
This work was supported by grants of the Land Sachsen-
Anhalt (1795A/0084 and 1919A/0025) and by a grant of
Deutsche Forschungsgemeinschaft (436 RUS 171141/95),
Germany, by Estonian Science Foundation grants N 2635 and
2092 and by INTAS grant 94 - 4738. The authors cordially
thank Dr. Yves U sson, Grenoble, for assistance with confocal
microscopy.
References
I. Veksler VI, Kuznetsov AV, Sharov VG, Kapelko VI, Saks VA:
mitochondrial respiratory parameters in cardiac tissue: A novel method
of assessment by using saponin-skinned fibers. Biochim Biophys Acta
892: 191-196,1987
phosphorylation mechanism for phosphate and adenosine diphosphate.
Proc Nat! Acad Sci USA 57: 1409-1414, 1967
8. Wiseman RW, Jeneson JAL, Kushmerick MJ: Why is the sensitivity
of mitochondria to ADP over tenfold lower in permeabilized fibers
than in vivo? Biothermokinetics ofthe living cell. Biothermokinetics
Press, Amsterdam, 1996 pp 124-127
9. Altschuld RA, Wenger WC, Lamka KG, Kindig OR, Capen CC,
Mizuhira V, Vander Heide RS, Brierley GP: Structural and functional
properties of adult rat heart myocytes lysed with digitonin. J Bioi Chern
260: 14325-14334, 1985.
10. Saks VA, Vasil' eva E, Belikova YO, Kuzuetsov AV, Lyapina S, Petrova
L, Perov NA: Retarded diffusion ofADP in cardiomyocytes: Possible
role of mitochondrial outer membrane and creatine kinase in cellular
regulation of oxidative phosphorylation. Biochim BiophysActa 1144:
134-148, 1993
II. Saks VA, Behkova YO, Kuznetsov AV: In vivo regulation of
mitochondrial respiration in cardiomyocytes: Specific restriction for
intracellular diffusion ofADP. Biochim BiophysActa 1074: 302-311,
1991
12. Penman S: Rethinking cell structure. Proc Nat! Acad Sci USA 92:
5251-5257, 1995
13. Lin A, Krockmalnic G, Penman S: imaging cytoskeleton-mitochondrial
membrane attachments by embedment-free electron-microscopy of
saponin-extracted cells. Proc Nat! Acad Sci USA 87: 8565--8569, 1990
14. KunzWS, Kuzuetsov AV, Schulze W, Eichhorn K, Schild L, Striggow
F, Bohnensack R, NeuhofS, GrasshoffH, Neumann HW, Gellerich FN:
Functional characterization of mit ochondrial oxidative phosphorylation
in saponinskinned human muscle fibers. Biochim BiophysActa 1144:
46--53, 1993
15. Vignais PV: Molecular and physiological aspects of adenine nucleotide
transport in mitochondria. Biochim Biophys Acta 456: 1-38, 1976
16. MinayevaA, Ventura-Clapier R, Veksler V: Ca2+ uptake by sarcoplasmic
reticulum ATPase in situ strongly depends on bound creatine kinase.
Pflugers Arch - Eur J Physiol432: 904-912,1996
17. Livingston BE, Altschuld RA, Hohl CM: Metabolic compart-
mentalization in neonatal swine myocytes. Pediatric Research 40: 59-
65, 1990
18. Kunz WS, Kuzuetsov AV, Gellerich FN: Mitochondrial oxidative
phopshorylation in saponin-skinned human muscle fibers is stimulated
by caffeine. FEBS Lett 323: 188-190, 1993
19. Ouhabi R, Boue-Grabot M, Mazat J-P:ATP synthesis in permeabilized
cells: assessment ofthe ATP/O ratio in situ. In: E. Gnaiger, F.N.
Gelierich, M.Wyss (eds). Modem Trends in Biothermokinetics,
Innsbruck University Press, vol. 3 1994 pp 141-144
20. Vercesi AE, Bemardes CF, Hoffmann ME, Gadelha FR, Docampo R:
Digitonin permeabilization deos not affect mitochondrial function and
allows the determination of the mitochondrial membrane potential of
Trypanosoma cruzi in situ. J Bioi Chern 266: 14431-14434, 1991
100
21. Saks VA, Kapelko VI, Kupriyanov VV, Kuznetsov AV, Lakomkin VL,
Veksler VI, Sharov VG, Javadov SA, Seppet EK: Quantitative
evaluation of relationship between cardiac energy metabolism and post-
ischemic recovery of contractile function. J Mol Cell Cardiol 21:
(Suppl) 67-78,1989
22. Veksler V, Ventura-Clapier R: In situ study of myofibrils, mitochondria
and bound creatine kineses in experimental cardiomyopathies. Mol
Cell Biochem 133/134: 287-298,1994
23. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova YO, Kuzuetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kineses in in vivo regulation of cellular
respiration - a synthesis. In: V. Saks, R. Ventura-Clapier (eds) Cellular
bioenergetics: Role of coupled creatine kineses, Kluwer Academic
Publishers, Dordrecht-Boston-London, 1994 pp 155-192
24. Saks VA, Kuznetsov AV, Khuchua ZA, Vasilieva EV, Belikova YO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis: Possible involvement of mitochondrial
cytoskeletal interactions. J Mol Cell Cardiol 27: 625--645, 1995
25. Saks VA, Belikova YO, Vasileva EV, Kuzuetsov AV, Fontain E, Keriel
C, Leverve X: Correlation between degree of rupture of outer
mitochondrial membrane and change of kinetics of regulation of
respiration by ADP in permeabilized heart and liver cells. Biochem
Biophys Res Comm 208: 919-926, 1995
26. Veksler VI, Kuznetsov AV, Anflous K, Mateo P, van Deursen J, Be
Wieringa, Ventura-Clapier Z: Muscle creatine kinase deficient mice.
II Cardiac and skeletal muscles exhibit tissue-specific adaptation of
the mitochondrial function. J Bioi Chern 270: 19921-19929,1995
27. Saks VA, Ventura-Clapier R, Aliev MK: Metabolic control and 37
metabolic capacity: Two aspects of creatine kinase functioning in the
cells. Biochim Biophys Acta 1274: 81-88,1996
28. Kakol I, Borovikov YS, Szczesna D, Kirillova VP, Levitsky or:
Conformational changes of F - actin in myosin - free ghost single
fibre induced by either phosphorylated or dephosphorylated heavy
meromyosin. Biochinl Biophys Acta 913: 1-9,1987
29. Kunz WS, Kuznetsov AV, Winkler K, Gellerich FN, Nenhof S,
Neumann HW: Measurement of fluorescence changes ofNADH and
of fluorescent flavoproteins in saponin-skinned human skeletal muscle
fibers. Anal Biochem 216: 322-327, 1994
30. Winkler K, Kuznetsov AV, Lins H, Kirches E, Von Bossany P,
Dietzmann K, Frank B, Feistner H, Kunz WS: Laser-excited
fluorescence studies of mitochondrial function in saponin-skinned
skeletal muscle fibers of patients with chronic progressive external
ophthalmoplegia. Biochim Biophys Acta 1272: 181-184, 1995
31. Kunz WS, Gellerich FN: Quantification of the content of fluorescent
flavoprateins in mitochondria from rat liver, kidney cortex, skeletal
muscle and brain. Biochim Med Metab Bio150: 103-110, 1993
32. Granger BL, Lazarides E: Desmin and vimentin coexist at the perifery
of myofibril Z-disc. Cell 18: 1053-1066, 1979
33. QuistorffB, Haselgrove JC, Chance B: High spatial resolution readout
of3D metabolic organ structure: an automated, low-temperature redox-
ratio scanning instrument. Anal Biochem 148: 389-400, 1985
34. Kuznetsov AV, Sikk P, Kaambre P, Kay L, Daneshrad Z, Rossi A,
Kadaja L, Peet N, Seppet E, Saks VA: Stricking differences between
the kinetics of regulation of respiration by ADP in slow - twitch
and fast - twitch muscles in vivo. Eur J Biochem 241: 909-915,
1996
35. Khuchua ZA, Vasileva EV, Clark JF, Korchazhkina OV, Branishte T,
Kapelko VI, Kuznetsov AY, Ventura-Clapier R, Steinschneider AYa,
Lakomkin VL, Runge EK, Saks VA: The creatine kinase system and
cardiomyopathy. Am J Cardiovasc Pathol 4: 223-234, 1992
36. Saks VA, Belikova YO, Kuzuetsov AV, Khuchua ZA, Branishte TH,
Semenovsky ML, Naumov VG: Phosphocreatine pathway for energy
transport: ADP diffusion and transport. Am J Physiol Suppl 261: 30-
38, 1991
37. Kuzuetsov AV, Winkler K, Kirches E, Lins H, Feishner H, Kunz
WS: Application of inhibitor titrations for the detection of oxidative
phosphorylation defects in saponin-skinned muscle fibers of
patients with mitochondrial diseases. Biochim Biophys Acta 1997
(in press)
Molecular and Cellular Biochemistry 184: 101-105, 1998.
1998 Kluwer Academic Publishers.

Cytoskeleton and mitochondrial morphology and

function

L. Rappaport, P. Oliviero and lL. Samuel

Unite 127 INSERM, IFR Circulation, University D Diderot, H6pital Lariboisiere, 41 Boulevard de La Chapelle,
Paris 75475, cedex 10, France

Abstract
It has been well established that the cytoskeleton is an essential modulator of cell morphology and motility, intracytoplasmic
transport and mitosis, however cytoskeletal linkage to the organelles has not been unequivocally demonstrated. Indeed,
cytoskeleton appears to be essential in detennining and modulating gene phenotype as a function of cellular environment.
According to recent studies, the organization of the cytoskeleton network together with associated protein( s) could be essential
in regulating mitochondrial function and particularly the penneability of the mitochondrial outer membrane to ADP. The aim
of this chapter is to summarize the main properties of the cytoskeletal environment of mitochondria and the possible role(s)
of this network in mitochondrial function in myocytes. (Mol Cell Biochem 184: 101-105,1998)

Key words: heart, microtubules, desmin, ADP, porin, mitochondria-associated proteins

Introduction Ventura-Clapier in this book). The aim of this chapter is to

Although the cytoskeleton is an essential modulator of cell
morphology and motility, intracytoplasmic transport and
mitosis, cytoskeletal linkage to the organelles has not been
unequivocally demonstrated [1, 2]. Indeed, the EM analysis
of resinless sections after treatment with saponin (which
permeabilizes plasma membranes, releases soluble proteins
and preserves many cytoplasmic membranes) shows a highly
intricate cytostructure of filaments which anastomoses with
the membrane surface of the organelles [3]. Concerning the
nucleus, the cytoskeleton could modulate gene phenotype
as a function of cellular environment by determining
junctions between the extracellular cell matrix and the
nucleus [4]. According to recent studies, the cytoskeleton
network and/or associated proteins could be essential in
regulating mitochondrial function as well; mitochondrial
function could be modulated by physical effects, such as
mitochondrial shape, stretching or contraction of the
mitochondrial membrane [5]. Mitochondrial function could
also depend on specific interactions between mitochondrial
outer membrane enzymes and associated, or regulatory,
cytoskeletal protein( s) ([6-8] and chapters of V Saks and R
summarize (1) the main knowledge concerning the cyto-
skeletal environment of mitochondria and (2) the proposals
of possible role(s) of this network and associated proteins
in regulating mitochondrial function.
The cytoskeleton
The tenn cytoskeleton is often considered roughly synony-
mous with the three filaments most frequently imaged by
fluorescence microscopy (micro filaments of actin, micro-
tubules and intennediate filaments) [1, 9].
The microtubules are essentially composed oftubulin, and
their assembly and function are regulated by microtubule
associated proteins, the major classes of which are the
microtubule associated motors (kinesin, dyne in etc.) and
the structurally associated proteins (tau, Microtubule
Associated Proteins MAPS), which modulate tubule stability
and spatial arrangement. In addition, a variety of other
proteins can interact with microtubules and enhance the
fonnation of the bundles. The major constituent of micro-
filaments is actin which, as tubulin in the microtubules, exists

Address for offPrints: L. Rappaport, Unite 127 INSERM, IFR Circulation, University D Diderot, Hiipital Lariboisiere, 41 Boulevard de La Chapelle, Paris
75475, cedex 10, France
102
in an equilibrium between polymerized and depolymerized
states. The third cytoskeleton structure consists of so-called
intermediate filaments, which are rather stable fila-
ments, the type of their constitutive protein, such as desmin
or vimentin, being specific of the tissue origin. However,
these three structures are only a small part (perhaps 10-15%)
of the cytostructure, according to biochemical measurement,
the remainder being occupied by a microtrabecular lattice [2].
The extracellular matrix alters cytoskeletal organization,
stiffness and apparent viscosity by binding integrins. This
promotes formation of molecular links with the cyto-
skeleton and transmission of mechanical stresses across
these linkages to induce structural rearrangements within a
continuous tensionally integrated cytoskeletal lattice [10].
Tensegrity models, postulated by Ingber, emphasize
how the mechanical forces generated through such cell-
extracellular matrix interactions could produce altered
cytoskeletal pattern and as a result nuclear morphology, and
ultimately induce changes in the pattern of gene expression
[11]. Indeed, the cytoskeleton appears to modulate cell-
surface events and the transduction of signals from external
to internal cellular sites, and also to participate in essential
nuclear events such as transcription [4, 12].
Cytoplasmic Ca 2+ and phosphoinositoside levels as well
as pH changes, which all are involved in several signal
transduction pathways, are able to change the degree of
polymerization of microtubule proteins. Polymerization-
depolymerization of cytoskeleton components may be
involved in the passage from microscopic to macroscopic
order at the cellular level and participate to the synchron-
ization of cellular activities in the cytoplasm of living cells
[13]. Indeed, these authors found that the catalytic activity
of several enzymes from carbon catabolism, e.g. glucose 6P
dehydrogenase and pyruvate kinase (PK), is activated in a
concentration dependent manner by microtubular protein
either in its polymerised or non polymerised state. A large
fraction of PK activity coprecipitated with polymerized
microtubule proteins indicating interaction between the
enzyme and the macromolecular structure and particularly
with MAPS. Brain microtubules induce an increase in
cooperativity (as measured by a Hill coefficient) and an
increase in Vmax' Besides, it has been proposed that in highly
organized systems, the substrates and products of inter-
connected enzymatic reactions move from one enzyme to
the other without intermediate release and in this way, they
may perform very rapid signal transmission from one part
of the cell to another [14]. The structural basis of this kind
of intracellular signaling system is poorly understood. Most
probably, it is based on a specific organization of the
cytoskeleton and its structural and functional interactions
with mitochondria and other components ofthe intracellular
metabolic pathways such as creatine kinase and adenylate
kinase systems [15].
The cytoskeletal environment of mitochondria
Mitochondria are associated with the three major cyto-
skeletal structures [16-18].
Mitochondria often display a subcellular distribution
corresponding to that of the cytoplasmic microtubular
network [19, 20]. Treatment of cultured cells with chemical
agents that depolymerize microtubules often leads to altered
mitochondrial distribution [16, 20]. An isozyme of the
microtubule-based motor protein, kinesin, has been co-
localized with mitochondria in neuronal cells [21].
Filamentous cross-bridgings between the mitochondrial
outer membrane and microtubules have been detected in
frozen frog neurons by EM [22]. These cross-bridgings are
influenced by ATP hydrolysis and may also depend on the
presence of additional cytoplasmic proteins [6, 23]. MAPs
bind in vitro to specific sites of the outer membrane of
purified brain mitochondria and mediate cross-bridgings
between these cell organelles and the microtubules [23].
The MAPs binding sites are regionally concentrated,
suggesting that specialized domains are involved in the
association of mitochondria to microtubules in situ [6].
MAPs and intermediate filament subunits seem to share
common binding sites on the brain mitochondrial membrane,
interactions between them depending on ATP hydrolyzing
enzymes.
The significance of cytoskeleton and mitochondria
relationships
A number of studies have implicated the cytoskeleton in
the positioning and movement of mitochondria [20, 24, 25].
However, the significance of the interrelations with inter-
mediate filaments is poorly understood. Axonal mito-
chondrial transport occurs along both microtubules and
microfilaments in vivo, but with different velocities and
net transport properties [26, 27]. Intermediate filaments,
in rat Leydig cells have been shown to establish direct
contact with mitochondria and lipid droplets, suggesting
possible mechanisms for the transport of cholesterol from
droplets to mitochondria, and hence into the steroidogenic
pathway [28].
The binding of MAPs to mitochondria induces the re-
organization of a voltage-dependent channel protein, porin,
[29,30] which is one ofthe MAPs-binding molecules of the
outer membrane, thus suggesting a physiological regulatory
relationship between the two cellular components (possibly
controlling the targeted delivery of ATP through the outer
membrane pores and participation in the regulation of the
nucleotide flux of nucleotides through the two membranes
at the site of contact) [31]. In addition to porin, hexokinase
and MAPS, the outer membrane fractions of brain mito-
 103

chondri a contain an actin binding protein, fodrin, which Cytoskeleton and mitochondria in muscle
mediates interactions between actin filaments and other
subcellular elements such as the plasma membrane [32]. Slow freezing processes permit the observation of numerous
An F5 protein, which has a large number of potential strand-like connections between mitochondria and the
phosphorylation sites, could also play also a role in membrane- myofibrillar surface within rat myocardial tissue. It has been
cytoskeleton interactions and in dynamic aspects of synaptic proposed that mitochondrial function and particularly the
structure or function [33]. The family of annexins appears as permeability of outer membrane for ADP, which depends on
an interesting candidate for modulating the eventual cyto- porin pores, is controlled by some protein(s) connected to
skeleton dependent function of mitochondria. These the cytoskeleton and associated in a structural complex ([7,
proteins associate with phospholipid membranes in a 8] and chapters ofV Saks and R Ventura-Clapier in this book).
calcium dependent manner. They modulate the differentiation Annexins V and VI, which have been recently shown in the heart
of cytoskeleton organization and activity of membrane such to be associated by a calcium dependent mechanism to both
as exocytosis or ionic fluxes [34]. In myocardium, annex ins the cytoskeleton and the mitochondria membranes could be
V and VI are associated to both cytoskeleton and mitochondria. one of these proteins [35].
Their binding depends on Ca2+ [35]. In the heart, the relative density of microtubules is 1000
In addition to the dynamic properties of mitochondria fold less than in brain [41, 42]. With electron microscopy
found in most cell types, some proteins have recently been and immunolabelling analysis, it has been clearly shown
shown to influence a dynamic equilibrium between different that microtubules run along myofibrils and organelles [43]
mitochondrial morphologies in yeast [36]. According to (Fig. I). In hypertrophying heart the density of micro-
these studies mitochondrial morphology, division and tubules increases mostly along nuclei but also around
inheritance all depend on interactions between mitochondria mitochondria, myofibrils and plasmalemma. Microtubules
and the cytoskeleton, which, in yeast, requires the presence have been said to be implicated in myofibrillogenesis and
of a protein MdmlOp [37]. This mitochondrial outer mem-
brane protein, with a large carboxyl-terminal domain facing A
the cytosol, could maintain mitochondria in an elongated
shape by attaching the organelle to an external framework,
such as the cytoskeleton [38]. To determine the mechanism
by which mitochondrial shape is established and to understand
its role in cell function, Burgess et al. have screened a
collection of yeast mutants for those with defective mito-
chondrial morphology. In one mutation (mmmi mutant) the
shape of mitochondria alone was affected (but not that of
the cytoskeleton or other organelles). The mutation
affected the expression of a mmmi protein, normally
located in the mitochondrial outer membrane, which would
maintain mitochondrial shape by mediating binding to the B
cytoskeleton or some other framework. Furthermore the
mmmi mutants were defective in transmitting the altered
mitochondria to daughter cells, suggesting that mitochondrial
shape is critical for normal cell function [38]. The involve-
ment of microtubules in mitochondria positioning was
indicated by the behavior of mitochondria in two distinct
yeast mutants each disrupted for microtubule function [39].
Finally, in meiotic yeast bearing temperature sensitive
mutations in the actin-encoding gene ACT I, defects of
mitochondrial morphology and motility are observed, thus
supporting a role for the actin cytoskeleton in the control
of mitochondrial position and movement [40]. Thus micro-
tubules and/or microfilaments together with mitochondrial Fig. 1. Electron micrograph of longitudinal ultrathin frozen sections of rat
outer membrane proteins appear critical in determining the papillary muscle double labelled with tubulin antibodies (small gold
particles) and desmin antibodies (large gold particles). In (A) desmin labelling
morphology, and probably function, of mitochondria.
is located along the Z lines and ends of mitochondria. In (B) microtubules
run along myofibrils and are located close to mitochondria. (Panel A x
15000; Panel Bx 35 000). From courtesy ofSC Watkins).
104
in regulating muscle contraction [41, 43, 44]. MAPS-I, 2,
4 have been described in the heart (ref. in [41]) but no clear
binding sites have been yet identified, as they have in
neurons, where they are involved in a translocation process
permitting mitochondrial migration throughout the cell (see
above). Interestingly, the binding of MAPs to the outer
membrane of mitochondria induces the reorganization of
the voltage-dependent pore of porin [29] and thus could
participate to the control of nucleotide flux across the outer
membrane of mitochondria [31]. However, in permeabilized
cardiomyocytes, alterations of either microtubules with
colchicine or taxol, or micro filaments, with cytochalasin,
did not change the affinity of mitochondria for ADP [8].
This result has to be compared with studies of digitonin-
permeabilized hepatocytes [45]. In these cells the increase
in activity of a mitochondrial outer membrane enzyme, the
camitine palmitoyitransferase-l (CPT -1), after stimulation
by the phosphatase inhibitor okadaic acid (OA), involves
interactions between mitochondrial outer membrane and
cytoskeleton, but is not altered by drugs inducing the
disruption of microtubules and microfilaments. Thus,
modulation of a mitochondrial outer membrane enzyme
activity could involve specific interactions between CPT-l
and regulatory cytoskeletal proteins.
In striated muscle, the intermediate filaments of desmin
form a sarcoplasmic network surrounding the Z-discs, linking
them together and integrating the contractile apparatus with
the sarcolemma and the membrane of organelles, such as
nuclei and mitochondria [43, 46]. Generated desmin null
mice are viable and fertile [47,48]. However, EM analysis
reveals structural abnormalities including loss of the lateral
alignment of myofibrils and abnormal mitochondrial organ-
ization. The consequences are particularly severe in the
heart, with progressive myocardial degeneration and
necrosis accompanied by extensive calcification. In the heart
and soleus skeletal muscle of these desmin deficient mice
the kinetics of respiration regulation by ADP is changed due
to the appearance of a population of mitochondria with high
ADP affinity in some cells with disorganized structure,
suggesting a close connection between structural organization
of the organelles, the cellular mechanism of ADP regulation
and energy fluxes [49]. One hypothesis is that the structural
disorganization of the intermediate filaments results in the
dissociation, or removal, of a factor X, normally
associated with the mitochondrial surface [49] and/or of
cytoskeletal-associated protein( s) as already described in
brain or yeast [38-41]. This factor would control the
permeability of the outer mitochondrial membrane porin
pore for adenine nucleotides in muscle cells with high
oxidative activity. Thus, the permeability of the outer mito-
chondrial membrane for ADP in muscle could be controlled
by unknown cytoplasmic protein(s), probably related to the
cytoskeleton [8]. Interestingly, according to the phospho-
creatine circuit theory, metabolic channelling occurring
between myofibrils and membrane-bound cytosolic creatine
kinase isoenzymes is also probably of prime importance for
the export of high-energy phosphates out of the mito-
chondria, and thus for the energy metabolism of the whole
cell ([2, 7], see the chapter by Walliman's group in this
volume).
In conclusion, there is agreement that the cytoskeleton
mediates mitochondrial movement and positioning. In
addition, as in nucleus, the cytoskeleton may interfere with
mitochondria activity by modulating mitochondrial shape
and thereby function. However, more probably and according
to recent results, the cytoskeleton network may participate
in mitochondrial activity via outer membrane and cyto-
skeleton associated proteins.
Acknowledgements
The authors are grateful to Dr. J.E Leterrier, Dr. D. Charle-
magne, Dr. V Saks and Dr. R. Ventura-Clapier for fruitful
discussions. The work was supported by INSERM grant 94
EW 10 and CNRS.
References
1. Cleveland DW, Mooseker MS: Cytoskeleton current opinion. Cell BioI
6: 1-140,1994
2. Ovadi J: Ultrastructure of the cell. In: Cell Architecture and Metabolic
Channelling. Springer-Verlag, New York, Berlin, 1995, pp 22-27
3. Lin A, Krockmalnic G, Penman S: Imaging cytoskeleton-mitochondrial
membrane attachments by embedment-free electron microscopy of
saponin-extracted cells. Proc Nat! Acad Sci USA 87: 8565-8569, 1990
4. Boudreau N, Myers C, Bissel MJ: From laminin to lamin; regulation of
tissue-specific gene expression by the ECM. Trend Cell BioI 5: 1-4,1995
5. Bereiter-Hahn J, Vohn M: Dynamics of mitochondria in living cells;
shape changes, dislocations, fusion and fission of mitochondria.
Microscopy Res Tech 27: 198-219, 1994
6. Leterrier JF, Rusakov BA, Nelson DB, Linden M: Interactions between
brain mitochondria and cytoskeleton: Evidence for specialized outer
membrane domains involved in the association of cytoskeleton-
associated proteins to mitochondrial in situ and in vitro. Micro Res
Tech 27: 233-261,1994.
7. Saks VA, Kuznetsov AV, Khuchua ZA, Vasilyeva EV, Belikova JO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis; possible involvement of mitochondrial-
cytoskeleton interactions. J Mol Cell Cardiol27: 625--{)45, 1995
8. Saks VA, Tiivel T, Kay L, Novel-Chate V, DaneshradZ, RossiA, Fontaine
E, Keriel C, Leverve X, Ventura-Clapier R, Anflour K, Samuel JL,
Rappaport L: On the regulation of cellular energetics in health and
disease. Mol Cell Biochem 160: 195-208, 1996
9. Penman S: Rethinking cell structure. Proc Natl Acad Sci USA 92: 5251-
5257,1995
10. Wang N, Ingber DE: Control of cytoskeletal mechanics by extracellular
matrix, cell shape and mechanical tension. Biophys J 66: 2181-2189, 1994

I!. Ingber DE: Cellular tensegrity: Defining new rules of biological design
that govern the cytoskeleton. J Cell Sci 104: 613-627, 1993
12. Roskelley CD, Srebow A, Bissel JA: A hierarchy of ECM-mediated
signalling regulates tissue specific gene expression. Curr Opin Cell
BioI vol 5: 736-743, 1995
13. Cortassa S, Aon MA: Entrainment of enzymatic activity by the
dynamics of cytoskeleton. In: HV Westenhoff, JL Snop, FE Shen, JE
Wighes, BN Khatodenko (eds). Biothermo Kinetics of the Living Cell.
Biochtermokinetics Press, Amsterdam, 1996, pp 337-342
14. Mitchell P: The ninth Sir Hans Krebs Lecture. Compartmentation and
communication in living systems. Ligand conduction: a general
catalytic principle in chemical, osmotic and chemiosmotic reaction
systems. Eur J Biochem 95: 1-20, 1979
15. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova YO, Kuznetsov A:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration. A synthesis. J Mol Cell Biochem 1331134: 155-192, 1994
16. Ball EH, Singer SJ: Mitochondria are associated with microtubules and
not with intermediate filaments in cultured fibroblasts. Proc Nat! Acad
Sci USA 79: 123-126, 1982
17. Mose-Larsen P, Bravo R, Fey SJ, Small JV, Celis JE: Putative association
of mitochondria with a sUbpopulation of intermediate-sized filaments
in cultured human skin fibroblasts. Ce1l31: 681-692, 1982
18. Bereiter-Hahn J: Behavior of mitochondria in the living cell. Intern Rev
Cyto1122: 1-62,1990
19. Leterrier JF, Rusakov DA, Linden M: Statistical analysis ofthe surface
distribution of microtubule-associated proteins (MAPs) bound in vitro
to rat brain mitochondria and labelled by 10 rnn gold-coupled antibodies.
Bulletin de l'Associations des Anatomistes 78: 46-51,1994
20. Heggeness MH, Simon M, Singer SJ: Association of mitochondria
with microtubules in cultured cells. Proc Natl Acad Sci USA 75: 3863-
3866,1978
21. Nangaku M, Sato-Yoshitake R, Okada Y, Noda Y, Takedamura R,
Yamasaki H, Hirikawa N: KIFIB, a novel microtubule plus end-directed
monomeric motor protein for transport of mitochondria. Cell 79: 1209-
1220,1994
22. Hirokawa; Cross-l inker system between neurofilaments, microtubules
and membranous organelles in frog axons revealed by quick-freeze,
deep etching method. J Cell Bioi 94: 129-142,1982
23. Leterrier J, Eyer J, Weiss DG, Linden M: In vitro studies of the physical
interactions between neurofilaments, microtubules and mitochondria
isolated from the central nervous system. In: Living Cell In Four
Dimensions. Proc Conf Amer Inst Physics. 226: 91-105, 1991
24. Drubin DG, Jones HD, Wertman KF: Actin structure and function:
Roles in mitochondrial organization and morphogenesis in buding yeast
and identification of the phalloidin-binding site. Mol BioI Cell 4: 1277-
1294,1993
25. Martz D, Lasek RJ, Brady ST, Allen RD: Mitochondrial motility in axons:
Membranous organelles may interact with the force generating system
through multiple surface binding sites. Cell Moti! Cyto 4: 89-101, 1984
26. Smith DS, Jarlfors U, Cameron BF: Morphological evidence for the
participation of microtubules in axonal transport. Ann NY Acad Sci
253:472-506,1975
27. Morris RL, Hollenbeck PJ: Axonal transport of mitochondrial along
microtubules and F -actin in living vertebrate neurons. J Cell BioI 131:
1315-1326,1995
28. Almahbobi G, Williams LJ, Han XG, Hall PF: Binding oflipid droplets
and mitochondria to intermediate filaments in rat Leydig cells. J Reprod
Fert98:209-217,1993
29. Linden M, Nelson BD, Loncar D, Leterrier JF: Studies on the interaction
between mitochondria and the cytoskeleton. J Bioenerg Biomem 21:
507-518,1989
105
30. Linden M, Gellefors P, Nelson BD: Purification of a protein having
pore forming activity from the rat liver mitochondrial outer membrane.
Biochem J 208: 77--S2, 1982
31. Brdiczka D: Contact sites between mitochondrial envelope membranes.
Structure and function in energy- and protein-transfer, (review) Biochim
BiophysActa 1071: 291-312,1991
32. Levine J, Willard M: Fodrin: Axonally transported polypeptide
associated with the internal periphery of many cells. J Cell BioI 90:
631--643, 1994
33. Arai M, Cohen JA: Subcellular localization of the F5 protein to the
neuronal membrane-associated cytoskeleton. J Neurosci Res 38: 348-
357,1994
34. Towle CA, Treadwell BY: Identification of a novel mammalian annexin.
CDNA cloning, sequence analysis, and Ubiquitous expression of the
annexinXI gene. JBiol Chern 267: 5416-5423, 1992
35. Sun J, Bird CH, Salem HH, Bird P: Association of annexin V with
mitochondria. FEBS Lett 329: 79--S3, 1993
36. Sogo FL, Yaffe MP: Regulation of mitochondrial morphology and
inheritance by Mdml Op, a protein of the mitochondrial outer membrane.
J Cell BioI 126: 1361-1373, 1994
37. McConnell SJ, Yaffe MP: Nuclear and mitochondrial inheritance in
yeast depends on novel cytoplasmic structures defined by the MOM I
protein. J Cell BioI 118: 385-395, 1992
38. Burgess SM, Delannoy M, Jensen RE: MMMI encodes a mitochondrial
outer membrane protein essential for establishing and maintaining the
structure of yeast mitochondriaU Cell BioI 126: 1375-1391, 1994
39. Yaffe MP, Harata D, Verdes F, Eddison M, Toda T, Nurse P: Microtubules
mediate mitochondrial distribution in fission yeast. Proc Natl Acad Sci
USA93: 11664-11668, 1996
40. Smith MG, Simon VR, O'Sullivan H, Pon LA: Organelle-cytoskeletal
interactions: Actin mutations inhibit meiosis-dependent mitochondrial
rearrangement in the budding yeast Saccharomyces cerevisiae. Mol
Bioi Cell 6: 1381-1396, 1995
41. Rappaport L, Samuel JL: Microtubules in cardiac myocytes. lnt Rev
Cytoll: 1310--1143, 1988
42. Samuel JL, Schwartz K, Lompre AM, Delcayre C, Marotte F,
Swynghedauw B, Rappaport L: Immunological quantitation and
localization of tubulin in adult rat heart isolated myocytes. Eur J Cell
Bio131: 99-106,1983
43. Watkins SC, Samuel JL, Marotte F, Bertier-Savalle B, Rappaport L:
Microtubules and desmin filaments during the onset of heart
hypertrophy in rat: A double immunoelectron microscope study. Circ
Res 60: 327-336, 1987
44. Sun J, Bird CH, Salem HH, Bird P: Association of annexin V with
mitochondria. FEBS Lett 329: 79--S3, 1993
45. Velasco G, Sanchez C, Geelen MH, Guzman M: Are cytoskeletal
components involved in the control of hepatic camitine palmitoyl-
transferase I activity? Biochem Biophys Res Comm 224: 754-759, 1996
46. Lazarides E, Granger BL: Transcytoplasmic integration in avian
erythrocytes and striated muscles; the role of intermediate filaments.
Mol Cell Bioi 2: 143-162, 1983
47. Milner DJ, Weitzer G, Tran D, Bradley A, Capetanaki Y: Disruption of
muscle architecture and myocardial degeneration in mice lacking
desmin. J Cell Bioi 134: 1255-1270, 1996
48. Li Z, Colucci-Guyon E, Pinyon-Raymond M, Mericskay M, Pournin S,
Paulin D, Babinet C: Cardiovascular lesions and skeletal myopathy in
mice lacking desmin. Devel Bioi 175: 362-366, 1996
49. Kay L, Li Z, Mericskay M, Olivares J, Tranqui L, Fontaine E, Tiivel T,
Sikk P, Kaambre T, Samuel JL, Rappaport LR, Usson Y, Leverve X,
Paulin D, Saks VA: Study of regulation of mitochondrial respiration in
vivo. An analysis of influence of ADP diffusion and possible role of
cytoskeleton. Biochim Biophys Acta 1322: 41-59, 1997.
Molecular and Cellular Biochemistry 184: 107-121, 1998.
1998 Kluwer Academic Publishers.

Energetics of swelling in isolated hepatocytes: A

comprehensive study

Anne Devin, Pascal Espie, Bernard Guerin and Michel Rigoulet

Institut de Biochimie et Gimetique Cellula ires du CNRS, Universite de Bordeaux 2, I rue Camille Saint-Saens, 33 077
Bordeaux cedex, France

Abstract
Cell swelling is now admitted as being a new principle of metabolic control but little is known about the energetics of cell
swelling. We have studied the influence of hypo- or hyperosmolarity on both isolated hepatocytes and isolated rat liver
mitochondria. Cytosolic hypoosmolarity on isolated hepatocytes induces an increase in matricial volume and does not
affect the myxothiazol sensitive respiratory rate while the absolute value of the overall thermodynamic driving force over
the electron transport chain increases. This points to an increase in kinetic control upstream the respiratory chain when
cytosolic osmolarity is decreased. On isolated rat liver mitochondria incubated in hypoosmotic potassium chloride media,
energetic parameters vary as in cells and oxidative phosphorylation efficiency is not affected. Cytosolic hyperosmolarity
induced by sodium co-transported amino acids, per se, does not affect either matrix volume or energetic parameters. This
is not the case in isolated rat liver mitochondria incubated in sucrose hyperosmotic medium. Indeed, in this medium, adenine
nucleotide carrier is inhibited as the external osmolarity increases, which lowers the state 3 respiration close to state 4
level and consequently leads to a decrease in oxidative phosphorylation efficiency. When isolated rat liver mitochondria
are incubated in KCI hyperosmotic medium, state 3 respiratory rate, matrix volume and membrane electrical potential
vary as a function oftime. Indeed, matrix volume is recovered in hyperosmotic KCI medium and this recovery is dependent
on Pi-Kentry. State 3 respiratory rate increases and membrane electrical potential difference decreases during the first
minutes of mitochondrial incubation until the attainment of the same value as in isoosmotic medium. This shows that matrix
volume, flux and force are regulated as a function of time in KCI hyperosmotic medium. Under steady state, neither matrix
volume nor energetic parameters are affected. Moreover, NaCI hyperosmotic medium allows matrix volume recovery but
induces a decrease in state 3 respiratory flux. This indicates that potassium is necessary for both matrix volume and flux
recovery in isolated mitochondria. We conclude that hypoosmotic medium induces an increase in kinetic control both
upstream and on the respiratory chain and changes the oxidative phosphorylation response to forces. At steady state,
hyperosmolarity, per se, has no effect on oxidative phosphorylation in either isolated hepatocytes or isolated mitochondria
incubated in KCI medium. Therefore, potassium plays a key role in matrix volume, flux and force regulation. (Mol Cell
Biochem 184: 107~121, 1998)

Key words: volume, hepatocytes, mitochondria, oxidative phosphorylation, potassium, osmolarity

Introduction intestinal absorption of water ([ 1] for review) and physio-

To serve their functions, cells have to accumulate a number
of osmotically active substances which tend to swell or
shrink the cell ([ 1] for review). Moreover, during intestinal
absorption, portal venous blood may become slightly hypo-
or hypertonic. In fact, liver swelling has been shown during
logical fluctuations of the portal amino acid concentration
are accompanied by parallel alterations ofliver cell volume
[2]. The degree of amino acid induced cell swelling seems
largely to be related to the steady state of intra/extracellular
amino acid concentration gradient and to the electro-
chemical Na+ gradient, which acts as a driving force for Na+

Address/or offprints: M. Rigoulet, Institut de Biochimie et Genetique Cellulaires du CNRS, Universite de Bordeaux 2, 1 rue Camille Saint-Saens, 33 077
Bordeaux cedex, France
108
coupled transport ([ 1] for review). N a+co-transported amino
acids induce an amino acid accumulation and consequently
a cytosolic hyperosmolarity [3]. Furthermore, in order to
maintain a certain constancy in their volume which is
necessary for their survival, cells have to highly regulate it
[4--{)]. Mechanisms of cell volume regulation have been
previously described [4--{)]. In spite of the activity of these
numerous systems involved in the regulation of cell volume
([1] for review), a residual increase or decrease in volume
exists in conditions of swelling or shrinkage. The extent
of these residual deviations of cell volume apparently
triggers metabolic cell functions ([1] for review). Indeed,
an increase in cell volume, both in perfused liver and in
isolated hepatocytes, affects fluxes through some meta-
bolic pathways, acts like an anabolic signal and stimulates
and inhibits the biosynthesis and degradation pathways
respectively, whereas a decrease in cell volume acts as a
catabolic signal. Thus, cell swelling induced by either a
hypoosmotic medium or exposure to amino acids (i)
inhibits glycogenolysis, glycolysis [4, 6], proteolysis [1,
7,8], glutamine synthesis and urea synthesis from NH/ [7,
9, 10]; (ii) stimulates glycogen synthesis [3], protein
synthesis [7], amino acid uptake [7, 9,11], glutaminase [7]
and urea synthesis from amino acids [12]. While the
intermediary metabolism has been extensively studied and
some mechanisms linked to cell volume changes and
alterations of metabolic function have been proposed [13-
15], little is known in this field regarding mitochondria in
situ and the bioenergetic parameters. Indeed, cell volume
variations, per se, are linked to an increase or a decrease
in cytosolic osmolarity which could induce modifications
of mitochondrial matrix volume.
Furthermore, the influence of matrix volume variations
on mitochondrial metabolism has been studied in isolated
rat liver mitochondria ([16] for review, [17]), and it appears
that in various conditions of increasing matrix volume X-
adrenergic agonist pretreatment of rat, hypoosmotic
incubation conditions, intramitochondrial increase in
[Ca2+]), the mitochondrial metabolism is strongly affected.
For instance, in the presence of glucagon, mitochondrial
respiratory rate, pyruvate metabolism and citrulline syn-
thesis are stimulated [18]. When external osmolarity was
decreased, fatty acid oxidation was stimulated while
cytochromes were more reduced [19]. In conditions of
matrix volume condensation, an inhibition of mitochondrial
substrate oxidation has been observed and interpreted as a
consequence of an osmotically sensitive diffusion of
quinones through the mitochondrial membrane [20]. It has
also been shown that under conditions of matrix con-
densation, state 3 respiratory rate was inhibited, as was
ATPase activity [21], andADP/O ratio was decreased [22].
All these studies were done on isolated rat liver mito-
chondria in either sucrose or NaCI medium. However, the
mitochondria in mammalian cells are exposed to a cytosol
that is high in potassium and low in sodium [23], and it is
now well established that potassium, in addition to its
osmotic function, is fully involved in matrix volume
regulation by the intermediary of K+ influx and/or efflux
by the way ofK+/H+ exchanger (which extrudes an excess
of K+), K+-channel and K+-Ieak (see Fig. I for the mito-
chondrial K+-cycle) [23]. Thus, potassium is involved in
both mitochondrial enzymatic matrix reactions and the
control of mitochondrial volume [24, 25]. It seems relevant
to note that in situ, mitochondria can dispose of potassium
in such a concentration that is isoosmotic for isolated
mitochondria. So, by replacing sucrose with KCI, we
mimicked the potassium concentration in the cytosol
surrounding isolated rat liver mitochondria.
The study of the energetics of cytosolic hypo- or hyper-
osmolarity was thus conducted in various incubation con-
ditions: (i) in isolated hepatocytes either in hypoosmotic
extracellular medium, in order to study the influence of
cytosolic hypoosmolarity on oxidative phosphorylation, or
in the presence of sodium co-transported amino acids in
order to study the influence of cytosolic hyperosmolarity
on oxidative phosphorylation; (ii) in isolated rat liver
mitochondria in either sucrose or KCI media of various
osmolarities. In this study, we show that many mitochondrial
functions are regulated by potassium fluxes which buffer the
volume and energetic parameter variations when external
medium osmolarity is altered. Hence, in order to compare
the energetics of isolated rat liver mitochondria with the
energetics of isolated hepatocytes, a relevant medium has
to be used.
Materials and methods
Isolated hepatocytes
Preparation and incubation of hepatocytes
Hepatocytes from 20-24 h starved male Wistar rats (200-
250 g) were isolated by the collagenase method of Berry
+
1
Fig. 1. The mitochondrial K+ cycle (according to Garlid KD (1994) J Bioener
Biomem 26: 537-542). (I) Respiratory chain (2) ATPsynthase (3) proton leak
(4) potassium/proton exchanger (5) potassium leak (6) potassium channels.

and Friend [26] as modified by Groen et al. [27].
Hepatocytes (8-10 mg dry wt/ml) were incubated in 20
ml stopped plastic or glass vials in a shaking waterbath. The
basic incubation medium was a Krebs-Henseleit-bicar-
bonate buffer (144 mM Na+) at pH 7.4 [28] containing 2.5
mM Ca2+, 2% defatted bovin serum albumin, a respiratory
substrate (20 mM glucose, 4 mM octanoate or 4 mM BSA-
oleate as indicated), 0.1 !-lM TPMP+ (tetraphenylmethyl-
phosphonium), 0.5 mM mannitol, and 0.2 mg/ml insulin.
Amino acids (proline, alanine, glutamine and ~-amino
isobutyrate) when added were used at 10 mM, except for
leucine 2.5 mM. Hypoosmotic, Na+-depleted media were
obtained by decreasing the Na+ concentration ofthe buffer
as indicated. All media were in equilibrium with a gas phase
containing 02:C0 2 (95:5). The temperature was 37C.
Measurement of respiration rate
After 20 min incubation, the cell suspension was transferred
to an oxygen electrode for determination of respiration rate.
Myxothiazol and oligomycin, were used at respectively 2 Ilg/
ml and 25 !-lg/ml to determine the myxothiazol- and oligo-
mycin-sensitive respiration rate. Aminooxyacetate (AOA)
was used at 1 mM.
Measurement of cell and mitochondria volume in situ
Hepatocytes were incubated in the presence of (i) either
2.5 !-lCi/mI3Hp and 0.1 !-lCi/ml [14C]carboxymethylinulin
(added 1 min before sampling) for measurement of cell
volume; (ii) or 2.5 IlCi/m13Hp and 0.1 !-lCi/ml [14C]mannitol
for measurement of mitochondrial volume [29]. After 25 min
incubation, I ml of cell suspension was transferred to an
Eppendorftube and centrifuged for 5 sec at 7600 g. Within
10 sec, 100 !-ll of the supernatant was removed into a
scintillation vial containing 5 ml of a liquid scintillation
cocktail (Beckman), and the remainder was aspired. The pellet
was resuspended with 300 !-ll HCIO 4 10%, centrifuged, and
200 !-ll of the supernatant treated as before. The volume of
each isotope in the pellet was (dpmc.3/2)/( dpm/l 00) in Ill.
Measurement of the mitochondrial and cytosolic
membrane potential in situ
Hepatocytes were incubated in the presence of (i) I !-lCi/ml
[3H]TPMP+ for measurement of TPMP+ accumulation; (ii)
0.1 IlCi/mI36Cl- for measurement ofCl- distribution [30]. Cell
samples were treated as above. The PH]TPMP+ accumulation
ratio and the 36Cl- distribution were calculated as in Nobes et
al. [31]. Thus, the mitochondrial membrane potential was:
109
(see [31-34]). Vc ,V m ,a,e acy ,am refer respectively to cellular
and mitochondrial volume, and apparent activity coefficients
for TPMP+ in the extracellular medium, the cytoplasm and
the mitochondrial matrix. We redetermined ae (because of
our different extracellular medium), as well as acy and am [31,
35]. We take ae = 0.85, acy = 0.43, am = 0.38.
Compartmentation of nucleotides, phosphate and
amino acids
The mitochondrial and cytosolic distribution of amino acids
was studied by using the digitonin fractionation method as
described by Zuurendonk and Tager ([36], see also [37]).
Measurement of intramitochondrial redox potential
The intramitochondrial nicotinamide adenine dinucleotide
reduced form/nicotinamide adenine dinucleotide oxidized
form ratio (NADHINAD+) was determined by the metabolite
indicator method [37] assuming the ~-hydroxybutyrate
(~OHBut) dehydrogenase reaction was in near equilibrium
(K = [AcetoAcetate] [NADH]/[~OHBut] [NAD+] i.e. 4.93
1O~~: [38]).
Assays
~-hydroxybutyrate and acetoacetate were measured spectro-
photometrically or fluorometrically in neutralised HClO 4-
extracts. Pi was measured according to the method of
Berenblum and Chain [39]. The amino acids (in neutralised
HCl0 4 -extracts) were derived in phenylthiocarbamyl-amino
acids and separated by HPLC at 254 nm [40].
Isolated mitochondria
Mitochondria preparation
Liver mitochondria were prepared from male Wistar rats (250-
300 g body wt ) starved overnight. Rats were killed by cervical
dislocation and liver was rapidly removed and put into an ice-
cold isolation medium containing 225 mM sucrose, 20 mM
Tris-HCl (pH 7.2) and I mM EGTA (ethylene glycol-bis(~
aminoethyl ether)N, N, N', N'-tetraacetic acid). Mitochondria
were isolated according to [41] in the same medium. The
mitochondrial pellet was finally resuspended in the isolation
medium. Protein concentration was estimated by the biuret
method using bovin serum albumin as standard [42].
Mitochondrial respiration and A TP/O measurements
The rate of oxygen uptake at various osmolarities was
measured polarographically at 26C using a Clark-type
110
oxygen electrode connected to a microcomputer giving an
online display of rate values. Respiration buffer contained 5
I!M TPMP+, 51!M DMO (5,5-dimethyl oxazolidine-2,4 dione),
51!M mannitol, 20 mMTris-HCl (PH 7.2), 1 mM EGTA, 6 mM
glutamate and malate as respiratory substrates, 5 mM Tris-Pi
and an amount of either sucrose or KCl to make up the
osmolarity to the required value. Before use, sucrose,
purchased from Merck, was passed through the Dowex 50W
column in order to remove contaminating cations. From 1-4
mg of mitochondrial protein were used as a function of
experimental conditions after checking that the respiratory
rate was always directly proportional to the mg of protein,
whatever the osmolarity.
Oxygen concentrations in the different osmolarity media
were determined with NADH quantitated spectrophoto-
metrically and yeast mitochondria [43].
Phosphorylation rate was measured by 32[P]Pi incorpor-
ation in adenine nucleotides as previously described [44].
Phosphorylation rate measurements were done in respiratory
buffer supplemented with 1 mMADP, at various osmolarities.
The ATP/O ratio stoichiometries were determined from the
yield of oligomycin-sensitive phosphorylation rates versus
respiratory rates.
Measurement of matrix space, ApH and L!lfI, using
radiolabelled elements
Routinely, the protonmotive force under different steady
states was determined as follows: Matrix space was
determined using 3[H]water and 14[C]mannitol, an inner
membrane impermeable sugar, in the respiratory buffer.
After equilibration (3 min) 0.5 mM ADP was added in order
to reach state 3 respiration. After 15 sec, mitochondria were
rapidly centrifuged (20 sec) and then treated as described
previously [44]. For state 4 measurements, the rapid
centrifugation was done when the net ATP synthesis flux was
nul (state 4). The delay after ADP addition necessary to obtain
state 4 depended on the medium osmolarity and was checked
by oxygen consumption determinations. ApH was measured
by distribution of 14[C]DMO according to [45] and A\jI by
3[H]TPMP+ distribution.
TPMP+ is a lipophilic cation which is partially linked to
membranes; we applied to our measurements a correction
coefficient of 0.38, determined in our laboratory [46], for
this binding. Since our experimental conditions implied
changes in membrane structures, we also measured A\jI in
parallel experiments using 86Rb (known as a marker which
does not bind to membranes) in the presence of valinomycin
for two different osmolarities. We observed the same linear
relationship between 3[H]TPMP+ and 86Rb accumulation
whatever the osmolarity. So we considered that the correction
factor used for TPMP+ binding was available whatever the
osmolarity.
Mitochondrial NAD(P)H fluorescence measurements
NAD(P)H measurements make it possible to qualitatively
evaluate the reducing equivalent supply to the respiratory
chain. NADH + NADPH fluorescence was monitored at
26C with a Kontron fluorimeter as previously described
[47]. Excitation wavelength was 340 nm and fluorescence
emission wavelength was 465 nm. Mitochondria (0.1 mg
protein) were suspended in the respiratory buffer at various
sucrose osmolarities to a final volume of 3 ml. Results are
expressed as a percentage of NAD(P)H fluorescence: 0%
was an endogenous signal corresponding to a steady state of
mitochondria placed in respiratory buffer without added
substrate, state 3 fluorescence corresponds to mitochondria
in the presence of 6 mM respiratory substrates and 50 I!M
ADP. State 4 fluorescence was measured when the net flux
of ATP synthesis was nul. Maximal NAD(P)H fluorescence
(100%) was measured in the presence of 1 I!g/ml antimycin.
This technique does not make it possible to determine
quantitatively the NADH concentration because free and
bound NAD(P)H are simultaneously measured and the part
played by the former and the latter can not be evaluated;
furthermore, maximal NAD(P)H fluorescence depends on
the enzyme involved and the NAD+ content could not be
quantitated.
Measurements of L!lfI time dependence
A\jI measurements as a function of time were done using
Rhodamine 123 as a probe of inner mitochondrial membrane
A\jI (as described in [48]); excitation wavelength was 485 nm
and emission wavelength was 525 nm. These measurements
were done with a KONTRON fluorimeter, in 2 ml res-
piratory buffer supplemented with 1l!M Rhodamine 123, 1
mg of mitochondrial protein and 0.5 mM ADP for state 3
establishment.
Time dependence of matrix volume
As matrix volume measurements using radiolabelled ele-
ments required 3 min to reach the equilibrium of dis-
tribution of these elements, time and respiration state
dependence of matrix volume were measured using a
KONTRON fluorimeter. Indeed, it is now well known that
an increase in mitochondrial matrix volume is
accompanied by a decrease in light scattering at 540 nm.
This general phenomenon was monitored when excitation
and emission wavelength were 540 nm. Relative matrix
volume measurements were done in 2 ml respiratory buffer
and 0.5-1 mg of mitochondrial protein supplemented with
0.5 mM ADP if required, in the same conditions as those
used for measurements of respiratory flux time
dependence.
 III

Results Indeed, both NADHINAD+ ratio and mitochondrial inner

membrane electrical potential difference (L1p) decreased when
Hypoosmoiarity external osmolarity was decreased. From a thermodynamic
point of view, the respiration rate is controlled by two
With octanoate as substrate, isolated hepatocytes incubated associated forces: the span of redox potential over the
in various hypoosmotic Na+-depleted media swell (Fig. 2). respiratory chain (L1E'h) and the protonmotive force (L1p). The
This swelling induces a decrease in cytosolic osmolarity and, protonmotive force has two components: the L1\jf and the inner
as expected, the mitochondria swell to balance this drop. mitochondrial membrane proton chemical potential difference
Table 1 shows that the respiration rate (estimated as the (L1pH). The way to assess L1pH in isolated hepatocytes is
oxygen consumption sensitive to myxothiazol at a con- controversial but it is only a minor component of
centration that completely inhibited the electron transfer protonmotive force. So we calculated the overall thermo-
through the respiratory chain) did not significantly vary dynamic driving force over the electron transport chain via
when the Na+ concentration in the extracellular medium is the L1\jf.1t appears that this overall thermodynamic driving force
decreased. Moreover, thermodynamic parameters were over the electron transport chain (expressed as 2L1E' h-1 OL1\jf)
strongly affected when external osmolarity was decreased. increases whereas the respiratory rate is maintained. Two
explanations, which are not exclusive, may be proposed: (i)
Table 1. Energetic parameters of isolated hepatocytes as a function of an increase in kinetic constraints at the level of the respiratory
external osmolarity chain limiting the increase in electron flux in response to the
decrease inL1\jf; (ii) changes in mitochondrial volume which
External JO, nat/min/mg NADHINAD+ -L'.\\f(mV) 2L'.E'h- IOL'.\\f
osmolarity (x 10-3) (volts)
could induce changes in the physical environment of
dw
membranal enzymes and consequently in their activity,
144mM 34.6 I.l 3l.62.3 160.3 4.3 -0.55 thereby inducing a change in the respiratory flux response to
120mM 33.4 1.5 27.62.1 142.8 5.7 -0.79
associated forces.
90mM 33.7 1.7 18.72.9 128.5 3.4 -0.88
To test these hypotheses and to have a better understanding
Cell substrate was 4 mM octanoate. Hypoosmotic Na+ depleted media were of the energetics of rat liver mitochondria in conditions of
obtained by decreasing the Na+ concentration of the buffer from 144 mM to
external hypoosmotic medium, we studied the influence of
120 or 90mM as indicated. Hepatocytes were incubated in closed vials for
25 min for steady state attainment. Values are means S.D. for at least 3 hypoosmotic KCl medium on isolated rat liver mitochondria.
cellular preparations. L'.E' h is the difference in redox potential across the Isolated rat liver mitochondria incubated in KCl hypo-
electron transport chain and lOis the W /0 stoichiometry of the electron osmotic media swell during the first minutes of mito-
transport chain. chondrial incubation (Fig. 3a). Moreover, this increase in
matrix volume (i) is independent of the addition of substrates
4.-----------------------------------~ which means that endogenous substrates are sufficient to
.....11\
.......
c: -+~
maintain a L1pH allowing K+/W carrier functioning; (ii) is
increased in the presence of antimycin A and oligomycin

II
(Fig. 3b). This correlates well with the hypothesis of a

-....."
0 3
A. functioning K+/W exchanger in hypoosmotic medium [24].
Indeed, the increase in matrix volume would decrease the

-
0
intramatricial [Mg2+] and this would allow the extrusion of
2
I potassium associated with the intrusion of protons. More-
'"~ over, this intrusion of protons would need a L1pH across the
;, mitochondrial inner membrane, which is abolished in the
'-"
0_ presence of antimycin A + oligomycin. This induces an
-
II
e
:lI increase in matricial volume under hypoosmotic conditions.
0
:> In KCl isoosmotic media, the volume does not vary until
O+-~_r~~._~._~_.~--._~~~~ substrate addition (Fig. 3c).
80 90 100 110 120 130 140 1 SO Under steady state, the matrix volume is increased in
NaCI osmolarity (mM) hypoosmotic medium (Fig. 4). Furthermore, when external
osmolarity is decreased, state 4 respiratory rate increases
Fig. 2. Cellular (e) and matrix (0) volume as a function of external NaCI and state 3 respiratory rate decreases (Fig. 5a). Moreover,
osmolarity on isolated hepatocytes. Cell substrate was 4 mM octanoate.
oxidative phosphorylation efficiency (i.e. ATP/O ratio) is
Hepatocytes were incubated in closed vials for 25 min for steady state
attainment. Volumes were determined as described in Materials and methods
constant whatever the external osmolarity (not shown). The
section. The values presented are from at least 3 different experiments carried NAD(P)H level, which is representative of the reduced
out with 3 different cellular preparations. Results are expressed S.D. equivalent supply to the respiratory chain, slightly decreased
112
L
light
scattering
(10%)
1 minute
Glutamate/malate
Fig. 3. Hypoosmolarity influence on time course of matrix volume in KCI
hypoosmotic medium. Matrix volume evolution in (a) hypoosmotic medium,
(b) hypoosmotic medium in the presence of antimycin and oligomycin, (c)
isoosmotic medium. Matrix volume evolution was measured by light
scattering as described in Materials and methods section. This experiment
is representative of3 such experiments.
from iso- to hypoosmolarity (Fig. 5b). From iso- to hypo-
osmolarity, ~pH slightly decreased, while ~'" was not
significantly modified (not shown). Consequently, the
protonmotive force slightly decreased from iso- to hypo-
osmolarity (Fig. 5c). These results correlate well with those
obtained in isolated hepatocytes (see Table I) and show that
oxidative phosphorylation efficiency is not affected by
external hypoosmolarity. Moreover, it is generally admitted
that there is a linear relationship between respiratory rate
and ~p when state 3 respiratory rate is titrated with oligo-
mycin. If one considers Fig. 6, given that from iso- to
hypoosmolarity, the protonmotive force and respiratory rate
are decreased in state 3 and protonmotive force is highly
decreased in state 4 for a slight increase in respiratory rate,
it could be hypothesized that there is a shift in the JO/~p
relationship between iso- and hypoosmolarity. Indeed, at the
same respiratory rate, the protonmotive force maintained in
a hypoosmotic medium was always lower than that measured
in the isoomotic medium.
Hyperosmolarity
Cytosolic hyperosmolarity in isolated hepatocytes was
induced by sodium cotransported amino acids [3). Sodium
co-transported amino acids induce an increase in cell volume
1.50
'<ii'
=
'il 1.25
-0
....'"
Q.
1.00
ciIl
~
:5
~
0.75
5
=
i~
0.50
C
oW
I 0.25
~
0.00
100 150 200 225
Osmolarity KCI (mosM)
Fig. 4. Matrix volume as a function of external osmolarity in KCI medium.
Matrix volumes were determined as described in Materials and methods
section. Mitochondria (4 mg/ml) were incubated in the presence of glutamate
and malate as respiratory substrates (6 mM) and Pi (5 mM). The values
presented are from at least 3 different experiments carried out with 3 different
mitochondrial preparations. Results are expressed S.D.
linked to water entry, in order to compensate the increase
in cytosolic osmolarity. However, this water entry is not
sufficient to compensate the hyperosmolarity induced by
amino acids. Moreover, we measured the ratio between
amino acid concentration in the mitochondria and in the
cytosol (Table 2) which shows that this ratio decreases to a
great extent in the presence of sodium co-transported amino
acids, which indicates cytosolic hyperosmolarity. Under
steady state, matrix volume does not significantly vary
whatever the cytosolic hyperosmolarity (i.e. whatever the
ratio between amino acid concentration in the cytosol and
in the mitochondria). Myxothiazol-sensitive respiratory rate
increases in the presence of metabolisable amino acid,
owing to a decrease in NADHINAD+ ratio. This increase in
respiratory rate is also related to a slight decrease in
mitochondrial membrane potential. If one considers the
overall thermodynamic force over the electron transport
chain, there is a good correlation between the increase in
respiratory rate and the increase in the overall thermo-
dynamic driving force over the electron transport chain
(see Chapter 3 in this volume). It appears that amino acids
stimulate oxidative phosphorylation because they are sub-
strates of intermediary and energetic metabolisms. Indeed,
when the amino acid under consideration is not metabolisable
(AlB) or when its metabolism is inhibited (alanine + AOA),
neither respiratory rate, NADH/NAD+ ratio nor mito-
chondrial membrane electrical potential difference vary
(Table 2). So, cytosolic hyperosmolarity, per se, does not
affect mitochondrial volume and energetics.
 113

These results concerning the energetics of hyperosmolarity external hyperosmolarity induces a decrease in state 3
do not correlate with what was previously shown in isolated respiratory rate and an increase in membrane electrical
mitochondria incubated in either sucrose or NaCI hyper- potential difference. In NaCI medium, state 3 respiratory
osmotic medium [21, 22]. Indeed, in sucrose medium, rate is slightly decreased by hyperosmolarity. To understand
the discrepancies between our results and those previously
i 120 described regarding isolated rat liver mitochondria, we
1 a checked mitochondrial energetic parameters as a function
e
CI.
100 of external osmolarity in either sucrose or KCI medium.
DIl When external osmolarity was increased in sucrose
~ 80 medium, matrix volume decreased (Fig. 7), this could
] indicate osmometric mitochondrial behaviour. Moreover,
0 60 state 3 respiratory rate decreased while both protonmotive
1 force and NAD(P)H level increased (not shown but see [49]).
f 40 It has previously been shown that when external osmolarity
t' is increased, the ADP/O ratio tends to decrease together with
t
"S.
20 the respiratory flux [22]. There is an inhibitory effect of an
increase in osmolarity on overall oxidative phosphorylation
t 0 efficiency: the ATP/O ratio decreases to almost half of the
50 100 150 200 250 control value (225 mosM) at 500 mosM sucrose [49]. So,
JATP is more markedly affected by hyperosmolarity than J0 2
Kel osmolarity (mosM) even ifboth fluxes are decreased. These facts indicate that in
50 state 3 and in hyperosmotic sucrose medium, phosphorylation
b processes became more controlling. We measured the flux
control coefficient by adenine nucleotide carrier in either
iso- or hyperosmolarity on both J0 2 and JATP' This deter-
mination was done by the inhibitor titration method [50-52]
using carboxyatractylate, a quasi-irreversible inhibitor of
adenine nucleotide carrier [53]. Indeed, when an irreversible
inhibitor is used, the flux control coefficient can be cal-
culated directly from an inhibition curve:
dJ/J
Cj = dI 11m , where 1m is the amount of inhibitor required for
total inhibition of the enzyme.
In isoosmolarity (225 mosM sucrose) as previously
30
100 150 200 shown, adenine nucleotide carrier exerts nearly the same
250
Kel osmolarity (mosM)
Fig. 5. (a) State 3 (.) and 4 (D) respiratory rates as a function of external
180
osmolarity in KCI medium. Respiratory rates were determined as described
c in Materials and methods section. The values presented are from at least 3
170 different experiments carried out with 3 different mitochondrial preparations.
Results are expressed S.E.M. (b) State 3 NAD(P)H level evolution as a
function of external osmolarity in KCI medium.NAD(P)H level was

--
160
determined as described in Materials and methods section. Mitochondria
> (0.1 mg/ml) were firstly incubated in the absence of respiratory substrates
S 150 in order to determine the endogenous NAD(P)H level. Then, 6 mM glutamate
.t and malate were added. State 3 NAD(P)H level was measured in the presence
140 of 0.25 mM ADP. The values presented are from at least 3 different
experiments carried out with three different mitochondrial preparations.
Results are expressed SEM. (c) State 3 ,1.p evolution as a function of
130 external osmolarity in KCl medium. ,1.p was determined as described in
Materials and methods section. Mitochondria (4 mg/ml) were incubated in
120 the presence of 6 mM respiratory substrates and 5 mM Pi. State 3 was
induced by the add of 1 mM ADP. The values presented are from at least 3
100 150 200 250 different experiments carried out with 3 different mitochondrial preparations.
Kel osmolarity (mosM) Results are expressed S. D.
114

120 energetic parameters are unchanged. So we studied the effect

of external hyperosmolarity in KCI medium on isolated rat
225mosMKCI
iii 100 1 liver mitochondria energetics.
..
c:
'ii
fIII.
T 150 mosMKCI To assess an actual steady state of oxygen consumption
under the different osmotic conditions in KCI medium, we
80
cr. 100 mosMKCI checked the time dependence of respiratory rates. Figure
.....e 8a shows that in KCI medium in isoosmolarity, state 3
c:
's
.. respiratory rate did not significantly vary during 10 min of
60
.....
a mitochondrial incubation. However, in hyperosmolarity,
1'1
.5 state 3 respiratory rate increased for 4 min of mitochondrial
N 40
a incubation, reaching the steady state rate equal to the state
~

3 respiratory rate in isoosmolarity. This time dependence

20
150mosMKCI
~SM KCI I I-Qt &5 mos~ KC
o salt used, and considering that the respiratory rates were in
160 180 200 220
t:.p (mV)
Fig. 6, JO/.1.p as a function of external osmolarity in KCI medium. State 3
(.) and 4 (D). Values used for this representation are from Figs 4 and 5b.
control (almost 0.5) on J0 2 and JATP [54]. From iso- to
hyperosmolarity, adenine nucleotide carrier control on J02
decreased (0.34 0.09), although on JATP it increased (0.97
0.14) when J0 2 decreased. The main interpretation is that
when external osmolarity increases, the phosphorylating
respiration moves nearer to state 4 respiration, as described
previously for external temperature decrease [55]. Three
experimental results sustain this interpretation: (i) kinetic
constraints which increase on J ATP when the osmolarity
increases from 225-400 mosM; (ii) a decrease in ATP/O
ratio and J0 2 when external osmolarity increases; (iii) state
3 respiration and ATP/O titrations with carboxyatractylate
in either iso- or hypoosmolarity show that a unique relation-
ship between J0 2 andATP/O exists whatever the osmolarity
[49]. Moreover, when cytosolic osmolarity is increased by
the means of amino acids, matrix volume and mitochondrial
of respiratory fluxes was not observed in sucrose medium
(not shown). To check if time dependence was linked to the
steady state in sucrose medium, we measured respiratory
rates in both sodium chloride and potassium glucuronate
medium. Glucuronate is a non-permeant anion and sodium
movements across the inner mitochondrial membrane occur
differently from those of potassium [56]. Comparison
between Figs 8a and 8b shows that whatever the anion
considered, in potassium medium state 3 respiratory rate is
time-dependent in hyperosmolarity and has the same steady
state value whatever the osmolarity considered. Figure 8c
shows that this time course response of state 3 respiratory
rate does not exist in sodium chloride medium. Indeed, in
this medium, state 3 respiratory rate in both iso- and hyper-
osmotic medium decreases as a function of time and state
3 respiratory rate in hyperosmotic medium is lower than in
isoosmotic medium. So it appears that in sucrose medium,
in which respiratory rates are not time-dependent, there is a
simple and instantaneous osmotic response of flux to
external osmolarity. In hyperosmotic ionic medium, flux
response to external osmolarity is dependent on the cation
considered: its restoration is complete in K+-salt media
while in NaCI medium, flux response to external osmolarity
seems to be more swift (Fig. 8c).

Table 2. Influence of cytosolic hyperosmolarity induced by sodium co-transported amino-acids on cellular and matricial volume and on mitochondrial
energetic parameters

Additions Cellular volume Matricial volume Mitochondrial NADHlNAD+ J02 myxo sensitive [a.a.]mito
(~g.mg-l cell (~I.mg-lcell membrane.1.'I' (x 10-3) (natO.min-1mg-1cell [a.a.]cyto
proteins) proteins) (mY) proteins)

Octanoate, control 2.75 O.l5 0.35 0.03 1604 31.62.3 36.S 1.2 2.%
alanine 3.36O.l2 0.34 0.06 l526 19.71.1 43.S1.9 1.42
glutamine 3.11 O.ll 0.330.04 l564 23.72.1 42.S3.7 1.45
glutamine 3.47 0.09 0.35 0.05 1505 IS.71.2 46.03.9 1.4
+Ieucine
AIB 2.S90.05 0.33 0.05 l665 31.1 2.2 36.63.0 1.12
alanine + AOA 3.44O.l5 0.330.07 1626 30.1 1.4 35.6 1.7 n.d.

Cell substrate was 4 mM octanoate. Amino-acids were added at a concentration of 10 mM except leucine which was added at a concentration of2.5 mM.
Hepatocytes were incubated in closed vials during 25 min for steady state attainement. AIB - aminoisobutyrate.

1.00 - r - - - - - - - - - - - - - - - - - - ,
0.75
0.50
0.25
0.00 +-----r-----r-----r----~
200 300 400
external osmolarity (mosM)
Fig. 7. Matrix volume as a function of external osmolarity in sucrose medium.
Mitochondria were incubated for 4 min in the presence of 6 mM glutamate
and malate and 5 mM Pi. Matrix volumes were determined as described in
Materials and methods section. The values presented are from at least 3
different experiments carried out with 3 different mitochondrial preparations.
Results are expressed S.D.
Matrix volume time dependence was monitored by light
scattering. Indeed, it is well known that light scattering
decreases when matrix volume increases and reciprocally.
But such a method is not a quantitative one and only makes
it possible to assess matrix volume evolution i.e. an increase
or a decrease in matrix volume. Figure 9a shows that under
isoosmotic conditions (A), matrix volume slightly increases
in the presence of respiratory substrates, and that this
increase is stimulated by Pi addition. ADP addition induces
a slight shrinkage of mitochondrial matrix (not shown).
Under hyperosmotic conditions (B), matrix volume is less
than in isoosmotic medium in the absence of respiratory
substrates, indicating a primary shrinkage of mitochondrial
matrix. Glutamate/malate addition induces an increase in
matrix volume which is highly stimulated when external Pi
is added. Again, ADP addition induces a shrinkage ofmito-
chondrial matrix, which is greater compared to that in
isoosmotic medium (not shown). Furthermore, final light
scattering in the presence of substrates and Pi is the same
between iso-and hyperosmolarity. It should be stressed that
whatever the osmolarity, mitochondrial volume varies for
4 min of mitochondrial incubation after substrate addition.
These experimental results raise the question of how the
matrix volume increases in both iso- and hyperosmotic KCI
media. Indeed, in both, this swelling is necessarily linked to
the entry of osmotic active substances which could be K-Pi
under our experimental conditions. Firstly, Fig. 9b shows that
this swelling does not occur in the absence of respiratory
substrates. It is strongly inhibited in the presence of uncoupler
115
(not shown), so it is an energy-dependent phenomenon.
Moreover, when Pi is replaced by either acetate or thio-
cyanate, the swelling is partly decreased in acetate medium
and drastically decreased in thiocyanate medium (not
shown). Therefore, it is greatly dependent on anion entry
by an electroneutral mechanism depending on .1pH. To
discriminate the way K+ enters the mitochondrial matrix, we
studied the influence of known inhibitors of a part of the
mitochondrial K+ cycle on this swelling. Neither MgC1 2
(inhibitor of K+/W exchanger [24]), Pindolol (inhibitor of
K+/W exchanger [57]) nor glybenclamide (inhibitor ofKATP
channels [58, 59]) had any effect on this swelling. Quinine [60]
aspecifically inhibits the respiratory chain for concentrations
inhibiting K+/H+ exchanger, so it could not be used under our
experimental conditions. As there was no clear correlation
between matrix volume and respiratory rate evolution as a
function of time, it was of interest to estimate inner mito-
chondrial membrane electrical potential difference as a
function of time. Under state 3 respiration, (i) in isoosmotic
medium (Fig. lOa), respiratory rate and .1", were not sig-
nificantly time-dependent; (ii) in hyperosmotic medium
(Fig. lOb), in the first 4 min, the increase in J0 2 was related
to a decrease in .1",. Time courses of either respiratory rate,
matrix volume or .1", all show that in order to study the
steady state on isolated mitochondria incubated in KCI
medium, 4 min of mitochondrial incubation are necessary.
So all the following experiments were done after 4 min of
mitochondrial incubation.
Under steady state, in KCI medium, the lack of volume
variations from iso- to hyperosmolarity seemed to indicate
osmotically active substance movements across the inner
mitochondrial membrane which make it possible to equilibrate
osmolarity changes (Table 3). From iso- to hyperosmolarity,
state 3 respiratory rate, NAD(P)H level, ATP/O ratio and
protonmotive force were in steady state (Table 3). It appears
that when isolated rat liver mitochondria were incubated
in hyperosmotic KCI medium, potassium movements
occurring in the first minutes of mitochondrial incubation,
make it possible to equilibrate matrix volume, flux and force.
These results are in agreement with those obtained on
isolated hepatocytes for which cytosolic hyperosmolarity
per se has no influence on the energetic status of the cell.
Discussion
It is now widely accepted that under physiological conditions
i.e. massive absorption of water and/or post-prandial period,
the liver is submitted to volume variations. This can induce
hepatocyte volume variations and consequently cytosolic
osmolarity variations, which can modify matrix volume and
mitochondrial energetics. So the influence of either hypo-
or hyperosmolarity on mitochondrial energetic parameters
116

120 has been studied on both isolated hepatocytes and isolated

rat liver mitochondria in either sucrose or KCl medium in
...... Cii'loo
II) c:
...
1\1 'ijj
order to have a better understanding of this phenomenon.
In most of the conditions studied, isolated rat liver mito-
~e 80 chondri a behave as an osmometer and changes in matricial
......o
a.
1\1
C) volume can modify mitochondrial metabolism [16-19, 61,
E
'5. ....... 60 62]. However, numerous anionic and cationic channels or
...~ .E 40 exchangers have been described as participating in matrix
("/') volume regulation [24]. Figure 11 shows that in sucrose
... 0
.......

... ...
II)

-
medium, there is a linear relationship between matrix volume
1\1
1\1 20 and 1Iosmolarity from iso- to hyperosmolarity, while in
III c:
hypoosmotic medium, matrix volume is less than expected
0 by this linear relationship. This indicates the activity of the
0 2 4 6 8 10 K+/H+ exchanger as previously proposed, in such conditions,
time (minutes) (i.e. increase in matrix volume inducing a decrease in
intramatricial [Mg2+]). This can induce an extrusion of
potassium associated with an intrusion of protons [24]. In
hypoosmotic KCl medium, again the increase in matrix
volume is less than expected for an actual osmometer and
120 is increased in the presence of antimycin A and oligomycin
b

-...
(see Fig. 9). This also is in favor of a functioning K+/W
100 carrier in hypoosmotic medium. Moreover, in KCl medium,
......
II)

1\1
III
c:
'ijj
80
the situation is different compared to sucrose medium, since
there is no relationship between matrix volume and 11
~ ...a.0
......
0
1\1 C) 60
osmolarity, and this is linked to the regulation of matrix
volume in KCl hyperosmotic medium. This points to the
E
'5. .......
c:
importance of cations and anions in the regulation of matrix
....~ 40 volume during osmotic changes.
("/') ~a
...... ...
II)
20
Preliminary studies on mitochondrial swelling in hypo-

-
1\1 1\1 osmotic sucrose medium or after rat hormonal pretreatment
c:
III (see [16] for review, [17]) have shown that an increase in
0
matrix volume induces an increase in the state 3 oxygen
0 2 4 6 8 10
consumption rate of isolated rat liver mitochondria. More-
time (minutes) over, measurement of the redox state of flavoproteins and
ubiquinones shows that the flavoprotein reduction state
diminishes when the matrix volume increases [62]. The
interpretation of these experimental results is that a respiratory
chain activation between flavoproteins and ubiquinones is
120 linked to the increase in matrix volume [62]. In our study, in

_100
.......
II) III
c:
1\1
...
'ijj
80
Fig. 8. Time dependence of respiratory rates in various iso- or hyperosmotic
medium. (a) Time course of state 3 respiratory rate in iso- 225 mosM (0) or
~ ...a.0 ~
...
0
C)
60
hyperosmotic 400 mosM (A) KCl medium. State 3 respiratory rates were
measured in the presence of 0.5 mM ADP. Respiratory rates were determined
....
1\1
E as described in Materials and methods section. The values presented are
'5. .......
III . from at least 3 different experiments carried out with 3 different mitochondrial
....
II)
E 40 preparations. Results are expressed S.E.M. (b) Time course of state 3
("/') .......
...... 0...
II)
respiratory rate in iso-225 mosM (0) or hyperosmotic 400 mosM (A)

-
1\1 1\1 20 potassium glucuronate medium. Respiratory rates were determined as
c: described in Materials and methods section. The experiment is representative
III
of3 such experiments. (c) Time course of state 3 respiratory rate in iso- (0)
0
or hyperosmotic (A) NaCI medium. Respiratory rates were determined as
0 2 4 6 8 10 described in Materials and methods section. The experiment is representative
time (minutes) of 3 such experiments.
 117

(a)
light
scattering
(10%) L
1 minute
200
l-
I + 4 a 100
"iii'
c:
'iii
...c.
~
,... CI
> E
E 150 ......
B
....
<I
50 c:
'E
......
9
0
c
~ 2
OJ

~
(9
c:
.......
A N
0...,
100 I I 0
0 2 4 6 8 10 12
time (minutes)
(b)

-
4OOmosM

200~--------------------~

b 100
225mosM

l/_mOS_M_ _ _ __
....~
<I
150

Fig. 9. Time course of matrix volume in KCl medium. (a) Matrix volume
evolution in (A) isoosmotic KCl medium (B) hyperosmotic medium. Matrix
volume evolution was measured by light scattering as described in
Materials and methods section. This experiment is representative of 3
such experiments. (b) Matrix volume evolution in iso (225 mOsM), hypo
(100 mOsM) and hyperosmotic (400 mOsM) KCI medium. Matrix volume
evolution was measured by light scattering as described in Materials and
methods section but respiratory substrates were not added to mitochondrial
buffer. This experiment is representative 00 such experiments.
KCI medium, under steady state hypoosmotic incubation
conditions, state 3 respiratory rate decreased during the first
minutes of mitochondrial incubation while the rhodamine
signal did not significantly vary. Indeed, under hypoosmolarity,
state 3 respiratory rate, NAD(P)H level and protonmotive
force decreased and oxidative phosphorylation efficiency did
not significantly vary. It should be stressed that NAD(P)H level
also decreased in hypoosmolarity, but our measurement was
not a quantitative one and did not allow us to check the redox
span between NADHINAD+ and O/HP couples as in cells.
So the main influence of hypoosmolarity in such an ionic
medium on oxidative phosphorylation is a kinetic control
upstream the respiratory chain. The present results obtained
on isolated rat liver mitochondria incubated in KCl medium
1004-~--~--~~~~~~~~0
o 2 4 6 8 10 12
time (minutes)
Fig. 10 Time course of state 3 respiratory rate ("') and d'l' (D) in: (a)
isoosmotic KCl medium. Respiratory rate and d'l' were determined as
described in Materials and methods section. The values presented are from
at least 3 different experiments carried out with 3 different mitochondrial
preparations. Results are expressed S.E.M. (b) hyperosmotic KCI medium.
Respiratory rate and d'l' were determined as described in Materials and
methods section. The values presented are from at least 3 different
experiments carried out with 3 different mitochondrial preparations. Results
are expressed S.D.
are in agreement with those obtained in hepatocytes [46].
Indeed, in isolated hepatocytes incubated in hypoosmotic
Na+ depleted media, a large increase in the mitochondrial
volume was not directly involved in the activation of
respiration. Moreover, results of the quantification of the
various bioenergetic parameters point to an inhibition of
respiration. Indeed, the same respiratory rate is obtained
in hypoosmolar incubation media for a higher overall
thermodynamic driving force over the electron transport
118
Table 3. External KCl hyperosmolarity influence on isolated rat liver mitochondria energetic parameters
JO, Matrix volume
(natO/min/mg prj (Ill/mg prj
Isoosmolarity 108 13 0.78 0.18
KCl (225 mosM)
Hyperosmolarity 122 17 0.84 0.15
KCl (500 mosM)
Respiratory substrates were glutamate and malate (6 mM). Mitochondrial energetic parameters were determined as described in Materials and methods
section. Values are means S.D. of at least 3 such experiments.
chain. Thennodynamically, the increase in the absolute value
of the term 2.-1E' h -1 OLnjl could be associated with an
increase in the respiration rate. However, this was not so
and considering our results obtained on both isolated
hepatocytes and isolated mitochondria, this could be
explained by the fact that an increase in mitochondrial
volume induces a modification in the response of the
respiratory chain to its associated forces. In the literature,
an increase in matrix volume in isolated rat liver mito-
chondria is often associated with an activation of the
respiratory chain. We confinned that state 3 respiratory rate
is stimulated in hypoosmotic sucrose medium [49] but its
two associated forces i.e. NAD(P)H level and .-1p decrease
when external osmolarity is decreased. This does not
confirm the hypothesis of an activation of the respiratory
chain. Moreover, the hypoosmotic KCl medium influence
on oxidative phosphorylation is related to the energetics
of hypoosmolarity in isolated hepatocytes. So, the main
2.---------------------------------~
o+-~_.~~._~_.--~~~~--~~~~
o 2 4 6 8 10 12 14
l/osmolarity (osM-l)
Fig. 11. Matrix volume evolution as a function of 1Iosmolarity in either
KCl or sucrose medium under state 4 (0, 0) and 3 (eo .) respectively.
Mitochondria were incubated for 4 min in the presence of 6 mM glutamate
and 6 mM malate and 5 mM Pi. Matrix volumes were determined as described
in Materials and methods section. The values presented are from at least 3
different experiments carried out with 3 different mitochondrial preparations.
i\p (mV) NAD(P)H level ATP/O(nmol
(%i\F/F) ATP/nat 0)
176 6 48 6 2.67 0.17
178 8 49 8 2.44 0.26
influence of hypoosmolarity on the overall oxidative phos-
phorylation pathway is an increase in kinetic constraints at
the level of the respiratory chain, limiting the increase in
electron flux in response to the decrease in .-1\jf.
To study the influence of hyperosmolarity on isolated
mitochondria, two distinct media were used in this work: a
sucrose medium and a KCI medium. In sucrose medium,
isolated rat liver mitochondria behave like an osmometer and
matrix volume decreases when external osmolarity is
increased. State 3 respiratory rate andATP synthesis flux also
decrease when external osmolarity is increased but not to
the same extent leading to a decrease in oxidative phos-
phorylation efficiency. NAD(P)H level and.-1p are increased
when external osmolarity is increased. Furthermore, when
external osmolarity increases, adenine nucleotide carrier
control on J0 2 decreases, and phosphorylation and res-
piratory flux titration with carboxyatractyloside give an
unique relationship between ATP/O ratio and J0 2 , whatever
the osmolarity. These results show that the main phenomenon
observed when external osmolarity increases is an inhibition
of ATP synthesis, and consequently an increase in both
protonmotive force and NAD(P)H level linked to a decrease
in state 3 oxygen consumption rate. However these results
do not exclude the sensitivity of the respiratory chain to
osmolarity, which has previously been shown [20]. On
isolated rat liver mitochondria, it has previously been shown
that when external osmolarity increases, there are kinetic
constraints applied to respiratory chain proton pumps [20].
Indeed, after segmental analysis of the electron transport
chain, the authors concluded that there is an osmotically
sensitive diffusion of quinones through the bilayer, con-
sidering that phosphorylation, per se, is not sufficient to
account for the osmotic sensitivity of respiration. However,
these kinetic constraints on respiratory chain do not playa
great part in phosphorylation conditions, in which a decrease
inATP/O ratio and an increase in adenine nucleotide carrier
control on JATP are related to an increase in.-1p and NAD(P)H
level when external osmolarity increases. Thus, our thenno-
dynamic and kinetic study of the effect of external sucrose
osmolarity on bothATP synthesis and respiratory flux points
to the major role of kinetic constraints on .-1p consuming

systems, particularly the adenine nucleotide carrier: when
external osmolarity increases, state 3 moves nearer to state
4 respiratory rate and oxidative phosphorylation efficiency
decreases.
These results cannot be connected to those obtained on
isolated hepatocytes. Indeed, when cytosolic osmolarity is
increased by the means of sodium co-transported amino
acids, energetic modifications are linked only to the
metabolism of the amino acid under consideration. If one
considers non metabolisable amino acids or those of which
the metabolism is inhibited, neither respiratory rate nor
protonmotive force are modified. Moreover, amino acid
accumulation does not lead to a decrease in matrix volume
which could be expected for a real osmometer. Furthermore,
when cytosolic osmolarity is increased by means of external
hyperosmolarity (i.e. 100 mM sucrose in Krebs-Henseleit
buffer), cell volume decreases from l.37 0.11 Ill/mg dry
wt (control) to 0.99 0.16 Ill/mg dry wt (100 mM sucrose)
while mitochondrial volume is not affected (0.25 0.03 Ill/
mg dry wt (control) and 0.22 0.03 Ill/mg dry wt (100 mM
sucrose)). This points to the importance of matrix volume
regulation in isolated hepatocytes. To investigate further the
influence of hyperosmolarity on oxidative phosphorylation,
we used a KCl medium in order to check whether oxidative
phosphorylation response to cytosolic hyperosmolarity on
isolated hepatocytes was really distinct from that in isolated
mitochondria, or if this was due to regulation of matrix
volume and thermodynamic parameters via numerous
anionic and cationic channels or exchangers which have been
described as occurring during matrix volume regulation.
In KCl medium, matrix volume in hyperosmolarity is
efficiently regulated in such a way that under steady state, it
is completely recovered in comparison to its value in
isoosmolarity. Furthermore, the mechanisms of matrix
volume regulation depend on both the energetic status of
isolated rat liver mitochondria and the presence of Pi (see
Fig. 9). The increase in matrix volume is probably linked to
Pi-K salt accumulation in the mitochondrial matrix. In such
a hypothesis, the fact that ADP addition induces a shrinkage
is probably linked to a shift in free Pi in phosphorylating
conditions. Moreover, the latter induces a slight decrease
in ~pH which could modify the ratio between external and
internal Pi. Potassium fluxes under our experimental
conditions have never been inhibited by the use of known
inhibitors of a part of the mitochondrial potassium cycle.
The main interpretation is that potassium enters the
mitochondrial matrix via both potassium leak and potassium
channels (see Fig. I). Taken together, these results indicate
that the matrix volume regulation is essentially due to an
entry ofK-Pi. In addition to matrix volume regulation, state
3 respiratory flux varies in hyperosmotic KCl medium during
4 min of isolated mitochondria incubation. Firstly, under
hyperosmotic KCI incubation conditions, respiratory rate
119
increases in the first 4 min. This increase is observed only
in hyperosmotic potassium salts medium (see Fig. 8).
Moreover, under steady state, state 3 respiratory rate in
NaCI hyperosmotic medium is not regulated with the same
efficiency as in hyperosmotic KCl medium while matrix
volume is the same in NaCI medium compared to KCI
medium (not shown). Related to the increase in respiratory
rate in KCl medium, the membrane electrical potential
difference decreases in hyperosmotic KCl medium until 4
min of mitochondrial incubation. So in hyperosmotic KCl
medium and during the first minutes of mitochondrial
incubation, it can be hypothesized that there is a potassium
energy-dependent entry which decreases 110/ and con-
sequently stimulates state 3 respiratory rate. So it appears
that in hyperosmotic conditions, matrix volume flux and
forces regulation need the input of potassium salts. Mean-
while, matrix volume recovery is not dependent on the cation
under consideration because sodium is also suitable for this
restoration of matrix volume in hyperosmotic medium.
When oxidative phosphorylation activity is concerned, it
appears that potassium,per se, is necessary for both flux and
force regulation. Indeed, in NaCI medium, state 3 respiratory
rate andADP/O ratio decrease in hyperosmolarity, while we
show in this study that oxidative phosphorylation efficiency
is not affected by KCI hyperosmotic medium. It has been
proposed that the respiratory rate in phosphorylation state
strongly depends on matrix volume [16-22]. In this work, we
show that matrix volume recovery is necessary but not
sufficient for respiratory rate and protonmotive force
maintenance. So, matrix volume intervenes in the maintenance
of oxidative phosphorylation activity even ifthere is no direct
relationship between matrix volume and flux regulation.
Conclusion
This work was initiated in order to understand the dis-
crepancies observed between the influence of cytosolic
hypo- or hyperosmolarity on mitochondria in situ and what
was known on isolated rat liver mitochondria behaviour in
hypo- or hyperosmotic sucrose medium. In this paper, we
report that on isolated hepatocytes incubated in hypoosmotic
media, a large increase in mitochondrial volume is not
directly involved in the activation of respiration. Indeed,
the lower the external osmolarity, the higher the overall
thermodynamic driving force for the same respiratory rate.
This points more to an inhibition of respiration at the level
of the electron transport chain rather than to a stimulation,
as previously proposed on isolated rat liver mitochondria.
Results obtained on isolated rat liver mitochondria incubated
in KCl medium are in agreement with those obtained on
hepatocytes. Indeed, under hypoosmolarity and when steady
state is reached, state 3 respiratory rate, NAD(P)H level and
120
protonmotive force decrease while oxidative phoshorylation
efficiency does not vary. This shows that the main influence
of hypoosmolarity in such a ionic medium on oxidative
phosphorylation is a kinetic control upstream and on the
respiratory chain.
Furthermore, studies on isolated hepatocytes in con-
ditions of cytosolic hyperosmolarity (i.e. sodium co-
transported amino acids) have shown that neither J02 , matrix
volume nor mitochondrial membrane electrical potential
vary when amino acids are not metabolized. This is clearly
in accordance with the results obtained on isolated rat liver
mitochondria incubated in hyperosmotic KCl medium.
Indeed, in this medium, when steady state is reached, under
state 3, neither respiratory rate, NAD(P)H level, oxidative
phosphorylation efficiency nor protonmotive force vary.
This is linked to flux, force and matrix volume regulation
on isolated rat liver mitochondria incubated in KCl hyper-
osmotic medium. Indeed, in this medium, isolated rat liver
mitochondria matrix volume and oxidative phosphorylation
activity tends to be regulated as in cells.
Acknowledgements
The authors wish to thank Dr. R. Cooke for his contribution
to the editing of the manuscript and Pro X. Leverve for
stimulating discussion. This work was supported by grants
from the Pole Medicament d' Aquitaine.
References
1. Haussinger D, Lang F: Cell volume in the regulation of hepatic function:
A mechanism for metabolic control. Biochim Biophys Acta 1071: 331-
350,1991
2. Hallbrucker C, Vom Dahl S, Lang F, Haussinger D: Control of hepatic
proteolysis by amino acids. The role of cell volume. Eur 1 Biochem 197:
717-724,1991
3. BacquetA, Hue L, MeijerAl, Van Woerkom GM, Plomb PIAM: Swelling
of rat hepatocytes stimulates glycogen synthesis. 1 Bioi Chern 265:
955-959, 1990
4. Graf 1, Haddad P, Haussinger D, Lang F: Cell volume regulation in liver.
Renal Physiol Biochem II: 202-220, 1988
5. Haussinger D, Stehle T, Lang F: Volume regulation in liver: Further
characterization by inhibitors and ionic substitution. Hepatology II:
243--254, 1990
6. Lang F, Stehle T, Haussinger D: Water, K+, W, lactate and glucose
fluxes during cell volume regulation in perfused rat liver. Pfliiggers
Arch 413: 209--216,1989
7. Haussinger D, Halbrucker C, Vom Dahl S, Lang F, Gerok W: Cell
swelling inhibits proteolysis in perfused rat liver. Biochem. 1272:
239--242,1990
8. Haussinger D, Halbrucker C, Vom Dahl S, Decker S, Schweizer V, Lang
F, Gerok W: Cell volume is a major determinant of proteolysis control
in liver. FEBS Lett 283: 70-72, 1991
9. Haussinger D, Lang F1: Exposure of perfused liver to hypotonic conditions
modifY cellular nitrogen metabolism. Cell Biochem 43: 355-361, 1990
10. Hallbrucker C, vom Dahl S, Lang F, Haussinger D: Interactions between
cell volume and hepatic nitrogen metabolism. Cont Nephrol 92: 175-
181,1991
II. Bode B, Kilberg M: Amino acid dependent increase in hepatic system
N activity is linked to cell swelling. 1 Bioi Chern 266: 7376-7381, 1991
12. Haussinger D, Lang F: The mutual interaction between cell volume
and cell function: A new principle of metabolic regulation. Biochem
Cell Bioi 43: 1-4,1991
13. Meijer Al, Bacquet A, Gustafson L, Van Woerkom GM, Hue L:
Mechanism of activation of liver glycogen synthase by swelling. 1
Bioi Chern 267: 5823--5828, 1992
14. BacquetA, Gaussin V, Bollen M, Stalmaus W, Hue L: Mechanism of
activation ofliver acetyl-CoA carboxylase by cell swelling. Eur 1 Biochem
217: 1083--1089,1993
15. Wilkel H, Busch GL, Haussinger D, Lang F: Alkalinisation of acidic cellular
compartments following cell swelling. FEBS Lett 338: 27-30, 1994
16. Halestrap PA: The regulation of the matrix volume of mammalian
mitochondria in vivo and in vitro and its role in the control of
mitochondrial metabolism. Biochim Biophys Acta 973: 355-382, 1989
17. Halestrap PA, Davidson MA, Potter WD: Mechanisms involved in the
hormonal regulation of mitochondria I function through changes in the
matrix volume. Biochim Biophys Acta 1018: 278--281, 1990
18. Armston EA, Halestrap PA, Scott DR: The nature of the changes in
liver mitochondrial function induced by glucagon treatment of rats.
The effects of intramitochondrial volume, aging and benzyl alcohol.
Biochim Biophys Acta 681: 429-439, 1982
19. Otto AD, Ontko AI: Structure-function relations between fatty acids
oxidation and the mitochondrial inner membrane-matrix region. Eur 1
Biochem 129: 479-485,1982
20. Mathai Cl, Sauna EZ, lohn 0, Sitaramarn V: Rate limiting step in electron
transport. Osmotically sensitive diffusion of quinones through voids
in the bilayer. 1 Bioi Chern 268: 15442-15454, 1993
21. Nicholls DG, Lindberg 0: Inhibited respiration and ATPase activity of
rat liver mitochondria under conditions of matrix condensation. FEBS
Lett 25: 61--64,1972
22. Sitaramam V, Rao NM: Oxidative phosphorylation in rat liver mito-
chondria: Influence of physical parameters. Ind 1 Biochem Biophys 28:
401-407,1991
23. Garlid KD: Mitochondrial cation transport: A progress report. 1 Bioenerg
Biomembr26: 537-542,1994
24. Garlid KD: On the mechanism of regulation of the mitochondrial K+/W
exchanger. 1 Bioi Chern 255: 11273-11279, 1980
25. Gunter TE: Transport of calcium by mitochondria. 1 Bioenerg
Biomembr 26: 465--469, 1994
26. Berry NM, Friend DS: High-yield preparation of isolated rat liver
parenchymal cells: A biochemical and fine structural study. 1 Cell Bioi
43: 506-520, 1969
27. Groen AK, Sips Hl, Vervoorn RC, Tager 1M: Intracellular com-
partrnentation and control of alanine metabolism in rat liver parenchymal
cells. Eur 1 Biochem 122: 87-93,1982
28. Krebs HA, Heiseleit K: Urea formation in the animal body. Hoppe-
Seyler's Z Physiol Chern 21 0: 33--66, 1932
29. Quilan PT, Thomas AP, Armston AE, Halestrap AP: Measurement of the
intrarnitochondrial volume in hepatocytes without cell disruption and its
elevation by hormones and valinomycin. Biochem 1 214: 395-404, 1983
30. Nobes CD, Brand MD: A quantitative assessment of the use of36CI-
distribution to measure plasma membrane potential in isolated
hepatocytes. Biochim Biophys Acta 987: 115-123, 1989
31. Nobes CD, Hay WW, Brand MD: The mechanism of stimulation of
respiration by fatty acids in isolated hepatocytes. 1 Bioi Chern 265:
12910-12915,1990

32. Scott ID, Nicholls DG: Energy transduction in intact synaptosomes.
BiochemJ 186: 21-33,1980
33. Berry MN, Gregory RB, GrivellAR, Henly DC, Nobes CD, Phillips JW,
Wallace PC: Intracellular mitochondrial membrane potential as an
indicator of hepatocytes energy metabolism: Further evidence for
thermodynamic control of metabolism. Biochim Biophys Acta 936: 294-
306,1988
34. Hoek JB, Nicholls DG, Williamson JR: Determination of the mitochondrial
protonmotive force in isolated hepatocytes. J Bioi Chern 255: 1458-
1464,1980
35. Murphy MP, Brand MD: The control of electron flux through
cytochrome oxidase. Biochem J 243: 499--505, 1987
36. Zuurendonck PF, Tager JM: Rapid separation of particulate components
and soluble cytoplasm of isolated rat liver cells. Biochim Biophys Acta
333: 393--399, 1974
37. Akerboom TPM, Van der Meer R, Tager JM: Techniques for the
investigation of intracellular compartmentation. Tech Metab Res B205:
1-33, 1979
38. Krebs HA, Veech RL: In: S Papa, JMTager, E Qualliariello, EC Slater
(eds). Pyridine Nucleotide Interrelations. Adriatica Editrice, Bari, 1969,
pp 329--382
39. Berenblum I, Chain E: Studies on the colorimetric determination of
phosphate. Biochem J 32: 286-294, 1938
40. Henrickson RL, Meredith SC: Amino acid analysis by reverse phase
high performance liquid chromatography: Precolumn derivatization
with phenylisothiocyanate. Anal Biochem 136: 65-74,1984
41. Cooper C, Lehninger AL: Oxidative phosphorylation by an enzyme
complex from extracts of mitochondria. J Bioi Chern 177: 751-766, 1956
42. Gomall AG, Bardawill CJ, David MM: Determination of serum proteins
by means ofthe biuret reaction. J Bioi Chern 177: 751-766, 1948
43. Rigoulet M, DevinA, Averet N, Vandais B, Guerin B: Mechanisms of
inhibition and uncoupling of respiration in isolated rat liver mito-
chondria by the general anesthetic 2,6-diisopropylphenol. Eur J
Biochem 241: 280-285, 1996
44. Rigoulet M, Guerin B: Phosphate transport andATP-synthesis in yeast
mitochondria: Effect of a new inhibitor: The tribenzyl phosphate. FEBS
Lett 102: 18-22, 1979
45. Rottenberg H.: The measurement of membrane potential and L'.pH in
cells, organelles and vesicles. Meth Enzymol55: 547-569, 1979
46. Espie P, Guerin B, Rigoulet M: On isolated hepatocytes mitochondrial
swelling induced in hypoosmotic medium does not affect the
respiration rate. Biochim Biophys Acta 1230: 139--146, 1995
47. Koretsky A, Balaban RS: Changes in pyridine nucleotides level alter
oxygen consumption and extra-mitochondrial phosphate in isolated
mitochondria: A lip NMR and NAD(P)H fluorescence study. Biochim
121
Biophys Acta 893: 398-408, 1987
48. Emaus RK, Grunwald R, Lemasters JJ: Rhodamine 123 as a probe of
transmembrane potential in isolated rat liver mitochondria: Spectral
and metabolic properties. Biochim Biophys Acta 850: 436-448, 1986
49. Devin A, Guerin B, Rigoulet M: Dependence of flux size and
efficiency of oxidative phosphorylation on external osmolarity
in isolated rat liver mitochondria: Role of adenine nucleotide
carrier. Biochim Biophys Acta 1273: 13-20, 1996
50. Groen KA, Wanders AJR, WesterhoffVH, van der Meer R, Tager MJ:
Quantification of the contribution of various steps to control mito-
chondrial respiration. J Bioi Chern 257: 2754-2757, 1982
51. Gellerich FN, Kunz WS, Bohnensack R: Estimation of flux control
coefficient from inhibitor titration by non-linear regression. FEBS Lett
274: 167-170,1990
52. Groen AK: Quantification of control in studies on intermediary
metabolism. Ph.D. Thesis, Amsterdam, 1984
53. Vignais PV, Vignais M, Defaye G: Adenosine diphosphate translocation
in yeast mitochondria. Nature of the receptor site for carboxy-
atractyloside. Biochemistry 12: 1508-1519, 1973
54. Brown GC: Control of respiration and ATP synthesis in mammalians
mitochondria and cells. Biochem J 284: 1-13, 1992
55. Quentin E,Averet N, Guerin B, Rigoulet M: Temperature dependence
of the coupling efficiency of rat liver oxidative phosphorylation: role
of adenine nucleotide translocator. Biophys Biochem Res Com 202:
816-821,1994
56. Brierley GP, Baysal K, Jung DW: Cation transport systems in
mitochondria: Na+ and K+ uniports and exchangers. J Bioenerg
Biomembr 26: 519-526, 1994
57. Beavis AD: On the inhibition of the mitochondrial inner membrane
anion uniporter by cationic amphiphiles and others drugs. J Bioi Chern
264: 1508-1515,1989
58. Sturgess NC, Ashford MLJ, Cook DL, Hales CN: The sulphonylurae
receptor may be an ATP sensitive potassium channel. Lancet 8453:
474-475,1985
59. Inoue I, Nagase H, Kishi K, Higuti T: ATP-sensitive K+ channel in the
mitochondrial inner membrane. Nature 352: 244-247, 1991
60. Garlid KD, DiResta DJ, Beavis AD, Martin WH: On the mechanism by
which dicyclohexylcarboxydiimide and quinine inhibits K+ transport
in rat liver mitochondria. J Bioi Chern 261: 1529--1535,1986
61. HalestrapAP, Quilan TP, Whipps ED, Armston EA: Regulation ofthe
mitochondrial matrix volume in vivo and in vitro. The role of calcium.
Biochem J 236: 779--787,1986
62. HalestrapAP, Dunlop LJ: Intramitochondrial regulation offatty acids
~-oxidation occurs between flavoproteins and ubiquinone. A role for
changes in matrix volume. Biochem J 239: 559--565, 1986
 PART II

ENERGY TRANSFER NETWORKS: MOLECULAR PHYSIOLOGY

OF KINASES, LESSONS FROM TRANSGENIC MICE,
MATHEMATICAL THEORIES
Molecular and Cellular Biochemistry 184: 125--140, 1998.
1998 Kluwer Academic Publishers.

Functional aspects of the X-ray structure of

mitochondrial creatine kinase: A molecular
physiology approach

Uwe Schlattner, l Michael Forstner, l Michael Eder, l Olaf Stachowiak, 1

Karin Fritz-Wolf2 and Theo Wallimann 1
lSwiss Federal Institute of Technology, Institute of Cell Biology, ETH Zurich, Switzerland; 2Max-Planck Institutfur
medizinische Forschung, Abt. Biophysik, Jahnstr. 29, Heidelberg, Germany

Abstract
Mitochondrial creatine kinase (Mi-CK) is a central enzyme in energy metabolism of tissues with high and fluctuating energy
requirements. In this review, recent progress in the functional and structural characterization of Mi-CK is summarized with
special emphasis on the solved X-ray structure of chicken Mib -CK octamer (Fritz-Wolf et al., Nature 381, 341-345, 1996).
The new results are discussed in a historical context and related to the characteristics of CK isoforms as known from a large
number of biophysical and biochemical studies. Finally, two hypothetical functional aspects ofthe Mi-CK structure are proposed:
(i) putative membrane binding motifs at the top and bottom faces ofthe octamer and (ii) a possible functional role ofthe central
20 Achannel. (Mol Cell Biochem 184: 125-140, 1998)

Key words: active site, catalytic mechanism, creatine kinase, high-energy phosphate transport, membrane binding, metabolic
channeling, mitochondrial energetics, octamer/dimer equilibrium, X-ray structure

Abbreviations: ANT - adenine nucleotide translocator; Cr - creatine; cCr - cyclocreatine; CK - creatine kinase; Mi-CK -
mitochondrial CK isoforms; Mib-CK - sarcomeric Mi-CK isoform; MM-CK, BB-CK - cytosolic CK isoforms; PCr - N-
phosphoryl-creatine; SR - sarcoplasmic-endoplasmic reticulum; TSAC - transition state analogue complex

Introduction enzyme simply fulfilling the role of a temporal ATP buffer

The 'energy-rich' compound, N-phosphoryl creatine (PCr)
was first identified in 1928 in muscle [1] and was initially
thought to represent the direct chemical energy source for
muscle contraction. However, the discovery of the enzyme
ATP:creatine phosphotransferase or creatine kinase (CK,
EC 2.7.3.2) in 1934 by Lohman [2] showed that the N-
phosphoryl group ofPCr is transphosphorylated onto ADP
to regenerate ATP [3], the latter being the universal energy
currency in all living systems [4]. CK was first purified in
1954 by Kuby et al. [5] from rabbit muscle and its properties
were described by Kuby and Noltman [6]. At this time and
for many years to come, CK was considered a strictly soluble
or back-up system, as still depicted in current textbooks of
biochemistry and cell biology.
However, after the discovery of (i) the cytosolic isoforms
ofCK [7], i.e. muscle-type MM-CK and brain-type BB-CK,
(ii) the developmental transition of CK isoforms during
muscle cell differentiation from ubiquitous BB-CK to
sarcomeric muscle-specific MM-CK via the transitory MB-
CK hybrid [8] and (iii) the identification of a novel mito-
chondrial CK isoform (Mi-CK), strictly confined to mito-
chondria [9], a different view of the enzyme's function
emerged. Right after the discovery of Mi-CK, it was
postulated that ATP, formed by oxidative phosphorylation
in muscle, would be transphosphorylated to PCr which then

Address/or offprints: T. Wallimann, Institute of Cell Biology, ETH Ziirich-Honggerberg HPM F42, CH - 8093 Zurich, Switzerland
126
would diffuse to the myofibrils and be converted into ATP
again ([10], p.170 and [11,12]). The proposal that creatine
may stimulate mitochondrial respiration was born soon after
[13] and this concept has been corroborated by numerous
consecutive studies showing, for example, a functional
coupling ofMi-CK with the adenine nucleotide translocator
(ANT, for a review see [14]).
After the discovery of the isoenzyme-specific attachment
of MM-CK to the myofibrillar M-band of sarcomeric
muscle [15, 16], the idea gained support that the ATP-
consuming contractile apparatus was connected to sites of
energy production, e.g. mitochondria, via PCr and the CK
isoforms located at these respective sites. Several models
have since been proposed to describe the complex cellular
energy supply system: The PCr-shuttle [17-23] or PCr-
circuit [24-27] (for a collection of relevant publications and
references see a special issue of Mol Cell BioI [28]). The
idea behind the PCr-circuit model is that the PCr/CK system
not only serves as a temporal energy buffer for ATP [29],
but also as a spatial energy transport system connecting
cellular sites of ATP-consumption, i.e. ATPases, to sites of
energy production, i.e. glycolysis and/or oxidative phospho-
rylation. The latter function is especially important in highly
polar cells such as photoreceptor cells or spermatozoa [30]
where diffusion limitations ofATP andADP can be overcome
by PCr via the CK system [31].
In addition, the PCr-circuit represents a regulatory system
for the physiologically required maintenance and adjustment
of subcellular ATP/ADP ratios. For example, CK associated
with the sarcoplasmic-endoplasmic reticulum (SR) [32] is
thermodynamically important for efficient Ca2+-pump activity,
as shown both in vitro with isolated SR vesicles [33--35], as
well as in situ with skinned muscle fibres [36]. Probably, the
role ofCK in the energetics ofCa2+-homeostasis in excitable
cells [30] represents the most crucial function of the CK
system in general. This has recently become evident from
analysis of the muscle phenotype of transgenic knock-out
mice lacking the expression of both cytosolic MM-CK and
sarcomeric Mib -CK. Skeletal muscles of these mice show
problems with Ca2+-release and uptake and display signifi-
cantly prolonged relaxing times [37].
The temporal buffer function of the CK system, which seems
generally accepted now (see above), has been corroborated
elegantly by a transgenic approach with mice expressing BB-
CK in the liver, where normally no or only very little CK
activity is found. In the perfused livers of such transgenic
animals,ATP levels and pH remain constant much longer after
a fructose overload than in control1ivers [38]. In addition, the
transgenic CK livers, if perfused with creatine, are protected
by endogenously formed PCr from hypoxia and ischemia [39]
and, due to a reinforced hepatic energy metabolism, regenerate
significantly better, after 70% hepatectomy, compared to
control livers [40].
The major point of discussion about the multiple functions
of the CK system, at the present time, is related to the spatial
buffer or transport function via PCr-shuttling. Results
obtained by in situ 31P-NMR saturation transfer measure-
ments investigating the CK-mediated reaction flux at
different work loads of muscle [41] or under creatine
depletion [42] have been interpreted as proof against the PCr
shuttle. However, the unexpectedly anomalous behaviour
in 31P_NMR terms of transgenic mice expressing graded CK
levels [43] suggested that some of the CK-flux, measured
in situ by saturation transfer 31 P-NMR, may evade detection
by this method [44]. This problem seems to be exacerbated
by the fact that, due to compartmentation of adenine
nucleotides in vivo, some ATP may be invisible by NMR
as well [45], indicating that a fresh look at the NMR data is
needed and that this problem can only be solved by a
multidisciplinary approach. Within this context, the dynamic
interplay between the CK and the adenylate kinase system
is also of relevance [46].
The recently solved X-ray structure of octameric Mi-CK
[47] sheds new light on some important functional aspects
of the CK system discussed above and seems to be consistent
with the proposed energy channeling function of this enzyme
[48]. The new insight into the atomic structure of the Mi-CK
molecule allows now for a detailed structure/function analysis
concerning the molecular physiology of this enzyme, its
catalytic site and mechanism, its octamer/dimer equilibrium,
as well as its interaction with mitochondrial membranes and
the relevant proteins thereof. Some of these new and intriguing
aspects ofthe Mi-CK structure are described here and certain
prominent key features ofthe molecule are discussed below
within the physiological framework of mitochondrial and
cellular bioenergetics.
The first part will present a short overview of the chicken
Mib -CK structure in the context of our present biochemical
and biophysical knowledge about CK isoforms, including
active site residues and catalytic mechanism. In the second
part, models will be developed, based on the Mi-CK structure,
to elucidate the unknown structural basis of some Mi-CK-
specific functions, e.g. the strong binding to mitochondrial
membranes and the functional coupling and metabolite
channeling to ANT of the inner mitochondrial membrane and
porin (or voltage-dependent anion channel, VDAC) of the
outer mitochondrial membrane. Two structural features and
their possible functional consequences will be discussed in
detail: (i) the identical fourfold top and bottom faces of the
octamer which contain putative membrane binding motifs
likely to be involved in binding to mitochondrial membranes
and, possibly, interaction with ANT or porin and (ii) the
central, 20 A wide channel which may be of significance for
the exchange of energy metabolites between mitochondria
and cytosol under the assumption that Mi-CK follows a 'back
door' mechanism.

Crystal structure of chicken Mih-CK
CK is one of the enzymes with the longest history in crystallo-
graphic studies. Crystallization attempts date back to the 1950s,
crystallization of cytosolic [5, 49-54] and mitochondrial [55,
56] isoenzymes have been reported by different groups.
Despite of all these results, the first successful solution of the
crystal structure of a guanidino kinase could only be reported
recently [47], using chicken Mib -CK crystallized in our group
[55].
The reason for the successful crystallization may have been
the choice of the animal species, i.e. the chicken, with a body
temperature of 42C, as well as the choice of the mitochondrial
CK isoform (Mib-CK), which had never before been used in
crystallization trials. Octameric chicken Mib -CK from
sarcomeric muscle turned out to be remarkably stable against
denaturation and was surprisingly insensitive towards
proteolysis, so that enzymatically active protein, mostly
octameric Mib -CK, could be recovered from crystallization
setups even after months of storage at 25C.
Mib -CK was crystallized by the hanging drop approach
using protein purified from chicken heart tissue [55, 56] or,
later, from overexpressing E. coli bearing expression plasmid
pRF23 [57]. Several tetragonal crystal forms were observed,
two of which proved useful for crystallographic studies ofthe
enzyme in the presence or absence of substrates (free enzyme
and Mib-CKxATP: space group P42 , a=126.0 A, c=144.7 A;
free enzyme: space group P42 12, a= 171 A, c= 150 A). The use
of a single isomorphous derivative and a complex theoretical
approach involving phase extension (W. Kabsch, unpublish-
ed) allowed the calculation of an electron density map to 3
A resolution. It should be noted here that the intrinsic non-
crystallographic symmetry of octameric Mib -CK was crucial
for the solving the crystal structure. The polypeptide chain
was fitted to the electron density map using the interactive
program 0 [58] and the model refined in several steps using
the X-PLOR software [59].
The course of the polypeptide chain of a monomer is
shown in Fig. 1. The structures of Mib -CK-ATP and of the
free enzyme were found to be very similar. Moreover,
structural differences between the four crystallographically
independent monomers in the octamer are rare, comprising
residues 1-10, 6Q..-{)6, 109-120 and 313-327. Uncertainties
in the structure assignment comprise the five N-terminal
residues and the loops 6Q..-{)5 and 316-326 (see Fig. 1), which
have a high temperature factor above 40 A2. This factor is a
displacement parameter for an atom in a crystal, containing
input from disorder in atomic positions as well as from thermal
vibrations and thus indicates flexibility inside a structure. The
overall dimensions of the monomer are approximately 36 Ax
42 A x 69 A. Assignment of the secondary structure by the
procedure DSSP [60] is given in Fig. 2. Of the 380 residues,
33% are a-helical and only antiparallel ~-strands were found.
127
This is in good agreement with the results obtained by circular
dichroism spectroscopy [61]. The monomer consists of a small
(residues 1-112) N-terminal (domain I) and a large (residues
113-380) C-terminal domain (domain II) with theATP binding
site located in the cleft between the two domains (see Fig. 2).
The small domain I is built from a short 3 IO -helix and five a-
helices of which helix a 3 is somewhat distorted. The large
domain II contains an eightstranded antiparallel ~-pleated
sheet, flanked by seven a-helices. As seen in Fig. 2, some of
the strands in the large ~-sheet are unusually connected.
Moreover, the Mib -CK fold is different from all other kinases
for which a three-dimensional structure has hitherto been
determined, although there are some similarities in the fold of
domain II to glutathione S-transferase (E. Di Iorio, ETH-
Zurich, personal communication). Figure 2 also shows the
unusual arrangement of the ~-strands in the eight-stranded~
pleated sheet of domain II, which can also be considered to
consist of two nearly equally sized subdomains IIa and lIb,
separated by ~-strand 1.
As CK is known to be remarkably stable towards proteo-
lysis, it is interesting to note that the main nicking site of CK
susceptible to proteases, Asp323 [62] (Fig. 2), is located on
the highly flexible loop 316-326. After partial unfolding of
muscle-type cytosolic M-CK by denaturants, another region
around amino acids 158-166 becomes sensitive to staphylo-
coccal proteinase V8 and therefore was postulated to represent
a domain/domain boundary [63]. In Mib-CK, this region forms
an extended stretch between the a 6 and the ~2 regions as shown
in Fig. 2 but cannot be considered a domain boundary. Recent
experiments employing two genetically engineered fragments
of Mib-CK [64], nevertheless, suggest that this region is
probably defining the boundaries offolding entities within
the Mib -CK structure since, more recently, a Mib -CK
fragment consisting of residues 167-380 was shown to be
an independent folding entity, whereas the fragment 1-166
could not fold to its native conformation (M. Forstner et al.,
unpublished).
The octamer is held together by numerous interactions of
its monomers. Mib-CK octamers are quite stable and the
dissociation into dimers upon dilution takes hours to weeks
[65]. This process is greatly accelerated by the addition of a
transition state analogue complex of the substrates (TSAC =
creatine, Mg-ADP and nitrate), which leads to dissociation
within minutes [66]. The dimers of all CK isoforrns are very
stable and disintegrate into monomers only under chaotropic
conditions. From this information it was possible to define
the dimeric building blocks of the octamer from the number
of interactions between the monomers, under the condition
that the monomer-monomer interactions forming the dimer
have to be more numerous and extensive than the interactions
holding the dimers together to yield the octamer. The validity
of this approach was supported by the finding that the Mi b-
CK dimer could be used to model the structure of dimeric M-
128

Fig. 1. Monomer structure and flexible regions ofthe Mib-CK monomer. The course of the polypeptide chain is shown in the 'backbone trace' representation
as provided by the macromolecular rendering software RASMOL v. 2.6 [123]. Domain I is shown in green and domain II in yellow; the bound ATP is
colored in blue. Flexible segments of the polypeptide chain with temperature factors above 40;"'2 are colored in red. The position of the flexible loops (60-
65 and 314-331, comprising the highly flexible stretch 316-326) and of the N-and C-termini is indicated. For His61, His92, Trp223 and Cys278, the side
chains are drawn in a 'ball-and-stick' representation (C atoms in gray, N atoms in blue, 0 atoms in red, S atoms in yellow).

ps P1

Fig. 2. Chain fold of the Mib-CK monomer. Schematic representation of secondary structure of Mib-CK. First and last amino acid residues in the helices and
sheets were determined by the use of the automatic procedure DSSP [60]. The molecule can be partitioned into two domains. The small domain I consists ofa
3'0 helix [PII-DI4] and five a-helices of which helix a3 is distorted [a\:C24-C28; a2:P31-L37; a3:L48-D57; a4:S76-V79; a5:A81-R91]. The large p-
domain contains a central p-stranded antiparallel p-sheet [Pl:VI21-R130; P2:G166-S170; P3:G2II-N215; P4:F220-1224; P5:T230-K237; P6:R287-K293;
P7:L312-G316; P8:V328-N333] surrounded by seven a-helices [a6:RI43-AI58; a7:EI76-DI85; a8:PI95-T198; a9: M241-R262; alO:P295-K299;
all :R302-L31O; aI2:E341-E363]. Two protease sensitive sites are indicated (Xl: V8-protease sensitive site at amino acid residue 166; X2: multiple proteases-
sensitive site on the flexible loop 316-326 at amino acid residue Asp323, see text for details).
 129

Fig. 3. The tryptophan residues of Mib-CK. A dimer of Mib-CK is shown in the 'backbone trace' representation as provided by the rendering program
RASMOL v. 2.6 [123] with one monomer being colored red, the other blue. The side-chains of the tryptophan residues Trp206, Trp213, Trp223, Trp264, and
Trp268 are shown in a 'spacefill' model representing the van der Waals-surface. The involvement of Trp206 in the monomer-monomer contact is evident
from the figure, the position of Trp223 in the active site is also obvious and Trp264 is located at the dimer/dimer interphase (see Fig.3b in [47] and [67]). The
N- and C-termini are indicated.

CK and compare the calculated scattering function of the Moreover, the contribution of Trp206 (Fig. 3) in one of the
model to the experimentally obtained small-angle scattering monomer/monomer contact regions to the stability of the
curve of M-CK in solution (Forstner et al., submitted). dimers has been proven by site-directed mutagenesis and

Fig. 4. Charged amino acids in the putative Mi-CK membrane binding motifs. The 'spacefill' model of the chicken Mib-CK octamer, with identical top and
bottom faces for symmetry reasons, is shown (left) in a slightly tilted side view and (right) in a view onto the top (or bottom) membrane binding face (side
length 96 A) with the central 20 Awide channel. Basic amino acids (positively charged) are depicted in red, acidic amino acids (negatively charged) are
shown in blue. The four dimers forming the octameric structure are marked in different grey levels. The model representing the van der Waals-surface was
prepared using the rendering software RASMOL v. 2.6 [123].
130
fluorescence spectroscopy [67] and the necessity for the
presence ofTrp264 [67], as well as of the intact N-terminus
for octamer formulation has been demonstrated as well [68].
Active site residues and catalytic mechanism
Data analysis of the ATP containing co-crystals of Mib -CK
reveals additional electron density for one nucleotide in each
monomer which is not present in the native map. It is located
in the cleft between the domains in the highly conserved
region of all guanidino kineses. The adenine base fits into a
pocket formed by the residues His 186, Met23 5, Serl23,
Arg125, His291, Gly289, and the pair Arg287 ,Asp330 which
forms a salt bridge. The phosphate groups of ATP interact
with four arginine residues at positions 125, 127,287, and
315. Two additional residues, Arg91 and Arg336, are found
at a distance of 5 A from the y-phosphate. This region, also
including His92, leaves enough room to accommodate
creatine in an orientation that would allow a direct in-line
transfer of the phosphoryl group between the bound substrates.
By a number of experiments with chemical modification
ofSH-groups using different SH-reagents in the presence or
absence of substrates, or the affinity label epoxycreatine [69],
Cys278 was identified as the highly reactive 'essential'
cysteine residue (see also Fig. 1). However, further studies
of this type yielded contradictory results concerning the
essentiality of Cys278 for enzyme catalysis (see [70]).
Nevertheless, until recently, Cys278 has been suggested in
the literature as an important residue for catalysis [71].
Thorough kinetic analysis of several replacement mutants of
Cys278 has solved this long debated issue by providing clear-
cut evidence that Cys278 is not essential for catalysis per se,
but is important for the synergism of substrate binding [72].
NMR studies on the role ofhistidines at the active site have
revealed four such residues at distances of 12, 12, 14 and> 18
Afrom the Cr3+ in the paramagnetic ~,y-bidentate Cr3+-ATP
complex bound to the enzyme [73]. One of these histidine
residues, with an altered intrinsic pK near 7, was postulated
to act as an acid-base catalyst [74]. We find His92, His291,
His229, His61 and His186 at distances of 12, 12, 14, 17,
and IS A, respectively, from the 07 atom ofATP (see Fig. 1).
Recent studies by our [75] and other groups [76] have shown,
however, that none of these residues can be considered
essential for the functioning ofCK, although some ofthem
certainly are of importance for substrate binding in the
active site ofCK. However, replacement mutants ofHis61
had a strong negative effect on the enzyme activity, e.g. the
respective His to Ala mutant showed approximately 1 and
10% of wild type activity in the forward and backward
reaction, respectively [75]. The importance ofHis61 for the
catalytic activity of CK is consistent with the assumption
that the flexible loop formed by residues 60-65 and which
is found rather distant from the ATP in the crystal structure
might move towards the active site to promote catalysis [75].
Spectroscopic studies of cytosolic M-CK and its complexes
with adenosine phosphates, as well as quenching studies,
were suggestive for the occurrence of a tryptophan residue
at or near the cosubstrate binding site [77]. Specific chemical
modification of tryptophan residues revealed one of them as
essential for enzymatic activity. These results are in agreement
with an earlier lH-NMR study indicating an interaction
between protons of the adenine ring of bound ADP and one
or more aromatic side chains of the protein. Trp206 and
Trp223 are conserved throughout the whole family of
guanidino kinases [78] and replacement of Trp223 resulted
in a complete inactivation of the enzyme, indicating that
Trp223 is situated near the active site and is crucial for
catalytic activity and, therefore, was termed the 'active site
tryptophan' [67]. By NMR, the active site tryptophan was
found to be in the vicinity 5 A) of the nucleotide substrate
[7Sa].
Incidentally, four of the five tryptophan residues of Mi b-
CK (Trp206, Trp213, Trp223, Trp264, see Fig. 3) are found
rather closely clustered and oriented in such a way that energy
transfer, indicated by fluorescence studies [67] may indeed
be possible, whereas Trp268 is closer to the C-terminal
membrane interaction face of the octamer. In the X-ray
structure of the wildtype enzyme, Trp223 was found at a
distance of about loA from the adenine ring. We assume,
however, that in the presence of all substrates or at the
transition state, the nucleotide would move 'down' towards
Trp223, thereby bringing the substrate closer to this residue,
thus explaining the distance measured by NMR. In this
context it is necessary to remember that the Mg-nucleotide
rather than the free nucleotide is the substrate for CK and that
only the Mg-nucleotide can induce structural changes
essential for catalysis [79].
In comparison with other kinase structures, Mib -CK has
acquired a different folding pattern for catalyzing phosphoryl
transfer, a result which is of significance for the evolutionary
positioning of CK. Furthermore, Mi-CK is the only isoform
in the creatine kinase system which forms octamers, and the
structure supports the hypothesis that Mi-CK has evolved to
serve an additional function as a structural protein that
mediates the adhesion between inner and outer mitochondrial
membranes and between cristae membranes via electrostatic
interactions due to charged residues located at its four-fold
faces [25-27]. These properties ofMi-CK will be discussed
in detail in the next chapters.
Membrane binding of Mi-CK
In view ofthe physiological function ofCK, one ofthe most
important properties of this enzyme is the distinct subcellular
 131

localization of its different isoforms (for a review see [25, complex formation via charged phospholipids [81] or by a
26]). This isoenzyme-specific localization has been related different, hydrophobic binding mode [94].
to the functional coupling of CK to energy providing and
energy consuming processes of distinct subcellular s~tes and
is part ofthe PCr-shuttle or -circuit model ofC~ fun.ctlOn (see Early studies on Mi-CK membrane binding domains
above). Mitochondrial CK isoforms, sarcomenc MIb-CK and
Mi-CK-membrane interactions have been studied mainly with
ubiquitous Mia -CK differ from their cytosolic counterparts n~t
respect to binding mode, binding constants and possible
only by their organellar compartmentation, but also by t?eIr
binding partners ofMi-CK; little is known about the structural
membrane-binding property. While cytosolic CKs are mamly
domains of Mi-CK involved in membrane binding. In a first
soluble but certain fractions of the enzyme are also bound to
attempt to identify membrane binding domains, peptide
particular subcellular structures (e.g. myofibrils, sarcoplas~ic
fragments of rat sarcomeric Mib -CK, obtained by cyanog.en
reticulum and plasma membrane), Mi-CKs are known to bmd
bromide digestion, were analyzed in a binding assay WIth
strongly to the outer surface of the inner mitochondrial
artificial cardiolipin vesicles [98]. Only one peptide, containing
membrane [80] and, in case ofMib -CK, to bridge mitochond-
the 25 N-terminal amino acids, showed high affinity to
rial membranes (for reviews see [81] and Stachowiak et aI.,
cardiolipin. This approach, however, is prone to artifacts since
this issue). Membrane binding of Mi-CK is considered the
an isolated peptide can behave differently compared to the
structural basis of functional coupling between Mi-CK and
same amino acid sequence in its native protein environment
oxidative phosphorylation which occurs via the metabolic
(see also below). To identify amino acids involv.ed in
channeling between Mi-CK and the adenine nucleotide
membrane binding, the intact Mi-CK molecule was subjected
translocator (ANT) of the inner mitochondrial membrane (e.g.
to chemical modification of selected amino acids (Met, Cys,
[82], see also below). Similarly, a function~l intera~tion
His, Lys and Arg) [98]. It occurred that, while a modi~c~ti?n
between the outer mitochondrial membrane ponn and MI-CK
at Met, Cys, or His residues still allowed binding to cardlOiIpm,
has been proposed [25].
modification ofLys or Arg residues drastically reduced binding
It is generally accepted now that Mi -CK binds to membranes
by over 90%. Although these results must be interpreted with
mainly by electrostatic interactions, although the participation
caution since chemical modifications could introduce overall
of hydrophobic interactions cannot be ruled out [83]. As
structu:al changes, they corroborate evidence on electrostatic
binding partners in the mitochondrial membranes, cardiolipin
interaction between basic amino acids of Mi-CK with
and other phospholipids were identified by different methods
negatively charged cardiolipin. . .
[83-87]. Highest affinity was determined for cardiolipin
A different early approach for the structural charactenzatlOn
which contains a two-fold negatively charged headgroup and
ofMi-CK and its membrane interaction sites applied a variety
occurs mainly in the inner mitochondrial membrane. However,
of electron microscopic techniques for single molecule
it is still debated whether the octamer only binds to the lipid
imaging [100--102]. These methods visualized for the first
surface [88] or penetrates partially into the membrane bilayer
time the octamer structure ofMi-CK, conserved throughout
[87]; the latter interaction would be favored by cardiolipin
different species from sea urchins to mammals [103], thus
which is capable to form non-bilayer structures [89, 90]. Such
confirming ample biochemical and biophysical evidence on
a 'dipping-in' could also explain the resistance ofa certain
the octameric structure of membrane-bound Mi-CK in vitro
Mi-CK fraction against detachment by high ionic strength or
(e.g. [65, 102]) and in vivo (e.g. [104]). In addition, th~se
detergents and the arrangement of the bulky enzyme in the
studies revealed the fourfold symmetry of the octamer WIth
rather narrow mitochondrial intermembrane space.
dimers arranged along the symmetry axis (compare to the
Although an interaction between Mi-CK and ANT or porin
atomic structure in Fig. 4). Most importantly, they showed
has been shown on the functional level (for a review see [25]),
that top and bottom faces of the octamer are identical and
direct evidence for a structural interaction in vivo, e.g. by
confer the 'sticky' character to the protein which is responsible
crosslinking studies, has not been obtained so far [81, 91].
for binding to artificial cardiolipin-rich membranes and other
Furthermore, reconstituted ANT did not bind to Mi-CK [85]
negatively charged surfaces [100--102]. Only under very
and electrostatic repulsion of the basic Mi-CK isoforms is
particular conditions in solution, due to the absence of
suggested by the basic pI of ANT [84] and positive charges
surfaces, top and bottom faces attach to each other and form
exposed to the intermembrane space by porin [92]. However,
Mi-CK filaments [101].
in vitro complex formation between Mi-CK, ANT and porin
has clearly been demonstrated [93-95]. Since ANT is situated
in a cardiolipin membrane patch [96, 97], the binding ofMi- Putative membrane-binding motifs of Mi-CK
CK to such patches is assumed to bring the ANT and Mi-CK
in close vicinity to each other [25, 98, 99]. The observed So far, information on the structural basis ofMi-CKmembrane
interaction of Mi-CK with porin might occur by a similar binding has been scarce. The recently solved octameric
132

structure of the sarcomeric Mib -CK from chicken (Fig. 4 and
[47]) together with a number of new sequence data (reviewed
in [78]) provide now the means for a closer analysis of the
membrane binding face ofMi-CK. They allow to reconsider
some of the earlier results quoted above and to designate new
putative membrane-binding motifs. Those motifs should
fulfill the following criteria: (i) since strong membrane-
binding is specific for mitochondrial CK isoforms, the
responsible sequence motif should be localized in Mi-CK
specific, conserved sequence stretches; (ii) the primary
structure should contain a positive net charge in order to
interact with the negative charges of cardiolipin or other
phospholipids and finally, (iii) the charged amino acids
should be properly exposed at the top and bottom faces of
the octameric Mi-CK structure known to interact with
membranes [102].
A comparison of28 known CK amino acid sequences [78]
reveals the high degree of sequence conservation within the
CK family, including six blocks of a near 100% sequence
identity, separated by seven less conserved blocks. For their
entire amino acid sequence, the four CK isoform classes
(cytosolic B- and M-CK, sarcomeric and ubiquitous Mi-CK)
show about 60% sequence identity, but this number increases
to about 80% if mitochondrial and cytosolic isoforms are
considered separately. Hence, as much as 20% of conserved
Mi-CK sequences differ from cytosolic CK isoforms. Some
of these conserved Mi-CK sequences are boxed in longer
stretches and contain basic amino acids, among them the N-
terminus (Lys5 - Cys24, sequence and numbering according
to chicken Mib-CK), the C-terminus (Asp352 - Lys380) and
some internal stretches, the longest being Asp96 - Arg 119,
Lys263 - Glu275 and Pr0305 - Leu320.As mentioned above,
the N-terminus (Thrl - Met25) was already proposed to be
the cardiolipin-binding domain due to the binding properties
of the corresponding peptide [98]. It contains the positively
charged residuesArg 19, Lys20 and His21 which are absolutely
conserved in all Mi-CKs. However, the octameric CK X-ray-
structure clearly shows that the N-terminus is not exposed at
the surface but buried inside the Mi-CK octamer (see Fig. 2
in: Stachowiak et al., this issue) and therefore not able to
interact with membranes. Kaldis et al. have shown the
involvement of the N-terminus in another Mi-CK-specific
property, that is the dimer-dimer-interaction inside the
octamer [68].
A closer analysis of the octamer structure of chicken Mi b-
CK reveals two sequence stretches which are properly
exposed at the top- and bottom faces of the octamer, carry a
positive net charge and are also Mi-CK specific (see Fig. 2
in: Stachowiak et al., this issue, and Figs 4 and 5, herein).
They encompass the C-terminus (Asp357 - Lys380) and a
short internal stretch (Ala 107 - Gln1l5). The C-terminal part
ofMi-CK as defined by its location at the molecules surface
consists of the distal part of an a-helix (Asp357 - Glu363)
and a stretch without defined secondary structure. This
sequence shows an accumulation of six or seven basic amino
acids resulting in a positive net charge (Fig. 5). In chicken
Mib -CK, the C-terminus contains six positive charges which
are well exposed at the binding face (Fig. 5): five lysines

108 115 360 370 380

+ -++ -+ + ++
Mib-CK ASK I T H G Q DCEKKLEKGQDIKVPPPLPQFGRK
Q R K
+ + -++ -+ + ++
ASKIKFGH DCERRLEKGQDIRI PSPVPQFRH
RS Y R P LVHGK
TIS
T

+ -++ +
M-CK ENLKGGDD EMEKKLEQNQPIDDMIPAQK
KG S

- ++ +
B-CK DNLQGGDD EMEQRLEKGQSIDDLMPAQK
Q P V
A

Fig. 5. Amino acid sequence of the putative membrane binding motifs. The homologous sequence stretches of the four CK isoforms, sarcomeric Mib-CK,
ubiquitous Mia-CK, as well as cytosolic muscle-type M-CK and brain-type B-CK, are compared. For each CK isoform, the sequence from chicken is shown
in the first line; differences occurring in other species are noted below. Charged amino acids are printed in bold letters and conserved charges are noted
above the sequence. Sequence comparison was taken from [78), the amino acid numbering corresponds to the sarcomeric chicken Mib-CK.

(Lys360, Lys361, Lys364, Lys369, Lys380) and one arginine
(Arg379). Out ofthe four negative charges (Asp357, Glu359,
Glu363, Asp367), only the two asparagines (Asp357 and
Asp367) are pointing to the binding face (marked in blue in
Fig. 4). The homologous stretches of cytosolic CK isoforms
(B-CK and M-CK) do not show such a density of positive
charges. It has to be mentioned that the C-terminal sequence
is not absolutely identical between sarcomeric and ubiquitous
Mi-CK. While the number of basic amino acids is well
conserved, their position and a sequence stretch at the very
C-terminus show differences which could be responsible for
divergent membrane binding properties of the two Mi-CK
isoforms (Fig. 5). In both isoenzymes, charged residues at the
C-terminus (Arg379 and Lys380) are separated from the other
charges (Lys 360, Lys361, Lys364, Lys369) by a stretch of
8-9 residues containing 6-7 hydrophobic amino acids (Fig.
4). These stretches may possibly enter a phospholipid bilayer
and thus contribute to the strengthening of the Mi-CK-
membrane interaction.
The second sequence motif, a short internal amino acid
stretch (Ala 107 - Gin 115) located at the interconnection
between helices of the small and large Mi-CK domain [47]
harbors one to three positive charges depending on the Mi-
CK isoform and the species considered. In chicken Mib -CK,
one lysine (Lys 11 0) and one histidine (His 113) are found
(Fig. 4). By contrast, B-CK does not contain these positive
charges and the single positive charge in M-CK is surrounded
by several negative charges (Fig. 5). In all Mi-CKs, the
cumulative positive net charge of both motifs is about four
and is mainly due to lysines (Fig. 4). A similar result, i.e. four
Iysines involved in Mi-CK-membrane binding, was obtained
by a pure modeling approach based on the pH-dependence
of the binding process [105].
Although the basic amino acids mentioned are not conti-
guous on the Mi-CK primary structure, their arrangement on
the octamer surface is such that they form clusters of two or
four positive charges (Fig. 4). This could have a special
significance, given the proximity of the two negatively
charged phosphates of the cardiolipin headgroup. A similar
'collar' oflysines has been identified in ANT and proposed
to mediate the contact to cardiolipin [106]. Another common
feature of both putative membrane-binding motifs is the
relatively high temperature factor compared to the main part
ofthe Mi-CK structure [47]. The flexibility indicated by this
fact may be useful for a 'docking' onto the negatively charged
phospholipid heads.
Taken together, the properties of the two sequence motifs
discussed fully support their putative function in membrane-
binding, provided that the Mi-CK is in the octameric form
and membrane binding occurs primarily due to electrostatic
interactions with negatively charged membrane phospholipids
and, possibly, a minor participation of hydrophobic forces.
This binding model could explain the ability of the Mi-CK
133
octamer to bridge two membranes in vitro by its top and
bottom faces and to interact functionally with ANT and porin
in vivo by a close co-location with these membrane proteins
inside cardiolipin/phospholipid membrane patches. Formation
of the proposed multienzyme complex for energy export
(ANT, Mi-CK and porin) could be a transient and dynamic
process, similar to the formation of contact sites. A detailed
site-directed mutagenesis study of the putative membrane
binding motifs ofMi-CK is underway and should provide a
clue for defining the interaction sites at the Mi-CK-membrane
interface.
Metabolic channeling - a mechanism for Mi-CK in
mitochondrial contact-sites?
One of the most important requirements of a cell is to save
metabolic energy and to control the flux of metabolites for
the biosynthesis of compounds necessary to keep the plethora
of biochemical reactions going. These requirements are of
particular importance for reactions which directly influence
the metabolic status of a cell and its future fate, e.g. energy-
conversion reactions like the turnover ofATP, ADP, PCr and
Cr catalyzed by CK isoforms. One of the mechanisms that
evolved to reach a maximum of energetic efficiency and
metabolic control is metabolic channeling: a precursor which
is synthesized by enzyme (or subunit) A is passed to enzyme
(or subunit) B without getting into contact with the bulk
solution (for a review see [107]). Metabolic channeling has
been well described for tryptophan synthase which possesses
a 25 A hydrophobic tunnel linking the active sites of two
different subunits [108] and thus passing the substrate indole
from the (X- to the ~-subunit. However, channeling can occur
between huge covalently linked enzyme-complexes such as
the fatty acid synthase, as well as between the structurally less
tight coupled glycolytic enzymes glyceraldehyde-3-phosphate
dehydrogenase and aldolase [109].
The sequestering of intermediate products in a micro-
compartment provides several advantages for the reaction
sequence, but also for cellular metabolism in general. Since
these intermediate products immediately serve as substrates
for the subsequent reaction and do not equilibrate between
microcompartment and bulk solution, which often would
favor the backward-reaction, the difference in Gibbs' free
energy for their formation is decreased. Competing side-
reactions may be excluded and counteracting catabolic and
anabolic pathways are separated thus preventing interference
which would perturb metabolism. Furthermore, sequestered
intermediates are present at high local concentrations and an
apparently low Km for these intermediates can be observed
with the channeling complex compared to the non-channeling
situation, e.g. in vitro measurement with isolated components.
Finally, channeling intermediate products of a reaction
134
cascade in a microcompartment also allows to strictly control
the whole process by feed-back regulatory mechanisms such
as substrate activation, product inhibition and cooperativity
and the over-all-flux ofthe reactions can be controlled very
accurately, e.g. by a limiting step in the reaction sequence (for
an overview and further information on the topic see 'Cell
Architecture and Metabolic Channeling' by Ovidi [107] and
'Channeling in Intermediary Metabolism' by Agius and
Sherratt [110]).
As already mentioned before, there is functional as well as
in vitro evidence for complexes containing Mi-CK, porin and
ANT, located in the contact sites of the mitochondrial envelope.
An important function ofthese ANT/Mi-CKJporin complexes
could be metabolite channeling [48]. In contrast to tryptophan
synthase, the function of a Mi-CK complex would not be a
reaction cascade where an intermediate is transferred, but the
transport of energy-rich phosphate compounds likeATP, ADP
and PCr between subcellular compartments. This means that
ATP reaching the mitochondrial intermembrane space viaANT
is directly channeled to the Mi-CK octamer and used for
synthesis of PCr which then is channeled out of the mito-
chondrion into the cytosol via Mi-CK and porin. Vice versa,
ADP is back-channeled into the matrix via ANT.
A prerequisite for this mode of action is a physical
interaction of Mi-CK and ANT or, at least, a close vicinity of
both leading to 'leaky' channeling. The finding that externally
added pyruvate kinase is not able to use the ADP produced by
the Mi-CK reaction [82, 111] and the fact that about 50% of
Mi -CK activity is resistant against iodo acetate treatment [93]
strongly supports the model ofMi-CK in close vicinity to ANT
and porin. Kottke and co-workers further showed that in
mitoplasts supplied with ATP and creatine, the blockage of
porin by a polyanion strikingly decreases Mi-CK activity [93].
In addition, creatine-stimulated mitochondrial respiration due
to ADP synthesis via Mi-CK is not abolished by the addition
of externalADP traps like PEP plus pyruvate kinase, showing
that the ADP generated by Mi-CK is located in a micro-
compartment close to the inner mitochondrial membrane and
does not equilibrate with the bulk ADP pool [82, 111]. An
interaction of Mi-CK with porin is indicated by the fact that
isolated porin is able to induce octamer formation of an N-
terminal Mi-CK deletion mutant which is unable to build up
octamers in free solution [94]. Finally, complex formation of
octameric Mi-CK with porin and ANT has recently been
demonstrated in vitro [94, 95].
There are several advantages of the porin/Mi-CKJANT
arrangement: (i) ATP originating from the matrix space does
not equilibrate with the cytosolic ATP and is present at very
high local concentrations which facilitate the formation ofPCr
from Cr; (ii) since ATP is directly turned over after being
exported out of the matrix, it will generate high local con-
centrations of ADP in the intermembrane space which stimu-
lates respiration and favors re-import of ADP into the matrix
because of a steeper ADP gradient across the inner membrane
[Ill]. This model also allows the ANT to act as an obligatory
antiporter, since for every exported ATP there is an ADP
molecule present which can be bound by the translocase. The
thermodynamic advantage of such a complex is due to (i) the
removal of the translocation product ATP, favoring further
export ofthe latter viaANT, and (ii) the removal of once formed
PCr from equilibrium by direct channeling into the cytosol via
porin, favoring PCr synthesis by Mi-CK [111,112]. Possible
transport and diffusion pathways involved of such a channel-
ing mechanism are discussed in detail in the next chapter (see
also Fig. 6).
Metabolite access to Mi-CK - a 'back door' hypothesis
As it has been outlined above, Mi-CK displays functional and
possibly also structural coupling to ANT and porin, two
membrane proteins involved in the transport of Mi-CK
substrates and products. If this coupling involves metabolic
channeling, a vectorial transport of substrates and products
would be expected [48]. The X-ray structure of chicken Mi b-
CK [47] provides a new and suitable tool to develop models
for metabolite transport in the multiprotein complex bridging
the two mitochondrial membranes as suggested from in vitro
studies [93-95]. In this respect, a specific structural feature
of the octameric structure is of particular interest, i.e. the
central pore or channel of approximately 20 A in diameter
penetrating the octamer. This channel has already been
detected by electron microscopic studies of chicken Mib -CK
[10 I] and has been proposed to connect the active site ofMi-
CK with the surroundings, supposing that active sites were
exposed towards the central channel. This assumption was
plausible at a time where still no three-dimensional high-
resolution structure of CK was available.
However, the X-ray structure ofMib-CK shows that the most
obvious and direct metabolite access to the active site ofMi-
CK is from the 'outside' of the molecule (see Fig. 3c in [47])
and not from the channel. The CK active sites, one for each
monomer, are exposed on the concave side of the 'banana-
shaped' dimers (Fig. 3), and four such dimers form an octamer
by contacting each other via the convex 'backside' (involving
an N-terminal stretch plus the region around Trp264, see Fig.
3 and also Fig. 3b in [47]). If the Mi-CK octamer binds to the
inner and outer mitochondrial membranes by its identical top
and bottom faces, as shown by Schnyder [102], the active sites
face the intermembrane space surrounding the octamer and the
channel is oriented orthogonally to the membranes and thus is
excluded from direct interaction with the intermembrane space.
For this kind of organization between Mi-CK and membranes,
one could propose two different models which would explain
a compartmentalized CK reaction as indicated by the functional
coupling of Mi-CK to ANT and porin (see Fig. 6).
 Cyto 01 lr 13

OM

A 1M

1M
Fig. 6. Model considerations for microcompartmentation of the Mi-CK
reaction. Two different models for microcompartmentation of substrates and
products of the creatine kinase (CK) reaction, as well as for vectorial export Matrix pace
of phosphocreatine out of the mitochondrion are shown. Both involve porin
of the outer mitochondrial membrane (OM), adenine nucleotide translocator
(ANT) of the inner membrane (1M) and Mi-CK octamers in the inter-
membrane space (lMS). The fluxes ofthe metabolites are depicted by arrows Cyto 01
colored in red (ADP), blue (ATP), green (Cr) and brown (PCr). (A): Possible
'back door' mechanism. Cr and ATP enter the CK active sites from the IMS.
After phosphoryltransfer, PCr leaves the active site via the suggested 'back
door' of the enzyme, being guided into the central20A wide channel of the
Mi-CK octamer. From there, it is exported into the cytosol via porin situated
on top of the central channel (see also text). This model suggests that the B 1M
central hole in the octamer and porin indeed serve as a channel connecting
the CK active site with the cytosol. (B): Microcompartmentation without the
need for a 'back door' mechanism. Reactants and products ofthe CK reaction
enter and leave the enzyme from the 'front door'. Octamers situated close to 1M
each other are suggested to form a microcompartment including porin and
ANT for the transport of the metabolites to and from the cytosol or ADP
mitochondrial matrix, respectively. In this model, the central hole ofMi-CK Matrix pace
has no channeling function and would serve merely a structural role. The
vectorial transport of PCr to the cytosol would be regulated by the
accumulation of the product in the IMS, resulting in a change of ion selectivity
ofporin (see also text).
O Porin fl K
ill
B

y - Phosphate

"bad<-
door" ..... - - -

"bad<-
door"

Fig. 7. Structural model of an alternative leaving-route for the phosphocreatine product through a 'back-door' exit. We propose a possible pathway of
phosphocreatine (PCr) through an exit between two CK-monomers leading to the central main channel of the octamer. A 'backbone trace' representation of
a 'banana-shaped' Mi-CK-dimer is shown with bound ATP (A). The dimer is rotated about 90 (B) and additional 20 (C) around its longitudinal axis to
provide a better 3-dimensional visualization. The C-termini are marked in green, the N-termini and residues 47-591194-202 are shown in red (monomer I)
and blue (monomer 2). The alternative leaving-route for PCr is depicted in each representation with a dashed arrow; the broad arrow in (B) indicates the
'back door'. Note that this 'back door' is deduced only from the X-ray structure ofMi-CK and is postulated on the basis of biochemical considerations (see
text). The figure was prepared using the rendering software RASMOL v. 2.6 [123].
136
In the first model, depicted in Fig. 6A, access of substrates
and products to the active site occurs from different sides of
the enzyme, that is from the intermembrane space and from
the central channel. In this case, a role of the channel in the
transport of metabolites would require (i) direct structural
interaction with porin (and/or ANT) and (ii) accessibility of
the channel from the active sites of CK, positioned on the
outside of the four side-faces of the MiCK octamer, by a so
called 'back door' pathway. The existence of an alternative
route for substrate, product or water ('back door' mechanism)
has been proposed recently for myosin [113], acetylcholine
esterase [114] and the tryptophan synthase a2~2 multi enzyme
complex [115]. Such a 'back door' connecting the active site
with the channel could, indeed, exist in Mi-CK, as shown in
Fig. 7 for a Mi-CK dimer. Two helices in each ofthe monomers
(residues 47-59 in domain I and 194-202 in domain II) and
the N-termini form an opening face to the central channel and
a small access route to the active site. However, this pathway
would be probably too narrow for adenine nucleotides having
a relatively large size and bulky conformation. Therefore,
access of ATP and exit of ADP to and from the active site,
respectively, is assumed to occur via the intermembrane
space. It then follows that ANT of the inner mitochondrial
membrane should be located in the vicinity of CK octamers
without direct contact to the central channel, thus releasing
ATP into the intermembrane space which would subsequently
diffuse into the active site pocket of CK (Fig. 6A).
The described 'back door' pathway, however, would be best
suited for PCr, since (i) the 'back door' directly points to a
putative PCr/Cr binding site (facing the y-phosphate of ATP)
and (ii) the PCr molecule has a more elongated structure which
could best fit through the 'back door' (due to the repulsion of
charges, PCr becomes more elongated than Cr, which has a
more or less cyclic structure). Thus, the model suggests that
PCr is dissociating from the active site directly into the central
channel of the octamer, thereby separating from ADP which
cannot follow due to its larger size and also for electrostatic
reasons. In addition, it is entirely conceivable that domain
movement upon substrate binding leads to further opening of
the 'back door', thus favoring PCr product release. In this
model, the porin in the outer mitochondrial membrane would
be in direct interaction with the central channel to transport
synthesized PCr into the cytosol (Fig. 6A).
Although it is tempting to believe in the existence of such a
'back door' that would enable vectorial transport of PCr out
of the mitochondrion or facilitate the proposed microcompart-
mentation of the CK reaction and its interaction with the
transport proteins, alternative possibilities as the one depicted
in Fig. 6B may be envisaged as well. In this model, several
CK octamers arranged in close proximity (only two are drawn
due to the limitation of the two-dimensional representation)
enclose a part of the intermembrane space, thereby forming a
micro compartment including ANT and porin. The central
channel of the Mi-CK octamer would then have no other
function than a mere structural role, stabilizing the convex
'backsides' (see Fig. 3) ofthe four dimers in the octamer and
probably offering space for structural rearrangements of the
monomers within the octamer during catalysis, as has bee
shown to take place upon Mg2+-nucleotide binding [79].
Several arguments, however, are in favor of a 'back door'
mechanism (Fig. 6A). One of the strongest is the great
thermodynamic advantage of such a mode of reaction due to
the continuos withdrawing of one of the products from the
active site environment. This mechanism would facilitate a
non-equilibrium system, compared to the alternative model
which releases all Mi-CK products into the microcompartment
of the incoming substrates (Fig. 6B). Furthermore, it has been
shown that product release is the rate limiting step of the CK
reaction in the forward direction [74] and that PCris the 'sticky'
product. This would be in line with the certainly longer time
for the diffusion of PCr out of the active site through a 'back
door' compared to the fast release of ADP through the large
open cleft in the CK structure. Another supportive argument
comes from the inhibition mode ofcyclocreatine (cCr). cCr, a
competitive inhibitor of the CK reaction, has been shown to
bind to the active site and to be phosphorylated [116]. The
'back door' model offers a simple explanation for this
inhibitory mechanism. Although the cCr structure mimics the
cyclic nature of Cr and cCr can be phosphorylated, the
conformational change upon phosphorylation leading to a
more elongated structure (see above) cannot take place like in
the case ofPCr. Therefore, phosphorylated cCr could be unable
to use the release pathway ofPCr through the 'back door' and
thus would block the active site of Mi-CK.
The main argument against a 'back door' mechanism comes
from the classical enzymological studies of the CK reaction.
The forward reaction of CK was described to follow a random
bi-bi mechanism [6, 74], while the proposed 'back door'
pathway would rather imply a sequential scheme, involving
binding ofthe substrates, domain closure induced by Mg-ATP
[79] together with a possible further opening ofthe 'back door',
PCr release via the 'back door' and, finally, Mg-ADP release
into the intermembrane space. However, this mechanism is not
necessarily in contradiction with the available kinetic data,
since all ofthem have been obtained with dimeric MM-CK in
free solution. Enzyme kinetics of Mi-CK might therefore be
different, especially under in vivo conditions in the inter-
membrane space when attached to mitochondrial membranes.
As of yet, there is no direct experimental proof for the proposed
'back door' mechanism in Mi-CK, and it should be noted
once more that the appealing hypothesis presented herein is
mere speculation based on structural data. Clearly, more
studies addressing this intriguing question are necessary,
including detailed enzyme kinetics of octameric Mi-CK, as
well as a direct analysis of the 'back door' pathway, e.g. by
introducing bulky residues with site directed mutagenesis.

Conclusions and outlook
It has taken more than 60 years since the first description of
CK [2] until the first X-ray structure ofa member of the CK
isoform family could be solved [47]. This step forward has
been recognized as a significant achievement and landmark
in CK research [71, 117]. As outlined above, the octameric
structure has already provided some important insights and
stimulating clues into the functioning of the enzyme and, after
all, is also rewarding from a purely aesthetical point of view.
However, having the structure of the Mi-CK octamer at hand
with many details becoming obvious now, we are brought
back to real life again. Some difficult longstanding questions,
e.g. about the kinetic mechanism of the enzyme and the
amino acid residues involved in catalysis, have still to be
answered in detail. Are there true catalytic residues or does
CK belong to the category of'conzymes' designed to merely
bring the substrates into close alignment [71]?
To solve this puzzle, an X-ray structure of CK in the
presence of reagents known to induce the enzymes' transition
state, e.g. Mg-ADP, creatine and nitrate [118], will be of
utmost importance. Such a structure of the 'closed con-
figuration' of the enzyme should reveal not only the exact
molecular outline of the active site pocket of CK caught in
the act of catalysis, but should also shed some light on the
expected hinge bending and domain movements taking place
upon binding of Mg 2+-nucleotide [79]. After having learned
that each subunit of CK has its own catalytic site, a further
important task will be the elucidation of possible coopera-
tivity between monomers in the dimer and, even more
interestingly, in the octamer. The latter question, of course,
will be most relevant to the issues of vectorial metabolite
channeling and directed export of PCr, derived by Mi-CK
from matrix generated ATP, out of the mitochondrion (see
above).
The understanding of the molecular details ofCK catalysis
and the possible design of CK-specific inhibitors are likely
to be of great clinical relevance, since an involvement of CK
in signal transduction [119, 120], as well as in tumor growth
and malignancy has been demonstrated and since certain
creatine analogues were shown to display anti-cancer activity
on explanted human tumors [121] or to enhance significantly
the effect of chemotherapy on cancer cells [122].
The precise knowledge ofCK structure and function at the
molecular level may elucidate the in vivo function of this
isoenzyme system in cells and organs with high and fluct-
uating energy requirements and thus may complement the
work done with transgenic mice lacking the expression of one
or several CK isoform in vivo [37]. Since every novel result
produces a myriad of new questions, much of what has to be
learned is still ahead of us.
137
Note added in proof
Recent 3 1P_NMR studies comparing CK-mediated fluxes of
inactive versus actively swimming sea-urchin sperm [124],
as well as mathematical modeling of cardiomyocyte energy
metabolism, including the known CK-'knock-out' models
[125], are in support of a CKJPCr-shuttle in these cells.
With the help of the chicken Mib-CK coordinates, the X-
ray structures of cytosolic B-CK (Rao JKM and Wlodawer
A, personal communication) and the related lobster arginine
kinase (Dumas C and Janin J, personal communication) have
been solved. Arginine kinase from horseshoe crab has been
crystallized in the presence of TSAC substrates [126] and
may provide the first transition state structure of a guanidino
kinase. A new study ofMi-CK membrane binding properties
[127] suggested a partial disorganization of the used phos-
pholipid bilayer upon interaction with Mi-CK and revealed
a decreased Mi-CK binding after treatment with p-hydroxy-
mercuribenzoat. The latter may be due to a modification of
the C-terminal Cys358 (see Fig. 5), which could influence
the binding capacity of the C-terminus.
Acknowledgments
Dr. W. Kabsch is gratefully acknowledged for discussion. This
work was supported by the Swiss National Science Found-
ation (SNF grant No. 31-33907.92, toT.w.), by ETH graduate
training grants for M.F., M.E. and O.S. and by financial
support from the Swiss Society for Muscle Diseases, as well
as by private sponsoring from Careal Holding AG and
Synergen AG, Switzerland.
References
I. Eggleton P, Eggleton GP: Further observations of phosphagens.
J Physiol65: 15-24, 1928
2. Lohman K: Dber die enzymatische Aufspaltung der Kreatinphos-
phorsaure; zugleich ein Beitrag zum Chemismus der Muskelkontraktion.
Biochem Z 271: 264-277, 1934
3. Lehmann H: Dber die Umesterung des Adenylsauresystems mit
Phosphagenen. Biochem Z 286: 336-343,1936
4. Lipmann F: Metabolic generation and utilization of phosphate bond
energy. Adv Enymoll: 99-162,1941
5. Kuby SA, Noda L, Lardy HA: Adenosinetriphosphate-creatine
transphosphorylase. I. Isolation of the crystalline enyme from rabbit
muscle. J Bioi Chern 209: 191-201,1954
6. Kuby SA, Noltman EA: ATP-Creatine Transphosphorylase. In: PD
Boyer, H Lardy, K Myrback (eds). The Enymes, Vol. 6. Academic
Press, NY, 1963, pp 515-521
7. Burger A, Eppenberger M, Wiesman U, Richterich R: Isoenzyme der
Kreatinkinase. Helv Physiol Acta 21: C6-CIO, 1963
8. Eppenberger HM, Eppenberger M, Richterich R,Aebi H: The ontogeny
of creatine kinase isozymes. Dev Bioi 10: 1-16, 1964
138
9. Jacobs H, Heldt HW, Klingenberg M: High activity of creatine kinase
in mitochondria from muscle and brain. Evidence for a separate
mitochondrial isoenzyme of creatine kinase. Biochem Biophys Res
Commun 16: 516--521,1964
10. Klingenberg M: Muskelmitochondrien. In: K Kramer, 0 Krayer, E
Lehnartz, A von Muralt, HH Weber (eds). Ergebnisse der Physiologie,
biologischen Chemie und experimentellen Pharmakologie. Springer,
Berlin, 1964, pp 132-189
II. Gercken G, Schlette U: Metabolite status of the heart in acute insuffciency
due to l-fluoro-2,4 dinitrobenzene. Experientia 24: 17- 19, 1968
12. Naegle S: Die Bedeutung von Creatinphosphat undAdenosintriphosphat
im Hinblick auf Energiebereitstellung, -transport und -verwertung in
normaIen und insuffzienten Herzmuskeln. Klin Wochenzeitschrift 48: 332-
341, 1970
13. Bessman Sp, Fonyo A The possible role of the mitochondrial bound
creatine kinase in regulation of mitochondrial respiration. Biochem
Biophys Res Comm 22: 597-420, 1966
14. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kineses in in vivo regulation of cellular respiration
- a synthesis. Mol Cell Biochem 133/134: 155-192, 1994
15. Tumer D, Wallimann T, Eppenberger HM:A protein that binds specifically
to the M-line of skeletal muscle is identified as the muscle form of creatine
kinase. Proc Nat! Acad Sci USA 70: 702-705, 1973
16. Wallimann T, Tumer DC, Eppenberger HM: Localization of creatine kinase
isoenzymes inmyofibrils. I Chicken skeletal muscle. J Cell Bioi 75: 297-
317,1977
17. Wallimann T: Creatinkinase-Isoenzyme und Myofibrillen-Struktur. Ph.D.
Thesis ETH No. 5437, Swiss Federal Institute of Technology, Ziirich,
1975
18. Saks VA, Rosenstraukh LV, Smimov VN, Chazov EI: Role of creatine
phosphokinase in cellular function and metabolism. Can J Physiol
PharmacoI56:691-706,1978
19. Bessman SP, Geiger PJ: Transport of energy in muscle: The phos-
phorylcreatine shuttle. Science 211: 449-452,1981
20. McClellan G, Weisberg A, Winegrad S: Energy transport from mito-
chondria to myofibril by a creatine phosphate shuttle in cardiac cells. Am
J Physiol245: C423-C427, 1983
21. Bessmann SP, Carpenter CL: The creatine-creatine phosphate energy
shuttle. Annu Rev Biochem 54: 831-862, 1985
22. Wallimann T, Eppenberger HM: Localization and function of M-line-
bound creatine kinase. M-band model and creatine phosphate shuttle. In:
JW Shay (ed). Cell and Muscle Motility Vol. 6, Plenum Publ, NY, 1985,
pp239-285
23. Kammermeier H: Why do cells need phosphocreatine and a phospho-
creatine shuttle. J Mol Cardiol19: 115-118, 1987
24. Wallimann T, Schnyder T, Schlegel J, Wyss M, Wegmann G, Rossi AM,
Hemmer W, Eppenberger HM, QuestAFG: Subcellular compartmentation
of creatine kinase isoenzymes, regulation of CK and octameric structure
ofmitochondrial CK: Important aspects of the phosphoryl-creatine circuit.
In: RJ Paul, G Elzinga, KYamada (eds). Progress in Clinical and Biological
Research. Vol. 315: 'Muscle Energetics', A.R. Liss, NY, 1989, pp 159-
176
25. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands: The
'phosphocreatine circuit' for cellular energy homeostasis. Biochem J 281:
21-40,1992
26. Wyss M, Smeitink J, Weyers RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochirn Biophys
Acta 1102: 119-166, 1992
27. Wallimann T: Dissecting the role of creatine kinase.The phenotype of
'gene knockout' mice deficient in a creatine kinase sheds new light on
the physiological functions of the 'phosphocreatine circuit'. Current Bioi
I: 42-46,1994
28. Saks VA, Ventura-Clapier R: Cellular Bioenergetics: Role of Coupled
Creatine Kinases. Mol Cell Bioi 133/134, 1994
29. Meyer R, Sweeney HL, Kushmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol246: C365-377, 1984
30. Wallimann T, Hemmer W: Creatine kinase in non-muscle tissues and cells.
Mol Cell Biochem 133/134: 193-220, 1994
31. Kaldis P, Kamp G, Piendl T, Wallimann T: Functions of creatine kinase
isoenzymes in spermatozoa. Adv Dev Biochem 5: 275-312, 1997
32. Baskin RJ, Deamer DW: A membrane-bound creatine phosphokinase in
fragmented sarcoplasmic reticulum. J Bioi Chern 245: 1345-1347, 1970
33. Levitsky DO, Levchenko TS, Saks VA, Sharov VG, Smimov: The role of
creatine phosphokinase in supplying energy for the calcium pump system
of heart sarcoplasmic reticulum. Membrane Biochem 2: 81-96, 1978
34. Rossi AM, Eppenberger HM, Volpe P, Cotrufo R, Wallimann T: Muscle-
type MM-creatine kinase is specifically bound to sarcoplasmic reticulum
and can support Ca2+-uptake and regulate local ATP/ADP ratios. J Bioi
Chern 265: 5258-5266,1990
35. Korge P, Byrd SK, Campbell KB: Functional coupling between sarco-
plasmic-reticulum-bound creatine kinase and Ca2+-ATPase. Eur J Biochem
213:973-980,1993
36. Minajeva A, Ventura-Clapier R, Veksler V: Ca 2+-uptake by cardiac
sarcoplasmic reticulumATPase in situ strongly depends on bound creatine
kinase. PfliigersArch Eur J Physiol432: 904-912,1996
37. Steeghs K, BendersA, Oerlemans F, deHaanA, HeerschapA, Ruitenbeek
W, Jost C, vanDeursen J, Perryman B, Pette D, Briickwilder M, Koudijs
J, Jap P, Veerkamp J, Wieringa B: Altered Ca2+-homeostasis in muscles
with combined mitochondrial and cytosolic CK deficiencies. Cell 89: 93-
103, 1997
38. Brosnan 1M, Chen L, Wheeler CE, vanDyke TA, Koretzky AP: Phos-
phocreatine protects ATP from a fructose load in transgenic mouse liver
expressing creatine kinase. Am J Physiol260: CII91-C1200, 1991
39. Miller K, Halow J, Koretzky AP: Phosphocreatine protects transgenic
mouse liver expressing creatine kinase from hypoxia and ischemia. Am J
Physiol265: CI544-C1551, 1993
40. Satoh S, Tanaka A, Hatano E, Inomoto T, Iwata S, Kitai T, Shinohara H,
Tsunekawa S, Chance B, Yamaoka Y: Energy metabolism and regeneration
in transgenic mouse livers expressing creatine kinase after major
hepatectomy. GastroenterolllO: 1166--1174, 1996
41. McFarland EW, Kushmerick MJ, Moerland T: Activity of creatine kinase
in a contracting mammalian muscle of uniform fibre type. Biophys J 67:
1912-1924, 1994
42. Wiseman RW, Kushmerick M: Creatine kinase equilibrium follows
solution thermodynamics in skeletal muscle: 3IP_NMR studies using
creatine analogs. J Bioi Chern 270: 12428-12438, 1995
43. VanDeursen J, Ruitenbeek W, Heerschap A, Jap P, Terlaak H, Wieringa
B: Creatine kinase (CK) in skeleteal muscle energy metabolism: A study
of mouse mutants with graded expression in muscle CK expression. Proc
NatiAcad Sci USA 91: 9091-9095,1994
44. Wallimann T: 3IP-NMR-measured creatine kinase reaction flux in muscle:
A caveat! J Muscle Res Cell Motil 17: 177-181, 1996
45. Williams JP, Headrick JP: Differences in nucleotide compartmentation
and energy state in isolated an in situ rat heart: Assessment by lIP_NMR
spectroscopy. Biochim BiophysActa 1276: 71-79, 1996
46. Dzeja PP, Zeleznikar RJ, Goldberg ND: Suppression of creatine kinase-
catalyzed phosphotransfer results in increased phosphoryl transfer by
adenylate kinase in intact skeletal muscle. J Bioi Chern 271: 12847-12851,
1996
47. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of
mitochondrial creatine kinase. Nature 381: 341-345, 1996
48. Wyss M, Wallimann T: Metabolite channelling in aerobic energy
metabolism. J theor Bioi 158: 129-132, 1992

49. Keutel HJ, Jacobs HK, Okabe K, Yue RH, Kuby SA: Studies on adenosine
triphosphate transphosphorylases. VII. Isolation of the crystalline
adenosine triphosphate - creatine transphosphorylase from calf brain.
Biochemistry 7: 4283-4290, 1968
50. McPherson A: A preliminary crystallographic investigation of rabbit
muscle creatine kinase. J Mol Bioi 81: 83-86, 1973
51. Burgess AN, Liddell JM, Cook W, Tweedlie RM, Swan IDA: Creatine
kinase. A new crystal form providing evidence of subunit structural
homogeneity. J Mol Bioi 123: 691-695,1978
52. Takasawa T, Fukushi K, Shiokawa H: Crystallization and properties of
creatine kinase from equine skeletal muscle. J Biochem 89: 1619-- 1631,
1981
53. Gilliland GL, Sjolin L, Olsson GL: Crystallization and preliminary X-ray
diffraction data of two crystal forms of bovine heart creatine kinase. J
Mol Bioi 170: 791-793, 1983
54. Hershenson S, Helmers N, Desmueles P, Stroud R: Purification and
crystallization of creatine kinase from rabbit skeletal muscle. J Bioi Chern
261:3732-3736,1986
55. SchnyderT, Sargent DF, Richmond TJ, Eppenberger HM, Wallimann T:
Crystallization and preliminary X-ray analysis of two different forms of
mitochondrial creatine kinase from chicken cardiac muscle. J Mol Bioi
216:809--812,1990
56. Schnyder T, Winkler H, Gross H, Eppenberger HM, Wallimann T:
Crystallization of mitochondrial creatine kinase. Growing oflarge protein
crystals and electron microscopic investigation of microcrystals consisting
of octamers. J Bioi Chern 266: 5318--5322, 1991
57. Furter R, Kaldis P, Furter-Graves EM, Schnyder T, Eppenberger HM,
Wallimann T: Expression of active octameric chicken cardiac mito-
chondrial creatine kinase in Escherichia coli. Biochem J 288: 771-775,
1992
58. Jones TA, Zou JY, Cowan SW, Kjelgaard M: Improved methods for
building protein models in electron density maps and the location of errors
in these models. Acta CrystA47: 110--119, 1991
59. Brunger AT: X-PLOR Version 3.1 Manual. Yale University Press, New
Haven, USA, 1992
60. Kabsch W, Sander C: Dictionary of protein secondary structure: pattern
recognition of hydrogen-bonded and geometrical features. Biopolymers
22:2577-2637,1983
61. Gross M: The tryptophan residues of mitochondrial creatine kinase: Roles
in enzyme structure and function. Ph.D.Thesis ETH No. 10719, Swiss
Federal Institute of Technology, Ziirich, 1994
62. Wyss M, James P, Schlegel J, Wallimann T: Limited proteolysis of creatine
kinase. Implications for the three-dimensional structure and for conform-
ational substates. Biochemistry 32: 10727-10735, 1993
63. Morris GE, Jackson PJ: Identification by protein micro sequencing of a
proteinase-V8-c1eavage site in a folding intermediate of chick muscle
creatine kinase. Biochem J 280: 809--811, 1991
64. Gross M, Wyss M, Furter-Graves EM, Wallimann T, Furter R: Recon-
stitution of active octameric mitochondrial creatine kinase from two
genetically engineered fragments. Protein Sci 5: 320--330, 1996
65. Schlegel J, Zurbriggen B, Wegmann G, Wyss M, Eppenberger HM,
Wallimann T: Native mitochondrial creatine kinase (Mi-CK) forms
octameric structures. 1. Isolation of two interconvertible Mi-CK isoforms:
Dimeric and octameric Mi-CK. J Bioi Chern 263: 16942-16953, 1988
66. Gross M, Wallimann T: Kinetics of assembly and dissociation of the
mitochondrial creatine kinase octamer. A fluorescence study. Biochemistry
32: 13933--13940, 1993
67. Gross M, Furter-Graves EM, Wallimann T, Eppenberger HM, Furter R:
The tryptophan residues of mitochondrial creatine kinase: Roles ofTrp-
223, Trp-206, and Trp-264 in active-site and quaternary structure
formation. Protein Sci 3: 1058--1068,1994
68. Kaldis P, Furter R, Wallimann T: The N-terminal heptapeptide of
mitochondrial creatine kinase is important for octamerization. Bio-
139
chemistry 33: 952-959, 1994
69. Beuchter DD, Medzihradsky KF, BuriingameAL, Kenyon GL: The active
site of creatine kinase. Affinity labeling of cysteine 282 with epoxycreatine.
J Bioi Chern 267: 2172-2178,1992
70. Hou LX, Vollmer S: The activity ofS-thiomethyl modified creatine kinase
is due to the regeneration of free thiol at the active site. Biochim Biophys
Acta 1205: 83--88, 1994
71. Stroud RM: Balancing ATP in the cell. Nature Struct Bioi 3: 567-569,
1996
72. Furter R, Furter-Graves EM, Wallimann T: Creatine kinase: The reactive
cysteine is required for synergism but is nonessential for catalysis.
Biochemistry 32: 7022-7029,1993
73. Rosevear PR, Desmeules P, Kenyon GL, MildvanAS: Nuclear magnetic
resonance studies of the role of histidine residues at the active site of
rabbit muscle creatine kinase. Biochemistry 20: 6155-6164,1981
74. Cook PF, Kenyon GL, Cleland WW: Use of pH studies to elucidate the
catalytic mechanism of rabbit muscle creatine kinase. Biochemistry 20:
1204--1210,1981
75. Forstner M, MiillerA, Stolz M, Wallimann T: The active site histidines of
creatine kinase. A critical role of His61 situated on a flexible loop. Prot
Sci 6: 331-339, 1997
76. Chen LH, Borders CL, Vasquez JR, Kenyon GL: Rabbit muscle creatine
kinase: Consequences of the mutagenesis of conserved histidine residues.
Biochemistry 35: 7895-7902,1996
77. Messmer CH, Kagi JHR: Tryptophan residues of creatine kinase: A
fluorescence study. Biochemistry 24: 7172-7178,1985
78. Miihlebach SM, Gross M, Wirz T, Wallimann T, Perriard JC, Wyss M:
Sequence homology and structure predictions of the creatine kinase
isozymes. Mol Cell Biochem 133/134: 245-262,1994
78a. Vasitk M, Nagayama K, Wiithrich K, Mertens M, Kagi JHR: Creatine
kinase. Nuclear magnetic resonance and fluorescence evidence for
interaction of adenosine-5 '-diphosphate with aromatic residue(s).
Biochemistry 18, 5050--5055,1979
79. Forstner M, Kriechbaum M, Laggner P, Wallimann T: Changes of creatine
kinase structure upon ligand binding as seen by small-angle scattering. J
Mol Struct 383: 217-222,1996
80. Jacobus WE, Lehninger AL: Creatine kinase of rat heart mitochondria.
Coupling of creatine phosphorylation to electron transport. J Bioi Chern
248:4803-4810,1973
81. SchnyderT, Rojo M, Furter R, Wallimann T: The structure of mitochondrial
creatine kinase and its membrane binding properties. Mol Cell Biochem
133/134: 115-123, 1994
82. Saks VA, Kuznetsov AV, Kupriyanov VV, Miceli MV, Jacobus WE:
Creatine kinase of rat mitochondria: The demonstration of functional
coupling to oxidative phosphorylation in an inner membrane matrix
preparation. J Bioi Chern 260: 7757-7764, 1985
83. Stachowiak 0, Dolder M andWallimannT: Membrane-binding and vesicle
cross-linking kinetics of the mitochondrial creatine kinase octamer.
Biochemistry 35: 15522-15528,1996
84. Lipskaya TY, Tempi VD, Belousova LV, Molokova EV, Rybina IV:
Investigation of the interaction of mitochondrial creatine kinase with the
membranes of the mitochondria. Biochimia USSR 45: 877-886, 1980
85. Miiller M, Moser R, Cheneval D, Carafoli E: Cardiolipin is the membrane
receptor for mitochondrial creatine phosphokinase. J Bioi Chern 260:
3839--3843, 1985
86. Brooks SPJ, Suelter CH: Asociation of chicken mitochondrial creatine
kinase with the inner mitochondrial membrane. Arch Biochem Biophys
253: 122-132, 1987
87. Rojo M, Hovius R, Demel RA, Wallimann T: Interaction of mitochondrial
creatine kinase with model membranes. A monolayer study. FEBS Lett
281: 123--129, 1991
88. Cheneval D, Carafoli E, Powell GL, Marsh D: A spin-label electron spin
resonance study of the binding of mitochondrial creatine kinase to
140
cardiolipin. Eur J Biochem 186: 415-419,1989
89. De KruijffB, Verkleij AJ, van Echteld CJA, Gerritsen WJ, Noordam PC,
Mombers C, Rietveld A, de Gier J, Cullis PR, Hope MJ, Nayar R: Non-
bilayer lipids and the inner mitochondrial membrane. In: HG Schweuger
(ed). International Cell Biology. Springer-Verlag, Berlin, 1981, pp 559-
671
90. Rietveld AG, Koorengevel MC, de Kruijff B: Non-bilayer lipids are
required for efficient protein transport across the plasma membrane of
Escherichia coli. EMBO J 22: 5506-5513,1995
91. Font B, Eichenberger D, Goldschmidt D, Vial C: Interaction of creatine
kinase and hexokinase with the mitochondrial membranes, and self-
association of creatine kinase: crosslinking studies. Mol Cell Biochem
78: 131-140, 1987
92. De Pinto V, Palmieri F: Transmembrane arrangement of mitochondrial
porin or voltage-depemdant anionchannel (VDAC). J Bioenerg Biomembr
24:21-26,1992
93. Kottke M,Adams V, Wallimann T, Nalam VK, Brdiczka D: Location and
regulation of octameric mitochondrial creatine kinase in the contact sites.
Biochem BiophysActa 1061: 215-225,1991
94. Brdiczka D, Kaldis P, Wallimann T: In vitro complex formation between
the octamer of mitochondrial creatine kinase and porin. J Bioi Chern 269:
27640-27644, 1994
95. Beutner G, RiickA, Riede B, Welte W, Brdiczka D: Complexes between
kinases, mitochondrial porin and adenylate translocator in rat brain
resemble the permeability transition pore. FEBS Lett 396: 189-195, 1996
96. Beyer K, Klingenberg M: ADP/ATP carrier protein from beef heart
mitochondria has high amounts of tightly bound cardiolipin, as revealed
by 31p nuclear magnetic resonance. Biochemistry 24: 8321--8326, 1985
97. Drees M and Beyer K: Interaction of phospholipids with the detergent-
solubilizedADPIATP carrier protein as studied by spin-label electron spin
resonance. Biochemistry 27: 8584--8591, 1988
98. Cheneval D, Carafoli E: Identification and primary structure of the
cardiolipin-binding domain of mitochondrial creatine kinase. Eur J
Biochem 171: 1-9, 1988
99. Saks VA, Khuchua ZA, Kuznetsov AV: Specific inhibition of ATP-ADP
translocase in cardiac mitoplasts by antibodies against mitochondrial
creatine kinase. Biochim BiophysActa 891: 138--144, 1987
100. SchnyderT, EngelA, LustigA, Wallimann T: Native mitochondrial creatine
kinase (Mi-CK) forms octameric structures. II. Characterization of dimers
and octamers by ultracentrifugation, direct mass measurement by STEM
and image analysis of single Mi-CK octamers. J Bioi Chern 263: 16954--
16962, 1988
101. SchnyderT, Gross H, Winkler H, Eppenberger HM, Wallimann T: Structure
ofthe mitochondrial creatine kinase octamer: High resolution shadowing
and image averaging of single molecules and formation oflinear filaments
under specific staining conditions. J Cell Bioi 112: 95-101, 1991
102. Schnyder T, Cyrklaff M, Fuchs K, Wallimann T: Crystallization of
mitochondrial creatine kinase on negatively charged lipid layers. J Struct
Bioi 112: 136-147, 1994
103. Wyss M, Maugham D, Wallimann T: Re-evaluation of the structure and
physiological function of guanidino kinases in fruitfly (Drosophila),
sea urchin (Psamechinus miliaris) and man. Biochem J 309: 255-261,
1995
104. Quemeneur E, Eichenberger D, Goldschmidt D, Vial C, Beauregard G,
Potier M: The radiation inactivation method provides evidence that
membrane-bound mitochondrial creatine kinase is an oligomer. Biochem
BiophysResComm 153: 1310-1314, 1988
105. Fedosov SN, Belousova LV, Plesner IW: A model of mitochondrial creatine
kinase binding to membranes: Adsorption constants, essential amino acids
and the effect of ionic strength. Biochem Biophys Acta 1153: 322-330,
1993
106. Bogner W, Aquila H, Klingenberg M: The transmembrane arrangement
of the ADPIATP carrier as elucidated by the lysine reagent pyridoxal 5-
phosphate. Eur J Biochem 161: 611--620,1986
107. Ovadi J: Cell Architecture and Metabolic Channeling. Springer-Verlag,
Heidelberg, 1995
108. Anderson KS, KimAY, Quillen JM, Sayers E, Yang XJ, Miles EW: Kinetic
characterization of channel impaired mutants of tryptophan synthase. J
Bioi Chern 270: 29936-29944, 1995
109. Chumachenko YV: Some evidence in favour of the partnership between
rabbit muscle aldolase and glyceraldehyde 3-phosphate dehydrogenase
in the consecutive reactions. Ukr Biokhim Zh 66: 52-70, 1994
110. Agius L, Sherratt HSA (eds).: Channelling in Intermediary Metabolism.
Portland Press, London and Miami, 1997
III. Gellerich FN, Schlame M, Bohnensack R, Kunz W: Dynamic compart-
mentation of adenine nucleotides in the mitochondrial intermembrane
space of rat-heart mitochondria. Biochim BiophysActa 890: 117-126, 1987
112. Gellerich FN, Kunz W: Cause and consequences of dynamic compart-
mentation of adenine nUcleotides in the mitochondrial intermembrane
space in respect to exchange of energy rich phosphates between cytosol
and mitochondria. Biomed Biochim Acta 46: 545-548, 1987
113. Yount RG, Lawson D, Rayment I: Is Myosin a 'back door' enzyme?
Biophys J 68: 44s--49s, 1995
114. Gilson MK, Straatsma TP, McCammon JA, Ripoll DR, Faerman CH,
Axelsen PH, Silman 1, Sussman JL: Open 'back door' in a molecular
dynamics simulation of acetylcholinesterase. Science: 263,1276-1278,
1994
115. Hyde CC, Ahmed SA, Padlan EA, Miles EW, Davies DR: Three-
dimensional structure of the tryptophan synthase a2~2 multienzyme
complex from Salmonella typhimurium. J Bioi Chern 263: 17857-17871,
1988
116. Boehm EA, Radda GK, Tomlin H, Clark JF: The utilisation of creatine
and its analogues by cytosolic and mitochondrial creatine kinase. Biochim.
Biophys.Acta 1274: 119-128, 1996
117. Kenyon GL: Creatine kinase shapes up. Nature 381: 281-282, 1996
118. Milner-White EJ, Watts DC: Inhibition of adenosine 5 '-triphosphate-
creatine phosphotransferase by substrate-anion complexes. Evidence for
the transition-state organization of the catalytic site. Biochem J 122: 717-
740, 1971
119. Chida K, Tsunenaga M, Kasahara K, Kohno Y, Kuroki T: Regulation of
creatine phosphokinase B activity by protein kinase C. Biochem Biophys
Res Commun 173: 346-350,1990
120. Hemmer W, Skarli M, Perriard JC, Wallimann T: Effect of okadoic acid
on protein phosphorylation patterns of chicken myogenic cells with special
reference to creatine kinase. FEBS Lett 327: 35-40, 1993
121. Martin KJ, Chen SF, Clark GM, Degen D, Wajima M, Von Hoff DD,
Kaddurah-Daouk R: Evaluation of creatine analogues as a new class of
anticancer agents using freshly explanted human tumor cells. J Nat! Cancer
Inst86:608--613,1994
122. Teicher BA, Menon K, Northey D, Liu J, KuDe DW, Kaddurah-Daouk
R: Cyclocreatine in cancer chemotherapy. Cancer Chemother Pharmacol
35: 411-416,1995
123. Sayle RA, Milner-White EJ: RASMOL: Biomolecular graphics for all.
Trends Biochem Sci 20: 374, 1995
124. Van Dorsten FA, Wyss M, Wallimann T, Nicolay K: Activation of sea-
urchin sperm motility is accompanied by an increase in the creatine kinase
exchange flux. Biochem J 325: 411-416, 1997
125. Aliev MK, Saks VA, Compartmentalized energy transfer in cardio-
myocytes: Use of mathematical modeling for analysis of in vivo regulation
of respiration. Biophys J 73: 428-445, 1997
126. Zhou G, Parthasarathy G, Somasundaram T, Ables A, Roy L, Strong SJ,
Ellington WR, Chapman MS: Expression, purification from inclusion
bodies, and crystal characterization of a transition state analog complex
of arginin kinase: A model for studying phosphagen kinases. Prot Sci 6:
444-449, 1997
127. Vacheron MJ, Clottes E, Chautard C, Vial C: Mitochondrial creatine kinase
interaction with phospholipid vesicles. Arch Biochem Biophys 344: 316-
324, 1997
Molecular and Cellular Biochemistry 184: 141-151, 1998.
1998 Kluwer Academic Publishers.

Oligomeric state and membrane binding behaviour

of creatine kinase isoenzymes: Implications for
cellular function and mitochondrial structure

Olaf Stachowiak, Uwe Schlattner, Max Dolder, and Theo Wallimann

Swiss Federal Institute o/Technology, Institute o/Cell Biology, ETH Ziirich-Honggerberg, Ziirich, Switzerland

Abstract
The membrane binding properties of cytosolic and mitochondrial creatine kinase isoenzymes are reviewed in this article.
Differences between both dimeric and octameric mitochondrial creatine kinase (Mi-CK) attached to membranes and the unbound
form are elaborated with respect to possible biological function. The formation of crystalline mitochondrial inclusions under
pathological conditions and its possible origin in the membrane attachment capabilities of Mi-CK are discussed. Finally, the
implications of these results on mitochondrial energy transduction and structure are presented. (Mol Cell Biochem 184: 141-
151, 1998)

Key words: Mi-CK, dimer/octamer equilibrium, membrane binding domains, lipid vesicle cross-linking, VDAC (porin), ANT

Introduction from mitochondria to sites of energy consumption are long (for

Cells with high and fluctuating energy metabolism usually co-
express a cytosolic and a mitochondrial creatine kinase
isoenzyme (CK, EC 2.7.3.2) to transphosphorylate the high-
energy phosphoguanidino compound phosphocreatine (PCr),
which is used as an energy-storage and -transport metabolite,
toADP, thereby restoring theATP-pool ofthe cell (for review
see [1-4]). Since the concentrations ofATP andADP and their
local ratio in the cell control many metabolic processes (by
allosteric interactions and phosphorylation potential), these
parameters have to be held relatively constant within their
physiological range. At rest, ATP and ADP are present in the
3-5 mM range and 10-30 J.lM range, respectively. At times of
high workload, the cell can immediately regenerate ATP as it
is consumed from the PCr pool (30-40 mM) via the action of
creatine kinase. This is called temporal energy buffering. A
further advantage of creatine kinase function is providing
metabolic capacity [5, 6] which means that using the CK
system, the cell is able to provide facilitated diffusion ofATP
andADP to sites of demand thereby creating a spatial buffering
effect. This concept is especially important for polar cells, e.g.
spermatozoa and photoreceptor cells where diffusion distances
review see [2, 7, 8]).
The CK system can be found in vertebrates, whereas in
invertebrates other guanidinokinases are expressed, such as
arginine, glycocyamine, lumbricine and taurocyamine kinase
[9-11].
To exert its function in an optimized manner, the CK
system is compartmentalized, e.g. there are dimeric CK
isoforms which are exclusively found in the cytosol compart-
ment, like MM-CK (muscle), BB-CK (brain) and MB-CK,
the only known heterodimeric in vivo CK-isoform which is
expressed during muscle development and to some extent
also in adult cardiac muscle, whereas mitochondrial CK (Mi-
CK) isoforms are strictly localized within the intermembrane
space of mitochondria [3, 12]. In contrast to the cytosolic
isoforms, Mi-CKs can occur as dimers and octamers, the latter
being built up by association of two dimers to instable
tetramers, which then react in a fast step to form octamers
[13]. Whereas the cytosolic CK isoforms mainly use PCr to
reproduce ATP at sites of high energy consumption, such as
the myofibrillar actomyosin ATPase in muscle [14, 15], the
Ca2+-ATPase of the sarcoplasmic reticulum [16, 17] or the
rods of photoreceptor cells [18], the mitochondrial isoform

Address/or offprints: T. Wallimann, Swiss Federal Institute ofTechnology, Institute of Cell Biology, ETH Ziirich-Hiinggerberg, CH-8093 Ziirich, Switzerland
142
is mainly responsible for the turnover of ATP which is
exported out of the mitochondrial matrix by the adenine
nucleotide translocase (ANT, synonymesADTl orAAC; for
review see [19, 20]), thereby forming PCr which is exported
into the cytosol via porin of the outer mitochondrial membrane
and then used at sites of high energy demand by the reverse
CK reaction yieldingATP and Cr. This postulated exchange
of PCr/Cr and ATP/ ADP between the cytosol and the
mitochondrion is referred to as 'creatine phosphate shuttle'
or 'phosphocreatine circuit' [1,2,21]. In the above mentioned
photoreceptor cells, Mi-CK is localised in the mitochondria
clustered exclusively in the inner segment of the cell body
[22], whereas BB-CK is localised within the outer cell-
segments which do not provide energy by oxidative phospho-
rylation [18]. For there are several energy-consuming
conversions taking place in the outer segments, such as the
photic process, CK is supposed to provide energy-buffering
capacity by regenerating ATP being an important prerequisite
for the reformation of GTP [18]. Hence, this is an example
for a complete PCr circuit within polarized cells. A similar
function of the CK system with respect to facilitated diffusion
of ATP and transport of high-energy phosphate from the
mitochondria to the dynein-ATPase is postulated to take place
in spermatozoa of certain species, e.g. sea urchin (for review
see [8]).
To understand the role of octameric Mib-CK between the
mitochondrial membranes, it has to be taken into consideration
that the mitochondrial isoenzyme forms highly symmetrical
cube-like octamers (P422 symmetry) possessing identical top
and bottom faces [23-26]. This particular arrangement is
supposed to be one of the Mi-CK properties facilitating
membrane cross-linking at mitochondrial contact-sites [27].
The interconvertible oligomeric states of Mi-CK
For both types of mitochondrial creatine kinase isoenzymes
(Mia-CK, mainly expressed in non-muscle tissues and in
brain, and sarcomeric Mib -CK mainly expressed in skeletal
muscle and in heart) the oligomeric state of the enzyme is
supposed to be important with respect to the formation of
membrane contact-sites, e.g. a close approximation of (i)
inner and outer mitochondrial membrane and (ii) adjacent inner
membranes only (inter-cristae contacts). In the following
chapters, we will concentrate on chicken Mib -CK which has
a more basic pI than Mia-CK [28, 29]. For this isoform the
dimer/octamer interconversion has been extensively studied
in vitro by fluorescence spectroscopy [13]. The decay of the
octamer in the presence of the transition-state analog complex
(TSAC) consisting of Mg-ADP, Cr and nitrate leads to a
fluorescence decrease of about 25%, caused by the quenching
of the dimer/dimer interface tryptophan residue (Trp264)
which upon dimerisation becomes exposed to the aequous
solution [30, 31]. With the rate constant of the decay being 0.19
min- 1 and the remaining octamer-content at equilibrium being
about 5%, TSAC is very efficient at quantitatively dissociating
the octamer (see Fig. 1). The reason for this effect is the
trigonal-planar nitrate ion which mimicks the transition state
of the terminal phosphate of ATP when being transferred to
creatine (see planar transition-state in SN2 reactions) thereby
locking the enzyme in a conformation which destabilizes the
octamer. The treatment of octameric Mi-CK with radicals like
peroxynitrite completely abolishes its activity and also the
TSAC effect, showing that an intact acitve-site is a pre-
requisite for the substrate analog-induced decay [31a].
Although a number of parameters like pH, ionic strength and
substrates have been shown to affect the dimer/octamer
equilibrium [29, 32], the in vivo signal for the decay of the
octameric isoform into dimers has not yet been identified.In
a)
F/cps
Oetamer
EB
TSAC Dimer
tf s
b)
Ffcps
Oetamer
Tetramer
EB
Dimer
slow fast
tf s
Fig. 1. Schematic representation of the TSAC-induced octamer decay of
Mi-CK in free solution monitored by fluorescence spectroscopy (a) and
reoligomerisation of the dimers upon removal of the TSAC complex (by
addition of EDT A, to complex TSAC-magnesium, (b) The bimolecular
association rate constant was determined to 318 M-i s 1, the dissociation
rate constant was shown to be 0.19 min 1 [13].

vitro measurements have proven the destabilizing effect of
high pH, high salt concentration, dilution of the protein and
low temperature on the octamer, the latter being a consequence
of the hydrophobic contribution of the dimer/dimer interfaces
[33]. It has to be pointed out that the assembly ofthe octamer
could be an almost irreversible reaction in vivo. Whereas the
disassembly of the octamer happens in an all-or-nothing
fashion, the reformation of the latter takes place via the
tetramer, which is a short-lifetime intermediate in solution and
therefore cannot be detected by gel-permeation chroma-
tography [13]. It may be speculated that the octameric and/
or tetrameric isoform is stabilized by mitochondrial mem-
branes, which have a high affinity for Mi-CK. Rebinding
experiments with dimeric CK [32] suggest a role of the
membrane in reoligomerisation of dimeric Mib -CK. As far as
the kinetic parameters are concerned, there are only subtle
differences in Km and Vmax between octameric and dimeric
isoforms ofMib-CK [34] (with three-fold higher Km(Cr) for
the octamer) showing that the formation of the octamer has
structural rather than catalytic advantages. One of the effects
which could be relevant may be of cooperative nature.
Nevertheless, experiments elucidating cooperative substrate
binding kinetics within the octamer are still missing. They
could be a crucial step towards a better understanding ofMi-
CK regulation between the mitochondrial inner and outer as
well as the cristae membranes.
On treating Mib -CK with chaotropic agents, such as guan-
idinium hydrochloride, the monomer is yielded. However, as
soon as the enzyme is solved under non-denaturing conditions,
an active dimer is reformed again, showing thatthe monomer
is functionally irrelevant under in vivo conditions and may
only be found as an intermediate state towards the dimer after
the TOM (translocase of outer membrane )-mediated import
ofCK pre-protein sequence into the intermembrane space and
the cleavage of the signal-peptide [35] (for review on
mitochondrial protein import see [36]).
Membrane binding and functional coupling of cytosolic
CK isoenzymes
The interaction of cytosolic CK isoforms with membranes
plays a particularly important role in the energetic supply of
ion-pumping systems. An important feature of these (dimeric)
isoenzymes is their functional coupling to ATPases like the
Ca2+-ATPase in the SR membrane [16, 17,37-39] as well as
the ouabain-sensitive Na+/K+-ATPase in kidney [40] and in
the postsynaptic membrane of electrocytes in the electric
organ of Torpedo, an electric fish [41]. Originally derived
from myogenic cells, the electric organ is composed oflarge
(each 0.1 mm of height), disc-shaped electrocytes which are
aligned on top of each other to form the columns of the organ.
The ventral postsynaptic membrane of the electrocyte
143
contains a large number of nicotinic acetylcholine receptors
(receptor-gated ion channels) which allow sodium influx into
the cell on binding acetylcholine. This happens when the fish
is producing an electric discharge for the purpose of killing
its prey or defending itself. To restore the intracellular resting
conditions after the discharge, the sodium ions are extruded
from the cell by Na+/K+-ATPase action [41]. The ATPase is
located within the tube-shaped, invaginated parts ofthe dorsal
electrocyte membrane face (called canniculi) and is driven
by the hydrolysis of ATP generated by the CK-catalysed
reaction of ADP and PCr [42]. Accordingly, the role of CK
in electrocytes is to keep the local ATP/ ADP ratios high and
thus make ion pump function near its maximal efficiency by
co-localisation of CK with the ATPase [2, 7, 42, 43]. Since
the drop of PCr concentration on rechargement of the
electrocyte can be abolished by specifically blocking the Na+/
K+-ATPase with ouabain, it can be concluded that the high
intracellular PCr concentration in electrocytes is mainly used
for the fueling of sodium extrusion out of the cell [43, 45].
This is corroborated by the immunohistochemical localisation
of CK along the ventral postsynaptic membrane as well as on
the dorsal membrane system of electrocytes where high
concentrations of Na+/K+ ATPase are found [44]. Direct
evidence for the tight coupling between the CK reaction and
the Na+/K+-ATPase was found by in vivo 3IP_NMR saturation
transfer measurements [45] which showed a highly increased
flux through the creatine kinase reaction on recharging the
electric organ.
Confusion arose as antibody labeling experiments with
heterologous anti-chicken B-CK antibodies indicated that CK
in the electric organ was brain-type B-CK, but later the same
protein was identified as genuine M-CK [44]. Sequence
comparison with 20 other known CK-sequences then showed
that electric organ CK, first called 'acetylcholine receptor-rich
membrane-associated peripheral v 2 -protein', contained
indeed both B-and M-CK sequence motifs [46].
In muscle, MM-CK has been proven to be specifically
associated with the SRmembrane [37, 38, 39] and the T-tubule
system by immuno-gold labeling of SR membrane vesicles as
well as in situ immuno-gold labeling of permeabilized muscle
[39]. Treatment of isolated SR vesicles with high-salt and
low-salt/EDTA did not cause a striking release of CK
activity [39], leading to the conclusion that CK is either
tightly associated with the membrane or bound to an SR
protein component. Regarding Ca 2+-ATPase function in the
sarcoplasmic reticulum, M-CK is able to maitain a high local
ATP/ ADP ratio necessary for ATP-dependent calcium
pumping [39, 47] and provides much more efficient energy
supply than external systems like pyruvate kinase [17] due
to its colocalisation and coupling with the pump [44,47]. This
effect is of special importance in 'emergency situations'
where a high cytoplasmic Ca 2+ concentration has to be
diminished by sequestering Ca 2+ as rapidly as possible into
144

the SR [17, 48]. Additionally, the role ofMM-CK in the SR
may not be exclusively restricted to energy-supply for
calcium pumping. As shown by Ferris and co-workers,
external ATP can allosterically regulate the inositol 1,4,5-
trisphosphate receptor-mediated efflux of Ca2+ from the SR
[49] which would suggest a regulatory impact of the CK
reaction in the SR. Furthermore, the phosphate-induced
hampering of calcium uptake was shown to be minimized by
addition of PCr also proving a regulatory role of the CK
system [50]. Very recent investigations which were carried
out with intact SR-structures in skinned muscle-fibres also
support the finding that CK-produced ATP, probably being
formed in a microenvironment close to the ATP binding-site
of the calcium pump, is much more efficient in driving
calcium uptake into the SR than externally generated ATP
[17].
Unlike the CK isoforms mentioned above, sea-urchin
sperm-tail CK (TCK) from Strongylocentrotus purpuratus
is able to insert into lipid bilayers via the rare C'4-lipid
compound myristic acid [51] which is attached to the amino-
terminus of the enzyme and seems to playa role in targeting
the enzyme to the flagellum during spermiogenesis. Two co-
existing pools of TCK (TCKI and TCKII), both with a
molecular weight of approximately 145 kD have been
identified [51]. Whereas TCKII turned out to be much more
efficient in liposome binding in vitro due to its myristoylation,
non-myristoylated TCKI was shown to be convertible to
TCKII in vitro by administration ofN-myristoyl transferase
[52]. Hence, the rate ofmyristoylation turns out to be a crucial
factor controlling the membrane association behaviour ofthe
entire TCK pool in the sperm cell. These data indicate that a
certain proportion, varying in quantitative terms within
different cell types, of dimeric 'cytosolic' CK can indeed be
specifically bound to cellular membranes.
Membranes as receptors for Mi-CK
In contrast to the cytosolic CK isoforms, the mitochondrial
CK isoenzymes (Mia-CK and Mib-CK) are the only known

Fig. 2. Putative membrane binding motifs of sarcomeric Mi-CK. Side view of the chicken sarcomeric Mi-CK (Mib-CK) octamer; the polypeptide chain is
shown in the backbone trace representation provided by the rendering software RASMOL v. 2.6. Putative membrane binding motifs which are symmetrically
exposed at the top and bottom faces of the octamer are indicated in red and blue: The C-terrninal domain (red, Asp 357 - Lys 380, including 5 Lys and 1 Arg)
and a short internal stretch (blue, Ala 107 - Gin 115, including 1 Lys and 1 His). The N-terrninal sequence (Tier 1 - Met 25) which is buried inside the octamer
is shown in yellow (for further details see [23, 89]).

members of the guanidino kinase family which can also be
found in the octameric state. The following considerations
will be restricted to chicken Mib -CK, the basic mitochondrial
isoform (pI> 9) from sarcomeric muscle because most ofthe
experiments done so far have been performed with this
isoform. Mi-CKs are known to bind strongly to the outer
surface of the inner mitochondrial membrane [12] and, in the
case of sarcomeric Mi-CK (Mib-CK), to bridge inner and
outer mitochondrial membranes [25,27]. One characteristic
feature of octameric Mib-CK is its high affinity for both the
inner and outer mitochondrial membrane which was measured
by membrane surface-pressure determination [27, 53] and
rebinding experiments with Mi-CK-depleted mitoplasts [32].
Furthermore, acidic phospholipids such as cardiolipin
facilitate the 2D-crystallization of the Mib -CK octamer as
shown by Schnyder et al. [24, 26]. In contrast to flagellar
sperm TCK, where myristoylation provides a membrane
anchor that inserts into the membrane (see above), such
evidence is missing for Mib -CK. Most probably, the C-termini
of Mi-CK, which contain clusters of basic amino acids (Lys
and Arg, see Fig. 2), are responsible for the tight binding of
octameric Mib -CK to mitochondrial membranes, or at least
play an important role in membrane docking of the enzyme
[89]. As a membrane receptor for Mib-CK, cardiolipin (CL),
a characteristic (acidic) phospholipid of the inner mitochondrial
membrane, has first been identified [54].
Interaction of Mi-CK with outer mitochondrial membranes
and porin (VDA C)
The mitochondrial envelope can form contact sites, a close
superposition of the two mitochondrial boundary membranes
which can be visualized by freeze- fracturing (for review see
[55]). Contact sites are dynamic structures and their formation
correlates with activity of oxidative phosphorylation [56] and
is regulated by [ADP] [57]. Mitochondrial contact sites are
enriched in Mi-CK and also contain ANT and a third protein,
the outer membrane porin or voltage dependent anion channel
(VDAC) [58, 59]. VDAC is responsible for the permeability
of the outer mitochondrial membrane (for reviews see [60,
61]) but can adopt a low conductance cation-selective state
which could account for a metabolic compartmentation of
nucleotides in the intermembrane space, especially at the
contact sites (for review see [62]). Extramitochondrial
creatine, which could permeate cation-selective VDAC,
effectively stimulates the mitochondrial phosphocreatine
synthesis via Mi-CK and finally oxidative phosphorylation
[63], suggesting a functional coupling between VDAC and
Mi-CK [2].
It was proposed that contact sites operate as micro-
compartments or multi enzyme complexes for energy export
(for reviews see [55, 64]). Such a compartment would confer
145
a thermodynamic advantage to the Mi-CK reaction, since
products of the Mi-CK reaction, phosphocreatine and ADP,
are constantly removed via VDAC and ANT [65]. The
microcompartment model suggests that Mi-CK, which is in
general only attached to the inner mitochondrial membrane,
binds in contact sites to the outer mitochondrial membrane
as well. It should be noted, however, that Mi-CK is not a
prerequisite for contact site formation which also occurs in
tissues exempt from Mi-CK, e.g. the liver.
The micro-compartment-model ofMi-CK is strongly sup-
ported by in situ localization ofMi-CK showing Mi-CK along
the cristae membranes and between inner and outer mem-
branes [22] as well as by in vitro studies demonstrating that
Mi-CK interacts not only with inner but also with outer mito-
chondrial membrane preparations [53] and, most importantly,
is able to bridge two membranes [27]. By this bridging
capability, Mi-CK is well suited for contact site localization.
However, despite the body of evidence supporting an interac-
tion of Mi-CK with the outer membrane and functional in-
teraction with VDAC, it is still debated how these interactions
occur and what their structural basis is. A spatial association
of Mi-CK with VDAC has been shown in digitonin-treated
mitochondria [59] and in vitro complex formation ofMi-CK
with VDAC [66] and ANT [59, 67] has been demonstrated.,
In the complexed state, VDAC induces octamerization of a
polymerization-deficient Mi-CK mutant [66].
Although these results suggest a rather direct physical
interaction between Mi-CK and VDAC, this could not be
confirmed by crosslinking experiments so far [25]. However,
since top and bottom faces of the Mi-CK octamer are identical,
it can be assumed that Mi-CK binding to the outer membrane
can also occur by electrostatic interaction of Mi-CK with
negative charges of outer membrane phospholipid headgroups.
According to current transmembrane models ofthe mammalian
VDAC structure [68], only three negatively charged amino
acids together with several positive charges of VDAC are
facing the intermembrane space. In addition, the N-terminal
negative charges of VDAC are not accessible in VDAC-
containing Triton micelles [68]. Therefore, it has been
suggested that the association between Mi-CK and VADC
could be due to additional hydrophobic interactions [66].
In an alternative model, maintaining the well known
electrostatic binding behaviour of Mi-CK, a close co-
localization between Mi-CK and VDAC could be mediated
via charged phospholipids, in analogy to the presumed Mi-CK-
ANT interaction [25]. Information on the lipid composition of
mitochondrial outer membranes is scarce but its cardiolipin
content is assumed to be low at least in liver mitochondria
[69]. Thus, for the interaction of the enzyme with outer
membrane, binding ofMi-CK to different other phospholipids
[53] could become important. Certainly further research is
needed to characterize the in vivo interaction between Mi-CK
and VDAC in the outer mitochondrial membrane.
146
Interaction of Mi-CK with inner mitochondrial membranes
and ANT
Binding of Mi-CK to membranes involves low and high
affinity binding sites [70]. While reconstitutedANT did not
increase Mi-CK binding [54] and other membrane proteins
inhibited binding [71] quite efficiently, several studies
uncovered an interaction of Mi-CK with membrane phos-
pholipids [54, 71, 72]. However, it is still debated which
phospholipids are involved and how the Mi-CK/phospholipid
interface is organized.
Out of different phospholipids examined in artificial
membrane systems, only cardiolipin was found to convey high
affinity binding sites for Mi-CK [54]. Mi-CK-binding to
mitochondria could be drastically inhibited by a prior
modification of cardiolipin by (i) the cardiolipin-specific drug
adriamycin [54] or (ii) treatment with phospholipase A2, but
not with phospholipase C which is not acting on cardiolipin
[70]. Preferential binding of Mi-CK to cardiolipin could
explain the specificity ofMi-CK for mitochondrial membranes,
since no other cellular membrane contains this phospholipid
in significant amounts [69]. It was concluded that cardiolipin
is the only membrane receptor for Mi-CK [54]. However, when
Mi-CK-membrane interactions were determined by their
influence on the surface pressure of model membranes, a
pressure increase was not only detectable with cardiolipin
membranes but also, to a minor degree, with other charged
phospholipids (e.g. phosphatidylserine or phosphatidyl-
inositol), preparations of inner or outer mitochondrial
membranes and even with microsomal membranes known to
be devoid of cardiolipin [53]. Obviously, not only cardiolipin
but also several other negatively charged phospholipids are
capable to interact with Mi-CK in vitro. Accordingly, the
binding of Mi-CK to cardiolipin cannot be called absolutely
specific, but, because of its high abundance in the inner
membrane and its two negative charges, it nevertheless seems
to be the most important interacting lipid. This is especially
relevant in the presence of divalent cations such as magnesium
or calcium which have been shown to induce a clustering of
acidic phospholipids such as cardiolipin [73], thereby causing
phase separation in lipid model membranes where fatty acid
chains of different length and degree of saturation are present
in the acidic and non-acidic lipid species.
It has to be pointed out that in vitro also the cytosolic
isoforms BB- and MM-CK bind to membrane preparations
from the inner mitochondrial membrane [53].
Further membrane binding properties of Mi-CK were
studied and quantified by different biophysical methods in
artificial cardiolipin membranes. Among the most important
results is the fact that these membranes favor the assembly
of sarcomeric Mi-CK dimers into octamers [32]. It remains,
however, an open question in which way the Mi-CK octamer
interacts structurally with the membrane lipid bilayer since
conflicting results were obtained by different methods.
Measurements of membrane surface pressure showed a
pressure increase upon Mi-CK binding, indicating a partial
penetration ofMiCK into the membrane bilayer [53]. Insertion
ofMi-CK into the membrane could have a shielding effect and
would explain why neither high ionic strength nor detergents
were able to detach Mi-CK quantitatively from mitochondrial
membranes. These results are not in accordance with ESR
studies performed with cardiolipin-bound Mi-CK [74], which
indicate that the lipid chains of cardiolipin are not strikingly
restricted in their mobility on binding of Mi-CK. This would
exclude an insertion ofMi-CK into the bilayer.
Functional coupling of Mi-CK to oxidative phospho-
rylation occurs via ANT, the only but abundant transport
system for ATP and ADP in the inner mitochondrial mem-
brane (for a review see [19,20]). This suggested a close
arrangement between ANT and Mi-CK [75-78]. So far, the
large body of evidence for a functional interaction between
Mi-CK and ANT has been accompanied by experiments
showing a lack of direct structural interaction, e.g. reconstituted
ANT did not bind to Mi-CK [54], ANT could not be found
unequivocally in crosslinking products with Mi-CK [25,79]
and finally, the basic overall pI ofANT and Mi-CK suggested
an electrostatic repulsion between both proteins. However,
the presence of ANT in the close vicinity ofMi-CK has been
demonstrated by the effect of Mi-CK antibodes on Mi-CK
and ANT activity in mitoplasts. Certain Mi-CK antibodies
which bind to Mi-CK without affecting its enzymatic activity
were able to inhibit adenyl ate transport by ANT [78].
The apparent discrepancy between functional and structural
evidence can be reconciled by the fact that ANT tightly binds
about 6 moles of cardiolipin [80] and that the transporter may
even be located inside a much larger cardiolipin membrane
patch [81]. Such a cluster of cardiolipin could in tum bind Mi-
CK [70,82] and this would result in a close co-localisation of
Mi-CK and ANT [78, 83]. Mi-CK and ANT could thus
interact by frequent collision within the cardiolipin domain
in the inner mitochondrial membrane [78].
Factors influencing Mi-CK membrane attachment
Dissociation and binding studies of Mi-CK with mitoplasts
have shown a crucial influence of ionic compounds on the
interaction ofMi-CK with the inner mitochondrial membrane
(e.g. [71, 72, 84, 85]). Dissociation ofMi-CK from mitoplasts
is determined by ionic strength of the medium [72] and by
ionic composition [84, 85]. The effect of most ions was
reported to correlate with their ionic strength, indicating a
competition for charges in the membrane. One of the most
powerful and rather specific Mi-CK-dissociating ions is
phosphate, which most effectively removes the enzyme from
mitochondrial inner membranes, most probably by competing

with the negatively charged cardiolipin headgroups [71, 85-
87]. Monovalent and especially divalent cations (e.g. Ca2+ and
Mg 2+) are less effective [54]. Similar ion-effects were
obtained when the functional coupling ofMi-CK to oxidative
phosphorylation was monitored in mitochondria in vitro.
Coupling was reduced by high phosphate buffer [87], probably
due to the dissociation of Mi-CK from inner mitochondrial
membranes, but not in a physiological salt solution with higher
ionic strength [84]. Such a lack of correlation between ionic
strength and the dissociating effect could be due to binding of
these ions to the active center or other sites of Mi-CK rather
than to the membrane binding sites, thus inducing structural
changes and/or dissociation ofMi-CK octamers into dimers
and finally leading to a loss of membrane binding. The CK
substrates MgADP and MgATP, chloride ions and negatively
charged organic mercurials may also act in this way.
The vast majority of results is in line with a predominantly
electrostatic interaction of MiCK-with mitochondrial
membranes. Only a few binding studies pointed to a limited
influence of non-ionic interactions [53, 72], although
hydrophobic interaction chromatography revealed that Mi-CK
is more hydrophobic than the cytosolic isoforrns [3]. A major
role of hydrophobic interactions is disproved by the effect of
Triton X-100 alone, which was not able to release more than
one half ofMi-CK from mitoplasts at concentrations known
to solubilize most membrane proteins [70]. Binding of Mi-
CK to cardiolipin may be purely ionic [32, 74], because
application of basic pH and high ionic strength is sufficient
for Mi-CK release from the membrane. Moreover, the cardio-
lipin-content is a factor that governs dissociation of bound
octameric Mi-CK from the membrane [88], but does not alter
the association properties of the octamer. This effect can be
sufficiently explained by ionic interactions which, having a
short range, only come into play when Mi-CK and the acidic
phospholipid are already in close vicinity (Debye length less
then 1.5 nm at 50 mM NaCl), e.g. after membrane binding
of Mi-CK. These considerations, however, do not exclude a
hydrophobic component being relevant in the attraction and
binding process ofMi-CK.
Quantitation of Mi-CK membrane binding and cross-
linking
The membrane binding and dissociation kinetics of octameric
Mib -CK have recently been extensively studied in a novel
approach using immobilised model membranes and plasmon
resonance as detection method [88]. The existence of two
independent binding sites for the octamer with strikingly
different affinity to CL-containing vesicles was shown. The
data obtained were corroborated by earlier experiments with
mitoplasts which also proved the existence of two kinds of
binding sites [70]. For vesicles with two different CL-contents
147
(16% and 100%) no major differences in the fast association
(high affinity) rate constants were seen (pseudo first-order,
both approximately 0.11 S-I, see Fig. 3), whereas dissociation
was much slower in the case of membranes made of pure CL.
In this case, more Mi-CK octamer (about 84%) stayed bound
to the vesicles under nonequilibrium conditions, whereas
vesicles with a lower CL content only retained 66% of the
once bound CK, showing the very high affinity ofMi-CK to
membranes. These data also indicate that the membrane
binding behaviour of octameric Mi-CK is controlled by the
rate of dissociation at different CL-contents rather than by
the rate of association. The thereof calculated equilibrium
constants (a high- and a low-affinity constant for each CL-
content) strongly indicate that under in vivo steady-state
conditions most of the octamer must be membrane associated.
Furthermore, high- and low-affinity constants differed by
four orders of magnitude, thus the low affinity equilibrium
constant represents very loosely associated CK molecules.
Similar results were obtained with mitoplasts, which had
been depleted of Mi-CK [32], showing that even under
conditions favouring the dissociation (e.g. dilution) most of
the octamer stayed bound to the membrane and with chemical
cross-linking studies using glutaraldehyde [71], which
indicated the presence of two distinct binding sites for Mi-
CK. In free solution, the injection of octameric Mi-CK leads
to lipid vesicle cross-linking resulting in the formation of
bridged Mi-CKIvesicle aggregates [88]. The kinetics of the
cross-linking reaction can be followed using a constant-angle
5' 700
~
0;;
~ 600
.~
'" 500
<1)
'"0
~
0..
'"~ 400
300
200
100
0
0 80 160 240 320 400 480
time [s1
Fig. 3. Membrane binding and dissociation kinetics of octameric Mib -CK
and phosphatidylcholine vesicles containing 16% cardiolipin in a buffer
(pH 7.0) containing 50 mM NaCI and 10 mM TES. Association (a) and
dissociation kinetics (b) of the Mi-CK octamer reacting with CL vesicles
of 180 nm diameter are shown. About 66% ofthe associated octameric Mi-
CK stays bound in the dissociation phase (b). For experimental details see
(88).
148
(90) light-scattering approach indicating an extensive
formation of enzyme cross-linked vesicle aggregates (Fig. 4).
The rate constant for the cross-linking reaction was two
orders of magnitude slower than the fast association rate
constant for the reaction ofMi-CK with immobilised vesicles
and was determined to 2.55 * 10-3 S-1 for membranes made of
pure CL. Therefore, the cross-linking reaction is significantly
slower than the fast association ofMi-CK to the membrane
and has to be regarded as the second step in the vesicle/CK
reaction, with the first step being mainly association of the
octamer with one membrane. As this reaction is merely
reversible by dilution, the stability of the Mi-CK-induced
membrane contact sites must be very high under steady-state
conditions, e.g. as long as the octamer does not decay. This
could have major importance for the formation of energy
transduction contact sites in vivo, which are the structural
basis of complexes involved in metabolic channeling [59, 89].
Possible functions of the dimerloctamer equilibrium in
vivo
The above considerations raise the question ofthe particular
importance of the dimer/octamer equilibrium, because
dimeric Mi-CK has a weaker affinity to the membrane and
interacts less strongly at high pH conditions [32] than the
octamer. Most probably, pH and phosphate concentration are
two signals also triggering the binding kinetics of the two
oligomers ofMib-CK in vivo: Whereas a high pH still enables
the octamer to stay bound, it also induces its decay into dimers
in free solution. Therefore, a pH of about 8 which is supposed
'" 3.6 10 ,
~
7
ft :-
:.
0 3.4 107 ~.
N
"<t
.......
.e 3.2 107
'"~
B
..s
3 107
2.8 101 ' ..
2.6 107
0 500 1000 1500
time [s]
Fig. 4. Mi-CK induced vesicle cross-linking. Fixed angle (90) light
scattering can monitor the cross-linking reaction of CL vesicles (100%
CL) with Mi-CK octamers in free solution. The time constant at 25C for
this association (pseudo first-order reaction) was determined to 2.55 10. 3
S-I [88].
to be relevant in respiring mitochondria, combined with high
phosphate concentrations could control the selective re-
binding of the octamer to the mitochondrial membranes,
where it is stabilised and resistant against dimerisation.
Dimers would stay in the intermembrane space where they
could diffuse more easily and assemble octamers at sites of
more acidic pH. The major structural difference between
dimer and octamer is that the octamer has a cube-like shape
with identical top and bottom faces, each exposing four C-
termini which is sufficient for membrane binding at two
adjacent sides of the cube [89]. Thus, octameric Mi-CK can
form stable cross-links between membranes (see also Fig. 5)
thereby inducing membrane contact sites in vitro [27],
whereas dimeric Mi-CK and the (dimeric) cytosolic CK-
isoforms fail to induce bridged membrane structures [27, 88],
the reason probably being that there is more than one C-
terminus necessary at each side of the molecule to stabilize
an intermembrane contact, or that a different membrane
binding mode of dimeric Mi-CK has to be considered, e.g.
both C-termini ofthe dimeric molecule are bound to the same
membrane (Fig. 5). These different binding modes of dimeric
and octameric Mi-CK still remain a matter of speculation
since there are hitherto no detailed experimental membrane
binding data available for dimeric CK.
Conclusion
Although significant new insight into the structure/function
relationship ofMiCKImembrane interaction has been gained
by quantitative biophysical approaches [88, 89], the so far
existing data on Mi-CK membrane binding capabilities still
leave some important questions unanswered. One major point
of discussion is related to mitochondrial ultrastructure: EM-
pictures of mitochondria shown in textbooks display separated
inner and outer membranes, where Mi-CK octamers would
Fig. 5. Schematic model of possible binding modes of Mi-CK. Whereas
the octamer can use its four C-termini exposed at each face of the molecule
for lipid interaction, thereby connecting two opposing membranes (a), the
dimer only posseses two C-termini which are supposed to interact with the
same membrane surface, thus lacking the crosslinking effect (b).

easily fit in between. However, high-pressure frozen samples
which are supposed to contain less artefacts display an
intermembrane space of 4-5 nm only [90], but the height of
the MiCK octamer in the protein crystals is 8.6 nm. Hence,
there are problems to fit the bulky enzyme into the small
compartment between the mitochondrial membranes. There
are two possible explanations for that phenomenon, one
being (i) that these EM-pictures show an artefact-stricken
intermembrane space which is too narrow due to shrinking
effects, or (ii) that octameric Mi-CK partially inserts into the
membrane or at least forms a kind of membrane 'hole' or
indentation which harbour the enzyme. The latter hypothesis
is supported by the cardiolipin-inherent property to form non-
bilayer phases [91] and by the fact that Mi-CK causes a
surface pressure increase upon binding to membranes [27].
Another unsolved problem is the kind and stoichiometry of
possible complexes formed between Mi-CK and proteins of
the inner and outer mitochondrial membrane, as a direct
biophysical evidence for the postulated (and functionally
shown) ANT/Mi-CK interaction is still missing. These
considerations are of special importance for the formation and
regulation ofthe permeability transition pore [67,92] and the
establishment of energy transduction contact sites. Further-
more, the binding mode for dime ric Mi-CK may differ from
the octameric oligomer inasmuch as the dimer is not able
to induce bridged membrane structures (e.g. contact sites,
[88]).
Acknowledgements
This work was supported by a Swiss National Science Founda-
tion (grant Nr. 3133907.92, toT.w. and grant Nr. 31-50'824.97
to T.w. and U.S.), by an ETH graduate stipend (to O.S.) and
by sponsoring from Careal Holding, Zurich (for M.D.).
Note added in proof
Recent experiments studying the binding characteristics of
Mi-CK with liposomes [93] fully confirmed that Mi-CKI
phospholipid interaction is largely of electrostatic nature.
Interestingly, while the nucleotide substratesATP andADP had
no influence on Mi-CK binding to liposomes, para-hydroxy-
mercuri-benzoate, a negatively charged organomercurial,
partially decreased Mi-CK binding and the anti-cancer agent
adriamycin efficiently prevented Mi-CK binding. This latter
finding may explain some ofthe cardiotoxic side-effects often
observed with this drug. Most interestingly, differential
scanning calorimetry performed by the same authors suggests
a partial disorganisation of the phospholipid bilayer upon
interaction with Mi-CK, which is fully in line with our earlier
experiments where binding ofMi-CK to a monolayer of outer
149
mitochondrial phospholipids was shown to result in a
significant increase in surface pressure [27, 53], also indicat-
ing that Mi-CK is locally disturbing the topology of the
phospholipids or is inserting to a certain extent into the
monolayer.
References
1. Bessman SP, Geiger PJ: Transport of energy in muscle: The
phosphorylcreatine shuttle. Science 211: 449--452, 1981
2. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine
kinase isoenzymes in tissues with high and fluctuating energy demands:
The phosphocreatine circuit for cellular energy homeostasis. Biochem
J 281: 21--40,1992
3. Wyss M, Smeitink J, Wevers RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119-166, 1992
4. Wallimann T: Dissecting the role of creatine kinase. Curr Bioi 4: 42-
46, 1994
5. Saks VA, Ventura-Clapier R, Aliev MK: Metabolic control and
metabolic capacity: Two aspects of creatine kinase functioning in cells.
Biochim Biophys Acta 1274: 81--88, 1996
6. Sweeney HL: The importance of the creatine kinase reaction: The
concept of metabolic capacitance. Med Sci Sports Exerc 26: 30-36,
1994
7. Wallimann T, Hemmer W: Creatine kinase in non-muscle tissues and
cells. Mol Cell Biochem 133/134: 193-220, 1994
8. Kaldis P, Kamp G, Piendl T, WallimannT: Functions of creatine kinase
isoenzymes in spermatozoa. Adv Develop Bioi: 5: 275-312, 1997
9. Blethen SL, Kaplan NO: Characteristics of arthropod arginine kinases.
Biochemistry 7: 2123-2135,1968
10. Suzuki T, Furukohori T: Evolution of phosphagen kinase. J Mol BioI
237: 353-357, 1994
II. Stein LD, Ham DA, David JR: A cloned ATP:Guanidino kinase in the
trematode Schistosoma mansoni has a novel duplicated structure. J
BioI Chern 265: 6582-6588,1990
12. Jacobus WE, Lehninger AL: Creatine kinase of rat heart mitochondria:
Coupling of creatine phosphorylation to electron transport. J BioI Chern
248: 4803--4810,1973
13. Gross M, Wallimann T: Kinetics of assembly and dissociation of the
mitochondrial creatine kinase octamer. A fluorescence study.
Biochemistry 32: 13933-13940, 1993
14. Wallimann T, SchlosserT, Eppenberger HM: Function ofM-line-bound
creatine kinase as intramyofibrillar ATP regenerator at the receiving
end of the phosphorylcreatine shuttle in muscle. J Bioi Chern 259:
5238-5246, 1984
15. Ventura-Clapier R, Veksler V, Hoerter JA: Myofibrillar creatine
kinase and cardiac contraction. Mol Cell Biochem 133/134: 125-
144, 1994
16. Rossi AM, Volpe P, Eppenberger HM, Wallimann T: Ca 2+ pumping is
supported by SR-bound creatine kinase. In: U Carraro (ed). Sarcomeric
and non-sarcomeric muscles: basic and applied research prospects for
the 90's. Unipress Padova, 1988
17. Minajeva A, Ventura-Clapier R, Veksler V: Ca2+ uptake by cardiac
sarcoplasmic reticulum ATPase in situ strongly depends on bound
creatine kinase. Pflugers Arch Eur J Physiol432: 904-912, 1996
18. Hemmer W, Riesinger I, Wallimann T, Eppenberger HM, QuestAFG:
Brain-type creatine kinase in photoreceptor cell outer segments: Role
of a phosphocreatine circuit in outer segment energy metabolism and
phototransduction. J Cell Sci 106: 671-684, 1993
150
19. Klingenberg M: The ADP/ATP carrier in mitochondrial membranes.
In: A.N. Martonosi (ed). The enzymes of biological membranes, Vol
4. Plenum, 1985, pp 511-552
20. Klingenberg M: Dialectics in carrier research: The ADP/ATP carrier
and the uncoupling protein. J Bioenerg Biomembr 25: 447-457, 1993
21. Bessman SP: The creatine phosphate energy shuttle-the molecular
asymmetry ofa pool. Anal Biochem 161: 519-523, 1984
22. Wegmann G, Huber R, Zanolla E, Eppenberger HM, Wallimann T:
Differential expression of brain-type and mitochondrial creatine kinase
isoenzymes during development of the chicken retina: Mi-CK as a
marker for differentiation of photoreceptor cells. Differentiation 46:
77- 87, 1991
23. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of
mitochondrial creatine kinase. Nature 381: 341-345,1996
24. Schnyder T, Winkler H, Gross H, Eppenberger HM,Wallimann T:
Crystallization of mitochondrial creatine kinase. J Bioi Chern 266:
5318-5322, 1991
25. Schnyder T, Rojo M, Furter R, Wallimann T: The structure of
mitochondrial creatine kinase and its membrane binding properties.
Mol Cell Biochem 133/134: 115-123, 1994
26. Schnyder T, Cyrklaff M, Fuchs K, Wallimann T: Crystallization of
mitochondrial creatine kinase on negatively charged lipid layers. J
Struct Bioi 112: 136-147,1994
27. Rojo M, Hovius R, Demel R, Nicolay K, Wallimann T: Mitochondrial
creatine kinase mediates contact formation between mitochondrial
membranes. J Bioi Chern 266: 20290--20295, 1991
28. Schlegel J, Wyss M, Schurch U, Schnyder T, Quest A, Wegmann G,
Eppenberger HM,Wallimann T: Mitochondrial creatine kinase from
cardiac muscle and brain are two distinct isoenzymes but both form
octameric molecules. J Bioi Chern 263: 16963-16969, 1988
29. Wyss M, Schlegel J, James P, Eppenberger HM, WallimannT:
Mitochondrial creatine kinase from chicken brain: Purification,
biophysical characterisation, and generation of heterodimeric and
heterooctameric molecules with subunits of other creatine kinase
isoenzymes. J Bioi Chern 265: 15900--15908,1990
30. Gross M, Furter-Graves EM, Wallimann T, Eppenberger HM, Furter
R: The tryptophan residues of mitochondrial creatine kinase: Roles of
Trp-223, Trp-206, and Trp-264 in active-site and quarternary structure
formation. Protein Science 3: 1058--1068, 1994
31. Gross M: The tryptophan residues of mitochondrial creatine kinase:
Roles in enzyme structure and function. Ph.D. Thesis, No. 10719, Swiss
Federal Institute of Technology, Ziirich, 1994
31a. Stachowiak 0, Dolder M, Wallimann T, Richter C: Mitochondrial
creatine kinase is a prime target of peroxynitrate-induced modification
and inactivation. Submitted, 1998
32. Schlegel J, Wyss M, Eppenberger HM, Wallimann T: Functional
Studies with the octameric and dimeric form of mitochondrial creatine
kinase. J Bioi Chern 265: 9221-9227, 1990
33. Gross M, Wallimann T: Dimer-dimer interactions in octameric
mitochondrial creatine kinase. Biochemistry 34: 6660--6667, 1995
34. Kaldis P, Wallimann T: Functional differences between dimeric and
octameric mitochondrial creatine kinase. Biochem J 308: 623-627,
1995
35. Hossle JP, Schlegel J, Wegmann G, Wyss M, Bohlen P, Eppenberger
HM, Wallimann T, Perriard JC: Distinct tissue specific mitochondrial
creatine kineses from chicken brain and striated muscle with a
conserved CK framework. Biochem Biophys Res Commun 151: 408-
416, 1988
36. Lill R, Neupert W: Mechanisms of protein import across the
mitochondrial outer membrane. Trends Cell Bioi 6: 56-61, 1996
37. Baskin RJ, Deamer DW: A membrane-bound creatine phosphokinase
in fragmented sarcoplasmic reticulum. J Bioi Chern 245: 1345-1347,
1970
38. Levitsky DO, Levchenko TS, Saks VA, Sharov VG, Smirnov VN:
The role of creatine phosphokinase in supplying energy for calcium
pump system of heart sarcoplasmic reticulum. Membr Biochem 2: 81-
96, 1978
39. Rossi AM, Eppenberger HM, Volpe P, Cotruto R, Wallimann T:
Muscle-type creatine kinase is specifically bound to sarcoplasmic
reticulum and can support Ca2+ uptake and regulate local ATP/ADP
ratios. J Bioi Chern 265: 5258--5266, 1990
40. Guerrero ML, Beron J, Spindler B, Groscurth P, Wallimann T, Verrey
F: Metabolic support ofNa+-pump in epically permeabilized A6 kidney
cell epithelia: Role of creatine kinase. Am J PhysioI272: C697--C706,
1997
41. Blum H, Nioka S, Johnson RG: Activation of the Na+/K+-ATPase in
Narcine brasiliensis. Proc NatiAcad Sci USA 1984: 1247-1251,1990
42. Borroni E: Role of creatine phosphate in the discharge of the electric
organ of Torpedo marmorata. J Neurochem 43: 795-798, 1984
43. Kammermeier H: Why do cells need phosphocreatine and a phospho-
creatine shuttle? J Mol Cardiol19: 115-118,1984
44. Wallimann T, Waltzhony D, Wegmann G, Moser H, Eppenberger HM,
Barrantes FJ: Subcellular localisation of creatine kinase in Torpedo
electrocytes: Association with acetylcholine receptor-rich membranes.
J Cell Bioi 100: 1063-1072, 1985
45. Blum H, Balschi JA, Johnson RG: Coupled in vivo activity of creatine
phosphokinase and the membrane-bound (Na+/K+)-ATPase in the
resting and stimulated electric organ of the electric fish Narcine
brasiliensis. J Bioi Chern 266: 10254-10259, 1991
46. Miihlebach S, Gross M, Wirz T, Wallimann T, Perriard JC, Wyss M:
Sequence homology and structure predictions of the creatine kinase
isoenzymes. Mol Cell Bioi 133/134: 245-262, 1994
47. Korge P, B.yrd SK, Campbell KB: Functional coupling between
sarcoplasmic reticulum-bound creatine kinase and Ca2+-ATPase. Eur
J Biochem 213: 973-980, 1993
48. Korge P, Campbell, KB: Local ATP regeneration is important for
sarcoplasmic reticulum Ca2+pump function. Am J Physiol267: C357-
C366, 1994
49. Ferris CD, Huganir RL, Snyder SH: Calcium flux mediated by purified
inositol I ,4,5,-trisphosphate receptor in reconstituted lipid vesicles is
allosterically regulated by adenine nucleotides. Proc Natl Acad Sci
1984:2147-2151,1990
50. Steele DS, McAinish AM, Smith GL: Effects of creatine phosphate
and inorganic phosphate on the sarcoplasmic reticulum of saponin-
treated rat heart. J Physiol (London) 483: 155-166, 1995
51. Quest AFG, Chadwick JK, Wothe DD, McIlhinney RAJ, Shapiro BM:
Myristoylation of flagellar creatine kinase in the sperm phosphocreatine
shuttle is linked to its membrane association properties. J Bioi Chern
267: 15080--15085, 1992
52. Quest AFG, Hervey DJ, McIlhinney RAJ: Identification of a non-
myristoylated pool of sea urchin sperm flagellar creatine kinase:
Myristoylation is necessary for efficient lipid association. J Bioi Chern,
submitted
53. Rojo M, Hovius R, Demel R, Wallimann T, Eppenberger HM, Nicolay
K: Interaction of mitochondrial creatine kinase with model membranes.
FEBS Lett 281: 123-129, 1991
54. Miiller M, Cheneval D, Carafoli E: The mitochondrial creatine
phosphokinase is associated with inner membrane cardiolipin. In: N.
Brautbar (ed). Myocardial and Skeletal Muscle Bioenergetics. 1985,
Plenum, pp 151-155
55. Brdiczka D: Contact sites between mitochondrial envelope membranes.
Structure and function in energy- and protein-transfer. Biochim
Biophys Acta 1071: 291-312,1991
56. Knoll G, Brdiczka D: Changes in freeze-fractured mitochondrial
membranes correlated to their energetic state: Dynamic interactions of
the boundary membranes. Biochim Biophys Acta 733: 102-110, 1983

57. Bucheler K, Adams V, Brdiczka D: Localization of the ATP/ADP
trans locator in the inner membrane and regulation of contact sites
between mitochondrial envelope membranes by ADP. A study on
freeze- fractured isolated liver mitochondria. Biochim Biophys Acta,
1056: 233-242,1991
58. Adams V, Bosch W, Schlegel J, Wallimann T, Brdiczka D: Further
characterization of contact sites from mitochondria of different tissues:
Topology of peripheral kineses. Biochim BiophysActa 981: 213-225,
1989
59. Kottke M, Adams V, Wallimann T, Nalam VK, Brdiczka D: Location
and regulation of octameric mitochondrial creatine kinase in the contact
sites. Biochim Biophys Acta 1061: 215-225,1991
60. Moran 0, Sorgato MC: High-conductance pathways in mitochondrial
membranes. J Bioenerg Biomembr 24: 91-98, 1992
61. Benz R: Permeation of hydrophilic solutes through mitochondrial outer
membranes: Review on mitochondrial porins. Biochim Biophys Acta
1197: 167-196,1994
62. Benz R, Brdiczka D: The cation-selective substate ofthe mitochondrial
outer membrane pore: Single-channel conductance and influence on
intermembrane and peripheral kinases. J Bioenerg Biomembr 24: 33-
39, 1992
63. Saks VA, Belikova YO, Kuznetsov AY, Branishte TH, Semenovsky
ML, Naumov VG: Phosphocreatine pathway for energy transport: ADP
diffusion and cardiomyopathy. Am J Physiol261: 30--38,1991
64. Brdiczka D: Function of the outer mitochondrial compartment in
regulation of energy metabolism. Biochim Biophys Acta 1187: 264-
269, 1994
65. Brdiczka D, Wallimann T: The importance of the outer mitochondrial
compartment in regulation of energy metabolism. Mol Cell Biochem
133/134: 69--83,1994
66. Brdiczka D, Kaldis P, Wallimann T: In vitro complex formation
between the octamer of mitochondrial creatine kinase and porin. J
Bioi Chern 269: 27640--27644, 1994
67. Beutner G, Ruck A, Riede B, Welte W, Brdiczka D: Complexes
between kinases, mitochondrial porin and adenylate translocator in
rat brain resemble the permeability transition pore. FEBS Lett 396:
189--195,1996
68. De Pinto V, Palmieri F: Transmembrane arrangement of mitochondrial
porin or voltage-dependent anion-channel (VDAC). J Bioenerg
Biomembr 24: 21-26,1992
69. Hovius R, Lambrechts H, Nicolay K, de KruijffB: Improved methods
to isolate and subfractionate rat liver mitochondria: Lipid composition
of the inner and outer membrane. Biochim Biophys Acta 1021: 217-
226,1990
70. Schlame M, Augustin W: Association of creatine kinase with rat
heart mitochondria: high and low affinity binding sites and the
involvement of phospholipids. Biomed Biochim Acta 44: 1083-1088,
1985
71. Lipskaya TY, Tempi VD, Belousova LV, Molokova EV, Rybina TV:
Investigation of the interaction of mitochondrial creatine kinase with
the membranes of the mitochondria. Biochimia USSR 45: 877-886,
1980
72. Brooks SPJ, Suelter CH: Association of chicken mitochondrial creatine
kinase with the inner mitochondrial membrane. Arch Biochem Biophys
253: 122-132, 1984
73. Killian lA, Koorengevel MC, Bouwstra lA, Gooris G, Dowhan W,
De Kruijff B: Effect of divalent cations on lipid organization of
cardiolipin isolated from Escherichia coli strain AH930. Biochim
Biophys Acta 1189: 225-232, 1994
74. Cheneval D, Carafoli E, Powell GL, Marsh D: A spin-label electron
spin resonance study of the binding of mitochondrial creatine kinase
to cardiolipin. Eur 1 Biochem 186,415-419,1989
151
75. Kupriyanov VV, Elizarova GY, Saks VA: Determination of the molar
content of creatine kinase in heart mitochondria using SH reagents.
Biochimia USSR 46: 762-772,1981
76. Barbour RL, Ribaudo J, Chan SHP: Effect of creatine kinase activity
on mitochondrial ADP/ATP transport. Evidence for a functional
interaction. J Bioi Chern 259: 8242-8251, 1984
77. Kuznetsov AV, Saks VA: Affinity modification of creatine kinase and
ATP-ADP translocase in heart mitochondria: Determination of their
molar stoichiometry. Biochem Biophys Res Comm 134: 359--366, 1986
78. Saks VA, Khuchua ZA, Kuznetsov AV: Specific inhibition ofATP-ADP
translocase in cardiac mitoplasts by antibodies against mitochondrial
creatine kinase. Biochim Biophys Acta 891: 138-144, 1984
79. Font B, Eichenberger D, Goldschmidt D and Vial C: Interaction of
creatine kinase and hexokinase with the mitochondrial membranes,
and self-association of creatine kinase: Crosslinking studies. Mol Cell
Biochem 78: 131-140, 1984
80. Beyer K, Klingenberg M: ADPI ATP carrier protein from beef heart
mitochondria has high amounts of tightly bound cardiolipin, as revealed
by lip nuclear magnetic resonance. Biochemistry 24: 3821-3826, 1985
81. Drees M, Beyer K: Interaction of phospholipids with the detergent-
solubilizedADP/ATP carrier protein as studied by spin-label electron
spin resonance. Biochemistry 27: 8584-8591, 1988
82. Saks VA, Khuchua ZA, Vasilyeva EY, Belikova OYu, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration - a synthesis. Mol Cell Biochem 1331134: 155-192, 1994
83. Cheneval D, Carafoli E: Identification and primary structure of the
cardiolipin binding domain of mitochondrial creatine kinase. Eur J
Biochem 171: 1-9, 1988
84. Saks VA, Khuchua ZA, Kuznetsov AV, Veksler VI, Sharov VG: Heart
mitochondria in physiological salt solution: not ionic strength but salt
composition is important for association of creatine kinase with the
inner membrane surface. Biochem Biophys Res Comm 139: 1262-
1271,1986
85. Vial C, Marcillat 0, Goldschmidt D, Font B, Eichenberger D: Interaction
of creatine kinase with phosphorylating rabbit heart mitochondria and
mitoplasts. Arch Biochem Biophys 251: 558-566,1986
86. Hall N, Deluca M: Binding of creatine kinase to heart and liver
mitochondria in vitro. Arch Biochem Biophys 20 I: 674--677, 1980
87. Hall N, Deluca M: The effect of inorganic phosphate on creatine kinase
in respiring rat heart mitochondria. Arch Biochem Biophys 229: 477-
482, 1984
88. Stachowiak 0, Dolder M, Wallimann T: Membrane-binding and lipid
vesicle crosslinking kinetics of the mitochondrial creatine kinase
octamer. Biochemistry 35: 15522-15528, 1996
89. Schlattner U, Forstner M, Eder M, Stachowiak 0, Fritz-Wolf K,
Wallimann T: Functional aspects of the X-ray structure of mito-
chondrial creatine kinase: A molecular physiology approach. Mol Cell
Biochem, in this issue
90. Piendl T: Cryoimmobilisation-based electron microscopy of mito-
chondria in situ and in vitro: A new model of contact sites for energy
export. Ph.D. thesis Nr. 11830, Swiss Federal Institute of Technology,
ETH Zurich, 1996
91. Rietveld AG, Koorengevel MC, De Kruijff B: Non-bilayer lipids are
required for efficient protein transport across the plasma membrane
of Escherichia coli. EMBO J 14: 5506--5513, 1995
92. Tikhonova IM,Andreyev AY, Antonenko YN, KaulenAD, Komrakov
AY, Skulachev VP: Ion permeability induced in artificial membranes
by the ATPI ADP antiporter. FEBS Letters 337: 231-234, 1994
93. Vacheron MJ, Clottes E, Chautard C, Vial C: Mitochondrial creatine
kinase interaction with phospholipid vesicles. Arch Biochem Biophys
344: 316--324,1997
Molecular and Cellular Biochemistry 184: 153-167, 1998.
1998 Kluwer Academic Publishers.

Molecular characterization of the creatine kinases

and some historical perspectives

Wenning Qin, Zaza Khuchua, Judy Cheng, Jaime Boero, R. Mark

Payne and Arnold W. Strauss
Departments of Pediatrics and Molecular Biology and Pharmacology, Washington University School of Medicine and St.
Louis Children's Hospital, St. Louis, MO 63110, USA

Abstract
Over the last 15 years, molecular characterization ofthe creatine kinase (CK) gene family has paralleled the molecular revolution
of understanding gene structure, function, and regulation. In this review, we present a summary of advances in molecular analysis
of the CK gene family with a few vignettes of historical interest. We describe how the muscle CK gene provided an essential
model system to examine myogenic regulatory mechanisms, leading to the discovery of the binding site for the MyoD family
of basic helix-loop-helix transcription factors essential in skeletal myogenesis and the characterization of the MEF2 family of
factors with an AIT rich consensus binding site essential in skeletal myogenesis and cardiogenesis. Cloning and characterization
of the four mRNAs and nuclear genes encoding the cytosolic CKs, muscle and brain CKs, and the mitochondrial (Mt) CKs,
sarcomeric MtCK and ubiquitous MtCK, has allowed intriguing study of tissue-specific and cell-specific expression of the
different CKs and analysis of structural, functional, regulatory, and evolutionary relationships among both the four CK proteins
and genes. Current and future studies focus on understanding both cellular energetics facilitated by the CK enzymes, especially
energy channelling from the site of production, the mitochondrial matrix and inner membrane, to various cytosolic foci of
utilization, and regulation of MtCK gene expression at the cell and tissue-specific level as models of regulation of energy
producing genes. (Mol Cell Biochem 184: 153-167,1998)

Key words: creatine kinase, transcription factors, myogenesis, mitochondrion, energy, gene family

Abbreviations: BCK - brain creatine kinase; bHLH - basic helix-loop-helix; bp - base pair; cDNA - complementary DNA;
CK -creatine kinase; MCK - muscle creatine kinase; MEF2 -myocyte-specific enhancer-binding factor 2; sMtCK - sarcomeric
mitochondrial creatine kinase; uMtCK - ubiquitous mitochondrial creatine kinase; UTR - untranslated region

Introduction human genome structure and sequence and may allow

The description of the double helical structure of DNA [1]
and subsequent discoveries of DNA restriction enzymes [2],
cloning of DNA in bacterial plasmid vectors [3, 4], the
complex intron-exon structure of genes [5], and the chemistry
to sequence DNA [6, 7] resulted in Nobel prizes for the
investigators and a revolution in our understanding of gene
structure, function, and regulation and in the molecular
genetics of human diseases. This molecular biologic revo-
lution will soon result in complete determination of the
effective somatic gene therapy.
Over the last 20 years, the creatine kinase (CK) gene family
played a key role in this molecular revolution [8-10]. Genes
within this family have been isolated from various animal
species and characterized. Combined with the structural
characterization of the chicken heart mitochondrial creatine
kinase (MtCK) by crystallography and x-ray diffraction ([11]
and in this issue), these primary sequence data from cDNA
cloning have begun to elucidate MtCK interactions in
mitochondrial energy channelling [12].

Present address: R.M. Payne, Division of Pediatric Cardiology, Department of Pediatrics, Bowman-Gray School of Medicine, Winston-Salem, NC 27157-
1081, USA
Addressfor correspondence: A.w. Strauss, St. Louis Children's Hospital, One Children's Place, St. Louis, MO 63110, USA
154
Moreover, the mouse muscle creatine kinase (MCK) gene
has provided an essential model system in analysis of the
molecular basis of myogenesis, one of the most elegantly
studied developmental programs during ontogeny (reviewed
in [13, 14]). Investigation of mouse MCK transcriptional
regulation led to the discovery of the MyoD DNA-binding
activity and the consensus binding site for the helix-loop-
helix family of myogenic regulatory factors and to the
identification of the consensus binding site for the illyocyte-
specific .!,<nhancer-binding factor 2 (MEF2) and the subse-
quent cloning ofMEF2 mRNAs and genes. These discoveries
provide a molecular understanding of skeletal myogenesis in
organisms from flies to humans and crucial clues to current
dissection of cardiac and smooth muscle organogenesis.
Thus, investigation ofthe CK genes has extended knowledge
far beyond their role in energy generation.
Since 1982, work from numerous researchers has identi-
fied four independent nuclear genes in the CK gene family,
each encoding a discrete CK subunit and named for the
differential tissue expression pattern and intracellular com-
partmentalization unique to each gene product [15]. These
four CK genes encode cytoplasmic muscle CK (MCK),
cytoplasmic brain CK (BCK), ubiquitous mitochondrial CK
(uMtCK), and sarcomeric mitochondrial CK (sMtCK) pro-
teins. In this review, we will present information on CK
mRNA (Fig. 1), protein (Fig. 2), and gene (Fig. 3) structures
in humans, discuss mechanisms regulating CK gene ex-
pression, and outline a brief history of the landmark
discovery of the MyoD and MEF2 transcription factors
from the vantage point of the CK gene family. We will
discuss the CK gene family as a model gene in the analysis
of cardiac and skeletal muscle myogenesis, exciting areas
of developmental biology. The roles of the mitochondrial
CKs in understanding tissue-specific gene expression and
energy metabolism will be discussed. Some historical
vignettes presented along the way may provide some insight
into the development of the CK field.
The Muscle Creatine Kinase gene
Characterization of the human MCK mRNA
The molecular revolution among members of the CK gene
family began with the isolation of a chicken MCK cDNA
in the early 1980s [16-18]. Total RNA was isolated from
chick skeletal muscle and fractionated through a sucrose
gradient to enrich for MCK mRNA. Following creation of
this size-fractionated cDNA library, MCK cDNA clones
were identified by ability to hybridize to and select for an
mRNA whose translation product was the MCK protein.
Shortly after, MCK cDNAs were isolated from other
species, including fish [19, 20], rabbit [21], rat [22], chicken
[23], mouse [24], dog [25], and human [26, 27]. In our own
work to isolate the dog MCK cDNA in the pre-PCR era of
1984 [25], it is of historical interest to note that a chick MCK
probe generously provided by Charles Ordahl and his
coworkers [18] was essential to the initiation of our entire
CK molecular biologic project summarized in this review.
Others [26] also benefited from similar scientific collabora-
tions and interactions in this field.
The human MCK cDNA is 1561 bp long [26, 27], includ-
ing 77 bp of the 5'-untranslated region (UTR), 1146 bp of the
coding region, and 338 bp of the 3'-UTR. The translation
initiation codon is at nucleotide 80, the stop codon is at
nucleotide 1144, and the polyadenylation sequence, AATAAA,
is at nucleotide 1464, 15 bp before the poly(A) sequence
(Fig. 1). This mRNA is predicted to encode a 381 amino acid
polypeptide with a calculated molecular weight of 43,051
daltons (Fig. 2). The active site residue cysteine is located
at position 282. Comparison ofthe coding region nucleotide
sequences among species including fish, birds, and mammals
show 77-91 % homology [28]. At the amino acid level, the
identity is 82-98%; and the active site cysteine is also
conserved [15]. The high degree of protein sequence con-
servation beyond the active site throughout evolution sug-
gests either that these homologous residues are essential in
detennining structure of the enzyme or that MCK may have
non-enzymatic roles, such as interacting with other proteins
or intracellular structures. The 3'-UTR is not conserved
among MCK cDNAs from different species.
Characterization of the human MCK gene
The human MCK gene was isolated in 1988 in our laboratory
[29] by screening a 'A phage human genomic DNA library with
a canine BCK probe. The human MCK gene spans 17.5 kb
of the genome and contains 8 exons separated by 7 introns
(Fig. 3). The translation initiation codon is located in exon
2 so that the 5'-UTR is split by an intron. This somewhat
unusual feature is common to genes expressed in muscle
[30]. All the intron-exon splice sites conform to the GT -AG
consensus rule. The single MCK gene is located in chromo-
some 19 [27] and, in human, no pseudogenes have been
reported. Analysis of the 5'-flanking region of the human
MCK gene demonstrates features of a highly regulated gene.
The sequence, TATATAA, is located at position -31, and a
GCCAAT box at position -56 (relative to its transcription
start point). Other consensus sequences important for
tissue-specific gene expression are present in the human
MCK gene 5'-flanking region, including an inverted GC box,
GGCGGG, at-274; two CArG boxes, CCATACAAGG and
CCTTGTAAGG, at -179 and -895; a MEF2 consensus,
CTAAAAATAA, at-739; and many E boxes (CANNTG).
Indeed, the human MCK 5'-flanking sequence, from -2620
to +203 bp, contains all necessary and sufficient elements
directing a tissue-specific and developmentally regulated
expression in muscle cells [29].
 155

sMtCK
I~~I 8

uMtCK
2 4

MCK
~~_2~3~_4~5~1_6~1_7_*~8~

BCK
~2.~1~3~4~5~1~6~1~7*~8~
Fig. I. Human mitochondrial and cytosolic CK exon-intron organizations. The exons are numbered, drawn to scale, and aligned at exon 8 of sMtCK (exon 6
of uMtCK, MCK and BCK), which is identical in all known human CK isoforms. Hatched areas are untranslated regions; open boxes are translated coding
sequence. The asterisk denotes the location of the conserved active site cysteine.

ALIGNMENT OF HUMAN CREATINE KINASE PROTEIN SEQUENCES

10 20 30 40 50 60 70 80 90 100
PEPVRAASERRRL1PPS AE1PDLRKHNNCMASHLTPAV1ARLCDKTTPTGWTLDQCIQTGVDNPGHPFIKTVGMVAGDEETYE VFADLFDPVIQERHNG1 uMtCK
QKVCAEVR-QP--F--- -D------------EC----I-SK-RN-V--N-1-----------------------------s-- ----------KL----- sMtCK
MPFSNSHNTLKLRF-AE D-F---SA---H--KV---EL--E-RA-S--S-F---DV-----------y-M---C---G--S-- --K-----I-ED--G-- BCK
MPFGNTHNKFKLNYKPE E-----S----H--KV--LEL-KK-R--E--S-F-V-DV-----------F-M---C---D--S-- --KE----I-SD--G-- MCK
** * _....... ilt * .. t*' ** ..
(62\) (47\)

110 130 140 150 170 180 190 200

DPRTMKHTTDLDASK IRSGYF DERYVLSSRVRTGRSIRGLSLPPACTRAERREVERVVVDALSGLKGDLAGRYYRLSEMTEAEQQQLID DHFLFDKPVSP uMtCK
---V----------- -TQ-Q- --H-------------------------------N-AIT--E-----------K------QD------ ----------- sMtCK
K- SDE-K---NPDN LQG-DDL-PN-----G---------FC---II-S-G---AI-KLA-E---S-D--------A-KS----------- ----------- BCK
K- -D--K---NHEN LKG-DDL-PN-----P-------K-y----II-S-G---A--KLS-E--NS-T-EFK-K--P-KS---K------- ---Q------- MCK

(55\)

210 230 240 250 I 270 280 $ 290 300

LLTAAGMARDWPDARGIW HNNEKSFLIWVNEEDHTRVISMEKGGNMKRVFERFCRGLKE VERLIQERGWEFMWNERLGYILTCPSNLGTGLRAGVHIKLP uMtCK
---C-------------- --YD-T----I------------------------------ -------------------------------------VRI- sMtCK
--L-S------------- --DN-T--V-------L-----Q------E--T---T--TQ I-T-FKSKDY-----PH------------------------ BCK
--L-S----H----P--- --DN----V-------L-----E------E--R---V--QK I-EIFKKAGHP----QH---V------------G---V--A MCK
tt . . . . . . * _ _ .t.
(79\) (70\ ) (56\)

310 320 330 340 350 360 370 380

LLSK DSRFPKILENLRLQKRGTGGVDTAATGGVFDISNLDRLGKSE VELVQLVIDGVNYLIDCERRLERGQDIRIPTPVIHTKH HUMAN uMtCK
K--- -P--S--------------------VAD-Y----I--I-R-- --------------V---KK------- HUMAN sMtCK
N-G- HEK-SEV-KR---------------V-----V--A----F-- -----H-V---KL--EM-Q---Q--A-DDLM-AQK HUMAN BCK
H-S- HPK-EE--T----------A-----V-S---V--V----S-- -RK----V---KLMVEM-KK--K--S-DDMI-AQK HUMAN MCK
.... _.*-_ .. * ...
(60\ ) (38\ )

Fig. 2. Predicted human creatine kinase amino acid sequences. The amino acid sequences predicted from cDNA clones encoding uMtCK, sMtCK, BCK and
MCK [26, 53, 55] are aligned. - indicates identical amino acids among all four human CK gene products; * indicates identical amino acid sequences among
other known CKs; ! indicates the points of intron-exon boundaries in the BCK or MCK genes; % indicates the degree of sequence identity within MtCK gene
exons as compared to the corresponding cytosolic CK sequences; $ indicates the reactive cysteine residue in the active site. The sites of MtCK intron-exon
boundaries are indicated by space in the sequence.
156
2 3 4
II
2 3
Fig. 3. Human creatine kinase gene structures. Exon-intron structure is drawn in scale. Three human CK genes are aligned at exon 8 of sMtCK (dashed
vertical line ). Heightened solid boxes indicate coding regions of exons. Shortened open boxes indicate nontranslated regions of exons [30, 52, 58].
Myogenic regulatory factors: the mouse MCK gene,
MyoD, E box, and Helix-loop-helix transcription
factors
The mouse MCK gene was isolated in 1986 in Hauschka's
laboratory [31]. Study of the transcriptional regulation of the
mouse MCK gene led to the discovery of the consensus
binding site, CANNTG, for the MyoD family of basic helix-
loop-helix (bHLH) proteins, master regulatory factors
controlling skeletal myogenesis [10]. This was accomplished
by two lines of independent research which converged at the
cloning ofthe cDNA encoding MyoD [32] and the identifica-
tion of a consensus binding site in the mouse MCK gene 5'-
flanking region as a crucial element regulating its expression
in muscle [8].
The MCK gene is highly and virtually exclusively expressed
in heart and skeletal muscles, and the transcriptional mech-
anisms regulating its expression have been intensively investi-
gated over the last decade. Deletion of sequence to -1256 bp
in the mouse MCK gene 5 '-flanking region (relative to its
transcription start point) had little effect on expression of CAT
reporter gene in vitro after transfection into muscle cell lines.
However, deletion to -1020 bp dramatically reduced ex-
lkb
56 7 8 9 10 11
II I I I I sMtCK
uMtCK
4 5
MCK
56:
4\ \7
\ 2~1 !;\?
,1111 III BCK
pression [31]. This 206 bp fragment, located between -1256
and -1050 bp of the mouse MCK 5'-flanking region, was,
thus, shown to be largely responsible for the tissue and
differentiation-dependent expression ofthe MCK gene both
in vitro in cultured MM14 and C 2 C l2 myocytes and in
transgenic mice [33, 34]. In gel electrophoretic mobility shift
and DNase I footprinting assays, a 25 bp fragment within this
enhancer, CCCCCCAACACCTGCTGCCTGAGCC, was
identified as the region to which nuclear proteins unique to
differentiated, cultured muscle cells bound. This sequence-
specific DNA-binding activity was initially named MEF 1, for
myocyte-specific ~nhancer-binding factor 1 [8]. Based on
sequence comparison between MEF 1 binding site in the
mouse MCK enhancer with regulatory regions of other
muscle specific genes, the authors proposed the following
consensus sequence for muscle-specific binding: CIGNGI
AGIACACIGCIGTGC/TC/TNCIG, with a core sequence of
CANNTG.
At the same time, a cDNA cloned by subtraction screening
in differentiating 10 TI!2 cells was shown to induce myo-
genesis in a variety of cell types [32]. The encoded myocyte
differentiating protein was designated MyoD. Detailed

comparison showed that MEF 1 and MyoD share many
similarities, including antigenicity and DNA binding speci-
ficity [10]. Subsequently, numerous studies proved that
MyoD is a sequence-specific DNA-binding protein which
binds to the consensus sequence, CANNTG, within the MCK
enhancer to activate gene transcription during skeletal
myogenesis (reviewed in [13]).
The MyoD family includes four transcription factors,
MyoDl [32], myogenin [35, 36], Myf5 [37], and MRF41
herculin/Myf6 [38-40]. Proteins within this family share
80% homology within a 70 amino acid stretch comprised
of a basic region and a helix-loop-helix domain (bHLH).
The basic region mediates the DNA-binding and trans-
cription activation function, and the HLH domain forms the
interface for dimer formation either homologously or with
other ubiquitously expressed HLH proteins [13]. These
sequence-specific DNA-binding proteins, with overlapping
pattern of expression during ontogeny, control skeletal
myogenesls.
The landmark discovery of MyoD opened the door to
the many subsequent studies documenting the crucial
roles of this and other bHLH transcription factors in gene
regulation, development, and organogenesis and serves as
a fitting memorial to the contributions of Hal Weintraub.
Cardiomyogenic regulatory factors: the mouse MCK
gene, MEF2, AfT-rich consensus binding site, and the
MADS box transcription factors
Study of the transcriptional regulation of the mouse MCK
gene also led to the identification of another group of
transcription factors, myocyte-specific ~nhancer-binding
factor 2 (MEF2), which regulate gene expression in a wide
variety of tissues, and in skeletal muscle, cooperate with the
MyoD family ofbHLH proteins to specify myogenic differ-
entiation [9, 41-44].
Nested deletion of the 5'-flanking region and in vitro
transfection studies of reporter constructs demonstrated that
the region between-1351 to-l050 bp of the mouse MCK
gene exhibited the properties of a classical enhancer, con-
ferring cell type-specific and developmentally-regulated
expression of the MCK gene irrespective of position, orienta-
tion, and distance from the MCK promoter and the hetero-
logous simian virus 40 promoter [41]. Further studies
employing site-directed mutagenesis, DNase I footprint,
diethylpyrocarbonate methylation interference, and gel
electrophoresis mobility shift assays identified a myocyte-
specific DNA-binding activity recognizing an AfT-rich
sequence stretch within this enhancer element. This A/T -rich
sequence, C/TTAAAAATAACC/TC/TC/T, was also noted in
the control regions of several other muscle-specific genes [9].
Interestingly, serum response factor (SRF) also binds to an
A/T-rich consensus, CQA+T)6GG, the serum response
element (SRE), found in the regulatory regions of many
157
cellular immediate early genes (reviewed in [45]) and
perplexingly, muscle specific genes (CArG box, [46]).
Using a DNA probe encoding the conserved SRF DNA-
binding domain to screen Agtll cDNA libraries from
human placenta and lurkat T cells, Pollock and Treisman
[42] identified several novel transcription factors that share
homology with SRF in the DNA binding domain. These new
transcription factors are named RSRF, for related to .s.erum
response factor. These RSRFs also bound to a consensus
sequence, CTA(A/T)4 TAG, reminiscent of the SRE in its
rich A/T content, but with a longer A/T core and distinct
flanking sequences.
By using the AT-rich consensus sequence from the mouse
MCK 5'-flanking enhancerregion to screen aAgtll expression
library generated from primary human skeletal myocytes, Yu
and coworkers [43] isolated several cDNAs which were
alternatively spliced isoforms of the Pollock and Treisman
clones described above [42]. These were designated as the
second type of myocyte-specific ~nhancer-binding factors,
the MEF2s. It is now known that, in vertebrates, four mej2s
exist (mej2A, B, C, and D) and the mej2A, C, and D mRNAs
also exist in alternatively spliced isoforms, demonstrating the
diversity ofthis family of transcription factors (reviewed in
[14]). Sequence comparison revealed that MEF2 proteins
share homology in the DNA binding and dimerization domains
to the MADS box family of transcription factors, which
includes the yeast mating type-specific transcription factor
MCM1, the plant homeotic gene productsAG andJ2EF A, and
humanSRF [47].
The MEF2 proteins share greater than 80% amino acid
homology in the 56-amino-acid MADS box at their N-
termini. Adjacent to the MADS box is a 29-amino-acid
domain, the MEF2 domain, that is unique to the MEF2
proteins among the MADS box transcription factors. To-
gether, the MADS box and the MEF2 domain mediate the
DNA binding/dimerization function; the C-terminal sequence
confers the transactivation function of the MEF2 proteins.
Many workers have established that the MEF2s are sequence-
specific DNA-binding proteins that bind to the A/T -rich
consensus sequence, CTA(A/T)4TAG, to regulate gene
expression in both muscle and nonmuscle tissues. In skeletal
muscle, they cooperate with the MyoD family of bHLH
transcription factors to specify muscle-specific gene expres-
sion (reviewed in [14]). Because a single MEF gene exists
in Drosophila, it was possible to ablate MEF simply [48]. The
phenotype demonstrated a crucial role of MEF2 in fly
myogenesis and dorsal vessel (the primitive fly 'heart')
formation. Gene ablation studies in mice of each MEF gene
are ongoing and demonstrate similar crucial roles in mam-
malian cardiac organogenesis.
Thus, historically, the mouse MCK promoter played an
essential role in the isolation of this crucial family of
cardiomyogenic transcription factors, the MEFs.
158
The Brain Creatine Kinase gene
Characterization of the human BCK mRNA
BCK cDNA was successfully isolated from several animal
species in the mid-1980s by different approaches, including
screening a rabbit brain cDNA library with a rabbit MCK
cDNA probe [49], BCK antibody screening of a rat Agtll
expression library [50], identification of the translation
product ofthe rescued chicken mRNA [51], probing a human
astrocytoma cDNA library with a 26 bp best-guess oligo-
nucleotide designed against the highly conserved region
around the reactive cysteine in several BCK and MCK proteins
[52], and via hybridization to a Agtl 0 library of a human cell
line, NCI-H378, with a rabbit BCK cDNA probe [53].
The human BCK cDNA is about 1.5 kb long and encodes
a polypeptide of381 amino acids (Figs 1 and 2). The reactive
cysteine conserved among the guanidino kinases is codon-
283. At its 3'-end, a poly(A) tail of about 100 bp is added 17,
18, or 19 nucleotides away from the polyadenylation signal,
AAUAAA, a result of imprecise cleavage-polyadenylation
ofthe primary transcript. A noticeable feature of the human
BCK cDNA is the high degree of homology of its 3'-UTR
with those of other species (from 58% in chicken to 88% in
dog). This extraordinary degree of sequence identity suggests
a regulatory role for the 3'-UTR in mRNA stability, pro-
cessing, or translation, possibly through interaction with
regulatory proteins (Cheng and Strauss, unpublished data).
Conserved sequences among 3'-UTRs have been shown to
have similar functions in regulation of mRNAs encoding
proteins of iron metabolism [54].
Structural characterization of the human BCK gene
The human BCK gene was isolated by screening a human
genomic AEMBL3 library with an 163 bp fragment of an
aberrant cDNA clone (pACK-101) consisting mostly of
intron 5 and some exon 6 sequences of the human BCK
cDNA [52]. There appears to be only one BCK gene per
haploid human genome [52, 56], and it is located on
chromosome 14 [57]. As graphically depicted in Fig. 3, the
human BCK gene is compact, as compared to the other three
CK genes. It has eight exons separated by seven short
introns. The most striking feature ofthe BCK gene structure
is the small size of the introns, which range from 72 bp
(intron III), 79 bp (intron VII), to 669 bp (intron V), with
introns III and VII meeting only the minimal length require-
ment for accurate splicing. However, all 7 pairs of exon-intron
junctions conform to the 5'-and 3'-splice site consensus
sequences. The translation initiation codon, AUG, is located
in exon 2, and preceded by the consensus translation
initiation signal, CCGCc. Nucleotide composition analysis
demonstrates that the human BCK gene has a high G+C
content. Overall, G+C constitutes 69.5% ofthe nucleotides,
with the 5'-flanking region also high in G+C (69%) [52, 58].
This feature is typical ofthe 5'-flanking regions of so-called
housekeeping genes which are widely expressed in many
tissues and cell types.
Potential regulatory elements of the human BCK gene
Sequence analysis of the 5'-flanking region of the human
BCK gene identified elements potentially important for
regulating gene transcription. A TfA-rich sequence, TTAA,
is centered at -26 from the presumptive transcription start
point, and a second TfA-rich sequence, TATAAATA, is
centered at -62. In addition, there are two possible CAT
sequences, GCCAATG (-52) and GGCCAATG (-82); three
copies of the Sp I consensus sequences, GGGCGG (-46, -
120, and -144); and a GCACCACCC sequence (-98) [52,
58, 59]. Although nonconsensus to the TATA box, the TTAA
sequence at -26, preceded by a CAT sequence at -52 and
located at the expected position, was likely the functional
TATA box which initiated transcription in vivo [60]. Similar
5'-flanking sequence organization and use of the TATA box
were found for the rat BCK gene [61, 62].
BCK tissue distribution and regulation of expression
BCK is abundantly expressed in brain, heart, smooth muscle,
uterus, placenta, colon [29], as well as in tumors [63] and
most permanent cell lines [64, 65]. In skeletal muscle, BCK
expression is very low. An absolute absence of B-CK is
observed in bone marrow [52]. Immunocytochemistry in
mouse and rabbit using a monoclonal antibody against a 17-
mer synthetic peptide derived from the N-terminus of the
human BCK demonstrated further that BCK expression is
restricted to a specific subset of cells in various tissues,
including neurons, but not glia, in brain; the inner segments
of photoreceptor cells and the outer plexiform layers of the
retina; and parietal cells in the stomach, enterocytes in the
gut, and epithelial cells of the urogenital system [66].
There are several interesting features regarding regulation
ofBCK gene expression. The second TfA-rich sequence in
the human and rat BCK genes, CTATAAATAG, conforms to
the TATA box consensus, is preceded by a CAT box and
located in the immediate BCK promoter, but is also homo-
logous to the AfT -rich consensus binding site of the MEF2
family of MADS box transcription factors [9, 42, 43]. In
promoter of the Xenopus MyoDa (XMyoDa), the TATA box
overlapped precisely the MEF2 site at-28 (CTATAAATAG)
and transcription of the minimal XMyoDa promoter in
nonmuscle cells was activated by expression of Xenopus
MEF2 and required binding of both MEF2 and TFIID to the
TfA-rich consensus [67]. Indeed, this second AfT -rich
consensus sequence found in rat BCK promoter bound the
same activity as that for the MCK enhancer AfT -rich con-
sensus, TARP (TA-richrecognition12rotein) [68]. The MEF2
gene is expressed in neuronal tissues [69, 70]. It remains to
be determined if this TATA box is used for the BCK gene

under certain circumstances, how MEF2 interacts, along with
TFIID, to regulate BCK gene expression in brain.
Another feature of BCK gene regulation is that, during
myogenesis, the cytosolic CK isozymes undergo an isotype
switch characterized by a transition from the embryonic
dimeric BB form to the adult MM form, with the MB hybrid
as the intermediate [71-74]. Studies from in vitro transfection
of C2C 12 cells indicate the presence of a suppressor between
-ll50 to -388 bp [64, 75]. A positive element between +41
to +57 bp was also identified which affected reporter gene
expression by 10 fold, based on results from in vitro transient
transfection of rat neonatal cardiomyocytes and in vivo
transfection by injection into the adult rat heart [75]. In gel
electrophoresis mobility shift assay, the fragment, +25 to +57
bp, contained site( s) to which a nuclear protein from neonatal
rat heart, C22 12 cells, and in much lower abundance, adult rat
heart extracts, bound specifically. The abundance of this
binding activity appeared to parallel that of the BCK protein,
providing the possibility that this DNA binding activity might
regulate developmental regulation of the BCK gene. In
addition, study ofthe rat BCK promoter revealed that mouse
p53, a tumor suppressor protein, repressed BCK expression
in Hela human cervical carcinoma cells, but activated MCK
gene in CV-I monkey kidney cells [76]. It would be inter-
esting to see whether p53 is involved in the isotype switch
in cardiac and skeletal myocytes.
A third feature ofBCK expression is its modulation by a
variety of steroid hormones [77-82]. BCK expression in
uterus and diaphyseal bones of female rats was induced after
estrogen or progesterone administration. In male rats,
testosterone had the same effect on diaphyseal bones.
Although the 1.7 kb BCK 5'-flanking region lacked an
estrogen response element, -GGTCAnnnTGACC-, this region
was responsive to ~-estradiol in Hela cells, and rat primary
fibroblasts cotransfected with estrogen receptor expression
vector [81,82]. Sequences between-75 and--45 bp, where a
T/A-rich sequence and a CCAAT sequence are located, are
crucial for the estrogen response. Estrogen receptor might
interact with the T/A-rich binding protein to stimulate BCK
gene expression. It is currently unknown how this interaction
occurs.
Regulation ofBCK, which is the more widely expressed of
the cytosolic CK isozyme genes, has not been studied in as
much detail its sister MCK gene. Further studies of gene
regulation may prove enlightening in understanding energy
production in rapidly dividing cells and in the ultimate among
terminally-differentiated cells, the neuron.
Two Mitochondrial Creatine Kinase genes
A mitochondrial isoform of the creatine kinases was first
discovered in 1964 by Jacobs and coworkers when an in-gel
159
kinase reaction detected an activity migrating to the cathode
end of the loading well [83]. Twenty-five years later, it was
shown that the MtCKs from heart and brain were two distinct
isozymes that are different in their N-terminal amino acid
sequences, peptide maps, immunoreactivity, and kinetic
parameters [84, 85]. The existence of two distinct MtCK
isozyme-encoding genes was conclusively demonstrated by
the cloning and characterization oftwo independent cDNAs
and nuclear genes encoding these MtCK isozymes in our
laboratory [30, 55, 86].
Our original isolation of an MtCK clone was fortuitous
[86]. We were interested in cloning the cytoplasmic CK genes
from humans and, so, screened a human phage genomic
library with a canine heart MCK probe. Among many
genomic clones isolated with restriction enzyme fragments
hybridizing to these canine MCK coding regions, we selected
one fragment with the most intense signal on Southern blot
for sequence analysis. The exonic sequences within this clone
differed from both M and BCK cDNAs, raising the possibility
that this encoded an MtCK. To obtain an MtCK cDNA clone,
we used a 312 bp fragment ofthis genomic clone to screen a
human placental MtCK cDNA. Characterization of this
cDNA clone proved that we had found the elusive MtCK.
This fortunate quirk of fate formed the basis for our sub-
sequent isolation and characterization of both human MtCK
cDNAs and genes [30, 55, 86].
The two different MtCK genes and proteins have received
various names, depending upon biochemical properties, sites
of tissue expression, and species derivation. Because of its
wide (albeit not universal) distribution oftissue expression,
we have used the term, ubiquitous MtCK (uMtCK), to refer
to the MtCK gene product originally isolated from placenta.
Others refer to this as the acidic form of MtCK or Mi a -CK
[87]. The expression of the second MtCK gene product was
restricted to heart and skeletal muscle, hence the name,
sarcomeric MtCK (sMtCK) [55]. However, as others have
noted, this gene product, with a more basic pI, may be
expressed in non-sarcomeric tissues such as spermatozoa
[88]. These investigators prefer the designation of basic
MtCK or Mib -CK.
The ubiquitous Mitochondrial Creatine Kinase gene
Structural characterization of the human uMtCK mRNA
As noted above, we first isolated the human gene encoding
the uMtCK and used a gene fragment to probe a human
placental cDNA library. The human MtCK eDNA thus
characterized was 1.5 kb in length, including 1252 bp of
coding sequence flanked by 160 bp of5'-UTR and 133 bp of
3'-UTR (Fig. I). This was the first complete MtCK cDNA to
be characterized, although Perriard, Wallimann, Eppenberger
and coworkers reported a partial chick muscle MtCK cDNA
160
HumanuMtCK NL\GPFSRLLSARGLRLLALAGAGSLAAGELLRPEPVRA
HumansMtCK NL\SIFSKLLIGRNASLLEATMGTSVLTTGYLLNRQKKVCA
Fig. 4. Comparison of predicted ubiquitous and sarcomeric MtCK transit peptide sequences. The identical amino acids are highlighted. Conservative
substitutions are underlined.
in 1988 [84]. This human uMtCK cDNA encoded a precursor
MtCK of 416 amino acids, including a 38 amino acid transit
peptide responsible for the mitochondrial targeting and import
(Figs 2 and 4) [86]. We also isolated the rat uMtCK cDNA
[28] and used this to demonstrate a high degree of homology
in both the coding region (92% nucleotide identity) and the
3'-UTR (82% identity) to the human homolog. The mouse
[89] and chick [87] uMtCK cDNAs are also highly homo-
logous within the mature protein coding regions (about 90%
overall). Among mammals, the transit peptides are also highly
conserved across species, suggesting that mitochondrial
uptake may require tissue-specific sequences.
Comparison of the amino acid and nucleotide sequence
identities among the human CK gene products reveals that
uMtCK is 63-66% identical to the cytosolic isozymes
(Table 1). This is also true in other species [15]. These
results support the concept that the CKs form a group of
evolutionarily conserved enzymes and that the gene
duplication event separating mitochondrial from cytosolic
genes occurred early in phylogeny.
Structural characterization of the human uMtCK gene
Comparison of the human placental MtCK cDNA clone with
the genomic MtCK clone revealed the structure of the uMtCK
gene (Fig. 3) [86]. The human uMtCK gene is located on
chromosome 15 [90]. The gene spans only 5.5 kb in the genome
and contains 9 exons varying in length from 86-310 bp
clustered in two groups (exons 1-6 and 7-9) separated by
an intron of 1.7 kb. The other 7 introns are less than 600 bp
in length. The nucleotide sequences surrounding intron-exon
junctions correspond well to those predicted by homology
data and the AG-GT splice junction rule for intron-exon
boundaries. In retrospect, it is likely that the compact nature
of this gene led to its fortuitous hybridization to the canine
Table 1. Percent amino acid and nucleotide sequence identities among
human creatine kinase gene products
Identity
sMtCK uMtCK MCK BCK
sMtCK 80 65 62
uMtCK 73 63 66
MCK 64 64 94
BCK 65 63 78
Numbers above and to the right of the blank diagonal are amino acid
sequence identity; numbers below and to left are coding region nucleotide
sequence identity.
MCK probe, as mentioned above. Although the human
uMtCK gene has a G+C content of 48%, similar to that of
the human BCK gene, the 5'-flanking region of the uMtCK
gene is G+C rich (62%). Moreover, the 5'-flanking region
has no T/A-rich region, no CAAT and TATAmotifs, but it
does have two potential Sp 1 binding sites (CCGCCC or
GGGCGG, -206 and -51 bp). A CpG island encompasses
exon 1. These are features typical of ubiquitously expressed,
housekeeping genes.
Subsequently, the structures ofthe mouse [89] and chicken
[87] genes were reported. All three known uMtCK genes are
compact, especially within the mature protein coding region
exons, and share an identical intron-exon boundary structure
and identical exon sizes (except for the UTR-encoding first
and last exons). The sixth exon of all three uMtCK genes is
identical in size, a feature shared among all known cytosolic
and mitochondrial CK genes (Fig. 1).
These results demonstrate the highly conserved evo-
lutionary nature of the uMtCK genes among birds and
mammals. including humans.
Tissue expression of the uMtCK gene
The discovery that two different MtCK genes existed sug-
gested that tissue-specific, or cell-specific, expression
would occur. Although many earlier studies ofMtCK protein
levels had been done both with gel electrophoresis to separate
isoforrns and by immunologic techniques, interpretation is
somewhat confused by the potential for antibody cross
reactivity or nucleic acid cross hybridization and by the
assumption prior to 1988 that only one MtCK gene product
existed.
We examined uMtCK expression in rodents [28, 91, 92]
and noted that expression ofuMtCK mRNA, as analyzed by
quantitative RNA blot, is 10-100 fold less than the other
three CK mRNAs in all tissues. This suggests that uMtCK
mRNA is very stable and that translation may be very
efficient, as uMtCK protein levels may become comparable
to other CKs in some tissues. uMtCK mRNA is abundantly
expressed in conjunction with BCK in intestine, brain,
kidney, placenta and, during pregnancy, uterus [91, 93]. Very
low level expression of uMtCK was detected in aorta and
sarcomeric tissues. No expression was detected for liver and
lung. Perriard and coworkers [87] recently demonstrated
similar high levels of chicken uMtCK mRNA expression in
brain, spinal cord, and gut coordinated with BCK expression.
Chicken liver did not express either B or uMtCK. More

detailed analysis of cell-specific uMtCK mRNA expression,
particularly in brain, is underway in our laboratory.
Transcriptional regulation of the uMtCK gene
No analysis of a uMtCK gene promoter has been reported to
date. Such analyses are hindered by the lack of permanent
cell lines that express mitochondrial proteins, including the
uMtCK, to a high level, as compared to differentiated and
energy-consuming tissues, such as brain. Furthermore, cell
lines similar to mature neurons do not exist. To study the
mechanisms and genomic sequences regulating uMtCK gene
expression in this variety of tissues, we compared the 5'-
flanking region of the human uMtCK gene to that ofthe mouse
recently isolated by Wieringa and coworkers [89]. Alignment
of the 5'-flanking regions of the two genes revealed 71%
nucleotide identity over a 600 bp region encompassing bp -
1344 to -753 in the human and bp -1300 to -707 in the
mouse, relative to their respective transcription start points.
We are now testing the functional importance of this cross-
species sequence conservation in vivo in mice by including
or excluding this region in the transgenic constructs (Boero,
Khuchua, Qin, and Strauss, work in progress). We also
noticed ,in our sequence analysis that a consensus nuclear
hormone receptor binding site (NRRE), TGACCTtTGTCCT,
is present within this region. This sequence, with two
consensus hexamer binding sites separated by 1 nucleotide,
is a DR (direct repeat)-1 NRRE. By using gel electrophoresis
mobility shift, competition, and supershift assays, we identi-
fied a brain protein binding to this site and the binding activity
was a RXRaJPPAR (retinoic-X receptor/peroxisomal proli-
ferator activating receptor) heterodimer. These preliminary
results suggest that the heterodimeric brain nuclear hormone
receptor, RXRa/PPAR, may regulate expression of the
uMtCK gene in the central nervous system.
Our reported data suggested that, similar to BCK, the
uMtCK gene is also regulated by estrogen [91]. Rodent
uMtCK mRNA expression was normally low in uterus, but
rose during pregnancy and rapidly fell (by more than 10 fold)
after delivery. In placenta, uMtCK RNA was induced by 20
days of rat gestation and also fell dramatically after delivery.
Our data from in vitro transfection studies of a human endo-
metrial adenocarcinoma cell line, AN3-CA, suggested that
estrogen response element was not located within the -2580
to +140 bp of the human uMtCK gene 5'-flanking region
(Cheng and Strauss, unpublished data). It remains to be
determined whether the estrogen response element is located
elsewhere in the gene.
The 3'-UTRs of both the BCK and uMtCK are strikingly
conserved across species, including chicken, mouse, rat,
rabbit, dog, and human, suggesting a role of these 3'-UTRs
in regulating gene expression ([94, 95] and Cheng and
Strauss, unpublished data). For example, rat and human BCK
3'-UTRs are 77% homologous within a 200 bp region, and
161
rat and human uMtCK 3'-UTRs are 81 % identical over a
130 bp fragment. In contrast, 3'-UTR sequence homology is
less than 50% among the four CK mRNAs within a species.
RNA gel mobility shift assays identified a 94 kDa protein
present in rodent brain and uterus that binds specifically to
both the BCK and uMtCK 3'-UTRs (Cheng and Strauss,
unpublished data). We are currently investigating the
functional significance of this binding.
We conclude that the uMtCK is highly conserved across
the evolutionary tree in size and structure; that expression
levels vary greatly by tissue, during development, and during
the estrus cycle and pregnancy; and that the DNA sequences
and protein factors regulating uMtCK gene expression
probably include members of the steroid superfamily of
transcription factors.
The sarcomeric Mitochondrial Creatine Kinase gene
As noted in the previous sections, evidence emerged about
10 years ago that two different isozymes of MtCK existed,
with specific patterns of tissue expression. Suggestive results
from Wallimann, Perriard, and Eppenberger's group with
chick MtCK protein characterizations and from our own
molecular studies clearly documented this phenomenon. That
tissue-specific isoforms, encoded by distinct genes, for
mitochondrial proteins existed was a novel concept at that
time. Subsequent studies with nuclearly-encoded cytochrome
oxidase subunits, the adenine nucleotide translocator, and
porins have proved the generality ofthis. This tissue-specific
(or, perhaps cell-specific) expression of proteins involved
in the production and channelling of energy from the site of
synthesis in the mitochondrial matrix through the mito-
chondrial membranes and intermembranous space to sites
of utilization in the cytosol is consistent with the hypo-
thesis that channels formed by multiple proteins exist [12],
especially in high energy requiring tissues such as muscle
and brain, and that these channels are different in various cell
and tissue types because of differential expression of
isozyme genes.
Investigation of this cell-specific energy 'channel hypo-
thesis' is now a significant focus in our laboratory.
Structure of the human sMtCK mRNA
At virtually the same time (1987-88), two laboratories
independently cloned sMtCK cDNAs. Perriard's group used
the traditional approach of protein purification and antibody
screening to isolate a partial chicken sMtCK cDNA [84]. As
noted above, we [55] used a probe from the fortuitously
isolated human uMtCK gene to screen a human heart cDNA
library under low stringency conditions and identified over-
lapping positive clones. The human sMtCK cDNA contains
an open reading frame of 1257 bpflanked by 198 bp of the
162
5'- and 143 bp of the 3'-UTRs (Figs 1 and 2). Polyadenyl-
ation at the 3' end is signaled by the consensus sequence,
AATAAAA. We subsequently isolated and characterized
rodent (both rat and mouse) sMtCK cDNAs [28]. The 3'-
UTR of mammalian sMtCK mRNAs is also highly conserved.
As with the BCK and uMtCK mRNAs, this extraordinary
degree of sequence identity suggests a potential role in
regulation of transcription, translation, or mRNA stability
or localization. This putative role remains to be investigated.
The encoded protein is 419 amino acid in length including
a transit peptide of 39 amino acids (Figs 2 and 4). Across
species, the sMtCK mature protein coding region is highly
conserved (about 90% at the amino acid level). The recent
and exciting determination of the sMtCK structure ([ 11], and
in this book) emphasizes that numerous regions of the sMtCK
subunits are essential not only for enzymatic function, but for
oligomer formation and for interaction with the inner mito-
chondrial membrane. Given these results, conservation
throughout the sMtCK subunit is not surprising.
Structure of the human sMtCK gene
We used the human sMtCK cDNA to screen a genomic
library and isolated the complete sMtCK gene [30] (Fig. 3).
The human sMtCK gene is located on chromosome 5, spans
37 kb of the genome, and contains 11 exons ranging in length
from 77 to 241 bp. Reminiscent of the organization of many
muscle-specific genes [96], the first exon is untranslated and
followed by a relatively large intron (10.7 kb). The second
exon is also untranslated, and intron 2 is also large, 7 kb in
length. The translation start codon is located in exon 3,
approximately 17 kb from the transcription start point. In
contrast to the more widely expressed BCK and uMtCK
genes, the sMtCK gene is much larger (37 kb versus 5-7 kb);
contains a substantial, G+C rich 5'-UTR of 345 bp; and has
features typical of genes expressed tissue-specifically,
including a TATAA box at position -35 and three CCAAT
sequences from-135 to-180. To date, this is the only sMtCK
gene characterized, although the mouse gene has been
isolated during gene ablation studies (see Wieringa and
coworkers in this volume).
Two distinct, yet closely related, MtCK genes
Alignment of the predicted amino acid sequences of the two
human MtCK isozymes revealed high homology, with 73%
nucleotide and 80% predicted amino acid identities (Fig. 2
and Table 1). From residues 12-373,86% identity is present,
with 21 of the 49 differences representing conservative
changes. The area surrounding Cys-278 in the active site [97]
is conserved among all CK isozymes [15]. However, dif-
ferences are spread throughout the coding region of the two
MtCKs. Moreover, the 5'- and 3'-UTRs of the two MtCK
cDNAs, though similar in overall lengths, have widely
divergent nucleotide sequences (less than 40% identity). This
sequence divergence clearly demonstrates that the two MtCK
isozymes are indeed encoded by two distinct, although
closely, related genes.
Both ubiquitous and sarcomeric MtCKs have an N-terminal
extension of 39 residues, the transit peptide, which leads
the protein across the outer mitochondrial membrane [86]
(Fig. 4). These transit peptides share only 46% identity,
although 5 of the 22 differences represent conservative
changes. Both peptides are rich in amino acids with hydroxyl-
containing and basic side chains, features typical of the
transit peptide for mitochondrial matrix targeting. We have
shown that this transit peptide can target the human uMtCK
precursor protein into mitochondria [86].
N-terminal sequences of the mature MtCKs are similar in
18 of the first 26 residues (69%). However, the mature N-
termini ofMtCK isozymes are conserved in a tissue-specific,
rather than a species-specific manner. This region is likely
necessary for oligomer formation [11] and, thus, may be
essential in channel formation.
The sMtCK gene is expressed in oxidative cardiac and
skeletal myocytes
To support the extraordinary and continuous requirement for
energy, cardiomyocytes express unique genes associated with
specific energy metabolism pathways [30, 98-100]. The
sMtCK gene represents one particularly obvious example of
this phenomenon. By RNA blot, protein blot, and immuno-
cytochemistry, we demonstrated that the sMtCK gene is
exclusively expressed in cardiomyocytes, slow-oxidative,
and fast-oxidative-glycolytic skeletal myocytes of heart and
skeletal muscle and not in any other organs tested, including
brain, lung, stomach, intestine, liver, kidney, bladder, testis,
and uterus. The high level of expression in cardiac and
skeletal myocytes provides a model system to study gene
expression in the heart as compared to that in skeletal muscle
(Qin and Strauss, unpublished data).
Transcriptional regulation of the human sMtCK gene
To analyze the regulatory elements and transcription factors
required for tissue-specific expression of the human sMtCK
gene, we isolated and sequenced 4.6 kb of the 5'-flanking
region of the human gene [30]. As described above, this
region possesses features of a highly regulated gene: a
TATAA box at nucleotide -36, three CCAAT sequence
homologues at positions -134, -155, and -175, and a non
GC-rich promoter region prior to the transcription start point.
The 5'-nested deletion constructs were created and transfected
into three types of cells: rat neonatal cardiomyocytes, C2 C I2
mouse skeletal myocytes, and NIH3T3 mouse embryo
fibroblasts. The results demonstrated that the human sMtCK
genomic sequence, -1308/+6 bp, directed high level ex-
pression of the CAT reporter gene in cardiac and skeletal
myocytes and that the sequence between -921 and -485 is

mainly responsible for the high level expression of the CAT
gene. Guided by the in vitro transient transfection results,
we generated three transgenic mouse lines carrying the
human sMtCK gene 5'-flanking region, -921 to +6 bp, -760
to +6 bp, and -485 to +6 bp, linked to the 5' end ofthe human
growth hormone reporter gene. Our results demonstrated that
the sequence, -921 to +6 bp, contains all necessary and
sufficient elements specifying gene expression to the cardiac
and skeletal myocytes, and the sequence between -921 and
-760 bp is required for a high level, myocyte-specific gene
expression (Qin et ai., JBC, in press).
Sequence analysis revealed that within this 160 bp region,
from -921 to -760 bp, several known transcription factor
binding sites exist and include: four GATT motifs (GATT
at -910 and -893 bp; AATC at -883 and -863 bp), one E
box (CACGTG at -827 bp), and one MEF2 consensus
(ATATTTTTAA at -782 bp). Preliminary results from gel
electrophoresis mobility shift, site-specific mutagenesis,
competition, and supershift studies with nuclear proteins from
the heart established that GATT at -893, AATC at -863, E
box at-827, and the MEF2 at-782 bp can bind, respectively,
GATA related proteins, USF -1 (upstream stimulatory factor
1), and MEF2, and the GATT-binding protein is not a GATA4.
Identity of the GATT-binding protein and the functional
significance as related to cardiac-specific gene expression is
now under further investigation (Qin and Strauss, un-
published data).
A model gene for studying cardiac myogenesis
The heart is the first organ to form in a vertebrate embryo
[101]. In the last few years, cardiac myogenesis and morpho-
genesis have begun to be understood at the molecular level
with the cloning and characterization of essential cardiac
transcription factors. These include Nkx2.5 [102, 103],
eHAND and dHAND [104-107], GATA4 (l08-110), and
MEF2 [42,43]. Studies in Drosophila also demonstrated that
dorsal tube formation, the heart equivalent in Drosophila,
requires the signaling pathway ofthe segment polarity gene,
wingless [Ill, 112], and a homeodomain-containing gene,
tinman [113, 114]. The spacial and temporal pattern of
expression of these genes in flies and vertebrates suggests
their crucial roles in cardiac myogenesis and morphogenesis,
which is further demonstrated in gene knockout (D-MEF2
of Drosophila [48]; Nkx2.5 in mice [115]) and mRNA ablation
studies (dHAND and eHAND [107]). However, numerous
questions remain as to how these transcription factors are
placed in the pathway culminating in formation of the heart,
how they interact with each other, and how they specify gene
expression to the heart. It is our hope that study of the human
sMtCK gene expression in the heart will add to our under-
standing of cardiac myogenesis and morphogenesis.
The discovery that cell or tissue-specific isoforms of
mitochondrial protein genes, such as the MtCKs, exist
163
emphasizes the need for understanding the mechanisms
regulating this specificity ofexpression ofenergy producing
and transducing genes and the need to understand how
differences in energy transduction or channelling occur as
a result of this differential tissue expression.
The creatine kinase gene family
Cellular energy metabolism and the creatine kinases
ATP is the universal carrier of metabolic energy. In non-
photosynthetic eukaryotic cells, ATP is predominantly
synthesized in mitochondria, where most of the energy-
yielding oxidative reactions are localized, including oxidative
phosphorylation, the Krebs or citric acid cycle, ~-oxidation
of fatty acids, and the oxidative degradation of amino acids
to acetyl-CoA.
Creatine kinases are guanidino kinases that catalyze the
reversible transfer of a high energy phosphate between ATP
and creatine, generatingADP and the high-energy phosphate
carrier, creatine phosphate. The creatine phosphate shuttle
hypothesis proposes that the CK isozymes playa central role
in cellular energy metabolism by transporting energy from
site of production, the mitochondrion, to various locations of
utilization, including the sarcomere, sarcoplasmic reticulum,
and plasma membrane [116, 117]. This cellular function is
collectively performed by the cytosolic CKs and the mito-
chondrial CKs. By directly coupling oxidative phosphory-
lation to energy utilization, the CK isozymes effect cellular
energetics, supporting the high and sustained energy require-
ment of tissues and vital organs as the heart and brain.
Features of this CK gene family
Sequence and structural analysis demonstrated that the four
CK genes are highly conserved and form a gene family. In
humans, the sMtCK gene coding region shares 73% nucleo-
tide identity with uMtCK, 64% with MCK, and 65% with
BCK (Table I). At the amino acid level, the sMtCK protein
shares 80% homology with uMtCK, 65% with MCK, and
62% with BCK (Table 1). Further structural analysis indicates
that, between the two MtCK isoforms, the length of each
coding region exon and the locations of exon junctions within
the MtCK cDNAs are absolutely identical (Figs 1 and 3). For
example, ex on 3 of the uMtCK gene corresponds precisely
to exon 5 of the sMtCK gene. However, the uMtCK gene
spans only 5.5 kb and the clustering of exons (2-6 and 7-9)
apparent in the uMtCK gene does not occur in the sMtCK
gene. Similarly, genes encoding the two cytosolic CK
isoforms share identical size and location of their coding
region exons (Figs 1 and 3). However, the cytosolic CK gene
exons differ from the MtCK gene exons in length and location
with one exception. Exon 6 in the BCK, MCK, and uMtCK
genes and ex on 8 in sMtCK gene are identical. This genomic
164
structure of the CK gene family suggests that two specific
gene duplication events occurred. From a common primordial
gene, the first duplication event resulted in an ancestral
mitochondrial and an ancestral cytosolic CK gene. The
second duplication event gave rise to nuclear genes encoding
the two MtCKs and two cytosolic CK isoforrns.
Coordinate regulation of CK gene expression
In humans, the four CK genes are localized on different
chromosomes, with the sMtCK gene on chromosome 5
[30], the uMtCK gene on chromosome 15 [90], the MCK gene
on chromosome 19 [90], and the BCK gene in chromosome
14 [27]. Extensive analyses by in-gel kinase reaction, protein
blot, and mRNA blot demonstrated that the cytosolic CKs
and the MtCKs are often coordinately expressed. That is,
when the sMtCK gene is expressed in a tissue, the MCK gene
product is also detected at similar levels. The known tissues
that coordinately express the sMtCK gene with MCK gene
are heart and skeletal muscle. This theme of a mitochondrial
CK teamed up with a cytosolic CK is echoed by the uMtCK
and BCK genes which are highly and coordinately expressed
in brain, uterus, placenta, and other organs. We conclude that
shared, upstream, trans-acting mechanisms must exist to
coordinate the expression of these genes located on different
chromosomes. In the case of the sMtCK gene and the MCK
gene, transcription factors activating gene expression in
muscle, such as MyoD and MEF2, are likely the mediating
factors. The mechanisms coordinately expressing the
uMtCK and the BCK genes are presently unknown.
Conclusions
The molecular characterization of CK mRNAs and genes
over the last 15 years has paralleled the revolution in mole-
cular biology and has provided crucial tools for isolation
of transcription factors essential in skeletal myogenesis
and cardiac development. Our discovery of two nuclear
genes encoding mitochondrial CKs with different patterns
of expression, together with evidence of other tissue-
specific mitochondrial protein isoforrn expression, suggests
that the mitochondrial energy channeling hypothesis
deserves investigation. The coordinate expression of
cytosolic CK isoform genes with respective mitochondrial
CK isoform genes in a tissue and cell-specific manner
emphasizes the need to understand regulation of these
energy producing and transducing genes in tissues with high
energy requirements and to isolate transcription factors
mediating this regulation. The CKs have provided fertile
ground for understanding biological processes at an
unprecedented molecular level.
References
1. Watson JD, Crick FH: Molecular structure of nucleic acids: A structure
for deoxyribose nucleic acid. Nature 171: 737-738, 1953
2. Nathans D, Smith HO: Restriction endonucleases in the analysis and
restructuring of DNA molecules. Annu Rev Biochem 44: 273-293, 1975
3. Lederberg 1: Viruses, genes and cells. Bacteriol Rev 21: 133-139, 1957
4. Berg P: The viral genome in transformed cells. Proc R Soc Lond B Bioi
Sci 177: 6~76, 1971
5. Sharp PA: Splicing of messenger RNA precursors. Science 235: 766-
771, 1987
6. Maxam AM, Gilbert W: A new method for sequencing DNA.
Proc Natl Acad Sci USA 74: 560-564, 1977
7. Sanger F, Nicklen S, Coulson AR: DNA sequencing with chain-
terminating inhibitors. Proc Natl Acad Sci USA 74: 5463-5467,1977
8. Buskin IN, Hauschka SD: Identification of a myocyte nuclear factor
that binds to the muscle-specific enhancer of the mouse muscle creatine
kinase gene. Mol Cell Biol9: 2627-2640, 1989
9. Gossett LA, Kelvin DJ, Sternberg EA, Olson EN: A new myocyte-
specific enhancer-binding factor that recognizes a conserved element
associated with multiple muscle specific genes. Mol Cell Bioi 9: 5022-
5033, 1989
10. Lassar AB, Buskin lN, Lockshon D, Davis RL, Apone S, Hauschka SD,
Weintraub H: MyoD is a sequence-specific DNA binding protein requiring
a region of myc homology to bind to the muscle creatine kinase enhancer.
Cell 58: 823--831, 1989
II. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of
mitochondrial creatine kinase. Nature 381: 341-345, 1996
12. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells:
Role of coupled creatine kinases in vivo regulation of cellular respiration
-a synthesis. Mol Cell Biochem 133/134: 15~192, 1994
13. Weintraub H, Robert D, Tapscott S, Thayer M, Krause M, Benezra R,
Blackwell TK, Turner D, Rupp R, Hollenberg S, Zhuang Y, Lassar A:
The myoD gene family: Nodal point during specification of the muscle
cell lineage. Science 251: 761-766, 1991
14. Olson EN, Perry WM, Schulz RA: Regulation of muscle differentiation
by the MEF2 family of MADS box transcription factors. Dev Bioi 172:
2-14,1995
15. Muhlebach SM, Gross M, Wirz T, Walliman T, Perriard JC, Wyss M:
Sequence homology and structure predictions of the creatine kinase
isoenzymes. Mol Cell Biochem 133/134: 24~262, 1994
16. Schweinfest CW, Kwiatkowski RW, Dottin R: Molecular cloning of a
DNA sequence complementary to creatine kinase M mRNA from
chickens. Proc Nat! Acad Sci USA 79: 4997-5000, 1982
17. Rosenberg UB, Kunz G, Frischauf A, Lehrach H, Mahr R, Eppenberger
HM, Perriard JC: Molecular cloning and expression during myogenesis
of sequences coding for M-creatine kinase. Proc Natl Acad Sci USA
79:6589-6592,1982
18. Ordahl RP, Evans GL, CooperTA, Kunz G, Perriard1: Complete cDNA-
derived amino acid sequence of chick muscle creatine kinase. J Bioi
Chem259: 15224-15227, 1984
19. Giraudat J, Devillers-Thiery A, Perriard J-C, and Changeux J-P: Complete
nucleotide sequence of Torpedo marmorate mRNA coding for the
43,000-dalton v, protein: Muscle-specific creatine kinase. Proc Nat!
Acad Sci USA 81: 7313-7317,1984
20. West BL, Babbitt PC, Mendez B, Baxter JD: Creatine kinase protein
sequence encoded by a cDNAmade from Torpedo californica electric
organmRNA. Proc Nat! Acad Sci USA 81: 7007-7011,1984
21. Putney S, Herlihy W, Royal N, Pang H, Aposhian HY, Pickering L,
Belagaje R, Biemann K, Page D, Kuby S, Schimmel P: Rabbit muscle
creatine kinase phosphokinase cDNA cloning, primary structure, and
detection of human homologues. J Bioi Chern 2591: 4317-14320, 1984

22. Benfield PA, Zivin RA, Miller LS, Sowder R, Smythers GW, Henderson
L, Oroszlan S, Pearson ML: Isolation and sequence analysis of cDNA
clones coding for rat skeletal muscle creatine kinase. J Bioi Chern 259:
14979-14984, 1984
23. Kwiatkowski RW, Schweinfest CW, Dottin RP: Molecular cloning and
the complete nucleotide sequence of the creatine kinase-M cDNA
from chicken. Nucleic Acids Res 126: 6925-6934, 1984
24. Buskin IN, Jaynes JB, Chamberlain JS, Hauschka SD: The mouse muscle
creatine kinase cDNA and deduced amino acid sequences: Comparison
to evolutionarily related enzymes. J Mol Evo122: 334-341, 1985
25. Roman D, Billadello J, Gordon J, Grace A, Sobel B, Strauss A: Complete
nucleotide sequence of dog heart creatine kinase mRN A: Conservation
of amino acid sequence within and among species. Proc Nat! Acad Sci
USA 82: 8394-8398, 1985
26. Perryman MB, Kerner SA, Bohlmeyer TJ, Roberts R: Isolation and
sequence analysis ofa full-length cDNA for human M-creatine kinase.
BiochemBiophys Res Comm 140: 981-989, 1986
27. Nigro JM, Schweinfest CW, RajkovicA, Pavlovic J, Jamal S, Dottin RP,
Hart JT, Kamarck ME, Rae PMM, Carty MD, Martin-Deleon P: cDNA
cloning and mapping of the human creatine kinase M gene to 19q13.
AmJHum Genet 40: 115-125, 1987
28. Payne RM, Haas RC, Strauss AW: Structural characterization and
tissue-specific expression of the mRNAs encoding isoenzymes from
two rat mitochondrial creatine kinase genes. Biochim Biophys Acta
1089: 352-361, 1991
29. Trask RV, Strauss AW, Billadello JJ: Developmental regulation and
tissue-specific expression of the human muscle creatine kinase gene. J
Bioi Chern 263: 17142-17149, 1988
30. Klein SC, Haas RC, Perryman B, Billadello JJ, StraussAW: Regulatory
element analysis and structural characterization of the human
sarcomeric mitochondrial creatine kinase gene. J Bioi Chern 266: 18058-
18065,1991
31. Jaynes JB, Chamberlain JS, Buskin IN, Johnson JE, Hauschka SD:
Transcriptional regulation of the muscle creatine kinase gene and
regulated expression in transfected mouse myoblasts. Mol Cell Bioi 6:
2855-2864,1986
32. Davis RL, Weintraub H, Lassar AB: Expression of a single transfected
eDNA converts fibroblasts to myoblasts. Cell 51 : 987-1000, 1987
33. Jaynes JB, Johnson JE, Buskin IN, Gartside CL, Hauschka SD: The
muscle creatine kinase gene is regulated by multiple upstream elements,
including a muscle-specific enhancer. Mol Cell Bioi 8: 62-70, 1988
34. Johnson JE, Wold BJ, Hauschka SD: Muscle creatine kinase sequence
elements regulating skeletal and cardiac muscle expression in transgenic
mice. Mol Cell Bioi 9: 3393-3399, 1989
35. Edmondson DG, Olson EN: A gene with homology to the myc similarity
region ofMyoDI is expressed during myogenesis and is sufficient to
activate the muscle differentiation program. Genes Dev 3: 628-640,
1989
36. Wright WE, Sassoon DA, Lin VK: Myogenin, a factor regulating
myogenesis, has a domain homologous to MyoD. Cell 56: 607-617,
1989
37. Braun T, Buschhausen-Denker G, Bober E, Tannich E,Arnold HH: A
novel human muscle factor related to but distinct from MyoD I induces
myogenic conversion in 1OTl/2 fibroblasts. EMBO J 8: 701-709, 1989
38. Rhodes SJ, Konieczny SF: Identification ofMRF4: A new member of the
muscle regulatory factor gene family. Genes Dev 3: 2050-2061,1989
39. Braun T, Bober E, Winter B, Rosenthal N, Arnold HH: MyJ-6, a new
member of the human gene family of myogenic determination factors:
evidence for a gene cluster on chromosome 12. EMBO J. 9: 821-831,
1990
40. Miner JH, Wold B: Herculin, a fourth member of the MyoD family of
myogenic regulatory genes. Proc Nat! Acad Sci USA 87: 1089-
1093, 1990
165
41. Sternberg EA, Spizz G, Perry WM, Vizard D, Weil T, Olson EN:
Identification of upstream and intragenic regulatory elements that confer
cell-type-restricted and differentiation-specific expression on the muscle
creatine kinase gene. Mol Cell Bioi 8: 2896-2909, 1988
42. Pollock R, Treisman R: Human SRF-related proteins: DNA-
binding properties and potential regulatory targets. Genes Dev 5:
2327-2341, 1991
43. Yu YT, Breitbart RE, Smoot LB, Lee Y, Mahdavi V, Nadal-Ginard B:
Human myocyte-specific enhancer factor 2 comprises a group of tissue-
restricted MADS box transcription factors. Genes Dev 6: 1783-1798,
1992
44. MoIkentin JD, Black BL, Martin JF, Olson EN: Cooperative activation
of muscle gene expression by MEF2 and myogenic bHLH proteins.
Cell 83 : 1125-1136, 1995
45. Treisman R: The SRE: A growth factor responsive transcriptional
regulator. SemCancerBiol1:47-58, 1990
46. Minty A, Kedes L: Upstream regions of the human cardiac actin gene
that modulate its transcription in muscle cells: Presence of an
evolutionarily conserved repeated motif. Mol Cell Bioi 6: 2125-2136,
1986
47. Schwarz-Sommer Z, Huijser P, Nacken W, Saedler H, Sommer H: Genetic
control of flower development by homeotic genes inAntirrhinum majus.
Science 250: 931-936, 1990
48. Lilly B, Zhao B, Ranganayakulu G, Paterson BM, Schulz RA, Olson
EN: Requirement of MADS domain transcription factor D-MEF2 for
muscle formation inDrosophiia. Science 267: 688-693,1995
49. Pickering L, Pang H, Biemann K, Munro H, Schimmel P: Two tissue-
specific isoenzymes of creatine kinase have closely matched amino
acid sequences. Proc Nat! Acad Sci USA 82: 2310-2314,1985
50. Benfield PA, Henderson L, Pearson ML: Expression of a rat brain
creatine kinase ~-galactosidase fusion protein in Escherichia coli
and derivation of the complete amino acid sequence of rat brain creatine
kinase. Gene 39: 263-267, 1985
51. Kwiatkowski RW, Ehrismann R, Schweinfest CW, Dottin RP:Accumul-
ation of creatine kinase mRNA during myogenesis: Molecular cloning
of a B-creatine kinase eDNA. Dev Bioi 112: 84-88, 1985
52. Mariman ECM, Broers CAM, Claesen CAA, Tesser GI, Wieringa B:
Structure and expression of the human creatine kinase B gene. Genomics
1: 126-137, 1987
53. Kaye FJ, McBride OW, Battey JF, Gazdar AF, Sausville EA: Human
creatine kinase-B complementary DNA: Nucleotide sequence, gene
expression in lung cancer, and chromosomal assignment to two distinct
loci. J Clin Invest 79: 1412-1420,1987
54. Harford JB, Klausner RD: Coordinate post-transcriptional regulation
of ferritin and transferrin receptor expression: The role of regulated
RNA-protein interaction. Enzyme 44: 2841, 1990
55. Haas RC, Strauss AW: Separate nuclear genes encode sarcomere-
specific and ubiquitous human mitochondrial creatine kinase iso-
enzymes. JBiol Chern 265: 6921-6927,1990
56. Villarreal-Levy G, Ma TS, Kerner SA, Roberts R, Perryman MB: Human
creatine kinase: Isolation and sequence analysis of cDNA clones for
the B subunit, development of subunit specific probes and deter-
mination of gene copy number. Biochem Biophys Res Comm 144: 1116-
1127,1987
57. Chern CJ, Tan P, Park H: Chromosomal mapping of human creatine
kinase (brain type) using human-rodent somatic cell hybrids. Cytogenet
Cell Genet 27: 232-237,1980
58. Mariman ECM, Schepens JTG, Wieringa B: Complete nucleotide
sequence of the human creatine kinase B gene. Nucleic Acid Res 17:
6385,1989
59. Daouk GH, Kaddurah R, Putney S, Kingston R, Schimmel P: Isolation
of a functional human gene for brain creatine kinase. J Bioi Chern
263: 2442-2446, 1988
166
60. Mariman E, Wieringa B: Expression of the gene encoding human
brain creatine kinase depends on sequences immediately following
the transcription start point. Gene 102: 205-212, 1991
61. Hobson GM, Mitchell MT, Molloy GR, Pearson ML, Benfield PA:
Identification ofa novel TA-rich DNA binding protein that recognizes
a TATA sequence within the brain creatine kinase promoter. Nuclei
Acids Res 16: 8925-8944,1988
62. Hobson G, Molloy GR, Benfield PA: Identification of cis-acting
regulatory elements in the promoter region of the rat brain creatine
kinase gene. Mol Cell Bioi 10: 6533--6543, 1990
63. Gazdar AF, Zweig MH, Carney ON, Van Steirteghen AC, Baylin SB,
Minna JD: Levels of creatine kinase and its BB isoenzyme in lung
cancer specimens and cultures. Cancer Res 41: 2773-2777, 1981
64. Ritchie ME, Trask RV, Fontanet HL, Billadello JJ: Multiple positive
and negative elements regulate human brain creatine kinase gene
expression. Nucleic Acids Res 19: 6231--6240, 1991
65. Kuzhikandathil EV, Molloy GR: Transcription of the brain creatine
kinase gene in glial cells is modulated by cyclic AMP-dependent
protein kinase. J Neurosci Res 39: 7082, 1994
66. Sistermans EA, de Kok YJM, Peters W, Ginsel LA, Jap PHK, Wieringa B:
Tissue- and cell-specific distribution of creatine .kinase B: A new and
highly specific monoclonal antibody for use in immunohistochemistry.
Cell Tissue Res 280: 435--446, 1995
67. Leibham 0, Wong MW, Cheng TC, Schroeder S, Weil PA, Olson EN,
Perry M: Binding ofTFIID and MEF2 to the TATA element activates
transcription oftheXenopus MyoDa promoter. Mol Cell Bioi 14: 686-
699,1994
68. Horlick RA, Hobson GM, Patterson JH, Mitchell MT, Benfield PA: Brain
and muscle creatine kinase genes contain common TA-rich recognition
protein-binding regulatory elements. Mol Cell Bioi I 0: 4826--4836, 1990
69. Lyons GE, Muhlebach S, MoserA, Masood R, Paterson BM, Buckingham
ME, Perriard J: Developmental regulation of creatine kinase gene
expression by myogenic factors in embryonic mouse and chick skeletal
muscle. Development 113: 1017-1029, 1991
70. Schulz RA, Chromey C, Lu MF, Zhao B, Olson EN: Expression ofthe
D-MEF2 transcription factor in the Drosophila brain suggests a role
in neuronal cell differentiation. Oncogene 12: 1827-1831, 1996
71. Adamson ED: Isoenzyme transition of creatine phosphokinase,
aldolase, and phosphoglycerate mutase in differentiating muscle cells.
J Embryo Exp Morph 35: 355--367, 1976
72. Ingwall JS: Creatine and the control of muscle-specific protein synthesis
in cardiac and skeletal muscle. Circ Res 38: 1115-123, 1976
73. Dym H, Yaffe 0: Expression of creatine kinase isoenzymes in myogenic
cell lines. Dev Bioi 68: 592-599,1979
74. Trask RV, Billadello JJ: Tissue-specific distribution and developmental
regulation ofM and B creatine kinase mRNAs. Biochim Biophys Acta
1049: 182-188,1990
75. Ritchie ME: Characterization of human B creatine kinase gene regulation
in the heart in vitro and in vivo. J Bioi Chern 271: 25485-25491, 1996
76. Zhao J, Schmieg FI, Simmons DT, Molloy GR: Mouse p53 represses
the rat brain creatine kinase gene but activates the rat muscle creatine
kinase gene. Mol Cell Bioi 14: 8483...s492, 1994
77. Notides A, Gorska J: Estrogen-induced synthesis ofa specific protein.
Proc Natl Acad Sci USA 56: 230--235, 1966
78. Reiss NA, Kaye AM: Identification of the major component of the
estrogen-induced protein of rat uterus as the BB isozyme ofCK. J Bioi
Chern 256: 5741-5749, 1981
79. Somjen 0, Weisman Y, Binderman I, KayeAM: Stimulation of creatine
kinase-BB activity by I alpha, 25-dihydroxycholecalciferol and 24R, 25-
dihydroxycholecalciferol in rat tissues. Biochem J 219: 1037-1041, 1984
80. Somjen 0, KayeAM, HarellA, Weisman Y: Modulation by vitamin 0
status of the responsiveness of rat bone to gonadal steroids.
Endocrinology 125: 1870--1876, 1989
81. Pentecost BT, Mattheiss L, Dickerman HW, Kumar SA: Estrogen
regulation of creatine kinase-B in the rat uterus. Mol Endocrinol 4:
1000--1010,1990
82. Sukovich DA, Mukherjee R, Benfield PA: A novel, cell-type-specific
mechanism for estrogen receptor-mediated gene activation in the
absence of an estrogen-responsive element. Mol Cell Bioi 14: 7134-
7143,1994
83. Jacobs H, Heldt HW, Klingenberg M: High activity of creatine kinase
in mitochondrial from muscle and brain and evidence for a separate
mitochondrial isoenzyme of creatine kinase. Biochem Biophys Res
CommunI6:516-521,1964
84. Hossle JP, Schlegel J, Wegmann G, Wyss M, Bohlen P, Eppenberger
HM, Wallimann T, Perriard J: Distinct tissue specific mitochondrial
creatine kinases from chicken brain and striated muscle with a conserved
CK framework. Biochem Biophys Res Commun 151: 408--416, 1988
85. Schlegel J, Wyss M, Schurch U, Schnyder T, Quest A, Wegmann G,
Eppenberger HM, Wallimann T: Mitochondrial creatine kinase from
cardiac muscle and brain are two distinct isoenzymes but both form
octameric molecules. J Bioi Chern 263: 16963--16969, 1988
86. Haas RC, Korenfeld C, Zhang Z, Perryman B, Roman 0, StraussAW:
Isolation and characterization of the gene and cDNA encoding human
mitochondrial creatine kinase. J Bioi Chern 264: 2890--2897, 1989
87. Muhlebach SM, WirzT, Brandle U, Perriard JC: Evolution of the creatine
kinases: The chicken acidic type mitochondrial creatine kinase gene
as the first nonmammalian gene. J Bioi Chern 271: 11920--11929,1996
88. Kaldis P, Stolz M, Wyss M, Zanolla E, Rothen-Rutishauser B, Vorherr T,
Wallimann T: Identification of two distinctly localized mitochondrial
creatine kinase isoenzymes in spermatozoa. J Cell Sci 109: 2079-2088, 1996
89. Steeghs K, Peters W, Bruckwilder M, Croes H, vanAlewiljk 0, Wieringa
B: Mouse ubiquitous mitochondrial creatine kinase: Gene organization
and consequences from inactivation in mouse embryonic stem cells.
DNA Cell Bioi 14: 539-553,1995
90. Stallings RL, Olson EN, Strauss AW, Thompson LH, Bachinski LL,
Siciliano MJ: Human creatine kinase genes on chromosome 15 and 19,
and proximity of the gene for the muscle form to the genes of
apolipoprotein C2 and excision repair. Am J Hum Genet 43: 144-151,
1988
91. Payne RM, Friedman DL, Grant JW, Perryman MB, StraussAW: Creatine
kinase isoenzymes are highly regulated during pregnancy in rat uterus
and placenta. Am J Physiol265: E624-E635, 1993
92. Payne RM, StraussAW: Developmental expression of sarcomeric and
ubiquitous mitochondrial creatine kinase is tissue-specific. Biochim
BiophysActa 1219: 33-38, 1994
93. Friedman DL, Perryman MB: Compartmentation of multiple forms of
creatine kinase in the distal nephron of the rat kidney. J Bioi Chern
266(33): 22404-22410, 1991
94. Billadello JJ, Kelly DP, Roman DG, StraussAW: The complete nucleotide
sequence of canine brain creatine kinase mRNA: Homology in the
coding and 3' noncoding regions among species. Biochem Biophys
Res Commun 138: 392-398, 1986
95. Papenbrock T, Wille W: The 3' non-coding region of the mouse brain
B creatine kinase mRNA: A sequence with exceptional homology among
species. Nucleic Acids Res 14: 8690, 1985
96. Konieczny SF, Emerson CP Jr: Complex region of the muscle-specific
contractile protein (troponin I) gene. Mol Cell Biol7: 3065-3075,1987
97. Smith DJ, Maggio ET, Kenyon GL: Simple alkanethiol groups for
temporary blocking of sulfhydryl groups of enzymes. Biochemistry
14: 766-771, 1975
98. Li K, Warner CK, Hodge JA, Minoshima S, Kudoh J, Fukuyama R,
Maekawa M, Shimizu Y, Shimizu N, Wallace DC: A human muscle
adenine nucleotide trans locator gene has four exons, is located on
chromosome 4, and is differentially expressed. J Bioi Chern 264(24):
13998-14004,1989

99. Lenka N, BasuA, MullickJ,Avadhani NG: The role of an E box binding
basic helix loop helix protein in the cardiac muscle-specific expression
of the rat cytochrome oxidase subunit VIII gene. J Bioi Chem 271(47):
30281-30289,1996
100. Weller PA, Price M, Isenberg H, EdwardsYH, JeffreysAJ: Myoglobin
expression: Early induction and subsequent modulation of myoglobin
and myoglobin mRNA during myogenesis. Mol Cell Bioi 6: 4539-4547,
1986
IO\. Rossant J: Mouse mutants and cardiac development. New molecular
insights into cardiogenesis. Circ Res 78: 349--353, 1996
102. Lints TJ, Parsons LM, Hartley L, Lyons I, Harvey RP: Nkx-2.5: A
novel murine homeobox gene expressed in early heart progenitor
cells and their myogenic descendants. Development 119: 419-431,
1993
103. Komuro I, Izumo S: Csx: A murine homeobox-containing gene
specifically expressed in the developing heart. Proc Natl Acad Sci
USA 90: 8145--8149, 1993
104. Cross JC, Flannery ML, Blanar MA, Steingrimsson E, Jenkins NA,
Copeland NG, RutterWJ, Werb Z: Hxtencodes a basic helix-loop-helix
transcription factor that regulates trophoblast cell development.
Development 121: 2513--2523, 1995
105. Cserjesi P, Brown D, Lyons GE, Olson EN: Expression of the novel
basic helixloop-helix gene eHAND in neural crest derivatives and
extraembryonic membranes during mouse development. Dev Bioi 170:
66~78, 1995
106. Hollenberg SM, Sternglanz R, Cheng PF, Weintraub H: Identification
ofa new family of tissue-specific basic helix-loop-helix proteins with a
two-hybrid system. Mol Cell Bioi 15: 3813-3822,1995
107. Srivastava D, Cserjesi P, Olson EN: A subclass of bHLH proteins
required for cardiac morphogenesis. Science 270: 1995-1999, 1995
167
108. Arceci RJ, KingAAJ, Simon MC, Orkin SH, Wilson DB: Mouse GATA-
4: A retinoic acid-inducible GATA-binding transcription factor
expressed in endodermally derived tissues and heart. Mol Cell BioI 13:
2235--2246, 1993
109. Tamura S, Wang XH, Maeda M, Futai M: Gastric DNA-binding proteins
recognize upstream sequence motifs of parietal cell-specific genes.
Proc Nat! Acad Sci USA 90: 10876-10880, 1993
110. Kelley C, Blumberg H, Zon U, Evans T: GATA-4 is a novel transcription
factor expressed in endocardium of the developing heart. Development
118: 817--827, 1993
Ill. Wu X, Golden K, Bodmer R: Heart development in Drosophila requires
the segment polarity gene wingless. Dev Bioi 169: 619--628, 1995
112. Ranganayakulu G, Schulz RA, Olson EN: Wingless signaling induces
nautilus expression in the ventral mesoderm oftheDrosophila embryo.
Dev Bioi 176: 143--148,1996
113. Bodmer R: The gene tinman is required for specification of the heart
and visceral muscles in Drosophila. Development 118: 719--729,1993
114. Azpiazu N, Frasch M: Tinman and Bagpipe: Two homeobox genes that
determine cell fates in the dorsal mesoderm of Drosophila. Genes Dev
7: 1325--1340, 1993
115. Lyons I, Parsons LM, Hartley L, Li R,Andrews TE, Robb L, Harvey RP:
Myogenic and morphogenetic defects in the heart tubes of murine
embryos lacking the homeo box gene Nkx2-5. Genes Dev 9: 1654-
1666,1995
116. Bessman SP, Carpenter CL: The creatine-creatine phosphate energy
shuttle. Annu Rev Biochem 54: 831-862, 1985
117. RossiAM, Eppenberger HM, Volpe P, Cotrufo R, Wallimann T: Muscle-
type MM creatine kinase is speci ficaJ1y bound to sarcoplasmic
reticulum and can transport Ca2+ uptake and regulate local ATP/ ADP
ratios. J BioI Chem 265: 5258--5266, 1990
Molecular and Cellular Biochemistry 184: 169-182, 1998.
1998 Kluwer Academic Publishers.

Adenylate kinase: Kinetic behavior in intact cells

indicates it is integral to multiple cellular processes
Petras P. Dzeja, Robert 1. Zeleznikar and Nelson D. Goldberg
Department of Biochemistry, University of Minnesota, Medical School, Minneapolis, Minnesota 55455, USA

Abstract
Monitoring the kinetic behavior of adenylate kinase (AK) and creatine kinase (CK} in intact cells by ISO-phosphoryl oxygen
exchange analysis has provided new perspectives from which to more fully define the involvement ofthese phosphotransferases
in cellular bioenergetics. A primary function attributable to bothAK and CK is their apparent capability to coupleATP utilization
with its generation by glycolytic and/or oxidative processes depending on cell metabolic status. This is evidenced by the
observation that the sum of the net AK- plus CK-catalyzed phosphoryl transfer is equivalent to about 95% of the total ATP
metabolic flux in non-contracting rat diaphragm; under basal conditions almost every newly generatedATP molecule appears
to be processed by one or the other of these phosphotransferases prior to its utilization. Although CK accounts for the transfer
of a majority of the ATP molecules generated/consumed in the basal state there is a progressive, apparently compensatory,
shift in phosphotransfer catalysis from the CK to the AK system with increasing muscle contraction or graded chemical
inhibition of CK activity. AK and CK appear therefore to provide similar and interrelated functions. Evidence that high energy
phosphoryl transfer in some cell types or metabolic states can also be provided by specific nucleoside mono- and diphosphate
kinases and by the phosphotransfer capability inherent to the glycolytic system has been obtained. Measurements by ISO_
exchange analyses of net AK- and CK-catalyzed phosphoryl transfer in conjunction with 31p NMR analyses of total
unidirectional phosphoryl flux show that each new energy-bearing molecule CK or AK generates subsequently undergoes
about 50 or more unidirectional CK-or AK-catalyzed phosphotransfers en route to an ATP consumption site in intact muscle.
This evidence of multiple enzyme catalyzed exchanges coincides with the mechanism of vectorial ligand conduction suggested
for accomplishing intracellular high energy phosphoryl transfer by the AK and CK systems. AK-catalyzed phosphotransfer
also appears to be integral to the transduction of metabolic signals influencing the operation of ion channels regulated by
adenine nucleotides such as ATP-inhibitable K+ channels in insulin secreting cells; transition from the ATP to ADP liganded
states closely coincides with the rateAK-catalyzes phosphotransfer transformingATP (+AMP) to (2)ADP. (Mol Cell Biochem
184: 169-182,1998)

Key words: adenylate kinase, creatine kinase, phosphoryl transfer, regulation of cell energy metabolism, ISO-phosphoryl oxygen
exchange, ligand conduction, K+ ATP channel, insulin secretion

Introduction generation to consumption sites and vice versa? Both of

Mechanisms for regulating muscle energy metabolism so
that the rate of ATP generation closely corresponds with the
rate of its utilization remain to be defined [1-4]. The problem
is at least 2 fold; first, how ATP-consuming processes
communicate with spatially separated ATP-generating sites
to signal for rapid regeneration of spent ATP and second,
how ATP and metabolites are efficiently transferred from
these processes occur by mechanisms that maintain virtually
constant cytosolic concentrations of ATP and its metabolic
products [4, 5]. Also unresolved is how numerous intracellular
components sensitive to ATP and/or ADP are regulated in
the face oflittle or no changes in cytosolic ATP and/or ADP
concentrations [6, 7]. The latter problem is underscored by
the fact that the effective concentration of ATP required to
elicit a change in the function of these components is often

Footnotes: IDzeja P, Walseth T, Goldberg N, unpublished observation. 'Dzeja P, Wieringa B, Goldberg N, unpublished observation.
Address for offprints: N.D. Goldberg, Department of Biochemistry, University of Minnesota, Medical School, Minneapolis, Minnesota 55455, USA
170
two or more orders of magnitude lower than the average
cytosolic concentration of adenine nucleotide effectors [7].
Several models that provide for intracellular signaling,
communication and/or regulation have been proposed [2, 3,
8, 9]. Most derive from classical in vitro Michaelis-Menten
kinetic properties of individual enzymes combined with
allosteric regulatory mechanisms relying on altered
cytosolic concentrations of metabolite effectors [2- 4].
Although simple diffusion has been tacitly assumed [8, 9]
to overcome spatial separation of effectors or substrates
from their targets or catalytically transforming enzymes
there are numerous arguments against it serving as a major
mechanism for accomplishing this [10-12]. It is not very
likely that circumstances in living cells equate with those
in reconstructed in vitro systems [13-15] (i.e. enzyme and
metabolite concentrations, enzyme-enzyme interactions,
multi component competition for substrates, substrate
cycling, etc.). Coupled enzyme catalyzed reactions in vivo,
for example, apparently do not conform to classical
Michaelis-Menten kinetics since increases in cellular
metabolic fluxes involving series of coupled enzyme-
catalyzed reactions are not obligatorily accompanied by
changes in the concentrations of intermediates [4, 5, 11].
Obviously, the boundaries for formulating models of
metabolic management and regulation are detennined, in great
measure, by the infonnation experimental technologies make
available. Until just recently this infonnation was limited to
analyses in cell-free systems or to static measurements in
tissues or cells at points in time when their viability was
tenninated.
The development of noninvasive procedures for monitoring
dynamic catalytic behavior of enzymes involved in high-
energy phosphoryl metabolism in the environment of the
intact cell cytosol elp NMR, 180-phosphoryl oxygen
exchange [5, 10, 11]) has provided different perspectives from
which to develop new concepts of metabolic regulation.
Among the new insights these procedures have provided is
that the dynamics of specific catalytic events have much
greater fidelity to cell responsivity than does the con-
centration of product a reaction may generate [5, 7, 16, 17].
The latter had been the primary if not the sole change that it
was technically possible to measure and the only change,
therefore, sought to be correlated with an elicited response
even if it required unphysiological experimental conditions
to achieve [1,2]. Another new perspective gained from these
advanced analytical technologies is a different biological
importance of phosphotransfer catalyzed by AK [5, 17]
which confirms and extends earlier speculations [18-21]
that this catalytic process like that promoted by CK is
integral to the management of cellular energy metabolism.
These measurements ofthe dynamic behavior of AK and of
CK catalysis in the intact cell environment [5, 12, 22]
indicate that the functional significance of these two
operationally distinct enzyme systems are similar, inter-
related and probably interdependent.
The information gleaned from the new analytical tech-
nologies monitoring intact cell metabolic dynamics
presents a strong case for a mechanism of transferring and
distributing high energy phosphoryls within the cell interior
which differs very markedly from previously adopted
diffusion-based mechanisms [5, 10, 11]. The suggested
mechanism, termed 'vectorial ligand conduction' [5, 23]
assures that the rate of ATP generation will coincide almost
precisely with its rate of utilization, that thennodynamically
inefficient concentration gradients are not required and that
virtually constant cellular concentrations of ATP and its
metabolites are maintained.
From the vantage point of observed intact cell metabolic
dynamics it has also become apparent that other phos-
photransfer reactions such as those catalyzed by specific
nucleoside monophosphate kinases other than AK and
probably nucleoside diphosphate kinases are also very likely
to provide high energy phosphoryl transfer capability. Other
results strongly suggest that the enzymes comprising the
glycolytic pathway may also be added to the list of alternative
cellular mechanisms that provide high energy phosphoryl
transfer capability. Experimental evidence has also been
obtained indicating that AK in addition to the other specific
nucleoside monophosphate kinases are integral to the
process of transducing metabolic signals into cellular
responses. This appears to be accomplished by their action
to promote transfonnation of regulatory adenine nucleotide
ligands from triphosphate to diphosphate species through the
rate they catalyze phosphoryl transfer. The latter is suggested
to depend on the frequency of nucleoside monophosphate
delivery from the signaling site.
The importance of adenylate kinase-catalyzed
phospho transfer in the transfer and distribution of
energy intracellularly
For most of the nearly 60 years since its discovery [24] AK-
catalyzed phosphotransfer has been thought to be essential in
adenine nucleotide de novo biosynthesis and in regenerating
ATP from AMP produced by biosynthetic reactions [25].
However, in relation to cell bioenergetics it has traditionally
been viewed as having a 2 fold 'housekeeping' role. One, for
rapid equilibration of cytosolic adenine nucleotides in
compliance with the 'energy charge' concept [26] which is
no longer thought to be of primary regulatory significance
[27]. And second, for 'salvaging' a molecule of ATP from
accumulatingADP (i.e. 2ADP ~ ATP + AMP) when energy-
providing metabolic processes fall short of keeping pace
with the rate ofATP consumption [1, 17]. The' ATP salvaging'
role is also of questionable physiological relevance because

the experimental conditions under which this has been
observed (i.e. anoxia-induced severe metabolic compromise,
metabolic poisons, etc. [1, 5]) represent metabolically
compromised conditions probably unrepresentative of the
physiological role AK may play in maintaining a homeostatic
state. For example, the often used indicator of accelerated
AK-catalyzed phosphotransfer was the increased rate of
AMP accumulation under the severely metabolically
compromised conditions alluded to above [26]. It is
commonly overlooked that homeostasis is characterized by
a remarkable constancy of the cellular concentration ofATP
and that of its metabolites in spite of a relatively rapid basal
rate of ATP metabolic turnover rate [5, 17]. This constancy
is maintained even whenATP metabolic flux is increased by
orders of magnitude [5]. If so, mechanisms of metabolic
regulation, operating independently of altered cellular adenine
nucleotide levels, may exist for precisely coordinating rates
of ATP regeneration with those of its consumption.
With the advent of the analytical procedure based on
monitoring ISO-phosphoryl oxygen exchanges that measure
net rates of phosphotransfer it became apparent that
increased rates of AK-catalyzed phosphoryl transfer occur
when specific demands are made of cellular energy meta-
bolism [17]. It also became obvious that these increased
rates of AK-catalyzed phosphotransfer are not manifest as
increases in cellular AMP concentration but rather as
increases in the rate of AK-catalyzed AMP generation
balanced by equally rapid rates of AMP phosphorylation to
(~180)ADP [5]. An example of this circumstance is shown
in Fig. 1 where it is clearly demonstrated that there are
proportional increases in the rate of AMP phosphorylation
corresponding to increasing frequency of muscle contraction
(e.g. rat diaphragm) while tissue concentrations ofATP, ADP
and AMP exhibit virtual constancy. This early observation
in seeking the metabolic importance of AK catalysis
provided a new guiding principle; that previously undetected
dynamics of specific enzyme catalyzed transformations may
be more closely related to mechanisms governing metabolic
and specialized cell functions than are altered concentrations
of metabolite effectors such as ATP, ADP and AMP [5, 7].
That AK may not playa mere 'housekeeping' role but one
more integral to the regulation of cellular bioenergetics is
not a new concept. This was the view advocated by Bessman
and his colleagues [18, 29] who suggested from their
experimental evidence that AK- and creatine kinase (CK)-
catalyzed phosphoryl transfer are interrelated and that the
former participates in high energy phosphoryl transfer at
both mitochondrial and myofibrillar ends of what they termed
a 'CK shuttle'. The view offered by Dzejaet al. [19] was that
mitochondrial and cytosolic AK isoforms comprise a high
energy phosphoryl transport system separate from that ofCK.
In the latter proposal the combined actions of mitochondrial
and cytosolic AK (i.e. AK2 and AK I, respectively) were
171
envisioned to provide a mechanism for transferring the two
high energy phosphoryls (i.e. ~ and y) in one molecule ofATP
from its cellular generation to utilization sites. This exclusive
property of AK catalysis, to provide for utilization of the ~
phosphoryl of ATP which is energetically equivalent to the y-
phosphoryl of ATP, doubles the energetic potential of a
molecule ofATP and is clearly in harmony with the parsimony
of cellular metabolism [28]. In addition, it underscores an
early recognized potential importance that AK-catalyzed
phosphoryl transfer may have in the management of cellular
energy metabolism [20, 28-31]. The feature of central
importance to both of these concepts, is that they envisioned
enzyme-catalyzed phosphoryl transfer as a mechanism for
either circumventing or overcoming the limitations that
diffusion-based mechanisms would pose as the sole or
major means of transferring adenine nucleotide high energy
phosphoryls. This point was emphasized experimentally by the
demonstration that AK can increase the rate and efficiency
of high energy phosphoryl transfer between two spatially
separated metabolic processes in a reconstituted system [19].
However, these concepts of the involvement ofAK in cellular
energy transfer and/or metabolism remained speculative for
lack of quantitative information about the kinetic behavior of
AK catalysis in the physiological environment of intact cells
especially with regard to metabolic circumstances when this
phosphotransferase activity may be called into play. Some of
this information was provided when the dynamic behavior AK
catalysis was monitored in intact cells by l80-phosphoryl
oxygen exchange analysis [17].
One of the early surprising results from this new analytical
approach was that the net catalytic activity of AK in intact
muscle cells appeared to be less than 111000 of the total
activity measured by conventional cell-free assay [17].
Initially, this was interpreted to be due to the limited
availability of ADP to AK in the intact cell compared to cell-
free assay conditions where nearly saturating concentrations
of ADP are available simultaneously and continuously to all
enzyme molecules in the assay medium. The major reason
for the seemingly slower rates observed in intact muscle by
ISO-phosphoryl oxygen exchange analysis compared to in
vitro measurements that subsequently became appreciated is
that the l80-based procedure measures only net phosphoryl
transfer catalyzed by AK [i.e. only the appearance of each
newly synthesized molecule of (~180)ADP and upon its
phosphorylation, (~180)ATP]. It will be pointed out later how
this measurement of net phosphoryl transfer by the l80-based
procedure can be used in conjunction with measurements by
3 lp NMR employing saturation transfer technology, which
determines total unidirectional enzyme-catalyzed phos-
phoryl flux, to develop unique mechanistic insights about how
phosphoryl transfer may be accomplished intracellularly.
Another previously unrecognized characteristic of AK-
catalyzed phosphoryl transfer in the natural environment of
172

100 100
! AK flux

I , r. . . . . ..
-g --...
....................... ,...........................; .......................... i .......................... ~ ......................
80 80 ~

gJ )( C
:2 :J
;:: 'E
,------I---:.'-:.i,-.. . - . . . . . . -, .".

-
0

-.
CI
60 60 U)
U Q. CIS
:J C
c CIE ...
:ii!

-
0
. i .

-1
CI
c ........... _.II!IP..t-......._."'.._.......~...............-. ......i.......................... ~..........................~ .......................
CI Q.
'2 0 40

- ; I 40 CIS Q
: -+--- -.------- :
CI
'tJ
oct
E
C ! : --- [ATP]
>.
c
CI
E

+. . . . . . . . . . ..
'tJ
oct ~c
-
-t
20 -r--j-t................. 20
~ _ _ ~ _ _ _ :_ - -~ [ADP]
i! I ~M~
o 0

control
o 1 2 3 4
Contractile frequency (Hz)

Fig. 1. The relationship of adenyl ate kinase velocity to frequency of muscle contraction and adenine nucleotide concentrations in intact rat diaphragm. The
AK flux is equivalent to adenylate kinase velocities assessed from the rates of appearance 180 in the ~-phosphoryls of ATP and ADP. The control values
represent the concentration of each adenine nucleotide after a 4 min preincubation and the O-Hz value is the concentration after an additional 4 min incubation
without stimulation. These results show that there is a virtual constancy of cellular adenine nucleotide concentrations even in the face of a very dynamic
adenine nucleotide flux that increases almost I O-fold with increasing frequency of muscle contraction. (Modified from ref. [5] with permission).

the intact muscle cell uncovered by the ISO-based tech- dispute since they clearly show the metabolism of ADP
nology is that this phosphotransfer activity is restricted to as well as other intermediates such as Pi to be highly
discrete, nonexchangeable, subcellular compartments (Fig. compartmentalized subcellularly [17, 22]. An additional
2). With increasing frequency of muscle contraction these enzyme-related property that may be peculiar to these
AK metabolically active compartments increase in size andl phosphotransferases that is not accounted for by these
or number (Fig. 2). This emphasizes the highly structured simplistic equilibrium constant-based estimates is the
environment of the cell cytosol and argues against diffusion dynamic physical association that would be predicted to
serving as a major transfer mechanism. occur between ADP and the abundant AK and CK enzyme
An additional point these results make underscoring the proteins that are envisioned to provide for the intracellular
power of the 180-exchange technology is that the percentage transfer of ADP by the suggested mechanism of ligand
of the total mass of ADP that is metabolically active in conduction [5, 12]. This mechanism would very likely
contracting skeletal muscle can be as high as 65% (e.g. at involve 'kinetic entrapment' [32] and distribution of a large
4 Hz). That is, at least 65% of the total cellular ADP arises percentage of the cytosolic ADP among reversible ADP-
from AK-catalyzed phosphotransfer since this percentage binding/consuming reactions which would continue to
appears as (~ISO)ADP at isotopic equilibrium. The 35% that function as the metabolically active ADP pool detected by
does not exhibit ~- (or fJ,-) phosphoryl labeling with 18 0 and the ISO-based analytical procedure (see section on Ligand
is, therefore, metabolically inactive with respect to AK- Conduction).
catalyzed phosphoryl transfer, was found to be represented An early observation consistent with AK involvement in
by the ADP tightly associated with muscle actin [17]. This the process of energy transfer was that the net rates of AK-
indicates that the metabolically active pool of ADP in catalyzed phosphoryl transfer increase in direct proportion to
skeletal muscle is in the range of 650 J.!M (i.e. total cellular increases in the frequency of muscle contractility (Fig. 1)
ADP was determined to be about 1 mM) [17]. This value is [17]. This provided evidence that a significant portion of
markedly higher than the 10--50 J.!M range for 'free' ADP contraction-generated ADP may be coupled through AK-
estimated indirectly from equilibrium constants of the CK catalyzed phosphotransfer to the synthesis of a new ATP
reaction [4, 11]. The latter estimates assume a homogeneous equivalent for the following reasons. First, with stimulated
cell cytosol which 180-phosphoryl oxygen exchange analyses contraction the net rate of AK-catalyzed phosphoryl transfer
 173

100 - 0 - OHz

Yi
0
....
QC)
90 I3-ADP - - - - 1 Hz
.c - - - 0 - 2Hz
i 80 --..tr- 4Hz non-metabolic ADP
"1:1
GI
70 ~ 35 % (actin bound)
!oJ

a." 60 ,.--------(r-----c:t
...
GI 65 % metabolic (2 and 4 Hz)
~ 50
GI
~

~ 40
....
Q
Q
30
fD
.! 20
ii
~ 10
c!: 25 - 30 % metabolic (0 Hz)

2 4 6 8 10 12 14 16 18 20 22 24 26

Minutes

Fig. 2. The metabolism of adenine nucleotides by adenylate kinase is highly compartmentalized In intact cells. Shown is the rate of appearance of"O-labeled
~-phosphoryls in ADP arising from adenylate kinase-catalyzed phosphoryl transfer in intact rat diaphragm muscle contracting at frequencies of 0-4 Hz. The
rate of adenylate kinase-catalyzed phosphotransfer increases with increasing contractile frequency and so does the fraction of the total cellular ADP that
undergoes lSO-labeling which is indicated by the plateau when isotopic equilibrium is achieved. A maximum of only 65% of the total cellular ADP becomes
labeled and this represents the fraction that is metabolically active with respect to adenyl ate kinase-catalyzed phosphotransfer. The 35% remaining unlabeled
is metabolically inactive ADP that was found to be bound to actin. (Modified from ref. [17] with permission).

increased incrementally from the equivalent of 2-3% of the
total ATP flux in the resting muscle to 23% of the ADP
specifically attributable to the contraction-related process
at 4 Hz [17]. The ADP representing the rate limiting or flux
generating element for AK catalysis in intact muscle (see
above), therefore, undoubtedly derived from a muscle
ATPase involved in the contractile process. Second, these
increased rates of ADP production were matched by equivalent
rates of AMP generation and AMP phosphorylation by (y-
18 0) ATP to produce (~ 18 0) ADP and (~ 18 0 )ATP without any
detectable change in the concentration of cellular ATP,
ADP or AMP signifying that steady state conditions were
maintained. This translates into the increased rates of
contraction-generated ADP triggering equally increased
rates of AK-catalyzed formation of AMP, AK-catalyzed
phosphorylation of AMP and correspondingly increased rates
of new (yt 80)ATP generation to phosphorylate this AMP.
It was subsequently determined that in fully oxygenated,
metabolically competent rat diaphragm muscle the newly
generated ATP serving as the reactant in the AK-catalyzed
phosphorylation of AMP most probably derives from
anaerobic glycolysis. This was concluded after observing
that there is a stoichiometry of equivalence between AK-
catalyzed phosphorylation of AMP and lactate production
over a greater than 20-fold range of enhanced rates of
compensatory ATP generation by anaerobic glycolysis when
oxidative ATP production is impaired by various strategies
in intact diaphragm muscle [5]. From this information a
model was proposed whereby AK-catalyzed phosphotransfer
functionally couplesATP generation by glycolysis withATP
utilization by specific cellular ATPases in intact muscle [5,
17]. These experimental results also suggested that whereas
AK appeared to be paired with the transfer of glycolytically
generatedATP, CK-catalyzed phosphotransfer was apparently
coupled to the transfer of oxidatively produced ATP. This
conclusion stemmed from the observation that impairment
of oxidative phosphorylation by oxygen deprivation dim-
inished CK catalysis while causing reciprocal, apparently
compensatory and equivalent increases in the rates of both
anaerobic glycolysis and AK-catalyzed phosphotransfer [5].
This pairing of AK catalysis with glycolytic ATP generation
was subsequently found to be too restrictive a view of the
plasticity of AK-catalyzed phosphotransfer which was later
found to catalyze the transfer of energy-rich phosphoryls also
deriving from ATP produced by oxidative phosphorylation
under other circumstances [12].
The interrelationship of AK- and CK-catalyzed
phosphoryl transfer
The mounting evidence from these results, that AK may
provide a cellular function integral to high energy phosphoryl
metabolism which is similar to and possibly interrelated with
174

the function attributed to CK catalysis was assessed more
directly by the following strategy. Net rates ofCK- andAK-
catalyzed phosphoryl transfer were quantified in relation to
total cellular ATP metabolic flux when CK activity in intact,
non contracting diaphragm muscle was progressively inhibited
by increasing concentrations of 1,2dinitrofluorobenzene
(DNFB) [12]. Graded inhibition by DNFB of up to 98% of
the intracellular CK activity resulted in a progressive shift
in phosphotransferase catalysis from the CK to the AK
system. Whereas CK- and AK-catalyzed phosphotransfer
accounted for the equivalent of 88 and 7%, respectively of
the totalATP metabolic rate in the non DNFB treated muscle,
with increasing DNFB concentration finally resulting in 98%
inhibition of CK activity, AK catalytic rate increased by a
magnitude that accounted for virtually the totalATP metabolic
flux (Fig. 3). Over the range ofCK-inhibitory concentrations
of DNFB used the total net phosphotransfer catalyzed by the
sum of the decreasing CK and increasing AK activities
continued to approximate the total ATP metabolic rate (Fig.
3). Furthermore, since oxidative phosphorylation continued
to provide a majority of the cellular ATP when CK was
inhibited by 98%, it became apparent that the AK system can
also transfer energy-rich phosphoryls deriving from ATP
produced by this oxidative process. These results clearly
indicate that AK and CK operate as interrelated phosphoryl
transfer systems through which a majority of newly generated
ATP is processed prior to its utilization [12].
On the other hand, the expression of CK in these muscle
cells was very likely intended to provide for specific cellular
function(s) and/or their efficient operation which AK may
not take over in toto. How phosphotransfer function is
shifted from the CK to the AK system is not presently
understood but it appears to involve the recruitment of the
activity of preexisting AK-enzyme protein. There is no
detectable increase in the total activity of AK assayable in
extracts of muscle obtained after DNFB-treatment. This
ostensibly rapid and efficient switch from CK- to AK-
catalyzed phosphotransfer is, however, consistent with the
coexistence of AK and CK in discrete subcellular locales
[31, 44]. This may indicate that when ADP from ATP
hydrolysis cannot be processed by CK and accumulates to a
small but criticallocaIized concentration favoring the Km-
requirement of AK for ADP, the metabolic processing of
this ADP shifts to the AK system.
The ligand conduction mechanism of high energy
phosphoryl transfer
Over the past several decades, simple or facilitated diffusion
mechanisms have been among the most prevalent to explain
how utilizable energy is transferred within the cell cytosol
[8, 9]. However, diffusional mechanisms of transferring
ATP and its mono- and diphosphate species have received

--e-- ATP turnover rate

--0- net CK flux
150
---.-. net AK flux
=
....
.~

Q
r...
c..
~ 100
8

-
50
Q
8
=
'-'

OT---~--or--~---.--~---r--~---
o 50 100 150 200
DNFB, j.J.M

Fig. 3. Progressive shift of cellular high energy phosphoryl transfer function from creatine kinase to adenylate kinase with acute inhibition of creatine kinase
by increasing concentrations of dinitrofluorobenzene (DNFB) in intact rat diaphragm muscle. The majority of the total cellular ATP metabolic turnover can be
accounted for by the sum of the net phosphoryl transfer catalyzed by creatine kinase plus adenylate kinase as the rate of the former progressively decreases
and that of the latter increases. When 98% of the creatine kinase becomes inhibited at 200 >tM DNFB phosphoryl transfer catalyzed by adenyl ate kinase
accounts for virtually all the cellular A TP metabolic flux. (Reproduced from ref. [12] with permission).

criticisms that characterize it as a kinetically and thermo-
dynamically inefficient process because it requires the
creation of relatively high concentration gradients [33- 35].
This circumstance would result in impairment of ATPase
activity by end product (Pi' ADP, W) inhibition [36], inability
to maintain the free energy of ATP hydrolysis above the
threshold level at energy consumption sites [27] and the
dissipation of energy during the transmission ofATP [19, 33].
Additionally, ATP accumulation in the narrow intermembrane
and intracristae mitochondrial spaces would interfere with
the export of ATP from the mitochondrial matrix by locking
the adenine nucleotide translocator into an ATP bound state
[37]. Excessive ATP accumulation at this site has also been
demonstrated to inhibit mitochondrial respiration through
an interaction with cytochrome c [38].
To date, no reliable information is available about the
diffusivity in the cell interior of ATP, its most commonly
produced metabolite, ADP or reactants involved in ADP
. metabolic transformation. Using 31p NMR technology [39,
40] attempts have been made to characterize the diffusion
process in living cells but unequivocal conclusions cannot
be drawn from any of the results. For example, distinguishing
between signals representing localized metabolite dynamics
and those that may reflect or contribute to through-going-
diffusion are not yet possible in the present state of
refinement of this technology [41].
The concept that CK-catalyzed phosphotransfer operates
as a phosphoryl 'shuttle' system [18] added an important new
dimension to developing an understanding of how high
energy phosphoryls may be transferred intracellularly. The
importance of enzyme-catalyzed phosphotransfer as a
process involved in energy transfer envisioned for CK was
subsequently suggested for phosphoryl transfer catalyzed by
AK [18, 19]. The phosphoryl shuttle concept was extended
and operationally redefined by the view that the near-
equilibrium phosphotransfer reactions catalyzed by AK and
CK probably occur as a series of sequential transfers
comprising intracellular flux transfer chains or 'circuits' [5,
10, 11]. These chains of enzyme-catalyzed phosphoryl
transfers have more recently been suggested [5, 11, 12] to
operate by a mechanism based on Mitchell's general
principle of 'vectorial ligand conduction' [23]. It predicts
that each molecule entering at one end of a chain of near
equilibrium reactions promotes sequential equilibrations
resulting in almost simultaneous release of an equivalent
molecule at the remote end [5, 12]. A mechanism similar in
principle had also been proposed for proton conduction
pathways (e.g. 'proton wires') traversing biological mem-
branes [42]. The ligand conduction mechanism contrasts with
the traditional diffusion theory and also with the 'metabolite
channelling' concept [43] by operating independently of
cytosolic ligand concentrations and without movement of a
specific ligand through the entire length ofthe pathway [5,12].
175
The experimental evidence supporting the operation of a
ligand conduction mechanism is several-fold. First, the
results from 180-phosphoryl oxygen exchange analysis in
intact muscle [5,12,17,22], summarized in the two
preceding sections, indicate that AK and CK are functionally
coupled to and, therefore, likely to be physically associated
with sites where ADP is generated from contraction-related
processes as well as sites where ATP regeneration from
ADP occurs. Immunocytochemical localization [44] of CK
and AK in association with actomyosin ATPase strengthens
this argument. Also shown by analysis of intact cell metabolic
behavior is that the sum of the net phosphotransfer rates
catalyzed by CK plus AK are nearly equivalent to the total
ATP metabolic flux in diaphragm muscle [12]. Combining
this information a conclusion that can be drawn is that
phosphotransfer catalyzed by AK and CK may represent a
mechanism functionally linking spatially separated sites of
ATP utilization and ATP generation .
That multiple AK- and CK-catalyzed phosphotransfer
reactions are very likely involved in linking these two
processes can be deduced from the experimental results that
we [5, 12] and others [10] have reported. The ISO-phosphoryl
oxygen exchange analytical procedure we use has an inherent
limitation which provides the unique advantage of detecting
only net phosphoryl flux [5,12]. In the case of AK-catalyzed
phosphotransfer, this translates into detecting only the
generation of each newly generated molecule of (PI80)ADP
[and after its phosphorylation, PI80)ATP]. In the case ofCK
catalysis it only detects each newly generated molecule of
C80)CrP. The 180-exchange procedure cannot detect any
subsequent enzyme-catalyzed transfers of the 180-labeled p-
phosphoryl of ADP or the 180-labeled phosphoryl of CrP
resulting from reversible enzyme-catalyzed exchanges with
(yI 80)ATP. In contrast, in situ measurements of the kinetics
of AK and CK catalysis by 31p NMR using saturation
transfer technology [2, 45--48] assesses total unidirectional
phosphoryl exchange rates catalyzed simultaneously by all
enzyme molecules of CK or AK [2, 45--48]. In the context
of the ligand conduction mechanism of phosphotransfer, the
total unidirectional phosphoryl flux (determined by 31p NMR
analysis) divided by the net phosphoryl flux (measured by
180-phosphoryl oxygen exchange analysis), provides an
estimate ofthe number of AK- or CK-catalyzed equilibration
reactions associated with transferring each new molecule of
(PI80)ADP or C80)CrP, respectively [12]. From the limited
experimentation in which both of these analyses have been
conducted on comparable skeletal muscles for CK catalysis
the number of CK-catalyzed equilibrating phosphotransfer
reactions accompanying the generation (and presumed
transfer) of each newly generated C80)CrP was provisionally
estimated to be in the range of 50 [12]. This is probably a
minimal estimate because 31p NMR measurements have
recently been judged to underestimate total phosphoryl flux
176
in intact tissues due to sensitivity limitations [49]. In-
formation from 31p NMR analyses about AK-catalyzed
phosphotransfer in situ is more limited and less quantitative
[47, 48]. It indicates that the number of equilibrating
phosphotransfers attributable to AK catalysis exceeds by
orders of magnitude [47,48] the range of net AK-catalyzed
phosphoryl transfers determined by ISO-phosphoryl oxygen
exchange in several muscle and non-musical types of cells
[7, 12, 17]. The interpretation of these marked differences
between total and net phosphoryl transfer is that a newly
generated molecule of CSO)CrP and probably (~ISO) ADP
undergo processing by multiple equilibrating phosphotransfer
reactions. This interpretation combined with the conclusion
that these two enzyme systems appear to be in close proximity
to ATP utilizing and to ATP generating sites and that these
two sites may be functionally coupled can be viewed as
signifying that multiple AK- and CK-catalyzed equilibration
reactions are integral to the coupling process, thus, to the
energy transfer function.
The latter view is supported by immunocytochemical
studies showing that both CK and AK are physically arranged
as linear arrays in close proximity to actomyosin ATPases
in skeletal muscle [44]. Additionally, it is well recognized
that there is an apparent over abundance of very active
cytosolic CK and AK enzyme protein to which ADP would
be expected to bind with relatively high affinity [11, 12]. This
ADP could also be envisioned to undergo rapid metabolic
processing by these phosphotransferases. The latter feature
of cellular CK and AK is fully compatible with the vectorial
ligand conduction mechanism but represents a major
drawback for a diffusion-based mechanism; ADP binding to
this abundance of CK and AK enzyme protein would be an
impediment to a diffusional mechanism. Along these same
lines, the' local' or 'spatial' buffering effects that have been
attributed to CK-catalyzed phosphotransfer [8, 10] would
be difficult if not impossible to achieve without local
disequilibria 'spreading' by way of the abundant CK activity
promoting the transfer function ascribed to it. It would
probably be more correct to view this function of phosphoryl
transfer as the extended utility of the apparently localized
'buffering' effect initiating the 'flux wave' of sequential CK-
(or AK-)catalyzed equilibration reactions. Another feature of
the proposed ligand conduction mechanism underscoring its
potential thermodynamic efficiency is its capability to operate
with minimal or no concentration gradients of reactants [5,
12]. This could explain why sought after changes in cellular
adenine nucleotide concentrations are most often not
observed even with marked increases in metabolic flux [1, 5].
The envisioned operation of the ligand conduction
mechanism is as follows. In the intact muscle cell each
molecule creating a disequilibrium at a flux-generating site
or proximal end would enter the chain of near equilibrium
reactions and promote sequential equilibrations resulting in
almost simultaneous release of an equivalent molecule at the
distal end (Fig. 4). During muscle contraction dynamic
enzyme-substrate complexes existing in the micro environ-
ment of a contraction-associated APT'ase responds to the
local displacement of equilibrium by 'pushing' neighboring
substrates to create a 'pressure' that is propagated as a 'flux
wave' through the network of near equilibrium phosphoryl
transfer reactions catalyzed by AK or CK. This equilibrating
wave would end when an equivalent of the energy-spent
metabolite or newly generated energy-bearing molecule is
released at a cellular locale where ATP generation or
utilization, respectively, occurs. An equal number of
phosphotransfer reactions would be expected to occur in
the reverse direction of these near-equilibrium, enzyme-
catalyzed reactions in order for steady state to be maintained.
Propagation of a wave of serial equilibrating reaction, of
the type described, in chemical and biological systems, has
been calculated to proceed at rates that are orders of
magnitude faster than the diffusion rate of reactants [50,
51]. Thus, at steady state, the time for accomplishing these
virtually, simultaneous phosphotransfers resulting in the
coincident transfer of the energy-related component
should be negligible.
The role of AK in the transduction of metabolic signals
There are numerous intracellular components the operational
status of which appears to be governed by ATP and/or ADP
liganding [52-54]. Binding of the nucleoside di- or tri-
phosphates produces a conformational change relayed to the
regulated protein component carrying out a specific cellular
function [55]. Among theseATP/ADP sensitive proteins are
the ATP-inhibitable potassium channels (K+ATP) which have
become a major focus of attention because of the central
role they are believed to play in insulin secretion [52, 54].
K+ AT? channels are also determinants of cardiac and neuronal
responses to altered cellular metabolic states [6]. In all
instances K+ ATP channel involvement appears to be integral
to transducing metabolic signals into membrane potential
changes that secondarily influence ion conductance by other,
voltage-dependent, channels (e.g. L-type Ca2+ channels).
Regulation of the K+ATP channel is manifest as shifts in the
probability of its open or closed state; when liganded with
ATP the frequency of the 'closed' state predominates (K+
conductance decreases and the membrane becomes de-
polarized) and in the ADP-liganded state the 'open state'
predominates and K+ conductance increases [6, 52]. The
mechanism by which transition from the ATP- to ADP-
liganded state is achieved remains to be elucidated; neither
the regulatory nor operational channel components exhibit
ATP phosphohydrolase activity. Another element of the
puzzle is that the cellular concentration of ATP (e.g. 3-5
 177

AMP signal for and effector of glycolysis and ox. ph .

.,IIIIE lllltlll.llllllullllllllllllllllllllllllltIlllIllllllllIllIIIIIIIIHUIIIIIIIIIIII'"IIIIIIIIIIIIIIIIIIIIIII'III'IIIIIIIIII

Glycolysis
(-) f
JP3tAMP ATP~",AMPfATP
and/or
Ox. Ph. .:c./ ;~;.~~. .~ ~
AAiR:
.!~ ,', ~

!
ADP AD ADP ADP AD
Glucose Energy-rich 13- and ,,(-phosphoryl transfer
IIttllllllllllllllllllllllllllllllllllllllllllll..IIIIII ..... IIIIIU.... IIIIIIIIIIII .. IlIlItIll.III ... ttlllllll~:iUI'.

Fig. 4. The originally suggested operation of the adenylate kinase-catalyzed high energy phosphoryl conduction system. This more basic version
of the scheme presented in ref. [5], shows how sequentially arranged molecules of adenylate kinase catalyzing rapid equilibration between
substrates can comprise a vectorial ligand conduction pathway for high energy phosphoryl transfer and signal communication between ATP
production sites and cellular ATPases. For detailed explanation see text.

mM) far exceeds the K j ATP value for channel closure (e.g.
15 11M [6, 7, 52]. This translates into a requirement of a
change in the intracellular ATP concentrations of almost two
orders of magnitude to favor channel closing and, depending
on the ATP concentration a relatively large alteration in the
level of ADP to induce channel opening. Since the con-
centrations of these nucleotides exhibit a remarkable
constancy and have not been shown to undergo changes that
correlate with altered states ofK+ ATP channel function, it was
not surprising when it was found that the transition from
closed to open state was closely related to a dynamic of
nucleotide metabolism which was identifiable by 18 0_
phosphoryl oxygen exchange analysis [7]. The most prom-
inent metabolic alteration associated with glucose-induced
insulin secretion (i.e. transition from the ADP- to ATP-
liganded state) was suppression in the rate of AK-catalyzed
phosphorylation ofAMP by ATP [7]. This suppression ofAK
catalysis occurred as graded decreases of up to 50% in the
rate of (~180)ADP and (~180)ATP appearance (Fig. 5); the
extent of this diminished rate of AK-catalyzed phospho-
transfer was inversely proportional to the magnitude of the
insulin secretory response and to the broad range of
increasing stimulatory glucose concentrations [7]. This
reduction in AK catalytic activity was demonstrated to
precede the glucose-induced insulin secretory response, to
depend on the internalization and metabolism of glucose
which are required characteristics of a process integral to
transduction of the glucose signal, and to occur prior to or
at the same time as closure of the K+ ATP channel. This
glucose-induced suppression of AK-catalyzed phosphoryl
transfer was also found to be independent of events such as
Ca2+ influx that are the consequence of and occur subsequent
to K+ ATP channel closure [7].
A major conclusion from these results is that the rate of
AMP 'delivery' to the locale of the channel should represent
a critical factor determining the rate of AK-catalyzed ATP
to ADP transformation. Assuming that this AMP derives
from a metabolic event, at a site remote from the channel,
generating ADP that is transformed by AK to AMP then
transferred by the AK-catalyzed phosphoryl transfer system
to the channel site, the signaling event can be traced to the
rate of ADP generation by some ATP consuming process.
Although there are reports [56] that a specific Na+/K+ ATPase
is inhibited by glucose in insulin secreting cells which could
be the sought after initiating metabolic signal that becomes
transduced it has been observed that overallATP consumption
increases with glucose stimulation of HIT cells [7, 57].
Alternatively, a key factor in diminishing the rate of AMP
'signal' generation could result from a shift in the system
that metabolically processes the ADP. In other words, a
shift from processing ADP by AK to either the CK or the
glycolytic system. Increased CK-catalyzed phosphoryl flux
[57] and especially increased glycolytic flux [57] have been
observed to accompany glucose-stimulation of insulin
secreting cells.
Even though the AMP signaling mechanism remains to be
defined these results identify AK-catalyzed phosphotransfer
as a likely determinant of the ATP to ADP transition affecting
channel function even in the environment of a high, relatively
constant intracellular concentration of ATP. This AK-
catalyzed transition has been speculated to involve adenine
nucleotides in the channel microenvironment or possibly
nucleotide species directly in association with the channel.
Mechanistically, suppression of AK catalytic rate would
decrease ATP transformation to ADP and thereby extend the
duration of the ATP-liganded or closed state of the channel
[6,7]. This apparent involvement of AK in the regulation of
K+ ATP channel function was likened to the role AK was
suggested to play in regulating glycolytic flux when coupling
glycolytic ATP generation with the rate of ATP utilization
[5]. Both K+ ATP channels and glycolytic flux can be directly
affected by ATP and ADP liganding and/or allosteric
interaction [5, 7]. The latter through specific glycolytic
enzymes (e.g. phophofructokinase and glyceraldehyde
178

700 100

-0
c 600
...
0.
~
0.
SOO 90
-
C(

cc.

- 400
Q
E
c 80 -",
o 8
300 8.
.=
~

QA
200 c-
:;)
100
70 2~
:::l. 8co
0. ...
0 60
0 2 3
mY gluco

700 100

c 600
0.
-0 90
Q
~

0.
500 =s.
cc.

-
CIt 400
E
.= 300
80 -'"
o~
~
.= --
8

QA
200

.. .
70 c
~
:::l. 100 30
8co
0. ...
0 60
0 2 3
mM gluco

120
C

c---.=.
Q 100

OA 80

-.
=0
z:.co
.= ....
0
60

_....
_0.
40
E<....
.cc.
M
E_
20

~o 0
0 2 3
mY gluco
Fig. 5. The reciprocal relationship between glucose-induced insulin secretion and decreased adenylate kinase-catalyzed phosphoryl transfer In intact HIT-
Tl5 cells. Panels A and B show the decreases in the rate of appearance l'O-labeled ~-phosphoryls in ATP and ADP, respectively compared to the magnitude
of the insulin secretory response as a function of increasing stimulatory glucose concentrations. Panel C shows the percentage of the maximal inhibition
(achieved at 2.8 mM glucose) of (~I'O) ATP appearance over the range of stimulatory glucose concentrations tested. (Reproduced from ref. [7] with
permission).

phosphate dehydrogenase) inhibitable and stimulable by ATP
and ADP, respectively [58).
Numerous circumstantial arguments can be made for AK-
catalyzed phosphoryl transfer serving as a mechanism for
effecting ATP to ADP regulatory transitions. For example,
a feature of AK-catalyzed phosphotransfer adding to its
attractiveness as a mechanism for ATP to ADP regulatory
transitions is that unlike several regulatory processes
involving ATP, phosphoryl transfer requires very low or no
expenditure of energy. Also, it provides for precise control
tantamount to a 'digital' type of regulation whereby each
molecule of AMP signals for a fixed (i.e. equivalent)
stoichiometry of ATP to ADP conversion.
This can assure a quantitative response as has been
demonstrated in the case of regulating glycolytic flux [5] and
could also be envisioned for the control of channel 'opening'
or 'closing'. It also has advantages over allosteric mechanisms
of regulation from the standpoint of the range or amplitude
of the regulated response; allosteric mechanisms would
require large changes in the cellular concentrations of
effector metabolites (e.g. ATP, AMP) incompatible with cell
viability, to effect, for example, a 20-100 fold increase in
glycolytic flux. The phosphoryl transfer mechanism requires
an increase in the rate of AMP 'delivery' by the AK phospho-
transfer system well within the range of its catalytic potential.
Additionally, to complement its theoretical thermodynamic
efficiency, this can be accomplished with minimal or no
detectable change in cellular adenine nucleotide concentra-
tions as already demonstrated [5, 12]. Another advantage is
that there is amplification potential if AMP were to serve
as the signaling molecule; a single AK catalytic cycle with
AMP as a reactant results not only in the removal of one
inhibitory molecule of ATP but also the generation of two
'activating' molecules of ADP.
Coinciding with the role we have suggested AK plays in
governing ATP/ ADP sensitive cell components is a more
recent awareness that AK is not distributed solely in the
soluble cell cytosol but that a significant fraction occurs in
association with the cell plasma membrane and intracellular
membranes [59, 60). AK activity is readily detected in the
isolated plasma membranes of HIT cells and islets I. Patch
clamp experimentation has also identified AK as a tightly
associated plasma membrane constituent and provided
support for AK involvement in the regulation of K+ ATP
channel function [61]. Using isolated inside-out patches of ~
cell membranes, AMP addition to medium on the intracellular
surface was shown to readily reverse ATP- induced channel
closure and promote transition to a greater 'open' state
probability [61]. This effect of AMP requires that AK occur
in close physical proximity to the channel or a functional
channel component in order to effect a relatively rapid
catalytic conversion of ATP to (2)ADP that would in-
fluence channel behavior. AK has also been implicated in
179
the regulation of large conductance Ca 2 +-activated K+
channels [62] and CFTR chloride channels [63] both of
which require ATP to open [52-55]. In excised inside-out
membrane patches from cells in which these channels occur
the addition of ADP produces the same effect on channel
behavior as added ATP. However, ADP is ineffective when
the activity of endogenous AK apparently associated with the
membrane patch is inhibited by diadenosine pentaphosphate
(ApsA) [62, 63]. AK involvement in governing the function
of an ATP sensitive cellular component was also indicated
by the demonstration [64] that a single mutation in the AK
gene is responsible for the impaired ability of glycine betaine
to produce an osmoprotectant effect in S. typhimurium. This
resulted from the loss of function in this bacteria of an ion
transporter [64] which is a member of a family of trans-
porters, occurring in both prokaryotes and eukaryotes, with
a consensus sequence for adenine nucleotide binding; they
have been designated the ABC family of proteins [53].
Noteworthy, is that the K+ ATP channel regulator, also known
as the sulfonylurea receptor, as well as the CFTR Cl-
conductance channels are all members of this ABC protein
family [53, 54]. The mechanism by which AK may be
involved in regulating the operation of these transporters/
channels remains to be elucidated.
Other phosphotransfer systems that appear to provide
high energy phosphoryl transfer capability
During the course of the investigations focused on AK and
CK it became increasingly apparent that enzyme catalyzed
phosphoryl transfer represented a process fundamental to
the transfer of ATP between its intracellular generation and
utilization sites. It also became apparent that the CK and AK
phosphotransfer systems may not be the sole providers of
this energy transfer function. The fundamental aspect of
enzyme-catalyzed phosphoryl transfer as opposed to a
diffusion-based mechanism for energy transfer was strikingly
emphasized when the acute inhibition of CK activity by
DNFB was found to be almost completely compensated for
by increased rates of AK-catalyzed phosphoryl transfer [12].
In other words, if diffusion accounted for a significant
amount of ATP transfer or distribution within the cell, this
would be difficult to reconcile with the compensatory increase
in the rate of another enzyme-catalyzed phosphotransfer
system that equalled the loss due to the impairment of the
system. However, the participation of additional or auxiliary
systems for this purpose was indicated in at least some other
cell types when it was found that the rates of other nucleoside
mono- and diphosphate kinases (NMK' s) precisely paralleled
the kinetic behavior of AK in HIT cells upon stimulation of
insulin release by glucose; the activities of guanylate kinase
(GK), uridylate kinase and cytidylate kinase all exhibited
180
decreases of almost 50% in their relatively rapid, basal
catalytic rates of phosphoryl transfer as did AK [7, 57]. It
appeared, therefore, that in this cell type every member of
the NMK family was participating in a process of phosphoryl
transfer similar or identical to that deduced for AK in the
transduction of the glucose metabolic signal. AK and CK did
not, therefore, appear to be exclusive systems for carrying
out the function of transferring energy-rich phosphoryls
within the cell cytosol.
Relevant to developing this thesis is the observation [65]
that the outcome of acute inhibition of CK, for example,
by treatment of muscle with DNFB is very different than
chronic inhibition resulting from deletion of genes encoding
for the two major muscle forms ofCK, M-CK and ScCKmit.
In the double deletion mutants the chronic loss of CK-
catalyzed phosphoryl transfer capability, which results in
nearly 'normal' appearing and functioning animals, AK-
catalyzed phosphotransfer increases no more than about 2-fold
and accounts for only a fraction of the net phosphotransfer
rate attributable to CK in wild type mice. This leaves about
70% of the totalATP metabolic flux unaccounted for by only
the net phosphoryl flux AK catalyzes [65]. This signifies that
either a significant fraction of cellular phosphoryl transfer
is accomplished by a diffusion type of mechanism in these
double gene deleted mutants or that other phosphotransfer
systems provide compensatory phosphotransfer capability.
Although this question has not been entirely resolved the
information emerging indicates that the latter is probably
closer to the truth than the former. One enzyme system that
appears to provide significant, but incomplete compensation,
at least in gastrocnemius muscle, is phosphoryl transfer
catalyzed by GK, the rate of which increases several fold in
the muscle from these mutants [65]. The identity of a
phosphotransfer system that fully compensates for the
absence of the deleted CK activity has not been firmly
established but evidence has been obtained that the major
compensatory phosphotransfer process in the chronically
CK-impaired muscle is glycolysis 2 Although traditionally
viewed as a pathway that primarily provides two carbon
fragments from the catabolism of hexoses for oxidative
metabolism and the production of two moles of ATP during
this process [58], the glycolytic pathway has the attributes
of a biochemical process that could also provide this
specialized but fundamentally important cellular function of
spatially directed energy-rich phosphoryl transfer. In
simplest terms, the y-phosphoryls from the two moles of
ATP used to phosphorylate glucose and fructose-6-phosphate
remain equally or more energetic as the y or ~ phosphoryls
of ATP as they appear in the form of the phosphoryls of
phosphorylated triose phosphates traversing the glycolytic
pathway. When these energy-rich phosphoryls finally appear
as the phosphoryl of phosphoenolpyruvate, they can be used
to phosphorylate ADP through the phosphotransfer reaction
catalyzed by pyruvate kinase. Attributing high energy
phosphoryl transfer capability to glycolysis would require a
spatial arrangement of the enzymes comprising this pathway
that would accomplish the transfer of the y-phosphoryl of a
newly generated molecule of ATP from the site of its
synthesis such as the mitochondria to a site where it is
required to support an energy-consuming process. Consistent
with this requirement is the localization of hexokinase (and
glucokinase when present) on the outer mitochondrial
membrane in several cell types [66]. Interestingly, the extent
of hexokinase association with the mitochondrial membrane
depends on hormonal state or physiological status [66].
We have observed an increase in glycolytic flux in CK-
double gene deleted mutants 2 If this represents a major
compensatory enzyme-catalyzed phosphoryl transfer mech-
anism will require considerably more experimental effort
to ascertain. If the phosphoryl function suggested for
glycolysis can be borne out by further experimentation it
would explain the very robust rate of glycolytic P/y-ATP
exchange observed by 3 I p NMR analysis [67, 68] and provide
an expanded perspective of the metabolic importance of the
glycolytic process.
Conclusion
The development of an analytical procedure based on
determining the rate of exchange of metabolic intermediate
phosphoryl oxygens with 18 0 from 180-water enriched
medium has provided quantitative information about the
kinetic behavior of the phosphotransfer reactions catalyzed
by AK and CK in the environment of intact muscle and
other types of cells. These in situ enzyme-catalyzed
phosphotransfer reactions have been assessed with respect
to total cellular ATP metabolic flux to determine if there is
a quantitative basis on which to characterize their metabolic
function. In resting skeletal muscle a strong case can be made
for AK- and CK-catalyzed phosphoryl transfer providing
the function of processing the vast majority of the ATP
generated prior to its consumption on the basis of an
approximately equivalent stoichiometry between the net
phosphoryl transfer by these two enzyme systems and ATP
metabolic turnover. High energy phosphoryl transfer by AK
and CK is suggested to be accomplished by a vectorial ligand
conduction mechanism. This mechanism predicts that a
disequilibrium created at the ATP-generating or ATP utilizing
end of a chain of AK or CK enzyme molecules initiates the
propagation of a wave of serial equilibrating phosphotransfer
reactions resulting in the very rapid liberation of an equivalent
energy-spent or newly generated energy-bearing molecule
at the remote end of the chain. That this proposed ligand
conduction mechanism operates in skeletal muscle is
strongly suggested by the marked difference between total

unidirectional enzyme-catalyzed phosphoryl flux determined
by lip NMR analyses and net phosphoryl flux assessed by
180-based analyses. This difference between total and net
phosphoryl transfer indicates that each newly generated
energy-bearing molecule undergoes 50 or more uni-
directional AK- or CK-catalyzed phosphotransfers en route
to an ATP consumption site. The results also show that the
function of these two phosphotransfer enzymes to transfer
energy-rich phosphoryls intracellularly is interrelated and
interdependent; an apparent compensatory, quantitative and
graded shift from CK- to AK-catalyzed phosphotransfer
occurs when muscle CK activity is progressively and acutely
inhibited by DNFB. On the other hand, chronic elimination
of muscle CK activity by genetic deletion is not entirely
compensated for by increased net rates of AK-catalyzed
phosphotransfer but apparently by other systems which have
the capability to provide this function. Other nucleoside
mono- and diphosphate kinases such as guanylate kinase and
the phosphotransfer function glycolytic metabolism can
provide, have provisionally been identified, to compensate
for the chronic loss of CK activity. Evidence has also been
obtained that AK-catalyzed phosphotransfer is integral to
the transduction of metabolic signals that influence the
operation of ion channels regulated by ATP-/ADP-liganding
such as ATP-inhibitable K+ channels in insulin secreting
cells. AK appears to playa role in regulating the rate of
the transition from the ATP- to ADP-liganded state in
correspondence with the rate AK catalyzes AMP generation
at a remote signaling site.
Acknowledgements
This work was supported by United States Public Health
Service Grant GM 28818, the Alsam Foundation, and the
American Heart Association Minnesota Affiliate.
References
I. Hochachka PW: Muscles as Molecular and Metabolic Machines. CRC
Press, London, 1994
2. FromAHL, Zimmer SO, Michurski SP, Mohanakrishnan P, Ulstad VK,
Thoma WJ, Ugurbil K: Regulation of the oxidative phosphorylation
rate in the intact cell. Biochemistry 29: 3731-3743,1990
3. Brown GC: Control of respiration and ATP synthesis in mammalian
mitochondria and cells. Biochem J 284: 1-13, 1992
4. Katz LA, Swain JA, Portman MA, Balaban RS: Relation between
phosphate metabolites and oxygen consumption of heart in vivo. Am
J Physiol256: H262-H274, 1989
5. Zeleznikar RJ, Dzeja PP, Goldberg NO: Adenylate kinase-catalyzed
phosphoryl transfer couples ATP utilization with its generation by
glycolysis in intact muscle. J Bioi Chern 270: 7311-7319, 1995
6. Nichols CG, Lederer WJ: The regulation of ATP-sensitive K+ channel
activity in intact and permeabilized rat ventricular myocytes. J Physiol
423: 91-110,1990
181
7. Olson LK, Schroeder W, Robertson RP, Goldberg NO, Walseth TF:
Suppression of adenylate kinase catalyzed phosphotransfer precedes
and is associated with glucose-induced insulin secretion in intact HIT-
Tl5 cells. J Bioi Chern 271: 16544-16552, 1996
8. Meyer RA, Sweeney HL, Kushmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol246: C365-C377, 1984
9. Kushmeric MJ: Skeletal muscle - a paradigm for testing principles of
bioenergetics. J Bioenerg Biomem 27: 555-569, 1995
10. Wallimann T, Wyss M, Brdiczka 0, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands: The
'phosphocreatine circuit' for cellular energy homeostasis. Biochem J
281:21-40,1992
II. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova YO, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration-a synthesis. Mol Cell Biochem 133/134: 155-192, 1994
12. Dzeja PP, Zeleznikar RJ, Goldberg NO: Suppression of creatine kinase-
catalyzed phosphotransfer results in increased phosphoryl transfer
by adenyl ate kinase in intact skeletal muscle. J Bioi Chern 271: 12847-
12851,1996
13. Welch GR, DeMoss JA: Enzyme organization in vivo: Thermodynamic-
kinetic perspectives. In: PA Sere, RW Estabrook (eds). Microenviron-
ments and Metabolic Compartmentation. Academic Press, New York,
1978, pp 323-344
14. Harold FM: Biochemical topology: from vectorial metabolism to
morphogenesis. Biosc Rep II: 347-385, 1991
15. Mathews CK: The cell-bag of enzymes or network of channels? J
Bacteriol175: 6377-6381, 1993
16. Dawis SM, Graeff RM, Heyman RA, Walseth TF, Goldberg NO:
Regulation of cyclic GMP metabolism in toad photoreceptors. Definition
ofthe metabolic events subserving photo excited and attenuated states.
J Bioi Chern 263: 8771-8785,1988
17. Zeleznikar RJ, Heyman RA, GraeffRM, Walseth TF, Dawis SM, Butz
EA, Goldberg NO: Evidence for compartmentalized adenylate kinase
catalysis serving a high energy phosphoryl transfer function in rat
skeletal muscle. JBiol Chern 265: 300-311,1990
18. Bessman SP, Carpenter CL: The creatine-creatine phosphate energy
shuttle. Annu Rev Biochem 54: 831-862, 1985
19. Dzeja P, KalvenasA, ToleikisA, PraskeviciusA: The effect of adenylate
kinase activity on the rate and efficiency of energy transport from
mitochondria to hexokinase. Biochem Int 10: 259-265, 1985
20. Schoff PK, Cheetham J, Lardy HA: Adenylate kinase activity in
ejaculated bovine sperm flagella. J Bioi Chern 264: 6086-6091, 1989
21. Gellerich FN: The role of adenylate kinase in dynamic compartmentation
of adenine nucleotides in the mitochondrial intermembrane space. FEBS
Lett 297: 55-58, 1992
22. Zeleznikar RJ, Goldberg NO: Kinetics and compartmentation of energy
metabolism in intact skeletal muscle determined from 18 0 labeling of
metabolite phosphoryls. J Bioi Chern 266: 15110-15119, 1991
23. Mitchell P: Foundations of vectorial metabolism and osmochemistry.
Biosci Rep II: 297-346, 1991
24. Colowick SP, Kalckar HM: The role of myokinase in transphos-
phorylations. I. The enzymatic phosphorylation ofhexoses by adenyl
pyrophosphate. J Bioi Chern 148: 117-126, 1943
25. Noda LH: Adenylate kinase. In: PO Boyer (ed). The Enzymes, 3rd ed,
vol8.Academic Press, New York, 1973, pp 279-305
26. Atkinson DE: The energy charge of the adenylate pool as a regulatory
parameter. Interaction with feedback modifiers. Biochemistry 7: 4030-
4034,1968
27. Kammermeier H: Meaning of energetic parameters. Basic Res Cardiol
88: 380-384,1993
28. Davies RE: On the mechanism of muscular contraction. Essays Biochem
1:29-55,1965
182
29. Criss WE: Relationship ofATP:AMP phosphotransferase isozymes to
tissue respiration. Arch Biochem Biophys 144: 138-142, 1971
30. Tanabe T, Yamada M, Noma T, Kajii T, Nakazawa A: Tissue-specific
and developmentally regulated expression of the genes encoding
adenyl ate kinase isoenzymes. J Biochem 113: 200-207,1993
31. Savabi F: Interaction of creatine kinase and adenylate kinase systems
in muscle cells. Mol Cell Biochem 133/134: 145--152, 1994
32. Verkhovsky MI Morgan JE, PuusteinA, Wikstrom M: Kinetic trapping
of oxygen in cell respiration. Nature 380: 268-270, 1996
33. Mainwood GW, Rakusan K:Amodel for intracellular energy transport.
Can J Physiol Pharmacol 60: 98-102, 1982
34. Jacobus WE: Theoretical support for the heart phosphocreatine energy
transport shuttle based on the intracellular diffusion limited mobility
ofADP. BiochemBiophys Res Commun 133: 1035--1041, 1985
35. van Deursen J, Heerschap A, Oerlemans F, Ruitenbeek W, Jap P, ter
Laak H, Wieringa B: Skeletal muscle of mice deficient in muscle creatine
kinase lack burst activity. Cell 74: 621--631,1993
36. Drew JS, Harwalkar VA, Stein LA: Product inhibition ofthe actomyosin
subfragment-I ATPase in skeletal, cardiac, and smooth muscle. Circ
Res71: 1067-1077, 1992
37. Bandlow W, Strobel G, Zoglowek C, Oechner U, Magdolen V: Yeast
adenyl ate kinase is active simultaneously in mitochondria and
cytoplasm and is required fornon-fermentative growth. Eur J Biochem
178: 451--457,1988
38. Craig DB, Wallace CJ: ATP binding to cytochrome c diminishes electron
flow in the mitochondrial respiratory pathway. Prot Sci 2: 966--976,
1993
39. Yoshizaki K, Watari H, Radda GH: Role of phosphocreatine in energy
transport in skeletal muscle of bullfrog studied by 3IP-NMR. Biochim
BiophysActa 1051: 144-150, 1990
40. Hubley MJ, Rosanske RC, Moerland TS: Diffusion coefficients ofATP
and creatine phosphate in isolated muscle: Pulsed gradient 31p NMR
of small biological samples. NMRin Biomed 8: 72-78,1995
41. Xu Z, Stebbins JF: Cation dynamics and diffusion in lithium ortho-
silicate: two- dimensionallithium-6 NMR. Science 270: 1332-1334, 1995
42. Nagle JF, Morowitz HJ: Molecular mechanisms for proton transport in
membranes. Proc Natl Acad Sci USA 75: 289-302, 1978
43. Sere P: Complexes of sequential metabolic enzymes. Ann Rev Biochem
56: 89-124, 1987
44. Wegmann G, Zanolla E, Eppenberger HM, Wallimann T: In situ
compartmentation of creatine kinase in intact sarcomeric muscle: The
actomyosin overlap zone as a molecular sieve. J Muscle Res Cell Moti!
13:420--435,1992
45. Zahler R, Bittl JA, Ingwall J:Analysis of compartment at ion ofATP in
skeletal and cardiac muscle using P nuclear magnetic resonance
saturation transfer. Biophys J 51: 883-893, 1987
46. Ingwal JS: Whole-organ enzymology of creatine kinase system in heart.
BiochemSocTrans 19: 1006--1010,1991
47. Gupta RK: Saturation transfer 31 P NMR studies of the intact human red
blood cell. Biochim Biophys Acta 586: 189-195, 1979
48. Neeman M, Rushkin E, Kaye AM, Degani H: 3IP_NMR studies of
phosphate transfer rates in T4 7D human breast cancer cells. Biochim
Biophys Acta 930: 179-192, 1987
49. Wallimann T: P-31-NMR-measured creatine reaction flux in muscle-a
caveat. J Muscle Res Cell Mot 17: 177-181, 1996
50. Goldbeter A, Nicolis G: An allosteric enzyme model with positive
feedback applied to glycolytic oscillations. Progr Theor Biol4: 65-160,
1976
51. Gray P, Scott SK: Chemical Oscillations and Instabilities. Non-linear
Chemical Kinetics. Clarendon Press, Oxford, 1990, pp 292-300
52. Ashcroft SJH,Ashcroft FM: Properties and functions ofATP-sensitive
K-channels. Cell Signal 2: 197-214, 1990
53. Higgins CF: The ABC of channel regulation. Cell 82: 693--696, 1995
54. Dukes 10, Philipson LH: K+ channels: Generating excitement in
pancreatic beta-cells. Diabetes 45: 845-853, 1996
55. Carson MR, Welsh MJ: Structural and functional similarities between
the nucleotide-binding domains of CFTR and GTP-binding proteins.
BiophysJ69:2443-2448,1995
56. Levin SR, Kasson BG, Driessen JF: Adenosine triphosphatases of rat
pancreatic islets: comparison with those of rat kidney. J Clin Invest 62:
692-701,1978
57. Olson LK: Dynamics of High Energy Nucleotide Phosphoryl
Metabolism in Intact HIT-T 15 Cells During Glucose-Induced Insulin
Secretion. Ph. D. Thesis, University of Minnesota, 1991, pp 97-98
58. Ottaway JH, Mowbray J: The role of compartmentation in the control
of glycolysis. CurrTop Cell Regul12: 107-208, 1977
59. Watanabe K, Fukumoto H, Isoi K: Intracellular localization of ATP:
AMP phosphotransferase in Escherchia coli. Biochem Biophys Res
CommunI34:527-531,1986
60. Nagy AK, Shuster TA, Delgado-Escueta AV: Rat brain synaptosomal
ATP:AMP-phosphotransferase activity. J Neurochem 53: 1166--1172,
1989
61. Larsson 0, Ammala C, Bokvist K, Fredholm B, Rorsman P: Stimulation
of the K ATP channel by ADP and diazoxide requires nucleotide
hydrolysis in mouse pancreatic ~-cells. J Physiol463: 349-365, 1993
62. Lee K, Rowe ICM,Ashford MW: Characterization of anATP-modulated
large conductance Ca 2+-activated channel present in rat cortical
neurones. J Physiol488: 319-337, 1995
63. Quinton PM, Reddy MM: Control of CFTR chloride conductance by
ATP levels through non-hydrolytic binding. Nature 360: 79-81, 1992
64. Gutierrez JA, Csonka LN: Isolation and characterization of adenyl ate
kinase (ADK) mutations in Salmonella typhimurium which block the
ability of glycine betaine to function as an osmoprotectant. J Bacteriol
177:390--400,1995
65. Dzeja P, Noronha L, Zeleznikar R, Wieringa B, Goldberg N: Creatine
phosphate turnover in skeletal muscles ofM-CK, ScCKmit and MCK/
ScCKmit deficient mice. FASEB J 10: LB 114, 1996
66. Lynch RM, Fogarty KE, Fay FS: Modulation of hexokinase association
with mitochondria analyzed with quantitative three-dimensional cofocal
microscopy. J Cell Bioi 112: 385--395, 1991
67. Kingsley-Hickman PB, Sako KY, Mohanakrishnan P, Robitaille PML,
From AHL, Foker JE, Ugurbil K: 31p NMR studies of ATP synthesis
and hydrolysis kinetics in the intact myocardium. Biochemistry 26:
7501-7510,1987
68. Portman MA: Measurement of unidirectional Pi ~ ATP flux in lamb
myocardium in vivo. Biochim Biophys Acta 1185: 221-227, 1994
Molecular and Cellular Biochemistry 184: 183-194, 1998.
1998 Kluwer Academic Publishers.

Cytoarchitectural and metabolic adaptations in

muscles with mitochondrial and cytosolic creatine
kinase deficiencies
Karen Steeghs, l Frank Oerlemans, l Arnold de Haan, 3 Arend
Heerschap,4 Lia Verdoodt,5 Martine de Bie,5 Wim Ruitenbeek,7 Ad
Benders,2 Carolina Jost,1 Jan van Deursen,1 Peter Tullson,6 Ronald
Terjung,6 Paul Jap, l Wim Jacob,5 Dirk Pette 8 and Be Wieringa 1
Departments of ICell Biology and Histology, 2Biochemistry, 4Diagnostic Radiology, 7Paediatrics, Faculty of Medical
Sciences, University of Nijmegen, PO Box 9101, 6500 HB, Nijmegen; 3Institute for Fundamental and Clinical Human
Movement Sciences, Faculty of Human Movement Sciences, Vrije University, Van der Boechorststraat 9, 1081 BT
Amsterdam, The Netherlands; 5Center for Electronmicroscopy, University of Antwerp (UIA), Universiteitsplein 1, B-
2610 Antwerpen, Belgium; 6Department of Physiology, State University of New York Health Science Center at
Syracuse, Syracuse University, Syracuse, New York 13210, USA; 8Department of Biology, University of Konstanz, PO
Box 55601M641, Konstanz D-7834, Germany

Abstract
We have blocked creatine kinase (CK) mediated phosphocreatine (PCr) ~ ATP transphosphorylation in mitochondria and
cytosol of skeletal muscle by knocking out the genes for the mitochondrial (ScCKmit) and the cytosolic (M-CK) CK isoforms
in mice. Animals which carry single or double mutations, if kept and tested under standard laboratory conditions, have
surprisingly mild changes in muscle physiology. Strenuous ex vivo conditions were necessary to reveal that MM-CK absence
in single and double mutants leads to a partial loss of tetanic force output. Single ScCKmit deficiency has no noticeable
effects but in combination the mutations cause slowing of the relaxation rate. Importantly, our studies revealed that there is
metabolic and cytoarchitectural adaptation to CK defects in energy metabolism. The effects involve mutation type-dependent
alterations in the levels of AMP, IMP, glycogen and phosphomonoesters, changes in activity of metabolic enzymes like AMP-
deaminase, alterations in mitochondrial volume and contractile protein (MHC isoform) profiles, and a hyperproliferation of
the terminal cysternae of the SR (in tubular aggregates). This suggests that there is a compensatory resiliency of loss-of-
function and redirection offlux distributions in the metabolic network for cellular energy in our mutants. (Mol Cell Biochem 184:
183-194,1998)
Key words: skeletal muscle mitochondria, creatine kinase, metabolic adaptation

Introduction + ATP, a nodal event in the network for high energy phos-
phoryl transfer in vertebrates. Although many concepts have
Creatine kinases (CK; EC 2.7.3.2)
form a small family of been formulated to explain CK's role in high energy-
mitochondrial and cytosolic isoenzymes which catalyse the phosphoryl homeostasis [1, 2], relatively little is known
reaction phosphocreatine (PCr) + ADP + H+ ~ creatine (Cr) about its involvement in maintaining the integrity of

Present address: J. van Deursen, Department of Genetics, St. Jude Children's Research Hospital, PO Box 318 Mephis, Tennessee 38101, USA
Addressfor offprints: B. Wieringa, Department of Cell Biology and Histology, Faculty of Medical Sciences, University ofNijmegen, PO Box 9101,6500 HB
Nijmegen, The Netherlands
184
metabolic-energy compartmentalization [3-5] and the
communication between intracellular sites of ATP con-
sumption and production. Depending on cell-type one can
find different CK isoforms in mitochondria (i.e. the dimeric
or octameric sarcomeric (Sc) or ubiquitous (Ub) mitCKs)
and cytosol (i.e. the homo and heterodimeric MM-, BM- or
BB-CKs) in all tissues with large fluctuations in energy
metabolism, such as muscle and CNS [2]. Individual CK
family members may appear at different stages and locations
during animal development but their functional activities
appear usually coordinately at different cellular locations.
It is generally assumed that CK' s help in keeping subcellular
levels of ATP and its hydrolysis products ADP, AMP, Pi and
H+ delicately balanced. In turn, these levels - or the ratios
thereof-may govern the dynamic kinetics of many ATPases
in different microenvironments [6-9] and may determine the
rate of glycolysis and oxidative phosphorylation (OXPHOS)
[10, 11] and the activity of ion-pumps ([2, 12, 13] and refs
therein). The precise flux distribution through the CK
reaction of high-energy phosphoryl metabolites in distinct
mitochondrial, cytosolic or nuclear compartments is
however very difficult to study as it is intertwined by high-
energy phosphoryl interconversion reactions catalyzed by
adenylate kinase (AdK) [14], nucleoside-mono and -di-
phosphate kinases, hexokinase or glycerol kinase. Individual
reaction steps in this elaborate and complex network might
well be facilitated by the subcellular clustering of these
kinases in multienzyme complexes [1], which may keep
metabolites from entering the cellular pool. Interestingly, the
allocation of CK's to distinct cellular sites is dynamically
influenced in response to variation in energy demand and
may be an important physiological mechanism(s) for
regulating the enzymes' catalytic properties [2, 15].
In order to obtain a better understanding of the subcellular
partitioning of components of the CKlPCr system, and to
unravel its role in the compartmentalization of energy
homeostasis in different cell types it will become of utmost
importance to apply experimental methods which preserve
the integrity of the cells delicate organization, and the
communication between specific cellular 'aggregulons' or
microcompartments. Fortunately, the rapid development of
techniques to manipulate gene expression in vivo, in cell's
and experimental animals, is providing us with powerful new
tools for these kind of studies; Normal or mutant genes of
interest can be either overexpressed, or their expression
can be directed to specific tissues using 'conventional'
transgenesis by micro-injecting DNA into the fertilized egg
[16]. In an alternative approach, gene targeting in mouse
embryonic stem (ES) cells creates the possibility to produce
mice carrying predesigned mutations in the germline [17].
This technique, mostly applied as a gene 'knock-out'
mutagenesis method, has already been used to generate
hundreds of new mouse lines [18] and is currently one of
the methods of choice for revealing unknown gene functions
in the context of the whole animal (see [19-23] for review).
We and others have applied the methodology for altering the
expression levels of the different CK isoenzymes in vivo.
Conventional transgenic techniques have been applied to
direct B-CK and UbCKmit isoenzyme expression to mouse
liver [24], as well as to induce ectopic B-CK expression in
striated muscle tissue [25-27]. Our group has generated
mice completely deficient in M-CK subunits and mice
expressing reduced levels ofM-CK, by gene targeting. The
biological consequences of these mutations have been
characterised [28, 29]. Likewise, animals lacking either the
ubiquitous mitochondrial CK subform (UbCKmit), or the
sarcomeric mitochondrial CK (ScCKmit) isoenzyme, with
suprisingly little phenotypic effects, have been generated [30,
31]. Here, we summarize old and new data regarding the
genotype-phenotype relationship in muscles of mice with
single or combined M-CK and ScCKmit deficiency, under
normal laboratory conditions. One particular intriguing
observation, the developmental adaptation of metabolic and
cytoarchitectural characteristics of muscle in response to
mutation-induced defects, is discussed most extensively.
Nodal links between the CK-circuit and muscle (cell)
physiology
In muscle, MM-CK mediated ATP production is mainly
coupled to the local activity ofSR- and plasma-bound Ca2+-
ATPases [6, 8, 9, 32], the Na+/K+-ATPase [33,34] and the
myosin ATPase involved in actin-myosin sliding during
contraction in the sarcomeric M-band [35-39]. The MM-CK
mediated reaction is also topologically coupled to the
glycolytic metabolic reaction cascade in the I-band thin-
filament region [35]. The ScCKmit reaction is thought to
be involved in the transphosphorylation, channelling and
transport of high-energy phosphoryl groups from mito-
chondria to the cytosol in muscle [40]. In addition, the
ScCKmit mediated reaction may fuel mitochondrial ATPases
(Ca2+ or K+-ATPases), in analogy to the cytosolic situation,
but this is still a rather hypothetical possibility.
CK-mutant mice
To assess the effects of CK absence on these different
functions in muscle we have generated mice lacking either
the mitochondrial CK (ScCKmit), the cytosolic CK (M-
CK) or a combination of both enzymes, by interbreeding
ScCKmit [-/-] [13,30,31] and M-CK [-/-] [28] deficient
mice in all possible combinations. By genotyping 215
offspring from intercrosses between double heterozygous
(i.e. ScCKmit [+/-]; M-CK [+/-]) animals eleven wild-types

and ten ScCKmit [-1-]; M-CK [-1-] double mutants were
identified. Contrary to expectation, these double deficient
mice (henceforth indicated as CK [-1-]) were not overtly
different from wild type, or single mutant animals, and bred
normally. Zymogram assays of cardiac and skeletal muscle
extracts of 6 week old CK mutants confirmed that the
anticipated CK profiles were present in the ScCKmit and M-
CK single and the CK double mutants muscles (Fig. 1). No
ectopic expression of any of the two other members of the
CK gene family, brain-type (B-CK) or ubiquitous mito-
chondrial CK (UbCKmit) was seen in any mutant. Only in
heart muscle we saw a slight increase in the level of BB-CK
isoenzyme but this can be explained by the lack of capture
of B-type subunits in MB-CK heterodimers as noted before
[28]. Residual total CK activity in heart and skeletal muscle
extracts was lowered to 0.3 and 2% of wild-type levels,
respectively, and can be entirely attributed to the BB-CK
and UbCKmit content, which in skeletal muscle is found
especially in capillary or vasculature endothelium and
satellite cells. Despite the fact that adenylate kinase (AdK)
activity was inhibited by the inclusion ofP 1,P5-( die adenine-
5')-pentaphosphate (ApSA) in our zymogram assays often a
weak band of AdK activity was seen in the extracts from
double knock-out animals. Although many explanations are
possible this may suggest that AdK activity is increased in CK
[-1-] mice, consistent with the idea that there is functional
overlap and interrelated coupling between the CK and AdK
circuits in the OXPHOS and glycolytic networks for storage
and transport of high energy phosphoryl groups [41].
Currently, in collaboration with Dr. N. Goldberg, we are
studying this phenomenon in more detail by the use of 18 0
incorporation into the ~ and y phosphoryls of ATP and the
1 2 3 4 5 6 7 8
-sc
J
-MM
- AK
- BM
- BB
Fig. 1. Zymogram gel analysis of skeletal (lanes 1-4) and cardiac (lanes 5-
8) muscle extracts from wild-type (+1+) (lanes I, 5), M-CK [-1-] (lanes 2, 6),
ScCKmit [-1-] (lanes 3, 7) and CK [-1-] double mutant (lanes 4,8) mouse.
Note the absence of the ScCKmit and MM-CK isoforms in the different
mutants and note also the weak signal of residual AdK activity (not fully
inhibited by Ap5A) in the CK [-1-] mutant.
185
phosphoryl groups of various other metabolites in the
structurally intact tissues.
Metabolite levels
Chemical methods were used to examine whether CK
deficiency had influenced the levels of energy-related
metabolites in muscles at rest [13, 30, 31]. Concentrations
of ATP in CK [-1-] muscles (18.1 2.6 ~M/g dry wt) were
somewhat lower than in wild-type and ScCKmit [-1-]
muscles (24.5 4.8 and 22.9 2.0 ~M/g, respectively),
but only marginally lower than in M-CK [-1-] muscles
(19.1 2.1 ~M/g). Another series of measurements showed
that AMP and IMP levels (0.51 0.08 or 0.66 0.1 0 ~M/g,
respectively) in the gastrocnemius-soleus-psoas complex of
M-CK [-1-] and CK [-1-] mutants were roughly 5-fold higher
than the levels in wild type and ScCKmit mice (0.11 0.03
or 0.15 0.04 ~M/g, for AMP or IMP, respectively). It is
important to note that any direct estimates on the primary
effects of CK absence on the degree of overproduction of
AMP and IMP may be unreliable, as adaptation in the primary
metabolism of these compounds, attributable to other
enzymes in the adenylate metabolic network, may involve
significant redirection of flux distribution (see below).
Most surprisingly, we observed that the concentration of
PCr in the medial gastrocnemius ofCK [-1-] animals was only
slightly lower than in wild-types (51.6 5.2 vs. 62.1 2.2
~moles/g dry wt). Levels of total Cr (Cr + PCr) were similar
in all muscles (with non normalised values ranging between
86.2-106.5 ~moles per g dry wt), indicating that Cr import
mechanisms per se are not affected. This unexpected
presence of PCr was confirmed by in vivo 31p nuclear
magnetic resonance (NMR) spectroscopy (Fig. 2). Fully
relaxed resting state spectra of CK [-1-] hindlimb muscles
showed a clear PCr peak, appearing at exactly the chemical
shift position expected for PCr (at 2.44 0.02 ppm from
the position ofyATP). The relative positions ofthe threeATP
resonances in double mutant skeletal muscle spectra were
similar to wild-type spectra. Relative peak areas (as % of
total) of energy metabolites in wildtype, ScCKmit [-1-] and
M-CK [-1-] single mutants were all similar [13], but CK
[-1-] mutant muscles showed a significantly lower PCr
signal (35.6% vs. 46.5-47.3%) and somewhat higher ATP
signals with respect to the total phosphate signal area.
Consequently, this results in a 30% lower PCrlATP ratio. We
think that NMR data give a better reflection of the actual
metabolite levels than the chemical method, because minor
energy metabolite changes can be induced by the freezing
procedure. In another report [42] we have provided evidence
that the PCr pool does not exchange with ATP and is
metabolically completely inert, even under conditions of
complete hypoxia. This leaves the question how and when
186
PCr
ATP
Pi
"'''1'''''''''1''''1 "'1''''1' i i 'iii *Ii , I ' 'I .... '
30 ZO 10... -10 -ZO
Fig. 2. 3IP_NMR recordings of phosphate metabolites in lower hind limb
musculature at rest of wild-type and CK [-/-] mice. Spectra in the top panel
(wild type) and the bottom panel (CK [-/-]) represent the average of 48 free
induction decays of 70 pulses and 5 sec repetition time, giving a time
resolution of 4 min per spectrum. Positions of signals from the phosphoryl
groups in PCr, the y, a- and ~- groups in ATP (from left to right), the
inorganic P;' and the phosphomonoesters (outermost left signal) can be
seen as separate peaks.
the PCr was formed in our mutants and whether it can
accumulate in a time-dependent fashion during development
or growth. At this moment it is even not clear if PCr was
imported via circulation or via gap-junctional contact with
cells from muscle-adjacent tissues, or formed by phosphoryl-
transfer-enzymes with low substrate specificity.
The peak area measurements also show that relative
levels of inorganic phosphate (P) and phospho-mono-
esters (PME) have increased significantly, about 2-3 fold,
in CK [-/-] muscles, more than in the muscles with single
CK mutations. The abnormal accumulation of phospho-
mono-esters points to a deregulation or adaptation of the flux
through the glycolytic pathway, which however does not
lead to significant accumulation of lactate. We come to
this conclusion because pH calculations from the res-
onance positions in the 31p spectra yielded similar values
for wild-type (7.29 0.07) and CK mutants muscles at
rest (7.17 0.05 for ScCKmit [-/-], 7.24 0.02 for M-
CK [-1-] and 7.21 0.05 for CK [-I-D. Currently, in order
to better understand the basis and origin of the phospho-
monoester accumulation we are using high resolution NMR
spectroscopy of metabolites in a perchloric acid extract of
clamp frozen muscle to identify the precise chemical origin
of the signals.
Earlier, we had reported that resting-state glycogen content
and the ability to consume this glycogen was about 60%
increased in M-CK muscle compared to wild-type [28] mainly
due to an increase in fast-type fibers. A similar increase was
found for CK [--1-] double mutants, but ScCKmit [--1-] animals
had near wild-type glycogen levels [13]. Taken together our
data suggest that CK absence causes a considerable re-
direction of flux distribution in various branches of primary
metabolism. This may lead to compensatory adaptation in the
steady-state levels of AMP, IMP and inorganic phosphate as
well as the levels of phosphomonoester carbohydrate-
metabolites. It should be emphasized again that all muscles
studied were from animals kept under normal housing
conditions. How the steady-state metabolite levels can fluctuate
during development, growth and activity-induction of the
different CK-mutants is subject for further study.
PhYSiology and cytoarchitectural adaptations
-3.
Details on the physiological consequences of CK ablation
have been published [13, 31, 42]. All observations relate to
the testing of muscle performance upon artificial electrical
stimulation, ex vivo. From our studies, as illustrated in Fig. 3
we had concluded that M-CK is the dominant governing
principle in the reduction of tetanic force output (i.e.
generation of burst activity; [13, 28D in skeletal muscle. If
combined with M-CK deficiency, but not singly, ScCKmit
deficiency leads to increase in relaxation time [42].
In a first in vivo survey we found no overt effects on
animal activity or behaviour, in keeping with the fact that a
number of biochemical parameters (in f.e. heart mito-
chondrial activity, oxygen consumption) were essentially
unchanged [13]. Apparently, muscle physiology is not
challenged to threshold levels under normal laboratory
housing conditions, but we might have missed subtle effects.
Therefore, to examine the basic physiological functions in our
mutants in somewhat more detail we started computerized
telemetric analyses [43,44] of heart beat rate, core tem-
perature and gross locomotor activity. Figure 4 shows the
typical recordings of one CK [--1-] double knock out and one
wild type animal during a continuous monitoring period of
2 days with successive light (14 h) and dark (10 h) intervals.
After averaging the measurements on three CK [--1-] and two
wild type animals there were no significant differences.
Average body temperature ranged from 36. 7C (dark period)
to 35. 9C (light period) in mutant and wild type animals, heart

120
110
-
100
~
o Consequences for mitochondria and ER membranes
90
Q)
o 80
~
o pensated by an adaptation in mitochondrial volume in type
U. 70
60
o 5 10 15 20
Contraction number
Fig. 3. Contractile characteristics of wild-type and CK-mutant medial
gastrocnemius muscles during high-intensity exercise. The isometric force
production profiles (as % of the initial force) in wild-type (open squares),
ScCKmit [-1-] (black squares), M-CK [-1-] (black circles) and CK [-1-]
double-mutants (open circles) in muscles are shown. The muscle complex
was stimulated 4 times per sec at 100 Hz with an intermittent stimulus duration
of 165 msec; the brief periods between contractions were sufficient for
complete relaxation. Note that the ScCKmit force profile overlaps with the
wild type profile. Recovery, as seen in the M-CK [-1-] profile is absent in
the CK [-1-] profile.
beat rates varied between average 469-521 (light-dark) in CK
[-/-] mice to average 513-561 (light-dark) beats/min in wild
types. Locomotor activity varied between 44 (dark) to 26
(light) for CK [-/-] and 57 (dark) to 27 (light) crossings/min
for wild type. Although these figures suggest that CK [-/-]
mice may be slightly less active during the dark period, the
differences are subtle and obviously many more recordings
on a large series of animals are necessary to make this
statistically significant. As is evident from Fig. 4 the fluctu-
ations in minima and maxima of body temperature, heart beat
rate and activity within individual dark or light intervals were
considerable. We therefore limited ourselves to obtaining a
qualitative impression because of several technical and
logistic reasons. One reason being that the current state of
sophistication of the equipment is still rather limited and in
combination with the' difficult-to-standardize surgical
transplantation-location' of the transmitter and electronic
leads can be the cause of considerable experimental variation.
Another - more interesting - variable can be introduced by
the genetics of the knockout procedure, as all our mutants and
controls have a mixed and variable background of 129/Sv x
C57BLl6 inbred genes. There is tight coupling of brain
function to all three variables measured (heart rate, temper-
ature and activity) and it is now becoming increasingly clear
that effects of genetic background cannot be ignored in these
studies [45]. Interestingly, both groups of mice show the
normal physiological response towards light and dark periods.
We thus may conclude that CK absence does not form a
187
serious obstacle to normal heart and skeletal muscle physio-
logical-functioning under laboratory conditions.
Previously, we had observed that M-CK absence is com-
2 (fast) fibers of the gastrocnemius-psoas-soleus (GPS)
complex of M-CK [-/-] mice, but that the mitochondrial
ultrastructure is otherwise fully normal [28, 29]. To see
whether other compensatory mechanisms were involved in
ameliorating the complete loss of the Cr-PCr shuttle
function in CK [-/-] animals we repeated the morphometric
inspection and examined the fiber ultrastructure of diaphragm,
heart, intercostal, gastrocnemius and soleus muscles.
Although we did not quantify the effect in absolute terms, the
enlarged intermyofibrillar mitochondrial volume (1.5-2 fold)
was clearly apparent in all muscles, except heart [13, 31].
Biochemical activity determination showed an increase from
353 76 to 470 42 mU/mg protein for COX and from 69
15 to 117 21 mU/mg protein for CS (both mitochondrial
marker enzymes) in 600 g whole-tissue extracts from hind-
limb. In contrast, COX and CS levels in heart (955 303 vs.
937 354 for COX and 236 57 vs. 263 86 for CS,
respectively) did not differ between wild type and CK [-/-]
mice, in keeping with the ultrastructural analyses. Mito-
chondria in CK [-/-] skeletal muscles were often packed in
rows and their sizes were highly variable, ranging from
extremely large (more than 5 11m in length) to very small.
Moreover, mitochondria-rich fibers of all CK [-/-] muscles,
including heart, diaphragm and large glycolytic skeletal
muscles, were distinctly different from other mutants and
wild type in that they contained large(r) numbers of lipid
droplets [13]. In semi-thin 1 11m sections, these lipid
droplets can often be seen as strings of beads. Electron
microscopic examination revealed the lipid droplets to be
located immediately adjacent to, and occasionally inside,
the intermyofibrillar mitochondria. Association to the
sub sarcolemmal mitochondrial population was only rarely
observed. Furthermore, mitochondria containing glycogen,
lipofuscin granules, and other lysosomal structures were
present in increased numbers in CK [-/-] muscles (data not
shown). These deposits are common to diseased muscle and
occur to a much lesser extent in normal muscle during ageing
[46]. Are these changes cause or consequence of an altered
mitochondrial physiology or is a cytoplasmic factor the
principal governing principle? To address these questions we
next examined whether the number of mitochondrial contact
sites had undergone changes in double CK mutants (Fig. 5A).
Contact sites are fusions between the inner and outer
mitochondrial membranes which are dynamically regulated
multi-subunit structures that assemble in coordinance with
188

10 15 20 25 30 40 45
5
35

38
r'
/:" \ ;t \
J

/1\~
:AJ \. 1/
, /\\f\\
. f\ , i\ 1\ " l r\\ II,
I.

\
\
I I
\ (\ I"~ 1, "."
./
) "
. \ I \\ II. /,/f\\'!\\(I,N I\} :'1

i
35
I' \ A, /' '''\ " \ \'.1\ : '., v
,/ J jJ \

34

~ 33
~0-----5~-----1~0------1~5----~----~~----~----~------~-----4~5---
20 : 25 30 35 40
~ 600

500 WILD TYPE

ij
400
300
~
.;; 200
I II ! ,.Il l

\UI U
':;3

!\ !\ I\:! 11,\ .]\

U 100 11 1\
< ,,/\I\i\/I /, ~

10 15 20 25 35 40
'C
'5
~
6
'WJI JIA.J \.1 \1\:1\\' iI ,/1 It ,~,~Ij'\~~iy)tM \\I~ J\~jl \~II\JMl \ iliA, \''rV~~\ \.1\. I

.
4
e
~
2
,,

~ 15 20 ~ 25 30 35 40 45
5 10

39
38
~
'""" 37'
,.oJ\. /~\A I' \
\ .. , ,,/ I,
e 36
I

~
~ 35
~
.. 34~----~----~------~----~--~~~--~~---T.~----~~--~~--
0 5 10 15 20 25 30 35 40 45
500

400

o
. 15 20 25 30 35 40 45
5 . 10
--

Fig. 4. Representative telemetric recordings of heart rate (top panels), body temperature (OC, middle panels) and locomotor activity (bottom panels) in one
single wild type and CK [-/-) animaL Intermittant periods with dark (10 h; black dots) and light (14 h) housing conditions are indicated. Note the (normal) drop
in body temperature and activity during the light period in both animals,

metabolic activity [14,47,48]. A direct function formitCK in
the assembly of these structures has been suggested by the
extensive studies of Wallimann and coworkers [40, IS]
showing that mitCK-octamers may physically bridge the gap
between the mitochondrial inner- and outer membrane. For
contact site surface density measurement we cytometrically
compared [49] skeletal and myocardial muscles of wild type
and CK [-1-] mutants. Suprisingly, even though different
metabolic pathways are used (as described above) we found
no significant difference in contact site densities (Ss)
between wt (0.37 O.OS) and CK [-1-] (0.36 0.02) animals
in femoral quadriceps muscle. The data for myocardial
muscle were Ss = 0.31 0.01 and Ss = 0.36 0.02 for the
wt and CK [-1-], respectively. The slight increase seen in CK
[-1-] mitochondria - ifreal- may originate from the fact that
CK [-1-] mice are more affected by emotional stress upon
handling, prior to anaesthetizing the animals. The larger
S.E.M. value for skeletal muscle is due to the less efficient
fixation . Our data indicate that the formation of contact sites
per se is not blocked by the absence of ScCKmit in our
mutants and we are currently investigating the dynamics of
contact formation under regimes where we stimulate or
inhibit OXHPOS, to examine whether CKmit has any role
at all in the dynamic process of contact site formation.
Another distinct feature of CK [-1- ] animals was the
omnipresence of conspicuous, darkly staining, elongated
Fig. SA. Electron micrograph ofa transverse section ofa heart muscle cell.
The right side of the surrounding membranes of the mitochondrion at the
top are obliquely cut. Contact sites are designated by arrows.
189
inclusions within the large fibers of gastrocnemius and
intercostal muscles (Fig. SB). Within these fibers, the
inclusions are seen at both subsarcolemmal and inter-
myofibrillar localizations. Ultrastructural examination of
various fibers revealed that the inclusions consist of closely
packed, longitudinally oriented clusters of membranes,
known as tubular aggregates (TAs). Besides the well-ordered
aggregates, unorganized, more dilated membrane structures
with varicosities containing electron-grey material were
also observed. TAs are a distinct pathological structure in
skeletal muscle consisting of aggregated terminal cisternae
or longitudinal components of the sarcoplasmic reticulum
(SR) [46]. They may be functionally equivalent to hypertrophy
of the SR terminal cisternae and have been shown to be
highly reactive with antibodies against Ca 2+-ATPases and
calsequestrin. The maximal calcium content ofTA-containing
fibers is increased [SO]. Remarkably, TA-like structures were
not found in single mutations and have never been reported
in creatine-depleted muscles, when rodents are fed with
creatine analogues [2]. In humans, the occurrence ofTAs has
been associated with conditions caused by mutations in
genes for the dihydropyridine receptor subunits, the
ryanodine receptor (Ca 2+-channels), the adult sodium
channel gene SCN4A [46, SI , S2] , and a variety of other
muscle diseases [S3] . In mice, TAs have been observed in
murine dystrophy heterozygotes [S4] and in congenic mice
Fig. 58. Cell-morphological abnormalities and the formation of tubular
aggregates in CK [-1-] muscles. Shown is an electron micrograph revealing
the different aspects of a section chosen from a d arkly toluidine-blue stained
area from CK-deficient gastrocnemius muscle. Note the hyperproliferation
of tightly packed, longitudinally oriented tubes of the SR (doe more detailed
analyses see [42]).
190
of the MRL +/+ substrain [55]. From study of the latter model
it is clear that appearance ofTAs is genetically predisposed,
and gender (i.e. hormone) and age related. In our double
mutants we observed that terminal cisternae ofthe SR, which
are located in the direct surroundings of, or connected with
tubular inclusions, were heavily dilated. This suggests that
the TAs in our CK [-/-] muscles are indeed derived from the
SR. The myofibrillar compartment adjacent to the tubular
aggregates showed no structural abnormalities, and triads in
those myofibrils also appeared normal. TAs are clearly a
manifestation of disease, and almost certainly a secondary
adaptation to the abnormal cell physiological responses
evoked by CK absence. Evidently, they develop already under
normal laboratory conditions, and it will be interesting to see
whether their formation interferes with normal development,
and in tum may affect muscle strength or even cause disuse
of muscle, especially if animals are kept under regimes for
higher activity, involving more strenuous conditions for
muscle labour.
Enzymatic adaptations
To explain the changes in contractile performance ofM-CK
[-/-] and CK [-/-] muscles one could postulate either direct
effects of CK absence or invoke (secondary) perturbations
in the metabolic network for production and use of high-
energy phosphoryls. This may involve changes in enzymatic
activities, concentrations, or both (see [56] for a review on
'metabolic network flexibility'). As analyses of metabolic
flux distributions are intrinsically difficult we performed
only a merely random survey of activities of nodal enzymes
in the network directly coupled to the CK reaction. Mutant
skeletal muscle fibers may temper decreases in the free
energy of ATP hydrolysis by the transphosphorylation of
free ADP to ATP and AMP through the action of adenylate
kinase. Due to the near-equilibrium nature of this reaction,
net forward flux can be sustained by increasing substrate (i.e.
free ADP) or decreasing product (free AMP) concentrations.
AMP formed by the adenyl ate kinase reaction can be de-
aminated by the enzyme AMP deaminase, a tetrameric
enzyme composed of identical 80 kD subunits that catalyzes
the non-equilibrium deamination of AMP to IMP and
ammonia. IMP is largely retained within the myocyte until
subsequent reamination to AMP via the purine nucleotide
cycle [57]. The AMP deaminase reaction is thought to be
controlled, at least in part, by AMP concentration, allosteric
modulators such as ADP and inorganic phosphate and ATP
turnover associated with muscle contraction [58]. As many
of these parameters are affected in CK-mutant muscles we
studied the activity and structure ofthe enzyme in our mouse
models. In hindlimb muscle composed of a mixture of fiber
types, AMP deaminase activity assayed under optimal in
vitro conditions [59] was decreased -60% in the M-CK
deficient versus wild-type muscle. When examined at low
AMP concentrations 0.2 mM), AMP deaminase from M-
CK deficient mice displays a marked increase in affinity
coupled with a decreased Vmax compared with the enzyme
from wild type muscle. This kinetic behaviour is character-
istic of negative cooperativity that can be exhibited by
oligomeric enzymes and is accompanied by changes in the
apparent subunit molecular weight determined with SDS-
PAGE and Western blots. In muscle from wild-type mice,
about 80% of the immunoreactive protein was found to have
an apparent molecular mass of 80 kD. In contrast, mixed
muscle from the M-CK deficient had less than half as much
80 kD AMP deaminase protein coupled with an accumulation
of smaller species with apparent molecular weights of 60
and 56 kD (data not shown). This loss of 80 kD AMP
deaminase is consistent with the decline in in vitro activity.
Interestingly, these changes in molecular weight and enzyme
activity are also found in ~-GPA treated rats [60] and may be
related to post-translational changes such as proteolysis or
covalent modifications. How the overall-purine metabolism
in functionally intact muscles, during rest and during intense
exercise, is influenced by the loss of CK isoforms and the
secondary changes inAMP deaminase structure and activity,
remains to be determined.
We postulated earlier that also at the ATP-consumption
sites in the metabolic network adaptive changes in the
profiles of enzyme-activities could be involved. Suprisingly,
however, biochemical assay of whole muscle extracts
showed that neither the activity nor the concentration of SR
Ca2+-ATPases was conspicuously different between mutants
and wild type animal. SR Ca2+-ATPase activity (expressed as
mU/mg protein; n = 3 animals each) was 76.5 9.5 in wild
type, 75.5 9.8 in M-CK [-/-] and 76.6 10.0 in CK [-/-]
mutants. SR Ca2+-ATPase content ranged between 94.3 11.0
and 95.2 11.8 pmollmg protein in these animals. We feel
that in addition to these measurements our studies should
include the analyses of SERCA- and PMCA-isoenzyme
profiles, to rule out that isoenzyme switches are involved.
This is a goal for the near future. As myosin-ATPase is the
other principal determinant of energy demand in working
muscle and alterations in myosin isoenzyme profiles could
result in shifts in intrinsic ATPase activity, shortening
velocity, and energetic economy we also studied the
myosin heavy chain isotype (MHC) distribution [61, 62].
Comparison of the MHC patterns in the representative
slow soleus muscle, intermediate type gastrocnemius and
diaphragm, and fast psoas and extensor digitorum longus
(EDL) muscle, showed that transitions in isomyosin heavy
chain types do not occur in any of these types of muscle in
ScCKmit [-/-] or, more strikingly, CK [-/-] mice. In contrast,
small but significant changes became apparent in fast muscles
from M-CK [-1-] mice. Profiles in M-CK [-1-] EDL, psoas and
 191

A B
_ "'Hc I "'HC Iia flIl!m "'HC lid ~ IofHC lib _ "'HC I a "'HC lIa flIl!m "'He lid ~ "'He lib
70 70

60 60

50 50

40 40

30 30

20 20

10 10

o o
Conlrol MCK[/) ScCKmIlI+1 CKI/)
Conllol MCK[/) ScCKmII[/) CK[/)

c o
_ "'HC I "'HC lIa flIl!m "'HC lid ~ IofHC lib 90 _ "'He I a "'He lIa flIl!m !.tHC lid ~ .. He lib
90
80
80
70
70
60
60
50
50
40
40
30
3'0
20
20
10 10

o L..-_'::= 01..---=
Conllol MCK[/) ScCKmll[/) CK(/) Conlrol MCK!/) ScCKmll!/1 CK!/I

Fig. 6. Diagrams of myosin heavy chain (MHC) distribution in extracts from several muscles of a wild-type, ScCKmit [-1-], M-CK [-1-], and CK
[-/-] mouse. Electrophoretic MHC separations of (A) diaphragm, (B) soleus, (C) gastrocnemius, and (D) psoas muscle extracts were densitometrically
scanned. Figures at the left indicate the levels of specific MHC isoforms as a percentage of total (100%). Every bar represents three gels from one
muscle. Note that the shift in the MHC-profile is most prominent in M-CK [-1-] animals.

gastrocnemius display a slight fast-to-slow shift, with 40-50%
increase in MHCIId and 20-25% decrease in the MHCIIb
content, compared to wild-type and ScCKmit [-/-] and CK [-
/-] mutants (Fig. 6). Changes in MHC in cardiac atria and
ventricles were not observed for any of the mutants. As
judged from the (rather small) magnitude of changes we expect
the involvement of the MHC-ATPase isotype to play only a
minor role in altering muscle performance, perhaps more so in
the relaxation and long-term performance of skeletal muscle,
aspects which indeed differ between CK [-1-] and MCK [-/-
] mutants. Obviously, under normal laboratory conditions
there is no stringent selection on these aspects of muscle
physiology.
At this point we would like to conclude that the deviation
from the situation in wt in our mutants can be most easily
explained as a direct effect of CK loss on (local) ATP. As
many Ca 2 +-ATPases operate close to thermodynamic
equilibrium, this may have immediate effects on [Ca2+]i levels
in muscle and evoke a cascade of secondary cellular events
[42]. In addition, theATP supply for the myosinATPase cycle
and actomyosin sliding may be somewhat perturbed in our
mutants. Importantly, pleiotrophic compensatory cytoarch-
itectural and enzymatic changes may obscure the phenotypic
consequences, and do so in a mutation-type and fiber-type
dependent manner. In any future studies of thresholds and
energy networks, particularly during development of our CK
mutants, we must therefore focus on obtaining a fully detailed
picture of this background. This is a challenging and lab-
ourious endeavour but the knowledge may ultimately help
us to discriminate between direct consequences of ablation
of CK functions and indirect effects which most likely
represent the fraction of phenotypic functions that cannot
be compensated by other genes and pathways.
192
Materials and methods
Generation and genotyping of CK deficient mice
Procedures involved in the construction of the different
targeting vectors, transfection of wild-type ES cells and the
generation ofM-CK [-!-] and ScCKmit [--1-] nullizygous mice
have been described in detail elsewhere [13, 28, 31, 63].
In vitro and in vivo measurement of CK isoenzyme and
mitochondrial enzyme activities and metabolite levels
Zymogram analyses and measurements of mitochondrial
indicator enzyme activities (cytochrome c oxidase and citrate
synthase) were done as described previously [13, 28, 31,
42]. Procedures for the extraction and concentration
measurements of creatine and adenosine metabolites have
been described [31]. In vivo 3 1P_NMR spectra of mouse hind
limb muscles at rest were recorded on an Oxford Instruments
magnet (4.3 T), equipped with a S.M.I.S. spectrometer, and
working at 73 MHz. Probe characteristics and experimental
conditions were essentially as described previously [28, 31,
64, 65]. pH values were calculated from the chemical shift
of the Pi signal as described previously [64, 65].
Physiological measurements of skeletal muscle function
Contractile characteristics and force measurements of medial
gastrocnemius muscles were performed as described in detail
elsewhere [66, 13, 31] and further assessed using three tetani
(duration 150 msec; stimulation frequency 100 Hz). Force
signals were digitized (1000 Hz) and analyzed for peak force,
time to peak force, and halftime of relaxation (time for force
to fall from half to a quarter at the end of stimulation [67]).
Telemetric monitoring of heart beat rate, body core tem-
perature and gross locomotor activity was performed with the
use of a telemetry system connected to a computer data
acquisition program (Data Sciences, st. Paul, MN, USA).
After intraperitoneal surgical implantation of a transmitter
(TAIOETA-F20) mice were allowed to recover for 21 days,
during which the mice regained their initial, pre-surgical
weight. Monitoring was performed during two successive days
with light/dark periods of 14110 h as described [43, 44].
Light microscopical, ultrastructural and biochemical
fiber typing
Cross sections of the gastrocnemius-plantaris-soleus
(GPS) muscle groups were stained according to standard
histochemical procedures as described [28]. Sample prep-
aration, pre-and post-fixation and Epon embedding was
exactly as described [28]. Ultrathin sections, double
contrasted with uranyl acetate, were examined in a Philips
electron microscope EM 301 or a JEOL TEM 1010. Details
of procedures used in the morphometric and qualitative
aspects of mitochondrial contacts sites will be published
elsewhere (Verdoodt et al. submitted). For analysis of
myosin heavy chain (MHC) composition, crude extracts
from individual muscles from a wild-type, ScCKmit [-!-],
M-CK [-!-], and CK [-!-] double mutant animals were prepared
and electrophoretic profiles on gradient polyacrylamide gels
were evaluated as described [68].
SR Ca 2+ -ATPase determinations
The procedures for determining activity and concentration
of SR Ca 2+-ATPase (measured by Ca 2+-dependent ATP
hydrolysis and steady-state phosphorylation) have been
described elsewhere [42].
Acknowledgements
This work was supported by a program grant from the Dutch
Organization for Scientific Research (Medical Sciences).
References
I. Bessmann SP, Carpenter LC: The creatine-creatine phosphate energy
shuttle. Annu Rev Biochem 54: 831-862, 1985
2. Wallimann T, Wyss M, Brdiczka 0, Nicolay K, Eppenberger HM:
Intracellular compartmentalization, structure and function of creatine
kinase isoenzymes in tissues with high and fluctuating energy demands:
The phosphocreatine circuit for cellular energy homeostasis. Biochem
J281: 21-40,1992
3. Miller OS, Horowitz SB: Intracellular compartmentalization of adenosine
triphosphate. JBiol Chem261: 13911-13915, 1986
4. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuznetsov AV:
Metabolic compartmentalization and substrate channelling in muscle
cells. Role of coupled creatine kinases in in vivo regulation of cellular
respiration - a synthesis. Mol Cell Biochem 1331134: 155-192, 1994
5. Han J-W, Thieleczek R, Varsanyi M, Heilmeyer LMG: Compartmentalized
ATP synthesis in skeletal muscle triads. Biochemistry 31: 377-384, 1992
6. RossiAM, Eppenberger HM, Volpe P, Cotrufo R, Wallimann T: Muscle-
type MM creatine kinase is specifically bound to sarcoplasmic
reticulum and can support Ca'+ uptake and regulate local ATP/ADP
ratios. J Bioi Chern 265: 5258-5266, 1990
7. Hardin C, Raeymakers L, Paul RJ: Comparison of endogenous and
exogenous sources of ATP in fuelling Ca uptake in smooth muscle
plasma membrane vesicles. J Gen Physiol99: 21-40,1992
8. Korge P, Campbell KB: LocalATPregeneration is important for sarcoplasmic
reticulum Ca' +pump function. Am J Physiol267: C357--C366, 1994
9. Minajeva A, Ventura-Clapier R, Veksler V: Ca'+ uptake by cardiac
sarcoplasmic reticulum ATPase in situ strongly depends on bound
creatine kinase. Pfiigers Arch Eur J Physiol432: 904-912, 1996

10. Erecinska M, Wilson D: Regulation of cellular energy metabolism. J
MembrBiol70: 1-14,1982
II. Balaban RS: Regulation of oxidative phosphorylation in the mammalian
cell. Am J Physiol258: C377-C389, 1990
12. Sistermans EA: Murine creatine kinase B: Cell-Type Specific Expression
and Biochemical Role. Ph.D. Thesis, University Nijmegen, 1996
13. Steeghs K: Consequences of Creatine Kinase Deficiencies in Mice.
Ph.D. Thesis, University ofNijmegen, 1995
14. Brdiczka D, Wallimann T: The importance of the outer mitochondrial
compartment in regulation of energy metabolism. Mol Cell Biochem
133/134: 69-83, 1994
15. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of
mitochondrial creatine kinase. Nature 381: 341-345, 1996
16. Jaenisch R: Transgenic animals. Science 240: 1468-1474, 1988
17. Thomas KR, Capecchi MR: Site-directed mutagenesis by gene targeting
in mouse embryo-derived stem cells. Cell 51 : 503-512, 1987
18. Brandon EP, Idzerda RL, McKnight GS: Targeting the mouse genome:
A compendium of knockouts. CUff Bioi 5: 625, 758 and 873,1995
19. Capecchi MR: The new mouse transgenetics: Altering the genome by
gene targeting. Trends Genet 5: 70-76, 1989
20. Evans MJ: Potential for genetic manipulation of mammals. Mol Bioi
Med6:557-565,1989
21. Joyner AL: Gene targeting and gene trap screens using embryonic
stem cells: New approaches to mammalian development. Bioessays 13:
649-656, 1991
22. Rossant J: Gene disruption in mammals. CUff Opin Genet Dev I: 236-
240,1991
23. Smithies 0: Animal models of human genetic diseases. Trends Genet
9: 112-117, 1993
24. Satoh S, TanakaA, Hatano E, Inomoto T, Iwata S, Kitai T, Shinohara H,
Tsunekawa S, Chance B, Yamaoka Y: Energy metabolism and re-
generation in transgenic mouse liver expressing creatine kinase after
major hepatectomy. Gastroenterology 110: 1166-1174, 1996
25. Koretsky AP: Insights into cellular energy metabolism from transgenic
mice. Physiol Rev 75: 667--688,1995
26. Brosnan MJ, Chen L, Van Dyke TA, Koretsky AP: FreeADP levels in
transgenic mouse liver expressing creatine kinase: Effects of enzyme
activity, phosphagen type, and substrate concentration. J Bioi Chem
265:20849-20855,1990
27. Brosnan MJ, Raman SP, Chen L, Koretsky AP: Altering creatine kinase
isoenzymes in transgenic mouse muscle by overexpression of the B
subunit. AmJPhysiol264: CI51-C160, 1993
28. Van Deursen J, Heerschap A, Oerlemans F, Ruitenbeek W, Jap P, ter
Laak H, Wieringa B: Skeletal muscles of mice deficient in muscle creatine
kinase lack burst activity. Cell 74: 621--631, 1993
29. Van Deursen J, Ruitenbeek W, HeerschapA, Jap P, ter Laak H, Wieringa
B: Creatine kinase in skeletal muscle energy metabolism: a study of
mouse mutants with graded reduction in M-CK expression. Proc Natl
Acad Sci USA 91: 9091-9095,1994
30. Steeghs K, Peters W, Croes H, Bruckwilder M, vanAlewijck D, Wieringa
B: Mouse ubiquitous mitochondrial creatine kinase: Gene organization
and consequences from inactivation in mouse embryonic stem cells.
ONACeli Bioi 14: 539-553, 1995
31. Steeghs K, HeerschapA, de HaanA, RuitenbeekW, Oerlemans F, van
Deursen J, Perryman B, Pette D, Bruckwilder M, Koudijs J, Wieringa B:
Use of gene targeting for compromising energy homeostasis in
neuromuscular tissues: The role for sarcomeric mitochondrial creatine
kinase. J Neurosci Meth 71: 29-41, 1997
32. Levitsky DO, Levchenko TS, Saks VA, Sharov VG, Smirnov VN: The role
of creatine phosphokinase in supplying energy for the calcium pump
system of heart sarcoplasmic reticulum. Membr Biochem 2: 81-96, 1978
33. SaksVA, Lipina NV, SharovVG, SmirnovVA, Chazov EI, Grosse R:
The localization ofMM-isoenzyme of creatine kinase on the surface
membrane of myocardial cells and its functional coupling to
193
ouabain-inhibited Na+/K+ ATPase. Biochim BiophysActa 465: 550-
558,1977
34. Grosse R, Spitzer E, Kupriyanov VV, Saks VA, Repke KRH: Coordinate
interplay between (Na+/K+)-ATPase and CK optimizes (Na+/K+)-antiport
across the membrane of vesicles formed from the plasma membrane of
cardiac muscle cells. Biochim Biophys Acta 603: 142-156, 1980
35. Wegmann G, Zanolla E, Eppenberger HM, Wallimann T: In situ
compartmentalization of creatine kinase in intact sarcomeric muscle:
The acto-myosin overlap zone as a molecular sieve. J Muscle Res Cell
MotiI13:42~35, 1992
36. Turner DC, Wallimann T, Eppenberger HM: A protein that binds
specifically to the M-line of skeletal muscle is identified as the muscle
form of creatine kinase. Proc Natl Acad Sci USA 70: 702-705, 1973
37. Ventura-Clapier R, Saks VA, Vassort G, Lauer C, Elizarova GV: Reversible
MM-creatine kinase binding to cardiac muscle. Am J Physiol253: C444-
C455,1987
38. Ventura-Clapier R, Veksler V, Hoerter JA: Myofibrillar creatine kinase
and cardiac contraction. Mol Cell Biochem 1331134: 125-144, 1994
39. Sata M, Ugiura S, Yamashita H, Momomura S-i, Serizawa T: Coupling
between myosin ATPase cycle and creatine kinase cycle facilitates
cardiac actomysin sliding in vitro. Circulation 93: 310-317, 1996
40. Wyss M, Smeitink J, Wevers R, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119-166,1992
41. Zeleznikar RJ, Dzeja PP, Goldberg ND: Adenylate kinase-catalyzed
phosphoryl transfer couples ATP utilization with its generation by
glycolysis in intact muscle. J Bioi Chern 270: 7311-7319, 1995
42. Steeghs K, Benders A, Oerlemans F, de Haan A, Heerschap A,
Ruitenbeek W, Jost C, van Deursen J, Perryman B, Pette D, Bruckwilder
M, Koudijs J, Jap P, Veerkamp J, Wieringa B:Altered Ca'+ responses in
muscles with combined mitochondrial and cytosolic creatine kinase
deficiencies. Cell 89: 93-103,1997
43. Kramer K, vanAcker SABE, Voss H-P, Grimbergen JA, van der Vijgh
W JF, Bast A: Use of telemetry to record electrocardiogram and heart
rate in freely moving mice. J PharmacolToxicol Meth 30: 209-215, 1993
44. Wan R, Diamant M, de Jong W, De Wied D: Changes in heart rate and
body temperature during passive avoidance behaviour in rats. Physiol
Behavior47: 493--499,1990
45. Gerlai R: Gene-targeting studies of mammalian behavior: Is it the
mutation or the background genotype? TINS 19: 177-190,1996
46. Dubowitz V: Muscle Biopsy: A Practical Approach. Lavenham Press,
Lavenham, England, 1985
47. Biermans W, Bakker A, Jacob W: Contact sites between inner and
outer mitochondrial membrane: A dynamic micro compartment for
creatine kinase. Biochim Biophys Acta 1018: 225-228, 1990
48. Bakker A, Bernaert I, De Bic M, Ravingerova T, ZiegeIhOffer A, Van
Belle H, Jacob W: The effect of calcium on the mitochondrial contact
sites: A study on isolated rat hearts. Biochim Biophys Acta 1224: 583-
588,1994
49. Baddeley AJ, Gundersen HJG, Cruz-Orive LM: Estimation of surface
area from vertical sections. J Microscopy 142: 259-276, 1986
50. Salviati G, Pierobon-Bormioli S, Betto R, Damiani E, Angelini C, Ringel
SP, Salvatori S, MargrethA: Tubular aggregates: Sarcoplasmic reticulum
origin, calcium storage ability, and functional implications. Muscle
and Nerve 8: 299-306, 1985
51. Hoffman EP, Lehmann-Horn F, Rudel R: Overexcited or inactive: Ion
channels in muscle disease. Cell 80: 681--686, 1995
52. Meijer AEFH: Histochemical features of tubular aggregates in diseased
human skeletal muscle fibers. J Neurol Sciences 86: 73-82, 1988
53. Rosenberg NL, Neville HE, Ringel SP: Tubular aggregates, their
association with neuromuscular diseases, including the syndrome of
myalgias/cramps. Arch Neurol42: 973-976, 1985
54. Craig 1D, Allen IV: Tubular aggregates in murine dystrophy hetero-
zygotes. Muscle Nerve 3: 134-140, 1980
194
55. Kuncl RW, PestronkA, Lane J, Alexander E: The MRL +/+ mouse: A
new model of tubular aggregates which are gender- and age related.
Acta Neuropathol78: 615-620, 1989
56. Stephanopoulos G, Vallino JJ: Network rigidity and metabolic engineer-
ing in metabolite overproduction. Science 252: 1675-1681, 1991
57. Tullson PC, Arabadjis PG, Rundell KW, Terjung RL: IMP reamination
in rat skeletal muscle fiber types. Am J Physiol270: C 1069-C1074 (Cell
Physiol. 39), 1996
58. Sahlin K, Gorski J, Edstrom L: Influence ofATP turnover and metabolite
changes on IMP formation and glycolysis in rat skeletal muscle. Am J
Physiol259: C409-C412 (3 Pt I), 1990
59. Tullson PC, Wieringa B, Terjung RL: Skeletal muscle creatine kinase
deficiency decreases AMP deaminase activity and apparent molecular
weight. Med Sci Sports Exer 28: S76, 1996
60. Tullson PC, Rundell KW, Sabina RL, Terjung RL: Treatment with the
creatine analog p-guanidinoproprionic acid alters skeletal muscle AMP
deaminase activity in vitro and in vivo. Am J Physiol 270: C76--C85
(Cell Physiol), 1996
61. Sweeney HL, Kushmerick MJ, Mabuchi K, Sreter FA, Gergely J: Myosin
alkali light chain and heavy chain variations correlate with altered
shortening velocity of isolated skeletal muscle fibers. J Bioi Chem 263:
9034--9039, 1988
62. Pette D, Vrbova G:Adaptation of mammalian skeletal muscle fibers to
chronic electrical stimulation. Rev Physiol Biochem Pharmacol120:
115-202,1992
63. Van Deursen J, Wieringa B: Targeting ofthe creatine kinase M gene in
embryonic stem cells using isogenic and nonisogenic vectors. Nucl
Acids Res 20: 3815-3820,1992
64. HeerschapA, BergmanAH, Vaals JJ, Wirtz P, Loermans HMT, Veerkamp
JH: Alterations in relative phosphocreatine concentrations in pre-
clinical mouse muscular dystrophy revealed by in vivo NMR. NMR
Biomed I: 27-31, 1988
65. Heerschap A, van Vaals n, Bergman AH, Den Boer JH, van Gerwen
PHJ, Gravenmade EJ: In vivo lIP_NMR studies offat salivary glands.
Magn Reson Med 8: 129, 1988
66. De Haan A, Jones DA, Sargeant AJ: Changes in velocity of
shortening, power output and relaxation rate during fatigue of rat
medial gastrocnemius. Pfliigers Arch 413: 422-428, 1989
67. Edwards RHT, Hill DK, Jones DA: Metabolic changes associated with
the slowing of relaxation in fatigued mouse muscle. J Physiol251: 287-
301,1975
68. Hamalainen N, Pette D. The histochemical profiles offast fiber types
JIB, TID, and IIA in skeletal muscles of mouse, rat, and rabbit. J
Histochem Cytochem 41: 733-743, 1993
Molecular and Cellular Biochemistry 184: 195-208, 1998.
1998 Kluwer Academic Publishers.

In situ measurements of creatine kinase flux by

NMR. The lessons from bioengineered mice
Klaas Nicolay, l Ferdi A. van Dorsten, l Torsten Reese, l Marijn J.
Kruiskamp,l Johannes F. Gellerich2 and Cees J.A. van Echteld3
lDepartment of in vivo NMR, Bijvoet Center, Utrecht University, Utrecht, The Netherlands; 2Neurological Clinic,
Martin-Luther University, Halle-Saale, Germany; 3Heart Lung Institute and ICIN, Utrecht University Hospital,
Utrecht, The Netherlands

Abstract
P-31 nuclear magnetic resonance (NMR) is uniquely suited to measure the kinetics of the phosphoryl-exchange reaction
catalyzed by creatine kinase in intact mammalian tissue, especially striated muscle. Recently developed transgenic mouse
models ofthe creatine kinase iso-enzyme system open novel opportunities to assess the functional importance of the individual
iso-enzymes and their relative contribution to the total in situ flux through the CK reaction. This chapter reviews the most
recent findings from NMR flux measurements on such genetic models of CK function. Findings in intact mouse skeletal and
cardiac muscle in vivo are compared to data from purified mitochondrial and cytosolic creatine kinase in vitro. The relevance
of findings in transgenic animals for the function of CK in wild-type tissue is described and the perspectives of transgenic
techniques in future quantitative studies on the creatine kinase iso-enzyme system are indicated. (Mol Cell Biochem 184: 195-
208,1998)

Key words: creatine kinase, skeletal muscle, heart muscle, nuclear magnetic resonance, magnetization transfer, transgenic
mice, energy metabolism, enzyme flux

Introduction is usually designated as MM-CK. In brain and many other

The creatine kinase system
Creatine kinase (ATP:creatine phosphotransferase; EC
2.7.3.2) is very abundant in excitable tissues, including
muscle and brain [1, 2]. In skeletal muscle, the enzyme can
reach concentrations of up to 1 mM monomeric units.
Creatine kinase (CK) catalyzes the exchange of a high-energy
phosphate group between phosphocreatine and ATP:
phosphocreatine (PCr2-) + Mg.ADP- + H+ <----> creatine
(Cr) + Mg.ATp2-
Five different CK iso-enzymes have been identified in
mammalian tissues. In adult striated muscle, the most
abundant CK species is present as a dimer in the cytosol and
tissues, the cytosolic isoform is of the so-called brain (or B)
type which also exclusively occurs in a dimeric configuration
(BB-CK). Shortly after birth, muscle harbors considerable
amounts of the B-type isoform which largely disappears in
adulthood, especially in skeletal muscle. The co-expression
of M- and B-type CK results in the formation of MB-CK
heterodimers [1]. Furthermore, there are two different
mitochondrial isoenzymes: the sarcomeric type in muscle
and the ubiquitous type in most other tissues harboring CK
[2]. Mitochondrial CK (Mi-CK) is located in the inter-
membrane space and primarily occurs as an octamer.
Although their relative abundance may widely differ, there
is always co-expression of a mitochondrial CK species with
at least one cytoplasmic isoform.
Mitochondrial CK is encoded on the nuclear DNA, trans-
lated with a cleavable targeting sequence and delivered to the

Present address: T. Reese, Sandoz, Preclinical Research, Basel, Switzerland

Address for offprints: K. Nicolay, Bijvoet Center, Utrecht University, Bolognalaan 50, NL-3584 CJ Utrecht, The Netherlands
196
intennembrane space by the mitochondrial protein import
machinery. Most of the mature Mi-CK is bound to the outer
leaflet of the inner membrane [1-4]. In muscle, part of the
MM-CK is freely dispersed in the cytosol while another part
is associated with various subcellular structures, including the
myofibrillar M-band, glycolytic multi-enzyme complexes
and the sarcoplasmic reticulum membrane [1, 3]. Most
probably, the cytoplasmic CK's are targeted to their cytosolic
destinations and kept in place by specific interactions.
Based on CK iso-enzyme diversity and its distinct sub-
cellular distribution, it has been proposed that the CK system
has the following major functions (reviewed in refs [1-3]):
(a) to provide temporal ATP buffering; (b) to ensure local
ATP buffering; and (c) to transport free-energy equivalents via
the so-called CKlPCr shuttle. The transientATP buffering role
is pronounced during periods of excessive cellular activity
when ATP demand temporarily exceeds ATP delivery. The
proposed local ATP buffering role of CK is mainly based on
the in vitro finding that phosphocreatine (PCr) supports
efficient free-energy delivery to certain ATPases, including
the myofibrillar myosin ATPase and the Ca2+-ATPase of the
sarcoplasmic reticulum [1]. Of course, the driving force for
the energy-requiring process is generated through ATP
hydrolysis by the appropriate ATPase but the 10caIATP/ADP
ratio is kept high by ATP regeneration via the nearby-bound
creatine kinase enzyme. The maintenance ofhighATP/ADP
ratios, even during periods of high cellular activity, is of
thermodynamic and kinetic importance since it aids in
keeping both the free-energy of ATP hydrolysis and the
ATPase activity high.
The third CK function of supporting free-energy transport
through the cell is usually referred to as the phosphocreatine
circuit or shuttle [1-3]. In this model, PCr acts as a free-energy
carrier which connects sites of free-energy delivery with
sites of free-energy utilization through diffusion. The
mitochondrial CK isofonn plays a key role in the model [1]
in that the coupling of its activity to oxidative phosphorylation
yields PCr as an endproduct of mitochondrial activity. The CK
species associated with cytosolic ATPases are at the PCr
utilizing/creatine (Cr) producing side of the circuit. The
circuit model is partly based on the estimated free-energy
transporting capacity of the ATP/AD P couple as compared to
that of the PCrlCr couple. ATP and ADP are expected to
diffuse less rapidly through the cytoplasm. More importantly,
however, the free concentration of ADP is several orders of
magnitude lower than that of Cr resulting in a much lower
diffusion capacity of ADP relative to Cr. Mathematical
modelling as carried out by Meyer et al. [5] lends support to
this idea that the CK system greatly facilitates the diffusion
of free-energy equivalents through the cell, and thereby also
accelerates and smoothes transitions between different work
states and dampens oscillations in the concentrations of ATP
andADP [2].
The P-31 NMR technique
P-31 nuclear magnetic resonance (NMR) is widely used as
a non-invasive tool for assessing the energetic status of intact
mammalian tissues, including skeletal muscle [6] and heart
muscle [7]. This application is mainly based on the deter-
mination of the (relative) concentrations of phosphocreatine,
ATP and inorganic phosphate from the P-31 spectra. P-31
NMR also is able to measure the in vivo flux through the
creatine kinase reaction, by using magnetization transfer (MT)
techniques. Such methods employ selective perturbation of
either the PCr or the y-ATP peak in the P-31 NMR spectrum
with a frequency-selective radio-frequency field which leads
to a modification ofthe intensity of the resonance concerned.
Via the chemical exchange reaction catalyzed by CK, the
frequency-selective perturbation is partially transferred to the
exchanging partner; PCr when the irradiating field is at the y-
ATP frequency, and y-ATP when irradiating at the PCr fre-
quency. This MT experiment enables the quantitation of the
forward and reverse flux through the creatine kinase reaction.
An important requirement of the NMR assay ofCK flux is that
the system is in a steady-state, i.e. that the metabolite levels
remain constant during the experiment. The MT experiment
can be carried out in many different ways.
In wild-type tissue, all CK iso-enzymes may be expected
to contribute to the measured CK flux. A differentiation
of the relative contributions of the mitochondrial and the
cytosolic CK species has not been possible until recently.
This chapter reviews P-31 NMR studies of transgenic
mouse models of CK which do in principle enable such
separate measurements of the cytoplasmic and mitochondrial
components of the CK system in the in vivo situation. The
overview of studies on intact skeletal and heart muscle is
complemented by a discussion of experiments on purified
MM- and Mi-CK. The latter studies allow for a full kinetic
analysis under well-defined in vitro conditions and thereby
serve as a reference frame for measurements in the in vivo
situation.
The transgenic creatine kinase mouse model
Wieringa and co-workers have recently developed extremely
versatile transgenic mouse models which offer unique
possibilities for creatine kinase function research [8-11].
Using homologous recombination in embryonic stem cells,
mice were generated that are modified in the expression of
one or more members of the striated muscle-specific CK
isoenzyme family. First the full MM-CK knock-out mouse
which is homozygous for the elimination of the M-type CK
was generated [8]. Figure 1 shows a schematic representation
of the consequences of this MM-CK null-mutation for the
muscle cell. It is evident that in MM-CK deficient muscle,
 197

the free-energy buffering role of CK can only be supported
by the mitochondrial isoform (except for the case of
myocardium where also a small amount of the brain type B-
CK is present [8]). The part of the phosphocreatine shuttle
that connects the mitochondrion with sites of free-energy
utilization, is no longer operative.
Wieringa and coworkers have also generated mice which
are deficient in mitochondrial CK [11] and mice which lack
both MM- and Mi-CK isoforms [11] in skeletal muscle and
heart. (For a detailed discussion on the transgenic technique,
the reader is referred to the contribution by Wieringa et al.).
It is important to note that the transgenic elimination ofthe
CK species so far examined has no significant compensatory
effect on the expression of any ofthe other CK isoforms. This
is a remarkable finding since the transgenic modification does
affect the expression of several enzymes in other metabolic
pathways: a number of glycolytic enzymes and mitochondrial
marker enzymes are expressed to a higher specific activity
in MM-CK knock-out skeletal muscle [8, 9].
P-31 nuclear magnetic resonance studies of creatine
kinase flux
In Vitro studies on purified CK enzymes
The kinetic properties of purified preparations of the
cytoplasmic and mitochondrial iso-enzymes of creatine
kinase were investigated using a combination ofP-31 NMR
and spectrophotometry [12]. An example of a P-31 NMR
experiment on purified MM-CK is shown in Fig. 2. Selective
irradiation of the resonances of PCr (Fig. 2B) and y-ATP
(Fig. 2D) results in a reduction of the signal intensity of the
y-ATP and PCr peaks, respectively, which can be used to
quantitate the unidirectional fluxes through the CK reaction.

A
Cr ......
___- -
cr~ATP~

pcr_--I.~Pcr~ADPJ ~

B
~-_I.~ Cr ATP~

....--I.~ PCr ADPJ~

LJI Mi-CK + MM-CK

Fig. 1. Schematic view of the consequences of the absence of the muscle-type creatine kinase iso-enzyme in the MM- CK -/- transgenic mouse.
(A) Wild-type situation with MM-CK and Mi-CK in the cytoplasmic and mitochondrial compartments, respectively. (B) The MMCK -/-
transgenic case where the cytoplasmic creatine kinase isoform is completely absent. To the left, the mitochondrion is depicted with its inner and
outer boundary membranes. The broken arrow (to the right) represents the process of ATP hydrolysis to drive cellular work, e.g. myosin ATPase
activity involved in muscle contraction. In the wild-type case (A), MM-CK is considered to playa role both in local ATP buffering (by
functionally coupling to ATPase activity) and in the transport of free-energy equivalents from the mitochondrion to sites of free-energy
utilization. Mitochondrial CK is involved in the rephosphorylation of Cr to PCr, using ATP from mitochondrial oxidative phosphorylation as the
free-energy donor. In the MM-CK -/- mutant (B), the CK/PCr circuit is no longer operative because the cytoplasmic member of the creatine
kinase system is absent. The temporal free-energy buffering role of the CK system may continue to function through the activity of Mi-CK. It
should be noted that the ATP buffering activity of Mi-CK requires diffusion of ADP into the mitochondrial intermembrane space, followed by
ATP diffusion in the opposite direction [3, 55].
198

per

P, ATP
/1"-
y a f3

I I
5 o -5 -10 -20 o -5 -10 -15
8 (p.p.m.) o (p.p.m.)
Fig. 2. In vitro magnetization transfer P-31 NMR experiment of the flux catalyzed by a purified preparation ofMM-CK. The sample contained 10 mM
phosphocreatine, 10 mM creatine, 5 mM ATP, and circa 160 units MM-creatine kinase (from rabbit muscle) in a buffer with 100 mM HEPES-Na, 5 mM NaP"
0.5 mM EGTA and 1 mM ~-mercapthoethanol at pH 7.4 and 25C. The P-31 NMR spectra are the sumof32 transients, acquired with a 90 degree excitation
pulse. During the recycle delay of 20 sec, a low power saturation pulse was given at the frequeJ;lcies indicated by the arrows. Spectra A and C represent
control experiments for irradiation ofPCr (B) and y-ATP (D), respectively, which enables corrections to be made for direct spillover of saturation power.
An exponential line broadening of 4 Hz was used. The procedure for converting the spectral information to CK fluxes in the forward and reverse direction
is detailed elsewhere [12].

Saturation transfer experiments were done as a function of 3

the MM-CK or Mi-CK activity, the Vmax' present. For both
isoforms, we observed a linear dependence of the steady-state 2.5
,.-..
flux on V max (Fig. 3). This is, of course, what one would expect U)

for a well-behaved enzymatic assay. The ratio between the -~ 2

steady-state flux and Vmax was approximately twice as high
for Mi-CK as for MM-CK under identical conditions [12]. We 51.5

also determined a set of Michaelis-Menten parameters, like
the apparent affinities (Km's) for the substrates creatine,
phosphocreatine, Mg-ATP and Mg-ADP [12]. The values
obtained were essentially the same for Mi- and MM-CK,
except for the KmADP that was substantially lower for Mi-CK
than MM-CK, i.e. 20 vs. 80 11M. The kinetic parameters were
subsequently used to calculate the expected CK flux, as based
on the kinetic model developed earlier for the M isoform by
Morrison and Cleland [13]. These authors have shown that
the kinetic scheme ofMM-CK is ofthe so-called random-order,
rapid equilibrium type [13]. Our modelling using the Morrison-
Cleland model demonstrated that the predicted flux closely
agreed with the measured flux, for both CK isoforms (solid
lines depicted in Fig. 3). Moreover, the calculations
indicated that the difference in flux-to-V max ratio between
the two iso-enzymes is mainly the result of the higher
affinity for ADP of the mitochondrial species.
a
~
1
~
U 0.5
o 2 4 6 8 10
Vrnax (mM/s)
Fig. 3. The dependence of the steady-state flux through purified mito-
chondrial and cytoplasmic creatine kinase under in vitro conditions. Mi-
CK (open circles) or MM-CK (closed circles) were added to buffer containing
CK substrates to the activities indicated on the x-axis whereafter P-31 NMR
was used to measure the flux in the forward direction with saturation transfer
P-31 NMR (as indicated on the y-axis). The solid lines represent the forward
fluxes through the respective iso-enzyme as calculated from their kinetic
constants, using the Morrison-Cleland kinetic model. For full details see
the legend to Fig. 2 and Van Dorstenet al. [12). (Reproduced with permission
from ref. [12]. Copyright Elsevier Science Publishers).

The in vitro studies suggest that, like the cytoplasmic
enzyme, the mitochondrial CK species has a random-order,
rapid equilibrium reaction mechanism. The most important
result was that the steady-state flux through each ofthe most
abundant muscle CK iso-enzymes closely corresponded to
the flux predicted from their Michaelis-Menten kinetic
parameters [12]. This finding represents an essential
reference frame for the interpretation of in vivo P-31 NMR
experiments.
In vivo skeletal muscle
We will now turn to in vivo P-31 NMR measurements of
CK flux in intact mouse skeletal muscle. In principle, the
magnetization transfer experiment is done similarly as in the
in vitro studies on purified CK enzymes, except that the
specificity of the exchange processes being measured has
to be considered. CK is the only enzyme that is able to
phosphorylate creatine and to dephosphorylate phospho-
creatine. Therefore, measuring the PCr peak intensity
changes upon selective saturation of the y-ATP peak provides
kinetic information that is related to the forward flux through
the CK reaction only. By contrast, ATP is a substrate for many
enzymes (among which the myosin-ATPase and the ion
pumping ATPases which jointly constitute a respectable
turnover number) and, consequently, the y-ATP peak changes
accompanying PCr irradiation cannot be simply interpreted
in terms of the reverse CK flux. For these reasons, we will
restrict the discussion of in vivo P-31 NMR data to
measurements of the forward CK flux which was arrived at
by selective perturbation of the y-ATP resonance.
Figure 4 shows a montage of magnetization transfer
experiments on resting hindleg skeletal muscle from a wild-
type mouse (Fig. 4A) and from transgenic mice that were
homozygous for MM-CK deficiency (Fig. 4B), for Mi-CK
deficiency (Fig. 4C) and for the combined MM-CK/Mi-CK
deficiency (Fig. 4D), respectively. The control spectra that
were obtained with irradiation downfield from the PCr peak
were surprisingly similar for the different mice. This implies
that the Na+ -dependent creatine uptake system is functioning
normally and that even in the case of the double knock-out
mutant sufficient CK activity persists to ensure a largely
normal degree of creatine phosphorylation. The origin of the
CK activity in the double knock-out is remarkable since
MM- and Mi-CK activity is nihil [11]. During embryonic
development, the B-type CK is expressed to a high activity
and thus PCr is readily formed during that period. It is
possible that the trace amount of B-CK remaining in
adulthood is involved in preserving this PCr pool, as
previously suggested by Steeghs et al. [11]. These authors
showed that the PCr in the double knock-out is no longer
utilized as a temporal energy buffer upon stimulation of
skeletal muscle [11]. This suggests that in this case the
presence of PCr is of no functional relevance although a
199
minor role involving the remaining B-type CK cannot be
fully excluded.
The series of spectra in Fig. 4 in which the selective
irradiation was at the y-ATP resonance show that the NMR-
detectable flux is strongly reduced by elimination of the M-
type CK (compare Fig. 4B with Figs 4D and 4H) but is not
significantly changed by Mi-CK deficiency (compare Fig.
4B with Fig. 4F, and Fig. 4D with Fig. 4H). In qualitative
terms, this is a plausible result: the mitochondrial CK
species only contributes around two percent to the total CK
activity in homogenates of wild-type tissue [1, 2, 8,14], the
majority being provided by cytoplasmic MM-CK. Our NMR
findings are in full agreement with those of Wieringa's group
[8-11]. Quantitative aspects of the relation between CK
activity in the tissue and NMR-detected flux will be dealt
with in a later section.
Although the steady-state CK flux becomes largely
undetectable by M-CK deficiency, the activity of the
remaining mitochondrial CK is readily detected by spectro-
photometry in tissue homogenates as well as by P-31 NMR
in vivo during work. Figure 5 depicts P-31 spectra that were
acquired before and during electrical stimulation ofthe sciatic
nerve to initiate muscle contraction in a MM-CK -/- mouse.
At rest (Fig. 5A), the PCr-to-ATP ratio is high and the Pi-to-
PCr ratio low, as usual. Isometric contractions lead to a
reduction in the PCr level and a corresponding increase in
Pi (Fig. 5B). Preliminary comparisons of the response of
wild-type and M-CK deficient skeletal muscle to twitch
contractions as monitored by P-31 NMR [8, 14] have shown
that the rate and extent of PCr hydrolysis and recovery are
surprisingly similar. This implies that, in the mutant, Mi-CK
catalyzes net synthesis of ATP during the initial phase of
muscle contraction and net PCr synthesis during the initial
phase of recovery. It is often assumed that, during work,
Mi-CK is displaced from equilibrium in the wild-type
situation and is tailored for the net synthesis of PCr, using
ATP synthesized by mitochondrial oxidative phosphorylation
(see refs [16, 17] and references cited therein). The NMR
data on transgenic tissue convincingly show that Mi-CK can
work either way and can be readily accessed by each of its
substrates. This is in line with expectations based on in vitro
experiments on isolated mitochondria. Obviously, direct data
on the Mi-CK equilibrium in the wild-type situation are not
available as yet because of the dominant role of the cyto-
plasmic isoform. It is generally accepted that in wild-type
skeletal muscle the high activity of the cytoplasmic CK
species maintains the CK reaction near-equilibrium,
essentially independent of work state. The NMR data from
the M-CK knock-out muscle strongly suggest, but do not
proof, that also the Mi-CK isoform alone is able to maintain
the CK reactants close to equilibrium, despite of its con-
siderably lower Vmax [14]. Steeghs et al. [11] have recently
reported that the PCr in skeletal muscle ofthe double MM-
200

per

NTP

c D

E F

10 o -5 -10 -15 -20 -25 10 5 o -5 -10 -15 -20 -25

ppm ppm

Fig. 4. P-31 NMR analysis of mouse hindleg skeletal muscle in vivo. Representative 81 MHz P-31 NMR spectra from the following mice are depicted: (A, B)
wild-type; (C, D) MM-CKknock-out; (E, F) Mi-CK knockout; (G, H) double MM-CKlMi-CK knock-out. The mice were anaesthetized with 1.3 % isoflurane in
a O,fN,o mixture (1 :2) delivered through a face mask. A three-turn solenoidal radiofrequency coil was placed around the upper hind leg of the mouse, so that
the NMR signal essentially originated from the gastrocnemius-plantaris-soleus muscle complex. Body temperature was maintained at 37C. A transient P-31
NMR saturation transfer protocol was used to determine the pseudo first-order unidirectional rate constant for the creatine kinase reaction from PCr to A TP.
They-ATP resonance was selectively irradiated for specific durations resulting in a time-dependent exponential decay of the PCr signal. Selective saturation,
at positions indicated by the arrows in the spectra, was achieved by applying low-power RF pulses with durations from 0-15 sec in a randomized order. The
total repetition time was maintained at 20 sec. Care was taken to minimize the saturation power (approximately 150 ~W), whilst selective
saturation of y-A TP was complete with a 0.3 sec pre-saturation pulse. Alternatingly, with each single saturation-transfer measurement, control
spectra were recorded in which the selective irradiation was placed at a frequency [v(PCr) - v(y-A TP)] down field from the PCr resonance (spectra
A, C, E, and G). Time-dependent P-31 NMR saturation transfer data were analyzed, as originally described by Forsen and Hoffman [56]. The PCr
signal, M,(PCr), as a function of the irradiation time (t) for y-ATP saturation, was fitted to a mono-exponential function:

M (PCr) 1 - k,o,' T1,pp (PCr)' (

~t ( T bP/PCr)
I - exp --t JJ ,
where Mo(PCr) is the PCr signal at time zero, i.e. without y-ATP saturation, k'n< is the apparent pseudo first-order forward CK rate constant and T1,pp(PCr) is
the apparent spin-lattice relaxation time of PCr, measured in the presence of saturation of y-ATP. Peak assignments: P;, inorganic phosphate; PCr,
phosphocreatine; U-, p- and y-ATP, the first, middle and terminal phosphates in adenosine triphosphate, respectively.

CKiMi-CK knock-out is resistant to contraction, while MiCK
deficient tissue has similar properties as the wild-type tissue.
So far, no measurements on CK flux have been reported for
hindleg skeletal muscle in M-CK or Mi-CK deficient mice
during work. It will be interesting to see whether or not the
flux through Mi-CK in the MM-CK -/- mutant remains
undetectable also under these conditions and, when detect-
able, how it responds to workload changes.
Recently, Koretsky and coworkers [18, 19] have studied
skeletal muscles of transgenic mice which overexpress the
B-CK subunit, resulting in the formation of dimeric MB-
and BB-CK isoforms and a 50% increase in the total tissue

(B)
PCr
NTP
P a.
(A)
I I I I I I
10 5 0 -5 -10 -15
Chemical shift (ppm)
Fig. 5. P-31 NMR of hind leg skeletal muscle of a MM-CK deficient mouse,
at rest (A) and during electrical stimulation (B). The experimental set-up is
similar to that described in the legend to Figure 4, except that the sciatic
nerve was exposed to allow for electrical stimulation of the muscle. Isometric
twitch contractions were delivered at a frequency of I Hz. Spectra of 64
scans which were acquired at a repetition time of 1 sec were continuously
recorded throughout the exercise and recovery periods. Typical spectra
from such a series are shown here.
CK activity. The flux through the CK reaction as measured
by P-31 NMR increased twofold, i.e. disproportionate to
the increase in specific CK activity. B-CK overexpression
did not significantly affect the metabolic and contractile
characteristics of the muscle, except that the rise-time of
force during an isometric tetanus was slightly faster. These
data suggest that CK activity in normal skeletal muscle is
sufficient to keep up with changes in ATP demand, except
perhaps during the early phase of intense contractile
activity [19].
Isolated myocardium
We have recently carried out P-31 NMR studies [14, 2~22]
on isolated hearts of wild-type and M-creatine kinase knock-
out mice which were perfused according to Langendorff. The
myocardial preparations were studied at three different
workloads: during pacing at 400 and 600 beats min-I, and
during KCl-induced arrest. Phosphocreatine and ATP
concentrations in M-CK deficient hearts appeared not
significantly different from those in wild-type hearts.
Using P-31 NMR saturation transfer, the flux mediated
by the mitochondrial creatine kinase (Mi-CK) was clearly
detected in M-CK deficient hearts. This is an interesting
finding that contrasts the situation in M-CK deficient
skeletal muscle where no appreciable CK flux was detected,
and enables the study of the relationship between (Mi-)CK
flux and mitochondrial activity as varied by changes in rate-
pressure product. The detection of Mi-CK related flux is
201
probably partly due to its relatively high specific activity in
heart where it represents approximately 25% of total CK
activity [14, 2~22]. It is possible that a small part of the
CK flux in M-CK -/- myocardium is related to the B-type
iso-enzyme that represents a few percent of the total CK
activity in wild-type hearts. M-CK deficiency has no effect
on the expression of the other CK isoforms in heart [8].
In M -CK -/- myocardium, the CK flux ranged from circa
6~90 % of that measured in wild-type tissue, depending on
the work state. These preliminary data suggest that in wild-
type mouse hearts the relative contribution of Mi-CK to the
total CK flux is larger than that of M -CK [14, 2~22].
The NMR-based flux measurements on isolated wild-type
and M-CK deficient mouse hearts are compared to the
results ofbio-mathematical modelling in the chapter by Aliev
et al. Previous modelling studies by the same workers [23]
have led to the suggestion that in the heart the CK system is
at equilibrium only during the diastolic phase of the contract-
ion cycle and that it is at equilibrium only in the cytoplasmic
compartment and not in the mitochondrial intermembrane
space. Future NMR experiments should address this issue and
particularly aim at assessing the kinetics of workload trans-
itions. Unfortunately, the low intrinsic sensitivity of the NMR
experiment requires signal averaging over many cardiac
cycles. This particularly applies to mouse heart with its small
tissue mass. More sophisticated NMR protocols should be
developed to get data in distinct phases of the cardiac cycle.
Relationship between NMR-detected CKflux and CK
activity
We have extensively studied the relationship between the
CK flux as detected by P-31 NMR and the activity ofCK that
is present in the tissue. The NMR-detected flux catalyzed by
purified MM- and Mi-CK quantitatively agrees with their
activity under in vitro conditions. The question is, whether
this is also the case in vivo and, if not, what the reason for
the discrepancy may be and what can be learnt from that. Even
if a quantitative correspondence between the CK flux and
tissue activity were absent, this might lead to novel in-
formation on the status of CK in the intact tissue, provided
that the reason for the discrepancy is identified.
The usual procedure to address the flux-V max relationship
is to measure in situ CK flux by NMR whereafter the tissue
is homogenized and enzymatic analysis of the homogenate
is performed. This approach has led to conflicting findings.
When comparing CK flux in rat brain, heart and skeletal
muscle with the flux estimated from Vmax and CK reactant
concentrations, Bittl et al. [24] concluded that the CK rate
equation correctly predicts the in vivo reaction velocity.
Several others, however, have reported discrepancies
between the CK flux measured by NMR and the flux pre-
dicted on the basis of the CK V max in the tissue homogenate.
McFarland et al. [25] measured a considerably lower CK
202

flux in cat soleus than that predicted. We also observed a large reached in the intact muscle. The loss of cellular integrity and
discrepancy between in situ CK flux and predicted flux in changes in ionic conditions upon tissue solubilization could
mouse skeletal muscle [14, 21]. change the enzyme's Vmax and apparent substrate affinity by
Van Deursen et al. [9] have recently made puzzling eliminating effector molecules that may be involved in
observations that nevertheless provide challenging oppor- regulating CK activity in the intact tissue. McFarland and
tunities for future studies. By crossbreeding wild-type and Kushmerick [25] have previously proposed such an effector
MM-CK deficient species, Van Deursen generated mice that role of certain anions, in particular carbonate. Such phen-
had intermediate expression levels of MM-CK, at 16, 34 omena are easily eliminated when grinding up the tissue. It
and 50% of wild-type activity [9]. Figure 6 depicts the seems rather improbable that these considerations per se are
relationship between the CK mediated flux, as measured in sufficient to explain the striking data of Van Deursen et al. [9]
vivo in hindleg skeletal muscle by P-31 NMR, and the level that the NMR-detected CK flux in resting skeletal muscle does
of MM-CK expression [9]. The most striking observation not scale linearly with the level ofMM-CK expression (Fig. 6).
was that MM-CK flux was below the detection limit up to A satisfying explanation for this finding cannot be given at
34% expression. CK flux was readily detected in skeletal present. More experimentation is required, for example address-
muscle with 50% of the wild-type CK activity. In fact, the ing the issue as to whether MM-CK localization is altered as a
magnitude of the magnetization transfer was almost the function of its expression level: is the enzyme homogeneously
same for the cases of 50% and wild-type expression levels distributed among its possible locations in the cell, irrespective
(Fig. 6). of the expression level or does its distribution change with
increasing expression level? This information is of great
vmax (U.mg- 1 protein) interest since CK localization may affect the 'NMR visibility'
of its flux. Combined cell biological, biochemical and bio-
0 5 10 15 20 25 30 35 40
30 physical experiments will solve these issues in the future, in
particular when tools are developed to modify the expression
25 40 level and cellular location ofM-type CK in a more sophist-
~ icated manner than currently possible.
~
Cil20
-.... 30 ~
Ol
~ (5 Functional relevance of NMR-based CK flux
.. 15 E measurements
X 20 ~
::J P-31 NMR offers the unique possibility to measure the
u:: 10 X

jlO
::J
u::
5
~ 0
0
0 20 40 60 80 100 120
Fig 6. The forward flux through the CK reaction in the hindleg muscles of
mice with reduced M-CK expression [9). The x-axis represents the CK activity
as measured in the tissue homogenate, and is either expressed as mM.sec
lor in units per mg of tissue protein. The forward CK flux (data points) was
measured using inversion transfer. The solid line represents the expected
steady-state flux, as estimated from the tissue CK activity, the kinetic
parameters of M-CK, and the concentrations of CK reactants, and was
based on the Morrison-Cleland kinetic model. The calculation was done
analogous to the procedure used in Fig. 3. Note that the NMR-detected CK
flux does not scale linearly with the activity measured in the homogenate.
For full details see Van Deursen et al. [9]. (Reproduced with permission
from ref[9]. Copyright National Academy of Sciences, USA).
It goes without saying that the above approach of com-
paring in vivo NMR data with homogenate CK assay data is
associated with considerable uncertainties. It is possible that
the CK activity assay in tissue homogenates gives an
erroneous impression of the maximal CK activity that can be
kinetics of the CK-catalyzed chemical exchange reaction in
the intact tissue. An obvious question to addre:ss is what the
physiological significance is of the steady-state flux thus
measured. For simplicity, we will assume that the uni-
directional fluxes quantitated from the NMR magnetization
transfer experiment are the actual forward and reverse fluxes
in the cell. Being intrinsically insensitive, NMR inevitably
samples a relatively large tissue mass. In most studies on
Langendorf-perfused myocardial preparations, for example,
the NMR signal is sampled from the entire organ and there
is no spatial discrimination of the signal to specific parts
of the ventricular tissue. Recent developments in NMR
measurement strategies make it possible to collect many
spectra from a single heart, each of which originate from a
different region (for an example, see ref. [26]). This opens
up fascinating opportunities for assessing regional diff-
erences in metabolism in relation to myocardial function and
dysfunction. These developments are becoming increasingly
available also to in vivo studies of human and animal heart
and skeletal muscle. Despite the technical advances the
measured NMR signal remains the superposition of signal
from molecules that are in many different cells and that are
at different locations inside these cells. Even in the case of
perfect tissue homogeneity, the magnetization transfer

studies of CK flux can therefore not be interpreted in tenns
of exchange that is occurring at distinct sites in the cell. In
wildtype muscle, it is impossible to resolve experimentally
which part of the measured CK flux is provided by the
cytoplasmic or the mitochondrial CK species, let alone that
we could differentiate among myofibrillar M-CK and
sarcolemma M-CK. Novel developments in genetic engin-
eering may make it possible in the foreseeable future to
knock out specific subclasses of CK species associated with
specific cellular entities and then to assay the kinetic
properties of specific CK subsets with in vivo NMR.
Since the magnetization transfer P-31 NMR has no
intrinsic resolution at the cellular level, the flux data can
only be interpreted in terms of the flux through the entire
CK system. In combined experimental and modelling studies,
Ingwall and coworkers [27] have previously attempted to
separate the flux through cytoplasmic CK from that through
mitochondrial CK. These authors developed a kinetic model
to evaluate the consequences of the possible existence of a
distinct pool of ATP that is unsaturatable in magnetization
transfer experiments. The approach was prompted by their
finding that time-dependent saturation transfer data are
often poorly described by a single-exponential decay. It was
claimed that the best way to explain this behaviour is to
assume an unsaturatable ATP pool [27, 28] which was
initially supposed to be NMR-invisible as well [28]. The
modelling indicated that the distinct ATP pool can only
realistically be in the mitochondrion. The majority oftheATP
is saturatable, NMR visible and present in the cytoplasm.
The modelling of an extensive set of cardiac data from rat
and rabbit [27] suggested that the flux from the unsaturatable,
presumably mitochondrial, ATP pool to PCr (i.e. the flux
through MiCK) is directly proportional to the rate of
mitochondrial oxidative phosphorylation, as estimated from
the rate of oxygen consumption. The flux in the reverse
direction seemed negligible at all workloads. It was also
suggested that the calculated Mi-CK flux is very similar
to the measured rate of mitochondrial ATP synthesis, in
agreement with a close functional coupling between oxidative
phosphorylation and Mi-CK activity.
It is of interest to compare the modelling results of
Ingwall's group with our preliminary findings in transgenic
mouse myocardium. We found a moderate change ofMi-CK
flux in MM-CK deficient heart with workload and the Mi-CK
flux seemed very much higher than the expected maximal rate
of mitochondrial respiration. It should be stressed, however,
that the findings in wild-type and in transgenic myocardium
should be compared with caution: because of the absence
of CK in the cytoplasmic compartment, a PCr/CK shuttle
is no longer operative in MM-CK deficient heart (Fig. I). This
will have consequences for the response of Mi-CK flux to
alterations of workload since a net flux through Mi-CKis no
longer possible. Furthermore, the elimination of MM-CK
203
from the cytoplasmic compartment may lead to adaptations,
further complicating a direct comparison of wild-type and
transgenic tissue.
Accepting the fact that the NMR flux measurements in
wild-type tissue allow no direct differentiation ofthe fluxes
through the cytoplasmic and mitochondrial species of CK
but rather yield the total flux through the entire CK system,
what do these overall CK fluxes tell us about CK function
in the cell's energy metabolism. Let us first consider the
magnitude of the in vivo CK flux as determined by NMR
techniques, in relation to the capacity of the major free-
energy delivering and free-energy utilizing pathways in the
cell. It is often found that the CK flux greatly exceeds the
maximal rates of ATP hydrolysis and synthesis. In cardiac
muscle CK flux is approx. an order of magnitude more rapid
than the maximal rate of ATP turnover [29]. Consequently,
the over-all CK reaction may be considered to be near-
equilibrium. (It should be stressed that this does not
necessarily mean that the CK reaction is close to equilibrium
everywhere in the cell at all times.) This high CK activity
emphasizes the prominent role of the enzyme in free-energy
buffering.
The next point to consider is whether the CK flux changes
at all with changes in cellular activity. Most studies in
skeletal muscle and heart muscle indicate that, in striated
muscle, there is not a dramatic change in CK flux with
increased workload. In cat soleus, McFarland et al. [25]
found that while kfor increased somewhat with workload, the
concomitant decrease of the PCr level caused the CK flux
to remain essentially unchanged. Similar data have been
reported by Brindle et al. [30] for rat gastrocnemius. It
should be realized that even if CK flux is independent of
workload, a major proportion of high-energy transport may
still proceed via PCr, due to the high CK equilibrium constant
and high CK activity [5]. The extent to which CK flux varies
with workload does therefore not present a critical test of
the functional relevance of the CK/PCr shuttle model.
Sauter and Rudin [31] have studied the CK flux in rat brain
and compared this to EEG activity as modulated by bi-
cuculline. A linear correlation was found between the CK
kfor and the EEG intensity, as a measure of brain activity, and
the rate of deoxyglucose-6-phosphate formation, as a
measure of the glycolytic flux. PCr levels decreased at high
EEG intensities whereas ATP remained constant for the
activity range tested.
We have recently studied the relationship between CK flux
and the activity of sea urchin spenn cells [32]. Depending on
the composition of the medium, the cells were either
maintained in a quiescent or maximally activated state.
Activation was accompanied by a several-fold increase in
oxygen consumption. In inactive spenn no CK exchange flux
was detected, while in activated sperm a steady-state CK
flux of 3.1 0.5 mM.sec l was measured. The estimated
204
intracellular free [ADP] was increased from circa 10 11M in
quiescent sperm to 110 11M in the activated state. Tombes and
Shapiro [33] have previously shown that uncompromised CK
function is vital to sustain motility of the sperm flagellum.
Selective inhibition of CK using fluorodinitrobenzene led
to diminished motility starting from the distal end of the
flagellum, i.e. at the largest distance from the sperm head
where the major part of the sperm's free-energy requirement
is delivered by mitochondrial oxidative phosphorylation.
The sea urchin sperm cell is therefore regarded the prototype
of a eukaryotic cell which is dependent on the CK/PCr
shuttle for its proper functioning.
These examples suggest that alterations in over-all (NMR
detected) CK flux with cellular activity can qualitatively be
understood in terms of kinetic regulation of the enzyme by
the concentrations of its substrates. Considering the cellular
levels of CK reactants and the kinetic constants of the
various CK isozymes, it is plausible that the free ADP level
has a strong regulatory effect on CK flux [34]. This is in
agreement with the common finding that the increases in CK
kfor with workload are parallelled by an increase in free [ADP].
Alternative mechanisms of CK flux regulation, for example
involving covalent modifications or non-covalent interactions
with (non-substrate) effector agents, may playa role as well.
Although some studies described here may give the
impression that there is a substantial redundancy in CK flux
in mammalian tissues, several lines of evidence suggest that
the high CK flux capacity of a tissue is of substantial
functional importance. For various forms of heart failure,
it has been observed that there is a close relationship
between the product of PCr content and CK flux (termed
energetic reserve), and the viability of the tissue. In post-
ischemic ferret heart Neubauer et al. [35] observed that the
flux through the CK reaction was strongly diminished and
was related to the myocardial contractile reserve. The
mechanism underlying such a relationship between CK flux
and tissue function remains to be elucidated.
The relevance of findings in transgenic animals for
energy metabolism in the wild-type situation
Transgenic mice offer extraordinary opportunities for
assessing the in vivo function of proteins. Changes brought
about by the transgenic action are specific to the gene that
is manipulated and, therefore, these must be the primary
cause of the functional effects that are observed. In present
day transgenic technology, however, the genetic defect is
present from the embryonic stage, leaving ample time for
long-term adaptations in response to the deficiency in the
chosen gene product [36]. The transgenic elimination ofthe
M-type creatine kinase leads to distinct phenotypic changes
in hindleg skeletal muscle [8, 37]: the transgenic tissue
lacks the ability to perform burst exercise and gains more
endurance character. Such phenotypic changes are very
informative and provide important information on the role
of the M-type CK in myocellular energy metabolism.
However, part ofthe consequences ofM-CK deficiency may
have been obscured by the compensatory adaptive response
of the tissue. MM-CK deficient skeletal muscle has more
mitochondria and a higher glycogen content [8], in agreement
with the above change in phenotype (see also chapter by
Wieringa et al. which specifically deals with adaptations in
response to CK deficiency). Ventura-Clappier and coworkers
[38,39] have also reported a number of adaptive changes in
ventricular tissue, including an increased expression of
several glycolytic enzymes. Such alterations make it
difficult to draw direct conclusions on the importance ofthe
MM-CK species in the wild-type situation.
The transgenic CK models demonstrate that muscle tissue
has a remarkable plasticity to CK deficiency when present
from the embryonic stage. This may well explain the apparent
controversy that M-CK deficiency has very mild effects on
myocardial functional performance while the loss of
contractile reserve in heart failure models is correlated with
the diminished CK flux [35]. The reduced tolerance to CK
deficiency in adult myocardium may de due to the limited
plasticity of mature heart muscle whereby compensatory
mechanisms playa less prominent role. These data also
suggest that the CK deficient mouse does not represent a
relevant model for neuromuscular diseases.
It has recently been suggested that the free-energy trans-
porting role ofCK may be partly taken over by the adenylate
kinase (AK) iso-enzyme system [40] which also has a
member in both the cytoplasm and the mitochondrial inter-
membrane space. This suggestion seems plausible since both
CK and AK are able to rapidly propagate a signal of altered
ATPIADP ratio from sites of free-energy utilization to sites
of free-energy delivery. This so-called metabolic wave
propagation involves the transmission of a net Cr signal in
the case of CK and a net AMP signal in the case of AK, both
directed towards the mitochondrion for stimulating oxidative
phosphorylation. The latter can be accomplished by trans-
lating the Cr or AMP signal in the intermembrane space by
the mitochondrial iso-enzymes ofCK andAK, respectively.
This process can also be viewed as implying that CK (and
similarly AK) is involved in the (indirect) shuttling of ADP
across the outer mitochondrial membrane. We have recently
reported evidence that seems to give a clue as to why
transport of Cr rather than direct transport of ADP may be
of considerable functional importance. Studies on isolated
mitochondria showed that at physiological colloidosmotic
pressures the diffusion of ADP across the outer membrane
is restricted [41-44]. As a consequence, ADP diffusion
gradients of up to some 15 11M are formed under high flux
conditions, especially when CK is involved. It is not clear

yet whether the restriction effect is imposed by the outer
membrane pore protein (the only way in and out the
mitochondrion for charged metabolites) or related to the
diffusion properties of the intermembrane space. The
diffusion restriction will, of course, not only apply to ADP
but similarly hold for PCr, ATP and Cr. However, the latter
remains without a functional effect because the concentrations
of PCr, ATP and Cr (several millimolars) are orders of
magnitude higher than that of ADP (typically 10-40 11M).
Since both CK and AK are represented on either side of the
mitochondrial outer membrane, the transfer of Cr and AMP
from the cytoplasm into the intermembrane space can be
viewed as the indirect transfer of ADP, the substrate of
oxidative phosphorylation.
In this respect it is of interest to mention the recent paper
by Veksler et al. [39] who studied skinned fibers from
skeletal muscle and cardiac muscle from M-CK deficient
animals. Their data are suggestive of an increased permeability
of the mitochondrial outer membrane in ventricular tissue
and slow-twitch skeletal muscles and not so much in fast-
twitch skeletal muscles. The facilitation of ADP transport
across the outer membrane might partially compensate for the
impairment of energy transport caused by M-CK deficiency.
It is suggested that the lack of such a mitochondrial effect
in the fast-twitch skeletal muscles can be explained from the
dominant free-energy buffering function of M -CK in such
tissue [39].
The question arises to what extent the transgenic CK
models are to be preferred over the more classical models
ofCK function which mostly involve the feeding of creatine
analogues [45--48]. The analogues act by competing with
endogenous creatine for the creatine uptake system in the
cell membrane which can lead to a sizable reduction of the
intracellular creatine pool. More direct interference with CK
activity via the introduction of inhibitors which often works
fine in vitro is cumbersome for in vivo studies: some of the
agents have a very limited specificity (e.g. the often used SH-
reagents may easily inhibit other enzymes as well) and they
have to be given systemically which provides minor chances
of privileged delivery to the target tissue. Ingwall and
coworkers [49,50] have successfully employed this inhibitor
approach, using iodoacetate, for studies on CK function in
isolated heart preparations.
Many animal models for myopathies are characterized by
lowered overall levels of Cr en PCr and by an increased
sensitivity to energetic stress conditions. Therefore, the
feeding of experimental animals with Cr analogues seems a
versatile approach to test whether lowered Cr levels cause
impaired muscle function, and to investigate the physiological
role of PCr, Cr and CK in intact muscle. However, creatine
analogue feeding has severe limitations (for a review see ref.
[45]). Chronic Cr depletion is accompanied by massive
metabolic and morphological adaptations. Moreover, Cr
205
depletion is never complete and the Cr analogues still serve
as (somewhat poor) substrates to CK. The transgenic models
of CK function therefore seem superior to those using Cr
analogue feeding, especially because the genetic approach
allows better control of the CK activity in the tissue.
Perspectives of transgenic models in in vivo studies of
mammalian bio-energetics
Transgenic mouse models have absolutely revolutionized the
field of CK research [36] and are expected to lead to many
important new insights in the fundamental role of the enzyme
in tissue function. Most studies have sofar been devoted to
striated muscle. Koretsky [51] has expressed the brain-type
B-CK gene in mouse liver, using a liver specific expression
vector and showed that the enzyme was functionally integrated
into the tissue's metabolism. This is a very interesting
finding since the liver harbours no endogenous CK. Sofar,
knock-out mice homozygous for the deficiency in the B-CK
which would, obviously, primarily affect the brain have not
been described. The availability of such mice would give
exciting possibilities for assessing the role of CK in this
complex 'organ'.
Transgenic models of CK have tremendous potential in
quantitative studies of the regulation of mammalian bio-
energetics that have hitherto not been exploited. Several
theoretical and conceptual frameworks have been recently
developed that aim at the quantitative description of cellular
energetics [52, 53]. One of the prominent approaches in this
field is called Metabolic Control Analysis (MCA) which
provides the tools for the quantitative determination of the
control of a certain enzyme on the flux through the pathway
it participates in. A recent addition to the original MCA
method involves the possibility to account for metabolic
channelling [53]. This is very important for CK function
research because Mi-CK and myofibrillar CK are probably
involved in local, unidirectional phosphoryl group transfer,
with limited equilibration with the bulk pool ofCK reactants.
The classical application of MCA involves the stepwise
reduction of the activity of the enzyme tested, either by
reduction of its content or by gradual inhibition with a
specific inhibitor (which is most often done in in vitro
studies) whereafter the effect this has on the pathway flux
is quantified. The best results are obtained when the reduction
in enzyme activity is done in very small steps because this
allows the precise measurement of so-called flux control
coefficients. Such techniques are increasingly applied to in
vivo systems as well. Molecular biological tools are then used
to modulate the expression of the enzyme of interest whereby
the control exerted by the enzyme can in principle be deter-
mined. Sofar in vivo studies using MCA have been mainly done
on micro-organisms. Westerhoff and coworkers [54] have for
206
example determined the control of the H+-ATPase of the E.
coli cell membrane on the cell's energy status. With the
generation of transgenic mice the MCA approach can be
extended to intact tissue in the intact animal. Van Deursen's
approach ofthe gradual expression ofM-CK striated muscle
[9] shows that such experiments are indeed feasible. The
steps in M-CK expression level were somewhat larger than
commonly employed in MCA (Fig. 6) but nevertheless enable
the assessment of the control of M-CK on a variety of
processes. One of the preliminary (and puzzling) conclusions
would be that MM-CK has no measurable control on its own
flux: Van Deursen found that the elimination of50% of the M-
CK activity only led to a small reduction in the CK flux as
measured by P-31 NMR (Fig. 6). Another conclusion could
be that the degree of control, expressed in the control co-
efficient, depends on the expression level and possibly the
cellular distribution of the CK iso-enzymes. The application
of techniques such as MCA to transgenic mouse models of
CK or of any other enzyme or protein will undoubtedly
contribute to an improved understanding of the functional,
metabolic and physiological consequences of the gene
deficiency. The possibility to determine the energetic status
and the CK flux of tissues in vivo in a noninvasive manner
will keep the NMR technique at the forefront in this field of
research.
Acknowledgements
The in vivo NMR studies described were carried at the
Netherlands in vivo NMR facility which is part of the
Bijvoet Center of Utrecht University and is supported by the
Netherlands Organization for Scientific Research (NWO). The
facility is part of the European Union-Large Scale facility for
Biomolecular NMR. Torsten Reese gratefully acknowledges
the financial support from the Human Capital and Mobility
programme of the European Union. Johannes Gellerich was
supported by the EO-Large Scale facility programme. Marijn
Kruiskamp is supported by the Foundation for the Life
Sciences (SLW) with financial aid of NWO.
We are indebted to Be Wieringa and Frank Oerlemans
(Department of Cell Biology and Histology, Nijmegen
University) for providing breeding couples of the transgenic
mice. The in vitro studies on purified CK and sea urchin
sperm cells were done in collaboration with Theo Wallimann
(ETH, Zurich).
References
1. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine
kinase isoenzymes in tissues with high and fluctuating energy
demands: The 'phosphocreatine circuit' for cellular energy
homeostasis. Biochem J 281: 21--40, 1992
2. Wyss M, Smeitink J, Wevers RA, Wallimann T: Mitochondrial
creatine kinase: A key enzyme in aerobic energy metabolism.
Biochim Biophys Acta 1102: 119-166, 1992
3. Wallimann T: Dissecting the role of creatine kinase. Curr Bioi I: 4246,
1994
4. Rojo M, Hovius RC, Demel R, Nicolay K, Wallimann T: Mitochondrial
creatine kinase mediates contact formation between mitochondrial
membranes. J Bioi Chern 266: 20290--20295, 1991
5. Meyer RA, Sweeney HL, Kushmerick MJ: A simple analysis of the
phosphocreatine shuttle. Am J Physiol246: C365--C377, 1984
6. Radda GK: Control, bioenergetics and adaptation in health and disease:
Non-invasive biochemistry from nuclear magnetic resonance. FASEB
J6:3032-3038,1992
7. Schaefer S, Balaban RS: Cardiovascular magnetic resonance spectro-
scopy. Kluwer Academic Publishers, Boston, 1992
8. Van Deursen J, Heerschap A, Oerlemans F, Ruitenbeek W, Jap P, Ter
Laak H, Wieringa B: Skeletal muscles of mice deficient in muscle creatine
kinase lack burst activity. Cell 74: 621--631, 1993
9. Van Deursen J, Ruitenbeek W, HeerschapA, Jap P, Ter Laak H, Wieringa
B: Creatine kinase in skeletal muscle metabolism: A study of mouse
mutants with graded reduction in M-CK expression. Proc Natl Acad
SciUSA91:9091-9095,1994
10. Van Deursen JMA: The role of the creatine kinase/phosphocreatine
system studied by gene targeting. Thesis, Nijmegen University,
Nijmegen, 1994
11. Steeghs KGJ, Benders A, Oerlemans F, De Haan A, Heerschap A,
Ruitenbeek W, Jost C, Van Deursen J, Perryman B, Pette D, Briick-
Wilder M, Koudijs J, Jap P, Veerkamp J, Wieringa B: Altered Ca 2+
responses in muscles with combined mitochondrial and cytosolic
creatine kinase deficiencies. Cell 89: I-II, 1997
12. Van Dorsten FA, Furter R, Bijkerk M, Wallimann T, Nicolay K: The in
vitro kinetics of mitochondrial and cytosolic creatine kinase detennined
by saturation transfer lip NMR. Biochim Biophys Acta 1274: 59--66,
1996
\3. Morrison JF, Cleland WW: Isotope exchange studies ofthe mechanism
of the reaction catalyzed by adenosine triphosphate: Creatine
phosphotransferase. J Bioi Chern 241: 673--683, 1966
14. Van Dorsten FA: 31P-NMR analysis of creatine kinase kinetics. From
purified enzyme to intact skeletal and cardiac muscle. Thesis, Utrecht
University, Utrecht, 1996
15. Hall N, DeLuca M: Developmental changes in creatine phosphokinase
isoenzymes in neonatal mouse hearts. Biochem Biophys Research
Commun 66: 988--994,1975
16. Saks VA, Ventura-Clappier R, Aliev MK: Metabolic control and
metabolic capacity: two aspects of creatine kinase functioning in the
cell. Biochim BiophysActa 1274: 81-88,1996
17. Saks VA, Khuchua ZA, Vasilyeva EY, Belikova 0 Yu, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration -a synthesis. In: VA Saks, R Ventura-Clappier (eds). Cellular
Bio-Energetics: Role of Coupled Creatine Kinases. Kluwer Academic
Publishers, Dordrecht, The Netherlands, 1994, pp 155-192
18. Brosnan MJ, Raman SP, Chen L, Koretsky, AP: Alteration of the creatine
kinase isoenzyme distribution in transgenic mouse muscle by over-
expression of the B subunit. AmJ PhysioI264: CI51--C160, 1993
19. Roman BB, Foley JM, Meyer RA, Koretsky AP: Contractile and
metabolic effects of increased creatine kinase activity in mouse skeletal
muscle. AmJ Physiol39: C I 236--C I 245, 1996
20. Van Dorsten FA, Nederhoff MGH, Nicolay K, Van Echteld CJA:
P31-NMR saturation transfer measurements of creatine kinase
flux in wild-type and MM-CK deficient Langendorff-perfused

mouse hearts. Proc. 4th Annual Scientific Meeting Inter-
national Society for Magnetic Resonance in Medicine, New
York, April 27-May 3, 1996, P 432
21. Van Dorsten FA, ReeseT, NederhoffMGH, Van Echteld CJA, Nicolay
K: In vivo fluxes through mitochondrial and cytoplasmic creatine
kinase. P-31 NMR of skeletal and cardiac muscle from transgenic
mice. Proc. 4th Annual Scientific Meeting International Society for
Magnetic Resonance in Medicine, NewYork,ApriI27-May 3, 1996,
p24
22. Van Echteld CJA, Van Dorsten FA, Nederhoff MGJ, Nicolay K: 31p
NMR study of creatine kinase (CK) flux in wild-type and transgenic
M-CK deficient perfused mouse hearts. Int Soc for Heart Research,
XVII European Congress, Bologna, J Mol Cell Cardiol28: A22, 1996
23. Saks VA, Aliev MK: Is there the creatine kinase equilibrium in working
heart cells? Biochem Biophys Res Commun 227: 360-367, 1996
24. Bittl JA, DeLayre J, Ingwall JS: Rate equation for creatine kinase predicts
the in vivo reaction velocity: 11 P NMR surface coil studies in brain,
heart, and skeletal muscle of the living rat. Biochemistry 26: 6083-
6090,1987
25. McFarland EW, Kushmerick MJ, Moerland TS: Activity of creatine
kinase in a contracting mammalian muscle of uniform fiber type.
BiophysJ67: 1912-1924,1994
26. Von Kienlin M, Le FurY, Behr W, Roder F, HaaseA, Horn M, Braeker B,
Neubauer S: In the isolated perfused heart, regional metabolic
heterogeneity can be detected during global low-flow ischemia. A IIp_
3D-CSI-NMR-study. Proc. 3rdAnnuai Scientific Meeting Society for
Magnetic Resonance, Nice, August 19-25, 1995, P 296
27. Zahler R, Ingwall JS: Estimation of heart mitochondrial creatine kinase
flux using magnetization transfer NMR spectroscopy. Am J Physiol
262: H 1022-HI028, 1992
28. Zahler R, Ingwall JS:Analysis of compartmentation ofATP in skeletal
and cardiac muscle using lip nuclear magnetic resonance saturation
transfer. Biophys J 51: 883-893, 1987
29. Matthews PM, Bland JL, Gadian DG, Radda GK: A lIP_NMR saturation
transfer study of the regulation of creatine kinase in the rat heart.
Biochim Biophys Acta 721: 312-320, 1982
30. Brindle KM, Blackledge MJ, Challiss RAJ, Radda GK: lip NMR
magnetization transfer measurements of ATP turnover during steady-
state isometric muscle contraction in the rat hind limb in vivo.
Biochemistry 28: 4887-4893, 1989
31. Sauter A, Rudin M: Determination of creatine kinase kinetic parameters
in rat brain by NMR magnetization transfer. Correlation with brain
function. J Bioi Chern 268: 13166-13171, 1993
32. Van Dorsten FA, Wyss M, Wallimann T, Nicolay K: Activation of sea
urchin sperm motility is accompanied by an increase in the creatine
kinase exchange flux. Biochem J 325: 411-416, 1997
33. Tombes RM, Shapiro BM: Metabolite channelling: A phosphoryl-
creatine shuttle to mediate high energy phosphate transport between
sperm mitochondrion and tail. Cell 41 : 325--334, 1985
34. Van Dorsten FA, Reese T, Gellerich JF, Van Echteld CJA, Nederhoff
MGJ, Muller H-J, Van Vliet G, Nicolay K: Fluxes through cytosolic and
mitochondrial creatine kinase, measured by P-31 NMR. Mol Cell
Biochem 174: 33-42, 1997
35. Neubauer S, Hamman BL, Perry SB, Bittl JA, Ingwall JS: Velocity of the
creatine kinase reaction decreases in post-ischemic myocardium: A
lIP-NMR magnetization transfer study of the isolated ferret heart. Circ
Res 63: 1-15, 1988
36. Koretsky AP: Insights into cellular energy metabolism from
transgenic mice. Physiol Rev 75: 667-688, 1995
37. De Haan A, Koudijs JCM, Wevers RA, Wieringa B: The effects of
MM-creatine kinase deficiency on sustained force production of mouse
fast skeletal muscle. Exp Physiol 80, 491--494, 1995
38. Ventura-Clappier R, Kuznetsov AV, D'Albis A, Van Deursen J,
207
Wieringa B, Veksler VI: Muscle creatine kinase-deficient mice. I.
Alterations in myofibrillar function. J Bioi Chern 270: 19914--
19920, 1995
39. Veksler VI, Kuznetsov AV, Anflous K, Mateo P, Van Deursen J, Wieringa
B, Ventura-Clappier R: Muscle creatine kinase-deficient mice. II. Cardiac
and skeletal muscles exhibit tissue-specific adaptation of the mito-
chondrial function. J BioI Chern 270: 19914--19920, 1995
40. Dzeja PP, Zeleznikar RJ, Goldberg ND: Suppression of creatine kinase-
catalyzed phosphotransfer results in increased phosphoryl transfer
by adenylate kinase in intact skeletal muscle. J BioI Chern 271: 12847-
12851,1996
41. Laterveer FD, Gellerich FN, Nicolay K: Macromolecules increase the
channelling of ADP from externally associated hexokinase to the matrix
of mitochondria. Eur J Biochem232: 569--577, 1995
42. Laterveer FD, Nicolay K, Gellerich FN: ADP delivery from adenylate
kinase in the mitochondrial intermembrane space to oxidative phos-
phorylation increases in the presence of macromolecules. FEBS Lett
386: 255--259,1996
43. Gellerich FN, Zierz S, Laterveer FD, Nicolay K: Effect of colloidosmotic
pressure on ADP transport into mitochondria. Experimental determination
of ADP concentration gradients across the mitochondrial outer
membrane. In: HV Westerhoff, JL Snoep, FE Sluse, JE Wijker, BN
Kholodenko (eds). BioThermoKinetics of the Living Cell. 1996, pp
295--299
44. Laterveer FD, Nicolay K, Gellerich FN: Experimental evidence
for dynamic compartmentation of ADP at the mitochondrial
periphery: Coupling of mitochondrial adenyl ate kinase and
mitochondrial hexokinase. Mol Cell Biochem 174: 43-51, 1997
45. Wyss M, Wallimann T: Creatine metabolism and the consequences
of creatine depletion in muscle. In: VA Saks, R Ventura-Clappier
(eds). Cellular Bio-Energetics: Role of Coupled Creatine Kinases.
Kluwer Academic Publishers, Dordrecht, The Netherlands, 1994,
pp 51-68
46. Van Deursen J, Jap P, HeerschapA, Ter Laak H, Ruitenbeek W, Wieringa
B: Effects of the creatine analogue ~-guanidinopropionic acid on
skeletal muscles of mice deficient in muscle creatine kinase. Biochim
BiophysActa 1185: 327-335,1994
47. O'Gorman E, Beutner G, WallimannT, Bridczka D: Differential effects
of creatine depletion on the regulation of enzyme activities and on
creatine-stimulated mitochondrial respiration in skeletal muscle, heart
and brain. BiochimBiophysActa 1276: 161-170, 1996
48. Boehm EA, Radda GK, Tomlin H, Clark JF: The utilization of creatine
and its analogues by cytosolic and mitochondrial creatine kinase.
Biochim Biophys Acta 1274: 119--128, 1996
49. Hamman BL, Bitt! JA, Jacobus WE, Allen PD, Spencer RS, Tian R,
Ingwall JS: Inhibition of the creatine kinase reaction decreases the
contractile reserve of isolated rat hearts. Am J Physiol 38: HI03O-
Hl036,1995
50. Tian R, Ingwall JS: Energetic basis for reduced contractile reserve in
isolated rat hearts. Am J Physiol39: HI207-HI216, 1996
51. Koretsky AP, Brosnan MJ, Chen L, Chen J, Van Dyke T: NMR
detection of creatine kinase expressed in liver of transgenic mice:
Determination of free ADP levels. Proc Nat! Acad Sci USA 87: 3112-
3116, 1990
52. Westerhoff HV, Van Dam K. In: Thermodynamics and Control of
Biological Free-Energy Transduction. Elsevier, Amsterdam, The
Netherlands, 1987
53. Kholodenko BN, Cascante M, Westerhoff HV: Control theory of
metabolic channelling. In: VA Saks, R Ventura-Clappier (eds). Cellular
Bio-Energetics: Role of Coupled Creatine Kinases. Kluwer Academic
Publishers, Dordrecht, The Netherlands, 1994, pp 313-331
54. Jensen PR, Michelsen 0, Westerhoff HV: Experimental deter-
mination of control by W-ATPase in Escherichia coli. J Bioenerg
208

Biomembr 27: 543-554, 1995 56. Forsen F, Hoffman RA: Study of moderately rapid chemical
55. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of exchange reactions by means of nuclear magnetic double
the mitochondrial creatine kinase. Nature 381: 341-345, 1996 resonance. J Chern Phys 39: 2892-2901, 1963
Molecular and Cellular Biochemistry 184: 209-229, 1998.
1998 Kluwer Academic Publishers.

Mathematical model of compartmentalized energy

transfer: Its use for analysis and interpretation of
31P-NMR studies of isolated heart of creatine
kinase deficient mice
Mayis K. Aliev,3 F.A. van Dorsten,4 M.G. Nederhoff,5 C.J.A. van
Echteld,5 Vladimir Veksler,6 Klaas Nicolay and Valdur A. Saks 1,2
lLaboratories of Bioenergetics, University of Joseph Fourier, Grenoble, France; 2Institute of Chemical Physics and
Biophysics, Tallinn, Estonia; 3Laboratory of Experimental Cardiac Pathology, Cardiology Center, Moscow, Russia;
4Department of in vivo NMR spectroscopy, Bijvoet Center for Biomolecular Research, Utrecht University, Utrecht;
5Heart-Lung Institute and ICIN, Utrecht University Hospital, Utrecht, The Netherlands; 6INSERM Unit 446, Universite

Paris - Sud, Chatenay- Malabry, France

Abstract
A mathematical model of the compartmentalized energy transfer in cardiac cells is described and used for interpretation of
novel experimental data obtained by using phosphorus NMR for determination of the energy fluxes in the isolated hearts of
transgenic mice with knocked out creatine kinase isoenzymes. These experiments were designed to study the meaning and
importance of compartmentation of creatine kinase isoenzymes in the cells in vivo. The model was constructed to describe
quantitatively the processes of energy production, transfer, utilization, and feedback between these processes. It describes
the production of ATP in mitochondrial matrix space by ATP synthase, use of this ATP for phosphocreatine production in the
mitochondrial creatine kinase reaction coupled to the adenine nucleotide translocation, diffusional exchange of metabolites
in the cytoplasmic space, and use of phosphocreatine for resynthesis of ATP in the myoplasmic creatine kinase reaction. It
accounts also for the recently discovered phenomenon of restricted diffusion of adenine nucleotides through mitochondrial
outer membrane porin pores (VDAC). Practically all parameters of the model were determined experimentally. The analysis
of energy fluxes between different cellular compartments shows that in all cellular compartments of working heart cells the
creatine kinase reaction is far from equilibrium in the systolic phase of the contraction cycle and approaches equilibrium
only in cytoplasm and only in the end-diastolic phase of the contraction cycle.
Experimental determination of the relationship between energy fluxes by a 3lP_NMR saturation transfer method and workload
in isolated and perfused heart of transgenic mice deficient in MM isoenzyme of the creatine kinase, MM -/- showed that in the
hearts from wild mice, containing all creatine kinase isoenzymes, the energy fluxes determined increased 3-4 times with
elevation of the workload. By contrast, in the hearts in which only the mitochondrial creatine kinase was active, the energy
fluxes became practically independent of the workload in spite of the preservation of26% of normal creatine kinase activity.
These results cannot be explained on the basis of the conventional near-equilibrium theory of creatine kinase in the cells,
which excludes any difference between creatine kinase isoenzymes. However, these apparently paradoxical experimental results
are quantitatively described by a mathematical model of the compartmentalized energy transfer based on the steady state
kinetics of coupled creatine kinase reactions, compartmentation of creatine kinase isoenzymes in the cells, and the kinetics
of ATP production and utilization reactions. The use ofthis model shows that: (1) in the wild type heart cells a major part of
energy is transported out of mitochondria via phosphocreatine, which is used for complete regeneration of ATP locally in the
myofibrils - this is the quantitative estimate for per pathway; (2) however, in the absence of MM-creatine kinase in the
myofibrils in transgenic mice the contraction results in a very rapid rise of ADP in cytoplasmic space, that reverses the

Addressfor offPrints: V. Saks, Laboratory of Bioenergetics, Joseph Fourier University, BP 53X - 38 041 Grenoble Cedex, France
210

mitochondrial creatine kinase reaction in the direction ofATP production. In this way, because of increasing concentrations of
cytoplasmicADP, mitochondrial creatine kinase is switched off functionally due to the absence of its counterpart in PCr pathway,
MM-creatine kinase. This may explain why the creatine kinase flux becomes practically independent from the workload in the
hearts of transgenic mouse without MM-CK. Thus, the analysis of the results of studies of hearts of creatine kinase-deficient
transgenic mice, based on the use of a mathematical model of compartmentalized energy transfer, show that in the PCr pathway
of intracellular energy transport two isoenzymes of creatine kinase always function in a coordinated manner out of equilibrium,
in the steady state, and disturbances in functioning of one of them inevitably result in the disturbances of the other component
of the PCr pathway. In the latter case, energy is transferred from mitochondria to myofibrils by alternative metabolic pathways,
probably involving adenyl ate kinase or other systems.
The merits of several mathematical models of muscle energy metabolism are also discussed. (Moll Cell Biochem 184: 209--229,
1998)

Key words: creatine kinase, transgenic mice, energy transfer, mathematical model, compartmentation, heart, energy
metabolism

Introduction sumption in isolated perfused hearts (refs [12-14], see the

Development of experimental research in the field of
cellular bioenergetics has led to clear requirement for new
quantitative methods of description of intracellular energy
fluxes in order to aid in the interpretation of experimental
results. Traditional simplified approaches which are based
on using the creatine kinase equilibrium for calculations of
intracellular ADP concentration and explain the regulatory
mechanisms of energy metabolism in muscle cells in terms
of homogeneous solution kinetics appear to be insufficient
[1-10]. Indeed, the basis of cardiac bioenergetics is
continuous production of high energy phosphates in mito-
chondrial oxidative phosphorylation (to the lesser extent-
in the glycolysis), their transfer to intracellular sites of
energy utilization, and finally conversion of the chemical
energy stored in these molecules into the mechanical
(contraction), osmotic (ion transport) and chemical (bio-
synthesis) work by the correspondingATPases. These energy
fluxes are directly determined by the work performed, and
the main characteristics of the cardiac energy metabolism
is very high efficiency of the feedback regulation of
respiration, which is the basis for control of oxidative
phosphorylation. That means that under normoxic conditions,
elevation of the workload of the heart results in rapid
proportional elevation of the rates of respiration and
oxidative phosphorylation without any significant change in
the intracellular level of metabolites such as ATP, phos-
phocreatine and creatine [4, 5]. These data can therefore not
be interpreted on the basis of experiments in vitro with isolated
mitochondria, in which the respiration rate is always depend-
ent on the ADP concentration in accordance with simple
Michaelis-Menten relationship [11]. The proposed explan-
ation of simultaneous regulation of respiration and contraction
by calcium ions has also not yet been found not to be
completely satisfactory, since inhibition of calcium uptake by
ruthenium red does not change the rate of oxygen con-
chapter by Hansford and Zorov in this volume).
Recent experimental and theoretical research has shown
that the regulation of respiration in the intact cell may be
in principle very different from that which operates in
homogeneous enzymatic systems and have been used in
attempts to explain the energy metabolism of the cells. Most
probably, in the highly organized intracellular medium the
enzymes and organelles including mitochondria are involved
in multiple direct structural and functional interactions.
Indeed, the concept of the organization of cell metabolism
is rapidly changing: modern methods, such as confocal
microscopy [15], embedment-free electron microscopy
[16], and the skinned fiber technique [7, 17] developed for
studies of the structure and function of the cell in vivo
directly demonstrate that the cell is a highly inhomogeneous
system. The cell is organized into compartmentalized and
microcompartmentalized enzyme systems which often
involve metabolic channelling of substrates. Cardiac,
skeletal muscle and brain cells for example contain a
complex specialized creatine kinase/phosphocreatine
energy transfer system which is considered to fulfil
important roles in the energy metabolism of skeletal and
cardiac muscles [ 9, 10, 18].
Already 20 years ago Peter Mitchell, the founder of the
chemiosmotic theory of oxidative phosphorylation, form-
ulated a general theory of vectorial ligand conduction
according to which the substrates and products of inter-
connected enzymatic reactions move from one enzyme to
another without intermediate release and, in this way, very
rapid signal transmission from one part of the cell to another
may occur [19]. This kind of macroosmotic process of
vectorial ligand transduction in heart cells may be performed
by structurally organized creatine kinase and adenyl ate
kinase systems for rapid feedback signaling between
contraction and respiration processes (ref. [17], see also the
chapter by Van Beek et al. in this volume). The structural

basis of this kind of intracellular signaling system is poorly
understood. New methods of the quantitative description of
energy metabolism are an absolute requirement for the
interpretation of experimental data and for the design of
critical new experiments.
This volume includes three approaches to mathematical
modelling of cellular energy metabolism, one of which is
described here. Furthermore, this chapter includes the
description of novel experimental results on the studies of
compartmentation phenomena in cardiac cells by means of
molecular biology, transgenic technology and NMR which
produced the unexpected results from the point of view of
traditional equilibrium theories and thus demonstrated the
needs for interpretation of the experimental data. The novel
experimental data can be accounted for in quantitative
manner by a mathematical model of compartmentalized
energy transfer which is totally based on experimental data
and includes all relevant processes.
Materials and methods
LangendorfJ perfusion of isolated hearts from wild and
MM-CK -/- transgenic mice
Hearts were obtained from male, wild-type C57B I mice
and MM-CK deficient mice [20] and perfused with a
modified Krebs-Henseleit medium, containing 124 mM
NaCl, 4.7 mM KCl, 1.3 mM CaCI 2 , 1.0 mM MgCI 2 , 24 mM
NaHC0 3, 10 rnM glucose and 5 mM Na-pyruvate (pH 7.35
at 37C), at a hydrostatic pressure of 76 mm Hg. The left
ventricular developed pressure (LVDP) and heart rate were
monitored with a fluid-filled catheter. Two pace-electrodes
were attached to the right ventricular outflow tract. The
heart was then placed into a 10 mm diameter NMR tube,
together with a reference capillary filled with methylene
diphosphonate.
31 P nuclear magnetic resonance spectroscopy
31P_NMR spectra were recorded at 202.5 MHz, using a 90
flip angle. First, a fully relaxed 31P_NMR spectrum was
recorded from 64 scans with an interpulse delay of 10 sec.
Next, a transient 31P-NMR saturation transfer protocol was
employed by selectively irradiating the enzyme y-ATP
resonance for specific durations, ranging from 0.0-6.0 sec,
and observing the phosphocreatine (PCr) signal, keeping
the repetition time constant at 6.1 sec. Finally, a control
experiment was performed in which the selective irradiation
was placed at a frequencyv(PCr) -v(y-ATP) downfield from
the PCr resonance. Direct radio frequency spillover was
generally less than 10%. Each heart was studied successively
211
at 400 and 600 beats.min-1 and during cardiac arrest, as
induced by 20 mM KCI. Workload is expressed as rate-
pressure product, i.e heart rate times LVDP. The intracellular
pH was assessed from the chemical shift difference between
intracellular Pi and PCr. 31P_NMR signal were quantified by
iterative fitting of the time-domain signal with the method of
variable projection (VARPRO), using prior knowledge [21].
Kinetic analysis
The time-dependent 31P_NMR magnetization transfer was
used to determine the pseudo first-order unidirectional rate
constant for the forward creatine kinase reaction, k for ' as
previously described [21, 22].
Enzyme activities and metabolite determinations
For biochemical analysis, hearts were freeze-clamped.
Creatine kinase activity was determined according to standard
procedures, after homogenization of the tissue in a medium
containing 0.1% Triton X-IOO. For metabolite assays, the
frozen hearts were extracted with ice-cold perchloric acid
(5% w/v), followed by centrifugation and neutralization.ATP
and PCr were determined spectrophotometric ally, using
standard coupled enzyme assay at pH 6.7 [21]. The creatine
concentration was also determined using a spectrophoto-
metric assay [23]. Metabolite concentrations were calculated
on the assumption that the intracellular water volume is 0.348
ml per g wet wt tissue.
Mathematical model of compartmentalized energy
transfer in the cells
Figure 1 shows the general scheme of the compartmentalized
energy exchange between mitochondrial and myofibrillar
compartments. This scheme is based on the wealth of
experimental data extensively reviewed in refs [6-9, 18].
The differential equations for plane diffusion of ATP,
ADP, creatine (Cr), phosphocreatine (PCr) and inorganic
phosphate (P) along the chosen diffusion path were solved
numerically [26] with diffusion constants for myoplasm
taken from Meyer et al. [27]. The computations on the basis
of Runge-Kutta-Nistrom algorithm [28] gave the metabolite
concentrations for each segment ofthe diffusion path at 0.01
msec time intervals. In mitochondria ATP is synthesized in
the matrix by the ATP-synthase and after translocation
across the inner mitochondrial membrane is used for
phosphocreatine synthesis in the coupled mitochondrial
creatine kinase reaction; in the myofibrillar compartment
creatine kinase is considered to be distributed homo-
212

Mitochondrion

Myofibril

i
Cytoplasm
Myofibril cor~O""'k__2_ 4 6 8 10 12 <tE--j
j(; I)iffusio.DpatiJ --~>I ~ 0.1 urn

II Myosin Pi ::. ...

~ADP
~cr"-:'"

~;;
t ~. "'.':
0_

:!;
l Myoplasmi CK I :.: .c:ij
-lr c~J' c: ..
O.c
'5;
ATP
,ge
Myofibril ( 010 ) Cytoplasm :::E Intermembrane
( 11 ) space (13)

Fig. 1. The general scheme of compartmentalised energy transfer in cardiac cell. In the model, the one-dimensional diffusional exchange of ATP, ADP, PCr,
Cr and Pi between myofibrils and mitochondria is considered along their radii and in a layer of cytoplasm. The total diffusion path length of 1.3 11m includes
ten 0.1 11m space units (j) in myofibril and 3 units in cytoplasm, mitochondrial outer membrane and intermembrane compartments. The lower part of the Fig.
shows compartmentation ofmyoplasmic CK (CKmyo) in myofibril1ar and cytoplasmic spaces, of mitochondrial CK (CKrnit) in the intermembrane space, of
adenine nucleotide translocase (ANT) in mitochondrial inner membrane and of mitochondrial ATP synthase (Syn) in mitochondrial matrix space. Myofibril1ar
myosin provides ATP hydrolysis during contraction. CKrnit and ANT are proposed to be coupled by high local ATP concentration, arising from restricted
ATP diffusion in the narrow gap (microcompartment) between them. Arrows indicate diffusion fluxes of metabolites within the different compartments and
across the mitochondrial outer membrane. Reproduced from ref. [56] with permission.

geneously in the cytoplasmic and myofibrillar spaces. The
calculated concentrations of substrates along the diffusion
pathway have been used for calculations of enzymatic rates
at each spatial location and each time point to take into
account the concentration changes due to enzymatic
activities. Functional coupling of mitochondrial creatine
kinase (CKmit) to the adenine nucleotide translocase (ANT)
was modelled by means of a high local ATP concentration
in the narrow space (100 A) - microcompartment - between
coupled molecules creatine kinase and ANT. The ATP
concentration in this microcompartment rises due to an
approximately 106-fold local retardation ofATP diffusion. The
degree of diffusion restriction from this local compartment
was estimated from the deviation of the mass action ratio
of CKmit from the solution creatine kinase equilibrium
constant value, as determined experimentally by Saks et al.
[29] and Soboll et al. [30, 31] in coupled rat heart mito-
chondria under phosphorylating conditions.
During contraction, the kinetics of myofibrillar ATP
hydrolysis were predicted from dP/dt change in isovolumic
rat hearts: a linear increase inATP hydrolysis rate up to 30th
msec, foIIowed by a linear decrease to zero at 60 msec at a
heart rate of 333 beats/min. The total amount of ATP
hydrolyzed in one cardiac contraction cycle was taken to be
0.414 mmollkg wet wt. This corresponds to an oxygen
consumption of23 mmol/minlkg wet wt, again at a heart rate
of 333 beats/min.
The main equations
Diffusion between myofibri/s and mitochondria: The plane
diffusion of metabolites along the chosen diffusion path
(Fig. I) was calculated numericaIIy by the explicit finite-
difference method [26] using the Runge-Kutta-Nistrom
algorithm [28]. This approach predicts the metabolite
concentration C for the next time point (t + ~t) from the
metabolite concentration at a given time t:
C'M' = 0' + r x (C ' - 2 x C' + C ') (I)
J ~ J ~
In this equation C denotes the concentration; the super-
scripts t and (t + ~t) indicate time; the subscripts j - I, j and j
+ 1 indicate the spatial position withj ranging from zero to 13
(Fig. I). The diffusion constant r is equal to Dx x ~t/~J2; Dx
is diffusion coefficient for metabolite x; ~t is the time
increment; ~I is the distance between two adjacent points
along the diffusion path (Fig. I).
The method is originaIIy designed for calculations in cases
where the metabolite concentrations at both boundaries of
a plane sheet are fixed [28]. To eliminate this inconvenience,
the myofibril was treated as a symmetric system with maximal
 213

(or minimal) concentrations being in the myofibril core (j = 0,

Fig. 1). In such a system C I' is equal to C-I', and Ct'" in ..- myo+mit
4.0
I

equation 1 is determined by changing values of Co' and Ci = y

~

C ,.
-I'
At the other boundary (j = 13, Fig. 1) diffusion stops at the
mitochondrial inner membrane. This case can be modelled
by taking the real C\3' equal to assumed additional C 14':
Restricted diffusion through the mitochondrial outer
membrane: Diffusion through the outer membrane was
simulated by changing the value of diffusion constant (r) in the
membrane space (the space betweenj = 11 andj = 12 in Figs 1
and 2A). According to Crank (equation 8.45 in ref. [26]), the
basic equation 1 can be rewritten for the concentrations at both
faces of the membrane (CII'M' and Cit"") as
Cit"" = C l2 ' + r x resm X (C II ' - C 12') - r x (C 12'
According to these equations, diffusion in the spaces just
before (between CIO' and C I / ) and after (between C l2 ' and
C[3') the membrane occurs with unchanged diffusion
constant (r), while in the membrane space (between C I/ and
C I2 ' ) it takes place at the changed diffusion constant (r x
resm). Resm (membrane resistance, resm ::; 1 ) reflects the
degree of the decrease of the diffusion coefficient in the
membrane space.
Creatine kinase: Activities of mitochondrial creatine kinase
(CKmit) and myoplasmic creatine kinases (CKmyo), v nc,cKmi'
and vne,CKmYO, respectively, were calculated from the steady
state equation of Saks et al. (equation 13 in ref. [32]) using
the experimentally determined values of the kinetic constants
[33] and the substrate concentrations as calculated at each
location. For CKmit theATP concentrations were local ones,
while ADP, Cr and PCr concentrations were those in the
mitochondrial intermembrane space (j = 13, Fig. 1).
E REV-myo
3
OJ
.Y
(Ia)
3.0
T""
I
CIl
0
2.0
E FORW-myo
E
x
(1 b) :J 1.0
~
U
0.0
0 1 2 3
ATP synthesis (mmol*s-1*(kg wm))
Fig. 2. Unidirectional averaged steady-state fluxes through cellular creatine
kinases at different simulated workloads in the isolated rat heart. Different
workloads at a constant heart rate of 333 beats per min were modelled by
taking the values of ATP hydrolysis in the myofibrils (ATPh in Table 1) to
(I c) be 0.0018, 0.018, 0.09, 0.\8, 0.27, 0.36 and 0.44 mmol x bear! x (kg wmt
! in each separate simulation. The parameters used in the simulations are
- C l3 ') indicated in 'Materials and methods'. The diffusion restrictions in the
mitochondrial outer membrane correspond to apparent Km for ADP of
(ld) 400 uM (resATP = reswp = 0.007). The values of all unidirectional fluxes were
integrated through the entire cardiac cycle and cytoplasmic diffusion
pathway (see Fig. 1) to obtain average data for comparison with those of
fue "P-NMR measurements. The intracellular concentrations of the
metabolites were as those in ref. [56]. Workload dependencies of
unidirectional mean steady-state fluxes furough CKmyo plus CKmit (myo +
mit, thick line) and CKmyo are depicted in the reverse (Rev-myo, dashed
thin line) and forward (FORW-myo, dotted thin line) directions. The forward
direction is phosphocreatine synthesis, reverse reaction - its utilization. The
mean total unidirectional fluxes through CKmyo plus CKmit are equal in
both directions. Therefore, the differences between thick and thin lines
indicate the contribution of CKmit in unidirectional mean fluxes in the
forward and reverse directions. The inequality of forward and reverse mean
unidirectional fluxes catalysed by particular creatine kinase isoforms again
indicates their steady state displacement from equilibrium.
where K d, K p and K a are the dissociation constants for ADP,
Pi and ATP, respectively; k f and kr are the rate constants for
the forward (ATP synthesis) and reverse (ATP hydrolysis)
reactions, respectively.
The steady state activity of Syn (vne,syn) for this scheme
can be expressed by equation 2,

ATP-synthase: The activity oftheATP synthase (Syn) in the V

net
syn =Vf max X [ADP] x [P]/K
ld
x KP x Den) - V rmax X [ATP]/
matrix space was estimated from a simple kinetic scheme (Ka X Den) (2)

Kd Kp kf Ka where vfmax and vrmax are the maximal rates in the forward
E+ADP<->E.ADP<->E.ADP.P=EAfP<-->
I
E +ATP. (ATP synthesis) and reverse (ATP hydrolysis) reactions,
Kr respectively; Den designates the Denominator of the
expression:
214
Den = 1+ [ADP]lKd + [ADP] X [PJI(Kd X K) + [Jffi>]lKa (2a)
[ADP], [PJ and [ATP] designate the concentrations of these
metabolites in the matrix space. The steady state con-
centration of Pi in the matrix space has been assumed to be
10-fold higher [34] than its level in the mitochondrial
intermembrane space for all time points.
Adenine nucleotide translocase: ATP export by the Adenine
Nucleotide Translocase (ANT) was taken to be equal toATP
production by the ATP-synthase: for this case the maximal
rate ofATP production by the ATP-synthase was assumed to
be equal to the maximal rate ofANT activity. Thus, to a first
approximation, it was assumed that ANT is at equilibrium,
providing a constant ATP/ADP ratio between matrix and
intermembrane space at a fixed membrane potential according
to the concept of Wilson and Erecinska and several others,
as reviewed by Hassinen [35]. The ANT -provided steady state
concentration of ADP in matrix space has been assumed to
be about 25-fold higher than its level in the mitochondrial
intermembrane space, due to the electrogenity of ADP-ATP
exchange reaction [36]. The sum of concentrations of ADP
and ATP in the matrix space, ANPm' was taken constant, at
10 mM. Accordingly, the steady state ratio betweenATP and
ADP concentrations in the mitochondrial matrix (m) and
intermembrane (im) spaces [23], (ATPim/ADPim)/(ATP ml
ADP m)' was assumed of a constant value, 25. In conclusion,
it was assumed that:
ADP m= ADP im X ANP j(ADP im + ATP imI25) (2b)
ATP m =ANP m -ADPm (2c)
Diffusional coupling between mitochondrial creatine
kinase and adenine nucleotide translocase: Functional
coupling of mitochondrial creatine kinase and the adenine
nucleotide translocase is a well documented phenomenon [6-
9, 18,24,25,29-32]. For its mathematical modelling we have
developed a probability approach which describes the
fluxes of adenine nucleotides out of mitochondria and into
mitochondria. [24, 25]. It is the most detailed model ofcoupled
reactions in mitochondria, however, its incorporation into the
complete model of cellular energy metabolism is difficult
because of the very large number of calculations required. This
is why we have simplified, in this stage of modelling, the
mathematical description of this central process of com-
partmentalized energy transfer functional coupling in
mitochondria. For the simplification we have used the
experimental data of Fossel et al. [37] which show that the
coupling between enzymes becomes of functional relevance
if the distance between them is smaller than 10 nm, or 100
A. This was taken to be the maximal size of a gap (micro-
compartment) between mitochondrial creatine kinase and
the adenine nucleotide translocase in which the local ATP
concentration is increased due to metabolic channelling. The
functional coupling was modelled by assuming a specific
retardation ofATP diffusion in 100 A narrow space between
coupled molecules (Fig. 1). The ATP concentration in this
gap was calculated in the following way. In the space between
translocase and creatine kinase, theATP transported by ANT
during the time interval dt increases the ATP concentration
by ([AATP]ANT). On the other hand, ATP partly diffuses out
of the gap decreasing the ATP concentration by (A[ATP]Dif).
Also, part oftheATP is consumed by mitochondrial creatine
kinase that decreases the ATP concentration by (A[ ATP]CKm)
(Fig. 1). The net change in local ATP concentration in the
gap, (A[ ATP]loc)' is determined by the difference between
ATP influx and outflux,
A[ATP]loc =A[ATP]ANT-A[ATP]CKm -A[ATP]Dif (3)
Equation 3 was used for the calculation of the local ATP
concentration ([ATP]lo) in the gap. For this purpose the
terms of equation 3 were calculated or expressed as a
function of [ATP]loc; their back introduction into equation 3
gave an easily solvable quadratic equation with [ATP]loc as
the only unknown. The final expression is too long to present,
so it seems reasonable to show only the principle of these
calculations.
A[ ATP]IOC the change in the local ATP concentration in the
gap, is the difference between local ATP concentrations at
time t + At, ([ATPt), and the preceding time point t,
([ATP]IOCO) ('0' meaning old concentration):
A[ ATP]loc = [ATP]locO - [ATP]loc (4)
The flux of ATP from ANT has been assumed equal to the
rate of net ATP synthesis by the ATP-synthase. So for the
time interval At the change in 10calATP concentration due to
ATP influx from ANT, A[ATP]ANT' is
(5)
A[ATP]cKm' ATP consumption by CKmyo during the time
interval At, is
(6)
VnetCKmit is calculated from [ATP]IOC and the ADP, Cr and PCr
concentrations in the intermembrane space according to the
kinetic equation for the creatine kinase reaction [32, 33].
A[ATP]Dii' the diffusional outflux ofATP from the gap, can
be estimated from the equations for restricted diffusion.
ATP diffusion in the gap (space betweenATPloc andATPout)
is restricted by the given coefficient, resg (resistance for

ATP diffusion in the gap space). The value of this coefficient
is estimated from experimental data of changes in creatine
kinase reaction characteristics induced by oxidative phos-
phorylation. The ATP concentration profile in this gap is
assumed to be symmetrical. So according to equations 1c
and I a the diffusional changes in [ATP]out and [ATP]loc during
time interval L1t can be expressed as:
L1[ATPtu, = rl x ([ATP] 13'- [ATP]out')-rl x resgx ([ATPtui
- [ATP]loc') (7)
L1[ ATP]loc = r I x resg x ([ ATPLui - [ATP]loc') x 2 (8)
In these equations r1 = D AT? x L1t!L11]2; D AT? is the diffusion
coefficient for ATP; L1t is the time increment; L11] is the
distance between two adjacent points along the diffusion
path.
In the steady state, L1[ATPtu, = L1[ATP]loc' Equalizing the
right sides of equations 7 and 8 gives the expression for
[ATPLut':
[ATPtu,' = (ATPI3'+ 3 xATP 1oc' x resg)/(I + 3 x resg) (9)
Substituting [ATP]oui in equation 8 by its expression in
equation 9, we obtain equation 10:
L1[ATP]IOC = D[ATP]Dif= 2 x rl x resgx ([ATP]!3'-
[ATP]luc')/ (l + 3 x resg) (10)
Kinetics of ATP hydrolysis by myofibrils during con-
traction: The proposed kinetic scheme of ATP hydrolysis
in the myofibrils: a linear increase in ATP hydrolysis rate
up to 30 msec (t]), followed by a linear decrease to zero at
60 msec (t2), has a simple analytic solution:
whereATP h is the total amount ofATP hydrolyzed per cardiac
cycle; Acl is the coefficient of the acceleration in the ATP
hydrolysis rate in the to - t] time period; Dcl is the coefficient
of the deceleration in the ATP hydrolysis rate in the time
period from tl to t2"
The Acl and Dcl coefficients are interrelated:
( 12)
Taken together, equations 11 and 12 allow the calculation of
the Acl (and hence Dcl) values:
(13)
Ael and Del, multiplied by time (t) in the time period
between 0 and t 2, give the ATP hydrolysis rate at the level of
215
myofibrils. Hence the amount of ATP (L1ATP h) hydrolyzed
during the L1t time interval is:
L1ATP h = Acl x t x L1t (14)
for the time period 0 to t], and
L1ATP h = Dcl X (t2 - t) x L1t (15)
for the time period from t] to t2.
Choice of the parameters for modelling
The values ofthe parameters used in the modelling procedure
are listed in Table 1. In the model we directly calculate the
metabolite concentrations at each point along the diffusion
path. Such calculations of diffusion events in a plane sheet
[26] suppose a proportionality between the unit of diffusion
path length and the space volume, corresponding to this
unit. In other words, if the diffusion path lengths in the
myofibrillar, cytoplasmic and mitochondrial intermembrane
spaces (MIS) are chosen to be in proportion 10:2: I, such
proportionality should also hold for the relative volumes of
these compartments. From the morphometric data of Page
[38], taking into account that 81.5% of volume in the intact
rat heart is taken up by cardiomyocytes [38, 39], the volumes
of myofibrillar, cytoplasmic and MIS compartments can be
estimated as 38'0.6, 79.1 and 39.4 ml per kg of wet weight
of tissue, respectively. The fractional volumes (FV, LlL of
tissue) of these compartments are 0.400:0.080:0.040, or
10:2: I. At 1.0 11m radius of cardiac myofibrils [27] the unit
of diffusion path length may be taken as 1.0 I1mll 0 parts
i.e. 0.1 11m. Therefore, the diffusion path lengths in the
myofibrillar, cytoplasmic and MIS compartments can be
taken as 1.0, 0.2 and 0.1 11m, respectively, again at a
proportion of 10:2: 1 (Fig. 1). In the model the minimal
diffusion path (0.1 11m) for the mitochondrial outer membrane
was artificially placed into cytoplasmic compartment (Fig. 1)
to keep the total diffusion accessible space unchanged,
which is 380.6 + 79.1 + 39.4 = 499.1 ml per kg of tissue.
With this total diffusion accessible space of about 0.5 Llkg
of tissue, the known metabolite contents in tissue (mmol/
kg wwt) should be multiplied by two to yield the real
concentration values (mM) in the diffusion spaces (Table 1).
These concentrations were taken from chemical and NMR
data on metabolite contents and concentrations in rat heart
muscle (see ref. [17] and Table 2).
In compartmentalized systems, each enzyme operates
exclusively in its compartment. Therefore, the maximal
activities of enzymes should be normalized to the volumes
of corresponding compartments. This was done by dividing
the value of the maximal activity of enzyme in tissue
(mmol/s/kg wwt) by the fractional volumes (FV) of the
compartments, listed in Table I. Such normalization has been
216

Table 1. Parameters used for modelling

Event Parameter Equation Dimension Value References

Diffusion I, \a-Jd, 7-10 cm'/s DATP=DADP= 1.45 x 10-<> [60)

D pc , = Dcc = 2.6 x 10-<> [60)
Dp;=3.27xl0"' [60)
1,la-ld em 1 X 10-4
7-10 em 1 X 10-5
1, la-Jd, 7-10 1 X 10-5
resm lc,ld 0.0070rO.014 *1)
resg 7-10 3xlO-<i
C'J 1, Ja-ld mM ATp t = 9-ADp t
J J
ADpt=MDP. *2)
J J
PCrt = 23-Cr'
J J
Crt = 2 + L'.PCr. *2)
J J
Pi/ = ADP/ + (Cr/- 2)

ATP ATP h 11--13,15 mmoVkgwwt 0.414 [61]

hydrolysis tl 11-13, IS 3xl0-2
III t2 11-13,15 s 6 x 10-'
myofibrils FV LlL of tissue 0.4

CKmyo V, (ADPproduction) 13 [19) mmol/s/kg wwt 6.886

V_I (ATP production) 13 [19) mmol/s/kg wwt 29.333
K, 13 [19) mM K;, (ATP) = 0 9 [62]
K;b (Cr) = 34.9 *3)
Kb (Cr) = 15.5 [63]
K;d (PCr) =4.73 *4)
Kd (PCr) = 1.67 [63]
K;c (ADP) = 0.2224 *5)
Klb =K;b [60]
FV 0.44

CKmit VI (ADPproduction) 13 [19) mmolls/kg wwt 3.175

V_I (ATPproduction) 13 [19) mmol/s/kg wwt 13.333
Kx 13 [19) mM K;, (ATP) = 0.75 [33]
K;b (Cr) = 28.8 [33]
~(Cr)=5.2 [33]
K;d (PCr) = 1.6 [33]
Kd (PCr) = 0.5 [63]
K;c (ADP) = 0.2048 *5)
K;b = K;b [60]
FV LlL of tissue 0.004 *6)

ATP V I (ATP production) 2 mmolls/kg wwt 3.175

synthase VI (ATPhydrolysis) 2 mmol/s/kg wwt 0.114
K, 2,2a mM K, (ATP) = 0.462 *7)
Kd (ADP) = O. J *7)
Kp (P) = 2.4 *7)
FV LlL of tissue 0.004 *6)
2,2a mM ADP' from eqn. 2b
P'=P
1 113
'x 10
ATP' = 10-ADP'

* I) resm values of 0.007 and 0.014 provide an apparent Km for ADP of 400 and 200 11M, respectively, at state 3 respiration; *2) L'.ATPj and L'.PCrj represent
integralATP and PCr expenditures, respectively, before time t at spatiallocationj; *3) K;b was calculated from thermodynamic equation K;bX K, = K;, x Kb at
K, = 0.4 mM [62]; *4) K;d was calculated from thermodynamic equation K;dX Kc = K;cx Kd with KjKc = 2.833 [62); *5) K;, value has been settled to obtain the
apparent equilibrium constant of CK reaction 6.25 x 10-3; *6) this coefficient takes into account the extremely narrow space between mitochondrial creatine
kinase and ANT. *7) values were taken from our analysis of data of Holian et al. [64] and Gyulai et al. [65].
 217

Table 2. Substrate concentrations in perfused mice hearts at different heart mitochondrial respiration 400 !lM [5, 6] in permeabilized
rates cardiomyocytes or skinned fibers (see Table 1).
Wild type
Most important to relevant modelling, however, is a
Substrate Arrest 400bpm 600bpm reliable determination of the restricted diffusion coefficient
PCr 18.8 I.S 22.7 1.8 20.2 2.2 for ATP from the gap, or microcompartment, between ANT
ATP 6.1 0.6 8.0 0.6 and CKmit. For this, we used the experimental data obtained
PCr/ATP 2.86 0.31 2.S2 0.2 on isolated heart mitochondria when the creatine kinase
Creatine 14.3 I.S 10.3 1.8 12.8 2.2
ADP* 0.034 0.007 0.04 0.01 0.038
reaction was studied under phosphorylating conditions. It
0.009 was shown by Saks et al. [29,32], Soboll et al. [30,31] and
Jacobus and Saks [33] that oxidative phosphorylation shifts
MM-CK deficient the creatine kinase mass action ratio from the equilibrium
Substrate Arrest 400bpm 600bpm value, and also alters ofthe kinetic constant, Ka, for dissociation
PCr 21.9 4.6 26.6 3.1 21.2 2 ofATP from ternary complex E.Cr.ATP, decreasing it from 0.15
ATP 8.0 2.0 11.5 2.3
PCr/ATP 2.79 0.07 2.00 0.18
to 0.014 mM [33]. These changes were well explained by a
Creatine 16.9 4.6 12.2 3.1 17.6 2.0 probability model [24]. In this work we assumed that the
ADP* 0.027 0.009 O.OSI 0.01 0.OS9 0.004 above mentioned changes in the creatine kinase character-
*Va1ues calculated assuming the creatine kinase equilibrium [S7]. All
istics are due to elevated 10calATP concentrations in the gap
concentrations are given in mM on the assumption that 1 g of wet wt between ANT and CKmit , which increases due to the activity
corresponds to 0.348 ml of intracellular water. Total creatine concentration ofANT and decreased diffusion ofATP out ofthis gap. Thus,
was 38.79 mM. we used the model described here giving the different values
ofresg coefficient, and calculated the rates of mitochondrial
done also for the amount ofATP, hydrolyzed in the myofibrils creatine kinase reaction. The calculated results with the curve
during contraction (Table I). obtained on the basis of experimental data. The best fit was
The maximal activities of enzymes in in vivo rat heart found at the value of resg = 3 x 10"" and this value was used
at 37C were taken from multiple biochemical data on in all calculations.
enzyme distribution, their stoichiometries and catalytic
constants (see ref. [17]). The accepted maximal rate ofATP Arrangement of calculations
export by mitochondrial ANT (Table I) at an ATP/02 ratio In this model, we calculate the time- and space-dependent
of 6 corresponds to 71.1 ml of 02 minll 00 g of tissue changes in metabolite levels in heart cell. Based on the
which is the upper limit of the range estimated by Elzinga known metabolite levels at a given time t, we determine the
et al. [40] (54-72 ml of O/minll 00 g oftissue). This value rates of concentration changes as resulting from diffusion
is in good agreement with many other data [17, 41]. With and enzyme activities. These rates, multiplied by the time
the commonly accepted mitochondrial protein content of interval Llt, estimate the concentration changes during each
rat heart, 60 mg/g wwt, the predicted maximal rate of ATP time interval. These values, when summed up with pre-
production by mitochondria in vivo is 3.18 !lmol ATP/mini existing concentration values at each point along the
mg of protein. The experimental value for isolated mito- diffusion pathway, give the new space profile of con-
chondria is somewhat lower, 2.1-2.4 )lmole/minlmg ofprotein centrations at a new time point, t + Llt. The calculation cycles
[42]. The accepted value of maximalATP production by CKmit are repeated until a steady state is reached, when the time
plus CKmyo (42.7 mmol/s/kg wet wt, see Table I), is in good profiles ofthe concentrations of any metabolite are identical
agreement with the experimental value ofVentura-Clapier et al. in each contraction cycle.
[43],9.9+0.5 !lmol/minlmg of fibre protein at30C, if we correct The maximal enzyme activities were constantly normalized
the value by temperature coefficient (1.75, see ref. [44]), the by the fractional volume of the compartment (Table I).
actual substrate concentrations during the measurement [43,44] Besides diffusion, the Runge-Kutta algorithm also was used
and the protein content of fibres (llO mg/g wet wt). Bittlet al. in numerical calculations of the rates of enzymatic events.
[44] estimated the maximal creatine kinase activity in heart as In these cases the steady state equations for enzymatic
40.6 mmoles/s/kg wet wt. velocities (equations 2 and 13 in ref. [32]) were used as
Based on essential similarity in structure and equal differential ones. The Runge-Kutta algorithm provides high
diffusivity of ADP and ATP molecules, the in vivo perme- accuracy and stability in numerical calculations.
abilities of the mitochondrial outer membrane for ATP and The equations for steady state activity were used in the
ADP have been assumed to be equal, resmATP = resm ADP ' For calculations of the immediate changes of enzyme activities
the case of restricted permeability of the outer mitochondrial during short Llt time intervals. This simplified approach is
membrane, the values of these constants were modified to justified for enzymes with rapid equilibrium binding
fit the experimentally determined apparent Km for ADP of mechanism and rate limitation at the catalytic step, such as
218

creatine kinase [45, 46]. The application of steady state
relations for the prediction of dynamic metabolite levels in
the mitochondrial matrix is based on the data of Davies and
Lumeng [47].
The attained data were used for the estimation of diffusion
fluxes through the mitochondrial outer membrane. The flux
(F) of metabolite x was calculated from its concentrations,
Cxll and C x l2, on both sides of the mitochondrial outer
membrane:
(16)
FV = 0.04 is an assumed fractional volume of mitochondrial
outer membrane.
Results
Workload dependence of the creatine kinase fluxes of
wild and MM-CK -/- mice
expression of the mitochondrial creatine kinase is not altered
by elimination ofMM-CK. The metabolite contents were not
different in the wild type and transgenic mice and did not
change with the workload (Table 2).At the arrest, the creatine
kinase flux was identical in both types of the hearts. However,
at the increased heart rate these fluxes became very different.
In the wild type mice heart the increase of workload from 0 to
about 30 000 mmHg x min- l was accompanied with 2-fold
increase in the creatine kinase flux. On the contrary, in the MM-
deficient mice the creatine kinase flux changed only very
slightly, not statistically significantly, and at workload of30,000
mmHg x min- l this flux was statistically significantly lower in
MM -CK deficient mice hearts than in that from wild-type mice
(p < 0.05, Table 3). Thus, the creatine kinases in different
compartments of the heart cells respond differently to the
workload changes. Why?
Analysis of intracellular compartmentalized energy
fluxes: use of the mathematical model

Table 2 shows the results of determination of the metabolite The mathematical model described in the Materials and
contents in the isolated perfused hearts of wild type and MM- methods section was used to analyze the energy and meta-
CK deficient mice at different workloads. In Table 3 the bolite profiles and fluxes in and between different cellular
creatine kinase flux, the rate of phosphoryl transfer from PCr compartments, and in particular, to understand the counter-
to ATP as measured using 31-P magnetization transfer NMR, intuitive experimental 3lP_NMR data given in Table 3.
is given at different workloads ofthe heart from wild and MM-
CK deficient mice. In the wild mouse heart all isoenzymes Verification of the model: comparison with published
of creatine kinase are present and the phosphocreatine experimental data
pathway for energy transfer is fully operative. In the To evaluate the possibilities of the model in describing
transgenic mouse the expression of creatine kinase in the available experimental data, we calculated first the time-
cytoplasmic compartment has been specifically eliminated averaged values of the creatine kinase reaction rates - or
by genetic manipulations (ref. [20] see also the chapters by unidirectional fluxes- for the whole range ATP synthesis and
Steegth et al. and Klaas Nicolay et al. in this volume), utilization rates, known in the literature - from 0-2.5 Ilmoles
leaving only the mitochondrial creatine kinase. Total ATP per sec per g of wet wt (mass) [41, 50-52]. Time-
creatine kinase activity in these MM-CK -/- mouse heart averaged means that the fluxes which change very significantly
was 26% of that of the wild type, in complete accordance within the contraction cycle (see below) were averaged for
the earlier estimations of the content of mitochondrial the whole cardiac cycle and the mean values for every
creatine kinase in mammalian hearts [48, 49]. Thus, the workload (ATP synthesis rate) were calculated. This
corresponds to the procedure which is commonly used in
the NMR studies of creatine kinase fluxes where the 3lp
Table 3. Creatine kinase fluxes in MM-CK deficient and control mice hearts signal is averaged over many contraction cycles. Figure 2
gives the calculated fluxes for the rat heart. In the normal
Heart type RPP,IOJ Per, k for Flux <1Flux
heart with all creatine kinase isoenzymes fully active, total
nunHgx mM (s ') (mM s ') (mMs')
mm-' steady state creatine kinase flux catalyzed by both isoenzymes,
CK flux myo+mlt. ' in the forward (PCr synthesis) and reversed
Wild 0 13.9 l.l 0.240.OS 3.8 0.5
(ATP synthesis) directions is equal (ref. [53], see the chapter
33.02.6 JS.O 1.6 0.39 0.04 6.6 1.3 2.8
MM-CK 0 18A3.3 0.170.02 3.5 OA
by Klaas Nicolay in this volume). This conclusion is equally
deficient 29.72.8 J9.72.3 0.220.03 4.3 0.6 0.8 valid for the equilibrium and for the steady state [53]. This
total unidirectional flux is shown in Fig. 2 by a thick line,
Values are expressed as means S.D. Number of experiments was from 5-7.
and it changes by factor of four over the range of ATP
Concentrations are given in mM in intracellular water. V m" for the creatine
kinase reaction was in wild type mice 29 3.0 and in MM-CK deficient mice production-utilization rates studied. When each CK iso-
7.7 0.6 mM x s-', respectively (at 37C). RPP means rate - pressure product. enzyme is separately analyzed, something more interesting
 219

Total in the cell is seen. These calculations are shown for myoplasmic
(myofibrillar) creatine kinase in Fig. 2 forreversed (REV, dashed
4 A line), and forward (FORW, dotted line) directions. While at a
ATP production-utilization rate of zero both these fluxes are
3 equal for both individual creatine kinase isoenzymes, indica-
tive of thermodynamic equilibrium, the difference between

-
.,...
I-
E
2

1 _ -El -
- 0 - _D
these fluxes progressively increases with elevation of the
workload. That means that the creatine kinase reaction is
shifted more and more out of equilibrium. The deviation from
~ TG-CKmit equilibrium varies through the contractile cycle (see below).

-
~

.,......
IU)
...
C)
0
Through CKmyo
The difference between myoplasmic CK fluxes, REV -myo-
FORW-myo, is the steady state rate ofATP regeneration by
myofibrillar CK, and Fig. 2 shows that it is equal to theATP
0 4 B production - utilization rate at any value of the latter
parameter. Thus, myofibrillar creatine kinase delivers, at the

-
E
E expense of PCr, all ATP necessary for contraction.
3
U)
On the other hand, the difference between the total CK flux
CD catalyzed by both isoenzymes (thick line in Fig. 2) and FORW-
>< 2
::::J myo (Fig. 3) is the rate of the forward CK reaction in the
0+-
mitochondria, i.e. the steady state rate of the aerobic synthesis
:::: 1 ofPCr. Again, Fig. 2 shows that this rate linearly increases with
0
the workload - that is, with theATP production rate. Analysis
as 0 of the data in Fig. 2 shows that practically for the whole range

-
c::
0 Through CKmit of the rates studied the rates of PCr production in the
0 mitochondrial intermembrane space and ATP production in

c
CD
~ 4 mitochondrial matrix are almost identical, as expected on the
'0 basis of input parameters.
c::
::::> 3 These calculated values can be compared with the results
of experimental studies. Most data on the quantitative
FORW-wild determination of the relationship between myocardial
2
creatine kinase fluxes and workload come from Ingwall's
TG-CKmit laboratory [SO-S2]. The agreement between our calculated
1 _ -El -0- _D
and their experimental data seems to be almost quantitative,
REV-wild in numbers. Indeed, Bittle and Ingwall [SO] and Perry et al.
OC~~~:;;;;~~ [Sl] have shown that in isolated mature rat and rabbit hearts,
0 1 2 an increase in the workload results in an increase in the
-1 -1 creatine kinase flux determined by a 31 P-NMR magnetization
Workload (mmol*s *(kg wm) )
transfer method. In the study by Perry et al. [SI] this
Fig. 3. Simulated workload dependencies of unidirectional energy fluxes augmentation of the flux was about four-fold for the same
through CK in the heart cells of wild type (continued lines) or transgenic range of ATP synthesis rates as used in our calculations.
mice with functioning CKmit (dashed lines, TG-CKmit) or CK m," (dotted lines,
TG-CKmyo). The concentrations of metabolites are shown in Table 2. The
More important is the estimation of the heart mitochondrial
parameters used in the simulations are indicated in 'Materials and methods'. creatine kinase flux by Zahler and Ingwall, based on a
The diffusion restrictions in the mitochondrial outer membrane correspond consideration of magnetization transfer NMR spectroscopy
to apparent Km for ADP of 400 uM (wild, res ATP = res ADP = 0.007), 200 uM and mathematical modelling assuming separate mitochondrial
(knocked-out CK myo ' res ATP= res ADP = 0.0175), 100 uM (knocked-out CKmn and cytoplasmic ATP pools in the cells [S2, S4]. Using this
resA"IP = res ADP = 0.035). The values of steady state fluxes are integrated
through the entire cardiac cycle for the forward (thin lines, FORW -) or the
method it was suggested that the mitochondrial creatine
reverse (thick lines, REV-) direction of the CK reaction. It is important to kinase flux almost linearly increases with the increase in the
note that for the complete system with both isoforrns ofCK (wild type) the ATP synthesis rate, and the values of ATP production flux in
total fluxes through them are equal in both directions, while the fluxes for the mitochondrial matrix and the PCr synthesis flux were
particular isoforrns are different in the forward and reverse directions. The
practically equal both changing from 0.5 (arrested heart) to
inequality of forward and reverse mean unidirectional fluxes in particular
creatine kinase isoforms indicates their steady displacement from equilibrium.
3 Ilmoles per g dry wt per sec at the maximal workload [54].
In the transgenic animals the modelled mean integral unidirectional fluxes This is in agreement with close functional coupling between
are always equal in both directions. mitochondrial creatine kinase and adenine nucleotide trans-
220
locase [54]. And this is exactly what we have found in our
model calculations on the basis of the model (Fig. 2). Thus,
the model describes well the published experimental data.
Therefore, it can be reasonably used for analysis of the
paradoxical experimental results described in Table 3.
Quantitative analysis of the transgenic mice
experiments
The application ofthe model described was used for detailed
analysis of the data given in Table 3. The results of these
analysis are shown in Figs 3-8. The parameters ofthe model
were the same as used for rat heart, assuming similarities
of enzyme activities, compartmentation etc., but total
substrate contents used were taken from experiments with
the isolated mouse heart shown in Table 2. Fig. 3 gives the
time-averaged unidirectional fluxes as a function of the
workload on the heart, but this time these are the calculated
dependencies. The agreement with the data given in Table
3 is remarkable. In Fig. 3A the thick line is the total
unidirectional flux calculated for the wild type mouse heart
using experimentally determined metabolite contents shown
in Table 2 (except for ADP, which is calculated as indicated
below). The dashed thick line is the total unidirectional flux
calculated for the transgenic MM-CK -/- with only mito-
chondrial creatine kinase in operation. The model predicts
that in this case the flux is practically independent from
workload. And this is exactly what we see in the experiments
(Table 3).
The dotted line in Fig. 3A is the prediction of the creatine
kinase flux in mice which are deficient in mitochondrial
creatine kinase, when only MM-CK is active. In this case the
flux is more dependent on the workload as expected.
Figure 3B gives calculations of the fluxes for the myo-
plasmic MM creatine kinase in the wild type mouse heart.
The explanations for this Fig. are identical to those given
above for Fig. 2.
Very interesting modelling results are shown in Fig. 3C,
i.e. a calculation of the creatine kinase fluxes in the
mitochondrial compartment for wild type and MM-CK
deficient mice. In wild type myocardium mitochondrial
creatine kinase progressively shifts from equilibrium with
elevation of workload: the rate of PCr production (forward
reaction) increases and that PCr utilization decreases to
practically zero. Thus, at increased workloads the creatine
kinase reaction in mitochondria becomes practically
unidirectional, and is driven by the adenine nucleotide
translocase in the direction of aerobic PCr synthesis with a
rate exactly equal to the rate of ATP production. It is also
clear from Figs 3B and 3C that both mitochondrial and MM-
CK are not in equilibrium if the ATP production rate is
different from zero.
The dashed line in Fig. 3C shows the mysterious behavior
of mitochondrial creatine kinase when it is the single
isoenzyme of creatine kinase left in the cell as the result of
genetic manipulation: its response to workload changes is
drastically reduced in comparison with that in the wild type
heart in which the mitochondrial creatine kinase cooperates
with MM-CK in the complete phosphocreatine shuttle.
The mystery of the behavior of mitochondrial creatine
kinase, however, is easily solved by the analysis of the
intracellular events in various compartments when the model
is used for calculations of compartmentalized energy fluxes
and changes in the metabolite profiles of the cell within one
contraction cycle.
Figure 4 in fact gives a key for the understanding: as
usually in the energy metabolism ofthe cell, the cytoplasmic
ADP concentration is the decisive factor. This Fig. shows
the calculated metabolic profiles within a contraction cycle
for low and medium workloads for three types of mouse
cells: wild type (thick lines), MM -CK -/- (dashed lines) and
mitochondrial creatine kinase (dotted lines) knock-outs. The
ADP concentration changes very significantly in the systolic
phase of contraction, and the amplitude of changes increases
with the workload (Figs 4A and 4D). This strongly changes
all ADP-dependent parameters such as free energy changes
for ATPase and creatine kinase systems etc. If MM-CK is
knocked out, contraction results in a very rapid rise of the
free ADP concentration, which may exceed the initial level
by an order of magnitude at high workload (Fig. 4D).
Correspondingly, there is much less creatine released in
the hearts (Fig. 4F). The inorganic phosphate, which
concentration changes are not very different between these
hearts, in the MM-CK -/- heart comes from ATP, while in
the wild type heart it comes mostly from PCr (compare Figs
4E, 4D and 4F). While changes in the Pi concentrations may
be significant due to its concentration of circa 0.5-1 mM,
the relative changes in PCr, creatine and ATP levels in the
contraction cycle do not exceed 3-4 % of their initial
value. By the end of diastolic phase the ADP as well other
metabolites concentrations return to their initial levels.
The creatine kinase fluxes in the different cellular
compartments in these three types of mouse heart are given
in Fig. 5 for two different workloads. Unidirectional creatine
kinase fluxes with positive values are those of ATP synthesis
(reversed CK reaction), negative values are those of ATP
utilization for PCr production. Figs 5A and 5D show the
fluxes in the myofibrillar (myoplasmic) compartment at two
different workloads. The flux in the direction ofATP synthesis
follows the contraction cycle by rapid rephosphorylation of
ADP, that keeping its level low (Fig. 4D) at the expense ofPCr,
as expected. While in the wild type hearts MM -CK is clearly
out of equilibrium in the systolic phase, the fluxes in both
directions become largely identical in diastolic phase (quasi-
equilibrium). The time-averaged fluxes given in Fig. 3 were
calculated from the area between the lines corresponding to
flux in consideration and abscissa axis.
 221

.. ,
,,
-
300 300
\
A

\
,, D

--
\

. ,,
~
~ 200 200
.... , ,
Q. ,, '\
:", '\

.. , ..
.. ..
-
'\
0 ''-
.. .. ..
'\
100 I '\ 100 ''-
c( ("
I
~.""'" ........... I
""' ,1", ,.""
,. . I,. I I I , I I I I I I I I ~,"'.,

--
0 0

2
B
--
:::I 200 200

ii:
.5 100 100
CD
Q
C
Cd
s::.
0
0 0

-
---...
2
:::I 200 C 200

0
.5 100 100
CD
Q
C
Cd
s::.
0
0 0
0 60 120 180 0 60 120 180

Time (ms)
Fig. 4. Simulated dynamics of metabolite levels in the myofibrillar core of the heart cell of wild type (solid lines) or transgenic animals with knocked-out CKmyo
(dashed lines) or CK . (dotted lines). Data were calculated for two workload levels: moderate, 0.5 mmol*s-I *(kg wmtl, (A-C) and high, 1.0 mmol*s-' *(kg wmt
" (D-F). For Pi andC~onlythe changes in concentrations are shown. The diastolic levels of Pi were: 1.151 mM (B) and 3.040 mM (E) for wild type, 0.251 mM
(B) and 0.398 mM (E) for knocked-out CKmyo' 6.070 mM (B) and 9.895 mM (E) for knocked-out CKmyo types. The diastolic levels ofCr were: 3.143 mM (C) and
5.027 mM (F) for wild, 2.228 mM (C) and 2.359 mM (F) for knocked-out CKmyo ' 8.045 mM (C) and 11.858 mM (F) for knocked-out CKmyo types. The parameters
used in the simulations are indicated in 'Materials and methods'. The diffusion restrictions in the mitochondrial outer membrane correspond to apparent Km
for ADP of400 uM (wild, res ATP = res ADP = 0.007), 200 uM (knocked-out CKmyo ' res ATP = res ADP = 0.0175), 100 uM (knocked-out CKmt " res ATP = res ADP 0 035).

In the hearts with only MM-CK present these areas phosphorylation via adenine nucleotide translocase. This rate
approach each other, showing the quasi-equilibrium state of increases with increase of workload which also drives the
the enzyme during the whole contraction cycle. This is due system further out of equilibrium.
to increased rates of PCr production (negative fluxes) In MM-CK -/- hearts we see that the rate of ATP pro-
probably due to an increased average creatine concentration. duction by mitochondrial creatine kinase is increased in
Most relevant to our case are, however, Figs SB and SC, response to contractile activity, in a workload dependent
which describe the experimental conditions given in the manner. This is due to the very significant increase in
Table 3, i.e. changes in the fluxes in the mitochondrial cytoplasmic ADP concentration (Fig. 4). Note that in this
compartment in wild type and MM-CK -/- hearts. In wild case the rate of PCr production changes very slowly with
type hearts mitochondrial creatine kinase produces PCr with increase in workload, in contrast to wild type heart ( Figs
a practically constant rate and is strongly and permanently SB and SE). However, the amplitude of ATP production -
out of equilibrium due to its coupling with oxidative cytoplasmic ADP rephosphorylation - by mitochondrial
222

Unidirectional fluxes through CKmyo

6 6 r---~----~--------~

A D
3 3
'11'.1.,'1111111111111

--
"11111111111111'111111

...,
O~------------------~ O~------------------~

" ' 1 11 ' 1 1 " 1 ' 1 1 " 1 1 1 ' 1 1 1 1 1 1 , 1 1 " 1 1 1

E , ".,1"'1".111111111,.1111111111

-3 '----..---~--,....----' -3 L-~_~_~--,-~~........J

-'.
~
Q Unidirectional fluxes through CKmit
~ ,-,
... ,'''''-''-- ...... B '" . .... .... E
,,
'
o ~ ...
, --- .. _-- ~' -.. _---
-
E
E

"iii
0

----- --.
o ,t:::::::::::::=::=::==::j

_.----------------
-
CD -1 -1
r.
c
~ Net fluxes through ATP synthase
,------------------, ,---~-------------.
a.
~ 2 c 2 F
,, .... --
:~,
.. .. ..
- ..
.. ., -.: .........
..... ,,'~ .......... ... ""
.. ...........
7 .......................:.. :,'~
,

O'--~~---.-------.-~_--.-J

o 60 120 180 0 60 120 180

Time (ms)
Fig. 5. Simulated dynamics of energy fluxes in different compartments of the heart cells of wild type (solid lines) or transgenic animals with knocked-out
CKmyo (dashed lines) or CKmit (dotted lines). Data were calculated for two workload levels: moderate, 0.5 mmol *S-I *(kg wm)-I, (A-C) and high, 1.0 mmol *S-I
*(kg wm)-I, (D-F) ATP hydrolysis rates in the myofibrils. Fluxes for forward direction ofCK reaction (PCr production) are indicated by negative values of ATP
synthesis, and by positive values for the reverse direction. The model parameters were as in Fig. 3.

creatine kinase chances in response to increased workload.
Because of increased ATP production, creatine kinase
approaches the quasi-equilibrium state (the average values of
fluxes in both directions become equal). Thus, progressively
increased ADP production in the contraction cycle, due to
absence of MM-CK in myofibrillar space, in response to
elevation of the workload slows down the rate of aerobic PCr
production, due to the reversal of the mitochondrial creatine
kinase reaction in the direction of ATP synthesis. This
direction, however, is a futile one for mitochondrial creatine
kinase, since production of ATP is the sole function of
mitochondrial oxidative phosphorylation itself. In other
words, in this case mitochondrial creatine kinase is not
performing its normal function and does not respond to
increases to the workload, but rather becomes indifferent to
what happens in the cell. It may only be, in some relaxed
manner, of some functional relevance to give some ATP
buffering. And this is exactly what has been observed in 31p_
NMR experiments on MM-CK deficient heart (Table 3).
Finally, Figs 5C and 5F show that in all cases the rate of
mitochondrial ATP synthesis responds normally to the
elevation of workload - in all cases the area below the ATP
production curve is proportionally increased. It is interesting
to note, however, that the oscillations in the ATP production
rate are more significant in the absence of mitochondrial
creatine kinase (dotted lines in Figs 5C and 5F) in spite of
the largest variations in ADP in the cytoplasm in the ab-
sence ofMM-CK (dotted lines, see Figs 4A and 4B). This
 223

emphasizes the role of mitochondrial creatine kinase and Wild

creatine in the regulation of mitochondrial oxidative phos-
phorylation. 4 ATP+PCr export 4
Figure 6 summarize all the conclusions made. The energy

-- --
fluxes between mitochondria and cytoplasm in the three
different types of hearts are depicted. In the wild type 2 2
cardiomyocytes, with all creatine kinases present and fully ..... .....
I
I
active, the energy is exported from the mitochondrion by 0 0 E

---
E PCr export
PCr almost exclusively, due to high permanent rate of aerobic
:: ::
-
- -
PCr production in mitochondria (dotted line in Fig. 6A). Since CKmit
C) C)
the rate of PCr utilization in the myofibrils is equally high, -2 ~
~
the ADP concentration in the cytoplasm remains low (Fig.
.....
TG-CKmyo
4). The functional relationship between mitochondrial and ..... I

B
I

cytoplasmic processes is principally changed if any of the VJ VJ

4 4
isoenzymes of creatine kinase is switched off. The most 0 0

- -
dramatic changes are observed when the cell is deprived of E E
MM-CK in the cytoplasm (Fig. 6C). In the case of the E 2 \ 2 E
I

-
absence of mitochondrial creatine kinase (Fig. 6B) there is I PCr export \

-
)( )(
. some apparently paradoxical production of PCr in the ::::J
::::J 0 0
intermembrane space. This is, however, a purely diffusional I \
"
~K.!!!yo_
process, since MM-CK is reversed in the diastolic phase and c: ~

...
0

produces PCr, and it follows ATPase activity only in systolic
phase and uses PCr for rapid rephosphorylation of ADP,
producing creatine. This is, in turn, rephosphorylated, but
with a low rate, to PCr during the diastolic phase (dashed line
in Fig. 6B). If integrated throughout the contraction cycle,
net PCr exchanges with the intermembrane space become
zero. Obviously, the rates ofPCr import and export are low,
and the energy is transported from the mitochondria in the
form of ATP, and probably directly enters the adenyl ate
kinase pathway, in details described by Dzeja et al. in this
volume.
In the case of MM-CK deficiency, the situation is
completely opposite with respect to mitochondrial PCr
export. In the diastolic phase PCr is produced and exported.
However, in the systolic phase PCr is rather used in the
mitochondrion due to reversal of the creatine kinase
reaction (dashed line in Fig. 6C) caused by the very high peak
values ADP (Fig. 4). And again free energy is exported
exclusively via ATP, since net PCr approaches zero.
Thus, in the absence of the complementary part of the
PCr circuit, both MM-CK and mitochondrial CK start to
fulfil the dual function of PCr production and utilization
in the quasi-equilibrium manner within the same con-
traction cycle. MM -CK keeps the cytoplasmic ADP concent-
ration reasonably low, at the expense ofPCr. However, in the
MM-CK knock-out, mitochondrial creatine kinase is com-
pletely switched off (it is more correct to say that it cannot
function properly because of high cytoplasmic ADP levels
and does not significantly respond to changes in workload.
This is the explanation of the quasi-paradoxical effect,
observed in the experiments with the transgenic mice, des-
cribed in the Table 3.
-
0 -2
VJ
::::J TG-CKmit Q)
Z
ATP export
0 4 4
2 2
0 0
-2
0 60 120 180
Time (ms)
Fig 6. Simulated dynamics of diffusional A TP and PCr export through the
mitochondrial outer membrane in the hearts of wild type (A) or transgenic
animals with functioning CK my " (B) or CK,mt (C). These data are compared
with net fluxes (thick interrupted lines) through CK mit (A, C) or CKmvo (B),
The shaded area indicates the PCr outflux. The model parameters were as in
Fig. 2, workload in all simulations was maximal, 2 mmol*s I *(kg wm)l,
Modeling of the influence of the restricted permeability
of the outer mitochondrial membrane porin pores on
ADP concentration and energy fluxes
Recent experimental work on the permeabilized cardio-
myocytes or hepatocytes and also on skinned muscle fibers
has shown that the permeability of the outer mitochondrial
porin pores for ADP is significantly decreased in the cells in
vivo (see the chapter by Saks et al. in this volume). This
observation was generalized by the Bordeaux group by
showing the same phenomenon in the yeast cells, who also
224
confirmed our conclusion ofthe role of the porin pores in the
control of the diffusion of ADP across the outer mitochondrial
membrane using porin-deficient mutants (see the chapter by
Averet et al. in this volume). Probably, this kind of control is
due to the expression of some protein controlling the porin
pores in the cells [7, 17]. Veksler et al. have studied the
respiration regulation in the skinned fibers from the MM-CK
deficient mice and observed increased permeability of the
mitochondrial outer membrane for ADP, as expressed by
several-fold decrease of the apparent Km for ADP [55]. This
is why we decided to model the influence the changes the
permeability ofthe outer mitochondrial membrane for adenine
nucleotides on the most important parameters of the energy
metabolism ofthe cells: cytoplasmic ADP concentration and
energy fluxes from mitochondria. The results of this modelling
are shown in Figs 7-9. Figure 7 shows that increasing the
outer mitochondrial membrane permeability significantly de-
creases the peak value of the ADP concentration in the
systolic phase in the MM-CK knock-out mice. This is related
to more rapid rephosphorylation of ADP in mitochondria,
that increasing the export ofATP, Fig. 8A, but decreasing that
of PCr, Fig. 8B. This result confirms that alteration of the
outer mitochondrial membrane permeability for ADP is a
reasonable mechanism of adaptation in the MM- CKknock-
out cells: by this mechanism the cells maintain lower
cytoplasmic ADP concentrations.
It is interesting that in the cells deficient with respect to
mitochondrial creatine kinase the alteration of the outer
membrane permeability has no effect on the already high
export of ATP and low diffusional export of PCr - Fig. 9.
In more details, these effects are described in our recent
publication [56].
Discussion
The results of this paper show that the mathematical model
of the compartmentalized energy transfer in cardiac cells
explains well the paradoxical experimental results of 31p_
NMR flux determination in the isolated hearts from transgenic
mice deficient in MM CK. The transgenic technique allows,
for the first time, to manipulate the creatine kinase activity
in different cellular compartments in vivo, and thus to
study the physiological importance of the intracellular
compartmentation of the creatine kinase isoenzymes. The
experimental results cannot be explained by the traditional
equilibrium concept of creatine kinase. Indeed, according
to this concept (assuming that 26% of the total CK activity
remaining in the MM -CK -/- hearts is sufficient to maintain
creatine kinase equilibrium and that there is no diffusion
limitation for ADP transport in the cell), calculated ADP levels
are the same in wild type and MM-CK deficient hearts (see
Table 2), and since all experimentally determined metabolite
concentrations are the same in wild-type and MM-CK deficient
hearts (Table 2), the fluxes also should be equal in both types
of heart. The experimental data, however, clearly show that
this is not the case. From the traditional viewpoint, this is a
paradox. Our analysis shows, however, that it is only an
apparent paradox, well explained on the basis of the compart-
mentation theory.
The mathematical model of compartmentalized energy
transfer, presented in this chapter, is based exclusively on
experimentally determined parameters (i.e. enzyme activities,
kinetic characteristics of the enzymes, the distribution of
creatine kinase isoenzymes - compartmentation - in the
cell) and it uses experimentally determined values of
intracellular (cytoplasmic) substrate concentrations. It
calculates the value of only one unknown parameter - ADP
concentration in the cell, and from the input parameters
predicts all energy fluxes and metabolite profiles. Thus,
this is a very realistic model, accounting for all known
experimental data. Its application for the analysis of the
energy metabolism in normal muscle cells shows, first, that
in the working muscle the creatine kinase reaction is not in
equilibrium but in the dynamic state. The classical equilibrium
concept completely ignores the compartmentation of the
creatine kinase isoenzymes, and considers the total activity
ofCK as the only important parameter [2, 27,57]. However,
considering the beautiful structural work ofTheo Wallimann' s
laboratory regarding mitochondrial creatine kinase and its
molecular physiology, and the review of molecular genetics
of this isoenzyme from Arnold Strauss laboratory (see the
corresponding chapters in this volume), the models, based on
the simple equilibrium concept of creatine kinase and ignoring
the creatine kinase isoenzyme differentiation in normal heart
muscle cells, are clearly gross oversimplification. The
mitochondrial isoenzyme of creatine kinase has a special
function in cellular energetics and plays a vital role in the
regulation of mitochondrial respiration. It appeared, for a
functional purpose, in the cells before gene duplication
gave rise to the appearance of different MM and BB iso-
forms of creatine kinase (ref. [58], see the chapter by Qin
et al. in this volume).
This work shows that the quantitative, mathematical model
based on firm experimental date and accounting for the
compartmentation of the creatine kinase isoenzymes,
including mitochondrial creatine kinase and its functional
coupling with the adenine nucleotide translocase, is in
concordance with the published relationship between energy
fluxes through creatine kinase and workload (see explanation
to the Fig. 3). Furthermore, the model gives full explanation
of the experimental observations on the isolated hearts of
transgenic mice (see Figs 3---{)).
Interesting conclusions which can be made concern the
behavior of the creatine kinase system of the cell. The most
interesting finding of our analysis is that in the fully active
 225

500,-------------------------------~

" ,,
,
400 \
\
\

,
\
\

~
\

300
\
\

\, OM permeability:
'-' ,,
,,
,
'''"low
""
.... med ""
100
...~!~h
....................
"
'" ' - '

'.
O+-~--~--~-,--~--~_.~~~~=~~~=~~

o 60 120 180
Time (ms)
Fig. 7. Modeling the influence of alteration of the permeability of the mitochondrial outer membrane for ADP on the cytoplasmic ADP concentration in the
MM-CK deficient mice. Thick solid line (control) corresponds to the wild type cardiomyocytes. High permeability for adenine nucleotides, ADP and ATP,
means that nucleotide diffusion in membrane segment does not differ from that in cytosolic segment; low permeability means that the diffusion constant in the
membrane segment is taken to be 0.0175 of the corresponding value in the cytosolic segment (this corresponds to apparent Km for ADP equal 140 JlM, as
described in ref. [55]); and intermediate permeability means that the diffusion constant in the membrane segment is increased two times as compared to the low
permeability state.

B
4

OM permeability:
2

-2

o 60 120 180 o 60 120 180

Time (ms) Time (ms)
Fig. 8. ATP (A) export and PCr (8) export from mitochondria into cytoplasm at different permeability of mitochondrial outer membrane for adenine nucleotides
in cardiomyocytes deficient in MM-CM. Explanations as in Fig. 7.
226
4 4
OM permeability:
2 2
-2
o 60 120
Time (ms)
Fig. 9. ATP (A) and PCr(B) export from mitochondria into cytoplasm in cardiomyocytes deficient in mitochondrial creatine kinase. Explanations as in Fig. 7.
PCr pathway for energy transfer which in normal, wild type
cells transfers almost all of the free energy from the
mitochondrion into the cytoplasm, the mitochondrial and
MM creatine kinase isoenzymes cooperate in a coordinated
manner in the steady state, out of equilibrium, and inhibition
of one of them immediately influences the function of the
other species (Fig. 10). Thus, mitochondrial creatine kinase
can function normally, in the direction of aerobic PCr
production, only if the myofibrillar MM-CK is fully active.
In this case mitochondria respond to changes in cytoplasmic
creatine concentrations as a signal from the myoplasmic
space (see Fig. 4), and mitochondrial creatine kinase
functions as an amplifier of this signal in the functionally
coupled creatine kinase-ANT -oxidative phosphorylation
systems [8], and the normal end-product of mitochondrial
energy free energy liberation is PCr. However, if the signal
comes to the mitochondria as a manifold increased cyto-
plasmic ADP level, mitochondrial creatine kinase turns to
function in the futile cycle not related to the workload. This
is the case of MM-CK deficient mice (Fig. lOB). The high
cytoplasmicADP level may however, switch on the adenylate
kinase pathway (see Chapter by Dzejaet al. in this volume).
In some sense, MM-CK in the mitochondrial creatine kinase
deficient mouse (theoretical case in Figs 4-8) seems to be
more effective, if it is the only CK (Fig. 10C): it keeps
cytoplasmic ADP levels lower (see Fig. 4) than those in MM-
CK -/- mice. This may give some explanations to the con-
A B
OM permeability:
Control
180
clusion from structural analysis that this was the MM-CK which
appeared first during evolution [58]. The appearance of
mitochondrial isoform of creatine kinase completed the
formation of the PCr circuit as a very effective energy
transfer and feedback regulation system, and this process is
repeated again and again in the CK - related developmental
changes during maturation of any individual animal [59].
The factors participating in and regulating the expression
of different creatine kinase isoenzymes are described in
chapter by Qin et al. of this volume.
Thus, the combination of transgenic technology and
quantitative, mathematical modelling of the compart-
mentalized energy transfer processes, described in this
chapter, gives valuable information of the integrated and
highly organised intracellular processes of free energy
transduction and their regulation. And it further evidence that
in normal, wild type heart cells the CK-PCK pathway (PCr
circuit, according to Wallimann) plays a central role in the
energy transfer and metabolic regulation.
For the analysis of other mathematical models of energy
transfer, see the chapters by Saks and by Kemp et al. in this
volume.
 227

AI

'\M---4- Cr - - _

~::t-- PCr -.-~~

81

-.::!~:;..;....-- ATP -----.....

Cr

J":;;Qo--- ATP --I..-c;:......

Fig. 10. Schematic presentation of different states of the CK-PCr system in the heart: A - wild type mouse, PCr circuit is fully active and the
creatine kinase isoenzymes function in well-coordinated manner out of equilibrium in the steady state; adenine nucleotides are mostly functionally
compartmentalized in mitochondria and myofibrillar spaces. Therefore, the creatine kinase flux increases linearly with elevation of the workload.
B - MM-CK -/- mouse, the mitochondrial creatine kinase is performing a futile cycle (see the text) and to some extent buffering changes of ATP,
in quasi-equilibrium manner; energy is transferred by adenylate kinase system or some other pathways in energy transfer network. C - theoretical
(not studied experimentally in this work) case of mitochondrial creatine kinase knock-out, MM-CK is in quasi-equilibrium performing dual
function of PCr production in diastolic phase and its utilization in the systolic phase of the heart cycle.
228
Acknowledgements
This work was supported by INTAS grant 94-4738, Estonian
Science Foundation grant N 1938, Russian Foundation for
Basic Research grant N 97-04 49554).
References
1. Hochachka PW, Matheson GO: Regulating ATP turnover rates over
broad dynamic work ranges in skeletal muscles. J Appl Physiol 73:
1697-1703,1992
2. Kushmerick MJ: Skeletal muscle: A paradigm for testing principles of
bioenergetics. J Bioenerg Biomembr 27: 555-569, 1995
3. Sahlin K: Control of bioenergetic processes in contracting human
skeletal muscle. Biochem Soc Trans 19: 353-358, 1991
4. Balaban RS: Regulation of mitochondrial oxidative phosphorylation in
mammalian cells. Am J Physiol258: C377--C389, 1990
5. Wan B, Dounen C, Duszyusky J, Salama G, VaryTC, Lanoue KF: Effect
of cardiac work on electrical potential gradient across mitochondrial
membrane in perfused hearts. Am J Physiol265: H453-H460, 1993
6. Saks VA, Ventura-Clapier R, (eds).: In: Cellular Bioenergetics: Role of
Coupled Creatine Kinases. Kluwer Academic Publishers, Dordrecht-
Boston-London, 1994,pp 1-346
7. Saks VA, Kuznetsov AV, Khuchua ZA, Vasilieva EV, Belikova YO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis: Involvement of mitochondrial-cytoskeletal
interactions. J Mol Cell Cardiol27: 625-645, 1995
8. Saks VA, Ventura-Clapier R,Aliev MK: Metabolic control and metabolic
capacity: Two aspects of creatine kinase functioning in the cells.
BiochimBiophysActa 1274: 81-88,1996
9. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger H:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands: The
'phosphocreatine circuit' for cellular energy homeostasis. Biochem J
281: 21--40,1992
10. Koretsky AP: Insights into cellular energy metabolism from transgenic
mice. PhysiolRev 75: 667-688, 1995
II. Chance B, Williams GR: Respiratory chain and oxidative phos-
phorylation. Adv Enzymol17: 65-134, 1956
12. McGormac JC, Halestrap AP, Denton DM: Role of calcium ions in
regulation of mammalian intramitochondrial metabolism. Physiol Rev
70:391--425,1990
13. GuntherT, Pfeifer DR: Mechanisms by which mitochondria transport
calcium. Am J Physiol 258: C755--C786, 1990
14. Fiolet JWT, Baartscheer A, Schumacher CA: Intracellular [Ca2+] and
V 02 after manipulation of the free energy of the Na/Ca exchanger in
isolated rat ventricular myocytes. J Mol Cell Cardiol27: 1513-1525,
1995
15. Lipp E, Niggli P: Subcellular features of calcium signalling in heart
muscle: What do we learn? Cardiovasc Res 29: 441--448, 1995
16. Penman S: Rethinking cell structure. Proc Nat! Acad Sci USA 92: 5251-
5257,1995
17. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova 0, Yu, Kuzuetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration -a synthesis. Mol Cell Biochem 133/134: 155-192, 1994
18. Wyss M, Smeitnik J, Weyers RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism Biochim Biophys
Acta 1102: 119-166, 1992
19. Mitchell P: Compartmentation and communication in living systems.
Ligand conduction: A general catalytic principle in chemical, osmotic
and chemiosmotic reaction systems. Eur J Biochem 95: 1-20, 1979
20. Van Deursen J, Heerschap A, Oerlemans F, Ruitenbeek W, Jap P, Ter
Laak H, Wieringa B: Skeletal muscles of mice deficient in muscle creatine
kinase lack burst activity. Cell 74: 621-631, 1993
21. Van Dorsten FA, Frnter R, Bijkerl<: M, Wallimann T, Nicolay K: The in vitro
kinetics of mitochondrial and cytosolic creatine kinase determined by
saturation transfer 31P_NMR. Biochim Biophys Acta 1274: 59-66, 1996
22. Forsen S, Hoffman RA: Exchange rates by nuclear magnetic
multiple resonance. III Exchange reactions in systems with several
non-equivalent sites. J Chern Phys 260: 1189-1196,1964
23. Lamprecht W, Trautschold I: In: Methods of Enzymatic Analysis.
Academic Press, NewYork,1974,pp 1772-1776
24. Aliev MK, Saks VA: Quantitative analysis of the phosphocreatine
shuttle. I A probability approach to the description of phosphocreatine
production in the coupled creatine kinase-ATP/ADP translocase-
oxidative phosphorylation reactions in heart mitochondria. Biochim
BiophysActa 1143: 291-300,1993
25. Aliev MK, Saks VA: Mathematical modeling of intracellular transport
processes and the creatine kinase systems: A probability approach.
Mol Cell Biochem 133/134: 333-346, 1994
26. Crank J: The Mathematics of Diffusion, 2nd edn. Clarendon Press,
Oxford,I975,pp 137-144
27. Meyer RA, Sweeney HL, Kushmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol246: C365--C377, 1984
28. Ebert K, Ederer H: Computeranwendungen in der Chemie, VCH
Verlagsgesellschaft mbH, Weinheim, Section 9.8, 1985
29. Saks VA, Kuzuetsov AV, Kupriyanov VV, Miceli MV, Jacobus WE:
Creatine kinase of rat heart mitochondria. The demonstration of
functional coupling to oxidative phosphorylation in an inner membrane-
matrix preparation. J Bioi Chern 260: 7757-7764, 1984
30. Soboll S, Conrad A, Keller M, Hebish S: The role ofthe mitochondrial
creatine kinase system for myocardial functioning during ischemia and
reperfusion. Biochim Biophys Acta 1100: 27-32, 1992
31. Soboll S, Conrad A, Hebisch S: Influence of mitochondrial creatine
kinase on the mitochondrial/extramitochondrial distribution of high
energy phosphates in muscle tissue: Evidence for a leak in the creatine
shuttle. Mol Cell Biochem 133/134: 105-113, 1994
32. Saks VA, Chernousova GB, Gukovsky DE, Smirnov VN, Chazov EI:
Studies of energy transport in heart cells. Mitochondrial isoenzyme of
creatine kinase: Kinetic properties and regulatory action of Mg2+ ions.
Eur J Biochem 57: 273-290, 1975
33. Jacobus WE, Saks VA: Creatine kinase of heart mitochondria:
Changes in its kinetic properties induced by coupling to oxidative
phosphorylation. Arch Biochem Biophys 219: 167-178, 1982
34. Giesen J, Karnmermeier H: Relationship of phosphorylation potential
and oxygen consumption in isolated perfused rat heart. J Mol Cell
Cardiol12: 891-907, 1980.
35. Hassinen IE: Mitochondrial respiratory control in myocardium. Biochim
BiophysActa853: 135-151,1986
36. Klingenberg M, Rottenberg H: Relation between the gradient of the
ATP/ADP ratio and the membrane potential across the mitochondrial
membrane. Eur JBiochem 73: 125-130,1977
37. Fossel E, Hoefeler H: A synthetic functional metabolic compartment.
The role of propinquity in a linked pair of immobilised enzymes. Eur J
Biochem 170: 165-171, 1987
38. Smith HE, Page E: Morphometry of rat heart mitochondrial sub-
compartments and membranes: Application to myocardial atrophy after
hypophysectomy. J Ultrastruct Res 55: 31--41, 1976
39. Anversa P, Olivetti G, Melissari M, LoudAV: Morphometric study of
myocardial hypertrophy induced by abdominal aortic stenosis.
Lab Invest 40: 341-349, 1979

40. Elzinga G, van der Laarse WJ: MVO,max of the heart cannot be
determined from uncoupled myocytes. Bas Res Cardiol 85: 315-317,
1990
41. Williamson JR, Ford G, Illingworth J, Safer B: Coordination of
citric acid activity with electron transport flux. Circ Res 38(suppl
I): 139-151, 1978
42. Saks VA: Creatine kinase isoenzymes and the control of cardiac
contraction. In: WE Jacobus, JS Ingwall (eds). Heart Creatine Kinase.
The Integration of Isozymes for Energy Distribution. Williams and
Wilkins, Baltimore-London, 1980, pp 109--126
43. Ventura-Clapier R, Saks VA, Vassort G, Lauer C, Elizarova GV: Reversible
MM-creatine kinase binding to cardiac myofibrils. Am J Physiol253:
C444-C455,1987
44. Bittl JA, DeLayre J, Ingwall JS: Rate equation for creatine kinase predicts
the in vivo reaction velocity: IIp_ NMR surface coil studies in brain,
heart, and skeletal muscle of the living rat. Biochemistry 26: 6083-
6090,1987
45. Cleland WW: Enzyme kinetics. Annu Rev Biochem 36: 77-112, 1967
46. Kenyon GL, Reed GH: Creatine kinase: Structure-activity relationships.
Adv Enzymol54: 367-426, 1983
47. Davies EJ, Lurneng L: Relationship between the phosphorylation
potentials generated by liver mitochondria and respiratory state under
conditions of adenosine diphosphate control. J BioI Chern 250: 2275-
2292,1975
48. Van Dorsten F: lIP-NMR analysis of creatine kinase kinetics. Utrecht
University, The Netherlands, 1996
49. Saks VA, Chernousova GB, Voronkov Yu I, Smirnov VN, Chazov EI:
Study of energy transport mechanism in myocardial cells. Circ Res 34
and 35(suppl III): 138--149,1974
50. Bitt! JA, Ingwall JS: Reaction rates of creatine kinase andATP synthesis
in the isolated rat hears. J Bioi Chern 260: 3512-3517, 1985
51. Perry SB, McAuliffe J, Balschi JA, Hickey PR, Ingwall JS: Velocity of
the creatine kinase reaction in the neonatal rabbit heart: The role of
mitochondrial creatine kinase. Biochemistry 27: 2165-2172, 1988
52. Zahler R, Bittl J, Ingwall JS: Analysis of compartmentation of ATP in
skeletal and cardiac muscle using lip nuclear magnetic resonance
saturation transfer. Biophys J 51: 883-893, 1987
53. Saks VA,Aliev MK: Is there the creatine kinase equilibrium in working
heart cells? Biochem Biophys Res Comm 227: 360-367, 1996
54. Zahler R, Ingwall JS: Estimation of heart mitochondrial creatine kinase
flux using magnetization transfer NMR spectroscopy. Am J Physiol
262: HI022-Hl028, 1992
229
55. Veksler VI, Kuzuetsov AV, Anflous K, Mateo P, Van Deursen J, Wieringa
B, Ventura-Clapier R: Muscle creatine kinase deficient mice. " Cardiac
and skeletal muscle exhibit tissue-specific adaptation of the mitochondrial
function. ] BioI Chern 270: 19921-19929, 1995
56. Aliev MK, Saks VA: Compartmentalized energy transfer in cardio-
myocytes. Use of mathematical modelling for analysis of an in vivo
regulation of respiration. Biophys J, 73: 428-445,1997
57. Veech RL, Lawson JWR, Cornell NW, Krebs HA: Cytosolic phos-
phorylation potential. J BioI Chern 254: 6538--6547, 1979
58. Muhlebach SM, Gross M, WirzT, Wallimann T, Perriard J-P, Wyss M:
Sequence homology and structure predictions of the creatine kinase
isoenzymes. In: VA Saks, R Ventura-Clapier (eds). Cellular Bioenergetics:
Role of Coupled Creatine Kinases. Kluwer Academic Publishers,
Dordrecht-Boston-London, 1994,pp 245-262
59. Hoerter JA, Ventura-Clapier R, Kuznetsov AV: Compartmentation of
creatine during perinatal development of mammalian heart. In: VA Saks,
R Ventura-Clapier (eds). Cellular Bioenergetics: Role of Coupled
Creatine Kinases. Kluwer Academic Publishers, Dordrecht-Boston-
London, 1994, pp 277-286
60. Kupriyanov VV, Steinschneider A, Ya, Runge EK, Kapelko VI, Zueva
M, Yu, Lakomkin VL, Smirnov VN, Saks VA: Regulation of energy flux
through the creatine kinase reaction in vitro and in perfused rat heart.
Biochim Biophys Acta 805: 319--331, 1984
61. Sievers R Parmley WW, James T, Wikman-Coffelt J: Energy levels at
systole vs. diastole in normal hamster hearts vs. myopathic hamster
hearts. Circ Res 53: 759--766, 1983
62. Saks VA, Ventura-Clapier R, Huchua ZA, Preobrazhensky AN, Eme1in
IV: Creatine kinase in regulation of heart function and metabolism. I.
Further evidence for compartmentation of adenine nuc1eotides in
cardiac myofibrillar and sarcolemmal coupled ATPase-creatine kinase
systems. Biochim Biophys Acta 80: 254-264, 1984
63. Saks VA, Chernousova GV, Vetter R, Smirnov VN, Chazov EI: Kinetic
properties and functional role of particulate MM-isoenzyme of creatine
phosphokinase bound to heart muscle myofibrils. FEBS Lett 62: 293-
296,1976
64. Holian A, Owen CS, Wilson DF: Control of respiration in isolated
mitochondria: Quantitative evaluation of the dependence of respiratory
rates on [ATP], [ADP], and [PJArch BiochemBiophys 181: 164-171,
1977
65. Gyulai L, Roth Z, Leigh JS, Chane B: Bioenergetic studies of mito-
chondrial oxidative phosphorylation using llPhosphorus-NMR J BioI
Chern 260: 3947-3954,1985
Molecular and Cellular Biochemistry 184: 231-247, 1998.
1998 Kluwer Academic Publishers.

Functional coupling of creatine kinases in muscles:

Species and tissue specificity
R. Ventura-Clapier, A. Kuznetsov, V. Veksler, E. Boehm and
K. Anflous
Cardiologie Cellula ire et Moleculaire U-446INSERM,
Faculte de Pharmacie Universite Paris-Sud, 5 rue Jean-Baptiste Clement, 92296 Chtitenay-Malabry, France

Abstract
Creatine kinase (CK) isoenzymes are present in all vertebrates. An important property of the creatine kinase system is that
its total activity, its isoform distribution, and the concentration of guanidino substrates are highly variable among species and
tissues. In the highly organized structure of adult muscles, it has been shown that specific CK isoenzymes are bound to
intracellular compartments, and are functionally coupled to enzymes and transport systems involved in energy production
and utilization. It is however, not established whether functional coupling and intracellular compartmentation are present in
all vertebrates. Furthermore, these characteristics seem to be different among different muscle types within a given species.
This study will review some of these aspects.
It has been observed that: (I) In heart ventricle, CK compartmentation and coupling characterize adult mammalian cells. It is
almost absent in frogs, and is weakly present in birds. (2) Efficient coupling of MM-CK to myosin ATPase is seen in adult
mammalian striated muscles but not in frog and bird heart where B-CK is expressed instead ofM-CK. Thus, the functional efficacy
of bound MM-CK to regulate adenine nucleotide turnover within the myofibrillar compartment seems to be specific for muscles
expressing M-CK as an integral part of the sarcomere. (3) Mi-CK expression and/or functional coupling are highly tissue and
species specific; moreover, they are subject to short term and long term adaptations, and are present late in development. The
mitochondrial form ofCK (mi-CK) can function in two modes depending on the tissue: (i) in an ADP regeneration mode and
(ii) in an ADP amplification mode. The mode of action of mi-CK seems to be related to its precise localization within the
mitochondrial intermembrane space, whereas its amount might control the quantitative aspects of the coupling. Mi-CK is highly
plastic, making it a strong candidate for fine regulation of excitation-contraction coupling in muscles and for energy transfer in
cells with large and fluctuating energy demands in general. (4) Although CK isoforms show a binding specificity, the presence
ofa given isoform within a tissue or a species only, does not predict its functional role. For example, M-CK is expressed before
it is functionally compartmentalized within myofibrils during development. Similarly, the presence of ubiquitous or sarcomeric
mi-CK isoforms, is not an index of functional coupling ofmi-CK to oxidative phosphorylation. (5) Amongst species or muscles,
it appears that a large buffering action of the CK system is associated with rapid contraction and high glycolytic activity. On the
other hand, an oxidative metabolism is associated with isoform diversity, increased compartmentation, a subsequent low buffering
action and efficient phosphotransfer between mitochondria and energy utilization sites.
It can be concluded that, in addition to a high variation of total activity and isoform expression, the role of the CK system
also critically depends on its intracellular organization and interaction with energy producing and utilizing pathways. This
compartmentation will determine the high cellular efficiency and fine specialization of highly organized and differentiated
muscle cells. (Mol Cell Biochem 184: 231-247, 1998)

Key words: mitochondria, compartmentation, myofilaments, contraction, ATPase, translocase, ventricle, atria, calcium
sensitivity, oxygen consumption, oxidative capacity, creatine, rigor tension, active tension

Present address: A. Kuznetsov, Department of Transplant Surgery, D. Swarovski Research Laboratory, University Hospital ofInnsbruck,A-6020 Innsbruck,
Austria
Address for offPrints: R. Ventura-Clapier, Cardiologie Cellulaire et Moleculaire U-446 INSERM, Faculte de Pharmacie Universite Paris-Sud, 5 rue Jean-Baptiste
Clement, 92296 Chiitenay-Malabry, France
232
Introduction
Understanding the regulation of energy transfer in muscle cells
is an important field of bioenergetics. It comprises fundamental
concepts of cell organization, multi enzyme complexes and
compartmentation of cellular functions. It has direct implic-
ations in the understanding of adaptive strategies and
pathophysiological mechanisms for a variety of disorders in
heart and skeletal muscle.
Differentiation and maturation lead to a highly complex
cellular specialization and organization in adult mammalian
muscles. Each muscle type has a specific program which leads
to a morphological and biochemical organization designed to
sustain a particular cellular activity. These programs are
adaptable and can be modified according to environmental
factors and whole body needs (hypokinesia, exercise, pressure
load, hormonal changes) [1-5]. Compartmentation of enzymes
implicated in cellular energy production, utilization and
transfer seems to be a characteristic of these highly differ-
entiated organized adult mammalian cells. Among these,
muscle cells represent a unique model of high and fluctuating
energy demand. Moreover, the specific organization and
function of different muscle cell types include a wide range of
possible metabolic pathways and regulation. The potential
importance of studying cellular energy transfer in normal and
pathological cells and tissues lies in the widespread and
harmful incidence of muscle and cardiac disorders in human
health.
In muscle cells, specialized cellular functions are highly
organized within structural and functional compartments.
Calcium transport, for example, occurs within the sarco-
plasmic reticulum compartment (SR) which contains calcium
release channels in contact with the sarcolemma membranes
and T-tubules, as well as the Ca-MgATPase that pumps
calcium from the sarcoplasm. Contraction is ensured in the
myofibrillar compartment (which occupies ",,50-60% of
muscle cell volume) by Ca 2+, leading to the sliding of two
types of filaments. The thick filament possesses myosin
ATPase activity, while the thin filament bears the regulatory
sites for calcium. Interaction between these two filaments
transforms chemical energy into mechanical work.
Energy production also occurs within compartments:
glycolytic complexes utilize glucose and glycogen to produce
anaerobic ATP, whilst mitochondria produce aerobic ATP by
oxidizing mitochondrial substrates. Mitochondria appear to
be clustered in sites of high ATP demand and are organized
into highly ordered elongated bundles, occupying a significant
portion of the cell volume (up to 40% in hearts of small
mammals). Electron microscope analysis of resinless
sections reveals a highly intricated cytostructure of
filaments within the cytoplasm leading to the association
of intracellular structures with an outer boundary of
membrane proteins [6]. This complex organization of
proteins and organelles, and a consequent interaction
between cell functions represents the highest degree of
cellular efficiency achieved by evolution.
In this context, the question arises as to the organization
and efficiency of the energy transfer. Although excitation-
contraction coupling and ion movements have been ex-
tensively studied, less is known concerning energy transfer
in muscle cells. The relative importance ofthe two metabolic
pathways of energy production (glycolysis and mito-
chondria) varies between muscle types depending on the
contractile pattern. In view of the high and fluctuating energy
demand of muscle cells, it is worthwhile to think that special-
ized pathways have been selected throughout evolution to
optimize the efficiency of production, transfer and utilization
of energy without large fluctuations of concentrations in
substrates and products. In order to efficiently link energy
production and utilization, muscle cells contain complex and
specialized energy transfer systems.
The family of creatine kinase (CK) isoenzymes is one
such system and is able to catalyze the reversible transfer
of a phosphate moiety between creatine and ATP. This
reaction is a dead end reaction in the sense that CK is the
only enzyme which interferes with creatine (Cr), and its
phosphorylated counterpart, phosphocreatine (PCr). Creatine
and CK are present in variable amounts in different muscle
types. In birds and mammals, four different isoforms of CK
are expressed in a tissue specific and developmentally
regulated manner. A major part of muscle CK exists as
dimers composed of two subunits, M and B, giving three
isoenzymes, MM, BB, and MB. In addition, there is a fourth
isoenzyme in the mitochondria (mi-CK), which differs
biochemically and immunochemically from the cytosolic
forms and can form both octameric and dimeric structures.
Mi-CK is coded by two different genes in a tissue specific
manner. One form (mi-CKs or mib -CK) is present in tissues
presenting sarcomeric structures (striated muscles) and is
co-expressed with M-CK, while the other (miCK or mi -
CK) is ubiquitous and co-expressed with B-CK [7, 8]. N~
clear kinetic differences exist between the four different
isoforms. They differ only by their subcellular distribution.
Studies using subcellular fractionation or histochemical
localization have revealed that CK isoenzymes are present
in the cytosol, or bound to the intracellular compartments.
M-CK has been found in myofibrils and described as a
structural protein of the M-band participating in the
connections between myosin filaments inside muscle fibers
(for review see [9]). This bound M-CK has been shown to
be functionally coupled to the myosin ATPase. Myofibrillar
CK can use PCr to rephosphorylate all of the ADP produced
by myosin ATPase and can provide enough energy for max-
imal force and normal contractile kinetics even in the
absence of MgATP (for review see [10]. Similarly, M-CK
is strongly bound to SR membranes where it is functionally
 233

coupled to Ca2+ ATPase, and ensures efficient energy pro-
vision of the SR by the local regeneration ofATP [11-14]. Mi-
CK is found on the outer surface of the inner mitochondrial
membrane. It can be localized in the vicinity oftheATP-ADP
carrier, structurally and functionally coupled to the adenine
nucleotide translocase so that ATP generated by oxidative
phosphorylation, after transport through the inner mito-
chondrial membrane, is transphosphorylated to PCr (for
review see [8, 15] and articles in this issue). The sites of
energy production and energy utilization are integrated
through near equilibrium reactions of CK in the cytosol
promoting sequential equilibration resulting in almost instant-
aneous transfer of phosphoryl groups and metabolic signal
[15-17]. This chain of near equilibrium reactions has been
suggested to operate by a mechanism based on Mitchell's
general principle of 'vectorial ligand conduction' [18]. This
principle of phosphotransfer reaction can be extended to
adenylate kinase system and is reviewed and illustrated in
this book by Dzeja et al. Phosphotransfer catalyzed by the
CK system in ventricular cells is illustrated in Fig. 1.
Skeletal muscle is an extremely heterogeneous tissue.
Multigene families and alternative transcript splicing create
multiple thick and thin filament protein isoforms. Activity
pattern influences fiber type and metabolic profile [1,4]. The
creatine kinase/phosphocreatine system is considered to
fulfil important roles in the energy metabolism of skeletal
and cardiac muscles (for reviews see [8, 15, 17]) and its
organization and function in the different muscle types is of
particular interest. Fast-twitch (white) muscles, exhibiting
a rapid and brief activity pattern, are mainly glycolytic and
contain high amounts of PCr and CK. By contrast, slow-
twitch (red) skeletal muscle or cardiac muscle, exhibit
prolonged and sustained activity associated with well
developed oxidative metabolism and relatively low con-
tents of PCr and CK. The localization of the different CK
isoenzymes appears different in these two kinds of muscles,
there is an abundance of muscle form creatine kinase (M -CK)
in the cytosol of fast-twitch muscle, and compartmentation
of the mitochondrial isoenzyme in slow-twitch muscle and
ventricle.
Although guanidino kinases in general and creatine kinase
in particular are present throughout the animal kingdom,
functional compartmentation of CK isoenzymes has so far
been mainly studied in muscles of adult mammals. Lower
vertebrates also contain several isoenzymes, all giving rise
to dimeric molecules. During evolution, a first gene
duplication resulted in a primordial cytosolic and a primordial
mitochondrial CK isoenzyme. Further gene duplications have
given rise to the multiple cytosolic and mitochondrial
isoforms that exist today [19]. However, the existence and
efficiency of functional coupling of CK isoenzymes within
specific intracellular structures in different classes of
vertebrates, have not been documented. Within a given class,
it is also interesting to see whether animals having different

mb. i

erJ(Q;\ iran leT =-

/
mito hondria myo Ilam

o Irani kin .. r }inA1P

Fig. 1. Creatine kinase system in adult mammalian ventricular cells. In mitochondria, due to structural and functional coupling betwem mitochondrial CK (mi-
CK) and adenine nucleotide translocase (ANT), the newly produced phosphoryl group of ATP is channelled to creatine via mitochondrial creatine kinase,
with ADP channeling back to mitochondrial matrix through translocase. This phosphoryl group enters the chain of near equilibrium reactions to the sites of
energy utilization (i.e. myofilaments), where it is transferred to ADP produced by myosin ATPase with simultaneous release of metabolic signal and transfer
of the signal back to the mitochondrial matrix.
234
metabolic requirements such as the chicken (sedentary) and
pigeon (flying) will exhibit specific organization and
regulation of energy transfer. Within a given animal, we can
ask whether muscles having different roles and activity
patterns exhibit differential organization of the CK system
and compartmentation.
The aims of the present article, based on skinned fibers
studies of the functional coupling of CK isoenzymes
within the two main intracellular compartment in cardiac
cells, namely mitochondria and myofibrils, are to compare
(1) different classes of vertebrates (mammals, birds,
amphibians), (2) within given classes, to compare different
species, (3) within a given species to compare different
tissues, and (4) to draw conclusions concerning the regulation
of energy fluxes in muscle cells and role of the creatine
kinase system in general.
Materials and methods
Biochemical determinations and data evaluation
Total CK and adenylate kinase (AK) activities were determined
after homogenization of tissues, in a solution containing 5
mM Hepes (pH 8.7), 1 mM EDTA, 1 mM dithiothreitol
(DTT) and 0.1 % Triton X-I 00. Homogenates were incubated
on ice for 1 h with shaking every 15 min. Whole homogenates
were used for enzymatic activity.
CK activity was determined using the coupled enzyme
assay of glucose-6-phosphate dehydrogenase (G-6-PDH) and
hexokinase (HK) producing NADPH. NADPH production
was followed spectrophotometric ally at 340 nm (Gilford
Spectrophotometer) and 30C. The solution used for CK
activity assay contains 20 mM Hepes (pH 7.4),5 mM MgCI 2 ,
20 mM glucose, 0.5 mM DTT, 1 mMADP, 0.6 mM NADP,
20 mM phosphocreatine (PCr), 100 J.!M diadenosine
pentaphosphate (APsA for adenylate kinase (AK) inhibition),
2 IV/ml HK and 2.5 IV/ml G-6-PDH. AK activity was
determined as described above but the medium did not
contain AP sA and PCr.
Mechanical experiments and myofibrillar CK efficacy
The general method is derived from Ventura-Clapier et al.
[20] and reviewed in [10]. Muscle fiber bundles were
dissected in a zero-Ca 2+ Krebs solution, pH 7.4. Bundles
were incubated for I h in a relaxing solution (pCa 9, see
solutions below) containing I % Triton X-IOO to solubilize
the membranes; these were then transferred to the relaxing
solution without detergent and kept at 4C until use. After
the skinning procedure, one bundle was mounted between
two tubes. The tubes are connected to a transducer (model
AE 801, SensoNor Microelectroniks, Horten, Norway) and
a vibrator to generate quick length changes as previously
described [21]. Length and tension changes were monitored
on a digital storage oscilloscope (OS4020, Gould Inc.,
Cleveland, OH, USA). Tension tracings were digitized at 20
kHz (12-bit analog/digital converter), analyzed on-line using
a PC compatible computer and stored on a videotape.
Solutions were calculated using the computer program of
Fabiato [22]. All solutions were calculated to contain
(mM): EGTA 10, imidazole 30, Na+ 30.6, Mg2+ 3.16 and
dithiothreitol 0.3; ionic strength was adjusted to 0.16 M with
potassium acetate. pH was adjusted to pH 7.1 with acetic
acid. In relaxing and rigor solutions, pCa was 9. In activating
solution pCa was 4.5. Relaxing and activating solutions also
contained 3.16 mM MgATP and 12 mM PCr. Rigor solutions
were obtained by mixing two solutions of pMgATP 2.5 and
6 or pMgATP 4 and 6.
pCa/tension relations were determined under isometric
conditions by briefly placing each fiber into solutions of
increasing calcium concentration until maximal tension was
reached (Fig. 2A). Data were fitted using linearization of the
Hill equation for relative tensions above 10% and under 90%
as follows: T = [Ca2+]n/(K +[Ca2+]n) were T is relative tension,
K a constant and n the Hill coefficient. Resting tension was
measured at pCa 9 and at 120% of slack length (La)' Active
tension (expressed as mN/mm2) was the total tension at pCa
4.5 minus resting tension. pMgATP/rigor tension relations
were established by stepwise decreasing ATP concentration
until maximal rigor tension was obtained either in the
presence or in the absence ofPCr (Fig. 2B). Data were fitted
using the Hill equation. The pMgATP for half-maximal rigor
tension, pMgATPsa = (-Log]OK)/n, was calculated for each
experimental condition using linear regression analysis. The
difference in pMgATP so with or without PCr was taken as
an index of myofibrillar CK efficacy.
In situ study of mitochondrial respiration and mi-CK
The general method is derived from Veksler et al. [23] and
reviewed in the issue by Saks et al. Bundles of muscle fibers,
200-300 11m in diameter were isolated and permeabilised
with saponin. Five to ten mg of fiber bundles were incubated
for 30 min in a solution S (Skinning) containing: 2.77 mM
Ca~EGTA, 7.23 mM KzEGTA (1 00 nM free Ca 2+), 6.56 mM
MgCl 2 (1 mM free Mg2+), 5.7 mM Na2ATP, 15 PCr, 20 mM
taurine, 0.5 mM DTT, 50 mM Kmethane-sulfonate (160 mM
ionic strength), 20 mM imidazole (pH 7.1), and 50 I1g/ml
of saponin. Bundles were washed for 10 min in a solution R
(Respiration) composed as solution S, but containing 5 mM
glutamic acid, 2 mM malic acid and 3 mM K 2 HP04 instead
of high energy phosphates. All these steps were carried out
at +4C with vigorous stirring.
 235

A 1 B 1

0.8
=
.!: =
=
'"= 0.6 ';1
.a
GI
....=
GI
0.5
~
~
0.4 c
... ...
GI GI
.~ .~
0.2
~ ~
0 0
6 5 4 3 2
7 6 5 4 pMgATP
pCa

c D
fibers
[ADPJ (11M)

ADP 25
fibers 100 11M
,.. Creatine ~ ~
20mM - creatine
ADP
Vi Vo ImM

+ creatine
Oxygen Oxygen
25 nmoles 25 nmoles

1 min 1 min

Fig. 2. Experimental protocols. (A) pCa/tension relationship are obtained by bathing TritonX-100 treated fibers in solutions of increasing calcium concentration.
Curves are plotted using the Hill equation. pCa 50 (the value of pCa for half maximal tension development) is indicated by an arrow. (B) Dependence of rigor
tension on pMgATP (-Log[MgATP]). Fibers are successively bathed in solution with low calcium (pCa 9) and decreasing concentrations ofMgATP in the
presence or in the absence of PCr. Curves are plotted using the Hill equation. pMgATP 50 values are indicated by arrows. (C) Experimental protocol for the
measurement of respiration of saponin skinned fibers. Fibers are bathed in solution R (see Materials and methods) and 100 J.lM ADP, 20 mM creatine, I mM
ADP are successively added. Va' basal respiration; Vadp, respiration in the presence of 100 J.lM ADP; VCr respiration in the presence of creatine; V mo>' maximal
respiration. (D) Protocol for oxygraphic measurement of the Km of respiration for ADP in the presence or in the absence of creatine. Increasing concentrations
of ADP are successively added in the respiratory chamber.

Respiration of skinned fibers was determined at 22C, in
3 ml of medium R containing 2 mg/ml of bovine serum
albumin, using a Yellow Spring Instruments Oxygraph. After
measurements, fibers were removed, dried and weighed.
Calculations were made assuming that the solubility of
oxygen at 22C was 230 nmoles oxygen/ml equilibrated with
ambient air. Rates of respiration are given in I1moles of
oxygen/min/g dry weight (g dw). Two different experimental
protocols were used. The first protocol (Fig. 2C) consisted
of addition of a submaximal ADP concentration (100 11M)
followed by creatine addition (20 mM) in order to reveal
creatine stimulated respiration and calculate the percentage
of stimulation of respiration by creatine (VCr%); the final
addition of I mM ADP allowed the measurement of maximal
respiration rates and the calculation of acceptor control
ratio (ACR) as well as the maximal oxidative capacities of a
given tissue. The second protocol (Fig. 2D) is for the
determination of the Km of mitochondrial respiration for
ADP in the presence and in the absence of creatine, in order
to estimate the affinity of mitochondria for ADP and the role
of mitochondrial CK.
Results and discussion
Species differences in functional coupling of creatine
kinases
We have compared the in situ myofibrillar and mitochondrial
properties and role ofbound-CKs in ventricular tissues from
different species: one amphibian (frog), two birds (chicken
and pigeon) and various mammals (ferret, rabbit, guinea-pig,
236
rat, hamster and mice). Heart weight to body weight ratio
(HW/BW) and total cardiac CK activities for these different
species are shown in Fig. 3. An important property of the
creatine kinase system is that total activity, isoform
distribution, isoform localisation, and concentrations of
creatine are highly variable. Total CK activity shows great
species variation (Fig. 3B). Gene expression of creatine
kinase and creatine transporter are developmentally regulated,
vary with species and change during pathology [24]. Hearts
oflarge mammalian species like human, sheep, dog and cat
contain more total CK activity and have less mi-CK activity
than hearts from smaller species (rabbit, ferret, guinea-pig,
rat and mouse) [24]. Hearts of birds contain a relatively low
total CK activity, and express B-CK instead of the M-CK
isoform (M-CK is present however in skeletal muscles of
birds). Lower vertebrates such as frog also contain several
CK isoenzymes loci, all giving rise to dimeric molecules
of around 80,000 but in most cases they seem to differ from
those of mammals and birds [19]. Christensen et at. [25] did
A We have first compared the mechanical properties of
Frog Pigeon Rabbit Rat Mouse
Chicken Ferret Guinea-pig Hamster
B the absence of Ca. This leads to rigor tension development.
Frog Pigeon Rabbit Rat Mouse
Chicken Guinea-pig Hamster
Fig. 3. Species specificity. (A) Heart weight to body weight ratio (HW/
BW) in mg/g. (B) Total cardiac creatine kinase activity in IU/g wW.
not find a clear relation between CK activity and total creatine
concentration among the different species. The variation in
total concentration of phosphorylated adenylates in heart
muscle of different vertebrates is considerably less than the
variation in total creatine.
In addition to specific activity variations, the role of CK
also crucially depends on efficient coupling to ATPases (at
sites of energy consumption) and oxidative phosphorylation
(at sites of energy production). The question arises as to
whether this ability to undergo functional coupling with
energy producing and energy utilizing reactions is present
throughout species expressing CK isoenzymes. At present,
this has been mainly described for ventricular muscle (and
to a lesser extent slow-twitch skeletal muscle) of adult
mammals.
Myofibrillar function and creatine kinase in heart of
different species
skinned cardiac fibers from the different species. Maximal
contractile capacities of muscle fibers can be obtained from
skinned fibers by measuring maximal force developed per unit
cross sectional area at a saturating calcium concentration.
Figure 4A summarizes data concerning the maximal force
developed by cardiac skinned fibers at pCa 4.5. Maximal
tension ranged between 20 and 40 mN/mm2 Figure 4B depicts
calcium sensitivity of tension development for the different
species. Calcium sensitivity was fairly constant except for
cardiac fibers from frog and chicken which had a higher
calcium sensitivity in accordance with the data of Fabiato [26].
To estimate the importance of bound creatine kinases, we
used an experimental test elaborated on the basis on muscle
function. When myosin is deprived of ATP it has a high
affinity for actin so that both molecules can interact even in
This process is enhanced by MgADP. Thus rigor tension
development as a function of MgATP is a good estimate of
MgATP and/or MgADP concentration in the vicinity of
myosin ATPase. In the absence of PCr, rigor tension
develops with a half maximal effect at ",300 ~M MgATP. We
have shown that this value can be shifted to a very low
[MgATP] in the presence of PCr 10 ~M) due to effective
local rephosphorylation ofMgADP by MM-CK bound to the
myofilaments. In order to compare the efficacy of bound CK
in cardiac myofilaments of the different species, pMgATP/
tension relationships were established for each species in
the presence and in the absence of phosphocreatine and
pMgATP 50 values were calculated (Fig. 5). While in the
absence ofPCr no noticeable difference could be observed,
the pMgATP 50 in the presence of PCr strongly differed. In
frog heart, the pMgATP50 with PCr was low and not different
 237

A 80 6

60
D
~+PCr
control
...
.,.
o 5
""'
N

S 40
..
Z
S
'-'
4
~ 20
~
0
Frog Rabbit Rat Mouse Frog Pigeon Rabbit Rat Mouse
Chicken Ferret Guinea-pig Hamster Chicken Ferret Guinea-pig Hamster

Fig. 5. Species specificity. Sensitivity of rigor tension to ATP and role of

myofibrillar creatine kinase in ventricular fibers. Rigor tension was recorded
in solutions of decreasing MgATP concentration in the presence or absence
B of PCr and the pMgA TP (-Log[MgATP]) for half maximal tension
development (pMgA TP 50) was calculated from the Hill equation. *p < 0.05;
***p < 0.001; statistically different from pMgATP,o values in the absence
of Per.

o
'"
Frog Pigeon Rabbit Rat Mouse
Chicken Ferret Guinea-pig Hamster
Fig. 4. Species specificity. Mechanical properties of Triton X-IOO treated
ventricular fibers. (A) Maximum active tension development measured at
maximal calcium concentration (pea 4.5) expressed in mN/mm'. (B) Tension
was measured in solutions of different calcium concentration and pea for
half maximal tension development (pea,o) was calculated from the Hill
equation.
from that in the absence of PCr, showing no role for myo-
fibrillar CK in controlling adenine nucleotides within the
myofilaments. In birds compared to frog, a slightly higher
value was observed, which was different from the value in
the absence of PCr. In all mammals, half maximal rigor
tension developed for MgATP values more than one order
of magnitude lower in the presence than in the absence of
PCr (",6 11M versus ",320 11M), evidencing a strong control
of myofibrillar bound CK on adenine nucleotide com-
partmentation. Though lO% of total CK activity is found in
the myofibrillar fraction of Triton X-IOO treated frog
ventricular cells [27], the exact location and affinity of CK
in frog myofibrils is not known. On the other hand, chicken
heart cells do not express any M-CK, contain relatively high
levels of BB-CK and are devoid of the electron dense
material typical of the M-band region in which M-CK is
usually located; 2% oftotal CK activity in a chicken heart is
located within the Z-line region of each sarcomere [28]. Since
the amount ofM-CK is",3 mg per g ofmyofibrillarprotein in
rat heart [29] and 1 mg per g ofBB-CK in chicken heart [28],
and since BB-CK has a two times lower specific activity than
M-CK, this may explain the lower efficacy of myofibrillar CK
in hearts of birds than mammals.
It can be concluded that the functional efficacy of myo-
fibrillar bound CK to regulate adenine nucleotides within
the myofibrillar compartment seems to be specific to
mammalian heart due to the expression and association of
M-CK as an integral part of myofibrillar structures. In this
tissue, bound CK will ensure efficient control of adenine
nucleotide concentration within the myofibrillar com-
partment allowing optimal function of myosin ATPase and
contraction.
Mitochondrial function and creatine kinase in the heart
of different species
One of the unique advantages of skinned fibers in studying
tissue oxygen consumption is that it allows the estimation
of the maximal oxidative capacities of a given tissue at
constant substrate concentration and at saturating oxygen
([23], for review see Saks et al. in this issue). Since the
whole mitochondrial population is kept in its natural
environment, the rate of oxygen consumption can be
normalized to fiber dry weight. Oxidative capacities have
been evaluated in cardiac skinned fibers of the different
species (Fig. 6A). Oxidative capacities were higher in cardiac
tissues of small mammals like guinea-pig, rat and mice
238

compared to larger mammals like rabbit and ferret (except in amount of mitochondria [34]. Despite lower animal size, pigeon
hamster for unknown reasons). It is generally believed that heart have oxidative capacities higher than chicken but lower
increased aerobic capacities are simply achieved by increased than rat, in agreement with lower cytochrome oxidase content
mitochondrial volume density [31, 32]. Barth et al. [33] have of pigeon heart [25].
examined the mitochondrial content of hearts of 10 mammalian The acceptor control ratio, which is the ratio between
species. Volume densities of mitochondria varies between 22- maximal respiration in the presence of 1 mM ADP and basal
37% among species but is a very constant and specific value respiration withoutADP, and indicates the coupling between
for any particular species, the smallest species having the oxidation and phosphorylation, is fairly constant among
highest content. These authors have shown a close species (Fig. 6B).
correlation between the mitochondrial volume density, heart The extent of stimulation of respiration by creatine in the
rate and the rate of basal oxygen consumption, inversely presence of a submaximal ADP concentration, an index of
related to animal size. It is also recognized that capillary length mi-CK coupling, exhibited clear species variations (Fig. 7).
density is correlated with mitochondrial volume density in In no species was respiration rate maximal in the presence
heart of different species [30]. Frog heart having lower of 100 flM ADP, showing that in all species studied, the
oxidative capacities than rat heart, also contains a lower sensitivity for bulk ADP of cardiac mitochondria in situ is
much lower than for mitochondria in vitro. Moreover,
A 35
respiration rates at 100 flM ADP were approximately one
third of maximal for all species, showing similar affinity for
'"~""' 30 ADP. In frog heart, no stimulation of respiration by creatine
"Cl
OJ)
-...... was observed, though the presence of mitochondrial forms
25
6' of creatine kinase have been documented in all vertebrates
..,
en
20 [8, 19]. Mi-CK in isolated mitochondria of frog heart has
"0
8 been shown to regenerate ADP in the presence of creatine
2; 15
~
[27]. Thus, the presence of mitochondrial CK does not
0
.~ 10 automatically imply functional coupling with oxidative
.....~ 5
phosphorylation. Such dissociation has already been
~ observed in guinea-pig uterus [35, 36] and rat atrial tissue
"""'0x 0 [37,38, and see below]. In chicken heart, the stimulation of
respiration by creatine was less than 20% while in pigeon it
8'"
Frog Pigeon Rabbit Rat Mouse
;> Chicken Ferret Guinea-pig Hamster reached 60%. Coupling ofmi-CK to respiration is therefore
present in birds but varies among species. Heart and skeletal
muscle of birds express a specific mi-CK isoenzyme (Mib-CK)

B 10
120

100
0
.~
.... 6 80
"0
E
0 4
60
u
....
0
'-" 40
0. ....
<I)
u 2 U
u ;>
-< 20

Frog Pigeon Rabbit Rat Mouse

Chicken Ferret Guinea-pig Hamster Frog Pigeon Rabbit Rat Mouse
Chicken Ferret Guinea-pig Hamster
Fig. 6. Species specificity. (A) Oxidative capacities of ventricular fibers
were measured in the presence of I mM ADP and 20 mM creatine and at Fig. 7. Species specificity. Stimulation of respiration by creatine in
saturating concentrations of substrates and oxygen. Values are given as ventricular fibers. The respiration rate was measured before and after addition
flmoles of oxygen consumed per g of fiber dry weight. (8) Acceptor of20 mM creatine in the presence of sub maximal ADP concentration (100
control ratio was calculated as the ratio between maximal respiration flM of ADP). The extent of stimulation is expressed as percent increase in
rate (V m,) and basal respiration rate (Vo)' respiration rate (VCr).

[39], which is present in dimeric or octameric forms, the octamer
being able to form contact sites between the inner and the
outer mitochondrial membranes as in mammals ([40], see also
Wallimann et al. in this issue). In chicken heart, the specific
activity ofmi-CK is ",,0.6--0.85 IU/mg mitochondrial protein
compared t04.5 IU/mg in rat heart (i.e. 15%) [15,41]. In chicken,
the degree of coupling estimated by the percentage of increase
in respiration by creatine, was 20% compared to 70--80% in
mammalian heart, suggesting that the degree of coupling is
proportional to mi-CK specific activity within mitochondria.
The higher stimulating effect of creatine in pigeon heart would
thus be due to a higher amount of mi-CK. In pigeon ventricle,
higher oxidative capacities, higher efficiency of mi-CK
coupling and higher heart weight to body weight ratio, may
account for higher energy requirements for this flying bird
compared to chicken. Degree of creatine stimulated respiration
among mammalian species was identical in accordance with
the similar specific activity ofmi-CK in these species [15].
One interesting feature of mi-CK coupling is its extreme
variability during development, and in pathological states. In
any cardiomyopathic states of different origins studied so
far, mi-CK coupling has been found to be profoundly altered
[42]. In addition, its binding and coupling seem to also be
the targets for short term regulation during, for example,
hypoxic and ischaemic states, by inorganic phosphate or
hydrogen ions [43,44].
What are the determinants of the appearance of functional
coupling ofCK isoenzymes. Although different CK isoforms
are present in hearts of all vertebrates, efficient functional
coupling of CK isoenzymes in sites of energy production
and utilization is only fully encountered in adult mammals.
However, CK isoenzymes are expressed early during cell
differentiation. In muscle, the B-form of CK is down-
regulated when cells differentiate and is replaced by M-CK.
In cardiac muscle, despite the presence of M-CK in fetal
heart, compartmentation in myofilaments appears only in the
late phase of fetal development [45,46]. Moreover, there
are species differences in the development of CK com-
partmentation. While M-CK and mi-CK compartmentation
are present before birth in the heart of guinea-pig, which is
mature at birth, it appears in the first postnatal weeks in
immature species like rabbit [46]. Thus, compartmentation
ofCK isoenzymes is not associated with birthper se but with
the general process of cell maturation. The late appearance
of mi-CK in neonatal rat heart signals the shift from
glycolytic (fetal) to oxidative (mature) metabolism [47].
Increased oxidative metabolism, compartmentation of both
Ca handling and energy turnover, leads to a highly structured
cell organization which is essential for the efficiency of
cardiac function. It is noteworthy that frog and bird cardiac
cells, similarly to fetal mammalian cardiac cells, have
smaller diameter, lesser cellular organization, and do not
contain T tubules, as compared to adult mammalian heart
239
cells [34,48]. It appears that one can not predict association
of CK isoenzymes with intracellular structures and building
up of functional compartmentation only from the presence
of specific isoforms, on neither an evolutionary nor
developmental point of view. Conversely, functional com-
partmentation seems to have evolved late after the duplication
of the CK isoenzymes, and correlates with the appearance
of a highly ordered and compartmentalized intracellular
architecture in mammalian cardiac cells.
Tissue specificity in functional coupling of creatine
kinases in rat
Another question of interest is whether compartmentation
and functional coupling are identical in the different muscle
types within one species. We have examined myofibrillar and
mitochondrial properties in two skeletal muscles having
different activity patterns (slow-twitch, soleus, and fast-
twitch, gastrocnemius) and the two cardiac tissues (atria and
ventricles) of the rat.
Myofibrillar function and creatine kinase in different
muscle types
In adult rat, the different muscle types exhibit functional,
structural and metabolic specificity. The maximal potency to
develop force greatly differs between muscles. Ventricular
muscle has low capabilities while the force increases in slow
twitch skeletal muscle, being highest in fast twitch muscles
(Fig. 8A). Force developed per unit cross sectional area is
lower in oxidative tissues at least in part due to the higher
relative volume occupied by mitochondria versus myofibrils.
When comparing ventricular, slow and fast skeletal muscles,
the maximal contractile capacities appear correlated with
total CK activities (Fig. 8B).
Skeletal muscles are classified according to their myosin
isoform content which characterizes myosin ATPase
activity and contraction speed. Yamashita and Yoshioka [49]
suggested that the high shortening speed of fast-twitch fibers
is produced by high myosin ATPase activity and associated
with high CK-MM activity to supply ATP for contraction.
Watchko et al. [50] have shown that total CK activity does
relate to the MHC phenotype during development of four
rat skeletal muscles, as indexed by the ratio of adult to
developmental MHC isoform content, and proposed that the
role of CK as an energy buffer is greatest in muscles
expressing the fast lIb MHC isoform. Cells containing high
levels of CK activity also contain high amounts of PCr or
total creatine. It was also shown recently that human skeletal
muscle fibers having the highest proportion of the fast IIX
and IIA type of myosin heavy chain (MHC) isoforms
240

A differed in skeletal muscles (Fig. 9A). The efficacy of

62 myofibrillar bound creatine kinase was also compared. When
studying the responses to MgATP, especially in the presence
60 of PCr, no difference could be observed among striated
muscles (Fig. 9B). Thus, despite clear contractile capacity
58
and activity pattern differences among these muscles,
56
myofibrillar CK was as efficient in controlling adenine
~
0:1
nucleotides in the myofibrillar compartment. Indeed, when
U 5.4 comparing rat heart and chicken skeletal muscle, the amount
Co
of bound M-CK is the same [10, 28]. This observation is
5.2 also in accordance with the structural role of M-CK in the
M-band of myofilaments in these muscles [9], suggesting
50
that MM-CK content and efficacy would be linked to the
EDL number ofCK binding sites within the myofilaments. Indeed,
efficacy of myofibrillar creatine kinase increases during perinatal
development in parallel with the amount of bound M-CK and

B 6 D control
~ + PCr
A
3000
5 2500
0
'" I 2000

~
:2
4
~ 1500
Co ~ 1000
3 500
Ventricle Atria Soleus EDL
0
Fig. 8. Tissue specificity. (A) Total creatine kinase activity in IU/g ww in
Ventricle Atria Soleus EDL
muscles. (B) Mechanical properties of Triton X-I 00 treated rat muscle fibers.
Maximum active tension development measured at maximal calcium
concentration (pCa 4.5) expressed in mN/mm'.
150
B
contain highest resting amount of PCr [51]. Based on the
general correlation between PCr and total CK activity in 100
different muscle types, Iyengar [52] suggested a scheme
in which the amount of creatine bound to CK isoenzymes :;-
determines the level of creatine and PCr in the cells. The
e
. 50
recent experiments on mice having a knock-out ofthe gene :z
E
.....,
coding for the M-CK subunit, where normal PCr and
~
creatine levels are present despite 95% decrease in CK E
!- 0
activity [53], do not confirm such an hypothesis. This
correlation could be better explained by a co-regulation of Ventricle Atria Soleus EDL
expression of the creatine transporter and M-CK in muscles,
by a common nuclear factor. Fig. 9. Tissue specificity. Myofibrillarproperties. (A) Tension was measured
It appears therefore, that the more powerful and faster the in solutions of different calcium concentration and pCa for half maximal
tension development (pCaso) was calculated from the Hill equation. (B)
muscle, the higher the amount of total CK and creatine.
Sensitivity of rigor tension to MgATP and role of myofibrillar creatine kinase.
There is, however, the question of the function of CK in Tension was measured in solutions of decreasing MgATP concentration
myofilaments and mitochondria in these different tissues. and the pMgATP (-Log[MgATP]) for half maximal tension development
The calcium sensitivity of tension development was (pMgATP50) was calculated from the Hill equation. ***p < 0.001; statistically
identical in the two cardiac tissues (atria and ventricles) but different from respective pMgA TP 50 values in the absence of PCr.
 241

maturation ofthe M-band [45]. Dissociation between expression A

and localisation ofM-CK in myofilaments suggests that the
binding ofM-CK is dependent on the late expression of another --
~
-c
35
30
a-
component of the M-line. Another corollary of the structural co
basis of myofibrillar CK efficacy is its stability during path- 25
<I)

ological or adaptive states as compared with mi-CK (see [42] OJ

"0 20
for review). E
::L
Despite different activity pattern different, the role of
'-'
c
15
.2
myofibrillar CK is similar in the different muscle types. e
.c..
10
<II 5
....f0 0
Mitochondrial function and creatine kinase in different ><
<II Ventricle Atria Soleus EDL
muscle types E
>
The different muscle types also greatly differ in their
mitochondrial content and the role of oxidative metabolism
in energy supply for contraction. This was reflected by the
oxidative capacities of fibers, expressed per g of tissue dry
B
weight, which was also greatly different among the distinct 600
o control
rat muscle types, demonstrating the relative importance of 500 ~ + creatine
oxidative metabolism in the different tissues (Fig. lOA).
In these rat muscle tissues, the role of mitochondrial CK i::t 400
'-'
was further assessed by measuring the apparent Km's for Cl... 300
ADP of mitochondrial respiration in the presence and in
the absence of creatine (Fig. lOB). It can be seen that the
~
c2
. 200
apparent Km for ADP is =250 11M in atrial and ventricular E 100
~
fibers, consistently higher in slow skeletal muscle (=500 11M),
and very low in fast twitch skeletal muscle 20 11M). Clearly, 0
a tissue specificity of regulation of mitochondrial respiration Ventricle Atria Soleus EDL
is present among different muscle types (see also Saks et
al. in this issue). How this diversity of regulation can be Fig. 10. Tissue specificity. Mitochondrial properties. (A) Oxidative
capacities were measured at saturating concentrations of substrate and
explained and what controls the sensitivity of mitochondria
oxygen and calculated as !!moles of oxygen consumed per g of fibers dry
to adenine nucleotides is still a matter of debate. A growing weight. (B) Apparent Km of mitochondrial oxygen consumption for ADP.
body of evidence implicates the mitochondrial outer The respiration rates were measured after addition of increasing
membrane as a regulator of mitochondrial function, by concentrations of ADP in the presence and in the absence of20 mM creatine.
limiting the rate of metabolite flux under a variety of conditions Data were fitted to Michaelis-Menten equation using a non-linear fit.
[54,55]. The permeability of the outer mitochondrial membrane
to adenine nucleotides is thought to be controlled by the mitochondria were first described in mouse [62] and in rat
voltage dependent anion channel (VDAC or porin). Two fast twitch skeletal muscle [63]. The structural basis for this
isoforms of porin have been described in mammals and more absence of regulation is presently unknown. It can be
particularly in humans, but their role is not yet established speculated that either these muscles express a different
[56]. The low affinity of mitochondria for ADP within isoform of porin having different permeability properties, or
permeabilised fibers or cardiomyocytes, compared to iso- the unknown cytosolic factor is missing in fast skeletal
lated mitochondria was first described by Saks et al. [57,58]. muscle, or both. Due to the very high buffering capacity of
The same observations have been reported for in situ cytosolic CK system in these muscles (2000 IU/g of tissue
mitochondria of rat liver [59], and yeast [60]. It was proposed compared with 600 IU/g for cardiac tissue), ADP will be
that a proteinic cytosolic factor possibly related to cyto- effectively buffered thus damping the ADP signal and
skeleton controls the permeability ofporin pores in the outer preventing regulation of respiration rate by cytosolic ADP.
mitochondrial membrane [15, 61]. Interesting is the discovery This would be of minor importance in what is a highly
of tissue heterogeneity in the regulation of accessibility of glycolytic tissue.
ADP to mitochondrial matrix. In situ mitochondria of glyco- Another example of tissue specific regulation of mito-
lytic fast skeletal muscle are completely devoid of control chondrial function is shown when studying the role ofmito-
ofADP permeability. Very low apparent Km values for in situ chondrial creatine kinase. The apparent Km for ADP of
242
mitochondrial respiration was decreased by creatine to less
than 100 11M (about 3 times) in ventricles and slow-twitch
skeletal muscle in accordance with the mi-CK specific
activity (in units per mitochondrial protein) in soleus being
close to the ventricular one [15]. No difference in Km's in
the presence and in the absence of creatine was observed
in fast-twitch skeletal muscle. Comparing in vivo 02
consumption rate and ATP and PCr concentrations using
3 1P_NMR spectra, Kushmerick et al. [64] showed that
mechanisms of control of cellular respiration were both
qualitatively and quantitatively different in fast and slow twitch
skeletal muscles. In contrast to fast muscle, mitochondrial
regulation in slow twitch muscle could not be explained by
cytosolic ADP concentrations calculated from the CK
equilibrium reaction. However, this can be attributed to the
control of respiration by mitochondrial kinases and ADP
compartmentation in slow oxidative muscles. No important
role might be attributed to mi-CK in fast skeletal muscle
since, in these muscles, accessibility of cytosolic ADP to
mitochondria is high and no further kinetic effect of creatine
could be observed. These results show that mitochondria
differ quantitatively but also qualitatively among different
muscle types.
The functional coupling between mi-CK and oxidative
phosphorylation has been largely documented (for review
see: [8, 15, 17, 65]). It was shown that creatine exerts an
effective acceptor control on mitochondrial respiration, and
PCr is the end product of oxidative phosphorylation at
physiological conditions. It is generally admitted that
positively charged creatine kinases are fixed to negatively
charged cardiolipin molecules at the surface of the inner
membrane, which surround adenine nucleotide translocase
(ANT), therefore connecting these proteins into one
complex ([66, 67], see also Wallimann et al. in this issue).
In addition, mi-CK is present as dimeric or octameric forms
and, due to their identical top and bottom faces, octamers
can simultaneously interact with two opposing membranes
[68], thus forming contact sites between the inner and the
outer mitochondrial membranes. Mi-CK has also been
consistently observed along the cristae membrane, bound to
the outer surface of the inner membrane and not in close
contact with the outer membrane. Three possible ex-
planations have been mainly proposed for this coupling. It
was proposed to result from a tight structural and functional
coupling between the ATP/ADP translocase and creatine
kinase, with direct channelling of newly produced ATP to
mi-CK, phosphorylation of creatine to PCr with ADP
production, and channelling of this ADP back to the mito-
chondrial matrix. In this reaction, by increasing the 10calATP
concentration and removing the product ADP, ANT drives
the creatine kinase reaction towards phosphocreatine
production (for review see [15]). In addition, the increasing
evidences of low permeability of the outer mitochondrial
membrane for adenine nucleotides suggests that this
membrane creates a diffusion barrier within the inter-
membrane space, and concentration gradients, building up
some dynamic, rate dependent compartmentation of ATP in
the intermembrane space [15, 69, 70]. The discovery that
octameric mi-CK interacts with both the inner and the outer
mitochondrial membrane within the contact sites and the
isolation of ANT/mi-CKIporin complexes forming highly
ordered multi-enzyme complexes allowing efficient substrate
channelling has propelled the contact sites as the structural
basis for compartmentation ([8, 71], and Wallimann et al.
in this issue). These three explanations may not be exclusive
and experimental evidences for all mechanisms have been
provided.
Surprisingly, rat atrial tissue does not exhibit mi-CK
coupling to oxidative phosphorylation (Fig. 9B). At present
we do not know whether this observation is general for atrial
tissues or whether this is specific to rat heart. Mi-CK is
present in rat atria, although at a lower specific activity than
in ventricle (0.86 0.07 vs. 3.24 0.15 IU/mg). The lower
content of mi-CK, however, is not sufficient to account for
the difference since a similar specific mi-CK activity in
chicken heart (see above) is able to induce a 20% increase
in respiration rate by creatine. The presence in rat atria of both
mi-CK and low permeability for ADP without functional
coupling is striking. On the other hand, in smooth muscles,
which express the other mi-CK isoform, mi-Cku' mi-CK
also appears not to be coupled to oxidative phosphorylation
[35, 36]. However, atrial tissue expresses the same mi-CK
isoform (mi-CKs) and presents the same proportion of
octameric form as ventricular tissue [38], suggesting that
neither the mi-CK isoform nor the oligomeric state explain
the absence of coupling. This suggests that the three
dimensional arrangement of proteins within the inter-
membrane space is of primary importance. Whether mi-CK
is localized close to the ANT or within contact sites in rat
atrial mitochondria is not known. Nevertheless, CK is not
the only kinase present in mitochondria, and another pathway
may be involved in the regulation of energy fluxes within
atrial cells. This tissue has a lower oxidative capacity and
depends more on the energy from glycolysis. It is possible
that, as the CKIAK (adenyl ate kinase) ratio is lower in atria,
AK plays a more important role in energy transfer than in
ventricles [38]. It has been shown that AK may take over a
substantial proportion of energy-rich phosphoryl transfer
when CK is inhibited: from studies of the kinetic behavior
of AK in intact diaphragm muscle, it was suggested that AK-
catalyzed phosphotransfer functionally couples ATP-con-
suming processes to anaerobic glycolytic ATP generation
[70]. Interestingly, higher AK and lower CK specific
activities have been also found in uterine mitochondria as
compared to ventricular ones [73]. The difference between
atrial and ventricular mi-CK function highlights the im-

portance of local architecture as a determining factor of
mitochondrial regulation. As a consequence of spatial
organization, mi-CK can function according to two modes
in the mitochondrial intermembrane space. Either the ADP
regeneration mode encountered in smooth muscle or atria,
where mi-CK is active in the intermembrane space and can
regenerate ADP, and the ADP amplification mode where
CK is kinetically coupled to oxidative phosphorylation and
can enhance respiration rate.
Clearly, the diffusion barrier through the outer mito-
chondrial membrane, compartmentation of adenine nucleo-
tides and role ofCK and other mitochondrial kinases provide
a means for mitochondrial specialization. It may seem, at first,
paradoxical that access of ADP to oxidative phosphorylation
machinery is restricted in oxidative tissues where a fund-
amental prerequisite is that energy production is matched
to energy utilization. However, it can be proposed that, on
the contrary, this low permeability of mitochondria to bulk
ADP enables oxidative phosphorylation to be driven more
efficiently by specialized kinases which are bound to the outer
or inner mitochondrial membranes, and are functionally
coupled to oxidative phosphorylation. Even within striated
muscles, the diversity of regulation of the access of
phosphate acceptor to mitochondrial matrix through the
mitochondrial membranes provides a means to adapt
mitochondrial function to the specific needs of each muscle,
apart from simple variations in oxidative capacities. This is
achieved through expression of tissue specific kinases,
channels and spatial organization of these proteins within the
intermembrane space [70]. In this manner, each metabolic
pathway, containing the counterpart kinase would be directly
connected to the mitochondria which could in tum respond
effectively and quickly to any energetic imbalance, with low
interference with the others. Such pathways have already
been identified in other tissues and open new and exciting
fields of investigation for mitochondrial regulation [70]. A
comparative approach based on recognized tissue specificity
of mitochondrial regulation may help in understanding the
structural basis of adenine nucleotide compartmentation in
mitochondria.
Creatine kinase system in muscles
It can be seen that there is a tissue specific expression and
organization of the CK system in muscles. Two main
functions can be ascribed to the CKJPCr system. First it is
involved in the temporal and spatial buffering of ATP and
ADP during fast transitions in contractile activity. This is
particularly important in fast twitch skeletal muscles which
exhibit a tenfold change in ATP consumption within a few
seconds. In these muscles (e.g. E.D.L.) the high buffering
capacity is supported by high MM-CK activity and high
243
content of creatine while the amount ofmi-CK is negligible
(less than 5%). Newsholme et al. [74] have compared the
maximum activities of arginine kinase and CK in various
muscle of different species to the calculated maximum rate
ofATP turnover. The highest activities of arginine kinase and
creatine kinase are present in muscles that depend on
glycolysis rather than the tricarboxylic cycle for energy
production [75]. Most of the CK activity in these muscles
is cytosolic with only 5-10% being associated with the
sarcoplasmic reticulum and myofibrils. CK activity is
considerably greater (around 10 times) than the maximum
rate of ATP turnover [74]. These muscles develop burst of
intense activity at the expenses of energy reserves and are
highly fatiguable. This type of activity sharply depends on
high amount of rapidly mobilizable energy sources and
efficient ADP removal. In these muscles, the PCr/CK
system functions as an important reserve of energy rich
phosphates, and phosphotransfer will function between
glycolytic complexes associated with intracellular structures
and ATPases.
On the other hand, oxidative tonic muscles having pro-
longed or cyclic activity like postural skeletal muscles (i.e.
soleus) or ventricle are strictly dependent on efficient
energy production by mitochondria and have slower rates of
tension development. In general, oxidative muscles contain
lower CK activity compared with the glycolytic ones.
Among oxidative muscle, cardiac muscle can be considered
as the most oxidative and maximal ATPase activity is only a
few times lower than CK rates [76]. Oxidative muscles rely
more on simultaneous energy production and energy transfer
from mitochondria than on energy reserves. Accordingly,
they possess a high amount of mitochondria, a high specific
activity of mi-CK and a high relative proportion of bound
CK' s, but with lower total CK and creatine contents. Indeed,
in ventricular muscle, at least 20% of total CK is associated
with myofibrils [77] whereas mitochondrial CK represents
20-30%. Additional binding sites in the sarcoplasmic
reticulum, plasma membrane and nucleus results in soluble
CK being at most 50% of total activity. Taken that total CK
activity is 3 times less in heart than in skeletal muscle and
that half of this CK is bound, this indicates that cytosolic
CK is 6-7 times less in ventricle than in white skeletal
muscle. In these muscles, changes in workload should be
reflected in oxygen consumption and the response time of
mitochondria is of the order of a few seconds ([78] and
review by van Beek et al. therein). Changes in workload and
increase in metabolic rates can proceed without marked
changes in PCr and adenine nucleotide contents, supporting
the alternative view that compartmentation and high energy
phosphoryl transfer through phosphotransferases regulate
metabolic rates. There have been extensive examination of
unidirectional NMR determined fluxes of the CK reaction
following increased workloads, and conflicting results have
244
appeared. Increased unidirectional CK fluxes with workload
was observed in heart muscle [76, 79], but not in skeletal
muscle [80] thus disputing the phosphocreatine shuttle
models of cellular energetics. However, increased
unidirectional CK fluxes are not a prerequisite for establish-
ing the CK shuttle hypothesis. Ifwe consider that myofibrillar
or bound CKs work close to equilibrium at rest, upon
muscle activation, bound CK at producing site will be
shifted towards PCr formation while bound CK at the
utilization site will be shifted in the direction of ATP
formation. In this way, at nearly constant PCr and ATP
concentrations, the total unidirectional fluxes during the
active states will be equal to that in the resting state. Thus,
NMR data have to be taken cautiously and do not give at
present decisive arguments for or against the relevance of
CK shuttle in vivo. The clear and abundant demonstration
of functional coupling are more informative in this sense
(see also discussions in [81, 82]).
It is noteworthy that, maneuvers which are known to
induce phenotypic changes in skeletal muscle from a
glycolytic fast skeletal muscle profile to a slow oxidative
profile, as training or electro-stimulation also induce a
change in CK profile. The activity and relative proportion
of mi-CK increases almost linearly with the duration of
electro-stimulation of a fast-twitch rabbit muscle; this is
correlated with increased citrate synthase activity and
decreased level of glycolytic enzymes while total CK
activity is decreased [83]. An increase in the expression of
mi-CK was also described in skeletal muscle of marathon
runners after intense training [84]. As a confirmation of the
participation of the CK system in the muscle phenotype,
tissue specific adaptations to alterations of the CK system
have been recently described. Genetic null mutation of M-
CK in mice is compensated by a facilitation of adenine
nucleotide transportation between mitochondria and
cellular ATPases in oxidative muscles (decreased apparent
Km for ADP), where the function of CK in energy transport
is essential [62]. In fast skeletal muscle, the depression of
the energy buffering function is overcome by an increase in
energy producing potential (increase oxidative capacities
due to higher mitochondrial density) [53, 62]. Animals can
be depleted of creatine by long term feeding of guanidino-
propionic acid, a slowly metabolizable analog of creatine,
thus decreasing the efficacy of the creatine kinase shuttle.
Such studies have also shown tissue specific metabolic
adaptations. Skeletal muscle and brain increased their
oxidative capacities [86, 87], while cardiac muscle hyper-
trophied and decreased its contractile speed [88]. Thus, in
addition to a clear tissue specificity ofCK isoform expression
and organization among muscles having different metabolic
and functional requirements, there exists a tissue specificity
of adaptation to CK system disturbances.
It is also of interest that, striated and smooth muscles,
although having obvious structural and functional differences,
exhibit certain similarities in their CK system. Smooth
muscle, although expressing mainly B-CK isoform, exhibit
a correlation between the amounts of creatine and total CK
[52, 89]. Phasic muscles have higher enzyme and substrate
contents than tonic smooth muscles [89]. In smooth muscle,
BB-CK is structurally associated with and functionally
coupled to the contractile machinery [35, 90-92], while mi-
CK is compartmentalized but not functionally coupled to
oxidative phosphorylation [35, 36]. There is much evidence
that creatine kinase system can be highly hormonally
regulated and is an integral part in the developmental process
of the uterine muscle occurring during gestation. The CK
system respond by changes in isoenzymes specific activity,
intracellular distribution and localization in preparation of
the increase in energy demand for delivery [35].
Conclusions
Although creatine kinase is present among all vertebrates,
compartmentation of creatine kinase isoenzymes and
efficient functional coupling to energy producing and utilizing
reactions appears in birds, but is only fully developed in adult
mammalian ventricular muscle cells. This is associated with
the increased degree of cellular organization and complexity
of the highly differentiated mammalian muscle cells.
Moreover, the creatine kinase system exhibits a high degree
of plasticity, both in a quantitative manner with variable
amounts of enzyme and substrates, and in a qualitative
manner, with the presence of specific isoenzymes, their
subcellular localization and ability to undergo functional
interactions with various enzymes and transporters involved
in cellular energy transduction. The creatine kinase system
can fulfil two main roles. With a high spatial and temporal
buffer capacity, CK can provide fast and quickly mobilizable
energy for a given amount of time, which is proportional to
the amount of available substrate. In this way, it allows for fast
and powerful but brief muscle activity characteristic of fast
glycolytic muscles. With a lower buffering capacity but higher
mitochondrial activity, it participates in efficient regulation
and fine tuning of energy production and utilization, allowing
slower but sustained and adjustable mechanical activity
characteristic of slow oxidative cardiac and skeletal muscles.
In the cytosol of more glycolytic frog and fetal heart, the
CK system mainly plays a role as an energy rich phosphate
buffer.
In adult mammals, the increased efficiency of the cardiac
muscle, is achieved by cell hyperplasia, increased intra-
cellular compartmentation and organization, increased
aerobic metabolism and elaboration of specialized energy
transduction pathways for excitation-contraction coupling,
formed by compartmentalized creatine kinase isoenzymes

connected through a cytosolic pool offree enzyme. The CK
system can no longer be considered only as a reservoir of
energy rich phosphate. On the contrary, it has evolved to be
an integral part ofthe muscle phenotype, participating in the
high energetic efficiency of highly differentiated adult
mammalian cells.
From simple buffer function in frog heart, to compart-
mentation of CK to sites of energy utilization and production
in adult mammalian heart, the CK system shows an increasing
complexity throughout evolution leading to complex ex-
pression patterns and intracellular organization, that are
undoubtedly linked to muscle function and fine regulation
of cellular energy fluxes.
Acknowledgements
The authors wish to acknowledge R. Fischmeister, 1. Hoerter
and V. Saks for helpful discussions, and P. Mateo and P.
Lechene for engineering assistance, and F. Boussac for
secretarial assistance. RV-C is supported by Centre
National de la Recherche Scientifique. This study was
supported by grants from Association Franc;aise contre les
Myopathies, Fondation pour la Recherche Medicale and
Centre National d'Etudes Spatiales.
References
I. Booth FW, Thomason DB: Molecular and cellular adaptation of muscle
in response to exercise: Perspectives of various models. Physiol Rev
71: 541-585, 1991
2. Swynghedauw B: Developmental and functional adaptation of con-
tractile proteins in cardiac and skeletal muscles. Physiol Rev 66: 710-
771,1986
3. Lompre A-M, Mercadier J-J, Schwartz K: Changes in gene expression
during cardiac growth. Intern Rev Cyto1124: 137-180, 1991
4. Pette 0, Staron RS: Cellular and molecular diversities of mammalian
skeletal muscle fibers. Rev Physiol Biochem Pharmacoll16: 1-76, 1990
5. Schiaffmo S, Reggiani C: Molecular diversity of myofibrillarproteins: Gene
regulation and functional significance. Physiol Rev 76: 371-422, 1996
6. Penman S: Rethinking cell structure. Proc Nat! Acad Sci USA 92: 5251-
5257,1995
7. Payne RM, Haas RC, Strauss AW: Structural characterization and
tissue-specific expression of the messenger RNAs encoding iso-
enzymes from two rat mitochondrial creatine kinase genes. Biochim
Biophys Acta 1089: 352-361,1991
8. Wyss M, Smeitink J, Wevers RA, Wallimann T: Mitochondrial creatine
kinase - A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119-166,1992
9. Wallimann T, Eppenberger HM: Localization and function ofM-line-
bound creatine kinase. M-band model and creatine phosphate shuttle.
Cell Muscle Mot 6: 239-285, 1985
10. Ventura-Clapier R, Veksler V, Hoerter JA: Myofibrillar creatine
kinase and cardiac contraction. Mol Cell Biochem 133: 125-
144, 1994
II. Levitskii DO, Levchenko TS, Saks VA, Sharov VG, Smirnov VN:
245
Functional coupling between Ca2+-ATPase and creatine phosphokinase
in sarcoplasmic reticulum of myocardium. Biochimiya 42: 1766-1773,
1977
12. RossiAM, Eppenberger HM, Volpe P, Cotrufo R, Wallimann T: Muscle-
type MM creatine kinase is specifically bound to sarcoplasmic
reticulum and can support Ca- 2+ uptake and regulate local ATP/ADP
ratios. J Bioi Chern 265: 5258-5266, 1990
13. Korge P, Byrd SK, Campbell KB: Functional coupling between
sarcoplasmic-reticulum-bound creatine kinase and CaY-ATPase. Eur
J Biochem 213: 973-980,1993
14. Minajeva A, Ventura-Clapier R, Veksler V: Ca'+ uptake by cardiac
sarcoplasmic reticulum ATPase in situ strongly depends on bound
creatine kinase. Pfliigers Arch-Eur J Physiol432: 904-912, 1996
15. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells
- Role of coupled creatine kinases in in vivo regulation of cellular
respiration- A synthesis. Mol Cell Biochem 133: 155-192, 1994
16. Zeleznikar RJ, Dzeja PP, Goldberg NO: Adenylate kinase-catalyzed
phosphoryl transfer couples ATP utilization with its generation by
glycolysis in intact muscle. J BioI Chern 270: 7311-7319, 1995
17. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands -
The phosphocreatine circuit for cellular energy homeostasis. Biochem
1281: 21-40,1992
18. Mitchell P: Foundations of vectorial metabolism and osmochemistry.
BiosciRep 11:297-346, 1991
19. Muhlebach SM, Gross M, Wirz T, Wallimann T, Perriard JC, Wyss M:
Sequence homology and structure predictions of the creatine kinase
isoenzymes. Mol Cell Biochem 133: 245-262,1994
20. Ventura-Clapier R, Mekhfi H, Vas sort G: Role of creatine kinase in force
development in chemically skinned rat cardiac muscle. J Gen Physiol
89: 815-~m, 1987
21. Mayoux E, Coutry N, Lechene P, Marotte F, Hoffmann C, Ventura-
Clapier R: Effects of acidosis and alkalosis on mechanical properties of
hypertrophied rat heart fiber bundles. Am J Physiol266: H2051-H2060,
1994
22. Fabiato A: Computer programs for calculating total from specified free
or free from specified total ionic concentrations in aqueous solutions
containing multiple metals and ligands. Meth Enzymol157: 378-417,
1988
23. Veksler VI, Kuznetsov AY, Sharov VG, Kapelko VI, Saks VA: Mito-
chondrial respiratory parameters in cardiac tissue: A novel method of
assessment by using saponin-skinned fibers. Biochim Biophys Acta
892: 191-196,1987
24. Ingwall JS: Whole-organ enzymology of the creatine kinase system in
heart biochemical Society. Transactions 19: 1005-1010, 1991
25. Christensen M, Hartmund T, Gesser H: Creatine kinase, energy-rich
phosphates and energy metabolism in heart muscle of different vertebrates.
J CompPhysiol-Biochem SystEnvPhysiol164: 118-123, 1994
26. Fabiato A: Calcium release in skinned cardiac cells: Variations with
species, tissues, and development. Fed Proc 41: 2238-2244,1982
27. Legssyer A, Arrio-Dupont M: Mitochondrial isoenzyme of creatine
phosphokinase in frog heart. Comp Biochem Physiol89B: 251-255, 1988
28. Wallimann T, Turner DC, Eppenberger HM: Localisation of creatine
kinase in myofibrils. II. Chicken heart muscle. J Cell BioI 75: 318-325,
1977
29. Ventura-Clapier R, Saks VA, Vassort G, Lauer C, Elizarova G: Reversible
MM creatine kinase binding to cardiac myofibrils. Am J Physiol253:
C444-C455,1987
30. Hoppeler H, Kayar SR: Capillarity and oxidative capacity of
muscles. News in Physiol Sci 3: 113-116, 1988
3 1. Diamond JM: Evolutionary physiology - The red flag of optimality.
246
Nature 355: 204-206, 1992
32. Suarez RK: Upper limits to mass-specific metabolic rates. Ann Rev
Physiol58: 583--605, 1996
33. Barth E, Stammler G, Speiser B, Schaper J: Ultrastructural quantitation
of mitochondria and myofilaments in cardiac muscle from 10 different
animal species including man. J Mol Cell Cardiol24: 669-681, 1992
34. Hoerter J, Mazet F, Vas sort G: Perinatal growth of the rabbit cardiac
cell: Possible implications for the mechanism of relaxation. J Mol Cell
Cardiol13: 725--740,1981
35. Clark JF, Khuchua Z, Kuznetsov A, Saks VA, Ventura-Clapier R:
Compartmentation of creatine kinase isoenzymes in myometrium of
gravid guinea-pig. J Physiol466: 553-572, 1993
36. Clark JF, Kuznetsov AY, Khuchua Z, Veksler V, Ventura-Clapier R, Saks
V: Creatine kinase function in mitochondria isolated from gravid and
non-gravid guinea-pig uteri. FEBS Lett 347: 147-151, 1994
37. Vannier C, Veksler Y, Mekhfi H, Mateo P, Ventura-Clapier; R: Functional
tissue and developmental specificities of myofibrils and mitochondria
in cardiac muscle. Can J Physiol Pharmacol 74: 23-31, 1996
38. Anflous K, Veksler V, Mateo P, Samson F, Saks V, Ventura-Clapier R:
Mitochondrial creatine kinase isoform expression does not correlate
with its mode of action. Biochem J 322: 73-78, 1997
39. Hossle HP, Schlegel J, Wegmann G, Wyss M, Bohlen P, Eppenberger HM,
Wal1imann T, Perriard JC: Distinct tissue specific mitochondrial creatine
kinase from chicken brain and striated muscle with conserved CK
framework. Biochem Biophys Res Commun 151 :408-416, 1988
40. Schnyder T, Rojo M, Furter R, Wallimann T: The structure of
mitochondrial creatine kinase and its membrane binding properties.
Mol Cell Biochem 133: 115--123,1994
41. Schlegel J, Zurbriggen B, Wegmann G, Wyss M, Eppenberger HM,
Wallimann T: Native mitochondrial creatine kinase forms octameric
structures 1. Isolation of two interconvertible mitochondrial creatine
kinase forms, dimeric and octameric mitochondrial creatine kinase:
Characterization, localization, and structure function relationships. J
Bioi Chern 263: 16942-16953, 1988
42. Veksler V, Ventura-Clapier R: In situ study ofmyofibrils, mitochondria
and bound creatine kinases in experimental cardiomyopathies. Mol
Cell Biochem 133: 287-298,1994
43. Veksler V, Ventura-Clapier R: Ischaemic metabolic factors: high
inorganic phosphate and acidosis modulate mitochondrial creatine
kinase functional activity in skinned cardiac fibres. J Mol Cell Cardiol
26: 335--339, 1994
44. Saks VA, Kapelko VI, Kupriyanov vv, Kuzuetsov AY, Lakomkin VL, Veksler
VI, Sharov VG, Javadov SA, Seppet EK, Kairane C: Quantitative evaluation
of relationship between cardiac energy metabolism and post-ischemic
recovery of contractile function. J Mol Cell Cardiol 21: 67-78, 1989
45. Hoerter J, Kuznetsov A, Ventura-Clapier R: Functional development of
the creatine kinase system in perinatal rabbit heart. Circ Res 69: 665--
676, 1991
46. Hoerter JA, Ventura-Clapier R, Kuznetsov A: Compartmentation of
creatine kinases during perinatal development of mammalian. Mol Cell
Biochem 133/134: 277-286, 1994
47. Payne RM, Strauss AW: Expression of the mitochondrial creatine kinase
genes. Mol Cell Biochem 133: 235--243,1994
48. Sommer JR, Jennings, RB: Ultrastructure of cardiac muscle. In: HA
Fozzard, E Haber, RB Jennings,AM Katz, HE Morgan (eds). The Heart
and Cardiovascular System. 1991, pp 3-50
49. Yamashita K, Yoshioka T: Profiles of creatine kinase isoenzyme
compositions in single muscle fibres of different types. J Muscle Res
Cell Mot 12: 37-44,1991
50. Watchko JF, Daood MJ, LaBella JJ: Creatine kinase activity in rat skeletal
muscle relates to myosin phenotype during development. Pediatric Res
40: 53-58, 1996
51. Sant'Ana Pereira JAA, Sargeant AJ, Rademaker ACHJ, de Haan
A, Van Mechelen W: Myosin heavy chain isoform expression
and high energy phosphate content in human muscle fibres at
rest and post-exercise. J Physiol 496: 583-588, 1996
52. Iyengar MR: Creatine kinase as an intracellular regulator. J Muscle Res
Cell Mot 5: 527-534, 1984
53. van Deursen J, HeerschapA, Oerlemans F, Ruitenbeek W, Jap P, Laak
H, Wieringa B: Skeletal muscles of mice deficient in muscle creatine
kinase burst activity. Cell 74: 621--631, 1993
54. Gellerich FN, Bohnensack R, Kunz W: Role of the mitochondrial outer
membrane in dynamic compartmentation of adenine nucleotides. In: A
Azzi et al. (eds). Anion carriers of mitochondrial membranes. Springer-
Verlag, Berlin Heidelberg, 1989, pp 349-359
55. Lee A-C, Xu X, Colombini M: The role of pyridine dinucleotides in
regulating the permeability of the mitochondrial outer membrane. J
Bioi Chem271: 26724--26731,1996
56. Blachy-Dyson E, Zambronicz EB, Hong Yu W, Adams Y, McCabe ER,
Adelman J, Colombini M, Forte M: Cloning and functional expression
in yeast of two human isoforms of the outer mitochondrial membrane
channel, the voltage-dependent anion channel. J Bioi Chern 268: 1835-
1841,1993
57. Saks VA, Belikova YO, Kuznetsov AV: In vivo regulation of mitochondrial
respiration in cardiomyocytes: Specific restrictions for intracellular
diffusion ofADP. Biochim BiophysActa 1074: 302-311, 1991
58. Saks VA, Vasil'eva E, Belikova YO, Kuznetsov AY, Lyapina S, Petrova
L Perov NA: Retarded diffusion of ADP in cardiomyocytes: possible
role of mitochondrial outer membrane and creatine kinase in cellular
regulation of oxidative phosphorylation. Biochim Biophys Acta 1144:
134--148,1993
59. Fontaine EM, Keriel C, Lantuejoul S, Rigoulet M, Leverve XM, Saks
VA: Cytoplasmic cellular structures control permeability of outer
mitochondrial membrane for ADP and oxidative phosphorylation in
rat liver cells. Biochem Biophys Res Comm 213: 138-146, 1995
60. Averet N, Rigoulet M, Fitton V, Bunoust 0, Dejean L, Guerin B: In
permeabilised yeast cells the retarded diffusion controlling the
mitochondrial respiration is due to porin channel closing. Biochim
Biophys Acta, EBEC Reports 9: 119, 1996
61. Saks VA, Tiivel T, Kay L, Novel-ChatY, Daneshad Z, RossiA, Fontaine
E, Keriel C, Leverve X, Ventura-Clapier R, Anflous K, Samuel JL,
Rappaport L: On the regulation of cellular energetics in health and
disease. Mol Cell Biochem 160/161: 195--208, 1996
62. Veksler VI, Kuznetsov AY, Anflous K, Mateo P, van Deurscn J, Wieringa
B, Ventura-Clapier R: Muscle creatine kinase-deficient mice .2. Cardiac
and skeletal muscles exhibit tissue-specific adaptation of the mito-
chondrial function. J Bioi Chern 270: 19921-19929, 1995
63. KuznetsovAY, Tiivel T, SikkP, KaambreT, Kay L, DaneshradZ, Rossi.
A, Kadaja L, Peet N, Seppet E, Saks VA: Striking differences between
the kinetics of regulation of respiration by ADP in slow-twitch and
fast-twitch muscles in vivo. Eur J Biochem 241: 909-915, 1996
64. Kushmerick MJ, Meyer RA, Brown TR: Regulation of oxygen
consumption in fast-twitch and slow-twitch muscle. Am J Physiol263:
C598-C606, 1992
65. Gellerich FN, Khuchua ZA, Kuznetsov AV: Influence of the mito-
chondrial outer membrane and the binding of creatine kinase to the
mitochondrial inner membrane on the compartmentation of adenine
nucleotides in the intermembrane space of rat heart mitochondria.
Biochim Biophys Acta 1140: 327-334, 1993
66. Muller M, Moser R, Cheneval D, Carafoli E: Cardiolipin is the membrane
receptor for mitochondrial creatine phosphokinase. J BioI Chern 260:
3829-3843, 1985
67. Stachowiak 0, Dolder M, Wallimann T: Membrane-binding and lipid
vesicle cross-linking kinetics of the mitochondrial creatine kinase octamer.
Biochemistry 35: 15522-15528, 1996
68. Rojo M, Wallimann T: The mitochondrial ATP/ADP carrier:

Interaction with detergents and purification by a novel procedure.
Biochim Biophys Acta 1187: 360-367,1994
69. Gellerich FN, Kapischke M, Kunz W, Neumann W, Kuznetsov A,
Brdiczka D, Nicolay K: The influence ofthe cytosolic oncotic pressure
on the permeability of the mitochondrial outer membrane for ADP:
Implications for the kinetic properties of mitochondrial creatine kinase
and for ADP channeling into the intermembrane space. Mol Cell
Biochem 133: 85-104, 1994
70. Brdiczka D, Wallimann T: The importance of the outer mitochondrial
compartment in regulation of energy metabolism. Mol Cell Biochem
133: 69--83, 1994
71. Brdiczka D: Contact sites between mitochondrial envelope membranes
- structure and function in energy-transfer and protein-transfer.
Biochim Biophys Acta 1071: 291-312,1991
72. Zeleznikar RJ, Dzeja PP, Goldberg ND: Adenylate kinase-catalyzed
phosphoryl transfer couples ATP utilization with its generation by
glycolysis in intact muscle. J Bioi Chern 270: 7311-7319, 1995
73. Clark JF, Kuznetsov AY, Radda GK: ADP-regenerating enzyme systems
in mitochondria of the guinea-pig myometrium and heart. Am J Physiol
272: C399-C404, 1997
74. Newsholme EA, Beis I, Leech AR, Zammit VA: The role of creatine
kinase and arginine kinase in muscle. Biochem J 172: 533-537, 1978
75. Brecker JL, Winters RB: MyosinATPase activity during avian cardiac
and skeletal muscle development. Mech Ageing Develop 46: 47-58,
1988
76. Kupriyanov VV, Steinschneider AY, Rhuge EK, Kapelko VI, Lakornkin
VL, Smimov VN, Saks VA: Regulation of energy flux through the creatine
kinase reaction in vitro and in perfused rat heart. 31P-NMR studies.
Biochim Biophys Acta 805: 319--331, 1984
77. Ventura-Clapier R, Saks VA, Vassort G, Lauer C, Elizarova G: Reversible
MM creatine kinase binding to cardiac myofibrils. Am J Physiol253:
C444-C455, 1987
78. Van Beek JHGM, WesterhofN: Response time of cardiac mitochondrial
oxygen consumption to heart rate steps. Am J Physiol 260: H613-
H625,1991
79. Bitt! JA, Ingwall JS: Reaction rates of creatine kinase andATP synthesis
in the isolated rat heart. A 31P-NMR magnetization transfer study.
J Bioi Chern 260: 3512-3517, 1985
247
80. McFarland EW, Kushmerick MJ, Moerland TS: Activity of creatine
kinase in a contracting mammalian muscle of uniform fiber type.
Biophys J 67: 1912-1924, 1994
81. Saks VA, Ventura-Clapier R,Aliev MK: Metabolic control and metabolic
capacity: two aspects of creatine kinase functioning in the cells. Biochim
Biophys Acta 1274: 81-88, 1996
82. Wallimann T: P-31-NMR-measured creatine kinase reaction flux in
muscle: A caveat! J Muscle Res Cell Mot 17: 177-181, 1996
83. Schmitt T, Pette D: Increased mitochondrial creatine kinase in
chronically stimulated fast-twitch rabbit muscle. FEBS Lett 188: 341-
344,1985
84. Apple FS, Rogers MA: Mitochondrial creatine kinase activity alter-
ations in skeletal muscle during long distance running. J Appl Physiol
61: 482-485,1986
85. Dzeja P, Noronha L, Zeleznikar R, Wieringa B, Goldberg N: ASBMBI
ASIPI AAljoint meeting, New-Orleans 1996, p 35
86. Shoubridge EA, Chaliss RAJ, Hayes DJ, Radda GK: Biochemical adaptation
in the skeletal muscle of rats depleted of creatine with the substrate
analogue b-guanidinopropionic acid. Biochem 1232: 125-131, 1985
87. O'Gorman E, Beutner G, Wallimann T, Brdiczka D: Differential effects
of creatine depletion on the regulation of enzyme activities and on
creatine-stimulated mitochondrial respiration in skeletal muscle, heart,
and brain. BiochimBiophysActa 1276: 161-170, 1996
88. Mekhfi H, Hoerter J, Lauer C, Wisnewsky C, Schwartz K, Ventura-
Clapier R: Myocardial adaptation to creatine deficiency in rats fed with
b-guanidinopropionic acid, a creatine analogue. Am J Physiol 258:
HI151-HI158,1990
89. Ishida Y, Riesinger I, Wallimann T, Paul RJ: Compartmentation ofATP
synthesis and utilization in smooth muscle: Roles of aerobic glycolysis
and creatine kinase. Mol Cell Biochem 133: 39--50, 1994
90. Clark JF, Khuchua Z, Boehm E, Ventura-Clapier R: Creatine kinase
activity associated with the contractile proteins of the guinea-pig
carotid artery. J Muscle Res Cell Mot 15: 432-439, 1994
91. Clark, JF, Khuchua Z, Ventura-Clapier R: Creatine kinase binding and
possible role in chemically skinned guinea-pig taenia coli. Biochim
Biophys Acta 1100: 137-145,1992
92. Clark JF: The creatine kinase system in smooth muscle. Mol Cell
BiochemI33:221-232,1994
Molecular and Cellular Biochemistry 184: 249-289, 1998.
1998 Kluwer Academic Publishers.

Theoretical modelling of some spatial and temporal

aspects of the mitochondrion/creatine kinase/
myofibril system in muscle
Graham 1. Kemp,1,2 David N. Manners, 1Joseph F. Clark, 1Mark E.
Bastin1and George K. Radda 1
iMRC Biochemical and Clinical Magnetic Resonance Unit, Oxford Radcliffe Hospital, Oxford; 2Department of
Orthopaedic and Accident Surgery, Royal Liverpool University Hospital, Liverpool, UK

Abstract
After discussing approaches to the modelling of mitochondrial regulation in muscle, we describe a model that takes account, in a
simplified way, of some aspects of the metabolic and physical structure of the energy production/usage system. In this model,
high-energy phosphates (ATP and phosphocreatine) and low energy metabolites (ADP and creatine) diffuse between the
mitochondrion and the myofibrillar ATPase, and can be exchanged at any point by creatine kinase. Creatine kinase is not assumed
to be at equilibrium, so explicit account can be taken of substantial changes in its activity of the sort that can now be achieved by
transgenic technology in vivo. The ATPase rate is the input function. OxidativeATP synthesis is controlled by juxtamitochondrial
ADP concentration. To allow for possible functional 'coupling' between the components of creatine kinase associated with the
mitochondrial adenine nucleotide translocase and the myofibrillar ATPase, we define parameters <p and \jf that set the fraction of
the total flux carried by ATP rather than phosphocreatine out of the mitochondrial unit and into the ATPase unit, respectively. This
simplification is justified by a detailed analysis of the interplay between the mitochondrial outer membrane porin proteins,
mitochondrial creatine kinase and the adenine nucleotide translocase. As both processes of possible' coupling' are incorporated
into the model as quantitative parameters, their effect on the energetics of the whole cell model can be explicitly assessed. The
main findings are as follows: (1) At high creatine kinase activity, the hyperbolic relationship of oxidative ATP synthesis rate to
spatially averaged ADP concentration at steady state implies also a near-linear relationship to creatine concentration, and a sigmoid
relation to free energy ofATP hydrolysis. At high creatine kinase activity, the degree of functional coupling at either the mitochondrial
or ATPase end has little effect on these relationships. However, lowering the creatine kinase activity raises the mean steady state
ADP and creatine concentrations, and this is exaggerated when <p or\jf is near unity (i.e. little coupling). (2) At high creatine kinase
activity, the fraction of flow at steady state carried in the middle ofthe model by ATP is small, unaffected by the degree of functional
coupling, but increases withADP concentration and rate ofATP turnover. Lowering the creatine kinase activity raises this fraction,
and this is exaggerated when <p or \jf is near unity. (3) Both creatine and ADP concentrations show small gradients decreasing
towards the mitochondrion (in the direction of their net flux), whileATP and phosphocreatine concentration show small gradients
decreasing towards the myosin ATPase. Unless <p = \jf "" 0 (i.e. complete coupling), there is a gradient of net creatine kinase flux
that results from the need to transform some of the 'adenine nucleotide flux' at the ends of the model into 'creatine flux' in the
middle; the overall net flux is small, but only zero if <p = \jf. A reduction in cytosolic creatine kinase activity decreases ADP
concentration at the mitochondrial end and increases it at the ATPase end. (4) During work-jump transitions, spatial average
responses exhibit exponential kinetics similar to those of models of mitochondrial control that assume equilibrium conditions for
creatine kinase. (5) In response to a step increase in ATPase activity, concentration changes start at the ATPase end and propagate
towards the mitochondrion, damped in time and space. This simplified model embodies many important features of muscle in
vivo, and accommodates a range of current theories as special cases. We end by discussing its relationship to other approaches to
mitochondrial regulation in muscle, and some possible extensions ofthe model. (Mol Cell Biochem 184: 249-289, 1998)

Key words: creatine kinase, heart, skeletal muscle, mitochondria, respiration, energy metabolism
Addressfor offPrints: G.J. Kemp, Department of Orthopaedic and Accident Surgery, Royal Liverpool University Hospital, Prescot Street, Liverpool L69 3BX, UK
250
1. Introduction
In skeletal muscle, ATP turnover can change by two orders
of magnitude in a fraction of a second. This requires precise
co-ordination ofATP supply with demand by the transmission
ofa signal from sites ofATPusage to sites ofATPproduction.
In high intensity exercise glycogenolysis is a major com-
ponent, but its control is not well understood in quantitative
detail [1, 2]. Oxidative (mitochondrial) ATP synthesis in
largely aerobic exercise has received more attention [3-11],
as we will briefly review. It is well known that oxidation rate
bears a hyperbolic relationship to [ADP] in vitro. The ADP
concentration for half-maximal oxidation (Km) is, classically,
20-30 JlM [12, 13], although lower values are reported e.g.
10-20 JlM [14-16], around 10 JlM [17,18], or even less than
5 JlM [16, 19], and the reasons for these discrepancies are not
understood. There have been two main experimental ap-
proaches to the problem in skeletal muscle in vivo. In
'aerobic' exercise, where glycogenolytic ATP synthesis and
net proton efflux can be ignored, the steady-state power
output [4, 20], the directly measured rate of oxygen con-
sumption [5, 6] and the Pi-to-ATP flux measured by magnetic
resonance saturation transfer [3] can all be used as estimates
of or surrogates for the rate of oxidativeATP synthesis. These
all have an approximately hyperbolic (Michaelis-Menten)
relationship to free [ADP] [3,5,6,20]: in particular, measure-
ments of oxygen consumption during exercise in rabbit [6]
and cat muscle [5], and measurements of power output during
aerobic exercise in human muscle [4, 20] suggests Km "" 30
JlM, although measurements of Pi-to-ATP magnetisation
transfer during steady state exercise in rat muscle suggest a
higher value, at least 90 JlM [3]. These data are supported by
measurements made during recovery from exercise: when
ischaemia is maintained after exercise, the lack of recovery
of Pi, phosphocreatine, ADP or AMP [21] implies that there
is no glycogenolyticATP synthesis (surprisingly, as all these
metabolites are clamped at their exercise values, and most
phosphorylase may be in the active a form [2]). It can
therefore be assumed that in aerobic recovery, only oxidative
ATP production funds phosphocreatine resynthesis. Further-
more, as there is no mechanical work, these can be taken as
equal, apart from basal ATP requirements (which are small:
see Table I), and a suprabasal component probably reflecting
post-contractile calcium pumping, which is also small [7]. So
the initial rate of phosphocreatine resynthesis is a reasonable
estimate ofthe end-exercise rate of oxidativeATP synthesis.
This is found to have a hyperbolic (Michaelis-Menten)
relationship to free ADP concentration [22, 23]: in human
muscle, analysis suggests Km "" 30 JlM [23,24], while in rat
muscle, measurements of d[PCr ]/dt as a function of [ADP] in
recovery from exercise suggest that Km is larger than this, at
least 90 JlM [22] (as in the magnetisation transfer measure-
ments in steady state exercise [3]).
There are still unexplained aspects. There is no explanation
of overshoots in phosphocreatine [5], ADP [25] and free
energy of ATP hydrolysis [26] during recovery, or of some
anomalous data from muscle during experimental acidosis
[27]. However, the case against the [ADP] control model in
exercise in perfused canine gracilis [11, 28] and stimulated
frog leg [7], rests heavily on the resting [ADP], and therefore
on the resting [PCr]/[TCr], and cannot be regarded as proved
[29, 30]. Thus although are some anomalies, oxidative ATP
synthesis in skeletal muscle behaves largely as if under the
control of free ADP concentration, acting at the adenine
nucleotide translocase of the inner mitochondrial membrane
according to a classical Michaelis-Menten hyperbolic rela-
tionship [12] (Section 2.3 will discuss some alternative
views).
The situation in cardiac muscle is apparently different.
High rates of oxidative synthesis are also found that increase
with heart rate and cardiac work (although the dynamic range
is smaller than in muscle). The role of changes in ADP
concentration (and other phosphorus metabolites) in the
response to work jumps has been controversial [14, 31].
Large changes inATP synthesis are reported to occur without
significant changes in phosphocreatine or ADP concentration
(e.g. [32-36]) and it has been suggested that at least part of
this may be because of changes in intramitochondrial con-
centration of calcium, that can stimulate pyruvate, succinate
and a-oxoglutarate dehydrogenases [37, 38]. However, in a
smaller number of studies changes in phosphocreatine and
ADP concentration have been observed in association with
changes in cardiac work (e.g. [39,40]), and these apparent
discrepancies remain unresolved. (The possible roles of
calcium-stimulation of intramitochondrial dehydrogenases
are reviewed in [38], and their implications for the present
work are discussed in Section 10.6.) Smooth muscle has a
very wide range of contractility and metabolic characteristics
and may depend significantly on glycolytic ATP synthesis
[41]. Generally, maximal rates ofATP turnover are lower than
in skeletal and cardiac muscle, but oxidative ATP synthesis
is still required to change in response to changes in tension
development and maintenance [42]. Details of its regulation
are, however, not well understood.
In this paper we confine ourselves to skeletal muscle, and
begin by briefly describing the aspects of the problem that
we hope to illuminate. Theories of mitochondrial control and
bioenergetic regulation in vivo are complicated by the
metabolic and physical structure of the energy production/
usage system. There are three important factors: interactions
between ATP turnover and the creatine kinase reaction;
functional 'coupling' between mitochondrial creatine kinase
and the adenine translocase, and between the myosinATPase
and its associated creatine kinase; and physical movement of
metabolites between sites ofATP usage and production. The
importance ofthe first of these, cytosolic creatine kinase [8,
 251

Table 1. Some definitions, symbols and values

Symbol Definition (and value)

General variables
x, f, t linear distance (~x = width of box = 10 11m), radial distance, time
[X]j general metabolite concentration in box j = I ... N
Xt"O pre-existing steady-state value of x
J~
J
diffusion flux of X out of box j = I ... N

ATP turnover fluxes

U ATP usage rate
Q (= Qc + QR) mitochondrial ATP synthesis rate ('coupled' and 'regulated' components)
QMAX maximum Q (= 1.0 mM sec-I for human skeletal muscle [24, 134, 135])
QMAx"' QMAX S calcium-independent and calcium-stimulated components ofQ MAX
U B=QB basal ATP turnover rate (= 0.01 mM sec-I for human skeletal muscle [136])
L ADP influx rate into mitochondrion

Creatine kinase fluxes'

C, CE net and unidirectional cytosolic creatine kinase fluxes (Cj = value in box j = I ... N)
C MAX maximum cytosolic flux (= 55 mM sec-I in skeletal muscle [53])
ImitoC, IATP,,,C sums of C in 'mitochondrial half and 'ATPase half' of model
W (=Wc + W R) mitochondrial creatine kinase flux ('coupled' and 'regulated' components)
W R.MAX maximum value ofW R(= 0.12 mM sec-I [79]b)

Kinetic. coupling and equilibrium parameters of creatine kinase

KCK ' Kc/ equilibrium constant (= 1.66 x 10' M-I [57]) and apparent equilibrium constant (= KCK[W] = 166 at pH 7) for cytosolic
creatine kinase
<p,1jI parameters describing overall coupling of mitochondrial and ATPase
S,p parameters describing coupling of mitochondrial CK to translocase and vice versa
Dissociation constants for mitochondrial creatine kinase (see Table 2)

K p' K", K jq, K ib, Kia' KIP' Kii' Klb Primary kinetic constants for cytosolic creatine kinase (see Table 2)
Koo-K06 Secondary kinetic constants for cytosolic creatine kinase (see Table 2)

Other symbols'
c oxygen consumption rate
Cp,Rm 'capacitance' and 'mitochondrial resistance' [8]
fb buffering factor= d[IPCr]/d(2[IATP] + [IADP]) [43]
~G free energy of ATP hydrolysis (standard free energy ~Go = -31.8 kJ mol-I)
Dp" DA, Dc cytosolic diffusion constants for Pi (= 3.3 x 10-10 m'sec- I), for ATP and ADP (= 2.8 x 10- 10 m'sec- I) and for phospho-
creatine and creatine (= 2.0 x 10-10 m'sec- I) [87])
E, Fq , FE generalised error signal and transfer functions in feedback loop
h, g, [T], Bx h = [W], g = [Mg2+], [T] = [PCr] + [ATP], Bx = [IX]/[Xn-]
kD second-order rate constant of collision between two creatine kinase molecules
K m, Kmapp ADP affinity of translocase (= 30 11M - see text) and whole mitochondrion
kp outer membrane porin permeability to ADP [89]
ks' kT apparent proportionality and rate constants relating Q to [PCr]
Ku value ofU at which Ca signal is half-maximal
M Jacobian matrix in which elements are M..IJ = ilFJilx.
1 J
P(T.A)'n' peT Dlo,)out mitochondrial binding probability (= 0.9), coupling probability [76]
PATPase' Pmito fraction of creatine kinase associated with the ATPase and with mitochondria
q thermodynamic coupling parameter [111]
radii of ATPase and myocyte in two-dimensional model
product of gas constant (= 8.31 J mol-I K- I) and absolute temperature (310 K)
= [IADP]/[IATP]

Concentrations and pH for human skeletal muscle

Total creatine [TCr] = 42.5 mM, total adenine nucleotides = 8.3 mM, pH = 7.00 [25]
Inhibitory [anion] = 25 mM [53]. [Mg2+] = 1 mM [118, 120]

'Nomenclature: C and C MAX are measured in direction of phosphocreatine breakdown, E is measured in either direction, and W and W RMAX are measured
in direction of phosphocreatine synthesis. bCalculated from values given per mitochondrial weight, assuming 0.6 I intracellular water (kg wet weight)-I
and 0.3 kg mitochondria (kg wet weighty-I [81]; measured in direction of phosphocreatine synthesis. 'Also A, B, C, D and E are terms in Eq. 48--53. kJ'
Is are terms in Eq. 91; A is defined in Eq. 51. Various kinetic and equilibrium constants are defined in Tables 2 and 3.
252
10, 11,43], is well-established, but there is no general account
of the effects of a severe reduction in creatine kinase of the
kind that can be obtained by genetic manipulation in vivo [44,
45] or by pharmacological means in vitro [46]. Nor, despite
the wealth of detail on the mechanisms of functional coupling
[47--49], is there a general account of the importance these
mechanisms have in vivo. In this paper we present an ap-
proach to a general account of the workings of these mech-
anisms in vivo, using a model that, although necessarily
simplified, offers a semiquantitative analysis of their inter-
actions. Some of this work has been presented in abstract
form [50, 51].
The structure ofthe paper is as follows. Section 2 considers
the cytosolic creatine kinase reaction as being at equilibrium,
and discusses the implications of this view. Section 3 reviews
and extends some approaches to modelling mitochondrial
control assuming that creatine kinase is at equilibrium.
Section 4 extends this by relaxing the assumption that creatine
kinase is at equilibrium. Section 5 considers the 'creatine
kinase circuit' and presents a general analysis of functional
'coupling' between mitochondrial creatine kinase and the
adenine translocase, and between the myosin ATPase and its
associated creatine kinase. Section 6 describes an approach
to the question of spatial diffusion of metabolites. Sections
7 and 8 presents some results of the complete model, at steady
state (Section 7) and during work-jump transitions (Section
8). Section 9 discusses this approach in relation to other
approaches to mitochondrial control in vivo. Section 10
discusses some limitations of the present approach and
possible extensions of the model. Lastly, Section 11 presents
a summary and conclusions.
2. Cytosolic creatine kinase considered
at equilibrium
2.1 The cytosolic creatine kinase reaction
As shown in Fig. 1 (which is based on, for example [10, 11,
52]), the mitochondrion is embedded in the creatine kinase
equilibrium [10,11]. This enzyme (E.C. 2.7.3.2) catalyses the
reaction ATP + Cr <=> ADP + PCr, where ATP means
MgATp 2-, Cr means free creatine, ADP means MgADP- and
PCr means phosphocreatine2-. This can also be arranged in
vitro [17]. (For most of what follows we can safely avoid any
detailed consideration of the ionic forms involved in this
reaction, although we return to the subject in Section 10.2.)
In normal skeletal muscle [53], in heart [54] and in vascular
smooth muscle [55, 56], this is near equilibrium under all
normal circumstances. The classical creatine kinase equili-
brium [57] states that
Cytosolic
per
creatine kinase
Mitochondrion
Q u
---14- ADP
Regulation ofATP
synthesis
Myosin
ATPase
Figure 1. Simple model of ATP turnover in the cell. In this model we
ignore the spatial dimension. See Table I for definitions of symbols. Fluxes
of 'high energy compounds' (phosphocreatine and A TP) and 'low energy
compounds' (ADP and creatine) pass between the mitochondrion on the
left and the myosin ATPase of the myofilament on the right. Between them,
cytosolic creatine kinase may catalyse a net flux C from phosphocreatine
and creatine and from ADP and ATP; the exchange flux is E. The output of
the mitochondrion and the input of the ATPase are both ATP; the input of
the mitochondrion and the output of the ATPase are both ADP. Mito-
chondrial A TP synthesis rate (Q) is under the control of cytosolic ADP
concentration. The ATPase rate (U) is imposed independently of any system
variables or other parameters. For simplicity, the figure omits inorganic
phosphate, which results from ATP hydrolysis at the ATPase and then
travels to the mitochondrion to participate in the synthesis of A TP (see
Section 6.5).
[ADP]/[ATP] = [Cr]/([PCr] [W]KCK) = ([Cr]/[PCr])/Kc/ (I)
where KCK is the equilibrium constant and KCK' is defined for
convenience as KCK[H+], and total creatine (TCr) does not
vary in the short term (a more complete account of the
creatine kinase equilibrium is given in Section 10.2). As the
muscle cell membrane is relatively impermeable to inorganic
phosphate (Pi), and as changes in ATP concentration are
always small (Section 2.2), changes in phosphocreatine
concentration are mirrored by changes in [Pi] so thati1[Pi] '"
-L1[PCr]. Before developing a general account of ATP-
buffering when creatine kinase activity is not sufficiently high
for the equilibrium assumption to be entirely valid, we must
consider the equilibrium case in more detail.
2.2 A TP-bufJering by the creatine kinase equilibrium
An important consequence of the existence of the creatine
kinase equilibrium is that ATP concentration is buffered
against mismatch between ATP usage and supply [10, 11]:
in the absence ofthe creatine kinase system a work jump would
lead to a catastrophic rise in ADP concentration (equal to the

fall inATP concentration). In the presence of such a system,
a workjump leads to a fall in phosphocreatine concentration
with little change inATPconcentration [10,11]. ThusADPis
free to act as error signal in a stable feedback loop, suggesting
a straightforward regulatory mechanism. This can be seen as
follows. Let [TA] = [ATP] + [ADP] (this is always close to
[ATP]). Differentiating Eq. 1 [10] shows
d[ATP]/d[PCr] = ([TA]/[TCrD([ATP][PCr])2/Kc/
"" ([TCr] [ADPD/([Cr] [PCrD (2)
=[ADP] {(I/[PCrD + (1/[Cr]))
It can be seen that as ADP concentration is always much
smaller than the other metabolite concentrations, d[ ATP]I
d[PCr] "" 0, and almost all the change in the concentration of
total 'high energy' compounds occurs in phosphocreatine
rather thanATP concentration. Thus if we define [T] =[ATP]
+ [PCr],
d[PCr]/d[T] = 1/ {I +d[ATP]/d[PCr]} "" 1 (3)
d[ATP]/d[T] = 1/{1 +d[PCr]/d[ADP]}
(4)
= 1-d[PCr]/d[T] "" 0 4.
This defines the effect of creatine kinase in buffering ATP
concentration. A fuller analysis is summarised in Section
10.2.
2.3 Consequences of creatine kinase equilibrium for
mitochondrial control
Another consequence of the creatine kinase equilibrium [10,
43] is that for modest changes in pH, oxidation rate is likely to
correlate with several possible metabolic feedback signals [10].
For example, as will be discussed later (Section 5.2), we can
take Q as being controlled by ADP concentration, according
to the classical mechanistic ADP-control model [10],
(5)
(Note thatADP andATP are in competition at the translocase
[58], but as the affinity constant for cytosolic ATP is about
0.7 mM [7], this effect is constant at all reasonable [ATP] in
vivo, and is ignored in the operational definition ofKm). From
Eqs 1 and 5, Q can be expressed in terms of the concentrations
of ATP and creatine metabolites [10] as
Q/Q MAX = [Cr]/ {[Cr] + ([PCr]/[ATPDKmKc/} (6)
As KmKcK'/[ATP] "" 1 in human muscle, Eq. 6 can, to a first
approximation, be written as
253
Q/QMAX"" [Cr]/[TCr] (7)
This approximately linear relationship, which is consistent
with what the kinetic version of the phosphocreatine shuttle
hypothesis requires [7], has often been observed [7-9, 22,
23]. The creatine kinase equilibrium has been regarded [59]
as amplifying small changes in ADP into large changes in
phosphocreatine or creatine. However, it is better thought
of as amplifying small changes in phosphocreatine into
larger (fractional) changes in ADP concentration, thus
increasing the gain of the feedback loop controlling oxi-
dative ATP synthesis: at constant pH, (d[ADP]/[ADP])/
(d[PCr ]/[PCr]) = -[TCr]I[Cr] "" 9 in resting skeletal muscle
[60].
Oxidation rate will also correlate with the free energy of
ATP hydrolysis as required by a nonequilibrium thermo-
dynamic approach [61, 62] or an electrical analogue of this
[8]. This can be seen as follows. The free energy of ATP
hydrolysis is given approximately by
(AG - AGO)IRT = -In(phosphorylation potential) =
(8)
In {[Pi] [ADP]/[ATP]}
where AGO is the standard free energy of ATP hydrolysis, R
is the gas constant and T is absolute temperature (see Table
1). At human body temperature, this reduces to AG = --67.4
+ 2.58 In {[Pi] [ADP]/[ATP]} .As [Cr] "" [Pi] at rest and during
exercise (Section 2.1), there is a relationship [8] between AG
and creatine metabolite concentrations which (from Eqs 1 and
8) can be written
(AG - AGO)IRT "" In{[Cr]2/([PCr]KcK')} (9)
By differentiating Eq. 9, it can be shown [8] that for moderate
phosphocreatine depletion
dAG/d[PCr] = RT ([TCr] + [PCrD/([PCr][CrD = 6RT(0)
[TCr]
Thus from Eqs 7 and 10, to a first approximation [10],
(11)
Such an approximately linear relationship between oxi-
dation rate and free energy of ATP hydrolysis has often
been observed in vivo, both in aerobic exercise [9] and
during recovery from exercise [9, 22, 23], and in isolated
mitochondria in vitro [17,63]. This relationship is the basis
of an electrical analogy in which ATP turnover is the
analogue of current, AG is a voltage, dQdAG is the mito-
chondrial conductance (reciprocal resistance) and the
creatine kinase system functions as a capacitance (given
by Eq. 10) [8, 64].
254
There is also a relationship between AG and [ADP] which
can be written
(AG -AGO)/RT '" -In {([ATP]/[ADP]Y/(Kc/[TCr))
(12)
+ ([ATP]/[ADP))/[TCr])
This quadratic could be solved for ADP concentration as a
function of AG, and used with Eq. 5 to write Q as a function
of AG over a range beyond the linear region. A more general
analysis [65] has the same result, by showing that the
relationship between Q and AG is sigmoidal, and can be
written as
Q/QMAX = {const.exp[(AG - AGO)/RT] - 1}/
(13)
{const.exp[(AG - AGO)/RT] + QMA/QB}
where Q Bis the basal rate of oxidative ATP synthesis and
'const' refers to arbitrary or empirical constants. This
relationship has been observed during aerobic exercise
[20] but is difficult to distinguish from the linear appro-
ximation. Such a sigmoid relationship has also been
observed in isolated mitochondria in vitro when respiration
rate was changed by altering the amount of an ATP-
consuming reaction, but not when it was changed with an
uncoupler (which produces theoretically no change in the
flux through the translocase) [66]. Thus the sigmoidicity
of the relation between respiration rate and extramito-
chondrial AG lies in the translocase, confirming the direct
measurements of [67] that the translocase is not at equi-
librium [66].
2.4 Mitochondrial control and negative feedback
All of these models conform to a general scheme of negative
feedback [10], in which Q is a function of an error signal (),
say
(14)
where the error signal is function ofthe change in [PCr] from
some steady state value:
(15)
and where (ignoring changes in [ATP)), the quantity ([PCr]-
[PCr]t=o) can be considered the time integral ofthemismatch
between ATP production and ATP utilisation:
([PCr]- [PCr]t=o) =-J(U -Q)dt (16)
Thus from Eqs 14-16, the relationship between Q and this
mismatch is
Q =F {F [f(U -
q E
Q)dtJ} (17)
and the kinetics of Q and [PCr] are defined by
d[PCr]/dt = Q - U = Fq() - U = Fq {F([PCr]-
(18)
[PCr]t=o)} - U
Notice that this differs from the typical engineering scheme
of the linear feedback loop, in which output is proportional
to an error signal, which is itself the difference between the
output (appropriately weighted), and an input target value
[68]: in particular, there is no input value, and no steady state
error, as in the long run Q must equal U [10] (in effect, this
is a variety of integral control [68]). In the simple linear
model, for example, FE is the negative identity function (i.e.
=([PCr] - [PCr]t =0))' and F q represents multiplication by
QMA/[TCr] (see Eq. 7). In the electrical analogy [8], FE is
given by =AG =<5([PCr] - [PCr]t=o)dAG/d[PCr] and Fq is
to a first approximation the linear relationship given by
Eq. 11.
For present purposes, any of the approaches described in
this section could be used in the model without much effect
on the conclusions. These models will be illustrated when we
consider spatially averaged steady-state responses in the
model to be developed later (see Section 7.1 and Fig. 6)
3. Simple models of aerobic exercise
assuming the creatine kinase
equilibrium
Given this analysis, we can now address the modelling of
aerobic exercise in the whole cell in the simplified case
where the spatial dimension, the possible coupling of
creatine kinase, and possible departures of cytosolic creatine
kinase are all ignored: these will be introduced into the
analysis later.
3.1 Equilibrium creatine kinase, ignoring changes in ATP
concentration
If the small changes in [ATP] are ignored, the linear steady-
state correlations with phosphocreatine concentration can be
shown to imply that during transient disturbances phospho-
creatine concentration will alter with exponential kinetics [10,
11]. This can be seen as follows. In general, in purely aerobic
exercise,
U = Q-d[PCr]/dt-d[ATP]/dt
= Q - (d[PCr]ldt){ 1 + d[ ATP]/d[PCrJ} (19)
'" Q - d[PCr]/dt =Q + d[Cr]/dt
 255

From the approximately linear relationship Eq. 7, we have Q-Qt=O = LlU{l-exp(-krt)} (28)

dQ/d[Cr] =-dQ/d[PCr] '" ks (20) and from Eqs 19 and 28

where the slope constant ks = QMAX/[TCr]. From Eq. 20, -d[PCr]/dt = LlUexp(-krt) (29)

dQ/dt '" -ksd[PCr]/dt = ksd[Cr]/dt (21) Integrating Eq. 29 from t to 00 and substituting into Eq. 28,
we have
If as part of a physiological response U undergoes a step
change LlU from some basal value Ut=o' then from Eqs 19- Q - Qt=o = -k,([PCr] - [PCr]t=o) (30)
21 we have

dQ/dt + k,(Q - Qt =0) = ksLlU (22) 70

60 ll.G kJ/mol
d[Cr]/dt + ks([Cr] - [Cr]t=O) = ksLlU (23) 50
II)
Q)

Eqs 22 and 23 define exponential kinetics for both oxidation :cco 40 ADP I'M
.;:: 30 PCr mM
rate and creatine or phosphocreatine concentration, with rate
~ 20 CrmM
constant ks [10, 11].
10 ATP mM
The same argument can be made using the approximate
0
proportionality between oxidation rate and LlG (Eq. 11). We 0 50 100 150 200 250 300
can write this relationship as
Time
dQ/dLlG = I1Rm (24) Q)
C)
c:
co B
where Rm has been defined as the mitochondrial resistance or;
0
[8] (Eq. 11 shows that this is approximately equal to 6RT/ "iii 0.1
c:
QMAX). From Eq. 19, 0
ti
l!!
U = Q - (d[PCr]/dLlG)(dLlG/dt) (25) LL 0.01 300
0 50 100 150 200 250
Time
where the term (d[PCr]/dLlG) , which has been defined as the
Q)
'capacitance' (here called Cp) of the creatine kinase equili- C)
c:
brium [8] is approximately equal to [TCr]/(6RT) (Eq. 10). co
Thus differentiating Eq. 24 and substituting into Eq. 25, then
or;
0 C
substituting for Q using Eq. 24 gives "iii 0.1
c:
0
ti
dQ/dt = [l/(RmCp)](Q- U) (26) !!! 0.01
LL
0 50 100 150 200 250 300
dLlG/dt = (LlG-LlGt=o)/(RmCp)-(LlU/C) (27) Time

Together Eqs 26 and 27 define exponential kinetics for LlG
(voltage) and Q (current), with rate constant l/(RmCp). It
follows from Eqs 10 and 11 that this rate constant is also
approximately equal to ks = QMAx/[TCr] [10]. These approxi-
mations (with ks = kr = QMAX/[TCr] - see Eq. 7) have been
used in the thick lines in Figs 2B and 2C.
The analysis can also be reversed: observed exponential
kinetics can be shown to imply a linear steady-state relation-
ship with phosphocreatine concentration [7, 10]. This can be
seen by supposing that oxidation rate changes with rate
constant ~ following a step change LlU, then
Fig. 2. Some implications of simple model of mitochondrial control. The
figure shows some predictions of the model shown in Fig. 1, assuming that
cytosolic creatine kinase is always at equilibrium. Panel (A) shows the
time course of changes in metabolite concentrations and free energy of
ATP hydrolysis following a work jump (U1Q MAx changes from zero to 0.4).
Panels (B) and (C) show semilogarithmic plots of the kinetics of these
following work jumps of various sizes (U/QMAX changes from zero to 0.1,
0.2,0.3,0.4,0.5,0.6, 0.7). The thick lines represent k = QMA/[TCr], derived
from the equilibrium approximation (Section 3.1 and Eq. 7). In (B), lines
to the left of these show phosphocreatine concentration and the lines to the
right show ADP concentration (largely superimposed); in (C), lines to the
left show A TP concentration and lines to the right show free energy of
ATP hydrolysis.
256
which is a form ofEq. 20 [7,10]. Similarly if [PCr] changes
exponentially, so
[PCr]t=o -[pCr] =([pCr]t=o -[PCrtJ{ l-exp(--!st)} (31)
then Eq. 30 follows by differentiating Eq. 31 and using Eq. 19.
Lastly, if L1G changes exponentially, then as [PCr] is
approximately proportional to L1G (see Eq. 11), Eq. 30 follows
again from Eq. 31. Thus if oxidation rate, creatine or phos-
phocreatine concentration, or L1G change exponentially, then
at steady state oxidation rate is proportional to creatine (or
phosphocreatine concentration) with proportionality constant
equal to the exponential rate constant [10].
Recovery from exercise is entirely aerobic, and can be
considered as a special case of work jump. An extension of
this model has been used to simulate the recovery of pH and
phosphocreatine and ADP concentration after exercise [69,
70]. In this approach Eq. 5 is used to calculate Q from the
ADP concentration; as external work is not done during
recovery, and as [ATP] changes can be ignored, oxidative
ATP synthesis is used to fund phosphocreatine resynthesis,
plus basal requirements: thus Q + QB = d[PCr]/dt. This
simulation could be run at constant pH, but also reproduces
the observed recovery kinetics following acidifying exercise
[69,70] ifit assumed that the proton load generated as a con-
sequence of phosphocreatine resynthesis is handled by a pH-
dependent proton efflux from the cell [29]. In such a model
where ADP is the error signal in a feedback loop [10], the
kinetics of [ADP] are largely stable against changes in
perturbations such as pH, and yet are highly sensitive to
alterations in mitochondrial gain (e.g. in mitochondrial
myopathies) [69, 70].
3.2 Equilibrium creatine kinase, allowing for change in
ATP concentration
To take this further, the analysis above (Section 2.1 and 2.2)
can be used for a simulation of aerobic exercise. At all times
v + d[T]/dt = Q (32)
We can take Q as being controlled by ADP concentration,
according to Eq. 5. Thus ifin the basal state V = VB' then QB
=VB and ADP and phosphocreatine concentration are defined
by
(33)
[ADP]B = Km/{(QMAx/V)-l}
(34)
[PCr]B = [TCr]/ {1 + ([ADPJi[ ATP]B)KcKL1'}
Following a step change from VB to V x'
(35)
Given the starting values defined by Eqs 33 and 34, and the
relationships in Eqs 3 and 4, we have
d[ ATP]/dt = (d[T]/dt)(d[ATP]/d[TD (36)
d[PCr]/dt = (d[T]/ dt)( d[PCr]/d[TD (37)
where d[T]/dt is from Eq. 35 and the second terms are from
Eqs 3 and 4. This gives the results shown in the thin lines in
Fig. 2: for successive increments of time Q is calculated from
ADP concentration using Eq. 5, then d[T]/dt is obtained from
Q using Eq. 35, and from this d[ATP]/dt and d[PCr] are
obtained using Eqs 36 and 37. These are used to calculate new
values of ATP and phosphocreatine concentration; creatine
concentration is obtained as [TCr] - [PCr], and ADP con-
centration either as [TA] - [ATP] or from [PCr], [Cr] and
[ATP] using Eq. l. (This is essentially equivalent to two
earlier approaches, one using a linear model of mitochondrial
contro1[11] and one using theAD P -contro1assumption [64 D.
In this figure, Fig. 2A shows the time course ofthe various
changes following a single work jump. Figures 2B and 2C
show semilogarithmic plots ofthe responses following work
jumps of varying size: these are approximately linear (imply-
ing near-exponential kinetics), and the rate constant (given
by the slope) decreases somewhat with increasing work
jumps. Nevertheless, for phosphocreatine (Fig. 2B) the rate
constants are fairly close to QMAX/[TCr] (Figs 2B and 2C),
as predicted by Eq. 7; so also are the rate constants for L1G
(Fig. 2C). The rate constants for ADP concentration (Fig. 2B)
are somewhat smaller than QMA/[TCr]. The rate constants for
ATP concentration (which the simple model says nothing
about) (Fig. 2C) are rather larger than QMA/[TCr], but the size
ofthe changes inATP concentration is negligible (Section 2.2)
Notice that in the absence of creatine kinase, d[ ATP]/dt =
-d[ADP]/dt = Q- V. Simplifying by assuming that [ADP] <
Km , the rate constant of changes in both variables would be
QMA/Km' which would imply a halftime of around 25 msec
rather than the 30-100 sec seen in vivo, in the presence of
creatine kinase. Thus without the buffering effect of creatine
kinase, the high gain ofthe system would require extremely
rapid mitochondrial response in order to avoid instability.
4. Aerobic exercise with nonequilibrium
creatine kinase
At least two attempts have been to avoid assuming that
creatine kinase is necessarily at equilibrium all the time. One,
a simulation of bioenergetic and ionic events in the myo-
cardial cell, combined a linear thermodynamic model of

mitochondrial control with a first-order approximation to the
kinetics of creatine kinase [62]. Another model, also using
linear approximations to mitochondrial control and the
kinetics of cytosolic creatine kinase, contained in addition a
linear approximation to the kinetics of mitochondrial creatine
kinase [71]. We modify and extend this approach here.
4.1 The rate expression for creatine kinase
One problem with more detailed treatments is that the kinetics
of creatine kinase are highly nonlinear: however, as we shall
see, the necessary kinetic properties have been published [53,
72,73]. We can now add to the earlier analysis the possibility
that creatine kinase may not always be at equilibrium. Let C
be the net flux through creatine kinase in the direction of
phosphocreatine breakdown and ATP synthesis (called the
reverse direction in [72, 73], but the forward direction in [53,
74] and in apparently all magnetisation transfer studies); then
Eqs 35-37 are replaced by the more exact forms
d[ATP]/dt = Q - U + C (38)
d[PCr]ldt =-C (39)
An equation for net flux is given in [72, 73] and in modified
form in [75]. For simplicity these treatments ignore inhibition
by formation of dead-end complexes. A recent magnetisation-
transfer study of skeletal muscle makes it clear that these must
be accounted for in a fully quantitative treatment [53]. A
suitable equation for unidirectional flux (in the direction of
phosphocreatine breakdown) which takes account of these
complexes, is given in [53, 74]; based on this and on the
expressions for net flux in [72, 73] we can write an equation
for the net chemical flux (C) through creatine kinase (in the
direction of phosphocreatine breakdown) as
C/C MAX {([ADP] [PCr]) - ([ATP] [Cr]/Kc/)}/M (40)
in which the full expansion of M is given by
M/(Kja~) = 1 + [ADP]/Kja + [PCr]/Kjb +
[ADP][PCr]l(Kja~) + [ATP]/Kjq + [Cr][ATP]/(KpKjq)
+ [Cr][ADP]I(K1pKja) + [PCr] [ATP]/(K1bK jq) +
[anion] [Cr] [ADP]I(Kr;KlpK;a)}'
and C MAX is the maximum activity measured in the direction
of phosphocreatine breakdown, or ATP synthesis. This
expression uses one current notation for the kinetic constants
[53]; an alternate notation [72, 73] is given in Table 2. Note
that the final, anion-inhibition term is necessary to explain
magnetisation transfer measurements of exchange flux
through creatine kinase in vivo [53]: in particular the lack of
257
a clear increase in exchange flux with increasing ATP
turnover, and the fact that fluxes in vivo are 10 fold lower than
creatine kinase activity in a model solution [53]. (Again, we
avoid any detailed consideration of the ionic forms in this
expression - see Section 10.2.) In evaluating M it is con-
venient to define some secondary constants as Koo = lIK;a'
KOI = l/K;b' K02 = lI(K;a~)' K03 = l/K;q' K04 = l/(KpK;q)' K05
= {1 + [anion]/KrJ/(KrpK) and K06 = l/(KlbK;q)' so that the
expression for M becomes
M/K;a~) = 1 + (K03 + K 04 [Cr] +K06 [PCr])[ATP] +
K01[PCr] + (Koo + Kos[Cr] + K02 [PCr])[ADP].
In practice, the anion-containing complex is the only dead-
end complex that makes any real difference [53], and so one
can neglect the terms [Cr][ADP]/(KrpK;a) and [PCr][ATP]/
(KlbK;q) in Eq. 40, and set K06 to zero.
If desired, the unidirectional (exchange) flux (C E) in the
directions of phosphocreatine synthesis or breakdown can be
obtained [53] by modifying Eq. 40 as follows:
CE =CMAX[ATP] [Cr]/(KCK'M)
phosphocreatine synthesis i e. ATP hydrolysis (41)
CE =CMAX[ADP][PCr]1M
phosphocreatine breakdown i.e .. ATP synthesis (42)
where M is as given above and again C MAX is measured in the
direction of phosphocreatine breakdown (see note a to Table
1). At steady state these two exchange fluxes are of course
equal.
4.2 Modelling aerobic exercise with nonequilibrium
creatine kinase
A simple simulation of the response to a work jump could
be performed as follows: for successive increments of time
Q is calculated from ADP concentration using Eq. 5, using
the expression for creatine kinase flux (Eq. 40), d[PCr]ldt is
obtained from Eq. 39 (in which Q does not feature) and
d[ATP]ldt is obtained from Q using Eq. 37. These are used
to calculate new values of ATP and phosphocreatine con-
centration; creatine concentration is obtained as [TCr] - [PCr]
and [ADP] as [TA] - [ATP]. The results at high creatine
kinase are, as expected, very similar to the equilibrium model.
Formally, at t =0, the instant ofa work jump, d[ATP]ldt =Q
- U and d[PCr]/dt = 0, but this phase is very short, and the
fall inATP concentration very small. We do not present such
data here, the results are similar to the spatially averaged
results from the full model, which will be presented below
(Sections 7 and 8). Before presenting these results, we must
discuss our treatment of functional coupling.
258

Table 2. Kinetic constants for creatine kinase

Affinity constants for cytosolic creatine kinase,b Symbol and value here [53] Other symbol and value
[73]

Michaelis constants enzyme-ATP and Cr Kp = 3.8 X lO-J M ~ = 1.9 X 10-2 M

enzyme-ADP and PCr K. = 1.11 X 10-J M K9 = 3.2 X 10-2 M
(enzyme-PCr and ADP) K, =4 x 1O-5 M Kg = 5.9 X 10-5 M
(enzyme-Cr and ATP) K q = 5.7 X 10-5 M K4 = 2.6 x 10-< M

Dissociation constants enzyme and ATP K = 3.5 X lO-J M KI = 6.3 x 10-< M

'q
enzyme and PCr Kib = 3.9 X lO-J M K6= 7.2 X 10-2 M
enzyme and ADP K i, = 1.35 X 10-4 M K, = 1.80 X 10-4 M
(enzyme and Cr) K =0.55M K 2=4.9X 10-2M
'p

Inhibitory constants enzyme-ADP and Cr Kip = 5.8 X 10-2 M

(enzyme-PCr and ATP) K lq = 3.2 X lO-J M
enzyme-ATP and PCr Klb = 3.9 X lO-J M
(enzyme-Cr and ADP) K I, = 1.5 X 10-5 M
enzyme-Cr-ADP and anion K h z2x10-4M

Secondary constants K= IIKi 7.4 X lO J M-I

KOl = IIK,b 2.6 X 102M-I
K02 = II(Ki,Kb) 6.7 X 106 M-2
KOJ = IIK iq 2.9 X 102 M-I
K04 = II(KpK,q) 7.6 X 104 M-2
K05 = {I + [anion]/K,JlK,pKi'l 16 X IO'M-2
K06 = II(K lbK iq) 7.3 X 104 M-2

Kinetic constants for mitochondrial creatine kinase Symbol and value here Other symbol and value [79]
enzyme-ATP and Cr Kmit A = 0.7 mM K,=0.7mM
enzyme-Cr and ATP Kmitc = 5.0 mM K,=5.0mM

'In the text we use symbols of [53,74], and (rabbit muscle) values of [53] (modified from [74,137]); in the right-hand column of the Table we also give
alternate nomenclature and (human muscle) values from [72]. bBracketed values are not used explicitly in the text because of the following equalities [73],
expressed in the notation of [53, 74]: KiqKp = KipKq' KibK, = Ki'~' ~.Kip = K,pKi, and KlbKiq = K,qKib (expressed in the notation of [72, 73], these equalities
are KIKJ = K 2K 4, K6Kg = K,K., K.K2 = KpK, and KyKI = KgK6)' Note that the ratio of maximum flux in the direction of phosphocreatine synthesis to that in
direction of phosphocreatine breakdown (as measured in [53]) can be expressed in terms of kinetic and equilibrium constants [73] either as KiqK/([W]KI4Ki'~)'
using the kinetic notation of [53], or by K,K/(K,K9 [W]K I4), using the kinetic notation of [72] (K'4 in both expressions is an equilibrium constant [115]
defined in Table 3). Values in [53], used here, are based on [74,137], except that in [74], IIK ip = Kq = K i, = O. The logic of the nomenclature in [74], used here,
is that subscripts a, b, p and q refer to MgADP, PCr, Cr and MgATP respectively; the initial subscripts i and I refer to dissociation from a binary complex and
an abortive (dead-end) complex respectively, and lack ofi and I means a Michaelis constant.

5. The 'creatine kinase circuit'
5.1 Introduction: coupling and diffusion
Another problem is that, although the bulk cytosolic creatine
kinase reaction is near equilibrium (see Section 2.1), there is
evidence that the creatine kinase system can operate as a
circuit carrying a net flux [49]. In addition to high con-
centrations of cytosolic creatine kinase, there is a distinct
mitochondrial creatine kinase on the inner mitochondrial
membrane that exerts acceptor control over oxidative phos-
phorylation in isolated mitochondria in heart [76] and smooth
muscle [77]. This mitochondrially generatedATP has good,
perhaps preferential access to mitochondrial creatine kinase
[78-82]. Similarly a significant fraction of cytosolic creatine
kinase is bound to the M-line ofmyofibrils in smooth muscle
[56, 83] and heart [48] and phosphocreatine is a good
substrate for myofibrillar ATPase activity [48]. Thus was
proposed the concept ofthe phosphocreatine shuttle [47,49],
in which 'high-energy bonds' travel from the mitochondrion
to myosin ATPase (and other ATPases) largely in the form
of phosphocreatine. This implies functional coupling at the
mitochondrial end between the adenine nucleotide trans-
locase of the inner mitochondrial membrane and mito-
chondrial creatine kinase, and at the opposite end, between
a component of cytosolic creatine kinase and the contractile
proteins. There would be an obligatory link between net flux
through creatine kinase and the rate of ATP turnover [53].
In the classical form, however, the model is incompatible
with several observations in skeletal muscle: in skeletal
muscle, replacement of creatine with the nonmetabolisable
analogue ~-guanidinoproprionate does not inhibit oxidative
ATP synthesis in response to work [84]; flux through creatine
kinase measured by saturation transfer is essentially inde-

pendent ofATP turnover rate in skeletal muscle [3,53,84,85]
and heart muscle [54, 86]; and measured diffusion coefficients
of phosphocreatine, creatine, ATP and ADP show that their
mean square diffusion lengths are larger than the radius of a
myofibril, and so net-flux 'shuttling' is at least not obligatory
in skeletal, cardiac and smooth muscle [87]. Nevertheless, in
the presence of creatine kinase near -equilibrium, most of the
flux of high-energy phosphates will be carried by phospho-
creatine (and backwards by creatine) than by ATP (and
backwards by ADP), simply because of the greater concent-
ration of phosphocreatine and creatine compared to ADP [52,
64]. However, there is no available account of how the balance
between these fluxes might be altered at lower cytosolic
creatine kinase activity, and by possible changes in the
functional coupling of e.g. mitochondrial creatine kinase and
the adenine nucleotide translocase. Next (Section 5.2) we shall
deal with the question of functional coupling. Spatial aspects
will be introduced in Sections 6 and 7.
5.2 A general analysis of coupling in the mitochondrion
Although much analytical [78, 82] and theoretical [76, 79]
work has been done on the kinetics of the mitochondrial
enzyme, it remains to be put quantitatively in the more general
context, and the functional role of these components in the
whole cell has been the subject of controversy. For example,
it has been proposed that the mitochondrial translocase-
creatine kinase system acts as an amplifier, receiving either
ADP and creatine signals that would travel, in effect, as waves
[81]. However, it is not clear that this could occur in the
system as currently envisaged. To answer such questions we
will propose that whole-cell metabolism can be modelled by
taking the mitochondrion as an ADP-regulated device whose
relative output of phosphocreatine andATP is set as a system
parameter. We must now examine these assumptions in more
detail, making use of the scheme set out in Fig. 3, which is
based on, for example [47, 59, 78---80, 88].
First, we examine the implications for our model of
diffusion-limitation ofADP at the outer membrane, and
coupling between mitochondrial creatine kinase and the
translocase in both directions. In the simple equilibrium
model described in Section 3.2 we approximate all these
factors by a simple hyperbolic relationship (Eq. 5). It would
be desirable to calculate the effect of various degrees of
kinetic limitation and mitochondrial coupling on the Km There
has been no detailed theoretical analysis of this, but one can
be supplied in the present framework.
We must first consider the role of the outer membrane
porins. In permeabilised cardiomyocytes, Km for ADP is
increased considerably relative to isolated mitochondria,
although the Km for creatine is unaffected [14, 81]. The
explanation is that ADP movement into the mitochondrion
259
is restricted by outer mitochondrial membrane voltage-
dependent anion selective pores (poring) [31, 81, 89]; not that
while similar gradients exist for the other metabolites, their
absolute concentrations are so much larger that their con-
centration gradients can be ignored). This effect is restored
to isolated intact mitochondria by incubation in media of high
oncotic pressure, which mimics cytosol [89]. It is likely that
there is a close physical association between the porin
proteins and both the adenine nucleotide translocase and the
mitochondrial creatine kinase [88]. Suppose the porin is
unidirectional with a first order permeability coefficient k p .
For a given flux L of ADP through the porin, the ADP
concentration gradient between the cytosol and the inter-
membrane space (suffix im) is given [89] by
[ADP]o - [ADP]1m = Ukp (43)
In vitro under appropriate conditions this concentration
difference can be as large as 13 flM at maximal rates of ATP
synthesis [89, 90].
Now consider the mitochondrial creatine kinase. In vitro,
a high creatine concentration can overcome the reduction in
Km due to the outer mitochondrial membrane [31]. There is
a close physical association between octameric creatine
kinase and the tetrameric adenine nucleotide translocase, and
plentiful evidence of functional interaction [47,49,81,88].
In some analyses [76, 79], near-complete coupling is posited,
so that there is a high probability of activation of the trans-
locase by ADP derived from mitochondrial creatine kinase
without leak into the bulk cytosol, and a high probability
that the translocase supplies the mitochondrial creatine
kinase with ATP from the mitochondrial matrix [76]. It is
implicit in these treatments [76, 79] that mitochondrial
creatine kinase is far from equilibrium, so that the flux is
essentially unidirectional in the direction of phospho-
creatine synthesis.
We can model this in more general terms (Fig. 3). ADP
enters the intermembrane space from the cytosol through
the outer-membrane porin (at rate L) and from the mito-
chondrial creatine kinase (at rate W) and it leaves for the
mitochondrial matrix at rate Q; thus on any but the smallest
time scale,
Q=W+L (44)
Now suppose that a fraction p of ATP from the translocase
is fed straight to the mitochondrial creatine kinase (the
remainder passing into the general intermembrane space), so
that a component of the mitochondrial creatine kinase flux
is given by W Q = pQ, and the remaining component W R is
formally independent of Q
(45)
260

Q = QR + Qc = L+ W L = (l-p)Q - WR

...............
......... ...
...............
.....
...............

~m~~~~~~~

\ Q
\
,, L

...
Q - ---.----~---~--
Q

--+0
MITOCHONDRIAL
MATRIX INTERMEMBRANE SPACE CYTOSOL

Inner membrane Outer membrane

Fig. 3. Model of the 'functional anatomy' of the mitochondrion. The figure shows a detailed model ofthe mechanisms operating within the mitochondrial
box in Fig. 5. See Table I for definitions of symbols. Porin proteins convey ADP and creatine in through the outer mitochondrial membrane, and ATP and
phosphocreatine out; these only represent an appreciable permeability barrier to ADP (because of its low absolute concentration). The adenine nucleotide
translocase conveys ADP through the inner membrane into the mitochondrial matrix, and ATP out. The mitochondrial creatine kinase, which runs in the
direction of ADP and phosphocreatine synthesis, is physically and functionally associated with the translocase: to preserve generality we assume, first, that
mitochondrial creatine kinase receives a fraction p of ATP straight from the translocase, the remainder passing into the bulk intermembrane space; second,
that the translocase receives a fraction e of ADP straight from mitochondrial creatine kinase, the remainder passing into the bulk intermembrane space.

In this equation WR is under the control of intennembrane Now consider oxidative ATP synthesis rate Q. Suppose
metabolite concentrations. To a first approximation (but see first that this is under the control of the intennembrane space
Section 10.8), mitochondrial creatine kinase can be regarded ADP concentration ([ADPlm), according to a modification
as far from equilibrium [79], and so W R is given by ofEq.5

(47)

where KmitA and Kmitc are the dissociation constants of the

Combining Eqs 43-45, then substituting for [ADP]im using
mitochondrial enzyme for cytosolic ATP and creatine, and
Eq. 47 we have a quadratic of the fonn AQ2 - BQ + C = 0,
their concentrations can be taken as essentially the same as
where
cytosolic concentrations. More complex expressions can be
written which also includes the concentrations ofthe products
phosphocreatine [76] and also ADP [76, 91] (described
briefly in Section 10.8), but these refinements make little
difference to the present analysis.

The solution of this is
Q = {B- [B2-4AC]1/2}/(2A) (48)
from which it follows that QIQMAX is no longer a strict
hyperbolic function of [ADP]o' although resembling one
sufficiently closely for us to define an apparent Km as
Kma pp , the value of [ADP]o at which QIQMAX = 1/2. In
general this is given (from Eq. 48 for Q = Q MAX /2) by the
solution of
(49)
and for the present case
K mapp = K m+ {(1/2)Q MAX (1-p)- W Rp
}/k (50)
This analysis shows directly how an outer-membrane diffusion
barrier can raise the apparent Km to ADP and how an increase
in the mitochondrial creatine kinase flux can reduce the
apparent Km despite this [14, 31]. Recent work showing that
the reduction in apparent Km in skinned cardiomyocytes and
cardiac fibres depends largely on the attachment of mito-
chondrial creatine kinase to the inner membrane (being lost
when it is detached), rather than on its mere presence in the
intermembrane space [14]: this arguably favours a large
component of direct channelling, although this is not the only
interpretation [92].
We might also wish to account separately for coupling in
the other direction, i.e. creatine kinase to translocase (Fig. 3).
Suppose then that a fraction 8 of the ADP leaving creatine
kinase is fed straight through the translocase (the remainder
passing into the general intermembrane space), so that Q
consists of this directly coupled component (Qc = 8W) as well
as a component (QR) controlled by intermembrane ADP
concentration:
In this expression the factor A is a technical device to make
the overall maximal value ofQ (with this creatine kinase-to-
translocase shunting) equal to the model value ofQMAX' It can
be shown that this requires A = 1 - p8 - 8(WR,MAX(eft)/QMA)
where W R,MAX(eft) is the maximum value ofWRat the existing
intermembrane [ATP] and maximum [Cr] = [TCr] (i.e.
complete phosphocreatine depletion): from Eq. 46,
(52)
The values of the terms in Eq. 48 for this general case are
(from Eqs 43-45, and substituting for [ADPlm using Eq.
51 ):
261
and so from Eq. 49,
K mapp = DKm+ E(Q MAXp
/k ) (53)
in which
liD = {(l - 8p )/2 - 8(W/QMA)} IA - 1
E = -{W /QMAX) (1- p)/2 + (1- p)(1- 8p)1
{2[A + 8(WR/QMA)] -I}
This shows how Kmapp depends on the ratio to QMAX of W R'
which itself depends on creatine concentration (Eq. 46); as
creatine concentration is related to ADP concentration by the
cytosolic creatine kinase equilibrium (Eq. 1), in principle
Km app could be calculated as an explicit function of ADP
concentration, but this would be difficult. We return to this
later in the context of the complete model (Section 6.5).
5.3 Results: implications of this analysis for the
mitochondrion
First we consider the implications for the control of oxidative
ATP synthesis. Figures 4A and 4B show the relationships
between the rate of oxidative ATP synthesis and the con-
centrations of ADP and creatine. Comparing line A (no porin
barrier or mitochondrial creatine kinase), with line B in Fig.
4A (porin barrier), it can be seen that the outer membrane
permeability barrier raises the apparent Km of oxidativeATP
synthesis for ADP. This produces a decrease in the 'sensi-
tivity' of the controlled process to its controller ADP con-
centration, i.e. a decrease in the flux control coefficient of
ADP; this can also be thought of as a decrease in mito-
chondrial 'gain'. As can be seen from Eq. 53, the occurrence
of coupling from oxidative ATP synthesis to mitochondrial
ATP synthesis (p > 0, line C in Fig. 4A) and of coupling from
mitochondrial ATP synthesis to oxidative ATP synthesis
(8 > 0, line D in Fig. 4A) both ameliorate this loss of gain,
indeed may increase gain (reduce apparent Km) beyond its
value in the absence of a permeability barrier. These altera-
tions only slightly alter the near-linear relationship between
the rate of oxidative ATP synthesis and the creatine con-
centration and the flux through mitochondrial creatine kinase
and the rate of oxidative ATP synthesis (Fig. 4B).
Next we consider the implications for the control of
mitochondrial creatine kinase (see Fig. 4C). None of these
alterations in parameters much affect the relationship between
the flux through mitochondrial creatine kinase and the rate
262

1.0
0.8
~
i
-
:::; 0.6
0 case 0
0
0.4
0.2
A
0.0 I
0 50 100 150 200 250 300

[ADP] !-1M

B
1.0
0.8

-
~
:::; case A
0.6
0 case 0
0.4
0
0.2
0.0
0 10 20 30 40

[Creatine] mM

II +=-_-..-C_--.----_-.-----.,
0.0 0.2 0.5 0.8 1.0

Fig. 4. Detailed analysis of mitochondrial model. The figure shows the relationships between the rate of oxidative A TP synthesis (expressed as QIQMAX) and
(A) ADP concentration and (B) creatine concentration, and also (C) the relationship between the rate of mitochondrial creatine kinase activity (expressed as
W/WR.MAX> and the rate of oxidative ATP synthesis (expressed as QIQMAX) Results are shown for four cases: case A, results calculated assuming no outer
membrane permeability barrier or mitochondrial creatine kinase (Km = 30 flM); case B, results calculated assuming an outer membrane permeability barrier
but no mitochondrial creatine kinase (QMA/kp = 50); case C, results calculated assuming an outer membrane permeability barrier and mitochondrial creatine
kinase coupled 'one-way' to oxidative A TP synthesis (QMA/W MAX = 0.14, P = 0.5, e = 0.0); and case D, results calculated assuming an outer membrane
permeability barrier and mitochondrial creatine kinase coupled 'both ways' to oxidative ATP synthesis (QMAX/WR.MAX = 0.14, P = e = 0.5). Dissociation
constants of mitochondrial creatine kinase are given in Table 2 (and in more detail in Table 4).

of oxidativeATP synthesis (Fig. 4C), which remains approxi-
mately a straight line through the origin, as is found experi-
mentally [54, 79]. It follows from the results in Figs 4B and
4C that the relation between the mitochondrial creatine kinase
flux and the cytosolic creatine concentration resembles the
relation between oxidative ATP synthesis rate and creatine
concentration i.e. near-linear, despite that fact that the model
assumes that the component of mitochondrial creatine kinase
flux which is independent of oxidative flux, W R' is half-
maximal at 5 mM creatine concentration (see Table 2). This is
because the flux through mitochondrial creatine kinase is
dominated by the component coupled to oxidative ATP
synthesis. If, hypothetically, W R MAX is made large compared
to QMAX (not shown), then the relationship between the
mitochondrial creatine kinase flux and the creatine con-
centration becomes more obviously hyperbolic; the relation
between the mitochondrial creatine kinase flux and the rate
of oxidative synthesis (as in Fig. 4C) continues to approxi-

mate a straight line through the origin, although the slope is
increased. This is because of the constraints imposed by the
cytosolic creatine kinase equilibrium on the concentrations
of creatine (which controls the non-coupled component of
mitochondrial creatine kinase flux) and ADP (which controls
the rate of oxidativeATP synthesis and thus also the coupled
component of mitochondrial creatine kinase flux).
This analysis is complicated. For the purposes ofthe whole
model (Section 6.5), it will be sufficient to define an overall
mitochondrial coupling parameter as the fraction of high-
energy phosphates leaving the whole mitochondrial unit as
ATP rather than as phosphocreatine:
<P= l-W/Q (54)
This goes from zero (complete coupling) to 1 (no coupling).
In terms of the detailed model of mitochondrial function
presented in Section 5.2, from Eqs 45 and 54 it follows that
<P= 1-p- W/Q (55)
It can be seen from Fig. 4C that W /Q is negligible for all
realistic cases, and so
<p"'l-p (56)
5.4 Relationship to other analyses o/mitochondrial
creatine kinase
A probabilistic model has been developed which explains
many of the kinetic observations made in vitro [76, 91]. In
this model, it can be shown (from Eqs. 23-25 in [76]) that
(in our terms) the quantity 1I P (= Q/W) is equal to the product
of P(T Dloc)ou!' the probability of matrix ATP binding to
translocase, and P(T.A)in' the probability of ATP from
translocase binding to mitochondrial creatine kinase. The
published analysis assumes 'complete channelling', so that
P(TDDlo)oU! is taken as unity [76]. It is inferred from experi-
mental fits that P(T.A)in '" 0.9 [76]; as this is close to 1, we
can make the intuitively satisfying simplification that p = 1 -
<P '" 11 {P(T Dlo)oU!} This approach has been extended by
incorporating a number of refinements relating to the possible
reversal of the mitochondrial creatine kinase i.e. creatine
synthesis (we describe these briefly in Section 10.8) and to
the possible exchange via the translocase of other species than
matrix ATP and intermembrane space ADP [76, 91]. A
numerical solution successfully predicts the effect of in-
creased respiration rate in shifting the mitochondrial creatine
kinase away from equilibrium [91]. It also allows calculation
of the dependence ofQ on ADP concentration in the presence
and absence of a given creatine concentration; this shows that
activation of mitochondrial creatine kinase in vitro by high
263
concentrations of creatine can, in effect, lower the Km for
ADP and also give rise to largeADP-independent activity of
mitochondrial creatine kinase [91] (although the intrinsic Km
of the translocase for ADP does not feature in this analysis,
as the translocase is treated as first-order with rate constants
for the exchange processes [91]).
An earlier kinetic analysis [79] predicts for fairly low rates
of ATP synthesis a linear relationship between Wand Q that
amounts to Eq. 45 withp = 1/(1 + [ATPVKm\f andWR '" O.
This analysis assumes that cytosolic ATP and ATP from
mitochondrial creatine kinase compete, so that in the absence
of cytosolic ATP all ATP molecules passing through the
translocase would be trapped by the mitochondrial creatine
kinase. Thus at very low ATP concentration in vitro, Eq. 45
approximates to W", Q, and this is confirmed experimentally
[79]. This conforms to the present assumption (Eq. 45) with
p'" 1 (and thus, from Eq. 56, <P '" 0). However, on this model
p '" 0 at physiologicalATP concentration, which would make
]pero independent of Q, and determined instead by [Cr]o'
which is apparently not compatible with the probability-based
approach [76, 91].
Mitochondrial creatine kinase flux has been measured by
magnetisation transfer studies of intact perfused heart [54].
Measurements of whole heart creatine kinase represent the
weighted mean of unidirectional mitochondrial creatine
kinase (0--25%) and reversible cytosolic creatine kinase [54].
The mitochondrial Pi pool has distinct NMR properties (due
to altered relaxation times) and the flux through mito-
chondrial creatine kinase can therefore be inferred if an
assumption is made about the size of this pool [54]. This
apparent mitochondrial creatine kinase flux increases with
work, showing an empirical correspondence to Eq. 54 with
<P '" 1, and to Eq. 45 with P '" 1 and WR '" 0 [54]. However, it
is not clear that this is the result of direct coupling; rather, it
may be the result of changes in creatine concentration
according to Eq. 46.
Thus in summary, Eq. 54 with <P '" 0 is implicit in the
probability-based model [76, 91], and is empirically (but
perhaps not mechanistically) appropriate to the perfused heart
data [54] and empirically (but perhaps not conceptually)
congruent with the earlier analyses [76, 79]. In the present
work we have judged that Eq. 54 offers a fairly general
account of coupling, able to accommodate theories of high
coupling (<p '" 0) and low coupling (<p '" 1).
It should be realised that these coupling parameters are not
necessarily fixed for a given tissue. For example, in skinned
cardiac fibres the functional coupling of the mitochondrial
creatine kinase to oxidativeATP synthesis (and the structural
association with the inner membrane) is decreased at high [Pi]
and increased at low pH (both of which changes occur in vivo
during ischaemia) [93]. Furthermore, the amount of mito-
chondrial creatine kinase may be altered in e.g. cardiac
myopathy [94].
264
5.5 Coupling at the ATPase end
In smooth muscle [56, 83], cardiac muscle [48, 95] and
skeletal muscle [48, 96, 97] creatine kinase is bound to the
contractile proteins in the M-line in the myosin-containing,
thick-filament A-band of the cell. In skinned fibres or
isolated myofibrils, ADP produced during contraction is
rephosphorylated within the myofibrillar complex by
myofibrillar creatine kinase [81]. Myofilament-bound
creatine kinase can produce ATP fast enough to sustain
sub maximal tension and significant relaxation from rigor
in smooth muscle [56] and cardiac muscle [98]. Formation
of ATP is kinetically preferred, and such ATP is directed
preferentially to the myosin ATPase [48]. This is supported
by isotope labelling and enzyme-competition experiments
[99, 100].
Such a close association between a component of creatine
kinase and myosin ATPase, and also with other ATPases at
the sarcolemma and sarcoplasmic reticulum (reviewed in [59,
80]), acts in the same way as the coupling at the mitochondrial
end between mitochondrial creatine kinase and the adenine
nucleotidase (Section 5.2), so that there is a high probability
of ATP from the associated creatine kinase (formed from
phosphocreatine) reaching the ATPase, and of ADP from its
hydrolysis by the ATPase reaching the creatine kinase (to
form creatine). In vitro, added phosphocreatine can produce
stronger contractions than ATP, and can intensify the con-
tractions produced by ATP [48]. However, in the present
model we consider only the case where ATPase activity is set
as a (time-variable) system parameter, independent oflocal
ATP and phosphocreatine concentration. Our only concern
is therefore the fraction of flux carried 'into' the ATPase by
phosphocreatine versus ATP, and so as at the mitochondrial
end, we will retain generality by defining a coupling para-
meter \If as the fraction of high-energy phosphates entering
the ATPase unit as ATP rather than as phosphocreatine (see
Section 6.5), without attempting to model this by a detailed
analysis of intramyofilament mechanisms of the kind des-
cribed in this Section.
6. Modelling the whole system: the
spatial dimension
To elucidate the functional importance in vivo of the cellular
components discussed above, we set out to incorporate them
into a quantitative theoretical model: i.e. an appropriate
theory of mitochondrial control (Section 2.3); a generalised
theory of coupling between mitochondrial creatine kinase
and the adenine translocase (Section 5.2), and between the
myosin ATPase and its associated creatine kinase (Section
5.5); the detailed kinetics of cytosolic creatine kinase
(Section 4); and the last aspect to be developed here, spatial
diffusion. The introduction of coupling at each end of the
system necessarily introduces a spatial dimension, as one
must distinguish at least the cytosolic creatine kinase nearest
the ATPase from that in the bulk cytosol. Here we have
taken the view that as a compromise between tractability and
plausibility we must begin by modelling a single spatial
dimension.
6.1 The full model
As shown in Fig. 5 (which is based on, for example [47, 81,
89]) the muscle cytosol is modelled initially as a one-
dimensional bar running between a mitochondrion making
ATP and a myofibril using it. Thus we only consider one
spatial dimension, which will simplify the argument while
still allowing us to make all the main points (a possible
extension to two and three dimensions is outlined in Section
10.3). 'High-energy' phosphates (ATP and phosphocreatine)
and 'low-energy' metabolites (ADP and creatine) can be
exchanged at any point via the reaction catalysed by creatine
kinase (Sections 2 and 4). In general, for metabolite X (=
ADP ATP, phosphocreatine and creatine)
d[X]/dt = JX C = (d[X]ot)diffusion
function([ADP], [ATP], [PCr], [Cr]]) (57)
in which JX is the in-line (diffusive) flux and C is the per-
pendicular (chemical) flux from X to Y, say; the chemical
flux is a function of the concentrations of all four meta-
bolites, which has been given above (Section 4). In Eq. 57,
the symbol represents plus for ATP and creatine and minus
for ADP and phosphocreatine. All fluxes J and Care
expressed per unit volume.
This function is (as we saw in Section 4.1), highly non-
linear, which renders Eq. 57 analytically insoluble. We
therefore model the system numerically. We have chosen to
do this by dividing the model along the spatial dimension
into boxes within which metabolites are assumed to be
uniformly concentrated (Fig. 5). As a compromise between
ease of computation and the need for sufficient spatial
resolution to develop our argument, we have taken N = 10.
Although the model is clearly highly simplified in its spatial
architecture, we have tried to relate its parameters to the
situation in vivo as far as possible. Each box has width ilx,
which we have taken as 10 fJM (we have not here considered
modelling muscle with different mitochondrion/myofibril
distances, but clearly all spatial gradients will be increased
as this distance is made smaller). The volumes of the boxes
play no part in the interpretation of the results, but as a
mathematical device we regard each box volume as unity,
so that concentrations and fluxes can be expressed as mM
 265

Myosin
Mitochondrion ATPase
box / box j box j+/ box N

JCr o JCr I J Cr j_ I J Cr. J Cr j+ I J Cr N _I JCr N

J
Crj Crj Crj+ I CrN

t CI JPCr I J PCr j _ 1 t Cj t Cj + 1 tCN

JPCr O J PCr. J Pcr j + I J PCr N-I JPCr N
J
PCrl PCrj PCrN
PCrj+ I

'"
<p JATP O JATP. JATPN
J
ATP I ATP j + 1
ATP j ATP N

tCI t Cj + I
JADP o JADP I JADP j _1 tC j JADP. t CN JADP N
J
ADP I ADP j ADP j + 1 ADP N

t
.--------- --
Low-energy compound flux
I Creatine ATPase rate
Regulation of I kinase
ATP synthesis
------------~ I fluxes
High-energy compound flux
t
Fig. 5. Model of ATP turnover and diffusion in the cell. The figure shows the model analysed in this paper. See Table I for definitions of symbols. We
assume a linear spatial dimension, with fluxes of 'high energy compounds' (the phosphates phosphocreatine and ATP) and 'low energy compounds' (ADP
and creatine) passing between the mitochondrion on the left and the myosin ATPase of the myofilament on the right. The model is divided into N boxes each
containing creatine kinase: in each box it is the same flux C passing between phosphocreatine and creatine and between ADP and ATP (the net flux through
cytosolic creatine kinase), displayed separately for clarity. The ratio of the adenine nucleotide flux to the total flux (adenine nucleotides plus creatine
compounds) is set by the system parameters <p at the mitochondrial end and \jf at the ATPase end. The total flux of high energy compounds out of the
mitochondrion (Q) is under the control of ADP concentration adjacent to the mitochondrion, and necessarily equals the total flux oflow energy compounds
into the mitochondrion. The total flux of high energy compounds into the ATPase is imposed independently of any system variables or other parameters, and
necessarily equals the total flux oflow energy compounds out of the ATPase. Like Fig. 1, the figure omits inorganic phosphate, which results from ATP
hydrolysis at the ATPase and then travels to the mitochondrion to participate in the synthesis of ATP (see Section 6.5).

and mM sec- 1 where necessary. The j-th box contains
creatine kinase of maximum activity CMAX (measured in the
direction of phosphocreatine breakdown andATP synthesis)
and metabolite concentrations [Cr],J [PCr],J [ADP].J and
[ATP]..J The net flux through the creatine kinase reaction
('perpendicular' to the linear axis) in the direction of
phosphocreatine breakdown and ATP generation is C,J a
function of the j-th metabolite concentrations. We measure
fluxes ofthe 'high energy phosphates' phosphocreatine and
ATP as positive away from the mitochondrion, and the
opposite fluxes of ADP and creatine as positive towards the
mitochondrion. The rate of oxidative ATP synthesis at the
mitochondrion is controlled by the adjacent ADP con-
centration, according to Eq. 5.
6.2 The diffusion term
Consider now the diffusion term. In a continuous system, by
analogy with the one-dimensional heat equation [10 I] we can
write
(58)
in which Dx is the relevant diffusion coefficient and (c)2[X]/
dX 2) is the one-dimensional Laplacian of [X]. The steady-state
solution (for d[X]/dt = 0) is d[X]/dX = JXlD x ' i.e. the steady-
state concentration gradient is linear. We must next find a
difference approximation to Eq. 58, to develop the difference
approximation ofEq. 57. The change in [X] (content per unit
266
volume) in boxj due to diffusion is the excess of molecules
diffusing in over those diffusing out, per unit volume. Thus
(59)
From Fick's first law of diffusion,
(absolute flux)/(area) = -Dx(d[X]/dx) (60)
As the box volume is unity, the area of the planes between
each pair of boxes is lIL1x (i.e. volume/width). Thus Eq. 60
can be rewritten as
(61)
and thus from Eqs 58 and 61,
(d[X]/dt)diffusion = -(D/L1x) {(d[X]/dx)j_1 - (d[X]/dx)) (62)
and, since d[X]/dx can be approximated as ([X]j+ 1- [X]Y L1x,
Eq. 62 can be written as
(d[X]/dt)diffusion =-(D/L1x){([X]j - [X]j)/L1x-
([X]j+1 + [X]YL1x}
= -(DxlAx){(2[X]j[X]j_I-[X]j+l) (63)
which is the difference approximation to Eq. 58. In these
expressions Dx is taken as the same (D A) for ADP and ATP,
and a different value (Dc) for phosphocreatine and creatine
[87].
6.3 The numerical model
Using Eq. 63, a difference approximation to Eq. 57 can now
be written as
([X]t l - [X]jn)/L1t = (11L1x){ (D/L1x)([X]j+1 n_
[X]jn) - (D/L1x)([X]jn - [X]H n)} cjn
= (D X /&2)([XlJ+I n-2[Xln+J [XlJ-I n)c nn (64)
where nand n+1 are successive time steps separated by L1t,
and represents plus for ATP and creatine and minus for
ADP and phosphocreatine. Eq. 64 is the direct approxi-
mation of the original differential equation. This is stable
only when the time step is limited by the convergence
criterion DxL11/L1x2 :'0: 112, essentially because the method
relies only on values at previous time points [101]. A better
approximation uses previous and current values in the
calculation. The Crank-Nicholson method is one such that
is stable for any value of DxL11/L1x 2, although it will not be
accurate if this quantity is too large. The Crank-Nicholson
equivalent of Eq. 64 is
(65)
and rearranging
(u + 2)[x]n+1 - [Xl n+l_ [Xl n+1 (L1x /D )cn+1
x J J-n J+I X J
- (u x - 2)[XlnJ - [X].J-I n - [X].J+I n (L1x 2/D X) cnJ = 0 (66)
where ux = 2L1x2/(Dx L1t). We will write Eq. 66 as F([X]) = 0
for each of phosphocreatine, creatine,ADP andATP. For boxes
1 and N (at the boundary) slightly different equations apply
which keep flow terms explicit. In all there are 4N equations
and 4N unknowns ([PCr], [Cr], [ADP], [ATPDj ~ I ... N' These
must be solved simultaneously at every time step. One scheme
iterates from an initial guess using a multidimensional version
of the Newton-Raphson approximation (one dimension per
unknown), as follows. We write a trial solution vector X as
{[Cr]o'" [Cr]N' [PCr]o'" [PCr]N' [ATP]o' .. [ATP]N' [ADP]o
... [ADP]N} arranged as a 1 x 4N column vector. We must first
solve the set of simultaneous linear equations M.~X =-E
where E is the set of 4N equations like Eq. 66 and M is the
Jacobian ofF -
overx,
-
whose elements are M I,J.. = dP/dx
1 J
where
I :'0: i, j :'0: N, specifying numbers for a particular box. Then
Xi+1 = Xi + OX until the value ofx converges to the required
tolerance.
6.4 Constraints
Lastly, we must consider the constraints at the two ends of
the system, the mitochondrion and the ATPase. By con-
servation, the total flow of high-energy phosphates out ofthe
mitochondrion (JPCro + JATPO) is equal to the rate of oxidative
ATP synthesis (Q).As discussed above (Section 2.3), we take
this rate as controlled by the ADP concentration in the first
box, according to a version ofEq. 5:
(67)
Considering the detailed analysis of mitochondrial control in
Section 5.2 we may regard this as an approximation, with an
approximate Km app that is itself a function of mitochondrial
coupling and outer-membrane permeability. We shall ignore
this complication here, which will have no appreciable effect
on any of the conclusions to be drawn. As explained (Section
5.3), we allow for possible coupling between mitochondrial
creatine kinase and the adenine nucleotide translocase using
an overall coupling parameter <p that sets the fraction of the
total flux Q carried out of the mitochondrion by ATP rather
than phosphocreatine (Eq. 54). In the context of the whole
model, W = JPcro and Q = JADPO' so

(6S)
This goes from zero (complete coupling) to 1 (no coupling).
At the ATPase end, flow between the myofibril and the last
box in the cytosol is set by the ATPase rate U. This is the
perturbed parameter. As explained (Section 5.5), we allow
for possible coupling between creatine kinase and the ATPase
using a coupling parameter that sets the fraction of the total
flux Q carried into the ATPase by rather than phospho-
creatine. In terms of the whole model,
(69)
This goes from zero (complete coupling) to 1 (no coupling).
Taking into account the overall conservation constraints, from
Eqs 67--69, we have
JATP O = JADPO = (cpQMA/{1 + K)[ADP]o}
JPcro = Jcro = (l-CP)QMA/{l + Km/[ADP]o}
]ATP N= J ADP N= 'VU
JPcr N= ]crN = (1- 'V)U (70)
This system of equations can now be solved. One technical
problem is that the terms in the Jacobian must be around unity,
so all variables are normalised to the values at t = O.
6.5 The role of inorganic phosphate
A flux of Pi must occur from ATPase to mitochondrion at a
rate equal to those of phosphocreatine and ATP. Pi concen-
trations are always high compared with those of ADP
(normally [Pi] "" [Cr] [9, 10]), and the diffusion coefficient
is higher than that of the creatine and adenine nucleotide
entities (see Table 1) so to a first approximation we can
assume [Pi] is the same throughout the cytosol. Also, Pi
transport into the mitochondrial can be assumed to be
unhindered. As the myocyte is relatively impermeable to Pi,
andATP changes very little, we can take mean [Pi] at any time
as [Pi]B + [PCr] - [PCr]B. In 'thermodynamic' models of
mitochondrial control, Pi has a theoretical influence on Q by
its contribution to ~G (Section 2.3) However, this is severely
constrained. In skeletal muscle, basal [Pi] "" basal [Cr], and
so [Pi] "" [Cr] throughout exercise. It follows from Eq. 1 that
[ADP]/[ATP] is approximately proportional to [Pi]/[PCr]
unless pH varies greatly, and that~G can be written approxi-
mately as a function of ADP or phosphocreatine concentra-
tion, independent of [Pi] (Section 2.3). In 'kinetic' models
of mitochondrial control, [Pi] has little independent role, as
its effective Km at the mitochondrion is about 1 mM [102,
103], much lower than concentrations found during exercise
[10]. Thus no separate account needs to be taken of it here.
267
7. Results: The whole system modelled
at steady state
7.1 Spatial average values
Consider first some finding for mitochondrial control,
metabolite concentrations and coupling. Figure 6 shows the
mitochondrial transfer functions applying to spatial-average
data in the whole model (Fig. 6 should be contrasted with
Fig. 4, which shows the effects of intramitochondrial
mechanisms on the transfer functions for mitochondria
themselves). It can be seen from Fig. 6 that the standard
Michaelis-Menten relationship between oxidative ATP
synthesis rate and spatially averaged ADP concentration
(Fig. 6A) is equivalent, at this high creatine kinase activity,
to a near linear relationship to creatine concentration (Fig.
6B), and to a near-linear (but more properly sigmoid)
relation to free energy of ATP hydrolysis (Fig. 6C). These
equivalencies result from the constraints imposed by
creatine kinase on the metabolite concentrations, and
approximate to the results of models of mitochondrial
control that assume equilibrium conditions for creatine
kinase [10]. Figure 6 also shows that lowering the creatine
kinase activity alters these relationships: both the mean ADP
concentration (Fig. 6A) and the mean creatine concentration
(Fig. 6B) required for a given rate of ATP synthesis are
higher. This can be also be seen in Figs 7A and 7B, which
show that these effects only become appreciable at creatine
kinase activities that are a few per cent of normal. The
reason for these effects will be discussed when spatial
gradients are considered (Section 7.2). The effects of
changes in functional coupling are complex, and are con-
sidered later (Section 7.3)
Now consider some implications for the exchange flux.
Although we shall demonstrate the presence oflocalised net
fluxes (Section 7.2), these are very small compared to the
exchange flux seen here, as creatine kinase is always near
equilibrium (Section 2. I). At high creatine kinase activity,
exchange flux shows a biphasic and approximately sym-
metrical dependence onATP turnover (Fig. SA), and a highly
asymmetrical dependence on ADP concentration (Fig. SB).
As has been pointed out [53], this amounts to relatively little
change in exchange over the physiological range of ATP
turnover rates. Note that without the anion-complex term in
the creatine kinase rate expression (Eq. 40), this flux increases
monotonically with both functions (dotted line in Figs 7A and
7B) [53]. It is interesting that at low creatine kinase activity
the exchange flux becomes a less symmetrical function of
ATP turnover rate (Fig. SA), and this is related to the ADP
concentration gradients that will discussed below (shown in
Fig.9B).
268

1.0 Decreasing CK

~ 0.7 A
-
::!!:
0 0.5
0
0.2

0.0
0.00 50 100 150 200
Mean [ADP] jJM

Decreasing CK

~ B
Q
o

0.00 10 20 30 40

Mean [Creatine] mM

Decreasing CK
C
-
~
::!l
0
0

0.0
-60 -50 -40 -30
Mean free energy of A TP hydrolysis kJ/mol

Fig. 6. Relationships between oxidation rate and spatially averaged metabolite concentrations at steady-state. The figure shows the relationship between
oxidation rate and spatially averaged values of (A) ADP concentration, (3) creatine concentration and (C) the free energy of ATP hydrolysis. We assume for
simplicity that qJ ='If =0.5. In all panels we show results for three different assumptions about the activity of creatine kinase (100, 10, 3 and 1% of normal
activity), as indicated by the arrow.

7.2 Spatial relationships
First consider the transport of creatine and adenine compounds.
Figures 8C and 8D show the fraction of flow at steady state
carried, in the middle of the model, by ATP, i.e. JATP/(JPCr +JATP)
as a function of ATP turnover and ADP concentration. This
fraction is always small. It can be shown [52] that for cytosolic
creatine kinase at equilibrium, this ratio is given, in general,
by JATP/(JPCr+ JATP) = 1/{1 + (JPCr/JATP)} where
JPCr/JATP = (Dc/DA){[PCr] + ([Cr]/KCK')}/{[ADP] +
([ATP]/KCK')} '" ([TCr]/[ATP])(Dc/DA)KcK'/ {1 +
(Kc/[ADP]I[ATP]W
This is of the order ofJATP/(JPCr + JATP) '" 0.01, thus most of
the flux goes as phosphocreatine and creatine rather thanATP
and ADP under almost all circumstances [7, 52] (notice that
this is contrary to prediction of the modelof[71]). Figure 8C
shows how JATP/(JPCr + JATP) increases with equilibriumADP
concentration, and Fig. 8D shows the equivalent increase
with rate of ATP turnover. These effects can also be seen in
Fig. 7C, which shows that they only become appreciable at
creatine kinase activity below about 10% of normal. These
effects on the adenine nucleotide fraction of flux are in
accordance with the earlier analysis [52], which however did
not deal with the case where creatine kinase activity was too
(71) low to permit the assumption of equilibrium.
Fig. 7. Effects of creatine kinase activity at steady state. The figure shows the relationship between creatine kinase activity (expressed as a percentage of
nonnal activity) and some steady-state variables. Panels (A) and (8) show, in different form, some data which are also shown in Figs 6A and 68; they
demonstrate the relationship between creatine kinase activity and spatially averaged values of (Al creatine concentration and (8) ADP concentration. Panel
(C) shows, in different fonn, some data which will also be shown in Figs SC and SD; it demonstrates the relationship between creatine kinase activity and the
steady state fraction of flow carried in the middle ofthe model by ATP. In all panels the solid lines show results at different values of ATP turnover (QIQMAX
= 0.1,0.2,0.3,0.4,0.5,0.6,0.7 and O.S), assuming that <p = \jI = 0.5. The lines with symbols show the results at QIQMAX = 0.3 of different assumptions about
<p and \jI: for circles, we assume <p =\jI = 1; for squares, we assume <p =\jI = 0; for upright triangles, we assume <p = 1; \jI = 0; for inverted triangles, we assume
<p = 0, \jI = 1 (last two are not shown in (A) because they are indistinguishable from the first two).

Next consider the spatial concentration gradients of
creatine andADP (Figs 9Aand 9B). With the given diffusion
coefficients [87] and the cylindrical muscle model of [52], it
can be shown that even in the absence offacilitated diffusion
due to creatine kinase at equilibrium [52], ADP concentration
gradients across the radius of the idealised cell would be small
even at high rates of ATP turnover. In accordance with this,
both creatine and ADP concentration show gradients de-
creasing towards the mitochondrion (in the direction of their
net flux), whileATP and phosphocreatine concentration (not
270

Decreasing CK
0.20
A
~

-
0.15
:iE
()
w 0.10
()
0.05

0.00
0.00 0.20 0.40 0.60 0.80

Q/Q MAX
0.20
Decreasing CK
~

-
0.15
:iE
()
0.10
()
w
0.05 B
000
0.00 50 100 150 200

Mean [ADP] IJM

a.. 100 Decreasing CK

l-
e:(
(/) 10
CO
~
0 1.0
C
Ii=
'#. 0.1
0.00 0.25 0.50 0.75 1.00

Q/Q MAX

a... 100
Decreasing CK
~ 10
If)
III
><
::J 1.0
Ii=

0~
0.1
0.00 50 100 150 200

Mean [ADP] IJM

Fig. 8. Creatine kinase exchange flux and fraction of flow carried by AIP at steady state. Panels (A) and (B) show the spatially averaged steady-state
creatine kinase exchange flux (measured in the direction of phosphocreatine synthesis, expressed as a fraction of maximum) plotted as a function of (A) A IP
turnover rate, and (B) spatial average ADP concentration. Panels (C) and (D) show in logarithmic form the fraction of flow at steady state carried in the
middle of the model by AIP, i.e. ]ATP/(JPc, + JATP) plotted as a function of(C) AIP turnover rate, and (D) spatial average ADP concentration. All panels show
data from three levels of creatine kinase activity, 100, 10, 3 and 1% of normal, as indicated by the arrow. We assume for simplicity that <p = 'If = 0.5.

shown) show gradients decreasing towards the myosin
ATPase. As implicit in other analyses [11,52, 104, 105], these
spatial gradients are very small.
It is evident from these results that the model does not fulfil
an important requirement ofa 'true shuttle' hypothesis, which
is that changes in [ADP] be confined to microenvironments
near the mitochondrion and the ATPase [7].
A surprising result is the nonlinearity of the ADP con-
centration gradient at low creatine kinase activity (Fig. 9B).
In a simple linear model such as this, one would expect all
the concentration gradients to be linear, just as in a two-
dimensional model of the sort analysed in [52], these would
be quadratic. The ADP concentration profile can only be
understood with reference to the spatial gradient of net flux
 271

::2: Fig. 9C. The profile of this net flux could not be predicted

::~ ~
E
&::
analytically, but Fig. 9C shows that, as one might expect, when
111
Q) creatine kinase activity is high this process is completed in
E the box adjacent to the ends; at low creatine kinase activity it
fI) -0.40
fI) is spread over the next few boxes. The sum ofthese net fluxes
~ -0.80 i i
';:' 1 2 3 4 5 6 7 8 9 10 C in each half of the model must nevertheless remain equal to
~ Mitochondrial end ATPase end the difference between JPCr at the end and JPCr in the middle. In
the 'mitochondrial half the sum of C (measured in the direction
of creatine synthesis), which we can write as L mito C, must be
equal to the difference between JPcro' which is (l-<p)Q, and
~
the 'middle' value of JPCr, which is approximately equal to Q;
6
thus L milo
. C ""---<pQ. In the 'myosinATPase half the equivalent
B
::::I..
c: 4
CIS sum LATPaseC is equal to the difference between JPCrN, which
Q) 2
E is (1-\jf)U and at steady state (1-\jf)Q, and the middle value
fI)
fI)
0 of JPCr; thus LATPase C "" \jfU = \jfQ at steady state. When <p = \jf,
~ -2 then at steady state the overall spatial average of the net
ii' -4 Decreasing CK cytosolic creatine kinase flux, <p "* LoverallC must be zero.
However, if <p "* \jf, this is not so; at steady state the overall
Cl
~ -6
1 2 3 4 5 6 7 8 9 10 spatial average of the net creatine kinase flux must be such as
Mitochondrial end ATPase end to balance the difference between phosphocreatine leaving
the mitochondrion and entering myosin ATPase, so LoverallC
= Q(\jf= <p). This is negative (net phosphocreatine synthesis)
.!!! 0.10
if the mitochondrial creatine kinase is less coupled to ATP
:E
E 0.05
C Decreasing CK
synthesis than the myofibril-associated creatine kinase is
>< coupled to the myosin ATPase (<p > \jf), because in this case a
::I 0.00
Ii= greater fraction of phosphocreatine is used by the myofibril!

-
~ creatine kinase complex than is produced by the
() -0.05
Q)
Z -0.10 I
1 2 3 4 5 6
Mitochondrial end
Fig. 9. Spatial gradients at steady state. The figure shows the spatial profile
of the concentrations of (A) creatine and (B) ADP, both expressed as
differences from the overall mean value, and (C) the net flux through
creatine kinase (measured in the direction of phosphocreatine synthesis).
Data in (A) are shown for five values of ATP turnover (Q/QMAX = 0.1, 0.2,
0.3, 0.4 and 0.5) and at 100% normal creatine kinase activity only (results
for 10% and 1% normal are very similar). Data in (B) and (C) are shown for
one value of A TP turnover (Q/QMAX = 0.1 ) and four levels of creatine
kinase activity, 100, 10,3 and 1% of normal, as indicated by the arrow. We
assume for simplicity that <p = \jf = 0.5, except in the dashed line in panel
(B), which shows results for <p = \jf = 0 (complete coupling), 1% creatine
kinase activity.
through cytosolic creatine kinase. As most of the flux in the
middle of the model must be carried as creatine compounds
[52], unless \jf and <p near 1 (i.e. near-complete coupling),
some of the 'adenine nucleotide flux' at the ends ofthe model
must be changed into 'creatine flux' in the middle. The only
means by which this can occur is a net flux through cytosolic
creatine kinase in the direction of phosphocreatine and ADP
synthesis near the mitochondrial end and creatine and ATP
synthesis near the ATPase end, and this is what is shown in
mitochondrion/creatine kinase complex. However, net flux is
always small compared to the exchange flux and overall net flux
7 8 9 10 (i.e. cytosolic, mitochondrial and ATPase-associated creatine
ATPase end
kinase) must be zero at steady state.
As the net flux through creatine kinase is a function ofthe
metabolite concentrations in each box, the metabolite profiles
(especially ADP concentration in Fig. 9B) are defined by (and
in tum define) the net flux profile in Fig. 9C. This is why the
ADP concentration gradient is influenced by the activity of
cytosolic creatine kinase such that a reduction in activity
decreases ADP concentration at the mitochondrial end and
increases it at the myosin ATPase (Fig. 9B): more flux is
required to go through adenine nuc1eotides as the shuttling
effect of creatine kinase [52] is reduced, but less creatine
kinase activity is available to achieve this. Thus the magni-
tude of the net flux through creatine kinase is decreased in
the first and last boxes (Fig. 9C) when creatine kinase activity
is reduced, but as L mlO t C and L ATP C are fixed (for given Q,
ase
<p and \jf), net flux in the second and second-to-Iast boxes are
higher, and so the net flux profile is 'smoother' at low creatine
kinase activity. As ADP concentration at the mitochondrial
end is fixed by the mitochondrial gain, a larger ADP con-
centration gradient implies a larger spatial mean ADP
concentration. This is why reduced creatine kinase activity
raise the mean ADP concentration required for a given ATP
turnover (Section 7.1).
272
An interesting point is that at low creatine kinase activity,
when these 'end-effects' produce a significant gradient of
ADP concentration at either end of the model, the zero-point
is not in the middle ofthe model (Fig. 9B). This is presumably
because of the asymmetry in the creatine kinase rate ex-
pression introduced by the anion-complex term (Eq. 40):
because of the need to harmonise the creatine and adenine
nucleotide fluxes (as discussed above) the creatine kinase
reaction does, contrary to the simplest view, carry local net
fluxes, and so the nature of the rate equation does influence
spatial gradients of concentration.
This question has been approached slightly differently
[87]: with measured values of the diffusion coefficients Dx'
the exchange flux of creatine kinase and the concentration
of a metabolite, the mean lifetime of X (before chemical
transformation by creatine kinase) is given by [X]/E; then for
random movement,
(mean diffusion distance )2 =2Dilifetime) =
2Dx[X]/E (72)
Even for ADP this is 1.8 ~m in frog muscle, larger than the
diameter of the myofibril (about 1 ~m): calculated values are
around 3 ~m for smooth muscle and cardiac muscle. Thus
even ADP can easily cross the cell without being transformed
into something else, despite the very activity of creatine
kinase [87). A similar calculation could be performed for the
present model, but its relation to the present kinetic analysis
is difficult to specify in detail.
7.3 Effects offunctional coupling
Figure 7 also shows the effects of altered functional coupling
on the mean spatial concentrations of ADP and creatine, and
on the fraction of flux carried in the middle of the model by
adenine nucleotides. It can be seen that compared with the
'median' case where <p = 'V, with high degrees of coupling at
the mitochondrial end (i.e. most output as phosphocreatine
rather thanATP, of which the limiting case is <p = 0) and at the
ATPase end (i.e. most input as phosphocreatine rather than
ATP, of which the limiting case is 'V = 0), the creatine kinase
activity has little effect on the concentration gradients or on
the adenine nucleotide fraction, which remains small (circles
in Fig. 7). Mechanistically, when most of the output of the
mitochondrion and the input to the ATPase occur as phospho-
creatine, then most of the flux is the middle of the model
automatically goes mainly as phosphocreatine, and no signi-
ficant net flux through cytosolic creatine kinase is necessary
to convert the end-model fluxes of adenine nucleotides into the
mid-model flux of phosphocreatine (although there is net flux
through the mitochondrial creatine kinase, and through
whatever component of cytosolic creatine kinase is separately
considered as being coupled to the ATPase; this flux is
approximately equal to Q, and in opposite directions at either
end). Conversely, with low degrees of coupling at the mito-
chondrial end (i.e. most output as ATP rather than phospho-
creatine, of which the limiting case is <p = I) and at the ATPase
end (i.e. most input as ATP rather than phosphocreatine, of
which the limiting case is 'V = 1), the creatine kinase activity
has a very great effect on the concentration gradients and on
the adenine nucleotide fraction (squares in Fig. 7). When most
of the output of the mitochondrion and the input to the ATPase
occur as ATP, then most of this flux at the ends has to be
converted (because of the constraints upon diffusion of ADP,
imposed by its low concentration) into a flux mainly of
phosphocreatine, and the process is therefore maximally
dependent on the activity of cytosolic creatine kinase.
Thus the degree of coupling has no significant con-
sequences at normal skeletal-muscle activity of creatine
kinase. This is consistent with recent results in a transgenic
mouse line which show that MM -CK can be replaced by MM-
BB without detectable effect on muscle energetics, even
though this form does not bind tightly to myofibrils [106).
However, in experimental situations where total creatine
kinase activity is very low [46, 93], the effects on spatial-
averaged concentrations could be large. When cytosolic
creatine kinase activity is low, it makes mechanistic sense for
both ends of the system to manifest a high degree coupling.
The adenine nucleotide flux through the mitochondrial
adenine nucleotide translocase has to be converted into a flux
largely of creatine compounds; if this is not done by the
mitochondrial creatine kinase, it must be done by the cyto-
solic creatine kinase at the expense of an increase in the
concentrations of creatine and ADP, and (not shown) and
decrease in those of ATP and phosphocreatine
8. Results: the whole system modelled
during work-jump transitions
8.1 The kinetics of spatial-average concentrations
As expected, spatial-average responses to a step increase in
ATPase rate approximate to those of models of mitochondrial
control that assume equilibrium conditions for creatine kinase
(see Section 3). In particular, overall phosphocreatine
concentration shows exponential kinetics [7-9, 11). ADP
shows near-exponential kinetics.
Figures 10 and 11 show the difference between the con-
centrations of creatine and ADP from their equilibrium values
(on a logarithmic scale) as a function of time after a step
increase in ATPase rate. Figures llA and 11 B show that the
rate constants (i.e. the negative slope of the lines) for creatine
and ADP are relatively independent of creatine kinase
 273

100
iii
.......
....
<:
10
~ 1.000
'i:'
0.100
A
~
0.010
0 1000 2000 3000
Time

....
a..
iii
<: 1000
100
l%CK: <P=1J1=0
<P = IJ1 = 0.5
0
~ 10 <p=IJ1=1
1.00
6:' 0.10
0
~ 0.01
0 1000 2000 3000

Time

100 I%CK
iii
<:
0;:::; 10
~ 1.000 down
'i:' 0.100
~ up
0.010
0 1000 2000 3000

Time

iii
...-<: 1000
a.. 100
0
I%CK
D
~ 10 down
1.00
6:' up
0 0.10
~ 0.01
0 1000 2000 3000
Time
Fig. 10. Time course of responses to step increases in ATPase activity. The figure shows the difference between the concentrations of (A) and (C) creatine,
and (8) and (D) ADP and their equilibrium values (on a logarithmic scale) as a function of time after a step change inATP turnover. Panels (A) and (8) show
the effects of different degrees of coupling: <P = IJ1 =0.5 (partial coupling), <P = IJ1 = I (no coupling), and <P =IJ1 = 0 (complete coupling), at I % creatine kinase
activity, compared with results at 100% activity (where coupling changes have no visible effect). The 1% activity<p = 'l' = I response is almost superimposed
upon the 100% activity line. The step is an increase of Q/Q MAX from 0.1 to 0.5. Panels (C) and (D) compare the responses to a step up ofQ/QMAX from 0.1 to
0.5, with those to a step down of Q/QMAX from 0.5 to 0.1.

activity. It can be seen from Figs 11 C and 11 D that the
apparent rate constant of changes in creatine concentration
is fairly independent of step size, while the effects on the
kinetics of ADP concentration are complex, but potentially
large. Figures lOA and lOB demonstrate that at low creatine
kinase activity, the degree of functional coupling can affect
both the rate constant of concentration changes (the negative
slopes) as well as the equilibrium values (which influence the
starting values in these plots). At normal creatine kinase
activity the effect is negligible (not shown), as we would
expect. Lastly, Figs 10C and 10D show that the apparent rate
constants of the responses to an increase and a decrease in
274

100
OJ
c:
,...Ji;
..... 10
()
....... 1.00
~
0.10
~
0.01
0 1000 2000 3000
Time

......." 1000
OJ

a.. 100
0
~ 10

(L
0
1.00
0.10
B
~
"-' 0.01
0 1000 2000 3000
Time

100 Increasing work jump

OJ

....."
,...Ji; 10
()
....... 1.00
~ 0.10
2.
0.01
0 1000 2000 3000
Time

1000
100
D
10
1.00
~ 0.10 Increasing work jump

~ 0.01 -t-------,.------...----------.
0.00 1000 2000 3000

Time
Fig. 11. Time course of responses to step increases in ATPase activity. As in Fig. 10, the figure shows the difference between the concentrations of (A) and
(C) creatine, and (B) and (D) ADP and their equilibrium values (on a logarithmic scale) as a function of time after a step change in ATP turnover. Panels (A)
and (B) show the effects of different degrees levels of creatine kinase activity, 100, 10 and 1% of normal. The step is an increase ofQIQMAX from 0.1 to 0.5.
Panels (C) and (D) compare the responses to steps up of various sizes: of QIQMAX from 0.1 to 0.2, 0.1 to 0.5 and 0.1 to 0.9.

work are not identical, and that the nature of this effect also
depends on the creatine kinase activity.
8.2 The space-time kinetics of concentrations
Figures 12 and 13 show the spatiotemporal response to a step
increase in ATPase rate. Fig. 12A shows the response of
creatine concentration at normal and low creatine kinase
activity, showing a near-exponential time course (cf. Figs 10
and 11) and the small influence of creatine kinase activity on
the equilibrium values. The response of ADP concentration
at normal creatine kinase activity qualitatively resembles that
of creatine, and is not shown; Fig. 12B shows how at low
creatine kinase activity (1 % normal) a large spatial con-
centration gradient develops in response to a work jump. It
 275

0.8 r B
0. 6

O.2C

Fig. 12. Responses of metabolite concentrations to step increase in ATPase activity. The figure shows the spatiotemporal response of the concentrations of (A)
creatine and (B) ADP, and of (C) the rate of change of creatine concentration to a step increase in ATPase rate. The spatial axis comes out of the page
(mitochondrial end nearest), while time goes from left to right. We assume <p = 'II = 0.5. In (B) and (C) we assume I % normal creatine kinase activity; in (A) we
assume both 100% and I % normal creatine kinase activity (lower and upper surfaces respectively). The increase in ATPase activity is from Q/QMAX =0.1 to 0.5.
276
can be seen in Fig. 12B how the concentration changes start
at the ATPase end and propagate towards the mitochondrion,
so that at any time, changes are larger at the ATPase end. This
can be seen more clearly in the time-derivative plot (Fig.
12C), which show the rate of increase of, for example,
creatine concentration starts high at the ATPase end and
diminishes in space, towards the mitochondrial end, and in
time, towards the new steady state at which all rates of change
are zero. The spatial gradient in d[Cr ]/dt diminishes both
because of the decline with time at the ATPase end, and
because ofthe initial increase with time at the mitochondrial
end; after a short time there is effectively no spatial gradient
of d[Cr]/dt, and its value then declines with time at the same
rate at all points (Fig. 12C).
Figure 13 supplements this analysis by showing the
spatiotemporal response of diffusion fluxes and the net flux
through creatine kinase. The case shown is that of 1% normal
creatine kinase activity. In Figs 13A and 13B, the fluxes
carried by phosphocreatine and by ATP (expressed here as
negative numbers) increase in response to a work jump. In
the case considered where", = <P = 0.5, the adenine nucleotide
flow is larger at the ends than in the middle, and so the
creatine flux is smaller at the ends. As some of the adenine
nucleotide flow at the ends has to be converted into creatine
flow in the middle and back again, there is in this case an
appreciable net flux through creatine kinase in opposite
directions at each end (Fig. 13C): at low creatine kinase
activity, this is increasingly smoothed out into a net flux
gradient across the whole model (cf. Fig. 9C).
9. Discussion: conceptual relationships
to other approaches
9.1 Relationships to other approaches to mitochondrial
control
We have discussed above some aspects of the relationship
between oxidative ATP synthesis and the flux through
mitochondrial creatine kinase (Section 5.4). In addition to
this, there are a number of current approaches to the regula-
tion of mitochondrialATP synthesis: in effect, there are many
candidates for the appropriate transfer function relating
oxidativeATP synthesis rate to its regulatory signal. We have
used the conventional ADP-control model [10, 17] and
discussed the equivalence of some other transfer functions
(Section 2.3). We discuss some other models here briefly,
mainly to elucidate the differences in the language and
concepts used.
First, we briefly consider metabolic control theory ap-
proaches. Metabolic control analysis has been applied to a
system comprising the isolated mitochondrion plus an ADP-
regenerating system considering [ATP]/[ ADP] the controlled
variable) [107]. This has some resemblance to the situation
in muscle, where the ADP-regenerating system is myosin
ATPase, but there are two main differences. First, the
buffering ofADP concentration by the creatine kinase system
influences kinetics, although not steady state values. Second,
the ATPase rate U is a parameter, not a system variable whose
value emerges from its workings. Conceptually, this has the
effect of forcing many of the metabolic terms described in
[107] to special-case values: in particular, the elasticity ofU
toward [ATP]/[ADP] is zero, while the elasticity ofQ towards
[ATP]/[ADP] is simply -Km/([ADP] + Km) The control
strength ofU on Q is 1, while the control strength ofQ on U
is zero. Thus at steady state Q is completely determined by
U, which is independently determined. The control strength
of U on [ATP]/[ ADP] = -{ 1 + [ADP]/Km}. Thus ADP con-
centration is determined by ATP usage rate and mitochondrial
properties [7].
We have simplified by considering the mitochondrion to
be, inside and including the inner membrane, a single
functional unit. In contrast, a detailed 'top-down' analysis of
metabolic control in isolated mitochondria can be used to
calculate separately the control of Q exerted by the respiratory
system (the dicarboxylate carrier which allows substrates into
the mitochondrion, the enzymes of the tricarboxylic acid cycle
plus the electron transport chain), the phosphorylating system
(the adenine nucleotide translocase and the Pi transporter,
which allow the substrates of ATP synthesis in and the
product out, plus theATP synthase) and the proton leak [108].
In one study it was found that the phosphorylation rate is
controlled largely by the phosphorylating system itself in
State 4 (equivalent to basal state in vivo) and at intermediate
rates, while in state 3 (equivalent to very high ADP con-
centration) control of phosphorylation was shared between
the phosphorylating system and the respiratory chain [108].
Thus under the State 4 and State 4-State 3 transitions
considered here, control is largely vested in the phos-
phorylating system (that is, its flux control coefficient over
itself is near to unity), and it is reasonable to take this as
dominated by theADP-affinity ofthe translocase. A recently-
reported study of resting perfused muscle found that the
proton leak had about 40% of the control over the respiration
rate, compared to about 20% for ATP turnover. ATP turnover
itself had about 85% control over itself [109]. However, in
contracting muscle it seems likely that virtually all the control
of respiration resides in the ATPases, dominated by the
actomyosin ATPase [110], given that the elasticity ofATPases
towards Pi and ADP is negligible under physiological
conditions [104].
Next consider phenomenological nonequilibrium thermo-
dynamics, an approach which has been used to conceptualise
the control of oxidation in vivo. In a general treatment by
nonequilibrium thermodynamics [61] both ATP flux and
 277

0.00

1 0
'0 ' - 0 .
": : f-
0
.$ -0.2
:t
o 30
;:;:::J - 0 .
(I
c
:.:::;
-0.40
eo
( .)

-0.50

...I/...J...
"- .10
~ -0
........
:t
-0.15

a..l
.20
-0
C

-0.25

.I/-J
O. 1
": : -f
...E
.. .;

x 0.0
.2
x 7
(.) - 0 .
~.
" '-

i
i
i
I
I
i
i
/
I
I
I
I
278
oxygen consumption are linear function of /lG and redox
potential. Taking account of the Onsager symmetry relation-
ships [61], the fundamental equations can be written as
Q = (dQldG)/lG + [dQld(redox)]/l(redox) (73)
c = (dcldG)/lG + [dcld(redox)]/l(redox)
= [dQld(redox)]/lG + [dcld(redox)]/l(redox) (74)
where /l(redox) is the difference in redox potential and c is
oxygen consumption rate. The degree of thermodynamic
coupling = q = [dQld(redox)]/[(dQldG)(dQldC)]l!2 and the
phenomenological stoichiometry = [(dQldG)/(dQdC )]112 =(dcl
dG)1/2 = [dQld(redox)] 1/2. The load is envisaged as propor-
tional to free energy ofATP hydrolysis, i.e. U = (dU/dG)/lG,
where (dU/dG) is the conductance of the load. In general [61]
oxidative phosphorylation works at optimal efficiency if (dUI
dG)/(dQldG) =dU/dQ =(l_ q 2)l!2. Creatine kinase and adeny-
late kinase are seen as matching the load impedance to that
of the supply [111]. This approach has been elaborated in
many ways [112].
There are two main differences between this approach and
that of the present model. First, we take Q as a function of
ADP concentration alone, and therefore in effect of/lG alone
(with slope dQld/lG ",--6QMAx/RT, as above). We assume no
dependence on redox potential, but the nonequilibrium
thermodynamic approach fails if dQld/l(redox) is taken as
zero (q becomes undefined), so we must assume that the [dQI
d(redox)]/l(redox) is small and near-constant (i.e. that
changes in redox potential alone do not change Q appre-
ciably). Thus Q = (dQldG)/lG plus a constant tenn which we
could think of as QB. Secondly, we take U as a parameter
independent of /lG, so load conductance dU/dG is zero. Thus
the optimum-efficiency criterion requires q = 1, i.e. full
thennodynamic coupling at all rates ofATP turnover, and this
is observed [112].
Nonequilibrium thermodynamic treatments depend on
approximate linearity in 'flow-force' relationships (in this case,
Q-to-/lG relationships). These can be derived from standard
enzyme kinetics, which predicts sigmoidal flow-force relation-
ships with extensive linear regions around their inflection
points [113]. In mitochondria, this sigmoidicity is due to
nonlinear force-flow relations in the adenine nucleotide
translocase [66]. Sigmoidicity means that, out of State 4,
neither respiration nor phosphorylation nor the translocase
itself, can be near equilibrium [66]: this accords with labelling
experiments which show that the forward flux through ATP
synthesis considerably exceeds the reverse flux [67]. Un-
fortunately the symmetries and linear force-flow relations of
nonequilibrium thennodynamic treatments hold, theoretically,
only near equilibrium. However, the empirical linearities,
properly seen as near-linear regions of sigmoidal relationships,
permit such approaches in practice [66]. The relationship
between kinetic and nonequilibrium-thermodynamic ap-
proaches remains controversial [114]. However, it seems clear
that at least some of the nonequilibrium thennodynamic control
of phosphorylation by /lG can be viewed as kinetic control of
the translocase by ADP concentration [66].
9.2 Relationship to other models of whole-cell metabolism
Now we must discuss the relationship to other models ofthe
whole system (mitochondria plus ATPase plus creatine
kinase). 'Linear' models of mitochondrial function lend
themselves readily to predictions of kinetic responses [8, 11],
as discussed elsewhere [10] and in Section 3.1.
A simple model of facilitated diffusion has been used to
analyse high-energy phosphate fluxes at steady state [52], and
this has been combined with an ADP-control model of
mitochondrial function in order to simulate kinetic responses
to work jumps with and without the capacitance function of
the creatine kinase equilibrium [64]. This has some re-
semblance to the present model, but no means of accom-
modating intennediate creatine kinase activities.
As part of a minimal model of ionic events and bio-
energetics of the myocardial cell [62], a model has been used
in which oxi~ative ATP synthesis is linear with free energy
of ATP hydrolysis (see above) and this was embedded in the
'phosphocreatine shuttle' system (in fact, simply near-
equilibrium creatine kinase). A first-order approximation was
used to calculate net fluxes through cytosolic creatine kinase,
and these in tum used to calculate changes in phosphocreatine
concentration. In fact the results were similar to simple
equilibrium-creatine kinase models of cellular energetics, as
one would expect, and generated exponential kinetics for
phosphocreatine concentration with little change in ATP
concentration [62].
An attempt has been made to model the whole mito-
chondrion/creatine kinase system [71] which appears to show
that increasing activities of mitochondrial creatine kinase
increased the ability of the system to buffer metabolite
concentrations against a step increase in load, even when
cytosolic creatine kinase was at high enough activity for the
reaction to be near-equilibrium. At high loads the 'shuttle'
flux became limited by the activity of mitochondrial creatine
kinase. The amount of mitochondrial creatine kinase with
which the system could be set up was limited by the fact that
large increases reduced resting ADP concentration to un-
reasonably low levels [71]. In contrast, we find mitochondrial
creatine kinase has no effect on resting ADP concentration
and does not affect energetic buffering. The reason for this
is the design of the system [71] in which mitochondrial
creatine kinase is put inside the mitochondrial penneability
barrier (the inner membrane), rather than just outside it [47,
49]. In this model [71] mitochondrial creatine kinase supplies

intramitochondrialADP (conveying low-energy bonds from
cytosolic creatine) parallel to the translocase, thus reducing
the required cytosolic ADP concentration [71]. Note, how-
ever, that the porins of the outer membrane also offer a
significant permeability barrier which the mitochondrial
creatine kinase short-circuits (by allowing low-energy bonds
to enter as creatine rather than ADP). This is the phenomenon
described and modelled above (Section 5.2).
Furthermore, in our model all effects of mitochondrial
constraints on ratio of phosphocreatine to ATP fluxes be-
comes diluted by cytosolic creatine kinase equilibrium near
to the mitochondrion. In [71], cytosolic creatine kinase
activity has a small influence on resting ADP concentration
by virtue of its contribution to the cytosolic end of this version
of the phosphocreatine shuttle, but as its activity is very large,
the effect is a small one. In our model, and in all 'creatine
kinase equilibrium' models of mitochondrial control [10, 20,
52,69], the activity of cytosolic creatine kinase can have no
effect on resting phosphocreatine or ADP concentration.
Thus it is not, as sometimes stated, that the creatine kinase
system maintains a low ADP concentration; rather, itsATP-
buffering ability [10, 11] permits ADP concentration to be low
without the risk ofATP being exhausted by a few milliseconds
work.
Finally, there are some other differences. The Fedosov
model [71] makes the translocase first-order, rather than
obeying Michaelis-Menten kinetics. Both mitochondrial and
cytosolic creatine kinase are also made first-order [71].
Furthermore, it is assumed that [ATP]/[ADP] controls the
ATPase rate, whereas we have argued thatATPase rate should
be considered a manipulable parameter, at least in cardiac and
skeletal muscle.
10. Discussion: limitations and possible
extensions of the model
10.1 Adenylate kinase
We have not modelled the effects of adenylate kinase, which
ensures a near-equilibrium relationship between its reactants
(ADP, ATP and AMP) which can be written approximately
[43,57,115] as
[AMP] = K AK [ADP]2/[ATP] (75)
To a first approximation, this could be used to calculate values
of [AMP] from the results of the present model. More
satisfactorily, we could modify the model by incorporating
the kinetics ofthe adenylate kinase reaction, and the diffusion
ofAMP. This will not be attempted here, but it should be clear
from what has been said how this could be done.
279
There is a complication. In addition to being near-equili-
brium throughout the cytosol, adenylate kinase is also
associated with creatine kinase at both the mitochondrial and
myofibrillar ends of the system [116]. The rate of exchange
of the ~-phosphate of adenine nucleotides (from ADP to
AMP to produce ATP and vice versa) catalysed by adenylate
kinase is only 2% of the rate of y-phosphate exchange
catalysed by creatine kinase (fromATP to creatine and vice
versa), increasing to 20% in contracting muscle [116], no
serious error results from leaving adenylate kinase out of the
present model. It may be more important in anaerobically
contracting muscle [116], where adenyl ate kinase may act to
couple glycolyticATP production to the contractile apparatus
[117]. Indeed there is a linear relation between the flux
through adenylate kinase and the rate of lactate production
in contracting rat diaphragm [117]. Adenylate kinase may be
a more important shuttle mechanism in smooth muscle [42].
This aspect too could be incorporated into the present model,
perhaps by defining parameters analogous to -<I> and 'II to
specify the fraction of adenylate flux passing as AMP at the
ends of the system. This will not be pursued here.
10.2 More detailed analysis of the creatine kinase and
adenylate kinase equilibria
In the present paper we have ignored the fact that many of
the metabolites with which we are concerned are present in
multiple forms: free and bound to magnesium and other
ions, and in different ionisation states [11,43, 115, 118-
120]. We have ignored all these complications on the
grounds that they make no appreciable difference. The
present model is inherently oversimplified, and we are using
it to make some very simple points. Nevertheless, we will
briefly describe how this complications may be incor-
porated. A more accurate statement of the creatine kinase
equilibrium than Eq. 1 is
[IADP]/[IATP] = ([Cr]/[IPCrD/(hKcK) (76)
in which the summation signs refer to the sum of free and Mg-
complexed forms [43, 115] and the quantity KCK can be shown
[115] to be given by
lI(KCK) =
{Ki(~K'4)} {I + gK,z + (h/K4)} {I + g~ +
(h/~)(1 + gK,o)}/{l + gK6 + (h/K,)(1 + gK9)} (77)
where the constants are as defined in [115] (see Table 3 ), and
for brevity we write h = [W], g = [Mgz+]. Analogously, for
the adenylate kinase (myokinase) equilibrium,
(78)
280

Table 3. Further dissociation and equilibrium constants"b

Symbol and definition Value and symbol in cited reference

Acid dissociation constants (M) [ 115] [43] [57,119]

Kl = h[ATp4-]l[HATpJ-] Kl = 1.08 X 10-7 IIK\TP = 1.12 X 10-7 K\TP = 1.09 X 10-7
K, = h[ADpJ-]/[HADP'-] K, = 1.20 X 10-7 IIK\DP= 1.80 x 10-7 K\DP = 1.05 X 10- 7
K J = h[AMP'-]/[HAMP-] KJ = 3.24 X 10-7 11K" AMP = 1.96 X 10-7
K4 = h[PCr']/[HPCr'-] K4 = 3.16 X 10-5 IIKH PC, = 3,02 X 10-5 K'rc, = 3.16 x 10-5
K5 = h[HPO/-]/[H,P04-] K5 = 1.76 X 10-7 11K" "P04 = 1.34 X 10-7 K'ri = 1.61 X 10-7

Stability constants (M-') [ 115] [43] [57, 119]

K6 = [MgATp 2-]/(g[ATp4-]) K6 = 2.63 X 104 KMgATP = 4.5 X 104 KbMgATr'- = 5.46 x 104
K7 = [MgADP-]/(g[ADpJ-]) K7 = 2.32 X 10J KMg ADP = 2.69 x 10J Kb MgADr'- = 1.6 x 10J
Kg = [MgAMP]/(g[AMP2-]) K8 =60.1 KMg AMP = 60
K. = [MgHATP-]/(g[HATpJ-]) K9 = 35.5 KMg ATPH = 567 Kb MgHATpl- = 46.1
KID = [MgHADP]/(g[HADp2-]) K'D = 32.4 KMg ADPH = 84 Kb MgHADP = 38.5
Kil = [MgHP0 4 ]/(g[HPO/-]) Kil = 93.1 KMg Pi = 93 Kb MgHP04 = 110
KI2 = [MgPCr ]/(g[PCr'-]) KI2 = 20.0 KMg pc , = 20 Kb MgP-C, = 23.8
KI5 = [Mg,ADP]/(g[MgATP'-])
Adenylate kinase equilibrium constants' [115] [43] [57,119]
K13 = [AMP'-][ATp4-]/[ADpJ-l' K IJ = 0.363
[AMP'-][MgATp2-]/([ADpJ-][MgADP-])
KAK = CLAMP] [IATP]/[IADPl' KMYK = 1.12

Creatine kinase equilibrium constants (M-'), [ 115] [43] [57, 119]

K'4 = [Cr][MgATP2-]/(h[PCr2-] [MgADP-]) Kl4 = 3.52 X 10' KCPK = 3.31 X 109
KCK = [Cr][(IATP]/(h[LPCr][IADP]) K eq" lI(hR cPK ) KCK = 1.66 X 109

'The left-hand column gives a defmition of the quantity and the symbol used here (based mainly on [115]); the right-hand columns give published values and
alternate symbols from several sources_ For brevity, h = [H+], g = [Mg'+]. bAlso: in [120] Kl (called 11K) = 9.2 x 10-" M, K6 (called KMgATP) = 3.48 x 10" M-',
~ (called ~gHATP) = 542 M-', and what we may define as K l5 = [M~ADP]/(g[MgATP2-]) (called ~g2ATP in [120]) = 4.0 M-l. Some further complications have
been ignored here: free [Mg2+] should, strictly, allow for MgCI+ (~gC' = 3.4 M-l) [120]); the Pi equilibria should, strictly, include calcium and sodium binding
(KbC,HP04 = 97.5 M-' [119], KbN,HP04 = 14.1 M-' [119]). 'RCPK is defined as (BADpBpc/BATP)(K/K7)/(hK,J in [II] (and incorrectly in [43]), and RAdk is defined as
{BATPBAMP I(BAOP)2}KIJK,1K6 in [II, 43], where the constants are given in the tenninology of [115] and Bx = [IX]/[X~] (i.e. total X relative to free ionic form)
[43].

where KAK can be shown [115] to be given by
KAK = K13 {l + gKg + hlKJ {I +gK6 +
(hlK[)( 1 + gK[o)} I {I + gK7 + (hlK2)(1 + gK9W
It is relatively simple to incorporate this more sophisticated
analysis into equilibrium models of mitochondrial regulation
in vivo (Section 3); an example of this approach is described
in [11, 43]. The ATP-buffering effect of the creatine kinase
equilibrium described in Section 2.2 can be expressed by the
'buffering factor' (fb) [11, 43] which is defined as
fb = d[LPCr]/d(2[IATP] + [IADP])
It can be shown by a detailed analysis of the creatine kinase
and adenyl ate kinase equilibria [43] that (adjusting the
terminology)
where, for convenience we define y = [IADP]/[IATP]. From
this expression, changes in [ATP] can be shown to be related
to changes in creatine metabolites by
(79)
[IATP]/[TA] = 11 {I + ([Cr]/[IPCr])/(hKcK ) + KAK([Cr]1
[IPCr])2/(hKcK)} (82)
The approximate considerations above show that 1/fb '"
2d[IATP]/d[IPCr] '" 0 [lO], and so [IATP]/[TA] '" 1. Detailed
calculations bear this out over all except large changes in
phosphocreatine concentration [43].
Similarly, it would be possible in principle to incorporate
(80) these refinements into a nonequilibrium analysis. In this more
accurate formulation the true substrates of creatine kinase are
MgADP- and MgATp2- [53]. Thus the net flux through
creatine kinase is more accurately given by the modifying
(Eq. 40) as follows:

fb = ([IPCr]2hK cK/([TCr]/[TA])} {1 + y + K AKy2}1 CIC MAX = {([MgADP--][PCr2-])~([MgATP2-][Cr]1

{I + 4KAKy + KAKy2} (81) ([W]K[4)}/M (83)

where K14 is defined in [115] and in Table 3, and
M/(Kia~) = 1 + (K03 + K04[Cr] +K06 [PCr2-])[MgATp 2-]
+ KoJPCr2-] + (Koo + K05 [Cr] + K02 [PCr])[MgADP-]
where M is as given above. Note that corresponding expres-
sions for exchange flux [53] can be obtained by modifying
Eq. 41 and 42:
E = CMAiMgATp2-][Cr]/([W]K[4M)
phosphocreatine synthesis i.e. ATP hydrolysis (84)
E = CMAX [MgADP-][PCr 2-]/M
phosphocreatine breakdown i.e. ATP synthesis (85)
A similar expression could be used to calculate net flux through
the adenylate kinase reaction, if desired. A full analysis of the
spatial aspects of the system would entail setting up diffusion
equations for each ionic form of each component (whether or
not involved in the chemical exchange). Finally the total
concentrations [IX] in each box would be calculated by
summing the concentrations of the ionic forms.
The situation is simpler if equilibrium is assumed, when
straightforward expressions can be used to relate con-
centrations of particular species to the total concentrations
of each metabolite. The following are most relevant:
[ADpJ--]I[IADP] = 1/{1 +gK7+ (h/K 2)(1 +gK[o)} ""
1/{1 + gK7} (86)
[MgADP-]/[IADP] = g[K7 + (hKIO/K2)]1 {[I + hK101
(K2~)][1 + g~ + (h/KJ(l + gK IO )]}
"" II {1 + 1/(gK7 )} (87)
[MgATp 2-]/[IATP] = g[K6 + (hK/K[)]/{[l +
hK9(K[K6)][1 + gK6 + (h/K[)(l + gK9)]}
"" 1/{1 + 1/(gK6)} (88)
in which the constants are in the nomenclature of [115] (see
Table 3) (Note that [IADP] does not include ADP bound to
proteins, notably actomyosin, which is by far the dominant
contribution to chemically-determined ADP content in muscle
[57]). At pH 7, assuming [Mg2+] "" 1 mM [118, 120] and the
values for the constants given in [115], these expressions give
[ADp 3-]/[IADP] = 0.24 ("" 0.3 in the approximation),
[MgADP-]/[IADP] = 0.56 ("" 0.7 in the approximation) and
[MgATp 2-]I[IATP] = 0.93 ("" 1 in the approximation). Also,
[PCr2- ]I[IPCr] = 1I {I + gK12 + (h/K4)} (89)
which is equal to 0.98, so near to unity that the difference can
probably be ignored for any conceivable analysis of the kind
presented in this paper.
281
There are two more implications. First, ADP and ATP are
transported across the inner mitochondrial membrane by the
translocase in their uncomp1exed ionised forms ADpJ-- and
ATp4- [58], which are only minor components under physio-
logical conditions, but mitochondrial Km values are cited
conventionally in terms of the total ADP concentration
[IADP] [7] (they are related [7] by Eq. 86 above). Second,
given that MgATp 2- is the form involved in active transport
and muscular contraction, only the hydrolysis of this form is
bioenergetic ally relevant, and so the correct expression for
the free energy of ATP hydrolysis is given by exp{(LlG-
LlG)/RT} = [MgADP-][Hl0 4-]/[MgATp2-] [115].
One obstacle to such a detailed consideration of the creatine
kinase and adenylate kinase equilibria is that the literature has
become confused by alternate nomenclatures. For the reader's
convenience, we list in Table 3 equivalent terms in some of
these nomenclatures, along with relevant published values.
10.3 Extension to two or three dimensions
For simplicity we have modelled only one spatial dimension.
However, the model should be readily generalisab1e to the
kind of geometry previously used to examine questions of
diffusion [52, 104l Consider first the system without creatine
kinase. The myosin ATPase is a cylinder of radius ra and the
cytosol is an annulus of radius rm bounded by a thin layer of
mitochondria. We need only consider a single disk of unit
thickness. Concentrations and fluxes are expressed relative
to volume; the volume of the whole disk is 1t(rm)2. In a
continuous system the two-dimensional equivalent ofEq. 58
can be written, by analogy to the heat equation written for
polar co-ordinates in a system with radial symmetry [101],
as
(90)
in which the term multiplying Dx is the two-dimensional polar
Laplacian of [Xl The general solution of Eq. 90 is given by
[101]
(91)
where k[ and k2 are constants to be found from the boundary
conditions.
We are interested principally in the steady state where
(d[X]dt) = O. Fick's first law (Eq. 60) still applies. However,
the area across which diffusion occurs is now 2m (as the disk
is of unit thickness) and as the volume of whole disk is 1t(rm),
the ratio (area/volume) = 2/rm and we can rewrite Eq. 60 as
JX = (absolute flux)/(volume) = -DxCd[X]/dr)(areal
volume) = -2(D xd[X]ldr)/rm (92)
282
At r = rm, JX = JXmand so from Eq. 92, d[X]jdr = -Jxmrm/(2Dx).
Differentiating the general solution Eq. 91 gives d[X]/dr = k/
r, and so d[X]m/dr = k/rm. Equating these two expressions for
d[X]m/dr, we obtain kl = Q(rm)2/(2Dx). Substituting this into Eq.
91 gives Is = [X]m + []Xm(rm)2/(2Dx)].ln(rm). Thus Eq. 92 can
be rewritten as
(93)
Notice that this differs from a previous treatment [52] in some
respects. InEq. 93 no solution is defined forr= 0, so the radius
ra of the myosinATPase cylinder must be finite. Equation 1 in
[52] is equivalent to the general heat-flow equation Eq. 90 only
if (in our terms) JX is identified with (d[X]/dt)diffusion; but at
steady state the latter is zero while the former is not. Thus the
solution given in [52] is a quadratic which can be written
(94)
which implies that at r = 0, d[X]m/dr = 0, and therefore (from
Eq. 92) that J = O. Amore complex treatment ofaradial model
has also been described [104].
Equation 93 gives the steady-state concentration profile
across the radius of this idealised cylindrical muscle in the
absence of creatine kinase, and can be compared with the
linear case where the steady-state concentration profile is
linear (Section 6.2). In the presence of creatine kinase at
equilibrium, the arguments of [52] (Section 7.2) show that
this should be true of the gradients of all metabolites, and that
these are related by the creatine kinase equilibrium ex-
pression. This can easily be applied to the two-dimensional
model. However, it is a finding of the present work that unless
creatine kinase activity (and the number ofboxes) is infinite,
there can be nonlinearity in the gradient of (most noticeably),
the concentration of ADP (Fig. 9B). It must be admitted,
therefore, that the detailed profile of ADP concentration at
low creatine kinase activity could not be obtained simply by
transforming the concentration gradients in the linear model
by a logarithmic function derived from Eq. 35.
We could pursue this further by developing a numerical
approximation. Instead of a continuous disk, imagine a
number of annuli of width dr. We can proceed by writing the
difference equivalent of Eq. 90 directly. Since d[X]./dr can
J
be approximated as ([X]j+l - [X])!&-, and a 2 [X]/ar can be
approximated by {([X]- J
[X]l)/&--([X]+l-
J- J
[Xl)/&-}/&-=
J
-{1/~) {(2[X]j - [X]j-l - [X]j+l)' the difference approximation
to Eq. 90 is
(a [x]/at)diffusion = Dx {([X]j-l' - 2 [X]j + [X]j+ 1)/~ + (1/
r)([X]j+l - [X])/&-} (95)
This is analogous to Eq. 62 in the one-dimensional model,
and can be used in a difference equation analogous to Eq. 64:
([X]t 1 - [X]/)/Llt = Dx {([X]j-l - 2 [X]j + [X]j+l)/
Llr + (IIr)([X]j+l - [X])/&-} q (96)
This, or its Crank-Nicholson equivalent, could be used in
order to model the system for two dimensions.
10.4 Enzyme-enzyme interactions
This model describes the interrelationship between the
chemical flux through creatine kinase and the diffusion flux
of metabolites through the cytosol. In this sense it is a
generalisation of the model of [52], which examines the
implications for diffusion ofthe creatine kinase equilibrium.
We have not considered direct transfer ofligands (in this case
ADP) between creatine kinase molecules. This kind of
'vectorial ligand conduction' has been postulated as im-
portant in the operation of both the adenylate kinase and
creatine kinase shuttles [14, 117]. The main point about this
mechanism is that ligand movement can occur without
reference to the ligand concentrations in bulk cytosol.
Certainly creatine kinase is in sufficient concentration in
cytosol that enzyme-enzyme interactions will occur at a high
rate compared to enzyme-substrate interactions. Theory exists
to calculate the rate of collision of two enzymes [121], which
can be adapted to the case of pairs of creatine kinase mole-
cules. Using the diffusion coefficient (5.78 x 10-7 cm 2sec- 1)
and radius (3.9 nm) of creatine kinase given in [121], we can
calculate the second-order rate constant of the rate of collision
between two creatine kinase molecules [121] as
kD = 161t.(Avogadro's number).(diffusion
coefficient).(radius) = 6.8 x 10 M.sec- 1 (97)
From this, the molar concentration of creatine kinase (about
0.5 mM in skeletal muscle [121]) and the dissociation
constant for ADP (see Table I) we can calculate [121] the rate
of formation of (for example) the complex enzyme-ADP-
enzyme. This is given by
2kD [enzyme-ADP][enzyme] ,., 2~[ADP][enzymeF/
Kia = 760 M.sec-1 (98)
This is 5 orders of magnitude faster than the rates of flux
through creatine kinase flux; thus two enzyme molecules will
associate with a single molecule of ADP far faster than an
ADP molecule will be transformed into ATP. It is possible,
then, that exchange of substrate between temporarily-
associating enzyme molecules may play an important role in
the reaction mechanism in vivo. However, the present
evidence is that creatine kinase behaves kinetically in vivo
as though it has access to a well-mixed pool of substrates [53,
122], and that there are no unexpected discrepancies between

the predicted fluxes in vivo and the rates measured by
chemical and magnetic resonance spin transfer [123, 124]
10.5 Replacement of phosphocreatine
Total creatine content can be manipulated in vivo by dietary
creatine supplementation [125]. This is obviously readily
modelled. Total creatine content can also be reduced by
replacement by a creatine analogue such as ~-guanidino
proprionic acid (~-GPA) [9,41,84]. In skeletal muscle this has
relatively little effect on mechanical function [9, 84] although
the kinetics of phosphocreatine concentration are altered in
accordance with a prediction of one of the linear models of
mitochondrial control [9]. Thus the kinetics of phosphocreatine
in the present model could be modelled by simply reducing
[TCr]. Changes in the phosphorylated form of ~-GPA (~
GPAP) also occur however [9] and to model these would
require the addition of further variables [~-GPA] and [~-GPAP]
and further equations describing the kinetics of their inter-
conversion by cytosolic creatine kinase (~-GPA is not phos-
phorylated by mitochondrial creatine kinase [126]). In vascular
smooth muscle, at least, there is also the complication that a
large fraction ofATP, apparently that produced glycolytically,
is not available for phosphorylation of~-GPA or creatine [41].
10.6 Calcium-activation of mitochondrial enzymes
We have not considered possible roles of calcium-activation
of mitochondrial dehydrogenases [37, 38], whose regulatory
contribution has not yet been demonstrated in skeletal muscle,
but which are thought to be important in heart [14, 38, 127].
The effect ofa calcium signal is (in terms of the 'top down'
approach alluded to above) is to stimulate the respiratory
system. The effect of this on the ATP synthesis rate would
depend on the relevant flux control coefficient under given
circumstances. To a first approximation it could be modelled
in the present system by making QMAX as a function ofU, so
that QMAX would be switched up by a closed-loop mechanism
as soon as the ATPase-rate (load) increased. This would mean
the change in ADP concentration needed for a work jump
might be substantially reduced; if this function were of the
form QMAX/U == constant, ADP concentration would not need
to change at all (perfect open-loop control), and this may
approximate the situation in cardiac muscle under some
circumstances [14, 31]. To retain a degree of generality we
could, rather, model this situation by
(99)
where Q MAX B was the (unstimulated) mitochondrial capacity
at basal ATP turnover, QMA/ is the increase produced by a
283
maximal calcium signal, and Ku is a term representing the
sensitivity of the calcium response expressed in units ofATP
turnover (i.e. ofthe calcium signal itself, and of the response
to it, combined). The present case would then be seen as
corresponding to Eq. 99, either with QMAX S "" 0, or (more
realistically) l/Ku "" O. A similar approach could be used to
formalise the effects of any other 'non-ADP', open-loop
influence on mitochondrial function, whether or not calcium
is involved.
10.7 Translocase kinetics
We have ignored the detailed work on the kinetics of the
translocase in liver mitochondria [67, 128]. In liver mito-
chondria in vitro, during transition from zero to maximalATP
synthesis, the forward flux fromADP and Pi toATP doubles,
while the reverse flux was reduced to less than 10% of its
zero-flux value [67]. The reaction is always far from equili-
brium except at low rates of net ATP synthesis (i.e. the
mitochondrial translocasel ATPase system is not at equili-
brium with extramitochondrial adenine nucleotides and Pi);
the ratio offorward to reverse flux was 3 at half-maximalATP
synthesis [67]. TheATP synthase itself is thought to be largely
reversible in vivo, and so the unidirectionality probably arises
from the properties ofthe translocase [67]. What controls the
reverse rate is unknown, but intramitochondrial [ATP]
influences the exchange flux through the translocase [128].
For present purposes, all these details can be combined into
the overall transfer function represented by Eq. 5. It would
be relatively easy to incorporate recent arguments that the
kinetics of the translocase with respect to [ADP] are co-
operative with a Hill coeffcient of about 2 [92, 129], in
accordance with recent detailed studies of the transport
kinetics [130].
10.8 Reversibility of mitochondrial CK
In Eq. 46 we took mitochondrial creatine kinase as essentially
irreversible. A more complete analysis would not make this
assumption, but would use the generalised rate expression
given in [91] (again there are several nomenclatures, and for
the reader's convenience translations between these are given
in Table 4, along with appropriate values). In essence the rate
expression is the same as for cytosolic creatine kinase, and
so W Ris given by
W/WR,MAX== {([ATPlm[Cr]jKMic/)-
([ ADP]im[PCrlm)} IN (100)
in which the full expansion ofN is given by
284

Table 4. Kinetic and equilibrium constants for mitochondrial creatine kinase

Definition (cf. Table 2) Values and symbols Alternate notation

mitochondrial form' mitochondrial form b cytosolic form'

[91) [82) [53,74) [72)

Michaelis constants
enzyme-ATP and Cr (Kmitc in Table 2) K =5.2x 10-3M K. = 5.2 x 10-3M
enzyme-Cr and ATP (Km\ in Table 2) K," = 0.15 X 10-3 M K,
enzyme-ADP and PCr K 'p = 1.0 X 10-3 M K.
enzyme-PCr and ADP K d = 0.13 X 10-2 M Kq

Dissociation constants
enzyme and ATP K i, = 0.75 X 10-3 M K i , = 0.75 X 10-3 M
enzyme and Cr K i" =26 X 10-3 M K ib =29x10-3 M
enzyme and PCr K. = 1.6 x 10-3 M K..p = 1.6 X 10-3 M
"p
enzyme and ADP Kid = 0.208 x 1O~ M K iq

Inhibition constants
enzyme-Cr and ADP K'd = 0.042 X 10-2 M K,q
enzyme-ADP and Cr K lor =5.2x 10-3 M KIb
enzyme-ATP and PCr Klop =24x 1O-3 M K,p = 20-50 X 10-3 M
enzyme-PCr and ATP K" = 11.25 X 10-3 M K"

Equilibrium constant ~iC: = 75 at pH = 7.4

'This column gives symbols and values for kinetic constants (from [91)) used in the more detailed treatment of mitochondrial creatine kinase kinetics
described in Section 10.8 and in [91); bThis column gives alternate symbols for the mitochondrial enzyme from an earlier kinetic analysis [82); 'These two
columns show alternate nomenclatures of the corresponding kinetic constants for the cytosolic enzyme in [72, 73) and [53) (see Table 2).

N/(KmitlqKmit)P
= 1 + [ADP].unIKmitla + [PCr].1mIKmitIb +
[ADP].1m[PCr].1mI(Kmitla Kmitb) + [ATP].1mIKrnitIq +
[Cr].1m[ATP].1mI(KrnitKmit)+[Cr].
p lq un
[ADP].1ml(KrnitIpKmit)+
13
[PCr].1m [ATP].1mI(Kmit IbKmitlq )
and K MiCK' is the apparent equilibrium constant. To point up
the similarities to the cytosolic case (Eq. 40), this expression,
although derived from [91], uses the notation of [53, 74] with
superscripts to denote the mitochondrion; translations are
given in Table 4. (Note that a difference in form from Eq. 40
is due to the fact that both the net rate and the maximum rate
are here measured in the direction of phosphocreatine
synthesis, and there is as yet no information about the anion-
complex term). Eq. 46 is a special case ofEq. 100 which
follows from neglecting the reverse direction.
11. Summary and conclusions
Theories of mitochondrial control and bioenergetic regulation
in vivo are complicated by the metabolic and physical
structure of the energy production/usage system. In vivo the
mitochondrion is embedded in the creatine kinase equili-
brium, which makes it hard to distinguish between candidate
models of mitochondrial control. Although the relative
simplicity of the system means that approximations (e.g. to
creatine kinase equilibrium) are useful, interactions make it
difficult to analyse quantitatively. Also, there is controversy
about the functional role in vivo of the creatine kinase
associated with the mitochondrial adenine nucleotide trans-
locase and the myofibrillar ATPase. In an effort to define the
roles of the essential elements of the system, we set out to
incorporate several components into a quantitative model: an
appropriate theory of mitochondrial control; a generalised
theory of coupling between mitochondrial creatine kinase and
the adenine translocase, and between the myosin ATPase and
its associated creatine kinase; the detailed kinetics of cytosolic
creatine kinase; and some aspects of spatial diffusion.
In summary, we have modelled the cell as follows. The
muscle cytosol is modelled simply as a one-dimensional bar
running between a mitochondrion making ATP and a myo-
fibril using it. Metabolites can diffuse along the bar depend-
ing on concentration gradients. High-energy phosphates (ATP
and phosphocreatine) and low energy metabolites (ADP and
creatine) can be exchanged at any point via the reaction
catalysed by creatine kinase. Oxidative ATP synthesis is
controlled by juxtamitochondrial ADP concentration. To
allow for possible coupling between mitochondrial creatine
kinase and the adenine nucleotide translocase, we define a
parameter <I> which sets the fraction of the total flux carried
out of the mitochondrial unit by ATP rather than phospho-
creatine. This simplification is justified by a detailed analysis
of the interplay between the mitochondrial outer membrane

porin proteins, mitochondrial creatine kinase and the adenine
nucleotide translocase. The ATPase rate is the input function.
To allow for possible coupling between creatine kinase and
the ATPase, we define a coupling parameter 'I' which sets the
fraction ofthe total flux into the ATPase unit which is carried
by ATP rather than phosphocreatine. Both <p and 'I' go from
zero (complete coupling) to 1 (no coupling).
In common with some other models, there are implications
for the 'buffering' ofATP concentration, and the steady-state
mitochondrial 'transfer functions'. Changes in ATP con-
centration are always small. The hyperbolic relationship
between oxidativeATP synthesis rate and spatially averaged
ADP concentration is equivalent, at high creatine kinase
activity, to a near linear relationship to creatine concentration,
and to a near-linear (but more properly sigmoid) relation to
free energy of ATP hydrolysis. These facts result from the
constraints imposed by creatine kinase on the metabolite
concentrations. The degree of functional coupling (between
. mitochondrial creatine kinase and oxidative ATP synthesis
or between myofibrillar creatine kinase and the myosin
ATPase) has little effect on these relationships at normal
creatine kinase activity, because cytosolic creatine kinase
'shorts out' the effect of changes in the proportion of adenine
nucleotides and creatine compounds at the ends of the system.
However, the tendency is for increasing degrees of coupling
to decrease the mean concentrations of both ADP and creatine
required for high rates of ATP synthesis. Lowering the
creatine kinase activity does alter these relationships: both the
mean ADP and the mean creatine concentration required for
a given rate of ATP synthesis are higher
Next we consider the creatine kinase exchange flux and
diffusion fluxes at steady state. At high creatine kinase
activity, exchange flux shows a biphasic and approximately
symmetrical dependence on ATP turnover, and a highly
asymmetrical dependence on ADP concentration, largely
independent of coupling. At low creatine kinase activity the
exchange flux becomes a much less symmetrical function of
ATP turnover rate. The fraction of flow at steady state carried
in the middle of the model by ATP is always small, but
increases with ADP concentration and rate of ATP turnover.
The degree of coupling has very little effect on this except
where creatine kinase activity was too low to permit the
assumption of equilibrium.
Next we consider gradients in metabolite concentrations
and in creatine kinase net flux at steady state. Both creatine
and ADP concentration show small gradients decreasing
towards the mitochondrion (in the direction of their net flux),
while ATP and phosphocreatine concentration show small
gradients decreasing towards the myosin ATPase. Sur-
prisingly in this linear model, the ADP concentration gradient
is nonlinear. This is entailed by the existence of a gradient of
net creatine kinase flux. This results from the need to
transform some of the 'adenine nucleotide flux' at the ends
285
of the model into 'creatine flux' in the middle. If 'I' = <p the
overall net flux is zero, if not it will be greater or less than
zero, although still very small compared to total exchange
flux. This is why the ADP concentration gradient is in-
fluenced by the activity of cytosolic creatine kinase such that
a reduction in activity decreases ADP concentration at the
mitochondrial end and increases it at the ATPase end. The
zero-point for these gradients is not necessarily in the middle
of the model, probably because of the asymmetry in the
creatine kinase rate expression introduced by the anion-
stabilised dead-end complex term
The model makes predictions about the response to work
jumps. During work-jump transitions, spatial-average res-
ponses approximate to those of models of mitochondrial
control which assume equilibrium conditions for creatine
kinase. In particular, overall phosphocreatine concentration
shows exponential kinetics. In response to a step increase in
ATPase activity, there is a rise in ADP concentration, a
stoichiometric fall in ATP concentration (trivial in com-
parison to its much larger concentration), a rise in creatine
concentration and a stoichiometric fall in phosphocreatine
concentration. These changes start at the ATPase end and
propagate towards the mitochondrion, so that at any time,
changes are larger at theATPase end. At the new steady state,
gradients exist across the cell in proportion to the new ATP
turnover rate: these are proportionately much smaller for
creatine than for ADP. The propagation of signal fromATPase
to mitochondrion can be seen more clearly in the time-
derivative plot (Fig. 12C), which show the rate of increase
of, for example,ADP concentration starts high at the ATPase
end and diminishes in space, towards the mitochondrial end,
and in time, towards the new steady state at which all rates
of change are zero. It has been argued that ADP for the
translocase must be generated between the outer and inner
membranes, and so that the translocase-creatine kinase
system acts as an amplifier, receiving either (weak)ADP and
creatine signals [81]. It has been further proposed [81] that
these signals cross the cytosol as a series of waves (ADP
produces rise in phosphocreatine at one point which in turn
raises ADP at the next points, and so on), which travel with
the frequency of contraction, but with amplitude damped by
high mobility of creatine and phosphocreatine, and at the rate
of creatine kinase exchange. It has not been clear whether
such a phenomenon would require a structured cytoplasm (or
what this would mean). We have no evidence from the present
model for any such wave phenomena.
In conclusion, the model is in many respects clearly
oversimplified (e.g. one-dimensional), but this provides the
simplest demonstration of its most important features. We
have endeavoured to encompass as many as possible of the
different models, theories and emphases now current either
by including them as special cases of general incorporated
mechanisms (particular values of parameters such as <p and
286
'l'), or by showing in detail how the model could be extended
to accommodate them (e.g. the extension beyond one dimen-
sion; different mitochondrial transfer functions and the
regulation of mitochondrial creatine kinase; the porin per-
meability barrier).
Appendix: Average exchange flux
To predict the results of total creatine kinase flux measure-
ments, we may define the ratio of mitochondrial creatine
kinase to total creatine kinase as Pmlto
. and ratio of myofibril-
associated creatine kinase to total creatine kinase as PATPase ,
SO that the measured overall exchange flux is equal to
E overall = P mito JPCr0 + P ATPase JPCrN + (1 - P mito -
P ATPase) E eyto
= Pmito(l- m 't"
)Q + PATPase (l-ltr)Q
't'
+ (l-Pmito -PATPase)Ecyto
where E eyto is the (spatial) mean creatine kinase exchange flux
in the cytosol. Clearly the relation between observed E overall
and Q depends on P. mlto
and PATPase To a first approximation
these can be estimated by the fraction of total creatine kinase
activity which is associated with the mitochondrial and
myofibril fractions. At the mitochondrial end, in heart muscle
Pmito "" O. 1-0.5 [54, 131, 132]; in skeletal muscle probably
Pmito "" 0.1 [53, 133]; in smooth muscle P mito varies from 0.03
in stomach to 0.6 in carotid artery [42]. At the myofibrillar
end, P ATPase"" 0.2 in smooth muscle [56, 77, 83] and probably
"" 0.2 in heart [48]; there appears to be no information on
skeletal muscle.
References
I. Chasiotis D: The regulation of glycogen phosphorylase and glycogen
breakdown in human skeletal muscle. Acta Physiol Scand Suppl
518: 1-68, 1983
2. Ren JM, Hultman E: Regulation of phosphorylase a activity in human
skeletal muscle. J Appl Physiol69: 919--923, 1990
3. Brindle KM, Blackledge MJ, Challis RAJ, Radda GK: 3\p NMR
magnetization-transfer measurements ofATP turnover during steady-
state isometric muscle contraction in the rat hind limb in vivo.
Biochemistry 28: 4887-4893, 1989
4. Chance B, Leigh J Jr, Clark BJ, Maris J, Kent J, Nioka S, Smith D:
Control of oxidative metabolism and oxygen delivery in human skeletal
muscle: A steady state analysis of the work/energy cost transfer
function. Proc Nat! Acad Sci USA 82: 8384--8388, 1985
5. Kushmerick MJ, Meyer RA, Brown TR: Regulation of oxygen
consumption in fast- and slow-twitch muscle. Am J Physiol263: C598--
C605,1992
6. Nioka S,Argov Z, Dobson GP, Forster RE, Subramanian HV, Veech
RL, Chance B: Substrate regulation of mitochondrial oxidative
phosphorylation in hypercapnic rabbit muscle. J Appl Physiol 72: 521-
528,1992
7. Mahler M: First-order kinetics of muscle oxygen consumption, and
an equivalent proportionality between QO, and phosphorylcreatine
level. J Gen Physiol 86: 135-165, 1985
8. Meyer RA: A linear model of muscle respiration explains mono-
exponential phosphocreatine changes. Am J Physiol254: C548--C553,
1988
9. Meyer RA: Linear dependence of muscle phosphocreatine kinetics on
total creatine content. AmJ Physiol257: C1149--C1157, 1989
10. Kemp GJ: Interactions ofmitochondrialATP synthesis and the creatine
kinase equilibrium in skeletal muscle. J Thear Bioi 170: 239--246, 1994
II. Funk C, Clark A, Connett R: A simple model of aerobic metabolism:
Applications to work transitions in muscle. Am J Physiol258: C995-
ClOO5,199O
12. Chance B, Williams GR: Respiratory enzymes in oxidative phos-
phorylation. III. The steady state. J Bioi Chern 217: 409-427, 1955
13. Chance B, Williams J: The respiratory chain and oxidative phos-
phorylation. Adv Enzymol17: 65-134, 1956
14. Saks VA, Kuznetsov AV, Khuchua ZA, Vasilyeva EV, Belikova JO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis: Possible involvement of mitochondrial-
cytoskeleton interactions. J Mol Cell Cardiol27: 625-645, 1995
15. Winkler HH, Bygrave FL, Lehninger AL: Characterization of the
atractyloside-sensitive adenine nucleotide transport system in rat liver
mitochondria. J Bioi Chern 243: 20--28, 1968
16. Jacobus WE, Moreadith RW, Vandergaer KM: Control of heart oxidative
phosphorylation by creatine kinase in mitochondrial membranes. Ann
NY Acad Sci414: 77~9, 1983
17. Gyulai L, Roth Z, Leigh JS, Chance B: Bioenergetic studies of
mitochondrial oxidative phosphorylation using 3\Phosphorus NMR. J
Bioi Chern 260: 3947-3954, 1985
18. Due ED, Vignais PV: Kinetics and specificity of the adenine nucleotide
translocation in rat liver mitochondria. J Bioi Chern 244: 3920--3931, 1969
19. PfaffE, Heldt HW, Klingenberg M: Adenine nucleotide translocation
of mitochondria. Kinetics of the adenine nucleotide exchange. Eur J
Biochem 10: 484-493, 1969
20. Jeneson JAL, WesterhoffHY, Brown TR, van Echteld CJA, Berger R:
Quasi-linear relationship between Gibbs free energy ofATP hydrolysis
and power-output in human forearm muscle. Am J Physiol268: C 1474-
CI484,1995
21. Quistorff B, Johansen L, Sahlin K: Absence of phosphocreatine
resynthesis in human calf muscle during ischaemic recovery. Biochem
J291:681-686,1993
22. Thompson CH, Kemp GJ, Sanderson AL, Radda GK: Skeletal muscle
mitochondrial function studied by kinetic analysis of postexercise
phosphocreatine resynthesis. J Appl Physiol 78: 2131-2139, 1995
23. Kemp GJ, Taylor DJ, Radda GK: Control of phosphocreatine resynthesis
during recovery from exercise in human skeletal muscle. NMR in Biomed
6: 66-72, 1993
24. Boska M: ATP production rates as a function of force level in the
human gastrocnemius/soleus group using lip MRS and IH MRI. Magn
Reson Med 32: 1-10,1994
25. Amold DL, Matthews PM, Radda GK: Metabolic recovery after exercise
and the assessment of mitochondrial function in vivo in human skeletal
muscle by means ofP-31 NMR. Magn Reson Med 1: 307-315, 1984
26. Wackerhage H, Hoffman U, Essfeld D, Leyk G, Mueller K, Zange J:
Effects of glycolysis on mitochondrial respiration. In: Proc 4th Ann
Meeting Internat Soc Magn Reson Med. New York, USA, 1996, p 1086
27. Harkema SD, Meyer RA: Effect of acidosis on control of respiration in
skeletal muscle. Am J Physiol 272: C491--C500, 1997
28. Connett RJ, Honig CA: Regulation of va, in red muscle: Do current
biochemical hypotheses fit in vivo data? Am J Physiol 256: R898--
R906, 1989

29. Kemp GJ, Radda GK: Quantitative interpretation of bioenergetic data
from 31p and IH magnetic resonance spectroscopic studies of skeletal
muscle: An analytical review. Magn Reson Quart 10: 43--63, 1994
30. Chance B, LaNoue K: Metabolic control in exercising skeletal
muscle. Am J Physiol 257: R288-R290, 1989
31. Saks VA, Vasil' eva E, Belikova Y0, Kuznetsov AV, Lyapina S, Petrova L,
Perov NA: Retarded diffusion of ADP in cardiomyocytes: Possible role
of mitochondrial outer membrane and creatine kinase in cellular regulation
of oxidative phosphorylation. Biochim Biophys Acta 1144: 134-148, 1993
32. Balaban R: Regulation of oxidative phosphorylation in the mammalian
cell. Am J Physiol258: C377--C389, 1990
33. FromAHL, Zimmer SD, Mishurski SP, Mohanakrishnan P, Ulstad VK,
Thoma WJ, Ugurbil K: Regulation of the oxidative phosphorylation
rate in the intact cell. Biochemistry 29: 3731-3743, 1990
34. Lewandowski ED: Nuclear magnetic resonance evaluation of metabolic
and respiratory support of work load in intact rabbit hearts. Circ Res
70: 576--582, 1992
35. Wan B, Doumen C, Duszynski J, Salama G, LaNoue KF:A method for
determining the electrical potential gradient across the mitochondrial
membrane in perfused rat hearts. Am J Physiol265: H445-H452, 1993
36. Doumen C, Wan B, Ondrejickova 0: The effect of 2, 3-butanedione
monoxime, verapamil and cardiac work on the mitochondrial membrane
potential in perfused rat hearts. Am J Physiol269: H515-H523, 1995
37. McCormack JG, Denton RM: The role of intramitochondrial Ca 2+
transport and matrix Ca2 + in signal transduction in mammalian tissues.
Biochim Biophys Acta 1018: 287-291,1990
38. Gunter TE, Gunter KE, Sheu S-S, Gavin CE: Mitochondrial calcium
transport: Physiological and pathological relevance. Am J Physiol267:
C313--C339,1994
39. Bittle JA, Balschi JA, Ingwall JS: Effects of norepinephrine infusion on
myocardial high energy phosphate content and turnover in the living
rat.J Clin Invest 79: 1852-1859, 1987
40. Giesin J, Kammermeier H: Relationship of phosphorylation potential
and oxygen consumption in isolated perfused rat hearts. J Mol Cell
Cardiol12: 891-907, 1980
41. Boehm EA, Clark JF, Radda GK: Metabolite utilization and com-
partmentation in porcine carotid artery: A study using ~-guanidino
proprionic acid. Am J Physiol268: C628--C635, 1995
42. Clark JF: The creatine kinase system in smooth muscle. Mol Cell
Biochem 1331134: 221-232,1994
43. Connett R: Analysis of metabolic control: New insights using a scaled
creatine model. Am J Physiol254: R949--R959, 1988
44. Veksler VI, Kuznetsov AV, Anflous K, Mateo P, van Deursen J, Wieringa
B, Ventura-Clapier R: Muscle creatine kinase-deficient mice. II. Cardiac
and skeletal muscles exhibit tissue-specific adaptation of the
mitochondrial function. J Bioi Chern 270: 19921-19929, 1995
45. van Deursen J, Ruitenbeek W, Heerschap A, Jap P, TerLaak H,
BeWieringa A: Creatine kinase (CK) in skeletal muscle energy
metabolism: A study of mouse mutants with graded reduction in muscle
CKexpression. Proc NatiAcad Sci USA 91: 9091-9095, 1994
46. Hamman BL, Bittl JA, Jacobus WE, Allen PD, Spencer RS, Tian R, Ingwall
JS: Inhibition of the creatine kinase reaction decreases the contractile
reserve of isolated rat hearts. Am J Physiol269: HI 030--H 1036, 1995
47. Wyss M, Smeitink J, Wevers RA, Walliman T: Mitochondrial creatine
kinase: a key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119--166, 1992
48. Ventura-Clapier R, Veksler V, Hoerter JA: Myofibrillar creatine kinase
and cardiac contraction. Mol Cell Biochem 133/134: 125-144,1994
49. Walliman T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands: The
'phosphocreatine circuit' for cellular energy homeostasis. Biochem J
281: 21-40,1992
287
50. Kemp GJ, Manners DN, Clark JF, Bastin ME, Radda GK: A
theoretical model of some spatial and temporal aspects of the
mitochondrion/creatine kinase/myofibril system in muscle. J
Molec Med 73: B43, 1995
51. Kemp GJ, Manners DN, Clark JF, Bastin ME, Radda GK: The regulation
of oxidative metabolism in exercising skeletal muscle: A theoretical
model of 31 p MRS-observable responses of the mitochondrion/creatine
kinase/myofibril system. In: Proc 4th Ann Meeting Internat Soc Magn
Reson Med. New York, USA, 1996,p 1060
52. Meyer RA, Sweeney HL, Kushmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol246: C365--C377, 1984
53. McFarland EW, Kushmerick MJ, Moerland TS: Activity of creatine
kinase in a contracting mammalian muscle of uniform fibre type.
BiophysJ67: 1912-1924,1994
54. Zahler R, Ingwall JS: Estimation of heart mitochondrial creatine kinase
flux using magnetization transfer NMR spectroscopy. Am J Physiol
262: H 1022-HI 028, 1992
55. Fisher MJ, Dillon PF: Direct determination of ADP in hypoxic porcine
carotid artery usingllPNMR. NMR in Biomed 1: 121-126, 1988
56. Clark JF, Khuchua Z, Ventura-Clapier R: Creatine kinase binding and
possible role in chemically skinned guinea-pig taenia coli. Biochim
BiophysActall00: 137-145, 1992
57. Veech RL, Lawson JWR, Cornell NW, Krebs HA: Cytosolic phos-
phorylation potential. J BioI Chern 254: 6538--6547, 1979
58. Klingenberg M: TheADP-ATP carrier in mitochondrial membranes. In:
AN Martonosi (ed). The Enzymes of Biological Membranes. Plenum
Press, New York, 1976, pp 383-438
59. Jacobus WE: Respiratory control and the integration of heart high-
energy phosphate metabolism by mitochondrial creatine kinase. Ann
Rev Physiol47: 707-725, 1985
60. Kemp GJ, Thompson CH, Taylor OJ, Hands LJ, Rajagopalan B, Radda
GK: Quantitative analysis by lip MRS of abnormal mitochondrial
oxidative in skeletal muscle during recovery from exercise. NMR in
Biomed 6: 302-310, 1993
61. Stucki JW: The optimal efficiency and the economic degrees of coupling
of oxidative phosphorylation. Eur J Biochem 109: 269--283, 1980
62. Daut J: The living cell as an energy-transducing machine. Biochim
Biophys Acta 895: 41-62,1987
63. Holian A, Owen CS, Wilson DF: Control of respiration in isolated
mitochondria: Quantitative evaluation ofthe dependence of respiratory
rates on [ATP], [ADP] and [Pi]. Arch Biochem Biophys 181: 164-171,
1977
64. Sweeney HL: The importance of the creatine kinase reaction:
The concept of metabolic capacitance. Med Sci Sports Exerc 26:
30-36, 1994
65. WesterhoffHV, van Dam K: Thermodynamics and Control of Biological
Free-Energy Transduction. Elsevier, Amsterdam, 1987
66. Wanders RAJ, Westerhoff HV: Sigmoidal relation between mito-
chondrial respiration and 10g([ATP]/[ADP])oo' under conditions of
extramitochondrial ATP utilization. Implications for the control and
thermodynamics of oxidative phosphorylation. Biochemistry 27: 7832-
7840,1988
67. LaNoue KF, Jeffries FMH, Radda GK: Kinetic control of mitochondrial
ATPsynthesis. Biochemistry 25: 7667-7675,1986
68. Healy M: Principles of Automatic Control. Hodder and Stoughton,
London, 1975
69. Styles P, Kemp GJ, Radda GK: A model for metabolic control which
reproduces the main features of recovery from exercise which are
observed by 11 P MRS. In: Proc II th Ann Meeting Soc Magn Res Med.
Berlin, Germany, 1992, p 322
70. Styles P, Kemp GJ, Radda GK: Modelling lip MRS observations in
patients with mitochondrial dysfunction. In: 12th Ann Meeting
Soc Mag Res Med. New York, USA, 1993, P 1130
288
71. Fedosov SN: Creatine-creatine phosphate shuttle modelled as two-
compartment system at different levels of creatine kinase activity.
Biochim Biophys Acta 1208: 238-246, 1994
72. Jacobs HK, Kuby SA: Studies on adenosine triphosphate trans-
phosphorylases. IX. Kinetic properties of the crystalline adenosine
triphosphate-creatine transphosphorylase from calfbrain. J Bioi Chern
245:3305-3314,1970
73. Jacobs HK, Kuby SA: Studies on muscular dystrophy. A comparison of
the steady state kinetics of the nonnal human ATP-creatine trans-
phosphorylase isoenzymes (creatine kinases) with those from tissues of
Duchenne muscular dystrophy. J Bioi Chern 255: 8477-8482, 1980
74. Schimerlik ML, Cleland WW: Inhibition of creatine kinase by
chromium nucleotides. J Bioi Chern 248: 8418-8423, 1973
75. Clark J, Kemp GJ, Radda GK: The creatine kinase equilibrium, free [ADP]
and myosin ATPase in vascular smooth muscle cross-bridges. J Theor
Bioi 173: 207-211,1995
76. Aliev MK, Saks VA: Quantitative analysis of the 'phosphocreatine
shuttle': 1. A probability approach to the description of phosphocreatine
production in the coupled creatine kinase-ATPI ADP translocase-
oxidative phosphorylation reactions in heart mitochondria. Biochim
BiophysActa 1143: 291-300,1993
77. Clark JF, Kuznetsov AY, Khuchua Z, Veksler Y, Ventura-Clapier R, Saks
V: Creatine kinase function in mitochondria isolated from gravid and
non-gravid guinea-pig uteri. FEBS Lett 347: 147-151,1994
78. Saks VA, Chernousova GB, Gukovsky DE, Smirnov VN, Chazov EI:
Studies of energy transport in heart cells: Mitochondrial isoenzyme of
creatine phosphokinase: Kinetic properties and regulatory action of
Mg2+ ions. Eur J Biochem 57: 273-290,1975
79. Saks VA, Lipina NY, Smirnov VN: Studies of energy transport in heart
cells. The functional coupling between mitochondrial creatine
phosphokinase and ATP-ADP translocase: Kinetic evidence. Arch
BiochemBiophys 173: 34-41, 1976
80. Saks VA, Kupriyanov VY, Elizarova GV, Jacobus WE: Studies of energy
transport in heart cells. The importance of creatine kinase localization
for the coupling of mitochondrial phosphoryJcreatine production to
oxidative phosphorylation. J Bioi Chern 255: 755-763, 1980
81. Saks VA, Khuchua ZA, Vasilyeva EV, YuBelikova 0, Kuznetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Mol Cell Biochem 133/134: 155-192, 1994
82. Jacobus WE, Saks VA: Creatine kinase of heart mitochondria:
Changes in its kinetic properties induced by coupling to oxidative
phosphorylation. Arch Biochem Biophys 219: 167-178, 1982
83. Clark JF, Khuchua Z, Ventura-Clapier R: The compartmentation and
role of creatine kinase in the myometrium of gestating guinea pigs. J
Physiol466: 553-572, 1993
84. Shoubridge E, Radda G: A 31P-nuclear magnetic resonance study of
skeletal muscle metabolism in rats depleted of creatine with the
analogue ~-guanidinopropionic acid. Biochim Biophys Acta 805: 79-
88,1984
85. Le Rumeur E, LeMoyec L, DeCertaines JD: Creatine kinase activity in
rat skeletal muscle with intennittent tetanic stimulation. Magn Reson
Med24: 335-342, 1992
86. Ugurbil KM, PeteinR, Maidan S, Michurski S, FromAHL: Measurement
of an individual rate constant in the presence of multiple exchanges:
Application to myocardial creatine kinase reaction. Biochemistry 25:
100-107, 1986
87. Yoshizaki K, Watari H, Radda GK: Role of phosphocreatine in energy
transport in skeletal muscle of bullfrog studied by 3IP-NMR. Biochim
BiophysActa 1051: 144--150,1990
88. Walliman T, Hemmer W: Creatine kinase in non-muscle tissues and
cells. Mol Cell Biochem 133/134: 193-220, 1994
89. Gellerich FN, Kapischke M, Kunz W, Neumann W, Kuznetsov A,
Brdiczka D, Nicolay K: The influence of the cytosolic oncotic pressure
on the permeability of the mitochondrial outer membrane for ADP:
Implications for the kinetic properties of mitochondrial creatine kinase
and for ADP channelling into the intermembrane space. Mol Cell
Biochem 133/134: 85-104,1994
90. Gellerich FN, Zierz S, Laterveer FD, Nicolay K: Effect of colloidosmotic
pressure on ADP-transport into mitochondria: Experimental detennination
of ADP concentration gradients across the mitochondrial outer
membrane. In: Proc 7th Internat Meeting BioThennoKinetics. Louvain-
la-Neuve, Belgium, 1996, pp 295-299
91. Aliev MK, Saks VA: Mathematical modelling of intracellular transport
processes and the creatine kinase systems: A probability approach.
Mol Cell Biochem 1331l34: 333-346, 1994
92. Wiseman RW, Ieneson JAL, Kushmerick MJ: Why is the sensitivity of
mitochondria to [ADP] in vivo over 10 fold lower in penneabilized
fibres than in vivo? In: Proc 7th Internat Meeting BioThennoKinetics.
Louvain-Ia-Neuve, Belgium, 1996, pp 124--127
93. Veksler Y, Ventura-Clapier R: Ischaemic metabolic factors - high
inorganic phosphate and acidosis modulate mitochondrial creatine
kinase functional activity in skinned cardiac fibres. J Mol Cell Cardiol
26: 335-339, 1994
94. Khuchua ZA, Ventura-Clapier R, Kuznetsov AY, Grishin MN, Saks VA:
Alterations in the creatine kinase system in the myocardium of
cardiomyopathic hamsters. Biochem Biophys Res Commun 165: 748-
757,1989
95. Saks VA, Rosenshtraukh LV, SmirnovVN, Chasov EI: Role of creatine
phosphokinase in cellular function and metabolism. Can J Physiol
PhannacoI56:691-706,1978
96. Walliman T, Turner DC, Eppenberger HM: Localization of creatine
kinase isoenzymes in myofibrils. 1. I Cell Bioi 75: 318--325, 1977
97. Walliman T, Eppenberger HM: Localization and function of M-line
bound creatine kinase. Cell Musc Motil6: 239-285, 1985
98. Ventura-Clapier R, Saks VA, Vassort G, Lauer C, Elizarova G: Reversible
MM-creatine kinase binding to cardiac myofibrils. Am I Physiol253:
C444-C455,1987
99. Saks VA, Ventura-Clapier R, Huchua ZA, Preobrazhensky AN, Emelin
IV: Creatine kinase in regulation of heart function and metabolism 1.
Further evidence for compartmentation of adenine nucleotides in
cardiac myofibrillar and sarcolemmal coupled ATPase-creatine kinase
systems. Biochim Biophys Acta 1984
100. Bessman SP, Yang WC, Geiger PI, Erickson-Viitanen S: Intimate
coupling of creatine phosphokinase and myofibrillar adenosine
triphosphatase. Biochem Biophys Res Commun 96: 1414--1420, 1980
101. Kreyszig E: Advanced Engineering Mathematics. Wiley, New York,
1988
102. Chance B, Connelly CM: A method for the estimation of the increase
in the concentration of adenosine diphosphate in muscle sarcosomes
following a contraction. Nature 179: 1235-1237, 1957
103. Coty WA, Pedersen PL: Phosphate transport in rat liver mitochondria.
Kinetics and energy requirements. J Bioi Chern 249: 2593-2598, 1974
104. Chase PB, Kushmerick MI: Effect of physiological ADP concentrations
on contraction of single skinned fibers from rabbit fast and slow
muscles. Am] Physiol268: C48O-C489, 1995
105. Hubley MJ, Locke BR, Moerland TS: Temperature effects on diffusion
of phosphorous metabolites in skeletal muscle. In: Proc 7th Internat
Meeting BioThennoKinetics. Louvain-Ia-Neuve, Belgium, 1996, pp 300-
304
106. Roman BP, Koretsky AP: The consequence of switching creatine kinase
isofonns in mouse skeletal muscle reveals that BB-CK can replace
MM-CK. In: Proc 4th Ann Meeting Internat Soc Magn Reson Med,
New York, USA, 1996, P 1076
107. Wanders RIA, Groen AK, van Roemund COOT, Tager 1M: Factors
determining the relative contribution of the adenine-nucleotide
translocator and the ADP-regenerating system to the control of
oxidative phosphorylation in isolated rat-liver mitochondria. Eur I
Biochem 147:417-424, 1984

108. Haffier RP, Brown GC, Brand M:Analysis of the control of respiration
rate, phosphorylation rate, proton leak rate and protonmotive force in
isolated mitochondria using the 'top-down' approach of metabolic
control theory. Eur JBiochem 188: 313-319,1990
109. Rolfe DFS, Brand MD: Proton leak and control of oxidative phosphorylation
on perfused, resting rat skeletal muscle. Biochim Biophys Acta
1276: 45-50, 1996
110. Jeneson JAL, Kushmerick MJ: On the control of respiration in striated
muscle. In: Proc 7th Internat Meeting BioThermoKinetics. Louvain-
la-Neuve, Belgium, 1996, pp 158-162
III. Stucki JW: The thermodynamic-buffer enzymes. Biochim Biophys Acta
109: 257-267, 1980
112. Stucki JW: Non-equilibrium thermodynamic sensitivity of oxidative
phosphorylation. Proc Roy Soc Lond B244: 197-202, 1991
113. van der Meer R, WesterhoffHY, van Dam K: Linear relation between
rate and thermodynamic force in enzyme-catalyzed reactions. Biochim
Biophys Acta 591: 488--493, 1980
114. Gnaiger E: Mitochondrial respiratory control: Energetics, kinetics and
efficiency. In: Proc 10th Conference Eur Soc Comp Physiol Biochem. Georg
Thieme Verlag Stuttgart-New York, Innsbruck,Austria, 1989, pp 6-17
115. Roth K, Weiner MW: Determination of cytosolic ADP and AMP
concentrations and the free energy ofATP hydrolysis in human muscle
and brain tissues with 31p NMR spectroscopy [see correction in Magn
Reson Med (1995), Volume 33, p. 282)] MagnReson Med22: 505-511,
1991
116. Savimbi F: Interaction of creatine kinase and adenylate kinase in muscle
cells. Mol Cell Biochem 133/134: 145-152,1994
117. Zeleznikar RJ, Dzeja PP, Goldberg ND: Adenylate kinase-catalyzed
phosphoryl transfer couple ATP utilization with its generation by
glycolysis in intact muscle. J Bioi Chern 270: 7311-7319, 1995
118. Gupta RK, Moore RD: lip NMR studies of intracellular free Mg2+ in
intact frog skeletal muscle. J Bioi Chern 255: 3987-3993,1980
119. Masuda T, Dobson GP, Veech RL: The Gibbs-Donnan near-equilibrium
system of heart. J Bioi Chern 256: 20321-20334,1990
120. Garfinkel L, Garfinkel D: Calculation of free-M g2+ concentration in
adenosine 5'-triphosphate containing solutions in vitro and in vivo.
Biochemistry 23: 3547-3552, 1984
121. Dillon PF, Clark JF: The theory of diazymes and functional coupling of
pyruvate kinase and creatine kinase. J Theor Bioi 143: 275-284,1990
122. Wiseman RW, Kushmerick MJ: Creatine kinase equilibration follows
solution thermodynamics in skeletal muscle. lip NMR studies using
creatine analogs. J Bioi Chern 270: 12428-38,1995
123. Kupriyanov VV, Lyulina NV, Steinschneider A, Zueva M, Saks VA:
Creatine kinase-catalyzedATP-phosphocreatine exchange: Comparison
of 31P-NMR saturation transfer technique and radioisotope tracer
methods. FEBS Lett 208: 89--93, 1986
289
124. van Dorsten FA, Reese T, van Echteld CJA, NederhoffMGF, Nicolay
K: In vivo fluxes through mitochondrial and cytoplasmic creatine
kinase. P-31 NMR of skeletal and cardiac muscle from transgenic mice.
In: Proc 4th Ann Meeting Internat Soc Magn Reson Med. New York,
USA, 1996, P 24
125. Harris RC, Soderlund K, Hultman E: Elevation of creatine in resting
and exercised muscle of normal subjects by creatine supplementation.
C1in Sci 83: 367-374, 1992
126. Clark JF, Khuchua Z, Kuznetsov AV, Vassil'eva E, Boehm EA, Radda
GK, Saks VA: Actions of the creatine analogue ~-guanidinoproprionic
acid on rat heart mitochondria. Biochem J 300: 211-216, 1994
127. Harris DA, Das AM: Control of mitochondrial ATP synthesis in the
heart. Biochem J 280: 561-573, 1991
128. Masiakos PT, Williams GD, Berkich DA, Smith MB, LaNoue KF: 31p
NMR saturation-transfer study of the in situ kinetics of the mito-
chondrial adenine nucleotide translocase. Biochemistry 30: 8351-8357,
1991
129.Jeneson JAL, Wiseman RW, Westerhoff HV, Kushmerick MJ: The
signal transduction function for oxidative phosphorylation is at least
second order in ADP. J Bioi Chern 271: 27995-27998, 1996
130.Sluse-Goffart CM, Evens A, Duyckaerts C, Sluse FE: Initial rate study
of ['4C]ADP uptake in ADP loaded mitoplasts and the kinetic
mechanism of the ADP/ATP carrier. In: Proc 7th Internat Meeting
BioThermoKinetics. Louvain-Ia-Neuve, Belgium, 1996, pp 218-224
131. Scholte HR: On the triple localization of creatine kinase in heart and
muscle cells. Biochim Biophys Acta 305: 413--427, 1973
132.Saks VA, Chernousova GB, Voronkov YI, Smimov VN, Chasov EI:
Study of energy transport in myocardial cells. Circ Res 34/35 suppl III:
139-149, 1974
133. Saks VA, Seppet EK, Lyulina NV: Comparative investigation of the role
of creatine phosphokinase isoenzymes in energy metabolism of skeletal
muscle and myocardium. Biokhimiya42: 579--588, 1977
134. Kemp GJ, Hands LJ, Ramaswami G, Taylor DJ, NicolaidesA, Radda
GK: Calf muscle mitochondrial and glycogenolytic ATP synthesis in
patients with claudication due to peripheral vascular disease analysed
using lip magnetic resonance spectroscopy. Clin Sci 89: 581-590,
1995
135. Bangsbo J, Graham T, Johansen L, Strange S, Christensen C, Saltin B:
Elevated muscle acidity and energy production during exhaustive
exercise in humans. AmJ Physiol263: R891--899, 1992
136. Harris RC, Hultman E, Kaijser L, Nordesjo L-O: The effect of circulatory
occlusion on isometric exercise capacity and energy metabolism of the
quadriceps muscle in man. Scand J Lab Clin Invest 35: 87-95, 1975
137. Watts DC: Creatine kinase (adenosine 5' -triphosphate-creatine phos-
photransferase). In: PD Boyer (ed). The Enzymes. Academic Press,
New York, 1967
Molecular and Cellular Biochemistry 184: 291-307, 1998.
1998 Kluwer Academic Publishers.

Quantitative studies of enzyme-substrate

compartmentation, functional coupling and
metabolic channelling in muscle cells

Valdur Saks, 1,2 Pierre Dos Santos,3 Frank N. Gellerich4 and Philippe
Diolez5
lLaboratories of Bioenergetics, Joseph Fourier University, Grenoble, France; 2Institute of Chemical and Biological
Physics, Tallin, Estonia; 3INSERM Unite 441 - Atherosclerose: Determinisme Cellula ire et Moleculaire, Consequences
Functionnelles, Bordeaux, France; 4Muskel Laboratory, Martin Luther University, Halle, Germany; 5Resonance
Magnetique Nucleaire des Systemes Biologiques, UMR 5536 CNRS/Universite Bordeaux II, Bordeaux, France

Diversity of our opinions does not proceed from some men being more rational
than others, but solely from the fact that our thoughts pass through
divers channels and the same objects are not considered by all.
Rene Descartes, 1596--1650
Discours de la methode

Abstract
Some historical aspects of development of the concepts of functional coupling, metabolic channelling, compartmentation and
energy transfer networks are reviewed. Different quantitative approaches, including kinetic and mathematical modeling of energy
metabolism, intracellular energy transfer and metabolic regulation of energy production and fluxes in the cells in vivo are analyzed.
As an example of the system with metabolic channelling, thermodynamic aspects of the functioning the mitochondrial creatine
kinase functionally coupled to the oxidative phosphorylation are considered. The internal thermodynamics of the mitochondrial
creatine kinase reaction is similar to that for other isoenzymes of creatine kinase, and the oxidative phosphorylation process
specifically influences steps of association and dissociation of MgATP with the enzyme due to channelling of ATP from adenine
nucleotide translocase. A new paradigm of muscle bioenergetics - the paradigm of energy transfer and feedback signaling networks
based on analysis of compartmentation phenomena and structural and functional interactions in the cell is described. Analysis
of the results of mathematical modeling of the compartmentalized energy transfer leads to conclusion that both calcium and
ADP, which concentration changes synchronously in contraction cycle, may simultaneously activate oxidative phosphorylation
in the muscle cells in vivo. The importance of the phosphocreatine circuit among other pathways of intracellular energy transfer
network is discussed on the basis of the recent data published in the literature, with some experimental demonstration. The results
of studies of perfused rat hearts with completely inhibited creatine kinase show significantly decreased work capacity and
respectively, energy fluxes, in these hearts in spite of significant activation of adenyl ate kinase system (Dzeja et al. this volume).
These results, combined with those of mathematical analysis ofthe energy metabolism of hearts oftransgenic mice with switched
off creatine kinase isoenzymes confirm the importance of phosphocreatine pathway for energy transfer for cell function and
energetics in mature heart and many other types of cells, as one of major parts of intracellular energy transfer network and
metabolic regulation. (Mol Cell Biochem 184: 291-307, 1998)

Key words: creatine kinase, mitochondria, respiration, contraction, regulation, thermodynamics, compartmentation, functional
coupling, metabolic channelling

Address/or offprints: V.A. Saks, Laboratory of Bioenergetics, Joseph Fourier University, BP 53X - 38041 Grenoble Cedex, France
292
Introduction
The best way to consider any problem in a general manner
is to start with its history. The reason what makes this often
necessary is an interesting phenomenon which many of us
may have observed after some decades of active scientific
research. The observation is that very often scientific work
is designed, performed with good results, sent for pub-
lication, accepted, published - and after some time seems
to be totally forgotten. And then, usually after some 10-15
years, similar work with similar results is made in another
laboratory by other people, who come to these ideas
independently, and again published. We may call this
phenomenon 'resurrection of results and ideas'. Any
scientist above forty can find abundant evidence for this
phenomenon from its own practice. We give here our own
examples of this experience.
In 1970 S. Nagle described facilitated diffusion of high
energy phosphates in the organized cytoplasm of muscle
cells [I]. In accordance with 'the law of resurrection', Meyer
et al. gave independently, in 1984, a quantitative description
of this idea [2], and later Dzeja, Goldberg (see chapter II in
this volume) and we used it [3] as the basis of hypothesis
of metabolic oscillations - a feedback signal between
contraction and respiration.
Second example. In 1978 Sjostrand gave an excellent
description of structural differences of mitochondria in
vivo and in vitro [4] - and only now these ideas are beginning
to be used for explaining the experimental data (see the
chapter lOin this volume).
All scientists involved in the creatine kinase research may
also confess that we have collectively ignored the paper by
Klingengberg et al. [5], one of the first descriptions of the
idea of the phosphocreatine pathway for energy transfer in
the muscle cells. One of the reasons of this may be that not
many of us were strong in German language in which this
paper was written.
The kinetics of the creatine kinase enzymes in all possible
versions was studied in many laboratories, including ours,
in seventies and early eighties [6, 7]. And now our colleagues
from The Netherlands have successfully performed the same
experiments with the same results [8], in good accordance
with the 'low of resurrection'.
When reading very interesting chapters of the recent book
'Channelling in intermediary metabolism' edited by Agius and
Sherratt (Portland Press, London and Miami 1997), an
experienced man in the area of research recalls with
appreciation the paper by Sols and Marco 'Concentrations of
metabolites and binding sites. Implications in metabolic
regulation.' Published in 1970 [9], who already then discussed
the significance and meaning of macromolecular crowding,
intracellular concentrations of enzymes and metabolites,
thermodynamic activities, and compartmentation. Now
these ideas are finding increasingly rapid and important
development.
Probably, most important example: in 1979 P. Mitchell
gave a Sir Hans Krebs Lecture in Dresden at the FEBS
Meeting describing in details the idea of vectorial ligand
transduction [10]. It took again about 10-15 years (see the
chapter 11 in this volume and ref. [3]) to recognize its value
and to start to use this profound concept of bioenergetics,
based on the recognition of the importance of cellular
organization, for explanation of one particular problem - of
the nature of intracellular feedback signal transduction for
regulation of respiration.
Sometimes funny things happen: some good results pub-
lished long time ago are excavated from journals and used,
without critical analysis, to explain new data, forgetting,
probably because of the lack of time, everything that happened
between, within decades of scientific research [11 ] (this is in
more details analyzed in chapter 6 of this volume).
And these are only few examples of the phenomenon of
'resurrection of ideas and results'.
In accordance with this general observation, the aim of
this paper is to accelerate resurrection of some results and
ideas published earlier in the field of the creatine kinase
research and in muscle cell bioenergetics, to give some
historical overview of the research in the field of metabolic
compartmentation to assist new generation of researchers
and also to analyze some important new quantitative
approaches, including mathematical modeling, in studies
of energy transfer networks in the cells. And finally we have
to explain why the studies of the energy transfer networks
including creatine kinase systems are becoming increas-
ingly important. For that, we analyze the function of the
heart in which the creatine kinase system has been totally
inhibited.
Internal thermodynamics of mitochondrial creatine
kinase, its functional coupling to the adenine
nucleotide translocase and oxidative phosphorylation
In the chapters 9 and 10 of this volume by Schlattner et al.
and by Stachoviak et al. from Theo Wallimanns laboratory
in Zuerich, the authors give fine analysis of the structure of
the mitochondrial creatine kinase and discuss possible ways
of its functional coupling with the adenine nucleotide
translocase. The mitochondrial isoenzyme was the first of
creatine kinases for which the molecular and spatial (X-
ray) structure was solved [12]. In connection with these
achievements in structural analysis of creatine kinase, we
recall our earlier studies on the kinetics and thermodynamics
of the coupled mitochondrial creatine kinase - they are very
complementary to the structural data from Wallimann's
laboratory and may be useful in further studies of the

molecular physiology of mitochondrial creatine kinase and
energy channelling from mitochondria into cytoplasm.
We used the isolated rat heart mitochondria as a source of
the bound mitochondrial creatine kinase [13]. The creatine
kinase reaction rates were determined in the sucrose medium
[14]. In the absence of oxidative phosphorylation, the reaction
rate in the forward direction (phosphocreatine and ADP
production) was measured by using the pyruvate kinase-lactate
dehydrogenase coupled enzyme assay, and in the reverse
direction (ATP production) by using the hexokinase - glucose-
6-phosphate dehydrogenase coupled enzyme assay [14].
Under respiring conditions, the steady state rate of the
forward creatine kinase reaction was calculated from the
rates of oxygen consumption in the presence of creatine as
described in details by Jacobus and Saks [14].
From these data, the dissociation constants ofthe enzyme-
substrate complexes were calculated [14]. The creatine
kinase reaction mechanism is known to be the random
binding quasi-equilibrium Bi-Bi type, according to Clelands
classification (Scheme 1) [6, 7]. The reaction mechanism
of the creatine kinase and the dissociation constants for
each step are shown in Scheme 1. The methods of these
calculations are described earlier [14].
Since dissociation of the creatine kinase enzyme-substrate
complexes represent quasi-equilibrium processes, they
obey, in general, the rules of the classical thermodynamics:
dGO = -2.303 RT log Kd
and:
log Kd =-dHo/2.303 RT + dso/2.303R
The reaction rate determinations were made in the tem-
perature range of 19-38C. Figures 1 and 2 show the results
of the complete kinetic analysis of the mitochondrial
+ Cr E.Cr
;/
E
" E.MgATP.Cr
/
+ MgATP E.MgATP
Scheme I. The schematic presentation of the mechanism of the creatine kinase reaction.
293
creatine kinase in the membrane-bound state in the absence
and in the presence of oxidative phosphorylation. Then
equations (1) and (2) were used to construct the free energy
profile of the creatine kinase reaction and for general
presentation of the data.
Figure lA shows the experimental results for the forward
reaction (phosphocreatine production). The oxidative
phosphorylation process does not influence the binding and
dissociation of the creatine: its dissociation constants from
the binary complex E.Cr, Kib, and from the ternary complex
E.MgATP.Cr, Kb are the same in the presence and in the
absence of oxidative phosphorylation at any temperature.
However, the situation with the second substrate, MgATP, is
very different. Depending on the temperature, the value of
Kia, dissociation constant for the binary complex E.MgATP,
is decreased two or three times under conditions of oxidative
phosphorylation. However, Ka, the dissociation constant of
MgATP from the ternary complex E.MgATP.Cr is changed
by order of magnitude. The Vmax values are not changed by
the oxidative phosphorylation (Fig. IB). Thus, the influence
of the oxidative phosphorylation on the creatine kinase
reaction concerns only the effective concentration of the
common intermediate, MgATP. Table 1 gives the empirical
equations for temperature dependencies of all of the constants
of dissociation of the enzyme-substrate complexes for the
forward creatine kinase reaction in the absence and in the
presence of oxidative phosphorylation. These equations may
(1) be used in future modelling of the coupling phenomena at
any temperature in the given range (see chapters by Aliev et
(2) al. and Kemp et al. in this volume).
Finally, Fig. 2 shows the temperature dependencies of the
kinetic constants of the reverse mitochondrial creatine
kinase reaction (this reaction was not studied in the presence
of oxidative phosphorylation because of the experimental
E.PCr + PCr
-
kl
... E.MgADP.PCr E
k_l
E.MgADP + MgADP
294

19V1

0.5

1
'*~ <> 19Kib (-)

:g.
.Q

r-l
~++~
't~ 10K ", (+)
-r 19K b (-)
~ 0.0
:g. 19K b (+)
r-l 0
,
os
:.::
01
r-l

os
.>i -0.5
:.::01
-1
r-l
19Ka (-)

3.2 3.3 3.4

-2 19Ka (+) I/T x 10 3

3.2 3.3 3.4 Fig. lB. The temperature dependencies of the maximal reaction rate ofthe
forward mitochondrial creatine kinase reaction; (-): in the absence of oxidative
liT x 10 3 phosphorylation; (+): in the presence of oxidative phosphorylation.
(Reproduced from ref. [13] with permission).
Fig. 1A. The temperature dependencies of the kinetic constants for the
forward creatine kinase reaction, catalyzed by the membrane-bound rat heart
optimized enzymes, which, according to Albery and Knowles,
mitochondrial creatine kinase. The data for dissociation constants of
substrates in the absence of oxidative phosphorylation (-) and in the have reached evolutionary perfection [17, 18]. The standard free
presence of oxidative phosphorylation (+) are shown energy change of the creatine kinase reaction (' external thermo-
dynamics ') calculated by this method is 10.5 kllmole and close
to other determinations [19].
difficulties concerning very high apparent affinity of the
mitochondrial oxidative phosphorylation for ADP). The
Table 1. Thermodynamic parameters of the forward and reverse
empirical equations for these constants are given in Table 1.
mitochondrial creatine kinase reactions in the membrane bound state of
All data given in Table 1 were used to calculate the free enzyme in the absence of oxidative phosphorylation
energy profile for the membrane-bound mitochondrial creatine
kinase reaction in the absence of oxidative phosphorylation. Kinetic constant Empirical equation ll.GO
This free energy profile gives the internal thermodynamics and oflgK = f (lIT) from kJ x mole ,
the reaction linear regression of data 25C
of the mitochondrial creatine kinase reaction and is shown
in Fig. 3. The difference between free energy levels of E + I Forward reaction, Cr + MgATP -7 PCr + MgADP + H+
PCr + MgADP and E + Cr + MgATP gives the standard free Kia 3.13-l0l9x liT 1.65
Ka 6.70-2241 x liT 4.7
energy change of the creatine kinase reaction.
Kib 11.5 - 3066 x liT -6.9
There are several remarkable peculiarities of this thermo- Kb 13.2 -3739x liT -3.7
dynamic profile. First, the activation energies for the Vrmx l4.62-4357x liT Ea= 83.42*
forward and reverse creatine kinase reactions calculated
from the Arrhenius equation are equal to each other (Ea = 82- II Reverse reaction, PCr + MgADP + W -7 MgATP + Cr
Kic 4.04-l569x liT 6.8
83 kllmole) and that conforms to the absence offree energy
Kc 6.6 -2407x liT 8.5
change at the step of phosphoryl transfer earlier discovered Kid 3.09-909x liT -0.24
by the NMR method [15]. The free energy profile shows that Kd 5.9 -1854x liT 1.83
the free energy changes occur at the steps of the substrate Vrmx l4.66-4288x liT Ea= 82.10*
binding and product dissociation. This is the common prop- *Ea - empirical activation energy determined according to the equation 5.
erty ofthe phosphoryl or phosphate transfer enzymes, including Other thermodynamic values were calculated from the data of linear
many kinases and myosinATPase [16, 17] acting as a kinetically regression according to the equations 1-3.

19V_ 1
19K
o chapters 9 and lOin this volume). This mechanism of
-1
-2
3.2 3.3 3.4
liT x 10 3
Fig. 2. The temperature dependencies of the kinetic constants for the reverse
creatine kinase reaction catalyzed by the membrane-bound rat heart
mitochondrial creatine kinase. The kinetic constants were detennined under
conditions of complete inhibition of mitochondrial oxidative phos-
phorylation. (Reproduced from ref. [13] with pennission).
Since in the presence of oxidative phosphorylation the
apparent kinetic constants of the mitochondrial creatine
kinase do not represent any more parameters of quasi-
equilibrium processes but are influenced by several coupled
processes and need mathematical models for their analysis
(see chapter by Aliev in this volume), it is not possible to
construct the free energy profile for this case. However,
the data shown in the Fig. 2 allow us to make some very
interesting conclusions and predictions concerning the
molecular physiology of the coupled mitochondrial creatine
kinase reaction. The very significant decrease in only one
of the dissociation constant, Ka, at the background of the
minor changes in the other one, Kia, means that formation
of the ternary complex E.Cr.MgATP from E.Cr is most
favored by direct channelling oftheATP molecules from the
adenine nucleotide translocase to the mitochondrial creatine
kinase under conditions of oxidative phosphorylation. When
ATP is channelled to the creatine kinase before binding of
creatine (a process characterized by Kia), it may significantly
dissociate from the E.MgATP complex before the formation
of the productive ternary complex, E.Cr.MgATP. In general,
this kind of mechanism of functional coupling which
includes direct channelling of ATP from translocase to
creatine kinase and ADP back from creatine kinase to
295
translocase, is shown in Fig. 4. Quantitatively, this process has
been described by Aliev and Saks by a probability model of
the functional coupling of creatine kinase and adenine
nucleotide translocase in mitochondria [20]. And remarkably,
after 15--20 years of these functional works, the conclusions
made are now directly confirmed by structural studies (see
functional (and structural) coupling is a good example of
metabolic channelling phenomenon, and it is the basis of the
aerobic phosphocreatine production in mitochondria in many
cells in vivo, and probably a powerful amplifYing mechanism
for the feedback regulation of respiration by creatine and by
changes (oscillations) of ADP concentrations [3].
The creatine kinases, compartmentation phenomena,
intracellular energy transfer networks and the problem
of respiration regulation in vivo
In spite of very advanced knowledge of particular problems,
such as membrane bioenergetics in mitochondria, structure
and function of actomyosin complexes, ion channels and
pumps, all kind of membrane ATPases, different signal
transmission pathways (G-proteins etc.), the principles of
regulation of energy fluxes within the integrated cellular
systems in vivo are not yet well understood. There are very
numerous works and reviews on cellular regulation of energy
metabolism in the cells, but the general agreement is not as
yet achieved on the question in which way the feedback signal
is given to cellular systems of ATP production to respond
to increased cellular work-contraction, ion transport,
biosynthesis etc. If there is some general agreement
regarding this question, it is that in many cells the regulatory
signal is not simply the change in cytoplasmic ADP con-
centration since it does not correlate with alteration of the
respiration rate [23-30]. Intensive studies of the role of
calcium ions in this process also did not give as yet any
decisive answer (see the chapters by Hansford and Zorov and
Mazat et al. in this volume). In fact, calcium alone cannot
be the solution of the problem since the energy transfer and
feedback signaling are associated with phosphoryl or
phosphate group transfer, and the mass conservation low
requires that the same amount of ADP as released in
myofibrils should be made available for rephosphorylation
in the mitochondria under steady state conditions. That
means that the signaling involves the group transfer, or
macroosmotic process, according to P. Mitchel [10]. And
it is important to note that diffusion of ADP seems to have
several serious obstacles in the cell, such as multiple binding
sites, macromolecular crowding [31, 32], problems of the
controlled permeability through the outer mitochondrial porin
pores [3] etc. Bereiter-Hahn and Voth studied morphological
changes of mitochondria, which acquire condensed form in
296
E MgADP . .. P ... Cr
0"
t!J
<l
10
E.Cr
E
+
Cr E.MgATP
+
MgATP
Fig. 3. The free energy profile of the mitochondrial creatine kinase reaction catalyzed by the rat heart mitochondrial creatine kinase in the absence of oxidative
phosphorylation.
response to direct microinjection of ADP into the cells, and
showed that spreading of these mitochondrial responses
from tip of the injecting capillary to more peripheral zones
is in fact very slow indicating slow diffusion of ADP in
cytoplasm [33]. This observation is relevant to our data on
the increase of the apparent Km for ADP in regulation of
oxidative phosphorylation in the cells in vivo by an order
of magnitude, in comparison with isolated mitochondria, due
to decreased permeability of outer mitochondrial membrane
[3], and to the results by described Van Beek et al. in
chapter 20 of this volume, showing that the response of
mitochondrial oxygen uptake to rapid work changes is
delayed with respect to rapid change of the level of PCr.
Rapidly growing body of experimental evidence shows the
unexpected importance of the porin pores, or VDAC, of
mitochondrial outer membrane for regulation of metabolic
fluxes in the cells in vivo [3, 34- 41]. Marco Colombini's
laboratory has recently shown that this channel in 'closed'
configuration almost completely blocks ATP flux and also
E
E.PCr +
PCr
+
MgADP
E.MgADP.PCr
decreases permeability for di- and trivalent anions while
monovalent ions are still permeable [34, 35]. Some proteins
in cytoplasmic fraction have been found to increase the
voltage-dependence of this channel [36, 37], and we have
proposed that some of these proteins may be associated with
cytoskeleton (see ref. [3] and chapter by Rappaport and
Samuel in this volume). Also, there are evidences of possible
importance of the complexes of mitochondrial porin with
kinases and adenine nucleotide translocase for stimulation
of mitochondrial oxidative phosphorylation and regulation
of permeability transition pore in brain [38]. Because of the
controlled permeability of porin channels for ADP, the
mitochondrial kinases connected to the outer membrane,
outer surface of the inner membrane or in the intermembrane
space become key enzymes in energy distribution between
different networks and signal transmission for mitochondrial
responses to energetic demands [3, 38--42].
Therefore, one ofthe most important questions in cellular
bioenergetics in vivo, an answer to which is in fact totally
 297

~ATPm
T~
MgATP Cr
- - - -..
-- PCr

~ Kib :(a MgATPm MgADP

C{ ~ '--/
c~ MgATP ox. phosph.

Fig. 4. Ihe schematic explanation of the specific effect of oxidative phosphorylation on the kinetics of mitochondrial creatine kinase reaction. I-adenine
nucleotide translocase. MgA IP is directed by the translocase by the mechanism of direct channelling to the active side the mitochondrial creatine kinase. In
the absence of creatine some dissociation of MgAIP from the binary enzyme-substrate complex is possible. If MgAIP is directed to the enzyme already
saturated with the creatine molecules the ternary complex is converted into the enzyme-product complex with the consequent dissociation of phosphocreatine
and MgADP; the latter is taken back into the mitochondrial matrix by adenine nucleotide translocase for rephosphorylation in the mitochondrial oxidative
phosphorylation to start new cycle of the coupled reactions. High turnover of these coupled reactions results in high steady-state concentration of the
ternary complex E.MgAIP.creatine and this results in apparent decrease of the dissociation constant Ka value.

unknown, is the value of ADP diffusion coefficient in the cell
cytoplasm. There are good experimental data for diffusion
coefficients for ATP, per and Pi in bulk water phase of cyto-
plasm determined by isotope or NMR pulse gradient methods
[43,44], but even these data are too general and do not take
into account the complex, fine structural organization of the
cell and existence of multiple compartments and micro-
compartments, where the diffusion is significantly restricted
[3, 31]. Also, one should take into account the structure of
the intracellular water phases [45,46]. For example, Walter and
Brooks have described phase separation in cytoplasm due to
macromolecular crowding as the basis for formation of multiple
microcompartments in the cells, and concluded that the
interface between these aqueous phases presents a barrier
for diffusion of molecules [45]. Further complications arise for
diffusion of ADP (and of any other substrate) if the active
centers of enzymes are fixed in multienzyme complexes (for
more details on this topic see recent monograph by J. Ovaldi
[31] and in several recent review [3, 32]). For example, this kind
of fixation, probably with participation of the cytoskeletal
system has been shown for glycolytic complexes and sarco-
lemmal or other membrane ATPase and ATP-dependent
potassium channel in heart cells: exchange of ATP and ADP
within these complexes (micro compartments) without the
release of adenine nucleotides into cytoplasm [47, 48]. And
because of diffusion problems for ADP and complex cellular
structural organization, simple calculations of diffusion
distances and transit times often used as arguments in dis-
cussions on the basis of ADP diffusion constant in diluted
solutions may be taken only as entertaining exercise but not
too convincing solution of this serious problem (see also
the chapter 20 by Van Beek et al. in this volume).
Probably, there are numerous ways of transmission of the
feedback signal within the metabolic networks in the cell,
depending on the type of the cell considered, and on the
metabolic state of the given cell. For example, mechanism
of regulation of respiration in the muscle cells seems to be
very tissue specific [49]. For fast twitch skeletal muscle the
classical paradigm based on homogeneous equilibrium meta-
bolic system [50, 51] may be a reasonable approximation,
especially in resting state. For other types of cells, notably
for the working heart cells, and also for liver and brain cells
and probably many others it does not provide any reasonable
explanation of regulation of respiration [3, 23- 27]. In these
cells manifold changes in respiration rate is seen at minor
changes in metabolic state [3, 23-27, 30]. Many data are
emerging now showing that the nature of this signal is
influenced or even determined by the structural interactions
in the cell. This regulation most possibly includes functional
interaction, or functional coupling between structurally
related enzyme systems in the compartments and micro-
compartments of the cell, and finally, both calcium and
metabolic wave propagation between microcompartments.
Thus, the classical paradigm of muscle bioenergetics
based on the concept of homogenous metabolic systems,
which are in equilibrium, has to be changed and replaced by
the paradigm based on the concept of metabolic networks for
energy transfer and signal transduction, which basically
298
function in non-equilibrium, metastable steady state. This
concept is developed in many chapters in this volume (Saks
et al., Schlattneret al., Dzejaet al.,Aliev et al., Van Beeket al.,
Kholodenko et al., K Nicolay et al., Rappaport and Samuel,
Ventura-Clapier et al.). According to this concept, not only free
diffusion of metabolites but mostly enzymatic, vectorial group
transfer (ligand conduction), due to fine structural organ-
ization ofthe cell interior by cytoskeleton, with participation
of different enzyme systems (creatine kinase, adenylate kinase,
glycolytic systems etc.) carry out the energy transfer between
different cellular compartments and feedback regulation
processes in cellular bioenergetics. The contribution of diff-
erent pathways in the process of energy transfer (total energy
fluxes) may vary, depending of metabolic state or altered
expression of enzymes (transgenic animals).
The concept of metabolic networks of energy transfer is
directly related to the phenomenon of compartmentation
of enzymes and metabolites. Two lines of experimental
research in the history of muscle biochemistry and physiology
led to the ideas of compartmentation of adenine nucleotides
in the cell: studies of the role of so-called peripheral
kinases, as hexokinase, creatine kinase and adenylate kinase
in the cells (peripheral - because in the mitochondria they
are associated with the outer surface ofthe inner membrane
or to the outer membrane), and studies, or attempts, to
establish any connection, or correlation between heart
muscle contractile function and tissue high energy phosphate
contents, (more precisely, from discovery of the lack of this
correlation). The first line of ideas was pioneered by
Bessmann et al. [52], and the works of both Klingenberg [5]
and Nagle [1] should be mentioned in connection to this
group. The second direction started to actively develop after
two publications - those by Gerken and Schlette, and by
Gudbjamasonet al. [53,54]. This was the beginning. Further
developments in this area resulted in detailed description of
the biochemistry and physiological roles of the coupled
creatine kinases reviewed in our first collective work [55].
The ideas ofthe metabolic channelling started to develop
on more general basis in cellular biochemistry in the works
of Kelety, Sols and Marco, as already mentioned, Srere,
Kurganov and others, as reviewed in the ref. [3], and the
current state of studies of these problems is reviewed in the
recent monograph by J. Ovaldi and in that edited by Agius
and Sherratt [31,32]. In this volume, the reader can find many
references to earlier works and recent reviews in this area
in the chapter by Kholodenko et al. in this volume.
In a rather unexpected way, the concept of metabolic
compartmentation and metabolic channelling (functional
coupling) has recently found its most direct experimental
confirmation in studies of excitation-contraction coupling
and calcium metabolism due to serious advancements in the
experimental techniques, mostly due to the use of confocal
microscopy [56---66]. The use of this technique allowed to
show that the calcium transients thought to represent
elevation of calcium concentration homogeneously in
cytoplasm during excitation-contraction coupling, in fact
represent summation of calcium sparks - triggered or
spontaneous local increases of calcium concentration in
micro compartments between cellular structures [57--60].
Also, by using digital imaging microscopy, Isenberg et al.
directly demonstrated the existence of deep calcium
gradients and diffusion limitations within the sarcomere in
ventricular myocytes [63]. Local increase in calcium
concentration may propagate in cardiomyocytes as a wave
[60, 61, 62]. References to these works are too numerous
and too rapidly increasing to be given here exhaustively (see
the chapter by Mazat et al. in this volume). One interesting
work in this line was recently published by groups of
Lewartowsky and Langer, showing direct channelling of
calcium between sarcoplasmic reticulum and Na-Ca exchanger
in sarcolemma in the process of calcium extrusion from the
cells [66]. All that means that determination, or calculation
of average cytoplasmic concentration of calcium is not
enough to understand in quantitative terms the mechanisms
of excitation-contraction coupling. In other words, all
cellular processes of calcium metabolism and its regulation
are now considered in terms of microcompartmentation.
Some years ago, Carmeliet analyzed electrophysiological
data and concluded that also for sodium ions there is a
'fuzzy subsarcolemmal space' where the ion concentration
is different from the average cytoplasmic one and its
diffusion limited [67]. That means the existence of the
compartmentation phenomenon also for this small ion in
the cell.
And all that is exactly what we have been telling about
energy metabolism and adenine nucleotides in the cells for
more than 25 years - that calculation of average cytoplasmic
concentrations of ATP and ADP does not help too much to
understand the mechanisms of regulation of respiration and
cellular energetics, and that functionally important pools of
adenine nucleotides are different from average concentrations
in the cell [3, 52-55, 68-71]. This conclusion has been
directly confirmed by Lanoue's group for the heart [72] and
is consistent with many other experimental data which have
shown dissociation of muscle function from the total
content of ATP in the cells (see ref. [3] for discussion). To
solve the problem of respiration regulation quantitatively,
it is necessary to obtain some information of local ATP and
ADP concentrations in cellular compartments and micro-
compartments. We clearly need some experimental methods
of direct determination of local concentrations of adenine
nucleotides in the cellular microcompartments (analogues
to calcium spark detection). To the author's knowledge,
if this method exists, it is not openly described and not
widely used as yet. That makes it necessary to combine 31p_
NMR flux determination with isotopic and other methods

(see the chapters by Dzeja et al. and Klaas Nicolay et al. in
this volume) to collect quantitative information on the
metabolic fluxes in the cell, and inevitably all conclusions
of the role of compartmentalization of adenine nucleotides
are still mostly indirect. One of interesting evidences for
intracellular compartmentation of cyclic adenine nucleotides
has been recently reported by Jurevicius and Fischmeister:
by using whole-cell patch-clamp recording and a double
capillary for extracellular microperfusion, they studied
distant and local effects of drugs on L-type calcium current
and found higher cAMP concentration close to calcium
channel than in rest of the cell [73, 74].
Strong evidence for compartmentation phenomenon
comes from studies of the glycolytic pathway of free energy
production: the glycolytically produced ATP (,glycolytic
ATP') has been found to be more effective thanATP produced
in mitochondria, in prevention of ischemic contracture and
membrane damage [47, 48]. Also, the important consequence
of compartmentation of the adenine nucleotides in the cell
is that the total contents of high energy phosphates are
dissociated from the function [75, 76].
Awaiting for the future developments of experimental
methods of direct determination of local concentrations
of adenine nucleotides in the cells, it is helpful to analyze the
problem of metabolic regulation of respiration quantitatively,
by using kinetic and mathematical models and new achieve-
ments in the Metabolic Control Analysis. And it seems to
be a very good idea to combine methods of mathematical
modeling with some powerful experimental approach, such
as transgenic animal technology (see below and a chapters
by Klaas Nicolay et al. and Aliev et al. in this volume).
Mathematical models of the energy transfer network
between different cellular compartments
Mathematical modeling of the metabolic systems is not a
new approach. Quantitative studies in muscle biochemistry,
by using mathematical models, were pioneered some 25-30
years ago by David Garfinkel in Philadelphia, USA. The
difficulties he and his co-workers met in these studies of
the whole metabolic systems - the large number of rate
equations and parameters used that made the models very
complicated for the analysis - promoted, in fact, the develop-
ment of the Metabolic Control Analysis. In details, these
problems and their historical aspects are well described by
David Fell in his 'Understanding the Control of Metabolism' ,
Portland Press, London and Miami 1997.
However, as it is mentioned by Korzenievski in this
volume, there are some limits for the use of the Metabolic
Control Analysis: it gives the methods of analysis of
regulation of a metabolic process in a given steady state,
and usually is not related to and not very much interested
299
in its mechanism. It is the great value of the kinetic and
mathematical models that they are based on the detailed
mechanisms of the reactions and processes involved. And
having learned some interesting lessons from the Metabolic
Control Analysis, we are now coming back to the math-
ematical models of metabolic processes. We may call them
'focused mathematical models'. This volume gives four
examples of these models: the kinetic model of interaction
between electron carriers in the respiratory chain by Demin
et al. the model of the oxidative phosphorylation reactions
in the muscle mitochondria by Korzenievski, the theoretical
model of the creatine kinase reactions in the spatial and
temporal energy buffering by Kemp et al. and the model of
compartmentalized energy transfer with the analyze of the
experiments on the isolated hearts of the creatine kinase
deficient mice by Aliev et al. All these models give some
reasonable results which can be experimentally proved, and
the model of the compartmentalized energy transfer explains
well the experimental results which look at first glance
paradoxical, and in this way increases or at least improves
our understanding of complicated intracellular situation. The
common property of all these models is that they are focused
on explaining quantitatively one or several processes which
are considered in details, while the others are generalized
and maximally simplified, or if possible, ignored. This is
analogous in some sense to the top-down analysis in which
the complex system is also divided into parts to simplify the
analysis (see chapter I by M. Brand in this volume). The
focused mathematical models require, however, that the
coefficients and constants included could be determined
experimentally. Otherwise they rapidly loose they value as
methods of quantitative research.
In the literature we can find multiple good examples of
this kind of focused models. One is the work by Meyer et
al. on the analysis of the role of the equilibrium creatine
kinase system in facilitated energy transfer [2], very good
examples are the models of oxygen diffusion and energy
metabolism by Michael Mahler [77], and the work by Jurgen
Daut on cellular energetics [78]. Thus, if the purposes of
modeling are well defined and focused on analysis of a given
process with well determined boundary conditions, these
mathematical models prove to be very effective method of
the research.
CK pathway for the energy transfer: the isoenzyme
compartmentation, the problem of the creatine kinase
equilibrium and cytoplasmic ADP concentration
One of the rather surprising results of the use of mathematical
modeling of the compartmentalized energy transfer in
cardiac muscle cells is the conclusion that the creatine
kinase equilibrium concept has only a very limited value
300
and application in muscle, particularly in the cardiac cell
physiology [21,22]. This very simple and convenient concept
may be well applied for the resting cell and in the diastolic
phase but not in the contracting heart muscle. In fact, this
conclusion is not the unexpected one. Kent Sahlin has
questioned the values of the estimation of the ADP con-
centration in the cells [79]. Meyer found that in skeletal muscle
the maximal activation of the contraction also causes the
disequilibrium of the creatine kinase reaction [80]. Calculations
made on the basis of exact experimental data on creatine kinase
isoenzyme composition and accounting for compartmentation
of its isoenzymes in different cellular compartments show that
for mitochondrial creatine kinase this equilibrium never exists,
and myofibrillar, or cytoplasmic creatine kinase may approach
the equilibrium state only in the systolic phase [21,22]. Thus,
all conclusions regarding the determination of the cytoplasmic
ADP concentration in the cells made on the basis of the
creatine kinase equilibrium may be considered critically and
with caution, if the muscle cell is performing significant work
and the oxygen consumption significantly exceeds the resting
level. Further, these calculations have also shown that
effective functioning of the phosphocreatine pathway in
steady state requires co-expression of different creatine kinase
isoenzymes in the different compartments of the cell. This
means that in the cells in which the phosphocreatine pathway
is active, the components of this pathway - creatine kinase
isoenzymes function in coordinated manner out of equilibrium,
in the metastable steady state (according to the terminology
introduced by Jurgen Daut [78]), ifthe cell is not in the resting
state (see the chapter by Aliev et al. in this volume). The way
in which mitochondrial creatine kinase functions is sensitive
to the changes in the myofibrillar end of the phosphocreatine
shuttle: in the MM-creatine kinase deficient mice mitochondrial
creatine kinase and aerobic phosphocreatine production are
strongly inhibited by high intracellular concentrations of ADP
(see the chapter by Aliev et al. in this volume). This con-
clusion is in concord with the results of the studies of
molecular biology of creatine kinases (see the chapter by in
this volume) and with recent experimental results by Veksler
etal. [81].
These calculations also show that if the compartmentalized
creatine kinase system is out of equilibrium, in the metastable
steady state, one may observe very significant (by order of
magnitude) variations of the myoplasmic ADP concentration
within the cardiac cycle under conditions when the total
cellular level of phosphorus metabolites, such as phospho-
creatine orATPchange within the range of 4-5% [21]. That is
within the range of experimental error of determination of these
metabolites. And this may explain many experimental results
from Balaban's and other laboratories showing constant PCr
levels or PCr/ATP ratios during significant variations of the
respiration rate [23-25], still leaving for ADP the leading role
in feedback regulation between contraction and respiration.
Many experimental data show variations of mitochondrial
calcium concentration within contraction cycle, synchron-
ously with cytoplasmic calcium, favoring the theory of
calcium regulation of mitochondrial oxidative phosphoryl-
ation in details described in this volume by Hansford and
Zorov. This activation of oxidative phosphorylation by
calcium may coincide with changes in ADP concentration
in cytoplasm, and therefore these both potentially important
regulatory factors may exert their action on the oxidative
phosphorylation in a well-coordinated manner. The result is
rapid and stable energy production exactly matching the
energy demand, and in this case the temporarily increased
ADP signal may be rapidly quenched by calcium-activated
oxidative phosphorylation. As a result, we may have some-
thing similar to 'ADP sparks' in cytoplasm, propagating as
a wave, in complete analogy with modern concepts of
calcium metabolism (see above). This is interesting
possibility of feedback regulation of respiration. This
conclusion fits with several groups of experimental data.
When calcium uptake into mitochondria is inhibited by
ruthenium red, the metabolic changes in response to altered
workload are enhanced in comparison with non inhibited
hearts, this showing the 'accumulation' of ADP signal
because of its decreased quenching [82]. However, the
relationship between workload and oxygen consumption
stays unchanged.
Thus, it is most possible that both ADP and calcium
simultaneously participate in the feedback regulation of
oxidative phosphorylation in the intact heart cells.
Experimental verification of this conclusion requires
further experimental and theoretical studies.
What happens when the creatine kinase system is
switched off (inhibited) in the muscle cells?
The phosphocreatine shuttle, or circuit is classically con-
sidered as a major energy transfer pathway in the cells with
high energy fluctuations, such as heart, skeletal muscle, brain
and many other types of the cells [3, 55, 68-71]. However,
this is not a single pathway of energy transfer, since many
types of cells such as liver etc. lack creatine kinase, and the
expression of the creatine kinase system is different in
different types of cells (see the chapter by Ventura-Clapier
et al. in this volume) and changes significantly during muscle
development [83]. The physiological importance of this
system has been studied in very numerous works by using
three different approaches: (1) inhibition of creatine kinase
in the perfused heart [84-86]; (2) substitution of creatine
in the cells by its analogs, such as guanidinopropionate or
guanidinobutirate by feeding it experimental animal during
2-3 weeks [87, 88]; (3) use of transgenic technology to
switch off different creatine kinase isoenzyme genes [89,
 301

90]. In general, the results of all these approaches are in
agreement with each other and may be summarized as it
follows: while the cells perfectly survive without the creatine
kinase system due to activation of other metabolic systems
and adaptive changes (see the paper by Dzeja et al. and
chapter 13 in this volume), the functional properties of the
cell are impaired if the creatine kinase system is defective
or inhibited. Thus, the physiological role of the creatine
kinase system is related to the specific function of the cell
but not to its viability.
We illustrate this conclusion by the experimental results
shown in Figs 5-8. The isovolumic isolated rat heart was
perfused with solutions containing different concentrations
of calcium ranging from 0.5-3.5 mM. This resulted in 3
fold increase of the developed tension, and this effect is
completely reversible (Fig. 5). However, when the hearts
were perfused with iodoacetamide which inhibits more than
90 % of the creatine kinase activity [86], the inotropic effect
of increasing calcium concentrations was completely lost
(Figs 6 and 7) . In the hearts perfused with pyruvate as a
substrate, iodoacetamide itself had no effect on the contract-
ility but eliminated the inotropic effect of increasing calcium

1 sec
r
A 180

120
100

rCa'. = 1,25 mM [Ca,. [3,5 mM]

B
_ _ _5-180
55
10

rCa,. = 3,5 mM

r
W'
1 sec 1 --~~~c
lCa'. = 0,5 mM rCa'. =1,25 mM

Fig. 5. Left ventricular pressure of Langendorff perfused rat heart, glucose perfusion. (A): Heart before and after a transition from 1.25 mM to 3.5 free calcium
perfusate concentration. Note the increase in left ventricular systolic pressure from 118-180 mmHg with almost no concomitant change in end diastolic
pressure; (B): Left ventricular pressure of the same Langendorff glucose perfused heart before and after a transition from 3.5 to 0.5 free calcium perfusate
concentration. Note the decrease in left ventricular systolic pressure from 180 mmHg to 55 mmHg; (C): Left ventricular pressure of the same Langendorff
glucose perfused heart switched back from 0.5 mm free calcium perfusate concentration to the initial concentration of 1.25 mM. Note the complete recovery
of the function and the absence of functional adverse effect of the protocol.
302
0-
t 100 sec
[Ca.,. =
1,25 mM [Ca.,. =
3,5mM
End of I A
(120 /lmo/es / 15 min)
Fig. 6. (A): Left ventricular pressure of Langendorff pyruvate perfused rat heart before and after a transition from 1.25 mM to 3.5 free calcium perfusate
concentration. Note the increase in systolic pressure from 100-180 mmHg; (8): Left ventricular pressure of the same heart before and after infusion of
iodoacetamide (120 ~M over 15 min). Note the absence of increase in end diastolic pressure and again the abolition of contractile reserve when free calcium
perfusate concentration is increased from 1.25--3.5 mM.
concentration (Fig. 6). In the case of glucose as a substrate,
this inotropic effect disappeared simultaneously with
elevation of the end diastolic pressure in the presence of
iodoacetamide (Fig. 7). The latter effect may be explained
by some inhibition of the enzymes of the glycolytic system
by iodoacetamide and decreased production of so called
glycolyticATP which latter effect may be explained by some
inhibition of the enzymes of the glycolytic system by
iodoacetamide and decreased production of so called
glycolytic ATP which has been found to be important for
maintaining heart muscle relaxation [48]. Figures 8A and
8B summarize all these data in coordinates of oxygen
consumption rate (as a measure of energy fluxes) vs work-
loads, represented, as usually for Langendorff perfused
hearts, by RPP, the rate-pressure product. The hearts with
inhibited creatine kinase system are able to perform only
a limited amount of work and develop correspondingly
limited energy fluxes, independently of the substrate used
(Fig. 8). Correspondingly, the maximal value of oxygen con-
sumption in these hearts treated with iodoacetamide is 35
flmole per min per gdw, in contrast with 65-70 flmole per min
per gdw in the control hearts. Taking into the account that
1 sec
180
=3,5 mM
Langendorff-perfused hearts usually perform decreased
levels of work in comparison with working heart models
which develop oxygen consumption rates up to 120 flmoles
per min per gdw [3], in the case of the latter models of
perfusion the differences may be expected to be even more
significant. Thus, the hearts treated with iodoacetamide to
inhibit the creatine kinase system have lost the contractile
reserve. Dzeja et al. have directly shown more than 10-fold
activation ofthe flux through adenylate kinase system under
these conditions (see Fig. 3 in chapter II in this volume),
which takes over the energy transfer function of the creatine
kinase system. However, this is not enough to maintain the
maximal contractile function. Thus, we may conclude that
under normal conditions, in the intact heart, the creatine
kinase system is indeed the major pathway of the energy
transfer network in the cells.
These results are very similar to those reported recently by
Matsumoto et al. and by Ingwalls group [84-86]. Very similar
results were obtained by Kapelko et al. in studies of rat hearts
in which the creatine kinase was intact creatine was substituted
by guanidinopropionate [87], and by Zweier and Jacobus who
used guanidinobutyrate as an analog of creatine [88]. Both these
 303

187

[Ca'. = 1,25 mM [Ca,. ;: 3,5 mM

i
---r-------r----- --~--------.

[Cal. = I A End of I A [Ca,. = 3,5 mM

1.25mM

Fig. 7. (A): Left ventricular pressure of glucose Langendorff perfused rat heart before and after a transition from 1.25--3.5 mM free calcium perfusate
concentration. Note the increase in left ventricular systolic pressure from 120--187 mmHg; (B): Left ventricular pressure of the same glucose Langendorff
perfused rot heart before and after infusion ofIodoacetamide (120 ~M in 15 min). Note the increase in diastolic pressure (from 8--30) and the complete abolition
of contractile reserve when perfusate free calcium concentration is increased from 1.25-3.5 mM.

types of hearts showed decreased contractile function. Acknowledgements

Similarly, the muscle contraction was significantly imp-
aired in transgenic mice with knocked out MM creatine
kinase or both MM and mitochondrial creatine kinases (see
chapters 13, 14 and refs [89, 90]). However, very significant
adaptive changes observed in the cells of these transgenic
animals (see chapter 13) make it difficult to evaluate quantit-
atively the importance of the phosphocreatine pathway for
contraction, since these strong adaptive changes give a
reason for a skepticism that these transgenic animal are in
fact new brand of animals. Nevertheless, this is a very
powerful technique which already has given us good lessons
(see chapter by Klaas Nicolay et al. in this volume).
The situation may be more serious for the cell when these
strong adaptive changes, for some reason, do not occur but
the creatine kinase system is impaired. This is a case of heart
pathology [55, 91].
The authors thank Drs. Robert Balaban (Bethesda, USA),
Jurgen Daut (Marburg, Germany) and MaisAliev (Moscow,
Russia) for active discussion of problems described in this
article. Technical assistance of Mrs. Lena Saareoja and Mr.
Toomas Tiivel is gratefully acknowledged.
References
1. Nagle S: Regelprobleme im Energiestoffwechel des Herzmuskels.
Klin Wsch 48: 1075-1089, 1970
2. Meyer RA, Sweeney HL, Kushmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol 246: C365-{;377, 1984
3. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova Yu 0, Kuzuetsov AV:
Metabolic compartmentation and substrate channelling in muscle
cells. Role of coupled creatine kinases in in vivo regulation of cellular
304

A. bO

50

i
"0
~ 40

]=
t'l
0 30
'0
S
::t
20
Ot'l
~
~
10

0
0 10000 20000 30000 40000 50000 60000 70000
RPP (mmHg/min)

PYRUVATE(IOmM)
y =8.3 x 10 -4 X + 6.15 r =0.99
PYRUV ATE (10 mM) after lodoacetamide (120 f.!molcs)

y = 7.95 x 10-4 X + 8.88 r =0.97

B. 70
o
60 .. ..
'
'

0.
i
"0
50 Q)...................
~
C
] 40
'0 .. ,
~/l, ...
t'l
0 .0
"S
S 30
::t

~ 20
~
~
10

0
0 10000 20000 30000 40000 50000 60000 70000

RPP (mmHg/min)

o GLUCOSE (16.7 mM)

y = 8.8 xlO-4 x + 7.63 r = 0.97

GLUCOSE (16.7 mM) after lodoacetamide ( 120 f.lmolcs)

y=7.5 xlO- 4 x +9,09 r=0.97

Fig. 8. Myocardial oxygen consumption of Langendorff perfused rat hearts as a function of rate pressure product without (circles) and with (triangles)
inhibition of creatine kineses by perfusion with iodoacetamide (120 l!M for 15 min). Different levels of rate pressure product were obtained by changes of the
free calcium concentration from 0.5-3.5 mM. (A): Pyruvate as substrate; (B): Glucose as substrate.

respiration - a synthesis. Mol Cell Biochem 133/134: 155-192, 1994
4. Sjostrand FS: The structure of mitochondrial membranes: A new
concept. J Ultrastruct Res 64: 217-245, 1978
5. Klingenberg M: Muskelmitochondrien. In: K Kramer, 0 Krayer, E
Lehnartz, A v Muralt, HH Weber (eds). Ergebnisse der Physiologic.
Biologischen Chemie und Experimentellen Pharmakologie. Springer-
Verlag, Berlin, 1964, pp 132-189
6. Morrison JF, Cleland WW: Isotope exchange studies of the mechanism
of the reaction catalyzed by adenosine triphosphate: Creatine
phosphotransferase. J Bioi Chern 241: 637--683, 1966
7. Saks VA, Chernousova GB, Gukovsky DE, Smirnov VN, Chazov EI:
Studies of energy transport in heart cells. Mitochondrial isoenzyme of
creatine kinase: Kinetic properties and regulatory action of Mg ions.
Eur JBiochem 57: 273-290,1975
8. van Dorsten FA, Furter R Bijkerk M, Wallimann T, Nicolay K: lli in vitro
kinetics of mitochondrial and cytosolic creatine kinase determined by
saturation transfer 3IP_NMR Biochim Biophys Acta 1274: 59--66, 1996
9. Sols A, Marco R: Concentrations of metabolites and binding sites.
Implications in metabolic regulation. Curr Top Cell Reg 2: 227-273,
1970
10. Mitchell P: Compartmentation and communication in living systems.
Ligand conduction: A general catalytic principle in chemical, osmotic
and chemiosmotic reaction systems. Eur J Biochem 95: 1-20, 1979
II. Wiseman RW, Jeneson JAL, Kushmerick MJ: Why is the sensitivity of
mitochondria to ADP over tenfold lower in permeabilized cells than in
vivo? Biothermokinetics of the Living Cell. Biothermokinetics Press,
Amsterdam, 1996,pp 124-127
12. Fritz-Wolf K, Schnyder T, Wallimann T, Kabsch W: Structure of
mitochondrial creatine kinase. Nature 381: 341-345, 1996
13. Saks VA, Ventura-Clapier RJ: Biochemical organization of energy
metabolism in muscle. J Biochem Org I: 9-29,1992
14. Jacobus WE, Saks VA: Creatine kinase of heart mitochondria:
Changes in its kinetic properties induced by coupling to oxidative
phosphorylation. Arch Biochem Biophys. 2 I 9: 167-178, 1982
15. Nageswara Rao BD, Cohn M: lIP_NMR of enzyme-bound substrates
of rabbit muscle creatine kinase. Equilibrium constants, interconversion
rates, and NMR parameters of enzyme-bound complexes. J Bioi Chern
256: 1716-1721, 1981
16. Burbaum n, Knowles JR: Internal thermodynamics of enzymes
Determined by equilibrium quench: Values of Kin, for enolase and
creatine kinase. Biochemistry 28: 9306-9317, 1989
17. Albery WJ, Knowles JR: Evolution of enzyme function and the
development of catalytic efficiency. Biochemistry 15: 5631-5640, 1976
18. Burbaum JJ, Raines RT, Albery J, Knowles JR: Evolutionary optimization
of the catalytic effectiveness of an enzyme. Biochemistry 28: 9293-
9305,1989
19. Lowson JWR, Veech RL: Effect of pH and free Mg2+ on the Keq of the
creatine kinase reaction and other phosphate hydro lases and phosphate
transfer reactions. J Bioi Chern 254: 6528--6537, 1979
20. Aliev MK, Saks VA: Mathematical modelling of intracellular transport
processes and the creatine kinase system: A probability approach.
Mol Cell Biochem 133/134: 333-346, 1994
21. Aliev MK, Saks VA: Compartmentalized energy transfer in cardio-
myocytes: Use of mathematical modelling for analysis of in vivo
regulation of respiration. Biophys J 73: 428--445,1997
22. Saks VA, Aliev MK: Is there the creatine kinase equilibrium in
working heart cells? Biochem Biophys Res Comm 227: 360-
367, 1996
23. Balaban RS, Kantor HL, Katz LA, Briggs RW: Relation between work
and phosphate metabolite in the in vivo paced mammalian heart.
Science 232: 1121-1123, 1986
24. Balaban RS, Heneman FW: Control of mitochondrial respiration in the
heart in vivo. Mol Cell Biochem 89: 191-197, 1989
305
25. Katz LA, Swain lA, Portman MA, Balaban RS: Relation between phosphate
metabolites and oxygen consumption of heart in vivo. Am 1 Physiol 256:
H265-H274, 1989
26. Mootha VK, Arai AE, Balaban RS: Maximum oxidative phosphorylation
capacity of the mammalian heart. Am J PhysioI272: H769-H775, 1997
27. Hochachka PW, Matheson GO: Regulating ATP turnover rates over
broad dynamic work ranges in skeletal muscles. J Appl Physiol 73:
1697-1703,1992
28. Hochachka PW, McClelland GB: Cellular metabolic homeostasis during
large scale change in ATP turnover rates in muscles. J Exp Bioi 200:
381-386,1997
29. Chance B, Leigh JS, Kent l, McCully K, Nioka S, Clark BJ, Maris JM,
Graham T: Multiple controls of oxidative metabolism in living tissues
as studied by phosphorus magnetic resonance. Proc Nat! Acad Sci
USA83: 9458-9462,1986
30. Ammann H, Noel J, Tejedor A, Boulanger Y, Gougoux A, Vinay P:
Could cytoplasmic concentration gradients for sodium and ATP exist
in intact renal cells? Can J Physiol Pharmacol 73: 421--435, 1995
31. OvaldiJ: Cell architecture and metabolic channelling. SpringerVerlag,
NewYork-Berlin-London-Paris, 1995, pp 1-250
32. Agius L, Sherratt HSA (eds).: Channelling in Intermediary Metabolism.
Portland Press, London and Miami, 1997, pp 1-342
33. Bereiter-Hahn J, Voth M: Dynamics of mitochondria in living cells:
Shape changes, dislocations, fusion and fission of mitochondria. Micro
ResTech27: 198-219, 1994
34. Rostovtseva T, Colombini M: VDAC channels mediate and gate the
flow ofATP: Implications for the regulation of mitochondrial function.
BiophysJ72:1954-1962,1997
35. Hodge T, Colombini M: Regulation of metabolic flux through voltage-
gating ofVDAC channels. 1 Membr Bioi 157: 271-279, 1997
36. Heiden M, Hilschmann N, Thinnes FP, Kroll K: Proteins of cytosol and
amniotic fluid increase the voltage dependence of human type-I porin.
JBioenergBiomembr28: 171-180, 1996
37. Reymann S, Flarke H, Heiden M, Jakob C, Stadtmuller U, Steinacker
P, Lalk VE, Pardowitz I, Thinnes FP: Further evidence for multi-
topological localization of mammalian porin (VDAC) in the plasma-
lemma forming part of a chloride channel complex affected in cystic
fibrosis and encephalomyopathy. Biochem Mol Medicine 54: 75-87,
1995
38. Beutner G, RuckA, Riede B, Welte W, Brdiczka 0: Complexes between
kinases, mitochondrial porin and adenylate translocator in rat brain
resemble the permeability transition pore. FEBS Lett 396: 189-195,
1996
39. Clark JF, Kuznetsov AY, Radda GK:ADP regenerating enzyme systems
in mitochondria of guinea pig myometrium and heart. Am J Physiol
272:C399-C404,1997
40. McCabe ERB: Microcompartmentation of energy metabolism at the
outer mitochondrial membrane: Role in diabetes mellitus and other
diseases. 1 Bioenerg Biomembr 26: 317-325, 1994
41. Laterveer FD, Nicolay K, Gellerich FN: ADP delivery from adenylate
kinase in the mitochondrial intermembrane space to oxidative
phosphorylation increases in the presence of macromolecules. FEBS
Lett 386: 255-259, 1996
42. Brdiczka 0, Wallimann T: The importance of outer mitochondrial
compartment in regulation of energy metabolism. Mol Cell Biochem
133/134: 69-84, 1994
43. Kushmerick MJ, Podolsky RJ: Ion mobility in muscle cells. Science
166: 1297-1298, 1969
44. Nicolay K, van der Toorn, Dijkhuizen RM: In vivo diffusion spectro-
scopy. An overview. NMRBiomed 8: 365-374, 1995
45. Walter H, Brooks DE: Phase separation in cytoplasm, due to macro-
molecular crowding, is the basis for microcompartmentation.
FEBS Lett 361: 135-139, 1995
306
46. Oplatka A: The role of water in the mechanism of muscular contraction.
FEBS Lett 355: 1-3, 1994
47. Weiss IN, Lamp ST: Glycolysis preferentially inhibits ATP-sensitive
K+ channels in isolated guinea pig cardiac myocytes. Science 238: 67-
69,1987
48. Opie LH: Cardiac metabolism - emergence, decline and resurgence.
Cardiovasc Res 26: 721-733, 1992
49. Kuzuetsov AV, Tiivel T, Sikk P, Kaambre T, Kay L, Daneshrad Z, Rossi
A, Kadaja L, Peet N, Seppet EK, Saks VA: Striking differences between
kinetics of regulation of respiration by ADP in slow-twitch and fast-
twitch muscles in vivo. Eur J Biochem 241: 909-915,1996
50. Veech RL, Lawson JWR, Cornell NW, Krebs HA: Cytosolic phos-
phorylation potential. J Bioi Chern 254: 6538--6547, 1979
51. Kushmerick MJ: Skeletal muscle: A paradigm for testing principles of
bioenergetics. J Bioenerg Biomembr 27: 555-569, 1995
52. Bessman SP, Geiger PJ: Transport of energy in muscle. The phosphoryl-
creatine shuttle. Science 211: 448-452, 1981
53. Gercken G, Schlette U: Metabolite status of the heart in acute
insufficiency due to l-fluoro-2,4-dinitrobenzene. Experientia 24: 17-
18,1968
54. Gudbjarnason S, Mathes P, Ravens KG: Functional compartmentation
of ATP and creatine phosphate in heart muscle. J Mol Cell Cardiol 1:
325-339, 1970
55. Saks VA, Ventura-Clapier R (eds).: Cellular Bioenergetics: Role of
Coupled Creatine Kinase. Kluwer Academic Publishers, 1994, pp 1-
346
56. Williams DA: Mechanisms of calcium release and propagation in
cardiac cells. Do studies with confocal microscopy add to our
understanding? Cell Calcium 14: 724-735, 1993
57. Rizzuto R, Bastianutto C, Brini M, Murgia M, Pozzan T: Mitochondrial
Ca'+ homeostasis in intact cells. J Cell Bioi 126: 1183-1194, 1994
58. Cheng H, Lederer WJ, Cannell MB: Calcium sparks: Elementary events
underlying excitation-contraction coupling in heart muscle. Science
262:740-744,1993
59. Cannell MB, Cheng H, Lederer WJ: The control of calcium release in
heart muscle. Science 268: 1045-1049, 1995
60. Lipp P, Niggli E: Modulation of calcium release in cultured neonatal
cardiac myocytes. Insight from subcellular release patterns revealed
by confocal microscopy. Circ Res 74: 979-990, 1994
61. Tanaka H, Kawanishi T, Matsuda T, Takahashi M, Shigenobu K:
Intracellular free calcium movements in cultured cardiac myocytes as
shown by rapid scanning confocal microscopy. J Cardiovasc Pharmacol
27: 761-760,1996
62. Minamikawa T, Cody SH, Williams DA: In situ visualization of
spontaneous calcium waves within perfused whole rat heart by confocal
imaging. Am J PhysioI272: H236--H243, 1997
63. Isenberg G, Etter EF, Wendt-Gallitelli MF, Schiefer A, Carrington WA,
Tuft RA, Fay FS: Intrasarcomere [Ca'+] gradients in ventricular
myocytes revealed by high speed digital imaging microscopy. Proc
Natl Acad Sci USA 93: 5413-5418, 1996
64. Mironneau J, Amaudeau S, Macrez-Lepretre, Boittin FX: Ca'+ sparks
and Ca'+ waves activate different Ca'+-dependent ion channels in
single myocytes from rat portal vein. Cell Calcium 20: 153-160, 1996
65. Fiolet JWT, Baartscheer A, Schumacher CA: IntraceJlular [Ca'+] and Vo,
after manipulation of the free-energy of the Na+/Ca'+-exchanger in
isolated rat ventricular myocytes. J Mol Cell Cardiol 27: 1513-1525,
1995
66. Janiak R, Lewartowsky B, Langer GA: Functional coupling between
sarcoplasmic reticulum and Na/Ca exchange in single myocytes of
guinea pig and rat heart. J Mol Cell Cardiol28: 253-264, 1996
67. Carmeliet E: A fuzzy subsarcolemmal space for intracellular Na in cardiac
cells? Cardiovasc Res 26: 433-442, 1992
68. Wa1limann T: 3IP-NMR-measured creatine kinase reaction flux in muscle:
A caveat! J Musc Res Cell MotiJ 17: 177-181, 1996
69. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine kinase
isoenzymes in tissues with high and fluctuating energy demands: The
'phosphocreatine circuit' for cellular energy homeostasis. Biochem J
281: 31-40,1992
70. Wallimann TW, Hemmer W: Creatine kinase in non -muscle tissues
and cells. Mol Cell Biochem 133/134: 193-220, 1994
71. Wyss M, Smeitnik J, Wever RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119-166,1992
72. Wan B, Dounen C, Duszyusky J, Salama G, Vary TC, Lanoue KF: Effect
of cardiac work on electrical potential across mitochondrial membrane
in perfused heart. Am J Physiol265: H453-H460, 1993
73. Jurevicius J, Fischmeister R: cAMP compartmentation is responsible
for a local activation of cardiac Ca'+ channels by ~-adrenergic agonists.
Proc NatiAcad Sci USA 93: 295-299,1996
74. Jurevicius J, Fischmeister R: Acetylcholine inhibits Ca'+ current by
acting exclusively at a site proximal to adenylyl cyclase in frog cardiac
myocytes. J Physiol491: 669-675, 1996
75. Hoerter JA, Lauer C, Vassort G, Gueron M: Sustained function of
normoxic hearts depleted inATP and phosphocreatine: A P-NMR study.
AmJPhysiol255: C192--C201, 1988
76. Neely JR, Grotyohann LW: Role of glycolytic products in damage to
ischemic myocardium. Dissociation of adenosine triphosphate levels
and recovery of function ofreperfused ischemic hearts. Circ Res 55:
816--824,1984
77. Mahler M: First-order kinetics of muscle oxygen consumption, and an
equivalent proportionality between Qo, and phosphorylcreatine level.
JGenPhysiol86: 135-165, 1985
78. Daut J: The living cell as energy-transducing machine. A minimal
model of myocardial metabolism. Biochim Biophys Acta 895: 41--62,
1987 .
79. Sahlin K: Control of energetic processes in contracting human skeletal
muscle. Biochem Exer 19: 353-358, 1991
80. Meyer RA, Foley JM: Cellular processes integrating the metabolic
response to exercise. In: LB Rowell, JT Shepherd (eds). American
Physiological Society Handbook of Physiology. Integration of Motor,
Circulatory, Respiratory and Metabolic Control during Exercise. 1996,
pp 841--868
81. Veksler VI, Kuznetsov AV,Anflous K, Mateo P, van Deursen J, Wieringa
B, Ventura-Clapier R: Muscle creatine kinase. Cardiac and skeletal
muscle exhibit tissue-specific adaptation of the mitochondrial function.
JBiolChem270: 19921-19929, 1995
82. McCormack JG, England PJ: Ruthenium red inhibits the activation of
pyruvate dehydrogenase caused by positive inotropic agents in the
perfused heart. BiochemJ 214: 581-589, 1983
83. Hoerter JA, Ventura-Clapier R, Kuznetsov AV: Compartmentation of
creatine kinases during perinatal development of mammalian heart.
Mol Cell Biochem 133/134: 277-286, 1994
84. Matsumoto Y, Kaneko M, KobayashiA, Fujise Y, Yamazaki N: Creatine
kinase kinetics in diabetic cardiomyopathy. Am J Physiol268: EI070-
E1076,1995
85. Hamman BL, Bitt! JA, Jacobus WE, Allen PD, Spencer RS, Tian R, Ingwall
JS: Inhibition of the creatine kinase reaction decreases the contractile
reserve of isolated rat hearts. Heart Cire Physiol 38: H103O-HI036, 1995
86. Tian R, Ingwall JS: Energetic basis for reduced contractile reserve in
isolated rat hearts. Am J Physiol 270: HI207-Hl216, 1996
87. Kapelko VI, Kupriyanov vv, Novikova NA, Lakomkin VL, Steinschneider
AY, Severina MY, Veksler VI, Saks VA: The cardiac contractile failure
induced by chronic creatine and phosphocreatine deficiency. J
Mol Cell Cardiol 20: 465-479, 1988
88. Zweier JL, Jacobus WE, Korecky B, Brandejs-Barry: Bioenergetic

consequences of cardiac phosphocreatine depletion induced by creatine
analog feeding. 1 Bioi Chern 266: 20296-20304, 1991
89. van Deursen l, Heerschap A, Oerlemans F, Ruitenbeek W, lap P, ter
Laak H, Wieringa B: Skeletal muscles of mice deficient in muscle creatine
kinase lack burst activity. Cell 74: 621--631, 1993
90. Steeghs K, Benders A, Oerlemans F, de Haan A, Heerschap A, Ruitenbeek
W, lost C, van Deursen l, Perrymann B, Pelle D, Bruckwilder M,
307
Koudijs J, lap P, Veerkamp Wieringa B: Altered Ca'+ responses in muscles
with combined mitochondrial and cytosolic creatine kinase deficiencies.
Cell 89: 93-103, 1997
91. Neubauer S, Hom M, Naumann A, Tian R, Hu K, Laser M, Friedrich l,
Gaudron, Schnackerz K, Ingwall JS, Ert! G: Impairment of energy
metabolism in intact residual myocardium of rat hearts with chronic
myocardial infarction. J Clin Invest 95: 1092-1100, 1995
 PART III

METABOLIC SIGNALLING AND CALCIUM:

REGULATION OF MITOCHONDRIAL OXIDATIVE
PHOSPHORYLATION IN VIVO
Molecular and Cellular Biochemistry 184: 311-320, 1998.
1998 Kluwer Academic Publishers.

Subtleties in control by metabolic channelling and

enzyme organization
Boris N. Kholodenko,1,2 Johann M. Rohwer,3 Marta Cascante2 and
Hans V Westerhoff3,4
lA.N. Belozersky Institute of Physico-Chemical Biology, Moscow State University, 119899 Moscow, Russia; 2Universitat de
Barcelona, Departmento de Bioquimica i Biologia Molecular, Marti i Franques 1, E-08028 Barcelona, Spain; 3E.C. Slater
Institute, BioCentrum, University of Amsterdam, Plantage Muidergracht 12, NL~1 018 TV Amsterdam; 4Department of
Microbial Physiology, BioCentrum, Faculty of Biology, Free University, De Boelelaan J087, NL~1081 HV Amsterdam, The
Netherlands

Abstract
Because of its importance to cell function, the free-energy metabolism of the living cell is subtly and homeostatically controlled.
Metabolic control analysis enables a quantitative determination of what controls the relevant fluxes. However, the original metabolic
control analysis was developed for idealized metabolic systems, which were assumed to lack enzyme-enzyme association and
direct metabolite transfer between enzymes (channelling). We here review the recently developed molecular control analysis, which
makes it possible to study non-ideal (channelled, organized) systems quantitatively in terms of what controls the fluxes,
concentrations, and transit times. We show that in real, non-ideal pathways, the central control laws, such as the summation theorem
for flux control, are richer than in ideal systems: the sum of the control of the enzymes participating in a non-ideal pathway may
well exceed one (the number expected in the ideal pathways), but may also drop to values below one. Precise expressions indicate
how total control is determined by non-ideal phenomena such as ternary complex formation (two enzymes, one metabolite), and
enzyme sequestration. The bacterial phosphotransferase system (PTS), which catalyses the uptake and concomitant phosphorylation
of glucose (and also regulates catabolite repression) is analyzed as an experimental example of a non-ideal pathway. Here, the
phosphoryl group is channelled between enzymes, which could increase the sum of the enzyme control coefficients to two, whereas
the formation of ternary complexes could decrease the sum of the enzyme control coefficients to below one. Experimental studies
have recently confirmed this identification, as well as theoretically predicted values for the total control. Macromolecular crowding
was shown to be a major candidate for the factor that modulates the non-ideal behaviour of the PTS pathway and the sum of the
enzyme control coefficients. (Mol Cell Biochem 184: 311~320, 1998)

Key words: metabolic channelling, non-ideal metabolism, control coefficient, enzyme-enzyme interactions, macromolecular
crowding, bacterial phosphotransferase system

Introduction forms, e.g., direct transfer of intermediates (,channelling'),

dynamic versus static channelling [3, 4], restricted diffusion
In a standard view cellular metabolism is presented in terms [5], local coupling [6] and macromolecular crowding [7, 8].
of so-called 'ideal' metabolic pathways. This paradigm implies Recently developed extensions of metabolic control analysis
that the catalysts (enzymes) interact only through bulk-phase (reviewed in [9]) make it possible to predict and analyse the
intermediates in well defined compartments containing well- implications of various types of enzyme organization for the
mixed metabolite pools. However, it has long been recognized regulation and control of cell function.
by biochemists that aspects of cellular metabolism are or- Enzyme-enzyme interactions can result in many special
ganized both structurally and temporarily (see [1, 2] for recent control properties. In this chapter we shall review the subtleties
reviews). Organized (non-ideal) metabolism comes in many of the control brought about by metabolic channelling and
Present address: B.N. Kholodenko, Department of Pathology, Anatomy and Cell Biology, Thomas Jefferson University, JAH, 1020 Locust Street, Philadelphia,
PA 19107, USA
Present address: J.M. Rohwer, Department of Biochemistry, University of Stellenbosch, Private Bag XI, Matieland, 7602, South Africa
Address/or offprints: B.N. Kholodenko, Department of Pathology, Anatomy and Cell Biology, Thomas Jefferson University, JAH, 1020 Locust Street,
Philadelphia, PA 19107, USA
312
enzyme organization. We shall show how, by stimulating
intensive and rational research, these subtleties have led to
increased insight into unsuspected aspects and mechanisms
of biological regulation. In particular, we shall show that the
high concentration of macromolecules in the cytoplasm of
living cells leads to completely new regulatory properties of
pathways involving protein-protein interactions.
High concentrations of enzymes and macromolecular
crowding in the cytoplasm drive the formation of specific
enzyme-enzyme complexes possibly involving the direct
transfer of metabolites. Hypotheses have been proposed for
the biological significance and formulated with respect to the
function of metabolic channelling. A reduction in required
hydration space (and hence, an apparent increase in accessible
cell volume) or a decrease in the bulk-phase concentrations
of unstable or toxic intermediates and in leakage reactions
have been suggested [3]. Here we shall review how metabolic
channelling can effect the bulk-phase pools of intermediates.
In addition a function of channelling in reducing response
times has been suggested (e.g. [10]). We shall discuss the
consequences of channelling for the transit time of a molecule
in a pathway.
Quantification of the control in ideal pathways: Metabolic
control analysis
Metabolic control analysis provides us with a tool for the
quantitative analysis of biological control and regulation. The
control exerted by an enzyme on the steady state flux or
metabolite concentration has been quantified as a control
coefficient [11, 12]. The control coefficient is the fractional
change in the pathway flux (J) or concentration (X) divided
by the fractional change in the activity of an enzyme (i),
extrapolated to infinitesimally small change:
Vi (aJlaptys (alnl.ll/ap )Sys
eJ J
Vi
(av/ap)enz (alnlvyap) enz
(aXlap)sys (alnlX]/ap )sys
eX =2 = direct protein-protein interactions. More subtle indirect
Vi
X (aV/ap)enz (alnlvyap )enz (1)
Here aJ and ax are the changes in J and X brought about by
the change (ap) in the parameter (P) which specifically affects
*
only the rate Vi of the enzyme i (av lap 0, av)ap = 0, for
*
any j i). aV i is the change in this net rate brought about by
the change (ap) when the concentrations of all the intermediates
are clamped at their steady-state values. The difference
between aJ and av is that the intermediate concentrations
are held constant to estimate av, but are allowed to reach new
steady-state values when determining oJ.
This difference is
specified by the subscripts enz and sys respectively.
ev is usually referred to as the control coefficient with
respect to the enzyme catalytic activity (VrnJ rather than
enzyme concentration (e). An advantage of the definition
given by Eq. 1 is that ev does not depend on a particular way
of modulation of the enzyme activity unless different
enzymes interact directly, e.g. associate in a complex [13].
Choosing the concentration (e) of an enzyme (i) as the
parameter (P) in Eq. 1, one arrives at the definition of enzyme
(-concentration) control coefficient (e), which was proposed
originally by Kacser and Burns [11]
(2)
The name' enzyme-concentration' control coefficient makes
explicit that this coefficient is measured by changing ei , e.g.
by addition of the enzyme or catalyst to a reconstituted
system (e.g. [14]) or by manipulating the expression of the
corresponding gene in an intact system (e.g. [15]).
In the paradigm of 'ideal' metabolism the enzymes are
independent catalysts, which are present at much lower
concentrations than the substrates they convert. As in
ordinary enzyme kinetics [16], the rate of a reaction is
proportional to the enzyme concentration. Therefore, in ideal
pathways the control coefficient (eJ with respect to the
enzyme activity is identical to the control coefficient referring
to the enzyme concentration (eJ However, this is not the
case for non-ideal metabolism [6, 17-21].
Molecular control analysis: A tool to quantifY control in
non-ideal pathways
For non-ideal metabolism the control coefficients (C) as
defined by Eqation 1 can depend on the choice of a modula-
tion parameter (P), i.e. on how the enzyme activity was
modulated [13, 22]. This dependence appears to be the result
of direct or indirect interactions between enzymes. Multi-
protein complexes and metabolic channelling result from
effects can also arise, e.g. when co-enzyme moiety conser-
vation involves different enzymes bound [23, 24]. Importantly,
in a channelled pathway it is no longer possible to assign an
(independent) single reaction to each enzyme.
A scheme of a 'dynamic' channel [25-28] of two con-
secutive reactions is depicted in Fig. 1. There are two routes
from a substrate molecule (S) to a product (P) of the reaction
pair: a 'classical' route through the bulk-phase intermediate (X)
and a channel path involving the enzyme-enzyme complex.
Figure 1 illustrates the problem: each enzyme participates in
two reactions rather than in one, and both enzymes 1 and 2
participate in the same channelled reaction. Therefore, in this

Fig. 1. A dynamic channel of two consecutive enzyme reactions. The upper
route represents the usual path from a substrate (S) to a product (P) of the
reaction pair through the bulk-phase intermediate (X), catalysed by free
enzymes; the lower route represents the channel. The elemental steps are
numbered as indicated.
non-ideal pathway it is no longer straightforward to modulate
the activity of a particular enzyme in such a way that only a
single reaction is changed. The other reaction involving the
second enzyme also tends to change.
To approach this problem we have quantified the control
with respect to every elemental process (step), defining the
elemental (microscopic) control coefficient [21]. This enabled
us to quantify the control exerted by the enzymes, estimating
the (macroscopic) control coefficient of an enzyme from the
microscopic contributions [29--31]. For instance, there are six
steps - or elemental processes - in the scheme of Fig. l. These
processes correspond to transitions between different states
of enzymes, or they can correspond to sequences of such
transitions that are not interrupted by branches. Consequently,
there are six elemental control coefficients, defined in terms
of simultaneous proportional modulation of the forward (k)
and reverse (k) rate constants for each step i (cf. [32]):
Ci
J _
-
(dlnl]l)
~~
_
,k.lk. - constant,
._ 1,2,... ,6
I -
dInk; ,ys / -/
It should be noted that this definition does not affect
microscopic reversibility. An equivalent, but more general
definition of the elemental control coefficient (C:) can be
given in a manner similar to Eq. 1. In this case, however, the
rate (v) becomes the rate of the ith elemental process rather
than the rate of the entire enzyme i reaction:
c' = l . (a]lap)m ,
C.X =
~
.
(aXlap)""
I ] (av/ap)proc I X (av/ap)proc
Most importantly, addressing the elemental processes, Eq. 4
defines the control coefficient C: as a general quantity, which
is independent of a special choice of a modulation parameter
(P) for both ideal and non-ideal cellular pathways [13, 22].
Quantification of control for any molecular system that can
be described by a kinetic scheme of transitions between states
of enzymes and enzyme-enzyme complexes has been referred
to as molecular control analysis [33] and has been used for
313
analysing the control within single enzymes [32-35] and
within 'relay' pathways such as the mitochondrial electron
transport chain and the bacterial phospho transferase system
[36-38]. Molecular control analysis descends to the micro-
level of molecular interactions and then mounts back to the
macro-level of enzyme-catalysed reactions. In the following
we shall use molecular control analysis to estimate various
modes of control exerted by enzymes in channelled pathways.
The concept of an elemental control coefficient makes it
possible to assign an analogue of the control coefficient with
respect to the enzyme activity (C) to the individual enzymes
of a channel. Let us suppose, that we simultaneously change
the elemental rate constants of all processes in which enzyme
i is involved, by the same factor. Considering the corre-
sponding change in the steady state flux] we define the
'impact' control coefficient, impC~i' as the sum ofthe elemental
control coefficients over the processes in which any of the
forms of enzyme i is involved (Ei-dependent processes) [21]:
impc~; = L
all Ei-dependent processes k
~ (5)
This coefficient evaluates the total impact enzyme i may have
on the flux] (the term 'E;-dependent' refers to any form of
enzyme i, monomeric or complexed, that contains Etmoiety)
[9,21].
Central to the special control properties of channelled
pathways is the fact that when two enzymes form a complex,
the corresponding elemental steps involved in the proteins'
interaction depend on either enzyme. For example, in the
dynamic channel of Fig. I the channel steps 5 and 6 both
contribute to the control exerted by either enzyme:
(3)
The impact control coefficient impc: (defined in terms of
modulation of the activities of all EI ~dependent processes)
does not correspond to the effect of just a change in the total
concentration of enzyme 1 at a constant concentration of
enzyme 2, which was the definition (Eq. 2) of the control
coefficient with respect to enzyme concentration, CJel .
Furthermore, the change in the form E I XE 2 , caused by
changes in activities of the elemental steps, interferes with
(4) the conservation of the total concentration of enzyme 2.
Consequently, in cases of direct enzyme-enzyme interactions
as well as in other non-ideal pathways there is a difference
between the control coefficients defined in terms ofmodulation
of activity and those defined in terms of modulation of the
enzyme concentration.
For any given channelling scheme, molecular control
analysis enables us to express the enzyme (-concentration)
control coefficients in terms of the elemental control
coefficients. For the dynamic channel of Fig. 1 this expression
reads (see Eq. 13 in [22]):
314
(7)
Here E 1XE 2 is the concentration of enzyme-enzyme complex,
e 1 and e2 are the total concentrations of the enzymes, and the
impact control coefficients impCl and impCl can be expressed
eI C2
in terms of elemental control coefficients through Eq. 6.
How to assess the contribution of different enzymes to the
control: The summation laws
In ideal metabolism the total control exerted by enzymes on
a flux is equal to one [11,12]:
I,C:= I
i I
(8)
In ideal pathways this 'summation theorem' is valid indepen-
dently of the structure ofa pathway and the kinetic properties
of enzymes. Indeed, in such pathways any steady-state flux
is a homogeneous first-order function of the enzyme
concentrations (i.e. if the enzyme concentrations simul-
taneously change by a factor ex, all fluxes in the system will
change by the same factor ex), and Eq. 8 is a simple conse-
quence of Euler's theorem [39-41].
The summation law has been used to assess which enzymes
in a pathway might be important in controlling the pathway
flux. For instance, ifin an ideal pathway of two consecutive
reactions one finds that the control coefficient of enzyme 1
is more (less) than 0.5, one can conclude that this enzyme
exerts more (less) control on the flux than enzyme 2. The
'summation theorem' given by Eq. 8 is the gold standard of
ideal metabolic pathways (e.g. of the enzymologist's test
tube). However, we shall see that new summation theorems
characterise channelled pathways.
Subtlety in control properties of a dynamic channel: Total
control over the flux can be as high as two or less than one
The conclusion that was drawn immediately above with
respect to the relative importance of enzymes in a two-
enzyme pathway for the flux, cannot apply to a channel where
the sum of enzyme control coefficients in Eq. 8 does not equal
one [6, 19,23]. For the dynamic channel of Fig. 1 the sum of
the enzymes' control coefficients can be estimated readily
from Eq. 7. In the case where the total concentrations of the
two enzymes are equal (e l = e2 = e) one obtains [21]:
1 + C~+C~
1 +EIXE/e
J is the total flux through the pathway; J ehan denotes the flux
through the channel. If most of the flux goes through the
channel,JchanlJ approaches one; the prediction of the Eq. 9 is
that the sum of the control coefficients is as high as two if
only a small proportion ofthe enzymes exists as the enzyme-
enzyme complex (i.e. E 1XE2 e) and iflittle control resides
in steps I and 4 (e.g. if the binding of Sand P is near-
equilibrium). On the other hand, if the complex does not
enhance the catalytic rate (so that a substantial fraction of the
flux continues to run through the bulk phase) and if most of
the proteins are complexed, i.e. E 1XE2 '" e, the sum of the flux
control coefficients can be well below one. This reflects
'enzyme sequestration': enzyme I is sequestered by enzyme
2 in the complex. These entirely new phenomena cannot
occur in pathways lacking the channel.
In a dynamically channelled pathway consisting of, say,
five enzymes, it is unlikely that the association of enzymes 4
and 5 is required for the combined catalytic action of enzymes
1 and 2. Therefore, in a pathway ofn enzymes with channels
formed by bimolecular complexes between adjacent enzymes
only, the maximum sum of the flux-control coefficients equals
two [36, 37]. However, if several enzymes can assemble in a
single channel in which a metabolite is handed over the
channelled path, then the sum of the enzyme flux-control
coefficients can reach as high as the number of the enzymes
in the channel [6,42].
Regulatory consequences of macromolecular crowding for
metabolic channelling
Traditionally, kinetic studies are performed in dilute solutions
of purified enzymes, i.e. in 'ideal' pathways. The cell,
however, cannot be reduced to the enzymologist's test tube.
The cytoplasm of even relatively 'simple' bacterial cells
contains a variety of macromolecular species that occupy a
substantial fraction of the total volume. Such a medium is
usually referred as 'crowded' [7]. Particularly in eukaryotes
the cytoplasm is structured by the presence of a network of
filaments, microtubules and membranes.
Macromolecular crowding has been shown to be able to
increase drastically the thermodynamic activities of individual
proteins (see [43] for review). For instance, at high concen-
trations of hemoglobin in solution (similar to those in the
erythrocyte), there is an almost exponential increase in the
solution osmotic pressure with hemoglobin concentration
[44]. Further studies have shown that the addition of inert
polymers (e.g. polyethylene glycol) as 'crowding agents'
causes a decrease in theKM-values of several enzymes and a
 315

dramatic stimulation of some macromolecular association 1.0

J!l
reactions (reviewed in [43]). c:
Q)
0.9
(a)
.(3
The source of nonideality in the dynamic and equilibrium
IE
behaviour of individual macromolecular species in crowded Q)
0 0.8 enzyme 1
0
media can be understood readily. To create a space for an
additional (test) molecule of a protein in crowded solution, the .... 0.7
(5
"E
background macromolecules should be reorganised. The 0
0 0.6
concomitant decrease in configuration entropy gives rise to the oJ<.
:::J
increase in the activity coefficient of the test macromolecule
q::
Q)
0.5
[43]. On the other hand, the effect of background macro- E
~ 0.4
molecules on the activity coefficients of small ligand molecules c:
is less significant, as an additional ligand molecule can be UJ 0.3
solvated without rearranging the positions of space-filling 0 20 40 60 80 100
macromolecules. To analyse what consequences macro- 1.7
molecular crowding may have on metabolic channelling we
1.6
(b)
assume, as a first approximation, that crowding increases the

-
thermodynamic activity of the individual proteins but does not
increase the thermodynamic activities of small (ligand) .... 1.5
(5
c:
molecules. 0
0 1.4
We shall consider how an increase in the concentration of x

inert macromolecules in the medium affects the control
properties of the channel in Fig. 1. An increase in the
thermodynamic activities of enzymes 1 and 2 at constant
activities of the substrate (S), product (P) and intermediate
(X) is equivalent to an equal relative increase in all the rate
constants except those corresponding to the association of the
enzymes in the ternary complexE,X'E2 These latter constants
should increase to a greater extent owing to the increased
activity of either interacting protein. Neglecting possible
changes in the volume excluded by macromolecules that can
accompany the association of enzymes in the complex, we
shall assume that when all the rate constants increase by a
factor y, the forward rate constant k5 and the reverse rate
constant k~, leading to the formation of the ternary complex
E,XE2 , will increase byi. Note that when the volume ofthe
enzyme complex is less than the sum of the volumes of the
separate enzyme molecules, the apparent association constants
k/k_5 and k~/k6 can increase even more than by a factor y.
Figure 2 illustrates numerically how the control properties
of a dynamic channel can change with relative increases in
the thermodynamic activities of enzymes. In a dilute solution,
in which the enzyme concentrations are low and background
macromolecules are absent, the sum of enzyme control
coefficients is close to the classical value of one (Fig. 2b).
Eq. 9 explains this result, as at such low activities of the
proteins both the channelled flux fraction and the concentration
of the ternary complex are negligible. Figure 2a shows that,
at the given magnitudes of the rate constants, the control
coefficient of enzyme 2 (0.6) appears to be greater than that
of enzyme I (0.4).
To approach the conditions in a real cell, one should
increase the concentration of the backgroundmacromolecules
(note that the concentration ofthe particular enzymes forming
-
:::J
q:: 1.3
Iii
0 1.2
I-
1.1
1.0
0 20 40 60 80 100
Increase in enzyme activities
Fig. 2. Changes in the control properties of a dynamic channel with an
increase in the thermodynamic activities of the enzymes. (a) The flux-control
coefficients of enzymes 1 and 2 (see Fig. 1); (b) their sum. An increase in
thermodynamic activities was modelled as equal relative increases in all
rate constants except the rate constants of enzyme-enzyme association (k,
and k..,), which increase with the same factor of proportionality but to the
power 2. The initial values ofthe rate constants were (dimensionless units):
kl = 0.2, k_l = 0.05, k2 = 10, k_2 = 2, k J = 0.001, k_J = 0.005, k4 = 20, k-4 =
0.01, k, = 0.25, k_, = 1, k6 = 4, k-f, = 1; the other parameters were: e 1 = e2=
O.I,S= 1O,P= 1.
the channel remains constant). This leads to an increase in
the activities of the channel enzymes and in the apparent
values of the association constants of the enzyme-enzyme
complex as explained above. As a consequence, the channelled
flux fraction increases and more enzyme-enzyme complex is
formed. In line with Eq. 9, Fig. 2 shows that the initial
increase in the channelled flux is accompanied by an increase
in the sum of the enzyme control coefficients. This sum
attains a maximal value much greater than unity [22, 31].
Interestingly, an equal increase in the thermodynamic
activities of the enzymes has switched their relative roles in
the control: the control coefficient of enzyme 1 has become
greater than that of enzyme 2 (Fig. 2a).
With a further increase in enzyme activities the sum of
control coefficients decreases again. This effect has been
recognised as 'enzyme sequestration' in channelled pathways
316
[21]. Importantly, these phenomena are absent from ideal
pathways, where proportional increases in the enzyme
activities always result in the same proportional increase of
the flux without any change in the control coefficients [11].
Phospho transferase system of Escherichia coli
The effect of enzyme sequestration is of great importance for
the control properties of group-transfer or relay pathways.
Group-transfer pathways in the cell are highly organised,
involving the transfer of a chemical group through a series
of proteins. Examples include the electron transport chain in
mitochondrial and bacterial membranes, which transfers an
electron from, e.g. NADH to oxygen, and the bacterial
phosphotransferase system shown in Fig. 3. Here a phosphate
group is transferred from phosphoenolpyruvate to a carbo-
hydrate molecule whilst transporting the latter across the
membrane. Such a group-transfer pathway can be considered
as a perfect dynamic channel in which a transferred group is
not released into the bulk aqueous phase until it reaches the
end of the reaction sequence.
Van Dam et al. [36] have shown that ifthe mean life-time
of enzyme-enzyme complexes is negligible compared to that
of non-complexed enzyme forms the total control of enzymes
over the group-transfer flux equals two. Experimental data
of Brand et al. [45] for the electron-transport chain of liver
mitochondria confirmed this theoretical prediction. One may
infer that the mean life-time of enzyme-enzyme complexes
will increase with an increase in enzyme concentrations. How
will the total control exerted by enzymes then change? A
(a)
~~
:;~~ PEP EI HPr-P IIAG'c IICsG'c-P',f~l Glc
~ ",lJ H~ x'v.'"J"'"CFf'-\
Glc-6-P
(b)
S P y EI yHPr-p
y IIA yIlCB-P-yGIC
S-P-EI EI-P-HPr HPr-P-IiA IIA-P-IICB IICB-P-Glc
s A E , - p A HPr A , A - p A IICB A G6P
Fig. 3. The phosphotransferase system of Escherichia coli. (a) Schematic
representation of the enzymes; (b) kinetic scheme with numbering of all
the elemental steps. SP refers to phosphoenolpyruvate, S to pyruvate, Glc
to glucose, and G6P to glucose 6-phosphate.
.!!3
lii 2.0 . , . . - : - - - - - - - - - - - - - - - ,
'0
Ie
(I)
8 1.5
-8
ec:
~ 1.0
<+=
(I)
E
~ 0.5
c:
(I)
'0
0.0 -t----.----r----,-------r-----.------l
U'J -2 -1 o 2 3 4
log (enzyme concentration)
Fig. 4. Numerical simulation of total flux control by the phosphotransferase
enzymes. The kinetic reaction scheme shown in Fig. 3b was analysed by
numerical simulation. The sum of the flux-control coefficients of four
phosphotransferase system enzymes decreased from the maximally possible
value of two to below one with proportional increases in the concentration
of phosphotransferase enzymes. Parameter values were (dimensionless
units): k, = 200, k_, = I, k2 = I, k_2 = 134, k3 = 10, k_3 =0.1, k4 =0.04, k-4 =
0.8, k5 = 4, k_5 = 0.4, k. =0.08, k-6 =6, k7 = 30, k_7 =0.05, k8 =0.05, k~ =
0.03,k9 =5 X 10-3, k... =10-4, klO =O.OI,k_1O = 10-5 The boundary metabolites
were clamped at the following values: SP = S =Glc = I, G6P = 0.1. Initial
values of the total enzyme concentrations were: e, = 10-1.8, e2 = 4 x 10-1.8,
e 3 = 2 x 10-1.8, e4 = 8 x 10-1.8.
theory developed recently [37] provides us with an analytical
expression for the sum of the enzyme control coefficients in
terms of the fractions of the enzymes involved in enzyme-
enzyme complexes and the elemental control coefficients of
the boundary substrates and products. With an increase in
enzyme concentrations the fraction of enzymes sequestered
by complexation increases, and the sum ofthe enzyme control
coefficients decreases (see also above). Numerical simulation
(Fig. 4) of a kinetic reaction scheme representing the
phosphotransferase system shows that the sum ofthe enzyme
flux-control coefficients can decrease from the maximally
possible value of two to below one with proportional
increases in the concentrations of the pathway enzymes. The
decrease of the control by enzymes to values below one is
explained by the control residing with the boundary meta-
bolites. Importantly, essentially the same results are obtained
when the apparent increase in enzyme concentration is caused
by increasing the concentration of the background macro-
molecules, i.e. owing to the phenomenon ofmacromolecular
crowding (data not shown, cf. Fig. 2).
Experimental evidence supports the theoretical predictions
and numerical simulations described above. Table 1 shows
the dependence of the flux through the phosphotransferase
system in vitro, and of the sum of the enzyme flux-control
coefficients and the response coefficients of phosphoenol-
pyruvate and pyruvate, on the total protein concentration as

Table 1. Dependence of the phosphotransferase system flux and of the
sum of the enzyme flux-control coefficients on the total enzyme
concentration
[PEP] [Pyr] [protein] J LC
i
J
ei
+ R PEP
J + R JPyr
mM mg/ml .LIM/min
2 2 0.5 0.76 2.05
4 4 1.0 3.17 1.91
6 6 1.5 6.10 1.64
8 8 2.0 9.34 1.37
Phosphorylation of ["C]-Iabelled methyl a-glucopyranoside, an analogue
of glucose, was measured in cell-free extracts of E. coli [55]. The
concentrations of extract, phosphoenolpyruvate (PEP) and pyruvate (Pyr)
were increased in the same relative proportions; this also keeps the relative
proportions of the phosphotransferase system enzymes constant. The sum
L CJ + ~p.cn
le' Ed
+ ~Pyr was determined by differentiation
~
of the data of
flux vs. protein concentration in double-logarithmic space. L-j CJei refers to the
sum of the four enzyme flux-control coefficients of the phosphotransferase
system.
. determined in cell-free extracts of E. coli. The flux always
increased with increasing enzyme concentrations; however,
the relative increase was less sharp for the higher enzyme
concentrations. This resulted in a decrease in the total flux
control exerted by the enzymes as their concentrations
increased, and points to increased complex formation between
them.
Subtlety in the control over intermediate concentrations
The present section is concerned with the consequences of
metabolic channelling for the concentrations of intermediates
in the bulk phase (i.e. for 'pool sizes '). To answer the question
if channelling can specifically affect pool sizes, one may
compare the effect of a parameter change enhancing the direct
transfer flux in a channelled pathway to the effect ofthe same
parameter change in a corresponding pathway lacking the
channel. The dynamic channel considered above (F ig. 1) and
the classical pathway in which the two enzymes do not
interact directly will serve as a 'test system' to examine this
question, by comparing the two.
At very low enzyme concentrations when the concentration
of a dynamic complex of two enzymes can be neglected, the
channelled flux fraction is almost zero. As a consequence,
the two pathways will pass almost the same flux under these
conditions and sustain the same concentration of the bulk-
phase metabolite (X). We have seen already that a natural way
of enhancing the channelled flux fraction is to increase the
enzyme concentrations or the concentration of background
macromolecules in the medium. To answer how pool size is
affected by such a proportional increase in the enzyme
concentrations one can rephrase this question in terms of
metabolic control analysis: what does the sum of the enzyme
control coefficients with respect to pool size amount to? In
317
the traditional pathway this sum equals zero in line with the
lack of effects of a proportional increase in the enzyme
concentrations on the intermediate pool. For channelled
pathways new summation theorems apply. As above, molecular
control analysis allows one to express the sum of enzyme
control coefficients over pool size (X) into the elemental
control coefficients. When the concentrations of the two
enzymes in a channel are equal, the sum ofthe enzyme control
coefficients with respect to pool size (X) is proportional to
the channelled flux fraction multiplied by the sum of the
elemental control coefficients of steps 2 and 3 [46]:
q+C~
(10)
1 + EjXE/e
Here C\ and CXe2 are the control coefficients of enzymes 1
and 2 over the bulk-phase metabolite X, and CX 2 and CX 3 are
the elemental control coefficients of steps 2 and 3 defined in
terms of identical relative modulations of the forward and
reverse rate constants (Eq. 4). J is the total flux, J ehan denotes
the flux through the channel, and e is the concentration of either
enzyme. Eq. 10 shows that whether the pool size increases,
remains constant or decreases with an increase in the enzyme
concentrations (i.e. the sign of the sum of the two enzyme
control coefficients on the left hand side of the Eq.) depends
on whether the sum of the elemental control coefficients of
steps 2 and 3 is greater, equal to or less than zero. It has been
shown [46] that all three cases are possible, and which of them
obtains depends on the particular Gibbs energy profile of the
reaction mechanism, i.e. on the magnitudes of the rate con-
stants. One may conjecture that when step 2 is in a state near-
equilibrium (L'lG2 is close to zero), whereas step 3 is far from
equilibrium (-L'lG3 RT), the control exerted by step 3 on the
concentration X is substantial and negative, and the control
exerted by step 2 is small. From Eq. 10 one then concludes that
in this case the pool size decreases (while the channel flux
increases) with an increase in enzyme concentrations.
Our demonstration here was not for constant total flux.
However, as was mentioned above, simultaneous equal
relative changes in all the (elemental) rate constants do not
change any concentration, whereas they do affect the pathway
flux. Therefore, our results imply that the channelled pathway
and the same pathway lacking the channel may pass the same
flux (although the enzyme concentrations in the pathways
differ), whereas the pool size and even the concentration of
all species is smaller in the case of channelling [46]. We
conclude that a subtlety in control of pool size by channelling
is that this pool may either decrease or increase depending
on the Gibbs energy profile of the reactions. Consequently,
at least for some kinetic conditions channelling does provide
a mechanism for decreasing pool size (cf. [47, 48]).
Importantly, an increase in the concentrations of the
background macromolecules rather than in the concentrations
318
80
70 2 the same rate at a steady state. The average time that a water
60
Q) 50
N
'iii
(5
40
0
a.. 30
20
10
0
0 20 40 60 80 100
Increase in enzyme activities
Fig. 5. Changes in pool size with an increase in the thermodynamic
activities of pathway enzymes. The dependence of the concentration of the
intermediate X on the activity of either enzyme is shown for the dynamic
channel of Fig. I (line I) and for the analogous pathway lacking the channel
(line 2). Parameters values are given in the legend to Fig. 2.
of particular enzymes forming a channel results in essentially
the same behaviour ofthe bulk-phase intermediate. Figure 5
shows that an increase in the apparent values of the thermo-
dynamic activities of enzymes with an increase in crowding
can lead to a dramatic decrease in the bulk-phase pool size
in a dynamic channel, whereas it does not affect pool size in
an analogous pathway lacking the channel. In fact, owing to
a substantial increase in the apparent effective concentrations
of the enzymes brought about by macromolecular crowding
[7,8,43], a relatively small increase in the concentrations of
two particular enzymes forming a channel may result in quite
a significant decrease in the pool sizes of small intermediate
molecules, thus evoking a threshold behaviour of pool sizes.
Subtlety in control over temporal phenomena: Channelling
mayor may not lead to stronger control over transit time
The striking changes in the control of fluxes and concentrations
that accompany channelling have remarkable consequences for
the control over the 'transit' time of metabolites in a channelled
pathway. The transit time can be defined as the average time it
takes for a molecule entering the steady-state system as a
substrate to reach the exit point and leave the system as a
product. Fora dynamic channel in which the entry reaction
(binding step I, Fig. I) and the exit reaction (product releasing
step 4) are irreversible, the transit time (r) equals the ratio of
the total amount of intermediates in the pathway (j) and the
total flux (J) at steady state (cf. [49]):
(11)
This definition of the transit time admits a close analogy with
the transit time of a molecule in a water pool, in which water
flows in through one tube and flows out via another tube with
molecule spends in the pool, i.e. the average turnover time,
equals the amount of water in the pool divided by the flow rate.
The control exerted by an enzyme on the transit time r can
be quantified as the fractional change in r divided by the
fractional change in the enzyme concentration (e),
extrapolated to infinitesimally small change at a system
steady state:
C:. =
1
(_dr_/r_
deJe i
J = (:::: J
ss I SS
(12)
Since the transit time r is determined as the ratio ofthe total
intermediate pool (j and the fluxJ, it follows from Eq. 11 that
the control coefficient over r equals the corresponding
control coefficient over (j minus the control coefficient over
J:
(13)
Because in 'ideal' metabolic pathways the control exerted by
the enzymes on any free metabolite concentration adds up to
zero, and the control exerted on the flux adds up to one, the
control exerted by all the enzymes on the transit time equals
-1 [50,51]. This value increases when the concentrations of
enzyme-bound metabolites are taken into consideration [17,
20,52]. Therefore, for any pathway in which direct enzyme-
enzyme interactions are absent, the total control over the
transit time must be less negative than -1. This means that
doubling enzyme concentrations decreases the transit time by
a factor of two or less.
The new summation theorems developed for flux- and
concentration-control coefficients in channelled pathways
[21,37,53] predict further subtleties in control over the transit
time. We shall illustrate this with the example of the dynamic
channel (Fig. 1). For the sake of simplicity we shall consider
the case where the concentrations of the two enzymes are
equal (e 1 = e2 = e) and the concentration of the bulk-phase
intermediate (X) significantly exceeds the concentration of
enzyme-bound intermediates. Then (j in Eq. 13 approaches
X Using Eqs 9, 10 and 13, the total control over the transit
time can be expressed in terms of the elemental control
coefficients of the channelled steps as follows:
-1 +(C;+Cn-(C~+C~)
(14)
We have already seen that whether the pool size (X) and the
flux (J) decrease or increase with the activities of the
channelled steps depends on the elemental rate constants

within the enzyme mechanisms. We have previously proved
[21,46] that the control exerted by the channelled steps onX
can be negative (C~ + C~ < 0), whilst simultaneously the
control over the flux J is positive (C~ + C~ > 0). In fact, the
former expression can assume values as low as ~l, and the
latter expression can assume values as high as + 1, so that
(from Eq. 14) the sum ofthe enzyme control coefficients over
transit time can be well below ~2. In other words, doubling
the enzyme concentrations in a dynamic channel can reduce
the transit time by a factor of four or more.
It is important to note that channelling does not necessarily
result in a stronger negative control by enzymes on the transit
time. Depending on the elemental rate constants within
enzyme mechanisms, channelling may cause an increase in
the pool size, i.e. the sum (C~+ C:) can be positive [46]. In
this case, Eq. 14 shows that in a dynamic channel the absolute
value of the control over the transit time can be less than in
an analogous unchannelled pathway.
To summarise, the transit time mayor may not be shorter
in a channelled pathway when compared to an analogous
pathway lacking the channel. Whether the transit time
increases or decreases with channelling, depends on whether
the total flux and the pool size increase or decrease. Recent
studies showing that at constant flux the free metabolite pool
can increase or decrease with channelling [46,48] imply that
the same is true for the transit time.
Discussion
In this paper we have discussed the subtlety in control properties
of organized metabolism, as compared to ideal (classical)
metabolic pathways. A remarkable feature of channelled
pathways is that the control exerted by pathway enzymes on the
flux can sum to well below or much higher than one, depending
on catalytic capability of the enzyme-enzyme complex and a
subtle phenomenon of enzyme-enzyme sequestration. Experi-
mental studies of the phosphotransferase system ofE. coli have
confirmed the theoretically predicted decrease in the total control
exerted by enzymes on the flux (from about two to less than one)
with increase in enzyme concentrations.
The concentration of bulk-phase intermediates is subtly
controlled by enzyme organization. Channelling can result
in either a decrease or an increase in the bulk-phase pool sizes,
depending on a particular Gibbs energy profile of the reaction
mechanism, i.e. on the magnitudes of the rate constants.
The phenomenon of macromolecular crowding [7,43] may
playa key role in facilitating the enzyme organization, e.g.
in metabolic channelling. An increase in total protein
concentration reduces the effective solvent volume. Macro-
molecular crowding then increases the apparent effective
concentrations of enzymes (resulting from an increase in their
319
thermodynamic activities) much more than those of the low
molecular weight metabolites, in some cases to dramatic
extents [7, 43]. As was mentioned above, under crowding
conditions a small increase in the concentrations of particular
enzymes forming a channel may result in quite a significant
drop in the pool sizes of small intermediate molecules. For
organisms which are subject to transient changes in the
cytoplasmic water content, as a result of changes in the
environment, this may be important physiologically. Bio-
technologically, implications for cells forced to overproduce
macromolecules may be of interest. Some of the protein burden
effect not explained by straightforward metabolic control
analysis [54], may well be due to macromolecular crowding.
Acknowledgement
We acknowledge the BBV Foundation (Spain), the NWO (the
Netherlands) and NIH grant AA11689-01 for support of this
work.
References
1. Ovadi J, Srere PA: Channel your energies. Trends Biochem Sci 17:
445--447, 1992
2. Welch GR, Easterby JS: Metabolic channelling versus free diffusion:
Transition time analysis. Trends Biochem Sci 19: 193-197, 1994
3. Ovadi J: Physiological significance of metabolite channelling. J Theor
Bioi 152: 1-22, 1991
4. Keizer J, Smolen P: Mechanisms of metabolite transfer between
enzymes: diffusional versus direct transfer. Curr Top Cell Regul 33:
391--405, 1992
5. WesterhoffHV, Welch GR: Enzyme organization and the direction of
metabolic flow: Physicochemical considerations. CurrTop Cell Regul
33: 361-390,1992
6. Kell DB, Westerhoff HV: Catalytic facilitation and membrane bio-
energetics. In: GR Welch (ed). Organized Multienzyme Systems.
Academic Press, New York, 1985, pp 63-138
7. Minton AP: A molecular model for the dependence of the osmotic
pressure of bovine serum albumin upon concentration and pH. Biophys
Chern 57: 65-70,1995
8. Garner M: The consequences of macromolecular crowding for
metabolic channelling. In: L Agius, HSA Sherratt (eds). Channelling
in Intermediary Metabolism. Portland Press, London, 1997, pp 41-52
9. Kholodenko BN, Westerhoff HV: The macroworld versus the
microworld of biochemical regulation and control. Trends Biochem
Sci 20: 52-54,1995
10. Easterby JS: Pathway dynamics and the analysis of metabolite
channelling. In: L Agius, HSA Sherratt (eds). Channelling in
Intermediary Metabolism. Portland Press, London, 1997, pp 71-90
II. Kaeser H, Burns JA: The control of flux. In: DD Davies (ed). Rate
Control of Biological Processes. Cambridge University Press, London,
1973, pp 65-104
12. Heinrich R, Rapoport TA: A linear steady-state treatment of enzymatic
chains. General properties, control and effector strength. Eur J Biochem
42: 89--95, 1974
13. Kholodenko BN, Molenaar D, Schuster S, Heinrich R, WesterhoffHV:
320
Defining control coefficients in non-ideal metabolic pathways.
Biophys Chern 56: 215~226, 1995
14. WesterhoffHV, Arents JC: Two completely rate-limiting steps in one
metabolic pathway? The resolution of a paradox using control theory
and bacteriorhodopsin liposomes. Biosci Rep 4: 23~31, 1984
15. Jensen PR, Michelsen 0, Westerhoff HV: Control analysis of the
dependence of E. coli physiology on the H+-ATPase. Proc Natl Acad
Sci USA 90: 8068~8072, 1993
16. Cornish-Bowden A: Fundamentals of Enzyme Kinetics. Portland Press,
London, 1995
17. Fell DA, Sauro HM: Metabolic control analysis. The effects of high
enzyme concentrations. Eur J Biochem 192: 183~187, 1990
18. Kacser H, Sauro HM, Acerenza L: Enzyme-enzyme interactions and
control analysis. I. The case of non-additivity: Monomer-oligomer
associations. Eur J Biochem 187: 481-491, 1990
19. Sauro HM, Kacser H: Enzyme-enzyme interactions and control
analysis. 2. The case of non-independence: Heterologous associations.
Eur J Biochem 187: 493~500, 1990
20. Kholodenko BN, Lyubarev AK, Kurganov BI: Control of metabolic
flux in a system with high enzyme concentrations and moiety-
conserved cycles. The sum of the flux control coefficients can drop
significantly below unity. Eur J Biochem 210: 147~153, 1992
21. Kholodenko BN, WesterhoffHV: Metabolic channelling and control
of the flux. FEBS Lett 320: 71~74, 1993
22. Kholodenko BN, Cascante M, WesterhoffHV: Control and regulation
of channelled versus ideal pathways. In: L Agius, HSA Sherratt (eds).
Channelling in Intermediary Metabolism. Portland Press, London,
1997, pp 91~114
23. Kholodenko BN, Sauro HM, Westerhoff HV: Control by enzymes,
coenzymes and conserved moieties: A generalization of the connectivity
theorem of metabolic control analysis. Eur J Biochem 225: 179-186, 1994
24. Sauro HM: Moiety-conserved cycles and metabolic control analysis:
Problems in sequestration and metabolic channelling. BioSystems 33:
55--67, 1994
25. Cori CF, Velick SF, Cori GT: Combination of diphosphopyridine
nucleotide with glyceraldehyde phosphate debydrogenase. Biochim
Biophys Acta 4: 160-169, 1950
26. Friedrich P: Dynamic compartmentation in soluble enzyme systems.
Acta Biochim Biophys Acad Sci Hung 9: 159~173, 1974
27. Srivastava DK, Bernhard SA: Biophysical chemistry of metabolic
reaction sequences in concentrated enzyme solution and in the cell.
Annu Rev Biophys Biophys Chern 16: 175~204, 1987
28. Ushiroyama T, Fukushima T, Styre JD, Spivey HO: Substrate
channelling of NADH in mitochondrial redox processes. Curr Top
Cell Regul33: 291~307, 1992
29. Kholodenko BN, Demin OV, Westerhoff HV: Channelled pathways
can be more sensitive to specific regulatory signals. FEBS Lett 320:
75--78, 1993
30. Kholodenko BN, Cascante M, Westerhoff HV: Dramatic changes in
control properties that accompany channelling and metabolite
sequestration. FEBS Lett 336: 381~384, 1993
31. Kholodenko BN, WesterhoffHV, Puigjaner J, Cascante M: Control in
channelled pathways. A matrix method calculating the enzyme control
coefficients. Biophys Chern 53: 247~258, 1995
32. Kholodenko BN, WesterhoffHV, Brown GC: Rate limitation within a
single enzyme is directly related to enzyme intermediate levels. FEBS
Lett 349: 131 ~ 134, 1994
33. Brown GC, Westerhoff HV, Kholodenko BN: Molecular control
analysis. Control within proteins and molecular processes. J Theor
Bioi 182: 389-396, 1996
34. Brown GC, Cooper CE: Control analysis applied to single enzymes:
Can an isolated enzyme have a unique rate-limiting step? Biochem J
294: 87~94, 1993
35. Kholodenko BN, Westerhoff HV: Control theory of one enzyme.
Biochim Biophys Acta 1208: 294-305, 1994
36. Van Dam K, Van der Vlag J, Kholodenko BN, Westerhoff HV: The
sum of the flux control coefficients of all enzymes on the flux through
a group-transfer pathway can be as high as two. Eur J Biochem 212:
791~799, 1993
37. Kholodenko BN, Westerhoff HV: Control theory of group-transfer
pathways. Biochim Biophys Acta 1229: 256--274,1995
38. Kholodenko BN, Westerhoff HV: How to reveal various aspects of
regulation in group-transfer pathways. Biochim Biophys Acta 1229:
275~289, 1995
39. Westerhoff HV, Van Dam K: Thermodynamics and Control of
Biological Free Energy Transduction. Elsevier, Amsterdam, 1987
40. Kholodenko BN: Control of molecular transformations in multienzyme
systems: quantitative theory of metabolic regulation. Mol Bioi (USSR)
22: 1238~1256, 1988. Engl trans I 22: 990-1005, 1989
41. Giersch C: Control analysis of metabolic networks. Homogeneous
functions and the summation theorems for control coefficients. Eur J
Biochem 174: 509~513, 1988
42. Kholodenko BN: Modern Theory of Metabolic Control. Viniti Press,
Moscow (in Russian), 1991
43. Garner MM, Burg MB: Macromolecular crowding and confinement
in cells exposed to hypertonicity. Am J Physiol266: C877--C892, 1994
44. Adair G: A theory of partial osmotic pressure and membrane equilibria
with special reference to application of Dalton's Law to hemoglobine
solutions in the presence of salts. Proc R Soc Lond A120: 573--603,
1928
45. Brand MD, Vallis, BPS, Kesseler A: The sum of the flux control
coefficients in the electron-transport chain of mitochondria. Eur J
Biochem 226: 819~829, 1994
46. Kholodenko BN, Westerhoff HV, Cascante M: Channelling and the
concentration of bulk phase intermediates as cytosolic proteins get
more concentrated. Biochem J 313: 921~926, 1996
47. Mendes P, Kell DB, Westerhoff HV: Channelling can decrease pool
size. Eur J Biochem 204: 257~266, 1992
48. Mendes P, Kell DB, WesterhoffHV: Why and when channelling can
decrease pool size at constant net flux in a simple dynamic channel.
Biochim Biophys Acta 1289: 175~186, 1996
49. Easterby JS: A generalized theory ofthe transition time for sequential
enzyme reactions. Biochem J 199: 155~161, 1981
50. Melendez-Hevia E, Torres NY, Sicilia J, Kacser H: Control analysis of
transition times in metabolic systems. Biochem J 265: 195~202, 1990.
51. Easterby JS: Temporal analysis of the transition between steady states.
In: A Cornish-Bowden and ML Cardenas (eds). Control of Metabolic
Processes. Plenum Press, New York, 1990, pp 281~290
52. Cascante M, Melendez-Hevia E, Kholodenko BN, Sicilia J, Kacser
H: Control analysis of transit time for free and enzyme-bound
metabolites. Biochem J 308: 895~899, 1995
53. Kholodenko BN, Cascante M, WesterhoffHV: (1995b) Control theory
of metabolic channelling. Mol Cell Biochem 143: 151~168, 1995
54. Snoep JL, Yomana LP, WesterhoffHV, Ingram LO: Protein burden in
Zymomonas mobilis: Negative flux and growth control due to
overproduction of glycolytic enzymes. Microbiology 141: 2329-2337,
1995
55. Rohwer JM: Interaction of functional units of metabolism. Control and
regulation of the bacterial phoshoenolpyruvate-dependent phos-
photransferase system. Ph.D. thesis, University of Amsterdam, 1997
Molecular and Cellular Biochemistry 184: 321-344, 1998.
1998 Kluwer Academic Publishers.

The dynamic regulation of myocardial oxidative

phosphorylation: Analysis of the response time of
oxygen consumption

lH.G.M. van Beek, l x. Tian,2 C.J. Zuurbier, l B. de Groot, 1

C.lA. van Echteld,3 M.H.l Eijgelshoven1,3 and lB. Hak l
ILaboratory for Physiology, Institute for Cardiovascular Research (ICaR- VU), Vrije Universiteit, Amsterdam,
The Netherlands; 2Department of Pathophysiology, Datong Medical College, Datong, People's Republic of China;
3Department of Experimental Cardiology, University Hospital Utrecht, The Netherlands

Abstract
Although usually steady-state fluxes and metabolite levels are assessed for the study of metabolic regulation, much can be
learned from studying the transient response during quick changes of an input to the system. To this end we study the transient
response of02 consumption in the heart during steps in heart rate. The time course is characterized by the mean response time
of02 consumption which is the first statistical moment of the impulse response function ofthe system (for mono-exponential
responses equal to the time constant). The time course of0 2uptake during quick changes is measured with 02 electrodes in the
arterial perfusate and venous effluent of the heart, but the venous signal is delayed with respect to 02 consumption in the
mitochondria due to 02 diffusion and vascular transport. We correct for this transport delay by using the mass balance of 02'
with all terms (e.g. 02 consumption and vascular 02 transport) taken as function of time. Integration of this mass balance over
the duration of the response yields a relation between the mean transit time for 02 and changes in cardiac 02 content. Experimental
data on the response times of venous [OJ during step changes in arterial [02] or in perfusion flow are used to calculate the
transport time between mitochondria and the venous 02 electrode. By subtracting the transport time from the response time
measured in the venous outflow the mean response time of mitochondrial 02 consumption (tmito ) to the step in heart rate is
obtained.
In isolated rabbit heart we found that tmito to heart rate steps is 4--12 s at 37C. This means that oxidative phosphorylation
responds to changing ATP hydrolysis with some delay, so that the phosphocreatine levels in the heart must be decreased, at
least in the early stages after an increase in cardiac ATP hydrolysis. Changes in ADP and inorganic phosphate (P) thus playa
role in regulating the dynamic adaptation of oxidative phosphorylation, although most steady state NMR measurements in the
heart had suggested thatADP and Pi do not change. Indeed, we found with 3 1P-NMR spectroscopy that phosphocreatine (PCr)
and Pi change in the first seconds after a quick change inATP hydrolysis, but remarkably they do this significantly faster (time
constant -2.5 s) than mitochondrial 02 consumption (time constant 12 s). Although it is quite likely that other factors besides
ADP and Pi regulate cardiac oxidative phosphorylation, a fascinating alternative explanation is that the first changes in PCr
measured with NMR spectroscopy took exclusively place in or near the myofibrils, and that a metabolic wave must then travel
with some delay to the mitochondria to stimulate oxidative phosphorylation. The tmito slows with falling temperature, intracellular
acidosis, and sometimes also during reperfusion following ischemia and with decreased mitochondrial aerobic capacity. In
conclusion, the study ofthe dynamic adaptation of cardiac oxidative phosphorylation to demand using the mean response time
of cardiac mitochondrial 02 consumption is a very valuable tool to investigate the regulation of cardiac mitochondrial energy
metabolism in health and disease. (Mol Cell Biochem 184: 321-344, 1998)

Key words: heart, oxidative phosphorylation, dynamic responses, metabolic wave, creatine kinase, compartmentation,
mitochondria, adenine nucleotides, oxygen consumption, NMR, stunning
Address for offPrints: J.H.G.M. van Beek, Laboratory for Physiology, Institute for Cardiovascular Research (ICaR-VU), Vrije Universiteit, Van der
Boechorststraat 7,1081 BT Amsterdam, The Netherlands
322
Introduction: the dynamic regulation of
respiration in vivo
Measuring metabolic transients
The regulation of oxidative phosphorylation in intact muscle
tissue is usually studied in the steady state. Often a change
in energy turnover is imposed experimentally, for instance
by increasing contractile activity, and when a steady state is
obtained metabolite levels and the steady metabolic flux in
the system are measured, for instance by measuring PCr and
02 consumption (for abbreviations see Table 1). It is then
Table 1. Symbols and abbreviations
Symbol Quantity
C 02 concentration
F Flow of perfusate
M(t) 02 consumption
Q Amount of 02 in organ
"- partition coefficient,
~Cti"j~C vessel lumen at steady
state
Wi local M/F divided by
MlF for the whole heart
Ri factor proportional to
local diffusion resistance
J metabolic flux
~ difference from previous
steady state
Subscripts
ww wet weight
dw dry weight
a arterial
v venous
Metabolite abbreviations
phosphocreatine
inorganic phosphate
analyzed whether the changes in metabolite levels can
stimulate oxidative phosphorylation sufficiently to explain
the increase of ATP synthesis. However, the time course of
adaptation ofATP synthesis to quickly changingATP hydro-
lysis was studied very rarely, at least in the heart. Here we
look at methods, which were recently developed in our
laboratory to study the dynamic response of mitochondrial
metabolism to changing metabolic demand in intact tissue.
We hope to convince the reader that obtaining the time course
of biochemical fluxes non-invasively by deconvolution ofthe
measured signal with the intervening transport function does
yield valuable insight on in vivo biochemistry.
The dynamics of02 consumption and of phosphate meta-
bolite levels during quick increases in contractile activity had
been studied quite extensively in skeletal muscle, but hardly
in cardiac muscle. The reason for this was probably that the
response in skeletal muscle is much slower than in cardiac
muscle, and the time constants of metabolite contents and of
02 consumption during steps in contractile activity are usually
greater than half a minute. Thus the time course could be
studied easily with sufficient time resolution in skeletal
muscle, but not in cardiac muscle where the time constant
turns out to be much smaller. However, recently we deve-
loped a method to determine the response time of 02 con-
sumption in isolated hearts, for instance during steps in heart
rate which cause immediate changes in ATP hydrolysis [1-
3].2 consumption is tightly coupled to resynthesis of ATP
in oxidative phosphorylation, and thus directly reflects mito-
Unit
chondrial function. The 02 uptake can be measured in the
mollL organ as a whole, but in order to derive the response time at
ml/gwjmin the level of the mitochondria the time course at the venous
mollgwjmin measurement site must be corrected for the delay caused by
mollgww
diffusion between mitochondria and capillaries and by
dimensionless
convective transport in the blood vessels. This calculation
based on an 02 transport model will be described. The rather
dimensionless complex model calculation would not be necessary if one can
measure the amount ofnonmetabolized 02 present at anyone
dimensionless
time during the transient. This can only partly be accomp-
molls lished by measuring myoglobin oxygenation with near
infrared spectroscopy [4,5]. Another way to follow the time
course of oxidative phosphorylation is by measuring the heat
that is generated [6, 7], which however is only feasible at
lower than physiological temperatures.
02 consumption is one side of the coin of oxidative
phosphorylation, butATP synthesis is the other. We designed
31P-nuclear magnetic resonance (NMR) spectroscopic proto-
cols to measure the time course of the tissue contents of the
phosphate metabolites ATP, PCr and Pi with as high a time
resolution as possible (-2 s). When ATP hydrolysis is in-
creased instantaneously by an upward step in heart rate, but
mitochondrial ATP synthesis follows with some delay, a
decrease in PCr should be measurable. Because the equilib-
rium constant of the creatine kinase reaction, PCr + ADP "'"
Cr + ATP, is strongly in the direction of ATP formation and
because this reaction is fast, the deficit of ATP formation in
the early phase should then show up as a decrease in PCr and
an increase in ADP concentration. The delay in mitochondrial
ATP synthesis exists, as shown below, and is given by the
response time ("time constant") of 02 consumption, which is
usually 5 s or larger and is virtually equal to the time constant
of mitochondrial ATP synthesis. However, glycolytic ATP
synthesis may also take place and may cause a discrepancy
between the time constant of the phosphate metabolites on
the one hand and of 02 consumption on the other. For this
reason the transient behavior of lactate efflux as a possible

result of glycolytic ATP synthesis to buffer per is also
discussed.
Regulation of cardiac oxidative phosphorylation in vivo
In skeletal muscle it had been found that per decreased and
Pi increased when contractile activity was increased [8, 9].
Sometimes it was also reported that per decreased and Pi
increased in the heart for substantial increases in energy
turnover [10], but often Pi and per (and by inference ADP)
were found to be unchanged in the steady state when 02
consumption was increased and phosphate metabolites
measured with NMR in the intact organ [11, 12]. The classic,
simple scheme for regulation of muscle metabolism is as
follows: ATP is broken down in the myofibrils during
contraction and the breakdown products ofATP, i.e. ADP and
Pi' diffuse to the mitochondria where ADP and Pi stimulate
oxidative resynthesis of ATP because they are direct sub-
strates for oxidative phosphorylation [13]. In the simplest
schemes it needed further to be taken into account that
changes in ADP in muscle are buffered by fast transfer ofthe
phosphoryl group of per, catalyzed by the enzyme creatine
kinase, leading to resynthesis ofATP, so that per rather than
ATP would decrease, see Fig. 1. For lack of change in the
pi
j, DpJI
p r
j{
A P P r
4 Increased ATP hydrolysis
p.
I I
Mito hondrion M otibril
Fig. 1. Simple classical model of ATP production and transport in muscle
cells. In the myofibrils, the contractile elements, A TP is broken down to
ADP and inorganic phosphat~ (P). ATP is quickly re-synthesized from
phosphocreatine (PCr), catalyzed by the enzyme creatine kinase, among
others in the cytosol, so that PCr rather than A TP declines during elevated
work loads. ADP and Pi diffuse to the mitochondria where they stimulate
oxidative phosphorylation. This simple scheme can be compared to schemes
in later figures on more up to date theories.
levels of phosphate metabolites in cardiac muscle, ADP and
Pi seemed not to be the regulators of oxidative phosphory-
lation in the heart during transitions in contractile activity. On
the other hand the considerable delay we found for the
adaptation of 02 consumption and mitochondrial ATP syn-
thesis, time constant greater than 5 s [3], must mean that the
high energy phosphates are decreased at least transiently.
Below we will see that the discrepancy we found between the
323
time constant of the phosphate metabolites on the one hand
and the response time of 02 consumption on the other
suggests that the phosphate metabolites may be compart-
mentalized in the cytosol. Thus the characterization of the
dynamic response of mitochondrial 02 consumption and
cytosolic phosphate metabolites sheds new light on the
organization of the cytosol. We will discuss these findings in
the light of theories for transport of high energy phosphate
groups through the cytosol, and will investigate whether the
data are consistent with the new theory of the metabolic wave,
which holds that oscillations of local concentrations ofAD P
and creatine are propagated through cytosolic micro-com-
partments and are coupled via cytosolic creatine kinase and
adenylate kinase [14].
Determination of the response time of
mitochondrial oxygen consumption
General strategy of the measurement
The 02 uptake can be readily measured at the level of the
whole heart by measuring the arterial and venous [OJ with
high time resolution and multiplying the arterial-to-venous
difference with the perfusate flow. This 02 uptake at the level
of the heart as a whole is delayed with respect to 02 con-
sumption in the mitochondria, due to diffusion between
capillaries and mitochondria and to the transport by the
perfusate flow through capillaries, venules and veins to the
site of the venous 02 electrode (Fig. 2). It appears that
~-----------------------~
,
I
I I
I I
Mitochondria
H,m M'I I
Atr ~V'
ery em ~venousLt:
0
~
I
c",;n";,, [2J
o
time -
Fig. 2. Schematic representation of the measurement of the transient of0 2
consumption during a step in heart rate. At t=O the electrically paced heart
rate is increased to a higher level. As a consequence the cardiac mitochondria
increase their 02 consumption gradually. The disturbance of the local 0,
concentration at the mitochondria is transmitted by diffusion and by vascular
transport to the venous exit of the heart, where an 02 electrode measures
the 0, tension, proportional to 02 concentration in a crystalloid perfusion
solution.
324
deconvolution of the venous 02 signal to obtain the full time
course of 02 consumption at the level of the mitochondria
would be very inaccurate for lack of detailed enough knowl-
edge of the 02 transport function. However, it is possible to
determine the mean response time of0 2consumption, which
is equivalent to the mean time (first statistical moment) of the
impulse response function, and gives the average delay of the
extra 02 consumption above the steady state value which is
caused by a brief pulse of extra ATP hydrolysis.
The theory to determine the mean response time is based
on mass balances and is mathematically and conceptually
very similar to indicator dilution theory which describes the
mean transit time of nonmetabolized molecules through a
blood vessel system [15-18]. The theory is explained in the
next paragraph, and may be a little bit intimidating to the non-
mathematically inclined readers. Therefore, a summary in
non-mathematical terms will be given in the paragraph
entitled "Non-mathematical explanation of the response
times and transit times" to which the reader may choose to
jump.
Theoretical basis of response time: integration of mass
balances
The uptake of oxygen in the heart can be measured by
monitoring the 02 concentration in the blood or saline
solution flowing into and out of the organ. However, during
a metabolic transient the 02 concentrations in the tissues also
change, and may contribute to 02 consumption. The mass
balance of oxygen for the heart as a whole is:
rate of 02 consumption =
02 brought in via arterial inflow - 02 carried out via
venous outflow
- rate of change of amount of 02 inside the organ.
The decrease in amount of 02 inside the organ must be
added to 02 taken up from the perfusate to obtain 02 con-
sumption at that moment. Diffusion of0 2 across the epicard-
ial and endocardial surfaces ofthe heart is usually negligible
relative to vascular transport [19, 20].
In the symbols of Table 1 the equation becomes:
M(t) = F '[C (t) - C (t)] _ dQ (1)
a v dt
Ifwe measure the uptake of0 2 from the perfusate, F '[Ca(t)
- Cy(t)], and if the amount of 02 inside the organ could be
measured as a function oftime, the simple Equation 1 would
suffice to calculate the 02 consumption inside the organ from
moment to moment. The 02 bound to myoglobin and hemo-
globin can be measured using near infrared spectroscopy, but
is not easy to quantify [21]. Furthermore, in a hemoglobin-
free saline perfused heart most of the 02 is not bound to
myoglobin, but is physically dissolved, as shown below. The
02 tension, P02' is directly proportional to the concentration
of physically dissolved 02 and is the force for diffusion into
tissue. The physically dissolved 02 can be measured locally
with an 02 electrode, but because ofthe large gradients of0 2
tension (e.g. from arterial to venous side of the vascular bed)
this single measurement is very hard to translate in the amount
of0 2present in the heart as a whole. We solved the problem
by using a model for 02 tension gradients in the tissue. We
could show by integrating the mass balance, Equation 1, over
time, that it was not necessary to solve the complex time
course of the 02 tension gradients in the model caused by
changes in 02 consumption. Equation 1 is rearranged and
integrated over the time during which 02 consumption M is
changing, for instance after a step in heart rate at t=O:
I[l-
00 LlC (t)
LlC)oo) J-dt-
00 LlM(t)
j[l- LlM(oo) J-dt
1 00 dQ LlQ
(2)
= F' I[ LlCJoo) ] = FLlCJoo)
Here LlC yindicates the change in venous concentration from
the steady state value before the step in heart rate, and LlM
the change in 02 consumption.
All terms have been divided by F[-LlC~oo)], which equals
the change in steady state 02 uptake from the perfusate, and
is the same as LlM( 00), the change in 02 consumption when
the new steady state has been reached. The leftmost term has
the dimension time and is called the venous response time,
tv' to a change in metabolism. Now we define'
00
tmilo = J[l LlM(t) ]'d (3)
o LlM(oo) t
This is the characteristic time ofthe response of metabolism
to the change in heart rate (or other stimulus of energy
turnover) at t=O. For a simple exponential response of 02
consumption, tmito is equal to the well known time constant.
For simplicity one may consider tmito to be a time constant,
but the precise definition of Equation 3 should be used for
more complicated responses than a mono exponential curve.
For a linear system this mathematical definition is equi-
valent to the first statistical moment (mean time) of the
impulse response function. The impulse response of a system
to a sudden brief disturbance of an input, e.g. an electrical
potential difference applied very briefly at the input of an
electrical circuit, is a method used by engineers to chara-
cterize the performance of a system. Because the math-
ematical definition for tmlto
. has the same form as for the mean
transit time of an indicator through the organ after a step
change in arterial infusion of the indicator, tmito has been called
the mean response time of metabolism [15]. However, the

mathematical definition of tmito is also equivalent to that for
the transient times which have been used to characterize
transitions of metabolite concentrations in biochemical
systems [22]. Please note how the integration of a mass
balance over the duration of the response yields the definition
of response times.
In Equation 2 the change in total amount of0 2in the organ,
AQ, normalized by the change in venous oxygen con-
centration AC v(00), is proportional to the difference between
tv and tmito ' The AQ is the volume integral of the local changes
in concentration, AC.
(4)
The normalization by ACv(oo) gives Equation 4 the math-
ematical form of a volume of distribution, a quantity well
known from indicator dilution theory which considers the
transport of molecules in the blood across an organ. Normally
Vd' the true volume of distribution in indicator dilution theory,
is determined by changing the arterial concentration and
measuring the venous concentration response. Here the
indicator molecules travel through the arterial vessels, the
microvascular exchange region and finally through the veins.
The changes in oxygen concentration resulting from in-
creased metabolism are not transmitted through the arteries
and on average are only transmitted through half the micro-
vascular exchange region to reach the venous exit ofthe heart.
To emphasize this difference in space to be traversed, the
"volume of distribution" for a change in metabolism is
indicated with the subscript m. V d,m is the volume of dist-
ribution between mitochondria and the venous site of 02
measurement, Vd is the volume of distribution between the
entrance of the coronary arterial tree and the same venous
measurement site.
Equation 2 may now be written as
t =t - Vd,m = t _.l.. AQ (5)
mito v F v F ACJoo)
showing that the response time of metabolism to a step
change in a stimulus of metabolism, such as heart rate, can
be calculated from the response time of uptake of the
metabolite from the blood if the change in amount of
metabolite in the organ caused by the change in metabolism
is determined. The change in amount of metabolite must be
divided by the change in metabolism F [-AC/oo)]. The last
term in Equation 5 represents the delay between the mito-
chondria and the venous measurement site, has the dimension
time, and is called the transport time:
t = Vd,m =~ . AQ (6)
transport F F ACv(oo)
325
The measured venous response time during a step in heart
rate can thus be corrected for 02 transport to give the mean
response time for changes in metabolism inside the organ
tmilo = tv -ttransport (7)
Direct determination of AQ by measuring the amount of 02
in the organ is not possible (see below), but it can be calcu-
lated from a model of the 02 tension profiles in tissue. On
the other hand, the AQ after either a step change in arterial
[02] (keeping perfusion flow constant) or after a step change
in perfusion flow (keeping arterial [02] constant) can be
measured from the washout of 02 by integrating the extra
venous [02] exceeding the steady state level for the current
02 consumption, arterial [02] and perfusion flow after the step
in 02 supply [3]. Using such washout data, the AQ for the
change in metabolism during the heart rate step can be
deduced from the AQ obtained from the washout of 02
following the arterial concentration step or the flow step. This
is done using a model for the 02 concentration profiles in
tissue which contains the concentration gradient from arteries
to veins, the diffusion gradient from blood vessels into tissue
and heterogeneity of tissue perfusion [3]. The AQ in the 02
profiles model is normalized to ACV<00), the change in venous
[02]' yielding the volume of distribution for 02' V d,02' for
arterial concentration changes
where V a and Vv denote arterial and venous intravascular
volumes, respectively, from which 02 diffusion is negligible.
Vi gives the volumes of microvascular exchange regions; N
distinct regions are summed to allow for heterogeneity in the
tissue. The partition coefficient Ai is the ratio of the 02
solubility in the tissue divided by the 02 solubility in the
perfusate; Ai may increase at low tissue 02 tensions where the
myoglobin-oxygen dissociation curve becomes steeper,
increasing the apparent 02 solubility. However, it should be
noted that in the model the relation between 02 tension and
02 concentration is linearized.
The "metabolic volume of distribution", Vd,m' which is
needed to correct the response time of 02 uptake measured
in the veins to obtain the true response time of 02 con-
sumption at the level of the mitochondria, see Equation 6,
differs from V d,02:
NI N
Vd,m =V+L
v
-VA,W.+LVw
;=1 2 i=lI
.. R.
I I I I I
(9)
Here Wi is the local metabolism-to-perfusion ratio in each
tissue unit, M/Fi' normalized by dividing it by the overall
metabolism-to-perfusion ratio, MIF. The weight Wi models
326
the heterogeneity of metabolism and perfusion; perfusion
heterogeneity has been found both in blood perfused myo-
cardium [23] and in saline perfused heart [24], but a method
to measure local 02 consumption using 13C-NMR has only
been developed very recently [25]. The heterogeneity of
perfusion can be described with fractal models [26,27], but
flow heterogeneity within 500 )..lm regions can be neglected
[28]. Initial results suggest that local flow and 02 con-
sumption are matched, so that Wi may be assumed to be equal
to 1. The factor Ri is proportional to diffusional resistance
from capillaries into the tissue. Using the Krogh model for
radial diffusion into tissue we found that Ri is usually
negligible with respect to the other terms [3]. The factor 112
in the second term on the right hand side indicates that the
distance between the sites of0 2 consumption and the venous
segment is on average half the length of the microvascular
exchange region (see Fig. 2). The full derivation of Equation
9 is given elsewhere [3]. Comparing Equations 8 and 9 shows
that the arterial volume and half the volume of distribution
of the exchange region need to be subtracted from Vd 02 to
obtain Vd,m' ' :::
This latter calculation of V d,m from the response to an
arterial concentration step is relatively complicated. How-
ever, a downward step in perfusion flow would produce an
identical arterial-to-venous 02 concentration gradient as an
increase in oxygen consumption. According to the model for
the 02 concentration profiles in tissue, the "volume of
distribution" for a flow step, Vd,!' is
i1Q _ N 1
Vd,j= i1C (00 ) - Vv + Li=l -2 V . A . w.
I I I
(10)
Vd,[is identical to Vd,m except for the radial diffusion gradient
term. If the extra venous oxygen concentration above the
level dictated by the existing 02 consumption and flow is
integrated over the transient response of venous 02 con-
centration after a step in perfusion flow, we have determined
Vd,fwhich is then set equal to Vd,m' Experimental evidence that
these extensive transport calculations are valid is presented
below, The correction for oxygen transport allows us to look
inside the intact heart, thus providing a means to study in vivo
biochemical transients non-invasively.
The derivation of the transport theory has been given in
detail before [3], and its similarity to indicator dilution theory
has been discussed [15]. In the next paragraph we try to
explain the theory in as non-mathematical terms as possible.
(Almost) Non-mathematical explanation of the response
fumes and transit times
After a heart rate step, 02 consumption increases not immedi-
ately, but rises slowly (Fig. 3), We characterize the slowness
of the increase with the response time, tmita' which is prop-
ortional to the hatched area above the curve of 02 con-
sumption. A special case would be that the response is
exponential (first rises steeply but then gradually levels off
to reach its final value very gradually), An exponential curve
is characterized by an RC-time or time constant, which is
equal to the time it takes to rise to 63% of the full response.
For an exponential curve the tmito equals the time constant. The
reason that we like to use tIDlto ' rather than the time constant is
that the response may be more complicated than an exp-
onential, see for instance Fig. 3. In fact we see this quite
clearly when we measure the time course of venous [OJ after
the heart rate step no change is measured for several seconds,
and only then the concentration starts to drop. The venous
response time, tv' is again proportional to the shaded area in
Fig 3. The difference between tv and tmito is the average time
taken for 02 molecules to travel between mitochondria and
the venous 02 electrode, which we call the transport time,
ttransport . It turns out that ttranspoIi is equal to the change in the
,-
0
I
'R I
6'
s::I
1~M(oo)
1
[Jl
:::
0
()
0
I
::: 0
1
0
.~
Ol~ ACv(oo)
C1)
()
:::
0 Cv ifOz amount were constant
()
6' 0
o time ---..
Fig, 3, Scheme of transient responses of oxygen consumption, venous
oxygen concentration and amount of oxygen inside the organ during an
upward step in heart rate at t=O, see Fig, 2, Because the consumption and
concentration have been normalized to 0 and I for the steady states at the
begin and end of the response, respectively, the total grey, hatched area is
the mean response time of 02 consumption, t milo ' and the measured venous
response time, tv' respectively, The 0, concentration, normalized according
to Equation 2, increases to 1, although the concentration in mol/L decreases,
In the middle panel the venous 02 concentration is indicated for the
hypothetical case that the amount of 0, in the organ is not changing, In
that hypothetical case the response time of venous 02 consumption would
be equal to t mito ' Because 0, content in the organ is decreasing and is also
used for 0, consumption during the transient (lower panel) the transient of
the venous 0, concentration is delayed, indicated by the hatched area
between the two C curves, which equals the transport time due to diffusion
y
and to vascular transport. By subtracting the transport time, which equals
the decrease in amount of 0, inside the organ divided by the increase in
oxygen consumption, the response time of 0, consumption is found from
the measured venous response time,
 327

amount of 02 present in the organ after the increase in heart
rate, divided by the change in 02 consumption caused by the
heart rate step (for details and derivation, see above). Further,
we can calculate the change in the 02 amount using an 02
transport model. This model for the 02 amount can be
calibrated by measuring the time course of venous [OJ after
a step in perfusion flow or after a step in arterial [OJ Thus
we have all the ingredients to calculate the "time constant"
of oxygen consumption in the mitochondria, tmito:
t.
milo
=t-t
v transport
Experimental determination o/the response time 0/2
consumption
Applying the transport theory explained above, the response
time of mitochondrial 02 consumption was determined in
rabbit hearts. Because we perfused isolated hearts by pump-
ing a constant flow of crystalloid solution, the coronary flow,
which had to be assumed constant for the theoretical develop-
ment, could be precisely controlled. The cardiac work load
was precisely imposed by pacing the heart. 02 concentrations
in coronary inflow and outflow were monitored with 02
electrodes, which had time constants of }-1.5 s. The setup
allowed fast switching of the arterial concentration and fast
withdrawal of part of the flow rate to create 02 supply steps
for determination of t transport . Although we tried hemoglobin
containing perfusates to optimize 02 carrying capacity,
perfusion attempts with whole blood, red blood cell suspen-
sions and with purified hemoglobin solutions did not yield
cardiac function good enough for determination oftmito [29].
The measured venous response time of02 uptake to heart
rate steps was 17.6 1.1 (SEM) s in empty beating rabbit
hearts at 37C (Fig. 4), from which the transport time,
calculated from arterial [02] steps and perfusion flow steps
to be 9.9 0.9 s, was subtracted. This yielded 7.7 0.7 s for
the response time of 02 consumption to 40 beats/min steps
in heart rate in the range 150--260 beats/min [3]. The response
times to upward and downward steps in rate were not
significantly different. An overview ofthe response times of
mitochondrial 02 consumption under various conditions is
given in Table 2, and many ofthese conditions are discussed
below.

""-:,;:.-- - - ---
c 17.rL2~~~----;;:":-
Table 2. Response time of cardiac mitochondrial oxygen consumption to
0 ~
'is end of steps in cardiac work load, t milo . In two early studies, indicated with an *,
to
L numerical the response time was not corrected for the time course of cardiac work
j.l
integration and t milo is underestimated by up to several seconds. For the rest of the
C
W
0 studies t mito in this Table has been corrected for the overshoot of RPP, and
C
0 may therefore differ from the value reported in the original paper. The RPP
0
ru is the product of heart rate and systolic left ventricular pressure, and is a
0 measure of cardiac work and energy turnover, which is linearly related to
10 cardiac oxygen consumption in the steady state. After a step in heart rate,
:J
0 0 which leads initially to a step in RPP, the LV pressure tends to decline,
c
w which is accounted for by subtracting the response time of the RPP signal.
>
1) This latter response time is negative because of the secondary gradual
W
N decline of pressure usually seen after a heart rate step [85). art. = arterial.
to ,
E
L t=D Temperature Condition Heart rate range tmito(s), Reference
+
0
Z (beats/min) mean SE no.
step in __ time ~ = 5 (sl
heart rate
37C normal 150--260 7.7 0.7* 3
28C low temperature 96-130 16.8 1.0 30
Fig. 4. The normalized venous 02 concentration after an upward step in 20C low temperature 44--69 27.53.0 30
heart rate step from 150 to 190 beats/min, at 37C in empty beating isolated 28C normal, lactate 60--120 10.4 0.5* 39
rabbit heart. The 0, concentration was normalized to 0 at the beginning of measurements,
the 02 transient (corresponding to -480 mmHg 0, tension; tension is <6s
tlactate

proportional to the 0, concentration) and to I at the maximum amplitude 28C normal pH 90--115 14.4 1.8 85
of the 0, transient (corresponding to -450 mmHg). Note that due to the art. pH 6.6, 84--108 22.4 0.4 85
normalization the response has been inverted. Using the normalized 40--75 min after
concentrations the measured venous response time tv is found by deter- start perfusion
mining the hatched area, corresponding to the first term of Equation 2, art. pH 6.6, 84--108 35.9 2.8 85
which is the time integral defining tv' In this case tv equals 16.3 s. After 85-120 min after
subtracting the transport time, which was equal to 7.6 s for this heart, the start perfusion
response time of mitochondrial 0, consumption t mito was found to be 7.2 s. 28C control 88-111 13.5 0.7 55
In this example the 0, concentration shows a second phase changing slowly low dose 89-110 13.5 1.2 55
in the direction of the baseline after about 70 s, which is sometimes seen at Ruthenium Red
this temperature. (Permission of the American Physiological Society to high dose 89-110 22.2 3.4 55
reproduce this figure from [3) was granted). Ruthenium Red
328

Temperature Condition Heart rate range t mito (s), Reference

Comparing the response time of 02 consumption with an
(beats/min) mean SE no. independent heat rate measurement

28C glucose 73~98 12.8 0.7 55 The correction ofthe measured venous response time is rather
pyruvate 73~96 11.8 1.4 55
complicated (see Equations 1-10), and an independent test
28C low heart rate 60--70 7.6 0.6 60
bigger step 60--120 12.1 1.0 60 of the correction for 02 transport was desirable. An independ-
reversed step 120--60 12.0 1.9 60 ent way to measure the time course of oxidative phosphory-
high heart rate 120--160 16.8 2.6 60 lation is determination of the rate of heat production after
LV volume step 60 13.3 1.7 60 contraction of myocardial tissue. This had been done at 20C
28C control 60--70 7.2 0.6 92
in papillary muscle isolated from the rabbit heart and the heat
60--120 14.3 0.6 92
4 min 0.2 ml/L 60--70 10.0 1.6 92 rate due to oxidative phosphorylation decayed with a time
oligomycin, constant of 25 s, after correction for heat transport by
-50% reduction conduction [6]. Because the heat measurements were not
of state 3 V 02 feasible at temperatures above 20C, we determined tmito from
60--120 13.8 1.7 92
the time course of 02 uptake in isolated rabbit heart also at
28C small rabbits, in 60--120 15.5 1.5 33
NMR 20C and found 26.9 3.0 s [30]. Thus, heat rate measure-
spectroscope ments gave an answer that was indiscernible from the
3JOC small rabbits, in 100--200 12.4 l.l 33 response time of 02 consumption, suggesting that the deter-
NMR mination oftmito from venous 02 measurements gives reliable
spectroscope
results.
28C glucose 60--70 8.10.9 61
60--140 13.42.5
pyruvate 60--70 5.4 1.4
60--140 10.1 1.2 Measuring the time course of tissue oxygenation directly
lactate 60--70 6.1 1.4 with near infrared spectroscopy
60--140 11.51.5
3JOC glucose 120--140 6.3 1.0 61
120--220 7.4 0.9 The complicated calculation given above in Equations 2-10
pyruvate 120--140 4.0 0.7 would not be necessary if the amount of 02 in tissue could
120--220 6.5 0.6 be measured directly, because then the contribution of 02
lactate 120--140 5.4 1.2 derived from the stores inside the organ in addition to the 02
120--220 7.0 0.9
directly supplied from the perfusate (see Equation 1) would
28C inhibited 60--120 8.9 0.8
glycolysis +
be known. Using near infrared spectrophotometry (NIRS)
pyruvate with two wavelengths the tissue oxygenation can be measur-
28C control 60--70 (upward) 11.2 0.6 91 ed in hemoglobin free perfused rat heart [5]. This relatively
60--120 (upward) 14.9 0.7 simple spectrometer cannot separate the signals resulting
70--60 10.5 0.6
from myoglobin and cytochrome c oxidase. However, we
(downward)
120--60 12.4 0.6 infused potassium cyanide in rabbit and rat heart, which leads
(downward) to full reduction of cytochrome c oxidase while retaining full
after 25 min 60--70 and myoglobin oxygenation. Potassium cyanide does not inter-
ischemia 60--120 unchanged fere with near infrared absorption. It was shown that three-
from
quarters of the NIRS signal is attributable to myoglobin
control

37C control 100--230 4.3 0.3 32 oxygenation [4]. We calculated the change in amount of0 2
after one hour 100--230 24 12% in tissue during steps in heart rate with the 2model discussed
normoxic increase above from the venous 02 tension measured with the 02
perfusion electrode. Given that rabbit heart contains 160 nmol myo-
after 15 min 62 10%
globin per ml [31], we estimated from the NIRS that myo-
ischemia increase
after 15 min 64 18%
globin provided only -20% of the calculated total change in
hypoxia increase 02 content. Because the remaining -80% consists ofphysi-
cally dissolved 02 which could not be measured with NIRS,
we were not able to determine ttransport directly by inserting the
measured change in 02 in Equation 6. Given the large oxygen
gradients and perfusion heterogeneities present in tissue, it
is also not practical to measure physically dissolved 02 with
electrodes. Further, inserting the electrodes in tissue would

probably disturb local tissue perfusion. Unmetabolized
radioactive oxygen tracers in 02 and in metabolic water
cannot be discerned by their radioactive emissions and NMR
spectroscopy of 170 has little sensitivity. Determining the time
course of02consumption directly using Equation 6 may only
be possible with NIRS at low tissue 02 tensions where most
02 is derived from oxymyoglobin. NIRS can be done with
more sophisticated equipment using more wavelengths to
allow independent assessment of myoglobin oxygenation and
of cytochrome c oxidase redox state, although the cost of such
equipment is considerable.
Consequence of delay in mitochondrial ATP synthesis for
the high energy phosphates
As indicated above, the first determination of tmito at 37C
gave a value of - 8 s [3]. In later studies at the same temp-
erature the response time in isolated rabbit hearts developing
pressure in a water-filled balloon in the left ventricle tended
to be somewhat lower. 4-{) s [32], but in younger rabbits with
lower body weights and smaller hearts the response time
might be longer, -12 s [33]. InTable 2 results of many studies
on the response time have been compiled. All these findings
indicate that there is a substantial time delay before mito-
chondrial 02 consumption is fully activated after a step in
heart rate. In isolated mitochondria it was found that 02
consumption and oxidative phosphorylation of ADP occur
within 100 ms of each other [34, 35]. Thus, the activation of
mitochondrial ATP synthesis after a step in heart rate takes
usually more than 4 s, which is mostly due to retardation in
the cytosol, given the intramitochondrial response time ofless
than 100 ms.
ATP hydrolysis, measurable as PCr hydrolysis due to the
fast phosphate group exchange to ATP in the creatine kinase
reaction, takes place instantaneously during the contraction
of heart muscle, and is therefore already fully changed during
the first contraction affected by the acceleration of heart rate.
This is clearly demonstrated when one measures the heat
developed in cardiac muscle by placing a papillary muscle
on a thermopile [6, 7], because the muscle's temperature goes
up immediately during a twitch, and the heat rate, calculated
from the temperature by correcting for heat conduction to the
environment, shows a large peak that goes down already
during force development. The initial heat peak measured
represents the enthalpy of the splitting of PCr during iso-
metric contraction [6] The immediate heat release from PCr
breakdown during contraction shows that high energy
phosphate breakdown in the heart responds instantaneously
to changes in contraction pattern and heart rate. We conclude
that ATP hydrolysis responds in the first accelerated heart
beat, but aerobicATP synthesis is activated with considerable
delay. Unless anaerobicATP synthesis is activated quickly to
329
fill the gap between ATP hydrolysis and aerobic synthesis,
we predict that high energy phosphate content must go down,
at least during the first half minute after a step in heart rate.
This prediction was tested by measuring the PCr, ATP and Pi
contents in the tissue with high time resolution using NMR,
see below.
Other dynamic measurements of
metabolism besides oxygen
Time resolved 31 P NMR spectroscopy of heart rate steps
The signal-to-noise ratio ofNMR spectroscopy is low, and
in order to be able to measure phosphate metabolites with -2
s time resolution the steps in heart rate were repeated 64 times
in each heart. The NMR acquisition was synchronized with
the electrical pacing and the NMR signals at fixed times after
the heart rate steps were averaged to increase the signal-to-
noise ratio [33]. In Fig. 5 we see that PCr decreases and Pi
increases immediately after the heart rate was accelerated,
suggesting that ATP synthesis follows ATP hydrolysis with
appreciable delay. The opposite behavior is seen when the
heart rate is decreased. This indicates that not only aerobic
ATP synthesis, but also total ATP synthesis is delayed with
respect toATP hydrolysis, causing the decrease in PCr levels.
A modified 02 electrode was designed that could be
inserted into the NMR magnet without loss of NMR signal
[36], and the time course of02consumption was determined
simultaneously with the time course of the phosphate meta-
bolites. One would expect that ATP synthesis has caught up
with the increased level of ATP hydrolysis at the moment
when 02 consumption has reached a new steady state after
the increase in heart rate, and not before. Instead we found
that PCr and Pi seem to stabilize much sooner than the 02
consumption as estimated from tmito ' the response time or
"time constant" of02consumption, which is 12.4 1.1 s [33],
corrected for 02 transport. The time constants estimated for
the changes in Pi and PCr are 2.3 1.2 sand 2.7 2.2 s,
respectively, much shorter than tmito .
The delay in increase ofATP synthesis after the change in
ATP hydrolysis means that the amount of high-energy
phosphate is decreasing. This is described in the following
equation [3]:
tJ.(PCr + ATP) = r[tJ.JATP_synthes;s(t)-tJ.JATPhYdro1ys;s(t)]-dt (11)
where J means the metabolic flux, and tJ.(PCr + ATP) means
the change in PCr plus ATP content from t=O. The terminal
phosphate group of ATP is exchanged very rapidly with
creatine to form phosphocreatine (PCr) so that in practice one
330

finds a change in PCr only [3]. Therefore, for a step change calculated the change inATP synthesis, Am, from the change
in ATP hydrolysis of amplitude Am, at t=O: in 02 consumption using an ATP/O ratio of 2.4 measured in
intact heart [37] to find that the decrease in PCr should be
AJ (t)
A(PCr) =-Am . HI
00

o
A TP-synthesls
Am
J-dt=-Am . t . (12) 9.3 /lmol/g dw, which is five times the decrease in PCr
milo actually measured with the NMR, see Fig. 5.
At 28C t mito was 15.5 s, and the time constant for the
This follows from the assumption that ATP synthesis is change in inorganic phosphate Pi and PCr, was 5 s; the change
immediately and linearly coupled to 02 consumption so that in PCr predicted from tmito was 5.2/lmol/g dw, while this was
the response time for ATP synthesis equals the response time 2. 7/lmol/g dw measured by NMR at 28C. The discrepancy
for 02 consumption. It is assumed for the moment that between predicted and measured changes in PCr in the first
glycolytic ATP synthesis does playa negligible role. For a half minute after a step in heart rate could be explained by
heart rate step from 100 to 200 beats/min in isolated hearts anaerobic glycolyticATP production in the first seconds after
from young rabbits at 37C t mito was 12.4 s [33], and the the heart rate step which stops again by the time 02 con-
increase in 02 consumption was 9.4 /lmol/g dw/min. We sumption has reached the level that matches the increased
ATP hydrolysis [3]. Such a glycolytic burst had already been
A found after the start of contraction in dog red skeletal muscle,
4 ~----------.-----------,
,-... by biochemical metabolite assays in quickly frozen tissue
~ 3
II!!I!Ir IlfhI1~ l [38], see below.
~:1.
'-'
is::
2
1
~!
r ~\Il14A1111ll
' Lactate ejJlux measurements
:
0
: Based on the hypothesis that a transient burst of glycolytic
: B ATP production and lactate synthesis explains the discrep-
26
ancy between the time courses of total ATP synthesis and
~ 24 mitochondrial ATP synthesis, the coronary venous efflux of

I:1.
'-'
22
20
~:IrJ!n!11hI]1111]~1!!~luJ~lullll
11
lactate during a heart rate step was measured in venous
samples taken every 2 s, in isolated rabbit heart at 28C [39].
. , When the heart rate was doubled from 60 to 120 beats/min
~ 18 ,
the lactate efflux increased from 0.23 0.10 to 0.45 /lmol'
16 g-I'min- 1 with a response time at the level of the venous
,-..
co C outflow of21 s. The transport time for lactate is not precisely

I~
.9 360
400
380
know, but is very likely larger than the 16 s for 02 transport
in the same preparation. The response time of lactate prod-
uction in the cytosol after a step in heart rate is therefore
probably less than 6 s. The increase in lactate efflux during
j 340 the heart rate step was gradual without the overshoot that was
5 320 predicted by the glycolytic burst hypothesis. However, in dog
....~ 300
co
skeletal muscle the intracellular lactate concentration was
0 80
o~ 40~ increased 0-5 s after start of contraction by electrical
200 BPM 100 BP stimulation, after which tissue lactate content decreased
again, but lactate efflux in the blood showed no overshoot
Time (s) and apparently most of the lactate stayed in the muscle [38].
Thus it cannot be excluded by measurements in the venous
Fig. 5. Time courses of (A) P; content, (B) PCr content, (C) venous 02 effluent that there is a glycolytic burst producing the extra
tension during heart rate steps imposed on isolated rabbit hearts at 37C.
ATP required to explain the simultaneous existence of a large
The hearts contracted against a water-filled balloon in the left ventricle.
NMR acquisitions were obtained every 1.8 s. Data points are averages of tmito' 12 s, and short time constants for the Pi and PCr change,
448 heart rate steps equally divided over 7 isolated hearts. At t=O the heart 2.5 s. However, if a glycolytic burst takes place the pyruvate
rate was 100 beats/min, BPM; at the arrows heart rate was switched to the and lactate produced must remain intracellular to explain the
indicated value. The heart rate steps were repeated cyclically, and NMR venous lactate measurements.
acquisitions averaged at each time point shown. Metabolite contents were
Glycolytic enzymes are located in and near the contractile
quantified by comparison with an external reference. Bars give SE.
(Reproduced from Circulation Research [33]. permission of the American elements [14]. Thus the machinery to resynthesize ATP
Heart Association was granted). glycolytic ally in the immediate vicinity of the myofibrillar

ATPase is present. This would buffer changes inADP and Pi
that could otherwise stimulate the mitochondria almost
immediately, see Fig. 6. It is therefore possible that the first
changes in PCr and Pi seen in the NMR (Fig. 5) took mostly
place close to the myofibrils and that changes close to the
mitochondria were delayed by glycolytic resynthesis ofADP
and Pi to ATP. This would delay stimulation of oxidative
phosphorylation by ADP and Pi and might help to explain that
the increase in 0 2 consumption is slower than the phosphate
metabolite changes.
Mil hondrion
Myofibril
I I I I
Fig. 6. Scheme of ATP production and consumption in muscle cells. The
scheme in Fig. I has been extended to include glycolytic re-synthesis of
ATP in or near to the myofibrils. This would buffer changes in ATP and Pi'
and might thereby delay the stimulation of the mitochondria. See Fig. I for
further explanation of symbols.
To test this idea we inhibited the glycolytic chain at the
level of glyceraldehyde 3-phosphate dehydrogenase with
iodoacetate [40]. In order to avoid side effects of iodoacetate
the dose was kept so low that lactate production during
perfusion with anoxia was not yet completely inhibited. To
supply the mitochondria with substrate, glycolysis was
bypassed by giving pyruvate as exogenous substrate. Under
these conditions PCr hardly decreased at 37C when heart
rate was doubled and the response of 02 consumption
appeared to be too fast to determine accurately. Therefore we
did the measurements at 28C, to slow down the response.
At this lower temperature we could indeed measure a small
change in Pi' with time constant 3.6 3.0 s (SE). The large
variability of the estimate was probably due to the small
amplitude of the response, i.e . a0.4 )lmol/g dw increase from
0.5--0.6 )lmol/g dw. The time constant of the phosphate
metabolites was of the same order as when glycolysis was
not inhibited and bypassed.
However, the t mito for the same conditions in the NMR
experiment was 12 s without glycolytic inhibition and bypass
but became shorter with glycolytic inhibition and bypass. 4.2
1.0 s. Thus after glycolytic inhibition and bypass the
response times of phosphate metabolites and of 02 con-
sumption become the same. With glycolysis inhibited, the
331
difference between the measured maximal decrease in PCr,
1.1 )lmol/g dry weight, seen only in the first 10 s after the
acceleration of heart rate, and the change calculated from t mito
as explained above, 1.6 )lmol/g dry weight, is not significant
any more. This result is compatible with the hypothesis that
glycolysis buffers changes ofATP, ADP and Pi so that changes
in these metabolites do reach the mitochondria with delay.
The short t mito with glycolysis blocked and pyruvate as
exogenous substrate might be explained, if pyruvate increases
the capacity ofthe cardiac mitochondria, as suggested before
[12,41], so that the mitochondrial response toADP and Pi is
enhanced. The loss of glycolyticATP synthesis could thus be
compensated by enhanced mitochondrial capacity.
Regulators of the fast response of
cardiac oxidative phosphorylation
Not ADP and ~, but Ca 2 ?
When non-invasive 31P_NMR measurements on the intact
heart in situ were done, it was often found that high energy
phosphates and Pi did not change when cardiac work load and
oxygen consumption were stimulated [11, 41, 42]. The
measurements were done in the steady state and the lack of
change in PCr meant thatADP did also not change, as inferred
from the creatine kinase reaction assuming equilibrium. Thus
the classic scheme for regulation of oxidative phosphory-
lation by the breakdown products of ATP [13] was not
applicable to the regulation of cardiac oxidative phosphory-
lation.
To explain how oxidative phosphorylation could be
stimulated otherwise, it was proposed that cytosolic calcium
would enter the mitochondria, where it would stimulate
mitochondrial dehydrogenases [43,44] and lead to enhanced
NADH synthesis, and NADH in turn would stimulate oxi-
dative phosphorylation [12]. Increases in NADH fluore-
scence, coming from the mitochondria [45] , immediately
after upward steps in heart rate were initially reported for
saline-perfused rat heart [46], but other reports show no
decrease in NADH fluorescence for rabbit heart [47, 48].
Decreases in NADH fluorescence were even reported in
isolated rabbit papillary muscle at 22- 23C [49], in isolated
rat cardiac myocytes at high 02 supply [50] and in isolated
rat heart vasodilated with adenosine [51].A possible increase
in NADH fluorescence at increased stimulation rates, as
found by Koretsky [46], was attributed to hypoxia caused by
low 02 supply [50, 51].
Calcium enters the cytosol immediately during excitation-
contraction coupling. Intramitochondrial Ca2+ can be measur-
ed with a calcium indicator whose fluorescence was quenched
in the cytosol, but not in the mitochondria, so that intra-
332
mitochondrial calcium transients could be measured in
isolated rat cardiac myocytes [52]. Calcium transport from
the cytosol into the mitochondria takes a considerable time
with a half time of -12 s for the decline of the mitochondrial
Ca2+concentration at 23C after electrical stimulation of the
cardiac myocytes is stopped. This half time for Ca2+transport
across the mitochondrial inner membrane is close to the half
time for 02 consumption found for the same temperature by
interpolation, -15 s [30]. These similar time courses indicate
the possibility that Ca2+ entry into the mitochondria may be
responsible for stimulating 02 consumption. The NADH
fluorescence changes with half decay time 19 s when twitch-
ing of rabbit papillary muscle was stopped at the same
temperature [49], but NADH increased when muscle stimu-
lation was stopped, which contradicts the calcium stimulation
theory. Very recently it was reported that the mitochondrial
NADH concentration initially falls after elevation of cardiac
workload, and subsequently recovers to a new steady state
level if cytosolic Ca2+ is increased [53]. When the elevation
of workload is not accompanied by increases in cytosolic Ca2+
the recovery of NADH concentration does not take place.
Apparently, Ca2+stimulates the intramitochondrial dehydro-
genases to minimize the decrease in NADH concentration,
but with a relatively slow time course. Ca2+ could not only
stimulate the mitochondrial dehydrogenases, but may also
lead to activation ofATP synthase, the mitochondrial enzyme
directly responsible for ATP synthesis [54]. Therefore we
investigated the role of Ca2+ further.
Because Ca2+ entry into the mitochondria is rather slow,
we reasoned that partial blockade of Ca2+ entry into the
mitochondria should further slow down the response of
oxidative phosphorylation to heart rate steps if calcium
stimulation of the intramitochondrial dehydrogenases or
other mitochondrial enzymes would be the main stimulus
pathway for the mitochondria. However, infusing ruthenium
red at -40% of the dose which completely blocked the
mitochondrial calcium transporters did not lead to any change
in tmito in isolated rabbit heart at 28C, suggesting that the
mitochondrial Ca 2+ channels are not rate limiting for the
response of oxidative phosphorylation during steps in heart
rate [55]. On the other hand, when the dose of ruthenium red
was so high that mitochondrial calcium uptake was strongly
blocked, tmito increased from 13.5 0.7 to 22.2 3.4 s. Not
only will blockade with ruthenium red prevent fast entry of
calcium during changes in cytosolic calcium, but it will
probably lead to low intramitochondrial Ca2+ concentration
in general because Ca2+ influx is inhibited, but the mechan-
isms for extruding Ca2+ from the mitochondria are not
specifically inhibited. Thus the increase in tmito during strong
blockade of the calcium entry channels is probably not due
to slow entry of calcium into the mitochondrial matrix, but
may be caused by a low background of activity of the
dehydrogenases (leading to low NADH levels) and perhaps
of other mitochondrial enzymes [54] causing a low capacity
for ATP synthesis. Such a low mitochondrial capacity might
cause a slow response of mitochondrial ATP synthesis to
quickly changingATP hydrolysis, as indicated by a model for
skeletal muscle ATP turnover [9].
The lack of effect of partial blockade ofthe Ca 2+channels
suggests that Ca2+ entry into the mitochondria is not rate
limiting for the transient adaptation of oxidative phos-
phorylation during cardiac workjumps. However, that does
not mean that Ca2+ may not playa role in a possible second
slower phase of adaptation. Indeed, PCr decreased quickly
when a heart rate step was applied to ferret heart, but in a
second slower phase partially returned to the control level,
suggesting that slow stimulatory processes also playa role
in the heart [56]. Recently, it was shown directly that slow
entry ofCa2+ leads to a second phase of reduction ofNAD+
[53]. However, here we focus mainly on the fast adaptation
phase characterized by tmito '
In summary, a key role for Ca 2+ for the regulation of
oxidative phosphorylation was proposed, but is not very
likely for the first fast phase ofthe response. The mechanism
of Ca2+ regulation in the steady state is also not completely
understood [57].
Limitation in carbon substrate supply at higher cardiac
workloads?
The stimulation by Ca2+ of the intramitochondrial dehydro-
genases is thought to enhance NADH production [43, 44],
but it is also possible that NADH production is limited "up-
stream" from the dehydrogenases. For instance, glucose must
be transported into the cell, then be metabolized to pyruvate
via the glycolytic chain, and the reducing equivalents
generated in glycolysis must be shuttled into the mitochondria
to prevent inhibition of glycolysis by accumulation of
NADH. Iflactate is provided as exogenous substrate it must
enter the cell, be oxidized to pyruvate and the reducing
equivalents must again be shuttled into the mitochondria.
Pyruvate needs to transported into the mitochondria to reach
pyruvate dehydrogenase, but there is no need to shuttle
reducing equivalents into the mitochondria. If one or more
of these steps of NADH entry or production in the mito-
chondria adapt slowly to increased need for reducing equi-
valents at the respiratory chain, this might limit the speed of
the response of oxidative phosphorylation. Indeed, in several
studies it was found that mitochondrial NADH fluorescence
decreases when cardiac contraction and 02 consumption was
stimulated [49-51]. On the other hand mitochondrial NADH
content rises when a high concentration of pyruvate rather
than glucose is provided to the heart [49], suggesting that
limitation ofNADH production with glucose as substrate is
alleviated. Indeed, when pyruvate or ~-hydroxybutyrate is

given to the heart, this leads to a clearly higher phosphory-
lation potential than when glucose is supplied, suggesting
improved mitochondrial function [41, 58, 59].
We found that t mito gradually increased if we stepped to
higher heart rates in isolated rabbit heart at 28C, when
glucose was the exogenous substrate [60]. A hypothesis to
explain this was that the reduction of NADH (and flavin
nucleotides) from glucose became increasingly more rate
limiting at higher heart rates This rate limitation might we
alleviated by giving pyruvate or lactate as exogenous sub-
strate. Thus we compared tmito for various substrates, stepping
from 60 beats/min to various heart rates up to 140 beats/min.
The tmito increased from 7.6 s (heart rate from 60 to 70 beats/
min) to 12.1 s (heart rate from 60 to 120 beats/min) with 11
mM glucose, but tmito did not change for any heart rate step
when 11 mM pyruvate or 11 mM lactate was provided instead
of 11 mM glucose [61, 62]. Thus it seems unlikely that
progressive rate limitation of NAD+ reduction to NADH is
the cause ofthe increase in t mito when stepping to high heart
rates. However, there seems to be a small rate limitation when
glucose is the substrate for a basal heart rate higher than 60
beats/min: when one steps from 89 to 110 beats/min at 28C,
tmito was significantly shortened from 12 to lOs when glucose
was replaced by pyruvate in the perfusate [55], although this
effect is less clear if the response time is corrected for the time
course of RPP, see Table 2. That rate limitation by NADH
production from glucose plays a small role at 28C is
compatible with the finding that the response time oflactate
production to a heart rate step from 60 to 80 beats/min is less
than 6 s [39], which is faster than the response of 02 con-
sumption, tmito ' which was 10.4 s for the same hearts.
At 37C tmito ' with glucose as exogenous substrate, does not
increase when stepping from 100-120 beats/min to in-
creasingly higher heart rates, but always remains in the range
5-7 s [61]. Despite this fact we investigated the effect of
substrate at this temperature: the tmito for lactate was the same
as for glucose, but t mito was significantly smaller, by 40%,
when pyruvate was given as exogenous substrate. Because
the main difference between lactate and pyruvate is that
reducing equivalents produced during the conversion of
lactate to pyruvate need to be transported into the mito-
chondria, the finding of a slower response with lactate and
glucose may indicate that activation of the malate-aspartate
shuttle (or perhaps the glycerol phosphate shuttle [63]) takes
time and slows the response of oxidative phosphorylation.
It remains to be seen whether regulation by NADH of the
gating ofVDAC, the mitochondrial outer membrane channel
[64] through which solutes reach the intermembrane space,
plays a role in these events.
Given the fact that tmito increases with heart rate at 28C,
we wondered whether elevating the cardiac work load by
other means would increase tmito as well, To answer this
question we stimulated the heart with the B-adrenergic agonist
333
dobutamine, and found no change in the response time to
heart rate steps, tmito ' despite increased 02 consumption and
developed pressure [65]. Thus a higher level of cytosolic Ca2+
during systole seems not the reason for the increase of tmito '
which remains an enigma. However, all the experiments to
determine tmito are routinely done with a low concentration
of adenosine, 10 11M, because this yields maximal vaso-
dilation to optimize 02 supply. The low adenosine concent-
ration chosen does have only minimal effects on glucose
uptake and metabolism. Despite this fact, we found that the
increase of tmito with heart rate disappeared if adenosine is
omitted [65]. The average tmito is the same whether adenosine
is absent or present, but only in the presence of adenosine
does t mito increase with heart rate. A convincing explanation
for this subtle effect has not yet been found. Given that tmito
does not increase with heart rate at 37C, adenosine may
cause an effect at 28C which is perhaps the same as caused
by higher temperature.
Cytosolic signal transduction via the creatine kinase
shuttle
Free energy in skeletal and cardiac muscle was proposed to
be transported by a phosphorylcreatine shuttle from mito-
chondria to the myofibrillar and other ATPases [66]. Creatine
kinase in different isoforms is present throughout the muscle
cell, not only in the cytosol, but notably localized in the
mitochondrial inner membrane space and in the myofibrils.
Thus PCr would transfer its "high energy" phosphate group
to ADP, thereby re-synthesizing ATP in the myofibril. The
creatine formed would diffuse out of the myofibrils, see Fig.
Mitochondrion
Myofibril
I I I I
Fig. 7. Scheme of ATP production and consumption in muscle cells. This
extension of the scheme in Fig. I emphasizes that isoforms of creatine
kinase are localized in the myofibrils and mitochondria. A TP is re-
synthesized efficiently from per, and as a consequence creatine is produced
during contraction in the myofibrils and may be the main form of phosphate
acceptor to diffuse across the cytosol. The hypothesis has been advocated
that diffusion of ADP through the cytosol is restricted. Unlike in the previous
Figure with glycolytic buffering, Pi is not involved in the phosphocreatine
shuttle.
334

7. Due to the fast reaction kinetics of creatine kinase and the
high equilibrium constant KCK ' -180 at 37C, which governs
the concentration [ADP] = [ATP] '[Cr] / {[PCr] 'KCK }' the
ADP that is produced is for the largest part re-synthesized to .
ATP, leading to a change in Cr and PCr that is much greater
than the change in free ADP concentration. Because Cr and
PCr are smaller molecules thanADP or ATP, they also diffuse
somewhat faster through the cytosol. One could say that a
small concentration change in ADP is amplified to a large
concentration change of Cr by the creatine kinase reaction.
This facilitates diffusion of phosphate acceptors from
myofibrils to mitochondria greatly [67]. We must now
consider the consequences of the presence of the creatine
phosphate shuttle for the dynamics of regulation of oxidative
phosphorylation to adapt to changed ATP hydrolysis in the
myofibrils. Can we predict how the response time t mito is
influenced by the presence of creatine kinase and the phos-
phorylcreatine shuttle?
The diffusion distance L between myofibrils and mito-
chondria in the muscle cell is not much more than 211m, and
the characteristic diffusion time td =U / D, where D is of order
1.5 . 10--{j cm2/s [67], is not more than 27 msec. However, this
time could be increased if many binding sites for the trans-
ported molecule exist on relatively immobile macromolecules
and the molecule remained bound at such sites for an
appreciable time, in other words if a large fraction of the
molecules is bound. To some extent this would indeed be
possible for ADP and Pi' where a large fraction is known to
be bound, but such large bound fractions do probably not
exist for ATP, PCr and Cr. If most of the transport of free
energy is via PCr and Cr, as is virtually undisputed [67], then
a diffusion time of less than 27 ms is applicable, which is
rather short relative to the response time of oxidative phos-
phorylation, and one can consider the cytosol a well mixed
compartment for practical purposes. Several mathematical
models for the response time include this assumption exp-
licitly or implicitly to simplify the modeling ofthe regulation
of oxidative phosphorylation in skeletal muscle [9,68] or
cardiac muscle [69]. These models show that, even if the
diffusion delay is negligible, then the powerful creatine
kinase buffer would delay the increase of ADP (Figs. 8 and
9). By inserting in vitro data on the relation between the
phosphorylation potential in the mitochondrial suspension on
the one hand and mitochondrial 02 consumption determined
for cardiac mitochondria on the other [70,71] response times
are predicted which are generally faster than measured in the

c::
.:2 - ....

"
\
0.
E
-;; PIO RATIO
..: ~c::
"0 0
c:: u
.c c::
0
u
8 '">-.en
. ><
0
,,
toG ATP

+
c.?
<l
"-
f-<
<

/ PCr
ATP synthesis

t PCr
+
PCr + ADP

t+
Cr + ATP

ATP hydrolysis

Fig. 8. Schematic representation of a simple model of regulation of the mitochondria via the phosphate metabolites. The model is modified from Meyer [9]:
the linearization is removed and real measurements on mitochondria are inserted. In the model the relation between the Gibbs free energy of A TP synthesis.
toG ATr' and mitochondrial respiration of isolated cardiac mitochondria measured by Gyulay et al. [70] or Holian et al. [71] are inserted. It can be shown that
this nonequilibrium thermodynamic approach is equivalent to an enzyme kinetic model. The creatine kinase reaction is assumed to deviate negligibly from
equilibrium. In this simple, non-realistic model the cytosol and myofibrils are assumed to form one well mixed compartment The nonlinear differential
equation for the model is solved numerically on a personal computer.
 335

i
~
20 ico 20
c: .6
g .s==-
'5 0
E
E
..:: ..::
IS 15 IS 15
"8- "8-
E E
::l

ei
::l
<II <II
c:
8
c:

g g
CD
10 10
5 0 10 20 30 5 0 10 20 30

30 25

:i :i

------
.s 25 - .sc: 20
IS 51
\1
lo
OJ
lo
c:
!!c:
e 20
8
15

~ ~
15 10
5 0 10 20 30 5 0 10 20 30
TIme(s) TIme (s)

Fig. 9. The gradual decrease in PCr concentration and increase in mitochondrial oxidative phosphorylation calculated from the model of Fig. 8. A t=O a step
in ATP hydrolysis was imposed on the model and toward the end of the response the 02 consumption catches up with metabolic needs. For the physiological
cardiac data that had been inserted the PCr decreases very moderately in the left panels (cytosolic pH 7.1), reflecting the high mitochondrial content of heart
muscle cells which causes a high feedback loop gain in the model of Fig. 8. The right hand panels show the response at cytosolic pH 6.5, assuming that the
mitochondrial capacity was decreased by 38% because this had been measured in brain mitochondria due to the acidosis. The creatine kinase equilibrium in
the model is shifted by the low cytosolic pH. Later measurements in cardiac mitochondria showed that the decrease in mitochondrial aerobic capacity was
less than found in the brain. This model calculation shows a substantial slowing of the response ofO, consumption when the mitochondrial aerobic capacity
is decreased.

15

ISOLATED
HEART

MODEL,
DATA
ISOLATED
MITOCHONDRIA
(HOLIAN)

MODEL,
__________~_or----------~D DATA
D----{]- ISOLATED
MITOCHONDRIA
(OYULAI)

r I i i
70 80 100 120
HEAR RATE (MIN-I)

Fig. 10. The response time of mitochondrial 02 consumption to a step increase in ATP hydrolysis_ We simulated with the model of Fig. 8 [69] experiments
on isolated rabbit heart where heart rate was stepped from 60 beats/min to the heart rate given on the abscissa [60]. The bars give the experimental SE.
336
experiment by Eijgelshoven et al. [60] (Fig. 10). The diff-
erence might be explained by additional delay due to transport
through the cytosol.
The mitochondrial content of heart muscle cells is large,
order 30%, causing a large feedback loop gain in the model
of Fig. 8, which by itself would cause the response to be fast.
The relatively slow response times seem to be partially
explained based upon the buffering action of creatine kinase,
which delays the rise of ADP at the mitochondria. The
combined increase inADP and P in the first minute after an
I
upward step in heart rate can account for 100% of the increase
in 02 consumption at 28C, although for only 30% of the
increase at 37C, based on enzyme kinetic calculations [40].
Although the simple models were to some extent success-
ful in explaining the order of magnitude of the amplitude and
time course of the regulation in skeletal and cardiac muscle,
they assumed a well mixed cytosol. This is, however, hard
to reconcile with the finding that PCr reaches a new level
much sooner (time constant -3 s) than 02 consumption (tmito
-12 s in same hearts), given that the mitochondria themselves
appear to react within 100 ms (see above). The discordance
of the time courses of phosphate metabolites and oxygen
consumption suggests that changes in PCr, Cr, and ADP in
or near the myofibrils take considerable time to spread
through the cytosol, or that metabolites are compartmental-
ized. It is hard to understand a large diffusion gradient in the
tissue and a large diffusion delay for creatine and PCr in
cardiac myocytes where the mitochondrial density is very
high and mitochondria are located close to the myofibrils
[67]. The classic version of the creatine phosphate shuttle
would predict fast and efficient transport of phosphate
acceptors to the mitochondria, without causing appreciable
delay above the characteristic diffusion times, which are less
than 27 ms. We will now briefly consider whether the reaction
rates in the creatine kinase reaction are high enough to sustain
fast operation of the shuttle.
The unidirectional reaction speeds of creatine kinase are
high and are usually believed to be one order of magnitude
above the ATP turnover rate [72, 73]. Conley [74] replaced
creatine in rat heart largely with p-guanidinopropionate,
which is an analogue of PCr that is phosphorylated in the
creatine kinase reaction, but three orders of magnitude slower
than creatine. The time course of decrease of p-guanidino-
propionate-phosphate, after a heart rate step was the same as
for PCr. This may be understood if the activity of creatine
kinase is in large excess, so that slowing the creatine kinase
reaction does not alter the time course of phosphoryl-group
transfer from a phosphagen.
Another important question regarding the speed of the
shuttle is to what extent intermediates in the shuttle, creatine
and PCr, are delayed by binding. About 15% oftotal creatine
in isolated perfused rat heart is not visible by NMR, which
may indicate that a small fraction of total creatine is bound
[75]. However, the same study shows that PCr is not bound.
If appreciable reversible binding of creatine takes place in the
diffusion trajectory between myofibril and mitochondrion,
the diffusion time may increase somewhat, but the fraction
of bound creatine should be considerably higher than 15%
in the diffusion trajectory to cause an increase considerably
above the characteristic diffusion time indicated above.
It appears that the classical form of the creatine kinase
shuttle [66] and the diffusion model of Meyer and Kush-
merick [67] predict a fast transfer of phosphate acceptors
through the cytosol. It is hard to explain the delay between
changes in PCr, Pi and ADP on the one hand and 02 con-
sumption on the other by diffusion delay, by slow chemical
reaction rates and by binding sites. Our conclusion for the
moment is that these views of phosphate acceptor transport
in the cytosol cannot explain by themselves the delay between
changes in phosphate metabolites and 02 consumption
measured in the heart [33]. It would seem a good idea to
measure the response time of 02 consumption and of phos-
phate metabolites after feeding a creatine analogue that reacts
less efficiently with creatine kinase, after blocking creatine
kinase or after genetic knock out one or more of the creatine
kinase isoforms. In that way we can test whether a fully
functional phosphorylcreatine shuttle does lead to quicker
adaptation of oxidative phosphorylation to ATP hydrolysis
in the cytosol.
Does 'metabolic wave' propagation delay the signal
between myojibrils and mitochondria?
Saks et al. [14] have proposed that a metabolic wave conducts
the feedback signal between myofibrils and mitochondria.
This means propagation of oscillations of local concent-
rations ofADP and creatine through the cytosol, mediated by
creatine kinase and adenylate kinase reactions. The picture
of a wave, travelling with a certain speed through the cytosol,
suggests immediately that there may be a delay between a
concentration change in the myofibrils and the arrival ofthis
change at the mitochondria. Therefore it is a reasonable idea
to explore whether the metabolic wave hypothesis may
explain the delay between the measured first changes in
phosphate metabolites (which would then take place in or
near the myofibrils) and the time course of the 02 which
would reflect arrival of the wave at the mitochondria. If this
explanation for the delay in oxygen consumption is correct,
we could measure the time of travel of the wave from the
delay between the first phosphate metabolite change and the
change of02consumption. In the metabolic wave hypothesis,
inorganic phosphate produced in the myofibrils would bypass
the reactions of the metabolic wave to diffuse to the mito-
chondria. However, as explained above Pi changes might be
buffered and delayed by glycolytic reactions at or near the

IIII
ConQQIOI'I
itochondrion
TP
IIII
Myofibril
I I I I
Fig. 11. The metabolic wave traveling from myofibrils to mitochondria.
The changes in A TP and ADP and Pi caused by contraction are gradually
transmitted to the mitochondria by coupling to PCr and Cr via the creatine
kinase reaction, and may also be coupled by the enzyme adenyl ate kinase
(AK) which converts 2 ADP molecules into ATP and AMP. AMP might
then also take part in the metabolic wave. Glycolytic buffering not only
involves ADP but also Pi and may take place in the myofibrils but also in
the cytosol. An important addition to the phosphocreatine shuttle hypothesis
is the idea that micro-compartments are formed in the cytosol, restricting
free diffusion [14].
myofibrils, see Fig. 11. In order to test whether a metabolic
wave travels through the cytosol to arrive with some delay
at the mitochondria, it would be very useful to measure the
full time course of 02 consumption, rather than the 'mean
time' tmito . The response may not be mono-exponential, but
may instead contain a certain lag phase before 02 con-
sumption starts, followed by a more or less steep transition
which coincides with arrival of the metabolic wave at the
mitochondria. However, assessment of the full time course
is difficult. One needs fast venous 02 measurements and a
very good and detailed model of 02 transport in tissue to
deconvolute the venous signal, which is virtually impossible
given the current state of 02 modelling. Alternatively, one
could measure the time course of all the 2 molecules present
in the tissue, but this also poses problems as already explained
above. However, one could try to develop an advanced and
well-calibrated near infrared spectroscopy method where the
total 02 bound to hemoglobin and myoglobin can be measur-
ed separately from the cytochrome c oxidase signal. If only
myocardial tissue is sampled and not blood in the ventricular
lumen, then it may be feasible to measure the full time course
of 02 consumption reasonably accurately or the time course
of reduction of cytochrome c oxidase under conditions where
this enzyme becomes more reduced with elevated cardiac
work load, and use these measurements as tools to study the
propagation of the metabolic wave.
Perhaps a very important factor in delaying a metabolic
wave is that the outer mitochondrial membrane has a limited
permeability, at least toADP[14, 76-78]. It was also suggest-
ed that the apparent Km of the mitochondria in vivo is much
337
higher than the Km of around 25 /-lM measured in vitro in
isolated mitochondria. Ifthis applies to in vivo conditions the
cytosolicADP concentration, reported to be over 50 /-lM [75],
may be an efficient regulator of oxidative phosphorylation,
while in the past it was often said that the ADP concentration,
at roughly twice the classic Km value, could not effectively
regulate oxidative phosphorylation. Given that bothADP and
P stimulate oxidative phosphorylation, the interaction term
b~tween the two may give the appearance that oxidative phos-
phorylation changes relatively slowly initially. To give an
oversimplified example: suppose bothADP and Pi change by
40%, then if oxidative phosphorylation would be proportional
to the product of ADP and Pi concentration (which is not
realistic and used just for the sake of the argument), oxidative
phosphorylation would increase by 96%. However, if ADP
and Pi are both halfway, and have changed by 20%, then
oxidative phosphorylation would have increased by 44%, less
than half of the final full change, 96%. This numerical
example indicates that some delay between metabolite change
and oxidative phosphorylation may be due to the fact that
several metabolites, whose concentrations are correlated,
stimulate oxidative phosphorylation simultaneously.
Recently, it was proposed that ADP stimulates oxidative
phosphorylation with a Hill coefficient close to two, and the
relation betweenADP and oxidative phosphorylation would
have a quadratic part, which steepens the slope below and
around the Km and leads to greater sensitivity to ADP [79].
This quadratic relation between ADP and oxidative phos-
phorylation rate would apparently slow down the response
of oxidative phosphorylation, similar to regulation by ADP
and Pi simultaneously. This moderate apparent slowing down
ofthe response due to quadratic and to interaction terms may
help to explain the delay between phosphate metabolite
changes and 02 consumption. However, the interaction term
between ADP and P was included in the model calculation
I
of Figs. 8-10, where we see that the calculated tmito from the
model tend to be still lower than actually measured values,
despite incorporation of the interaction terms.
An interesting aspect of delay caused by the metabolic
wave is that inevitably the concentrations of the wave's
intermediates in tissue must change. This can be derived
mathematically in analogy to the mass balance equations used
above, see Equations 11 and 12.
fl(er + ADP) = f[flJADP_entry(t) -flJADP_exJt)]'dt (13)
o
HereADP stands for the phosphate acceptor which enters the
conducting medium for the metabolic wave after ATP hydro-
lysis for contraction. The change in acceptor flux entering the
conducting medium is given by flJ ADP--entry . The change in flux
leaving the conducting medium entering the mitochondria is
flJ ADP-;:xit' The integral in Equation 13 shows that if the exit
338
of phosphate acceptor from the conducting medium follows
the entry with delay, the content of phosphate acceptor in the
medium changes appreciably.
For a step change in ATP hydrolysis of amplitude ~m, at
t=O, the total change in phosphate acceptor (Cr plusADP) in
the medium conducting the metabolic wave, is related to the
travel time of the wave twave defined in analogy to the response
times discussed above
M(t)
=-~m . Hl-
CD
~(Cr + ADP) ATP--xll ]-dt
o ~m
(14)
The change of phosphate acceptor concentration may not be
uniform, but may build up a gradient. Cr+ADP does not
increase if ADP+P i are buffered by glycolysis, see Fig. 11.
We hypothesize that glycolysis plays a similar role as creatine
kinase and adenyl ate kinase in the energy shuttle between
sites of ATP hydrolysis and synthesis.
In conclusion, delay in the stimulation of oxidative
phosphorylation caused by a metabolic wave might be
investigated by examining the full time course of 02 con-
sumption to reveal a possible lag phase in the response
indicative of wave-like propagation of the metabolic signal.
The metabolic wave could explain that the time course ofPCr
is much faster than the time course of mitochondrial 02
consumption.
Dynamic regulation of cardiac oxidative
phosphorylation in disease
The dynamic regulation of oxidative phosphorylation to
quickly changing load conditions has also been studied in
pathological conditions of the heart. The response time of
mitochondrial 02 consumption gives the delay with which
mitochondrialATP synthesis adapts to a change in work load
andATP hydrolysis. If this delay becomes larger this indicates
a potentially enhanced depletion of the high energy phos-
phates. A decreased phosphate potential may mean decreased
cardiac contractility and increased susceptibility to cardiac
arrhythmia. Indeed, the flux of energy metabolism in cardiac
myocytes may start to oscillate under certain conditions and
oscillations of the glycolytic chain have been shown to be
coupled to synchronous oscillations in excitability of the cells
and in membrane currents [80]. Mitochondrial metabolism
can also oscillate, at least in vitro [81].
Cardiac mitochondrial ATP synthesis seems to react to a
simple step in heart rate with a bi-phasic response causing
an overshoot in PCr concentration [56). It is possible in
principle that under certain conditions 02 consumption
oscillates locally, but out of phase in different small regions
of the heart, so that a non-oscillating, slightly fluctuating 02
uptake is measured for the heart as a whole. In that hypo-
thetical case oscillations, especially if out of phase in adjacent
regions could cause disturbances of electrical excitation wave
propagation or electrical impulse generating activity in the
heart. On the other hand, electrical excitation waves spreading
through the heart do synchronize metabolic adaptation and
cardiac metabolism. The metabolic disturbance caused for
instance by temporary speeding of the heart rate might in
itself lead to changes in metabolic state which alter excita-
bility or propagation characteristics of waves. An example
of oscillations in heart rate leading to changes in NADH
fluorescence was already shown by Britton Chance in 1966
[82].
The study of metabolic transients is not only of interest for
such reasons, but it was also suggested that the response time
of 02 consumption is inversely proportional to the mito-
chondrial capacity (Elzinga, personal communication). If this
holds, tmito becomes a diagnostic tool to assess the mitochond-
rial function in the intact heart, and the necessity to study
isolated mitochondria can be avoided, which is desirable
because isolation could cause major changes in mitochondrial
properties [14]. Indeed, the response time of02consumption
in isolated frog skeletal muscle cells of different types was
inversely proportional with the mitochondrial content of the
cells, and this relation could be extrapolated very well to
rabbit cardiac muscle at 20C [83; Van der Laarse, personal
communication]. Further, we found that tmito increased with
about 110% per 10C decrease in cardiac temperature, over
the range 20-37C. The QIO' the factor by which tmito increased
per 10C decrease, thus was 2.1 in isolated rabbit heart, which
is close to the factor by which state 3 2consumption changes
with temperature in isolated mitochondria [84]. This is in
accord with a mathematical model [9] which suggests that
tmito is inversely proportional to the mitochondrial capacity,
which in turn would be proportional to state 3 02 consump-
tion.
Intracellular acidosis
When the blood supply to the heart is interrupted, the pH in
the cytosol of heart muscle cells, normally about 7.1, may fall
to 6.5 or below because of CO 2 accumulation, net ATP
breakdown and extensive production oflactate which cannot
be removed for lack of blood flow. We studied the effects of
intracellular acidosis per se, without interruption of02supply,
by adding CO 2 to the perfusate of isolated rabbit hearts. To
ensure adequacy of 02 supply despite increased CO 2 levels
which replaced some of the 02 the experiment was done at
28C [85]. We chose CO 2 , and not lactic acid or low bicarbon-
ate levels in the perfusate, because CO 2 diffuses easily across

A B
g
~
~
&'j.g
l!l
13 B
.!::l &'j
0
[
1~
o 0
Z Z
0 100
Time (s)
Fig. 12. The effect of intracellular acidosis caused by increasing the perfusate CO, content from 5 to 23 %, so that perfusate pH was 7.3 in panel A and 6.6
in panel B. The transient of coronary venous 0, tension during a step in heart rate in isolated rabbit heart at 28C is shown. The response is slowed
substantially because of a much larger tmilo , but 0, transport through the heart was little affected, as evidenced by an increase in t",o,poo by only 1.5-3 s. The
measured venous response time, given by the hatched area was 19 s in panel A and 35 s in panel B (from [85], permission ofPfhigers Arch. publishers was
granted).
the cell membrane and alters pH in the cell quickly [86].
Lowering the perfusate pH from 7.30 0.03 (SD) to 6.59
0.02 by increasing CO 2 in the gas mixture from 5 to 23 %,
the cytosolic pH was decreased from 7.1 to 6.5, comparable
to what may happen during cardiac ischemia. During acidosis
tmito increased dramatically, from 14.4 1.8 s at normal pH
to 22.4 0.4 s (40-75 min after isolation of the heart) or 35.9
2.8 s (85-120 min after isolation), see Fig. 12. A decrease
of mitochondrial ATP synthetic capacity may contribute to
the increase of tmite' However, when the pH in the mito-
chondrial suspension was brought down to the same value
as during acidosis in the heart, the state 3 02 consumption
was decreased only by 10-15%, both for metabolic acidosis
(no CO 2 change), and respiratory acidosis brought about with
CO 2 [87]. Thus, although the inverse proportionality between
mitochondrial capacity and tmito' found in frog skeletal muscle
fibers (Van der Laarse, personal communication) and pre-
dicted by a model of muscle energy metabolism [9], was
invoked originally to explain most of the slowing of the
response [85], actual data on isolated rabbit heart mitochondria
suggest that decreased mitochondrial capacity can only partly
explain the slowing of the response to heart rate steps during
acidosis. Perhaps the Km of the mitochondria for ADP and/or
Pi is affected by acidosis, but this has never been investigated
to the best of our knowledge. In any case, either measurements
on isolated mitochondria give misleading results for the in vivo
situation, or the cytosolic signal that feeds the cytosolic ATP
demand back to the mitochondria is transmitted or interpreted
otherwise if the cytosol is acidotic, for instance during severe
depression of breathing or during ischemia. Whatever the
cause, during cytosolic acidosis the response time of mito-
chondrial 02 consumption is strongly affected.
Post-ischemic dysfunction: stunning of the heart
If the blood supply to heart tissue has been interrupted, e.g.
by coronary stenosis, but is restored soon, one may find that
339
560 &'jf
~
0,-,
19 ~
g
5~
~ l!l
13 B
.!::l &'j
0
[507 H
0,-,
19
~'I 1~ 1
o.~
547 o 0
477 ~~
100
0
Time (s)
the contractile force of the heart is still depressed despite
reperfusion although there is no tissue necrosis and cell
membrane damage, as indicated by microscopic examination
or by the absence of leak of cytosolic enzymes to the blood
plasma. To obtain this situation the interruption of blood flow
must be brief, e.g. 15 min in rabbit heart at 37C [32].
Although originally a decreased energy supply to the myo-
fibrils was considered one of the possible causes, in recent
years the consensus appeared to be that depressed mitochond-
rial function was not the cause ofthe reduced mitochondrial
function. Among others, it was shown that mitochondria
isolated from stunned myocardium had an undiminished state
3 02 consumption and phosphorylation rate [88]. However,
Elzinga (personal communication) hypothesized that al-
though the capacity of isolated mitochondria was undimin-
ished under maximal stimulation, the mitochondria would
require a stronger cytosolic signal to accomplish the same
oxidative phosphorylation flux. The stronger cytosolic signal
could then inhibit contractile function in Elzinga's view.
Because a stronger signal will take longer to build up, the
hypothesis that the mitochondria require a stronger cytosolic
signal might be tested by looking for an increase in the
response time of oxidative phosphorylation in stunned rabbit
myocardium. Testing this hypothesis in intact myocardium
is all the more necessary because it was shown recently that
reactive oxygen species damage the myofibrillar creatine
kinase, and may thus contribute to stunning [89]. It is not at
all excluded that the mitochondrial creatine kinase is also
affected. Indeed, mitochondrial creatine kinase activity
measured in vitro is significantly diminished after very brief
ischemia and the decrease correlates very significantly with
the depression of contractile force [90]. Testing mitochondrial
function in vivo would involve all enzymes including creatine
kinase, unlike typical state 3 measurements [88] on isolated
mitochondria done by adding ADP.
First we looked at the time constant of the recovery heat
of rabbit papillary muscle (see above), and compared a
normoxic control group with a group that had been made
340
anoxic for 40 min while electrical pacing was continued [7].
The contractile force was reduced by about 20% in post-
anoxic muscle relative to control, but cell membranes were
shown to be still intact. The time constant of the decay of
recovery heat production after 10 contractions was 24.9 2.2
s initially. After 40 min anoxia and 2 h normoxic recovery
the time constant was a little bit shorter, by 2-3 s (P<0.05).
However, the time constant in the control group, which never
was anoxic, was increased to 41.7 12.3 s. From the small
decrease in the response time of oxidative phosphorylation
after hypoxia, although significant, it was cautiously con-
cluded that there is no increase in the response time, given
the large change of the time constant in the control group.
Further the economy of contraction (force-time integral of
the muscle divided by developed heat) and the ratio of heat
released during recovery from contraction (produced by
oxidative phosphorylation) to heat released during isometric
contraction (which is the enthalpy ofPCr breakdown) were
the same in normoxic controls and post-anoxic test groups.
Thus the conclusion was warranted that stunning of cardiac
muscle after anoxia is not associated with a decrease in
mitochondrial function, and probably in mitochondrial
capacity, at least at 20C.
Further we tested the effect of 25 min total ischemia on the
isolated rabbit heart at 28C, because flow interruption
combined with low tissue temperatures is a common cardiac
surgical procedure and because 02 supply to the heart is non-
limiting at this temperature. When the heart was electrically
paced during the ischemic episode the developed pressure
was decreased to 47% at 20 min reperfusion, but complete
contractile recovery was found when the heart was not paced
during ischemia. The tmito which was 11 s, increased by -1 s
(P<0.05) in a non-ischemic time control group. The tmito after
ischemia was not different from control [91]. Also the 02
consumption was the same after ischemia when assessed at
the same level of developed pressure. Thus 25 min total
ischemia at 28C does not decrease mitochondrial function,
even when there is quite severe mechanical dysfunction. It
is remarkable that after a relatively long period of anoxia
leading to stunning, where many cellular functions have been
shown to be affected, the mitochondria and signals to the
mitochondria are still functionally intact, at least at this low
temperature.
More relevant for the normal situation in the patient were
experiments at 3 rc, where we interrupted flow for 15 min
or perfused at normal flow with anoxic perfusate for 15 min
[32]. The contractile force was reduced by -30% in both cases,
but cellular membranes were not damaged. Now tmito increas-
ed by 62 10% after ischemia, and by 64 18% after high-
flow hypoxia (P<0.05). At a given contractile performance
(product of heart rate and systolic pressure) 02 consumption
was unchanged. The recovery ratio and the isometric econ-
omy of contraction (see preceding paragraph) were un-
changed after anoxia or ischemia, indicating that there is no
mitochondrial uncoupling. However, tmito changed, indicating
decreased in vivo mitochondrial function, despite that this is
not found in mitochondria isolated after brief ischemia or
hypoxia at 37C. We hypothesize that it is not the capacity
of the mitochondria themselves which has been affected, but
that trans-cytosolic energetic signal transduction is retarded.
Indeed, an acute reduction of mitochondrial aerobic capacity
does not lead to an increase in tmito at this temperature (see
below). The impairment of trans-cytosolic energy transfer
might also hinder free energy to reach the myofibrils, and
might contribute to the impairment of contractility after the
brief ischemia or hypoxia.
Reduced mitochondrial capacity and
the response time
Amodel analysis for skeletal muscle energy metabolism had
suggested that tmito would be inversely proportional to the
mitochondrial aerobic capacity for ATP synthesis [9,69],
based on the assumption that the logarithm of ATP/( ADP x
Pi ) is linearly related to oxidative phosphorylation. Further
this model assumed equilibrium for the creatine kinase
reaction. The model predicted that tmito is directly proportional
to total creatine content. Inserting data on isolated cardiac
mitochondria and on the total cardiac creatine pool predicted
a tmito of the right order of magnitude [69], see above. We
tested the model prediction on the relation between aerobic
capacity and tmito by inhibiting the mitochondrial "enzyme ATP
synthase with the specific blocker oligomycin, infused in
isolated hearts at 28C. Control experiments showed that the
inhibition of state 3 respiration of the mitochondria by
oligomycin was not changed during isolation of the mito-
chondria. Consequently, the degree of inhibition could be
judged by dividing ADP stimulated state 3 respiration by the
respiration elicited with the uncoupler FCCP which bypasses
the blocked ATP synthase, revealing maximal respiration of
the mitochondria unaffected by oligomycin. The tmito in-
creased by 9.3 and 10.8 % per 10 % decrease in the state 3 /
FCCP respiration ratio in the isolated mitochondria, for heart
rate steps from 60 to 70 and 80 beats/min respectively [92].
However, for steps from 60 to 100 or 120 beats/min the
change in tmito was not significant. For small heart rate steps
the Transient Control Coefficient of the mitochondria [22]
was apparently quite large, because the sum of Transient
Control Coefficients for all enzymes in the cell should be -
1, close to the value of just the mitochondria on their own.
However, this strong control of the mitochondria over
transients was lost at higher heart rates. Recent pilot experi-
ments at 37C also show that an about 50% reduction of
mitochondrial aerobic capacity by oligomycin does not

increase tmito [93]. This suggests that, although the mito-
chondrial capacity is an important determinant of tmito under
some conditions (low heart rates and temperatures), the
signalling and conduction properties ofthe cytosol probably
playa more important role.
The steady state Flux Control Coefficient of mitochondrial
aerobic capacity over respiration of the heart as a whole was
quite large at 28C, 0.26, which means that per lO % decrease
in mitochondrial aerobic capacity the respiration of the whole
heart is decreased by 2.6%. Because the sum of the Flux
Control Coefficients of all enzymes in the cell is 1, this
suggests that up to 26% of the rate limitation of cardiac energy
turnover is located in the mitochondria [94]. Recent pilot
experiments at 37 C even suggest that the Flux Control
Coefficient may be substantially higher, up to 0.7 [95]. It
remains to be investigated whether disturbances in the
transmission pathway of the signal between myofibrils and
other sites ofATP hydrolysis to the mitochondria through the
cytosol, which retard the metabolic wave, have a similar
depressing effect on cardiac contractile function and energy
turnover as a reduction of mitochondrial capacity as obtained
with oligomycin. However, it appears that cardiac function
is very sensitive to depression of the mitochondrial capacity.
It is not excluded that the retardation of the signal to the
mitochondria seen during acidosis and after myocardial
stunning (see above) indicates problems in energy transfer
through the cytosol which contribute to the depression of
cardiac contractile function and energy turnover under these
conditions.
Conclusion
We hope to have convinced the reader that measurement of
the response time of mitochondrial 02 consumption, although
this is relatively complicated, can give unique information,
which is hard to obtain in any other way. The finding of
response times of mitochondrial 02 consumption of 5 sand
larger show that mitochondrial ATP synthesis is not immedi-
ately adapted to ATP hydrolysis, as was previously implied.
Indeed, NMR measurements at high time resolution show that
the phosphate metabolites do change in the early moments
after an increase in ATP hydrolysis. Further, the finding in
the same heart of a response time of 2 consumption of 12 s,
which is much longer than the time constant of inorganic
phosphate and PCr (-3 s), to which it should be compared,
indicates that metabolic changes do not arrive immediately
at the mitochondria, and that the phosphate metabolites in the
heart muscle cells are more or less compartmentalized.
341
Acknowledgements
This work was supported by several grants from the Nether-
lands Heart Foundation (including an Established Investi-
gator Grant), from the Foundation for Life Sciences NWO,
and from the Medical Research Council NWO, the Nether-
lands.
References
1. Van Beek JHGM, Elzinga G: Response time of mitochondrial oxygen
consumption to heart rate changes in isolated rabbit heart. Proc. Int.
Union Physiol Sci 16: 485,1986
2. Van Beek JHGM, WesterhofN: Response time of mitochondrial oxygen
consumption following stepwise changes in cardiac energy demand.
In: J Piiper, TK Goldstick, M Meyer (eds). Oxygen Transport to Tissue
XII. Plenum Press, New York, 1990, pp 415-423
3. Van Beek JHGM, WesterhofN: Response time of cardiac mitochondrial
oxygen consumption to heart rate steps. Am J Physiol 260: H613-
H625, 1991
4. Groot B de, Zuurbier CJ, Beek JHGM van: Response times of
mitochondrial oxygen consumption, tissue oxygenation and venous
oxygen concentration during steps in heart rate in isolated rabbit heart.
Pfliigers Arch. 430 (S4): R93, 1995
5. Beek JHGM van, Osbakken MD, Chance B: Measurement of the
oxygenation status of the isolated perfused rat heart using near infrared
detection. Adv Exp Med Bioi 388: 147-154, 1996
6. Mast F, Elzinga G: Recovery heat production of isolated rabbit papillary
muscle at 20 c C. PfliigersArch 411: 600--605,1988
7. Zuurbier CJ, Mast F, Elzinga G, Beek JHGM van: Mitochondrial
function is not decreased in stunned papillary muscle at 20 c C. J Moll
Cell Cardiol29: 347-355,1997
8. Kushmerick MJ, Meyer RA: Chemical changes in rat leg muscle by
phosphorus nuclear magnetic resonance. Am J Physiol 248: C542-
549, 1985
9. Meyer RA: A linear model of muscle respiration explains mono-
exponential phosphocreatine changes. Am J Physiol254: C548-C553,
1988
10. Massie B, Schwartz GG, Garcia J, Wisneski JA, Weiner MW, Owens
T: Myocardial metabolism during increased work states in the porcine
left ventricle in vivo. Circ Res 74: 64--73, 1994
II. Balaban RS, Kantor HL, Katz LA, Briggs RW: Relation between work
and phosphate metabolites in the in vivo paced mammalian heart.
Science 232: 1121-1123, 1986
12. Balaban RS: Regulation of mitochondrial oxidative phosphorylation
in mammalian cells. Am J Physiol258: C377-389, 1990
13. Chance B: Reaction of oxygen with the respiratory chain in cells and
tissues. J Gen Physiol49: 163-188, 1966
14. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuzuetsov AV:
Metabolic compartmentation and substrate channelling in muscle cells.
Role of coupled creatine kineses in in vivo regulation of cellular
respiration - a synthesis. Moll Cell Biochem 133/134: 155-192, 1994
15. Beek, JHGM van: Dynamics of cardiac mitochondrial oxygen
consumption deduced from whole organ oxygen uptake. In: JB
Bassingthwaighte, J Linehan (eds). Whole OrganApproach to Cellular
Metabolism: Capillary Permeation, Cellular Transport and Reaction
Kinetics, Springer, New York, in press
16. Lassen NA, Henriksen 0, Sejrsen P: Indicator methods for measurement
of organ and tissue blood flow. In: JT Shepherd, FM Abboud (eds).
342
Handbook of Physiology. Section 2: The Cardiovascular System. Vol.
III. Peripheral Circulation and Organ Blood Flow. Part I, 1983, pp
2Hi3
17. Meier P, Zierler KL: On the theory of the indicator-dilution method
for measurement of blood flow and volume. J Appl Physiol 6:731-
744, 1954
18. Roberts GW, Larson KB, Spaeth EE: The interpretation of mean transit
time measurements for multiphase tissue systems. J Theor Bioi 39:
447-475,1973
19. Loiselle DS, Van Beek JHGM, Mawson DA, Hunter PJ: The surface
of the heart leaks oxygen. News Physiol Sciences 10: 129-133,1995
20. Van Beek JHGM, Loiselle DS, Westerhof N: Calculation of oxygen
diffusion across the surface of isolated perfused hearts. Am J Physiol
263: HI003-1010, 1992
21. Beauvoit B, Chance B: Time-resolved spectroscopy of mitochondria,
cells and tissues under normal and pathological conditions. This
volume.
22. Heinrich R, Rapoport TA: Mathematical analysis of multienzyme
systems: II. Steady state and transient control. Biosystems 7: 130-
136, 1975
23. King RB, Bassingthwaighte 18, Hales JRS, Rowell LB: Stability of
heterogeneity of myocardial blood flow in normal awake baboons.
Circ Res 57: 285-295, 1985
24. Gorman MW, Wangler RD, Sparks HV: Distribution of perfusate flow
during vasodilation in isolated guinea pig heart. Am J Physiol 256:
H297-301, 1989
25. Beek JHGM van, Bussemaker J, Barends JPF, WesterhofN: A 13C-
NMR technique to determine absolute oxygen consumption in quickly
frozen small myocardial samples. FASEB J 10: A325, 1996
26. Van Beek JHGM: Fractal models of heterogeneity in organ blood flow.
In: S Egginton, H Ross (eds). Oxygen Transport in Biological Systems:
modelling of pathways from environment to cell. Cambridge Uni-
versity Press, Cambridge, 1992, pp 135-163
27. Van Beek JHGM, Roger SA, Bassingthwaighte 18: Regional myo-
cardial flow heterogeneity explained with fractal networks. Am J
Physiol257: H1670-1680, 1989
28. Van Beek JHGM, Barends JPF, and WesterhofN: The microvascular
unit size for fractal flow heterogeneity relevant for oxygen transport.
In: P Vaupel, R Zander, DF Bruley (eds). Oxygen Transport to Tissue
XV, 1994, pp 901-908
29. Hak 18, Biessels PTM, Beek JHGM van, Bakker JC, and Westerhof
N: Function of isolated rabbit hearts perfused with erythrocyte
suspension is not stable but improvement may be feasible with
hemoglobin solution. Adv Exp Med Bioi 345: 291-297,1994
30. Hak 18, Van Beek JHGM, Wijhe MH van, WesterhofN: Influence of
temperature on the response time of mitochondrial oxygen con-
sumption in isolated rabbit heart. J Physiol (Lond) 447: 17-31, 1992
31. Wittenberg JB: Myoglobin-facilitated oxygen diffusion: role of
myoglobin in oxygen entry into muscle. Physiol Rev 50: 559-636,
1970
32. Zuurbier CJ, Beek JHGM van: Brief ischemia results in decreased
mitochondrial function and contractility in isolated rabbit heart at 37C.
FASEB J 9, 1995
33. Eijgelshoven MHJ, Van Beek JHGM, Mottet I, Nederhoff MGJ, Van
Echteld CJA, Westerhof N: Cardiac high-energy phosphates adapt
faster than oxygen consumption to changes in heart rate. Circ Res 75:
751-759, 1994
34. Chance B: The energy-linked reaction of calcium with mitochondria.
J Bioi Chern 240: 2729-2748, 1965
35. Lemasters JJ, Hackenbrock CR: Continuous measurement and rapid
kinetics ofATP synthesis in rat liver mitochondria, mitoplasts and inner
membrane vesicles determined by firefly-Iuciferase luminescence. Eur
J Biochem 67: 1-10, 1976
36. Eijgelshoven MHJ, Lekkerkerk C, Van Beek JHGM, Van Echteld CJA:
Oxygen measurement inside an NMR magnet with a catheter electrode.
Magn Res Med 29: 559-562,1993
37. Kingsley-Hickman PB, Sako EY, Mohanakrishnan P, Robitaille PML,
From AHL, Foker JE, Ugurbil K:31P NMR studies of ATP synthesis
and hydrolysis kinetics in the intact myocardium. Biochem 26: 7501-
7510,1987
38. Connett RJ, Gayeski TEJ, Honig CR: Energy sources in fully aerobic
rest-work transitions: a new role for glycolysis. Am J Physiol 248:
H922-H929, 1985
39. Hak 18, Van Beek JHGM, van Wijhe MH, WesterhofN: Dynamics of
myocardial lactate effflux after a step in heart rate in isolated rabbit
heart. Am J Physiol265: H2081-2085, 1993
40. Eijgelshoven MHJ: Dynamics of cardiac energy metabolism. (thesis),
Free University, Amsterdam, 1995, ISBN 90-9007857-6
41. Katz LA, Koretsky AP, Balaban RS: Activation of dehydrogenase
activity and cardiac respiration: a JIP-NMR study. Am J Physiol255:
HI85-188,1988
42. Heineman FW, Balaban RS: Phosphorus-31 nuclear magnetic resonance
analysis of transient changes of canine myocardial metabolism in vivo.
J Clinic Invest 85: 843--852, 1990
43. McCormack JG, Denton RM: Ca'+ as a second messenger within
mitochondria. Trends Biochem Sci 11: 258-262, 1986
44. Hansford RG: Relation between mitochondrial calcium transport and
control of energy metabolism. Rev Physiol Biochem Pharmacol102:
1-72,1985
45. Eng J, Lynch RM, Balaban RS: Nicotinamide adenine dinucleotide
fluorescence spectroscopy and imaging of isolated cardiac myocytes.
Biophys J 55: 621-630, 1989
46. Katz LA, Koretsky AP, Balaban RS: Respiratory control in the glucose
perfused heart: a JIP-NMR and NADH fluorescence study. FEBS Lett
221: 270-276,1987
47. Heineman FW, Kupriyanov W, Marshall R, Fralix TA, Balaban RS:
Myocardial oxygenation in the isolated working rabbit heart as a
function of work. Am J Physiol262: H255-267, 1992
48. Heineman FW, Balaban RS: Effects of afterload and heart rate on
NAD(P)H redox state in the isolated rabbit heart. Am J Physiol 264:
H433-440,1993
49. Chapman 18: Fluorometric studies of oxidative metabolism in isolated
papillary muscle of the rabbit. J Gen Physiol59: 135-154, 1972
50. White RL, Wittenberg BA: NADH fluorescence of isolated ventricular
myocytes: effects of pacing, myoglobin and oxygen supply. Biophys
J 65: 196-204, 1993
51. Ashruf JF, Coremans JMCC, Bruining HA, Ince C: Increase of cardiac
work is associated with decrease of mitochondrial NADH in the
Langendorffrat heart. Am J Physiol269: H856-862, 1995
52. Miyata H, Silverman HS, Sollott SJ, Lakatta EG, Stem MD, Hansford
RG: Measurement of mitochondrial free Ca'+ concentration in living
single rat cardiac myocytes. Am J Physio1 261 :H1l23-HI134, 1991
53. Brandes R, Bers D: Intracellular Ca'+ increases the mitochondrial
NADH concentration during elevated work in intact cardiac muscle.
Circ Res 80: 82-87,1997
54. Harris DA, Das AM: Control of mitochondrial ATP synthesis in the
heart. Biochem J 280: 561-573,1991
55. Hak 18, Van Beek JHGM, Eijgelshoven MHJ, Westerhof N: Mito-
chondrial dehydrogenase activity affects adaptation of cardiac oxygen
consumption to demand. Am J Physiol264: H448-453, 1993
56. Elliott AC, Smith GL, Allen DG: The metabolic consequences of an
increase in the frequency of stimulation in isolated ferret hearts. J
Physiol (Lond) 474: 147-159, 1994
57. WanB, Doumen C, Duszynski, SalamaG, VaryTC, LaNoue KF: Effects
of cardiac work on electrical potential gradient across mitochondrial
membrane in perfused rat hearts. Am J Physiol 265: H453-460, 1993

58. From AHL, Zimmer SD, Michurski SP, Mohanokrishnan P, Ulstadt
VK, Thoma WJ, Ugurbil K: Regulation of the oxidative phosphory-
lation in the intact cell. Biochem 29: 3731-3743,1990
59. Kim DL, Heineman FW, Balaban RS: Effects of ~-hydroxybutyrate
on oxidative metabolism and phosphorylation potential in canine heart
in vivo. Am J Physiol260: HI767-1773, 1991
60. Eijgelshoven MHJ, Hak JB, Van Beek JHGM, WesterhofN: Adaptation
speed of cardiac mitochondrial oxygen consumption to demand slows
with higher hears rate. Am J Physiol265: HI893-1898, 1993.
61. Eijgelshoven MHJ, Tian X, Van Beek JHGM: Exogenous carbon
substrate supply affects the time course of adaptation of cardiac
mitochondrial oxygen consumption to demand in a temperature
dependent fashion. In: MHJ Eijgelshoven. Dynamics of cardiac energy
metabolism. (thesis), Free University, Amsterdam, 1995, ISBN 90-
9007857 -6, pp 73-98
62. Eijgelshoven MHJ, Tian X, Westerhof N, van Beek JHGM: Is the
adaptation speed of myocardial oxidative phosphorylation to heart rate
steps affected by the type of exogenous substrate? J. Physiol. 479P:
72-73P, 1994
63. Isaacs GH, Sacktor B, Murphy TA: The role ofthe a-glycerophosphate
cycle in the control of carbohydrate oxidation in heart and in the
mechanism of action of thyroid hormone. Biochim BiophysActa 177:
196-203, 1969
64. Zizi M, Forte M, Blachly-Dyson E, Colombini M: NADH regulates
the gating of VDAC, the mitochondrial outer membrane channel. J
Bioi Chern 269: 1614-1616, 1994.
65. Tian X, van Beek JHGM, Eijgelshoven MHJ: Effects of inotropic
stimulation and adenosine on the response time of cardiac mito-
chondrial oxygen consumption to heart rate steps in the isolated rabbit
heart. J Physiol (Lond) 479: P132, 1994
66. Bessman SP, Geiger PJ: Transport of energy in muscle: the phos-
phorylcreatine shuttle. Science 211: 448-452, 1981
67. Meyer RA, Sweeney HL, Kuskmerick MJ: A simple analysis of the
'phosphocreatine shuttle'. Am J Physiol246: C365-377, 1984
68. Mahler M: First order kinetics of muscle oxygen consumption and an
equivalent proportionality between Q02 and phosphorylcreatine level.
Implications for the control of respiration. J Gen Physiol 86: 135-
165,1985
69. van Beek JHGM, Eijgelshoven MHJ, Hak JB, WesterhofN: Control
loop gain of cardiac oxidative phosphorylation via phosphate
metabolites is high and partially explains the dynamic adaptation of
oxygen consumption to demand. FASEB J 7: A313, 1993.
70. Gyulay L, Roth Z, Leigh JS Jr., Chance B: Bioenergetic studies of
mitochondrial oxidative phosphorylation using llphosphorus NMR. J
Bioi Chern 260: 3947-3954, 1985
71. Holian A, Owen CS, Wilson DF: Control of respiration in isolated
mitochondria: quantitative evaluation ofthe dependence of respiratory
rates on [ATP], [ADP] and [PJ Arch Biochem Biophys 181: 164-
171,1977
72. U gurbil K, Petein M, Maidan R, Michurski S, FromAHL: Measurement
of an individual rate constant in the presence of multiple exchanges:
application to myocardial creatine kinase reaction. Biochem 25: 100-
107, 1986
73. Bittl JA, Ingwall JS: Reaction rates of creatine kinase and ATP synthesis
in the isolated rat heart. J Bioi Chern 260: 3512-3517, 1985
74. Conley CK, Kushmerick MJ: Buffering work transitions in myo-
cardium: role of a poorly metabolized creatine analogue. Book of
Abstracts. Society of Magnetic Resonance in Medicine. 9th Annual
Meeting, p. 902, 1990
75. Unitt JF, Schrader J, Brunotte F, Radda GK, Seymour AML: Deter-
mination offree creatine and phosphocreatine in the isolated perfused
rat heart by lH_ and lIP-NMR. Biochim Biophys Acta 1133: 115-
120,1992
343
76. Gellerich FN, Khuchua ZA, Kuznetsov AV: Influence of the mitochond-
rial outer membrane and the binding of creatine kinase to the
mitochondrial inner membrane on the compartmentation of adenine
nucleotides in the intermembrane space of rat heart mitochondria.
Biochim Biophys Acta 1140: 327-334, 1993
77. Gellerich FN, Wagner M, Kapischke M, Brdiczka D: Effect of
macromolecules onADP-<:hanneling into mitochondria. In: JP Mazat.
Proceedings of the 5th BTK-Workshop. Plenum, London, 1993
78. Saks VA, Vasileva E, Belikova YO, Kuznetsov AV, Lyapina S, Petrova
L, Perov NA: Retarded diffusion of ADP in cardiomyocytes: possible
role of mitochondrial outer membrane and creatine kinase in cellular
regulation of oxidative phosphorylation. Biochim BiophysActa 1144:
134-148, 1993
79. Jeneson JAL, Wiseman RW, WesterhoffHY, Kushmerick MJ:Amplified
ADP sensitivity of mitochondria. In: HV Westerhoff, JL Snoep, JE
Wijker, FE Sluse, BN Kholodenko (eds). BioThermoKinetics of the
Living Cell. BioThermoKinetics Press, Amsterdam, 1996, pp 94-95
80. O'Rourke B, Ramza BM, Marban E: Oscillations of membrane current
and excitability driven by metabolic oscillations in heart cells. Science
265: 962-966, 1994
81. Cristea E: Effect of amytal, pyridine nucleotides, and oxaloacetate on
the respiratory inhibition-activation cycles in mitochondria. Arch
Biochem Biophys 113: 273-280,1966
82. Chance B, Williamson JR, Jamieson D, Schoener B: Properties and
kinetics of reduced pyridine nucleotide fluorescence of the isolated
and in vivo rat heart. Biochem Z 341: 357-377, 1965
83. Van der Laarse WJ, Diegenbach PC, Elzinga G: Maximum rate of
oxygen consumption and quantitative histochemistry of succinate
dehydrogenase in single muscle fibres of Xenopus laevis. J Muscle
Res Cell Motill0: 221-228,1989
84. Gollnick PD, Bertocci LA, Kelso TB, Witt EH, Hodgson DR: The
effect ofhigh-i'ntensity exercise on the respiratory capacity of skeletal
muscle. Pfliigers Arch 415: 407-413, 1990
85. Hak JB, Van Beek JHGM, WesterhofN: Acidosis slows the response
of oxidative phosphorylation to metabolic demand in isolated rabbit
heart. Pfliigers Arch. 423: 324-329, 1993
86. Nakanishi T, Seguchi M, Tsuchiya T, Yasukuochi S, TakaoA: Effect of
acidosis on intracellular pH and calcium concentrations in the newborn
and adult rabbit myocardium. Circ Res 67: 111-123,1990
87. van Wijhe MH, van Beek JHGM: Effects of respiratory and metabolic
acidosis on oxidative phosphorylation in isolated rabbit heart
mitochondria. Europ J Physiol430: R174, 1995
88. Flameng W, Andres J, Ferdinande P, Mattheussen M, Van Belle H.
Mitochondrial function in myocardial stunning. J Moll Cell Cardiol
23: 1-11, 1991
89. Mekhfi H, Veksler V, Mateo P, Maupoil V, Rochette L, Ventura-Clapier
R: Creatine kinase is the main target of reactive oxygen species in
cardiac myofibrils. Circ Res 78: 1016-1027, 1996
90. Bittl JA, Weisfeldt ML, Jacobus WE: Creatine kinase of heart
mitochondria. The progressive loss of enzyme activity during in vivo
ischemia and its correlation to depressed myocardial function. J Bioi
Chern 260: 208-214, 1985
91. Zuurbier CJ, Beek JHGM van: The dynamic control of mitochondrial
oxygen consumption in isolated perfused rabbit heart is not affected
by a short period of ischaemia. J Physiol479: 71-72P, 1994
92. Beek JHGM van, Hak JB, Wijhe MH van. Eijgelshoven MHJ,
Westerhof N: The myocardial mitochondrial aerobic capacity is
partially limiting for cardiac oxygen consumption and contractility.
Europ J Physiol 430 (S4): R9, 1995
93. Groot B de, Dijk FJ, Beek JHGM van: Response of cardiac respiration
to steps in heart rate is not slowed after reduction of mitochondrial
capacity in normal and pressure overloaded hearts. Proceedings of
the Congress of the International Union of Physiological Sciences, St.
344

Petersburg, Russia, p. 063.06, 1997 Amsterdam, 1996, pp 119-123

94. JHGM van Beek, ill Hak, MH Wijhe van, N Westerhof: The control 95. Groot B de, Dijk FJ, Beek JHGM van: Limitation of contractile function
exerted by oxidative phosphorylation on respiration in the intact heart. and respiration by mitochondrial aerobic capacity in nonnal and
In: HV Westerhoff, JL Snoep, JE Wijker, FE Sluse and BN Kholodenko pressure overloaded isolated rabbit hearts. Europ J Physiol, abstract
(eds). Biothennokinetics of the Living Cell, BioThennoKinetics Press, 434: R45, 1997
Molecular and Cellular Biochemistry 184: 345-358, 1998.
1998 Kluwer Academic Publishers.

Is it possible to predict any properties of oxidative

phosphorylation in a theoretical way?

Bernard Korzeniewski
Institute of Molecular Biology, Jagiellonian University, al. Mickiewicza 3, 31-120 Krakow, Poland

Abstract
Two theoretical approaches applied to oxidative phosphorylation, namely Metabolic Control Analysis (MCA) [1-7] and Non-
Equilibrium Thermodynamics (NET) [8-11], turned out to be very useful tools for quantitative description and understanding
of control and regulation of this process. However, they were not able to predict any new properties of the considered system.
On the other hand, the previously developed dynamic model of oxidative phosphorylation [12-17], representing a kinetic
approach, allowed to formulate several interesting predictions which can be tested experimentally. The most important of
these predictions are: (1) Different steps of ATP-production must be directly activated to a similar extent asATP-consumption
during stimulation of ATP turnover by calcium-acting hormones as well as by neural signals during muscle contraction; (2) A
universal activator/regulatory mechanism responsible for such a precise balance of activation should be identified; (3) The
flux-force relationship for cytochrome oxidase can be inverse during the transition towards hypoxia and anoxia, when oxygen
concentration falls below 30 J..lM; (4) The flux-force relationship can depend on the way in which the thermodynamic force
is changed; (5) The pattern of metabolic control is completely different in normoxic and hypoxic conditions; in the latter
case cytochrome oxidase has the flux control coefficient close to unity. Thus, the kinetic model of oxidative phosphorylation
seems to be a useful scientific tool, offering some novel theoretical predictions, which then can be tested in the experimental
way. (Mol Cell Biochem 184: 345-358, 1998)

Key words: oxidative phosphorylation, dynamic model, muscle contraction, multi-site activation, hypoxia, cytochrome oxidase

Introduction determining the rate of ATP synthesis in response to

The whole general network of physicochemical events
(chemical reactions and physical processes) underlying the
process of oxidative phosphorylation has been known since
Peter Mitchell's formulation of the chemiosmotic theory [18].
There have been still some controversies concerning details,
for example the proton pomp stoichiometries or particular
mechanisms of the ATP/ADP carrier and ATP synthase
functioning. Nevertheless, in principle we have been able to
answer the question: how mitochondria synthesise ATP? Or,
more precisely, we have been able to answer the qualitative,
mechanistic aspect of this question.
Chance and Williams discovered that an addition of ADP
stimulated the respiration of isolated mitochondria incubated
in the presence of respiratory substrates and oxygen (so
called transition from state 4 to state 3) [19, 20]. They
proposed that the concentration of ADP was the main factor
increased energy demand (ATP hydrolysis). This was the first
hypothesis, which still remains a very attractive possibility,
concerning the regulation of oxidative phosphorylation.
However, this approach was also mostly intuitive and
essentially qualitative.
Later on, many measurements were performed; they
concerned changes in steady-state values as well as time
courses of several parameters (respiration rate, ATP
synthesis, L1p, ATP/ADP, NADH/NAD+, cytochrome c
reduction) during state 4 --? state 3 transition (increase in
[ADP] and/or energy demand) [21-24] and aerobiosis --?
anaerobiosis transition (decrease in oxygen concentration
from saturated value to zero) [25-27]. All such experiments,
although extremely useful and important for our understanding
of the process of oxidative phosphorylation, were not able
to relate changes in fluxes to changes in metabolite con-
centrations and enzyme activities in a quantitative way. There

Address for ojJjJrints: B. Korzeniewski, Institute of Molecular Biology, MagieUonian University, ai. Mickiewicza 3, 31-120 Krakow, Poland
346
was no 'language' able to 'translate' these basic studies into
quantitative concepts of control and regulation, to interpret
them in a quantitative way.
The first attempt to formulate oxidative phosphorylation
in a quantitative way was an application of Non-Equilibrium
Thermodynamics (NET) to this process [8-11]. NET
assumes a linear flux-force relationship and such a linear
dependence between oxidation and phosphorylation fluxes
and thermodynamic forces was actually established by
measurements during state 4 --7 state 3 transition [8, 9, 11],
although it appeared to be only a linear fragment of a more
general sigmoidal relationship [28]. Non-Equilibrium
Thermodynamics, although very useful for explanation of
some, especially energetic, aspects of oxidative ATP
production, has also some disadvantages. This approach is
essentially phenomenological (treats the system as a black
box) and therefore its explanatory power seems to be
relatively small. Furthermore, the validity of its general
applicability is based on many assumptions, for example
that the flux-force relationship is linear and that this
relationship does not depend on the way in which the force
is changed. The first of these disadvantages was overcame to
some extent by Mosaic Non-Equilibrium Thermodynamics
(MNET) [10].
Generally, the thermodynamic approach is a quest of a
unified description of a system. However, unlike in physics,
living organisms are complex biological systems and the
belief that they can be described satisfactory by any simple
mathematical formalism is not realistic.
Metabolic Control Analysis (MCA) turned out to be an
extremely useful tool for understanding of quantitative
aspects of oxidative phosphorylation control. It was
formulated in the 70's [1,2] and firstly applied to this
process in 1982 [3]. MCA allows to quantify the control
exerted by particular steps (enzymes, processes) over the
respiration (and ATP synthesis) flux by the measurement of
flux control coefficients of these steps [1, 2, 6]. The 'top-
down' approach to MCA, dealing with blocks of reactions
rather than single enzymes, is also an excellent tool for
studying quantitative aspects of mitochondrial respiration in
isolated mitochondria [29] and intact cells [30]. Nevertheless,
both methods are not free from some disadvantages either.
They are (especially the 'top-down' approach) phenomen-
ological and not mechanistic. What is more important, the
pattern of control obtained in the frame of Metabolic
Control Analysis is valid for only one particular steady-state
and cannot be extrapolated to other steady-states. In other
words, MCA is able to cope only with infinitesimal changes
in the fluxes and enzyme concentrations (activities), but
not with finite changes occurring in most physiological
conditions. There were made some attempts to overcome
these limitations [31-33], but the theoretical methods used
(although in some cases yielding very promising results)
were not universal since it could not be a priori determined
for which systems they were applicable.
In a sense, both Non-Equilibrium Thermodynamics and
Metabolic Control Analysis are mathematical tools (like
statistics) and not falsifiable scientific theories (in the
Popper's meaning). Therefore, they can be useful or not,
more or less applicable to a given system, but they cannot
be either true or false. This is because they are as they must
be, due to their unique logical (mathematical) structure. They
can serve to interpret the measured properties of oxidative
phosphorylation in a quantitative way; however, they cannot
predict any new properties of this process.
A quite different situation takes place in the case of kinetic
models. A kinetic model of such a complicated system as
oxidative phosphorylation can be never fully and definitely
completed, since during development of the model several
assumptions and simplifications have to be made. The
kinetic descriptions of many steps have to be deduced form
accessible, frequently incomplete, experimental data and
some reactions have to be grouped into blocks, represented
by only approximate phenomenological kinetic descriptions.
Therefore, a great number of alternative kinetic models of
oxidative phosphorylation can be constructed. In a sense, such
models can be 'true' or 'false', depending on how well they
are able to represent a broad range of properties of a modelled
system.
Kinetic models operate on a more basic level than Non-
Equilibrium Thermodynamics and Metabolic Control Analysis
and, at least in principle, all the properties described by these
approaches can be derived from a proper kinetic model.
Therefore, the kinetic approach allows a deeper insight into
the mechanisms underlying the control and regulation of
oxidative phosphorylation. Moreover, a proper kinetic model
allows to investigate how macroscopic properties of the
entire system emerge from microscopic properties of its
components. Another advantage of the kinetic model is that
such a model, if it is 'true', can produce some predictions
resulting from a given particular mathematical structure of the
model (there are many possible mathematical structures of
kinetic models, unlike in NET or MCA). These predictions
can concern different properties not yet measured experi-
mentally and therefore to provide some new insights, and not
just interpretations of the existing results. This makes the
kinetic models a potentially very powerful tool for studying
biochemical systems.
However, a great disadvantage of kinetic models is that they
can be 'false'; moreover, in a sense, they are always 'false'.A
kinetic model of a sophisticated metabolic system is, at best,
only an approximation good enough for a certain set of
properties and valid for a limited range of conditions. This
is because many assumptions and simplifications are made
and only selected parameters, important from the physio-
logical point of view, are taken into account. Therefore,

every model of this type should be carefully tested for as
broad a range of parameters and different conditions as
possible. If the model does not work for a new set of
conditions, it should be modified to cover these conditions
as well. For this reason, any model of this type is never
completely finished. At best, it can be at a given time a good
enough approximation to a valid description ofthe properties
of the considered system we are interested in. In this
situation, the reliability coefficient of the model is the ratio
of the number of different properties and conditions it has
been tested for to the number of parameters adjusted during
the construction of the model. Therefore, this is crucial to
verify any kinetic model of a complicated system in a
possibly broad range of conditions. For this reason, both very
reliable and very 'accidental' kinetic models can exist. Only
the former category can be reasonably used for making
predictions.
The dynamic model of oxidative phosphorylation dev-
eloped by myself and (at the first stage) by Wojciech
Froncisz [12-17] can be regarded as fairly (although by no
means perfectly) well verified by the experimental results,
especially those concerning state 4 ~ state 3 transition and
aerobiosis ~ anaerobiosis transition. The previous model
developed by Bohnensack [34, 35], although very important
as the first one in this field and successfully used in several
theoretical studies, was tested for a considerably smaller set
of the system properties. Furthermore, maybe just for this
reason, it had not been used for making any new predictions.
The present model, discussed in this paper, was able to mimic
in a (semi) quantitative way the following properties and
parameter values of oxidative phosphorylation:
changes in different parameter values (respiration rate, fip,
reduction level of cytochrome c, internal and external
ATP/ADP ratio and others) during state 4 ~ state 3
transition in isolated mitochondria [15],
time courses of these parameters after an addition of a
small amount of ADP to mitochondria in state 4 (state 4
~ state 3 ~ state 4 transition) [15],
flux control coefficients for different components of
oxidative phosphorylation with respect to the respiration
flux at different respiration rates between state 4 and state
3 [15, 36], obtained in the frame of Metabolic Control
Analysis (MCA) [3-5],
flux control coefficients for the oxidation, phosphorylation
and proton leak subsystems over the oxidation (res-
piration), phosphorylation and proton leak fluxes at
different respiration rates between state 4 and state 3 [15],
obtained in the frame ofthe 'top-down approach' to MCA
[29],
time courses of different parameter values during con-
sumption of oxygen by suspension of cells (or mito-
chondria) in a closed chamber (aerobiosis ~ anaerobiosis
transition) [12, 16],
347
kinetic responses of oxidation, phosphorylation and
proton leak subsystems to fip in hepatocytes incubated
with different respiratory substrates [14],
values of the respiration rate, cytochrome c reduction
level and external ATPI ADP ratio at different oxygen
concentrations [16],
linear flux-force relationships, similar to those obtained
for oxidative phosphorylation in the frame of Non-
Equilibrium Thermodynamics (NET) [16],
inhibitor titration curves for cytochrome oxidase,ATPIADP
carrier and other components of oxidative phosphorylation,
obtained experimentally in the frame of MCA [17].
The agreement between simulations and experiments was
at least semi-quantitative and in many cases very good.
Therefore, it can be assumed that the model is able to produce,
at least qualitatively (and probably semi-quantitatively),
correct predictions in the range of conditions it has been
tested for. Nevertheless, the theoretical results obtained with
the aim of the model should be treated with an appropriate
caution. In each case, they can be at best only propositions
which should be necessarily checked experimentally.
However, just for this reason, the model can be a very
useful tool helpful in showing what is worth to be studied
experimentally as well as in quantitative (not intuitive)
interpretation of the existing experimental data.
The dynamic model of oxidative phosphorylation discussed
contains kinetic descriptions of all the steps (enzymes,
processes) taken into account explicitly. Changes in time of
particular variables (for example metabolite concentrations)
are expressed as a set of ordinary differential equations. This
set is integrated numerically, by means of a computer. At the
present stage, the model exists in four versions: for liver
mitochondria, skeletal muscle mitochondria, heart mito-
chondria and hepatocytes. The detailed description of these
versions is given elsewhere [12, 13, 16, 17], and in appendix.
The present article presents the most interesting pre-
dictions of the model. They concern two groups of oxidative
phosphorylation properties, named: (1) Equal activation
rule. This set of predictions concern the problem of parallel
activation of different steps of oxidative phosphorylation by
external effectors having physiological significance. (2)
Strange 'low oxygen' world. Here the predictions concern
the control and regulation of oxidative phosphorylation at
low (hypoxic) oxygen concentrations, in the range around
or slightly above the (relatively very low: below 1 ~M)
apparent Km constant for oxygen.
Equal activation rule
The rate of respiration andATP synthesis can change in time
in a given cell due to the metabolic state of the cell,
especially to the current energy demand. The processes
348

responsible for ATP production, particularly mitochondrial A new theoretical method, called 'proportional activation
oxidative phosphorylation, have to 'know' in some way how approach', was proposed to elaborate these results more
much ATP they should produce in a given moment of time. quantitatively [44]. This approach applies to any system in
Therefore, when the cell receives an external signal informing which a block of reactions A produces a common metabolite
it that some metabolic processes are to be turned on or M, a block of reactions B consumes this metabolite, and M is
speeded up (and thus the ATP turnover increased), both the the only way in which both blocks can affect the activity of
metabolic processes consuming ATP and the reactions each other:
responsible for production of this chemical compound
must be directly or indirectly activated. The mechanisms A B
-----7) M -----7
underlying this activation and the way in which the regulatory
cybernetic signals are transferred is still a matter of debate.
An external activator X can increase the flux J through the
The physiological external regulators, determining the
system via direct activation of either only one of the blocks
energetic state of the cell, are mostly different hormones
or both of them. In the case when one of the blocks is
and neurotransmitters, the latter significantly increasingATP
activated exclusively, or to a greater extent than the other,
utilisation in neurones and myocytes.
the other block is (also) stimulated indirectly, through
The activation of oxidative phosphorylation (increase in
changes in M. On the other hand, M remains constant when
respiration and ATP synthesis) in hepatocytes by hormones
both blocks are activated to exactly the same extent.
acting through calcium ions, such as vasopressin, adrenaline
The so called 'proportional activation coefficient' was
and glucagon, causes some increase in the NADHINAD+
defined [44] as the ratio of the relative activation by X of
ratio, cyt. c2+/cyt. c3+ ratio and ATP/ADP ratio [37-41]. The
M-consuming subsystem B to the relative activation by X
simplest, qualitative, intuitive interpretation of these results
ofM-producing subsystem A:
is that calcium activates directly only the block of reactions
upstream of NADH, cyt. c and ATP, namely the substrate
dehydrogenation [42]. This causes an increase in the above-
mentioned ratios, chiefly in theATPIADP ratio, which in tum
P~ = (::)
x
(1)

activates ATP-consuming reactions, for example gluconeo-

genesis. where A and B are activities of systemA and B, respectively.
However, more quantitative calculations made within the The value of this coefficient can be determined experi-
model of oxidative phosphorylation in hepatocytes suggest mentally from the (initial, in the case of a non-linear curve)
that this interpretation is essentially oversimplified [13]. slope of the relationship between the change in flux M
Taking into account quantitative changes in each intermediate caused by X and the change in J caused by X (see Fig. 1):
metabolite M (M = ~p, NADHlNAD+, cyt.c2+/cyt.c 3+or ATPI
ADP) leads to the conclusion that Ca2+-acting hormones
( dMlM)
stimulate to a similar extent the M-producing block and M- dJ/J (2)
x
consuming block, the former being activated only slightly
more. Stimulation of only the M-producing reactions would and from the (initial) slopes of the kinetic responses of
cause [13]: (1) a much greater increase in M after activation subsystems A and B to changes in M:
than actually observed, (2) an increase in proton leak rather
A dvA/VA
than inATP consumption, (3) an increase in the flux control eM =
dM/M (3)
coefficient of the ATPI ADP carrier after activation (the
opposite effect was reported [43]).
B dviVB
This theoretical prediction inspired an experiment eM =
dM/M (4)
designed to test if the parallel activation of M-production
and M-consumption to a similar extent by Ca 2+-acting The proportional activation coefficient can be calculated as
hormones actually occurs [44]. Changes in the membrane follows [44]:
potential ~'P (which represented M) across the inner
mitochondrial membrane after addition of vasopressin to l-e~ ~
suspension of isolated hepatocytes were studied. Although
pAB
X
=
vasopressin caused an increase in the respiration rate by up
1-e~ ~ (5)
to 35%, essentially no changes in the membrane potential This method was used to fix the proportional activation
were observed [44]. These results proved that ~'P-producing coefficients for different external activators of oxidative
subsystem and ~'P-consuming subsystem were activated to phosphorylation using either the experimental data obtained
a similar extent. Thus the prediction was confirmed. in [44] or other data present in the literature. The proportional

_ -0 steady-state
_- -M" - after activation
rx
100 ............. .
100
flux J (%)
Fig. 1. Calculation of the proportional activation coefficient for the increase
in flux J caused by activation by an external effector X of M-producing
subsystem A and M-consuming subsystem B from experimental data. The
coefficient value is calculated according to Eq. 5, from the slope r~ of the
relationship between the change in metabolite M and change in flux J (both
caused by X) as well as from the slopes e~ and ~ of the kinetic responses
of subsystems A and B, respectively, to changes in M.
activation around different common metabolites M (~p, NADHI
NAD+, cyt.c2+/cyt.c 3+ or ATP/ADP) were determined. The
approximate values of the proportional activation coefficients
calculated are shown in Table 1. The most striking property of
the presented set of proportional activation coefficient values
is that all these values are (very) close to unity. This means that
the balance of activation on both sides ofM (M-production and
M-consumption) is (extremely) good.
Even for respiratory substrates, such as lactate + pyruvate
or fatty acids, which are not expected to be very important
'external' activators of oxidative phosphorylation, ATP
consumption is stimulated by about 112 to 2/3 of the extent
to which ATP production is activated. In the case of more
physiological signals increasing the rate of ATP synthesis,
such as Ca2+ -acting hormones and neural stimulation of heart
and skeletal muscle contraction (also mediated by calcium
ions), the stimulation ofATP-production seems to be almost
equal to stimulation of ATP-consumption. This (almost)
perfect balance is most striking in skeletal muscle, where,
during contraction, the respiration rate can increase 100-fold
(and ATP synthesis even much more, since in the resting
muscle there is some proton leak through the inner mito-
chondrial membrane, which contributes to the resting oxygen
consumption), while theATPIADP ratio decreases only 5-fold.
This remains in a great contrast with isolated muscle mito-
chondria, where, in the absence of parallel activation, the
maximal possible stimulation of respiration is only about 10-
fold and it needs a greater than IO-fold decrease in theATPI
ADPratio.
The above data justify a proposition of a general rule which
349
can be called 'equal activation rule'. This rule states, that in
the oxidative phosphorylation system different external
effectors activate the production and consumption of such
metabolites, asATP, ~p or NADH, to a similar extent.
The above conclusions bear on the still ongoing discussion
concerning the primary mechanism of adjustment of the
rate of energy (ATP) production to the rate of energy
consumption. Two main propositions have been formulated,
which can be called: 'negative feedback' end 'parallel
activation'. Both are presented in Fig. 2.
In the frame of the negative feedback paradigm it is
assumed that only theATP-consuming reactions are directly
activated. The increase in the ATP usage causes a fall ofATP
concentration as well as an increase in ADP and inorganic
phosphate concentrations. SinceATP inhibits andADP (and P)
stimulates respiration (and ArP synthesis) in mitochondria,
these changes speed up the energy supply and equilibrateATP
production withATP consumption, and therefore counteract
a complete exhaustion of ATP. Different primary signals,
constituting this negative feedback, have been proposed.
Their list includes ADP concentration [19, 20], the ATPI
ADP ratio [45, 46],ATPIADP*P j ratio [47], phosphorylation
potential [48], and adenylate energy charge [49] (I proposed
[50] that the problem which of these parameters is the
'true' regulator is mostly semantic, and that the external
phosphorylation potential can be regarded as the most
a, negative feedback X
$
A
+ M j
B
e
,...............,
b, parallel activation
,-------X----,
$ $
_ _ _J.z..'- -. . . . . . . M __--Ioj_ _-I.~
A B
, .. _-
xt_...._ _Jt_,M=const.
I~ :
_ ~_. ____ I
Fig. 2. Two possible general mechanisms adjusting the rate ofM production
(A) to the rate ofM consumption (B). In the case of negative feedback (a),
only M consumption is activated directly by an external effector X, while M
production is activated indirectly, through decrease in M. During parallel
activation (b), M production and M consumption are activated directly by
X to a similar extent and M concentration remains relatively unchanged.
350
'effective' signal, since the respiration rate depends in a near-
linear way on this potential in isolated mitochondria, where
no direct activation ofATP production occurs). Probably still
the most popular idea, especially among the people dealing
with skeletal muscle, is that this is the decrease in ADP
concentration which increases ATP production by mito-
chondria in the case of an enlarged energy demand.
The 'parallel activation' paradigm was firstly formulated
when the stimulation of the irreversible tricarboxylate acid
cycle dehydrogenases (pyruvate dehydrogenase, isocitrate
dehydrogenase and 2-oxoglutarate dehydrogenase) by
calcium ions had been discovered [51-54]. Later reports,
concerning stimulation by calcium also of the ATP synthase
[55] and ATP/ADP carrier [56], reinforced the 'multi-site
activation' hypothesis [7]. However, all these studies
provided very few data that would allow to assess the
quantitative significance of the parallel stimulation of
different steps of the ATP production block.
The crucial data came from in vivo studies on working heart
[57--60]. It was stated that after the stimulation of a heart,
ADP concentration increased only by up to 20%, when the
respiration rate increased 5-fold. The report that there is no
change in,1'I' during stimulation of respiration in hepatocytes
by vasopressin [44] was in line with those results. On the other
hand, ADP concentration increases up to 5-fold during skeletal
muscle contraction, what provokes many people to keep
regarding the negative feedback via ADP concentration as the
primary mechanism responsible for fitting of the rate of ATP
production to the rate of ATP consumption. However, the
increase in the respiration rate is much greater in skeletal
muscle than in heart, up to 100-fold, and the ratio of the
relative increase in flux to the relative increase in [ADP] does
not differ significantly in both cases. r~ is equal to 20%/500%
= 1125 in heart and 500%/10000% = 1120 in skeletal muscle.
In fact, when we calculate the proportional activation
coefficients, we obtain almost the same value for heart and
skeletal muscle (see Table 1), which is very close to unity.
This means that in both cases ATP-production and ATP-
consumption are activated to a very similar extent.
Generally, the proportional activation approach offers a
(semi-)quantitative method, the results of which frequently
contradict some intuitive interpretations and expectations.
This emphasises the significance of quantitative methods in
biochemistry and physiology.
What could be the advantages of the parallel (multi-site)
activation? The two main of them are listed: (1) Constant
concentration of different intermediate metabolites M. If
only ATP consumption were stimulated directly by an
external effector, the concentration of such intermediate
metabolites as acetyl-CoA, NADH, protonmotive force and
ATP would drop very significantly. These metabolites are
used by many reactions necessary for keeping the cell alive
and a decrease in M concentrations would seriously disturb
such processes. Therefore, it is much better to keep M
concentrations as constant as possible by parallel activation.
This problem was discussed in the context of an increase in
enzyme concentrations by genetic manipulation in order to
increase a given metabolic flux for commercial goals [61].
(2) Economy of action. It was probably easier for the
(essentially opportunistic) biological evolution to activate
parallely several steps by an external activator rather than to
stimulate just one step and to make other steps extremely
sensitive to intermediate metabolite concentrations.
Very recent simulations performed using the discussed
model of oxidative phosphorylation suggest that (almost) all
steps of this process must be directly activated to a similar
extent during stimulation of heart and skeletal muscle con-
traction in order to avoid changes in different intermediate
metabolites, especially in the external ATP/ADP ratio, two
orders of magnitude greater than the changes measured
experimentally [61 a]. This implies an existence of a very
universal and precise external activator (regulatory mechan-
ism). It is not clear if the main known candidate for such an
activator, namely calcium, exhibits all the necessary properties.
Calcium-ions hypothesis is not fully satisfactory, since their
effect on many steps (citric acid cycle dehydrogenases, ATP
synthase, ATP/ ADP carrier) in vitro is too small and/or
dubious to explain the phenomena occurring in vivo [51-56].
The proportional activation approach and the dynamic
model of oxidative phosphorylation do not answer the
question concerning the nature of the universal regulator and
mechanistic aspects of parallel activation. However, the
theoretical results obtained with the aid of these tools allow
to formulate a challenge for the future: to identify the
general factor activating (almost) all the enzymes (carriers,
processes) participating in oxidative phosphorylation to a
comparable extent. There exist three possibilities concerning
the nature of such an universal activator: (1) Calcium ions
are the factor sought for. Their action is, for some reasons,
more effective in vivo than in vitro; (2) Calcium ions act
via some protein, functionally analogous to calmodulin,
which is lost or inactivated during mitochondria preparation;
(3) Another activator/mechanism (for example protein
phosphorylation) of yet unknown nature exerts an effect
parallel to the effect of calcium ions in intact cells. The
second possibility can be supported by the fact that the
isolated ATP/ADP carrier is activated by calcium ions, but
the concentration of calcium needed is at least two orders
of magnitude greater than in intact cells [62]. This suggests
that something mediates in 'presenting' calcium ions to the
ATP/ADP carrier.
In conclusion, the dynamic model of oxidative phos-
phorylation and proportional activation approach not only
strongly support the idea of parallel activation, but also show
in a quantitative way that the balance of activation is very
precise (which is not intuitively obvious in the case of

Table 1. Proportional activation coefficients calculated for activation of oxidative phosphorylation by different external effectors around different intermediate
metabolites, as indicated [44].
External effector
Lactate/Pyruvate
Fatty acids
Vasopressin
Glucagon
Ca'+ -acting
hormones
generally
neural stimulation of heart contraction
neural stimulation of skeletal muscle contraction
skeletal muscle contraction, where [ADP] increases 5-fold).
The 'equal activation rule' applied to the particular case of
(heart or skeletal) muscle contraction states that parallel
activation by an external effector is the primary mechanism
. responsible for the adjustment of the rate of ATP production
to the current energy demand in the cell, while the negative
feedback viaADP concentration constitutes only a secondary,
'fine tuning' mechanism. This means that the majority of the
stimulation of ATP production is due to a direct activation
of this block and only a small part of this activation is due
to the decrease in the ATP/ADP ratio.
Strange 'low oxygen' world
The process of oxidative phosphorylation may proceed under
different conditions. In vivo, especially in skeletal muscle,
this process undergoes two main kinds of changes in
external conditions, namely varying energy demand and
decreasing oxygen concentration. These kinds of changes
can be represented in isolated mitochondria by state 4 ~
state 3 transition and aerobiosis ~ anaerobiosis transition,
respectively. The quantitative aspects of regulation of state
4 ~ state 3 transition have been extensively studied in three
main aspects, using appropriate approaches: (1) Kinetic
approach, involving changes in fluxes (respiration, ATP
synthesis) and metabolite concentrations (L\.p, external and
internal ATP/ADP ratio, NADH/NAD+ ratio etc.) [21-24];
(2) Thermodynamic approach, consisting in measurements
of the flux-force relationship between oxidation and
phosphorylation thermodynamic potentials and fluxes. It was
obtained that the flux-force relationship during state 4 ~
state 3 transition was nearly linear [8-11], which turned out
to be a linear part of a more general sigmoidal dependence
[28]; (3) Metabolic Control Analysis, in the frame of which
flux control coefficients of particular steps of oxidative
phosphorylation were measured [3-5, 29, 30]. Those studies
showed that proton leak was the main step controlling the
oxygen consumption flux in state 4, control was shared
351
Intermediate metabolite Proportional activation coefficient
L'lp 0.5--0.7
L'lp 0.5-0.7
L'lp 1.0-1.2
L'lp ~ 1.0
NADH/NAD+, cyt.c'+/cyt.c l +,
L'lp, ATP/ADP 0.8--1.0
ATP/ADP ~l.l
ATP/ADP ~l.l
between several steps in state 3, while ATP usage kept most
of the control in 'physiological' state 3.5 (about 50% of the
maximal respiration rate). All those studies were performed
at high, saturated oxygen concentrations.
On the other hand, relatively few experiments have been
performed at low oxygen concentrations, in the range of few
micromoles, which take place in intact tissues [25-27].
Moreover, all those studies concerned the kinetic approach,
especially the dependence of the oxygen consumption and
metabolite concentrations (cyt. C2+/cyt. C3+, ATPIADP) on the
oxygen concentration [25, 26]. It was shown that both
respiration and metabolite concentrations were relatively
constant until the oxygen level dropped below a few
micromoles. However, neither thermodynamic properties
nor the pattern of control were measured experimentally
at low oxygen concentrations or during aerobiosis ~
anaerobiosis transition, mainly because of technical diff-
iculties. Therefore, this field offered a good opportunity for
making some theoretical predictions using the dynamic
model of oxidative phosphorylation. The theoretical results
obtained exhibit many very interesting and unexpected
properties and these predictions, if correct, allow a first
glimpse into the strange 'low oxygen' world, in which
cytochrome oxidase seems to play the main role. This world
appears for the oxygen concentration near or slightly above
the apparent Km constant for oxygen (about 1 flM).
First of all, the flux-force relationship of cytochrome
oxidase was simulated. It is widely, although implicitly,
accepted [8-10] that a reaction (or a block of reactions)
displaced from thermodynamic equilibrium, exhibit the
following three properties: (1) The dependence of a flux
on a force is proportional (flux increases when force
increases), although not necessarily linear; (2) There exists
a unique flux-force relationship independent of the way in
which the thermodynamic force is changed; (3) The regulation
of the reaction rate via changes in metabolite concentrations
is the same in a given set of conditions when the system
undergoes changes caused by modification of different
external parameter values.
352

None of these intuitive 'rules' turned out to be fulfilled in to the oxygen concentration of about 30 J,lM) it becomes
the case of cytochrome oxidase. The total thermodynamic reversed (the flux decreases when the force increases).
span of the reaction catalysed by cytochrome oxidase X tot The increase in the respiration rate during state 4 ---7 state
is a sum of three thermodynamic potentials: 3 transition is caused by an increase in Xc (i.e. in the
reduction level of cytochrome c) as well as by an increase
(6) in X H (decrease in Llp), while Xo (oxygen concentration)
remains constant. Similarly, during aerobiosis ---7 anaero-
where biosis transition, the increase in Xc and X H compensates for
the decrease in the respiration rate caused by lowered oxygen
X H = - (2 + 2 u) . Llp (7) tension (and Xo)' However, the contribution of both thermo-
dynamic potentials to the regulation of oxygen consumption
Xc =-2' LlEh (8) is different in both cases [16]. This is shown in Table 2. The
eyt. e
calculations were made for state 3.5 and oxygen con-
(9) centration equal to 30 J,lM, which is the average concentration
in intact tissues [7, 27]. The increase in the rate of the
and u = Ll' /Llp, LlE m = 250 m V, LlEmo = 820 m V. The reaction catalysed by cytochrome oxidase caused by
ct,c 2
thermodynamic potentIals were calculated as follows: increased energy demand is mainly due to the decrease in
the protonmotive force (contribution greater than 80%),
while the increase in the reduction level of cytochrome c
plays only a minor role (contribution less than 20%). On the
(10)
other hand, both Llp and cytochrome c participate in the
compensation of a decrease in the respiration rate caused by
diminished oxygen tension. Their contribution is comparable,
(11) the 'regulatory strength' of the reduction level of cytochrome
c being even slightly higher. Without this compensation,
1 decrease in the oxygen consumption would be about ten-fold
Llp = Ll ' + LlpH = ~ LlpH
l-u (12) greater for the same change in oxygen concentration.
From the above theoretical results it can be concluded
that: (1) During aerobiosis ---7 anaerobiosis transition the
dependence of the oxygen consumption flux on the thermo-
(13) dynamic span of cytochrome oxidase is reverse at lower

Generally, changes in X H, Xc and Xo reflect changes in Llp, 80

5'
reduction level of cytochrome c and oxygen concentration, S 60
respectively, while change in the total thermodynamic span c
C'O
Q.
X tot is a resultant of those changes. en 40
The simulated changes in the total thermodynamic span u ,
E 20 \
of cytochrome oxidase X tot as a function of changes in the C'O
c
>-
respiration rate during state 4 ---7 state 3 transition and -0
0
0
aerobiosis ---7 anaerobiosis transition [16] are shown in Fig. E
- 1
(])
3. The reference point, shown in the centre of the diagram, ; -20 ~
is 'physiological' state 3.5 (about 50% of the maximal 0
-40
(])
respiration rate) at saturated oxygen concentration (240
1
OJ
c
J,lM). State 4 ---7 state 3 transition was simulated by changing C'O
-c
u -60
the rate constant (or concentration) of hexokinase between ~---'----~I----'----'-----'----'----'
o (state 4) and the value 6-fold greater (state 3) than in the 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
reference point. The transition to anaerobiosis was obtained respiration rate (arbitrary units)
by a decrease in the oxygen concentration from the saturated
value to zero. It can be clearly seen that, while during state Fig. 3. Simulated relationship between the respiration rate and changes in
the thermodynamic span of cytochrome oxidase during state 4 ~ state 3
4 ---7 state 3 transition the flux-force relationship is
transition (solid line) and aerobiosis ~ anaerobiosis transition (dashed
proportional and near-linear, during aerobiosis ---7 anaero- line). The reference state, represented as a point in the middle of the diagram,
biosis transition, this relationship is strongly non-linear and is state 3.5 (about 50% of the maximal respiration rate) at oxygen
below a certain value of the respiration rate (corresponding concentration of 240 11M.

Table 2. Contributions of changes in particular thermodynamic potentials
to changes in the respiration rate during state 4 --> state 3 and aerobiosis-->
anaerobiosis transitions. i1v H, i1v c' and i1v 0 indicate changes in the
respiration rate cased by changes in X H, Xc and Xo' respectively. Their
values are standardized for the resultant change i1v tot ' being the sum of
these changes, equal to I or -I Simulations were performed for state 3.5
(about 50% of maximal respiration) and oxygen concentration equal to 30 j.lM
Change in respiration State 4 --> state 3 aerobiosis --> anaerobiosis
rate transition transition
i1vH 0.83 4.0
i1v c 0.17 5.3
i1vo 0 -10.3
i1v tot -1
(below 30 J.1M), but still physiological oxygen concentrations.
This contradicts many common expectations based on the Non-
Equilibrium Thermodynamics (NET) applied to oxidative
phosphorylation, although not the essence of this theory. (2)
The flux-force relationship is completely different when the
flux is changed by lowered oxygen concentration and by
increased energy demand. This imposes a very important
limitation on the applicability of the Non-Equilibrium
Thermodynamics (NET) to the description of oxidative
phosphorylation, since this approach takes into account only
how much, but not in which way a particular thermodynamic
force is changed and therefore, according to NET, the flux-
force relationship should be identical in both cases discussed
above. For this reason, the kinetic approach seems to be
much more appropriate for quantitative description of
oxidative phosphorylation. (3) The regulation of cytochrome
oxidase is different in both transitions. The decrease in ~p
is the main factor causing the increase in the rate of the
reaction catalysed by cytochrome oxidase when the energy
demand increases. On the other hand, the reduction level of
cytochrome c is at least as important as ~p in compensation
of the decrease in the respiration rate brought about by
diminished oxygen pressure. Therefore, ~p seems to be
universal regulatory factor of cytochrome oxidase, while the
cyt.c 2+/cyt.c 3+ ratio probably plays an important role mainly
in compensation ofthe decrease in oxygen concentration [25].
The most surprising is perhaps the first of the above three
properties, i.e. the increase in the total thermodynamic flux
of the reaction catalysed by cytochrome oxidase when the
oxygen concentration (and respiration rate) decreases. This
can be interpreted as a 'thermodynamic cost' of kinetic
regulation. The purpose if this regulation is to keep the
respiration as constant as possible when oxygen pressure
decreases. This leads, because of kinetic properties of
cytochrome oxidase, to an 'overcompensation' of the
decrease in Xo (being a function of oxygen concentration)
by the increase in Xc (proportional to reduction level of
cytochrome c) and X H (negatively proportional to the
protonmotive force). The theoretical results obtained
353
suggest that, in the case of decrease in oxygen concentration,
it is more important for the cell to optimise the flux through
cytochrome oxidase rather than the thermodynamic force
ofthe reaction catalysed by this enzyme. The thermodynamic
efficiency is here sacrificed in order to maintain oxygen
consumption and ATP synthesis unchanged.
However, it is important for the cell to maintain not only
a high rate of ATP production, but also a high concentration
of this compound or, rather, a high value of the external
phosphorylation potential. When oxygen concentration
decreases below 30 J.1M, cytochrome oxidase becomes more
and more displaced from equilibrium, partially because ~p
decreases significantly. Since ATP synthase is supposed to
work near equilibrium, the value of the internal phosphoryl-
ation potential should follow changes in the protonmotive
force. The external phosphorylation potential is related to
the internal phosphorylation potential mainly through the
ATP/ADP carrier. It was shown [63] that this carrier can
compensate the increase in the thermodynamic span of
cytochrome oxidase during transition to anaerobiosis by a
decrease in its own thermodynamic span (displacement from
equilibrium), as can be deduced from its high positive
thermodynamic response coefficient.
The thermodynamic response coefficient describes the
relative change in the thermodynamic span of a given
reaction as compared to the change in the thermodynamic
span through the entire considered system during a particular
transition. This coefficient was defined as follows [63]:
(14)
where k, j, i denote the indices of the external metabolite
(substrate, product or effector) Mk , whose concentration is
changed, of the considered system Vj and of the enzyme ei
(being a component of this system), respectively. ~Gi and
~G stand for the infinitesimal changes in the Gibbs free
)
energy difference of the reaction catalysed by enzyme ei, and
in the Gibbs free energy difference of the whole considered
system U,) respectively. L1) denotes the relative flux through
the enzyme ei.
A negative value of the thermodynamic response co-
efficient of cytochrome oxidase was calculated for aero-
biosis ~ anaerobiosis transition when oxygen concentration
fell below 30 J.1M [63]. At the same time, thermodynamic
response coefficients of other steps, particularly ATP/ADP
carrier had relatively high positive values (thermodynamic
response coefficients of all the steps of any considered
system fulfil the summation property, since they sum up to
unity). This caused that, despite a decrease in ~p and the
internal phosphorylation potential at lower oxygen con-
centrations, the external phosphorylation potential remained
relatively constant. The ATP/ADP carrier acted as a 'thermo-
dynamic buffer', quickly diminishing its own displacement
354
from equilibrium when oxygen concentration decreased. As a
result, the concomitant action of cytochrome oxidase and the
ATP/ADP carrier caused that both the ATP synthesis and the
external phosphorylation potential remained relatively high at
low (a few micromoles) oxygen concentration. Therefore,
despite the fact that the thermodynamic efficiency of cyto-
chrome oxidase decreased significantly when oxygen con-
centration decreased, the efficiency of the whole system
remained much more constant due to the 'thermodynamic
buffering' by theATP/ADP carrier and some other steps [63].
Generally, even if the predictions produced by the model
were not confirmed experimentally, the simulations would
show on a particular example that there is no a priori reason
for the flux-force relationship to be proportional and not
reverse, and that this relationship can depend on the way in
which a given thermodynamic force is changed. Such a
possibility was not discussed explicitly in earlier studies in
the context of oxidative phosphorylation [8-10] and
imposes a serious limitation for an application of the Non-
Equilibrium Thermodynamics to real systems.
Besides thermodynamic properties, also the pattern of
flux control coefficients (defined in the frame of Metabolic
Control Analysis) of cytochrome oxidase and other steps
was studied at low oxygen concentrations by means of the
dynamic model of oxidative phosphorylation [17,36]. It was
obtained [17] that the dependence of the respiration flux
on the concentration (activity) of cytochrome oxidase
(corresponding to the titration of this enzyme with KCN)
exhibit a clear threshold: sudden decrease in the respiration
rate below a certain concentration of this enzyme. The relative
concentration at which this threshold occurred changed
with oxygen concentration and the energetic state of
mitochondria (these simulations were performed using the
version of the model for skeletal muscle mitochondria).
Both decrease in oxygen concentration and increase in
energy demand (amount of hexokinase added in the presence
of glucose) shifted the threshold towards higher con-
centrations (activities) of cytochrome oxidase [17]. This
theoretical result is able to offer an explanation why inborn
deficiencies of cytochrome oxidase, caused by mutations
in mitochondrial DNA (mtDNA), impair the process of
oxidative phosphorylation particularly in skeletal muscle
(tissue specificity), which undergoes extremely high
changes in ATP utilisation and frequently works under
hypoxic conditions.
The pattern of control exerted by particular steps of
oxidative phosphorylation over the oxidation flux in isolated
skeletal muscle mitochondria was simulated in state 4, state
3 and intermediate (perhaps most physiological) state 3.5
(about 50% of the maximal respiration rate) at normoxic
(240 11M) and hypoxic (311M [27]) oxygen concentration
[36]. The theoretical results obtained are presented in Table
3. In hypoxic conditions, 'state 4', 'state 3.5' and 'state 3'
are defined rather in terms of the amount of hexokinase
added (the same as in corresponding states at saturated
oxygen concentration) than in terms of respiration rates. At
low oxygen concentrations, the access of oxygen becomes
limiting at higher fluxes and therefore the respiration rate in
'state 3' does not differ significantly from the respiration rate
in 'state 3.5' (last row in Table 3).
The most interesting property of the data presented in
Table 3 is that the values of flux control coefficients of
particular steps are different in different conditions. Proton
leak is the main controlling factor in state 4 (both in normoxic
and hypoxic conditions): its flux control coefficient is equal
to about 0.7. Almost 90% of the control over oxygen
consumption in the intermediate state 3.5 is kept by ATP
consumption (hexokinase) at high oxygen concentration. On
the other hand, at low oxygen pressure, hexokinase shares
the control with cytochrome oxidase. Finally, in state 3
complex III exerts the greatest control over the respiration
flux at normoxic oxygen concentration, while under hypoxic
conditions cytochrome oxidase is the main controlling
factor, being responsible for more than a half of the control.
Generally, at low oxygen pressure and moderate or high
energy demand the control is shifted from other steps to
cytochrome oxidase.
A simulated dependence of the flux control coefficients
of cytochrome oxidase and ATP utilisation on oxygen
concentration is presented in Fig. 4. At higher oxygen levels
(> 10 11M) ATP consumption is the main controlling factor,
while at low oxygen concentrations I 11M) cytochrome
oxidase keeps almost all the control. A steep threshold
occurs at about 3 11M, where the flux control coefficient of
ATP usage suddenly decreases with diminishing oxygen
concentration, while the flux control coefficient of cyto-
chrome oxidase behaves in the opposite way, in order to
fulfil the summation property. The value corresponding to
the threshold in oxygen concentration depended on the
energy demand (amount of hexokinase): a decrease in the
rate constant of ATP utilisation brought about a decrease in
this value (data not shown).
More generally, the threshold effect in the respiration rate
and/or flux control coefficients was co-determined by three
factors: relative concentration (activity) of cytochrome
oxidase (as discussed above), energy demand and oxygen
concentration. The threshold value of cytochrome oxidase
activity, oxygen pressure and/or energy demand, at which
oxidative phosphorylation was impaired, was increased by:
(I) a decrease in cytochrome oxidase activity (at a given
oxygen concentration and energy demand), (2) an increase in
the rate constant of ATP usage (at a given active cytochrome
oxidase concentration and oxygen level) and (3) a decrease
in the oxygen concentration (at a given cytochrome oxidase
activity and energy demand). This has important implications
for the understanding of the effect of inborn enzyme
 355

Table 3. Simulated values of flux control coefficients of particular steps of oxidative phosphorylation in skeletal muscle mitochondria in state 4, state 3.5
(about 50% of maximal respiration rate) and state 3 for saturated oxygen concentration (240 flM) and hypoxic oxygen concentration (3 flM). Particular states
are defined in therms of the amount of hexokinase rather than the respiration rate The values written in bold characters correspond to the highest flux control
coefficients under particular conditions

Enzyme (process) Flux control coefficient

state 4 state 3.5 state 3

[0,] 240 flM 3 flM 240 flM 3flM 240 flM 3 flM

substrate
dehydrogenation 0.19 0.15 0.02 om 0.13 0.12
complex I 000 0.00 0.00 0.01 0.12 0.01
complex III 0.00 0.00 0.00 0.01 0.29 0.02
complex IV 0.13 0.14 0.01 0.29 om 0.54
proton leak 0.68 0.71 0.08 0.02 0.01 0.02
ATP synthase 0.00 0.00 000 0.05 0.14 0.08
ATP/ADP carrier 0.00 0.00 0.00 0.12 0.13 0.14
phosphate carrier 0.00 0.00 0.00 0.03 0.09 0.05
hexokinase 000 0.00 0.88 0.39 003 0.03

respiration rate 0.42 0.27 2.40 2.10 4.47 2.44

(arbitrary units)

deficiencies, causing diseases of skeletal muscle (myopathies).
Particularly, a defected cell with a decreased level of cyto-
chrome oxidase is not able to cope with so great variations in
energy demand and oxygen tension as a normal cell is.
model allows to show what is worth to be studied, to be
tested and to be looked for. Additionally, it offers a quan-
titative tool for elaboration of experimental measurements,
giving theoretical results which frequently contradict intu-
itive, qualitative interpretations.

Conclusions
The dynamic model of oxidative phosphorylation discussed
in the present paper seems to be a novel and useful tool for
studying quantitative aspects of cellular bioenergetics. It has
produced several interesting predictions concerning the
properties of mitochondrial metabolism which has not been
studied experimentally. Of course, all these predictions will
have to be corroborated experimentally. Nevertheless, the
Appendix
(concise description of the model)
Three versions ofthe dynamic model of oxidative phosphorylation, namely
for liver mitochondria, liver cells and muscle mitochondria, were used in the
simulations which gave the theoretical results discussed in the present paper.
The complete description of even one of these versions would occupy
several pages. Therefore, only a short description ofthe version for muscle

.
mitochondria is presented here as an example. The complete description of
all three versions can be found elsewhere [12, 13, 16, 17].
1.0 - - The model explicitly takes into account the following components of
cytochrome - ~ oxidative phosphorylation: substrate dehydrogenation (reduction ofNAD
\
C oxidase and/or UQ by different processes), respiratory chain (complex I, complex
(J) 0.8 hexokinase
'(3 \ III and complex IV), proton leak,ATP synthase,ATP/ADP carrier, phosphate
ti=
\
carrier, ATP utilisation (hexokinase + glucose system in isolated mito-
(J)

u
0 0.6 chondria) and adenylate kinase. The kinetics of each step is described by
an appropriate kinetic equation, expressing the dependence of the rate of
a
I- this step on different metabolite concentrations. The kinetic equations
C 0.4 for muscle mitochondria are presented in Table AI.
0
u The rates of changes of different metabolite concentrations,
X
::::J 0.2 constituting independent variables, in time are expressed as a set of
~
ordinary differential equations. Such a set is presented in Table A2.
This set is integrated numerically, by means of a computer. The Gear
integration procedure is used and the simulation programs are written
o 10 100 1000 in Microsoft FORTRAN programming language.
oxygen concentration (11M) Different levels of energy demand are fixed in the model by using different
rate constants ofATP utilisation, corresponding to different concentrations
Fig. 4. Simulated dependence of the flux control coefficients of cytochrome of hexokinase. Particularly, this constant is equal to zero in state 4.
oxidase and ATP usage (hexokinase) on oxygen concentration. State 3 corresponds to the concentration of hexokinase which causes
356

TableAl. Kinetic descriptions of the components of oxidative phosphory- Table A2. The set of differential equations, expressing the rates of
lation taken into account explicitly in the model metabolite concentration changes in time

Substrate dehydrogenation:

VOH = kOH K D ,KmN= 100'Po=06

(1 + NA;~ADHl
Complex I:
02 =-0.5 . V C4 or =0
H;= (2 + 2 u) VC4 + (4 - 2 u) . VC3 + 4 VCI - nA . VSN - u v EX -
Complex III: (I-u)' Vpl-VLK

ComplexlY:

ATP" = VEX - VUT + VAK

ATP synthase:
y-l
v =k - - ,y= 10~GsNFlz
SN SN y+ I
u = ~'P/~p
ATPIADP carrier:
nA= 2 5, H+/ATP stoichiometry
_ ADP c, ADP fi ) .

VEX - ~x ~DP" + ATP" . 1O'I"\z ADPfi + ATPfi . lO'I'i\Z

I no further increase in the respiration rate. At low (hypoxic) oxygen

( ) , KmAop = 3.5 11M
1 + KmAo.,lADP c,
Phosphate carrier:
v pI = k.l (Pi, . H, -Pi,' H),
ATP usage (hexokinase):
1 as a constant parameter.
VUT = ~T ' KmA = 150 11M
1 + KmA/ATP"
Proton leak:
VLK = kLKI . (I OkLK2 ' t>'I' -I) + kLK3 . ~P
Adenylate kinase:
vAK = k'AK . ADP c, ADP m, - k.AK . ATP m, . AMP,
Chosen parameters and variables:
KmN , - Michaelis-Menten constant of substrate dehydrogenation for the
NAD+/NADH ratio; Kmo, - Michaelis-Menten constant of cytochrome
oxidase for oxygen at constant Ap and reduction level of cytochrome c;
KmA - Michaelis-Menten constant of hexokinase for ATP; KmCI' KmC3 - Half
saturation' constants of complexes I and III for their thermodynamic spans;
~ECl' ~EC3 - Thermodynamic spans of complexes I and III; ~GSN' ~GEX
Thermodynamic spans of reactions catalysed by ATP synthase and ATPI
ADP carrier; a2+, c2+- Concentrations of the reduced form of cytochromes
a 3 and c; Po - Relative 'sensitivity coefficient' of substrate dehydrogenation
to the NAD+/NADH ratio; kLKI' k LK2 , kLK3 - Constants used in a
phenomenological description of proton leak; kfAK' k.AK - Forward and
backward rate constants of adenylate kinase.
subscripts: e - external; i-internal; t - total; f - free; m - magnesium
complex.
concentrations the respiration rate in state 3 (at the same rate constant
for ATP utilisation as in state 3 for normoxic conditions) is much
smaller than at high (saturating) oxygen concentrations, since at higher
values of the flux the access of oxygen becomes limiting. Different
concentrations of oxygen are simply fixed in particular simulations
.
The activation of particular steps corresponded to an increase in
appropriate rate constant values in a given (reference or 'resting') steady-
state. When the new ('active') steady-state is reached, the changed values
of fluxes and metabolite concentrations are recorded.
To obtain the flux control coefficients of a particular step in a given steady
state, the rate constant of this step is decreased by a relative factor of 0.01
(I %). When a new steady is reached, the flux control coefficient is calculated
by a division of a relative change in the respiration rate by this factor.
The general structure of the model versions for liver mitochondria
or cells is the same .as the structure of the model for muscle
mitochondria. The existing differences concern some details of kinetic
descriptions of particular steps, the number of respiratory chain
complexes involved (complex I is absent in the version for liver
mitochondria respiring on succinate) and the volumes in which
metabolite concentrations are determined (the model for liver cells
involves the cytosolic compartment).
References
I. Kaeser H, Bums JA: The control of flux. Symp Sos Exp Bioi 32: 61-104,
1973
2. Heinrich R, RapoportTA: A linear steady-state treatment of enzymatic
chains. General properties, control and strength. Eur J Biochem 42: 89-
95, 1974

3. Groen AI<, Wanders RJA, Westerhoff HV, van der Meer R, Tager JM:
Quantification of the contribution of various steps to the control of
mitochondrial respiration. J Bioi Chern 257: 2754--2757, 1982
4. Lettelier T, Malgat M, Mazat J-P: Control of oxidative phosphorylation
in rat muscle mitochondria: Implications for mitochondrial myopathies.
Biochim Biophys Acta 1141: 58--64, 1993
5. Jouaville L, Ichas F, Lettelier T, Mazat J-P: Control of ATP flux
in rat muscle mitochondria. In: S Schuster, M Rigoulet, R Ouabi,
J-P Mazat (eds). Modern Trends in BioThermoKinetics. Plenum
Press, New York, 1993, pp 319-325
6. Fell DA: Metabolic Control Analysis: A survey of its theoretical and
experimental development. Biochem J 286: 313-330, 1992
7. Brown GC: Control of respiration and ATP synthesis in mammalian
mitochondria and cells. Biochem J 284: 1-13, 1992
8. Rottenberg H: The thermodynamic description of enzyme-catalyzed
reactions. BiophysJ 13: 503-511, 1973
9. Rottenberg H: Non-equilibrium thermodynamics of energy conversion
in bioenergetics. Biochim Biophys Acta 549: 225-253, 1979
10. WesterhoffHV, van Dam K: Thermodynamics and control of free-energy
transduction, Elsevier, Amsterdam, 1987
II. Stucki JW: The optimal efficiency and the economic degrees of coupling
of oxidative phosphorylation. Eur J Biochem 109: 269-283, 1980
12. Korzeniewski B, Froncisz W: An extended dynamic model of oxidative
phosphorylation. Biochim Biophys Acta 1060: 210--223, 1991
13. Korzeniewski B, Froncisz W: Theoretical studies on the control of the
oxidative phosphorylation system. Biochim Biophys Acta 1102: 67-
75,1992
14. Korzeniewski B: Simulation of oxidative phosphorylation in hepato-
cytes. Biophys Chern 58: 215-224,1996
15. Korzeniewski B: Simulation of state 4 --j state 3 transition in isolated
mitochondria. Biophys Chern 57: 143--153, 1996
16. Korzeniewski B: Regulation of cytochrome oxidase: Theoretical studies.
Biophys Chern 59: 75-86,1996
17. Korzeniewski B, Mazat J-P: Theoretical studies on the control of
oxidative phosphorylation in muscle mitochondria: Application to
mitochondrial deficiencies. Biochem 1319: 143--148, 1996
18. Mitchell P: Coupling of phosphorylation to electron and hydrogen
transfer by a chemiosmotic-type of mechanism. Nature 191: 144--148,
1961
19. Chance B, Williams GR: Respiration enzymes in oxidative phosphorylation.
I. Kinetics of oxygen utilization. J Bioi Chern 217: 383--393, 1955
20. Chance B, Williams GR: The respiratory chain and oxidative phos-
phorylation. Adv Enzymol17: 65-134, 1956
21. Nicholls DG: The influence of respiration andATPhydrolysis on the
proton-electrochemical gradient across the inner membrane of rat-liver
mitochondria as determined by ion distribution. Eur J Biochem 50:
305-315,1974
22. Duszynski J, Bogucka K, Wojtczak L: Homeostasis of the protonmotive
force in phosphorylating mitochondria. Biochim Biophys Acta 767:
540--547,1984
23. Bohnensack R, Gellerich FN, Schild L, Kunz W: The function of the
adenine nucleotide translocator. Biochim BiophysActa 1018: 182-184,
1990
24. Smith CM, Bryla J, Williamson JR: Regulation of mitochondrial a-
ketoglutarate metabolism by product inhibition at a-ketoglutarate
dehydrogenase. J Bioi Chern 249: 1497-1505, 1974
25. Wilson DF, Ereciriska M, Drown C, Silver IA: The oxygen dependence
of cellular energy metabolism. Arch Biochem Biophys 195: 485--493,
1979
26. Kashiwagura T, Wilson DF, Ereciaska M: Oxygen dependence of cellular
metabolism: the effect of 0, tension on gluconeogenesis and urea
synthesis in isolated rat hepatocytes. J Cell Physiol 120: 13--18, 1984
27. Gnaiger E, Steinlechner-Maran R, Mendez G, Eberl T, Margerite R:
357
Control of mitochondrial and cellular respiration by oxygen. J Bioenerg
Biomembr 27: 583-596, 1995
28. Wanders RJA, Westerhoff HV: Sigmoidal relation between mitochondrial
respiration and log([ATP]/[ADPD,", under conditions of extramitochondrial
ATP utilization. Implications for the control and thermodynamics of
oxidative phosphorylation. Biochemistry 27: 7832-7840, 1988
29. Hafner RP, Brown GC, Brand MD: Analysis of the control of
respiration rate, phosphorylation rate, proton leak rate and
protonmotive force in isolated mitochondria using the 'top-down'
approach of metabolic control theory. Eur J Biochem 188: 313-
319, 1990
30. Brown GC, Lakin-Thomas PL, Brand MD: Control of respiration and
oxidative phosphorylation in isolated rat liver cells. Eur J Biochem 192:
355-362, 1990
31. Small JR, Kacser H: Responses of metabolic systems to large changes
in enzyme activities and effectors. I. The linear treatment of unbranched
chains. Eur J Biochem 213: 613--624, 1993
32. Small JR, Kacser H: Responses of metabolic systems to large changes
in enzyme activities and effectors. 2. The linear treatment of branched
pathways and metabolite concentrations. Assessment of the general
non-linear case. Eur J Biochem 213: 625-640, 1993
33. Small JR, Kacser H: A method for increasing the concentration of a
specific internal metabolite in steady-state systems. Eur J Biochem
226:649--656,1994
34. Bohnensack R: Control of energy transformation in mitochondria:
Analysis by a quantitative model. Biochim Biophys Acta 634: 201-
218,1981
35. Bohnensack R, Kuster U, Letko G: Rate-controlling steps of oxidative
phosphorylation in rat liver mitochondria; a synoptic approach of model
and experiment. Biochim BiophysActa 680: 271-280, 1982
36. Korzeniewski B, Mazat J-P: Theoretical studies on control of oxidative
phosphorylation in muscle mitochondria at different energy demands
and oxygen concentrations. Acta Biotheoretica 44: 263--269, 1996
37. Taylor WM, Van de Pol E, Bygrave FL: On the stimulation of respiration
by a-adrenergic agonists in perfused rat liver. Eur J Biochem 155: 319-
322, 1986
38. Kimura S, Suzaki T, Kobayashi S,Abe K, Ogata E: Effects of glucagon
on the redox states of cytochromes in mitochondria in situ in perfused
rat liver. BiochemBiophys ResComm 119: 212-219,1984
39. Strzelecki T, Thomas JA, Koch CD, LaNoue KF: The effect of hormones
on proton compartmentation in hepatocytes. J Bioi Chern 259: 4122-
4129,1984
40. Titheridge MA, Haynes RC: The hormonal stimulation ofureogenesis
in isolated hepatocytes through increases in mitochondrial ATP
production. Arch Biochem Biophys 20 I: 44--55, 1980
41. Quinlan PT, HalestrapAP: The mechanism ofthe hormonal activation
of respiration in isolated hepatocytes and its importance in the
regulation of gluconeogenesis. Biochem J 236: 789-800, 1986
42. Brand MD, Murphy MP: Control of electron flux through the respiration
chain in mitochondria and cells. Bioi Rev 62: 141-193, 1987
43. Tager JM, Groen AI<, Wanders RJA, Duszyriski J, Westerhoff HV,
Vervoom RC: Control of mitochondrial respiration in rat-liver cells. In:
RA Harris, NW Cornell (eds). Isolation characterization and use of
hepatocytes. Elsevier, Amsterdam, 1983, pp 313-322
44. Korzeniewski B, Harper M-E, Brand M: Proportional activation coefficients
during stimulation of oxidative phosphorylation by lactate and pyruvate
or by vasopressin. Biochim Biophys Acta 1229: 315-322, 1995
45. Akerboom TM, Bookelman H, Tager JM: Control of ATP transport
across the mitochondrial membrane in isolated rat-liver cells. FEBS
Lett 74: 50--54, 1977
46. Lemasters JJ, Sowers AK: Phosphate dependence of and actracty10side
inhibition of mitochondrial oxidative phosphorylation. J Bioi Chern
254: 1248-1251, 1979
358
47. Erecinska M, Stubbs M, Miyata Y, Ditre CM, Wilson DF: Regulation of
cellular metabolism by intracellular phosphate. Biochim Biophys Acta
462: 20-35, 1977
48. Hassinen JE: Mitochondrial respiratory control in the myocardium.
Biochim Biophys Acta 853: 135--151, 1986
49. Atkinson DE: The energy charge of the adenylate pool as a regulatory
parameter. Interaction with feedback modifiers. Biochemistry 7: 4030-
4034,1968
50. Korzeniewski B: What regulates respiration in mitochondria? Biochem
Molec Bioi Int39: 415--419,1996
51. Denton RM, McCormack JG: On the role of the calcium cycle in heart
and other mammalian mitochondria. FEBS Lett 119: 1-8, 1980
52. Hansford RG: Control of mitochondrial substrate oxidation. CurrTop
Bioenerg 10: 217-278, 1980
53. McCormack JG, Halestrap AP, Denton RM: Role of calcium ions in
regulation of mammalian intramitochondrial metabolism. Physiol Rev
70: 391-425,1990
54. McCormack JG, Denton RM: The role of mitochondrial Ca'+ transport
and matrix Ca'+ in signal transduction in mammalian tissues.
Biochim Biophys Acta 1018: 287-291, 1990
55. Harris DA, Das AM: Control of mitochondrial ATP synthesis in the
heart. Biochem J 280: 561-573, 1991
56. Moreno-Sanchez R, Hogue BA, Hansford RG: Influence ofNAD-linked
dehydrogenase activity on flux through oxidative phosphorylation.
Biochem J 268: 421-428, 1990
57. Balaban RS, Kantor HL, Katz LA, Briggs RW: Relation between work
and phosphate metabolite in the in vivo paced marrunalian heart. Science
232: 1121-1123, 1986
58. Katz LA, Swain JA, Portman MA, Balaban RS: Relation between
phosphate metabolites and oxygen consumption of heart in vivo. Am
J Physiol256: H265--H274, 1989
59. Balaban RS, Heineman FW: Control of mitochondrial respiration in the
heart in vivo. Mol Cell Biochem 89: 191-197, 1989
60. Heineman FW, Balaban RS: Control of mitochondrial respiration in the
heart in vivo. Annu Rev Physiol52: 523-542, 1990
61. Kacser H, Acerenza L: A universal method for achieving increases in
metabolite production. EurJ Biochem 216: 361-367, 1993
6la.Korzeniewski B: Regulation of ATP supply during muscle
contraction. Biochem (in press)
62. Kramer R, Mayr U, Heberger C, Tsompanidou S: Activation of
the ADP! ATP carrier from mitochondria by cationic effectors.
Biochim Biophys Acta 855: 201-210, 1986
63. Korzeniewski B, Froncisz W: Thermodynamic response paradigm and
its application to oxidative phosphorylation. In: S Schuster, M Rigoulet,
R Ouabi, J-P Mazat (eds). Modern Trends in BioThermoKinetics.
Plenum Press, New York, 1993, pp 33-38
Molecular and Cellular Biochemistry 184: 359-369, 1998.
1998 Kluwer Academic Publishers.

Role of mitochondrial calcium transport in the

control of substrate oxidation

Richard G. Hansford and Dmitry Zorov

National Institute on Aging, NIH, Baltimore, MD, USA

Abstract
This paper reviews the model of the control of mitochondrial substrate oxidation by Ca2+ions. The mechanism is the activation
by Ca 2+ of four mitochondrial dehydrogenases, viz. glycerol 3-phosphate dehydrogenase, the pyruvate dehydrogenase
multienzyme complex (PDH), NAD-linked isocitrate dehydrogenase (NAD-IDH) and 2-oxoglutarate dehydrogenase (OGDH).
This results in the increase, or near-maintenance, of mitochondrial NADHINAD ratios in the activated state, depending upon
the tissue and the degree of 'downstream' activation by Ca2+, likely at the level of the FIFoATPase. Higher values of the redox
span of the respiratory chain allow for greatly increased fluxes through oxidative phosphorylation with a minimal drop in
protonmotive force and phosphorylation potential. As PDH, NAD-IDH and OGDH are all located within the inner mitochondrial
membrane, it is changes in matrix free Ca 2 + [Ca2 +]m which act as a signal to these activities. In this article, we review recent
work in which [Ca2 +]m is measured in cells and tissues, using different techniques, with special emphasis on the question of the
degree of damping of [Ca2+]m relative to changes in cytosol free Ca2+ in cells with rapid transients in cytosol Ca2+, e.g. cardiac
myocytes. Further, we put forward the point of view that the failure of mitochondrial energy transduction to keep pace with
cellular energy needs in some forms of heart failure may involve a failure of [Ca2 +]m to be raised adequately to allow the activation
of the dehydrogenases. We present new data to show that this is so in cardiac myocytes isolated from animals suffering from
chronic, streptozocin-induced diabetes. This raises the possibility of therapy based upon partial inhibition of mitochondrial
Ca2 + effluxpathways, thereby raising [Ca2 +]m at a given, time-average value of cytosol free Ca2 +. (Mol Cell Biochem 184: 359-
369, 1997)

Key words: mitochondrial dehydrogenases, mitochondrial matrix free calcium, cardiac myocytes, cardiomyopathy, oxidative
phosphorylation

Introduction may then profoundly modify the characteristics of the

It is a characteristic of activated tissues that the level of free
Ca2+in the cytosol ([Ca 2+]J is raised above that seen in non-
stimulated tissues. Recent studies with single cells have
indicated that [Ca2+t generally oscillates above the baseline
value, with the frequency of the oscillations, and sometimes
also the amplitude, reflecting the degree or intensity of
activation [1-3]. In a muscle or nerve cell, the frequency is
of course determined by the frequency of depolarization of
the plasma membrane due to neurotransmitter action [4]: in
'non-excitable' cells such as hepatocytes the pattern of
oscillations is a consequence of binding of Ca2+-mobilizing
hormones to plasma membrane receptors [2, 5]. In both types
of tissue, Ca2+-induced Ca2+-release from reticulum stores
oscillations in [Ca 2+]c [6, 7]. Such activated cells have an
increased rate of ATP usage and require increased flux
through oxidative phosphorylation to maintain a balance of
energy supply and demand. It has been our point of view for
some time (Fig. 1) that Ca2+ acts as a messenger not only to
activateATP utilizing activities such as actomyosinATP-ase
and (indirectly) Na+ -K+ -ATP-ase but also to potentiate
oxidative phosphorylation [8, 9]; (for recent reviews see [10-
12]). The mechanism of the latter is the activation of four
mitochondrial dehydrogenases, viz: glycerol 3-phosphate
dehydrogenase [13, 14], the pyruvate dehydrogenase com-
plex [10, 15], NAD-isocitrate dehydrogenase [14, 16] and 2-
oxoglutarate dehydrogenase [14, 17]. This of course enhances
the provision of reducing equivalents to the mitochondrial

Address for offprints: R.G. Hansford, Gerontology Research Centre, National Institute on Aging, Hopkins Bayview, Baltimore, MD 21224, USA
360
electron transport chain. Somewhat less well established is
the Ca 2+-linked activation of the FjFOATP-ase (or ATP-
synthase: [18, 19]) and the adenine nucleotide translocase
(20).Activation ofthedehydrogenases will tend to maximize
values of the mitochondrial proton gradient (L1p) whereas
activation of the phosphorylation reaction will diminish L1p.
Both can be envisaged to lead to an increase in flux through
the overall process of oxidative phosphorylation.
The model of control of oxidative phosphorylation at the
dehydrogenase level was never intended to be an alternative
or competitor to the classic view of Chance and Williams [21]
of control by ADP as substrate for phosphorylation. Rather
it was intended as another level or stratum of control, which
operates in vivo but which had been ignored in the pre-
ponderance of in vitro mitochondrial experiments, which
use saturating (and non-physiological) concentrations of
oxidizable substrates (see [22] for a lengthy review of
dehydrogenase-level control). The metabolic control analysis
of Kacser and Bums [23] and Heinrich and Rappoport [24]
allows a more rigorous treatment of the relation between
control at the level of dehydrogenases, the respiratory chain
and the ATP-synthase/adenine nucleotide translocase. For a
thoughtful review the reader is referred to Brown [25].
Both ADP and Ca2+ ions act as allosteric activators of
dehydrogenases and facilitate the rest-to-work transition,
with quantitative importance ofthe two signals varying from
tissue to tissue. The model of control by Ca2+ arose initially
from the finding that dehydrogenases are activated by
physiologically-relevant concentrations of Ca2+, beginning
with the glycero13-phosphate dehydrogenase [13], which is
particularly active in insect flight muscle and squid mantle
muscle, but which is also a component of the redox shuttle
whereby some mammalian tissues oxidize glycolytically-
derived NADH. Examples are the B-cells of the pancreas
[26--29] and certain tumor cells [11]. The glycerol 3-phos-
phate dehydrogenase is an integral membrane protein, with
a Ca 2+ binding site which faces outwards on the inner
mitochondrial membrane [14, 30]. It thus senses [Ca2+L
directly-or, at least, [Ca2+] ofthe intermembrane space which
is probably a good surrogate for [Ca2+t By contrast, the other
Ca 2+-sensitive dehydrogenases are all located in the mito-
chondrial matrix and sense matrix free Ca2+ ([Ca2+]m - [11,
31-34 D. Their response to changes in [Ca2+L therefore
depends upon the fidelity with which changes in [Ca 2+L are
transduced into changes in [Ca2+]m by the proteins which
catalyze the transport of Ca2+ ions across the inner mito-
chondrial membrane, and this will be a major theme of this
article. These matrix enzymes include the pyruvate dehydro-
genase complex which provides acetyl groups from carbo-
hydrate for oxidation by the tricarboxylate cycle. Here, Ca2+
activates a phosphatase which dephosphorylates a trio of
serine residues on the a subunit of enzyme 1 of the complex
(the pyruvate decarboxylase), thereby restoring catalytic
activity [14, 15,35,36]. The other Ca 2+-sensitive matrix
dehydrogenases are the NAD-isocitrate dehydrogenase [14,
16] and the 2-oxoglutarate dehydrogenase [14, 17,37], both
of which have been implicated in exerting appreciable
control of flux through the tricarboxylate cycle [22]: i.e.
having measurable flux control coefficients [25]. Thus both
acetyl group production and oxidation are activated by Ca2+.
Indirect evidence in favor of the Ca2+ control hypothesis
comes from the finding that in adult heart there are only
minimal changes in calculated free ADP concentration when
flux through oxidative phosphorylation is increased several
fold [38--42]. Such changes inADP are not sufficient to drive
the extra flux on a Chance and Williams [21] mechanism,
making it likely by default that the Ca2+ control mechanism
is important. It is noted, however, that the nmr findings on
tissue ADP concentration were not the genesis of the Ca 2+
control model, as stated by Gunter et al. [43] in a recent piece
of revisionist writing.
The complete dependence of the Ca 2+ effects on the
tricarboxylate cycle enzymes upon mitochondrial Ca2+
transport pathways was first emphasized by Denton and
McCormack [44], who hypothesized that the Ca 2+ transport
systems had evolved to allow the control of these matrix
enzymes. Subsequent work delineated the relationships
between enzyme activity and the concentration of buffer Ca 2+
to which intact mitochondria were exposed and later, the
values of [Ca2+]m which were generated in these experiments
and measured using the fluorescent chelating agents fura 2
and indo 1. This work is summarized in previous review
articles [9-12, 43]. In one respect these experiments failed
to replicate the milieu experienced by mitochondria in vivo,
in that they did not expose mitochondria to oscillating Ca2+
levels, as are characteristic of the cytosol. Thus, the question
remained of whether [Ca2+]m follows changes in [Ca2+L with
fidelity in tissues, such as muscle, where [Ca 2+L transients
may be brief and, more generally, what the relation is between
frequency of [Ca2+l oscillations and the control ofmetabolism.
There have been exciting recent developments on this front,
and this issue will be addressed in more detail in this article.
We will then proceed to examine the altered mitochondrial
Ca 2+ homeostasis which occurs in some forms of cardio-
myopathy and which impinges upon energy metabolism. This
section contains new, unpublished data on these interactions
in cardiac myocytes from a model of chronic diabetes.
Before discussing this recent work it is necessary to
introduce mitochondrial Ca2+transport briefly. This topic has
been cogently reviewed by Gunter and Pfeiffer [45] and
Gunter et al. [43] for those seeking more detail. Calcium
enters mitochondria electrophoretic ally, driven by the large,
internally negative, membrane potential in a process which
is not allowed to come to electrochemical equilibrium. The
process has historically been referred to as a 'uniport', in
Peter Mitchell's terminology [46,47] and the carrier protein

as a 'uniporter'. The uniport process may be so active as to
subvert oxidative phosphorylation, as has been realized for
many years [48]. However, the affinity of the uniporter for
Ca2+is very low, e.g. Km = 10 11M for heart mitochondria [49],
when compared to measured concentrations of [Ca2 +]c in
suspensions of cells. Thus, the uniporter has generally been
considered to operate at a very small fraction of its Vmax in
living tissue, although this view can be challenged (please
see below). Remarkably the uniporter protein remains quite
ill-characterized and it has not been cloned. Calcium leaves
the mitochondria in exchange for H+ (predominant mechanism
in liver) or for Na+ (predominant mechanism in nerve and
muscle) in a process which has generally [50] been considered
to be electroneutral. As Na+/H+ exchange is active in
mitochondria, the net effect in either type of mitochondria is
that Ca2+efflux is driven by ~pH. Thus Ca2+uptake and efflux
are driven separately by the two components of ~p, viz, ~"',
and~pH -with the effect that the continuous cycling ofCa2+
across the membrane consumes ~p, albeit at a low rate. In
recent work, the electroneutral nature ofthe N a+ICa2+ exchange
has been challenged [51] and evidence has been advanced
that the process can maintain gradients of Ca2+ which are
greater than would be predicted from the magnitude of the
~pH, at least under certain experimental conditions [52].
These authors therefore consider that there is an additional
source of energy input, either because the process is an
electrophonetic 3Na+ICa2 exchange, or because of direct
coupling with electron transport. The exchange of Ca2+for n
H+ or n Na+ is referred to as an 'antiport' and the antiporter
protein has been purified and reconstituted [53]. The n Na+1
Ca 2+ exchanger is inhibited by a number of Ca2+-channel
blockers [54-56] and this is significant in view of the possible
advantages of inhibiting this carrier as a therapeutic
intervention in certain myopathies (see below).
Fidelity of response of mitochondrial Ca 2+ transport to
oscillations in [Ca 2+J c
A central question is whether [Ca2+]m follows the oscillations
in [Ca2+t which occur in excitable tissues or whether there
is instead a more gradual response to changes in frequency
ofCa 2+'transients'. More precisely, how much are transients
in [Ca2+] damped in relation to transients in [Ca2+] ? This is
still cont~versial, though more information is avail~ble than
at the time of writing a previous review [12]. In essence, the
answers depend upon the technique used.
Loading of isolated cardiac myocytes with the permeant
acetoxymethyl ester of the fluorescent chelating agent indo
1 results in the loading of indo 1 into the mitochondrion as
well as into the cytosol. Cytosolic fluorescence can then be
quenched by judicious exposure ofthe cells to Mn2+, leaving
a fluorescent signal emanating from the mitochondria and
361
allowing the monitoring of [Ca2+]m [57]. Use of this technique
with rat cardiac myocytes reveals no beat-to-beat variations in
[Ca2+]m. However, increase in frequency of stimulation of the
cells from 0.2-3 HZ, in the presence of isoproterenol, results
in an increase in [Ca2+]m from approximately 100 to 600 nM,
with a T 1/2 of around 30 sec [57]. These changes would
allow meaningful control of dehydrogenase activity, given
the finding of half maximal stimulation of the pyruvate
dehydrogenase interconversion by matrix Ca 2+ of 650 nM
[11]. Although care was taken in these studies to 'titrate' the
cytosolic indo-l with Mn2+, by watching the gradual dis-
appearance of fluoresence transients linked to excitation-
contraction coupling, it has to be admitted that miniscule
concentrations of free Mn2+ could bind to the Ca2+uniporter
or antiporter proteins and affect kinetics of mitochondrial
Ca 2+transport. Thus, this negative result concerning beat-to-
beat changes in [Ca2+]m has to be treated with caution.
By contrast, laser-scanning confocal microscopy of rabbit
cardiac myocytes loaded with the fluorescent chelating
agent fluo 3 shows clear beat-to-beat changes in [Ca 2+]m
[58]. The result is credible in the sense that the mito-
chondrial origin of the signal was confirmed by colocalization
with the signal from TMRM, a membrane potential-sensitive
probe. In cells not exposed to a positive inotropic activation,
the fluorescent transients from fluo 3 in cytosol and mito-
chondria were essentially identical, whereas in cells stimulated
by the ~-adrenergic agonist isoproterenol, the fluorescent
transients from the mitochondrial compartment were of
slightly smaller amplitude. Remarkably, the kinetics of rise
and decay appeared the same for [Ca 2+] in the two com-
partments, within the temporal resolution of the apparatus.
An odd feature of these results [58] is that fluorescent
transients were extremely prolonged, appearing to last for
approximately one second at 37C. This may indicate that
the cells were loaded with fluo 3 to the point that the
buffering by the indicator markedly changed the kinetics of
change in [Ca 2+]. This may cloud the interpretation ofthese
data.
The only other set of experiments suggesting beat-to-beat
variations in cardiac myocyte [Ca2+]m used electron probe
microanalysis (EPMA) and guinea pig ventricular myocytes
[59, 60]. Use of a paired stimulus protocol, designed to
replicate the degree of cell activation seen in vivo, produced
large cytosol transients in Ca2+and transients in [Ca 2+]m -total
which lagged those of cytosol Ca 2+. Thus, these authors
showed a rise in [Ca2+]m-total which began after a 20 msec
lag and almost reversed by 90 msec, the point at which
contraction was maximal. The maximum value of [Ca2+]m-
total which was achieved, 40 msec after the stimulation, was
3.2 0.25 mmol Kg-l dry wt, up from an end-diastolic value
of 1.34 0.24. In unstimulated cells [Ca2+]m-total was found
to be 0.51 0.34 mmol Kg-l dry wt, which is near the
detection limit of the method.
362
By contrast, Moravec and Bond [61], also using EPMA,
failed to show any beat-to-beat variation in [Ca2+]m -total when
studying hamster papillary muscle stimulated at 0.2 Hz. The
sensitivity of this technique is a problem, as the Ca peak in
the emission spectrum is very largely obscured by a much
larger K peak. In fact, not only is beat-to-beat variation not
visualized, but no increase in [Ca2+]m -total due to a prolonged
(5 min) positive inotropic (lQ--6M isoproterenol) stimulation
of isolated papillary muscle [62] or perfused heart is seen,
whereas such an increase has been demonstrated by a number
of other techniques (see [12] for a review). Moravec and
Bond [62] also fail to seen an increase in A-band Ca in
response to inotropic stimulation. It seems strange to argue
that troponin C is activated by Caz+, although no change in
A-band Ca is detected, whereas mitochondrial dehydro-
genases cannot be activated by Ca, because no change in
mitochondrial Ca is detected. After all, the ratio of free
ionized Caz+, which is the signal, to total Ca is essentially the
same, at about 1 X 10-3, in both cell compartments.
Thus, the question of whether [Ca2+]m oscillates on a beat-
to-beat basis in cardiac myocytes is still contentious, with
different techniques giving different answers. In addition,
there may be very real interspecies variation, with oscillations
being more likely in larger hearts with lower heart rates. In
previous articles [11, 12] the argument has also been advanced
that little beat-to-beat variation of [Ca2+]m would be expected
on the grounds of modelling based upon the measured rates
of uniport and antiport in isolated mitochondria [63, 64].
Though the conclusion may yet still be correct, the calculations
need to be re-done with the realization that there is a more
rapid mode of Caz+uptake shown recently for both liver and
heart mitochondria [65]. This rapid mode of uptake, or RAM,
is inferred to exist by extrapolating measured rates of Caz+
uptake obtained with Ca2+ pulses of varied duration back to
very short times. Thus mitochondria have the capability to
catalyse a short-burst ofCaz+uptake which is more active than
that described previously in steady-state uptake experiments.
It is not clear at this point whether RAM is catalyzed by the
uniporter protein or involves some other pathway. For the
existence of RAM to allow transients in [Ca2+]m to occur in
heart muscle, it would presumably have to be accompanied
by some equally active, and undescribed, mode ofCa2+efflux.
Response of [Ca 2+] m to microdomains of [ Ca 2+] c
Perhaps the most novel recent insight into mitochondrial Ca2+
transport in cells has come from the work of Rizzuto et al.
[66--68] involving the targeting of recombinant aequorin to
the mitochondrial matrix space. This has shown unex-
pectedly high values of[Ca2+]m in a variety of cell types, when
challenged with hormones which release Ca z+ from the
endoplasmic reticulum via IP 3-sensitive channels. Whenever
this mechnisim is triggered, transient values of [Caz+]m greatly
exceed values of [Ca2+]c measured by fura 2 fluorescence and
often reach 10 !lM or more [67, 68]. By contrast, depolari-
zation of the plasma membrane, using elevated [K+] graded
to achieve comparable values of [Caz+L to those achieved
using IP 3-mediated agonists, results in much smaller increases
in [Caz+]m. The authors interpret these findings to indicate that
opening oflP3-sensitive channels in the endoplasmic reticulum
exposes a fraction of the mitochondria to locally high
concentrations ofCaz+, which give rise to a disproportionately
high rate of mitochondrial Ca2+ uptake, perhaps due to the
sigmoidal kinetics of the mitochondrial uniporter [49]. By
contrast, entry ofCaz+through the plasma membrane gives a
more homogeneous rise in [Caz+]c. Clearly fura 2 becomes a
rather insensitive Caz+indicator when values ofCa2+approach
1 11M in activated cells and in the same region aequorin
exaggerates responses, owing to its dependence on a higher
order of [Ca2+] [69]. Despite this, these results seem clear and
compelling. They are supported by studies of He La cells
expressing mitochondrially targeted aequorin and per-
meabilized by digitonin, which also show large increases in
[Caz+]m in response to additions oflP 3 [67]: evidently the
structural alignment of mitochondria and reticulum is
preserved in these permeabilized cells. There is evidence in
intact cells that the rate of release ofCa2+ from the reticulum
is crucial: a matched value of [Ca2+L achieved with thap-
sigargin, which inhibits SERCA ATP-ases, and measured
with fura 2, does not generate as high a value of [Caz+]m as
does an IP 3-linked agonist [68]. The rate of rise of [Ca2+L is
slower with thapsigargin than with IP 3-mediated agonists and,
presumably, the more rapid efflux of Ca2+ through the IP 3-
sensitive channel results in steeper gradients of [Caz+] in the
vicinity of the reticulum.
By contrast, cell types with a sub-population of mito-
chondria underlying the plasma membrane, eg cardiac
myocytes, may be expected to show a sensitive response of
[Ca2+]m to the opening of plasma membrane Ca2+ channels,
and indeed there is some evidence for this in very recent
experiments with a highly differentiated rat pancreatic ~-cell
line (INS 1) stably expressing mitochondrially targeted
aequorin [70]. In these cells, [Ca2+]m increases in response to
the IP3-linked agonist carbachol, but increases to a greater
extent when the plasma membrane is depolarized with high
K+ in the presence of glucose: more precisely, higher values
of [Ca2+]m are generated at a given value of [Ca2+L, measured
with fura 2 fluorescence, when the source of the activator Ca2+
is voltage-dependent plasma membrane Ca2+ channels.
Presumably, the sensitive mitochondria are a sub-population
underlying the plasma membrane.
The pioneering experiments on mitochondrially-targeted
aequorin involved studies oflight emitted from large numbers
of cells. This means loss of information on the individual
pattern ofCaz+oscillation in response to cell activation, which

tends to vary among cells [71-73]. Recent work involving a
more effective expression of mitochondrially targeted
recombinant aequorin, as well as more sensitive detection of
luminescence, has allowed the measurement of [Ca 2+]m in
single cells [74]. Stimulation ofCHO.T cells withATP, which
activates purinergic (P 2U ) receptors and mobilizes IP 3 , gave
rise to rapid, large transients in [Ca2+]m of 15-20 flM followed,
when extracellular Ca2+ was present, by an elevated plateau
of [Ca2+]m of approximately 1 flM. Parallel measurements of
the activated form of pyruvate dehydrogenase (PDH A) in bulk
suspensions of cells indicated that it was increased upon
addition ofATP and remained elevated during the plateau of
[Ca2+]m' consistent with prior estimates of the sensitivity of
conversion of pyruvate dehydrogenase to values of [Ca2+]m
[31,34]. The appeal of this technique, of course, is that it
allows imaging of mitochondrial populations within a single
cell, with respect to values of [Ca2+]m.In this study [74], there
were no large differences in response of mitochondria to the
mobilization ofCa2+withATP, as a function of position within
the cell, with the exception that there was a population of
under-responders in the vicinity ofthe nucleus. Clearly, this
is a powerful technology. Hopefully, application to the cell-
types discussed above could establish whether there is indeed
a population of highly-responding mitochondria in the
vicinity ofthe reticulum in those cells, as well as establishing
more firmly the temporal relationship between [Ca2+L and
[Ca2+]m in individual cells with oscillating [Ca2+t
A direct test of the model of partial control of oxidative
phosphorylation by Ca2+ ions [9, 10, 75] involves demon-
stration of the dependence of flux through oxidative phos-
phorylation upon [Ca2+]m. Flux is most easily measured by
02-uptake of organs, or suspensions of cells or mitochondria
and results at this level of tissue organization have been
presented previously [11, 12]. Recent advances in measure-
ment of [Ca2+]m have involved mainly work with single, or
relatively few cells, which do not lend themselves to direct
determinations of flux. However, NADH levels are readily
measurable by fluorescence even in single cells and study of
NADH levels provides some understanding of the degree of
dehydrogenase activation which is occurring. A caveat is that
if indeed the Flo ATP-ase is activated by Ca2+, as suggested
in work on cardiac myocytes [18], then of course there could
be an activation of flux through oxidative phosphorylation
driven by Ca2+, but revealing no change in NADHINAD+
ratio, or even a decrease. This issue was addressed in a
previous review [12]. The most recent information on the
heart suggests that increased work is associated with a
decreased mitochondrial NADHINAD+ ratio. This is true both
of the isolated, perfused rat heart, studied by an improved
ratiometric fluorescence technique [76] and of electrically-
stimulated suspensions of cardiomyocytes [77]. This finding
is consistent with a greater degree of stimulation by Ca2+ at
the level of the F loATP-ase, or possibly the adenine nucleotide
363
translocase, than at the level of the dehydrogenases.
Alternatively, there may be a local rise inADP concentration,
as a consequence of Ca2+-linked activation of actomycin
ATP-ase, sufficient to explain some ofthe increase in flux,
despite the general agreement from nmr studies of the heart
that any such increase must be small (see [78]). Direct
activation of the ATP-synthase is strongly suggested by the
recent, intriguing work ofWan et al. [79] which showed that
mitochondrial membrane potential (il\jf) falls with increased
work in the intact isolated heart, as ADP concentration also
falls or remains unchanged. The 4-fold increase in flux
through oxidative phosphorylation essentially establishes that
kinetic modification oftheATP-synthase is occurring under
these conditions.
The use of NADH fluorescence to index dehydrogenase
activity in single cells has yielded exciting insights into the
relation between [Ca2+]c oscillations and mitochondrial
dehydrogenase activation in the last 2 or 3 years. Single
hepatocytes respond to activation by the Ca 2+-mobilizing
hormones vasopressin and angiotensin II by release of Ca2+
from reticulum stores which is mediated by IP 3 and results
in a series ofCa2+waves which propagate throughout the cell
[2, 80, 81]. Higher agonist concentrations are associated with
higher frequencies of these oscillations until eventually, at
saturating hormone concentrations, the oscillations fuse to
give a sustained high value of [Ca2+]c. The new information
comes in the response of [Ca 2+]m and ofNADH to these waves
of [Ca 2+L. In hepatocytes, oscillations of [Ca 2+L were
accompanied by oscillations in NAD(P)H fluorescence,
which essentially index changes in mitochondrial NADH [72,
73]. The rate of rise ofNADH (and of reduced flavin-73) was
similar to that of [Ca2+L; however, the rate of fall of NADH
was substantially delayed. This alone could reflect either a
limitation in efflux of Ca2+ from the mitochondria or some
hysteresis in the activation ofthe dehydrogenases themselves.
However, the work by Hajnoczky et al. [73] was able to
distinguish these two possibilities, by using rhod 2 to monitor
[Ca 2+]m directly. This showed that [Ca 2+]m follows [Ca 2+L
faithfully on the rising phase, but lags substantially on the
falling phase. Thus [Ca2+] efflux from the mitochondria is
relatively slow. Of great interest was the finding that the most
effective mechanism to raise [Ca2+]m and NADH was stimu-
lation by a hormone resulting in a high frequency train of
[Ca 2 +L pulses (0.5-1 per min). By contrast, the sustained
elevation of [Ca 2+L generated by saturating agonist was
associated with a partial return of values of [Ca 2+]m and
NADH towards resting values [73]. Thus, there seems to be
some sort of desensitization process that occurs. An alter-
native explanation, and one which is consistent with the
'micro domain' rationale of Rizzuto et al. [67] is that fura 2
is under-reporting the [Ca2+L of micro domains in the vicinity
of the endoplasmic reticulum and that higher transients of
[Ca2+]c are actually achieved in this domain when pulsatile
364
release ofCa2+ from the ER is occurring. Possibly continuous
exposure to IP 3 at the saturating concentrations of agonist
does not allow the re-filling of the reticulum stores. The
paper by Hajnoczky et al. [73], by virtue of measuring
[Ca2+]m' allows the conclusion that the slow falling phase in
transients of NADH fluorescence in single hepatocytes
reflects Ca2+ transport rather than the kinetics of inactivation
of dehydrogenases upon removal ofCa2+. This isnot surprising
as the activation of NAD-isocitrate dehydrogenase and 2-
oxoglutarate dehydrogenase is allosteric in nature and the
interconversion of pyruvate dehydrogenase, though potent-
ially more limiting, has been shown to respond with at T 112
of around 15 sec to changes in effectors of both kinase and
phosphatase, in prior experiments with isolated mitochondria
[82].
The oscillatory response of NADH to the hormonal
stimulation of single cells has been shown for adrenal
glomerulosa and pancreatic ~-cells, as well as for hepatocytes
[72, 73]. Evidently in these three cell types, activation of
dehydrogenases by Ca 2+ outweighs any putative activation of
the F loATP-ase, in distinction to what is seen in the heart
[76, 77]. In the case of the pancreatic ~-cells, a portion of
the increase in NADH may reflect activation ofthe glycerol
3-phosphate dehydrogenase by Ca 2+ and a more active
transfer of glycolytic reducing equivalents into the mito-
chondria [13, 26, 29,83].
The realization that waves of [Ca2+]c traverse activated cells
and the demonstration that [Ca2+]m cycles in synchrony [73]
in cells like hepatocytes which show slow oscillations (i.e.
slow relative to myocytes or neurons) raises the question of
whether mitochondria vary in energy status in different
regions of the cell. One recent study [84] did in fact show a
distribution in values of 1", among mitochondria in neuro-
blastoma cells, with some fraction being particularly
depolarized in response to bradykinin. Although bradykinin
releases Ca 2+ from the endoplasmic reticulum in an IP 3-
dependent manner, no particular inhomogeneities of
[Ca 2+]c were found with fura 2 [84]. Spatial inhomogeneities
in mitochondrial functioning appears to be a particularly
promising field, with the advent of the new microscope
technologies.
Altered regulation of mitochondrial calcium in
cardiomyopathy
Cardiomyopathies have long been associated with cellular
Ca2+ loading [85, 86], and this has often been assumed to be
the mechanism of deterioration of cell function. However, it
is clear that the disturbance of cellular energetics for whatever
reason will lead to Ca 2+ loading, as the chemical gradient
alone for Ca2+ ions, across the plasma membrane isapproxima-
tely 10 4 and thus it is hard to establish cause and effect. In
the cardiomyopathic Syrian hamster, a well-known genetic
model of cardiomyopathy, bulk assay of cardiac muscle from
affected hamsters shows an increased total Ca 2+, as do
mitochondrial preparations from these hearts [87,88].
However, cell-by-cell examination using electron probe
microanalysis reveals that the large majority ofmyocytes do
not have elevated cytosol total Ca2+ or [Ca2+]m_total, but that the
extra Ca2+ of bulk preparations derives from necrotic areas
and from a few cells on the periphery of these areas, which
are presumably mechanically stressed [89].
A publication from this lab [90] made the case that, to the
contrary, [Ca 2+]m is actually lower in myocytes from the
myopathic animal and that this compromises mitochondrial
substrate oxidation. This was demonstrated by examining
single isolated cardiac myocytes from cardiomyopathic
(strain B10 14.6) and control (FIB) animals, and investigating
the response of [Ca2+]m to raised rates of electrical pacing. It
was found that [Ca2+]m only rose from 248 l5nM at rest to
348 44nM with 4 Hz stimulation in the BIO 14.6 cells,
compared with a rise from 241 35 to 830 124 nM in cells
from healthy FIB animals. Developed pressure from isolated,
perfused hearts of the B 10 14.6 animals was lower, as had
been shown previously [91, 92] and, most interestingly, the
fraction of pyruvate dehydrogenase present as PDHA was
lower in glucose-perfused hearts from the myopathic animals
[90]. The inability to increase PDHA inresponse to work load
ofthe heart is consistent with the inability to increase [Ca2+]m
in response to electrical stimulation of myocytes from the B 10
14.6 animals. The basic lesion does not seem to be at the level
of the mitochondrial membrane, but rather in the excitation-
contraction coupling process. Thus, electrical stimulation of
cardiac myocytes from B 10 14.6 animals resulted in smaller
transients in [Ca2+t as sensed with the salt form of indo-1,
than those seen in control cells.
Two recent papers on the cardiomyopathic hamster have
added extra credence to this point of view. Keller et al. [93]
used the EPMA technique to study elemental composition of
different cell regions in papillary muscles and determined that
the Ca content of the junctional SR was lower in specimens
from cardiomyopathic hamsters, when compared to healthy
controls. The amount ofCa released on electrical excitation
was also smaller. KrUger et at. [94] studied [Ca 2+t directly
using fura 2 in work with isolated cardiac myocytes and
confirmed the finding ofDiLisa et al. [90], to the effect that
systolic transients in [Ca 2+t were smaller in cells from the
cardiomyopathic hamsters (B1O 14.6 strain). Although care
should be taken in comparing these studies, as Keller et al.
[93] used preparations from hamsters at the hypertrophic
stage of the disease, whereas DiLisa et at. [90] and KrUger
et at. [94] used myocytes from the hearts of animals in overt
heart failure, a consensus seems to be emerging of a more
limited loading of the SR in the myopathic animals, possibly
a consequence of decreased SR Ca 2+ ATP-ase activity or an
 365

indirect consequence of a lower density of SL Ca2+ channels EXCITATION STIMULUS

[94], giving rise to less Ca 2+ release and smaller systolic
transients. The significance of this, of course, for the purpose
of this review is that of decreased mitochondrial Ca2+ uptake,
~/
regardless of whether this follows the simple arithmetic mean
ofthe external [Ca2+] as shown by Leisey et al. [95] in in vitro //-------/~'~
experiments with mitochondria, or is more affected by the
height of the systolic transients, as would be expected from CONTRACTION SECRETION
the sigmoidal nature of the v versus s plot for the cardiac
mitochondrial uniporter [64]. ACTOMYOSI~ ~a+.K+.ATP.ASE
ATP.ASE " " /' '"
In chronic diabetes, there is evidence for a cardiomyopathy
,,

>-<
which is quite separate from the vascular changes which ATP + H20 ADP + Pi
,,
occur [96, 97]. Accompanying this is a profound inhibition ,
of flux through pyruvate dehydrogenase, as measured with
l3C-labelled metabolites and nrnr [98]. Although it has been
OXIDATIVE
known for some time that pyruvate dehydrogenase kinase PHOSPHORYLATION
activity is potentiated in diabetic heart [99, 100] and that
perfusion with dichloroacetate improves contractile function
[101] it seemed possible by analogy with our work on the
ACTIVATION
cardiomyopathic hamster [90] that the phosphatase activity
might also be decreased, as a consequence of low values of
Fig. I. Calcium signalling to both energy demand and supply.
[Ca2+]m' These inverse changes in kinase and phosphatase
activities would have a final common result; the reduction
of the content of active pyruvate dehydrogenase.
We have recently investigated Ca2+ homeostasis in single
isolated cardiac myocytes from rats made diabetic by a single
injection of streptozotocin (65 mg/kg body wt.) and studied
6-8 weeks later. Measurement of [Ca 2+]m was accomplished
by loading myocytes with indo 11AM and quenching the
fluorescent signal originating from the cytosol by exposing the
cells to low concentrations ofMn Cl 2 - in a technique developed [,
by Miyata et al. [57]. Figure 2 shows the results of such a study, ':""
E

when cells were subjected to a profound positive inotropic 'fa

stimulation by use of 1 x lO--{)M isoproterenol. It is seen that ~
.:<a
the elevation of[Ca2+]m which occurs in response to electrical o
....I
stimulation (3 Hz) was significantly less in the cells from hearts -<
of diabetic animals. This could have been a consequence of a ii:
c
less effective ~-adrenergic activation in the diabetic heart cells, z
o
as both a decreased density of ~-receptors [102, 103] and an J:
o
impaired stimulation by cell-permeant c-AMP analogues [104] o
!:::::
have been described in the diabetic heart. However, repetition ~
-<
of these experiments using 3.5 mM superfusate Ca2+, in place
of ~-adrenergic agonists, to achieve a positive inotropic state, ~
yielded a very similar result (Fig.3), with a lower value of
[Ca 2+]m achieved upon electrical stimulation of cells from
diabetic animals. These values of [Ca 2+]m are within the
region where modulation of dehydrogenase activation
occurs [11,31-34, 74] and thus the finding is likely to provide
at least part of the mechanism of the decreased flux through RESTING STIMULATED
pyruvate dehydrogenase in the diabetic heart.
Fig. 2. Intramitochondrial free Ca'+ -[Ca'+lm - as a function of electrical
The lowered values of [Ca 2+]m are likely to follow from
stimulation of cardiac myocytes from diabetic and control rats. Positive
smaller systolic transients in [Ca2+lc in myocytes from diabetic inotropic activation was achieved with the ~-adrenergic agonist isoproterenol
hearts, as there are now several studies showing this result (lO-iiM).
366
...m
E
E
..~
<Q
o a finding needs to be made more secure by studies of uptake
..J
'"
~
C
Z
o resolution.
:z:
u
g Provocatively, there have been findings of diminished
'"
~ cardiac myocytes from failing human hearts [94, 113], raising
RESTING STIMULATED
Fig. 3. Intramitochondrial free Ca2+-[Ca 2+l m - as a function of electrical
stimulation of cardiac myocytes from diabetic and control rats. Positive
inotropic activation was achieved with 3.5mM Ca 2+ in the superfusate.
[104-106]. Evidently, SR loading of Ca 2+ is reduced in
diabetes [106, 107], consistent with a lower activity of the
SR Ca 2+ATP-ase [108, 109]. In addition, it is possible that
the entry of Ca 2+ through voltage-dependent sarcolemmal
channels, to provide the trigger for SR release, may be
diminished, though there is evidence suggesting that the
inward Ca 2+current is unchanged [110]. There is no literature
finding on [Ca 2+]c in diabetes which provides a ready
explanation of our finding of raised [Ca2+]m in unstimulated,
non-contracting cells (Figs 2 and 3).
Although the diminished systolic transients in [Ca2+]c in the
myocytes from diabetic hearts provide a plausible mechanism
for our finding of relative failure to elevate [Ca2+]m' it is of
course also possible that there is also an intrinsic deficit at
the level of the mitochondrial membrane. This issue has been
addressed in the past, with findings of diminished activity of
Ca 2+ uptake by isolated mitochondria [Ill] and due to
mitochondrial activity in permeabilized myocytes [112], in
material from diabetic rats. It is hard to evaluate the work with
the isolated mitochondria as the diabetes-linked deficit was
seen only under massive loading condition, with no change
seen under the more limited loading conditions which are
more relevant to the physiology of the heart. Conceivably the
lesser accumulation of Ca2+under massive loading conditions
in the presence of phosphate reflects an increased frequency
of opening of the non-specific 'mitochondrial permeability
transition' (see [45]) in mitochondria from diabetic animals.
The work with saponin-permeabilized myocytes (Tanaka et
al. 1992) also used excessive concentrations of Ca2+, but does
show some decrease in uptake velocity at < I ~M Ca2+, in
preparations from diabetic animals. There is also evidence
for a slightly lower A\jf, as well as decreased rates of substrate
oxidation, as also noted by Pierce and Dhalla [111]. Thus, it
is possible that there is a lower intrinsic activity of Ca 2+
uniporter in the mitochondria from the diabetic heart, but such
at lower concentrations of Ca2+, such that proton pumping by
the respiratory chain is not limiting, and with higher time
systolic transients in [Ca 2 +lc, similar to those we have just
discussed for the Syrian hamster and the diabetic rat, in
the possibility that myocyte mitochondria in human heart
failure may contain insufficient Ca 2+ to activate matrix
dehydrogenases. In such a case, there could be therapeutic
benefit in partial inhibition of the mitochondrial Na+/Ca 2+
exchanger, or antiporter, to increase the gradient [Ca 2+]m/
[Ca2+lc [12]. In terms of the metaphor of the failing heart being
a struggling horse pulling a delivery cart up a hill [114], such
an approach would be equivalent to feeding the horse with
some readily digestible feed.
References
1. Rapp PE, Berridge MJ: The control of transepithelial potential
oscillations in the salivary gland of Calliphora erythrocephala. J Exp
BioI 93: 119-132, 1981
2. Woods NM, Cuthbertson KSR, Cobbold PH: Repetitive transient rises
in cytoplasmic free calcium in hormone-stimulated hepatocytes. Nature
319: 60{}-{)02, 1986
3. Berridge MJ, GalioneA: Cytosolic calcium oscillations. Faseb J 2: 3074--
3082, 1988
4. Katz B, Miledi R: The release of acetylcholine from nerve endings by
graded electrical pulses. Proc Ray Soc Lond (Bioi) 167: 23-38,1967
5. RooneyTA, Sass E, ThomasAP: Characterization ofcytosolic calcium
oscillations induced by phenylephrine and vasopressin in single fura-
2-loaded hepatocytes. J BioI Chern 264: 17\31-17141, 1989
6. Fabiato A, Fabiato F: Contractions induced by a calcium triggered
release of calcium from the sarcoplasmic reticulum of single skinned
cardiac cells. J Physiol (London) 249: 469--495, 1975
7. Busa W8, Ferguson JE, Joseph SK, Williamson JR, Nuccitelli R:
Activation of frog (Xenopus /aevis) eggs by inositol triphosphate. 1.
Characterization of Ca+ release from intracellular stores. J Cell BioI
\0 I: 677--682, 1985
8. Hansford RG, Cohen L: Relative importance of pyruvate dehydrogenase
interconversion and feed-back inhibition in the effect of fatty acids on
pyruvate oxidation by rat heart mitochondria. Arch Biochem Biophys
191: 65-81,1978
9. Hansford RG: Relation between mitochondrial calcium transport and
control of energy metabolism. Rev Physiol Biochem Pharmacol 102:
1-72,1985

10. McCormack JG, Halestrap AP, Denton RM: Role of calcium ions in
regulation of mammalian intramitochondrial metabolism. Physiol Rev
70:391-425,1990
II. Hansford RG: Dehydrogenase activation by Ca2+ in cells and tissues.
J Bioenerg Biomembr 23: 823-854, 1991
12. Hansford RG: Physiological role of mitochondrial Ca 2 + transport. J
Bioenerg Biomembr 26: 495--508, 1994
13. Hansford RG, Chappell JB: The effect of Ca2+ on the oxidation of
glycerol phosphate by blowfly flight-muscle mitochondria. Biochem
Biophys Res Commun 27: 686--692, 1967.
14. Nichols BJ, Denton RM: Towards the molecular basis for the regulation
of mitochondrial dehydrogenases by calcium ions. Mol Cell Biochem
149/150: 203-212, 1995
15. Denton RM, Randle PJ, Martin BR: Stimulation by Ca 2+ of pyruvate
dehydrogenase phosphate phosphatase. Biochem J 128: 161-163, 1972
16. Denton RM, Richards DA, Chin JG: Calcium ions and the regulation
of NAD+ linked isocitrate dehydrogenase from the mitochondria of
rat-heart and other tissues. Biochem J 176: 899-906, 1978
17. McCormack JG, Denton RM: The effects of calcium ions and adenine
nucleotides on the activity of pig heart 2-oxoglutarate dehydrogenase
complex. Biochem J 180: 533-544, 1979
18. Harris DA, Das AM: Control of mitochondrial ATP synthesis in the
heart. Biochem J 280: 561-573,1991
19. Scholz TD, Balaban RS: Mitochondrial F I-ATP-ase activity of canine
myocardium: Effects of hypoxia and stimulation. Am J Physiol266:
H 2396-H 2403, 1994
20. Moreno-Sanchez R: Contribution of the translocator of adenine
nucleotides and the ATP synthase to the control of oxidative phos-
phorylation and arsenylation in liver mitochondria. J Bioi Chern 260:
12554-12560,1985
21. Chance B, Williams GR: The respiratory chain and oxidative
phosphorylation. Adv Enzymol17: 65-134, 1956
22. Hansford RG: Control ofmitochondrial substrate oxidation. CUff Top
Bioenerg 10: 217-278, 1980
23. Kacser H, Bums JA: The control of flux. In: DD Davies (ed). Rate
Control of Biological Processes, Symposia of the Society for
Experimental Biology, No. XXVII. Cambridge University Press,
Cambridge, UK, 1973, pp 65-104
24. Heinrich R, RapoportTA: A linear steady-state treatment of enzymatic
chains: General properties, control and effector strength. Eur J Biochem
42: 89-95, 1974
25. Brown GC: Control of respiration and ATP synthesis in mammalian
mitochondria and cells. Biochem J 284: 1-13, 1992
26. MacDonald MJ: High content of mitochondrial glycerol-3-phosphate
dehydrogenase in pancreatic islets and its inhibition by diazoxide. J
Bioi Chern 256: 8287--8290, 1981
27. MacDonald MJ: Calcium activation of pancreatic islet mitochondrial
glycerol phosphate dehydrogenase. Horm Metab Res 14: 678-679,
1982
28. Rutter GA, Pralong W-F, Wollheim CB: Regulation of mitochondrial
glycerol-phosphate dehydrogenase by Ca2+within electroperrneabilized
insulin secreting cells INS-I. Biochim Biophys Acta 1\ 75: 107-113,
1992
29. Civelek VN, Deeney JT, Shalosky NJ, Tomheim K, Hansford RG,
Prentki M, Corkey BE: Regulation of pancreatic ~-cell mitochondrial
metabolism: Influence ofCa2+, substrate and ADP. Biochem J 318:
615-621,1996
30. Klingenberg M, Buchholz M: Localization of the glycerol-phosphate
dehydrogenase in the outer phase of the mitochondrial inner membrane.
Eur J Biochem 13: 247-252, 1970
31. Moreno-Sanchez R, Hansford RG: Dependence of cardiac mito-
chondrial pyruvate dehydrogenase activity on intramitochondrial free
Ca2+ concentration. Biochem J 256: 403-412,1988
367
32. Lukacs GL, KapusA, FonyoA: Parallel measurements of oxoglutarate
dehydrogenase activity and matrix free Ca2+ in fura-2 loaded heart
mitochondria. FEBS Lett 229: 219-223, 1988
33. McCormack JG, Browne HM, Dawes NJ: Studies of mitochondrial
. Ca2+-transport and matrix Ca2+ using fura-2-loaded rat heart mito-
chondria. Biochim Biophys Acta 973: 420-427, 1989
34. Wan B, LaNoue KF, Cheung JY, Scaduto RC Jr.: Regulation of citric
acid cycle by calcium. J Bioi Chern 264: 13430-13439, 1989
35. Reed LJ: Regulation of mammalian pyruvate debydrogenase complex
by a phosphorylation dephosphorylation cycle. CUff Top Cell Regul
18: 95--106,1981
36. Teague WM, Pettit FH, Wu T-L, Silberman SL, Reed LJ: Purification
and properties of pyruvate dehydrogenase phosphatase from bovine
heart and kidney. Biochemistry 21: 5585--5592, 1982
37. Panov A, Scarpa A: Independent modulation of the activity of cx-
ketoglutarate dehydrogenase complex by Ca2+ and Mg2+. Biochem 35:
427-432, 1996
38. Balaban RS, Kantor HL, Katz LA, Briggs RW: Relation between work
and phosphate metabolise in the in vivo paced mammalian heart.
Science 232: 1\21-\123, 1986
39. From AHL, Petein MA, Michurski SP, Zimmer SD, Ugurbil K: IIp_
NMR studies of respiratory regulation on the intact myocardium. FEBS
Lett 206: 257-261, 1986
40. Katz LA, Koretsky AP, Balaban RS: Respiratory control in the glucose
perfused heart. A llPNMR and NADH fluorescence study. FEBS Lett
221: 270-276,1987
41. Katz LA, Koretsky AP, Balaban RS: Activation of dehydrogenase
activity and cardiac respiration. A lip NMR study. Am J Physiol 255:
HI85--HI88,1988
42. Katz LA, Swain JA, Portman MA, Balaban RS: Relation between
phosphate metabolises and oxygen consumption of heart in vivo. Am
J Physiol256: H265--H 274, 1989
43. Gunter TE, Gunter KK, Sheu S-S, Gavin CE: Mitochondrial calcium
transport: Physiological and pathological relevance. Am J Physiol267:
C313--C339,1994
44. Denton RM, McCormack JG: On the role of the calcium transport
cycle in heart and other mammalian mitochondria. FEBS Lett 119: 1-
8,1980
45. GunterTE, Pfeiffer DR: Mechanisms by which mitochondria transport
calcium. Am J Physiol258: C755--C786, 1990
46. Mitchell P: Chemiosmotic coupling in oxidative and photosynthetic
phosphorylation. Bioi Rev Camb Philos Soc 41: 445-502, 1966
47. Mitchell P: Compartmentation and communication in living systems.
Ligand conduction: A general catalytic principle in chemical, osmotic
and chemiosmotic reaction systems. Eur J Biochem 95: 1-20, 1979
48. Lehninger AL, Carafoli E, Rossi CS: Energy-linked ion movements
in mitochondrial systems. Adv. Enzymol 29: 259-320, 1967
49. Scarpa A, Graziotti P: Mechanisms for intracellular calcium regulation
in heart. I. Stopped flow measurements of Ca2+ uptake by cardiac
mitochondria. J Gen Physiol62: 756-772, 1973
50. Brand MD: The stoichiometry of the exchange catalyzed by the
mitochondrial calcium/sodium antiporter. Biochem J 229: 161-166, 1985
51. Jung DW, Baysal K, Brierley GP: The sodium-calcium antiport of
heart mitochondria is not electroneutral. J Bioi Chern 270: 672-678,
1995
52. Baysal K, Jung DW, Gunter KK, Gunter TE, Brierley GP: Na+-
dependent Ca2+efflux mechanism of heart mitochondria is not a passive
Ca+/2Na+ exchanger. Am J Physiol: 266: C80O--C808, 1994
53. Li W, Shariat-Madar Z, Powers M, Sun X, Lane RD, Garlid KD:
Reconstitution, identification, purification, and immunological
characterization of the 1\0-kDa Na+ICa2+ antiporter from beef heart
mitochondria. J Bioi Chern 267: 17983-17989, 1992
54. Vaghy PL, Johnson JD, Matlib MA, Wang T, Schwartz A: Selective
368
inhibition of Na+-induced Ca2+ release from heart mitochondria by
diltiazem and certain other Ca'+ antagonist drugs. J Bioi Chern 257:
600()"'{)002, 1982
55. Chiesi M, Schwaller R, Eichenberger K: Structural dependency of the
inhibitory action of benzodiazepines and related compounds on the
mitochondrial Na+-Ca'+ exchanger. Biochem Pharmacol 37: 4399-
4402, 1988
56. Cox DA, Mat!ib MA: A role for the mitochondrial Na+ -Ca'+ exchange
in the regulation of oxidative phosphorylation in isolated heart
mitochondria. J Bioi Chern 268: 938-947, 1993
57. Miyata H, Silverman HS, Sollott SJ, Lakatta EG, Stem MD, Hansford
RG: Measurement of mitochondrial free Ca 2+ concentration in living
single rat cardiac myocytes. Am J Physiol261: HI123-HI134, 1991
58. Chacon E, Ohata H, Harper IS, Trollinger DR, Herman B, Lemasters
JJ: Mitochondrial free calcium transients during excitation-contraction
coupling in rabbit cardiac myocytes. FEBS Fett 382: 31-36, 1996
59. Isenberg G, Han S, Schiefer A, Wendt-Gallitelli: M-F: Changes in
mitochondrial calcium concentration during the cardiac contraction
cycle. Cardiovasc Res 27: 1800-1809, 1993
60. Wendt-Gallitelli MF, Isenberg G: Total and free myoplasmic calcium
during a contraction cycle: X-ray micro-analysis in guinea-pig
ventricular myocytes. J Physiol Lond 435: 349-372, 1991
61. Moravec CS, Bond M: Calcium is released from the junctional
sarcoplasmic reticulum during cardiac muscle contraction. Am J
Physiol260: H989-H997, 1991
62. Moravec CS, Bond M: Effect of inotropic stimulation of mitochondrial
calcium in cardiac muscle. J Bioi Chern 267: 5310-5316, 1992
63. Robertson SP, Rotter JD, Rouslin W: The Ca'+ and Mg2+ dependence
ofCa'+ uptake and respiratory function of porcine heart mitochondria.
J Bioi Chern 257: 1743-1748, 1982
64. Crompton M: The regulation of mitochondrial calcium transport in
heart. Curr Top Membr Transp 25: 231-276, 1985
65. Sparagna GC, Gunter KK, Sheu S-S, Gunter TE: Mitochondrial Ca'+
uptake from physiological-type pulses of calcium: A description of
the rapid uptake mode. J Bioi Chern 270: 27510-27515, 1995
66. Rizzuto R, Simpson AWM, Brini M, Pozzan T: Rapid changes of
mitochondrial Ca2+ revealed by specifically targeted recombinant
aequorin. Nature 358: 325-327, 1992
67. Rizzuto R, Brini M, Murgia M, Pozzan T: Microdomains with high
Ca2+ close to IP 3-sensitive channels that are sensed by neighboring
mitochondria. Science 262: 744-747,1993
68. Rizzuto R, Bastianutto C, Brini M, Murgia M, Pozzan T: Mitochondrial
Ca2+ homeostasis inintact ceils. J Cell Bioi 126: 1183-1194, 1994
70. Kennedy ED, Rizzuto R, Theler J-M, Pralong W-F, Bastianutto C,
Pozzan T, Wollheim CB: Glucose-stimulated insulin secretion
correlates with changes in mitochondrial and cytosolic Ca 2+in aequorin
expressing INS-l cells. J Clin Invest 98: 2524-2538, 1996
71. Rutter GA, Theler J-M, Murgia M, Wollheim CB, Pozzan T, Rizzuto
R: Stimulated Ca 2+ influx raises mitochondrial free Ca'+ to supra-
micromolar levels in a pancreatic ~-celiline. J Bioi Chern 268: 22385-
22390, 1993
72. Pralong W-F, Spat A, Wollheim CB: Dynamic pacing of cell metab-
olism by intracellular Ca2+ transients. J Bioi Chern 269: 27310-27314,
1994
73. Hajn6czky G, Robb-Gaspers LD, Seitz MB, Thomas AP: Decoding
of cytosolic calcium oscillations in the mitochondria. Cell 82: 415-
424, 1995
74. Rutter GA, Burnett P, Rizzuto R, Brini M, Murgia M, Pozzan T, Tavare
JM, Denton RM: Subcellular imaging of intramitochondrial Ca'+ with
recombinant targeted aequorin: Significance for the regulation of
pyruvate dehydrogenase activity. Proc Nat! Acad Sci USA 93: 5489-
5494, 1996
75. Denton RM, McCormack JG: Ca'+ transport by mammalian mito-
chondria and its role in hormone action. Am J Physiol 249: E543-
554, 1985
76. Scott DA, Grotyohann LW, Cheung JY, Scaduto Jr. RC: Ratiometric
methodology for NAD(P)H measurement in the perfused rat heart using
surface fluorescence. Am J Physiol267: H636-H 644, 1994
77. White RL. Wittenberg BA: NADH fluorescence of isolated ventricular
myocytes: Effects of pacing, myoglobin, and oxygen supply. Biophys
J 65: 196-204, 1993
78. Balaban RS: Regulation of oxidative phosphorylation in the mam-
malian cell. Am J Physiol258: C377-C389, 1990
79. Wan B, Doumen C, Duszyuski J, Salama G, Vary TC, LaNoue KF:
Effects of cardiac work on electrical potential gradient across mito-
chondrial membrane in perfused rat hearts. Am J Physiol265: H453-
H460, 1993
80. Kawanishi T, Blank LM, Harootunian AT, Smith MT, Tsien RY: Ca'+
Oscillations induced by hormonal stimulation of individual fura-2-
loaded hepatocytes. J Bioi Chern 264: 12859-12866, 1989
81. Rink TJ, Hallam TJ: Calcium signaling in non-excitable cells: Notes
on oscillations and store refilling. Cell Calcium 10: 385-395, 1989
82. Hansford RG: Studies on the effects of coenzyme A-SH: Acetyl
coenzyme A, nicotinamide adenine dinucleotide: Reduced nicotin-
amide adenine dinucleotide, and adenosine diphosphate: Adenosine
triphosphate ratios on the interconversion of active and inactive
pyruvate dehydrogenase in isolated rat heart mitochondria. J Bioi Chern
251: 5483-5489, 1976
83. Meglasson MD, Matschinsky FM: Pancreatic islet glucose metabolism
and regulation of insulin secretion. Diabetes Metab Rev 2: 163-214, 1986
84. Loew LM, Carrington W, Tuft RA, Fay FS: Physiological cytosolic
Ca'+ transients evoke concurrent mitochondrial depolarizations. Proc
Nat! Acad Sci 91: 12579-12583, 1994
85. Dhalla NS, Sulakhe PV, Fedelesova M, Yates J C: Molecular
abnormalities in cardiomyopathy. Adv Cardiol13: 282-300,1974
86. Markiewicz W, Wu SS, Parmley WW, Higgins CB, Sievers R, James
TL, Wikman-Coffelt J, James G: Evaluation of the hereditary Syrian
hamster cardiomyopathy by 31p nuclear magnetic resonance spectro-
scopy: Improvement after acute verapamil therapy. Circ Res 59: 597-
604, 1986
87. Proschek L, Jasmin G: Hereditary polymyopathy and cardiomyopathy
in the Syrian hamster. II. Development of heart necrotic changes in
relation to defective mitochondrial function. Muscle Nerve 5: 26-32,
1982
88. Wrogemann K, Nylen EG: Mitochondrial calcium overloading in
cardiomyopathic hamsters. J Mol Cell Cardioll0: 185-195, 1978
89. Bond M, laraki A-R, Disch CH, Healy BP: Subcellular calcium content
in cardiomyopathic hamster hearts in vivo: an electron probe study.
Circ Res 64: 1001-1012,1989
90. DiLisa F, Fan C-Z, Gambassi G, Hogue BA, Kudryashova I, Hansford
RG: Altered pyrovate dehydrogenase control and mitochondrial free
Ca2+in heart of cardiomyopathic hamsters. Am 1 Physiol264: H2188-
H2197,1993
91. Wikman-Coffelt 1, Sievers R, Parmley WW, Jasmin G: Cardio-
myopathic and healthy acidotic hamster hearts: Mitochondrial activity
may regulate cardiac performance. Cardiovasc Res 20: 471-481, 1986
92. Wikman-Coffeltl, StefenelliT, Wu ST, ParmleyWW, Jasmin G: [Ca2+l;
Transients in the cardiomyopathic hamster heart. Circ Res 68: 45-51,
1991
93. Keller E, Moravec CS, Bond M: Altered subcellular Ca'+ regulation
in papillary muscles from cardiomyopathic hamster hearts. Am J
Physiol268: HI875-Hl883, 1995
94. Kruger C, Erdmann E, Nabauer M, Beuckelmann DJ: Intracellular
calcium handling in isolated ventricular myocytes from cardio-
myopathic hamsters (strian BIO 14.6) with congestive heart failure.
Cell Calcium 16: 500-508,1994

95. Leisey JR, Grotyohann LW, Scott DA, Scaduto RC Jr: Regulation of
cardiac mitochondrial calcium by average extramitochondrial
calcium. Am J Physiol265: HI203-H1208, 1993
96. Regan TJ, Lyons MM, Ahmed SS: Evidence for cardiomyopathy in
familial diabetes mellitus. J Clin Invest 60: 885--889, 1977
97. Rodrigues B, McNeill JH: The diabetic heart: Metabolic causes for
the development of a cardiomyopathy. Cardiovasc Res 26: 913-922,
1992
98. Chatham JC, Forder JR: A 13C-NMR study of glucose oxidation in
the intact functioning rat heart following diabetes-induced cardio-
myopathy. J Mol Cell Cardiol25: 1203-1213,1993
99. Kerbey AL, Randle PJ: Thermolabile factor accelerates pyruvate
dehydrogenase kinase reaction in heart mitochondria of starved or
alloxan-diabetic rats. FEBS Lett 27: 188-192, 1981
100. Kobayashi K, Neely JR: Effects of increased cardiac work on pyruvate
dehydrogenase activity in hearts from diabetic animals. J Mol Cell
Cardiol15: 347-357, 1983
101. Nicholl TA, Lopaschuk GD, McNeill JH: Effects offree fatty acids
and dichloroacetate on isolated working diabetic rat heart. Am J
Physiol261: HI053-HI059, 1991
102. Latifpour J, McNeil JH: Cardiac autonomic receptors: effect oflong-
term experimental diabetes. J Pharmacol Exp Ther 230: 242-249,
1984
103. Nishio Y, KashivagiA, Kida Y, Kodoma M,Abe N, Saeki Y, Shigeta
Y: Deficiency of cardiac b-adrenergic receptors in streptozocin-
induced diabetic rats. Diabetes 37: 1181-1187, 1988
104. Yu Z, Quamme GA, McNeill JH: Depressed [Ca2+]. responses to
isoproterenol and cAMP in isolated cardiomyocytes from experi-
369
mental diabetic rats. Am J Physiol266: H2334-H2342, 1994
105. Noda N, Hayashi H, Satoh H, Terada H, Hirano M, Kobayashi A,
Yamazaki N: Ca2+ transients and cell shortening in diabetic rat
ventricular myocytes. Jpn Circ J 57: 449-457, 1993.
106. Lagadic-Gossmann D, Buckler KJ, Le Prigent K, Feuvray D: Altered
Ca2+ handling in ventricular myocytes isolated from diabetic rats.
Am J Physiol270: HI529--HI537, 1996
107. Bouchard RA, Bose D: Influence of experimental diabetes on
sarcoplasmic reticulum function in rat ventricular muscle. Am J
Physiol260: H341-H354, 1991
108. Ganguly PK, Pierce GN, Dhalla KS, Dhalla NS: Defective sarco-
plasmic reticular calcium transport in diabetic cardiomyopathy. Am
J Physiol244: E528-E535, 1983
109. Lopaschuk GD, Tahiliani AG, Vadlamuri R, Katz S, McNeil JH:
Cardiac sarcoplasmic reticulum function in insulin- or carnitine-
treated diabetic rats. Am J Physiol245: H969--H976, 1983
110. Jourdon P, Feuvray D: Calcium and potassium currents in ventricular
myocytes isolated from diabetic rats. J Physiol470: 411-429,1993
111. Pierce GN, Dhalla NS: Heart mitochondrial function in chronic
experimental diabetes in rats. Can J Cardioll: 48-54, 1985
112. Tanaka Y, Konno N, Kako KJ: Mitochondrial dysfunction observed
in situ in cardiomyocytes of rats in experimental diabetes. Cardiovase
Res 26: 409-414, 1992
113. Beuckelmann DJ, Nabauer M, Erdmann E: Intracellular calcium
handling in isolated ventricular myocytes from patients with terminal
heart failure. Circulation 85: 104~1055, 1992
114. Katz AM: Energetics and the failing heart. Hosp Pract Aug 26: 78-
90, 1991
Molecular and Cellular Biochemistry 184: 371-376, 1998.
1998 Kluwer Academic Publishers.

Modulation of cell calcium signals by mitochondria

Laurence S. Jouaville, Fran<;ois Ichas and Jean-Pierre Mazat

CJF INSERM 97-05, Universite Victor Segalen, Bordeaux 2, 146 rue Leo Saignat, F-33076 Bordeaux-Cedex, France

Abstract
It is now clearer and clearer that mitochondria playa role, and perhaps an active role, in cell calcium signalling. The fact that
mitochondria can exhibit a Ca2+-induced Ca2+ release (mCICR, Ichas et al. [37]) reinforces this concept and makes the
mitochondria an essential element in the relay of Ca2+ wave propagation. It must be emphasized that the modulation of cell
Ca2+ signals by mitochondria depends upon their energetic status, thus making mitochondria an essential link between energy
metabolism and calcium signalling inside the cell. (Mol Cell Biochem 184: 371-376, 1998)

Key words: cell calcium signalling, mitochondria, permeability transition pore (PTP), calcium waves

Introduction Calcium uptake and efflux by

Despite the fact that mitochondria have been known since
the sixties to be able to accumulate calcium, there was no
evidence until recently that they could participate in cytosolic
calcium movements during physiological processes. The
recent evidence that mitochondria are able to accumulate
calcium during cell calcium signals [1-5] has raised the
question of the physiological significance of their involve-
ment in this process. Given the effect of mitochondrial
calcium sequestration on mitochondrial metabolism, mito-
chondria can be considered as targets of calcium signals.
Indeed, it must be stressed that changes in Cam have been
shown to modulate the activity of key matrix enzymes
(dehydrogenases) involved in substrate oxidation ([6-8 and
the previous chapter of this book).
Here we emphasize another role of Ca 2+ uptake and
release by mitochondria: the modulation of Ca 2+ signals
emitted by other Ca 2+ stores.
We will first briefly describe the Ca2+ uptake and release
pathways in mitochondria, and more particularly the perm-
eability transition pore (or PTP).
We will then review some evidence implicating mito-
chondria in modulation of cell calcium signals. We will
finally demonstrate that mitochondria are themselves able
to generate calcium spikes induced by calcium (mCICR).
mitochondria
It is now accepted that there are three distinct mechanisms of
Ca2+transport operating in the inner mitochondrial membrane
of vertebrates (see Fig. 1). The influx takes place through a
specific uniporter. The efflux mechanisms are driven by Na+
or H+ exchangers. To these efflux pathways we have to add
the permeability transition pore (PTP or MTP), which is not
Ca 2+ specific but can be activated by Ca2+. We will describe
the properties of these pathways in tum (for a review see [9]):
The mitochondrial calcium uniporter
A uniporter is a mechanism which facilitates ion diffusion
in the direction of its electrochemical gradient, without the
help of any other ion or source of energy. The mitochondrial
Ca 2+ uniporter allows the rapid accumulation of Ca2+ inside
the mitochondrial matrix, and also other divalent cations in
the order Sr2+> Mn 2+> Ba2+>Fe 2+>Pb 2+> lanthanides. Ca2+
accumulation via the uniporter is dependent on the external
Ca2+ concentration and the transmembrane potential ~qt.
This dual dependency makes it difficult to precisely assess
the rates of Ca2+ uptake through the uniporter. Furthermore
this is an electophoretic process: the Ca 2+ entry decreases
the transmembrane potential and thus the uptake rate. Hence
it is likely that the measured rate of Ca2+ entry are under-
estimated. The highest reported rate is 1750 nmol/min/mg

Address for offprints: J.P. Mazat, Laboratoire GESBI, Universite Victor Segal en, Bordeaux 2, 146 rue Leo Saignat, F-33076 Bordeaux-Cedex,
France
372
Ruthenium Red
KCN
Oligomycin
Fig. 1. Ca'+ influx and effiux pathways in mitochondria with their specific inhibitors. Abbreviations: U - Ca'+ uniporter; RC - respiratory chain; PTP -
permeability transition pore; 1- Ca'+/2H+ exchanger (Na+ independent); D - Ca2+/2Na+ exchanger (Na+ dependent); COP - benzothiazepine COP-
37157; .:'1'1' - transmembrane potential.
prot in the heart [10]. The accumulation of Ca2+ via the
uniporter is seen as a co-operative process. Several studies
report a sigmoidal shape of the rate ofCa2+accumulation as a
function of external calcium concentration [11-16]. Several
inhibitors of the Ca 2+ uniporter have been described: (i)
competitive inhibitors, which are the other ions that can be
transported by the uniporter (Sr2+, Mn2+, Ba2+, Fe 2+, Pb2+ and
the lanthanides); (ii) Mg2+, the polyamines and W, which can
interfere with Ca2+transport by binding or by charge effects,
although they are not transported; (iii) polycations such as
ruthenium red and cobalt hexamine, which have a low inhibit-
ion constant; and (iv) some drugs such as local anesthetics,
~-blockers, guanidines and the diuretics. Two types of
activators are known: (i) some transported ions, such as Ca2+
itself and some lanthanides, such as Pr3+ ,which appear to bind
to an activator site; and (ii) specific polyamines, the spermine
particularly, which activates Ca2+ accumulation at low Ca2+
concentrations.
More recently, a new influx pathway has been described
that might correspond to a specific conformation of the
uniporter and that allows mitochondria to accumulate
significant amounts of calcium even when external calcium
concentration is submicromolar [17].
H+ Cyclosporin A
The Ca 2+12H+ exchanger
This exchanger is predominant in mitochondria isolated
from liver, kidney and smooth muscle [18, 19]. Sr2+, Mn2+
and Ba2 are also transported. The transport mechanism is
second order with a V max = 1.2 0.1 nmol/min/mg prot in
rat liver mitochondria. This rate is three orders of magnitude
lower than the maximal rate of Ca2+ uptake. The energy of
the efflux is taken from the proton-motive force generated
by the respiratory chain, and is thus inhibited by KCN and
low concentrations 'of uncouplers.
The Ca2+12Na+ exchanger
This exchanger is predominant in mitochondria isolated from
heart, skeletal muscle, brain, parotid gland, adrenal cortex, brown
fat and some other excitable tissues [20, 21]. This exchanger
can also transport Sr2+ but not Mn2+. Li+ can take the place of
Na+. The Ca2+ transport mediated through this mechanism is
inhibited by a great variety of inhibitors: Mg2+, Mn2+, external
Ca2+, benzothiazepines, trifluoperazine, diltiazem, verapamil,
bepridil, amiloride, tetraphenylphosphonium. External Na+,
spermine and spermidine are activators ofthe exchanger. In liver
[22] and in heart [23] this mechanism exhibits second order
kinetics towards Na+ and first order kinetics towards Ca2+. The
 373

maximal rate of this exchanger is greater in brain mitochondria

(30 nmoles/min/mg prot [24]) than in heart mitochondria (18
nmoles/min/mg prot [21]) and liver mitochondria (3 nmolesl
out
min/mg prot [22]).
It used to be accepted that the Ca 2+/2Na+ was a neutral ++
exchanger; it seems however that this exchanger could be
electrogenic with the stoichiometry: Ca2+/3Na+ [6].

The permeability transition pore (Fig. 2)

Isolated mitochondria can undergo a dramatic increase of M~+
permeability to ions and small molecules (PM < 1500) known
as the permeability transition (PT) [6]. This phenomenon is ADP
accompanied by a massive and irreversible loss of calcium,
Cyclosporin A
and an entry of protons giving rise to a collapse of the
transmembrane potential ~'P and ~pH. It has been known for
a long time under the name of 'Ca2+ induced uncoupling'
[25]. It can be experimentally monitored by the swelling of
mitochondria following the entry of sucrose.
At the end of the eighties, a mitochondrial channel (2-3
nm) presenting a large conductance (~ 1 nS,) was identified
by patch-clamp studies on liver mitoplasts [26--28], and
called mitochondrial Megachannel. This channel is inhibited
by cyclosporin A [29] and by several other agents that act
similarly on the PTP [25]. For this reason, it has been
suggested that the Megachannel and the PTP are the same
molecular entity [30]. This hypothesis has received additional
support [31].
A great variety of agents are able to affect the operation of
the PTP [2]. Some key factors are however crucial [31, 32]:

transmembrane potential (.1 P)
The PTP behaves as a voltage-dependent channel. High ~'P
(-180 m V) favours the closed conformation of the PTP and
depolarization increases its probability of opening. The PTP
site of voltage sensing can be modulated by adjacent dithiol
groups; their oxidation to disulfide increases the opening
probability at physiological ~'P.
matrix pH
The maximum opening probability is observed at a matrix
pH around 7.3. Opening ofthe PTP is inhibited at matrix pH
values lower than 7.0. It has been shown that this inhibitory
effect is mediated through the reversible protonation of
histidyl residues localized on the inner side of the inner
mitochondrial membrane [33].
divalent cations
There are at least two sites where divalent cations can affect
the PTP open-closed transitions: (i) an external site where
the binding of divalent cations, including Ca 2+ itself
Fig. 2. Activation and inhibition ofPTP opening. Abbreviations as in Fig. 1.
decreases the opening probability and (ii) an internal site
where Ca2+binding leads to an increased probability of pore
opening, though other cations such as Mg2+, Sr2+ and Mn 2+
have the opposite effect.
adenine nucleotides
The opening probability of the PTP is decreased by the
presence of adenine nucleotides. ADP is more efficient than
ATP. AMP has a negligible effect. ADP acts synergestically
with cyclosporin A.
cyclosporin A
Cyclosporin A is the most efficient inhibitor of the PTP. Its
inhibition is of the competitive type with Ca2+. It seems to
occur through its binding to a mitochondrial cyclophilin
(peptidylprolyl-cis-trans isomerase). Unlike the require-
ments for immunosupression, inhibition of calcineurin by
CsA-cyclophilin complex is not required for PTP inhibition
374
[31]. Cyclosporin appears to act on the mitochondrial PTP in
cardiomyocytes to block Ca2+efflux from mitochondria [34].
Even though the PTP is a non-selective channel, because
the inner mitochondrial membrane is impermeable to most
ions (particularly H+ and K+) except to Ca 2+ it behaves as a
selective channel for Ca2+. This is because mitochondria are
able to accumulate large quantities of Ca2+through the Ca2+
uniporter. This situation is close to the one observed with
the ryanodine receptor in the endoplasmic reticulum which
behaves as a specific Ca 2+ channel despite its high perm-
eability for sucrose [35] and its conductance for monovalent
cations (around 1 nS). Several analogies between the PTP
and the ryanodine receptor have been pointed out by Bernardi
and Petronilli [36]. As we will see later on the mitochondrial
PTP can as the ryanodine receptor participate to a Ca 2+-
induced Ca 2+ release.
Modulation of Ca2+ signals by
mitochondria
Several recent reports point out the role of mitochondria in
modulating cell calcium signals initially emitted by the
reticulum [37-43]. We showed [37] that antimycin, a
respiratory chain inhibitor, substantially decreases the
calcium spike induced by ATP in Ehrlich ascites tumour
cells. Cyclospoin A has the same effect indicating that the
PTP is involved in this effect.
Friel and Tsien [38] have studied the role of FCCP, a
mitochondrial uncoupler, in modulating the intracellular free
calcium [Ca2+ l responses elicited by either depolarization
or Ca 2+ release from a caffeine- and ryanodine-sensitive
i
store in bullfrog sympathetic neurons. This FCCP-sensitive
store slows both the rise in [Ca 2 +l during stimulation
(apparently by accumulating Ca2+ from the cytosol) and the
recovery following stimulation (by releasing the accumulated
Ca 2+into the cytosol). They propose several physiologically
beneficial functions for this process: neuroprotection by
buffering high Ca 2+ concentrations which can have both
short-term and longterm cytotoxic effects, prolongation of
the cytosolic Ca 2+signal and thus amplification of the effects
of a given stimulus and metabolic signalling that will reflect
the cell's recent history of [Ca 2+]i
Jouaville et at. [39] showed that energization of mito-
chondria by injection of respiratory substrates in Xenopus
laevis oocytes can strengthen the Ca 2+ wave activity
triggered by Ins(l ,4,5)P3 : Ca2+wave amplitude, velocity and
interwave period all increase. The effects of the substrates
pyruvate/malate are blocked by ruthenium red (at the
mitochondrial Ca 2+ uniporter), by rotenone (a complex I
inhibitor) and by antimycin (a complex III inhibitor), and can
subsequently be rescued at complex IV with tetramethyl-
phenylenediamine (TMPD) + ascorbate. They showed that
potential-driven mitochondrial Ca 2+ uptake is the major
factor in the modulation of the waves and not increasedATP
synthesis (which could act through an increase in the activity
of Ca2+ATPase of the reticulum for instance).
In rat chromaffin cells, Babcock et al. [42] demonstrated,
using patch clamp methods that CCCP and ruthenium red
decrease the rate of rapid Ca 2+ clearance after Ca 2+ entry
into the cell induced by step depolarization of the plasma
membrane. By simultaneous monitoring of mitochondrial
and cytosolic Ca2+ at high temporal resolution they found
that mitochondrial Ca2+uptake limits the rise and underlies
the rapid decay of cytosolic Ca2+excursions produced either
by Ca2+ entry or by mobilization of reticular stores. They
demonstrate that export of mitochondrial Ca 2 + prolongs
complete cytosolic Ca2+ concentration recovery. The Ca2+/
Na+ exchanger appears to take part in this phenomenon since
its inhibition reversibly hastens the final recovery of the
cytosolic calcium concentration.
Some other studies point out the close association of
mitochondria and reticulum. For instance, Simpson and
Russel [43] have examined the spatial and temporal nature of
Ca2+ signals activated via the phosphoinositide pathway that
underlie the oligodendrocyte Ca2+ response characteristics.
The results of their study demonstrate that stimulation of
phospho-inositide-coupled muscarinic acetylcholino-
receptors activates propagating Ca2+ wave fronts in oligo-
dendrocytes and that the characteristics of these waves are
dependent on mitochondrial location and function.
We showed [37, 44] that antimycin decreases the Ins(l ,4 ,5)P3-
dependent calcium spikes induced by extracellular ATP in
Ehrlich ascites tumor cells. Budds and Nicholls [41]
observed the same effect using FCCP, a mitochondrial
uncoupler. Additionally, we found that cyclosporin A also
inhibits the Ins(l ,4,5)P3 -dependent calcium spikes suggesting
that the PTP is involved. It thus appears that mitochondria
are active participants in cell Ca2+ signalling, essentially by
rapidly accumulating Ca2+and then releasing large quantities
ofCa2+. Indeed, we have reported results [37,44] that indicate
that mitochondria can take a more active role than being a
simple buffer of cytosolic calcium variations due to other
stores. We observed a reversible Ca2+-induced Ca2+ release
in rat liver mitochondria, that takes the form of Ca2+ spike.
This is an all-or nothing process with a threshold dependence
on both the frequency and the amplitude of the Ca 2+pulses
used as stimuli. This Ca 2+ spiking is accompanied by a
depolarization spike which is the mirror image of the Ca 2+
spike. Both spikes rely on the transient operation of the
PTP and can be initiated during IP 3 -mediated calcium
responses in cells.

IP3R
ER
Fig. 3. Proposed mechanism of active involvement of mitochondria in the propagation of calcium waves. Abbreviations: U - Ca2+ uniporter; M - mitochondria;
ER -endoplasmic reticulum; SERCA - sarco-endoplasmic reticulum Ca2 + ATPase; IP3R -~ InsP3 receptor.
Conclusion
It appears more and more clear that mitochondria playa role,
and perhaps an active role, in cell calcium signalling. This led
Babcock et al. [42] to propose the concept of an intracellular
calcium network, linking various Ca2+ stores in an interactive
manner. The fact that mitochondria can exhibit a Ca2+-induced
Ca 2+ release (mCICR [37]) reinforces this concept and makes
the mitochondria an essential element in the relay ofCa2+ wave
propagation (Fig. 3). It must be emphasized that the modu-
lation of cell Ca2+ signals by mitochondria depends upon their
energetic status. Mitochondria are thus a link between energy
metabolism and cell calcium signalling.
From another point of view, modulating the Ca 2+ signals
emitted by other Ca 2+ stores could be the means used by
mitochondria in order to communicate with the rest of the cell
(particularly the nucleus) and to inform it of its needs or
supplies.
Thus the entry of mitochondria into the field of cell
calcium signalling opens several new direction of research.
Acknowledgements
This work was funded by grants from la Region Aquitaine,
la FRM, l' AFM, and la Ligue Contre Ie Cancer (L.C.c.).
L.S.J. was supported by L.C.C .. The authors wish to thank
Dr. D. Fell for many valuable comments.
References
1. Rizzuto R, Simpson AWM, Brini M, Pozzan T: Rapid changes of
mitochondrial Ca 2+ revealed by specifically targeted recombinant
375
Ca2+
IP3R
ER
aequorin. Nature 358: 325-327, 1992
2. Rizzuto R, Brini M, Murgia M, Pozzan T: Microdomains with
high Ca H close to IP]-sensitive channels are sensed by neighboring
mitochondria. Science 262: 744-747, 1993
3. Rizzuto R, Bastianutto M, Brini M, Murgia M, Pozzan T:
Mitochondrial Ca 2+ homeostasis in intact cells. I Cell Bioi 126:
1183-1194, 1994
4. Rutter GA, Theler 1M, Murgia M, Wolheim CB, Pozzan T, Rizzuto
R: Increased Ca 2+ influx raises mitochondrial free Ca 2+ to
micromolar levers in a pancratic beta-cell line. I Bioi Chern 268:
22385-22390, 1993
5. Lawrie AM, Rizzuto R, Pozzan T, Simpson AWM: A role for
calcium influx in the regulation of mitochondrial calcium in
endothelial cells. J Bioi Chern 271: 10753-10759, 1996
6. Gunter TE, Gunter KK, Sheu SS, Gavin CE: Mitochondrial Ca 2+
transport: Physiological and pathological relevance. Am J Physiol
267: C313-C339, 1994
7. McCormack IC, Halestrap AP, Denton RM: Role of calcium ions
in regulation of mammalian intramitochondrial metabolism.
Physiol Rev 70: 391-495, 1990
8. McCormack JG, Denton RM: Signal transduction by intramitochondrial
Ca2+ in mammalian energy metabolism. News Physiol Sci 9: 71-76, 1994
9. Gunter TE, Pfeiffer DR: Mechanisms by which mitochondria transport
calcium. Am J Physiol258: C755-C786, 1990
10. McMillin-Wood J, Wolkowicz PE, Chu A, Tate CA, Goldstone MA,
Entman ML: Calcium uptake by two preparations of mitochondria from
11.
heart. Biochem Biophys Acta 591: 251-265,1980
Bragadin M, Pozzan T, Azzone GF: Kinetics ofCa 2 + carrier in rat liver
mitochondria. Biochemistry 18: 5972-5978,1979
12. Bygrave FL, Reed KC, SpencerT: Cooperative interactions in energy-
dependent accumulation of Ca2+ by isolated rat liver mitochondria.
Nature New Bioi 230: 89-91, 1971
13. Heaton GM, Nicholls DG: The Ca 2+ conductance of the inner
membrane of rat liver mitochondria and the determination of the Ca 2+
electrochemical gradient. Biochem I 156: 635-646, 1976
14. Huston SM, Pfeiffer DR, Lardy HA: Effect of cations and anions on the
steady state kinetics of energy-dependent Ca2+ transport in rat liver
mitochondria. IBiol Chem251: 5251-5258,1976
15. Reed KC, Bygrave FL: A kinetic study of mitochondrial Ca 2 + transport.
Eur JBiochem 55: 97-504,1975
16. Vinogradov A, Scarpa A: The initial velocities of Ca 2 + uptake by
376
rat liver mitochondria. J BioI Chern 248: 5527-5531, 1975
17. Sparagna GC, Gunter KK, Sheu SS, Gunter TE: Mitochondrial
calcium uptake from physiological-type pulses of calcium. A
description of the rapid uptake mode. J BioI Chern 270: 27510--
27515, 1995
18. Fiskum G, Lehninger AL: Regulated release of Ca2+I2H+ antiport.
J BioI Chern 254: 6236-6239,1979
19. Puskin JS, Gunter TE, Gunter KK, Russel PR: Evidence for more
than one Ca2 + transport mechanism in mitochondria. Biochemistry
15: 3834--3842, 1976
20. Crompton M, Capano M, Carafoli E: The sodium-induced efflux
of calcium from heart mitochondria. A possible mechanism for
the regulation of mitochondrial transport. Eur J Biochem 69:
453--4562, 1976
21. Crompton M, Heid I: The cycling of Ca'+, Na'+, and H+ across the
inner membrane of cardiac mitochondria. Eur J Biochem 91:
599-608, 1978
22. Crompton M, Kunzi M, Carafoli E: The calcium-induced and
sodium-induced effluxes of calcium from heart mitochondria. Eur
J Biochem 79: 549-558, 1977
23. Wingrove DE, Gunter TE: Kinetics of mitochondrial calcium
transport. II. A kinetic description of the sodium-dependent
calcium efflux mechanism of liver mitochondria and inhibition
by ruthenium red and by tetraphenylphosphonium. J BioI Chern
261: 15166-15171, 1986
24. Gavin CE, Gunter KK, Gunter TE: Mn'+ transport across biological
membranes may be monitored spectroscopically using Ca2 + indicator
dye antipyrylazo III. Anal Biochem 192: 44--48, 1991
25. Zoratti M, Szabo I: The mitochondrial permeability transition. Biochem
BiophysActa 1241: 139--176,1995
26. Petronilli V, Szabo i, Zoratti M: The inner mitochondrial
membrane contains ion-conducting channels similar to those found
in bacteria. FEBS Lett 259: 137-143, 1989
27. Tedeschi H, Kinnaly KW, Mannella CA: Properties of channels in the
mitochondrial outer membrane. J Bioenerg Biomembr 21: 451-559, 1989
28. Kinnaly KW, Zorov D,Antonenko Y, Perini S: Calcium modulation of
the mitochondrial inner membrane channel activity. Biochem
Biophys Res Commun 176: 1183-1188, 1991
29. Szabo I, Zoratti M: The giant channel of the inner mitochondrial
membrane is inhibited by cyclosporine A. J BioI Chern 267: 3376--3379,
1991
30. Szabo I, Zoratti M: The mitochondrial megachannel is the permeability
transition pore. J Bioenerg Biomembr 24: 111-117, 1992
31. Bernardi P, Broekemeir KM, Pfeiffer DR: Recent progress on
regulation of the mitochondrial permeability transition pore; a
cyclosporin-sensitive pore in the inner mitochondrial membrane.
J Bioenerg Biomembr 26: 509-517, 1994
32. Petronilli V, Cola C, Bernardi P: Modulation of the mitochondrial
cyclosporin A-sensitive permeability transition pore. J BioI Chern 268:
1011-1016,1993
33. Nicolli A, Petronilli V, Bernardi P: Modulation of the mitochondrial
cyclosporin A-sensitive permeability transition pore by matrix pH.
Evidence that the pore open-closed probability is regulated by
reversible histidine protonation. Biochem 32: 4461--4465, 1993
34. Altschuld RA, Hohl CM, Castillo LC, Garleb AA, Starling RC, Brierley
GP: Cyclosporin inhibits mitochondrial calcium efflux in isolated adult
rat ventricular cardiomyocytes. Am J Physiol262: H1699--H1704, 1992
35. Meissner G: Ryanodine activation and inhibition of the Ca'+ release
channel of sarcoplasmic reticulum. J BioI Chern 261: 6300--6306, 1986
36. Bernardi P, Petronilli V: The permeability transition pore as a mito-
chondrial calcium release channel; a critical appraisal. J Bioenerg
Biomembr28: 129--136, 1996
37. Ichas F, Jouaville LS, Sidash SS, Mazat J-P, Holmuhamedov EL:
Mitochondrial calcium spiking: A transduction mechanism based on
calcium-induced permeability transition involved in cell calcium
signaling. FEBS Lett 348: 211-215, 1994
38. Friel DD, Tsien RW: An FCCP-sensitive Ca ' + store in bullfrog
sympathetic neurons and its participation in stimulus-evoked changes
in [Ca'+],. J Neurosci 14: 4007--4024, 1994
39. Jouaville LS, Ichas F, Holmuhamedov EL, Camacho P, Lechleiter JD:
Synchronization of calcium waves by mitochondrial substrates in
Xenopus laevis oocytes. Nature 77: 438--441,1996
40. Herrington. J, Park YB, Babcock DF, Hille B: Dominant role of
mitochondria in clearance oflarge Ca'+ loads from rat adrenal chromaffin
cells. Neuron 16: 219--228, 1996
41. Budd SL, Nicholls DG: Are-evaluation of the role of mitochondria in
neuronal calcium homeostasis. J Neurochem 66: 403--411, 1996
42. Babcock DF, Herrington J, Goodwin PC, Park YB, Hille B: Mitochondrial
participation in the intracellular Ca2+ network. J Cell BioI 136: 833--
844, 1997
43. Simpson PB, Russel JT: Mitochondria support inositol 1,4,5-tris-
phosphate-mediated Ca' + waves in cultured oligodendrocytes. J BioI
Chern 271: 33493-33501, 1996
44. lchas F, JouaviJle LS, Mazat JP: Mitochondria are excitable organelles
capable of generating and conveying electrical and calcium signals. Cell
89: 1145-1153, 1997
 PART IV

BIOENERGETICS AND MEDICINE

Molecular and Cellular Biochemistry 184: 379-391, 1998.
1998 Kluwer Academic Publishers.

Mitochondrial function as a determinant of

recovery or death in cell response to injury
Fabio Di Lisa1 and Paolo Bemardi2
CNR Unit for the Study of Biomembranes and the Departments of I Biological Chemistry and 2Biomedical Sciences,
University of Padova, Viale G. Colombo 3,1-35121 Padova, Italy

Abstract
Many pathological conditions can be the cause or the consequence of mitochondrial dysfunction. For instance anoxia, which
is initiated by a critical reduction of oxygen availability for mitochondrial oxidations, is followed by a wide variety of
mitochondrial alterations. A crucial role in the evolution of cell injury is to be attributed to the direction of operation of the FII
ATPase, which may tum mitochondria into the major consumers of cellular ATP in the futile attempt to restore the proton
electrochemical gradient. On the other hand, functional mitochondria can paradoxically accelerate or exacerbate cell damage.
This concept is particularly relevant for the ischemic myocardium. Indeed, inhibition of the respiratory chain or addition of
uncouplers of oxidative phosphorylation can both limit the extent of enzyme release in the intact heart and prevent the onset
of irreversible morphological changes in isolated myocytes. From studies on different tissues in a variety of pathological
conditions a general consensus emerges on the role of intracellular Ca2+ overload as a pivotal link between cellular alterations
and mitochondrial dysfunction. Oxidative phosphorylation is reduced by a massive mitochondrial uptake of Ca2+, resulting in
a vicious cycle whereby the reduced ATP availability is followed by a failure of the mechanisms which extrude Ca2+ from the
sarcoplasm. In addition, the rise in [Ca2+l could promote opening of the cyclosporin-sensitive mitochondrial permeability
transition pore, leading to a sudden ~'"m dissipation. Here, we review the changes in intracellular and intramitochondrial ionic
homeostasis occurring during ischemia and reperfusion. In particular, we evaluate the potential contribution of the permeability
transition pore to cellular damage and discuss the mechanisms which can determine the cellular fate from a mitochondrial point
of view. (Moll Cell Biochem 184: 379-391,1998)

Key words: myocytes, ischemia-reperfusion, hypoxia, cyclosporin, membrane permeability, channels

Introduction exacerbate cell damage. For instance, as discussed in the

Cell injury is a perturbation of cellular vital processes which
initiates a series of events leading to functional and structural
alterations. In the most severe cases cell death is the final
outcome, but cell recovery is possible upon early interruption
of the damaging condition. This means that a point of no
return exists which can only be defined after the fact. Indeed,
we still ignore the primary molecular mechanisms which
make the injury irreversible.
It is quite obvious that maintenance of mitochondrial
function is essential for cell survival, and that its complete
loss inevitably leads to cell death. However, a number of
observations indicate that mitochondria can accelerate or
following Sections, inhibition of the respiratory chain or
addition of uncouplers of oxidative phosphorylation limit the
extent of enzyme release in different models of myocardial
damage such as post-ischemic reperfusion and calcium
paradox [1,2]. On the other hand, under pathological
conditions not only mitochondria cease to be the major ATP
producers of the cell: they can also become its major con-
sumers owing to the hydrolytic activity of the Fli ATPase
[3-5]. This inverse operation of the mitochondrial ATPase
could also precipitate other harmful conditions, and cause
massive ATP hydrolysis in the futile attempt to restore the
proton electrochemical gradient (~ilH) collapsed by the
opening of a proton conductive pathway.

Address for o./JjJrints: Fabio Di Lisa and Paolo Bernardi, CNR, Unit for the Study of Biomembranes, Viale G. Colombo 3, 1-35121 Padova, Italy
380
The permeability transition pore (MTP)
Since the beginning of work on oxidative phosphorylation
in isolated mitochondria, conditions have been described that
cause a Ca2+-dependent increase of mitochondrial perm-
eability to ions and solutes with molecular weights up to 1500
Daltons, with matrix swelling and inhibition of oxidative
phosphorylation. This phenomenon (the permeability trans-
ition) is today ascribed to opening of the permeability
transition pore, a regulated channel inhibited by submicro-
molar concentration of cyclosporinA (see ref [6] for a review
and references therein for the original papers). The pore
appears to coincide with the mitochondrial megachannel, a
high-conductance channel identified by patch clamp studies
of rat liver mitoplasts [7-9].
The molecular nature of the MTP is not known. Based on
the effects of two inhibitors of the adenine nucleotide
translocase (bongkrekate, which inhibits the permeability
transition, and atractylate, which promotes it) and of ADP
(which is inhibitory) it has been suggested that the translocase
itself is a pore component [10]. However, the vast majority
of pore inducers and inhibitors (including cyclosporinA) do
not affect the adenine nucleotide translocase at all. More
direct evidence for an involvement ofthe translocase (poss-
ibly as a complex with porin) was recently obtained in
electrophysiological studies. Brustovetsky and Klingenberg
have shown that highly purified adenine nucleotide trans-
locase can form a large conductance channel activated by Ca2+
and inhibited by H+ and bongkrekate (but not by cyclosporin
A) [11]. This channel displays a prominent voltage gating
leading to pore closure at high membrane potentials [11],
which is the predicted behaviour of both the permeability
transition pore [7] and ofthe megachannel [12]. In a different
approach, Brdiczka and Coworkers have isolated enzy-
matically active complexes of hexokinase bound to porin and
to the adenine nucleotide translocase by extraction of
mitochondria with low concentrations of detergent. When
reconstituted in lipid bilayers these complexes exhibited a
high conductance channel behaviour which could be in-
hibited by cyclosporin A [13]. Importantly, this preparation
also catalyzed Ca 2+-dependent and cyclosporin A-sensitive
transport ofATP and malate when reconstituted in liposomes
[l3].
Mechanistic aspects ofpore regulation
Irrespective of its molecular nature, over the last few years
several key features of pore regulation have been defined,
allowing a better description of its mode of operation, and
of the mechanism of action of a variety of pore inducers and
inhibitors.
Membrane potential
Although it had been known for quite some time that pore
opening can be triggered by uncouplers [14], its voltage
dependence was clearly recognized only recently [7]. The
voltage dependence appears to be an intrinsic property of the
pore, which can be observed both in studies on isolated
mitochondria (by indirectly modulating the membrane
potential with uncouplers), and in single channel experiments
in mitoplasts (by directly modulating the applied voltage) (see
[15, 16], respectively for reviews). At physiological mem-
brane potentials the pore favors the closed state, while it can
be opened by membrane depolarization. Based on these
findings, the existence of a sensor which decodes the voltage
changes into variations of the probability of pore opening has
been postulated [15].
Surface potential
Evidence that the pore can also sense changes of the surface
potential represents a substantial improvement in our under-
standing of how its responses are modulated by a variety of
heterogeneous agents [15, 17]. Amphipathic anions (like fatty
acids) favor pore opening with an effect that cannot be
explained by depolarization [15]; and conversely, polycations
(like spermine [18]), positively charged peptides [19] and
amphipathic cations (like sphingosine and trifluoroperazine)
inhibit pore opening, the latter acting independently of
inhibition of phospholipaseA2 [17]. These data are consistent
with the idea that a more positive surface potential favors pore
closure, while a more negative surface potential favors its
opening.
Gating (threshold) modulation
It appears that many effectors are able to modify the threshold
voltage (the' gating potential ') at which pore opening occurs.
Many pore inducers (for example Ca2+and oxidative agents)
shift the apparent gating potential to more negative (physio-
logical) values, thereby favoring pore opening, while many
pore inhibitors (like ADP and reducing agents) have the
opposite effect and favor its closure [15]. Thus, pore opening
can be obtained by either depolarization, or by changing the
threshold potential at which opening occurs.
Divalent cations
The permeability transition is greatly favored by accumul-
ation of Ca2+ ions in the matrix, while it is counteracted by
Me 2+ ions like Mg2+, Sr2+ and Mn2+. The Ca2+ requirement is
not as strict as it has been previously assumed. In general, it
is becoming clear that pore opening in vitro can be easily
achieved at micromolar Ca2+concentrations, and that the Ca2+
requirement should not be intended in the sense of a massive
overload. The general effects of Me2+ ions on the pore can
be rationalized with the existence oftwo sites; (i) an external
site: when this site (apparent 150 0.2 mM) is occupied by a

Me 2+ ion (including Ca 2 + itself), the probability of pore
opening decreases [20]; (ii) an internal site: occupancy of this
site by Ca2+ leads to increased probability of pore opening,
while other Me 2+ (Sr+, Mn2+) are inhibitory, and apparently
compete with Ca2+[7, 20]; the apparent Iso of this site is harder
to estimate but could be in the micromolar range [21]. The
well-known inducing effects of Pi have been partly explained
with its ability of decreasing the intramitochondrial free
[Mg2+] [22].
Adenine nucleotides and their transloease
The probability of pore opening is decreased by adenine
nucleotides, ADP being more potent thanATP. As mentioned
above, the pore is strikingly affected in opposite directions
by atractylate and bongkrekate. The former (which is be-
lieved to lock the translocase in the 'c' conformation) is a
relatively potent pore inducer while the latter (which is
believed to lock the translocase in the 'm' conformation) is
a pore inhibitor ([ 6] and references therein). It has been noted
that the change from the 'm' to the 'c' conformation of the
translocase is accompanied by a large decrease ofthe surface
potential [23,24]. This observation may explain the induction
of pore opening or closure by the changes of surface potential
induced by the conformational changes of the carrier.
Pore modulation by redox agents
Oxidative stress has long been known to increase the prob-
ability of pore opening [6]. In general, pore opening is
favored by oxidants of both pyridine nucleotides (like
acetoacetate and oxaloacetate), glutathione (like tert-butyl-
hydroperoxide), and of dithiols (like diamide), as well as by
dithiol crosslinkers (like phenyl arsine oxide and arsenite).
Two sites can be experimentally distinguished, which cont-
ribute to pore modulation in an additive fashion. (i) A first
site (the 'S-site') can be traced to a critical dithiol, the dithiol-
disulfide interconversion being the critical event (a higher
probability of opening is associated with the disulfide, which
is equivalent to dithiol cross-linking with arsenite or pheny-
larsine oxide) [25]. The immediate oxidant is oxidized
glutathione, and many pore inducers (like organic hydro-
peroxides and hydrogen peroxide) affect the pore at the S-
site through changes in the levels of reduced glutathione
rather than by direct oxidation [26]. The S-site can be blocked
by both N-ethylmaleimide [25, 27] and monobromobimane
[28], and the effects of oxidation or cross-linking can be fully
reverted by reduction with dithiothreitol [26, 29]. (ii) A
second site (the 'P-site') is in apparent oxidation-reduction
equilibrium with the pyridine nucleotides pool. Pyridine
nucleotides oxidation is matched by an increased open
probability ofthe pore under conditions where the glutathione
pool is demonstrably kept in the fully reduced state. While it
is also sensitive to N-ethylmaleimide, the P-site cannot be
blocked by the sulfhydryl reagents monobromobimane and
381
methylmethanethiosulfonate, nor by dithiothreitol or 2-
mercaptoethanol [29].
Cyclosporin A
Cyclosporin A inhibits the permeability transition pore with
high affinity, the concentration required for 50% inhibition
being in the submicromolar range ([ 15] and references
therein). It is now well established that inhibition is exerted
through a mitochondrial isoform of cyclophilin, a matrix
peptidyl-prolyl cis-trans isomerase [30--32]. It appears likely
that cyclophilin modulates the pore by a direct interaction,
and that cyclosporin binding displaces cyclophilin from its
putative binding site on the pore [31, 32]. As a result of
cyclosporin binding, the enzymatic activity of cyclophilin is
inhibited, but it remains unclear whether this is relevant for
inhibition of the pore. On the other hand, calcineurin is not
involved in pore inhibition by cyclosporin A, whose mito-
chondrial effects can therefore be dissociated from immuno-
suppression [27, 32].
Protons
The permeability transition pore is potently inhibited by
acidic pH values [33]. The inhibitory effect ofH+ is exerted
from the matrix side ofthe inner membrane [34], and is linked
to reversible protonation of his tidy1 residues [35]. The pore
is extremely sensitive to variations of matrix pH, in that at
pH below about 7.0 the pore is substantially inhibited while
at pH 6.5 it does not open even in fully depolarized mito-
chondria [22].
Figure 1 presents a schematic model summarizing our
current working hypothesis of how the pore might be reg-
ulated. We suggest that the open-closed transitions may be
affected either through the voltage sensor (transitions I :::>
2a), or through matrix cyclophilin (transitions 1 :::> 2b). The
first transition would render the pore responsive to changes
of the proton motive force (for example by inhibition of
respiration or by uncoupling), or to signalling molecules
active on the sensor and produced at the plasma membrane
(like sphingolipids) or resulting from metabolism of phos-
pholipids in the mitochondrial membrane itself (such as fatty
acids produced by phospholipase A2 activation); the second
transition, which could be afforded by conformational
changes effected by matrix acidification or by cyclosporin
and may involve the Sand P sites, would convey signals
arising from the matrix. Both transitions could be affected by
dynamic inner-outer membrane interactions at the contact
sites [13], and possibly by hitherto unsuspected biochemical
signalling pathways. It must be stressed that this model is only
a working hypothesis, although it has the merit to accomodate
a large number of experimental findings within the relatively
simple framework of the pore as a voltage-dependent channel
modulated by cyclophilin.
382

Iclosed I
o
CsA 0
@

Fig. 1. Modulation of the open-closed transitions of the permeability transition pore. A working hypothesis. We suggest that open-closed transitions can be
modulated through two basic mechanisms: (i) a membrane effect through the voltage sensor, either because of a change of voltage, or of the threshold
potential at which opening occurs (transitions 1 ~ 2a); (ii) a matrix effect through (un)binding of mitochondrial cyclophilin, as exemplified here for CsA
(transitions 1 ~ 2b). This suggests that two "closed" states may exist, one complexed (2a) and one uncomplexed with cyclophilin (2b). Although these
cannot be distinguished by measurements of sucrose permeation, they might correspond to pore conformations with different "open" conductance.

Mitochondria and myocardial ischemia an insufficient availability of oxygen for mitochondrial

It has been proposed that inappropriate MTP opening during
postischemic reperfusion might represent a key event in the
ensuing tissue damage, especially in the heart [36]. In fact,
dissipation of the mitochondrial membrane potential (A'I'm)
caused by MTP opening could well result in massive and
abrupt release of the accumulated Ca 2+ into the cytosol
leading to cell death. Before discussing the potential role of
MTP in ischemia-reperfusion injury, we report here our
current understanding of mitochondrial function in the
ischemic heart. The alterations of myocardial function and
structure which are produced by ischemia indeed provide an
extensive, yet complex, framework of the relationships
between mitochondrial derangements and cell damage.
Myocardial tissue is typically aerobic and its metabolism
is closely dependent upon oxygen availability, as confirmed
by the abundance of mitochondria (30% of the total cell
volume) and myoglobin. The high energy requirements of
contraction are met almost exclusively by mitochondrial
oxidative phosphorylation. This in tum leads to the high
sensitivity of myocardial cells to oxygen deficiency.
In the heart, oxygen supply is generally reduced by a
coronary obstruction which results in a critical reduction of
flow (ischemia). In the affected region deprived of oxygen,
drastic changes of metabolism and lack of adequate wash-
out determine abnormal accumulation of ions and meta-
bolises.
At a cellular level, the onset of ischemia is determined by
oxidations. As a consequence of the reduced or absent
oxidative phosphorylation, intracellular creatine phosphate
is rapidly depleted with a concomitant rise in Pi, both factors
stimulating glycolysis and lactate production. The accumu-
lation of lactate and the hydrolysis of ATP decrease intra-
cellular pH.
If the oxygen restriction is maintained, mitochondria
themselves become targets of ischemic damage, decreasing
the possibility of recovery for both metabolism and function.
A number of mitochondrial alterations have been described
as a consequence of either ischemia or post-ischemic reper-
fusion. A decreased function of NADH dehydrogenase has
been reported as a consequence of ischemia [37, 38]. NMR
studies indicate that oxidative phosphorylation is still active
in isolated and perfused hearts under ischemic conditions
[39], but oxygen consumption does not correlate with
performance, consistent with mitochondrial uncoupling [40].
Also the activities of the Fll' ATPase [41 ] and ofthe adenine
nucleotide translocase [42] are reduced. Finally, mitochondria
isolated from ischemic hearts have been shown to produce
more oxyradicals than mitochondria harvested from norm-
oxic hearts [43]. Since oxyradicals are able to elicit specific
damage at the level of respiratory chain components [44] and
of membrane permeability [45], a vicious cycle of increasing
damage is likely to be generated. In addition, the release of
cytochrome c in the coronary effluent after a brief period of
ischemia [46] could contribute to the development of apop-
tosis [47].

From the above observations it is clear that mitochondria
are damaged by ischemia, but a residual, although impaired
function is present even in mitochondria extracted from
severely damaged tissues [48]. Furthermore, reoxygenation
is associated with partial recovery of ATP content also in
irreversibly damaged myocytes [49]. These observations fit
quite well the results obtained with phosphorus NMR show-
ing that a partial recovery of phosphocreatine and ATP
contents can occur even when contractile performance and
tissue viability are compromised. The time course and the
degree of mitochondrial alteration appear not to be univocally
related to the evolution of the structural and functional
derangements induced by ischemia. Rather, a coupled
mitochondrial respiration seems to be a necessary prerequisite
for the occurrence of cell disruption. Indeed, inhibition of the
respiratory chain or addition of uncouplers of oxidative
phosphorylation are able to limit the extent of enzyme release
in different models of myocardial damage such as post-
ischemic reperfusion [2], calcium paradox (reintroduction of
Ca2+ after Ca 2+-free perfusion) and oxygen paradox (read-
mission of oxygen after anoxia) [1]. These findings, obtained
on perfused hearts and isolated myocytes, suggest that
restoration of ATP production by mitochondrial oxidative
phosphorylation is essential for cell recovery, but can also
contribute to the processes causing cell necrosis.
Mitochondrial information from single anoxic myocytes
Although mitochondria playa central role in several forms
of cell injury, little is known about their function in intact
cells. Indeed, most of our knowledge on mitochondrial
function derives from studies performed on isolated mito-
chondria. Few attempts have been made to characterizeL1'J1m
modifications induced by a lack of oxygen in intact cells. In
earlier reports it was suggested that mitochondrial Ca 2+
release and adenine nucleotide sequestration should be
responsible for the maintenance ofL1'J1min anoxic hepatocytes
[50]. However, later experiments performed both in cell
suspension [5] and in single hepatocytes [4] clearly demon-
strated that adenine nucleotide translocase is not inhibited.
In fact L1'J1mis maintained in the presence of fructose unless
oligomycin is added [4, 5].
Dynamic studies at the single cell level were made possible
by the availability of a variety of optical probes, with the aid
of microspectrofluorometry. This methodology was also used
for the characterization of energy linked processes in cardiac
myocytes. For instance, as reported by R. G. Hansford in
another Section of this Issue, a specific technique was
developed to monitor Ca2+ changes within mitochondria of
single adult beating cardiac myocytes [51]. In the same
laboratory, other indices of mitochondrial function were
characterized while simultaneous monitoring cell shortening.
383
In addition, all these parameters were investigated under
conditions of anoxia and reoxygenation by means of a
specially developed chamber in which oxygen was excluded
by a laminar flow of ultra pure argon [52].
Freshly isolated adult cardiac myocytes show a charac-
teristic elongated morphology (rod shape), a clear sarcomere
pattern and are quiescent (i.e., they do not display spon-
taneous contractions or waves) in the presence of> 1 mM
extracellular [Ca2+]. Rhythmic contractions can be evoked by
electrical stimulation. Under deenergizing conditions, rod
shaped myocytes can either change into square forms which
maintain the sarcomere pattern (rigor state or contracture) or
hypercontract into rounded dysfunctional forms in which the
typical sarcomeric striature is no longer distinguishable
(hypercontracture). In some ofthese hypercontracted round-
ed cells the sarcolemma is still intact, as judged from trypan
blue exclusion.
After a variable period of time, rod-shaped myocytes
exposed to glucose-free anoxia (P02 < 1 torr) rapidly assume
the square aspect ([52,53], reviewed in [54]). All cells exhibit
this transition, which is probably concomitant with ATP
depletion [55]. Recently, the relationship betweenATP levels
and rigor has been directly confirmed in single myocytes
injected with luciferase [56]. Ifanoxia is prolonged afterrigor
development, the morphology does not show any further
change. A similar pattern of changes, namely a rapid trans-
ition to rigor followed by no further morphological modi-
fication, is produced also by inhibition of the respiratory chain
or by addition of un couplers.
If reoxygenation takes place within 5 min after the onset
of rigor, all cells partially relengthen, retaining a clear
sarcomere pattern and the ability to twitch in response to
electrical stimulation. When cells are kept anoxic for longer
periods after rigor contracture, at reoxygenation recovery is
less frequent and part of the cells hypercontract into rounded
dysfunctional forms [57].
A relevant metabolic feature of the isolated myocyte is its
ability to tolerate a complete inhibition of the respiratory
chain if glycolytic rates are maintained by either exogenous
glucose or endogenous glycogen. The addition of rotenone
or cyanide does not affect shortening of electrically stimu-
lated cells, ionic homeostasis and L1'J1 m. Under these cond-
itions a further addition of iodoacetate rapidly induces cell
contracture with a simultaneous fall ofATP content andL1'J1m
[58]. The maintenance of L1'J1m in the absence of electron flow
through the respiratory chain can be explained only by the
utilization of the glycolytic ally produced ATP by the FII
ATPase operating in 'reverse'.
In this experimental model the relationships between cell
function and L1'J1m were investigated by using JC-l, a carbo-
cyanine derivative [58]. This fluorescent probe is characterized
by two emission peaks (539 and 597 nm with excitation at 490
nm) corresponding to monomer and aggregate forms of the
384
dye, respectively [59]. The responses of monomers and
aggregates show different ranges of sensitivity to the Ll\jfm
changes [58, 60]. In particular, the monomer emission is
modified by values of Ll\jfm below 140 mY, which hardly
changes the aggregate emission. Thus, JC-I represents a unique
'double sensor' which can provide semi-quantitative inform-
ation in both the low and high membrane potential range.
In order to gain further information about the bioenergetics
of single living anoxic myocytes, the Ll\jfm measurements
were compared to those obtained with another indicator of
energy metabolism, namely Mg2+. The intracellular con-
centration of this cation ([Mg2+]) was studied in mag-indo-l
loaded myocytes [57]. Inside the cells most Mg2+ is bound
to ATP, so that a fall in ATP content is reflected by a pro-
portional increase of [Mg2+]j.
Immediately at the onset of anoxia, a gradual decline of
the emission at 590 nm without changes of the shorter
wavelength emission took place, indicating that only a limited
reduction of Ll\jfm occurs before the asystole. Then, after the
failure of contraction, ATP content fell, as shown by a gradual
increase in [Mg2+l, which reached a plateau at the onset of
rigor [57]. Concomitantly, the rapid increase ofthe emission
at 530 nm indicated the collapse of Ll\jfm'
It has been demonstrated that mitochondrial ATPase
constitutes a relevant site for ATP hydrolysis in the anoxic
myocardium [3]. The respiratory chain inhibition consequent
to the lack of oxygen forces mitochondria to utilize the
glycolytically produced ATP to maintain their membrane
potential, albeit apparently at lower values. Consequently, the
complete collapse of Ll\jfmwould follow the glycolytic failure
and the ATP depletion following glycogen exhaustion. The
fact that an active mitochondrial ATPase is necessary for the
maintenance of Ll\jfm is demonstrated by the rapid and
complete collapse ofLl\jfmoccurring in oligomycin-pretreated
myocytes [58]. This mechanism does not seem to pertain
exclusively to cardiac myocytes. The acceleration of anoxia-
induced mitochondrial depolarization after oligomycin
administration has been shown also in carotid body type I
cells [61].
Cell length, Ll\jfmand Mg2+ are not modified by the prolong-
ation of anoxia following rigor. Conversely, both cytosolic
and mitochondrial [Ca2+] ([Ca 2+]c and [Ca2+]m) rise progress-
ively only after the onset of ATP-depletion contracture [62].
Upon reoxygenation, Ll\jfm is partially restored indepen-
dently of the recovery of mechanical function. Indeed, even
in the hypercontracted myocytes a further addition ofFCCP
is followed by sudden 'anti parallel , changes in the two
emission wavelengths of JC-l, i.e., enhancement of the
monomer and decrease of the aggregate fluorescence [58].
Irrespective of the functional outcome, reoxygenation is
always associated with a decrease in [Mg2+l. Since [Mg2+]j
increases in response to subsequent FCCP addition, oxidative
phosphorylation is likely restored also in irreversibly dam-
aged cells. As far as Ca 2+homeostasis is concerned, reoxygen-
ation is followed by a rapid fall of [Ca2+t towards pre anoxic
levels, whereas [Ca2+]m shows a modest and transient decrease
which is followed by a secondary rise. Interestingly, it appears
that anoxic levels of [Ca2+]m are inversely related to the
probability of cell recovery upon reoxygenation, confirming
the relationship between intramitochondrial Ca2+ levels and
worsening of the anoxic damage [63]. These experimental
observations are summarized in Fig. 2.
Anoxia and mitochondrial calcium overload
The association between myocardial damage and intracellular
calcium overload has been related to mitochondrial dys-
function since the pioneering observations of Shen and
Jennings [63]. Oxidative phosphorylation is likely reduced
by a massive mitochondrial uptake of Ca2+ since both pro-
cesses utilize the same energy source, namely Ll\jfm' The
reduced ATP availability is followed by a failure of the
mechanisms which extrude Ca2+ out of the sarcoplasm
creating a vicious cycle of Ca2+ overload and mitochondrial
impairment. In addition, the rise in [Ca 2+l could promote
opening of the cyclosporin-sensitive mitochondrial perm-
eability transition pore, leading to a sudden Ll\jfmdissipation.
However, this mechanism is not supported by the reported
studies on single myocytes. In fact, during reoxygenation, and
irrespective of functional or morphological recovery, Ll\jfm
and ATP contents were partially restored, with no major
changes in [Ca2+]m [62]. MTP opening might rather occur
during anoxia. It has been shown that the membrane perm-
eability transition correlates with a decrease of Ll\jfm' sug-
gesting a voltage-gated mechanism [7]. Thus, it could be
hypothesized that a drastic reduction in glycolytic ATP
production curtails the possibility to maintain Ll\jfm' whose
decline below a critical threshold would then promote MTP
opening. Such a mechanism could underlie the rapid Ll\jfm
collapse and [Ca2+]m rise which occur concomitantly with
rigor development. On the other hand, conditions which
determine cytosolic Ca2+ overload as a primary event could
lead to an abnormal elevation of [Ca2+]m followed by MTP
opening and Ll\jfm collapse. Indeed, it has been recently
demonstrated that Ca2+ readmission to myocytes previously
bathed in the absence ofCa 2+determines a biphasic response
of Ll\jfm' a transient increase being followed by a rapid and
persistent collapse [64].
Mitochondria, ATP and myofibrils: the apparent
mitochondrial paradox
Not only is mitochondrial function retained even when cell
morphology and function are irreversibly impaired, but this
 385

tglycolysis lack of mitochondrial

AlP production

+
ATP production

mitochondrial
cellular AlP
glycogen ATP utilization

+
maintenance
exhaustion

partial

~ d,m maintenance

t glycolysis lack of
AlP production
MTP
HAlP
6'm collapse
contracture

reoxygenation

partial
~ 61Vm recovery

+
mitochondrial
A TP production

t cytosolic ATP t cytosolic ATP

+ +
high [Ci+Jc low [ca2 J c
(~500 nM) 200 nM)
fat
1 min)
hyperconlracture
,..covery of myocyte
morpholollY and functlon
Irreversible loss
of structure and function

Fig. 2. Sequence of events observed in isolated cardiac myocytes leading to either recovery or irreversible damage during anoxia and reoxygenation. In the
first phase of anoxia no major changes apparently occur, due to the maintenance of cell morphology, function and metabolism by anaerobic glycolysis.
However, during anoxia mitochondria switch from ATP producers into A TP consumers; when glycolytic substrates are no longer available, they contribute
to a rapid fall of ATP content, resulting in myocyte contracture and L1'1fmcollapse. The prolongation of anoxia after contracture is mostly characterized by a
parallel rise in both [Ca2+lc and [Ca2+]m. Ifreoxygenation takes place before a relevant increase in intracellular [Ca'+];, L1'1fm recovery allows a relatively slow
increase of the A TP content, which in turn is necessary for myocyte relengthening and eventually for the recovery of mechanical activity. On the other hand,
in the presence of an increased [Ca'+];, the rise in ATP concentration from <10-4 M to I0-4-10-1M accelerates cross bridge cycling, resulting in rapid and
irreversible transformation of myocytes into rounded dysfunctional forms.
386
function is necessary to obtain the hypercontracture. Indeed,
FCCP-treated myocytes undergo rigor contracture and never
develop hypercontracture. In addition, if the respiratory chain
is inhibited with rotenone during anoxia and reoxygenation,
cells retain indefinitely the rigor state they would maintain
if anoxia were not followed by reoxygenation (F. Di Lisa and
R.G. Hansford, unpublished observations). Thus, in cardiac
myocytes the restoration of ATP production by mitochondrial
oxidative phosphorylation is essential for cell recovery, but
can also cause cell death. This apparent paradox could be
related to the different ATP requirements of contraction and
relaxation at the myofilament level.
The deleterious mixture of small amounts of ATP and of
an even modest increase in [Ca2+l, which are present in those
cells committed to hypercontracture at the beginning of
reoxygenation, is likely to alter their cross-bridge cycling.
When cardiac myocytes are permeabilized with digitonin, the
maintenance of the rod-shape morphology is correlated with
the ATP concentrations in the bathing solution [65]. When
permeabilized myocytes are exposed to solutions containing
no ATP, they shorten into square form with retained sarco-
mere pattern. These cells relenghten as soon as [ATP] in the
surrounding medium is increased to millimolar levels [66].
On the contrary, cells exposed to low but non-zero levels of
ATP (4-20 )..lM) undergo progressive, irreversible hyper-
contracture. Since these changes occur in the presence of
EGTA, contracture and hypercontracture are probably caused
by Ca 2+-independent cross-bridge cycling occurring in the
presence of different degrees of myofilament stiffness [66].
However, the ATP sensitivity of these changes is modulated
by Ca 2+, which increases the ATP requirements for the
maintenance or the recovery of the elongated morphology.
In the presence of high levels of ATP, hypercontracture can
be considered the result ofCa 2+ overload [65]. These obser-
vations provide a rationale for the paradoxical association of
mitochondrial function with the opposite events of recovery
of function and irreversible damage occurring during post-
ischemic reperfusion. Three major possibilities can be
envisaged: (i) under conditions of mitochondrial inhibition
(cyanide) or uncoupling, ATP is virtually absent and cells or
tissues are kept 'frozen' in a rigor state; (ii) when reperfusion
is accompanied by adequate recovery of mitochondrial
function, the restoration ofATP content allows cell elongation
or tissue relaxation, and [Ca2+]j rapidly returns to control
levels; (iii) when mitochondrial function is impaired, low
levels of ATP impair Ca 2+ homeostasis and enhance the
formation of rigor bonds which, in the absence of sufficient
energy for the relaxation process, produce the hypercon-
tracture.
These concepts are supported by the ability of a temporary
contractile blockade to inhibit reoxygenation-induced hyper-
contracture. Indeed, the addition of 2,3 butanedione mono-
xime (BDM) just prior to reoxygenation prevented the
hypercontracture so that 15 min later the cell could recover
the elongated morphology upon BDM washout [67]. Since
hypercontracture is similarly blunted by a decrease of
extracellular Ca2+, myocyte recovery cannot occur in the
presence of an increased [Ca2+]j. Indeed, during reoxygen-
ation levels of [Ca 2+]j which are tolerated by normoxic
myocytes [68] determine a destructive activation of cross-
bridge cycling owing to the gradual increase ofATP content.
Limits of studies on isolated myocytes
At the single cell level hypercontracture is the paradigm of
irreversible damage and is considered equivalent to the abrupt
rise in diastolic pressure in the absence of contractile activity
which is observed in the intact heart when reperfusion takes
place after a prolonged ischemia. However, a major dif-
ference emerges when the myocyte plasma membrane
permeability is compared under the two conditions. In the
intact heart irreversible damage is characterized by the loss
of cell membrane integrity, easily demonstrated by the release
of intracellular proteins in the coronary effluent. Conversely,
hypercontracted myocytes often retain the integrity of
sarcolemma, as demonstrated by the lack of release of small
molecules such as intracellularlytrapped fluorescent probes.
Nevertheless, shortly after reoxygenation in the vast majority
of hypercontracted myocytes the plasma membrane becomes
eventually permeable to extracellular large molecules owing
to the rupture of enlarging blebs. The difference between
single cells and intact hearts is likely explained by the lack
of cell attachments. The same morphological distortion which
determines the hypercontracture in isolated myocytes could
probably cause the rupture of the sarcolemma in the intact
heart by tearing off adjacent myoctes [69]. Further problems
are the severity of ischemia, which can be approached [70]
but not fully reproduced in cellular studies, which also fail
to mimick other relevant aspects of tissue ischemia such as
metabolite accumulation and extracellular ion retention.
Finally, some advantages of cellular studies (e.g., lack of both
neural modulation and humoral influences by non-myocyte
cell types) become drawbacks when it comes to extending
the value of the observations to the heart in situ. Thus,
although the studies performed on single cells allow to define
the relations between cell function and biochemical events,
the extrapolation of the resulting mechanisms to the intact
organ is never too cautious.
A relevant example is provided by the possible relations
betweeen [Ca2+]j and cell damage. As mentioned above, in
the intact heart reperfusion after a prolonged ischemic period
is associated with the loss of cell membrane integrity. This
phenomenon occurs in the first minutes of reperfusion and it
is quite difficult to establish if it is the cause or rather the
consequence of intracellular Ca 2+ overload. The results

obtained in single anoxic myocytes demonstrate that a
massive Ca2+overload is not a necessary prerequisite for the
onset of the irreversible damage. However, it must be pointed
out that these experiments were performed by exposing the
cells to almost complete anoxia (P02 < 0.003 kPa). Ifanoxia
is replaced with hypoxia (P02 > 0.1 kPa), in half of the cells
the transition to the square form is prevented, and hyper-
contracture is observed after a significant increase in [Ca2+t
[71]. In general, any conditions associated with intracellular
Ca2+ overload eventually result in hypercontracture. Under
these circumstances, the increase in [Ca2+t precedes i1",m
collapse [72]. It is tempting to speculate that in the heart in
situ a residual tissue oxygenation provided by collateral blood
flow could result in early alterations of [Ca2+]i' which could
precipitate an already impaired mitochondrial function. In
this case, opening of the permeability transition pore could
playa crucial role.
MTP role in situ: a critical evaluation
The feasibility of MTP opening cannot be easily predicted
by considering the changes which characterize the ischemic
myocardium. In fact, a complete picture of cytosolic and
mitochondrial components is complicated by the coexistence
of factors that can either promote or reduce the probability
of pore opening. The following list is meant to summarize
and critically evaluate the changes of known pore effectors
during ischemia and reperfusion (anoxia - reoxygenation in
single cells):
A. Inducers
(i) [Ca 2+] m increase occurs during prolonged ischemia as a
result of the equilibration with [Ca2+L in the absence of a
significant mitochondrial membrane potential. At the onset
of reperfusion mitochondria develop a membrane potential
and are exposed to increased [Ca 2+L. Since Ca 2+ uptake
utilizes the same energy source (i1jlH), it competes with ATP
formation. A severe calcium overload might alter both the
function and the structure of mitochondria, and contribute to
MTP opening.
(ii) Mitochondrial depolarization. Due to the lack of oxygen,
the membrane potential is obviously reduced or collapsed
during ischemia. i1",m is recovered during reoxygenation [58],
but the degree and the rate of this recovery have not been
determined yet. A quantitative measurement of i1"'m in situ
is hampered by the coexistence of a membrane potential
difference across the plasma membrane, which makes the
classic calibration by means of valinomycin-K+ diffusion
potential unreliable. Considering the possible relationship
with MTP, it has to be pointed out that, even at the single cell
387
level, i1",m determination is a gross average of the behaviour
of about 3000 mitochondria. Thus, depolarization could be
a reliable indicator of permeability transitions only ifthe vast
majority of mitochondria depolarized synchronously, which
appears extremely unlikely.
(iii) Inorganic phosphate (Pi). 31p NMR measurements have
unambiguously documented an increase in cytosolic Pi which
is inversely proportional to the reduction of creatine phos-
phate content [73]. Unfortunately, no such data are available
for mitochondrial Pi. Conceivably, during ischemia, due to
the reduced or absent i1iIH, Pi equilibrates within the
intracellular compartments, so that upon reperfusion matrix
[Pi] should be higher than in normoxic controls.
(iv) Long chain acyl-CoA content increases 2-3 fold during
ischemia due to the reduction ofp-oxidation flux [74]. This
accumulation occurs within the matrix space, since >90% of
cellular CoA is compartmentalized inside mitochondria.
(v) Oxygen radicals. As already mentioned, their formation
is increased, particularly at the moment of reoxygenation.
Monovalent reduction of oxygen by the mitochondrial
respiratory chain produces superoxide anions, which then
yield hydrogen peroxide through the superoxide dismutase
reaction. The latter species oxidizes mitochondrial gluta-
thione, and is therefore a likely candidate as an endogenous
mediator which can affect MTP open-closed transitions at the
S-site, and at the P-site because of the redox link through
glutathione reductase and nicotinamide adenine dinucleotide
transhydrogenase [75].
B. Inhibitors
(i) intracellular pH decrease. Acidosis is one ofthe hallmarks
of ischemic tissues and the characteristics of pH decrease
have been largely documented by 31p NMR (for a review see
[73]). Also in this case, as for Pi, there is no information
concerning matrix pH in situ. It is likely that during ischemia,
upon collapse ofi1pH, acidotic conditions exist in the matrix.
Upon reperfusion the recovery of i1"'m together with meta-
bolite washout should result in matrix alkalinization. Such
changes in matrix pH could contribute to the so called pH
paradox [76]. This term describes the ability of an increase
in intracellular pH to damage myocytes which were pre-
viously exposed to an acidic buffer. Under such conditions
the coupled activities of Na+/H+ and Na+ICa 2+ exchangers
result in [Ca2+t rise which is likely to promote MTP opening.
(ii) ADP. Free cytoplasmic [ADP] increases during ischemia
from control values of-30 /-lM to 0.2-0.5 mM [73].Afurther
increase is limited by the highly active adenyl ate kinase. On
the basis of considerations analogous to those presented for
388
Pi and pH, comparableADP accumulation should take place
also within the matrix space. (It has to be pointed out that in
the presence of a high ~jl H, even in normoxic myocytes free
[ADP] and [ADP]/[ ATP] inside mitochondria are higher than
in the cytosol due to the activity of adenylate translocase).
(iii) Mff+. NMR results obtained in isolated hearts [77] and
data collected by means of fluorescent indicators in single
cells [57] provide evidence for a 2-3 fold increase of
intracellular Mg2+ during ischemia (from 0.8 to 2.l mM, [57,
77]), a process which is most likely related toATP hydrolysis.
According to the recovery ofATP content, intracellular Mg2+
returns to normoxic values during reperfusion.
A role for MTP in either physiological or pathological
processes depends obviously on the equilibrium between
these components. If a membrane transition occurrs during
reperfusion, two different modes of operation can be hypo-
thesized:
(i) synchronous MTP opening in the majority of
mitochondria. Such a scenario, that could be brought about
by a sudden increase in [Ca 2+J;, would result in ~'Jfm
collapse and massive [Ca2+]m release. At a single cell level
it should be possible to demonstrate a rapid rise in [Ca2+]c'
enhanced by the failure of Ca2+ pumps occurring in the
absence of adequate ATP production. This hypothetical
behaviour is in sharp contrast with the experimental
findings showing a sustained rise in [Ca2+]m associated with
a rapid decrease of [Ca2+lc to normoxic values [62].
(ii) asynchronous mode. A fraction of mitochondria could
be exposed to conditions compatible with MTP opening.
Ca 2+ released through these pores would promote the
membrane transition in adjacent mitochondria, resulting
in a propagating wave of mitochondrial impairment as is
often observed in vitro. Although the latter hypothesis
remains a possibility, its study in situ is hampered by the
lack of adequate methodologies allowing a discrimination
between these two modes of pore opening.
A final comment is in order concerning the cellular effects
of cyclosporinA, since most investigations on the role ofthe
pore in ischemic cell death rely on the effects of this drug.
The results obtained with this approach can be safely inter-
preted only if cyclosporinA reaches mitochondria in an active
form, and at concentrations adequate to block the pore. This
problem relates both to metabolism of cyclosporin A, which
is known to yield pore-inactive products like N-Desmethyl4
cyclosporin A [15], and to the existence of a number of
cellular cyclophilins. Of particular concern is the abundant
cytosolic isoform ofthe enzyme (cyclophilinA) which could
sequester a large fraction of added cyclosporin A.
Another problem in interpreting the effects of cyclosporin
A at the cellular level is posed by the inhibition of other
cyclophilin-dependent signalling pathway(s), including
calcineurin-dependent signal transduction to the nucleus.
Since the effects of cyclosporinA on the pore do not involve
calcineurin, in cellular studies aimed at investigating the pore
it is advisable to use N-MethyIVal-4 cyclosporin, which
retains the pore inhibitory properties of cyclosporin A but
does not inhibit calcineurin [27, 32]. Even with this deri-
vative, however, a major concern is posed by the endoplasmic
reticulum isoform of the enzyme (cyclophilin B), which
regulates Ca 2+-dependent signalling through its specific
interactions with CAML (an integral membrane protein of the
endoplasmic reticulum) in a cyclosporinA-sensitive fashion,
in lymphocytes at least [78].
Conclusions
In summary, although occurrence of the permeability tran-
sition in ischemic damage is supported by a variety of data,
the evidence in favor of its causative role in the ischemia-
reperfusion syndrome is still insufficient. This largely
depends on the intrinsic problems related to extrapolation of
data obtained in vitro to those prevailing in vivo. We feel that
future investigations should be aimed at answering two major
questions: Is pore opening a necessary event in cell death?
Is it a primary switch marking a point or no return, or does it
rather follow the death decision as an effector mechanism?
An answer to these questions appears essentia~ for a better
understanding of the pathways leading to cell death. Despite
these open problems, we feel that mitochondria are major
players in the ischemia-reperfusion syndrome, and that the
practical implications of a 'mitochondrial therapy' of this
high-prevalence disease are or great perspective value.
Acknowledgements
This review article is dedicated to the memory of Howard S.
Silverman. Research in the Authors , laboratories is supported
by the CNR, MURST, and Telethon-Italy (Grant n. 847 to
PB).
References
I. Ganote CE, McGarr 1, Liu SY, Kaltenbach IP: Oxygen-induced enzyme
release. Assessment of mitochondrial function in anoxic myocardial
injury and effects of the mitochondrial uncoupling agent 2,4-
dinitrophenol (DNP). 1 Mol Cell Cardiol12: 387-408, 1980
2. Elz JS, Nayler WG: Calcium gain during postischemic reperfusion.
The effect of2,4-dinitrophenol. Am 1 Pathol131: 137-145,1988

3. Rouslin W, Erickson JL, Solaro RJ: Effects of oligomycin and
acidosis on rates of ATP depletion in ischemic heart muscle. Am J
Physiol250: H503-508, 1986
4. Imberti R, Nieminen AL, Herman B, Lemasters JJ: Mitochondrial
and glycolytic dysfunction in lethal injury to hepatocytes by t-butyl-
hydroperoxide: Protection by fructose, cyclosporin A and trifluo-
perazine. J Pharmacol Exp Ther 265: 392-400, 1993
5. Snyder JW, Pastorino JG, Thomas AP, Hoek JB, Farber JL: ATP
synthase activity is required for fructose to protect cultured
hepatocytes from the toxicity of cyanide. Am J Physiol 264: C709-
14,1993
6. Zoratti M, Szabo I: The mitochondrial permeability transition.
BiochimBiophysActa 1241: 139-176,1995
7. Bernardi P: Modulation of the mitochondrial cyclosporinA-sensitive
permeability transition pore by the proton electrochemical gradient.
Evidence that the pore can be opened by membrane depolarization.
J Bioi Chern 267: 8834-8839, 1992
8. Szabo I, Bernardi P, Zoratti M: Modulation of the mitochondrial
megachannel by divalent cations and protons. J Bioi Chern 267:
2940-2946, 1992
9. Szabo I, Zoratti M: The mitochondrial megachannel is the
permeability transition pore. J Bioenerg Biomembr 24: 111-117, 1992
10. Halestrap AP, Davidson AM: Inhibition of Ca 2+-induced large-
amplitude swelling of liver and heart mitochondria by cyclosporin
is probably caused by the inhibitor binding to mitochondrial-matrix
peptidyl-prolyl cis-trans isomerase and preventing it interacting with
the adenine nucleotide translocase. Biochem J 268: 153-160, 1990
II. Brustovetsky N, Klingenberg M: MitochondrialADPIATP carrier can
be reversibly converted into a large channel by Ca'+. Biochemistry
35: 8483-8488, 1996
12. Szabo I, Zoratti M: The mitochondrial permeability transition pore
may comprise VDAC molecules. I. Binary structure and voltage
dependence of the pore. FEBS Lett 330: 201-205,1993
13. Beutner G, Ruck A, Riede B, Welte W, Brdiczka D: Complexes
between kineses, mitochondrial porin and adenyl ate translocator in
rat brain resemble the permeability transition pore. FEBS Lett 396:
189-195,1996
14. Igvabvoa U, Pfeiffer DR: EGTA inhibits reverse uniport-dependent
Ca2+ release from uncoupled mitochondria. Possible regulation of
the Ca2+ uniporter by a Ca'+ binding site on the cytoplasmic side of
the inner membrane. J Bioi Chern 263: 1405-1412,1988
15. Bernardi P, Broekemeier KM, Pfeiffer DR: Recent progress on
regulation of the mitochondrial permeability transition pore; a
cyclosporin-sensitive pore in the inner mitochondrial membrane. J
Bioenerg Biomembr 26: 509-517, 1994
16. Zoratti M, Szabo I: Electrophysiology of the inner mitochondrial
membrane. J Bioenerg Biomembr 26: 543-553, 1994
17. Broekemeier KM, Pfeiffer DR: Inhibition of the mitochondrial
permeability transition by cyclosporin A during long time frame
experiments: Relationship between pore opening and the activity of
mitochondrial phospholipases. Biochemistry 34: 16440-16449, 1995
18. Lapidus RG, Sokolove PM: The mitochondrial permeability
transition. Interactions of spermine, ADP, and inorganic phosphate.
J Bioi Chern 269: 18931-18936, 1994
19. Rigobello MP, Barzon E, Marin 0, Bindoli A: Effect of poly cation
peptides on mitochondrial permeability transition. Biochem Biophys
Res Commun 217: 144-149,1995
20. Bernardi P, Veronese P, Petronilli V: Modulation ofthe mitochondrial
cyclosporinA-sensitive permeability transition pore. I. Evidence for
two separate Me'+ binding sites with opposing effects on the pore
open probability. J Bioi Chern 268: 1005-1010, 1993
21. Crompton M, Costi A: Kinetic evidence for a heart mitochondrial
pore activated by Ca2+, inorganic phosphate and oxidative stress. A
389
potential mechanism for mitochondrial dysfunction during cellular
Ca'+ overload. Eur J Biochem 178: 489-501,1988
22. Petronilli V, Cola C, Bernardi P: Modulation of the mitochondrial
cyclosporinA sensitive permeability transition pore. II. The minimal
requirements for pore induction underscore a key role for trans-
membrane electrical potential, matrix pH, and matrix Ca2+. J Bioi
Chern 268: 1011-1016, 1993
23. Rottenberg H, Marbach M: Regulation of Ca2+ transport in brain
mitochondria. I. The mechanism of spermine enhancement of Ca'+
uptake and retention. Biochim Biophys Acta 1016: 77-86, 1990
24. Rottenberg H, Marbach M: Regulation of Ca 2+transport in brain mito-
chondria. II. The mechanism of the adenine nucleotides enhancement
of Ca'+ uptake and retention. Biochim Biophys Acta 1016: 87-98,
1990
25. Petronilli V, Costantini P, Scorrano L, Colonna R, Passamonti S,
Bernardi P: The voltage sensor of the mitochondrial permeability
transition pore is tuned by the oxidation-reduction state of vicinal
thiols. Increase of the gating potential by oxidants and its reversal
by reducing agents. J Bioi Chern 269: 16638-16642,1994
26. Chernyak BV, Bernardi P: The mitochondrial permeability transition
pore is modulated by oxidative agents through both pyridine nucleo-
tides and glutathione at two separate sites. Eur J Biochem 238: 623-
630, 1996
27. Petronilli V, Nicolli A, Costantini P, Colonna R, Bernardi P:
Regulation of the permeability transition pore, a voltage-dependent
mitochondrial channel inhibited by cyclosporin A. Biochim Biophys
Acta 1187: 255-259, 1994
28. Costantini P, Chernyak BV, Petronilli V, Bernardi P: Selective
inhibition of the mitochondrial permeability transition pore at the
oxidation-reduction sensitive dithiol by monobromobimane. FEBS
Lett 362: 239-242, 1995
29. Costantini P, Chernyak BV, Petronilli V, Bernardi P: Modulation of
the mitochondrial permeability transition pore by pyridine
nucleotides and dithiol oxidation at two separate sites. J Bioi Chern
271: 6746-6751,1996
30. Connern CP, Halestrap AP: Purification and N-terminal sequencing
of peptidylprolyl cis-trans-isomerase from rat liver mitochondrial
matrix reveals the existence of a distinct mitochondrial cyclophilin.
Biochem J 284: 381-385, 1992
31. Connern CP, HalestrapAP: Recruitment of mitochondrial cyclophilin
to the mitochondrial inner membrane under conditions of oxidative
stress that enhance the opening of a calcium-sensitive non-specific
channel. Biochem J 302: 321-324, 1994
32. NicolliA, Basso E, Petronilli V, Wenger RM, Bernardi P: Interactions
of cyclophilin with the mitochondrial inner membrane and regulation
of the permeability transition pore, a cyclosporinA-sensitive channel.
J Bioi Chern 271: 2185-2192,1996
33. Hunter DR, Haworth RA: The Ca2+-induced membrane transition in
mitochondria. I. The protective mechanisms. Arch Biochem Biophys
195: 453-459,1979
34. Bernardi P, Vassanelli S, Veronese P, Colonna R, Szabo I, Zoratti M:
Modulation of the mitochondrial permeability transition pore. Effect
of protons and divalent cations. J Bioi Chern 267: 2934-2939,1992
35. NicolliA, Petronilli V, Bernardi P: Modulation of the mitochondrial
cyclosporin A-sensitive permeability transition pore by matrix pH.
Evidence that the pore open-closed probability is regulated by
reversible histidine protonation. Biochemistry 32: 4461-4465,1993
36. Crompton M, Costi A: A heart mitochondrial Ca 2+-dependent pore
of possible relevance to re-perfusion-induced injury. Evidence that
ADP facilitates pore interconversion between the closed and open
states. Biochem J 266: 33-39, 1990
37. Rouslin W, Ranganathan S: Impaired function of mitochondrial
electron transfer complex I in canine myocardial ischemia: loss of
390
flavin mononucleotide. J Mol Cell Cardiol 15: 537-542, 1983
38. Hardy L, Clark JB, Darley Usmar VM, Smith DR, Stone D:
Reoxygenation-dependent decrease in mitochondrial NADH: CoQ
reductase (Complex I) activity in the hypoxic/reoxygenated rat heart.
Biochem J 274: 133-137, 1991
39. Sako KY, Kingsley Hickman PB, From AH, Foker IE, Ugurbil K:
ATP synthesis kinetics and mitochondrial function in the postischemic
myocardium as studied by 31P NMR. J Bioi Chern 263: 10600--
10607, 1988
40. Neubauer S, Hamman BL, Perry SB, Bittl JA, Ingwall JS: Velocity
of the creatine kinase reaction decreases in postischemic
myocardium: a 3lP-NMR magnetization transfer study of the isolated
ferret heart. Circ Res 63: 1-15, 1988
41. Rouslin W: Protonic inhibition of the mitochondrial oligomycin-
sensitive adenosine 5'-triphosphatase in ischemic and autolyzing
cardiac muscle. Possible mechanism for the mitigation of ATP
hydrolysis under nonenergizing conditions. J Bioi Chern 258: 9657-
9661,1983
42. Hutter JF, Alves C, Soboll S: Effects of hypoxia and fatty acids on
the distribution of metabolites in rat heart. Biochim Biophys Acta
1016: 244--252,1990
43. Das DK, George A, Liu XK, Rao PS: Detection of hydroxyl radical
in the mitochondria of ischemic-reperfused myocardium by trapping
with salicylate. Biochem Biophys Res Commun 165: 1004--1009,
1989
44. ZhangY, Marcillat 0, Giulivi C, Ernster L, Davies KJ: The oxidative
inactivation of mitochondrial electron transport chain components
and ATPase. J Bioi Chern 265: 16330--16336, 1990
45. Castilho RF, Kowaltowski AJ, Meinicke AR, Bechara EJH, Vercesi
AK: Permeabilization of the inner mitochondrial membrane by Ca'+
ions is stimulated by t-butyl hydroperoxide and mediated by reactive
oxygen species generated by mitochondria. Free Rad Bioi Med 18:
479--486, 1995
46. Naro F, Fazini A, Grappone C, Citro G, Dini G, Giotti A, Malatesta
F, Franconi F, Brunori M: Release of cytochromes from hypoxic and
reoxygenated guinea pig heart. Cardioscience 4: 177-184, 1993
47. Liu XS, Kim CN, Yang J, Jemmerson R, Wang XD: Induction of
apoptotic program in cell free extracts: requirement for dATP and
cytochrome c. Cell 86: 147-157, 1996
48. Ferrari R, Di Lisa F, Raddino R, Visioli 0: The effects of ruthenium
red on mitochondrial function during post-ischaemic reperfusion. J
Mol Cell Cardiol14: 737-740, 1982
49. Siegmund BA, KoopT, KlietzT, Schwartz P, Piper HM: Sarcolemmal
integrity and metabolic competence of cardiomyocytes under anoxia-
reoxygenation. Am J Physiol 258: H285-H291, 1990
50. Aw TY, Andersson BS, Jones DP: Mitochondrial transmembrane ion
distribution during anoxia. Am J Physiol252: C356--C361, 1987
51. Miyata H, Silverman HS, Sollott SJ, Lakatta EG, Stern MD, Hansford
RG: Measurement of mitochondrial free Ca'+ concentration in living
single rat cardiac myocytes. Am J Physiol 261: H1123-HI134,
1991
52. Stern MD, Silverman HD, Houser SR, Josephson RA, Capogrossi
MC, Nichols CG, Lederer WJ, Lakatta EG: Anoxic contractile failure
in rat heart myocytes is caused by failure of intracellular calcium
release due to alteration of the action potential. Proc Nat! Acad Sci
USA 85: 6954--4958, 1988
53. Hohl CM, Ansel A, Altschuld RA, Brierley GP: Contracture of
isolated rat heart cells on anaerobic to aerobic transition. Am J Physiol
242: HI022-HI030, 1982
54. Silverman HS, Stern MD: ionic basis of ischaemic cardiac injury:
insights from cellular studies. Cardiovasc Res 28: 581-597, 1994
55. Haworth RA, Hunter DG, BerkoffHA: Contracture in isolated adult
rat heart cells: role ofCa'+, ATP and compartmentation. Circ Res 49:
1119-1128,1981
56. Bowers KC, AllshireA, Cobbold PH: Bioluminescent measurement
in single cardiomyocytes of sudden cytosolic ATP depletion
coincident with rigor. J Mol Cell Cardiol 24: 213-218, 1992
57. Silvennan HS, Di Lisa F, Hui RC, Miyata H, Sollott SJ, Hansford
RG, Lakatta EG, Stern MD: Regulation of intracellular free Mg'+
and contraction in single adult mammalian cardiac myocytes. Am J
Physiol266: C222-233, 1994
58. Di Lisa F, Blank PS, Colonna R, Gambassi G, Silverman HS, Stern
MD, Hansford RG: Mitochondrial membrane potential in single
living adult rat cardiac myocytes exposed to anoxia or metabolic
inhibition. J Physiol (Lond) 486: 113, 1995
59. Reers M, Smith TW, Chen LB: J-aggregate formation of a
carbocyanine as a quantitative fluorescent indicator of membrane
potential. Biochemistry 30: 4480--4486, 1991
60. Reers M, Smiley ST, Mottola Hartshorn C, Chen A, Lin M, Chen
LB: Mitochondrial membrane potential monitored by JC-I dye.
Methods Enzymol260: 406--417,1995
61. Duchen MR, Biscoe TJ: Mitochondrial function in type I cells isolated
from rabbit arterial chemoreceptors. J Physiol (Lond) 450: 13-31,
1992
62. Miyata H, Lakatta EG, Stern MD, Silverman HS: Relation of
mitochondrial and cytosolic free calcium to cardiac myocyte recovery
after exposure to anoxia. Circ Res 71: 605--613,1992
63. ShenAC, Jennings RB: Myocardial calcium and magnesium in acute
ischemic injury. Am J Pathol 67: 417--421,1972
64. Minezaki KK, Suleiman MS, Chapman RA: Changes in
mitochondrial function induced in isolated guinea-pig ventricular
myocytes by calcium overload. J Physiol (Lon d) 476: 459--471,1994
65. Altschuld R.A., Wenger WC, Lamka KG, Kindig OR, Capen CC,
Mizushira V, Vander Heide RS, Brierley GP: Structural and functional
properties of adult rat heart myocytes lysed with digitonin. J Bioi
Chern 260: 14325-14334,1985
66. Nichols CG, Lederer WJ: The role of ATP in energy deprivation
contractures in unloaded rat ventricular myocytes. Can J Physiol
Pharmacol 68: 183-194, 1990
67. Siegmund B, Klietz T, Schwartz P, Piper HM: Temporary contractile
blockade prevents hypercontracture in anoxic-reoxygenated cardio-
myocytes. Am J Physiol 260: H426--435, 1991
68. Murphy E, Jacob R, Lieberman M: Cytosolic free calcium in chick
heart cells. Its role in cell injury. J Mol Cell Cardiol 17: 221-231,
1985
69. Ganote CE, Kaltenbach JP: Oxygen-induced enzyme release: early
events and a proposed mechanism. J Mol Cell Cardiol II: 389--406,
1979
70. Vander Heide RS, Rim D, Hohl CM, Ganote CE: An in vitro model
of myocardial ischemia utilizing isolated adult rat myocytes. J Mol
Cell Cardiol22: 165-181, 1990
71. Silverman HS, Lakatta EG, Stern MD: Contrasting effect of anoxia
and graded hypoxia on single cardiac myocyte function. Am J
Cardiovasc Pathol 4: 256-264, 1992
72. Chacon E, Reece JM, Nieminen AL, Zahrebelski G, Herman B,
Lemasters JJ: Distribution of electrical potential, pH, free Ca'+, and
volume inside cultured adult rabbit cardiac myocytes during chemical
hypoxia: a multiparameter digitized confocal microscopic study.
Biophys J 66: 942-952, 1994
73. Allen DG, Orchard CH: Myocardial contractile function during
ischemia and hypoxia. Circ Res 60: 153-168,1987
74. Idell Wenger JA, Grotyohann LW, Neely JR: Coenzyme A and
carnitine distribution in normal and ischemic hearts. J Bioi Chern
253: 4310--8X, 1978
75. Hoek JB, Rydstrom J: Physiological roles of nicotinamide nucleotide
transhydrogenase. Biochem J 254: 1-10,1988
 391

76. Bond 1M, Harper IS, Chacon E, Reece JM, Herman B, Lemasters free magnesium in ischemic rat heart. 1 Bioi Chern 264: 5622-5627,
JJ: The pH paradox in the pathophysiology of reperfusion injury to 1989
rat neonatal cardiac myocytes. Ann N Y Acad Sci 723: 25-37,1994 78. Bram RJ, Crabtree GR: Calcium signalling in T cells stimulated by a
77. Murphy E, Steenbergen C, Levy LA, Raju B, London RE: Cytosolic cycJophilin B-binding protein. Nature 371: 355-358,1994
Molecular and Cellular Biochemistry 184: 393-400, 1998.
1998 Kluwer Academic Publishers.

Role of cellular energetics in ischemia-reperfusion

and ischemic preconditioning of myocardium

Hmo E. Hassinen, Klaus H. Vuorinen, Kari Ylitalo and Antti Ala-Rami

Department of Medical Biochemistry, University of Oulu, Kajaanintie 52 A, FIN-90220 Oulu, Finland

Abstract
A short period of ischemia followed by reperfusion produces a state of affairs in which the cells' potential for surviving longer
ischemia is enhanced. This is called ischemic preconditioning. The effects of preconditioning are also related to the reperfusion
damage which ensues upon tissue oxygenation. The role of the cellular energy state in reperfusion damage remains an enigma,
although ischemic preconditioning is known to trigger mechanisms which contribute to the prevention of unnecessary ATP
waste. In some species up to 80% ofATPhydrolysis in ischemia can be attributed to mitochondrial F1-Fa-ATPase (ATP synthase),
and a role for its inhibitor protein (IF 1) inATP preservation has been proposed. Although originally regarded as limited to large
animals with a slow heart beat, inhibition by IF) is probably a universal phenomenon. Coincidentally with ATPase inhibition,
the decline in cellular ATP slows down, but even so the difference in ATP concentration between preconditioned and non-
conditioned hearts is still small at the final stages of a long ischemia, when the beneficial effect of preconditioning is observable,
although the energy state during reperfusion remains low in hearts which do not recover. (Moll Cell Biochem 184: 393-400,
1998)

Key words: ATP synthase, phosphorylation potential, cytosolic pH, reperfusion damage, calcium, free radicals

Introduction or the shortage of their precursors, but some of them would

Ischemic damage is accentuated upon reperfusion and
reoxygenation. This reperfusion damage is significant in a
clinical setting, because modern efficient thrombolytic
therapy and coronary vascular surgery result in rapid reper-
fusion. Reperfusion involves re-energization after a pro-
longed period of de-energization, and the adverse effects of
this remain a dilemma.
Ischemia-related phenomena during
and after ischemia
Extended ischemia brings about several concomitant phe-
nomena. One has to distinguish between the irreversible
damage due to prolonged ischemia and viable myocardium,
which mayor may not recover upon reperfusion. The events
involved in ischemia may merely be secondary to the short-
age of energy, the accumulation of metabolic end products
certainly appear to have a role in preserving cell integrity or
saving energy. Typically, the mechanical work output in the
heart muscle declines, resulting in a lowering in energy
consumption. This drop in functional status, which recovers
upon reperfusion, is often called hibernation [1-3]. The
mechanical dysfunction may continue as a temporary con-
dition even during reperfusion, and in this case the condition
is referred to as stunning [4, 5]. A period of ischemia and
subsequent reperfusion may improve the ischemia tolerance
and survival of the cells during a subsequent bout of ischemia.
This is called ischemic preconditioning.
Reperfusion damage
Reperfusion can be followed by both structural and funct-
ional sequels such as loss of cellular integrity or arrhyth-
mias, and calcium overload has been suggested as one of the
main causes. The energy-dependent sarcolemmal, outward-
directed calcium pumps become de-energised during ische-

Address for offprints: limo Hassinen, Department of Medical Biochemistry, University of Oulu, Kajaanintie 52 A, FIN-90220, Finland
394
mia, which leads to calcium influx and an increase in
sarcoplasmic calcium. There is an electrogenic calcium
uniport in the mitochondrial inner membrane which keeps
the free calcium concentration in the mitochondrial matrix
higher than in the cytosol. In fact, a thermodynamic equili-
brium with the mitochondrial proton motive force would
drive the transmembrane calcium gradient up to extremes
(10 4 ), if the uniport were not counterbalanced by calcium
efflux via a sodium/calcium antiport coupled to the proton
gradient by means of a proton/sodium antiport.
Upon reperfusion/reoxygenation the mitochondria com-
mence respiration and buildup the protonic and electric
gradients of their inner membranes. This energizes the
electrogenic calcium uniport, producing a mitochondrial
calcium overload that eventually leads to mitochondrial
damage. The mechanism by which this takes place is un-
certain, but the mitochondrial permeability transition pore
(MTP) is a likely candidate. A non-selective 'megachannel'
opens upon calcium increase and the oxidation of matrix
nicotinamide nucleotides and dithiols [6). The production of
reactive oxygen species may playa role in the latter event
[7]. In any case, an open state is effectively excluded by the
high proton concentration prevailing in ischemia [8], but this
inhibition subsides due to proton elimination by washout and
metabolic elimination upon reperfusion. The MTP inhibitor
cyclosporin A protects isolated hearts from ischemialreper-
fusion damage [9], although the role ofMTP in reperfusion
damage has been doubted [10). There are even some reports
that suggest partial protection by blockers of the plasma
membrane voltage-gated calcium channels [11).
Preconditioning
Preconditioning means preparation of the tissue for prolonged
ischemia by a preceding short period of ischemia and an
intervening period of reperfusion. Since reperfusion injury
is by definition an adverse effect of sudden energization after
prolonged de-energization, measures which improve the
cellular energy state during ischemia may enhance cell
survival. It is not only ischemia that has a preconditioning
effect, as it can also be elicited by hypoxic preperfusion [12].
The situation is complex, and the phenomenon of precond-
itioning probably cannot be attributed to anyone individual
agent.
Several mechanisms have been considered. The precond-
itioning of the heart muscle that takes place as a result of brief
ischemia followed by reperfusion has been considered to be
mediated by adenosine and adenosine Al receptors, G i
proteins, K ATP channels, protein kinase C, free radicals,
including NO and reactive oxygen species, lipid bilayer
modification, induction of gene expression (proto-oncogenes
and heat shock proteins) and cellular energetics.
Adenosine
As stated above, phenomena such as myocardial hibernation
and stunning which suppress ATP consumption are poten-
tially beneficial for cell preservation. Adenosine, through its
Al receptor, has a negative inotropic and dromotropic effect
on the myocardium, which results in lowering of the energy
consumption rate. Views on the role of adenosine in pre-
conditioning are at variance, and contrasting evidence has
been presented [13, 14). Exogenous adenosine and adenosine
agonists have failed to provide protection against post-
ischemic dysfunction, and in an ischemia model comprising
isolated ventricular myocytes adenosine andA I agonists have
failed to protect the cells. Exposure to a l adrenergic agonists
has a preconditioning effect, but this is not accompanied by
purine release [15]. Thus purine release is not necessarily
connected with preconditioning.
Protein kinase C
Evidence in favor of the participation of protein kinase C
(PKC) in preconditioning includes immunofluorescence
staining with isomorph-specific antiserum, which has shown
translocation of the PKC - isomorph to the sarcolemma [16].
Diacylglycerol infusion also mimics ischemic precond-
itioning [17]. Staurosporine, an inhibitor ofPKC, blocks the
preconditioning elicited by transient ischemia or <XI adren-
ergic agonists [18], and this suggests that PKC is involved,
but as a typical G protein-mediated transducer, it needs a
triggering signal, which remains to be revealed. As explained
below, the ischemia tolerance induced by preconditioning is
biphasic, and the late phenomena seem to be related to
induction of the expression of certain stress proteins [19].
PKC or other protein kineses such as the MAP kinase
pathway which are known to have nuclear targets may be
involved in these nucleus-dependent mechanisms [20).
Mitochondrial ATP synthase
Mitochondrial ATP synthase is an ATP-driven, vectorial,
transmembrane proton pump (H+ -ATPase, known as Flo-
ATPase or complex V in the mitochondrion) which is in
equilibrium with the electrochemical proton gradient of the
mitochondrial inner membrane (L1J.l H +, having electric (mem-
brane potential) and chemical (L1pH) components) [21). An
equilibrium means that the activity of the enzyme is high
compared with the net flux of the reaction, and it also means
that, provided the proton gradient is high enough, protons
flow in the direction of ATP synthesis. Conversely, because
of the inherently highly negative Gibb's free energy change,
the reaction easily runs in the direction of ATP hydrolysis

when the mitochondrial electrochemical proton gradient
vanishes. The hydrolytic reaction is then limited only by the
proton conductance of the mitochondrial inner membrane and
reversed electron transfer in some segments ofthe respiratory
chain.
Proteins inhibiting mitochondrial ATP synthase
One of the inhibitor proteins of Flo-ATPase isolated from
the mitochondrial matrix, IF 1, binds to the enzyme in an
energy-dependent manner. This binding is promoted by a
decrease in the membrane potential and a decrease in matrix
pH [22]. A calcium-binding inhibitor (CaBI) has also been
described [23], and this is released from the enzyme upon an
increase in matrix [Ca 2+]f A role in the regulation of cellular
respiration can be envisaged for the calcium-sensitive inhib-
itor, but the situation for its potential-sensitive counterpart is
more complicated. Mitochondrial respiratory control is
expressed in the form of a dependence of oxygen consump-
tion on ADP and inorganic phosphate (P), i.e. a lowered
energy state. Therefore, inhibition ofATP synthase upon de-
energization is not feasible as a means of respiratory control.
In fact the controlling power of ATP synthase over cellular
respiration is relatively low.
Energy dependent inhibition and de-inhibition of ATP
synthase has been demonstrated in cardiomyocytes exposed
to electrical stimulation [24], and inhibition ofATP synthase
has also been observed in ischemic heart muscle [25, 26]
being reversible upon reperfusion [27]. Flo-ATPase be-
comes a major consumer ofATP during ischemia, and in the
dog heart this may contribute 80% of the ATP usage, which
then becomes oligomycin-sensitive. On the other hand, the
presence of a potential-sensitive inhibitor of Flo-ATPase
would in that case make the ATP synthase unidirectional in
function.
Recent crystallographic data on Flo-ATPase have been
interpreted as indicating that ATP synthase is probably a
mechanochemical engine in which the proton gradient is
exploited to rotate the F1 part of the enzyme, which contains
the active sites for nucleotide binding [21]. If this is so, the
IF1 protein could be visualized as a 'ratchet wheel' which
prevents a back reaction. There might also be other roles for
the IF1 protein, namely in inhibiting the proton conductivity
of the Flo complex [28], and it may act as a device which
increases the coupling efficiency of the complex [29]. This
is supported by the finding that IF 1expression is lacking from
a cell line derived from one of the first mitochondrial diseases
discovered, Luft's disease [30], typified by uncoupled
mitochondria.
One intriguing question concerns the amount ofIF 1present
in various tissues and species. Immunochemical data indicate
that its concentration in the heart muscle of small rodents is
395
low, so that any physiological role that it may have may be
limited to larger animals with a slower heartbeat [31]. The
data on the up and down-regulation of ATP synthase have
nevertheless been obtained with isolated or cultured rat
cardiomyocytes [24].
Discrepancies in reports on the presence of IFf
As mentioned in the preceding chapter, expression of the IF1
protein has been assumed to be restricted to large, slowly
beating hearts on one hand [31], although it is still found in
rat hearts on the other [24]. Although the IF1 levels differ, a
low level is not necessarily a sign of irrelevance. This is
elegantly demonstrated in the case of malonyl-C 0 A regulation
of fatty acid transport into mitochondria, for example.
Although malonyl-CoA appears to be present in significant
amounts only in lipogenic tissues, the low concentration
found in a non-lipogenic organ such as the heart muscle is
sufficient for regulation because the sensitivity of the heart
enzyme to the inhibitor is high.
Results concerning the species distribution ofIF 1function
may also differ for methodological reasons. Studies on the
rat heart have employed a coupled enzymatic assay ofATPase
with continuous monitoring of the appearance ofthe reaction
product, ADP [22]. Acid extraction of the reaction mixture
and subsequent determination of Pi was used in the earlier
studies [32], but this method is inherently less sensitive than
the above method for detecting small differences in rates
between samples. It has been pointed out recently that the
sample preparation and assay methods must be very fast in
order to obtain reliable values for the in situ activity ofF lo-
ATPase in the heart muscle [27, 33].
Kinetics ofATP synthase inhibition
The inhibition of ATP synthase by IF 1 is a relatively slow
process, because it is dependent on the binding and detach-
ment of a protein following changes in affinity. In a perfused
rat heart under ischemia the commencement of inhibition is
almost linear during the first 20 min, being 40% at this time
point [27]. After ischemia it recovers to 80% of the basal
value after 20 min of reperfusion. Considering precon-
ditioning by ischemia, this type of temporal pattern would
be suitable for ATP sparing, but even on this time scale, the
study of such patterns calls for rapid sampling and measure-
ment procedures.
Feasibility ofATP synthase inhibition in tissue protection
during reperfusion
The prevention of unnecessary ATP consumption by a
reversed ATP synthase reaction is feasible in ischemia and
mitochondrial de-energization, but an inhibition that outlasted
the ischemic period would be potentially harmful. The
distribution of enzyme activity along the pathway of oxi-
dative phosphorylation is in favor of a beneficial effect of
396

3.0 Energetics of ischemia-reperfusion

0)
1/)_
cue It is evident that if ischemic preconditioning is mediated by
Q..
t;(
O)Q.
e
'iii
2.5 a lowering ofFl a-ATPase activity, this effect must outlast
the intervening reperfusion. To be efficient, the precond-
.2: 0)
~E itioning ischemia must be short, and there is an optimum
~ Q; 2.0 length for the time interval between preconditioning and the
I/)Q.
subsequent sustained ischemia.
c.EE
'(3 When studied by 31 P NMR, which allows semicontinuous
>-- monitoring, the decline in cellular energy state, which can be
oEo
E
.2> ::::L
5-
1.5
conveniently expressed in terms of the phosphocreatine/

Q)
1.0 :J

o 3 6 9 12 15 18 21 1
u 100 .---...-..... ...........
Ischemia time (min) _'E
D..Ql
Fig. 1. Kinetics ofF,Fo-ATPase inhibition during global ischemia in the !;(15
rat heart. Oligomycin-sensitive ATPase was measured in excised rat heart 'ea.
~III
50
tissue kept ischemic for zero, 12 or 21 min. The linear regression line is
shown. R = 0.94, p < 0.001. Adapted from [27) by permission of the '0
publisher. ~ 0 L..........L_....-'--'---_......._ ..._'-......J---'--'-~
o 6 12 18 24 30 36 42 48 54 60
Q) 400
Fl a-ATPase inhibition in selected situations. Namely, FIFa- :J

ATPase activity is high enough to result a low controlling 1 300

power for ATP synthase in the regulation of oxygen con- if'~
:S;:Ql 200
sumption. That the activity is high is shown by the near-
~"
equilibrium of oxidative phosphorylation across complexes 2:e 100.
I, III andV and the adenylate and phosphatetranslocases [34]. a.
'0 0
~
ADPo + Pio + NADH j + CytCo 3+ ~ NAD j + + ~ 0 6 12 18 24 30 36 42 48 54 60

CytC o2+ + ATP o + Hp

15

where subscripts i and 0 refer to the matrix and inter- ~

U~
10
membrane spaces, respectively. Metabolic control is shared -~
..E
0.. __
with all the enzyme complexes of the mitochondrial res-
piratory chain plus the ATP synthase, and the greatest ~ 5

controlling power is usually exerted by cytochrome oxidase

0
(complex IV). At low AIlH+ values the system relaxes, and in 0 6 12 18 24 30 36 42 48 54 60
this case the rate of ATP hydrolysis is almost exclusively 8
controlled by the F la-ATPase activity and probably limited
by the proton conductance of the inner membrane, because 7
control is not shared with the respiratory complexes due to a
halt in terminal electron transfer. The inhibitor protein comes
into play here because of its putative role in regulation of the 6
proton conductance of the F l a complex [28]. The relative
control exercised by the respiratory complexes,ATP synthase 5~~~~~------~~~~~
and membrane transporters is highly dependent on the o 6 12 18 24 30 36 42 48 54 60
experimental conditions [35, 36]. In the case of isolated heart Time (min)
mitochondria oxidizing pyruvate in the presence of different Fig. 2. Behavior of the cellular energy state and cytosolic pH during
amounts of [Ca2+] to activate the pyruvate dehydrogenase ischemia and reperfusion. Filled symbols, solid line, preconditioning and
complex, the flux control coefficient ofATP synthase varies ischemia. Open symbols, broken line, control and ischemia. Adapted from
between 0.34 and zero [35] [27] by permission of the publisher.
 397

Glycogen/Glucose -+ H+

Ischemia
~ ATP/ADPPi
Ca"

C 2-
a Na+

Reperfusion

Dying cell
~/'-..
A Surviving cell
H'
Glycogen/Glucose -+ Glycogen/Glucose -+ H'

tATP/ADPPi Ca"
Ca"
ATP/ADP'Pi

Ado

Fig. 3. Recovery and failure of an ischemic/reperfused heart. In irreversibly damaged hearts the mitochondria do not attain aerobic function upon reperfusion.
An all-or-none phenomenon of recovery is envisioned as a result of the properties of the mitochondrial transition pore (MTP), since the probability of it
opening increases upon reperfusion due to excessive oxidation ofNADH; production of reactive oxygen species, oxidation ofthiol groups and decrease in
proton concentration.

inorganic phosphate ratio ([CrP]/[PJ), is rapid in ischemia,
almost reaching the steady state level of sustained ischemia
in 3 min. During this period, however, CrP is able to buffer
theATP concentration, which changes very little even though
the total phosphagen concentration (ATP + CrP) declines. The
temporal pattern of the changes in energy state is typified by
an overshoot' of the [CrP]/[PJ ratio above the basal value just
after the preconditioning ischemia. Thus, after an intervening
reperfusion, the energy state in the first three minutes of
sustained ischemia is higher in the preconditioned heart than
in non-preconditioned hearts. Although the final energy level
is equal in control and preconditioned hearts after prolonged
ischemia, recovery of the energy state after ischemia is
faster in the preconditioned heart. The [CrP]/[PJ ratio
overshoot after short ischemia is most probably caused by
a lag in the commencement of ATP consumption by the
heartbeat.
It remains to be established what is the primary cause of
reperfusion damage. Is it energetics or oxygen? In the first
case, the mitochondrial primary energy mode will be the
electrochemical potential of protons across the inner mem-
brane and the ATP/ADpPi system will only be secondary.
There are some data to indicate that the presence of oxygen
itself is necessary at least for myocardial stunning [37]. Our
own 31p NMR experience shows that in all cases where the heart
succumbs in reperfusion the cellular energy state remains
extremely low. This would indicate that re-energization of the
ATP/ADP' Pi system may not be necessary for the reperfusion
injury. NMR is problematic because accumulation of a signal
over a period of minutes is usually necessary for data
acquisition due to the low sensitivity ofthe method. It would
be important to be able to reveal the probably rapid energiz-
ation changes in the mitochondrial membrane during the
development ofreperfusion injury. This may be possible by
monitoring mitochondrial membrane potential by optical
methods in situ [38]. It is also possible that the beneficial
effect ofIF 1 binding to Flo is not necessarily due to a
reduction in ATP wastage but to prevention of the diss-
ipation of the mitochondrial electrochemical potential of
protons because of inhibition of the proton conductance of
Fo [28]. This would limit the range of membrane potential
changes, and also limit the respiratory burst with NAD(P)H
oxidation upon reperfusion.
The evidence for IF 1 participation in the energy sparing
brought about by preconditioning is only circumstantial, and
the obtaining of more direct evidence by other experimental
398
approaches such as use of exogenous inhibitors of Flo-
ATPase is hampered by the fact that these are highly lipophilic
compounds with poor water solubility, which makes the
inhibition practically irreversible and difficult to control in .
an intact tissue. A short oligomycin infusion before ischemia
retards the decline in ATP level [27], but the effect on the
[CrP]/[PJ ratio is less prominent and already shows a slow
decline during the preischemic oligomycin infusion. It may
be possible by careful titrations to find a concentration of
oligomycin which not only retards ATP decline but also
improves recovery from ischemia.
Whatever the role of the cellular energy state may be,
postischemic ATP levels and the [CrP]/[PJ ratio during the
reperfusion period are higher in the preconditioned heart. The
results are made difficult to interpret, however, by the fact
that recovery from ischemia is mostly an all-or-none phen-
omenon, i.e. the hearts which do not recover remain in
asystole or go into ventricular fibrillation at the onset of
reperfusion. NMR experiments show that the energy state in
terms of CrP/P i ratio (a variable measurable by NMR) may
also partially recover in hearts which do not attain con-
tractility, but in all cases the adenylate a and ypeaks remain
low and the ~ peak almost unmeasurable during reperfusion
in hearts which do not recover (Vuorinen and Hassinen,
unpublished). This indicates that probably all the adenyl ate
moiety is lost from the cells, and although there is enough
adenylate to make someATP for collection of CrP depending
on theATPIADP ratio, theATP concentration remains too low
for a true metabolic value.
As stated above, the MPT pore may be involved in reper-
fusion damage, and the hydrogen ion is one of its regulators,
in that the probability of the pore opening at pH values below
6.5 is negligible [8]. The cytosolic pH decreases to 5.4 in non-
preconditioned hearts but only to 5.95 in the preconditioned
heart during 21 min of continuous ischemia [27]. It is
significant that the pH decrease is less marked in the precond-
itioned heart and the pHjump (to normal) greater in the non-
preconditioned heart. It remains to be established what is the
contribution of this larger pH excursion in the non-precond-
itioned heart to its high susceptibility to reperfusion damage.
It should be noted, however, that skinned-fiber experiments
on mitochondria in the myocardium indicate that mito-
chondrial function is not compromised by reperfusion even
when the contractile function remains permanently depressed
[39].
Calcium
As described in the preceding chapters, calcium may be
implicated in reperfusion damage, which has been shown to
be alleviated by calcium channel blockers [11]. Little infor-
mati on is available on its role in the preconditioning phen-
omenon, however. [Ca 2+]f can be monitored by means of
fluorescent calcium-sensitive probes which can be loaded and
trapped in cells. The most widely used are Fura-2 and Indo-
l, with which calcium transitions can be detected by a dual
wavelength technique (either dual wavelength excitation or
dual wavelength emission detection). To study of ischemia-
related events using these indicators is nevertheless hampered
by the wavelength scale of their excitation and emission
bands, which coincides with those of the reduced forms of
the nicotinamide-adenine nucleotides. This may not result in
serious NAD(P)H interference in cells with a low mito-
chondrion content, but in mitochondrion-rich, intact tissues,
where the loading efficiency of the indicators may not be
optimal, the NAD(P)H signal during ischemia swamps the
Fura-2 or Indo-l fluorescence.
We have solved the NAD(P)H interference problem in myo-
cardium by using triple wavelength excitation fluorometry,
which is based on the fact that the lipoamide dehydrogenase
flavoprotein and NADH/NAD are in rapid equilibrium in the
mitochondrial matrix. The procedure employs simultaneous
readout of NADH and flavin fluorescence changes in an
ischemia or hypoxia protocol, calibration ofNADH fluore-
scence against flavin fluorescence and subsequent recording
of Fura-2 fluorescence changes simultaneously with the
changes in flavin fluorescence. NADH fluorescence is
calculated from the flavin fluorescence and subtracted from
the Fura-2 fluorescence signal.
The results show that there is an increase in cytosolic
[Ca 2+]f during ischemia and a recovery with a sharp [Ca2+]f
spike upon reperfusion. This is not affected by the precond-
itioning protocol, although [Ca 2+]f after this initial spike
remains higher in the non-conditioned hearts. The final [Ca2+]f
during the prolonged ischemia reaches the same values in
preconditioned and control hearts, although the initial [Ca 2+]f
increase is slower in the preconditioned hearts (K. Ylitalo and
1. Hassinen, unpublished).
Free radical biology
It has been suggested that the formation of reactive oxygen
species (ROS) may be involved in the development of the
preconditioned state [40]. Two contrasting principles have
been proposed. According to the first, ROS production during
the initial ischemia triggers the preconditioning mechanism
[37], while according to the second, preconditioning subdues
ROS formation and thus retards or prevents cell damage [41].
The altered ROS concentrations or tolerance may be related
to an increase in the expression of stress proteins, including
superoxide dismutase [42, 19].
Although a number of reports exist regarding ROS form-

ation and reperfusion, information on the heart muscle in
preconditioning is scarce. Our own laboratory has data on
lucigenin-enhanced chemiluminescence, an indicator of
superoxide anion (02~) formation, that show that there is
indeed a burst of02~ formation upon reperfusion even after
a short period of preconditioning ischemia, but that 02 ~
formation during reperfusion after more sustained ischemia
is the same in both preconditioned and non-conditioned
hearts. Within the time range of a typical preconditioning
ischemia-reperfusion-ischemia-reperfusion protocol, the
amount of ROS produced may not be sufficient to have a
significant effect on cellular function (K. Ylitalo, K. Peuh-
kurinen and I. Hassinen, manuscript in preparation).
Acknowledgement
Supported by grants from the Council for Health Sciences in
. the Academy of Finland and the Sigrid Juselius Foundation.
References
I. Rahimtoola SH, Griffith GC: The hibernating myocardium. Am Heart
J 117: 211-221, 1989
2. Burn S, Caplin JL: Myocardial hibernation. Br J Hosp Med 53: 395-
402, 1995
3. Heusch G, Schulz R: Myocardial hibernation: Adaptation to ischaemia
Eur Heart J 17: 824, 1996
4. Braunwald E, Kloner RA: The stunned myocadium: Prolonged,
postischemic ventricular dysfunction. Circulation 66: 1146-1149, 1992
5. Schulz R, Heusch G: Characterization of hibernating and stunned
myocardium. Eur Heart J 16 (suppl J): 19-25, 1995
6. Bernardi P: The permeability transition pore. Control points of a
cyclosporinA-sensitive mitochondrial channel involved in cell death.
Biocim Biopys Acta 1275: 5-9, 1996
7. Kowaltowski AJ, Castilho RF, Vercesi AK: Opening of the mito-
chondrial permeability transition pore by uncoupling or inorganic
phosphate in the presence of Ca 2 + is dependent on mitochondrial-
generated reactive oxygen species. FEBS Lett 378: 150--152, 1996
8. Nicolli A, Petronilli V, Bernardi P: Modulation of the mitochondrial
cyclosporin A-sensitive permeability transition pore by matrix pH.
Evidence that the pore open-closed probability is regulated by
reversible histidine protonation. Biochemistry 32: 4461-4465, 1993
9. Griffiths EJ, Halestrap AP: Protection by cyclosporin A of ischemia/
reperfusion-induced damage in isolated rat hearts. J Mol Cell Cardiol
25: 1461-1469, 1993
10. Darley-Usmar VM, Stone D, Smith D, Martin JF: Mitochondria,
oxygen and reperfusion damage. Ann Med 23: 583-588, 1991
II. Konrad T, B10echle C, Haller G, Broelsch CE, Usadel KH, Kusterer
K: Verapamil and flunarizine protect the isolated perfused rat liver
against warm ischemia and reperfusion injury. Res Exp Med (Berl)
195: 61--68,1995
12. Veitch K, Hombroeckx A, Caucheteux D, Pouleur H, Hue L: Global
ischemia induces biphasic response of the mitochondrial respiratory
chain. Anoxic pre-perfusion protects against ischaemic damage.
Biochem J 281: 709-715,1992
13. Asimakis GK, Inners-McBride K, Conti VR: Attenuation of post-
399
ischaemic dysfunction by ischaemic preconditioning is not mediated
by adenosine in the isolated rat heart. Cardiovasc Res 27: 1522-1530,
1993
14. Ogawa TT, Miura, K Shimamoto, Iimura 0: Activation of adenosine
receptors before ischemia enhances tolerance against myocardial
stunning in the rabbit heart. J Am Coli Cardiol 27: 225-33 1996
15. Goto M, Cohen MV, van Wylen DG, Downey JM: Attenuated purine
production during subsequent ischemia in preconditioned rabbit
myocardium is unrelated to the mechanism of protection. J Mol Cell
Cardiol28: 447-454,1996
16. Mitchell MB, Meng X, Brown JM, HarkenAH, Banerjee A: Precond-
itioning of isolated rat heart is mediated by protein kinase. Circ Res
76: 73-81, 1995
17. Speechly-Dick ME, Mocanu MM, Yellon DM: Protein kinase C. Its
role in ischemic preconditioning. Circ Res 75: 586-590, 1994
18. Liu Y, Ytrehus K, Downey JM: Evidence that translocation of protein
kinase C is a key event during ischemic preconditioning of rat
myocardium. J Mol Cell Cardiol26: 661--668,1994
19. Heads RJ, Yellon DM, Latchman DS: Differential cytoprotection
against heat stress or hypoxia following expression of specific stress
protein genes in myogenic cells. J Mol Cell Cardiol 27: 1669-1678,
1995
20. Bogoyevitch MA, Gillespie-Brown J, KettermanAJ, et at.: Stimulation
of the stress-activated mitogen-activated protein kinase subfamilies
in perfused heart. p38/RK mitogen-activated protein kineses and c-
Jun N-terminal kineses are activated by ischemia/reperfusion. Circ
Res 79: 162-173, 1996
21. Abrahams JP, Leslie AG, Lutter R, Walker JE: Structure at 2.8 A
resolution ofF,-ATPase from bovine heart mitochondria. Nature 370:
621--628,1994
22. Rosing J, Harris DA, Kemp A, Slater EC: Nucleotide-binding
properties of native and cold-treated mitochondrial ATPase. Biochim
Biophys Acta 376: 13-26, 1975
23. DasAM, Harris DA: Control ofmitochondrialATP-synthase in heart
cells inactive to active transition caused by beating or positive inotropic
agents. Cardiovasc Res 24: 411-417,1990
24. DasAM, Harris DA: Reversible modulation of the mitochondrialATP
synthase with energy demand in cultured rat cardiomyocytes. FEBS
Lett 256: 97-100, 1989
25. Jennings RB, Reimer KA, Steenbergen C: Effect of inhibition of the
mitochondrial ATPase on net myocardialATP in total ischemia. J Mol
Cell Cardiol23: 1383-1395, 1991
26. Rouslin W: Mitochondrial complexes I, II, III, IV and V in myocardial
ischemia and autolysis. Am J Physiol 244: 743-H748, 1983
27. Vuorinen KH, Ylitalo K, Peuhkurinen K, Raatikainen P, Ala-Rami A,
Hassinen IE: Mechanism of ischemic preconditioning in the rat
myocardium. Roles of adenosine, cellular energy state and mito-
chondrial Flo-ATPase. Circulation 91: 2810--2818,1995
28. Guerrieri F, Scarfo R, Zanotti F, Che Y-W, Papa S: Regulatory role of
the ATPase inhibitor protein on proton conduction by mitochondrial
W-ATPase complex. FEBS Lett 213: 67-72,1987
29. Vazquez-Contreras E, V azquez-Laslop N, Dreyfus G: The native Fof ,-
inhibitor protein complex from beef heart mitochondria and its
reconstitution in liposomes. J Bioenerg Biomembr 27: 109-116, 1995
30. Yamada EW, Huzel NJ: Distribution of the ATPase inhibitor proteins
of mitochondria in mammalian tissues including fibroblasts from
patient with Luffs disease. Biochim Biophys Acta: 1139:143-147,
1992
32. Tzagoloff A, Byington KH, MacLennan DH: Studies on the mito-
chondrial adenosine triphosphatase system. II. The isolation and
characterization of an oligomycin-sensitive adenosine triphosphatase
from bovine heart mitochondria. J BioI Chern 243: 2405-2412, 1968
33. Scholz TD, Balaban RS: Mitochondrial F,-ATPase activity of canine
400
myocardium: effects of hypoxia and stimulation. Am J Physiol 266:
H2396-H2403, 1994
34. Hassinen IE, Hiltunen K: Respiratory control in isolated perfused rat
heart. Role of the equilibrium relations between the mitochondrial
electron carriers and the adenylate system. Biochim BiophysActa 408:
319-330, 1975
35. Moreno-Sanchez R, Devars S, Lopez-Gomez F, Uribe A, Corona N:
Distribution of control of oxidative phosphorylation in mitochondria
oxidizing NAD-linked substrates. Biochim BiophysActa 1060: 284-
292,1991
31. Rouslin W: The mitochondrial adenosine 5'-triphosphatase in slow
and fast heart rate hearts. Am J Physiol 252: H622-H627, 1987
36. Wisniewski E, Kunz WS, Gellerich FN: Phosphate affects the
distribution of flux control among the enzymes of oxidative phos-
phtylation in rat skeletal muscle mitochondria. J Bioi Chern 268: 9343-
9346, 1993
37. Sun JZ, Tang XL, Park SW, Qiu Y, Turrens JF, Bolli R: Evidence for
an essential role of reactive oxygen species in the genesis of late
preconditioning against myocardial stunning in conscious pigs. J Clin
Invest 97: 562-76, 1996
38. Kauppinen R, Hassinen I: Monitoring of mitochondrial membrane
potential in the isolated perfused rat heart. Am J Physiol247: H508-
H516, 1984
39. Saks VA, Kapelko VI, Kuprianov W, Kuzuetsov AV, Lakomkin VL,
Veksler VI, Sharov VG, Javadov SA, Seppet EK, Kairane C: Quantit-
ative evaluation of relationship between cardiac energy metabolism
and post-ischmic recovery of contractile function. J Mol Cell Cardiol
21 (Suppll): 67-78,1989
40. Ambrosio G, Tritto I, Chiariello M: The role of oxygen free radicals
in preconditioning. J Mol Cell Cardiol27: 1035-9, 1995
41. Ferrari R, Ceconi C, Curello S, CargnoniA,Aflieri 0, PardiniA, Marzollo
P, Visioli 0: Oxygen free radicals and myocardial damage: protective
role ofthiol-containing agents. Am J Med 91: 95S-105S, 1991
42. Hoshida S, Kuzuya T, Fuji H, Yamashita N, Oe H, Hori M, Suzuki K,
Taniguchi N, Tada M: Sublethal ischemia alters myocardial antioxidant
activity in canine heart. Am J Physiol264: H33-H39, 1993
Molecular and Cellular Biochemistry 184: 401-408, 1998.
1998 Kluwer Academic Publishers.

Early ischemia-induced alterations of the outer

mitochondrial membrane and the intermembrane
space: A potential cause for altered energy transfer
in cardiac muscle?

A. Rossi, l L. Kayl and V Saks2

ILaboratoire de Bioenergetique, Universite Joseph Fourier, Grenoble, France; 2Laboratory of Bioenergetics, Institute
of Chemical and Biological Physics, Tallinn, Estonia

Abstract
Our aim was to carefully analyse the time-dependent changes that affect the mitochondrial function of myocardial cells during
and after an ischemic episode. To this end, variables characterizing mitochondrial function have been evaluated on myocardial
samples. from isolated rat hearts subjected to different conditions of ischemia. The technique ofpermeabilized fibers was used
in order to evaluate the mitochondrial function whilst retaining intracellular structure.
The earliest alteration that could be detected was a decrease in the stimulatory effect of creatine on mitochondrial respiration.
This alteration became more pronounced as the severity (or duration) ofthe ischemia increased. Afterwards, a significant decrease
in the apparent Km of mitochondrial respiration for ADP also appeared, followed by a diminution of the maximal respiration
rate which was partly restored by adding cytochrome c. Finally, for the most severe conditions of ischemia, the basal respiratory
rate also increased. These observations are indicative of a sequence of alterations affecting first the intermembrane space, then
the outer mitochondrial membrane, and finally the inner membrane. The discussion is focused on the very early alterations,
that could not be detected using the conventional techniques of isolated mitochondria. We postulate that these alterations to the
intermembrane space and outer mitochondrial membrane can induce disturbances both in the channelling of energy from the
mitochondria, and on the signalling towards the mitochondria. The potential consequences on the regulation of the production
of energy (ATP, PC) by the mitochondria are evoked. (Mol Cell Biochem 184: 401-408, 1998)

Key words: mitochondrial function, ischemia, outer membrane, creatine, energy and signal channelling

Introduction
The progressive alteration of mitochondrial function, which
occurs in the course of sustained ischemia of myocardial
tissue, has crucial consequences on the mechanical function
of the heart on reperfusion and on the viability of cardio-
myocytes. Numerous studies have been devoted to the
evaluation of the function of mitochondria isolated from the
tissue following various conditions of ischemia and reper-
fusion (for a review see [1, 2]. In summary, damage to
mitochondria can affect: (1) the components ofthe respiratory
chain, among which a loss of cytochrome c seems to be the
most precocious effect; (2) the transport of ADP and ATP
through the inner mitochondrial membrane mediated by the
adenine nucleotide translocator; (3) the permeability of the
inner mitochondrial membrane the increase of which leads
to a dissipation of the electrochemical potential, while the
components ofthe respiratory chain are damaged. Otherwise,
the respiratory energy can be diverted from the oxidative
production of ATP due to the cycling of Ca 2+ in excess
across the inner membrane at the expense of the proton
gradient. Finally, activation of a mitochondrial net ATP
degradation occurs, while 'pores' opening leads to mito-
chondrial swelling.

Present address: V Saks, Laboratoire de Bioenergetique, Universite Joseph Fourier, BP 53X-38041, Grenoble Cedex 9, France
Address/or offprints: A. Rossi, Laboratoire de Bioenergetique, Universite Joseph Fourier, BP 53X-38041, Grenoble Cedex 9, France
402
The alterations caused by ischemia at the level of the
outer mitochondrial compartment, i.e. the intermembrane
space, has so far not been studied in detail. The reason is
that this compartment is severely disintegrated during the
standard isolation of mitochondria, suppressing the structural
contact sites between the two membranes created through
several kinases [3]. However, the integrity of the outer
mitochondrial membrane and of contact sites, in regulation
of transport and exchange processes between mitochondria
and other cellular compartments appears to be increasingly
important [4-8]. The outer mitochondrial membrane seems
to be physiologically not freely permeable to adenine nucleo-
tides due to the properties of porin and (or) to a putative
unknown protein associated with the outer face of porin [9].
We can consider the functional association of mitochondrial
creatine kinase with both the adenylate translocator of the
inner membrane and porin of the outer membrane as a system
for molecular channelling of energy (in the form of phos-
phocreatine) out of mitochondria and of creatine, from the
cytosol, playing the role of signal [10, 11].
Experimental data on the alterations occurring during
ischemia at this step of energy channelling, and on molecular
signalling between mitochondria and other cellular structures
are scarce. A loss of mitochondrial CK activity during
ischemia has been shown [12] suggesting the occurrence
of some damage to the intermembrane mitochondrial space.
The permeabilized fiber technique [13], which allows the
study of the function of mitochondria, while the organelles
remain inside the cell, seems particularly suited to this
investigation. Utilizing this technique Veksler et at. [14] and
Saks et al. [15] have detected a decrease in the stimulatory
effect of creatine on mitochondrial respiration after
ischemia. Similarly, we have obtained experimental data on
fibers from isolated rat hearts subjected to long term cold
ischemia [16] and to normothermic (37C) ischemia [17],
that reflected alterations at the outer membrane level and
in the intermembrane space of the mitochondria. In this
paper we aim to interpret our results in the light of the
above considerations in order to identify the earliest
alterations of the mitochondrial function during ischemia
and to discuss the possible consequences on energy balance
and mechanical function of the ischemic and reperfused
myocardium.
Materials and methods
Animals
All experiments have been performed on the hearts taken
from male Wi star rats weighing 300-350 g, after anaesthesia
with sodium pentobarbital (50 mg/kg body wt).
Experimental protocols
Ischemia at 3 lOC
The hearts were perfused with a physiological solution (in
mM: NaCl129, KC15.6, MgCl z 2.4, CaClz 2.5, NaHC0 3 21,
glucose 11 and hexanoate 1), under a 100 KPa pressure, at
37C in an isovolumic mode. The functional variables were
continuously evaluated using an intraventricular latex
balloon. Ischemia was induced by reducing the coronary flow
to 0.1 mllmin for 15 or 30 min. The hearts were then
reperfused for 10 min before taking samples for fiber
isolation. In another group the hearts were subjected to a 60
min global total ischemia at 37C.
Preservation at 4C
The hearts were subjected to cardioplegia 'in situ' then
isolated and preserved for 6 or 15 h either simply immersed
in the cardioplegic solution (in mM: MgClz 16, NaCl 147,
KCl 20 and CaClz 0.5.) or low-flow (0.3 ml/min) perfused
by the same solution. The samples for fiber preparation were
taken at the end of the period of preservation.
Functional variables
The functional variables (left ventricular developed pressure,
dP/dt max, heart rate) were evaluated using an intraventricular
balloon with a diastolic pressure adjusted to 5 mmHg.
Mitochondrial respiratory function
Experimental groups
Controls 'in situ' : In a control group, the hearts were quickly
removed and immersed at 4C. The permeabilized fibers
were then immediately prepared for determination of
mitochondrial respiratory function.
Ischemia: At the end of the ischemia-reperfusion protocol,
a cardiac arrest was induced by perfusion of a St Thomas
cardioplegic solution, in mM: MgCl z 16, NaCl147, KC120
and CaC1 2 0.5. A volume of 20 ml of this solution at a
temperature of 4 C was sufficient to rapidly stop the cardiac
activity. Samples of myocardium were then taken and
immersed at 4C for fiber preparation.
Preservation: At the end of the preservation period, samples
of myocardium were then taken and immersed at 4C for
fiber preparation.
Permeabilized cardiac fiber preparation
This method has been exhaustively described and discussed
in earlier studies.
 403

All details are given in references [13, 18] and in the article
by Saks et al. in this volume [19].
The variables of mitochondrial function that could be
evaluated were:
(i) the respiratory rates: basal rate (with substrate but
without ADP:Vsubstrate)' maximal rate (with substrate + 1 or 2
mM ADP:V rna); (ii) the apparent Km for ADP; (iii) the
stimulatory effect of creatine (20 mM) evaluated with various
concentrations of ADP, and an index characterizing this
effect with 0.1 mM ADP: '(VCr-VADP)/vADP'; (iv) the
cytochrome c test: effect of the addition of cytochrome c on
the maximal rate (Vrna) evaluated in a solution containing
KCI ( 125 mM); (v) the Acceptor Control Index (ACR) = V max
IVsubstrate'
Results
The experimental results concerning the variables of the
mitochondrial function, evaluated using the permeabilized
fiber technique after various conditions of ischemia are
collected in Fig. 1. The data are drawn from the experiments
previously published in [16] and [17].
An important remark must be made before any further
interpretation. The experimental results concern a population
of mitochondria. Despite the fact that the fibers dissected
were always taken in the same small part of the myocardium,
one can suspect that the damage caused by ischemia can
diversely affect the cells according to their situation in the
tissue. Moreover inside a given cardiomyocyte the mito-
chondria can be affected to different extent because of their
position in the cell and (or) the age ofthe organelle. Therefore,
the variables evaluated represent a mean from a population
of mitochondria possibly diversely affected. Since it is not
possible to select an homogeneous sample of mitochondria,
we hypothesize that the sequence ofthe alterations observed
globally is the same as that concerning a single mitochondria.
The stimulatory effect of creatine is expressed in the figure
by the relative increase in respiration rate when creatine (20
mM) was added to fibers respiring under the effect of 0.1 mM
ADP. This index express the fact that creatine in excess enters
the intermembrane space of the mitochondria and induces
the production of ADP directly at the level of the adenine
nucleotide translocase while the transfer of ADP through the
outer membrane is under control. Indeed, it has been
previously shown [18] that the apparent Km, characterizing

Intact Damaged
membranes membranes

Outer Inner
II f"'jgl

1
175%
Vsubst.

100% It. i
u *
, Vmax (1 mM ADP)
+ cytochrome c

~ 0

V 0 0 o Vmax (1 mM ADP)
V 8: 0*

50%
v .* o o ACR =Vmax I Vsubst
*
v*
KmforADP

V *
v* V (VCr - VADP) I V ADP
0%
Cont. tfJ6h P15h 115mln 130mln tfJ15h 11h
'-y-I '-y-I 4C 37C
Preservation Ischemia
4C 37C

Figure I : Relative changes in mitochondrial respiratory variables as a function of severity of ischemia. All the variables evaluated are expressed as
percentages of the values measured on fibers taken from heart in situ. For clarity of the Fig. the means from 5-S separated hearts are reported without
indication of the standard deviation. The difference p < 0.05 from the control conditions has been checked using a statistical analysis on the absolute values.
For the definition of the variables evaluated see Materials and methods. Preservation 4C: low-flow perfusion for IS h (P 15h), immersion for 6 and IS h (NP6h
and NP 15h ). Ischemia 37C: coronary flow 0.1 ml/min for IS, 30 min (I15min and I3Omi.), global total for I h (llh)'
404
the control by ADP ofthe respiratory rate, is much higher (300
11M) when evaluated on permeabilized fibers than on isolated
mitochondria (20--30 11M). This observation has been inter-
preted as reflecting the existence of some diffusion barrier to
ADP, and several experimental works have led to the con-
clusion that the outer mitochondrial membrane is responsible
for this phenomenon [9,19,20].
Figure 1 shows the gradation ofthe alterations that occur
when the severity of the conditions of ischemia are
. .
mcreasmg.
As can be seen, for mild conditions of ischemia: 15 min at
37C, 6 h for arrested hearts at 4C, or even 15 h preservation
at 4C with low-flow perfusion, the only variable to be altered
was the stimulatory effect of creatine. At this stage of
ischemia the variables characterizing the respiratory rates
(Vsubstrate or Vrna) were unaltered. As there was no effect of
cytochrome c, one can assume that the outer mitochondrial
membrane was intact. However, some alteration in its
permeability for ADP could have occurred in some mito-
chondria since the apparent Krn for ADP was slightly (but not
significantly) lowered.
For more severe ischemic conditions: 30 min ischemia
at 37C or 15 h preservation at 4C without perfusion, the
stimulatory effect of creatine still decreased and a significant
diminution of the apparent Krn for ADP (by about 50%) also
occured.
Concerning the respiratory rate measured in the presence
of a high concentration of ADP (1 mM), which is the
maximal respiratory rate (V max' State 3), a significant
decrease occured only in the fibers from the hearts subjected
to the most severe conditions of ischemia. At least a part of
this effect can be accounted for by a loss of cytochrome c
since the addition of exogenous cytochrome c, partially (in
normothermic ischemic hearts) or completely (in preserved
hearts) restored the respiratory rate. The loss of cytochrome
c can be interpreted as a consequence of some rupture of
the outer mitochondrial membrane, a phenomenon which is
therefore not evidenced under the milder ischemia conditions
of the study. When the correction of the respiratory rate by
exogenous cytochrome c was complete it can be assumed
that no damage to the inner membrane had occured and that
the mitochondrial machinery, such as the respiratory chain,
was intact.
The respiratory rate, evaluated in the absence of ADP
(Vsubstrate)' was significantly altered (increased) only under the
strongest conditions of ischemia of our studies: i.e. 60 min
ischemia at 37C. This alteration means that the inner
mitochondrial membrane was altered under these conditions
leading to a loss of the proton-motive force. At this stage of
the ischemic process it can be assumed that the damage to
the mitochondria is irreversible since both the inner and the
outer mitochondrial membranes are altered.
From these results one can suppose that the alterations
to the different parts of the mitochondrion appearing during
ischemia follow a sequence of events that can be schematic-
ally ordered as follows.
In the first stage, the outer mitochondrial membrane is still
intact and the main alteration concerns the stimulatory effect
of creatine. At this stage the function of the myocardium on
reperfusion remains diminished by 20--50% (not shown).
In a second stage, the outer mitochondrial membrane
becomes damaged and all variables of the mitochondrial
function are modified except the rate of basal respiration. At
this stage the mechanical function is severely compromised
(10% residual).
Lastly, when the inner mitochondrial membrane becomes
damaged as evidenced by a rise in Vsubstrate value, the cardio-
myocyte does not recover contractile activity on reperfusion.
Discussion
Our own experimental results [16, 17] and those previously
published by Veksler et al. [15] and by Saks et al. [14] all
show that a very early alteration of the energy function of
myocardial myocytes is a decrease in the stimulatory effect
of creatine on mitochondrial respiration. This phenomenon
is indeed the first significant change to be detected using the
technique of permeabilized fibers at a stage when no other
alteration can be statistically shown. Since the technique
evaluates the mean activity of a population of mitochondria,
it cannot be excluded that these organelles can be differently
affected according to their localization in the cell or to the
position of the cell relatively to blood vessels. However, since
the extent of change observed in the effect of creatine was
considerably larger than the alterations of the other mito-
chondrial variables, and increased with the severity of the
conditions of ischemia, it seems reasonable to consider that
this ischemia-induced change occured prior to any other.
Afterwards, other alterations, that can be attributed to the
rupture of the outer mitochondrial membrane, are evidenced
by the partial or complete restoration of maximal respiratory
rate under the effect of adding cytochrome c. In the most
severe ischemic conditions an alteration in the inner
mitochondrial membrane and in the respiratory chain also
occured.
Let us focus this discussion on the earlier alterations. It is
generally admitted that, under physiological conditions, the
mitochondrial creatine kinase (CKm) simultaneously binds to
both the inner and outer mitochondrial membrane being
functionally coupled to Adenine Nucleotide Translocase
(ANT) of the inner and porin of the outer membrane (Fig. 2A).
This multi-enzyme complex allows phosphocreatine (PC)
synthesis far from equilibrium ofthe CKm reaction. It creates
a channelling for creatine, acting as a signal inducing the
production of ADP directly at the level of the ANT, and for

AI
PCr
BI
ADP
CI
Fig. 2. Schematic representation of the putative (hypothesized) alterations
at the level of outer mitochondrial, and intermembrane space in the course
of ischemia. (A) In physiological state adenine nucleotide translocase (ANT)
is supposed to be tightly coupled to the mitochondrial creatine kinase (CKm)
itself associated with a selective channel (porin) of the outer membrane.
Creatine, acting as a signal, induces the production of ADP directly at the
level of ANT. Meanwhile the PC produced at the level of CKm is conveyed
towards the cytosol through the associated channel. According to this
scheme PC is the final product of the mitochondrial function. Alternate (or
in addition) ADP can enter the intermembrane space by simple diffusion
through other porins not associated with the complex CKm-ANT and possibly
controlled by a factor x represented here with some connection with the
infra-structure (cytoskeleton) of the cell. This simplified scheme is prompted
from diverse publications, mainly these of the Walliman's group. It fits well
with the observations that can be made using the permeabilized fiber
technique. The addition of ADP to the preparation stimulates the respiration
but, due to the barrier of diffusion and to the size of the intermembrane
space, the extramitochondrial concentration of experimentally added ADP
must be high enough (300 ~M) to reach the level of the ANT, a value near
the Km evaluated in isolated mitochondria (Km 20 ~M). The stimulatory
effect of exogenous creatine can be explained by the production of ADP
directly at the ANT site due to the functional association with CKm. (B)
Early in the course of ischemia, alterations in the ionic composition of
cytosol and intermembrane space leads to some swelling of mitochondria, to
a detachment of CKm and cytochrome c from the outer face of the inner
mitochondrial membrane and to a rupture of the association between ANT,
CKm and porin. In addition the permeability of the outer membrane can be
405
PC, being the final product of mitochondrial function. In
addition there may be a diffusion limitation for ADP (and
adenine nucleotides) across the outer mitochondrial
membrane and it has been hypothesized [9] that a protein
factor, possibly associated to the cytoskeleton plays an
important role in the limitation of permeability to ADP. In
experiments with permeabilized fibers, due to the barrier of
diffusion for ADP, the added exogenousADP must reach high
concentrations (300-400 11M) to allow a diffusion rate of
ADP through the outer membrane sufficient to induce a rise
in the ADP concentration near the ANT to values in the
range ofthe Km for ADP evaluated in isolated mitochondria
(20 11M). In Fig. 2A, cytochrome c which is associated with
the outer face of the inner mitochondrial membrane is also
represented.
The decrease in the stimulatory effect of creatine during
ischemia can be interpreted as a consequence of a loss of
the activity of the mitochondrial creatine kinase. Such a
decrease of the stimulatory effect of creatine on respiration
during ischemia was previously detected, using the same
technique of permeabilized fibers, both by Saks et al. [14]
and by Veksler et al. [15]. However, these last authors
claimed that this effect was reversible on reperfusion while,
under our experimental conditions, the alterations persists
for at least 10 min of reperfusion. Others groups, using
different techniques, previously described ischemia-induced
CK alterations. Bitt! et al. [12] showed that a total ischemia
of the rabbit heart in situ resulted in an almost immediate
loss of CK activity. A further progressive decrease of this
activity, pe~sisting on reperfusion, was directly correlated
to contractile abnormalities. A decrease in the activity of
CKm was also evidenced by Neubauer et al. [21] in the
residual intact left ventricular tissue of chronically infarcted
rat heart. However, the same group did not detect any
decrease of tissue CKm activity in isolated ischemic ferret
heart [22].
It seems therefore that, despite variations that are probably
due to the differences in experimental protocols or species
used, an alteration to the CKm activity is a common early event
in ischemia. Thus, the question can be raised as to the nature
and cause of these alterations. In their experiments showing
increased through changes in the physiological structure of the porin and
(or) to the loss of control by factor x. As a consequence, creatine losses
its efficacy in stimulating the respiration through a channelled ADP
production and the diffusion of ADP through the outer membrane being
easier, the value of the apparent Km for ADP decreases. At this stage of
ischemia the production of A TP is not altered but both the signalling
towards the ATPase and the respiratory chain and the channelling of
energy from mitochondria are altered leading to a loss in the efficacy of
the regulation of the energy balance. (C) Later on during ischemia, rupture
of the outer membrane occurs leading to a loss of CKm and cytochrome c.
At this stage both the channelling of signal and energy and the respiratory
rate are affected.
406
an ischemia-induced CKrn activity decrease Bittlet al. [12] could
not detect any CKrn activity in other cellular fractions than
mitochondria and suggested that the decrease in CKm activity
resulted from some denaturation or degradation without
release of the enzyme. It may be that inactivation ofthis enzyme
occurs during ischemia due to the generation and the action
of oxygen free radicals. Indeed it has been demonstrated that
reactive oxygen species may inhibit myofibrillar CK [23] and
CKm [24] probably by modifying sulfhydryl groups in the
enzyme protein.
In addition, physico-chemical conditions prevailing
during ischemia can induce a release of CK from the inner
mitochondrial membrane. This can be the case of the
osmolar load leading to cell and mitochondria swelling
during ischemia [25]. This can indeed lead to an uncoupling
of the enzyme from the adenine nucleotide translocase (ANT)
and porin (Fig. 2B). An increase in the concentration of
inorganic phosphate (P) can be put forward as one of the
possible factors responsible for such an effect. Indeed during
ischemia intracellular Pi concentration rapidly approaches
values in the range used for in vitro solubilization of CKm from
the mitochondrial inner membrane [26,27]. During early ische-
mia, the solubilization of CKm from the inner membrane is not
followed by a loss of the enzyme through the outer membrane
because at this time a loss of cytochrome c cannot be
evidenced. However, an alteration in the tight connection
between ANT, CKm and porin can have a 'per se' crucial
consequence on the efficacy of energy channelling from the
mitochondria, and on the transmission of the signal (Creatine)
towards the mitochondria. A detachment ofCKm can result
in a decreased efficacy of creatine, considered as an extra-
mitochondrial signal, but also in an alteration of the function
of the external mitochondrial membrane. It can be supposed
that a deorganization of the association of the two mito-
chondrial membranes can also facilitate the access of cyto-
solic ADP to the intermembrane space via changes in the
porin structure. These changes are evidenced by the de-
crease of the apparent Krn for ADP that is detected later.
However, it cannot be excluded that this phenomenon begins
to appear earlier, even if it is not statistically significant, due
to its evaluation on a large population of heterogeneously
damaged organelles. Alternately (or in addition) one can
suppose that some putative protein factor controlling the
permeability of the porins [9] is also displaced (Fig. 2B) due
to changes in the chemical composition of the cell during
ischemia or to physical constraints provoked by the osmolar
load. The assumption that alterations in the permeability of
the outer mitochondrial membrane and in the ionic status of
the intermembrane space are among the first events during
myocardium ischemia is coherent with the concept develop-
ed by Di Lisa and Bernardi [28] about the fate of inner
mitochondrial membrane. Indeed these authors suggest that
the rise in the intracellular concentration of Ca2+ during
ischemia can promote the opening of a cyclosporine-sensitive
mitochondrial permeability transition pore leading to a
dissipation of the proton gradient, thus accelerating energy
unbalance and evolution towards irreversible damage to the
cell. We suggest that the alterations to the outer membrane
permeability occur earlier in the course of ischemia induced
damages probably at a time when the rise in intracellular
calcium concentration is not very pronounced (as suggested
by absence of contracture); these alterations are probably
reversible.
Later on, more dramatic changes occur leading to some
rupture ofthe outer membrane and to a leak of cytochrome c.
It is likely that the inner membrane is also affected (Fig. 2C).
Moreover, it cannot be excluded that at this stage the loss of
integrity of the outer membrane contributes by itself to the
evolution of necrosis. In this respect it is interesting to note
that some molecules which regulate apoptosis are located
on the mitochondrial membranes, and that the release of
cytochrome c by the mitochondria seems to participate in
apoptosis [29].
Last but not least the question arises: can such early
alterations at the outer mitochondrial membrane and
intermembrane space have consequences on the mechanical
activity of the cardiac muscle? In cases of severe ischemia
of isolated rabbit hearts (for 30 min) a progressive and
irreversible loss of mitochondrial creatine kinase activity
was detected and a close correlation between this decline
and the reduction in performance was evidenced [12]. The
creatine kinase reaction velocity decreased in proportion
to the duration of ischemia of the isolated ferret heart and
again a close linear correlation with cardiac performance
was demonstrated [22]. However, since the creatine kinase
reaction velocity is an order of magnitude greater than
maximum ATP synthesis rates, it cannot be concluded that
the high-energy phosphate transfer via creatine kinase is
directly limiting for the recovery of function in the post-
ischemic myocardium. Most of the experiments in which
the creatine kinase reaction velocity was decreased, either
by using creatine analogues to deplete PC stores [30, 31]
or by inhibiting the enzyme [32, 33], showed that, rather
than the capacity for maintaining moderate levels of
performance, the potential to increase and sustain a high
workload (i.e. the 'contractile reserve') was severely
compromised. Such an impairment of the energy reserve
was also observed in residual intact myocardium of chroni-
cally infarcted rats [21]. In this last experimental model the
mitochondrial creatine kinase isoenzyme activity decreased
and the mechanical recovery after acute stress (hypoxia) was
impaired. It seems therefore that the integrity of the PC shuttle
and of mitochondrial CK, while being not necessary to
maintain a low level of performance, is definitely required to
recruit the contractile reserve of the heart.
Most of the experimental studies have been directed

towards the role of 'energy transport' (or channelling) of the
so-called PC shuttle. However, the concerted coupling ofthe
three systems ANT, CKm and porin at the mitochondrial level
is also involved in the regulation ofthe import of creatine into
the mitochondria and regeneration of ADP. At last, the PC
shuttle can also be considered as necessary to reduce the
transient time of the system to reach new-steady-states upon
abrupt changes in workload. We suggest that the very early
alterations in creatine mitochondrial respiration control could
directly affect the signalling function of the multi-enzymatic
complex (including CKm) in the intermembrane space. This can
be particularly evidenced during the transient phase accomp-
anying acute changes in workload. A variable that is of great
interest to this respect is the transient response of oxygen
consumption as studied by Van Beek et al. [34, 35]. These
authors found that, following a step in heart rate, the mean
response time of mitochondrial oxygen consumption in
isolated rabbit hearts, which is about 7 sec [34] under normexic
conditions, increased following a brief (1 5 min) ischemia [35,
36] suggesting that a trans-cytosolic energetic signal trans-
duction was retarded. In agreement with these observations,
we may hypothesize that, while the mitochondrial capacity is
not affected by short periods of ischemia, allowing the
maintenance of the energy balance in a steady-state, the
response time is delayed creating some brief energy unbalance
during the transient phase from one workload level to another.
In summary, our own experimental results, in agreement with
other data, are coherent with the assumption that very early
ischemia induced disturbances occur in the intermembrane
space of the mitochondria. These alterations could be caused
by early alterations in the homeostasis ofthe cardiomyocyte.
They could consist of some deorganization of the postulated
intimate coupling of CKm with ANT and porin. As a con-
sequence both the features of the channelling of energy and
of the transfer of signals to the mitochondria are changed.
Since these alterations seem not to be rapidly reversible the
regulation of the energy transfer in the cardiomyocyte could
be markedly altered in the post-ischemic reperfused myo-
cardium.
Acknowledgements
The authors are very grateful to Dr Ernest Boehm, University of
Oxford, England, for correcting English. VA. Saks' s position
of invited professor in the laboratory, part-time, is supported
by the French Ministry of Higher Education and Research.
Part of this study was supported by [NTAS grant 94 4738 and
Estonian Science Foundation grant 2635.
407
References
I. Regitz v, Paulson DJ, Hodach RJ, Little SE, Shaper W, Shug AL:
Mitochondrial damage during myocardial ischemia. Basic Res Cardiol
79: 207-217, 1984
2. Piper HM, Noll T, Siegmund B: Mitochondrial function in the oxygen
depleted and reoxygenated myocardial cells. Cardiovasc Res 28: 1-
15, I 994
3. Brdiczka D, Wallimann T: The importance of the outer mitochondrial
compartment in regulation of energy metabolism. Mol Cell Biochem
1331134: 69--83, 1994
4. Brdiczka D: function of the outer mitochondrial compartment in
regulation of energy metabolism. Biochim Biophys Acta 1187: 264-
269,1994
5. Brdiczka D: Contact site between mitochondrial envelope membrane.
Structure and function in energy and protein transfer. Biochim Biophys
Acta 1071: 291-312,1991
6. Soboll S, Conrad A, Keller M, Hebisch S: The role of the mitochondrial
creatine kinase system for myocardial function during ischemia and
reperfusion. Biochim Biophys Acta 1100: 27-372,1992
7. Gellerich FN, Khuchua ZA, Kuznetsov AV: Influence of the mito-
chondrial outer membrane and the binding of creatine kinase to the
mitochondrial inner membrane on the compartmentation of adenine
nuc1eotides in the intermembrane space of rat heart mitochondria.
Biochim Biophys Acta 1140: 327-334, 1993
8. Rauch U, Schulze K, Witzenbichler B, Schultheiss HP: Alteration of
the cytosolic-mitochondrial distribution of high-energy phosphates
during global myocardial ischemia may contribute to early contractile
failure. Circ Res 75: 760-769,1994
9. Saks VA, Kuznetsov AV, Khuchua ZA,Vasilyeva EV, Belikova YuO,
Kesvatera T, Tiivel T: Control of cellular respiration in vivo by
mitochondrial outer membrane and by creatine kinase. A new
speculative hypothesis: Possible involvement of mitochondrial-
cytoskeleton interactions. J Mol Cell Cardiol27: 625-645, 1995
10. Wyss M, Smeitink J, Weyers RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta (Review) 1102: 119-166, 1992
II. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Intracellular compartmentation, structure and function of creatine
kinase isoenzymes in tissues with high and fluctuating energy demands:
the 'phosphocreatine circuit' for cellular energy homeostasis. Biochem
J 281: 21-40, 1992
12. Bittl JA, Weisfeldt ML, Jacobus WE: Creatine kinase of heart
mitochondria. The progressive loss of enzyme activity during in vivo
ischemia and its correlation to depressed myocardial function. J Bioi
Chern 260: 208-214, 1985
13. Veksler VI, Kuznetsov AV, Sharov VG, Kapelko VI, Saks VA:
Mitochondrial respiratory parameters in cardiac tissue: a novel method
for assessment by using saponin-skinned fibers. Biochim Biophys Acta
892: 191-196, 1987
14. Saks VA, Kapelko VI, Kupriyanov VV, Kuznetsov AV, Lakomkin VL,
Veksler VI, SharovVG, lavadov SA, Seppet EK, Kairane C: Quantitative
evaluation of relationship between cardiac energy metabolism and post-
ischemic recovery of contractile function. J Mol Cell Cardiol21: 6778,
1989
15. Veksler VI, KhatkevichAN, Elizarova GY, Kapelko VI: Mitochondrial
respiration in myocardial biopsy samples as a criterion of postischemic
recovery of the cardiac contractility. Basic Res Cardiol85: 307-314, 1990
16. Kay L, Daneshrad Z, Saks V, Rossi A: Alteration in the control of
mitochondrial respiration by outer mitochondrial membrane and
creatine during heart preservation. Cardiovasc Res 34: 547-556, 1997
17. Kay L, Saks Y, Rossi A: Early alteration of energy transfer in myocardial
408
ischemia J Moll Cell Cardiol 29: 3399-3411, 1997
18. Saks VA, Vasileva EY, Belikova YuO, Kuznetsov AY, Lyapina S, Petrova
L, Perov NA: Retarded diffusion of ADP in cardiomyocytes: Possible
role of mitochondrial outer membrane and creatine kinase in cellular
regulation of oxidative phosphorylation. Biochim Biophys Acta 1144:
134--148, 1993
19. Saks VA, Khuchua ZA, Vasilyeva EY, Belikova Yu, Kuznetsov AV:
Metabolic compartrnentation and substrate channelling in muscle cells.
Role of coupled creatine kinases in in vivo regulation of cellular
respiration-a synthesis. Mol Cell Biochem 133/134: 155-192, 1994
20. Saks V, Veksler VI, Kay L, Sikk P, Tiivel T, Tranqui L, Olivares J,
Kuznetsov AY, Winkler K, Wiedemann F, Kunz W: Permeabilized cell
and skinned fiber techniques in studies of mitochondrial function in
vivo. Mol Cell Biochem, this volume
21. Neubauer S, Horn M, Naumann A, Tian R, Hu K, Laser M, Friedrich J,
Gaudron P, Schnackerz K, Ingwall JS, Erti G: Impairment of energy
metabolism in intact residual myocardium of rat hearts with chronic
myocardial infarction. J Clin Invest 95: 1092-1100, 1995
22. Neubauer S, Hamman BL, Perry SB, Bittl JA, Ingwall JS: Velocity of
the creatine kinase reaction decreases in postischemic myocardium:
A 31 P-NMR magnetization transfer study of the isolated ferret heart.
Cir Res 63: 1-15,1988
23. Mekhfi H, Veksler Y, Mateo P, Maupoil V, Rochette L, Ventura-Clapier
R: Creatine kinase is the main target of reactive oxygen species in
cardiac myofibrils. Circ Res 78: 1016--1027, 1996
24. Yuan G, Kaneko M, Masuda H, Hon RB, Kobayashi A, Yamazaki N:
Decrease in heart mitochondrial creatine kinase activity due to oxygen
free radicals. Biochim Biophys Acta 1140: 78--84, 1992
25. Jennings RB, Reimer KA, Steenbergen C: Myocardial ischemia
revisited. The osmolar load, membrane damage, and reperfusion. J
Mol Cell Cardiol (Editorial Comment) 18: 769-780, 1986
26. Hall N, DeLuca M: The effect of inorganic phosphate on creatine kinase
in rat heart mitochondria. Arch Biochem Biophys 229: 477-419, 1984
27. Vial C, Font B, Goldschmidt D, Gautheron DC: Dissociation and
reassociation of creatine kinase with heart mitochondria: pH and
phosphate dependence. Biochem Biophys Res Commun 88: 1352-
1359, 1979
28. Di Lisa F, Bernardi P: Mitochondrial function as a determinant of
recovery or death in cell response to injury. Mol Cell Biochem (this
volume)
29. Yang J, Liu X, Bhalla K, Kim CN, IbradoAM, Cai J, PengTI Jones DP,
Wang X: Prevention of apoptosis by Bc1-2: Release of cytochrome c
from mitochondria blocked. Science 275: 1129, 1997
30. Kape1ko VI, Kupriyanov vv, Novikova NA, Lakomkim VL, Steinschneider
AY, Severina MY, Veksler VI Saks VA: The cardiac contractile failure
induced by chronic creatine and phosphocreatine deficiency. J Mol
Cell Cardiol 20: 465-479, 1988
31. Zweier JL, Jacobus WE, Korecky B, Brandeis-Barry Y: Bioenergetic
consequences of cardiac phosphocreatine depletion induced by creatine
analogue feeding. J BioI Chern 266: 20296--20304, 1991
32. Fossel ET, Hoefeler H: Complete inhibition of creatine kinase in
isolated perfused rat hearts. Am J Physiol252 (Endocrinol Metab 15):
EI24--E130, 1987
33. Hamman BL, Bittl JA, Jacobus WE, Allen PD, Spencer RS, Tian R,
Ingwall JS: Inhibition of the creatine kinase reaction decreases the
contractile reserve of isolated rat hearts. Am J Physiol269 (Heart Circ
PhysioI38): HI030--HI036, 1995
34. Van Beek mGM, WesterhofN: Response times of mitochondrial oxygen
consumption to heart rate steps. Am J Physiol260: H613-H625, 1991
35. Van Beek JHGM, Tian X, Zuurbier CJ, De Groot B, Van Echteld CJA,
Eijgelshoven MHJ, Hak JB: The dynamic regulation of myocardial
oxidative phosphorylation; analysis of the response time of oxygen
consumptio)1. Mol Cell Biochem, this volume
36. Zuurbier CJ, Beek JHGM van: The dynamic control of mitochondrial
oxygen consumption in isolated perfused rabbit heart is not affected
by a short period of ischemia. J Physiol479: 71-72P, 1994
Molecular and Cellular Biochemistry 184: 409--417, 1998.
1998 Kluwer Academic Publishers.

Metabolic control analysis and mitochondrial

pathologies
Thierry Letellier, Monique Malgat, Rodrigue Rossignol and lean-Pierre
Mazat
CJF-INSERM 97-05, Universite Victor 5egalen Bordeaux 2, 146 rue Leo Saignat, F-33076, Bordeaux-Cedex, France

Abstract
One of the main salient features recognized in mitochondrial diseases is the existence of a threshold in the degree of a
mitochondrial deficit for the expression of the disease. When expressed as a function of the degree of heteroplasmy, the
value of the threshold can be very high, around 90% (mutated DNAltotal DNA). This means that 10% of normal DNA is enough
to sustain a quasi normal mitochondrial oxidative phosphorylating flux. We have shown that most of the compensation is
done at the metabolic level: for instance a 70% deficit of cytochrome oxidase decreases the oxidative flux by just 10%.
Similar patterns are observed for the other complexes. Using Metabolic Control Anlaysis (MCA), we have shown that this
kind of result is inescapable: the threshold value can be correlated to the control coefficient of the deficient step. The value
of the threshold is reinforced by slight increases at the transcriptional and translational level as we show in a simple
mathematical model.
Finally we associate the threshold in the expression of a deficit, to the threshold in the energy demand of different tissues,
in order to describe various patterns of onset of mitochondrial diseases (double threshold hypothesis). (Mol Cell Biochem 184:
409-417,1998)

Key words: metabolic control analysis, mitochondrial diseases, heteroplasmy, double threshold hypothesis

Introduction fact that no precise mutation has been described to date in

The unique features of mitochondrial
genetics and pathologies. The concept
of heteroplasmy of mtDNA mutations
Several aspects of mitochondrial genetics have to be taken
into account. Firstly, human mtDNA is a 16,569 nucleotide
pair, closed circular molecule which codes for a small (12S)
and large (16S) ribosomal RNA (rRNA), 22 transfer RNAs
(tRNA) and 13 polypeptides, all of which are components
of the oxidative phosphorylation system. Several mutations
of mtDNA are known and give rise to now well described
pathologies. However, the complexes of mitochondrial
oxidative phosphorylation involve more than 100 subunits,
which means that most of them (except the 13 mtDNA-
encoded) are coded by the nuclear genome. Thus, despite the
the nuclear genome that affects mitochondrial oxidative
phosphorylation, such mutations are to be expected. Further-
more, the nucleus also codes for all the specific machinery
operating inside mitochondria for mtDNA replication,
mtDNA transcription, and mtRNA translation. Mutations
have also to be expected in these nuclear-encoded, mito-
chondrial enzymes. Some have been described, but not
localized, such as mutations regulating the number of copies
ofmtDNA (mtDNA depletion).
Secondly, mtDNA is particularly prone to mutation since
the mtDNA lacks protective proteins such as histones and
has a low-efficiency repair system. Hence, the mtDNA has
a more than ten times greater rate of mutation than the
nuclear DNA. This high rate of mutation can be used to study
evolution on short scales such as human evolution and
migrations.
Thirdly, mtDNAis inherited through the oocyte cytoplasm,

Addressfor offprints: T. Letellier, Laboratoire GESBI, Universite Victor Segalen, Bordeaux 2, 146 rue Leo Saignat, F-33076, Bordeaux-Cedex, France
410
and thus shows maternal inheritance; from the spermatozoon,
only the nucleus usually enters the oocyte. This means that
there is no mtDNA recombination at fertilization. Hence all
mtDNA variation is the product of sequential accumulation
of mutations along radiating maternal lineages.
Fourthly, since there are 2-10 copies of mtDNA per
mitochondria and some hundreds of mitochondria per cell,
there are some thousands of mtDNA molecules per cell. This
means that one mtDNA encoded gene exists at more than a
thousand times the copy in number of nuclear DNA encoded
gene, which in turn clearly implies regulation of the
expression of these different genes in the building of those
mitochondrial complexes composed of a mixture of nuclear
and mitochondrial subunits.
Fifthly, when a new mutation arises in a mtDNA mol-
ecule, it creates a mixture of mutant and normal mtDNAs
known as heteroplasmy. Thus, the expression of a mutation
in the mtDNA will be a function of the degree of hetero-
plasmy.
A new technique has been developed by King and Attardi
[3,4] to study the expression ofmtDNA mutations and the
relationships between the mitochondrial and nuclear
genomes. It consists of production of cell strains lacking
mtDNA by long term exposure to low concentrations of
ethidium bromide. These cells can survive in the presence
of pyruvate, to regenerate NAD, and uridine, because dihy-
droorotate dehydrogenase in UMP biosynthesis requires the
presence of CoQ/CoQH 2 These cells called rho 0 or
mtDNA-Iess can be fused with enucleated cells, the mtDNA
of which carries a mutation (Fig. 1) [5, 6]. The resulting
cells are called cybrids. In this way, it is possible to follow
the expression of the mitochondrial mutation and the
effects of its heteroplasmy in a wild type nuclear DNA
environment.
The threshold effect in the expression
of mitochondrial pathologies
One problem in mitochondrial diseases caused by mtDNA
mutations is to know how the heteroplasmy of the mito-
chondrial mutation is expressed in the two relevant mito-
chondrial fluxes: the respiratory rate and the rate of ATP
synthesis.
As a matter of fact, the answer is known: the group of
Attardi [6] for instance showed that, in the case of the
MELAS mutation 3243, 10%, or perhaps less, of wild type
DNA is enough to sustain a normal respiratory rate. Other
authors have already shown or have recently confirmed that,
in the case of other tRNA mutations or in the case of deletions,
10% of wild type DNA is usually enough to observe a normal
activity. For instance Sciaccoet al. [7] showed that (we quote):
'85-90% deleted mitochondrial DNA must be reached before
COX activity is impaired'. The phenotypic expression oftRNA
mutations are complex because they affect all the mitochon-
drially encoded subunits, but differentially according to their
amino-acids composition.
An example of an effect on a single complex was given by
Bindoffand Turnbull in Newcastle [16] who observed that
both in a patient with cytochrome c oxidase deficiency and
in an animal model, a copper-deficient rat, lowering the
activity of complex IV by over 50% did not affect the
respiratory flux. More recently, Kuznetsov and Kunz in
Magdeburg [17] showed that in a mouse mutant with a severe
copper deficiency the activity of COX is only about one half
of the normal activity, but that no difference was found in
maximal rates of respiration; however the control co-
efficient was higher in the mutants (0.8 instead of 0.30 for
the control value).
All these authors clearly demonstrate that there is a
threshold in the heteroplasmy of the mutation around 90%.
Above this threshold, when there is more than 90% of
mutation, the mutation leads to pathological behaviour.
Below, the fluxes of respiration and of ATP synthesis are
normal, and even the activities of some complexes are at the
normal level. This rises the question of the mechanism
leading to this threshold behaviour i.e. the question of the
effect of the heteroplasmy at each step of the process
leading from the mutation to the respiratory rate or to ATP
synthesis.
We will consider each process in turn and we will begin
with the last step which, as we have shown, is probably the
most important for generating the threshold. This will answer
the question: 'How does a variation in the activity of a
complex influence the flux of respiration or of ATP
synthesis' .
The explanation of the threshold effect
in terms of metabolic control analysis
(MeA)
It is rather easy to mimic a complex deficiency by use of
specific inhibitors. Figure 2A shows the effect of KCN on
complex IV activity alone and on the rate of oxygen con-
sumption at the same concentrations ofKCN. It can be seen
that even at 50% inhibition of complex IV one observes only
a very weak inhibition ofthe flux through the whole chain and
one has to go as far as 90% of inhibition of the isolated step
in order to obtain a substantial inhibition of respiration. This
is more obvious when represented as in Fig. 2B where we
have plotted the inhibition of the respiratory flux as a function
ofthe complex IV inhibition for the same KCN concentrations.
What we observe is a clear-cut threshold: until 90% inhibition
 411

rho 0 cells Pathological cells

o
o
.0 o
0
'.-
o

mitochondria without mtDNA

t enucleation

pathological cells
without nucleus

cybrids:
wild type nucleus
mutated mt DNA
with various degrees of heteroplasmy

Fig. 1. Construction of cybrid cells with a wild type nucleus from ro cells and the mutated mtDNA.

of complex IV, the respiratory rate decreases slowly; but,
beyond 90% of complex IV inhibition, the respiratory rate
abruptly decreases to reach the zero level.
The same pattern is observed in the case of the inhibition
of other complexes [8].
We have already interpreted this behaviour in the frame-
work of the metabolic control theory [10--12]. In this theory
an important parameter is defined: the control coefficient
which quantitatively expresses the effect on a flux of a
perturbation of a step. For instance a control coefficient of
412

120
A

100
~
0
~

=
"- 80
~
.~

13 60
-<
~
0
U
40
~

20

8
0
0 100 200 300 400 500 600

KCNJ.1M

120
B
... ...... .. AM.
100

,-..
O~ 80
~
'-'

~ 60
t'
=
f
.. 40
e
~
20

0
0 20 40 60 80 100 120

Fig. 2. (A) Complex IV inhibition by KCN of respiratory rate (.) and of isolated cytochrome c oxidase activity (0). The theoretical inhibition curve of the
global flux and the isolated step have been fitted according to [25] and [26]. (B) Respiratory rate as a function of the complex IV inhibition drawn using the
theoretical curves of Fig. 2A.

0.1 for complex IV means that a perturbation of 10% of the of every step of the network on anyone flux is equal to one:
activity ofthis complex will only entail a change of 1% in the N F
respiration rate. Thus the value of control coefficient appears L C =1 which means, for instance that:
i=l i
clearly as the initial slope ofthe curve in Fig. 2B.
A very important consequence of the definition of control CV(Oz)
Cplx)
+ CV(OZ)
Cplxlll
+ CV(Oz)
CplxIV
+ CV(Oz) +
ATPase" .....
= 1 or
coefficient is the summation theorem, stating that in a
metabolic network, the sum of the flux control coefficients CV(A,},P + CV(ATP) + CV(ATP) + CV(ATP) + = 1
CplxI Cplxlll CplxIV ATPase ......

Inhibitor
Fig. 3. Constraints on the flux inhibition curve as a function of the control
coefficient value. The curve represents the step inhibition curve. The initial
slope to the flux inhibition curve is drawn in accordance with a low control
coefficient by comparison with the initial slope to the step inhibition curve.
The final slope to the flux inhibition curve is imposed by the requirement for
this curve to reach the x-axis when the step is completely inhibited.
Experimental results have largely confirmed this theoretical
prediction ([9, 13, 14] for instance.
All the experimental results now show that most of the
control coefficients are very low. Most of them have to be
positive in the case of oxidative phosphorylation, and in order
to give a sum equal to 1 it is necessary that most of them
should be close to zero. This is an unescapable consequence
of the summation theorem (see [15] for a more general
discussion). The shape of Fig. 2B is also imposed by the
small value of the control coefficients as shown in Fig. 3:
at the beginning (small deficit), a quasi horizontal slope is
observed due to the low control coefficient. On the contrary,
at a very low activity of the step both curves must meet again,
due to the fact that the flux becomes zero when the step is
Table 1. The threshold value of the defect is expressed as a function of the
control coefficients of the different complexes of respiratory chain on the
rate of oxygen consumption (rat muscle mitochondria, after [9])
Control coefficient Threshold
(% of defect)
Complex! 0.07 90"10
Complex ill 0.23 60%
ComplexJV(COX) 0.14 70%
413
Table 2. Control coefficients of cytochrome c oxidase and of ATP/ADP
translocator in different tissues at state 3 respiration
Tissue Muscle Heart Liver Brain Kidney
Control coefficient of 0,20 0,12 0,D1 0,01 0,03
cytochrome c oxidase
Control coefficient of 0,04 0,04 0,01 0,07 0,06
ATP/ADP translocator
being completely inactivated. This unavoidable behaviour
leads to a sigmoid inhibition shape of the flux inhibition curve
and to a threshold effect when the flux is plotted as a function
of the inhibition of one of its steps. Furthermore, Table 1 shows
that the lower the control coefficient, the higher is the
threshold.
We have shown that this threshold effect is also observed
in a model of oxidative phosphorylation developed by
Bernard Korzeniewski in our laboratory [18].
This observation lead us to the following hypothesis: we
have demonstrated that the threshold effect observed on a
flux value when a specific activity is modulated is mainly the
consequence of the value of the control coefficient of the
step on the flux. We know that the control coefficient of a
given step can vary according to different types of mito-
chondria. This led us to propose the hypothesis that part of
the tissue specificity in the metabolic expression of mito-
chondrial mutations could be due to the differences in control
coefficients. Thus we measured the control coefficients of
the cytochrome c oxidase and of the nucleotide adenylic
translocator in different types of tissues: heart, muscle, liver,
kidney and brain. The results we obtain (Table 2) show that,
in some cases at least, the variation in the expression of a
mitochondrial mutation in different tissues could be due to
the tissue variation of the control coefficients of the various
complexes.
The reinforcement of the threshold at transcription and
translation steps
Next, we wi11look at the effect of the heteroplasmy in the
first steps of the expression of mitochondrial mutations, that
is on transcription and translation steps.
Unfortunately, there are, to our knowledge, only two,
incomplete, quantitative reports in the literature.
One of them comes from the laboratory of Serge Alziari
in Clemmont-Ferrand [19, 20] (Table 3) and concerns a deletion
in the mitochondrial DNA of a Drosophila species. The results
exhibit a slight elevation of the wild type compounds at each
step of the process: a wild type mRNA ratio slightly higher
than the corresponding wild type DNA ratio and a percentage
of complex activities slightly higher than the corresponding
414

Table 3. Expression of the heteroplasmy of a large deletion in the

mitochondrial genome of a Drosophila subobscura strain. After [19] and
[20] 100

Relative level in the mutant (%) ><

"S. 80
mtDNAtotal 150 2
mt DNA normal 30 -<
Z
QI:O 60
L\mtDNA 120
ND5mRNA 35 e
Eo<
NDlmRNA 45 ~ 40
ND4mRNA 55 ~
CytobmRNA 66
L\mRNA 45 20
Complex I activity 60
Complex III activity 70
0
Complex IV activity 107
0 20 40 60 80 100
Resp. (Glu-Mal) 70
Resp. (Succ.) 100 % WTmtDNA (WTmRNA)

Fig. 4. Percentage of WT mRNA (or WT tRNA) as a function of the

percentage in WT mtDNA according to equation (1).
Table 4. Expression of the heteroplasmy of a mtDNA deletion in
Iymphoblasts derived from a patient with Pearson's syndrome. After [21].
(ND4 and Cox II are included in the deletion) A simple model of the threshold as a function of
heteroplasmy
Relative level in the patient (%)

mtDNAtotal 140 We have developed a simple model which considers the

mtDNA normal 40 expression of a mitochondrial mutation according to
L\ mtDNA 100
Scheme 1. We have used simple functions to express the
ND4mRNA 55
COXIImRNA 40 result of each step of this scheme: - the normalized quantity
COX I mRNA 100 ofWT mRNA and WT tRNA (between 0-100%; 100% is
COXIVmRNA 100 the normal wild type (WT) quantity) is given as a function
COX VI mRNA 100 of the normalized quantity ofWT mtDNA (between 0-100)
COX activity 81-88
by the hyperbolic function, with KmtDNA = 100:
ATP synthesis 100
200* WTmtDNA
WTmRNA = (1)
KmtDNA + WTmtDNA
wild type mRNA ratio. Finally, this ends up with a quasi
This function is equal to 100 for mtDNA = 100 (normal
normal recovery ofthe respiratory flux.
situation); it gives a slight increase in mRNA or tRNA
The other comes from the laboratory of Coby van den
between 0-100 as it can be seen in Fig. 4). -the normalized
Bogert in Amsterdam [21] (Table 4), and shows the same
quantity of a complex is given by the same type offunction:
phenomenon with a normal rate of ATP synthesis despite
the fact that the content in wild type mitochondrial DNA 200 *WTmRNA
is only 40%. Cplx = (2)
KmRNA + WTmRNA

mRNA
}~
Subunits of
mtDNA Respiratory Chain ~ V02 ~ VATP
tRNA complexes

SCHEME 1
 415

This gives also a slight increase in the quantity of complex
compared to the corresponding quantity of mRNA. We
additionally hypothesise that the decreased quantity oftRNA
is not limiting until a threshold tRNA o IftRNA < tRNAo then
the quantity of complex becomes:
2*WTmRNA 2* WTtRNA
CPlx = 100* * (2')
KmRNA +WTmRNA KtRNA + WTtRNA
- the rate of respiration is given by a function simulating the
threshold effect described in Fig. 2. We have described this
type of curve simply by two straight lines (see Fig. 5):
IfCplx~ Cplx'h then V02 =C*Cplx + 100 (I-C) (3), where C is
the control coefficient ofthe step catalyzed by the complex.
IfCplx~ Cplx'h then we define V02 th = C*Cplxth , + 100 (I-C)
and
V02 = V02JCplxth * Cplx (3')
With these equations and for tRNAo = 0.2 (20% of tRNA is
enough to translate the quantity of WTmRNA) , we obtain the
curve of Fig. 6 which is very similar to the Fig. I in [6] and
exhibits a strong threshold in the expression of the mtDNA
mutation in the rate of oxygen consumption around 15% of
mtDNA.
It can be seen in the simulation that for instance, 30% of
WT mtDNA gives 46% of the corresponding WT mRNA
(involved in the mutation) and 63% of the corresponding
complex itself and then 96% in the rate of respiration.
The double threshold hypothesis
Douglas Wallace, some years ago, also proposed the concept
of a threshold in the expression of mitochondrial mutations
but ofa slightly different nature [1,22]. This threshold mainly
concerns the energy demand of a tissue, with a classification
ofthe tissues according to their energetic needs; this explains
why some tissues that have higher energy demands, such as
brain and muscles, are also more sensitive to mitochondrial
diseases. If the supply of ATP by mitochondria falls down
under the particular threshold ofthe tissue, pathological signs
appear. A decrease in the mitochondrial power due to aging
or to an increase in heteroplasmy can lead to pass under the
threshold.
In order to explain the tissue variability of mitochondrial
diseases, we propose combining both concepts of the
threshold into what we call the double threshold model
(Fig. 7).
The intercept of the two curves separates the normal
functioning of the mitochondria from the pathologic state.
It can be evidenced on Fig. 7 two types of intercept which
can lead to different types of pathologies. In the case of
intercept of type 1 (Fig. 7A) one can expect a progressive
installation of a pathological state as is observed in the
chronic encephalomyopathies in the adults (Keams-Sayre;
MELAS or MERRF). The intercept of type 2 (Fig. 7B) will
correspond to an abrupt passage towards a pathologic state
as observed in fatal infantile myopathies.

V02
100

V02th ~
80

60

40

20

20 t 40 60 80 100

Respiratory Complex
Cplxth

Fig. 5. Percentage of normal respiratory rate as a function ofthe percentage in the activity of a complex (equations) C is the control coefficient of the step on
the respiratory flux and Cplxth is the threshold as defined in the text.
416

FLUX A

N
0
100
..... - - IEnergetic threshold in tissue I

> 80

60

40

0% Defect
20
J~
0
0 20 40 60 80 100
WTmtDNA

FLUX
B

Fig. 6. Percentage of respiratory rate as a function of the percentage of

--1-----+---f~-- IEnergetic threshold in tissue 2
WT mtDNA resulting from the combination of the equations (I) to (3').

Conclusion
Mitochondrial metabolism and genetics exhibit several
particularities which force us to consider them differently
from cytosolic metabolism, particularly in the case of
inborn errors of metabolism affecting oxidative phos-
phorylation.
Several mutations of mitochondrial DNA have been
described that can lead to different pathologies as a function
of the heteroplasmy of the mutation in the various tissues.
One of the main salient features recognized in mito-
chondrial diseases is the existence of a threshold in the
degree of a mitochondrial deficit for the expression of the
disease. When expressed as a function of the degree of
heteroplasmy the value of the threshold can be very high,
around 90% (mutated DNA/total DNA). This means that
10% of normal DNA is enough to sustain a quasi normal
mitochondrial oxidative phosphorylating flux. We have
shown that most of the compensation is done at the metabolic
level: for instance a 70% deficit of cytochrome oxidase
decrease the oxidative flux by just 10%. Similar patterns are
observed for the other complexes in accordance with
metabolic control theory. The value of the threshold is
reinforced by slight increases at the transcriptional and
translational level.
In fact this result, though clearly apparent in mitochondria,
because the heteroplasmy phenomenon allows a complete
0% Defect
Fig. 7. The double threshold hypothesis.
range of variation of a mutation between 0--100%, is more
general and had already been depicted particularly by Kacser
[23]. It is an inescapable result ofMCA and more particularly
of the summation theorem as shown on Fig. 3.
The fact that different tissues not only have differences in
energy demands, but also have different types of mitochondria
and thus different control coefficients (and associated
threshold) at each step can explain the tissue differences
expression of mitochondrial mutations. The combination
of these two concepts leads us to the double threshold
hypothesis.
Acknowledgements
This work was supported by the Association Fran9aise contre
les Myopathies (A.F.M), the Universite Bordeaux II, the
Region Aquitaine, INSERM and the French Ministry of High
Education and Research. The authors wish to thank Dr. D.
Fell for many valuable comments.

References
I. Wallace DC: Diseases of the mitochondrial DNA. Ann Rev Biochem
61: 1175-1212, 1992
2. Torroni A, Wallace DC: Mitochondrial DNA variation in human
populations and implications for detection of mitochondrial DNA
mutations of pathological significance. J Bioenerg Biomemb 26: 261-
271,1994
3. King MP, Attardi G: Injection of mitochondria into human cells leads
to a rapid replacement ofthe endogenous mitochondrial DNA. Cell 52:
811--819,1988
4. King MP, Attardi G: Human cells lacking mtDNA: Repopulation with
exogenous mitochondria by complementation. Science 246: 500-503,
1989
5. Hayashi J-I, Otha S, Kikuchi M, Takemitsu M, Goto V-I, Nonaka 1:
Introduction of disease-related DNA deletions into HeLa cells lacking
mitochondrial dysfunction. Proc Nat Acad Sci USA 88: 10614--10618,
1991
6. ChomynA, MartinuzziA, Yoneda M, DagaA, Hurko 0, Johns D, Lai
ST, Nonaka I,Angelini C,Attardi G: MELAS mutation in mtDNA site
for transcription termination factor causes defects in protein syn-
thesis and in respiration but no change in levels of upstream and
downstream mature transcripts. Proc Nat Acad Sci USA 89: 4221-
4225,1992
7. Sciacco M, Bonilla E, Schon EA, DiMauro S, Moraes CT: Distribution
of wild-type and common deletion forms of mtDNA in normal and
respiration-deficient muscle fibers from patients with mitochondrial
myopathy. Human Mol Genet 3: 13-19, 1992
8. LetellierT, Malgat M, Mazat J-P: Control of oxidative phosphorylation
in muscle. Application to mitochondrial myopathies. Biochim Biophys
Acta 1141: 58-64,1993
9. Malgat M, Letellier T, Jouaville SL, Mazat J-P: Value of the control
theory in the study of cellular metabolism - Biomedical implications. J
BiolSyst3: 165-175, 1995
10. Kacser H, Burns JA: The control of flux. In: DD Davies (ed). Rate
Control of Biological Processes. Cambridge University Press,
Cambridge, UK, 1973, pp 65-104
11. Heinrich R, Rapoport TA: A linear steady-state treatment of enzymatic
chains. General properties, control and effector strength. Eur J Biochem
42: 89-95, 1974
417
12. Reder C: Metabolism Control Theory: A structural approach. J Theor
Bioi 135: 175-201, 1988
13. GroenAK, Wanders RJA, WesterhoffHV, Van der Meer R, Tager JM:
Quantification of the contribution of various steps to the control of
mitochondrial respiration. J Bioi Chem257: 275-457, 1982
14. Tager JM, Wanders RJA, Groen AK, Kunz W, Bohnensack R, Kuster
U, Letko G, Bohme G, Duszynski JWojtczak L: Control of mitochondrial
respiration. FEBS Lett 151: 1-9,1983
15. Mazat J-P, Reder C, LetellierT: Why are most flux control coefficients
so small? JTheorBiol182: 253-258, 1996
16. BindoffLA: PhD thesis, Newcastle upon Tyne University, 1990
17. Kuznetsov AV, Clark JF, Winkler K, Kunz WS: Change in flux control
coefficient of cytochrome c oxidase in copper deficient mottled brindled
mice. In: E Gnaiger, FN Gellerich, M Wyss (eds). What is Controlling
Life? Modern Trends in BioThermokinetics Vo13. Innsbruck University
Press, 1994,pp 141-144
18. Korzeniewsky B, Mazat J-P: Theoretical studies on the control of
oxidative phosphorylation in muscle mitochondria. Application to
mitochondrial deficiencies. Biochem J319: 143-148, 1996
19. Beziat F, Volz-LingenhoIA, Saint Paul N,Alziari S: Mitochondrial genome
expression in a mutant strain ofD. subobscura, an animal model for large
scale mtDNA deletion. Nucl Acid Res21: 387-392,1993
20. Debise R, Touraille S, Durand R,Alziari S: Biochemical consequences
of a large deletion in the mitochondrial genome of a Drosophila
subobscura strain. B.B.R.C. 196: 355-362, 1993
21. Spelbrink IN, Van Oost BA, Van den Bogert C: The relationship between
mitochondrial genotype and mitochondrial phenotype in Iymphoblasts
with a heteroplasmic mtDNA deletion. Human Mol Genet 3: 1989-
1997,1994
22. Wallace DC: Mitochondrial genetics: A paradigm for aging and
degenerative diseases? Science 256: 128-133, 1993
23. Kacser H, Burns JA: The molecular basis of dominance. Genetics 97:
639-666, 1981
24. LetellierT, Heinrich R, Malgat M, Mazat J-P: The kinetic basis of the
threshold effects observed in mitochondrial diseases: A systemic
approach. BiochemJ302: 171-174, 1994
25. Gellerich FN, Kunz WS, Bohnensack R: Estimation of flux control
coefficients from inhibitor titrations by non-linear regression. FEBS
Lett274: 167-170, 1990
26. Holzhiitter H, ColosimoA: SIMFIT: A microcomputer software-toolkit
formodelistic studies in biochemistry. CABIOS 6: 23-28,1990
Molecular and Cellular Biochemistry 184: 419--426,1998.
1998 Kluwer Academic Publishers.

Mechanisms of thyroid hormone control over

sensitivity and maximal contractile responsiveness
to p-adrenergic agonists in atria
Enn K. Seppet, 1 Allen Kaasik, 1 Ave Minajeva, 1 Kalju Paju, 1 Jonna J.
Ohisalo,2 Roland Vetter3 and Unno Braun 1
1Department ofPathophysiology, University of Tartu, Tartu, Estonia; 2Department of Medical Chemistry, University of

Helsinki, Helsinki, Finland; 3Max Delbrilck Center for Molecular Medicine, Berlin, Germany.

Abstract
This paper discusses the mechanisms of two basic effects of thyroid hormones on atrial responses to ~-adrenergic agonists, i.e.
increased inotropic sensitivity and decreased maximal contractile responsiveness. The increased sensitivity of atria to ~
adrenergic agonists under thyroid hormones appears to be related to increases in ~-adrenoceptor density and G /G j protein ratio,
leading to activation of G, -mediated pathway, but suppression of Grmediated pathway of adenylate cyclase regulation. Therefore,
the i/c concentrations of cAMP and corresponding inotropic responses achieve their maximums at lower doses of~-adrenergic
agonist. Thyroid hormones also decrease the expression of phospholamban, but increase the expression of sarcoplasmic reticulum
Ca2+-pump. As a result, the basal activity of sarcoplasmic reticulum Ca2+-pump increases, but its ~-adrenergic activation through
phosphorylation of phospholamban decreases. It is suggested that these changes are causal for decreased maximal inotropic
and lusitropic responses of atria to ~-adrenergic agonists. (Mol Cell Biochem 184: 419-426, 1998)

Key words: atria, thyroid hormones, ~-adrenergic effect, sarcoplasmic reticulum, phospholamban

Introduction synthesis of phospholamban (PLB), a SR Ca2+-pump regulating

The ~-adrenergic agonists increase cardiac contractility and
relaxation. These effects are mediated by cAMP-dependent
phosphorylation of several key proteins in the sarcolemma,
sarcoplasmic reticulum (SR) and myofibrils [1]. It has been
shown that thyroid hormones markedly modulate the action
of ~-adrenergic agonists. A shift from hypothyroidism to
hyperthyroidism is associated with decreased EC 50, i.e. with
increased sensitivity to ~-adrenergic agonist [2-7]. At the same
time, however, a dramatic decrease in maximal inotropic
response to ~-adrenergic agonist has been observed [4, 5, 8-
10]. The molecular mechanisms underlying these differential
effects of thyroid hormones are not fully understood. The
increased sensitivity to catecholamines has been related to
enhanced ~-adrenoceptor density [2, 11, 12]. On the other
hand, the diminished responsiveness of ventricular myo-
cardium to isoproterenol (ISO) has been attributed to decreased
protein [8-10]. However, in the experiments on ventricular
myocardium, the contribution of thyroid hormone-dependent
changes in myosin ATPase isoenzyme profile [13-16] to
modulation ofISO effects cannot be ruled out. Therefore, we
have studied the effects of ISO on contractile function and
underlying biochemical mechanisms in rat atria under different
thyroid states. Contrary to ventricular myocardium, the
expression of myosin is insensitive to thyroid hormones in atria
[13,14,16]. This feature allows to specifically outline the role
of hormone-dependent changes in SR in controlling contractile
function, without interference of the changes in myosin activity.
The results of these studies, in conjunction with previous
reports, allow to conclude that thyroid hormone-dependent
changes in atrial sensitivity and maximal inotropic response
to ISO are likely regulated via different mechanisms, through
altered regulation of adenylate cyclase and changes in PLBI
SR Ca2+-pump ratio, respectively.

Addressfor offprints: E.K. Seppet, Department of Pathophysiology, Faculty of Medicine, University ofTartu, 18 Ulikooli Street, EE2400 Tartu, Estonia
420

Effects of thyroid hormones on atrial contractile responses Effects of thyroid hormones on /3-adrenoceptors, G-
to isoproterenol proteins and adenylate cyclase activity in atria

Dose-response curves of atrial contractility have revealed the
decreased sensitivity to ISO under hypothyroidism, but
unchanged or increased sensitivity under hyperthyroidism
compared to euthyroid state [3-5, 7, 17-19]. In contrast,
hyperthyroidism greatly reduces the maximal inotropic
response of atria to ISO [4, 5, 10,20]. This may be due to
increased basal level of ~-adrenergic activation, resulting
from increased atrial ~-adrenoceptor density [5, 6] and/or
enhanced tissue content of endogenous catecholamines [3].
We have tried to rule out the contamination of endogenous
catecholamines by careful washing of atrial preparations
before the effects ofISO were studied. For this purpose, the
electrically paced atria were perfused without recirculation
during 2 h of adaptation period. Throughout that period, the
developed tension (DT) decreased until reaching its stable
plateau level. We found that inhibitors of protein kinase A
(PKA) and Ca2+-calmodulin-dependent protein kinase, added
cumulatively, suppressed the effects of ISO on DT to the
similar extent as after catecholamine washout (results not
shown). This was taken to indicate that, by the end of
adaptation period, the endogenous catecholamines had been
washed out from the atria. The ISO dose-response curves
obtained thereafter (Fig. lA) demonstrate that the maximal
positive inotropic response was significantly higher in
hypothyroid atria, but negligible in hyperthyroid atria
compared with euthyroid controls. At the same time, the
sensitivity to ISO was clearly lower in hypothyroid than in
euthyroid atria (Fig. IB), whereas the responses of hyper-
thyroid atria to ISO were too small to obtain reliable EC so ' It
should be noted that changes in relaxation fully paralleled
those in contractility (results not shown).
In order to elucidate mechanisms underlying the different
thyroid hormone effects on sensitivity and maximum ino-
tropic response to ISO various components of transsarco-
lemmal ~-adrenergic signaling were studied. Like in
ventricles [2, 11, 12], hypothyroidism is associated with
decreased and hyperthyroidism with increased ~-adreno
ceptor densities in atria [5, 6]. Probably, the upregulation of
~-adrenoceptors is based on increased transcription of
relevant gene under thyroid hormones, similarly to that
occurring in ventricles [21].
The ~-adrenoceptors are coupled to adenylate cyclase via
Gs proteins, which activates the enzyme. On the other hand,
the suppressors of adenylate cyclase act through Gi proteins
[22]. Treatment of hypothyroid rats with thyroid hormones
resulted in a 3-fold increase in atrial Gs levels. At the same
time, only a slight increase in Gia protein was observed (Fig.
2) [20]. In ventricular myocardium, however, a shift from
hypothyroidism to hyperthyroidism was associated with
decreased G ia protein levels without alterations in Gsa levels
[23, 24]. Experiments on isolated T3-treated neonatal rat
ventricular cardiomyocytes have shown that thyroid hormone
increases the Gsa levels by prolonging the half-life of that
protein and decreases the Gia protein levels by suppressing
transcription of its gene [25]. In general, these results show
that although thyroid hormones modulate expression of
individual G proteins differently in atria and ventricles, the
increased G/Gi ratio is a common response to these hormones
in both tissues.
Several observations suggest the important role ofthyroid
hormone-dependent changes in G/Gi ratio in a cascade of
adenylate cyclase regulation. Firstly, in hypothyroid atria,

A B
2000 hypothyroid c
0
(5
.... o euthyroid ~
"E 1500 A hyperthyroid .~
"0 75
0
(.) as
'0 '0
~ 1000 0~ 50
~ ~
n 500
U
as
.... 25
~
"E "E
0 0
() ()
0-
0
CD -10 -9 -8 -7 -6 CD -10 -9 -8 -7 -6
log{[lSQ] . M"} log{[lSQ] . M'}

Fig. 1. Dose-response curves for ISO, expressed as a percentage of the maximum activation by ISO (A), and as as a percentage of +dT/dt level before ISO
addition (B). Here and in the following figures, the results are expressed as the mean SEM. n =4--6. *-p < 0.05 compared to euthyroid atria (n =4--6). Methods:
Wistar rats were made hypothyroid and hyperthyroid as described previously [20]. Electrically paced (1 Hz) left atria were perfused in the medium gassed with
100% O 2 and containing (in mM): NaC1120, KCI5.4, CaC~ 1.0, NaH'pO. 0.42, MgCI2 1.05, D-glucose II, Na2EDTA 0.05, HEPES 5.0 (PH 7.4, 30C).
 421

150 on muscarinic receptor density [28]. Malbon et ai. [29] have

(IJ
~Ql
~100
'0
>.
.r=
"5
Ql 50
15
o
o
o ceptors, this may constitute an important mechanism for
Hypothyroid Hypothyroid+T.
Fig. 2. Effect of thyroid state on relative G, and G, protein levels in rat
atrial homogenates. *-p < 0.05 compared to euthyroid atria. (Modified from
ref. [20]).
activation of adenylate cyclase either by ISO or fluoride ion,
which directly stimulates G s proteins, is reduced compared
to euthyroid preparations [3]. Likewise, decreased accumu-
lation of cAMP in response to ISO occurs in hypothyroid
ventricles [23, 26]. It is known that thyroid state exerts no
effect on the maximal forskolin stimulated activity of
adenyl ate cyclase [23, 27]. Hence, the decreased effects of
ISO are restricted to decreased p-adrenoceptor levels and/
or diminished G/G j protein ratio. The second group of data
suggests, that thyroid hormone-dependent changes in G /Gj
protein system may affect adenyl ate cyclase even inde-
pendently of ligand-receptor number. For instance, the
sensitivity and maximum response of atria to adenosine Al
receptor agonist, acting via G j, were found to be elevated
in sequence hyperthyroidism ~ euthyroidism ~ hypothy-
roidism (Fig. 3) [20]. These changes were not due to the
altered Al receptor densities, for the values ofthat parameter
were similar in all three groups of muscles studied [20]. In
atria, hypothyroidism increased the sensitivity to carbachol,
shown that although no detectable difference in the number
of p-adrenoceptors was found in membranes from fat cells
under different thyroid states, the catecholamine-stimulated
adenylate cyclase activity varied directly with thyroid status.
Collectively these data show that the decreased G/Gj ratio
in hypothyroid atria leads to the suppressed effects of G s -
mediated agonists, but enhances the influence ofGj-mediated
agonists. Together with decreased number of p-adreno-
decreased sensitivity to ISO in hypothyroid atria. However,
it cannot increase the maximal inotropic response to ISO
under hypothyroidism.
Effects of thyroid hormones on cAMP-phosphodiesterases
in atria
The levels of intracellular cAMP are regulated not only by
the activity of adenylate cyclase, but also by the rate of cAMP
degradation catalysed by cyclic nucleotide phosphodiesterases
(PDEs). Therefore, the effect of hypothyroidism on isoenzyme
profile of cAMP-degrading PDEs was assessed in rat heart
[30]. It was found that, in response to altered thyroid state,
the cytosolic PDE activity remained unchanged in atria.
However, the activity of membrane-bound PDE decreased by
30% under hypothyroidism compared to euthyroid state (Fig.
4). This was due to corresponding decline in type IV isoenzyme
(also known as cAMP-specific PDE) activity, whereas type III
PDE (cGMP-inhibited PDE) activity did not change. Treatment
of hypothyroid rats with T3 reversed these changes in type IV
PDE activity. The mechanisms of decreased activity of type
IV PDE in hypothyroid atria are unclear. In addition to direct
transcriptional control, there is another possible reason for
decreased membrane-bound type IV PDE in hypothyroid

,
another GI-mediated mediator, without exerting any effect atria. In mammalian heart the type IV PDE is partly bound to

GI 100
::I b. hyperthyrOid
<U
> o euthyroid
OJ
2
80 !' hypothyrOid
'C
ea. 60
~
.~ 40
U
e 20
c
8 0
CD 7 -6.5 -6 -5.5 -5
log([PIA) . M '} EU HYPO HYPO+T.

Fig. 3. Effect of N'-(phenylisopropyl)-adenosine (PIA) on I /lM ISO- Fig. 4. Effect of thyroid state on membrane-bound cAMP-degrading
stimulated contractility (+dT/dt) in left atria from euthyroid, hypothyroid, phosphodiesterase isoenzyme activities in rat atria. Type III - type III PDE,
and hypothyroid+T] treated rats. *-p < 0.05 compared to euthyroid atria. type IV -type IV PDE, total-total PDE. *-p < 0.05 compared to euthyroid
(Modified from ref. [20]). atria. (Modified from ref. [30]).
422
SR [31]. Since the amount ofSR membranes was dramatically
reduced under hypothyroidism in atrial tissue (as suggested by
Figs 7 and 9), it could result in decreased activity of SR-
bound PDE as well.
To evaluate the role of thyroid hormone-dependent
changes in PDE isoenzymes, we have assessed the effects
of isoenzyme-selective inhibitors ofPDE on contractility of
atria isolated from rats with different thyroid states. Figure 5
shows that euthyroid atria responded to rolipram (type IV
PDE inhibitor) with marked concentration-dependent
potentiation, so that subsequent addition of siguazodan (type
III PDE inhibitor) exerted just a minute extra stimulation
(compared to values in the presence of 1 11M ISO, taken for
100%). In comparison with euthyroid atria, the hypothyroid
ones exhibited markedly lower stimulation of contractility
with rolipram, and the major part of activation was gained
only after subsequent addition of siguazodan. Positive
inotropic effect of rolipram in euthyroid atria reflects a
dominance of type IV over type III PDE. Thus, in condition
of inhibition of type IV PDE, the type III isoenzyme with
relatively low activity fails to effectively eliminate the cAMP
production that leads to development of the positive inotropic
effect. Hypothyroidism, however, by decreasing the type IVI
type III PDE ratio, sets these two isoenzymes towards their
equal role in hydrolyzing cAMP. Consequently, the system
becomes less sensitive to inhibition of single enzyme, as
another noninhibited isoenzyme effectively degrades cAMP.
It has been demonstrated that partial inhibition of PDE
increased the inotropic sensitivity of cardiac muscle to ISO
o euthyroid
100 hypothyroid
-
0 accelerated by PKA-dependent phosphorylation of SR Ca 2+
~
75
0
~
0
50
~ faster activation of contractile apparatus. On the other hand,
~
"13
~ 25 * *
c0 * which Ca2+ dissociates from troponin C [40]. Protein kinase
0 A also phosphorylates PLB that relieves the SR Ca2+-pump
0
o 10 20 10 20
[rolipramjM 6 [siguazodanjM6
Fig. 5. Effect ofrolipram (type IV POE inhibitor) and siguazodan (type III
POE inhibitor) on contractility (+dT/dt) in left atria from euthyroid and
hypothyroid rats. The results are expressed as a percentage of the effect of
1 11M ISO in the same atrial preparations. n = 5--6 in both group. *-p <
0.05 compared to euthyroid atria. Methods: Left atria from euthyroid and
hyperthyroid rats were prepared and adapted as described in Fig. 1.
Thereafter the rolipram was cumulatively added in increasing concentrations,
from 2-20 11M. This procedure was followed by addition of siguazodan,
from 10-20 11M. The contractile parameters were registered after muscle
stabilization at each of the concentrations of added drugs.
even when such an inhibition had no effect on basal contract-
ility [32, 33]. Thus, one may speculate that due to decreased
total activity of PDE, the i/c cAMP concentration remains
higher at any extracellular dose of, ~-agonist. This may lead
to increased sensitivity, as well as to increased maximal
responses of atria to ISO under hypothyroidism. However,
as indicated in Fig. 1B this is not the case. In fact, the
sensitivity of atria to ISO decreased in hypothyroidism. In
addition, the cumulative inotropic effects of rolipram and
siguazodan both in euthyroid and hypothyroid atria were
similar to that ofISO. This suggests that similar cellular levels
of cAMP were attained in the presence of ISO and PDE
inhibitors. Thus, it seems likely that decreased PDE activity
neither played a role in maximal responses to ISO nor
contributed to decreased sensitivity of atria to ISO under
hypothyroidism.
Effects of thyroid hormones and catecholamines on
intracellular Ca 2 + handling in atria
In cardiac cells, the inotropic effects of ISO are mediated
mainly by PKA [1] and, to a lower extent, by Ca2+-calmodulin-
dependent protein kinase [34, 35]. Activation of PKA by
increased cAMP levels results in enhanced i/c Ca2 +concentra-
tion, which, in turn, activates Ca2+-calmodulin-dependent
protein kinase [34, 35]. Eventually, several target proteins
become phosphorylated, that manifests in increased contract-
ility and rate of relaxation. The underlying mechanisms could
be considered as follows. Protein kinase A-dependent
phosphorylation of<XI subunit ofL-type Ca2+channel augments
the inward Ca2+ current [36], which stimulates Ca2+induced
Ca 2+ release from the SR [37]. The latter process is also
release channel (ryanodine receptor), which activates the
channel [1, 38]. Altogether, these changes ensure increased
rise and higher peak of i/c Ca2+ transient [39], that leads to
PKA phosphorylates troponin I that enhances the rate at
from its inhibition by nonphosphorylated PLB, and thereby
increases the SR Ca2+uptake rate as well as the load of this
intracellular store with Ca2+ [41]. This will also contribute
to an enhanced SR Ca2+release. The biochemically distinct
Ca2+-pump in the plasma membrane is also activated by
PKA-dependent phosphorylation [42, 43]. These changes
speed up the removal of Ca2+ ions from the cytoplasm, as
suggested by faster decay of i/c Ca 2+ transient [39], which
produces the higher rate of relaxation.
Most of the proteins involved in Ca 2+ handling are
regulated by thyroid hormones as well. They upregulate the
SR Ca2+-pump [8-11,44-46], Ca2+release channel [45,47],
 423

and sarcolemmal Ca 2+-pump [48-49], but downregulate phospholamban/SR Ca"-pump ratio

PLB [8 -11,44-46] and, in rat heart, L-type inward channels A 0.4

B
[7,49]. Recent studies have related the thyroid hormone-
dependent changes in PLB to regulation of ventricular 0.3
responsiveness to catecholamines [8-10]. We have outlined
the role of changes in PLB and SR Ca2+-pump in determining 0.2

the inotropic and lusitropic responses to ISO in rat atria [10]. J.. ,f). hypothyroid
0.1 .,0 euthyroid
Tissue expression of atrial myosin isoforms is insensitive to
_,0 hyperthyroid
thyroid hormones [13, 14, 16]. This modulatory mechanism
of contractile performance in rat ventricle can therefore not o 30 60 80 120 150 o 2 4 6 8 10
interfere with thyroid hormone-dependent changes in SR Ca2+ dT/dt (mN/s) Effect of ISO (fold of stimulation)

handling in atria. The results (Fig. 6) show that a shift from

hypothyroid to hyperthyroid state resulted in a 2.4-fold Fig. 7. Effect of thyroid state-dependent changes in phospholamban/SR
increase in levels ofSR Ca 2+-pump, but in a6.7-fold decrease Ca 2+-pump ratio on +dT/dt and --DT/dt (A), and on inotropic or lusitropic
effect ofISO (B). (Modified from ref [10]).
in PLB tissue contents, that produced a 14-fold decrease in
PLB/SR Ca2 +-pump ratio. These changes were associated
with augmented contractility and relaxation, but decreased
inotropic and lusitropic responses to ISO (Fig. 7). Figure 8
demonstrates the results of experiments that were designed
to gain insight into the role of the changes in PLB phos-
phorylation. It can be seen that while the SR Ca2+ uptake
increased 7.2 times, a 5-fold decrease in PKA-dependent
activation of that process was observed under shift from
hypothyroidism to hyperthyroidism. Thus, higher PLB
content in hypothyroid atria resulted in a larger extent of
activation of SR Ca2+-pump through PLB phosphorylation
Fig. 8. Effect of thyroid state on SR Ca 2 +-uptake (Al and its stimulation by
and consequently, greater maximal inotropic response to ISO.
12 ~g/ml PKA (B) at pCa 7.0. *-p < 0.05 compared to euthyroid atria.
In contrast, smaller amount of PLB in hyperthyroid atria PKA. (Modified from ref. [1 OJ).
results in higher basal rate of SR Ca2+-pump, but also limits
the extent of its extra activation by ISO.
In comparison with earlier studies on ventricles [9], these linking relaxation and contractility could be interpreted as
results suggest the greater regulatory importance of thyroid follows. In hypothyroid atria the SR Ca 2+-pump function is
state-dependent changes in PLB/SR Ca 2+-pump in atrial than inhibited due to the relatively high content of PLB. In
in ventricular myocardium. Moreover, relaxation and contrast, hyperthyroidism which removes such an inhibition
contractility appear to be governed by a common mechanism by downregulating PLB expression, results in maximal
- changes in PLB/SR Ca 2+-pump ratio. The role of SR in activation of SR Ca2+-pump. Hence, higher amounts of Ca2+
are sequestered by the SR which enhances the rate of
relaxation. On the other hand, more Ca 2+would be available
Q) 0 hypothyroid for release to activate contraction which explains the
C::::I
'Qi til 150 II1II hyperthyroid increased rate of tension development. In atria, which
0>
'--0
a..- cannot respond to thyroid hormones by increasing the
Ole
>>- 100 myosin ATPase activity [13, 14, 16], mechanisms leading
:;:..c:
u-
(1)::::1
to an accelerated Ca2+-induced Ca2+release from SR appears
OlOl
'-
00
.... to be of principal importance for increasing contractility.
c_ 50 Figure 9 demonstrates that thyroid hormone-dependent
::::I C
E Ol changes in SR Ca 2+-pump activity were closely associated
E~
-Ol with alterations in recirculation fraction of activator Ca 2+
S 0 (RFA) [50]. RFAis an index ofthe fractional amount ofCa2+,
which is sequestered during the diastole and released during
the subsequent systole by SR [51, 52]. Thus, much higher
Fig. 6. Effect of thyroid state on phospholamban (PLB) and SR Ca 2+-pump RFA in hyperthyroid than in hypothyroid atria indicates that
expression in rat atria. *-p < 0.05 compared to euthyroid atria. (Modified thyroid hormones increase the role of SR and decrease the
from ref. [10]). role of sarcolemma for control of activator Ca 2+. Obviously,
424

~m regulating contractility and relaxation in rat atria. It is shown

() 1.0
~
that thyroid hormones increase the ~-adrenoceptor density,
~
> 0.8
in association with increased G/G j ratio and PDE activity.
~
m These changes appear to be responsible for increased
15 0.6 sensitivity of atria to ISO. On the other hand, thyroid
c
0 hormones decrease the PLB/SR Ca 2 +-pump ratio due to
u downregulation of expression ofPLB and upregulation ofSR
c 0.4
0 Ca2+-pump. The latter changes lead to twofold consequences:
~ (i) to increased contractility and relaxation, and (ii) to
:; 0.2
~
'0 decreased inotropic and lusitropic responses to ISO.
Q)
II: 0
HYPO EU HYPO+T,
Acknowledgements
Fig. 9. Effect of thyroid state on recirculation fraction of activator Ca'+ in
rat atria. *-p < 0.05 compared to euthyroid atria. (Modified from ref. [50]). This work was supported by Estonian Science Foundation,
German Research Foundation (Ve 136/1-3), The Sigrid
such a shift in the balance between transsarcolemmal and SR Juselius Foundation, The Juho Vainio Foundation, The
routes of Ca2+ handling reflects activation of SR function Centre ofIntemational Mobility ofthe Finnish foreign office
under thyroid hormones. Higher rate of SR Ca2+uptake allows as well as INTAS-94 Grant.
SR to compete with sarcolemmal Na+-Ca2+ exchanger for
Ca 2+, and avoid extrusion ofCa2+ from the cell via Na+-Ca2+
exchange. It is known that thyroid hormones suppress the References
expression of Na+-Ca 2+ exchanger protein [53, 54] and
I. Karczewski P, Bartel S, Krause EG: Protein phosphorylation in the
decrease the density of slow inward Ca2+ channels in adult regulation of cardiac contractility and vascular smooth muscle tone.
rat myocardium [7, 49]. These changes, by limiting the CUIT Opin Nephrol Hypertens 2: 33-40, 1993
transsarcolemmal Ca2+fluxes, could favour the larger fraction 2. Bilezikian 'JP, Loeb IN: The influence of hyperthyroidism and
of Ca 2+ to be handled by SR. hypothyroidism on (J.- and ~-adrenergic receptor systems and adrenergic
In conclusion, the Fig. 10 shows the mechanisms of responsiveness. Endocr Rev 4: 378-388, 1983
3. Ishac EJ, Pennefather IN, Handberg GM: Effect of changes in thyroid
interaction of thyroid hormones and catech01amines in state alpha- and beta-adrenoceptors, adenyl ate cyclase activity, and
catecholamine levels in the atria. J Cardiovasc Pharmacol5: 396-405,
1983
4. Handberg GM: Influence of altered thyroid state on the inotropic
potency of isoproterenol-studies with isolated rat left atria paced at
different frequencies. J Cardiovasc Pharmacol 6: 943-948, 1984
5. FoxAW, Juberg EN, May JM, Johnson RO,Abel PW, Minneman KP:
Thyroid status and adrenergic receptor subtypes in the rat. J Pharmacol
Exp Ther 235: 715-723, 1985
6. Hammond HK, White FC, Buxton IL, Saltzstein P, Brunton LL,
Longhurst JC: Increased myocardial beta-receptors and adrenergic
responses in hyperthyroid pigs. Am J Physiol252: H283-H290, 1987
7. Hawthorn MH, Gengo P, Wei X, Rutledge A, Moran JF, Gallant S,
Triggle OJ: Effect of thyroid status on ~-adrenoceptors and calcium
channels in rat cardiac and vascular tissue. Naunyn-Schmiedeberg's
Arch Pharmacol337: 539--544, 1988
SARCOPLASMIC 8. Beekman RE, van Hardeveld C, Simonides S: On the mechanism of
RETICULUM the reduction by thyroid hormone of ~-adrenergic relaxation rate
CONTROL OVER
MAXIMAL
stimulation in rat heart. Biochem J 259: 229-236, 1989
RESPONSIVENESS 9. Kiss E, Jakab G, Kranias EG, Edes I: Thyroid hormone-induced
alterations in phospholamban protein expression. Regulatory effects
on sarcoplasmic reticulum Ca'+ transport and myocardial relaxation.
Fig. 10. Summary of the mechanisms of regulation of~-adrenergic signaling Circ Res 75: 245-251,1994
system in atria by thyroid hormones. AC - adenyl ate cyclase; ~AR - ~ 10. Kaasik A, Paju K,Vetter R, Seppet EK: Thyroid hormones increase
adrenergic receptor; CRC - Ca'+ release channel; G, - G, protein; G; - G; the contractility but suppress the effects of ~-agonist by decreasing
protein; POE- phosphodiesterase; PKA - cAMP-dependent protein kinase; the phospholamban expression in rat atria. Cardiovasc Res, in press
PLB - phospholamban; Tnl - Troponin I. Plus and minus signs indicate II. Banerjee SP, Kung LS: ~-Adrenergic receptors in the rat heart: Effects
the upregulation and downregulation, respectively, of the given protein of thyroidectomy. Eur J Pharmacol43: 207-208, 1977
levels by thyroid hormones. 12. Ciaraldi T, Marinetti GV: Thyroxine and propylthiouracil effects of
vivo on alpha and beta adrenergic receptors in rat heart. Biochem

Biophys Res Commun 74: 984-991, 1977
13. Banerjee SK: Comparative studies of atrial and ventricular myosin
from normal, thyrotoxic, and thyroidectomized rabbits. Circ Res 52:
131-136, 1983
14. Samuel J-L, Rappaport L, Syrovy I, Wisnewsky C, Marotte F, Whalen
RG, Swartz K: Differential effect of thyroxine on atrial and ventricular
myosin in rats. Am J Physiol 250: H333-H341, 1986
15. Rupp H, Berger H-J, PfeiferA, Werdan K: Effect of positive inotropic
agents on myosin isoenzyme population and mechanical activity of
cultured rat heart myocytes. Circ Res 68: 1164-1173, 1991
16. Bottinelli R., Canepari M, Cappelli V, Reggiani C: Maximum speed
of shortening and ATPase activity in atrial and ventricular myocardia
of hyperthyroid rats. Am J Physiol269: C785-C790, 1995
17. Nakashima M, Maeda K, Sekiya A, Hagino Y: Effect of hypothyroid
status on myocardial responses to sympathomimetic drugs. Jpn J
Pharmacol21: 819-825, 1971
18. Brodde O-E, Schumann H-J, Wagner J: Decreased responsiveness of
the adenylate cyclase system on left atria from hypothyroid rats. Mol
Pharmacol17: 180--186, 1980
19. Kunos G: Modulation of adrenergic reactivity and adrenoceptors by
thyroid hormones. In: G Konus (ed). Adrenoceptors and Catecholamine
Action. New York: John Wiley and Sons, Inc. 287-333,1981
20. Kaasik A, Seppet EK, Ohisalo JJ: Enhanced negative inotropic effect
of an adenosine AI-receptor agonist in rat left atria in hypothyroidism.
J Mol Cell Cardiol26: 509-517,1994
21. Bahouth SW: Thyroid hormones transcriptionally regulate the ~ 1-
adrenergic receptor gene in cultured ventricular myocytes. J Bioi Chern
266: 15863-15869, 1991
22. Gilman AG: G proteins: Transducers of receptor-generated signals.
Annu Rev Biochem 56: 615--649, 1987
23. Levine MA, Feldman AM, Robishaw JD, Ladenson PW, Ahn TG,
Moroney JF, Smallwood PM: Influence of thyroid hormone status on
expression of genes encoding G-protein subunits in the rat heart. J
Bioi Chern 265: 3553-3560,1990
24. Michel-Reher M, Gross G, Jasper JR, Bernstein D, OlbrichtT, Brodde
O-E, Michel MC: Tissue- and subunit regulation of G-protein
expression by hypo- and hyperthyroidism. Biochem Pharmacol 45:
1417-1423,1993
25. Bahouth SW: Thyroid hormone regulation of transmembrane signalling
in neonatal rat ventricular myocytes by selective alteration of the
expression and coupling of G-protein alfa-subunits. Biochem J 307:
831-841,1995
26. Sundaresan PR, Banerjee SP: Differential regulation of beta-adrenergic
receptor-coupled adenylate cyclase by thyroid hormones in rat liver and
heart: Possible role of corticosteroids. Hormone Res 27: 109-118, 1987
27. Hohl CM, Wetzel S, Fertel RH, Wimsatt DK, Brierley GP, Altschuld
RA: Hyperthyroid adult rat cardiomyocytes. I. Nucleotide content, ~
and a-adrenoreceptors, and cAMP production. Am J Physiol 257:
C948-C956,1989
28. Kragie L, Kwon YW, Smiehorowski R: Rat cardiac calcium channels
and their relationships with beta-adrenergic and muscarinic receptors
in hypothyroidism. Endocr Res 19: 57-71, 1993
29. Malbon CC, Moreno FJ, Cabelli RJ, Fain IN: Fat cell adenyl ate cyclase
and ~-adrenergic receptors in altered thyroid states. J Bioi Chern 253:
671--678, 1978
30. KaasikA, Elomaa V, Seppet EK, Ohisalo JJ: Low particulate type IV
phosphodiesterase activity in hypothyroid rat atria. J Mol Cell Cardiol
26: 1587-1592,1994
31. Kithas PA,Artman M, Thompson WJ, Strada SJ: Subcellular distribution
of high-affinity type IV cyclic AMP phosphodiesterase activity in rabbit
ventricular myocardium: Relations to the effects of cardiotonic drugs.
Circ Res 62: 782-789, 1988.
32. Gristwood RW, English TAH, Wallwork J, Sampford KA, Owen DAA:
425
Analysis of responses to a selective phosphodiesterase III inhibitor,
SK&F 94120, on isolated myocardium, including human ventricular
myocardium from 'end-stage' failure patients. J Cardiovasc Pharm 9:
719-727, 1987
33. Katano Y, Endoh M: Effects of a cardiotonic quinolinone derivate Y-
20487 on the isoproterenol-induced positive inotropic action and cyclic
AMP accumulation in rat ventricular myocardium: Comparison with
rolipram, Ro 20-1724, milrinone, and isobutylmethylxanthine. J
Cardiovasc Pharm 20: 715-722, 1992
34. Lindemann JP, Watanabe AM: Phosphorylation of phospho lamb an in
intact myocardium. Role of Ca2+ -calmodulin-dependent mechanisms.
J BioI Chern 260: 4516-4525,1985
35. Vittone L, Mundifta C, Chiappe de Cingolani G, Mattiazi A: Role of
Ca2+-calmodulin dependent phospholamban phosphorylation on the
relaxant effect of ~-adrenergic agonists. Mol Cell Biochem 124: 33-
42, 1993
36. Reuter H: Calcium channel modulation by neurotransmitters, enzymes,
and drugs. Nature 301: 569-574,1983
37. Fabiato A: Calcium-induced release of calcium from the cardiac
sarcoplasmic reticulum. Am J Physiol245: CI-C14, 1983
38. Valdivia HH, Kaplan JH, Ellis-Davies GC, Lederer WJ: Rapid
adaptation of cardiac ryanodine receptors: Modulation by Mg2+ and
phosphorylation. Science 267: 1997-2000, 1995
39. Morgan JP, Blinks JR: Intracellular Ca2+ transients in the cat papillary
muscle. Can J Physiol Pharmacol 60: 520--528, 1982
40. Solaro RJ, Moir AJ, Perry SV: Phosphorylation oftroponin I and the
inotropic effect of adrenaline in the perfused rabbit heart. Nature 262:
615--617,1976
41. Colyer J: Control of the calcium pump of cardiac sarcoplasmic
reticulum. A specific role for the pentameric structure of phos-
pholamban. Cardiovasc Res 27: 1766-1771, 1993
42. Vetter R, Haase H, Will H: Potentiating effect of calmodulin and cyclic
AMP-dependent protein kinase on ATP-dependent Ca2+-transport by
cardiac sarcolemma. FEBS Lett 148: 326-330,1982
43. Dixon DA, Haynes DH: Kinetic characterization of the Ca 2+-pumping
ATPase of cardiac sarcolemma in four states of activation. J Bioi Chern
264: 13612-13622, 1989
44. Nagai R., Zarain-HerzbergA, Brandl CJ, Fujii J, Tada MM, MacLennan
DH, Alpert NR, Periasamy M: Regulation of myocardial Ca2+ -ATPase
and phospholamban mRNA expression in response to pressure
overload and thyroid hormone. Proc Natl Acad Sci USA 86: 2966-
2970, 1989
45. Arai M, Otsu K, MacLennan DH, Alpert NR, Periasamy M: Effect of
thyroid hormone on the expression ofmRNA encoding sarcoplasmic
reticulum proteins. Circ Res 69: 266-276,1991
46. Kimura Y, Otsu K, Nishida K, Kuzuya T, Tada M: Thyroid hormone
enhances Ca 2 +pumping activity of the cardiac sarcoplasmic reticulum
by increasing Ca2+ATPase and decreasing phospholamban expression.
J Mol Cell Cardiol26: 1145-1154, 1994
47. Wibo M, Kolar F, Zheng L, Godfraind T: Influence of thyroid status
on postnatal maturation of calcium channels, beta-adrenoceptors and
cation transport ATPases in rat ventricular tissue. J Mol Cell Cardiol
27: 1731-1743, 1995
48. Seppet EK, Kadaja LY, Hata T, Kallikorm AP, Saks VA, Vetter R,
Dhalla NS: Thyroid control over membrane processes in rat heart.
Am J Physiol26: 66-71, 1991
49. Seppet EK, Kolar F, Dixon IMC, Hata T, Dhalla NS: Regulation of
cardiac sarcolemmal Ca2+ channels and Ca2+ transporters by thyroid
hormone. Mol Cell Biochem 129: 145-159, 1993
50. Kaasik A, Minajeva A, Paju K, Eimre M, Seppet EK: Thyroid
hormones differentially affect sarcoplasmic reticulum function in rat
atria and ventricles. Mol Cell Biochem, in press
51. Morad M, GoldmanY: Excitation-contraction coupling in heart muscle:
426
Membrane control of development of tension. Prog Biophys Mol Bioi
27:257-313,1975
52. Urthaler F, Walker AA, Reeves DNS, Hetner LL: Maximal twitch
tension in intact length-clamped ferret papillary muscles evoked
by modified postextrasystolic potentiation. Circ Res 62: 65-74,
1988
53. Magyar CE, Wang 1, Azuma KK, McDonough: Reciprocal regulation
of cardiac Na-K-ATPase and Na/Ca exchanger: Hypertension, thyroid
hormone, development. Am 1 Physiol269: C675-682, 1995
54. Boerth SR, Artman M: Thyroid hormone regulates Na+/Ca2 +exchanger
expression during postnatal maturation and in adult rabbit ventricular
myocardium. Cardiovasc Res 31: E 145-152, 1996
Molecular and Cellular Biochemistry 184: 427-437, 1998.
1998 Kluwer Academic Publishers.

Creatine supplementation in health and disease.

Effects of chronic creatine ingestion in vivo: Down-
regulation of the expression of creatine transporter
isoforms in skeletal muscle
Maria Lourdes Guerrero-Ontiveros and Theo Wallirnann
Institute for Cell Biology, Swiss Federal Institute of Technology, ETH-Honggerberg, CH-8093 Zurich, Switzerland

Abstract
Interest in creatine (Cr) as a nutritional supplement and ergogenic aid for athletes has surged over recent years. After cellular
uptake, Cr is phosphorylated to phosphocreatine (PCr) by the creatine kinase (CK) reaction using ATP. At subcellular sites
with high energy requirements, e.g. at the myofibrillar apparatus during muscle contraction, CK catalyzes the trans-
phosphorylation of PCr to ADP to regenerate ATP, thus preventing a depletion of ATP levels. PCr is thus available as an
immediate energy source, serving not only as an energy buffer but also as an energy transport vehicle. Ingestion of creatine
increases intramuscular Cr, as well as PCr concentrations, and leads to exercise enhancement, especially in sprint performance.
Additional benefits of Cr supplementation have also been noticed for high-intensity long-endurance tasks, e.g. shortening of
recovery periods after physical exercise.
The present article summarizes recent findings on the influence of Cr supplementation on energy metabolism, and introduces
the Cr transporter protein (CreaT), responsible for uptake of Cr into cells, as one of the key-players for the multi-faceted
regulation of cellular Cr homeostasis. Furthermore, it is suggested that patients with disturbances in Cr metabolism or with
different neuro-muscular diseases may benefit from Cr supplementation as an adjuvant therapy to relieve or delay the onset
of symptoms. Although it is still unclear how Cr biosynthesis and transport are regulated in health and disease, so far there
are no reports of harmful side effects ofCr loading in humans. However, in this study, we report that chronic Cr supplementation
in rats down-regulates in vivo the expression of the CreaT. In addition, we describe the presence of CreaT isoforms most
likely generated by alternative splicing. (Mol Cell Biochem184: 427-437,1998)

Key words: creatine supplementation, creatine transporter, creatine transporter isoenzymes, differential splicing, creatine
kinase, creatine metabolism, muscle diseases, bioenergetics, phosphocreatine

Abbreviations: Cr - creatine; PCr - phosphocreatine; CK - creatine kinase; CreaT - creatine transporter or carrier; Crn-
creatinine; ~-GPA=GPA - 3-guanidinopropionate; ~-GBA - 4-guanidinobutyrate; GABA - guanidino y-aminobutyric acid;
IGF-l - insulin-like growth factor I; CHO - carbohydrate; Mi-CK - mitochondrial CK isoenzyme

Introduction which is present in muscle and brain. The estimated daily

Creatine and creatine kinase
Creatine (Cr) is a naturally occurring guanidino compound
first described in 1832 by Chevreul [1]. For a 70 kg man,
the total body content ofCr is approximately 120 g, most of
requirement ofCr for an average individual is about 2 g [2].
Cr is ingested with food and is also endogenously synthe-
sized from arginine, glycine and methionine in the liver,
pancreas and, in humans, primarily in the kidney [2, 3]. After
its synthesis, Cr is released into the blood stream, from where
it is taken up by cells via the creatine transporter (CreaT) [4].

Address/or offprints: T. Wallimann, Institute for Cell Biology, Swiss Federal Institute ofTechnology, ETH-Honggerberg, CH-8-93 Zurich, Switzerlands
428
Without a continuous supply of muscle cells with creatine,
either synthesized by the body or supplemented externally,
Cr and thus also phosphocreatine (PCr) would be depleted in
muscle by continuous irreversible non-enzymatic conversion
of Cr into creatinine (Crn), the latter being released from
muscle cells and excreted by the kidney [5, 6] (for review see
Walker 1979). The total intracellular Cr concentration [PCr +
Crfree ] is an essential component of energy metabolism of
skeletal, cardiac and smooth muscle and brain, since it
determines the concentration of available PCr [7]. PCr serves
as a dynamic reservoir of 'high-energy' phosphate during
rapid utilization of ATP, delays and dampens fluctuations in
adenosine nucleotides, modulates glycolysis and oxidative
phosphorylation, and, through the Cr/PCr interconversion by
the CK reaction (PCr2- + MgADpl- + H+ f-7 MgATp2- + Cr),
also buffers intracellular changes in pH and, in conjunction
with ATP hydrolysis, influences inorganic phosphate levels
(PCr + ADP -t ATP + Cr -t ADP + P., net PCr breakdown
followed by hydrolysis of the ATP formed leads to an
I Introduction) are indicated by filled circles. The location oftwo consensus
increase inP) (for reviews see [8, 9]).
The creatine transporter
A saturable sodium- and chloride dependent, high affinity
uptake process, which moves Cr against a concentration
gradient, has been studied in a number of tissues and cell types
[4, 10--13]. The CreaT has recently been cloned [14-16].
Based on sequence homology, it belongs to the y-amino-
butyric acid/norepinephrine (GATINET) transporter gene
family. The members of this family are Na+ and Cl--dependent
twelve transmembrane-helix-spanning transporters res-
ponsible for the uptake of certain neurotransmitters (e.g.
dopamine, guanidinoy-aminobutyric acid (GABA), serotonin
and norepinephrine), and amino acids (e.g. glycine and
taurine) [15-18] . Northern blot analysis indicated the
expression ofCreaT mRNAin kidney, heart, skeletal muscle,
brain, testis and colon, but not in liver, pancreas, spleen,
uterus or intestine [15,16,19,20], essentially reproducing
the expression pattern ofCK (see [21]).In situ hybridization
showed high CreaT transcript levels in cerebellum, hippo-
campus and other regions of the brain, as well as in muscle
tissues [17]. From analysis ofthe CreaT cDNA one predicts
an Asn-linked plasma membrane glycoprotein of 635 amino
acids (70.5 KDa) with an isoelectric point of 6.38 and five
putative phosphorylation sites [14] (Fig. 1).
The CreaT gene (SLC6A8 solute carrier class 6, member
8) was mapped to human chromosome Xq28 telomeric to
G6PD [22]. Iyer et al. [23] confirmed these findings, and
additionally showed the existence of a novel autosomal,
testis-specific form of the human CreaT gene located on
chromosome 16p 11.2. The Xq28 locus has been linked to
the genes for several neuromuscular disorders such as Emery-
EXTERNAL
Fig. 1. Schematic representation of the creatine transporter demonstrating
its hypothetical orientation within the plasma membrane. Amino acids which
are conserved among members of the GATINET family (Na+- and Cl--
dependent, twelve transmembrane spanning domain transporters; see
N-linked glycosylation sites are marked with stylized drawings of a
polysaccharide side chain. The canonical phosphorylation sites (circled P)
are also indicated. The figure was adopted from Nash et af. [16].
Dreifuss muscular distrophy [24, 25], Barth syndrome [26,
27], infantile cardiomyopathy [28], and myotubular myo-
pathy [29]. Since CreaT plays an important role in Cr
homeostasis and muscular physiology, it was postulated as
a potential gene responsible for one of these muscular
disorders.
Regulation and inhibition of the creatine transporter
Several types of animal cells from heart, brain, skeletal and
smooth muscle, adipose tissue, as well as fibroblasts, and
macrophages in isolated tissues or in culture take up Cr [4,
10--13,30]. These cells transport Cr against a concentration
gradient (200: 1 in skeletal muscle) [31] through a saturable,
sodium dependent process, whose Km value varies between
25-110 11M [4, 10--13, 15-17, 32]. Under physiological
conditions, the CreaT is working at about its half maximal
rate since resting levels of Cr in plasma amount to 25-100
11M only [33, 34]. The sodium Cr cotransporter uses the
energy of the Na+ gradient and the membrane potential
maintained by the Na+,K+-ATPase, which itself depends on
the function of CK being in close vicinity of and functionally
coupled to this ion pump [35]. Kinetic data of the Na+
dependence of Cr transport in glial cells suggest a stoichio-
metry of 2 Na+ per Cr transported [13].
The total cellular Cr content depends on the balance of
uptake, retention and efflux of Cr. However, the mechanisms
by which cells regulate their intracellular Cr stores are

poorly understood. A variety of potential regulatory mechan-
isms have been suggested (for an extensive review see [2]).
One of the identified sites for regulation of intra- and
extracellular Cr occurs at the level of Cr biosynthesis in the
liver, pancreas and kidney. Here, Cr exerts a negative
feedback control on the arginine glycine amidinotransferase
(AGAT) [36], which is the enzyme catalyzing the first step
in de novo Cr biosynthesis, the formation of guanidino-
acetate from L-arginine and glycine.
Cr homeostasis is also regulated through control of
CreaT expression and activity. The saturable component of
Cr uptake is affected by Cr structural analogs and metabolic
inhibitors [4, 10]. The sensitivity of the CreaT to several
guanidinocarboxylic acids such as 3-guanidinopropionate (~
GPA), 2-amino-3-guanidinopropionate, 4-guanidinobutyrate
(~-GBA) and guanidinoacetate, has been widely investigated
[4, 10-13, IS, 16,20,32]. Other guanidino compounds like
guanidinosuccinate, GAB A, taurine and taurocyamine (or
~-guanidinoethanesulfonic acid) as well as several amino
acids (e.g. ~-alanine, glycine, arginine), and Cr derivatives
(PCr and creatinine) were also tested for their ability to
compete for Cr at the uptake site. The most efficient
blocking agent in all mammalian Cr transport systems studied
so far ([10-13, IS, 20] is ~-GPA, which acts as a competitive
inhibitor of Cr uptake systems and also of CK [4]. Inhibition
of Cr entry by ~-GPA varies from 45-96% in the different
cells and tissues measured and the inhibition constant values
range between 1S--4S0 J.lM depending on the tissue [4,10-
13, 30]. Other compounds exerting weaker inhibition on
CreaT include y-guanidinbutyrate and guanidinoacetate in
Torpedo and humans, N-ethylguanidinoacetic acid or
guanidinoacetate in rat skeletal muscle, and taurocyamine
in astroglia cells [4, 13, 16, 20, 32]. The competing
guanidino compounds as well as Cr interact reversibly with
the transporter protein. Compounds lacking an ami dine
group fail to inhibit Cr transport. Certain mono- and N,N-
disubstituted guanidines can compete with higher affinity
for uptake [4].
Oddom et al. [37] have identified a series of hormones
which influence net Cr uptake into a skeletal muscle cell
line. They have shown that catecholamines (noradrenalin,
isoproterenol and clenbuterol) can stimulate net Cr uptake
preferentially through ~2 receptors, probably via a cyclic
AMP-dependent mechanism. In addition, they demon-
strated that insulin (at supraphysiological concentrations)
and IGF-I can also stimulate net Cr uptake. Furthermore,
they have identified a modulation of total Cr content by
ouabain that inhibits - as well as by agonists that stimulate
the Na+,K+-ATPase.
429
Creatine supplementation comes of age: creatine
loading in man
Although studies with creatine supplementation have been
already performed at the end of the nineteenth century [1], (for
references see [34]), it would appear that only recently the
implications of Cr metabolism in bioenergetics, physiology and
human pathology are attracting considerable attention and are
actively being investigated in many research groups.
A century ago, studies with dietary Cr intake performed in
both humans and animals concluded that some ofthe ingested
Cr was retained in the body [1]. Recent studies have shown
that administration ofdifferent Cr doses for variable time periods
(typically 20 g daily for S days) result in an =20% (20 mmol/kg
dry muscle) increase in total muscle Cr. As a consequence, Cr
loading has a positive effect on exercise performance during
single and repeated bouts of exhaustive, short-lasting exercise
and increases the rate ofPCr resynthesis [7, 38--41].
The extent of Cr retention during Cr supplementation is
highly variable between subjects. This variations suggested
that Cr uptake is dependent on several factors including
differences in the composition of the diet, gender, muscle
fiber composition and on the initial total muscle Cr con-
centration 13, 31, 41]. The positive effect of Cr loading on
PCr resynthesis and improvement in exercise performance,
however, is basically only observed in those individuals who
show an increase of more than 20 mmol/kg dry muscle in
total Cr concentration upon dietary Cr-Ioading [42].
Improvement of the accumulation of creatine in muscle
Koszalka and Andrew [43] as well as Haugland and Chang [44]
reported the effect of insulin on Cr transport in rat skeletal
muscle. They have found that insulin enhances the transfer of
Cr from the circulation into skeletal muscle. Based on this result,
Greenet al. [4S] investigated the effect of carbohydrate (CHO)
ingestion on muscle Cr accumulation during Cr supple-
mentation in humans. Administration of both Cr and CHO
resulted in an increase of the skeletal muscle Cr retention. This
appeared to be mediated by insulin, since upon Cr supple-
mentation in conjunction with CHO, the concentration of
insulin augmented considerably compared to dietary Cr loading
alone. Insulin stimulates the Na+,K+-ATPase, in turn promoting
the Na+ -Cr-cotransport by maintaining or restoring the normal
transmembrane N a+ gradient and membrane potential [19, 37,
46, 47]. These findings suggested a potential mechanism for
optimising muscle Cr accumulation in humans, which can have
important implications for improving exercise performance and
for the treatment of some neuromuscular diseases related to
deficiencies in Cr metabolism and/or chronically lowered Cr
levels.
Another attempt to improve muscle Cr trapping was made
430
by Vandenberghe et al. [48]. These authors expected that
oral Cr uptake, combined with a physiological degree of
adrenergic stimulation by caffeine, might facilitate muscle
Cr accumulation. Caffeine has been shown to directly.
stimulate muscle Na+,K+-ATPase activity and to enhance
plasma epinephrine levels, another direct stimulus of the
muscle Na+,K+-pump [19, 37,46-48]. Surprisingly, however,
their results demonstrated that caffeine does not improve
the efficiency of oral Cr supplementation, neither increasing
muscle PCr levels nor improving exercise performance.
Caffeine fully abolished the ergogenic effect of muscle Cr
loading. Additional work is needed on the effects of Cr
intake together with caffeine on the systems that regulate
muscle energetics and improve performance during inter-
mittent exercise, e.g. the PCr-CK energy shuttle between
energy providing and energy consuming sites in the cell [8].
Creatine supplementation and exercise performance
Cr loading improves the ability to maintain power output
during exhaustive, high intensity tasks, especially when
repeated exercise bouts are carried out [3, 38-41, 49-52].
This appears to be due to a rise in the total Cr concentration
and thus to an increase in pre-exercise PCr availability and
to an increasing rate of PCr resynthesis during episodes of
muscle relaxation. In addition, an enhanced fatigue resistance
is also observed [39-41, 49, 50].
Some results related to metabolic response after Cr
supplementation are controversial. It appears, however, that
post-exercise plasma ammonia and hypoxanthine con-
centration [40, 41], and in some cases muscle lactate [49,
53], are lowered after ingestion of creatine. Low levels of
ammonia may indicate that enhanced Cr and PCr con-
centrations prevent a loss of adenine nucleotides via
adenylate kinase and AMP-deaminase, by keeping, via the CK
reaction, [ADP] and [AMP] low (see [8]).
For long endurance exercise, however, one cannot yet
clearly rationalize how a higher concentration of total Cr
could improve exercise performance. The enhancement of
intramuscular Cr and PCr levels upon dietary Cr intake
causes an increase in anaerobic capacity [41]. In addition,
increased Cr stimulates oxidative phosphorylation via
activation of mitochondrial creatine kinase (Mi-CK) [9].
Phosphorylation of Cr through Mi-CK increases the ADP
levels. Green et al. [45] reported higher ADP concentration
following Cr supplementation. ThisADP, in turn, could enter
into the mitochondrial matrix through the adenine nucleotide
translocator and facilitate ATP resynthesis by stimulating
mitochondrial respiration [45]. By this line of arguments,
one could explain the advantages provided by Cr ingestion
for enhancing aerobic output in a continuously working
muscle. In addition, although there is no hard evidence yet,
chronic elevation of Cr levels could lead to alterations in
expression and/or activity of enzymes such as phospho-
fructokinase, hexokinase or adenylate kinase, possibly in an
opposite way than observed with the Cr analogue ~-GPA
[54]. Cr may also delay fatigue by the higher availability and
rate of PCr resynthesis observed in athletes after ingestion
of Cr [50, 53].
Creatine levels in disease
Extensive studies on the PCr-CK system and its involvement
in regulation of energy metabolism have been performed
in the past. In contrast, only little is known about Cr
metabolism in disease, e.g. whether there are alterations in
the regulation of Cr synthesis, uptake or retention in
different tissues and in the whole organism compared to the
healthy state [55]. Consequently, research concerned with
the effects of Cr ingestion on disease is also scarce. It has
been demonstrated that several pathologic states such as
ischemia, anoxia, hypoxia, toxic cardiomyopathies, muscular
dystrophies or atrophies and mitochondrial myopathies are
related to energy deficiencies [56-60]. Some muscle
diseases were shown to be associated with disturbances in
Cr metabolism, low intracellular levels of Cr and phospho-
creatine and derangements of the CK system. In this section,
we shortly describe some reports related to the effects of
Cr supplementation both on animal and human disease states
and its potential role as an adjuvant therapeutic aid.
Recently, Cr treatment was tried with a patient with
MELAS, a maternally inherited disease characterized by
mitochondrial myopathy, encephalopathy with seizures, and!
or dementia, lactic acidosis, and stroke-like episodes [61],
who was given Cr in addition to his normal therapy [62].
His symptoms included exercise-induced muscle pain,
headache, chronic lactic acidosis, and episodes of cortical
blindness, visual hallucinations, dysphasia and arm paresis.
Cr was given orally at 1 g daily for 2 weeks and 4 g daily,
thereafter. The patient reported reduced headache, less
weakness, better appetite, and an improved work performance
during Cr treatment. Interestingly enough, a connection of CK
with mitochondrial myopathies has been clearly elaborated
in that mitochondrial CK (Mi-CK) has been shown to be the
major component of the crystalline intramitochondrial
inclusions seen in patients with these diseases [63], indicating
that a lowering of the cellular energy status, e.g. by a defect
in enzymes of oxidative phosphorylation or by Cr depletion,
leads to compensatory overexpression and finally a crystalline
deposition of Mi-CK between mitochondrial membranes
[54, 64, 85, 86].
As a second case, gyrate atrophy of the choroid and
retina is an autosomal recessive dystrophy characterized
by constriction of visual field, myopia, and posterior cataract

and total blindness. Concomitantly with the eye disease,
clinically mild but marked morphological atrophy of type II
skeletal muscle fibres is observed [65, 66]. Patients affected
by gyrate atrophy have 10-20 fold increased ornithine
concentrations in body fluids and significantly reduced
activity of ornithine aminotransferase [67-69]. High
ornithine concentrations inhibit arginine-glycine trans-
amidinase, the rate limiting enzyme in Cr biosynthesis. The
latter leads to deficient guanidinoacetate formation and
subsequent reduction of Cr and PCr production. Deficiency
of 'high-energy' phosphates cause the atrophy in the muscles
and eyes. Sipila et al. [70] treated patients suffering gyrate
atrophy with 1.5 g Cr daily during a year. The treatment
showed a reduction of the type II muscle fiber atrophy but
the chorioretinal degeneration continued. They suggested
that the latter might be due to poor entry of Cr into the
retina [66].
Additionally, StockIer et al. [59] described a case in which
. extrapyramidal movement disorder and extremely low
creatinine concentrations in serum and urine were observed.
This was caused by a generalized depletion ofCr in the brain.
This patient with an inborn error of Cr biosynthesis at the
level of guanidinoacetatemethyltransferase (GAMT), was
treated with Cr-monohydrate. Oral substitution of Cr-
monohydrate led to a significant increase of brain Cr, a
decrease of brain guanidinoacetate, a normalization of
creatinine in serum and urine, and a decrease in plasma
ammonia. Partial restoration of cerebral Cr concentrations
was accompanied by improvement of the patient's neurologic
symptoms [59].
Finally, based on reports on the amazing effects of
creatine supplementation on exercise performance in top
athletes, a growing number of muscle patients, especially in
Germany, started to take creatine as self-medication. Some
of these patients with different neuromuscular disorders
reported a marked improvement of their subjective status as
well as of objective parameters (see [71]); One of us (T.w.)
receives encouraging letters from a growing number of
different patients taking creatine (letters may be obtained
on request [87, 88]). This movement from the basis prompted
physicians to take creatine supplementation seriously and
to start with controlled studies [62, 89].
In an animal experiment, the effects of anoxia and hypoxia
on synaptic transmission, protein synthesis and levels of
energy metabolites were investigated in rat brain or guinea
pig hyppocampal slices. Anoxia or hypoxia were found to
abolish electrical activity as a result of membrane de-
polarization caused by a decline in neuronal energy-rich
metabolites PCr and ATP and a subsequent increase in
intracellular Ca2+ [58, 72]. Preincubation of hippocampal
slices with Cr before exposing them to anoxia and allowing
recovery increased the levels of free- and phosphorylated
Cr, slowed down the rate of decline in ATP and, in parallel,
431
enhanced electrophysiological recovery, as well as protein
synthesis [58].
The reported findings suggest that Cr is a substance with
therapeutic properties. If it is administered together with
established treatments for a specific disease, significant
improvements in a growing number of patients with different
diseases affecting cellular energetics in one way or another
may be achieved (for additional information and analysis of
the therapeutic role of Cr and PCr see the recent book by
Conway and Clark [73]. We propose that the possibility of
Cr treatment as an adjuvant therapy for a whole variety of
diseases of the central and peripheral nervous system, the
skeletomuscular and cardiovascular systems should be
seriously considered and studied in double-blinded tests in
the future.
A potential problem of chronic Cr supplementation,
however, could be a downregulation ofthe CreaT expression
which would antagonize such a treatment. To study the role
of CreaT in the regulation of intracellular creatine levels
and to answer the question of whether CreaT synthesis is
influenced by long-term creatine ingestion, as well as by the
Cr analogue 3-GPA, we decided to use an immunochemical
approach rising specific antibodies against the CreaT protein.
Despite the homologous transmembrane topology which the
CreaT shares with other neurotransmitter transporters, its
N- and C-terminal regions are clearly divergent and unique.
Therefore, we have chosen the latter amino acid sequences
as immunogens for the production of specific anti-CreaT
antibodies as a tool for the detection and quantification of
CreaT in different tissues under different experimental
conditions. The antibodies against synthetic peptides (15-
mers) corresponding to the N- and C-terminal regions ofthe
cDNA-derived CreaT amino acid sequence reacted on
Western blots with two distinct polypeptides of equal
prevalence in all CreaT positive tissues (see Results).
The study of the CreaT could help us to understand its
role in Cr metabolism, its regulation upon chronic Cr
supplementation, as well as its potential involvement in a
variety of neuromuscular diseases.
Materials and methods
Generation of antipeptide antibodies against C- and N-
terminal peptides of the creatine transporter
The l5-mer amino acid residue peptides corresponding to
the NH 2- and COOH-terminal sequences, 1-15 and 621-
635 respectively, (NH2-M-A-K-K-S-A-E-N-G-I-Y-S-V-S-
G; P-V-S-E-S-S-K-V-V-V-V-E-S-V-M-COOH) of the
creatine transporter were synthesized by ANAWA Trading
SA, Zurich according to standard solid phase procedures,
followed by HPLC purification. The CreaT peptides were
432
coupled separately to keyhole limpet hemocyanin as follows
[74]: hemocyanin (80 Ill, 100 mg/ml) in 50% glycerol was
mixed with 420 III of buffer A (140 mM NaCI, 1.6 mM KCI,
1.1 mM KHlO 4' 8 mM Na2HPO 4' pH 7.4) and supplemented
with 20 III of a freshly prepared solution of 0.1 M succinimidyl
suberate. Then, 6.3 mg of solid peptide was added, the mixture
stirred, and finally diluted with 500 III of buffer A. The material
was divided into 160 III aliquots and stored at-20cC. Rabbits
were injected intracutaneously with a mixture of 160 III of
coupled peptide, 340 III of buffer A and 500 III of either
Freund's complete adjuvant (day 0) or incomplete adjuvant
(days 15,29,41,66,98 and 170). Sera were analysed by
Western blots (see below) and stored at -20cC.
Preparation of rat skeletal muscle extracts
Skeletal muscle tissue (Quadriceps muscle) from rats fed
creatine, 3-GPA, as well as control animals fed a creatine-
free diet, was excised and proteins were extracted by hypo-
osmotic swelling of the tissue followed by phosphate
extraction at high pH. In short, the tissue was minced,
exposed to 2-3 vols. of bidistilled water for 15 min, then
incubated with 2-3 vols of phosphate buffer (50 mM
NaHl04, 10 mM 2-mercaptoethanol, plus 0.05-1 % Triton
X-I00 (depending on the tissues), freshly added, at pH>
8.75) for another 60-90 min, and finally centrifuged at
10,000 x g for 10 min. The supernatant served as a tissue
extract and was kept at -20 ce. The protein concentration was
determined by the method of Bradford [75] using bovine
serum albumin as a standard.
Electrophoretic techniques and immunoblotting
SDS polyacrylamide gel electrophoresis (PAGE) was
performed according to Laemmli [76] on an 8% gel, with
40 Ilg of tissue extract proteins being loaded per lane. After
electrophoresis, separated proteins were either stained
with Coomasie Blue or semy-dry blotted onto nitro-
cellulose paper (Schleicher and Schull, Dassel Germany).
Unspecific sites were blocked by a solution containing 3%
fat-free milk powder in phosphate-buffered saline. Papers
were labelled with polyclonal anti-C- or anti-N-termini
CreaT antibodies (at 1:500 dilution in blocking solution)
for 2-3 hat ncc. After 3 washes with the above blocking
buffer, the membranes were incubated with the secondary
antibody (goat anti-rabbit IgG-conjugated with horseradish
peroxidase and diluted 1:2000 in blocking buffer). For
detection, the peroxidase reaction was carried out using
luminol (2.5 mM luminol, 0.5 mM p-iodophenol, 50 mM
Tris-HCl, pH 7.5, and 0.15% HP) [77] and exposure to
X-ray film for 1-10 sec.
Materials and animal care and feeding
p-GPA was synthesised in the laboratory using the method
of Rowley et al. [78]. Purity was checked by TLC and lH_
NMR. Raw materials were purchased from Fluka. Creatine
monohydrate was kindly provided by Chemie Linz, Austria.
Female Sprague-Dawley rats were fed at three weeks of age
either on control diet free of creatine, on a diet containing
4% creatine plus 50 mM creatine in their water supply, or a
diet containing 2.5% p-GPA plus 1% p-GPA in water, for
3-6 months. Rats were killed by exposure to CO 2 gas and
then exsanguinated.
Results
Identification of two highly related creatine transporter
isoforms
We have produced specific antibodies against synthetic
peptides corresponding to the amino acid sequence of 15
residues at either the N- or C-termini of the cDNA-derived
CreaT amino acid sequence. Two proteins with a molecular
mass of70 and 55 kDa were consistently recognized by any
of the above antipeptide antibodies (see Fig. 2). As one can
see in the same figure (control sample lane I of B), the 70
and 55 kDa proteins are coexpressed nearly at a I: I ratio in
skeletal muscle tissue extracts as well as in a number of
other tissues. Both polypeptides are minor and almost
ubiquitously expressed in all the rat tissues tested so far
(data not shown). The highest levels of expression of these
two proteins occur in rat kidney, heart, brain, and skeletal
muscle extracts. This agrees with the distribution ofmRNA
in rat and human tissues [15, 16,20]. Based on the cDNA
analyses, the CreaT has an expected molecular weight of70
kDa. Since antibodies against the N - as well as the C-terminal
always reacted with the same two polypeptides in all positive
tissues, it is inferred that the 55 kDa polypeptide must be
highly related to the 70 kDa protein, representing most
likely a novel CreaT isoform arising by differential splicing
of the original CreaT mRNA. Preliminary results suggest
that the 55 kDa putative novel CreaT isoenzyme is not
generated by de-glycosylation, since incubation with
glycosidases does not convert the 70 kDa glycoprotein into
a 55 kDa polypeptide species (data not shown). Also, the 55
kDa polypeptide is unlikely a degradation product of the 70
kDa species, since both bands were always detected, although
sometimes varying in staining intensity, with both, the anti-
N as well as the anti-C-terminal antibody. In addition, there
was never a smaller difference polypeptide, expected to run
at approximately 15 kDa, detected on Western blots. These
data supports the hypothesis of the existence of at least two
CreaT isoforms.

Fig. 2. Effect of creatine and its analogue 3-guanidinopropionic acid (3-
GPA) on creatine transporter expression. Skeletal muscle proteins from rats
fed for 3 months with creatine (lane 2); 3-GPA (lane 3); as well as proteins
from control animals on a creatine free diet (lane I), were extracted as
described, subjected to electrophoresis on an 8% polyacrylamide SDS gels
and were either stained with Coomasie Blue to visualize total protein (A) or
semy-dry-blotted onto nitrocellulose membrane. The transfer was labelled
with 500 diluted anti-C-terminus peptide antibody to stain the creatine
transporter protein (B). The same immunostaining of both the 70 and 55 kDa
polypeptides were also obtained with the anti-N-terminal antibody (not
shown). Therefore, the 55 kDa protein species cannot be a degradation
product of the 70 kDa polypeptide but must be a highly related CreaT
isoform most likely generated by alternative splicing at internal site(s) of
CreaT mRNA. Note that the creatine carrier synthesis is down-regulated in
vivo by its substrate creatine (lane 2) whereas 3-GPA prevented this effect
of creatine.
Down-regulation in vivo of the expression of both
creatine transporter isoforms upon chronic creatine
supplementation
Extracellular creatine regulates creatine transport in L6 rat
muscle cells as measured by isotope studies [30]. Down-
regulation of Cr uptake in this cells can be partially reversed
when cells are maintained in medium lacking creatine. The
creatine analogue 3-GPA is a well known competitive
inhibitor of creatine entry. Long-term feeding of rats with
3-GPA decreases creatine levels in skeletal muscle. There-
fore, we studied the in vivo effect of dietary creatine as
well as of 3-GPA on the CreaT expression in skeletal
muscle (quadriceps) of rats chronically fed either 4%
creatine or 2.5% GPA. Dietary creatine, administered for
3-6 months, significantly lowers the levels of both the 70
and 55 kDa polypeptides as shown in Fig. 2 (lane 2B). In
contrast, in rats fed with the substrate analogue 3-GPA, the
expression levels of both proteins remain similar or are
slightly increased compared to those of control rats (Fig.
2, lane 3B).
Both the 70 and 55 kDa polypeptide bands respond in a
similar manner to creatine (down-regulation) and 3-GPA
(up-regulation) in skeletal muscle. Therefore, it is very
433
likely that the two proteins must be related to creatine
transport per se.
Further studies are needed to characterize the two CreaT
isoforms on a molecular level and to document the regulation
of the CreaT by creatine supplementation in other tissues
than muscle.
Discussion
Regulation of creatine homeostasis: role of the creatine
transporter
In this preliminary study, we have presented evidence for the
existence of two highly related CreaT isoforms which are
coexpressed in all tissues where the CreaT, and incidentally
also CK, is found, except for the liver which according to
our results also contains CreaT but no or very little CK. In
addition, we show that both of these highly homologous
proteins are down-regulated after chronic administration of
external Cr and that this down-regulation is prevented or even
counteracted by GPA.
Several lines of evidence support the hypothesis that
more than one CreaT exist. Two CreaT mRNA species of
4.0-4.3 and 2.2-3.0 kb, whose tissue expression patterns
differ from one another, were detected by Northern blot
analysis [20, 79]. Barnwell et al. [80] cloned and sequenced
two cDNAs: one, CreaTl, homologous throughout its
length with the rat CreaT sequence and a second, CreaT2,
encoding a new protein containing regions of perfect
homology with the CreaT amino-acid sequence and four
segments of unique sequences. Although no evidence was
presented, the authors suggest that CreaT2 mRNA may be
transcribed from the CreaTl gene, possibly by alternative
splicing [80]. Other evidence is provided by genomic local-
ization studies. These reveal the existence of an autosomal,
'testis-specific' form of the human CreaT gene besides the
previously identified and mapped X-linked CreaT gene.
This is reminiscent in the sense that there is also evidence
for a sperm-specific CK isoenzyme expressed in fish and
rooster testis [81-83].
In conclusion, the above experimental findings clearly
support the existence of two different CreaT mRNA species,
as well as of two CreaT transporter polypeptides with
identical N- and C-termini, as judged by anti-peptide-
specific antibodies. We believe that the two mRNA, most
likely being generated by alternative splicing, correspond to
the 70 and 55 kDa proteins detected by both the anti-N- as
well as the anti-C- terminal antipeptide antibodies. The
differential splicing, therefore, must take place in a region
of the mRNA coding for the amino acid sequence between
the N- and C-termini. This interpretation is consistent both
with the results from molecular biology, as well as with our
434
protein and immunological data.
The Cr analogue, p-GPA, competitively inhibits CreaT
activity. Long-term feeding (6--10 weeks) of rats withp-GPA
results in a marked decrease in PCr, Cr and ATP levels in
skeletal muscle [84]. On the other hand, Cr supplementation
augments Cr uptake and accumulation. Therefore, it is not
surprising that depending on the substrate and dosis used, and
the time of ingestion, different metabolic adaptations will
occur. Concerning the CreaT, it was suggested that extra-
cellular Cr may down-regulate the level of the CreaT
expression and its activity in a rat skeletal muscle cell line
[30]. In the present study, we have directly shown that, using
our anti-NH 2- and anti-COOH-terminal CreaT antipeptide
antibodies, the expression of both, the 70 and 55 kDa
proteins, is indeed down-regulated in vivo by long-term Cr-
feeding in rat skeletal muscle. The Cr-analogue p-GPA,
however, prevented this effect or even slightly up-regulated
the CreaT
These results can most likely also be extrapolated to
human athletes who chronically ingest Cr. Human muscle
appears to have an upper limit for its Cr content of 150-160
mmol/kg of dry muscle. [50]. The latter suggests that long-
term Cr intake influences the synthesis ofthe CreaT in order
to prevent the accumulation of excessive intramuscular Cr.
On the other hand, the down-regulation of the CreaT could
be interpreted as an undesirable side effect of Cr supple-
mentation. Therefore, it is unadvisable to consume Cr
continuously for longer periods of time, e.g. over 3 months
or so, or to abuse Cr ingestion in order to improve perform-
ance at any cost. A creatine-free period of one month, after
3 months of Cr ingestion, is thus advisable.
In the section devoted to Cr levels in disease, we dis-
cussed the benefits of Cr ingestion and point out that Cr
supplementation may be a promising adjuvant therapy for
patients with a variety of neuromuscular diseases.
Our results suggest that the effect of extracellular Cr on
the transporter protein is to control, by a negative feedback
repression mechanism, the synthesis of the CreaT itself.
Thus, down-regulation of CreaT expression is a mean to
control intracellular Cr homeostasis. The study ofthe CreaT
could also help us to understand its potential role in
neuromuscular diseases. Utilization of our antibodies on
biopsy material of patients with different neuromuscular
diseases, mitochondrial myopathies and dystrophies may
reveal altered expression of CreaT, defects in its regulation
or of mutations in the CreaT protein.
Note added in proof
Recently, Kekelidze et al. [90] reported that by using a
polyclonal antiserum against a CreaT fusion protein on
Western blots, also two distinct polypeptide bands with an
apparent Mr of 70 and 50 kDa could be observed in lysates
from HeLa cells transfected with rat CreaT cDNA, as well
as in homegenates of rat brain and muscle. These results
obtained by a different strategy are fully in line with our
observation that two CreaT species, which need further
characterization, are found in different tissues. Very
recently, neuroprotective effects of creatine and cyclo-
creatine were reported in animal models of Huntington disease
[91] and a remarkable protection of brain metabolism by creatine
[92] and GPA [93] was demonstrated in mice that showed
stabilized brain ATP and significantly enhanced survival
during hypoxia. Finally, a protective effect of creatine and
cyclo-creatine preventing calcium (40 ~m)- plus atractyloside
(5 ~m)- induced mitochondrial destabilization by the so-called
permeability transition have been demonstrated [94], linking
the CK system and its substrates to early events of apoptosis.
These results are bound to encourage further investigations.
For a recent review on the nutritional biochemistry of creatine
see also Greenhaff[95].
Acknowledgment
This work was supported by a Swiss Bundesstipendium to L.G.
and a grant from the Swiss Society for Muscle Diseases given
to TW., as well as by private sponsoring by SYNERGEN, CH-
6330 Cham, Switzerland. We would like to thank Elsy
Zanolla for technical assistance, Markus Gruber for help and
discussion, and Drs. Fred Damberger, Max Dolder, Max
Bronnimann and Uwe Schlattner for valuable discussion and
for carefully reading the manuscript. We are very grateful
to Dr. Eddie a 'Gorman for kindly making the GPA used for
these experiments.
References
1. Hunter A: Monographs on biochemistry: Creatine and creatinine.
Longmans, Green and Co., 1928
2. Walker JB: Creatine: Biosynthesis, regulation and function. In:
Advances in Enzymology and related areas of Molecular Biology, Vol.
50. John Wiley and Sons, 1979, pp 177-241
3. Maughan RJ: Creatine supplementation and exercise performance. Int
J Sport Nut 5: 94--101,1995
4. Fitch CD, Shields RP, Payne WF, Dacus JM: Creatine metabolism in
skeletal muscle. Specificity ofthe creatine entry process. J Bioi Chern
243:2024--2027,1968
5. Borsook H, Dubnoff JW: The hydrolysis of phosphocreatine and the
origin of urinary creatinine. J Bioi Chern 168: 493-510,1947
6. Fitch CD, Shields RP: Creatine metabolism in skeletal muscle. Creatine
movement across muscle membranes. J Bioi Chern 241: 3611-3614,
1966
7. Harris RC, Vim M, GreenhaffPL, Hultman E: The effect of oral creatine
supplementation on running performance during maximal short term

exercise in man. J Physiol 467: 74P, 1993
8. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM:
Significance of intracellular compartmentation, structure and function
of creatine kinase isoenzymes for cellular energy homeostasis: 'The
Phospho-Creatine Circuit'. Biochem J 281: 21-40, 1992
9. Wyss M, Smeitnik J, Wevers RA, Wallimann T: Mitochondrial creatine
kinase: A key enzyme of aerobic energy metabolism. Biochim Biophys
Acta 1102: 119-166, 1992
10. Daly MM, Seifter S: Uptake of creatine by cultured cells. Arch Biochem
Biophys203:317-324,1980
11. Ku CP, Passow H: Creatine and creatinine transport in old and young
human red blood cells. Biochim Biophys Acta 600: 212-227, 1980
12. Lolke JD, Somes M, Silverstein SC: Creatine uptake, metabolism, and
efflux in human monocytes and macrophages. Am J Physiol251 : C 128-
Cl35,1986
13. Moller A, Hamprecht B: Creatine transport in cultured cells ofrat and
mouse brain. J Neurochem 52: 544-550, 1989
14. Mayser W, Schloss P, Betz H: Primary structure and functional
expression of a choline transporter expressed in the rat nervous system.
FEBS Lett 305: 31-36,1992
15. Gulmbal C, Kilimann MW: A Na+-dependent creatine transporter in
rabbit brain, muscle, heart, and kidney. J BioI Chem 268: 8418-8421,
1993
16. Nash SR, Giros B, Kingsmore SF, Rochelle JM, Suter ST, Gregor P,
Seldin MF, Caron MG: Cloning, pharmacological characterization, and
genomic localization of the human creatine transporter. Recep Channels
2: 165-174, 1994
17. Schloss P, Mayser W, Betz H: The putative rat choline transporter
CHOT! transports creatine and is highly expressed in neural and
muscle-rich tissues. Biochem Biophys Res Comm 198: 637--645, 1994
18. Schloss P, Piischel AW, Betz H: Neurotransmitter transporters: New
members of known families. Curr Op Cell Biol6: 595-599, 1994
19. Sora I, Richman J, Santoro G, Wei H, Wang Y, Vanderah T, Horvath R,
Nguyen M, Waite S, Roeske WR, Yamamura HI: The cloning and
expression of a human creatine transporter. Biochem Biophys Res
Commun 204: 419-427, 1994
20. Saltarelli MD, Bauman AL, Moore KR, Bradley CC, Blakely RD:
Expression of the rat brain creatine transporter in situ and in transfected
He La cells. Develop Neurosci 18: 524-534, 1996
21. Wallimann T, Hemmer W: Creatine kinase in non-muscle tissues and
cells. Mol Cell Biochem 1331134: 193-220, 1994
22. Gregor P, Nash SR, Caron MG, Seldin MF, Warren ST: Assignment of
the creatine transporter gene (SLC6A8) to human chromosome Xq28
telomeric to G6PD. Genomics 25: 332-333, 1995
23. [yer GS, Krahe R, Goodwin LA, Doggett NA, Siciliano MJ, Funanage
VL, Proujansky R: Identification of a testis-expressed creatine
transporter gene at 16p 11.2 and confirmation of the linked locus to
Xq28. Genomics 34: 143--146, 1996
24. Emery AK: X-linked muscular dystrophy with early contractures and
cardiomyopathy (Emery-Dreifuss type). Clin Gen 32: 360-367,1987
25. Conzalez GG, Thomas NS, Stayton NS, Knight SJ, Johnson M, Hopkins
LC, Harper PS, Elsas LJ, Warren ST: Assignment of Emery-Dreifuss
muscular dystrophy to the distal region of Xq28: The results of a
collaborative study. Am J Hum Genet 48: 468-480,1991
26. Ades LC, GedeonAK, Wilson MJ, Latham M, Partington MW, Mulley
JC, Nelson J, Lui K, Sillence DO: Barth syndrome: Clinical features and
confirmation of gene localization to distal Xq28. Am J Med Genet 45:
327-334, 1993
27. Bolhuis PA, Hensels GW, Hulsebos TIM, Baas F, Barth PG:
Mapping of the locus for X-linked cardioskeletal myopathy with
neutropenia and abnormal mitochondria (Barth syndrome) to
Xq28. Am J Hum Genet 48: 481-485, 1991
28. GedeonAK, Wilson MJ, Colley AC, Sillence DO, Mulley JC: X-linked
435
fatal infantile cardiomyopathy maps to Xq28 and is possibly
allelic to Barth Syndrome. J Med Genet 32: 383-388, 1995
29. Thomas NST, Williams H, Cole G, Roberts K, Clarke A, Liechti-Gallati
S, Braga S, Gerber A, Meier C, Moser H, Harper PS: X-linked neonatal
centronuclear/myotubular myopathy: Evidence for linkage to Xq28
DNA marker loci. JMed Genet 27: 284-287,1990
30. Loike JD, Zalutsky DL, Kaback E, Miranda AF, Silverstein SC:
Extracellular creatine regulates creatine transport in rat and human
muscle cells. Proc Nat! Acad Sci USA 85: 807-811, 1988
31. Harris RC, Soderlund K, Hultman E: Elevation of creatine in resting
and exercised muscle of normal subjects by creatine supplementation.
Clin Sci 83: 367-374,1992
32. Gulmbal C, Kilimann MW: A creatine transporter cDNA from Torpedo
illustrates structure/function relationships in the GABAlNoradrenaline
transporter family. J Mol Bioi 241 : 317-324, 1994
33. Delanghe J, De Slypere JP, De Buyzere M, Robbrecht J, Wi erne R,
Vermeulen A: Normal reference values for creatine, creatinine, and
carnitine are lower in vegetarians. Clin Chern 35: 1802-1803, 1989
34. Balsom PD, Soderlund K, Ekblom B: Creatine in humans with special
reference to creatine supplementation. Sports Med 18: 268-280, 1994
35. Guerrero ML, Beron J, Spindler B, Groscurth P, Wallimann T, Verrey F:
Metabolic support ofNa+-pump in apically permeabilized A6 kidney
cell epithelia: Role of creatine kinase. Am J PhysioI272: C697--C706,
1997
36. Walker JB: Metabolic control of creatine biosynthesis. I. Effect of dietary
creatine. J BioI Chern 235: 2357-2361,1960
37. Odoom JE, Kemp GJ, Radda GK: The regulation of total creatine content
in a myoblast cell line. Mol Cell Biochem 158: 179-188, 1996
38. Balsom PD, Ekblom B, Soderlund K, SjOdin B, Hultman E: Creatine
supplementation and dynamic high-intensity intermittent exercise.
Scand J Med Sci Sports 3: 143-149, 1993
39. Birch R, Noble D, Greenhaff PL: The influence of dietary creatine
supplementation on performance during repeated bouts of maximal
isokinetic cycling in man. J Appl Physiol69: 268-270, 1994
40. Earnest CP, Snell PG, Rodriguez R,AlmadaAL: The effect of creatine
monohydrate ingestion on anaerobic power indices, muscular strength
and body composition. Acta Physiol Scand 153: 207-209, 1995
41. Greenhaff PL, Casey A, Short AH, Harris R, Soderlund K, Hultman E:
Influence of oral creatine supplementation of muscle torque during
repeated bouts of maximal voluntary exercise in man. Clin Sci 84: 565-
571,1993
42. Casey A, Constantin-Teodosiu D, Howell S, Hultman E, Greenhaff
PL: Creatine ingestion favorably affects performance and muscle
metabolism during maximal exercise in humans. Am Physiol Soc 271 :
E31-E37, 1996
43. Koszalka TR, Andrew CL: Effect of insulin on the uptake of creatine-I-
14C by skeletal muscle in normal and X-irradiated rats. Proc Soc Exp
BiolMed 139: 1265-1271,1972
44. Haughland RB, Chang DT: Insulin effect on creatine transport in
skeletal muscle. Proc Soc Exp Bioi Med 148: 1-4, 1975
45. Green AL, Hultman E, Macdonald lA, Sewell DA, Greenhaff PL:
Carbohydrate ingestion augments skeletal muscle creatine accumulation
during creatine supplementation in man. Am J Physiol271: E821-E826,
1996
46. Clausen T, Flatman JA: The effect of catecholamines on Na-K transport
and membrane potential in rat soleus muscle. J Physiol270: 383-414, 1977
47. Clausen T: Regulation of active Na+-K+ transport in skeletal muscle.
Physiol Rev 66: 542-580,1986
48. Vandenberghe K, Gillis N, Van Leemputte M, Van Hecke P,
Vanstapel F, Hespel P: Caffeine counteracts the ergogenic action
of muscle creatine loading. J Appl Physiol 80: 452-457, 1996
49. Soderlund K, Balsom PD, Ekblom B: Creatine supplementation and
high intensity exercise: Influence on performance and muscle
436
metabolism. Clin Sci 87(suppl): 120-121, 1994
50. Greenhaff PL: Kreatin: Seine Rolle in Bezug auf die korperliche
Leistungsfahigkeit sowie Ermiidung; seine Anwendung als ein
Sporternaahrungs-Supplement. In: F Browns, S Stromme, C Wenk, C
Williams (eds). Insider, News on Sport Nutrition, Isostar Sport Nutrition
Foundation 3: 1-4, 1995
51. Dawson B, Cutler M, Moody A, Lawrence S, Goodman C, Randall N:
Effects of oral creatine loading on single and repeated maximal short
sprints. Aus J Sci Med Sport 27: 56--61, 1995
52. Hultman E, Soderlund K, Timmons JA, Cederblad G, GreenhaffPL:
Muscle creatine loading in men. J Appl Physiol81: 232-237,1996
53. Bronnimann M: Spitzenleistungenohne "sauer" zu werden? Move 6:
54.
5-6,1995
'Gorman E, Beutner G, Wallimann T, Brdiczka D: Differential effects
of creatine depletion on the regulation of enzyme activities and on
creatine-stimulated mitochondrial respiration in skeletal muscle, heart,
and brain. BiochimBiophysActa 1276: 161-170, 1996
55. Chanutin A: The fate of creatine when administered to man. J BioI
Chern 67: 29--37,1926
56. Neubauer S, HamrnanBL, Perry SB, Bittl JA, Ingwall JS: Velocity ofthe CK
reaction decreases in postischemic myocardium: A 31P-NMRmagnetization
transfer study of the isolated ferret heart. Circ Res 63: 1-15, 1988
57. Whittingham TS, Lipton P: Cerebral synaptic transmission during
anoxia is protected by creatine. J Neurochem 37: 161&--1621, 1981
58. Carter AJ, Miiller RE, Pschom U, Stransky W: Preincubation with
creatine enhances levels of creatine phosphate and prevents anoxic
damage in rat hippocampal slices. J Neurochem 64: 2691-2699, 1995
59. StockIer S, Holzbach U, Hanefeld F, Marquardt I, Helms G, Requart M,
Hanicke W, Frahm J: Creatine deficiency in the brain:Anew, treatable
inborn error of metabolism. Pediatric Res 36: 409-413, 1994
60. Liao R, Nascimben L, Friedrich J, Gwathmey JK, Ingwall JS: Decreased
energy reserve in an animal model of dilated cardiomyopathy.
Relationship to contractile performance. Circ Res 78: 893-902, 1996
61. Ciafaloni E, Ricci E, Shanske S: MELAS: Clinical features, biochemistry,
and molecular genetics. Ann Neurol31: 391-398, 1992
62. Hagenfeldt L, von Dobeln U, Solders G, Kaijser L: Creatine treatment
inMELAS. Muscle Nerve 17: 1236,1994
63. StadhoudersAM, Jap PHK, Winkler HP, Eppenberger HM, Wallimann
T: Mitochondrial creatine kinase: A major constituent of pathological
inclusions seen in mitochondrial myopathies. Proc Nat! Acad Sci USA
91:5089--5093,1994
64. Eppenberger-Eberhardt M, Riesinger I, Messerli M, Schwarb P, Miiller
M, Eppenberger HM, Wallimann T: Adult rat cardiomyocytes cultured
in creatine-deficient medium display large mitochondria with
paracrystalline inclusions, emiched for creatine kinase. J Cell Bioi 113 :
289--302, 1991
65. Sipila I, Simell 0, Rapola J, Sainio K, Tuuteri L: Gyrate atrophy of the
choroid and retina with hyperornithonemia: Tubular aggregates and
type 2 fiber atrophy in muscle. Neurology 29: 996-1005,1979
66. Sipila I, Valle D, Brusilow S: Low guanidinoacetic acid and creatine
concentrations in gyrate atrophy ofthe choroid and retina (GA). In: PP
De Deyn, B Marescau, V Stalon, IA Qureshi (eds). Guanidino
Compounds in Biology and Medicine, Vol. John Libbey and Company
Ltd., 1992,pp379--383
67. Trijbels JMF, Sengers RCA, Bakkeren JAJM, De KortAFM, Deutman
AF: L-ornithine-ketoacid transaminase deficiency in cultured fibroblasts
of a patient with hyperomithinernia and gyrate atrophy of the choroid
and retina. Clin ChimActa 79: 371-377,1977
68. Shih VE, Berson EL, Mandell R, Schmidt SY: Ornithine ketoacid
transaminase deficiency in gyrate atrophy of the choroid and retina.
Am J Hum Genet 30: 174--179, 1978
69. Sipila I, Simell 0, Arjomaa P: Gyrate atrophy of the chorid and retina
with hyperomithinernia. J Clin Invest 66: 684-687, 1980
70. Sipila I, Rapola J, Simell 0, Vannas A: Supplementary creatine as
a treatment for gyrate atrophy of the choroid and retina. New
Engl J Med 304: 867-870, 1981
71. Bigge T: Mein Selbstversuch mit Kreatin-Monohydrat. Deutscher
Muskelreport3: 23-25,1996
72. Kass I, Lipton P: Mechanisms involved in irreversible anoxic damage
to the in vitro rat hippocampal slice. J Physiol332: 459-472, 1982
73. Conway MA, Clark JF: Biochemical basis for a therapeutic role of
creatine and creatine phosphate. In: Creatine and Creatine Phosphate:
Scientific and Clinical Perspectives. Harcourt Brace and Company,
Publishers, London, 1996, pp 91-123
74. Schneider WJ, Slaughter J, Goldstein JL,Anderson RGW, Capra JD,
Brown MS: Use of antipeptide antibodies to demonstrate external
orientation of the NH,-terminus of the low density lipoprotein
receptor in the plasma membrane of fibroblasts. J BioI Chern 97: 1635-
1640,1983
75. Bradford M: A rapid and sensitive method for the quantitation of
microgram quantities of protein utilizing the principle of protein-dye
binding. Anal Biochem 72: 24&--254,1976
76. Laemmli UK: Cleavage of structural proteins during the assembly of
the head of bacteriophage T4. Nature 227: 680-685, 1970
77. Schneppenheim R, Budde U, Dahlmann N, Rautenberg P: Luminography
- a new, highly sensitive visualization method for electrophoresis.
Electrophoresis 12: 367-372,1991
78. Rowley GL, Greenleaf AL, Kenyon GL: On the specificity of creatine
kinase. New glycocyamines and glycocyamine analogs related to
creatine. J Am Chern Soc 93: 5542-5551, 1971
79. Gonzalez AM, Uhl GR: 'Choline/orphan V8-2-lIcreatine transporter'
mRNA is expressed in nervous, renal and gastrointestinal systems.
Mol Brain Res 23: 266-270, 1994
80. Barnwell LF, Chaudhuri G, Townsel JG: Cloning and sequencing of a
cDNA encoding a novel member ofthe human brain GABAInoradrenaline
neurotransmitter transporter family. Gene 159: 287-288, 1995
81. Garber AT, Winkfein RJ, Dixon GH: A novel creatine kinase cDNA
whose transcript shows enhanced testicular expression. Biochim
BiophysActa 1087: 256-258,1990
82. Saudrais C, Garber AT, McKay DJ, Dixon GH, Loir M:'Creatine kinase
in trout male germ cells - purification, gene expression, and localization
in the testis. Mol Reprod Dev 44: 433-442, 1996
83. Kaldis P, Kamp G, Piendl T, Wallimann T: Functions of creatine kinase
isoenzymes in spermatozoa. Adv Devel Biochem 5: 275-312, 1997
84. Wyss M, Wallimann T: Creatine metabolism and the consequences of
85.
creatine depletion in muscle. Mol Cell Biochem 133/134: 51-66, 1994
'Gorman E, Fuchs KH, Tittmann P, Gross H, Wallimann T: Crystalline
mitochondrial inclusion bodies isolated from creatine depleted rat
soleus muscle. J Cell Sci 110: 1403-1411,1997
86. O'Gorman E, Piendl T, Miiller M, Brdiczka D, Wallimann T: Mito-
chondrial intermembrane inclusion bodies: The common denominator
between human mitochondrial myopathies and creatine depletion,
due to impairment of cellular energetics. Mol Cell Biochem 174: 283-
289,1997
87. Wallimann T: Einnahme von Kreatin als moegliche Hilfstherapie fuer
Patienten mit verschiedenen neuromuskulaeren Erkrankungen.
Mitteilungsblatt der Schweizerischen Gesellschaft fuer Muskelkranke
41: 6-9, 1997
88. Broennimann M, Wallimann T: Kreatin: Durchbruch in der Behandlung
neuromuskulaerer Krankheiten. Mitteilungsblatt der Schweizerischen
Gesellschaft fuer Muskelkranke 41: 3-11, 1997
89. Hanefeld F, Hobbiebrunken E: Kreatin zur Verbesserung von Kraft und
Ausdauer bei Muskelkranken. Deutscher Muskelreport 4: 10-11, 1996
90. Kekelidze TN, Hermes R, Goodwyn D, Blakely RD, Saltarelli
MD: Analysis of creatine transporter expression in rat brain and
cell lines using specific polyclonal antisera. Soc Neurosci 23 (Abstr.):

586.15, 1997
91. Matthews RT, Yang L, Jenkins BG, Ferrante RJ, Rosen BR, Kaddurah-
Daouk R, Beal MF: Neuroprotective effects of creatine and cyc1o-
creatine in animal models of Huntington's disease. J Neurosci 18:
156-163, 1998
92. Wilken B, Ramirez JM, Hanefeld F, Richter DW: Supplementary creatine
enhances hypoxic augmentation in vivo by preventing ATP depletion.
Soc Neurosci 23 (Abstr.): 171.23, 1997
437
93. Holtzman D, Meyers R, O'Gorman E, Khait I, Wallimann T, Alfred E,
Jensen F: In vivo brain phosphocreatine and ATP regulation in mice fed
a creatine analog. Am J Physiol 272: CI567-C1577, 1997
94. O'Gorrnan E, Beutner G, Dolder M, Koretsky A, Bridczka D, Wallimann
T: The role of creatine kinase in inhibition of mitochondrial permeability
transition. FEBS Lett 414: 253-257, 1997
95. Greenhaff PL: The nutritional biochemistry of creatine. Nutr Biochem
8: 610-618, 1997
Molecular and Cellular Biochemistry 184: 439-443, 1998.
1998 Kluwer Academic Publishers.

Review article

Clinical cardiac magnetic resonance spectroscopy-

present state and future directions

Stefan Neubauer, 1 Michael Hom, l Dietbert Hahn2 and Kurt Kochsiek 1

I Medizinische Klinik, 2Institut fur Rontgendiagnostik; Wurzburg University, Germany

Abstract
MR spectroscopy opens a window to the non-invasive evaluation of various aspects of cardiac metabolism. Experimentally,
the method has extensively been used since 1970 'so 31p_ MR allows the registration of cardiac high-energy phosphate metabolism
to non-invasively estimate the energetic state of the heart: ATP, phosphocreatine, inorganic phosphate, monophosphate esters
and intracellular pH can all be quantitated. In conjunction with extracellular shift reagents such as [DyTTHAP- or
[TmDOTP]5-, 23Na_ and 39K_MR allow the measurement of intra- and extra-cellular cation pools. IH-MR spectroscopy allows
the detection of a large number of metabolites such as, e.g. creatine, lactate, or camitine.
Human cardiac spectrocsopy has so far been confined to the 31p nucleus. Localization techniques (DRESS, ISIS, 3D-CSI etc.)
are required to confine the acquired signal to the heart region. Relative quantification is straightforward (phosphocreatine/ATP
ratio), absolute quantification (mM) is under development. Cardiac 3IP_MR spectroscopy has research application in at least
three clinical areas: (1) Coronary artery disease: A biochemical stress test for non-invasive ischemia detection (decrease of
phosphocreatine with exercise) and viability assessment via quantification ofATP may become feasible. (2) Heart failure: The
phosphocreatine /ATP ratio may provide an independent index for grading of heart failure, allow to monitor the longterm effects
of different forms of drug therapy on cardiac energy metabolism in heart failure, and may also hold prognostic information on
survival. (3) Valve disease: It is possible that the decrease of phosphocreatine/ATP can be used to guide the timing for the valve
replacement.
At the present time, no routine clinical applications can be defined for the use of human cardiac spectroscopy in patients
with cardiac disease. However, the technique holds great potential for the future as a non-invasive approach to cardiac
metabolism, and in coming years routine applications may become reality. (Mol Cell Biochem 184: 439-443, 1998)

Key words: MR spectroscopy, Cardiac metabolism, Cardiac energetics, Ion homeostasis, Heart failure, Coronary artery disease,
Valve disease

Introduction
The signal source for MR imaging is exclusively the IH
nucleus, more specifically the highly abundant IH nuclei in
water (HP) and fat (CH2and CH3 groups). MR spectroscopy,
however, also allows the study of other nuclei which have a
nuclear spin (i.e. an uneven number of protons, neutrons or
both). The main problem of MR spectroscopy is that of
sensitivity. All MR detectable nuclei have a substantially lower
MR sensitivity than IH and are present in concentrations
several orders of magnitude lower than concentrations of H 20
and fat protons. On the other hand, while IH-MR imaging
allows the registration of cardiac anatomy and function, MR
spectroscopy opens a window to the noninvasive evaluation
of various aspects of cardiac metabolism. Nuclei of biological
interest that can be studied with MR spectroscopy include 31p
(the most widely studied nucleus), 23Na, IH (protons from
metabolites other than water and fat), 39K, l3C and 19F,
Basic research aspects
Experimentally, MR spectroscopy has been used since the
mid-1970's as an increasingly versatile tool for the study of
cardiac metabolism in animal models (see [1,2] for reviews).
3IP_MR allows the registration of cardiac high-energy
phosphate metabolism to non-invasively estimate the energetic

Address for offprints: Stefan Neubauer, Medizinische Universitatsk1inik, Josef-Schneider-Stra~e 2, 97080 Wiirzburg, Germany
440
per
y- (1- Il-ATP
5 o -5 -10 -15 -20 ppm
Fig. 1. 3IP_MR spectrum of an isolated perfused rat heart obtained in 5 min
at 7 Tesla
state of the heart. Figure 1 shows a 31P_MR spectrum from an
isolated perfused rat heart, demonstrating some of the
principles relevant to spectroscopy of all MR visible nuclei. Six
resonances are identified (three 31P-atoms of ATP (Y-, (X-, ~-),
phosphocreatine (PCr), inorganic phosphate (P) and mono-
phosphate esters (MPE, which meet the following require-
ments: Present in sufficient concentration (for 31p > ca. 0.6
mM) and free in solution (immobilized metabolites give no
quantifiable MR signal due to very short T2 values). The
phenomenon that different metabolites resonate at distinct
frequencies, allowing their discrimination from each other,
is termed 'chemical shift' (expressed relative to the BI field
in ppm = parts per million) [3]: Different positions in the
molecule lead to subtle differences in the strength of the local
magnetic field, thus spreading the resonances frequencies of
31p metabolite over a range of ca. 30 ppm. The area under each
resonance is proportional to the amount of each 31P-nucleus
in the spectrum. Absolute metabolite concentrations can be
evaluated by comparing to an external 31P-standard. In
addition, intracellular pH can be quantitated from the
chemical shift difference between PCr and Pi' which is pH-
sensitive [1]. In conjunction with extracellular shift reagents
such as [DyTTHAP- or [TmDOTPp-, 23Na_ and 39K_MR
allow the measurement of intra- and extracellular cation pools.
The 13C nucleus has a low natural abundance (1.1 %); for this
reason, one has to supply 13C-Iabelled compounds such as,
e.g., 1-13C-glucose; we can then, however, follow the fate of
the labelled nuclei and analyze a number of intracellular
pathways, most prominently, citric acid cycle kinetics [4,
5]. IH-MR spectroscopy allows the detection of a large
number of metabolites such as creatine, lactate, carnitine
etc., however, even experimentally, this technique is just
beginning to evolve.
Human cardiac MR spectroscopy
Principles and methodological problems
Almost without exception, human cardiac spectroscopy has
been confined to the 31p nucleus. Usually, patients are
positioned prone (fewer motion artifacts and smaller distance
to 31P-coil than in supine position) in the magnet. A series of
IH scout images is obtained, which is used to select the
spectroscopic volume(s). Methodologically, clinical cardiac
spectroscopy faces a number of major technical problems.
Total examination time (including imaging and spectroscopy
scans) should not be more than one hour, and even this may
not be tolerated by patients with severe cardiac disease; thus,
time for signal acquisition is limited. The heart is rapidly
moving, requiring gating to the heart beat, and, possibly, to
respiration as well. Cardiac muscle is situated behind a layer
of chest wall skeletal muscle giving rise to a strong 31 P-signal
that must be suppressed; unlike in experimental studies, this
necessitates the use of additional localization techniques,
which, in turn, are unavoidably associated with substantial
signal loss. Localization techniques that have been applied
to the human heart are DRESS (depth-resolved surface coil
spectroscopy), Rotating frame, ID-CSI (chemical shift
imaging), ISIS (image-selected in vivo spectroscopy), and
3D-CSI; a full review is beyond the scope of this text and is
available elsewhere [6]. In the future, 3D-CSI may become
the standard for localization. Due to the relatively low MR
sensitivity of the 31p nucleus and small amounts of 31p_
metabolites in heart (relative to IH in HP), required voxel sizes
have been quite large, usually> 30 cm3. Figure 2 shows a
typical 31 P-MR spectrum of a healthy volunteer acquired from
a -60 cm3 voxel using ISIS and a 15 sec repetition time.
Compared to the rat heart spectrum, signal-to-noise is
considerably lower, and two additional resonances appear:
2,3-diphosphoglycerate (2,3-DPG), arising from the presence
of blood (erythrocytes) in the selected voxel and phos-
CP
Volunteer Severe DCM
10 5 0 -5 -10 15 20 10 5 0 -5 -10 15 20
ppm
Fig. 2. 3IP_MR spectra of a healthy volunteer and a patient with severe
DCM obtained in 30 min at 1.5 Tesla using ISIS localization. The PCrl
ATP ratio is substantially reduced in the patient. Adapted from Neubauer
Set al. [12] Reproduced with permission.

phodiesters (PDE), a signal due to membrane as well as serum
phospholipids. The 2,3-DPG resonances appear at the same
frequency as Pi' which therefore cannot be detected in blood-
contaminated human spectra; for the same reason, intracellular
pH cannot be determined. Traditionally, human lIP-spectra are
quantified in relative terms only, i.e. the PCrlATP and PDEI
ATP ratios are calculated. PCrlATP is considered an index of
the energetic state ofthe heart, while the meaning ofthe PDEI
ATP ratio is poorly understood; PDEIATP does not seem to
change with cardiac disease. Absolute quantification of PCr
andATP is technically difficult, although highly desirable and
theoretically achievable using a lIP-standard and estimates
of myocardial mass based on MR imaging (6). A recently
proposed approach for absolute quantification of IIp_
metabolites uses the IH signal as a calibration reference and
looks most promising (7). lIP-spectra need to be corrected
for the effects of partial saturation, which are the greater the
higher the pulse repetition rate is. Correction for the amount
of blood contamination (blood contributes signal fromATP,
2,3-diphosphoglycerate and phosphodiesters) will in the long
term hopefully be no longer a problem when voxel sizes are
reduced far enough to obtain spectra largely uncontaminated
from blood. For integration of human spectra, an algorithm
is required that allows for time domain or frequency domain
Lorentzian line fitting. The ideal approach to spectral
quantification, which should then informally be used in all
MR centres for the sake of data comparability, still has to be
worked out, however.
Present status
At present, cardiac 3IP_MR spectroscopy has research
applications in at least four clinical areas:
Coronary artery disease: In patients with LAD stenosis,
anterior wall PCrl ATP ratios are normal at rest [8], but
decrease during hand grip exercise and return towards
normal with recovery from exercise (9). This is true for
patients with reversible defects on thallium scintigraphy
(presumably viable myocardium) but not for those with
fixed thallium defects (presumably scar), where PCr/ATP
is reduced at rest but does not decrease further with exercise
(10). Most recently, absolute quantification ofPCr andATP
in patients with LAD stenosis was reported [11]. Absolute
ATP content was significantly reduced in patients with fixed
thallium defects but unchanged in those with reversible
thallium defects. This suggests that MR spectroscopy may
be suitable for the non-invasive assessment of myocardial
viability much like PET scanning, the main limitation at
present being large voxel sizes.
Heartfailure: Our own studies have mostly concentrated on
patients with symptomatic and asymptomatic stages of heart
failure (8, 12):
Figure 2 (right panel) shows the spectrum of a patient with
441
severe dilated cardiomyopathy (DCM). In this spectrum, the
PCrl ATP ratio is reduced. We have studied 19 patients with
DCM. In these patients, the PCrlATP ratio showed a trend
towards a decrease (1.78 0.51), but was not significantly
different from healthy volunteers (l.95 0.45). Thus, as a
group including all clinical stages, patients with DCM could
not be distinguished from volunteers based on 3IP_MR data.
However, when patients were grouped according to the
clinical severity of heart failure, a different picture evolved.
Figure 3 shows that the PCrlATPratio decreased progressively
in relation to the severity of heart failure. Furthermore,
linear regression between NYHA state and PCrlATP was
highly significant (r = 0.60, *p < 0.005). We also studied
six patients with DCM sequentially before and after 12 6
weeks of drug therapy (digitalis, diuretics, ACE-inhibitors,
beta-blockers). The six patients improved by 0.8 0.3
NYHA classes during treatment. Figure 4 shows the changes
of PCrl ATP during treatment. The initial PCrl ATP ratio was
1.51 0.32 and increased in all six patients after therapy to
a value of2.15 0.27; this increase was statistically significant
(*p < 0.01). Thus, in patients with DCM, the PCr/ATP ratio
was inversely correlated to the severity of heart failure and
could be improved by chronic drug therapy. In contrast, we
observed no changes ofthe PDEIATP ratio with heart failure
[8]. Thus, this index does not characterize patients with left
ventricular dysfunction. In an extended group of patients
[12], we observed a significant correlation (r = 0.54, P <
0.01) between PCrlATP and left ventricular ejection fraction
(Fig. 5). We also found a significant correlation (r = 0.51, P <
3.0
2.5
2.0 it
0..
~
<t
'" 1.5
0..
u
1.0
0.5
0
II II -ID ill ID-IV
Fig. 3. PCr/ATP ratios (CP/ATP in the figure, 'creatine phosphate' and
'phosphocreatine' are synonymous terms) in patients with dilated
cardiomyopathy graded according to the NYHA classification. For each
NYHA grade, raw and mean data are shown. Correlation between the NYHA
grade and PCr/ATP was highly significant (r = 0.60; *p < 0.005). Adapted
from Neubauer S et al. [8] Reproduced with permission.
442

0.0 1 ) between perlATP and left ventricular wall thickness

(Fig. 5). Patients with the thinnest LV walls also had the lowest
perlATP ratios. perlATP did not correlate with any other
2.S
hemodynamic index measured by right and left heart catheter-
ization [12].
Other studies agree that perlATP ratios are reduced in
patients with more advanced stages of heart failure of
ischemic [13] and non-ischemic origin [8, 13], but may
loS remain unchanged in early stages [8,14]. Thus, 31P-MRmay
provide an independent index for grading of heart failure and
may allow to monitor the long-term effects of different forms
of drug therapy on cardiac energy metabolism in heart failure.
At present, we are investigating whether the perlATP ratio
also holds prognostic information on survival of patients with
o.s heart failure [15].
before after therapy Valve disease and other forms of cardiac hypertrophy: In
patients with aortic or mitral valve disease (stenosis and/or
Fig. 4. PCrlATP ratios before and after 12 6 weeks of drug therapy in 6 incompetence) perl ATP ratios are reduced only when
patients with dilated cardiomyopathy. There was a significant increase of patients are in heart failure but not in early stages. Thus, it
PCr/ATP from 1.51 0.32 to 2.15 0.27 (*p < 0.01). Adapted from
was speculated that the decrease of perlATP can be used to
Neubauer S et al. [8] Reproduced with permission.
guide the timing for valve replacement [16]. Although this
is a promising perspective, larger prospective studies of
2.75
0 0 patients with valve disease before and after valve replacement
2.50 are needed to determine the value of31P-MR in this clinical
2.25 0 entity. Sporadic reports have shown reduced perlATP in
2.00 0 patients with hypertrophic cardiomyopathy, although, due to
~ 1.75 o 0
large voxel sizes and the heterogenic nature of the disease,
5 1.50 these findings may be difficult to interpret. Systematic studies
oi( 0 0

0 of cardiac hypertrophy due to long-standing hypertension or

0 0
due to heavy exercise in athletes ('physiologic hypertrophy')
1.00 0
remain to be performed.
0.75
Y = 0.02x + 0.94
0
0.50 L..LJ.....L...L!....L.......LL...L.L.L--L..J.--L...JL.....c..-LL..L....J Cardiac transplantation: Experimental reports had raised the
10 15 20 25 30 35 40 45 50 55 60 65 hope that 31p_MR can be used to monitor transplant rejection
EF(%)
and possibly, at least in part, substitute for repetitive biopsies.
2.75
0
However, although perlATP ratios are reduced following
2.50 cardiac transplantation, no reliable relation to the degree of
2.25 0 rejection has been demonstrated, and it is questionable at
2.00
present whether 31P_MR can playa role for this group of
0
patients ([6] for review)
1.75
~~ 0
0

t.l
1.50 0
Future Potential
1.25
0
0
0 0
First and before all, the future challenge will be to substantially
~

1.00 0
0 improve temporal and spatial resolution for clinical cardiac
spectroscopy. Using nuclear overhauser enhancement, higher
0.75
0 Y = 0.18x + 0.08 field strengths [17] and other improvements in hardware and
0.50
6 7 8 9 10 11 12 13 software design, true 3D metabolic imaging with voxel sizes
LV end-diastolic wall thickness (mm I of < 5 cm3 may be possible; a great effort on the side of the
research community as well as on the side of the manufacturers
Fig. 5. Linear regression analysis ofPCr/ATP vs.left ventricular ejection
will be required. However, if achievable, a non-invasive
fraction (EF) and of PCrl ATP vs. left ventricular end-diastolic wall
thickness. These correlations were highly significant (r = 0.54, P < 0.01
method to probe cardiac metabolism in human heart disease
and r = 0.51, p < 0.01, respectively). Adapted from Neubauer S et al. [12] that could well find its way into routine clinical cardiological
Reproduced with permission. practice may be established. Spectroscopic data will only be

meaningful in the context of an 'integrated cardiac MR exam'
that allows the evaluation of morphology, global and regional
function, perfusion, coronary anatomy and metabolism within
one examination. If cardiac spectroscopy can in fact rival, or
even surpass, PET in the evaluation of myocardial viability,
or if it can be made practical for detection of exercise-induced
ischemia with sufficient spatial resolution, the indication base
for the technique would in fact be overwhelming, given the
widespread availability ofMR scanners. Whether the technique
can also playa role in the routine clinical workup of patients
with heart failure, valve disease or after cardiac transplantation
will have to be established through thorough long-term
clinical studies that should best be performed in a multicenter
approach. Human cardiac MR spectroscopy of nuclei other
than 3Jp, most likely JH, 13C and 23Na, may also offer new
insights into human cardiac metabolism in heart disease; these
techniques should be established, although many method-
ological problems remain to be solved.
In summary, at the present time, no routine clinical
applications can be defined for the use of human cardiac
spectroscopy in patients with cardiac disease. However, the
technique holds great potential for the future as a non-
invasive approach to cardiac metabolism, and in coming
years routine applications may become reality.
References
l. Ingwall JS: Phosphorus nuclear magnetic resonance spectroscopy of
cardiac and skeletal muscles. Am. J. Physiol242: H729---H744, 1982
2. Neubauer S, Ertl G, Krahe T, Schindler R, Hillenbrand H, Lackner K,
Kochsiek K: Experimentelle und klinische Moglichkeiten der MR-
Spektroskopie des Herzens. Z Kardiol 80:25-36, 1991
3. Gadian DG: Nuclear Magnetic Resonance and Its Applications to
Living Systems. Oxford University Press, New York, 1982
4. Malloy CR, Sherry AD, Jeffrey FMH: Substrate metabolism in the
citric acid cycle ofthe heart by i3C NMR. Kluwer Academic Publishers,
Boston, Dordrecht, London 153-168,1992
5. Weiss RG, Chacko VP, Glickson JD, Gerstenblith G: Comparative
i3C and lip NMR assessment of altered metabolism during graded
reductions in coronary flow in intact rat hearts. Proc Nat! Acad
443
Sci USA 86: 6426--6430, 1989
6. Bottomley PA: MR spectroscopy of the human heart. The status
and the challenges. Radiology 191: 593-612, 1994
7. Bottomley PA, Atalar E, Weiss RG: Human cardiac high-energy phos-
phate metabolite concentrations by 1 D-resolved NMR spectro-
scopy. MRM 35: 664-670, 1996
8. Neubauer S, Krahe T, Schindler R, Hom M, Hillenbrand H, Entzeroth C,
Kromer EP, Riegger AJG, Lackner K, Ert1 G: Cardiac liP-magnetic
resonance spectroscopy of volunteers and patients with coronary
artery disease and dilated cardiomyopathy. Altered cardiac high-
energy phosphate metabolism in heart failure. Circulation 86:
1810-1818, 1992
9. Weiss RG, Bottomley PA, Hardy CJ, Gerstenblith G: Regional
myocardial metabolism of high-energy phosphates during isometric
exercise in patients with coronary artery disease. New Engl J Med
323: 1593-1600,1990
10. Yabe T, Mitsunami K, Okada M. Morikawa S, Inubushi T, Knoshita
M: Detection of myocardial ischemia by lip magnetic resonance
spectroscopy during handgrip exercise. Circulation 89: 1709-
1716, 1994
11. Yabe T, Mitsunami K, Inubushi T, Kinoshita M: Quantitative
measurements of cardiac phosphorus metabolites in coronary
artery disease by lip magnetic resonance spectroscopy. Circulation
92: 15-23, 1994
12. Neubauer S, Hom M, Pabst T, Godde M, Liibke D, Illing B, Hahn
D, Ert! G: Contributions of liP-magnetic resonance spectroscopy
to the understanding of dilated heart muscle disease. Eur Heart J
16 (supp\. 0): 115-118, 1995
13. Hardy CJ, Weiss RG, Bottomley PA, Gerstenblith G: Altered
myocardial high-energy phosphate metabolism in patients with
dilated cardiomyopathy. Am Heart J 122: 795-801,1991
14. De Roos A, Doornbos J, Luyten PR, Oosterwaal LJMP, van der Wall
EE, den Hollander JA: Cardiac metabolism in patients with dilated
and hypertrophic cardiomyopathy: Assessment with proton-
decoupled p-ll MR spectroscopy. JMRI 2: 711-719, 1992
15. Neubauer S, Hom M, Cramer M, Harre K, Newell JB Peters W,
Pabst T, Ertl G, Hahn D, Ingwall JS, Kochsiek K: In patients with
dilated cardiomyopathy the myocardial phosphocreatine-to-ATP
ratio is a predictor of mortality. Circulation 96: 2190-2196, 1997
16. Conway MA, Allis J, Ouwerkerk KR, Niioka T, Rajagopalan B,
Radda GK: Detection of low phosphocreatine to ATP ratio in
failing hypertrophied human myocardium by lip magnetic
resonance spectroscopy. Lancet 338: 973-976, 1991
17. Hetherington HP, Luney DJE, Vaughan JT, Pan JW, Ponder SL,
Tschendel 0, Twieg DB, Pohost GM. 3D lip spectroscopic imaging
of the human heart at 4.1 T. Magn Reson Med 33, 3: 427-431,1995
Molecular and Cellular Biochemistry 184: 445-455, 1998.
1998 Kluwer Academic Publishers.

Time-Resolved Spectroscopy of mitochondria, cells

and tissues under normal and pathological
conditions
Bertrand Beauvoit1 and Britton Chance2
lInstitut de Biochimie et Genetique Cellulaires du CNRS, Universite de Bordeaux II, 1 rue Camille Saint Saens, 33077
Bordeaux cedex, France; 2Department of Biochemistry and Biophysics, Medical School, University of Pennsylvania,
Philadelphia, PA 19104-6089, USA

Abstract
In this study, the detailed dependence of light scattering on tissue architecture and intracellular composition has been
investigated. Firstly, we simulated the reduced scattering coefficient (!-ls') of the rat liver using the Mie theory, the Rayleigh-
Debye-Gans approximation and electron microscopy data. Then, the reduced scattering coefficient of isolated rat liver
mitochondria, isolated hepatocytes and various rat tissues (i.e. perfused liver, brain, muscle, tumors) was measured at 780 nm
by using time-resolved spectroscopy and a sample-substitution protocol. The comparison of the isolated mitochondria data
with the isolated hepatocyte and whole liver measurements suggests that the mitochondrial compartment is the primary factor
for light propagation in hepatic tissue, thus strengthening the relevance of the preliminary theoretical study. Nevertheless,
the possibility that other intracellular components, such as peroxisomes and lysosomes, interfere with light propagation in
rat liver is discussed. Finally, we demonstrate that light scattering in normal rat tissues and tumors is roughly proportional to
the mitochondrial content, according to estimates ofthe mitochondrial protein content of the tissues. (Mol Cell Biochem 184:
445-455,1998)

Key words: mitochondria, transplantable tumors, rat liver, near-infrared spectroscopy, light absorption, light scattering

Abbreviations: NIR - near infrared; Q7 - Morris 7777 hepatoma; SDH - succinate dehydrogenase (succinate:oxidoreductase,
EC 1.1.99.1); TRS - Time-Resolved Spectroscopy; !-l. and !-ls' - absorption and reduced scattering coefficients (Napierian log
base) respectively; air - transport cross-sections; WAT - white adipose tissue

Introduction cellular components of the tissue [1]. Based on the theory of

In the past decade, non-invasive optical techniques have
assumed an increased importance due to the growing
utilization oflight in medicine and surgery. Considering the
potential clinical and fundamental applications of near-
infrared (NIR) spectroscopy, there has been much effort to
establish whether the optical properties of tissues are
intrinsically dependent on both functional and structural
changes in tissues [1].
When visible and near-infrared light enter tissue, it is
scattered and absorbed by the intrinsic cellular and non-
photon diffusion in a highly scattering medium - conditions
where the Beer-Lambert law no longer holds - time-
resolved spectroscopy (TRS) allows the independent deter-
mination of two macroscopic optical parameters: absorption
(!-l.) and reduced scattering coefficients (!-ls') [1-3]. The
total light absorption is the sum of the absorption due to
endogenous chromophores. For instance, in the NIR spectral
window, these compounds consist of: (i) hemoglobin and
myoglobin, in both deoxygenated and oxygenated states, (ii)
cytochrome aa 3, in both reduced and oxidized states, and
(iii) water. In most soft tissues, the light attenuation is

Address for offprints: B. Beauvoit, Institut de Biochimie et Genetique Cellulaires du CNRS, Universite de Bordeaux II, 1 rue Camille Saint Saens, 33077
Bordeaux-cedex, France
446
dominated by hemoglobin absorption. Furthennore, it is now
well established that the measurement ofthe /la' at different
wavelengths, is responsive to changes in tissue blood
volume, hemoglobin oxygenation and the metabolic rate
sustained by the tissue [3-5].
Elastic light scattering by soft tissues is associated with
microscopic heterogeneities in the tissue dielectric
constant. The detailed dependence of scattering on tissue
architecture and intracellular composition is not well
understood. Theoretically, the /ls' value of a highly multiple
scattering medium depends on several physical parameters
of the constituting particles, namely: (i) the shape, (ii) the
size compared to the wavelength value, (iii) the refractive
index relative to that of the surrounding medium, and (iv) the
number density as well as (v) particle interaction at high
concentration [6, 7]. The dependence of light scattering on
cell volume has been proved for many cell types and is now
used as a cell-sorting parameter in flow cytometry ex-
periments [8, 9]. However, the dependence oflight scattering
on intracellular organelle composition is unclear. A pre-
liminary approach, combining the detennination of the /ls' of
yeast cell suspensions by TRS and Mie theory calculations,
has clearly demonstrated that cell size is the primary factor
affecting light scattering, regardless of the differentiation
status of the mitochondria [lO]. These results could be
interpreted in two ways: (i) mitochondria per se do not
contribute to light scattering or (ii) the total volume of the
mitochondrial compartment and/or the number of mito-
chondria inside the yeast cell is small. In fact, the morphology
of yeast mitochondria varies from a giant branched mito-
chondrion [11, 12] to about 50 mitochondria particles per
yeast cell [12, l3]. Moreover, the volume fraction of the
mitochondrion in the yeast cell is relatively small (about
3-12% of the total cell volume) [l3].
The mitochondrial compartment in mammalian cells is
more dispersed than in microorganisms. The morphology
and the content of mitochondria depend especially on tissue
origin [14], the metabolic rate sustained by the organ [15,
16], honnonal treatment [17], and neoplastic transition (for
review see [18]). Fluorescence staining of various cultured
mammalian cells showed a filamentous structure (for
review see [19]) and a 'string of beads' appearance of the
mitochondrion [20]. In contrast to other mammalian cells,
hepatocytes exhibit a polydispersed fluorescent staining of
the mitochondrial compartment [21, 22]. Interestingly,
both the morphology and the content of hepatic cell
compartments (i.e. endoplasmic reticulum, mitochondria,
lysosomes, peroxisomes) are able to change drastically
following an short intoxication of the animal with specific
drugs. For instance, mitochondrial proliferation and fusion,
leading to giant mitochondria or megamitochondria, can be
induced by clofibrate feeding [23] and cuprizone treatment
[24, 25], respectively.
The goal of this study was to investigate the dependence
of the scattering properties of cells and tissues on the
mitochondrial compartment. In the first part, we calculated
the reduced scattering coefficient of rat liver by using
mathematical models and an electron microscopic des-
cription of the liver ultrastructure. Then, quantification of
the optical properties of small biological samples (e.g.
isolated mitochondria and cell suspensions, excised
tissues ... ) was performed at 780 nm using time-resolved
spectroscopy and a samples substitution procedure [26].
Both mathematical simulations and extrapolations from
isolated mitochondrial measurements to whole liver suggest
that the mitochondria are the primary factor for light
scattering in hepatic tissue. Finally, a wide range of
mitochondrial content in the tissue was analyzed by
measuring the optical properties of: (i) normal rat tissues
(i.e. perfused liver, brain, white adipose tissue, skeletal
muscle), and (ii) exteriorized solid tumors of different
histological type (i.e. hepatoma, glioma, mammary adeno-
carcinoma). We demonstrate a proportionality between
light scattering and the mitochondrial content ofthe tissue,
based upon estimates of the succinate dehydrogenase
activity and the mitochondrial protein content per tissue
wet weight.
Material and methods
Cell lines and tumor transplantation
The ascites l3762A tumor is a metastasizing mammary
adenocarcinoma of the Fisher 344 female rat, and was
adapted to this form of growth from the solid tumor
13762NF. To obtain 13762A cells for subcutaneous
implants or for mitochondrial isolation, l3762A ascites
tumor was grown by injecting 10-20 x 106 cells in RPMI
medium intraperitoneally into young Fisher 344 female
rats (100-150 g). One week later, the ascitic fluid was
collected and the cells were washed twice in RPMI. The
implantation of l3762A tumors was done by subcutaneously
injecting 0.5 x 106 live cells in 0.5 ml RPMI using a sterile
technique. Glioma 26 (G26), classified as a tumor arising
from immature glial cells, was originally induced by
methylcholanthrene treatment in the C57B 1/6 inbred
mouse. The 9L tumor is a gliosarcoma of the male Fisher
rat. MH7777 (Q7) is a Morris hepatoma which was induced
by N-2-fluorenylphenylphthalamic acid in the Buffalo
rat. The G26, 9L and Q7 tumors were maintained sub-
cutaneously as a solid tumor by passage in C57B 1/6 mice,
Fisher and Buffalo rats, respectively, at intervals of 2-3
weeks.

Administration of drugs to rat
Male Sprague-Dawley rats (150-200 g) were used for drug
intoxications. Cuprizone (bis cyclohexanone oxaldihydra-
zone), purchased from Fluka (Belgium), was subcutaneously
injected twice a week, for 35-38 days (1.4 g/kg body
weight, each injection). Clofibrate (ethyl p-chlorophenoxy
isobutyric acid), purchased from Fluka (Belgium), was
subcutaneously injected twice a week, for 35-38 days
(1.26 mllkg body weight, each injection). Diethyl ether,
purchased from Sigma (MO, USA), was subcutaneously
injected twice a week for 50-52 days (0.4 ml in com oil/kg
body weight, each injection).
Tissues preparation
Brain, skeletal muscle (gastrocnemius) and white adipose
tissue were excised from male Sprague-Dawley rats (250-
300 g rats for brain and muscle; 500-700 g rats for white
adipose tissue). After decapitation, either the brain, the
skeletal muscle or the adipose tissue was carefully re-
moved from the rat, washed twice with Hanks/Hepes buffer
(137 mM NaCl, 5.4 mM KCl, 4.2 mM NaHC0 3 , 1.2 mM
NaH 2 P0 4 , 1.25 mM CaCI 2 , 0.4 mM MgS0 4 , 0.5 mM MgCI 2 ,
10 mM Hepes (pH 7.4)) and placed in a plastic bag for TRS
measurements. The tumor-bearing rats were anesthetized by
intraperitoneal injection of pentobarbital (50 mg/kg body
mass). The tumors (3-7 g wet weight) were surgically
excised, washed several times in Hanks/Hepes buffer and
placed in a plastic bag for TRS measurements.
During liver perfusion, the rat (250-300 g male Sprague-
Dawley) was anesthetized by intraperitoneal injection of
pentobarbital (50 mg/kg body mass). The liver was perfused
at a flow rate of 30 mllmin (2.5 mllmin/g liver mass) with a
Hanks/Hepes buffer supplemented with 5 mM glucose. The
liver was removed from the carcass and placed in a plastic
bag for TRS measurements. Perfusion flow was stopped just
before measuring and the liver samples were stored on ice
between two TRS measurements. It was assumed that
perfusion cessation did not induce any change in optical
properties during the short period of the TRS measurements
(1-5 min).
Hepatocyte isolation
Rat liver parenchymal cells were isolated by using the two-
step procedure as described in Berry et al. [27]. Cells were
suspended in the complete Hanks/Hepes buffer at different
concentrations (10---60 mg cell dry mass/ml). Before TRS
measurements, plastic bags were filled with 25 ml of these
cell suspensions and stored on ice. Once the TRS was
447
measured, the dry weight ofthe different cellular suspensions
was determined.
Isolated mitochondrial preparations
Rat liver mitochondria were isolated according to Johnson
and Lardy [28]. Mitochondrial suspensions were made up to
5-15 mg proteins/ml. Before TRS measurements, plastic
bags were filled with 25 ml of these suspensions and stored
on ice. Brain and skeletal muscle mitochondria were isolated
according to references [29] and [15], respectively. Solid
tumor mitochondria were prepared according to the protocol
of reference [18]. Mitochondria were isolated from 13762A
ascites tumor cells by using the digitonin-permeabilization
procedure described in reference [30].
The protein concentration was estimated by the biuret
method after removing the lipid by an extraction with
diethylether-ethanol 3: 1 (v/v) [31]. Bovine serum albumin
was used as standard.
Enzyme activity measurements
The frozen tissues (20 mg wet wtlml) were homogenized
using a Dounce homogenizer in a solution containing 1 mM
EDTA and 50 mM potassium phosphate (pH 7.2) in the
presence of 0.5% (v/v) Triton X-IOO. Prior to making
measurements, the tissue homogenate was centrifuged for
1 min at 1000 g. An appropriate dilution of either the isolated
mitochondrial fraction or the tissue homogenate was made. The
activity of succinate dehydrogenase (succinate: oxidoreductase,
EC 1.1.99.1) and glycerol-3-phosphate dehydrogenase (sn-
Glycerol-3-phosphate: NAD 2-oxidoreductase, EC 1.1.1.8)
was measured by phenazine methosulfate (PMS)-mediated
reduction of 2,6 dichlorophenol indophenol (DCIP), in the
presence of succinate [32] and glycerol-3-phosphate,
respectively. The activity of NADPH cytochrome c oxido-
reductase was measured according to references [33, 34].
The activity of catalase (hydrogen peroxide:hydrogen
peroxide oxidoreductase, EC 1.11.1.6), acid phosphatase
(orthophosphoric monoester phosphohydrolase, EC 3.1.3.2)
and carboxyl esterase (carboxylic-ester hydrolase, EC
3.1.1.1) was measured according to references [35, 36].
Determination of absorption and reduced scattering
coefficients of small-volume samples by TRS
In principle, the photon diffusion equation, i.e. equation (2)
in Results and discussion section, is not applicable to
determine the !la and !ls' of a small-volume sample, due to
the violation ofthe boundary conditions [2]. An experimental
448

approach, the 'sample substitution method' or 'matching and a computer (Fig. 1). Photon kinetics measured by TRS
method' has been developed to solve this boundary problem were fitted to equation (2) with a compensation for instru-
[10, 26]. Briefly, experiments were performed in a two- ment response.
component system consisting of a small guest plastic-bag
containing the sample immersed in a large host medium
filled with Intralipid (Kabi Pharmatica, Clayton, NC, Simulations of transport cross-section
USA) (Fig. 1). If the absorption and scattering coefficients
of the two media are matched, then the boundary of the Mie theory
small sample vanishes, and this two-component system Graff et al. [37] have shown that the calculation of the
becomes identical to a semi-infinite medium. Under such transport cross-section by the Mie theory can be approx-
matching conditions, fitting the photon kinetics with imated by the following equation:
equation (2) leads to the correct J..l a and J..l,' values for the
sample [26]. For each sample, the matching conditions (Jtr = (Js (1- g) = nr2 3.28x.37 (m -1) 2.09

were determined by modulating: (i) the absorption co-

efficient of the Intralipid solution by adding black ink where (J, is the scattering cross-section, g, the average cosine
(Faber-Castell, No. 4416, NJ, USA) and (ii) the reduced of the scattering angle 8, x, the size parameter (x = 2nr/A),
scattering coefficient of the host medium by using different m, the refractive index relative to the surrounding medium,
Intralipid concentrations [10, 26]. r, the radius and A, the wavelength. This approximation is
The pulsed light source and the TRS detector were placed valid over a range of size parameters (5<x<50), of refractive
2.7-3 cm apart on opposite sides of the sample. They were indexes (1 <m<l.l) and of g values (g>0.90).
both inserted into the host medium about 8-10 mm below
the solution surface (Fig. 1). Pulses were injected into the Rayleigh-Debye-Gans approximation
host medium by a 200 J..lm fiber, at a 5 MHz repetition rate, The transport cross-section, i.e. (Jtr = (Js(1 - g), can also be
with a pulse width of 50 psec, and wavelength of 780 or
830 nm. The scattered light was collected with a 3 mm
diameter fiber connected to a single photon counting
system. This system consisted of a micro channel plate-
photomultiplier tube (MCP-PMT) (Hamamatsu, R1712U)
with a time resolution of 150 psec, a constant fraction
discriminator (CFD), a time to amplitude converter (TAC)
with

ps laser pulser
1 Optical fiber
calculated by using the following two formulae [8, 37]:
Source Detector and u =2 x sin (8)12.

1
Numerical integrations were done with a total of900 8 angles.
Guest
Host ...---+ sample
medium bag
Results and discussion
Theoretical background
Trigger
The attenuation oflight over a known path length, L, provides
a quantitative description of the concentration of light
absorbers via the Beer-Lambert relationship:
O.D.
J..l = - =[C]
Fig. 1. Time-resolved spectroscopy set-up for small biological samples. a L
Abbreviations used: PMT - photomultiplier tube; CFD - constant fraction (1)
discriminator; TAC - time-to-amplitude converter. where: J..l a is the absorption coefficient of the sample

(expressed in unit of cm-1); E, the extinction coefficient of the
absorber (cm-1 M-1); [C], the concentration of absorber (M);
D.D, the optical density of the sample; and L, the path length
(cm). When the sample scatters light, the latter relationship
no longer holds since the photon path lengths become
undetermined and significantly longer than the source-
detector separation. Nevertheless, in time-resolved spectro-
scopy (TRS), the distribution of path lengths traveled in
tissue is directly measured by monitoring the intensity of
light emitted by the sample as a function of time after an
incident light impulse. In such experiments, light propagation
in the tissue is described by the photon diffusion approx-
imation [2, 3]. Accordingly, the logarithm of the temporal
reflectance R (p,t) measured byTRS at the surface of a semi-
infinite medium can be expressed as [2, 3,26]:
5 p2
log R(p,t) = - 2 log t - /l. ct - 4Dct + constant (2)
where c is the speed of the light travelling in the medium,
D represents the diffusion coefficient and is equal to 11
[3(/l. + /ls')], /l. and, /ls' are the absorption and reduced
scattering coefficients, and p is the source-detector
separation.
Light propagation in biological materials is dominated by
scattering because of: (i) the relatively low light absorption
in the NIR window, and (ii) the inhomogeneities of both the
cellular structure, composition and particle size. By analogy
with the Beer-Lambert law for light absorption, the reduced
scattering coefficient of a multiple scattering medium is
related to a microscopic parameter, O"tr' by:
(3)
where O"tr is the transport cross-section of the particle and y
the number of particles per unit of volume, i.e. number
density. y is also written as 01V, so the reduced scattering
coefficient is given by the following equation:
0"
/l'=---.!!:..0
s V (4)
where 0"tr is the transport cross-section ofthe particle; V, the
single volume and 0, the volume fraction of the particles in
the total volume. This equation is valid if 0 is sufficiently small
(0 < 0.2), such as in biological suspensions. For 0 < 0.5, the
particles are densely packed and the whole solution may be
viewed as a homogeneous medium with the scattering
medium composed of the inter-particle space. In the limit
o ~ 1, the inter-particle space disappears and /ls' should
approach zero. This consideration developed by Ishimaru and
others for red blood cells [7, 38, 39] leads to the following
approximate expression of /ls':
449
(5)
Optical properties of isolated rat liver mitochondria,
isolated hepatocytes and perfused rat liver
The rat liver organ offered a unique opportunity for this study
since: (i) it is possible to isolate quite a homogeneous and
'intact' population of hepatocytes and mitochondria, (ii) the
liver ultrastructure, analyzed by electron microscopy, is well
documented, (iii) the mitochondrial compartment is the most
prominent cellular compartment, and (iv) quantitative and
morphological alterations of the liver subcellular structure by
specific drugs are well described in the literature.
Simulation of the scattering properties of subcellular
compartments of the liver
To compute the /ls' of a given scattering solution, one needs
to know: (i) the morphometric data of the constituting
particle, i.e. y(see equation (3)) or 0 and V (see equations (4)
and (5)) and (ii) the value of the transport cross-section
of the particle, i.e. O"tr (see equations (3), (4) and (5)).
Morphometrical descriptions of the normal and pathological
rat liver by using electron microscopy have been available for
about three decades [17,40-42]. Table 1 summarizes these
quantitative stereo logical descriptions of the ultrastructure
of the rat liver that we used for the calculations of /ls'.
Hepatocytes are by far the most prominent component of
the liver parenchyma (about 80% of the organ volume, see
Table 1) while non-hepatocytes account only for about 6%
of the liver volume [42]. Moreover, the mitochondrial
compartment represents the most important subcellular
compartment of the hepatocyte (about 22% of the cell
volume, see Table 1), in contrast to the other liver cells (about
4% of the cell volume in endothelial or Kupfer cells, [42]).
The mean mitochondrial particle volume is about 0.7 !lffi3 and
the number of particles ranges from 1300-2200 (Table 1).
Mie theory and Rayleigh-Debye-Gans (RDG) approx-
imation are able to calculate the transport cross-section of
a scattering particle. These two mathematical models treat the
scattering particle as a sphere - of radius r and homogeneous
relative refractive index m - illuminated by an unpolarized
light of wavelength').. [6-8]. For our purpose, we assumed
that: (i) the liver is a mixture of 3 types of scattering
particles randomly distributed: the whole hepatocyte, the
nucleus, the mitochondria and the peroxisomes, with regard
to their respective volume fractions listed in Table 1, (ii)
these particles are spheres of radius r with a corresponding
volume listed in Table 1 and (iii) there is no interference
between the different particles for light scattering, particularly
between hepatocyte cytoplasm and the organelles. We chose
450
Table 1. Morphometric data of the rat liver and simulation of scattering properties by Mie theory and RDG approximation*
Volume fraction Number density Single volume
(iJlll3)
Hepatocyte 80.5 4% of the 1.4 x 108 per g wet 5100 160
liver liver
Nucleus 7.6 2.4% of the I per 345 45
hepatocyte hepatocyte
Mitochondria 22.6 5.7% of the 1740462 per 0.67 0.14
hepatocyte hepatocyte
Peroxisomes 1.7 0.6% of the 496 179 per O.l6 0.03
(microbodies) hepatocyte hepatocyte
*.Average electron microscopy data compiled from references [17, 40-42]. Transport cross-section, i.e. crtr, and reduced scattering coefficient, i.e. were
sImulated as described in Material and methods and Results and discussion sections, assuming a refractive index value of 1.04 relative to the extracellular
medium for the hepatocyte and a value of 1.04 relative to the cytoplasm for the nucleus and the organelles.
a refractive index value for the hepatocyte equal to 1.04 relative
to the extracellular medium [43]. For the nucleus and the
organelles, it was assumed to be equal to 1.04 relative to the
cytoplasm, i.e. 1.08 relative to the extracellular medium [44-46].
Table 1 contains the calculated transport cross-section values
of the different scattering particles ofthe liver. Besides the fact
that there is a deviation between the Mie theory and RDG
approximation calculations (see Table 1 and reference [37]), the
hepatocyte cytoplasm has a higher transport cross-section
value than the nucleus (Table 1). The latter compartment has a
higher crtr value than the organelles (mitochondria and
microbodies). This phenomenon illustrates the size-depend-
ence of the transport cross-section [37].
Finally, we calculated the reduced scattering coefficient
due to the cell volume, the nucleus, the mitochondria and
the peroxisomes by taking into account: (i) their calculated
transport cross-section values and (ii) their respective single
volume and volume fraction listed in Table 1. For instance,
'" of the mitochondria in the liver is equal to 0.226 x 0.805
= 0.182. Table 1 summarizes the simulated ~s' given by the
total hepatocyte volume, nucleus, mitochondria and per-
oxisomes under in vivo conditions. The ~ , value from the
whole cells, the nuclei and the peroxiso~es, are included
between 1.2-2.2 cm-I , regardless of the mathematical
model. In contrast, the mitochondrial compartment gives by
far the largest ~s' values, between 12 cm-I and 14 cm-I by
using RDG and Mie models, respectively.
Measurement of the scattering properties of isolated
mitochondria, isolated hepatocytes and rat liver
We measured the reduced scattering coefficient of: (i) isolated
liver mitochondrial suspensions, (ii) isolated hepatocyte
suspensions and (iii) blood-free perfused rat livers, by using
time-resolved spectroscopy and the sample-substitution
protocol. Figure 2A shows that the ~s' value of isolated
mitochondria is proportional to the protein concentration, with
a slope value equal to 0.32 0.05 cm2/mg mitochondrial proteins
(Table 2). The corresponding value of the microscopic scattering
simulated crtr (iJlll') simulated IlS' (em-I)
Mie RDGapp. Mie RDGapp.
7.3O.l 5.6 O.l 2.2 O.l 1.7O.l
0.86 0.06 0.75 0.05 1.40.5 1.2 0.5
(6.3 0.7)x (5.4 0.7)x 14.04.5 12.1 4
10-3 10-3
(2 O.l5)x (1.4 O.l5)x 1.70.7 1.20.5
10-3 10-3
Il:'
coefficient (i.e. the transport cross-section, crt) for isolated
mitochondria is equal to (3.7 1.6) 10-3 ~2 (see Table 2). Figure
2B shows a linear relationship between the ~s' value of isolated
hepatocytes and the dry mass concentration, with a slope value
equal to 0.079 0.003 cm2/mg dry cell (Table 2). The corres-
ponding value of the transport cross-section (i.e. crt) for isolated
hepatocytes is equal to 15 2.1 ~m2 (see Table 2). Finally, the
experimental value obtained for the whole rat liver was equal to
15.9 2.4 cm-I (Table 2).
Contribution of mitochondria to the light scattering of
hepatocytes and whole liver
Assuming that the optical properties ofthe mitochondria are
not significantly altered during mitochondrial preparation
and by knowing the mitochondrial content of the hepatocyte
or the whole liver, one can estimate the contribution of the
mitochondria to the total light scattering of the hepatocyte
or the whole liver. Firstly, from the isolated mitochondrial
data and taking into account the mitochondrial content ofthe
hepatocyte (see Table 2), we estimated the scattering
coefficient of the cell (i.e. crt) due to the mitochondria. This
extrapolated value ofcrtrwas equal to 11.1 1.6 ~2 (see Table
2). The latter value represents about 73% of the total light
scattering measured on the isolated hepatocyte suspension
(11.1 ~2 compared to 15.0 ~2). Secondly, from the isolated
mitochondria data and taking into account the mitochondrial
content of the liver (see Table 2), we estimated the scattering
coefficient ofthe liver (i.e. ~s') due to the mitochondria. This
extrapolated value was equal to 14.7-16.4 cm-I (see Table 2),
which was very close to that measured on the perfused liver
(15.9 2.4 cm-I ). In summary, the comparison of the isolated
mitochondria data with the isolated hepatocyte and whole liver
measurements suggests that the mitochondrial compartment
is the primary factor for light propagation in hepatic tissue,
strengthening the relevance of the preliminary theoretical
study. However, the extrapolation procedure assumes that
the in situ mitochondrial compartment of the liver is comp-
arable to an aqueous suspension of isolated mitochondria.
 451

A B

6
,-.
..t
'eu S
'-'

-",
-",
::i. 4
::i.

2
4 6 8 10 12 14 10 20 30 40 SO 60

Mitochondrial proteins (mglml) Hepatocyte dry mass (mglml)

Fig. 2. Relationship between reduced scattering coefficient and concentration of isolated rat liver mitochondria (A) and isolated rat hepatocytes (B). The
reduced scattering coefficient (11,') was determined by using time-resolved spectroscopy and the sample-substitution protocol as described in Material and
methods section.

Effect of organelle pleomorphism on light scattering of
perfused rat liver
Morphological and quantitative disturbances of organelles
can be induced in the rat liver after a treatment of animals
with specific chemicals. For instance, diethyl ether has been
described as an endoplasmic reticulum-inducing agent [33];
clofibrate - a hypolipidemic drug - as a mitochondrial and
peroxisomal-proliferating agent [23]; and finally, cuprizone-
a copper chelating chemical- as a megamitochondria-induc-
ing agent [24, 25]. Prior to measuring the optical properties of
the perfused liver, the effectiveness of the intoxication was
checked by measuring the activity of marker enzymes such
as: (i) succinate and glycerol 3 phosphate dehydrogenases,
for mitochondria; (ii) catalase, for peroxisome; (iv) acid
phosphatase, for lysosome; and (v) NADPH cytochrome c
oxidoreductase and carboxylesterase, for endoplasmic reticu-
lum. The results are summarized in Table 3. It is to be noted
that: (i) cuprizone intoxication did not significantly change
the enzyme content of the liver (see also [49,50], except for
carboxylesterase, (ii) clofibrate treatment markedly increased
the mitochondrial enzyme activity (especially G3PDH) as well
as the peroxisomal, lysosomal and endoplasmic reticulum
enzyme activity, in accordance with [23, 35, 51-53] and (iii)

Table 2. Scattering properties of rat liver and isolated hepatocytes and mitochondria, measured by time-resolved spectroscopy

rlIl:1d[mass] measured a tt extrapolated a" measured 11: extrapolated 11:

(cm2 (mg proteinstl or (1lffi2) CIlffi2) (em-I) (em-I)
cm2(mg dry celltl)

Isolated mitochondria 0.32 0.05 (3.7 1.6) x 10-3*

Isolated hepatocytes 0.079 0.008 15.1 2.1 * 11.1 1.6"
Perfused liver 15.92.4 14.7 5.5
(n=4) 16.42.P

*Calculated from the following equation: a tt = ~ ~, where N is the number density (10). N is equal to (8.7 2.5) x 109 mitochondria per mg
N d[mass]
mitochondrial protein [47, 48] or to (5.1 0.5) x 105 cells per mg cell dry mass. "Calculated from the isolated mitochondria data and assuming the
mitochondrial content of hepatocytes to be equal to 0.18 mg of mitochondrial proteins per mg of cell dry mass. Calcu1ated from the measured
a" of isolated mitochondria and using equation (5) (see Results and discussion section). The single volume, V, is equal to 0.375 0.080 11m3 for
isolated mitochondria [47, 48] and the volume fraction of the mitochondria in the liver, 0, is equal to 0.182 0.055 (see Table 1). 'Calculated
from the isolated mitochondria data and assuming the mitochondrial content of liver to be equal to 48 5 mg of mitochondrial proteins per g of
wet liver.
452

diethyl ether injection induced an increase in the endoplasmic Table 4. Optical properties of perfused rat liver after treatment with various
reticulum enzyme activity (see also [33, 34]). drugs*

Table 4 shows the effect of drug intoxication on the optical

properties of the blood-free perfused rat liver. Diethyl ether
(i.e. an endoplasmic reticulum-inducing agent) treatment
slightly increased the reduced scattering coefficient of the
liver: 18.2 cm-l vs. 14.8 cm-l as control value, at 780 nm; and
14.6 cm-l vs. 12.4 cm-l as control value, at 830 nm (Table 4).
Moreover, clofibrate treatment induced a 2 fold increase
in the !ls' value: 23 cm- l vs. 14.8 cm- l as control value, at
780 nm; and 23.8 cm-l vs. 12.4 cm-l as control value, at 830
nm (Table 4). The latter results correlate with the biochemical
and cytological data exhibiting an increase (l.4--l.9 times) in
the mitochondrial number and volume fraction [23, 36, 51-
53] as well as a proliferation (by a factor 1.3-6.2) of the
microbodies (e.g. peroxisomes and lysosomes) [23,36,51,
52]. Interestingly, cuprizone intoxication also led to a 2 fold
increase in the !lS' value of the liver: 22.6 cm-l vs. 14.8 cm-l as
control value, at 780 nm; and 22.9 cm-l vs. 12.4 cm-l as control
value, at 830 nm (Table 4). This increase in light scattering
induced by cuprizone is therefore related to the increase in the
size ofthe mitochondria (by a factor 1.6) already observed by
electron microscopy in the liver of cuprizone-treated animals
[49].
Correlation between light scattering and the
mitochondrial content of normal tissues and
transplantable rodent tumors
Based upon the realization that various rodent tissues exhibit
different degrees of differentiation in their mitochondrial
compartment, we investigated the optical properties of both
nonnal and neoplastic tissues. We obtained a wide range of
mitochondrial concentrations in the tissue by analyzing: (i)
nonnal rat tissues (i.e. perfused liver, brain, white adipose
tissue, skeletal muscle) and, (ii) exteriorized rat and mouse
solid tumors of different histological type (i.e. hepatoma,
I\'at 780 nm 11,' at 830 nm
(cm') (cm")
Control (n =4) 14.8 1.2 12.4 1.6
Cuprizone (n =4) 22.62 22.93.6
Clofibrate (n = 3) 23 2 23.8 4.7
Diethyl ether (n = 3) 18.2 2.5 14.62.8
*Cuprizone, clofibrate and diethyl ether were administered to rats as
described in Material and methods section. The reduced scattering
coefficient of perfused livers was determined at 780 and 830 nm as described
in Material and methods section by using time-resolved spectroscopy and
a sample-substitution procedure.
glioma, mammary adenocarcinoma). For each sample, we deter-
mined the absorption and reduced scattering coefficients by
means of TRS, and measured the succinate dehydrogenase
specific activity on tissue homogenates as an indicator of
the mitochondrial content of tissues.
Figure 3A represents the relationship between the value
of !ls' measured by TRS and SDH activity with respect to
tissue and tumor type. Among the normal tissues, the
skeletal muscle had the lowest !ls' and also a low SDH
activity. The liver was characterized by a high !ls' value and
the highest ~DH content, and the brain had intennediary
values of !ls' and SDH activity (Fig. 3A). In general, the
tumors had a relatively low scattering coefficient (between
7.6-10.1 cm-l ) and a low SDH activity (between 0.5-2.5 U/g
wet weight) (Fig. 3A). The SDH activity and the !ls' values
of the gliomas (i.e. G26 and 9L) were respectively 4.0-
9.2 and 1.6-1.8 times lower than those of the rat brain. The
SDH activity and the !ls' values of the Q7 hepatoma were
respectively 3.4 and 1.8 times lower than those of the rat
liver. Thus, Fig. 3A shows that the reduced light scattering
coefficient is roughly proportional to the SDH content of
tissue. However, there is an exception to this relationship:
the white adipose tissue (WAT) is characterized by a high
scattering coefficient and a very low SDH content (Fig. 3A).

Table 3. Enzyme activity of perfused rat liver after treatment with drugs*

SDH G3PDH Catalase Acid Carboxyl NADPHcyt.c

(U/g wet wt) (U/gwetwt) (U/mg wet wt) phosphatase esterase oxidoreductase
(U/g wet wt) (Ulmg wet wt) (U/g wet wt)

Control (n =4) 13.3 1.3 1.8 0.3 25.72.2 5.20.4 0.180.04 7.2 1.3
Cuprizone (n =4) 15.41.7 1.9 0.2 25.5 4.3 5.80.2 0.27 0.03 7.4 1.1
Clofibrate (n =3) 19.4 1.1 3.70.3 36.5 7.0 6.60.4 0.300.02 9.7 1.4
Diethyl ether (n = 3) 13.9 1.2 1.80.2 24.94.6 5.3 0.4 0.320.02 13.9 2.2

*Drugs were administered to rats as described in Material and methods section. The activity of enzymes was measured on liver homogenates as described in
Material and methods section (one unit of enzyme activity is defined as one micromole of substrate or product transformed per min). Marker enzymes are:
(a) for mitochondria, SDH (succinate oxidoreductase, BC 1.1.99.1) and G3PDH (sn-glycerol-3-phosphate: NAD 2-oxidoreductase, BC 1.1.1.8); (b) for
peroxisomes, catalase (hydrogen peroxide:hydrogen peroxide oxidoreductase, EC 1.11.1.6); (c) for lysosomes, acid phosphatase (orthophosphoric monoester
phosphohydrolase, BC 3.1.3.2); and (d) for endoplasmic reticulum, NADPH cytochrome c oxidoreductase and carboxylesterase (carboxylic-ester hydrolase,
BC 3.1.1.1).
 453

A reported. To confirm the dependence of photon scattering on

20 ~ mitochondrial content, we estimated the mitochondrial protein
content of normal tissues and tumors. This was done by taking

15
TWAT
Iive+ the ratio between the value of the SDH activity measured on
the whole tissue and that measured on the mitochondrial
fraction isolated from the respective cells (13762A ascites
b":'Lv tumor cell) or solid tissues. As shown in Fig. 3B, the scattering
coefficient value correlates with the mitochondrial protein
13762A
content of the tissue: the greater the mitochondrial protein
=.,'" 10 +1: .. ' content, the higher the photon scattering. It should be

~.f~;~i:7
lilt
noticed that the linear regression of the latter graph gives a
slope value of 0.26 0.03 cm-l (mg proteins/g wet wttl.
Assuming that the tissue density is about 1.07 glml, the slope
skeL muscle
is equal to 0.28 0.04 cm-2/mg of mitochondrial proteins,
o 2 4 6 8 10 which is not significantly different to the value determined
Tissue SDH activity for isolated rat liver mitochondria from Fig. 2A (Table 2).
(U/g wet wt) The in vivo morphometric analysis of the mitochondrial
compartment in normal mammalian tissues has been well
20
B documented. For instance, the mitochondrial volume
fraction determined by means of electron microscopy is
equal to: 16.9--28.3% in liver [17, 40--42], about 8% in brain
[14],3.9--4.8% in skeletal muscle [14, 15] and about 1.7% in
WAT [56]. When compared, these values are in accordance
with our determinations of SDH activities and mito-
chondrial protein contents. The correspondence between
our biochemical assays and the electron microscopic data
III
=., of the literature strengthens the validity of our approach
for reliably evaluating the in vivo tissue mitochondrial
concentration. The neoplastic tissues we analyzed in this
work have a relatively low mitochondrial content, ex-
pressed as SDH activity or protein per g wet weight. This
o 10 20 SO 30 40 60 is consistent with the fact that they have a rapid growth rate,
which is a common characteristic of 'poorly differentiated
Tissue mitochondrial proteins tumors' (for review see [18]).
(mglg wet wt)
The fact that the IlS' of the white adipose tissue is un-
Fig. 3. Relationship between reduced scattering coefficient and mito- related to its mitochondrial content suggests that other
J.l:
chondrial content of various tissues. (A) as a function of the SDH activity parameters might be involved in photon migration in a tissue.
of the tissue. The dotted line represents the curve fitting corresponding to
Morphometric examinations of adipose tissue have demon-
J.l:
the formula y= 6.7 + l.lx (R2 = 0.86). (B) as a function of the mitochondrial
strated that the lipid droplets represent about 88% of the
protein content of the tissue. The dotted line represents the curve fitting
corresponding to the equation y = 3.6 + 36x/(98 + x) (R2 = 0.977). The cellular volume [56]. Such lipid particles are potentially
reduced scattering coefficient of tissues and tumors was determined by candidates for light scattering in the fat tissue because of a
using time-resolved spectroscopy and the sample-substitution protocol as combination of a large volume fraction and a high refractive
described in Material and methods section. The SDH activity was measured
index [57].
as described in Material and methods section. Mitochondrial protein content
was calculated from the SDH activity of the whole tissue divided by that of In conclusion, this article demonstrates that subtle changes
the isolated mitochondrial fraction. in mitochondrial content appear to produce large changes in
tissue light scattering, which can be detected and quantified
Although SDH activity has been described as a reliable by time-resolved spectroscopy. This emphasizes the use-
probe for studying mitochondrial biogenesis in skeletal fulness of a non-invasive method for metabolic studies
muscle [16], SDH activity measured on the whole tissue may under normal and pathological conditions. However, photon
not be a reliable marker for mitochondrial content when migration in tissue is sensitive to other intracellular com-
normal and neoplastic tissues are compared. For instance, a ponents, such as: (i) peroxisomes and lysosomes, in the liver
decrease [54] as well as an increase [55] in SDH in both of intoxicated animals (this study), and (ii) lipid droplets in
whole tumors or isolated mitochondrial fractions have been white adipose tissue (this study) or in fatty rat liver [57].
454
Acknowledgements
This work is supported in part by NIH grants CA50766 and
CA56679-0 IAI. We thank Dr. R. Cooke for his contribution
for editing the manuscript.
References
1. Wilson BC, Sevick E, Patterson MS, Chance B: Time-dependent optical
spectroscopy and imaging for biomedical applications. Proc IEEE 80:
918--930,1992
2. Patterson MS, Chance B, Wilson BC: Time resolved reflectance and
transmittance for the non-invasive measurement of tissue optical
properties. Appl Opt 28: 2331-2336,1989
3. Sevick EM, Chance B, Leigh J, Nioka S, Maris M: Quantitation of time-
and frequency-resolved optical spectra for the determination of tissue
oxygenation. Anal Biochem 195: 330--351, 1991
4. Chance B, Leigh J, Miyake H, Smith D, Nioka S, Greenfeld R, Finlander
M, Kaufmann K, Levy W, Young M, Cohen P, Yoshioka H, Boretsky R:
Comparison of time-resolved and un-resolved measurements of
deoxyhemoglobin in brain. Proc Natl Acad Sci USA 85: 4971--4975, 1988
5. Chance B, Nioka S, Kent J, McCully K, FountainM, Greenfeld R, Holtom
G: Time-resolved spectroscopy of hemoglobin and myoglobin in
resting and ischemic muscle. Anal Biochem 174: 698-707, 1988
6. Kerker M: The Scattering of Light, and other Electromagnetic Radiation.
Academic Press, New York, USA, 1969,pp 1-666
7. Ishimaru A: Wave Propagation and Scattering in Random Media.
Academic Press, New York, USA, 1978, pp 1-572
8. Kerker M: Elastic and inelastic light scattering in flow cytometry.
Cytometry4: 1-10, 1983
9. Steinkamp JA: Flow cytometry. Rev Sci Instrum 55: 1375-1400, 1984
10. Beauvoit B, Liu H, Kang K, Kaplan PD, Miwa M, Chance B: Character-
ization of absorption and scattering properties for various yeast strains
by time-resolved spectroscopy. Cell Biophys 23: 1-16, 1993
11. Hoffmann HP, Avers C: Mitochondrion of yeast: Ultrastructural
evidence for one giant, branched organelle per cell. Nature 181: 749-
751,1973
12. Miyakawa I, Aoi H, Sando N, Kuroiwa T: Fluorescence microscopic
studies of mitochondrial nuc1eoids during meiosis and sporulation in
the yeast, Saccharomyces cerevisiae. J Cell Sci 66: 21-38,1984
13. Stevens B: Mitochondrial structure. In: IN Strathern, EW Jones, JR
Broach (eds). The Molecular Biology of the Yeast Saccharomyces
cerevisiae. Life Cycle and Inheritance. Cold Spring Harbor Laboratory,
Cold Spring Harbor, New York, USA, 1981, pp 471-504
14. Else PL, Hulbert AJ: An allometric comparison of the mitochondria of
mammalian and reptilian tissues: The implications for the evolution of
endothermy. J Comp Physiol B 156: 3-11, 1985
IS. Schwerzmann K, Hoppeler H, Kayar SR, Weibel ER: Oxidative capacity
of muscle and mitochondria: Correlation of physiological, biochemical,
and morphometric characteristics. Proc Nat! Acad Sci USA 86: 1583-
1587,1989
16. Takahashi M, Hood DA: Chronic stimulation-induced changes in
mitochondria and performance in rat skeletal muscle. J Appl Physiol
74:934-941,1993
17. Jakovcic S, Swift HH, Gross NJ, Rabinowitz M: Biochemical and
stereological analysis of rat liver mitochondria in different thyroid
states. J Cell Bioi 77: 887-901, 1978
18. Pedersen PL: Tumor mitochondria and the bioenergetics of cancer
cells. Prog Exp Tumor Res 22: 190--274, 1978
19. Chen LB: Fluorescent labeling of mitochondria. Meth Cell Bioi 29: 103-
123,1989
20. Smiley ST, Reers M, Mottola-Hartshorn C, Lin M, Chen A, Smith TW,
Steele GD, Chen LB: Intracellular heterogeneity in mitochondrial
membrane potentials revealed by a J-aggregate-forming lipophilic cation
JC-1. ProcNat! Acad Sci USA 88: 3671-3675, 1991
21. Lemasters JJ, Di Guiseppi J, Nieminen AL, Herman B: Blebbing, free
Ca2 + and mitochondrial membrane potential preceding cell death in
hepatocytes. Nature 325: 78-81,1987
22. Reber BFX, Somogyi R, Stucki JW: Hormone-induced intracellular
calcium oscillations and mitochondrial energy supply in single
hepatocytes. Biochim Biophys Acta 1018: 190--193, 1990
23. Hess R, Staubli W, Riess W: Nature of the hepatomegalic effect
produced by ethyl-chlorophenoxy-isobutyrate in the rat. Nature 27:
856-858, 1965
24. Tandler B, Hoppel CL: Studies on giant mitochondria. Ann NY Acad
Sci 488: 65-81, 1986
25. Susuki K: Giant hepatic mitochondria: Production in mice fed with
cuprizone. Science 163: 81-82,1969
26. Liu H, Miwa M, Beauvoit B, Wang NG, Chance B: Characterization of
absorption and scattering properties of small-volume biological
samples using time-resolved spectroscopy. Anal Biochem 213: 378-
385,1993
27. Berry MN, Edwards AM, Barrit OJ, Grivell MB, Halls HJ, Gannon BJ,
Friend DS: Isolated hepatocytes. Preparation, properties and
applications. In: RH Burdon, PH Knippenberg (eds). Laboratory
Techniques in Biochemistry and Molecular Biology. Elsevier, New York,
USA, 1991, vol. 21,pp 1--460
28. Johnson D, Lardy H: Isolation of liver or kidney mitochondria. Meth
EnzyTllolI0:94-96,1967
29. Clark JB, Nicklas WJ: The metabolism of rat brain mitochondria.
Preparation and characterization. J Bioi Chern 245: 4724--4731, 1970
30. Moreadith RW, Fiskum G: Isolation of mitochondria from ascites tumor
cells permeabilized with digitonin. Anal Biochem 137: 360--367,1984
31. Rodriguez-Vico F, Martinez-Cayuela M, Garcia-Peregrin E, Ramirez H:
A procedure for eliminating interferences in the Lowry method of protein
determination. Anal Biochem 183: 275-278,1989
32. Hatefi Y, Stiggall DL: Preparation and properties of sucdnate:ubiquinone
oxidoreductase (complex II}. MethEnzyTllol 53: 21-27,1978
33. Ross WT, Cardell RR: Proliferation of smooth endoplasmic reticulum
and induction of microsomal drug-metabolizing enzymes after ether or
halothane. Anesthesiology 48: 325-331, 1978
34. Brown BR, Sagalyn AM: Hepatic microsomal enzyme induction by
inhalation of anesthetics: Mechanism in the rat. Anesthesiology 40:
152-161,1974
35. Berge RK, Skrede' S, Farstad M: Effects of clofibrate on the intracellular
localization of palmitoyl-CoA hydrolase and palmitoyl-L-carnitine
hydrolase in rat liver.'FEBS Lett 124: 43--47, 1981
36. Hardeman D, Zomer HWM, Schutgens RBH, Tager JM, van den Bosch
H: Effect of peroxisome proliferation on ether phospholipid bio-
synthesizing enzymes in rat liver. Int J Biochem 22: 1413-1418, 1990
37. GraaffR, Aarnoudse JG, Zijp JR, Sioot PMA, de Mul FFM, Greve J,
Koelink MH: Reduced light-scattering properties for mixtures of
spherical particles: A simple approximation derived from Mie
calculation. Appl Opt 31: 1370--1376,1992
38. Reynolds L, Johnson C, Ishimaru A: Diffusive reflectance from a
finite blood medium: To the modelling of fiber optic catheters.
Appl Opt 15: 2059-2067, 1976
39. Steinke JM, Shepherd AP: Comparison of Mie theory and the light
scattering of red blood cells. Appl Opt 27: 4027--4033, 1988
40. Loud AV: A quantitative stereological description of the
ultrastructure of normal rat liver parenchymal cell. J Cell Bioi
37: 27-39, 1968
41. Weibel ER, Staubli W, Gnagi HR, Hess FA: Correlated morphometric

and biochemical studies on the liver cell. 1. Morphometric model,
stereo logic methods, and nonnal morphometric data for rat liver. J Cell
Bioi 42: 68-84, 1969
42. Blouin A, Bolender RP, Weibel ER: Distribution of organelles and
membranes between hepatocytes and non-hepatocytes in the rat liver
parenchyma. A stereological study. J Cell Bioi 72: 441-455, 1977
43. Bolin FP, Preuss LE, Taylor RC, Ference RJ: Refractive index of some
mammalian tissues using a fiber optic cladding method. Appl Opt 28:
2297-2303,1989
44. Cross DA, Latimer P: Angular dependence of scattering from
Escherichia coli cells. Appl Opt II: 1225-1228, 1972
45. Fujime S, Takasaki-Ohsita M, Miyamoto S: Dynamic light scattering
from polydisperse suspensions of large spheres. Characterization of
isolated secretory granules. Biophys J 54: 1179-1184, 1988
46. Sioot PMA, Hoekstra AG, Figdor CG: Osmotic response oflymphocytes
measured by means of forward light scattering: Theoretical con-
siderations. Cytometry9: 636--641,1988
47. Gear ARL, Bednarek JM: Direct counting and sizing of mitochondria in
solution. J Cell Bioi 54: 325-345, 1972
48. Schwerzmann K, Cruz-Orive LM, Eggman R, Sanger A, Weibel ER:
Molecular architecture ofthe inner membrane of mitochondria from rat
liver: A combined biochemical and stereo logical study. J Cell Bioi
102: 97-103, 1986
49. Flatmark T, Kryvi H, Tangeras A: Induction of megamitochondria
by cuprizone (biscyclohexanone oxaldihydrazone). Evidence for
455
an inhibition of the mitochondrial division process. Eur J Cell
Bioi 23: 141-148, 1980
50. Wakabayashi T, Asano M, Kurono C, Ozawa T: Mechanism of the
formation of megamitochondria induced by copper-chelating agents.
II. Isolation and some properties of megamitochondria from the
cuprizone-treated mouse liver. Acta Path Jap 25: 39-49, 1975
51. Kurup CKR, Aithal HN, Ramasanna T: Increase of hepatic mitochondria
on administration of ethyl alpha-p-chlorophenoxyisobutyrate to the rat.
Biochem J 116: 773-779, 1970
52. Anthony LE, Schmucker DL, Mooney JS, Jones AL: A quantitative
analysis of fine structure and drug metabolism in livers of clofibrate-
treated young adult and retired breeder rats. J Lipids Res 19: 151-163,
1978
53. Lipsky NG, Pedersen PL: Perturbation by clofibrate of mitochondrial
levels in animals cells. J Bioi Chern 257: 1473-1481, 1982
54. Myers MW, Bosmann HE: Mitochondrial protein content and enzyme
activity of Reuber hepatoma H35. Cancer Res 34: 1989-1994, 1974
55. Matsuno T: Pathway of glutamate oxidation and its regulation in the
HuH13 line of human hepatoma cells. J Cell Physiol 148: 290-294, 1991
56. Loncar D, Afzelius BA, Cannon B: Epididymal white adipose tissue after
cold stress in rats. I. Non-mitochondrial changes. J Ultrastr Mol Str Res
101: 109-122, 1988
57. Kitai T, Beauvoit B, Chance B: Optical determination of fatty change of
the graft liver with near-infrared time-resolved spectroscopy.
Transplantation 62: 642--647, 1996
Molecular and Cellular Biochemistry 184: 457-460, 1998,

Index to Volume 184

Ala-Rami A, see Hassinen IE et ai,

Aliev MK, van Dorsten FA, Nederhoff MG, van Echteld CJA, Veksler V, Nicolay K, Saks VA: Mathematical model of
compartmentalized energy transfer: Its use for analysis and interpretation of II P-NMR studies of isolated heart of creatine
kinase deficient mice 209-229
Anflous K, see Ventura-Clapier R et ai,
Averet N, Fitton V, Bunoust 0, Rigoulet M, Guerin B: Yeast mitochondrial metabolism: From in vitro to in situ quantitative
study 67-79

Bastin ME, see Kemp GJ et ai,

Beauvoit B, Chance B: Time-Resolved Spectroscopy of mitochondria, cells and tissues under normal and pathological
conditions 445-455
Benders A, see Steeghs K et ai,
Bernardi P, see Di Lisa F
Boehm E, see Ventura-Clapier R et ai,
Boero J, see Qin Wet ai,
Brand MD: Top-down elasticity analysis and its application to energy metabolism in isolated mitochondria and intact cells 13-20
Braun U, see Seppet EK et al.
Bunoust 0, see Averet N et ai,

Cascante M, see Kholodenko BN et al.

Chance B, see Beauvoit B
Cheng J, see Qin W et ai,
Clark JF, see Kemp GJ et ai,

de Bie M, see Steeghs K et ai,

de Groot B, see van Beek JHGM et ai,
de Haan A, see Steeghs K et al.
Demin OV, Kholodenko BN, Skulachev VP: A model of 0'2- generation in the complex III of the electron transport chain 21-33
Devin A, Espie P, Guerin B, Rigoulet M: Energetics of swelling in isolated hepatocytes: A comprehensive study 107-121
Devin A, see ~everve X et ai,
Di Lisa F, Bernardi P: Mitochondrial function as a determinant of recovery or death in cell response to injury 379-391
Diolez P, see Saks V et ai,

Dolder M, see Stachowiak et ai,
Dos Santos P, see Saks V et al.
Dzeja PP, Zeleznikar RJ, Goldberg ND: Adenylate kinase: Kinetic behavior in intact cells indicates it is integral to multiple
cellular processes 169-182

Eder M, see Schlattner U et ai,

Eijgelshoven MHJ, see van Beek JHGM et ai,
Espie P, see Devin A et al.
Espie P, see Leverve X et al.

Fitton V, see Averet N et ai,

Fontaine E, see Rigoulet M et ai,
Forstner M, see Schlattner U et al.
Fritz-Wolf K, see Schlattner U et al.
458

Gellerich JF, see Nicolay K et al.

Gellerich FN, see Saks V et al.
Goldberg ND, see Dzeja PP et al.
Guerin B, see Averet N et al.
Guerin B, see Devin A et al.
Guerin B, see Rigoulet M et al.
Guerrero-Ontiveros ML, Wallimann T: Creatine supplementation in health and disease. Effects of chronic creatine ingestion
in vivo: Down-regulation of the expression of creatine transporter isoforrns in skeletal muscle 427--437

Hahn D, see Neubauer S et al.

Hak JB, see van Beek JHGM et al.
Hansford RG, Zorov D: Role of mitochondrial calcium transport in the control of substrate oxidation 359-369
Hassinen IE, Vuorinen KH, Ylitalo K, Ala-Riimi A: Role of cellular energetics in ischemia-reperfusion and ischemic pre-
conditioning of myocardium 393--400
Heerschap A, see Steeghs K et al.
Hom M, see Neubauer S et al.

Ichas F, see Jouaville LS et al.

Jacob W, see Steeghs K et al.

Jap P, see Steeghs K et al.
Jost C, see Steeghs K et al.
Jouaville LS, Ichas F, Mazat J-P: Modulation of cell calcium signals by mitochondria 371-376

Kaasik A, see Seppet EK et al.

Kay L, see Rossi A et al.
Kay L, see Saks VA et al.
Kemp GJ, Manners DN, Clark JF, Bastin ME, Radda GK: Theoretical modelling of some spatial and temporal aspects of the
mitochondrion/creatine kinase/myofibril system in muscle 249-289
Kholodenko BN, Rohwer JM, Cascante M, Westerhoff HV: Subtleties in control by metabolic channelling and enzyme
organization 311-320
Kholodenko BN, see Demin OV et al.
Khuchua Z, see Qin W et al.
Kochsiek K, see Neubauer S et al.
Korzeniewski B: Is it possible to predict any properties of oxidative phosphorylation in a theoretical way? 345-358
Kruiskamp MJ, see Nicolay K et al.
Kunz WS, see Saks VA et al.
Kuznetsov AY, see Saks VA et al.
Kuznetsov A, see Ventura-Clapier R et al.

Letellier T, Malgat M, Rossignol R, Mazat J-P: Metabolic control analysis and mitochondrial pathologies 409--417
Leverve X, Sibille B, Devin A, Piquet M-A, Espie P, Rigoulet M: Oxidative phosphorylation in intact hepatocytes: Quanti-
tative characterization of the mechanisms of change in efficiency and cellular consequences 53--65
Leverve X, see Rigoulet M et al.
Leverve X, see Saks VA et al.

Malgat M, see Letellier T et al.

Manners DN, see Kemp GJ et al.
Mazat J-P, see Jouaville LS et al.
Mazat J-P, see Letellier T et al.
Minajeva A, see Seppet EK et al.
 459

NederhoffMG, see Aliev MK et al.

Neubauer S, Hom M, Hahn D, Kochsiek K: Clinical cardiac magnetic resonance spectroscopy - present state and future
directions 439-443
Nicolay K, van Dorsten FA, Reese T, Kruiskamp MJ, Gellerich JF, van Echteld CJA: In situ measurements of creatine kinase
flux by NMR. The lessons from bioengineered mice 195-208
Nicolay K, see Aliev MK et at.

Oerlemans F, see Steeghs K et at.

Ohisalo JJ, see Seppet EK et al.
Olivares J, see Saks VA et at.
Oliviero P, see Rappaport L et at.
Ouhabi R, see Rigoulet M et al.

Paju K, see Seppet EK et at.

Payne RM, see Qin W et at.
Pette D, see Steeghs K et al.
Piquet M-A, see Leverve X et at.

Qin W, Khuchua Z, Cheng J, Boero J, Payne RM, Strauss AW: Molecular characterization of the creatine kinases: and
some historical perspectives 153-167

Radda GK, see Kemp GJ et al.

Rappaport L, Oliviero P, Samuel JL: Cytoskeleton and mitochondrial morphology and function 101-105
Reese T, see Nicolay K et at.
Rigoulet M, Leverve X, Fontaine E, Ouhabi R, Guerin B: Quantitative analysis of some mechanisms affecting the yield
of oxidative phosphorylation: Dependence upon both fluxes and forces 35-52
Rigoulet M, see Averet N et at.
Rigoulet M, see Devin A et al.
Rigoulet M, see Leverve X et at.
Rigoulet M, see Saks VA et at.
Rohwer JM, see Kholodenko BN et al.
Rossi A, Kay L, Saks V: Early ischemia-induced alterations of the outer mitochondrial membrane and the intermembrane
space: A potential cause for altered energy transfer in cardiac muscle? 401-408
Rossi A, see Saks VA et al.
Rossignol R, see Letellier T et at.
Ruitenbeek W, see Steeghs K et al.

Saks VA, Dos Santos P, Gellerich FN, Diolez P: Quantitative studies of enzyme-substrate compartmentation, functional coupling
and metabolic channelling in muscle cells 291-307
Saks VA, Ventura-Clapier R, Leverve X, Rigoulet M, Rossi A: What do we not know of cellular bioenergetics? - a general
view on the state of the art 3-9
Saks VA, Veksler VI, Kuznetsov AV, Kay L, Sikk P, Tiivel T, Tranqui L, Olivares J, Winkler K, Wiedemann F, Kunz WS:
Permeabilized cell and skinned fiber techniques in studies of mitochondrial function in vivo 81-100
Saks VA, see Aliev MK et al.
Saks VA, see Rossi A et al.
Samuel JL see Rappaport L et al.
Schlattner U, Forstner M, Eder M, Stachowiak 0, Fritz-WolfK, Wallimann T: Functional aspects of the X-ray structure of
mitochondrial creatine kinase: A molecular physiology approach 125-140
Schlattner U, see Stachowiak 0 et at.
Seppet EK, Kaasik A, Minajeva A, Paju K, Ohisalo JJ, Vetter R, Braun U: Mechanisms ofthyroid hormone control over
sensitivity and maximal contractile responsiveness to ~-adrenergic agonists in atria 419-426
Sibille B, see Leverve X et al.
460

Sikk P, see Saks VA et al.

Skulachev VP, see Demin OV et al.
Stachowiak 0, Schlattner U, Dolder M, Wallimann T: Oligomeric state and membrane binding behaviour of creatine kinase
isoenzymes: Implications for cellular function and mitochondrial structure 141-151
Stachowiak 0, see Schlattner U et al.
Steeghs K, Oerlemans F, de Haan A, Heerschap A, Verdoodt L, de Bie M, Ruitenbeek W, Benders A, Jost C, van Deursen J,
Tullson P, Terjung R, Jap P, Jacob W, Pette D, Wieringa B: Cytoarchitectural and metabolic adaptations in muscles
with mitochondrial and cytosolic creatine kinase deficiencies 183-194
Strauss AW see Qin Wet al.

Terjung R, see Steeghs K et al.

Tian X, see van Beek JHGM et al.
Tiivel T, see Saks VA et al.
Tranqui L, see Saks VA et al.
Tullson P, see Steeghs K et al.

van Beek JHGM, Tian X, Zuurbier CJ, de Groot B, van Echteld CJA, Eijgelshoven MHJ, Hak JB: The dynamic regulation
of myocardial oxidative phosphorylation: Analysis of the response time of oxygen consumption 321-344
van Deursen J, see Steeghs K et al.
van Dorsten FA, see Aliev MK et al.
van Dorsten FA, see Nicolay K et al.
van Echteld CJA, see Aliev MK et at.
van Echteld CJA, see Nicolay K et al.
van Echteld CJA, see van Beek JHGM et al.
Veksler V, see Aliev MK et al.
Veksler V, see Ventura-Clapier R et al.
Veksler VI, see Saks VA et at.
Ventura-Clapier R, Kuznetsov A, Veksler V, Boehm E, Anflous K: Functional coupling of creatine kinases in muscles:
Species and tissue specificity 231-247
Ventura-Clapier R, see Saks VA et al.
Verdoodt L, see Steeghs K et al.
Vetter R, see Seppet EK et al.
Vuorinen KH, see Hassinen IE et al.

Wallimann T, see Guerrero-Ontiveros ML

Wallimann T, see Schlattner U et al.
Wallimann T, see Stachowiak 0 et al.
Westerhoff HV, see Kholodenko BN et al.
Wiedemann F, see Saks VA et al.
Wieringa B, see Steeghs K et al.
Winkler K, see Saks VA et al.

Ylitalo K, see Hassinen IE et al.

Zeleznikar RJ, see Dzeja PP et al.

Zorov D, see Hansford RG
Zuurbier CJ, see van Beek JHGM et al.