(Methods in Molecular Biology 1299) Gary R. Skuse, Maureen C. Ferran (Eds.) - Cardiomyocytes - Methods and Protocols-Humana Press (2015)

Methods in
Molecular Biology 1299
Gary R. Skuse
Maureen C. Ferran Editors
Cardio-
myocytes
Methods and Protocols
METHODS IN MOLECULAR BIOLOGY
Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK
For further volumes:

http://www.springer.com/series/7651
Cardiomyocytes
Methods and Protocols
Edited by
Gary R. Skuse and Maureen C. Ferran

Rochester Institute of Technology, Rochester, NY, USA
Editors
Gary R. Skuse Maureen C. Ferran
Rochester Institute of Technology Rochester Institute of Technology
Rochester, NY, USA Rochester, NY, USA
ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-2571-1 ISBN 978-1-4939-2572-8 (eBook)
DOI 10.1007/978-1-4939-2572-8
Library of Congress Control Number: 2015935077
Springer New York Heidelberg Dordrecht London

# Springer Science+Business Media New York 2015
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
Cover Illustration: CytivaTM Plus Cardiomyocytes: The deconvolved Image was acquired on DeltaVision OMX by
Angela Williams (GE Healthcare). After 14 days of growth the cells were stained for cardiac Troponin I (Red) and
a-Actinin (Green)
Printed on acid-free paper
Humana Press is a brand of Springer

Springer Science+Business Media LLC New York is part of Springer Science+Business Media (www.springer.com)
Preface
Heart disease is responsible for untold global morbidity and mortality. Traditional medical
approaches to the treatment of heart disease often strive to ameliorate damage and to
prevent future damage, but they cannot reverse what has already happened. This is
especially apparent in the most extreme instances where heart transplantation is used to
replace the entire organ. Unfortunately, the supply of donor hearts cannot keep pace with
the demand; something must be done to enable us to repair damaged cardiac tissue and
generate whole organs when needed. Despite the fact that death rates from coronary heart
disease are falling, heart disease remains the leading cause of death worldwide.
This volume, Cardiomyocytes: Methods and Protocols, has been assembled for scientists
interested in basic and applied biomedical research directed toward understanding the
development, genetics, and function of cardiomyocytes. The methods and protocols
contained herein address cell culture techniques, cardiomyocyte differentiation and redif-
ferentiation, experimental induction of cardiomyopathies, introducing genes into cardio-
myocytes, genomic approaches to the understanding of cardiomyocytes, cryopreservation
of neonatal cardiomyocytes, and modeling of cardiomyocyte function. Among the chapters
of this work, readers will find complimentary areas of cardiomyocyte science that, taken
together, should inform individuals with a broad range of interests.
These collected contributions were written by current and nascent leaders in the field
of cardiomyocyte biology. Together the authors have provided a wealth of methods that
can be used to further explore the many aspects of cardiomyocyte biology that we need to
understand in order to better grasp the development and function of these cells and to
develop the next generation of effective therapies. The chapters are organized thematically
with regard to cardiovascular disease, modelling of cardiomyocytes function, isolation of
cells, induced differentiation of cells into cardiomyocytes, gene transfer into cardiac myo-
cytes, gene expression analysis, and the application of next-generation sequencing toward
furthering our understanding of cardiovascular disease.
Of course it is impossible to include contributions from every researcher who is
contributing to this important field. Instead we have compiled a collection of chapters
that together represent some of the leading and potentially most impactful work. We hope
you find them informative and useful in your own laboratories.
Rochester, NY, USA Gary R. Skuse

Maureen C. Ferran
v
Contents
Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Generating Primary Cultures of Murine Cardiac Myocytes

and Cardiac Fibroblasts to Study Viral Myocarditis . . . . . . . . . . . . . . . . . . . . . . . . 1
Barbara Sherry
2 Enrichment of Cardiomyocytes in Primary Cultures of Murine
Neonatal Hearts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Sreejit Parameswaran, Rajalakshmi Santhakumar,
Prasanna Vidyasekar, and Rama S. Verma
3 Deep Sequencing of Cardiac MicroRNA-mRNA Interactomes
in Clinical and Experimental Cardiomyopathy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Scot J. Matkovich and Gerald W. Dorn II
4 Next-Generation Sequencing Technology in the Genetics
of Cardiovascular Disease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Cecilia Vecoli
5 Computational Cardiac Electrophysiology: Implementing Mathematical
Models of Cardiomyocytes to Simulate Action Potentials of the Heart . . . . . . . 65
Michael M. Bell and Elizabeth M. Cherry
6 Methods of Myofibrillogenesis Modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Nancy K. Drew and Anna Grosberg
7 Using the Mechanical Bidomain Model to Analyze the Biomechanical
Behavior of Cardiomyocytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Bradley J. Roth
8 Fabrication of a Myocardial Patch with Cells Differentiated
from Human-Induced Pluripotent Stem Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Lei Ye, Joydeep Basu, and Jianyi Zhang
9 Efficient Differentiation of Cardiomyocytes from Human Pluripotent
Stem Cells with Growth Factors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Rajneesh Jha, Ren-He Xu, and Chunhui Xu
10 Isolation, Culturing, and Characterization of Cardiac Muscle Cells
from Nonhuman Primate Heart Tissue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Steven M. Hoynowski and John W. Ludlow
11 Mouse Embryonic Stem Cell-Derived Cardiac Myocytes
in a Cell Culture Dish. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Carley Glass, Reetu Singla, Anshu Arora, and Dinender K. Singla
12 Cryopreservation of Neonatal Cardiomyocytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Adam C. Vandergriff, M. Taylor Hensley, and Ke Cheng
13 Evaluation of Sarcomeric Organization in Human Pluripotent
Stem Cell-Derived Cardiomyocytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Chrishan J.A. Ramachandra and Winston Shim
vii
viii Contents
14 Electrotonic Coupled Metabolic Purification of Chick Cardiomyocytes . . . . . . 167

Winston Shim, Haiyang Yu, K.P. Myu Mai Ja,
Muhammad Parasuram, Kee Pah Lim, and Philip Wong
15 Gene Transfer into Cardiac Myocytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Sarah E. Lang and Margaret V. Westfall
16 Analysis of 4D Myocardial Wall Motion During Early Stages
of Chick Heart Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Madeline Midgett and Sandra Rugonyi
Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Contributors
ANSHU ARORA Biomolecular Science Center, Burnett School of Biomedical Sciences,

College of Medicine, University of Central Florida, Orlando, FL, USA
JOYDEEP BASU Tissue Engineering and Assay Development, Zen-Bio, Inc., Research
Triangle Park, NC USA; Process Research and Translation, Tengion, Inc.,
Winston-Salem, NC, USA
MICHAEL M. BELL School of Mathematical Sciences, Rochester Institute of Technology,
Rochester, NY, USA
KE CHENG Department of Molecular Biomedical Sciences, College of Veterinary Medicine,
North Carolina State University, Raleigh, NC, USA; UNC/NCSU Joint Department of
Biomedical Engineering, Chapel Hill, NC, USA
ELIZABETH M. CHERRY School of Mathematical Sciences, Rochester Institute of Technology,
Rochester, NY, USA
GERALD W. DORN II Center for Pharmacogenomics, Department of Internal Medicine,
Washington University School of Medicine, St. Louis, MO, USA
NANCY K. DREW The Edwards Lifesciences Center for Advanced Cardiovascular
Technology, Irvine, CA, USA
CARLEY GLASS Biomolecular Science Center, Burnett School of Biomedical Sciences,
ANNA GROSBERG The Edwards Lifesciences Center for Advanced Cardiovascular
Technology, Irvine, CA, USA
M. TAYLOR HENSLEY Department of Molecular Biomedical Sciences, College of Veterinary
Medicine, North Carolina State University, Raleigh, NC, USA
STEVEN M. HOYNOWSKI Zen Bio, Inc., Research Triangle Park, NC, USA
K.P. MYU MAI JA Stem Cell Laboratory, National Heart Research Institute Singapore
(NHRIS), National Heart Centre Singapore (NHCS), Singapore, Singapore
RAJNEESH JHA Department of Pediatrics, Emory University School of Medicine and
Children’s Healthcare of Atlanta, Atlanta, GA, USA
SARAH E. LANG Program in Cellular and Molecular Biology, University of Michigan,
Ann Arbor, MI, USA
KEE PAH LIM Stem Cell Laboratory, National Heart Research Institute Singapore
JOHN W. LUDLOW Regenerative Medicine, Zen Bio, Inc., Research Triangle Park, NC,
USA
SCOT J. MATKOVICH Center for Pharmacogenomics, Department of Internal Medicine,
Washington University School of Medicine, St. Louis, MO, USA
MADELINE MIDGETT Department of Biomedical Engineering, Oregon Health & Science
University, Portland, OR, USA
SREEJIT PARAMESWARAN Stem Cell and Molecular Biology Laboratory,
Department of Biotechnology, Indian Institute of Technology, Chennai,
Tamil Nadu, India
MUHAMMAD PARASURAM Stem Cell Laboratory, National Heart Research Institute
(NHRIS) Singapore, National Heart Centre Singapore (NHCS), Singapore, Singapore
ix
x Contributors
CHRISHAN J.A. RAMACHANDRA Stem Cell Laboratory, National Heart Research

Institute Singapore (NHRIS), National Heart Centre Singapore (NHCS), Singapore,
Singapore
BRADLEY J. ROTH Department of Physics, Oakland University, Rochester, MI, USA
SANDRA RUGONYI Department of Biomedical Engineering, Oregon Health & Science
University, Portland, OR, USA
RAJALAKSHMI SANTHAKUMAR Stem Cell and Molecular Biology Laboratory,
Department of Biotechnology, Indian Institute of Technology, Chennai,
Tamil Nadu, India
BARBARA SHERRY College of Veterinary Medicine, North Carolina State University,
Raleigh, NC, USA
WINSTON SHIM Stem Cell Laboratory, National Heart Research Institute Singapore
(NHRIS), National Heart Centre Singapore (NHCS), Singapore, Singapore; Duke-
NUS Graduate Medical School, Singapore, Singapore
DINENDER K. SINGLA Biomolecular Science Center, Burnett School of Biomedical Sciences,
REETU SINGLA Biomolecular Science Center, Burnett School of Biomedical Sciences,
ADAM C. VANDERGRIFF Department of Molecular Biomedical Sciences,
College of Veterinary Medicine, North Carolina State University, Raleigh, NC, USA;
UNC/NCSU Joint Department of Biomedical Engineering, Chapel Hill, NC, USA
CECILIA VECOLI Institute of Clinical Physiology, CNR, Pisa, Italy
RAMA S. VERMA Stem Cell and Molecular Biology Laboratory, Department of
Biotechnology, Indian Institute of Technology, Chennai, Tamil Nadu, India
PRASANNA VIDYASEKAR Stem Cell and Molecular Biology Laboratory, Department
of Biotechnology, Indian Institute of Technology, Chennai, Tamil Nadu, India
MARGARET V. WESTFALL Department of Cardiac Surgery and Program in Cellular
and Molecular Biology, University of Michigan, Ann Arbor, MI, USA
PHILIP WONG Stem Cell Laboratory, National Heart Research Institute Singapore
(NHRIS), National Heart Centre Singapore (NHCS), Singapore, Singapore; Duke-
NUS Graduate Medical School, Singapore, Singapore
CHUNHUI XU Department of Pediatrics, Emory University School of Medicine and
Children’s Healthcare of Atlanta, Atlanta, GA, USA; The Parker H. Petit Institute
for Bioengineering and Bioscience, Atlanta, GA, USA
REN-HE XU Faculty of Health Sciences, University of Macau, Taipa, Macau, China
LEI YE Division of Cardiology, Department of Medicine, University of Minnesota Medical
School, Minneapolis, MN, USA; Process Research and Translation, Tengion, Inc.,
HAIYANG YU Stem Cell Laboratory, National Heart Research Institute Singapore
JIANYI ZHANG Department of Medicine/Cardiology, University of Minnesota Medical
School, Minneapolis, MN, USA; Process Research and Translation, Tengion, Inc.,
Chapter 1
Generating Primary Cultures of Murine Cardiac Myocytes

and Cardiac Fibroblasts to Study Viral Myocarditis
Barbara Sherry
Abstract
Viruses can induce direct damage to cardiac myocytes and cardiac fibroblasts resulting in myocarditis and
impaired cardiac function. Cardiac myocytes and cardiac fibroblasts display different capacities to support
viral infection and generate a protective antiviral response. This chapter provides detailed protocols for
generation and characterization of primary cultures of murine cardiac myocytes and cardiac fibroblasts,
offering a powerful tool to probe cell type-specific responses that determine protection against viral
myocarditis.
Key words Cardiac myocyte, Cardiomyocyte, Cardiac fibroblast, Myocarditis, Primary cell culture,
Reovirus
1 Introduction
Viral myocarditis (cardiac damage) [1–3] is the second leading

cause of sudden death in young adults [4]. Most virus families
have been implicated in the human disease, although enteroviruses
such as coxsackievirus B3 and adenoviruses are generally the most
frequently cited [1–3, 5]. While immune-mediated damage is likely
a major component of enterovirus-induced myocarditis, the virus
also induces direct cytopathic effect [2]. Importantly, immunosup-
pressive therapy has proven only minimally beneficial for human
patients, leaving the role of immune-mediated damage in viral
myocarditis unclear [3, 6]. Reovirus induces a direct cytopathic
effect in cardiac cells [7] and induces myocarditis in mice lacking
B and T cells [8]. Thus, reovirus-induced murine myocarditis pro-
vides an outstanding model to study the direct effects of viral
infection in the heart and cardiac cells [9–11]. This chapter
describes generation of primary cultures of murine cardiac myo-
cytes and cardiac fibroblasts to study those direct viral effects.
While each year brings new technologies to stimulate cardiac
cell repair [12, 13], the unfortunate reality is that adult cardiac
Gary R. Skuse and Maureen C. Ferran (eds.), Cardiomyocytes: Methods and Protocols, Methods in Molecular Biology,
vol. 1299, DOI 10.1007/978-1-4939-2572-8_1, © Springer Science+Business Media New York 2015
1
2 Barbara Sherry
myocytes are largely nondividing and unable to self-renew after

cardiac damage [14–17]. Cardiac fibroblasts are the other major
cell in the heart and, in addition to maintaining their capacity for
cell division, provide a variety of critical functions [18, 19]. Unin-
fected cardiac myocytes and cardiac fibroblasts express different
levels of the antiviral cytokine interferon-β, providing an integrated
network of protection for the heart before the virus infects [11].
During infection, cardiac myocytes and cardiac fibroblasts display
cell type-specific differences in susceptibility to infection and in the
ensuing protective response [9, 11]. Indeed, the capacity for differ-
ent reoviruses to induce myocarditis correlates with their capacity
to induce cytopathic effect in cardiac myocytes but not cardiac
fibroblasts, confirming cell type-specific differences in the cardiac
cell response to viral infection, and validating the use of these
primary cultures [7]. The primary cell cultures described in this
chapter offer a powerful tool to probe cell type-specific responses
that determine protection against viral myocarditis.
2 Materials
2.1 Mice (See Note 1) 1. Timed-pregnant mice can be purchased or generated in-house
from wild-type or transgenic colonies (see Note 2).
2. Use a sufficient number of timed-pregnant females to generate
30–150 fetal and neonatal (<24 h old) mice. The fraction of
each is random (see Note 3).
2.2 On Lab Bench 1. Bench paper.

(See Note 4) 2. Large dissection scissors to open pregnant mice, small dissec-
tion scissors to remove fetuses, small dissection scissors to
remove hearts. Sterilize by autoclaving in advance or by dip-
ping in ethanol and flaming immediately before starting.
3. Two plastic Petri dishes each containing ~20 ml sterile PBS
(see Subheading 2.5).
2.3 In Tissue Culture 1. Two plastic Petri dishes each containing ~20 ml sterile PBS
Hood (See Note 4) (see Subheading 2.5).
2. Disposable sterile plastic transfer pipettes.
3. 5, 10, and 25 ml sterile pipettes.
4. Forceps, sterilized by autoclaving in advance or by dipping in
ethanol and flaming immediately before starting.
5. Micropipettes (e.g., Eppendorf, Pipetman) and both large and
regular orifice sterile tips.
Generating Primary Cultures of Murine Cardiac Myocytes and Cardiac Fibroblasts. . . 3
6. Two sterile small “flea” (7 2 mm) spin bars (VWR #58948-

976, autoclaved in glass test tubes).
7. Two 50 ml disposable sterile plastic conical tubes (“50 ml
tubes”), each with one of the sterilized small spin bars (above).
8. Rack to hold two 50 ml tubes.
9. Stirrer/hot plate (set temperature lowest possible; see more
about temperature below).
10. 1,000 ml beaker (to hold trypsin flask and one or two 50 ml
tubes) with thermometer (to maintain 36–37 C).
11. Sterile Whatman filter paper in Petri dish (stored as autoclaved
batch of filters in sterile covering).
12. BD Falcon cell strainer #352340 (BD Biosciences, San Jose
CA, USA).
13. Ice bucket with two to four 50 ml disposable sterile plastic
conical tubes each containing 21 ml trypsin inhibitor (see Sub-
heading 2.5); generally kept directly adjacent to tissue culture
hood (see Note 5).
2.4 Additional 1. Refrigerated tabletop centrifuge set at 4 C.

Equipment and 2. Hemocytometer and Eppendorf tubes for cell counts.
Supplies
3. Multiple 50 ml disposable sterile plastic conical tubes (“50 ml
tubes”).
4. Multiple 6-well tissue culture clusters.
5. Multiple 96-well, 48-well, or 24-well tissue culture clusters
(based on desired plating; see below).
2.5 Reagents 1. Ethanol for sterilizations.

(See Note 4) 2. PBS (without divalent cations, e.g., Mediatech #45000-446).
3. 60 or 120 ml 37 C 0.05 % trypsin-PBS for heart dissociations
(diluted from 10 [e.g., Gibco #15400-054] see Note 6).
4. Completed DMEM ¼ cDMEM (DMEM [Corning/Cellgro #
10-017-CV] + 7 % fetal calf serum [FCS; e.g., Atlanta Biolo-
gicals Premium Select #S11550] + 10 μg/ml gentamicin (e.g.,
Sigma #G1272)) (see Note 7).
5. Cold trypsin inhibitor (cDMEM + additional 13 % FCS) (see
Note 8).
6. 37 C 0.05 % trypsin-PBS for fibroblast trypsinization (e.g.,
Mediatech # 25-051-Cl, 1, see Note 9).
4 Barbara Sherry
3 Methods
3.1 Euthanizing Mice 1. Euthanize nonpregnant dams by IACUC-approved protocol.

and Extracting 2. Decapitate neonates by IACUC-approved protocol and dab
Neonatal and Fetal briefly on paper towel to remove blood (see Note 10).
Hearts, on Benchtop
3. Use small dissection scissors to cut through ribs to open tho-
with Bench Paper racic cavity.
4. Squeeze open gently to expose heart. If it is still beating, this
can facilitate identification.
5. Use same small dissection scissors to cut base of heart and
transfer to PBS dish (see Note 11).
6. Euthanize up to three pregnant females at a time (by IACUC-
approved cervical dislocation; not CO2), and spray abdomens
generously with 70 % ethanol (see Note 12).
7. Working with a single euthanized female, use the large dissec-
tion scissors to cut the abdominal skin and peel it back. Spread
the U-shaped uterus for good access (see Fig. 1).
8. Use small dissection scissors to cut one end of the U-shaped
uterus.
9. Remove one fetus and use small dissection scissors to decapitate
immediately by IACUC-approved protocol.
10. Gently roll the decapitated fetus back and forth across a
paper towel to ensure that all uterine fluids are removed (see
Note 13).
11. Proceed with next fetus until all fetuses have been removed
from first female.
12. Change gloves and use fresh dissection scissors to open the fetal
thoracic cavity and remove heart to PBS dish as for neonates
(see Fig. 1) (see Note 14).
13. When all fetuses have been removed from first female, proceed
with second female until all fetuses are removed.
14. When all hearts have been removed to PBS dish(es), transfer
the dish(es) to tissue culture hood.
3.2 Dissociating 1. Use a transfer pipette to remove most of the PBS from the Petri
the Hearts dish(es) with hearts.
2. Refill the dish with PBS and then remove most of this PBS to
facilitate access to hearts.
3. Use sterile forceps to transfer the hearts to a fresh Petri dish
containing PBS.
4. Use a transfer pipette to remove most of the PBS from this Petri
dish with hearts.
Fig. 1 Removing fetal mouse hearts. (a) Abdomen of euthanized pregnant mouse is sprayed with ethanol, skin
and membrane are pulled back, and U-shaped uterus is exposed. (b and c) Decapitated fetal mouse is held
between fingers. Thoracic cavity is cut, taking care to keep scissor tips pointed outward away from mouse to
avoid the heart. (d) Middle finger is gently pressed against back of the fetus to splay open the thoracic cavity
and better expose the heart for removal. (e) Scissor tips are used to cut heart away from mouse and transfer to
Petri dish. (f) A maximum of 100 hearts are transferred to a Petri dish containing PBS. Some contaminating
blood is common
5. Refill the dish with PBS and then remove most of this PBS to
facilitate access to hearts.
6. Use the same sterile forceps to transfer a group of ~15 hearts to
sterile PBS-wetted Whatman paper in a glass Petri dish. Gently
roll the group a little bit being sure to remove blood clots, and
then transfer them to a fresh Petri dish containing PBS (see
Note 15).
7. Repeat for the remaining hearts, pooling all hearts in a single
Petri dish containing PBS.
8. Remove most of the PBS to facilitate access to hearts.
6 Barbara Sherry
9. Use the same sterile forceps to transfer the hearts ~equally to

one or two 50 ml tubes containing only a spin bar. Use one
tube for <70 hearts and two tubes for >80 hearts.
10. Add 6.5 ml 37 C trypsin-PBS.
11. Stir in 36–37 C beaker bath (~150 ml H2O in 1,000 ml
beaker) on stirrer/hot plate until suspension is only barely
cloudy, ~1 min. Do not let temperature exceed 37 C (see
Note 16).
12. Remove tube(s) from the beaker bath to a rack in the hood.
The hearts will quickly settle.
13. Use a transfer pipette to carefully remove the supernatant to the
tube(s), and discard this supernatant (it contains mainly
fibroblasts).
14. Add 6.5 ml 37 C trypsin-PBS to each tube.
15. Replace the tube(s) in the 37 C beaker bath.
16. Stir 5 min (see Note 16 again).
17. Remove the tube(s) from the beaker bath to a rack in the hood.
The hearts will quickly settle.
18. Note the appearance of the supernatants and hearts, and use a
fresh transfer pipette to transfer the supernatant from the tube
of hearts to the first cold trypsin inhibitor tube. Invert the
trypsin inhibitor tube to mix and replace it on ice. Save this
transfer pipette and use it for all subsequent transfer steps to
avoid losing cells (see Note 17).
19. Add 6.5 ml 37 C trypsin-PBS to each tube of hearts.
20. Replace the tube(s) in the 37 C beaker bath.
21. Repeat steps 16–20, until four supernatants (~6 ml each) have
been pooled in the first cold trypsin inhibitor tube (four tryp-
sinizations if using one heart tube; two trypsinizations if using
two heart tubes; not including the first trypsinization from
which the supernatant was discarded).
22. Repeat steps 16–21 for a maximum of eight total trypsiniza-
tions (not including the first discarded supernatant), using the
remaining cold trypsin inhibitor tubes (see Note 18).
23. Centrifuge in a tabletop centrifuge at ~300 g for 10 min at
4 C (~1,200 rpm in a Sorvall tabletop centrifuge).
24. Gently decant the supernatants into an equal number of sterile
50 ml tubes (save the cell pellets for step 26) (see Note 19).
25. Centrifuge the supernatants retrieved in the previous step
in a tabletop centrifuge at ~300 g for 10 min at 4 C (see
Note 20).
26. Meanwhile, aliquot ~150 ml cDMEM into a flask or bottle as a
reservoir. Use a 5 ml pipette to transfer 2.5 ml cDMEM from
the reservoir to the pellet in the first tube, take up the pellet
gently to keep it intact, transfer it to the second tube with a
pellet, and continue transfers if additional tubes exist. Use the
same cell-laden pipette to retrieve 2.5 ml cDMEM from the
reservoir, rinse the first “empty” tube, serially rinse any remain-
ing tubes and pool this rinse with the cell suspension (final total
volume ~5 ml). Holding the pipette so that the tip is flat against
the tube bottom, use maximum force to retrieve and eject the
suspension to disperse the pellet. Repeat this “up and down”
~7 times to disperse the pellet. Place tube on ice while waiting
to proceed to step 27.
27. Gently decant the supernatants from step 25 into a waste
beaker to discard.
28. Handle the pellets and washes exactly as for step 26, and pool
with the cell suspension in step 26. The final volume will be
~10 ml total.
29. Continue to pipette up and down (~5 ml volumes) ~7 times
until there is no further obvious disaggregation. Estimate the
volume.
30. Remove two independent 10 μl samples to Eppendorf tubes
with 90 μl PBS and use a hemocytometer to count cells that are
not Red Blood Cells [RBCs] (see Note 21).
31. Calculate the number of cells (see Note 22).
3.3 Separating 1. Calculate the number of 6-well clusters to plate at ~1.25 106
Cardiac Myocytes from cells/well as a “pre-plating” to separate the rapidly adhering
Cardiac Fibroblasts fibroblasts from the more slowly adhering myocytes.
2. Add cDMEM to the cell suspension so the final volume will
result in 2 ml suspension per well.
3. Dispense 2 ml per well in 6-well clusters. Rinse suspension tube
with very small volume cDMEM and add to several already-
plated wells.
4. Incubate for 3 h in a 37 C, CO2 incubator to allow fibroblasts
to adhere (see Note 23).
5. Use a P1000 micropipette set at 700 μl with a large orifice tip to
use media in the first well to vigorously rinse that well seven to
eight times, squirting directly onto the “monolayer.” Transfer
700 μl of cell suspension to a 50 ml tube, leaving ~1.3 ml in the
well (see Note 24).
6. Similarly remove 700 μl from each of the five remaining wells in
that cluster, pooling all in the same 50 ml tube (leaving ~1.3 ml
in each well) and rocking the plate after each removal to ensure
that cells do not dry out.
8 Barbara Sherry
7. Repeat steps 5 and 6 for that first cluster, pooling the suspen-
sion in the same 50 ml tube and leaving ~600 μl in each well.
8. Remove the remaining ~600 μl from all six wells and pool in the
same 50 ml tube.
9. Using a 10 ml pipette, retrieve ~6 from the cDMEM reservoir
and add ~1 ml to each well.
10. Repeat steps 5 and 6 for that first cluster, pooling the suspen-
sion in the same 50 ml tube and leaving ~300 μl in each well.
11. Remove the remaining ~300 μl from all six wells and pool with
the cell suspension.
12. Retrieve ~2 ml from the cDMEM reservoir and add ~1 ml to
first wells in top and bottom rows.
13. Use a P1000 micropipette with a large orifice tip to transfer
media across the rows of wells and pool with the cell
suspension.
14. Using a 10 ml pipette, retrieve ~6 from the cDMEM reservoir
and add ~1 ml to each well (see Note 25).
15. Repeat the entire procedure for the remaining clusters. Aim to
generate a single tube of cell suspension if possible, by slightly
adjusting volumes above as necessary. If a single tube would
require major volume adjustments, generate two tubes of cell
suspension.
16. Centrifuge in a tabletop centrifuge at ~300 g for 10 min at
4 C (~1,200 rpm in a Sorvall tabletop centrifuge).
17. Meanwhile, incubate clusters containing fibroblasts and 1 ml
overlying media in a 37 C, CO2 incubator until ready to
trypsinize.
18. Gently decant the supernatants into an equal number of sterile
50 ml disposable sterile plastic conical tubes (save the cell
pellets for step 20).
19. Centrifuge the supernatants retrieved in the previous step
in a tabletop centrifuge at ~300 g for 10 min at 4 C
(see Note 20 again).
20. Meanwhile, resuspend pellets from step 18 and wash tubes
exactly as for step 26 in previous section.
21. Gently decant the supernatants from step 19 into a waste
beaker to discard.
22. Handle the pellets and washes exactly as for step 20, and pool
with the cell suspension in step 20.
23. Pipette the cell suspension through a cell strainer into a fresh
50 ml tube (using small volumes to avoid losing cells along the
length of pipette), carefully measuring volumes of each
addition.
Table 1
Plating cardiac myocyte cultures
Volume of cell
Vessel suspension to plate Cell number Example of purpose
96-well cluster 300 μl per well 1.5 10 myocytes
5
Viral replication by plaque assay
48-well cluster 1,000 μl per well 5 10 myocytes
5
RNA or protein harvest (small)
24-well cluster 2,000 μl per well 1 106 myocytes RNA or protein harvest (large)
8-well chamber slide 700 μl per well 3.5 10 myocytes
5
Microscopy
24. Rinse the source tube with 1 ml cDMEM and add to the
strainer.
25. Calculate the final volume.
26. Remove two independent 10 μl samples to Eppendorf tubes
with 90 μl PBS and use a hemocytometer to count cells that are
not RBCs.
27. Calculate the number of cells (see Note 26).
28. Add cDMEM to 5 105 cells/ml.
3.4 Plating Cardiac 1. Using large orifice micropipette tips, plate cardiac myocytes (see
Myocytes Table 1).
2. Incubate in a 37 C, CO2 incubator overnight.
3.5 Plating Cardiac 1. Use a transfer pipette to remove and discard media from all
Fibroblasts wells in one 6-well cluster.
2. Add 1 ml PBS to each well.
3. Repeat for up to two additional 6-well clusters.
4. Use a transfer pipette to remove PBS from all wells in first
6-well cluster.
5. Add 100 μl 0.05 % trypsin-PBS to each well.
6. Repeat for up to two additional 6-well clusters.
7. Incubate in a 37 C, CO2 incubator for ~7 min.
8. Remove cluster(s) and tap sides to release cells.
9. Check by microscope to ensure cells are rounded and floating.
If not, incubate several minutes longer.
10. Add 1,000 μl cDMEM to each well.
11. Use overlying media to rinse wells and pool in 50 ml tube.
12. Centrifuge in a tabletop centrifuge at ~300 g for 10 min
at 4 C (~1,200 rpm in a Sorvall tabletop centrifuge).
13. Decant supernatants into waste.
10 Barbara Sherry
Table 2
Infecting cardiac myocyte and cardiac fibroblast cultures
Final volume including

Vessel Volume of inoculum overlay (μl)
96-well cluster 100 μl per well 200–250
48-well cluster 350 μl per well 1,000–1,200
24-well cluster 700 μl per well 1,700–2,000
8-well chamber slide 200 μl per well 400–600
14. Use 5–10 ml cDMEM to resuspend the pellet.

15. Do not count the cells. Instead, resuspend fibroblasts in
two times the volume of media used for cardiac myocytes
(see Note 27).
16. Incubate in a 37 C, CO2 incubator overnight.
3.6 Infecting 1. After incubating cultures overnight, assess by light microscopy

Cultures 2 Days (see Notes 28 and 29).
Post-plating 2. Incubate overnight and assess by light microscopy again
(see Note 30).
3. Remove overlying media and infect the cultures (see Table 2)
(see Notes 30–32).
3.7 Assessing 1. Plate cultures on Poly-D-Lysine-coated 8-well chamber slides

Culture Purity by (BD Biosciences, #354632) (see Table 1) (see Note 33).
Immunofluorescent 2. One day post-plating, assess by light microscopy.
Microscopy
3. Two days post-plating, assess by light microscopy and then
carefully aspirate media and wash once with PBS.
4. Add 200 μl 4 % paraformaldehyde in PBS to each chamber and
incubate for 20 min at room temperature.
5. Carefully remove paraformaldehyde and wash three times with
ice-cold PBS (see Note 34).
6. Snap off chamber top. Never let chamber areas dry throughout
the rest of this protocol.
7. Add 0.25 % Triton X-100 in PBS and incubate for 10 min at
room temperature.
8. Use a transfer pipette to wash slide well with ice-cold PBS, bang
slide edge on paper towel to remove excess PBS, and use strips
of Whatman filter paper to wick off remaining PBS near cells.
9. Add 60 μl 5 % normal serum from the host of the secondary
antibody (see below) in PBS to each chamber area and incubate
for 60 min at room temperature.
10. Wash with ice-cold PBS as for step 8.

11. Add 60 μl 300 nM DAPI (Sigma, no. D8417, diluted in PBS)
to each chamber area and incubate for 15 min at room temper-
ature protected from light.
13. Add 60 μl mouse anti-alpha skeletal muscle actin (Abcam
#ab28052, diluted 1:50 in PBS 0.1 % BSA [IgG- and
protease-free; Jackson ImmunoResearch #001-000-161]) to
each chamber area and incubate for 60 min at room
temperature.
15. Add 60 μl fluor-tagged appropriate secondary antibody (e.g.,
Alexa 488-conjugated goat anti-mouse, Invitrogen # A11029,
diluted 1:1,000 in PBS 0.1 % BSA) to each chamber area and
incubate for 60 min at room temperature (see Note 35).
17. Add 60 μl chicken anti-vimentin (Thermo Scientific # PA1-
16759, diluted 1:5,000 in PBS 0.1 % BSA) to each chamber
area and incubate for 60 min at room temperature.
19. Add 60 μl fluor-tagged appropriate secondary antibody (e.g.,
Alexa 594-conjugated goat anti-chicken, Invitrogen #
A11042, diluted 1:1,000 in PBS 0.1 % BSA) to each chamber
area and incubate for 60 min at room temperature.
21. Use large orifice micropipette tip to add 60 μl ProLong Gold
antifade mounting reagent (Invitrogen # P36930) directly to
~4 chamber areas (i.e., ~15 μl per area) (see Note 36).
22. Place a large glass coverslip on the slide, starting at an angle to
prevent bubbles. Remove excess mounting reagent surround-
ing coverslip.
23. Use nail polish to seal the right and left (not the top or bottom)
edges of the coverslip on the slide.
24. Examine slides using a microscope (inverted or confocal) with
wide-field ultraviolet light illumination and appropriate filters.
25. Store slides at 4 C protected from light. Slides will last for at
least 1 year.
3.8 Assessing Virus 1. At the appropriate time post-infection (depending on the virus
Effects in Cardiac Cell and assay), cultures are suitable for quantitation of: frequency
Cultures of virus infection (e.g., immunocytochemistry), virus subcellu-
lar localization (e.g., confocal microscopy; see Note 35 again),
generation of infectious virus (e.g., plaque assay), generation of
12 Barbara Sherry
virus DNA or RNA (e.g., qPCR or qRT-PCR) or protein (e.g.,

immunoblot), virus-induced cytokine secretion (e.g., ELISA of
supernatants), virus-induced cytopathic effects (e.g., MTT
assay; see Note 37), and other viral parameters (see Note 38).
4 Notes
1. All procedures, including housing mice, mating mice, and

euthanizing mice and fetuses, must be approved in advance by
the appropriate Institutional Animal Care and Use Committee
(IACUC). Timed-pregnant mice are available from commercial
sources such as Charles River Laboratories.
2. Cardiac myocyte cultures derived from adults do not thrive for
more than approximately 1 day, and therefore it is necessary to
generate cultures from fetuses and/or neonates. If mating mice
in-house and using females that have not been synchronized for
estrus, house two females per male and separate after 2–4 days,
the shorter time yielding a smaller fraction of pregnancies. It is
convenient to mate mice on Monday or Tuesday and separate
them on Friday to generate cultures on a Monday (20 or 21
days after initiating the mating).
3. Mouse strains vary dramatically in litter size. Plan accordingly.
Cultures do not appear to differ regardless of the fraction of
fetuses vs. neonates, and can be 100 % one or the other.
4. Cardiac myocyte cultures are extraordinarily sensitive to con-
taminating residues that can build up in autoclaves. Use dispos-
able plastic rather than glass whenever possible. If autoclaved
glass must be used, be sure that the autoclave is used only for
clean items rather than disposal of waste.
5. The number of trypsin inhibitor tubes will depend on the
number of tubes used for trypsinizations (see Note 18).
6. Use 60 ml if trypsinizations are in a single tube; use 120 ml if
two tubes. Mix 6 or 12 ml 10 (0.5 %) trypsin/EDTA (Gibco
# 15400-054) with 54 or 108 ml PBS. It is convenient to place
this in a T75 tissue culture flask to fit in the beaker bath in the
tissue culture hood; keep it at 36–37 C.
7. Generally 500 ml DMEM (Corning/Cellgro # 10-017-CV;
contains glutamine but not sodium pyruvate) + 38 ml FCS +
538 μl 10 mg/ml gentamicin.
8. For each tube of 21 ml: 18.3 ml completed DMEM + 2,730 μl
FCS, for final 20 % FCS.
9. It is convenient to use Mediatech # 25-051-Cl, 1.
10. Two methods appear to be equally successful. Either decapitate
a single neonate and remove its heart before proceeding to the
next neonate or decapitate a group of five to ten neonates

(depending on personal speed) and remove their hearts before
proceeding to the next group.
11. The number of hearts per Petri dish should not exceed 100.
The PBS dish will accumulate some blood (see Fig. 1f).
12. CO2 will enter the bloodstream and could affect the quality of
cardiac cell cultures generated from fetal hearts. The ethanol
serves to sterilize the abdomen surface but also to wet the fur
and keep it from becoming airborne.
13. It is essential that uterine fluids do not contaminate the hearts.
These contaminants will result in a swirl of slime when centri-
fuging cardiac cells, and the resulting cultures will not thrive.
14. Fetal and neonatal hearts can be put in the same dish, but there
should be no more than 100 hearts per dish.
15. Rolling hearts on wetted Whatman paper removes the extreme
outer layer of the hearts and removes blood clots to increase the
purity of the cardiac myocyte and fibroblast cultures. The paper
should be barely damp. Do not roll them too much; the goal is
to remove all blood clots (with forceps facilitation) but minimal
cardiac tissue.
16. The stirring speed should be sufficient to generate a gentle
vortex down the center of the tube, and the tube should be
angled so that the spinbar is “beating” the hearts physically.
17. It is highly advisable to take good notes about the appearance
of the supernatant (cloudiness) and hearts (clumped or not;
easy or difficult to disperse), to help troubleshoot if culture
preparations are not reproducible.
18. Note the cloudiness of the supernatants each time; after peak-
ing it should decrease. In later trypsinizations, the supernatant
nearest the hearts may become “gooey.” It is critical that none
of this “goo” gets into the cold trypsin inhibitor tubes. If it is
not possible to remove cells from the trypsinization tubes
without also retrieving this “goo,” stop the trypsinizations.
Generally, eight trypsinizations (not including the first from
which the supernatant is discarded) are sufficient. If there is
excessive “goo” (see above), then stop earlier.
19. After centrifugation, carefully hold the 50 ml tubes up to the
light and look for a “swirl” of “slime” emanating from
the pellet. If there is a minor “swirl,” proceed as planned.
If there is a major, dominant swirl, consider discarding this
entire tube of cells. While this will reduce the yield of cells,
contaminating slime may jeopardize the culture. One cause of
“slime” is uterine fluids that contaminate gloves or scissors
when they touch the hearts during their removal.
14 Barbara Sherry
20. A significant fraction of cells do not pellet in the first centrifu-

gation; this second centrifugation captures them.
21. The capacity to distinguish cardiac myocytes from fibroblasts
by size and shape is controversial, but some would say that
fibroblasts appear smaller and have a jagged edge. The protocol
is successful regardless of whether both myocytes and fibro-
blasts or only myocytes are counted at this step. RBCs are
smaller, extremely round, and have a “donut-like” appearance.
Do not count RBCs. Count at least 100 cells per loading.
22. Calculate: the average number of cells per square 104 cells/ml
tenfold dilution in PBS total ml. There should be ~2–5 105
non-RBC cells per heart.
23. By eye, at the end of the incubation the wells should look like
they have a thick “monolayer” which starts to “rip” as you swirl
the cluster. The myocytes are not adherent at this stage; just
aggregated.
24. The wash here should be quite vigorous. The fibroblasts are
fairly firmly attached, and disengagement of overlying myo-
cytes requires direct disruption.
25. Check by microscope after removing the myocytes. The well
should contain mainly adherent cells with some rounded cells
but no “rips” across the monolayer.
26. Calculate: the average number of cells per square 104 cells/ml
tenfold dilution in PBS total ml. There should be
~1 105–3 105 myocytes per heart.
27. This generates fibroblasts that remain slightly subconfluent
2 days after plating. If fewer fibroblasts are needed, simply
harvest fewer wells and calculate volume of media accordingly.
For example: if two 6-well clusters are pre-plated during proce-
dure to separate myocytes from fibroblasts, harvest only one
6-well cluster to plate fibroblast cultures.
28. Cardiac myocyte cultures usually beat after 1 day, but not
vigorously. Usually only one or several myocytes per field
beat. The culture is very temperature sensitive and beating
will decrease dramatically as the cluster is examined at room
temperature. After returning the cluster to the incubator, it
may take several hours for it to resume its original beating
frequency. Cardiac myocyte cultures should have only minimal
contaminating RBCs. If there are many RBCs, modify proce-
dure for better removal of blood clots when rolling across
Whatman filter. Cultures can be washed gently 1 day post-
plating to remove RBCs, but be aware that this introduces a
variable.
29. Cardiac fibroblast cultures are difficult to visualize by light

microscopy because the cell borders are difficult to discern.
Decreasing the light can help to increase the contrast and
visibility.
30. After 2 days, cardiac myocyte cultures should beat well, either
as individual cells or as synchronous fields. Cardiac fibroblasts
will still be difficult to visualize, but should be subconfluent.
31. Cultures are generally inoculated with a small volume, incu-
bated for 1 h to allow cell attachment, and then overlaid with
additional media.
32. Cultures should not be maintained for longer than approxi-
mately 1 week post-infection. Contaminating dividing cardiac
fibroblasts in cardiac myocyte cultures may obscure myocyte-
specific phenotypes. Cardiac fibroblasts may change their phe-
notypes as they divide post-plating.
33. Cardiac myocytes do not adhere well to glass slides. Collagen
binding is known to stimulate signal transduction pathways,
and therefore is a suboptimal coating. Poly-D-Lysine-coated
slides (BD Biosciences, #354632) enhance adherence and are
inert to our knowledge. Cardiac fibroblasts are plated on Poly-
D-Lysine slides as well to allow direct comparisons.
34. Paraformaldehyde should be disposed of properly as hazardous

waste.
35. Cardiac myocytes may autofluoresce, particularly in the red
channel (excitation ~590 nm, emission ~620 nm) but rarely if
ever in the green channel (excitation ~495 nm, emission
~519 nm) or far-red channel (excitation ~650 nm, emission
~665 nm). Because use of red fluors is generally necessary for
double staining, be sure to include appropriate negative con-
trols (e.g., cardiac myocytes incubated with primary but not
fluor-tagged secondary antibody) to monitor autofluorescence,
which appears as one to five perfectly round bodies per cell
~half the size of the nucleus.
36. Using more than ~15 μl ProLong Gold per chamber will leave
slides too wet such that the nail polish in step 23 will mix with
the ProLong Gold and cover the chamber areas.
37. Assays for apoptosis can be employed, but cardiac myocyte
cultures may have high rates of spontaneous apoptosis. Be
sure to include appropriate control cell cultures.
38. For any assay, be sure to normalize results to internal controls
appropriately. Primary cultures may generate greater well-to-
well variability than cell lines do. Keep in mind that cardiac
fibroblasts divide while cardiac myocytes do not. Phenotypes
are likely to be cell type-specific; therefore it is important to
ensure that cultures are monitored for purity.
16 Barbara Sherry
Acknowledgements
This work was supported by NIH grant R01AI083333. I thank

Mary Ann Blum, Wrennie Edwards, and Lance Johnson for assis-
tance optimizing culture techniques; Efraı́n Rivera Serrano for
assistance optimizing immunofluorescent techniques; and
Dr. Jonathan Horowitz for photography. I thank Lance Johnson,
Kimberly Parks, Rachael Stebbing, and Efraı́n Rivera Serrano for
help preparing the chapter.
References
1. Cooper LT Jr (2009) Myocarditis. N Engl J 11. Zurney J, Howard KE, Sherry B (2007) Basal
Med 360:1526–1538 expression levels of IFNAR and Jak-STAT
2. Esfandiarei M, McManus BM (2008) Molecu- components are determinants of cell-type-
lar biology and pathogenesis of viral myocardi- specific differences in cardiac antiviral
tis. Annu Rev Pathol 3:127–155 responses. J Virol 81:13668–13680
3. Kindermann I, Barth C, Mahfoud F, Ukena C, 12. Nam YJ, Song K, Olson EN (2013) Heart
Lenski M, Yilmaz A, Klingel K, Kandolf R, repair by cardiac reprogramming. Nat Med
Sechtem U, Cooper LT, Bohm M (2012) 19:413–415
Update on myocarditis. J Am Coll Cardiol 13. Quijada P, Sussman MA (2014) Making it
59:779–792 stick: chasing the optimal stem cells for cardiac
4. Drory Y, Turetz Y, Hiss Y, Lev B, Fisman EZ, regeneration. Expert Rev Cardiovasc Ther
Pines A, Kramer MR (1991) Sudden unex- 12:1275–1288
pected death in persons less than 40 years of 14. Naqvi N, Li M, Calvert JW, Tejada T, Lambert
age. Am J Cardiol 68:1388–1392 JP, Wu J, Kesteven SH, Holman SR, Matsuda
5. Feldman AM, McNamara D (2000) Myocardi- T, Lovelock JD, Howard WW, Iismaa SE, Chan
tis. N Engl J Med 343:1388–1398 AY, Crawford BH, Wagner MB, Martin DI,
6. Mason JW, O’Connell JB, Herskowitz A, Rose Lefer DJ, Graham RM, Husain A (2014) A
NR, McManus BM, Billingham ME, Moon proliferative burst during preadolescence
TE, The Myocarditis Treatment Trial Investi- establishes the final cardiomyocyte number.
gators (1995) A clinical trial of immunosup- Cell 157:795–807
pressive therapy for myocarditis. N Engl J 15. Porrello ER, Mahmoud AI, Simpson E, Hill
Med 333:269–275 JA, Richardson JA, Olson EN, Sadek HA
7. Baty CJ, Sherry B (1993) Cytopathogenic (2011) Transient regenerative potential of the
effect in cardiac myocytes but not in cardiac neonatal mouse heart. Science 331:1078–1080
fibroblasts is correlated with reovirus-induced 16. Porrello ER, Olson EN (2014) A neonatal
acute myocarditis. J Virol 67:6295–6298 blueprint for cardiac regeneration. Stem Cell
8. Sherry B, Li X, Tyler KL, Cullen JM, Virgin Res 13:556–570
HW (1993) Lymphocytes protect against and 17. Senyo SE, Steinhauser ML, Pizzimenti CL, Yang
are not required for reovirus-induced myocar- VK, Cai L, Wang M, Wu TD, Guerquin-Kern
ditis. J Virol 67:6119–6124 JL, Lechene CP, Lee RT (2013) Mammalian
9. Irvin SC, Zurney J, Ooms LS, Chappell JD, heart renewal by pre-existing cardiomyocytes.
Dermody TS, Sherry B (2012) A single- Nature 493:433–436
amino-acid polymorphism in reovirus protein 18. Martin ML, Blaxall BC (2012) Cardiac inter-
mu2 determines repression of interferon sig- cellular communication: are myocytes and
naling and modulates myocarditis. J Virol fibroblasts fair-weather friends? J Cardiovasc
86:2302–2311 Transl Res 5:768–782
10. Li L, Sevinsky JR, Rowland MD, Bundy JL, 19. Porter KE, Turner NA (2009) Cardiac fibro-
Stephenson JL, Sherry B (2010) Proteomic blasts: at the heart of myocardial remodeling.
analysis reveals virus-specific Hsp25 modula- Pharmacol Ther 123:255–278
tion in cardiac myocytes. J Proteome Res
9:2460–2471
Chapter 2
Enrichment of Cardiomyocytes in Primary Cultures

of Murine Neonatal Hearts
Sreejit Parameswaran, Rajalakshmi Santhakumar,
Prasanna Vidyasekar, and Rama S. Verma
Abstract
In vitro culture of neonatal murine cardiomyocytes is vital for understanding the functions of the heart.
Cardiomyocyte cultures are difficult to maintain because they do not proliferate after birth. The mainte-
nance of primary cultures of viable and functional cardiomyocytes is considerably affected by the yield from
initial steps of isolation procedures. This protocol describes an efficient and rapid method for isolation and
maintenance of long-term cultures of neonatal murine cardiomyocytes by effectively shortening the trypsin
enzyme digestion period and the cardiomyocyte enrichment step.
Key words Primary cell culture, Neonatal mice, Murine cardiomyocyte enrichment,
Immunocytochemistry
1 Introduction
Neonatal murine cardiomyocyte cultures have been used as a model

for studying contraction, ischemia, and hypoxia at the cellular level
[1, 2]. Murine cardiomyocyte cultures have also been used to study
the toxicology of drugs and their transport [3, 4]. Cardiomyocyte
cultures are difficult to maintain, especially over long-term periods.
Therefore it is necessary to follow a protocol which can yield a
sufficient number of cells that are viable and functional. Various
groups working with adult murine cardiomyocyte cultures have
elaborated on issues like finding and formulating conditions for
proper tissue dissociation such as enzymes and duration of enzyme
treatment, in selecting various growth supplements and appropriate
serum concentrations for cell attachment, and also appropriate
attachment surfaces especially for long-term cell cultures [5, 6].
17
18 Sreejit Parameswaran et al.
Here, we demonstrate a protocol for the primary culture of

cardiomyocytes that consistently yielded long-term cultures in our
laboratory. Growth supplements such as carnitine or creatinine and
growth inhibitors of non-myocytes such as 9-β-D-arabinofuranosyl
cytosine and 5-bromo-20 -deoxyuridine [5, 7] are not used in this
protocol. This protocol reduces the processing time considerably
and relies on two critical steps: (1) cardiomyocyte enrichment and
(2) enzyme digestion procedure. The cardiomyocyte enrichment
step removes non-myocytes and is crucial to obtaining a homoge-
nous yield of cardiomyocytes for culture. The duration of enzy-
matic digestion that yields viable and functional cells is reduced to a
maximum of 3 h.
2 Materials
Prepare all solutions using UPW (ultrapure water—prepared by

purifying deionized water to attain a sensitivity of 18 MΩ cm at
25 C) and cell culture grade reagents. Prepare and store all
reagents at 4 C (unless indicated otherwise). Diligently follow all
waste disposal regulations when disposing waste materials.
2.1 Buffer 1. Dulbecco’s Phosphate Buffered Saline (DPBSA): Place a clear

Preparation glass 1 L container on a magnetic stirrer and set to 200 rpm
after adding 1 L of UPW. Add a 1 L packet of D-PBS powder
(Solution A, lacking Ca2+ and Mg2+, e.g., Sigma D5652). Stir
until completely dissolved and check pH. (pH should not vary
more than 0.1 pH units from 7.0.) Aliquot and sterilize by
autoclaving for 20 min at 121 C and 100 kPa in sealed bottles
with sterile indicating tape.
2. Balanced salt solution (BSS): Prepare 20 mM hydroxyethylpi-
perazineethanesulfonic acid–NaOH (HEPES-NaOH buffer)
pH 7.6 by dissolving 4.766 g of HEPES in 700 mL ultrapure
water and add approximately 5.5 g sodium hydroxide pellets to
adjust the pH to 7.6. In this buffer, dissolve each of the
following components separately—130 mM NaCl, 1 mM
NaH2PO4, 4 mM glucose, 3 mM KCl and make up the volume
to 1 L. Ensure the final pH of the solution is 7.6.
2.2 Medium Commercial media are available as powdered media, 1 working

Preparation strength solutions or 10 concentrates. Usual supplements to the
medium are serum and antibiotics. Other additions include gluta-
mine or sodium bicarbonate depending on the media formulation.
Enrichment of Cardiomyocytes in Primary Cultures of Murine Neonatal Hearts 19
1. MAINTENANCE MEDIUM—DMEM/F12 (1:1) with fetal

calf serum (20 %) and horse serum (5 %): Dissolve the entire
contents of the DMEM/F12 pack (Gibco, Carlsbad, CA)
gradually in 1 L of UPW using a magnetic stirrer for constant
mixing. When all the constituents are dissolved completely, the
pH of the medium should be checked (pH should not vary
more than 0.1 pH units from 7.4). Sterilize by filtration
through a 0.2 μm filter. Store at 4 C. Supplement the medium
with the following sterile components just before use—Fetal
calf serum (Gibco, Carlsbad, CA) 20 %, horse serum (Gibco,
Carlsbad, CA) 5 %, penicillin (100 U/mL), and streptomycin
(100 mg/mL; Cambrex, Verviers, Belgium), 2 mM L-gluta-
mine (Cambrex), 0.1 mM nonessential amino acids (Gibco,
Carlsbad, CA), 3 mM sodium pyruvate (Gibco, Carlsbad,
CA), and bovine insulin (1 μg/mL; USV, India).
2.3 Enzymes, 1. 0.05 and 0.5 % Trypsin-EDTA; (Invitrogen, Carlsbad, CA).

Antibodies, and Other 2. 70 % Ethanol prepared with UPW.
Reagents
3. Trypan Blue Dye.
4. 4 % Paraformaldehyde (PFA; HIMEDIA, India) in DPBSA:
Heat 45 mL of UPW to around 60 C and add 3.6 g of PFA
while constantly stirring in a heated water bath. Add approxi-
mately 600 μL of 2 N NaOH in fractions of 200 μL until the
solution becomes clear. Ensure pH is between 7.0 and 7.4.
Make up the volume to 90 mL using 2 DPBSA. (Final
concentration will be 4 % PFA in 1 DPBSA.) Filter the
solution using a 0.22 μm filter.
5. 0.25 % Triton-X (Merck, India) v/v in DPBSA.
6. Hoechst 33342 (H 33342, Sigma, St. Louis, MO).
7. Primary antibodies: mouse monoclonal anti-GATA-4 (Santa
Cruz Biotechnology, Santa Cruz, CA) and rabbit polyclonal
anti-Nkx-2.5 (Santa Cruz Biotechnology).
8. Secondary antibodies: R-Phycoerythrin (PE-R) labeled goat
anti-mouse IgG (Sigma, St. Louis, MO) and fluorescein iso-
thiocyanate (FITC) labeled goat anti-rabbit IgG (Sigma,
St. Louis, MO).
9. Gelatin pre-coated plates: 1 % Gelatin was prepared using UPW
and sterilized by autoclaving. Tissue culture plates were coated
with 1 % gelatin sufficient to cover the surface. The plates were
incubated for 30 min following which the gelatin was removed,
leaving a thin coating over the plates. The plates were kept
under UV for 90 min and stored at 37 C.
2.4 Animals 1. Mice—BALB/c or Swiss Albino neonatal mice 1–3 days old.
2.5 Instruments 1. Class II A Bio-safety cabinet.

and Equipment 2. CO2 Incubator.
3. Centrifuge with swing bucket rotor.
4. Water bath.
5. Florescence microscope with phase contrast.
6. Dissection/surgical instruments—forceps, scissors, scalpels,
razor blades.
7. 15-mL centrifuge tubes.
8. Petri dishes.
3 Methods
Carry out all procedures at room temperature unless specified

otherwise.
3.1 Excision of 1. Swab the neonatal mice completely with 70 % ethanol before
Ventricular Myocardial sacrificing by cervical dislocation (see Notes 1–7).
Tissue from Neonatal 2. Excise the whole heart and transfer it immediately to sterile
Mice ice-cold DPBSA.
3. Squeeze the blood out of the heart gently using sterile forceps.
4. Wash the excised hearts once more with sterile ice-cold
DPBSA.
5. Place the tissue in sterile ice-cold BSS for 10 min (see Note 8).
Remove the auricles carefully and discard. Place the excised
ventricles in a sterile 60-mm Petri dish.
6. Carefully mince the ventricles into small sections less than or
equal to 1 mm3 in 0.05 % Trypsin EDTA using a sterile scalpel.
Transfer the pieces into a sterile 15-mL falcon tube.
3.2 Isolation and 1. Disperse the cells by incubating the tissue with 0.5 % Trypsin
Culture of EDTA (1 mL of Trypsin EDTA per 100 mg of tissue) (see
Cardiomyocytes Note 9). Incubate it at 37 C in a water bath for 4 min with
intermittent pipetting.
2. Allow the cell suspension to stand for 1 min and then transfer
the supernatant into a fresh 15-mL centrifuge tube kept on ice
(see Note 10).
3. Add 2 mL of Maintenance Medium. Repeat steps 1–3 (diges-
tion process) three times.
4. Pool the supernatant obtained from each of the digestion steps.
5. Spin down at 2,205 g (RCF) for 10 min at 4 C and resus-

pend the cell pellet in Maintenance Medium.
6. Assess the viability of the cells by trypan blue exclusion test:
Mix equal volumes (~10–20 μL each) of 0.4 % trypan blue and
the cell suspension and incubate at room temperature for
3 min. Introduce a drop of the trypan blue/cell suspension
into one of the V-shaped wells of a hemocytometer. Using a
bright field microscope, focus on the counting grids and count
the unstained (viable) and stained (nonviable) cells separately.
Repeat the count using the other chambers of the hemocytom-
eter. Add the total number of unstained cells obtained from the
four corner square grids and multiply the sum by the dilution
factor (factor 2 for equal volumes) to obtain the total number
of viable cells present per mL of the cell suspension.
7. Plate the cells on tissue culture plates pre-coated with 1 %
Gelatin.
8. Incubate the cells at 37 C, 5 % CO2 for 3 h to allow differential
attachment of non-myocardial cells.
9. Transfer all non-adhered cells (cardiomyocytes) into a fresh
tube.
10. Assess the viability of cells by trypan blue exclusion test (see
Notes 11–13).
11. Plate the myocyte-enriched suspension on to tissue culture
plates at a density of 2 104 cells per cm2 and incubate at
37 C, 5 % CO2.
12. Replenish the medium after 72 h and thereafter every 48 h.
Observe for beating myocardial cells after 72 h using an
inverted phase contrast microscope (Fig. 1).
3.3 Characterization 1. Fix the cells by incubating them at room temperature in an

of Cardiomyocytes by adequate volume of 4 % PFA (see Note 14) for 20 min. Wash
Immunocytochemistry the cells once using DPBSA.
2. Incubate the cells in an adequate volume of Triton-X (0.25 %)
for 10 min. Wash gently, three to four times with DPBSA.
3. Prepare the dilutions of primary antibodies, Anti-GATA-4 and
Anti-Nkx-2.5, at 1:100 in cold DPBSA. Add the dilutions to
the cells and incubate for 90 min at room temperature. Wash
the cells gently, three to four times with DPBSA.
4. Prepare dilutions of fluorophore-conjugated secondary antibo-
dies, Anti-IgG (Mouse and Rabbit); at 1:200/300 (see Note
15) in cold DPBSA. Add the dilutions to the cells and incubate
for 90 min at room temperature. Wash the cells gently, three to
four times with DPBSA.
Fig. 1 Phase contrast photomicrograph of cultures showing different morpholo-

gies of beating cardiomyocytes in culture (a–d; magnification: 400)
5. Add an adequate volume of Hoechst 33342 (5 μg/mL)

and incubate the cells at room temperature for 30 min (see
Note 16). Wash the cells gently, three times with DPBSA.
6. Maintain the cells in DPBSA. Visualize the cells using a flores-
cence microscope under the 460–490 nm band-pass excitation
filters for FITC and the 490–578 band-pass excitation filters
for Phycoerythrin-R. Hoechst 33342 can be visualized using
355–465 band-pass excitation filters (Fig. 2).
Fig. 2 Immunocytochemical analysis of GATA-4 and Nkx-2.5 in cardiomyocytes.

(a) Phase contrast photomicrograph of culture. (b) Positive staining with anti-
GATA-4 antibody. (c) Positive staining with anti-Nkx-2.5 antibody (original
magnification, 200)
4 Notes
1. Handle the animals gently in order to minimize stress that may

affect the neuro-humoral state of the cells.
2. Spray the carcass of the neonatal mice liberally with 70 % etha-
nol (EtOH) before beginning experiment. Check the sterility
of all the culture media and reagents, including DPBSA, BSS,
and trypsin before use. Warm sterile media to 37 C.
3. Sterilize surgical blades and forceps with 70 % ethanol and
autoclave. Maintain aseptic conditions in the incubator.
4. Maintain aseptic conditions during all steps of culturing and
isolation. Use sterile latex or nitrile gloves. Glass coverslips
must be autoclaved in Petri dishes and coated with gelatin
before use.
5. Use fine razor blades or surgical scalpels for cutting the tissue
into small pieces. We recommend using disposable no. 22
surgical blades for this purpose.
6. Drying of tissue can lead to less yield. Therefore avoid getting
the tissue dry. Soaking or rinsing the surgical instruments in
DPBSA prior to cutting can help in making finer pieces of the
tissues.
7. The amount of tissue surrounding the heart can be minimized
by using sharp fine forceps or curved forceps which should be
placed under the heart. In this manner the lungs can be entirely
avoided or the amount of lung tissue can be minimized.
8. Since Glucose is included in the preparation of the Balanced
Salt Solution, the solution should be sterilized by filtration to
avoid caramelization of glucose. Alternatively, glucose may be
autoclaved separately at a higher concentration (e.g., 20 %) and
added later.
9. Enzyme digestion is an important step. Warm the trypsin to
37 C before use. Use fresh trypsin as it tends to lose enzymatic
activity gradually when kept at 4 C. Alternatively, aliquot
enzyme solutions in small quantities (5–10-mL aliquots) and
freeze them at 20 or 80 C.
10. Duration of trypsin treatment is important. During the trypan
blue exclusion assay if the viable cell count is less than 70 %, it
indicates overdigestion by trypsin. To stop the reaction and
prevent overdigestion, adding ice-cold complete medium is
also an option.
11. Since trypan blue is light sensitive, it must be stored in dark
bottle.
12. Cells must be counted within 3–5 min of mixing with trypan
blue, as prolonged incubation might cause cell death and
thereby can result in reduced viability counts.
13. Avoid excessive pipetting and only use intermittently so as to
avoid cell death and debris.
14. During the preparation of 4 % Paraformaldehyde, the solution
will never be free of precipitate unless the optimal pH is
attained and will carry a hazy appearance. Less than 1 mL of
2 N NaOH was sufficient to dissolve the precipitate under this
author’s laboratory conditions.
15. When preparing dilutions of antibodies for use on adherent cell
populations such as a cardiomyocyte primary culture, the dilu-
tion need be prepared according to the surface area of the TCP.
An adequate volume that just covers the surface of the plate is
sufficient. The dilutions given in Subheading 3 can vary accord-
ing to manufacturer and has to be determined by titrations.
16. GATA-4 and Nkx2.5 are transcriptional factors; therefore
staining for these proteins will require permeabilization with
triton-X. While H 33342 can pass through the membrane and
needs no permeabilization, the use of DAPI as a nuclear coun-
terstain can be facilitated using triton-X.
Acknowledgments
This work is supported by grants to R.S.V. by the Ministry of

Human Resource Development (MHRD—BIO/2005–2006/
007/MHRD/RAMS/859) and Department of Biotechnology,
Ministry of Science and Technology (DBT-BT/PR5392/MED/
14/693/2004).
References
1. Yamashita N, Nishida M, Hoshida S, Kuzuya T, 4. Wang GW, Kang YJ (1999) Inhibition of doxo-
Hori M, Taniguchi N et al (1994) Induction of rubicin toxicity in cultured neonatal mouse car-
manganese superoxide dismutase in rat cardiac diomyocytes with elevated metallothionein
myocytes increases tolerance to hypoxia 24 h levels. J Pharmacol Exp Ther 288(3):938–944
after preconditioning. J Clin Invest 5. Kruppenbacher JP, May T, Eggers HJ, Piper
94:2193–2199 HM (1993) Cardiomyocytes of adult mice in
2. Bahi N, Zhang J, Llovera M, Ballester M, long-term culture. Naturwissenschaften
Comella JX, Sanchis D (2006) Switch from 80:132–134
caspase-dependent to caspase-independent 6. Chlopclkova Š, Psotova J, Miketova P (2001)
death during heart development: essential role Neonatal rat cardiomyocytes—a model for the
of endonuclease G in ischemia-induced DNA study of morphological, biochemical and
processing of differentiated cardiomyocytes. J electrophysiological characteristic of the heart.
Biol Chem 281(32):22943–22952 Biomed Pap Med Fac Univ Palacky Olomouc
3. Limaye DA, Shaikh ZA (1999) Cytotoxicity of Czech Repub 145:49–55
cadmium and characteristics of its transport in 7. Nuss HB, Marban E (1994) Electrophysiological
cardiomyocytes. Toxicol Appl Pharmacol 154 properties of neonatal mouse cardiac myocytes in
(1):59–66 primary culture. J Physiol 479(2): 265–279
Chapter 3
Deep Sequencing of Cardiac MicroRNA-mRNA Interactomes

in Clinical and Experimental Cardiomyopathy
Scot J. Matkovich and Gerald W. Dorn II
Abstract
MicroRNAs are a family of short (~21 nucleotide) noncoding RNAs that serve key roles in cellular growth
and differentiation and the response of the heart to stress stimuli. As the sequence-specific recognition
element of RNA-induced silencing complexes (RISCs), microRNAs bind mRNAs and prevent their
translation via mechanisms that may include transcript degradation and/or prevention of ribosome
binding. Short microRNA sequences and the ability of microRNAs to bind to mRNA sites having only
partial/imperfect sequence complementarity complicate purely computational analyses of microRNA-
mRNA interactomes. Furthermore, computational microRNA target prediction programs typically ignore
biological context, and therefore the principal determinants of microRNA-mRNA binding: the presence
and quantity of each. To address these deficiencies we describe an empirical method, developed via studies
of stressed and failing hearts, to determine disease-induced changes in microRNAs, mRNAs, and the
mRNAs targeted to the RISC, without cross-linking mRNAs to RISC proteins. Deep sequencing methods
are used to determine RNA abundances, delivering unbiased, quantitative RNA data limited only by their
annotation in the genome of interest. We describe the laboratory bench steps required to perform these
experiments, experimental design strategies to achieve an appropriate number of sequencing reads per
biological replicate, and computer-based processing tools and procedures to convert large raw sequencing
data files into gene expression measures useful for differential expression analyses.
Key words microRNA, mRNA, Argonaute, Ago2, RNA-induced silencing complex (RISC),
Immunoprecipitation, Deep sequencing, Read alignment, Gene annotation, Differential expression
1 Introduction
Cellular homeostatic maintenance and responses to stress stimuli

involve RNA transcription, not only of coding mRNAs but of
regulatory noncoding RNAs, whose many potential functions
are gradually coming to light. MicroRNAs are 21–22 nucleotide
molecules in their mature, processed forms and comprise but one
subfamily of noncoding RNAs. However, their importance to
mammalian cellular function has been widely demonstrated, with
dramatic effects on phenotype resulting from deletion of the
enzymatic machinery required to process microRNA precursor
27
28 Scot J. Matkovich and Gerald W. Dorn II
transcripts into their active forms [1–4]. The mechanism of action of

microRNAs is straightforward, in principle: a microRNA is
incorporated into a protein complex including Argonaute family
members and TNRC6/GW182 (the RNA-induced silencing
complex, or RISC) and selects mRNAs with partially complemen-
tary sequences for degradation and/or translational suppression
[5–7]. Difficulties in understanding which microRNAs are actively
repressing individual mRNAs arise from the presence of ~1,000
different microRNAs in the mammalian genome [8], ~300 of
which are detectably expressed in the heart [9], and which each
could theoretically interact with a large number of different
mRNAs (from the ~10,000 detectably expressed in the heart
[10, 11]) considering that only partial sequence complementarity
is needed for effective regulation. The simultaneous alteration of a
large number of both microRNAs and mRNAs in clinical [12–16]
and experimental cardiomyopathies [11, 14, 17] means that it is
difficult, if not impossible, to determine which mRNAs in such
contexts are undergoing regulation that is largely dependent on
microRNAs, without knowledge of the changes induced in direct
microRNA-mRNA interactions.
Our approach to this challenging question involves measurement
of three distinct RNA fractions that underlie the microRNA regula-
tion of mRNAs and their ultimate effect on the gene expression
profile of the heart; that is, (a) the microRNAs themselves; (b) the
mRNAs captured at the RISC via interaction via microRNAs; and (c)
the global cohort of mRNAs. We use next-generation/deep sequenc-
ing methods to obtain quantitative RNA abundance information that
cannot be determined using qPCR techniques or microarrays, and
which have superior dynamic range and sensitivity compared to these
approaches [10, 18]. MicroRNAs and mRNAs are monitored in the
same heart or cardiomyocyte preparation, together with the RISC-
associated mRNAs that are targeted by microRNAs. Comparison of
microRNAs, RISC-associated mRNAs, and translationally compe-
tent mRNAs between control and diseased hearts permits the char-
acterization of cardiomyopathic alterations in the microRNA-mRNA
interactome. We do not utilize the CLIP approach (cross-linking with
immunoprecipitation) [19] popular in a number of in vitro studies of
mRNA interactions with mRNA-binding proteins. This is mostly due
to the technical problems inherent in performing UV- [19] or photo-
activatable-analog-based [20, 21] cross-linking in intact heart pre-
parations (although UV cross-linking has been performed in young
mouse brain [22, 23]), but partly out of concern that the contribu-
tion of RNAs that are only weakly associated with the RISC to the
microRNA-mRNA interactome may be overestimated. Nonetheless,
such methods are of undeniable importance for identifying micro-
RNA binding sites on target mRNAs [23]. Our methods for Ago2
immunoprecipitation without cross-linking and subsequent compar-
ison of RISC-bound and global mRNA content are based on studies
Deep Sequencing of Cardiac MicroRNA-mRNA Interactomes in Clinical. . . 29
by Karginov et al. [24], who evaluated microRNA targets in 293T

cells using recombinantly overexpressed Ago2 and microarrays for
differential expression analysis; our studies have shown it is possible to
gain sufficient Ago2 and bound RNA by immunoprecipitating the
Ago2 endogenous to mouse hearts.
In this chapter, we describe in detail laboratory procedures for
obtaining the necessary RNA fractions and preparing libraries for
deep sequencing analysis. Additionally, we describe the computer-
based workflows required to align the resulting sequence data to
annotated transcriptomes, perform gene-level quantitation, and
assess differential expression. Data visualization tools such as princi-
pal components analysis (PCA) and heatmaps, helpful in observing
important trends in highly multidimensional data sets such as RNA-
sequencing experiments, are not covered, but the interested reader
can become familiar with our application of these graphic formats
in published studies of cardiac RNA- and RISC-sequencing [9, 11,
25, 26]. Likewise, advanced analytical procedures that build upon
deep sequence microRNA and mRNA results to designate specific
microRNA-mRNA pairing events in experimental context are
beyond the scope of this chapter, but have been described [9, 11].
2 Materials
2.1 miR-Seq Library 1. Trizol (Invitrogen).

Preparation 2. Illumina TruSeq small RNA preparation kit (catalog # RS-200-
00xx, with the xx digits referring to the particular set of 12
indexing primers supplied). There are a large number of
reagents and specialized consumables required for completion
of the protocol that need to be purchased separately, and a list
can be found in the manual supplied with the kit.
2.2 PolyA+ 1. Invitrogen oligo(dT)20 Dynabeads (Invitrogen #610-06) for

Enrichment polyA+ RNA selection. New England Biolabs also make an
for mRNA-Seq Library alternative magnetic bead purification system which we have
Preparation used successfully (“Magnetic mRNA isolation kit,” #S1550S).
2. Dynal magnet (Invitrogen) or other magnet suitable for pelleting
magnetic beads in 1.5 mL microcentrifuge tubes.
2.3 Ago2 1. Anti-mouse Ago2 monoclonal antibody (Wako Pure, clone

Immunoprecipitation #2D4).
for RISC-Seq Library 2. Protein G-coupled Dynabeads (Invitrogen).
Preparation
3. Lysis and immunoprecipitation: Ice-cold buffer solution com-
prised of 50 mM Tris–HCl, 5 mM EDTA, 5 mM EGTA,
pH 7.5, with Roche Complete protease inhibitors, yeast
tRNA (Invitrogen) (1 μg/μL), and SUPERnase-IN (Ambion)
(1 U/μL); after initial homogenization, 0.5 % Nonidet P-40
(Sigma IGEPAL CA-630).
2.4 RNA- and 1. 5 fragmentation buffer, 50 mL: 1.21 g Tris–HCl base, 2.45 g
RISC-Seq Library potassium acetate, 1.61 g magnesium acetate tetrahydrate
Preparation (200 mM Tris–HCl acetate, 500 mM potassium acetate,
150 mM magnesium acetate, DEPC H2O). Adjust pH to 8.2
with 1 N acetic acid, and keep aliquots frozen at 20 C.
2. Zymo RNA Clean + Concentrator™-5 columns for RNA
purification, catalog #R1015.
3. Invitrogen SuperScript III first-strand synthesis system,
#18080-051 (50 reactions), which comprises SuperScript III
reverse transcriptase, random hexamer solution, and buffer
components including DTT and MgCl2.
4. 5 second-strand reaction buffer; 100 mM Tris–HCl pH 6.9,
450 mM KCl, 23 mM MgCl2, 0.75 mM β-NAD+, 50 mM
(NH4)2SO4. Keep frozen aliquots at 20 C.
5. E. coli DNA ligase (New England Biolabs, catalog #M0205S)
and DNA polymerase I (New England Biolabs, catalog
#M0209S).
6. Qiagen QIAquick PCR purification kit (catalog #28104),
MinElute kit (catalog #28004), gel purification kit (catalog
#28704).
7. Adapter oligonucleotides for multiplexed Illumina library
sequencing (Table 1).
Table 1
Adapter oligonucleotide and PCR primer sequences
Name Sequence 50 –30

Index Adapter A ACACTCTTTCCCTACACGACGCTCTTCCGATCT
Index Adapter B /5phos/GATCGGAAGAGCACACGTCTGAACTCCAGTCAC
Index PCR 1.0 AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGC
TCTTCCGATCT
Index PCR 2.0 GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT
Indexing primer X CAAGCAGAAGACGGCATACGAGATACGCCTCGTGACTGGAGTTCAG
ACGTGTGCTCTTCCGA
Indexing primer Y CAAGCAGAAGACGGCATACGAGATAGTTAAAGTGACTGGA
GTTCAGACGTGTGCTCTTCCGA
Indexing primer Z CAAGCAGAAGACGGCATACGAGATGAGGACCGTGACTG
GAGTTCAGACGTGTGCTCTTCCGA
Nucleotide sequences for adapters and index primers are proprietary designs of Illumina Corporation. Adapter B requires
50 phosphorylation (shown by the symbol /5phos/), and adapters A and B are mixed in equimolar proportions and
annealed prior to use. The underlined nucleotides in indexing primers X, Y, and Z reflect sequences which can be varied to
produce different indexing primers; these sequences are the reverse complement of what will be observed in raw sequencing
reads of the index region (i.e., primer X will be read as GAGGCGT, primer Y as TTTAACT, etc.). We order Adapters
A and B and Index PCR 1.0 and 2.0 as 100 nmol scale, HPLC-purified oligonucleotides from Integrated DNA
Technologies (http://www.idtdna.com) and indexing primers as 4 nmol scale “Ultramers”
8. Enzymes and reagents required for adapter ligation to cDNA

fragments: End-It DNA End-Repair kit (Epicentre catalog
#ER0720), Klenow 30 –50 exo (New England Biolabs
M0212S), 1 mM dATP solution in aliquots, Promega LigaFast
T4 DNA ligase and reaction buffer (catalog #M8221).
9. 2 % low-melt agarose (e.g., Amresco Agarose II) to prepare gels
for library size selection.
10. Indexing primers (Table 1) and post-size selection PCR reagent
(Phusion High-Fidelity PCR master mix, New England Biolabs
catalog #M0531S).
11. DNA-binding dye assay (such as the Invitrogen Qubit or
PicoGreen assays).
2.5 Suggested 1. FastQC (http://www.bioinformatics.babraham.ac.uk/projects/

Software Packages fastqc/).
for Data Analysis 2. E-miR (http://www.lgtc.nl/services/general_services/software_
tools/emir.php) [27].
3. miRdeep* (http://www.australianprostatecentre.org/research/
software/mirdeep-star) [28].
4. Bowtie (http://bowtie-bio.sourceforge.net/index.shtml) [29].
5. Tophat (http://tophat.cbcb.umd.edu/) [30].
6. HTSeq (http://www-huber.embl.de/users/anders/HTSeq/
doc/index.html). See especially the section on the included
Python script htseq-count.
7. DESeq (http://bioconductor.org/packages/release/bioc/
html/DESeq.html) [31]. Prior to publication of this article,
an extensively revised version (DESeq2) has also been released
(http://www.bioconductor.org/packages/2.13/bioc/html/
DESeq2.html), but has not yet been evaluated in the Matko-
vich laboratory workflow.
3 Methods
3.1 Isolation of Total 1. Use Trizol following the manufacturer’s directions precisely.
RNA Take the usual precautions for RNase-free benchtop work.
2. Flash-frozen intact tissue should be homogenized (e.g., an Ika
Ultra Turrax probe homogenizer or similar): keep tissue frozen
(in liquid nitrogen), wash down homogenization probe with a
few hundred μL of Trizol first, add, e.g., 0.5 mL Trizol directly
to tube as soon as it is removed from liquid nitrogen, and
homogenize. For mouse heart apices, we use a total of 1 mL
Trizol; the second aliquot of 0.5 mL is used to help wash tissue
pieces out of the probe. This results in good extraction of the
total RNA fraction. To enhance recovery of the small RNA
fraction, increase the length of the isopropanol precipitation

step at room temperature from the recommended 10 min to at
least 30 min (see Note 1). Resuspend the final precipitated
RNA pellet in 50–100 μL DEPC-treated or nuclease-free
H2O and heat at 65 C for 5 min to ensure complete dissolu-
tion. Store RNA preparations at 80 C.
3. Quantitate RNA using a standard (cuvette-based) UV spec-
trometer or a NanoDrop-style instrument and expect high
RNA purity; (OD260/OD280) 2.0. RNA integrity should
also be analyzed; we find that electrophoresing 1 μg of total
RNA on a 1 % TBE-agarose gel is generally sufficient. Sharp
28S and 18S bands should be evident with a 28S:18S intensity
ratio approaching 2.0. (Provided that the gel apparatus is trea-
ted beforehand with a suitable RNase inactivator such as RNa-
seZAP (Ambion) and that the gel and running buffer are
prepared with DEPC-treated H2O, there is no need to use a
formaldehyde-agarose gel unless Northern blot transfer of the
material is desired.) Another popular option, available in several
laboratories or core facilities, is the use of an Agilent BioAna-
lyzer to size-separate and quantitate RNA species via capillary
electrophoresis.
3.2 Preparation 1. To prepare small RNA (microRNA) sequencing libraries from

of Small RNA total RNA aliquots, we use the exact reagents and protocol
(microRNA) supplied by Illumina with their TruSeq small RNA preparation
Sequencing Libraries kits (catalog # RS-200-00xx, with the “xx” digits referring to
the particular set of indexing primers supplied; see below) and
will not repeat the steps involved here. However, we note that
we begin preparations using 1 μg of total cardiac RNA, as also
recommended in Illumina’s protocol, and have found that 14
cycles of PCR are optimal at the step that amplifies reverse-
transcribed, adapter-ligated microRNA species prior to size
selection on a 6 % polyacrylamide gel.
2. The Illumina TruSeq system allows for individual sequencing
libraries to be prepared with nucleotide “indexes,” such that
multiple libraries from different samples can be loaded on a
single sequencing lane and the sequencing reads so obtained
can be demultiplexed. We further discuss considerations for the
number of samples appropriate to load on a single lane (i.e., the
number of sequencing reads appropriate for each individual
sample) in Subheading 3.7, for RNA- and RISC-sequencing
libraries. However, our current experience is that five to eight
million aligned sequencing reads are more than enough for
accurate microRNA quantitation and comparison. Typically
we eliminate all microRNAs that represent less than one
millionth of the total sequencing reads obtained (~5) from
later analytic steps, as we regard these species to be too rare
to be of likely biological significance in our experiments.

Whether or not other noncoding RNA species will be present
in the sequencing libraries is highly dependent on the accuracy
of gel-based size selection during the preparation process. For
this reason, we also recommend combining and size-selecting
multiple microRNA-sequencing libraries for a particular ana-
lytical project on the same gel or lane.
3.3 Poly(A)+-RNA 1. Use oligo(dT)-coupled magnetic beads to purify polyadeny-

Enrichment lated RNA from total RNA samples. We typically use the
for mRNA-Seq Dynabeads mRNA purification kit and associated Dynal
magnet. Use no more than 100 μL of bead suspension per
5 μg total RNA sample (see Note 2). The following steps briefly
summarize the steps required for polyA+ RNA enrichment
using the Invitrogen kit.
2. Follow the protocol given with the kit to perform the first
round of isolation. Keep the supernatant (after initial binding
of RNA to beads) on ice, and after washing beads, elute polyA+
RNA in 15 μL 10 mM Tris–HCl, pH 8.2.
3. Re-prepare Dynabeads for binding by washing once in 1
“binding buffer”; it will be necessary to prepare a dilution of
the 2 buffer supplied with the kit.
4. Incubate the supernatant preserved from the initial binding
step at 65 C, 20 and then chill on ice. Perform an additional
round of RNA binding to beads.
5. Wash beads and elute further polyA+ RNA in another 15 μL
10 mM Tris–HCl, and add to the RNA eluted in the first
isolation. The aim is to maximize the amount of polyA+ RNA
obtained more so than maximally eliminating non-polyA+ spe-
cies; we find that there is minimal carryover of rRNA in our
final sequenced libraries.
6. Quantitate polyA+ RNA using a NanoDrop or low-volume
spectrophotometer. Store purified RNA at 80 C.
3.4 Ago2 1. For the analysis of RISC-bound RNA, Ago2 immunoprecipi-

Immunoprecipitation tates are prepared from intact tissue or cells and RNA is directly
Without Cross-linking extracted from these immunoprecipitates. Further rRNA
and RNA Extraction depletion is not performed on the very small amount of RNA
for RISC-Seq obtained from these procedures.
2. Flash-frozen mouse heart bases (approximately 2/3rd of the
heart) are homogenized in 500 μL ice-cold 50 mM Tris–HCl,
5 mM EDTA, 5 mM EGTA, pH 7.5, with Roche Complete
protease inhibitors, and yeast tRNA and SUPERnase-IN added
to final concentrations of 1 μg/μL and 1 U/μL, respectively.
Following homogenization, unbroken cellular material is
removed at 100 g, 50 , 4 C and Nonidet P-40 is added to a
final concentration of 0.5 % (w/v) to solubilize proteins (150 ,

4 C, end-over-end rotating) followed by removal of insoluble
material at 10,000 g, 150 , 4 C. The supernatant is added to
a previously prepared solution of 50 μL protein G-coupled
Dynabeads, to which 5 μg anti-mouse Ago2 monoclonal anti-
body has been previously bound according to the supplied
Dynabead protocol (see Note 3). Following 1 h rotational
incubation at 4 C, the beads are washed 3 with Dynabead
washing buffer, transferring the suspension to a fresh tube for
the last wash.
3. The last wash is performed in 200 μL of Dynabead washing
buffer. Prior to pelleting beads and removing supernatant,
extract 20 μL to a new tube and add an appropriate volume
of Laemmli denaturing sample buffer to prepare a protein
extract for SDS-PAGE. This will allow later SDS-PAGE analysis
to ensure that Ago2 was successfully immunoprecipitated.
4. Pellet beads from the remaining suspension, remove and dis-
card the supernatant, and add 500 μL of Trizol directly to the
Dynabeads and retained immunoprecipitated material to
extract RNA. Proceed with total RNA extraction as described
under Subheading 3.1, but see Note 4.
While there are various rRNA reduction procedures and
reagents available, e.g., RiboMinus™ from Invitrogen or Ribo-
Zero™ from Epicentre, we have not yet tested whether these leave
a sufficient yield of Ago2-RNA to allow successful library construc-
tion. Despite the availability of methodologies for RNA/cDNA
amplification (e.g., SMARTer® from Clontech), we do not attempt
to amplify the Ago2-immunoprecipitated RNA prior to fragmenta-
tion and cDNA synthesis. We noticed significant disruption to RNA
abundances measured after the use of such a technique [32].
3.5 For RNA-Seq This step chemically shears full-length RNA transcripts into
and RISC-Seq: RNA 200–500 nt fragments, permitting uniform reverse transcription
Fragmentation during subsequent stages. An important initial paper on RNA-Seq
methodology by Mortazavi et al. [33] documents full-length
RNA + oligo(dT) reverse transcription vs. fragmented RNA + ran-
dom primer reverse transcription, and concludes that fragmenta-
tion followed by random primer-based reverse transcription
improves the coverage of exonic elements with reduction of 30
bias in final sequencing libraries.
1. Take 100–500 ng polyA+ RNA, or the Ago2-
immunoprecipitated RNA; typically we use one-half of the
RNA obtained, thus preserving an aliquot should there be a
processing error in subsequent steps, and prepare a solution of
AAAA
AAAA
AAAA
mRNA AAAA
AAAA
Ago2 IP
Fragment mRNA
~200 nt
ds cDNA synthesis
No amplification
Ligate common sequencing adapters
Recover, amplify and index libraries

(PCR)
Illumina HiSeq
Fig. 1 Workflow for laboratory processing of input RNA to final sequencing

libraries
20 μL in “1 fragmentation buffer” (5 fragmentation buffer is

diluted appropriately with DEPC H2O).
2. Heat at 94 C, 2 min 30 s, then chill on ice (see Notes 5 and 6).
3. Purify the fragmented RNA from the acetate buffer using, e.g.,
Zymo RNA Clean + Concentrate™ columns, eluting in 8–9 μL
for maximum input into the SuperScript III first-strand cDNA
reaction mixture used at the next step.
4. All the following steps for cDNA synthesis and later sequencing
library preparation are the same, regardless of whether the
starting material is polyA+-selected mRNA or Ago2-
immunoprecipitated RNA (Fig. 1).
3.6 cDNA Synthesis Use the SuperScript III first-strand synthesis system. Each reaction
can accommodate 8 μL of fragmented polyA+ RNA. Follow the
3.6.1 First-Strand cDNA directions supplied, but do not treat with supplied RNase H at the
end of this procedure.
1. Combine 8 μL RNA, 1 μL 50 ng/μL random hexamers, 1 μL
10 mM dNTP mix.
2. Incubate at 65 C for 5 min, then place on ice for 1 min.
3. Add 10 μL “cDNA synthesis mix,” prepared as follows:
10 supplied RT buffer 2 μL

25 mM MgCl2 4 μL
0.1 M DTT 2 μL
RNaseOUT (40 U/μL) 1 μL
SuperScript III RT (200 U/μL) 1 μL
4. Incubate 10 min at 25 C, 50 min at 50 C, 5 min at 85 C, and

chill on ice.
3.6.2 Second-Strand To the 20 μL first-strand reaction, add on ice:

cDNA
DEPC-treated H2O 91 μL
5 second-strand reaction buffer 30 μL
10 mM dNTP mix 3 μL
E. coli DNA ligase (10 U/μL) 1 μL
E. coli DNA polymerase I (10 U/μL) 4 μL
E. coli RNase H (2 U/μL) 1 μL
1. Vortex gently to mix, and incubate for 2 h at 16 C. Do not allow

the temperature to rise above 16 C. Purify and concentrate the
synthesized cDNA via binding to DNA-binding and purification
columns (such as the Qiagen QIAquick PCR purification kit),
and elute the purified DNA in a volume of 34 μL of 10 mM
Tris–HCl, pH 8.5, for optimal input into Subheading 3.7.1
(see Note 6).
2. For RNase H, use the enzyme supplied by Invitrogen with
SuperScript III reverse transcription kits, or purchase separately
from New England Biolabs.
3.7 For RNA-Seq The protocol given below for generating sequencing libraries for
and RISC-Seq: Adapter the Illumina platform is based on the “ChIP sequencing protocol”
Ligation to cDNA available at http://bioinfo.mbb.yale.edu/array/resources.html
Fragments (Yale Center for Excellence in Genomic Science). The detailed
steps we describe in Subheadings 3.6 and 3.7 are specific for the
Illumina platform and will be different for different sequencing
platforms, requiring consultation of the manufacturers’ documen-
tation or other validated protocols.
3.7.1 End-Repair Use the “End-It DNA End Repair Kit.”

1. Combine and mix the following kit components in a microfuge
tube:
Double-stranded cDNA x μL
10 End-Repair Buffer 5 μL
2.5 mM dNTP Mix 5 μL
10 mM ATP 5 μL
Sterile water y μL
End-Repair Enzyme Mix 1 μL
Total reaction volume 50 μL
2. Incubate at room temperature for 45 min.

3. Purify on one QIAquick column using the QIAquick PCR
Purification Kit and protocol, eluting in 34 μL of the supplied
buffer EB (10 mM Tris–HCl, pH 8.5).
3.7.2 Addition of “A” Use Klenow (30 ! 50 exo), not the large Klenow fragment typi-
Base to 30 Ends cally used to fill in DNA ends.
1. Combine and mix the following components:
DNA from Step A 34 μL

Klenow buffer (NEB Buffer 2) 5 μL
1 mM dATP 10 μL

Klenow fragment (30 –50 exo ) 1 μL
2. Incubate for 30 min at 37 C.

3. Purify on one QIAquick MinElute column using the Qiagen
MinElute PCR Purification Kit and protocol. Elute in 10 μL
buffer EB (10 mM Tris–HCl, pH 8.5).
3.7.3 Ligation The adapter mix is a preannealed blend of Illumina-Index-AdapterA

of Common “Core” and 50 -phosphorylated Illumina-Index-AdapterB (see Table 1).
Adapters to cDNA For optimal performance of the ligation step, a reasonable
Fragments estimate of the cDNA fragment molarity should be calculated.
Moles of starting cDNA ¼ x g/(660 g/mol/bp y bp), where
bp ¼ typical length of heat + acetate-sheared fragments
(200–300 bp). For, e.g., 1 pmol starting ds cDNA, plan to add
10 pmol annealed adapter mix (tenfold molar excess). The aim is to
ensure complete ligation of cDNA fragments to core adapters,
without introducing an unnecessarily large quantity of these
adapters. The maximum yield of transcriptome-aligned sequencing
reads can be compromised if unligated or self-ligated adapters are
not completely removed.
1. Combine and mix the following components in a microfuge

tube:
DNA from Step B 10 μL

2 DNA ligase buffer (Promega) 15 μL
RNase, DNase-free water 2 μL
Adapter mix 1 μL
DNA ligase (Promega) 2 μL
2. Incubate for 15 min at room temperature.

3. Purify on one QIAquick MinElute column using the Qiagen
MinElute PCR Purification Kit and protocol. Elute in 10 μL
buffer EB (10 mM Tris–HCl pH 8.5).
3.8 For RNA-Seq Considerations for sample indexing and the number of separate
and RISC-Seq: Size samples to be combined and sequenced on a single lane:
Selection, PCR For standard mRNA work, it is likely that sample “indexing”
Recovery, Sample (conceptually similar to DNA barcoding) will be desired so that
Indexing, 8–12 indexed samples can be combined into one sequencing lane of
and Preparation ~240 million raw 50 nucleotide single-end reads—this represents
for Cluster Formation the current capacity of the Illumina HiSeq 2000-family sequencers
available at our core facility. Indexes are read separately and do not
detract from the 50 nucleotides available for sequence alignment.
The nature of RISC-Seq analysis involves a large amount of rRNA is
present in the Ago2 immunoprecipitate, which dilutes the reads
useful for determining the abundance of bound mRNAs. For this
reason we submit approximately half of the number of RISC-Seq
samples on a single lane in comparison to what we submit for
RNA-Seq on a single lane; in the final analysis, the number of
sequencing reads that map to mRNA species is largely similar.
Some examples of indexing primers are given in Table 1.
Particular core sequencing facilities may recommend particular
7-nucleotide indexes over others. Provided that the nucleotides
are chosen are varied, without >3 of the same nucleotide occurring
in series, we have observed no particular bias or difficulty in
recovering sequence reads resulting from the use of one index
sequence compared to another.
A common question in work of this nature relates to the
number of reads that are appropriate in order to perform accurate
quantitation of the starting RNAs and consequently to perform
differential expression analyses of a suitable power. This is a com-
plex and user-dependent issue, and will partly depend on the rarity
of genes of interest, especially if analysis of splice variants is
required. Some recommendations for appropriate sequencing
depth can be found in the following references: [34, 35].
For analysis of cardiac global mRNA, we allocate 20 million raw

sequencing reads to each library and typically find that 75 % of these
can be aligned to the transcriptome. For RISC-bound RNA
(including unwanted rRNA, as described above) 40 million
sequencing reads are allocated to each library, with an alignment
to the transcriptome of 40–50 %. Individual RNAs with less than
ten aligned reads are typically eliminated from downstream
analyses.
3.8.1 Size Selection 1. Gel-purify the DNA (using a 2 % low-melting agarose gel) by
of Sequencing Libraries via cutting a gel slice that does NOT include any DNA from a
Agarose Gel Purification potential adapter-adapter (self-ligated) band migrating at
~120 bp. The ligation product may not be readily visible at
this stage if the cDNA input was small (0.5 μg or less), and will
appear as a smear of fragments of different sizes in any case.
Isolate cDNA in the 150–300 bp range. If purifying multiple
libraries on a single gel, be very careful to avoid cross-
contamination by leaving at least two empty lanes between
each sample (and preferably three to four).
2. Purify the DNA from the agarose slice using a Qiagen Gel
Extraction Kit (see Note 7). Elute in 30 μL EB.
3.8.2 Recovery As for the “core” adapters described in Subheading 3.7.3,

and Amplification sequences are available in Table 1.
of Size-Selected Libraries, Components for PCR:
with Simultaneous Addition
of Index Sequences DNA from Step A 10 μL
Illumina Index PCR 1.0, (This will retain 20 μL of the gel-purified material from step A, should
Illumina Index PCR 2.0, there be a future need to add a different index sequence, or if
Illumina Index Reverse optimization of the number of PCR cycles is required. See Note 8.)
Primers
Phusion HF PCR master mix 25 μL
Index PCR 1.0, 25 μM 1 μL
Index PCR 2.0, 0.5 μM 1 μL
PCR index-specific reverse primer, 25 μM 1 μL
Nuclease-free H2O 12 μL
1. Amplify using the following PCR protocol:

30 s at 98 C.
[10 s at 98 C, 30 s at 65 C, 30 s at 72 C], 12 cycles total
for mRNA-Seq, 16 cycles total for RISC-Seq (see Note 8).
5 min at 72 C.
Hold at 4 C.
3.8.3 Quality Check Purify the amplified product on a QIAquick PCR purification
of Final Library column, similarly to the preceding steps in library generation, and
Preparations, elute in 30 μL buffer EB. Run 3–5 μL on a 2 % agarose gel (no need
and Preparation for low-melting agarose) to verify that the amplified fragment sizes
for Sequencing are similar to what was excised from the purification gel in step A.
Alternatively, an Agilent BioAnalyzer trace could be gained on the
amplified material. It is advisable to electrophorese the gel suffi-
ciently to allow good visualization of any material in the
100–120 bp range. Typically, detection of sharp bands in this
range indicates the presence of PCR primer-dimers or PCR ampli-
fication of core adapters that were not sufficiently removed after the
ligation step.
If these low-molecular-weight products represent a significant
fraction of the total sequencing library, then it is usually desirable to
perform a second size-selection step to remove them. Our experi-
ence has been that these products tend to be more evident when
beginning with a lower overall RNA input, such as occurs with
RISC-Seq. If the low-molecular-weight products are not removed,
they will compete with the desired cDNA fragments for binding to
the Illumina sequencing flowcell and decrease the number of useful
sequencing reads obtained as a result. One option is to perform
another gel purification round, this time on the PCR-amplified
material. Our core sequencing facility offers AMPure XP bead
purification (Beckman Coulter), on the basis of differential binding
of differently sized fragments, to achieve the same effect.
Quantitate the final libraries using a fluorescent DNA-binding
dye protocol (such as the Qubit or PicoGreen assays); we have
repeatedly found that this gives more accurate loading than Nano-
Drop or other UV-based measurement for sequencing library
quantitation. Prepare an equimolar mixture of the desired, indexed
libraries and dilute as appropriate for the chosen sequencing instru-
ment. As an example, we submit library mixtures at 10 nM to our
sequencing facility, which are then further diluted to 5 pM for
cluster formation on Illumina flowcells.
3.9 Sequence Data Typically, sequencing reactions and determination of nucleotide

Generation sequences (basecalling) are performed using highly specialized
reagents and equipment which are particular to the major platforms
(e.g., Illumina HiSeq, Applied Biosystems SOLiD, Life Technolo-
gies Ion Torrent, and the like). These steps are often performed at a
core facility of an academic institution or a commercial service.
As such, it is beyond the scope of this chapter to discuss detailed
protocols for generation of raw sequence data.
However, it is worthwhile to consider the format in which
basecalled sequencing read data are provided to the end user. It is
very helpful if demultiplexing (allocating reads according to
indexes) is performed prior to receipt of the data, as this process
can be very time-consuming on a desktop-class workstation.
A common format for distribution of sequencing read data, with

accompanying probabilities of basecalling accuracy, is the fastq (.fq)
format. We maintain archives of the original fastq data for our
RNA-seq experiments together with tables describing the experi-
mental design and index allocation. Ensuring that original data are
preserved at this level means that they are readily available for
alignment against new transcriptome versions (approved genome
releases) in the future.
3.10 Sequence Data In high-throughput analyses where hundreds to thousands of RNA

Quality Control species are being measured simultaneously, it can be challenging to
identify whether a particular biological or technical replicate is
behaving so differently from its counterparts that it should be
classed as an outlier. While this topic is covered in more detail
under Subheading 3.12, some simple determinations can help to
pinpoint problematic samples.
Analyzing fastq files with the freely available program FastQC can
point out anomalies such as a high percentage of self-ligated or
unligated sequencing adapters or indexing primers in the sequencing
read data, or an unduly large number of identical sequencing reads.
The program is written in Java and can thus be used in Windows,
Mac, and Linux/Unix operating environments.
It is also prudent to monitor the number of raw sequencing
reads present in each library, and the number which were success-
fully aligned to the transcriptome. These data are gained from the
procedures outlined in Subheadings 3.10 and 3.11, but should be
assessed before proceeding with downstream analyses. Inconsisten-
cies in these parameters reveal samples that were insufficiently
loaded on to the sequencer or that suffered degradation or other
problems during library preparation.
3.11 microRNA-Seq In this section and the following ones, a general workflow will be
Data Alignment described, together with recommendations for specific software
and Quantitation packages that we currently use. However, many of the programs
used for read alignment and quantitation are under continuous
development, or are superseded by more powerful alternatives,
and due to their “open-source” nature may become unavailable
or relocated to different web addresses without notice. The major-
ity of these programs are designed to be run in a Linux environ-
ment, either on a native Linux computer, a Linux emulator for
Windows (such as Cygwin), or the “Terminal” of Mac OS X.
Biological scientists that have not had previous exposure to Unix-
style operating systems may experience some level of discomfort
with the use of these programs at first, and some initial assistance
from a computational biologist or bioinformatician that is fluent in
the environment may be helpful in overcoming the initial learning
curve. A detailed description of Unix file management and com-
mands, read alignment procedures, and use of the popular statistical
software R and its associated Bioconductor programs is beyond the

scope of this chapter and would require a chapter of its own. We
have previously published example programs (Unix “shell scripts”)
that accomplish the steps used in our RNA-sequencing pipelines
(supplemental material of reference [25]). However, we encourage
biologists not to be dismayed by the unfamiliarity of these proce-
dures, and to at least develop a strong understanding of the
conceptual underpinnings, even if the bulk of the computational
work is outsourced. One helpful resource for the beginner, and
which allows many of the following procedures to be performed
via a friendly web interface, is the Galaxy project (http://
galaxyproject.org/).
microRNA-sequencing alignment and quantitation require the
following:
1. Trimming of adapter sequences from sequencing reads and
alignment of trimmed reads to annotated microRNAs and
other small RNAs.
2. Separation of microRNA-specific reads from other small RNA
reads.
3. Quantitation of microRNAs, typically as a proportion of the
total microRNA-specific reads, e.g., RpM (Reads per Million
microRNA reads).
We use the E-miR program [27], which is well documented
and runs in a Linux environment, to perform step 1. E-miR aligns
reads to all known small or noncoding RNAs in Ensembl databases.
It requires a separate alignment software module to be installed,
and for this we use Bowtie, a well-established and efficient program
which runs easily on desktop or laptop-class workstations [29]
(see Notes 9 and 10). Trimming of adapter sequences is necessary
because the length of the sequencing read is greater than that of the
mature microRNA sequence. We manually filter the resulting data
for microRNAs (step 2) by importing the output files from E-miR
into a spreadsheet program such as Microsoft Excel. Finally, we add
together all of the sequence reads for each sample and use this as the
denominator when calculating RpM.
Many other possibilities exist for analysis of small RNA-
sequencing data, such as the software miRdeep* [28].
3.12 mRNA- The successful alignment and quantitation of relatively short

and RISC-Seq Data sequencing reads to mRNA sequences is substantially more com-
Alignment plex than for microRNAs, in which the entire sequence of the
and Quantitation mature microRNA is encompassed by the length of the read.
Sequences aligning to different sections of the same mRNA must
be gathered together, and the alignment program must be aware
that sequences may span exon-exon boundaries.
Raw read data

(nucleotide sequence and position on flowcell lane)
FastQC
TopHat
(sequence alignment to genome/transcriptome)
HTSeq Cufflinks
(calculate reads per gene) (calculate FPKM)
DESeq Cuffdiff or other

(calculate differential expression) differential expression method
Fig. 2 Workflow for processing sequence reads to gene abundance data and
differential expression
Our workflow is as follows, and is the same for both mRNA and
RISC RNA sequencing (Fig. 2):
1. Alignment of sequencing reads to the desired genome/tran-
scriptome. We use the program TopHat [30, 36], which
employs the Bowtie aligner [29] at its core with specific con-
sideration for the splicing out of introns. TopHat can use a
genome-wide, annotated transcriptome database for its align-
ment procedure. Annotated genome-wide transcriptomes from
Ensembl or UCSC sources often contain multiple noncoding
RNA species in addition to mRNAs. While this is useful in some
circumstances, we prefer the alternative allowed by TopHat of
defining a more limited transcriptome and annotating only to
this. An advantage of such an approach is that it is easy to limit
alignments to mRNAs only (without inclusion of noncoding
RNAs), which are the only data we require for parallel mRNA-
and RISC-Seq analyses, and thus discarding all reads which
map to, e.g., the rRNA that is inevitably contained in RISC-
Seq libraries.
2. Conversion of aligned sequencing reads to gene expression
measures. The calculation of gene expression values (with or
without provision for determining isoform fractions and/or
new alternatively spliced transcripts) and the use of normaliza-
tion procedures to compensate for differences in library
sequencing depth and/or gene length [37, 38] have been the
subject of much discussion. These issues cannot be divorced
from debate over the most appropriate method for determina-
tion of significant differences in gene expression between treat-
ments, with three principal scientific groups involved in the
development of methods to achieve this. All have released

packages to calculate expression values and associated statistics
(HTSeq and DESeq [31]; Cufflinks and Cuffdiff [39]; and
edgeR [40]), and the strengths and weaknesses of each approach
have been the subject of energetic discussions in online forums
(e.g., http://seqanswers.com/forums/index.php).
HTSeq/DESeq, and edgeR, use the actual number of reads
aligned to a given mRNA, normalized only to the total reads
obtained for each individual sequence library, in their statistical
calculations. They can use either local (per gene) or global
(per gene set) dispersion (~ variance) to determine statis-
tical significance, and rely on a negative binomial distribution.
A variant of DESeq, DEXSeq [41], can calculate exon-level
differences for analysis of separate gene isoforms. Cufflinks and
Cuffdiff use a gene expression measurement that adjusts for both
library depth and gene length, the FPKM (Fragments per kb of
exon per million mapped reads), and offer to investigate isoform-
level differences, but estimate variances in a different manner
[39]. A workflow for the full “Tuxedo” suite (Bowtie, Tophat,
Cufflinks, and Cuffdiff) has been described by the programs’
authors [36].
Our internal workflow currently uses the HTSeq/DESeq
method for normalization of differences in sequencing depth
between individual samples, and reports gene expression as the
normalized number of reads for each gene (without any attempt
to compensate for gene length, or any attempt to discover novel
transcripts). For convenience, we do use Cufflinks to process
TopHat-aligned data into FPKM values, which give a sense of the
concentration (molarity) of each gene in the preparation that is not
possible when using HTSeq reads alone, but do not use Cuffdiff for
differential expression analyses. By monitoring HTSeq reads and
Cufflinks FPKM values, we eliminate genes from our data sets that
have insufficient reads for reliable determination and/or are likely
expressed at cellular levels of limited biological significance. Indeed,
we perform similar filtering of raw microRNA data prior to statisti-
cal analysis, which includes elimination of all RNAs in the output
files for which zero reads were obtained. Thus, false discovery rates
(which are based partly on the number of input RNAs) are calcu-
lated in accordance with the number of detected RNA species,
rather than in accordance with the (greater) number of RNA
species present in an annotation file.
3.13 Statistical Tools In many of our earlier publications using deep sequencing meth-
to Determine odologies, we used a fairly straightforward method of importing
Differential Gene Cufflinks-derived FPKM values into statistical software such as
Expression Partek Genomics Suite (http://www.partek.com/), log-
transformed the data (achieving a better fit to the normal
distribution) and calculated p-values and associated false discovery

rates for comparisons between treatment conditions. However, in
the light of studies comparing various strategies for read depth
normalization across different individual samples [42] and differ-
ential gene (mRNA) expression methods [37], we now use the
DESeq package [31], to compute differential microRNA and
mRNA expression. DESeq uses a negative binomial distribution
in comparison to the Gaussian distribution assumed by other tools
for log-transformed data, and is designed to be more robust in
contexts where relatively few genes are highly abundant and there
is a majority of less abundant genes. edgeR is based on similar
concepts [40].
DESeq comprises a series of data import and manipulation
commands for the R statistical environment. The software is freely
available and includes thorough documentation. One potential
stumbling block for the use of this and many related data analysis
packages involves correct formatting of the read data from each
sample (which is specific to each package). For small numbers of
samples, careful import into a spreadsheet program with any neces-
sary manual rearrangement may suffice. However, for import and
grouping of large numbers of samples, with the attendant possibi-
lities for operator error, it may prove valuable to employ a user-
friendly relational database such as Microsoft Access to import and
correctly group data in a more automated fashion. While there is
insufficient space to give detailed procedures here, mastering the
concepts of “crosstab query” and “lookup tables” should prove to
be useful in converting the output from programs such as HTSeq
and Cufflinks into the data tables required for statistical analysis.
An additional consideration is the number of biological
replicates that are necessary to measure a given differential gene
expression event (i.e., the experiment’s power). Once again, the
multidimensional nature of the data obtained from these assays
means that classical methods of analyzing experimental power are
not well suited to these tasks. In practical terms, we aim to obtain
six biological replicates of any given treatment condition, which is
readily achievable for most studies involving mouse hearts.
Our experience has been that the reduction in variance obtained
by using six (vs. three or four replicates), and subsequent increase in
power, far outweighs the additional considerations of cost, espe-
cially with continued decreases in the cost of sequencing per base.
4 Notes
1. On the issue of effective precipitation of small RNA using the

Trizol method, see ref. 43, which suggests that some small
RNAs may be selectively lost during Trizol extraction. While
the title may not sound especially hopeful, in this important
technical paper the authors only use a 10 min precipitation in

isopropanol with Trizol when comparing to microRNA extrac-
tion with Ambion’s miRVana kit using a column-binding step,
and the “small number of cells” referred to in the manuscript as
a difficulty are those sufficient for a total RNA yield of only
~1 μg. Our RNA extraction protocol using Trizol calls for at
least a 30 min isopropanol precipitation step. The extraction
deficit appears to be minimized at 10 μg yield and absent at
50 μg (see figure 1E of [43]), and considering the approximate
total RNA yield of at least 50 μg from even a mouse heart apex
(1/3rd of the heart) this should be of no or only minor
concern.
2. This quantity of oligo(dT)-Dynabeads (100 μL) has a binding
capacity of 1 μg of polyA+ RNA. Considering that 5 μg of total
RNA is unlikely to contain more than 100–150 ng of polyA+
RNA (2–3 %), it is likely possible to reduce the amount of
Dynabead suspension further if desired.
3. The Wako Pure monoclonal anti-Ago2 antibody recommended
here is targeted toward mouse, rat, and hamster Ago2. While
we have not used it, an alternate monoclonal antibody is avail-
able from Wako Pure for anti-human Ago2 immunoprecipita-
tion. For smaller quantities of tissue, less anti-Ago2 antibody
should be required for maximal RNA yield, although this will
need to be optimized by the user.
4. Since only small quantities of RNA are present in the Ago2
immunoprecipitate, it will not be possible to visualize an RNA
pellet after isopropanol precipitation and centrifugation. To
avoid unintentional loss of RNA at this stage, add 20 μg
RNase-free glycogen immediately prior to isopropanol precipi-
tation, and precipitate for 30 min at room temperature. After
washing with 75 % ethanol, resuspend the final dried pellet in
30 μL 10 mM Tris–HCl, pH 8.2. Store RNA at 80 C.
5. Time the length of heat treatment exactly to ensure reproducibil-
ity of chemical shearing between libraries prepared on different
occasions.
6. Out of an abundance of caution to protect against RNA degrada-
tion, we perform RNA fragmentation, column purification, and
first- and second-strand cDNA synthesis in a single working day.
7. Perform the steps listed as “optional” in the Qiagen booklet by
adding 1 gel volume of isopropanol to the melted agarose, and
an additional wash in 0.5 mL buffer QG prior to washing in
buffer PE.
8. 12 cycles of PCR at this step is optimal when starting with
100–200 ng of polyA+ RNA as input into Subheadings 3.5–3.7,
and using 1/3rd of the size-selected, gel-purified cDNA as
input into PCR. The lower input of RNA derived from Ago2
immunoprecipitates necessitates further amplification; our

rule-of-thumb has been that an immunoprecipitation from 2/
3rd of a mouse heart, from which a sequencing library is
prepared in the same way as for mRNA-Seq, requires at least
an additional four PCR cycles in order to be visualized and
accurately quantitated.
Overamplification and introduction of a degree of PCR bias
is possible at this stage, and the use of too many PCR cycles is
typically indicated by the formation of distinct bands within the
desired size range, rather than an even “smear” of fragments
across the range. If this is seen to occur, it is desirable to begin
PCR with a new aliquot of 10 μL of the size-selected,
gel-purified library and adjust the number of PCR cycles.
9. The E-miR version using Bowtie as its alignment module
appears to have a bug (at least as of late 2011) in which mature
microRNAs which originate from multiple genomic loci fail to
map correctly. We have overcome this by manually altering the
line in the E-miR Perl program which invokes the Bowtie
command, using appropriate parameters for Bowtie to allow
alignment of such “multiply-mapped” reads [29].
Specifically, we modify line 563 of the EmiR_Bowtie.pl file
as follows:
(from line 551)
###################################################
###################################################
## POST ALIGNMENT SUBS
###################################################
###################################################
#############
sub load_bowtie_output_files # and process
{
& time; print "Running Bowtie alignment using
$processors processors start ¼ $theTime. . .\n";
# system("cat *_TBDL | awk ’{print \">\"\$1\"\\n
\"\$1}’ > bowtie_temp.fa ");
$outfile ¼"bowtie_temp.fa_OUT";
(line 563) system("$bowtie_location -p $processors -n 1

-l 32 -f -m 1 - -best - -strata $genome_location bowtie_-
temp.fa $outfile");
(change line 563 to) system("$bowtie_location -p $pro-
cessors -n 1 -l 32 -f $genome_location bowtie_temp.fa
$outfile");
10. E-miR aligns reads against genomic regions in which micro-
RNA precursors are embedded. While we do find the program
useful, we have noticed using the Bowtie version of the
program (modified as described in Note 9) that mature micro-

RNAs which originate from more than one genomic locus
(e.g., miR-9 and miR-133a) can be randomly assigned to one
locus or the other in output tables. We manually add together
the microRNA reads from multiple loci to arrive at a single
value for the number of reads pertaining to that microRNA.
Acknowledgements
Related work in the authors’ laboratories is supported by the

NIH-sponsored Diabetes Research Center at Washington University,
grant 5 P30 DK020579 (to S.J.M.) and NIH grant R01 HL108943
(to G.W.D.).
References
1. Lynn FC, Skewes-Cox P, Kosaka Y et al (2007) 10. Matkovich SJ, Zhang Y, Van Booven D et al
MicroRNA expression is required for pancre- (2010) Deep mRNA sequencing for in vivo
atic islet cell genesis in the mouse. Diabetes functional analysis of cardiac transcriptional
56:2938–2945 regulators. Application to Gαq. Circ Res
2. Kuehbacher A, Urbich C, Zeiher AM et al 106:1459–1467
(2007) Role of Dicer and Drosha for endothe- 11. Hu Y, Matkovich SJ, Hecker PA et al (2012)
lial microRNA expression and angiogenesis. Epitranscriptional orchestration of genetic
Circ Res 101:59–68 reprogramming is an emergent property of
3. Chen JF, Murchison EP, Tang R et al (2008) stress-regulated cardiac microRNAs. Proc
Targeted deletion of Dicer in the heart leads to Natl Acad Sci U S A 109:19864–19869
dilated cardiomyopathy and heart failure. Proc 12. Matkovich SJ, Van Booven DJ, Youker KA et al
Natl Acad Sci U S A 105:2111–2116 (2009) Reciprocal regulation of myocardial
4. Melkman-Zehavi T, Oren R, Kredo-Russo S microRNAs and messenger RNA in human
et al (2011) miRNAs control insulin content cardiomyopathy and reversal of the microRNA
in pancreatic β-cells via downregulation of signature by biomechanical support. Circula-
transcriptional repressors. EMBO J tion 119:1263–1271
30:835–845 13. Ikeda S, Kong SW, Lu J et al (2007) Altered
5. van Rooij E, Olson EN (2007) MicroRNAs: microRNA expression in human heart disease.
powerful new regulators of heart disease and Physiol Genomics 31:367–373
provocative therapeutic targets. J Clin Invest 14. van Rooij E, Sutherland LB, Liu N et al (2006)
117:2369–2376 A signature pattern of stress-responsive micro-
6. Liu N, Olson EN (2010) MicroRNA regu- RNAs that can evoke cardiac hypertrophy and
latory networks in cardiovascular development. heart failure. Proc Natl Acad Sci U S A
Dev Cell 18:510–525 103:18255–18260
7. Chen CY, Zheng D, Xia Z et al (2009) Ago- 15. Putt ME, Hannenhalli S, Lu Y et al (2009)
TNRC6 triggers microRNA-mediated decay Evidence for coregulation of myocardial gene
by promoting two deadenylation steps. Nat expression by MEF2 and NFAT in human
Struct Mol Biol 16:1160–1166 heart failure. Circ Cardiovasc Genet
8. Kozomara A, Griffiths-Jones S (2011) miR- 2:212–219
Base: integrating microRNA annotation and 16. Margulies KB, Matiwala S, Cornejo C et al
deep-sequencing data. Nucleic Acids Res 39: (2005) Mixed messages: transcription patterns
D152–D157 in failing and recovering human myocardium.
9. Matkovich SJ, Hu Y, Dorn GW II (2013) Reg- Circ Res 96:592–599
ulation of cardiac microRNAs by cardiac 17. van Rooij E, Sutherland LB, Thatcher JE et al
microRNAs. Circ Res 113:62–71 (2008) Dysregulation of microRNAs after
myocardial infarction reveals a role of miR-29 31. Anders S, Huber W (2010) Differential expres-
in cardiac fibrosis. Proc Natl Acad Sci U S A sion analysis for sequence count data. Genome
105:13027–13032 Biol 11:R106
18. Ozsolak F, Milos PM (2011) RNA sequencing: 32. Matkovich SJ, Van Booven DJ, Eschenbacher
advances, challenges and opportunities. Nat WH et al (2011) RISC RNA sequencing for
Rev Genet 12:87–98 context-specific identification of in vivo micro-
19. Jensen KB, Darnell RB (2008) CLIP: cross- RNA targets. Circ Res 108:18–26
linking and immunoprecipitation of in vivo 33. Mortazavi A, Williams BA, McCue K et al
RNA targets of RNA-binding proteins. Meth- (2008) Mapping and quantifying mammalian
ods Mol Biol 488:85–98 transcriptomes by RNA-Seq. Nat Methods
20. Hafner M, Landthaler M, Burger L et al (2010) 5:621–628
PAR-CliP—a method to identify 34. Toung JM, Morley M, Li M et al (2011) RNA-
transcriptome-wide the binding sites of RNA sequence analysis of human B-cells. Genome
binding proteins. J Vis Exp 41:e2034 Res 21:991–998
21. Hafner M, Landthaler M, Burger L et al (2010) 35. Labaj PP, Leparc GG, Linggi BE et al (2011)
Transcriptome-wide identification of RNA- Characterization and improvement of RNA-
binding protein and microRNA target sites by Seq precision in quantitative transcript expres-
PAR-CLIP. Cell 141:129–141 sion profiling. Bioinformatics 27:i383–i391
22. Licatalosi DD, Mele A, Fak JJ et al (2008) 36. Trapnell C, Roberts A, Goff L et al (2012)
HITS-CLIP yields genome-wide insights into Differential gene and transcript expression
brain alternative RNA processing. Nature analysis of RNA-seq experiments with TopHat
456:464–469 and Cufflinks. Nat Protoc 7:562–578
23. Chi SW, Zang JB, Mele A et al (2009) Argo- 37. Bullard JH, Purdom E, Hansen KD et al
naute HITS-CLIP decodes microRNA-mRNA (2010) Evaluation of statistical methods for
interaction maps. Nature 460:479–486 normalization and differential expression in
24. Karginov FV, Conaco C, Xuan Z et al (2007) A mRNA-Seq experiments. BMC Bioinformatics
biochemical approach to identifying micro- 11:94
RNA targets. Proc Natl Acad Sci U S A 38. Oshlack A, Wakefield MJ (2009) Transcript
104:19291–19296 length bias in RNA-seq data confounds systems
25. Matkovich SJ, Hu Y, Eschenbacher WH et al biology. Biol Direct 4:14
(2012) Direct and indirect involvement of 39. Trapnell C, Williams BA, Pertea G et al (2010)
microRNA-499 in clinical and experimental Transcript assembly and abundance estimation
cardiomyopathy. Circ Res 111:521–531 from RNA-Seq reveals unannotated transcripts
26. Dorn GW II, Matkovich SJ, Eschenbacher WH and isoform switching during cell differentia-
et al (2012) A human 30 miR-499 mutation tion. Nat Biotechnol 28:511–515
alters cardiac mRNA targeting and function. 40. Robinson MD, McCarthy DJ, Smyth GK
Circ Res 110:958–967 (2010) edgeR: a bioconductor package for dif-
27. Buermans HP, Ariyurek Y, van Ommen G et al ferential expression analysis of digital gene
(2010) New methods for next generation expression data. Bioinformatics 26:139–140
sequencing based microRNA expression 41. Anders S, Reyes A, Huber W (2012) Detecting
profiling. BMC Genomics 11:716 differential usage of exons from RNA-seq data.
28. An J, Lai J, Lehman ML et al (2013) miR- Genome Res 22:2008–2017
Deep*: an integrated application tool for 42. Dillies MA, Rau A, Aubert J et al (2013) A
miRNA identification from RNA sequencing comprehensive evaluation of normalization
data. Nucleic Acids Res 41:727–737 methods for Illumina high-throughput RNA
29. Langmead B, Trapnell C, Pop M et al (2009) sequencing data analysis. Brief Bioinform
Ultrafast and memory-efficient alignment of 14:671–683
short DNA sequences to the human genome. 43. Kim YK, Yeo J, Kim B et al (2012) Short
Genome Biol 10:R25 structured RNAs with low GC content are
30. Trapnell C, Pachter L, Salzberg SL (2009) selectively lost during extraction from a small
TopHat: discovering splice junctions with number of cells. Mol Cell 46:893–895
RNA-Seq. Bioinformatics 25:1105–1111
Chapter 4
Next-Generation Sequencing Technology in the Genetics

of Cardiovascular Disease
Cecilia Vecoli
Abstract
In recent years, next-generation sequencing (NGS) technologies have revolutionized approaches to genetic
studies, making whole-genome sequencing a possible way for obtaining global genomic information. At
present, three most NGS platforms are used in genetics for clonally amplified templates. These technologies
share general processing steps but differing in specific technical details that determine their limits or
advantages. NGS has been recently shown to have great potential for identifying novel causative mutations
in different disorders. It is expected that the NGS will be increasingly important in the study of inherited
and complex traits such as cardiovascular diseases (CVDs). Indeed, the identification and characterization of
genes that enhance prediction of CVDs risk remain an important challenge for improving prevention and
treatment.
Key words Next-generation sequencing, Genetics, Cardiovascular disease, Hypertrophic

cardiomyopathy
1 Introduction
Sanger sequencing was used for the Human Genome Project [1], but
despite significant technical improvements to this “first-generation”
technology, new second-generation screening or next-generation
sequencing (NGS) technologies are required for sequencing mul-
tiple human genomes at adequate depth. Over the last 5 years,
NGS technology has revolutionized the genomic (and transcrip-
tomic) approach to biology reducing the cost of sequencing on a
per-base pair (bp) basis and increasing the output of sequencing
from a few hundred bps by each Sanger analysis to about 600
billion bps per NGS machine run [2, 3]. Whole-genome
sequencing has become a possible and efficient way to obtain
global genomic information [4]. Thus also in the field of cardio-
vascular diseases (CVDs), hundreds of loci associated with these
pathologies have been identified [5].
51
52 Cecilia Vecoli
At present, three companies have provided the most widely

used NGS systems for clonally amplified templates—Illumina,
Roche, and Life Technologies. The three technologies show impor-
tant differences in term of sample preparation, sequencing chemis-
try, sequencing read lengths, run time, and gigabase output for run.
The unique combination of specific protocols distinguishes the
three NGS technologies determining their limits or advantages
[6]. In this chapter an overview of the various NGS technologies
for clonal amplification is provided and an example of the NGS
approach for target resequencing of genes associated with hyper-
trophic cardiomyopathy (HCM) is reported as a basic example of
NGS approach in cardiovascular disease. Indeed, cardiovascular
disease is a class of many pathologies of the heart and blood vessels
(including inherited cardiomyopathies, coronary artery disease,
peripheral artery disease, hypertension, congenital heart disease,
and heart failure) difficult to cluster and define in a unique, well-
defined protocol of investigation.
1.1 NGS Platforms The three NGS technologies share general processing steps while
for Clonally Amplified differing in specific technical details (Fig. 1). The first common step
Template in NGS is the preparation of a “library” comprising genomic DNA
(or cDNA) fragments ligated to platform specific oligonucleotide
adapters.
The input nucleic acid can be genomic DNA, standard or long-
range PCR amplicons (or cDNA).
To achieve fragmentation, the input nucleic acid is fragmented
(by nebulization, sonication, or enzymatic digestion) to generate
random overlapping fragments typically in the size range of
150–600 bp depending on platform and application requirements
[7]. A common theme among NGS technologies is that the tem-
plate is immobilized or attached to a solid surface or support. The
immobilization of spatially separated template sites allows
thousands to billions of sequencing reactions to be performed
simultaneously. For the Roche and Life Technologies platforms,
clonal amplification uses emulsion PCR (emPCR) and requires
hybridizing the adapter modified fragment library to beads that
display oligonucleotides with sequences complementary to adapter
sequences [8–10]. Briefly, in emPCR, a reaction mixture consisting
of an oil–aqueous emulsion is created to encapsulate bead–DNA
complexes into single aqueous droplets. DNA fragments are ampli-
fied on the beads in emPCR, resulting in beads carrying tens of
millions of copies of the original DNA fragment. After PCR, the
emulsion is broken, and DNA-coated beads are purified, dena-
tured, and loaded into the wells of a “picotiter” plate. EmPCR
beads can be deposited into individual PicoTiterPlate (PTP) wells
(Roche) [11] in which the NGS chemistry can be performed or
chemically attached to a glass slide chemically cross-linked to an
amino-coated glass surface (Life Technologies) [12].
Fig. 1 Next-generation sequencing process steps for platforms requiring clonally amplified templates
(Illumina, Roche, and Life Technologies). Input DNA is random fragmented and ligated to platform-specific
oligonucleotide adapters. Individual library fragments are clonally amplified by emulsion PCR (Roche and Life
Technologies) or solid surface bridge amplification (Illumina). Flow cell sequencing of clonal templates
generates luminescent (Roche) or fluorescent (Illumina and Life Technologies) signals. From luminescent or
fluorescent signals images are generated that, then, will be algorithmically processed into sequence reads.
These data can be exported and next assembly and aligned thanks special bioinformatics tools
54 Cecilia Vecoli
The wells of the picotiter plate are large enough for only a
single bead to be loaded; each well carrying a bead will generate
an individual DNA sequence. Pyrosequencing is performed by
cyclical addition of individual nucleotides, sulfurylase, and lucifer-
ase. As each nucleotide is incorporated into the growing strand, an
inorganic pyrophosphate group is released and converted to ATP
by the sulfurylase. Luciferase uses the ATP to convert luciferin to
oxyluciferin, producing a light signal that is directly proportional to
the number of inorganic pyrophosphate molecules released and the
number of nucleotides incorporated [13].
In Life Technologies protocol, beads are deposited onto a slide
and primers hybridize to the adaptor sequence on the template
beads. Four fluorescently labeled probes compete for ligation to the
sequencing primer. Multiple cycles of ligation, detection, and cleav-
age are performed, with the number of cycles determining the even-
tual read length of up to 75 bp. More recently, Life Technologies
acquired Ion Torrent to release the PGM™ and Proton™ sequencers
that use a very similar approach to the original Roche/454 pyrose-
quencing. Indeed, the sequencing is performed on a semiconductor
chip that has wells into which individual emulsion PCR beads can be
loaded. Sequencing is carried out in a similar cyclical manner, but as
each nucleotide is incorporated hydrogen ions are released, changing
the pH of the well. This is detected by the ion sensors in the semi-
conductor chip, which then produces a “flowgram” format.
In contrast to emPCR, Illumina utilizes a unique “bridge
amplification” reaction that occurs on the surface of the flow cell
[14]. Certainly, the isothermal bridge amplification considerably
reduces the complexity of sample processing compared to emPCR
even if new front-end automation devices for the emulsion PCR
steps have been developed in recent years.
In Illumina chemistry, libraries are prepared from two oligonu-
cleotides that share complementarity at one end; when annealed
and ligated to DNA fragments they allow different sequences to be
added to the end of each fragment. The library of DNA fragments
is enriched by PCR ready for clustering and sequencing. Libraries
are denatured to pM concentration and are introduced to a specific
flow cell. The fragments hybridize to complementary oligonucleo-
tides on the surface of the flow cell and are copied by DNA
polymerase. These daughter molecules are then “bridge-amplified”
by repeated cycles of chemical denaturation and polymerase exten-
sion to produce discrete clusters each containing about 1,000
molecules. Sequencing-by-synthesis technology uses fluorescently
labeled and reversibly blocked terminator deoxynucleoside-
triphosphates in a cyclic sequencing reaction. Nucleotides are
incorporated by DNA polymerase into the growing DNA strand,
the flow cell is imaged to determine which nucleotide has been
incorporated into each individual cluster, and finally the terminator
is removed by chemical cleavage ready for the next round of incor-
poration, imaging, and cleavage [14, 15].
Next-Generation Sequencing Technology in the Genetics of Cardiovascular Disease 55
The results of the sequenced segments are called “reads,”

which could be 25–100 bps from one or both ends. The massive
capacity of NGS allows the sequencing of many randomly over-
lapping DNA fragments; therefore, each nucleotide in targeted
regions may be included in many reads, allowing repeated analysis
which provides depth of coverage. Increased depth of coverage
usually improves sequencing accuracy, because a consensus voting
algorithm is used in determining the final nucleotide calls.
The depth of coverage is a measure of the number of times that
a specific genomic site is sequenced during a sequencing run. In
amplicon resequencing, for example, the target might be 150
coverage, meaning that—on average—each targeted base is
sequenced 150 times. This does not mean that every targeted
base is sequenced every time; some segments may be read 200 or
more times, while others might only be read once or twice, or not at
all. The higher the number of times that a base is sequenced, the
better the quality of the data.
A major challenge in NGS technologies remains data analysis
and interpretation. Bioinformatics skills are critical for successful
analysis and interpretation of NGS data, and this is an area that will
present a significant challenge to the diagnostic laboratory. The first
step in finding DNA variations is to align the NGS reads with the
reference genome [16]. Indeed, the production of many reads (tens
or hundreds of Gbp for each run) has made necessary the develop-
ment of several bioinformatics tools for the correct alignment/
assembly and for the analysis of a large amount of data. Recently a
unified analytic framework to identify genetic variations among
multiple samples has been proposed [17]. For researchers with
bioinformatic training and knowledge, many open-source aligners
are available. A summary of the sources for downloading various
software packages is given at http://seqanswers.com/forums/
showthread.php?t¼43.
After analysis, differences between the subject’s sequence and
the reference sequence are reported in a list that usually contains
large numbers of variants. The next and most important step is to
determine the clinical significance of these variants by sorting out
benign SNPs and those that cause diseases.
In recent years three benchtop high-throughput sequencing
platforms—the 454 GS Junior (Roche), MiSeq (Illumina), and
Ion Torrent PGM (Life Technologies)—have been introduced
and are a real opportunity for the use of NGS in a clinical setting.
Indeed, these benchtop instruments offer several advantages over
the larger “whole-genome” sequencing instruments. They are gen-
erally much cheaper to buy, faster to run, and the volume of data
generated is smaller and, therefore, easier to manage. With these
platforms, data analysis has become very easy and costumer-made
without requiring the aid of a bioinformatics team. They can
perform multiple sequencing runs in a day, offering laboratories
56 Cecilia Vecoli
greater throughput and flexibility. This comes with a reduction in

the Gb of sequence data generated. Rather than hundreds of Gb,
only single digit or tens of Gb are generated. This is not the obvious
drawback that it may seem as the clinically relevant portion of the
genome is currently quite small.
1.2 NGS Defining an exclusive methodological strategy for NGS approach in

Technologies in assessing cardiovascular disease is a challenge for two reasons:
Cardiovascular
1. the complexity of cardiovascular disease.
Disease Gene Although there is a clear hereditary component in the etiology
Discovery of most cardiovascular diseases, many forms of CVD exist.
There are monogenic (rare) forms in which the mutation of a
single gene causes the pathology. Furthermore, in the clinical
practice the most common CVDs (i.e., coronary artery disease)
are complex traits that arise from elaborate gene–gene and
gene–environmental interactions that confer risk for disease in
a probabilistic manner. In these cases a series of polymorphic
variants in several genes increases the risk of developing the
disease. Thus, it is difficult to select a panel of genes or a panel
of amplicons as a valuable, suitable example for describing the
use of NGS.
2. the differences in NGS technologies.
As described above, the three most important NGS platforms
are totally different in terms of sample preparation and chemis-
try of sequencing. Furthermore, different models with differ-
ent performances are available within the same brand and each
platform is continuously upgraded. For instance, the early
Solexa-based sequencers from Illumina generated reads of
35 bp in 2007 and generated around 30 million sequences or
1 Gb of data from a flow cell. Read length has increased to
150 bp on the HiSeq 2500 system, which generates over 1.5
billion sequences (or 3 billion paired-end sequences) and
300 Gb of data from a single flow cell so far.
Thus, the panels of genes to sequence have to be selected
according to the platform’s performance and once that gene
selection has been made, different protocols have to be applied.
For example, Illumina helps in the study design and provides
specific protocols, from library preparation to the end of the
study ensuring the success of experiment. It is crucial to rigor-
ously follow the procedures. If you decide to use one of Illu-
mina’s kits for library development, the brand of instruments
to be used in the different phases of library preparation is also
often mentioned or recommended. It is not advisable to rear-
range the protocol: the success rate drops dramatically and the
company disclaims any responsibility.
Due to numerous problems related to the description of a
specific protocol for the use of the NGS in the cardiovascular
HiSeq HiSeq HiSeq 2500 HiSeq 1500 MiSeq

2500/2000 1500/1000
Output 600Gb 8.5Gb

(maximum)
2-11days 4-39 hours
Run Time
Max Read 2x100pb 2x500pb

Lenght
Pair-end reads 3 billion 750 million
(maximum)
100 ng –1 µg
Required input *
Fig. 2 Major characteristics of different models of Illumina’ sequencing plat-

forms. *Input DNA 100 ng to 1 μg with TruSeq
field, in the final section of this chapter, a specific application of

targeted resequencing of genes implicated in primary hypertro-
phic cardiomyopathy (HCM) using the MiSeq benchtop is
reported. MiSeq is a fully integrated personal sequencer
especially convenient for amplicon (re)sequencing and easily
available in a research/clinical diagnostic laboratory. The infor-
mation provided here is totally representative to illustrate some
concepts described previously (Fig. 2).
1.3 Next-Generation HCM is a heterogeneous disorder of the myocardium where a clear

System in genetic component is recognized although nongenetic factors,
Hypertrophic such as lifestyle, sex, and age, have a role in modulating clinical
Cardiomyopathy presentation [18–21]. Mutations in numerous genes, many of
which encode protein components of the sarcomere, can cause
HCM. Many gene mutations that cause HCM are missense variants
that lead to structurally abnormal polypeptides incorporation into
cardiac myofilaments that disrupt normal sarcomere function. The
genes strongly associated with HCM are listed in Table 1 [22]. For
each such gene, a range of different mutation frequencies has been
reported, and clinical manifestations have been highly heteroge-
neous, both of which limit the use of genetic information in clinical
practice. A targeted resequencing of the multiple causative genes
could be the right approach for research but also for diagnostic
development.
The approach describes below can be commonly used to
sequence many individuals in order to discover, screen, or validate
genetic variation within a population. In the example reported
here, a target resequencing of exons and flanking intronic bases of
genes has been performed.
Figure 3 shows the schematic flow chart of steps for target
resequencing using MiSeq Illumina platform.
Table 1
Genes recognized as strongly associated with HCM
Protein Gene
Myosin, heavy chain 7 MYH7
Myosin binding protein C MYBPC3
Troponin T type 2 TNNT2
Troponin I type 3 TNNI3
Cysteine and glycine-rich protein 3 CSRP3
Tropomyosin α TPM1
Myosin light chain 2 MYL2
Actin ACTC
Myosin light chain 3 MYL3
Protein kinase AMP activated, γ2 PRKAG2
Phospholamban PLN
Troponin C type 1 TNNC1
Titin TTN
Myosin, heavy chain 6 MYH6
Titin-cap TCAP
Caveolin 3 CAV3
?
Design Studio
Library Preparation
Libraries loading into the reagent

cartridge
Sequencing Run Setup
Data Quality Check
Data Analysis
Fig. 3 Schematic flow chart of steps for target resequencing using MiSeq
Illumina platform
Note: The first thing is to identify the Sample Preparation Kit that
best fits our sequencing project. In the Illumina web site there is a
Sample Preparation Kit Selector Tool (http://support.illumina.
com/training/sample_prep_kit_selector.ilmn) that helps in the
search. To resequence several amplicons, the TruSeq Custom
Amplicon is recommended.
2 Materials
1. DNA extraction kit: GenElute™ Blood Genomic DNA Kit,

Sigma Aldrich®.
Note: Other commercially available or laboratory validated
DNA extraction methods are compatible with TruSeq Custom
Amplicon kit.
2. TruSeq Custom Amplicon kit (Illumina). This kit has to be
custom designed by using the “Design Studio” web tool (see
below).
Note: For Illumina sequencing platform, other companies such as
Agilent Technologies Inc., can provide kits for the enrichment of
(costumer) selected genes (i.e., HaloPlex Target Enrichment
System) or for a target enrichment panel (or kit) designed specifi-
cally for inherited forms of cardiomyopathy (HaloPlex
cardiomyopathy).
3. TruSeq® Custom Amplicon Index Kit (96 Indices, 384 Sam-
ples) (Illumina). Each TruSeq® Custom Amplicon Index Kit
includes 96 unique indices for preparation of up to 384 samples.
– Standard Flow Cell, 500 cycles (Illumina).
3 Methods
3.1 DNA Extraction GenElute™ Blood Genomic DNA Kit (Sigma Aldrich) is a simple
and convenient way to isolate pure genomic DNA from fresh or
aged (older than 24 h) whole blood. It is important to carefully
follow the protocol provided by the manufacture’s company
(http://www.sigmaaldrich.com/technical-documents/protocols/
biology/genelute-blood-genomic-dna-kit.html). Briefly, the start-
ing material is lysed in a chaotropic salt-containing solution to
insure the thorough denaturation of macromolecules. The addition
of ethanol causes the DNA to bind when the lysate is spun through
a silica membrane in a microcentrifuge tube. A Prewash Solution is
provided to help remove contaminants that are associated with
aged (older than 24 h) whole blood samples. After washing to
remove contaminants, the DNA is eluted in 200 μL of a Tris-
EDTA solution.
60 Cecilia Vecoli
3.2 DNA After the extraction, the DNA quantification is strongly

Quantification recommended. It is very important to quantify the starting genomic
material using a fluorescence-based quantification method, such as by
using Qubit® 2.0 Fluorometer (Life Technologies). The genomic
DNA concentration for the TruSeq Custom Amplicon kit must be
from 150 ng (minimum) to 250 ng (recommended) in 5 μL.
3.3 TruSeq Custom Once the TruSeq Custom Amplicon kit has been designed the
Amplicon Kit Company will provide all the material (including probes) for
sequencing the selected amplicons.
3.3.1 Study Design
General consideration. During the study design, we must take
into account three general considerations (valid for any
applications):
1. Read Length: Decide which read length is suitable for each
experiment before preparing the library. The amplicon length is
user-selected.
2. Coverage: The coverage is referred to as the number of
times that a nucleotide base must be sequenced. Expressed in
percent, coverage is the total number of non-overlapping bases
covered by the attempted amplicons divided by the total
number of bases in the design.
3. Pooling indexed samples (Multiplexing): DNA samples col-
lected from the different subjects can be distinguished in
NGS reads by adding unique short oligos (6–10 bps) as bar-
codes when sequencing adapters are ligated. Barcoded samples
can be pooled and sequenced together, enabling a significant
reduction in the per sample cost of each NGS machine run.
This labeling allows sharing between several samples by tagging
each sample with a unique DNA “barcode.” In this way it is
possible to use a single NGS run to sequence either a handful of
genes from many patients or many genes from fewer patients.
TruSeq Custom Amplicon has integrated sample barcodes that
enable pooling of up to 96 samples per run. However, the real
number of samples that can be pooled together per sequencing
run depends on the number of amplicons and the desired depth
of sequencing coverage.
Design Studio for selected HCM genes. Go to the http://
designstudio.illumina.com/ web page. After registration, you can
start your project design. In the Design Studio for the genes listed
in Table 1, you can insert “Amplicon length” as 250 bp, and then
directly the name of the gene (i.e., MYH7) and choose the option
“exons only” to sequence all the exonic regions. You have to repeat
this operation for each gene associated with HCM. (Check the
“estimated amplicon” that cannot surpass 1536).
In this case, the study screened a cumulative target (the sum

length of targeted sequence minus overlapping regions in base
pairs) will be of 199,160 bp of genomic DNA sequence per patient,
covering coding regions of 16 genes known to be associated with
HCM. In this example, the number of targets (a count of the
number of targets that are in the project. A project can contain
numerous targets, and these targets may be discontinuous) was
263. The number of attempted amplicons was 1,476 and 47 gaps
were present (with a total gap distance of 5.035). The mean cover-
age was 97 %. An important parameter for evaluating each amplicon
is the score that reflects the predicted relative performance of
amplicons given Illumina’s evaluation of the amplicon probes. A
perfect score should be higher than 90 %.
In this project there were six targets with low score. We have to
use different parameters to optimize the target design. Indeed, in
general, several factors can affect the designability of a target
including the following:
– Homologs: including homologs in the same design can lead to
probes cross-binding in similar regions. Putting a highly homol-
ogous target sequence into a separate design can frequently
improve the designability.
– GC content (>80 %) especially if the region is greater than
500 bp in length.
– Homopolymer sequences and repetitive elements. In this region
the specificity of probes decreases.
– Presence of SNPs. The probes are designed, by default, on
regions with known variants (dbSNP) that result in a conserva-
tive design. Users can improve the designability by setting the
“SNP OFF” feature.
NGS technologies have also a reduced ability to detect inser-
tions and deletions because the sequence reads generated are short
and therefore sometimes are difficult to map to the reference
sequence. In general, increasing the size of the target to design
against can rescue previously “undesignable” regions. This target
increase leaves more flexibility to fit a higher scoring amplicon over
the desired target bases. Usually, Sanger sequencing is used to
provide data for bases with insufficient coverage by NGS.
3.4 Kit Protocol Kit details seem to be perfect. (Note: library preparation and their
concentration measurement can both be automated with compatible
systems like Agilent Bravo, Hamilton Banadu, Tecan, and Apricot
Designs.) This assay takes less than 8 h total, with 2–3 h of hands‐on
time and allows processing up to ~650 kb of cumulative genomic
sequence (1,536 amplicons 425 bp each) starting from between
150 and 250 ng of DNA, depending on the quality of DNA and
62 Cecilia Vecoli
number of targets in the final pool of oligonucleotide probes used

to generate amplicons in the assay.
The protocol of TruSeq Custom Amplicon kit is very detailed
and must be followed closely without rearranging any steps. Once
the protocol for library preparation is completed, to confirm that
the library has been successfully amplified, an aliquot (5 μL) of the
samples has to be run on a 4 % agarose gel. The expected PCR
product sizes for each amplicon length are indicated in the Library
preparation guide. For an amplicon length of 250 bp, we expected
to have a ~350 bp PCR product size. An accurate quantification of
library’s quality is necessary for optimal cluster density and highest
sequencing performance. Then the TruSeq Custom Amplicon pro-
tocol expects a special process to normalize the quantity of each
library to ensure more equal library representation in the pooled
sample. After that, equal volumes of normalized library (normali-
zation is a special step of the Illumina’s protocol performed
through special magnetic beads) are combined and diluted in
hybridization buffer. The diluted amplicon must be heat-denatured
and immediately loaded into the MiSeq reagent cartridge to ensure
efficient template loading on the MiSeq. Importantly, to properly
prepare the reagents, the cartridge must be thawed in room tem-
perature water for almost 1 h. Next, the cartridge must be inverted
several times to mix it thoroughly and then tapped on the bench to
dislodge any air bubbles. After these steps, the library (600 μL) can
be added to the cartridge. Before loading the cartridge you need to
clean and load the flow cell (single use). The flow cell is immersed in
storage buffer in a flow cell container. It is important to clean the
flow cell glass. Make sure that the glass is free of streaks, finger-
prints, and lint or tissue fibers. Then the run can start.
At the end of the sequencing the data can be automatically
analyzed and visualized by a special on-instrument software. Impor-
tant it is to control the quality of data as Q-Score (Phred-like quality
scores is a prediction of the probability of an error in base calling)
that in MiSeq is guaranteed >30. A Q-score ¼ 30 is equivalent to
the probability of an incorrect base call 1 in 1,000 times. The
MiSeq’s analysis software produces information such as alignment
and structural variants. Alternatively, all the data can be exported
and elaborated using different (sometimes free) software of offline
analysis. After that, the clinical/diagnostic significance of these
variants has to be determined by sorting out benign SNPs from
those that cause diseases.
Acknowledgements
This research was partially funded by a grant from Italian Ministry

of Research’s Fund for Basic Research (FIRB 2005).
References
1. Venter JC, Adams MD, Myers EW et al (2001) S, Ho CH, Irzyk GP, Jando SC, Alenquer ML,
The sequence of the human genome. Science Jarvie TP, Jirage KB, Kim JB, Knight JR, Lanza
291:1304–1351 JR, Leamon JH, Lefkowitz SM, Lei M, Li J,
2. Clark MJ, Chen R, Lam HY, Karczewski KJ, Lohman KL, Lu H, Makhijani VB, McDade
Euskirchen G, Butte AJ, Snyder M (2011) Per- KE, McKenna MP, Myers EW, Nickerson E,
formance comparison of exome DNA sequenc- Nobile JR, Plant R, Puc BP, Ronan MT, Roth
ing technologies. Nat Biotechnol 29:908–914 GT, Sarkis GJ, Simons JF, Simpson JW, Srini-
3. Zhang J, Chiodini R, Badr A, Zhang G (2011) vasan M, Tartaro KR, Tomasz A, Vogt KA,
The impact of next-generation sequencing on Volkmer GA, Wang SH, Wang Y, Weiner MP,
genomics. J Genet Genomics 38:95–109 Yu P, Begley RF, Rothberg JM (2005) Genome
sequencing in microfabricated high-density
4. Su Z, Ning B, Fang H, Hong H, Perkins R, picolitre reactors. Nature 437:376–380
Tong W, Shi L (2011) Next-generation
sequencing and its applications in molecular 11. Leamon JH (2003) A massively parallel Pico-
diagnostics. Expert Rev Mol Diagn 11:333–343 TiterPlate based platform for discrete picoliter-
scale polymerase chain reactions. Electrophore-
5. Faita F, Vecoli C, Foffa I, Andreassi MG sis 24:3769–3777
(2012) Next generation sequencing in cardio-
vascular diseases. World J Cardiol 4:288–295 12. Kim JB, Porreca GJ, Song L, Greenway SC,
Gorham JM, Church GM, Seidman CE, Seid-
6. Davey JW, Hohenlohe PA, Etter PD, Boon JQ, man JC (2007) Polony multiplex analysis of
Catchen JM, Blaxter ML (2011) Genome-wide gene expression (PMAGE) in mouse hypertro-
genetic marker discovery and genotyping using phic cardiomyopathy. Science 316:1481–1484
next-generation sequencing. Nat Rev Genet
12:499–510 13. Quail MA, Kozarewa I, Smith F, Scally A, Ste-
phens PJ, Durbin R, Swerdlow H, Turner DJ
7. Borgstrom E, Lundin S, Lundeberg J (2011) (2008) A large genome center’s improvements
Large scale library generation for high to the Illumina sequencing system. Nat Meth-
throughput sequencing. PLoS One 6:e19119 ods 5:1005–1010
8. McKernan KJ, Peckham HE, Costa GL, 14. Bentley DR, Balasubramanian S, Swerdlow HP
McLaughlin SF, Fu Y, Tsung EF, Clouser CR, et al (2008) Accurate whole human genome
Duncan C, Ichikawa JK, Lee CC, Zhang Z, sequencing using reversible terminator chemis-
Ranade SS, Dimalanta ET, Hyland FC, try. Nature 456:53–59
Sokolsky TD, Zhang L, Sheridan A, Fu H,
Hendrickson CL, Li B, Kotler L, Stuart JR, 15. Ruparel H, Bi L, Li Z, Bai X, Kim DH, Turro
Malek JA, Manning JM, Antipova AA, Perez NJ, Ju J (2005) Design and synthesis of a 30-
DS, Moore MP, Hayashibara KC, Lyons MR, O-allyl photocleavable fluorescent nucleotide
Beaudoin RE, Coleman BE, Laptewicz MW, as a reversible terminator for DNA sequencing
Sannicandro AE, Rhodes MD, Gottimukkala by synthesis. Proc Natl Acad Sci U S A
RK, Yang S, Bafna V, Bashir A, MacBride A, 102:5932–5937
Alkan C, Kidd JM, Eichler EE, Reese MG, De 16. Ruffalo M, Laframboise T, Koyuturk M (2011)
La Vega FM, Blanchard AP (2009) Sequence Comparative analysis of algorithms for next-
and structural variation in a human genome generation sequencing read alignment. Bioin-
uncovered by short-read, massively parallel formatics (Oxford, England) 27:2790–2796
ligation sequencing using two-base encoding. 17. DePristo MA, Banks E, Poplin R, Garimella
Genome Res 19:1527–1541 KV, Maguire JR, Hartl C, Philippakis AA, del
9. Wheeler DA, Srinivasan M, Egholm M, Shen Y, Angel G, Rivas MA, Hanna M, McKenna A,
Chen L, McGuire A, He W, Chen YJ, Makhi- Fennell TJ, Kernytsky AM, Sivachenko AY,
jani V, Roth GT, Gomes X, Tartaro K, Niazi F, Cibulskis K, Gabriel SB, Altshuler D, Daly MJ
Turcotte CL, Irzyk GP, Lupski JR, Chinault C, (2011) A framework for variation discovery
Song XZ, Liu Y, Yuan Y, Nazareth L, Qin X, and genotyping using next-generation DNA
Muzny DM, Margulies M, Weinstock GM, sequencing data. Nat Genet 43:491–498
Gibbs RA, Rothberg JM (2008) The complete 18. Bos JM, Towbin JA, Ackerman MJ (2009)
genome of an individual by massively parallel Diagnostic, prognostic, and therapeutic impli-
DNA sequencing. Nature 452:872–876 cations of genetic testing for hypertrophic car-
10. Margulies M, Egholm M, Altman WE, Attiya diomyopathy. J Am Coll Cardiol 54:201–211
S, Bader JS, Bemben LA, Berka J, Braverman 19. Soor GS, Luk A, Ahn E, Abraham JR, Woo A,
MS, Chen YJ, Chen Z, Dewell SB, Du L, Fierro Ralph-Edwards A, Butany J (2009) Hypertro-
JM, Gomes XV, Godwin BC, He W, Helgesen phic cardiomyopathy: current understanding
64 Cecilia Vecoli
and treatment objectives. J Clin Pathol 62: basic concepts and future molecular diagnos-
226–235 tics. Clin Biochem 42:755–765
20. Taylor MR, Carniel E, Mestroni L (2004) 22. Fokstuen S, Lyle R, Munoz A, Gehrig C, Lerch
Familial hypertrophic cardiomyopathy: clinical R, Perrot A, Osterziel KJ, Geier C, Beghetti M,
features, molecular genetics and molecular Mach F, Sztajzel J, Sigwart U, Antonarakis SE,
genetic testing. Expert Rev Mol Diagn 4: Blouin JL (2008) A DNA resequencing array
99–113 for pathogenic mutation detection in hypertro-
21. Rodriguez JE, McCudden CR, Willis MS phic cardiomyopathy. Hum Mutat 29:
(2009) Familial hypertrophic cardiomyopathy: 879–885
Chapter 5
Computational Cardiac Electrophysiology:

Implementing Mathematical Models of Cardiomyocytes
to Simulate Action Potentials of the Heart
Michael M. Bell and Elizabeth M. Cherry
Abstract
Mathematical models are now an important tool for studying cardiac electrophysiology and arrhythmias.
Utilizing these models to quantify behavior and make predictions requires solving the models computa-
tionally using numerical schemes. We discuss different solution methods and other computational con-
siderations for simulating cardiac action potentials in single cells and multicellular preparations.
Key words Cardiac action potential, Computational methods, Computer simulation, Mathematical
models, Numerical integration
1 Introduction
Since the second half of the twentieth century, mathematical mod-

els have been an important tool for studying cardiac electrophysi-
ology. The most widely used models consist of differential
equations that describe how the cell membrane potential and
other quantities change over the course of an action potential in
response to transmembrane ion currents. Most of these models are
derived from the pioneering work of Hodgkin and Huxley in
describing the squid giant axon [1]; indeed, neurons and cardio-
myocytes have many fundamental properties in common, and it is
not surprising that the same types of equations can be used to
represent both systems. The first adaptation of the Hodgkin-
Huxley model to cardiac cells was developed by Noble [2] to
describe cells of the cardiac Purkinje network (a specialized con-
duction system) and retained the same formalism. Over the years,
models were updated and expanded to incorporate new findings in
cardiac electrophysiology, such as newly discovered ion currents
and modifications of ion current responses as well as more intricate
intracellular calcium handling. Beeler and Reuter [3] published
65
66 Michael M. Bell and Elizabeth M. Cherry
one of the first models specifically representing ventricular

cardiomyocytes, and DiFrancesco and Noble [4] published the
first “second-generation” model that included differential equa-
tions describing intracellular and extracellular ion concentrations
as well as pump and exchanger currents necessary in this context to
prevent long-term ion concentration accumulation or depletion.
More recently, after being popularized by Clancy and Rudy [5],
Markov models for ion channels, which include discrete channel
states with transition probabilities, have become an important
alternative to Hodgkin-Huxley-style gating variables in models of
cellular cardiac electrophysiology. Many models describing differ-
ent species and regions of the heart have been developed and are
discussed in detail elsewhere [6, 7].
Mathematical models of cardiac electrophysiology can be valu-
able scientific tools to complement traditional experiments. For
example, models can be used to probe at high temporal and spatial
resolution states that cannot be observed easily experimentally.
Results can be reproduced exactly, which facilitates analysis of the
effects of changing a parameter. Ideally, modelers and experimen-
talists work together, with experimental data informing model
design and parameter selection as well as validation and hypotheses
generated from models being tested experimentally. However, it is
important to remember that a model must be a simplification of the
underlying system; a model that retains all the complexity of the
original system offers no advantages. As such, the usefulness of a
particular model depends quite strongly on the question being
asked, and limitations of a model should not be perceived as weak-
nesses, but rather as a candid discussion of how the assumptions
underlying the model affect its applicability to different conditions.
Utilizing a model of cardiac electrophysiology requires solving
its differential equations through numerical integration. Many
choices must be made even after selecting a mathematical model,
such as numerical integration method, temporal and (for multicel-
lular models) spatial resolutions, programming language or envi-
ronment, and program organization and structure. In large-scale
simulations [8], geometry, heterogeneity, and fiber anisotropy
become important considerations as well.
Implementation choices can affect performance quite signifi-
cantly; in some cases, careful programming choices have been
shown to yield codes that run several times faster than otherwise
[9] or even more than two more orders of magnitude faster [10].
Although advances in technology continue to increase available
computational resources, performance differences of this scale can
allow different questions to be asked and answered using numerical
approaches. At the same time, accuracy considerations are also
important, as insufficient resolution can produce spurious results
such as pinning of a reentrant wave trajectory to the underlying
grid, breakup of reentrant waves, or lack of breakup of reentrant
Computational Cardiac Electrophysiology. . . 67
waves [8, 10, 11]. Thus, it is important to know the consequences

of implementation decisions, especially for simulations involving
large multicellular systems or long times on the order of several
seconds or more. Note that we restrict our discussion below to
monodomain, as opposed to bidomain, formulations [12, 13].
2 Materials
1. Programming language or environment. Many different pro-

gramming languages can be used. Examples include C, C++,
Fortran, and Java. Python is another option, but its perfor-
mance is limited, so it is not recommended for large problems.
Other types of programming environments, such as MATLAB,
can be used as well. These environments often have built-in
useful functionality, such as native matrix operations, numerical
integration methods, and visualization support. CellML [14] is
another useful resource; it provides implementations for a large
number of models, and it can be used to generate code in a
number of languages. This flexibility, however, comes at the
price of reduced code readability.
2. Compiler (as needed). Most programming languages require a
compiler. Free compilers are available for Linux (and thus for
Macintosh) operating systems under the GNU license for many
languages, including C, C++, Fortran, and Java. Options for
free compilers are more limited for Windows. Other compilers
are available from Intel, Portland Group, and many more.
These compilers may include more advanced options.
3. Appropriate computer architecture. Most standard desktop
and laptop computers are well equipped to support implemen-
tations of most cardiac electrophysiology models in isolated
cells and in one-dimensional cables and rings. As the spatial
dimension increases, available memory (RAM) may be
exceeded, which will lead to significant computational slow-
downs. In such cases, advanced architectures, such as graphics
processing units [15] or parallel computing using multiple
processors with distributed memory, may be necessary.
3 Methods
Below, we describe how to solve basic differential equations for

cardiac electrophysiology using common numerical integration
schemes and discuss efficiency and accuracy considerations.
We focus on how numerical integration techniques can be imple-
mented directly, rather than on third-party packages.
3.1 Numerical 1. Specify initial conditions for all state variables according to
Integration of Model the model.
Components 2. Select a temporal resolution Δt (and, for multicellular prepara-
tions, a spatial resolution Δx); efficiency and accuracy consid-
erations discussed below may affect these choices.
3. For Hodgkin-Huxley-style models, integrate the gating variables
using a forward Euler (see Note 1), backward Euler (see Note 2),
or analytic (see Note 3) approach. For example, for an arbitrary
y 1 y
gating variable y that follows the differential equation dy
dt = τ y ,
where y 1 and τy are functions of the transmembrane potential V,
a solution could be obtained using the three methods as
y þβy
y iþ1 =y i þβ y 1 y i , y iþ1 = i1þβ1 , and y iþ1 =y 1 y 1 y i eβ ,
respectively, where β=Δt=τ y . For Markov models, calculate tran-
sition rates and integrate the channel state variables using a
forward Euler approach or possibly eliminate one differential
equation by substituting in the algebraic equation for that chan-
nel (see Note 4).
4. Calculate the transmembrane currents using the updated values
of the gating variables and/or Markov state variables.
5. Update intracellular and extracellular concentrations using
forward Euler for all associated differential equations.
6. Calculate the time-dependent stimulus current, if needed
(see Note 5).
7. For multicellular preparations, apply boundary conditions
(see Note 6).
8. For multicellular preparations, calculate the diffusive current
(see Note 7).
9. Update the transmembrane potential using a forward Euler
approach.
3.2 Efficiency The following practices are recommended for promoting efficiency:
Considerations
1. Calculate and store separate variables for combinations of con-
stants that occur inside the time loop (see Note 8).
2. Preserve memory contiguity by traversing higher-dimensional
arrays in order, if applicable (see Note 9).
3. Identify computationally expensive functions of one variable
(such as exponentials and raising a variable to a non-integer
power) and replace them with lookup tables (see Note 10).
4. Be aware of trade-offs between extensibility and efficiency
(see Note 11).
5. Note that specialized architectures like supercomputers, clus-
ters, or graphics processing units are useful or even necessary
for addressing large-scale, long-time, or high-resolution pro-
blems, but such a discussion is beyond the scope of this article.
3.3 Accuracy The accuracy achieved in a simulation depends on the temporal and
Considerations spatial resolutions used, the numerical method chosen, and the
model itself. The following practices are recommended for validat-
ing numerical solutions:
1. Use one-dimensional simulations to measure conduction
velocity achieved with different values of the spatial and tem-
poral resolution to verify that the velocity has converged
(see Note 12). Linear extrapolation to a time step of zero can
be used to determine the relative accuracy of velocity for a given
fixed spatial resolution [9].
2. For simulations in two or more spatial dimensions, especially
with simple finite-difference schemes for the spatial derivative,
check target or reentrant waves for indications of insufficient
resolution that appear as asymmetries in the wave shape [8].
For an isotropic simulation, consider using a spatial derivative
discretization whose error is rotationally symmetric to leading
order (see Note 13). Be aware of additional complications
that arise from anisotropy, fiber rotation, heterogeneity, curved
boundaries, and other geometric complexities; these may
benefit from the use of more sophisticated computational
methods [8].
3. Repeat results using finer temporal and spatial resolution to
verify that convergence has been reached. Unless a high degree
of accuracy is needed for a particular application, it usually is
more helpful to focus on qualitative similarity, as small quanti-
tative differences will be expected and, for studies in a chaotic
regime, will be unavoidable.
4. For further verification, benchmarks have been developed that
can be used to assess accuracy [16].
4 Notes
1. Many of the differential equations found in models of cardiac

electrophysiology are of the form dy dt = f ð y; t Þ with initial condi-
tion y ðt 0 Þ = y 0 . In general, this initial-value problem cannot be
solved in closed form, so it is necessary to use numerical tech-
niques to find an approximation to the true solution. This is
typically accomplished by discretizing the time interval of inter-
est into evenly spaced points t0, t1, t2, . . . where t iþ1 = t i þ Δt
and then finding the corresponding approximate values y0, y1,
y2, . . . where y i y ðt i Þ. One method for solving the differential
equation numerically is based on the usual definition of the
y ðtþΔt Þy ðt Þ
derivative as dy dt = limΔt!0 Δt . If instead of taking
the limit as Δt goes to 0 we fix Δt to be a small yet finite
value, we obtain the following finite-difference approximation
y ðtþΔt Þy ðt Þ
of the derivative: dy dt Δt . Evaluating this expression at
dy ðt i Þ
t=t i and noting that dt = f y i ; t i leads to the desired

approximation y iþ1 = y i þ f y i , t i Δt. Given an initial condi-
tion y 0 = f ðt 0 Þ, repeated iteration of the formula y iþ1 =
y i þ f y i , t i Δt produces the subsequent approximate
function values y1, y2, y3, . . .. This is known as the forward
Euler approximation and is an explicit method because the
value of y iþ1 can be determined using only the known values
of yi and f (yi, ti). Although the forward Euler method is
straightforward to implement and is sufficient for many equa-
tions in cardiac electrophysiology, certain equations require
such a small time step to produce accurate solutions using this
method that other numerical methods are preferred.
2. A slight modification to forward Euler leads to what is known
as the backward Euler method: y iþ1 = y i þ f y iþ1 , t iþ1 Δt.
Note that this method is an implicit method, as the value of
y iþ1 depends on the unknown value of f y iþ1 , t iþ1 . In gen-
eral, the backward Euler method is more computationally
intensive than the forward method, because an algebraic equa-
tion (typically nonlinear) must be solved at each step. However,
the backward version can produce accurate approximations to a
wider class of differential equations. For integrating gating
variables using this method, generally only the gating variable
itself is treated implicitly, with all other variables (such as the
transmembrane potential V) evaluated using the known values
at the current time step.
3. Note that a standard gating variable equation could be solved
analytically if there were no voltage dependence. Following this
approach, making the assumption that V does not change
significantly during one time interval (over Δt) gives the itera-
Δt=τ
tive solution y iþ1 = y 1 þ y i y 1 e y
, where y 1 and τy are
evaluated using the current voltage, V i =V ðt i Þ [17].
4. The systems of equations for Markov states are overdeter-
mined: in addition to having a differential equation for each
state variable, there is an algebraic constraint that the sum of
the states must be identically one. Thus, the final state may be
solved using the algebraic constraint (one minus the sum of the
values of all the other newly updated state variables) or using its
differential equation.
5. The stimulus current usually is a square pulse set to be twice the
diastolic threshold current needed to elicit an action potential.
This procedure avoids distorted action potentials that can occur
when the stimulus is just slightly above threshold. The stimulus
duration usually is fairly brief, often on the order of 1–2 ms.
In multicellular preparations, the stimulus typically is applied to
a small region in space. For models that track intracellular
concentrations of Na+, K+, and/or Cl ions along with Ca2+

ions, it is important to assign a charge carrier for the stimulus
[18]; the appropriate charge carrier often is K+ but may be a
different ion depending on the model [19].
6. Generally, for large-scale simulations, multicellular preparations
are approximated as a uniform continuum with a single diffu-
sion coefficient and then discretized for numerical solution.
Appropriate handling of the boundaries must be implemented.
Most commonly, boundaries are assumed to be no-flux or
reflecting. Periodic boundary conditions, in which one edge
joins another, also are fairly common. It is also possible to
implement multicellular models in other ways, such as with
each computational node representing a single cell connected
to neighbors by a fixed gap junction resistance or even by
discretizing cells subcellularly with separate intracellular and
intercellular resistance values [20].
7. The diffusive current is the second-order spatial derivative term
appearing in the differential equation for the transmembrane
potential V. For homogeneous and isotropic tissue, these terms
appear as ∂∂xV2 (with similar terms for any additional spatial
2
derivatives), and for a uniform spatial resolution, Δx can

∂2 V
be approximated using Taylor series as ∂x 2

1
ðΔx Þ2
ðV ðx Δx Þ 2V ðx Þ þ V ðx þ Δx ÞÞ. At the boundaries,
no-flux boundary conditions set the transmembrane potential
outside the boundary to be its reflection inside the boundary
(e.g., at x ¼ 0, ∂∂xV2 ðΔx1 Þ2 ð2V ð0Þ þ 2V ðΔx ÞÞ. If periodic
2
boundary conditions are used, the needed value is obtained

from the other side of the preparation (e.g., at x ¼ 0,
∂2 V
∂x 2
ðΔx1 Þ2 ðV ðL x Þ 2V ð0Þ þ V ðΔx ÞÞ), where the boundaries
are located at x ¼ 0 and x ¼ L x . Treatment of the diffusive
current in other situations, such as anisotropic domains with
fibers not aligned with the coordinate axes, may be found
elsewhere [8, 11].
8. For example, the ratio of the product of the universal gas
constant and temperature to Faraday’s constant is used in
calculating Nernst potentials of ions. For models where ion
concentrations vary over time (follow their own differential
equations), the Nernst potentials must be updated every time
step. As such, it is preferable to calculate and store this ratio
outside the time loop to avoid performing the same calculation
repeatedly.
9. How arrays with more than one dimension are stored in mem-
ory depends on the programming language or environment.
In many cases, multidimensional arrays are flattened in memory
to a one-dimensional array, but whether rows or columns vary
faster depends on the programming language. Thus, it is

important to nest loops appropriately. In some environments,
such as MATLAB, such consideration is not necessary when
using natively defined matrix operations.
10. Lookup tables have been an important programming consider-
ation from the earliest differential equation-based electrophys-
iology models—Hodgkin and Huxley actually used lookup
tables in implementing their model [1]. In this approach,
computationally expensive functions of one variable, such as
voltage-dependent exponentials, trigonometric functions, and
raising a variable to a non-integer power, are pre-calculated
outside the time loop and stored in arrays. When the quantity
calculated and stored in a table is needed, a simple array lookup
is performed instead either using the nearest value in the
table or linearly interpolating between the two nearest values.
In practice, a table resolution of 0.01 mV often is sufficient for
voltage-dependent tables [9]. For functions of other variables,
such as concentrations or currents, it may be necessary to
quantify the accuracy achieved by various table resolutions.
Because most cardiac electrophysiology models contain large
numbers of computationally expensive functions of one vari-
able, especially for calculating gating variable time constants
and steady-state values in Hodgkin-Huxley-type models and
transition rates in Markov-type models, the use of lookup tables
often results in significant speedup. It has been shown that
such tables can improve performance by a factor of 4–5 [9],
depending on the model.
11. Certain programming practices can improve readability and
make it easier to extend a code, but in practice may affect
efficiency. For example, use of separate functions to calculate
individual currents or other components of the model may
make the code more readable, but the overhead involved in
calling these functions may make the code somewhat less effi-
cient. Storing all state variables in one array, all parameters in
another, etc. can facilitate using different numerical integration
schemes or packages by making it easier to pass such informa-
tion as arguments to functions, but such naming and organiza-
tion conventions generally reduce readability and make it more
difficult to modify code. Third-party numerical integration
packages, such as CVODE, may make it easier to solve the
differential equations of models of this type, but because they
are designed with generality in mind, they often are not opti-
mized for solving cardiac electrophysiology models and thus
may not be particularly efficient. In many cases, these packages
allow for the use of adaptive time-stepping, where the size of
time steps used by the model is constantly evaluated and
updated. In a single cell simulation, such an approach can
improve efficiency; however, in multicellular systems, adapting

a single global time step often is not useful during simulations
of arrhythmias, because a fine time step is needed somewhere
in the computational domain nearly all the time (whenever a
wave front is present).
12. In most cases, the velocity is significantly more sensitive to the
spatial resolution than temporal resolution [8]. However, for
extremely coarse time steps, the velocity may decrease substan-
tially to the point of inaccuracy. Action potential duration
usually is not as sensitive to either spatial or temporal
resolution.
13. One such option in two dimensions is the following
nine-point discretization of ∇2 V ¼ ∂∂xV2 þ ∂∂yV2 : ∇2 V 6Δ
2 2
D
2
ð20 V ðx; y Þþ4V ðx þΔ, y Þþ4ðx Δ, y Þþ4ðx, y þΔÞþ4ðx, y ΔÞþ

V ðx þΔ, y þΔÞþV ðx þΔ, y ΔÞþV ðx Δ, y þΔÞ þ V ðx Δ, y ΔÞÞ,
where V(x, y) is the membrane potential in space, D is the
spatially invariant diffusion coefficient, and Δ is the uniform
spatial resolution.
Acknowledgment
This work was supported by NSF grant CCF-0926190.
References
1. Hodgkin AL, Huxley AF (1952) A quantitative (2011) Cardiac cell modelling: observations
description of membrane currents and its appli- from the heart of the cardiac physiome project.
cation to conduction and excitation in nerve. Prog Biophys Mol Biol 104:2–21
J Physiol 117:500–544 8. Clayton RH, Bernus O, Cherry EM, Dierckx
2. Noble D (1962) A modification of the H, Fenton FH, Mirabella L, Panfilov AV,
Hodgkin–Huxley equations applicable to Sachse FB, Seemann G, Zhang H (2011)
Purkinje fibre action and pace-maker potentials. Models of cardiac tissue electrophysiology:
J Physiol 160:317–352 progress, challenges and open questions. Prog
3. Beeler GW, Reuter H (1977) Reconstruction Biophys Mol Biol 104:22–48
of the action potential of ventricular myocar- 9. Cherry EM, Greenside HS, Henriquez CS
dial fibres. J Physiol 268:177–210 (2003) Efficient simulation of three-dimensional
4. DiFrancesco D, Noble D (1985) A model of anisotropic cardiac tissue using an adaptive mesh
cardiac electrical activity incorporating ionic refinement method. Chaos 13:853–865
pumps and concentration changes. Philos 10. Bueno-Orovio A, Cherry EM, Fenton FH
Trans R Soc Lond B Biol Sci 307:353–398 (2008) Minimal model for human ventricular
5. Clancy CE, Rudy Y (2002) Na(+) channel action potentials in tissue. J Theor Biol
mutation that causes both Brugada and 253:544–560
long-QT syndrome phenotypes: a simulation 11. Fenton FH, Cherry EM, Hastings HM, Evans
study of mechanism. Circulation SJ (2002) Multiple mechanisms of spiral wave
105:1208–1213 breakup in a model of cardiac electrical activity.
6. Fenton FH, Cherry EM (2008) Models of car- Chaos 12:852–892
diac cell. Scholarpedia 3:1868 12. Roth BJ, Wikswo JP (1986) A bidomain model
7. Fink M, Niederer SA, Cherry EM, Fenton FH, for the extracellular potential and magnetic
Koivum€aki JT, Seemann G, Thul R, Zhang H, field of cardiac tissue. IEEE Trans Biomed
Sachse FB, Beard D, Crampin EJ, Smith NP Eng 33:467–469
13. Roth BJ (2008) Bidomain model. Scholarpedia Philos Transact A Math Phys Eng Sci
3:6221 369:4331–4351
14. The CellML Project. www.cellml.org 17. Rush S, Larsen H (1978) A practical algorithm
15. Bartocci E, Cherry EM, Glimm J, Grosu R, for solving dynamic membrane equations.
Smolka SA, Fenton FH (2011) Toward IEEE Trans Biomed Eng 25:389–392
real-time simulation of cardiac dynamics. 18. Hund TJ, Kucera JP, Otani NF, Rudy Y (2001)
In: CMSB 2011: Proceedings of the 9th inter- Ionic charge conservation and long-term
national conference on computational methods steady state in the Luo-Rudy dynamic cell
in systems biology. ACM, Paris, France, model. Biophys J 81:3324–3331
pp. 103–110 19. Cherry EM, Hastings HM, Evans SJ (2008)
16. Niederer SA, Kerfoot E, Benson AP, Bernabeu Dynamics of human atrial cell models: restitu-
MO, Bernus O, Bradley C, Cherry EM, tion, memory, and intracellular calcium
Clayton R, Fenton FH, Garny A, Heidenreich dynamics in single cells. Prog Biophys Mol
E, Land S, Maleckar M, Pathmanathan P, Biol 98:24–37
Plank G, Rodrı́guez JF, Roy I, Sachse FB, 20. Henriquez CS, Papazoglou AA (1996) Using
Seemann G, Skavhaug O, Smith NP (2011) computer models to understand the roles of
Verification of cardiac tissue electrophysiology tissue structure and membrane dynamics in
simulators using an N-version benchmark. arrhythmogenesis. Proc IEEE 84:334–354
Chapter 6
Methods of Myofibrillogenesis Modeling

Nancy K. Drew and Anna Grosberg
Abstract
Organization in the heart is important on multiple length scales. Myofibrillogenesis processes control the
assembly of this multi-scale architecture. Understanding myofibrillogenesis might allow us to better control
self-assembly of cardiac tissues. One approach consists of creating phenomenological models and compar-
ing these models to in vitro data from primary myocytes. In this chapter, we present a method for building
these models to recapitulate different aspects of myofibrillogenesis. We present a specific example for a
cardiomyocyte model, but the same procedure can be used to model fibrillogenesis with other mechanisms
such as motility. In sum, the models allow for a better understanding of mechanisms behind self-assembly.
Key words Cardiomyocyte, Myofibrillogenesis, ECM, Modeling
1 Introduction
Multi-scale organization of the heart is essential for its function.

At the micrometer length scale are sarcomeres consisting of
organized, alternating parallel filaments of actin and myosin [1].
At the next length scale, multiple repeats of sarcomeres form the
internal structure of myofibrils. Myofibrils have a parallel alignment
and are organized into bundles in myocytes [1]. In isotropic tissues,
bundles of myofibrils are dispersed in various directions within
myocytes. In anisotropic tissues, myofibril bundles are aligned
[2]. Organization even extends to the extracellular matrix (ECM)
which, in vivo, is organized roughly along the myofibrils. Thus, on
a tissue scale, this leads to sarcomeres more aligned in anisotropic
tissues than in isotropic tissues [2]. In the heart, cardiac sheets at
the millimeter to centimeter length scale contain a laminar organi-
zation of myocytes, and these sheets form the muscular walls of the
heart [3, 4]. Cellular information from multiple scales and a cell’s
microenvironment help guide intracellular organization.
The intracellular architecture of a cardiomyocyte contains mul-
tiple structures that form an arrangement of myofibrils and focal
adhesions [5]. There are several integrin types in the cell, and these
75
76 Nancy K. Drew and Anna Grosberg
can be engaged (bound) or unengaged (unbound) [6]. Unbound

integrins can diffuse throughout the cell and membrane. Bound
integrins are part of the complexes that make focal adhesions. The
focal adhesions are then connected to actin filaments, which are
part of the myofibril [1, 6]. During the assembly of cardiomyocytes
in vitro or development in vivo, the structure of myofibrils matures
from pre-myofibrils, which are stress fiber-like, to nascent myofi-
brils, which have developed sarcomeres, to fully mature myofibrils
[7]. In cultured isolated myocytes, bundles of myofibrils span the
longest diagonal of the cell, and focal adhesions are at the periphery
of the cell, often concentrated at the corners [5]. The full process of
a cell building its internal structures, i.e., self-assembly, is controlled
by the internal mechanisms of the cell with outside cues providing
guidance.
Self-assembly of the intracellular structure can be influenced by
mechanotransduction, migration, and boundary conditions. How
the cell responds depends on stretching, migration cues, geometri-
cal boundaries, cell shape, and other physical or chemical cues [4].
Stretch affects actin and sarcomere orientation in engineered car-
diac tissues by aligning the cells with the direction of the stretch
[8]. A different dynamical effect can be seen in noncardiac motile
cells as they reorganize their internal structure to achieve move-
ment. The intracellular structure of motile cells includes actin stress
fibers, but not nascent myofibrils [9]. While cardiomyocytes are
natively nonmotile, we believe it is important to consider this
mechanism because of reports that induced pluripotent stem-cell-
derived cardiomyocytes retain some features of the precursor cells,
which are motile [10]. One of the static cues is cell shape that can be
controlled by extracellular matrix boundary conditions. It has been
experimentally observed that internal cellular structure organiza-
tion is a function of these geometrical cues. For example, in a square
or triangular cell, the myofibrils span the longest diagonals. While
in long rectangular cells, the fibrils are longitudinal to the cell. In a
symmetric case, a circle, the internal structure is more random [5].
Boundary conditions and motility cues can be manipulated in vitro
and are therefore an attractive variable for early fibrillogenesis
modeling.
Previously, several models have been proposed to recapitulate
cell motility. Cirit et al. made a model of leading edge dynamics
using signaling pathways and two versions of focal adhesions, new
and maturate [11]. The model does not predict the internal struc-
ture of the cell, but does show how extracellular matrix density
affects migration speed. Maree et al. created a 2D model of cell
motility examining the interactions between Rho GTPases and
phosphoinositides [12]. This model does not incorporate integrin
signaling and does not predict the internal structure of the cell; it
only provides a gradient of actin filaments. An accumulative
particle-spring model of actin-based motility validated with
Methods of Myofibrillogenesis Modeling 77
in vitro bead motility experiments was designed by Dayel et al.

[13]. Although this model lacks filament orientation and does not
predict internal structure, it shows how the shape of the cell affects
symmetry breaking. The cell shape has also been shown to greatly
affect self-assembly of the cell, which itself can be modeled.
A large variety of stationary cell myofibrillogenesis computa-
tional models have already been created. Novak et al. created an
elegant phenomenological model of the distribution of focal adhe-
sions in a cell [14]. This model predicts that the cell has a high
concentration of focal adhesions at the edge. Experimental data
from NIH3T3 fibroblasts were used to validate their model. The
model does not account for maturation of fibrils, myofibril length-
developed tension relationship, migration, substrate mechanical
properties, and others. Deshpande et al. designed a finite element
model of myofibril distribution, using the interactions between the
stress fibers, integrins, and substrate [15, 16]. Their model is able
to predict myofibril distribution for various cell shapes. Data
obtained from fibroblasts and epithelial cells was used for valida-
tion. This model cannot predict fiber distribution for migrating
cells and is computationally complex. A chemo-mechanical model
of the interactions between integrins and ligands was designed by
Paszek et al. [17]. The model predicts focal adhesions’ stresses and
strains. Unlike the previous two models, this model does not use a
specific cell type for validation but is more comprehensive about
focal adhesion mechanics. However, myofibrils and stress fibers are
not included in the Paszek et al. model. Grosberg et al. created a
phenomenological model of myofibrillogenesis based on focal
adhesions and myofibril interaction [18]. This model uses concepts
from Novak et al.’s model, but includes the components to make
the model cardiomyocyte specific. In vitro data from neonatal rat
ventricular myocytes was used to validate the model. The main
purpose of their model is to examine how boundary conditions
and symmetry breaking affect self-assembly. We will mainly focus
on the latter model in this chapter.
2 Materials
2.1 Modeling 1. Computer preferably with high RAM (over 8 GB).

2. Matlab or equivalent.
2.2 Experimental The in vitro experiments are essential to validate the models dis-
cussed in this chapter. The experiments can utilize a very wide range
of techniques and still be used for validation purposes. It is possible
to validate models based on published data alone. However, this
approach places a limit on some of the hypotheses that can be tested
with the models. We therefore provide a list of supplies needed to run
a generic set of in vitro experiments on cardiac myofibrillogenesis.

As the focus of this chapter is on modeling, we will provide only brief
lists of supplies and descriptions of experiments which will enable the
reader to evaluate if it is worth doing in their situation.
2.2.1 Cardiomyocytes 1. Neonatal Sprague Dawley rats (Charles River, Wilmington,

MA, USA).
Materials
2. Culture medium: Medium 199 (Life Technologies Corpora-
tion, Grand Island, NY, USA), HEPES buffer solution (Life
Technologies Corporation, Grand Island, NY, USA), MEM
nonessential amino acids (Life Technologies Corporation,
Grand Island, NY, USA), glucose (Sigma-Aldrich, Inc.,
St. Louis, MO, USA), L-glutamine (Life Technologies Corpo-
ration, Grand Island, NY, USA), vitamin B12/penicillin
solution (Sigma-Aldrich, Inc., St. Louis, MO, USA), FBS
(Life Technologies Corporation, Grand Island, NY, USA).
3. For isolation: Hank’s balanced salt solution (HBSS) (Life Tech-
nologies Corporation, Grand Island, NY, USA), trypsin
(Sigma-Aldrich, Inc., St. Louis, MO, USA), collagenase
(Worthington Biochemical Corporation, Lakewood, NJ,
USA).
Equipment 1. NuAire biosafety cabinet (model: NU-425-400) or equivalent.

2. Surgical instruments: iris scissors (Fisher Scientific Company
LLC, Hanover Park, IL, USA), eye dressing forceps (Fisher
Scientific Company LLC, Hanover Park, IL, USA), Stevens
tenotomy scissors (Fisher Scientific Company LLC, Hanover
Park, IL, USA), operating scissors (Biomedical Research
Instruments, Inc., Silver Spring, MD, USA), microdissecting
scissors (Biomedical Research Instruments, Inc., Silver Spring,
MD, USA), microdissecting forceps (Biomedical Research
Instruments, Inc., Silver Spring, MD, USA), or equivalent.
2.2.2 Micropatterning 1. Stamp template: mask example source (Front Range Photo-
Components Mask Co, Palmer Lake, CO, USA); silicon wafers (wafer source
example: Wafer World) spin-coated with SU-8 2002
Materials
photoresist (MicroChem Corp.) can be made using the
custom photomasks and a mask aligner (ABM Inc.) or can be
ordered from INRF UCI, Irvine, CA, USA; and polydimethyl-
siloxane (PDMS) (Sylgard 184, Dow Corning).
2. ECM: Millipore water, fibronectin (BD Biosciences).
3. Wash: phosphate-buffered saline (PBS) (Life Technologies
Corporation, Grand Island, NY, USA).
4. Blocking: Pluronic F-127 (Sigma-Aldrich, Inc., St. Louis, MO,
USA).
5. Coverslips (VWR, Radnor, PA, USA).
Equipment 1. Sonicator (Fisher Scientific Company LLC, Hanover Park, IL,

USA) or equivalent.
2. Spin coater (Specialty Coating Systems, Indianapolis, IN,
USA) or equivalent.
3. NuAire biosafety cabinet (model: NU-425-400) or equivalent.
4. Vacuum desiccator (Fisher Scientific Company LLC, Hanover
Park, IL, USA) or equivalent.
5. Ultraviolet ozone (UVO) cleaner (Jelight Company, Inc.,
Irvine, CA, USA) or equivalent.
2.2.3 Immunofluores- 1. Fixing solution: 4 % paraformaldehyde (VWR, Radnor, PA,

cent-Staining Components USA) and phosphate-buffered saline (Life Technologies Cor-
and Imaging poration, Grand Island, NY, USA) plus Triton-X 100 (Sigma-
Aldrich, Inc., St. Louis, MO, USA).
Materials
2. Wash: phosphate-buffered saline (PBS).
3. Primary stains: Alexa Fluor 488 phalloidin (actin stain) (Life
Technologies Corporation, Grand Island, NY, USA), Clone
hVIN-1 (primary vinculin stain) (Sigma-Aldrich, Inc.,
St. Louis, MO, USA), Clone EA-53 (primary alpha-actinin
stain) (Sigma-Aldrich,, Inc., St. Louis, MO, USA), DAPI
(nuclei stain) (Life Technologies Corporation, Grand Island,
NY, USA).
4. Secondary stains: Goat anti-mouse IgG Alexa Fluor 633 (sec-
ondary alpha-actinin stain) (Life Technologies Corporation,
Grand Island, NY, USA), Goat anti-mouse IgG Alexa Fluor
594 (secondary vinculin stain) (Life Technologies Corpora-
tion, Grand Island, NY, USA).
5. ProLong Gold (Life Technologies Corporation, Grand Island,
NY, USA).
Equipment 1. Leica DMI 6000B microscope or equivalent.

2. CooLSNAP HQ CCD camera or equivalent.
3. Chemical safety hood.
2.2.4 Traction Force 1. Tyrode’s solution at pH ¼ 7.4: HEPES, glucose, CaCl2,

Microscopy Components MgCl2, KCl, NaCl, NaH2PO4, deionized water.
Materials 2. Fluorescent beads (Molecular Probes, Eugene, OR, USA).
3. Coverslips (VWR, Radnor, PA, USA).
4. Polyacrylamide gels (0.1 % bis and 5 % acrylamide
(streptavidin-acrylamide), 90 μm thick) or equivalent soft
culture gel.
Equipment 1. Leica DMI 6000B microscope or equivalent.

2. Cascade 512b enhanced CCD camera or equivalent.
3 Methods
3.1 Formulation To model the interaction between myofibrils and focal adhesions,
of Model we start with a base model of focal adhesion distribution dynamics
[14]. The model is two dimensional, which is a reasonable approxi-
3.1.1 Base Model
mation for a cell that has spread to a relatively large extracellular
matrix (ECM) island. Unbound integrins are free to diffuse
throughout the model cell. Bound integrins are fixed to the ECM
as part of the focal adhesion complex. We will designate ρ∗ ðrÞ and
ρ(r) as the density at point r of unbound and bound integrins,
respectively. Their dynamics is then described by Novak et al. [14] as
∂ρ
¼ k0 þ k1 Fðr; t Þ ρ∗ k1 ρ; ð1Þ
∂t
∂ρ∗
¼ k0 þ k1 Fðr; t Þ ρ∗ þ k1 ρ þ D ρ ∇2 ρ∗ : ð2Þ
∂t
In both equations, k0 is the rate constant of adhesion assembly

without the contribution of force. The rate of adhesion assembly
driven by the force, on the focal adhesion at r, is described by k1.
The sum of k0 and k1F determines the conversion rate of unbound
integrins to bound integrins, i.e., the formation of focal adhesions.
Conversely, k1 is the rate constant of the disassembly of adhesions.
Since unbound integrins diffuse throughout the membrane,
Eq. 2 includes the change of ρ∗ ðrÞ due to diffusion, with the
coefficient Dρ.
In order to evaluate the force on the focal adhesions, Novak
et al. assume that each focal adhesion is connected to all other focal
adhesions. At first, this seems like a very unrealistic assumption.
However, because the cytoskeleton is interconnected within the
cell, this approximation allows for a good estimate of the forces
on the focal adhesion. The dynamics of stress fibers’ force at the
adhesion site are thus written as
ð 0 r0 r
∂Fðr; t Þ 0
¼ f1 ρðr; t Þρ r ; t 0 d2 r f 1 Fðr; t Þ:
ð3aÞ
∂t Ω r r
The first term describes the change in force due to the assembly of
myofibrils that connect to point r, while the second term describes
the change in force due to the disassembly of the fibril. In Eq. 3a f1
is the product of the rate of bundle formation and the force pro-
duced by filaments of actomyosin. The density of bound integrins
at point r and r0 at time point t is ρ(r, t) and ρ(r0, t), respectively.
The fraction in this equation is the unit vector between r and r0.
The integration is done over the area of the cell, Ω. The rate
constant f 1 describes bundle disassembly. Based on biological
insight, it is reasonable to assume that the initial formation of stress

fibrils is fast compared to the rate of focal adhesion assembly. This
implies that ∂F ∂t
0, and Eq. 3a can be rewritten as
ð 0 r0 r
f
Fðr; t Þ ¼ 1 ρðr; t Þρ r ; t 0 d2 r: ð3bÞ
f 1 Ω r r
While this framework predicts the focal adhesion distribution at the

cell edges, it lacks mechanisms essential to myofibrillogenesis in
cardiomyocytes. The elegance of this model is in its flexibility, and
we can adjust it for more functionality.
3.1.2 Cardiac To build a more comprehensive model applicable to cardiomyocyte

Myofibrillogenesis-Specific myofibrillogenesis, it is necessary to include mechanisms that
Model describe among other things the maturation of myofibrils from
pre-myofibrils to nascent myofibrils, a varying myofibril length-
tension relationship, and active myofibril parallel coupling. We
still assume integrins can occur in two forms, unbound ρ∗ ðrÞ and
bound ρ(r). To include the maturation of myofibrils in the model
without great computational expense, we virtually label the bound
integrins as ρp(r) and ρn(r), when they are attached to the pre-
myofibrils and nascent myofibrils, respectively. This necessitates a
split of Eq. 1 into two equations where k2 is the rate constant of
maturation of myofibrils and k2 is the rate constant of nascent
myofibrils conversion back to pre-myofibrils:
∂ρ p
¼ k0 ρ∗ þ k1 Fðr; t Þρ∗ k1 ρ p k2 Fðr; t Þρ p þ k2 ρn
∂t
¼ fchange of bound integrin connected to pre‐myofibril density with timeg;
ð4Þ
∂ρn
¼ k2 Fðr; t Þρ p k2 ρn
∂t
¼ fchange of bound integrin connected to nascent myofibril density with timeg:
ð5Þ
The mechanism of myofibril active parallel coupling is unknown;

therefore, we model it in a computationally conservative, phenom-
enological way. As such, we assume that a potential well exists such
that the free integrins are more concentrated on the side of the focal
adhesion in the direction closest to an almost parallel bundle. The
potential well will affect the diffusion term, introducing a bias in the
direction of interest. Incorporating this into Eq. 2, we get
∂ρ∗ h i
¼ k0 þ k1 Fðr; t Þ ρ∗ þ k1 ρ p þ D ρ ∇ eU μ=D ρ ∇ ρ∗ eU μ=Dρ :
∂t
ð6Þ
Now the last term in Eq. 6 includes a biasing potential field (U),
which over time, in the course of normal integrin recycling, forces
mutual fiber alignment. In order to follow conservation of mass,

throughout the simulation the total number of integrins is kept
constant. Thus, the total amount of integrin is
ð ð
fTotal integring ¼ ρ∗ þ ρ p þ ρn d2 r ¼ ρ d2 r ¼ ρA; ð7Þ
Ω Ω
where the average integrin density, ρ, can be calculated from the

initial conditions:
ð
fTotal integrin g 1
ρ¼ ¼ ρ∗ ðt ¼ 0Þ þ ρ p ðt ¼ 0Þ
A A Ω
þ ρn ðt ¼ 0Þd2 r: ð8Þ
The equation for mass conservation then becomes

ð ð
∗ 2
ρ d r¼ ρ ρ p ρn d2 r: ð9Þ
Ω Ω
We assume that formation of bound integrins is significantly slower

than the diffusion of unbound integrins:

U μ=D ∗ U μ=D 1 ∂ρ∗ ∗

∇ e ρ
∇ ρ e ρ
>> ∗
þ k0 ρ þ k1 Fðr; t Þ ρ k1 ρ p ,
D ρ ∂t

∇ eU μ=Dρ ∇ ρ∗ eU μ=Dρ 0:
ð10Þ
Solving Eq. 10 we get
ρ∗ ðrÞ ¼ C eμU =Dρ : ð11Þ
By combing Eqs. 9 and 11, we can explicitly solve for the constant:
ð
ρ ρ p ρn d2 r
C¼ Ωð : ð12Þ
eμU =Dρ d2 r
Ω
Thus, the density of unbound integrins at time t is

ð
ρ ρ p ρn d2 r
ρ∗ ðrÞ ¼ Ω ð eτU ¼ fdensity of free integring;
eτU 2
d r
Ω
ð13Þ
where the ratio between bias diffusion and free diffusion coeffi-
cients is τ=μ=D ρ . Thus, Eqs. 4, 5, and 13 define the dynamics of
integrins and myofibrils, which are influenced by force and the
biasing potential.
3.1.3 Focal Adhesions An infinite number of fibrils cannot attach to a focal adhesion due
and Fibril Connections to space limitations, but Novak et al. [14] do not account for the
physical space limitation within the focal adhesion. We believe a
focal adhesion complex can be better modeled as a surface reaction,
meaning after saturation, the presence of additional integrins does
not affect the production of force. Both types of bound integrins
are located in the same focal adhesion complex, and thus the pre-
myofibrils and myofibrils are competing for the same space. This
can be described in Langmuir isotherm form for all bound
integrins:
ρ p ðr; t Þ þ ρn ðr; t Þ
RðrÞ ¼ :
ρ0 þ ρ p ðr; t Þ þ ρn ðr; t Þ
The constant ρ0 corresponds to the total density of bound integrins

for which half the sites are full. Equations for the fraction of force-
bearing fibrils connected to the focal adhesions of both types are
ρ p ðr; t Þ
R p ðrÞ ¼
¼ ffraction of force bearing pre‐myofibril connections at FAg;

ð14Þ
ρn ðr; t Þ
Rn ðrÞ ¼
¼ ffraction of force bearing nascent myofibril connections at FAg:

ð15Þ
The fraction of bound integrins connected to fibrils will control the

amount of force on the focal adhesion.
3.1.4 Traction Force We changed the dynamics of myofibrils by incorporating myofibril

and Potential Field maturation in Eqs. 4 and 5. Now, the model includes two versions
of myofibrils, in immature and quasi-mature states. The pre-
myofibril is the immature state of the myofibril and produces less
force. These pre-myofibrils turn into nascent myofibrils, which are
the quasi-mature state. Nascent myofibrils are more stable and
produce more force [7, 19]. Using Eqs. 14 and 15, we can rewrite
Eq. 3b as
ð 0
0 r r 2 0
FðrÞ ¼ R p ðrÞ Rp r 0 1L d r
~
Ω ðrÞ r r
ð 0
0 r r 2 0
þ f 0 Rn ðrÞ Rn r 0 1L d r ¼ fnet force at FAg:
~ ðrÞ
Ω r r
ð16aÞ
Now, Eq. 16a describes the force at each focal adhesion at point r
and the equation is dimensionalized. The force contribution from
pre-myofibrils and nascent myofibrils is in the first and second
integral of Eq. 16a, respectively. The ratio between the strength
of pre-myofibril and nascent myofibril is defined as f0. Unlike
Eq. 3b, here we define a constant, L, which controls the contribu-
tion of fibril length to the amount of produced force. Specifically, if
L = 0, the fraction is a unit vector like in Eq. 3b and there is no
length-force dependence. Conversely, if L = 1, there is a linear
dependence of force on fiber length. The force in Eq. 16a can be
used to estimate the traction stress on the substrate [20]:
T ¼ dF=dA ð16bÞ
In the model, we group together focal adhesions whose nascent

myofibrils reach the same point by adding a biasing diffusion
potential field:
ðð " 0 2 #1
0 00 ½r r ½r r
00 0
00 0
U ðrÞ ¼ Rn r Rn r 1 þ ξ 00 0 d2 r d2 r
~ ðr0 Þ
~ ðrÞΩ
Ω r r
¼ f biasing diffusion potential fieldg:

ð17Þ
The purpose of the biasing potential is to 0 biasthe concentration of

free integrins such that the fibril
00 0 bundle r r0 would
move
00 toward
0
the thicker fibril bundle r -r if the fibrils r r and r r are
sufficiently close for this effect
00 to take place. The thickness of the
0
nascent myofibril bundle in r r is proportional to the product
Rn(r0 )Rn(r00 ). The term
00 in0 the figure parenthesis is the distance
between the bundle r r and the point of interest r. To include
the proximity property, we introduce the term in square parenthe-
00 0
sis, which is one when the distance between r and bundle r r is
zero, and it approaches zero when the distance approaches infinity.
The constant ξ controls the rate of decrease of the contribution to
the potential with growing distance between fiber bundles.
To increase the flexibility of the model, we include the possibil-
ity of concave and convex cells. In Eqs. 16a and 17 when consider-
ing convex cells, the integration is done over Ω e ðrÞ=Ω. However,
integration is done for pairs of points connected by fibers when
using concave cells. We define the 2D space at each r for pairs:
0 0
8 r; r ∈ Ω2 and 8m ∈ 0; r r , we define
n o ð18Þ
e ðrÞ r0 ¼ rm r þ r, rm ∈ Ω :
Ω
m
In Subheadings 3.1.2–3.1.4 we have provided all equations
necessary to calculate the distribution of focal adhesions. The equa-
tions can be discretized and solved with a program such as MatLab
(see Note 1). On a relatively fast personal computer with sufficient

RAM, this type of simulation should be complete with a run time
on the order of about 10 min. However, these equations do not
explicitly output the location and direction of myofibrils.
3.1.5 Fibril Distribution Like Novak et al. [14], we have approximated the fibers starting
with the assumption that each bound integrin is connected to all
other integrins with the same virtual label. We then approximate
the actual fibrils by calculating the average direction. In the pre-
myofibril, nascent, and total myofibril networks at each small area,
δA in a specific direction, n^ Δθ, we calculate the total length of
the fiber passing through, where n^ is a unit vector. For the pre-
myofibril network, we obtain
ðð
p
^Þ
S NN ðr; n ðα1 þ α2 Þ R p ðr þ α1 n^ ÞR p ðr α2 n^ Þ dα1 dα2 ;
ð19Þ
for the nascent myofibril network

ðð
^ Þ ðα1 þ α2 Þ½Rn ðr þ α1 n^ ÞRn ðr α2 n^ Þdα1 dα2 ; ð20Þ
S NN ðr; n
n
and the total network is

p
S NN
all
¼ S NN þ S NN
n
: ð21Þ
In both Eqs. 19 and 20, we integrate along a fiber with respect to

the spatial step α. Then, the length of the fiber in δA is given by
ðπ
^ Þdθ ¼ fLength of fiber in δA g;
S total ðrÞ ¼ S dθ ðr; n ð22Þ
0
and the length of fibers in the cell is

ð
S cell ¼ S total ðrÞd2 r ¼ fLength of fiber in cellg: ð23Þ
Ω
To derive the fiber density in δA at point r, Eq. 22 is normalized by

Eq. 23:
S total ðrÞ
S ðrÞ ¼ ¼ fFiber density in δA g: ð24Þ
S cell
Similarly, in a small area, δA, around point r and in direction n^ , the
fiber density distribution is the fiber length in the direction of n^
derived from Eq. 22 to be
S NN ðr; n ^Þ
^Þ ¼
S ðr; n ¼ fAngular fiber distributiong : ð25Þ
S total ðrÞ
We will assume that for any point r we can approximate the fibers as
straight rods within the area δA. The orientational order parameter
(OOP) classifies the orientational order of rod distribution and is
one for perfectly aligned rods and zero for isotropic systems [21].
The main orientation of rod distribution is defined as the director.
At each point r in the cell, we determine the director of fiber
distribution and OOP. The coefficients of the Fourier series of the
^ Þ were used to calculate the OOP
distribution of fiber density S ðr; n
and director
ð ð
2 π 2 π
a ðrÞ ¼ ^ Þ cos 2θ dθ,
S ðr; n b ðrÞ ¼ ^ Þ sin 2θ dθ; ð26Þ
S ðr; n
π 0 π 0
π pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
OOP ðrÞ ¼ a2 þ b 2
2
¼ fLocal orientational order parameterg, and ð27Þ
"sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi #
1 a b 1 a
n^ 0 ðrÞ ¼ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi, pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 a2 þ b 2 b 2 2 a2 þ b 2
¼ fMain fiber directiong: ð28Þ
To draw the fiber distribution, we assume that at any r the fiber is

parallel to n^ 0 (Eq. 28). The local parallel coupling degree (ψ),
a measure of how parallel the fibers are at a point, can be defined
as the product of Eq. 27 and the normalized nascent myofibril
density:
Sn
ψ ¼ OOPn : ð29Þ
max S n
Determining the fiber distribution is considerably more computa-

tionally expensive (run times of ~4 h). It is not necessary to perform
these calculations at every time step. However, the equilibrium
fiber distribution is essential for parameter fitting and validation.
3.1.6 Parameter Fitting In the work described by Grosberg et al. [18], the experiments were
and Validation done specifically for the modeling effort. It is possible to fit para-
meters using published data. We will give a very brief overview of
the experiments performed to fit and validate this model to point
the reader in the right direction. However, since this chapter is
focused on the modeling aspect, the experiments will not be dis-
cussed in depth.
To create a template for parameter fitting, a cardiomyocyte was
cultured on a stair-shaped ECM island and stained for actin fibrils.
The stain shows shadows where the sarcomeres have coupled the
aligned actin fibrils, giving a visual measure of the location of
greater parallel coupling [18]. The model parameters were fitted
such that the distribution of fibrils at equilibrium would match that

of the in vitro experiment and checked against known biological
data. For example, previous research proposed that the formation
of focal adhesion occurs on the seconds time scale, the rate of
pre-myofibril assembly on the minutes time scale, and nascent
myofibril realignment on the hours time scale [7, 22, 23, 9].
Thus, the parameters were set to f 0 = 2; ρ0 = 0:7; k1 = 60; k1 = 1;
k2 = 1:5;k2 = 0:33;ξ=200. The parameters A = 1;k0 = 1;Δt = 0:1 are
discussed in Note 2. The triangular, square, and circular cells were
used to validate the fit of the parameters.
3.1.7 Testing Hypotheses These types of models can be used to test biological hypotheses.
For example, Grosberg et al. hypothesized that longer fibers pro-
duce more force than shorter fibers. It is possible to test for this
because Eq. 16a includes a fiber length-developed tension coeffi-
cient L. In Eq. 16a force can be made independent of fiber length
by setting L=0 and force dependent on fiber length by setting L = 1.
Running the simulation with the two parameter options will result
in two different images of fibril distributions within the cell. These
can then be compared to in vitro images to test the hypothesis.
Grosberg et al. also hypothesized that there is active parallel
coupling between neighboring myofibrils. Testing for active align-
ment is possible by changing τ, the ratio between bias diffusion and
free diffusion, which is included in Eq. 17. By setting the parameter
τ=0 in Eq. 13, active alignment can be turned off and it can be
turned on by setting τ=150. Again, the simulation results would
be compared to the in vitro images to determine the validity of the
hypothesis. Some hypotheses are easier to test by comparing T
(Eq. 16b) to results of traction force experiments.
3.1.8 Motility There are no current models that combine all the functionalities of
myofibrillogenesis that we describe in previous sections and cell
motility. It should be possible to incorporate motility into this
model by using concepts from Dayel et al. [13]. The model
would need to be updated to include the change of cell shape
during cell migration. Thus, Ω, the cell geometry would change
at every time step. Since the cell is no longer stationary, the grid
inside the cell is constantly changing. At each time step, the model
needs to be updated with the points (r) that are actually in the cell.
This would present several challenges, such as a changing grid and
competing mechanisms on similar time scales. However, incorpor-
ating these changes would allow the model to predict the internal
structure for a migrating cell.
3.2 In Vitro This chapter is focused on modeling, and the in vitro experiments
Experiments are used for model validation. The work done by Grosberg et al.
included in vitro experiments on neonatal rat ventricular myocytes
[18]. The same model was also later validated for mouse
cardiomyocytes [4]. It is also possible to use published data to

validate and fit such models. Therefore, we provide a short descrip-
tion of experiments performed for the model that was the main
focus of this chapter.
3.2.1 Cardiac Myocyte Cardiomyocytes were isolated from neonatal Sprague Dawley rats.
Culture Heart tissue obtained from the rats was placed in a trypsin solution
overnight and collagenase was used to dissociate the cells from the
tissue [18]. A pre-plating separation technique was used to obtain
purified cardiomyocytes [18].
3.2.2 Micropatterning PDMS was spin-coated onto glass coverslips and incubated for
Substrates 8–12 h. Square, triangular, stair-shaped, and circular islands of
fibronectin were micropatterned on substrates using PDMS stamps
[24, 18]. These substrates were later used for immunostaining. For
traction force microscopy, substrates were coated with soft gels,
which were also micropatterned with ECM islands [18].
3.2.3 Traction Force Images of beating cardiomyocytes above a gel of fluorescent beads
Microscopy Data were collected as described previously [18]. Consecutive images
Measurement and Analysis were taken and used to calculate the displacement of the beads
positions [25]. A method described previously was used to calculate
traction force [18].
3.2.4 Immunofluorescent Stains for actin, vinculin, or alpha-actinin were done on cardiomyo-
Staining and Imaging cytes using a procedure similar to the one described in Grosberg
et al. [18]. Primary stains Alexa Fluor 488 Phalloidin (actin stain),
Clone hVIN-1 (vinculin stain), Clone EA-53 (alpha-actinin stain),
and DAPI (nuclei stain) were used on the myocytes. The secondary
stains used on the myocytes were Goat anti-mouse IgG Alexa Fluor
633 (alpha-actinin stain) and Goat anti-mouse IgG Alexa Fluor 594
(vinculin stain). After staining the myocytes, a microscope was used
to complete imaging.
4 Notes
1. Matlab or equivalent software can be used to solve the system

of model equations if they are discretized. We need to rewrite
every equation in the model. Discretizing the equations for
density of bound integrin connected to pre-myofibrils (Eq. 4)
and nascent myofibrils (Eq. 5), we obtain
h
ρ p ðt þ Δt Þ ¼ k0 ρ∗ þ k1 Fðr; t Þρ∗ k1 ρ p
i ð30Þ
k2 Fðr; t Þρ p þ k2 ρn Δt þ ρ p ðt Þ;
h i
ρn ðt þ Δt Þ ¼ k2 Fðr; t Þρ p k2 ρn Δt þ ρn ðt Þ: ð31Þ
The rest of the equations will be discretized using summations.

For example, the density of free integrins (Eq. 13) becomes

∑ ρ ρ p ρn
8r∈Ω
ρ∗ ðrÞ ¼ eτU : ð32Þ
∑ ðeτU Þ
8r∈Ω
Thus, Eq. 16a is rewritten as

0 h 0 i
FðrÞ ¼ R p ðrÞ ∑ R p r r r δA
eðrÞ
8r0 ∈Ω
0 h 0 i
þ f 0 Rn ðrÞ ∑ Rn r r r δA ð33Þ
eðrÞ
8r0 ∈Ω
where δA is the A divided by the number of points; Eq. 33

becomes
0 r0 r
FðrÞ ¼ R p ðrÞ ∑ R p r 0 δA
0 e r r1L
8r ∈Ω ðrÞ
0 r 0 r
þ f 0 Rn ðrÞ ∑ Rn r 0 δA: ð34Þ
e ðrÞ r r1L
8r0 ∈Ω
Thus, the equation for traction stress is now TðrÞ=FðrÞ=δA.

Then, rewriting Eq. 17 in a discrete setting, we obtain
0 00
U ðrÞ ¼ ∑ ∑ Rn r Rn r
eðrÞ 8r00 ∈Ω
8r0 ∈Ω eðr0 Þ
2 ( 0 ) 31 ð35Þ
r r r00 r0 2
41 þ ξ 00
r r0
5 δA :
2
We can expand the cross product term in Eq. 35 as
( )
r r0 r00 r0 2 00
00
1 0 00 0 2 0
r r0 ¼ 00 r r r r r r
r r 0 2
!
h 00 0
0
i 2
r r rr :
ð36Þ
The equations in Subheading 3.1.5 are also rewritten using

summations and we obtain

^ Þ Δα1 Δα2 ∑ ∑ ðα1 þ α2 Þ R p ðr þ α1 n^ ÞR p ðr α2 n^ Þ ;
S NN ðr; n
8α1 8α2
ð37Þ
^ Þ;
S total ðrÞ ¼ Δθ∑ S NN ðr; n ð38Þ
8n^
S cell ¼ δA∑ S total ðrÞ; ð39Þ

8r
S total ðrÞ
S ðrÞ ¼ ; ð40Þ
S cell
^Þ
S dθ ðr; n
S ðr, n^ ðθÞÞ ¼ ; ð41Þ
S total ðrÞ
2Δθ 2Δθ
aðrÞ ¼ ^ Þ cos 2θ,
∑ S ðr; n b ðrÞ ¼ ^ Þ sin 2θ; ð42Þ
∑ S ðr; n
π 8n^ π 8n^
π pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
OOP ðrÞ ¼ a2 þ b 2 ; ð43Þ
2
"sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi #
1 a b 1 a
n^ 0 ðrÞ ¼ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi, pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð44Þ
2 2 a2 þ b b 2 2 a2 þ b 2 2
2. The equations described in this chapter were nondimensiona-

lized for convenience. This means that some parameters are
arbitrarily set to unity as they only affect the units. For example,
in the model, spatial units are controlled by the parameter A.
The area (A) of the cell was set to one; thus, the units used in
the model differ from experimental. Similarly, we set k0 =1;
defining the units of density and time. The time step, Δt, is
determined such that the code runs efficiently, and the solution
converges. During the simulations of a circular cell, the total
computational time to reach steady state was t max =500, and for
all other shapes in the model, the time was t max =120. The
model uses arbitrary time units defined by the value of k0.
References
1. Alberts B, Johnson A, Lewis J, Raff M, Roberts 3. Chen JJ, Liu W, Zhang HY, Lacy L, Yang XX,
K, Walter P (2002) Molecular biology of the Song SK, Wickline SA, Yu X (2005) Regional
cell, 4th edn. Garland, New York ventricular wall thickening reflects changes in
2. Feinberg AW, Alford PW, Jin H, Ripplinger cardiac fiber and sheet structure during con-
CM, Werdich AA, Sheehy SP, Grosberg A, traction: quantification with diffusion tensor
Parker KK (2012) Controlling the contractile MRI. Am J Physiol Heart Circ Physiol 289
strength of engineered cardiac muscle by hier- (5):H1898–H1907
archal tissue architecture. Biomaterials 4. Sheehy SP, Grosberg A, Parker KK (2012) The
33:5732–5741. doi:10.1016/j.biomaterials. contribution of cellular mechanotransduction
2012.04.043 to cardiomyocyte form and function. Biomech
Model Mechanobiol 11(8):1227–1239. cell contractility. Proc Natl Acad Sci U S A

doi:10.1007/s10237-012-0419-2 103(45):17065. doi:10.1073/pnas.06058
5. Bray MA, Sheehy SP, Parker KK (2008) Sarco- 37103
mere alignment is regulated by myocyte shape. 16. Deshpande VS, Mrksich M, McMeeking RM,
Cell Motil Cytoskeleton 65(8):641–651. Evans AG (2008) A bio-mechanical model
doi:10.1002/cm.20290 for coupling cell contractility with focal adhe-
6. Ivaska J (2012) Unanchoring integrins in focal sion formation. J Mech Phys Solids
adhesions. Nat Cell Biol 14(10):981–983 56(4):1484–1510. doi:10.1016/j.jmps.2007.
7. Dabiri GA, Turnacioglu KK, Sanger JM, San- 08.006
ger JW (1997) Myofibrillogenesis visualized in 17. Paszek MJ, Boettiger D, Weaver VM,
living embryonic cardiomyocytes. Proc Natl Hammer DA (2009) Integrin clustering is
Acad Sci U S A 94(17):9493–9498 driven by mechanical resistance from the gly-
8. McCain ML, Sheehy SP, Grosberg A, cocalyx and the substrate. PLoS Comput Biol
Goss JA, Parker KK (2013) Recapitulating 5(12):e1000604. doi:10.1371/journal.pcbi.
maladaptive, multiscale remodeling of failing 1000604
myocardium on a chip. Proc Natl Acad Sci 18. Grosberg A, Kuo PL, Guo CL, Geisse NA,
U S A 110(24):9770–9775. doi:10.1073/ Bray MA, Adams WJ, Sheehy SP, Parker KK
pnas.1304913110 (2011) Self-organization of muscle cell
9. Hotulainen P, Lappalainen P (2006) Stress structure and function. PLoS Comput Biol
fibers are generated by two distinct actin 7(2):e1001088. doi:10.1371/journal.pcbi.
assembly mechanisms in motile cells. J Cell 1001088
Biol 173(3):383–394. doi:10.1083/jcb. 19. Rhee D, Sanger JM, Sanger JW (1994) The
200511093 premyofibril—evidence for its role in myofibril-
10. Robinton DA, Daley GQ (2012) The promise logenesis. Cell Motil Cytoskeleton 28(1):1–24
of induced pluripotent stem cells in research 20. Love AEH (1927) A treatise on the mathemat-
and therapy. Nature 481(7381):295–305. ical theory of elasticity, 4th edn. Cambridge
doi:10.1038/nature10761 University Eng. Press, Cambridge
11. Cirit M, Krajcovic M, Choi CK, Welf ES, Hor- 21. Umeno A, Ueno S (2003) Quantitative analy-
witz AF, Haugh JM (2010) Stochastic model sis of adherent cell orientation influenced by
of integrin-mediated signaling and adhesion strong magnetic fields. IEEE Trans Nanobiosci
dynamics at the leading edges of migrating 2(1):26–28. doi:10.1109/tnb.2003.810157
cells. PLoS Comput Biol 6(2):e1000688. 22. Balaban NQ, Schwarz US, Riveline D, Goich-
doi:10.1371/journal.pcbi.1000688 berg P, Tzur G, Sabanay I, Mahalu D, Safran S,
12. Maree AFM, Grieneisen VA, Edelstein-Keshet Bershadsky A, Addadi L, Geiger B (2001)
L (2012) How cells integrate complex stimuli: Force and focal adhesion assembly: a close rela-
the effect of feedback from phosphoinositides tionship studied using elastic micropatterned
and cell shape on cell polarization and motility. substrates. Nat Cell Biol 3(5):466–472
PLoS Comput Biol 8(3):e1002402. doi:10. 23. McKenna NM, Wang YL (1986) Possible
1371/journal.pcbi.1002402 translocation of actin and alpha-actinin along
13. Dayel MJ, Akin O, Landeryou M, Risca V, stress fibers. Exp Cell Res 167(1):95–105
Mogilner A, Mullins RD (2009) In silico 24. Tan JL, Liu W, Nelson CM, Raghavan S,
reconstitution of actin-based symmetry break- Chen CS (2004) Simple approach to micropat-
ing and motility. PLoS Biol 7(9):e1000201. tern cells on common culture substrates
doi:10.1371/journal.pbio.1000201 by tuning substrate wettability. Tissue Eng
14. Novak IL, Slepchenko BM, Mogilner A, Loew 10(5–6):865–872
LM (2004) Cooperativity between cell con- 25. Butler JP, Tolic-Norrelykke IM, Fabry B,
tractility and adhesion. Phys Rev Lett 93(26 Fredberg JJ (2002) Traction fields, moments,
Pt 1):268109 and strain energy that cells exert on their sur-
15. Deshpande VS, McMeeking RM, Evans AG roundings. Am J Physiol Cell Physiol 282(3):
(2006) A bio-chemo-mechanical model for C595–C605
Chapter 7
Using the Mechanical Bidomain Model to Analyze

the Biomechanical Behavior of Cardiomyocytes
Bradley J. Roth
Abstract
The mechanical bidomain model provides a macroscopic description of cardiac tissue biomechanics and also
predicts the microscopic coupling between the extracellular matrix and the intracellular cytoskeleton of
cardiomyocytes. The goal of this chapter is to introduce the mechanical bidomain model, to describe the
mathematical methods required to solve the model equations, and to predict where the membrane forces
acting on integrin proteins coupling the intracellular and extracellular spaces are large.
Key words Displacement, Extracellular matrix, Integrin, Intracellular cytoskeleton, Length constant,
Mathematical modeling, Mechanical bidomain model, Membrane, Pressure, Shear modulus, Strain,
Stress
1 Introduction
Experimental techniques are essential for studying the behavior of

cardiomyocytes, but mathematical modeling also has much to con-
tribute to our understanding of these cells. Some aspects of cellular
biomechanics can be analyzed by modeling single cells, but other
features can be appreciated only by examining the mechanics of the
intact tissue. Most biomechanical models are either microscopic
models of a cell or macroscopic models of tissue that do not
separate the behavior of the intracellular and extracellular spaces.
Such models do not account for the important mechanical cou-
pling between the extracellular matrix and the cytoskeleton.
Recently, a mechanical bidomain (see Note 1) model has been
developed that provides a macroscopic representation of cardiac
tissue, but also predicts the microscopic coupling between the
extracellular matrix and the cytoskeleton [1–4]. The goals of this
chapter are to introduce the mechanical bidomain model, to
describe the mathematical methods required for solving the
model equations, and to make predictions about the mechanical
behavior of cardiac tissue that can be tested experimentally.
93
94 Bradley J. Roth
2 Methods
The mechanical bidomain model predicts the intracellular and

extracellular displacements, u and w. The central hypothesis of
the model is that differences between u and w give rise to forces
across the cell membrane (Fig. 1, see Note 2).
2.1 Deriving the 1. Four factors contribute to the intracellular stress τi: (a) an
Equations of the intracellular hydrostatic pressure p; (b) an isotropic shear stress,
Mechanical Bidomain proportional to the intracellular shear strain εi with shear mod-
Model ulus ν; (c) an anisotropic Young’s modulus along the myocardial
fibers γ; and (d) an active tension acting along the fibers,
T (see Note 3). In two dimensions (see Note 4), the intracellular
stress tensor consists of three terms: τixx, τiyy, and τixy,
τixx ¼ p þ 2νεixx þ γεixx þ T ,

ð1Þ
τiy y ¼ p þ 2νεiy y , τix y ¼ 2νεix y :
2. The extracellular stress contains two terms: (a) an extracellular

hydrostatic pressure q (see Note 5) and (b) an isotropic shear
stress, proportional to the extracellular shear strain εe, with
shear modulus μ:
Fig. 1 The extracellular matrix is coupled to the intracellular cytoskeleton

through proteins in the cell membrane, such as integrins. Differences between
the extracellular displacement w and the intracellular displacement u result in
shear forces acting on these proteins across the cell membrane. These forces
may control phenomena such as tissue remodeling, mechanotransduction, and
mechanoelectrical feedback
Mechanical Bidomain Model for Analyzing Cardiomyocytes 95
τexx ¼ q þ 2μεexx , τey y ¼ q þ 2μεey y ,

ð2Þ
τex y ¼ 2μεex y :
3. The strains (see Note 6) in the intracellular and extracellular

spaces are related to the displacements by

∂ux ∂u y 1 ∂ux ∂u y
εixx ¼ , εiy y ¼ , εix y ¼ þ ; ð3Þ
∂x ∂y 2 ∂y ∂x

∂wx ∂w y 1 ∂w x ∂w y
εexx ¼ , εey y ¼ , εex y ¼ þ : ð4Þ
∂x ∂y 2 ∂y ∂x
4. Because the intracellular and extracellular spaces are incom-

pressible, the displacements can be obtained from the intracel-
lular and extracellular stream functions (see Note 7) ϕ and η:
∂ϕ ∂ϕ ∂η ∂η
ux ¼ , uy ¼ , wx ¼ , wy ¼ : ð5Þ
∂y ∂x ∂y ∂x
5. The equations of mechanical equilibrium consist of taking the

divergence of the intracellular and extracellular stress tensors
(see Note 8) and adding a term representing the coupling of
the intracellular and extracellular spaces via a spring constant K
(see Note 9):
∂τixx ∂τix y
þ ¼ K ðux wx Þ,
∂x ∂y
ð6Þ
∂τix y ∂τiy y
þ ¼ K uy wy ;
∂x ∂y
∂τexx ∂τex y
þ ¼ K ðux w x Þ,
∂x ∂y
ð7Þ
∂τex y ∂τey y
þ ¼ K u y w y :
∂x ∂y
6. Putting Eqs. 1–5 into Eqs. 6 and 7 gives the mechanical

bidomain equations (see Note 10):
3
∂p ∂ ϕ ∂3 ϕ ∂3 ϕ ∂T
þν þ þ γ þ
∂x ∂x ∂y ∂y
2 3 ∂x ∂y ∂x
2

∂ϕ ∂η
¼K ; ð8Þ
∂y ∂y
96 Bradley J. Roth
3
∂p ∂ ϕ ∂3 ϕ ∂ϕ ∂η
ν þ ¼ K ; ð9Þ
∂y ∂x 3 ∂x∂y 2 ∂x ∂x
3
∂q ∂ η ∂3 η ∂ϕ ∂η
þμ þ ¼ K ; ð10Þ
∂x ∂x 2 ∂y ∂y 3 ∂y ∂y
3
∂q ∂ η ∂3 η ∂ϕ ∂η
μ þ ¼K : ð11Þ
∂y ∂x 3 ∂x∂y 2 ∂x ∂x
2.2 Solving the 1. The mechanical bidomain equations can be solved using either
Equations of the analytical techniques or numerical methods (see Note 11).
Mechanical Bidomain Here, an analytical solution is derived for a simple example
Model that shows many of the key features of the model (see
Note 12). Consider a long strip of cardiac tissue of width 2L.
Let the x direction be along this strip and assume the fibers lie
along the x direction, and let y be the direction perpendicular to
the fibers (Fig. 2). Assume the active tension T is zero. The
tissue is passively sheared along its surfaces: to the right over
the surface y ¼ L and to the left over the surface y ¼ L. In
particular, assume that the shear force F is applied to the
extracellular space at the tissue boundary and is only
Fig. 2 A long, two-dimensional strip of cardiac tissue. The fibers (thin horizontal
lines) lie along the x-axis. A shear force is applied to the extracellular space at
the tissue boundary y =L; the intracellular space is free at the boundary. (a)
The intracellular and extracellular displacements, u and w, as functions of y. The
value of the length constant σ is small enough that the difference between u and
w is negligible on this scale. (b) The differences between the extracellular
displacement and the intracellular displacement, u w. The length of the
arrows is magnified compared to panel (a). The difference u w is
significant only within a few length constants of the tissue surface
transmitted to the intracellular space via the intra-extracellular

coupling (see Note 13).
2. Guess (see Note 14) a solution of the form (see Note 15)
y
ϕ ¼ Ay 2 þ B cosh , p ¼ 0; ð12Þ
σ
y
η ¼ C y 2 þ D cosh , q ¼ 0; ð13Þ
σ
where A, B, C, D, and σ are constant parameters yet to be
determined.
3. Plugging Eqs. 12 and 13 into the mechanical bidomain equa-
tions (Eqs. 8–11), one gets
y y y
B B D
ν sinh ¼ K 2Ay þ sinh 2C y sinh ; ð14Þ
σ3 σ σ σ σ σ
y y y
D B D
μ 3 sinh ¼ K 2Ay þ sinh 2C y sinh : ð15Þ
σ σ σ σ σ σ
4. In order to satisfy these equations for all values of y, the para-

meters are related by (see Note 16)
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ν νμ
C ¼ A, D¼ B; and σ ¼ : ð16Þ
μ K ðν þ μÞ
5. To determine A and B, apply the boundary conditions at y=L.

The stresses τixx, τiyy, τexx, and τeyy are all identically zero. At the
surface, the boundary conditions are τix y =0 and τex y =F . These
boundary conditions give
F σ2
A¼ , B¼ F: ð17Þ
2ðν þ μÞ ðν þ μÞ cosh Lσ
6. The solution is therefore (see Note 17)

F 1 2 2 coshð y=σ Þ
ϕðx; y Þ ¼ y σ ; ð18Þ
ðν þ μÞ 2 coshðL=σ Þ

F 1 2 ν 2 coshð y=σ Þ
ηðx; y Þ ¼ y þ σ : ð19Þ
ðν þ μÞ 2 μ coshðL=σ Þ
98 Bradley J. Roth
7. The resulting displacements ux and wx are (see Note 18)

F sinhð y=σ Þ
ux ¼ y σ ; ð20Þ
ðν þ μÞ coshðL=σ Þ

F ν sinhð y=σ Þ
wx ¼ yþ σ : ð21Þ
ðν þ μÞ μ coshðL=σ Þ
Figure 2a shows the displacement caused by the first term in

Eqs. 20 and 21, common to both spaces. If σ is small, the
contribution of the second term in Fig. 2a is negligible, and
the first term gives the displacement in both spaces to a high
level of accuracy.
8. The difference ux wx is proportional to the membrane force
(see Note 19):
F sinhð y=σ Þ
ux w x ¼ σ : ð22Þ
μ coshðL=σ Þ
Figure 2b shows ux wx, which is large only within a few

length constants of the tissue surface. If tissue remodeling is
triggered by membrane forces, then the remodeling should
occur only near the surface.
9. The shear strain in the intracellular and extracellular spaces is

F coshð y=σ Þ
εix y ¼ 1 ; ð23Þ
2ðν þ μÞ coshðL=σ Þ

F ν coshð y=σ Þ
εex y ¼ 1þ : ð24Þ
2ðν þ μÞ μ coshðL=σ Þ
Except near the tissue surface, the shear strain in each space is
approximately F/2(ν + μ). At the tissue surface, the intracellu-
lar shear strain is zero and the extracellular shear strain is F/2μ.
2.3 Lessons Learned 1. The mechanical bidomain model is a macroscopic model,

from the Mechanical appropriate for representing a multicellular tissue rather than
Bidomain Model single cells, yet it accounts for the microscopic coupling
between the cytoskeleton and the extracellular matrix.
2. The mechanical bidomain model introduces a new length con-
stant, σ, that depends on the strength of the coupling between
the intracellular and extracellular spaces. If the two spaces are
tightly coupled, σ will be small.
3. The tissue displacement can generally be divided into two
parts: a “monodomain” part that is the same in the intracellular
and extracellular spaces and a “bidomain” part that is generally
smaller than the monodomain part but is different in the two

spaces.
4. The central hypothesis of the mechanical bidomain model is
that membrane proteins such as integrins—which may control
important biological effects such as mechanotransduction and
remodeling—respond to differences between the intracellular
and extracellular displacements.
5. The most important contribution of mathematical modeling to
biology is to make predictions that can be tested experimen-
tally. The mechanical bidomain model makes such predictions.
For example, if effects such as mechanotransduction are modu-
lated by shear strain, then in this example these effects are
uniformly distributed throughout the tissue. If, however,
mechanotransduction is modulated by differences between
the intracellular and extracellular displacements (the central
hypothesis of the mechanical bidomain model), then these
effects are restricted to a thin layer near the tissue surface.
3 Notes
1. The electrical bidomain model [5] has been used for decades to
simulate pacing and defibrillation of the heart. The mechanical
bidomain model has many similarities to the electrical bidomain
model.
2. Figure 1 indicates that the extracellular matrix is coupled to the
cytoskeleton across the cell membrane by membrane proteins,
such as integrins. Differences between the intracellular and
extracellular displacements, u and w, result in a shear force
acting between the intracellular and extracellular sides of the
protein. Integrins may play a key role in mechanotransduction
[6], remodeling, and mechanoelectric feedback [7].
3. This separation of the stress into several components follows
the model of cardiac tissue biomechanics derived by Ohayon
and Chadwick [8]. They assumed a hydrostatic pressure
because the tissue is primarily water, an active tension caused
by the myocardial fibers and an isotropic shear associated with
the collagen matrix. The main difference between their model
and the mechanical bidomain model is that their model was a
“monodomain,” in which all three contributions to the stress
were used to describe the overall tissue behavior. In a “bido-
main,” the active tension is assigned to the intracellular space
and the collagen shear to the extracellular space.
4. In this analysis, the model is restricted to two dimensions (x, y).
It can be generalized to three dimensions without too much
difficulty. In this derivation of the model, the fibers are assumed
to lie in the x direction.
100 Bradley J. Roth
5. One interesting feature of the bidomain model is that there are

two pressures, p in the intracellular space and q in the extracellular
space. The monodomain pressure is p + q. The physical meaning
of the difference between p and q is still being studied [1–4].
6. This representation of the strain in terms of the displacement
follows the traditional definitions used in elasticity theory [9].
The model developed here assumes small strains, so the linear
theory of elasticity is appropriate. It could be generalized to
account for large strains and a nonlinear stress-strain
relationship.
7. The divergence of the displacement in an incompressible tissue is
∂u
zero: ∂u
∂x
x
þ ∂yy = 0. The virtue of the stream function ϕ is that it is
defined such that the divergence ofthe displacement
is zero
∂u y ∂ϕ
by construction [9]: ∂ux
∂x
þ ∂y
= ∂
∂x ∂y
∂
þ ∂y ∂ϕ
∂x
= ∂2 ϕ
∂x∂y

∂ ϕ
2
∂y∂x
= 0. The scalar function ϕ specifies both components of
the vector function u and guarantees that the tissue is
incompressible.
8. In a monodomain model, the equations of mechanical
∂τ ∂τ ∂τ
equilibrium are [9] ∂τ∂xxx þ ∂yx y = 0 and ∂xx y þ ∂yy y = 0. The phys-
ical meaning of these equations is that the sum of the forces in
both the x and y directions is zero everywhere in the tissue.
9. The coupling K(u w) is the key new term that distinguishes a
bidomain model from a monodomain model. This term is the
force acting across the cell membrane, like in Fig. 1. In this
version of the mechanical bidomain model, the coupling is a
simple Hookean spring; the force equals the negative of the
spring constant times the displacement of the spring from
equilibrium.
10. The first two equations (Eqs. 8 and 9) govern the intracellular
space and the second two (Eqs. 10 and 11) the extracellular
space. Equations 8 and 10 describe forces in the x direction and
Eqs. 9 and 11 in the y direction.
11. A numerical algorithm for solving the mechanical bidomain
equations would permit the solution of many more problems
than do analytical methods, which are generally restricted to
idealized geometries. However, numerical methods to solve
Eqs. 8–11 are not yet developed. One advantage of analytical
solutions is that they often illustrate the physical behavior
better than numerical solutions [2, 4].
12. Another analytical solution can be found in [4]. However, that
solution uses polar coordinates (r, θ) rather than Cartesian
coordinates (x, y), complicating the mathematics.
13. Determining the correct boundary conditions at the surface of
a bidomain remains an open question. Punal and Roth [2]
assumed the surface of the tissue is fixed, so the displacement at

the surface is zero. Roth [4] used a free boundary, so the stress
at the surface is zero. Here, at the boundary a stress is applied to
the extracellular space, and the intracellular space is free. One
can imagine that the stress is applied by many small contacts
with the extracellular matrix, but not directly with the intracel-
lular cytoskeleton.
14. Guessing a solution with several unknown parameters (A, B, C,
D, and σ in this case) is a time-honored way to find analytical
solutions to differential equations. It requires some insight to
determine the correct guess. However, if one can find a solu-
tion that satisfies the differential equation and boundary con-
ditions, it is the correct solution. This claim assumes the
solution is unique, which is a reasonable assumption for most
physical problems, but which has not yet been proven mathe-
matically for these equations.
15. The solution contains the hyperbolic cosine function,
cosh(y/σ). If σ is small (σ L), then near the tissue surface
y ¼ L the hyperbolic cosine function behaves like an exponen-
tial, ey/σ . The derivative
of the hyperbolic
cosine is the hyper-
bolic sine: dyd
cosh σy = 1σ sinh σy .
16. One of the most interesting implications of the bidomain
model is the introduction of a new length constant [1–4],
qffiffiffiffiffiffiffiffiffiffiffiffi
σ= K ðνμ
νþμÞ. If the intracellular and extracellular spaces are
tightly coupled, then K will be large and σ is small. At present
the value of K is not known. Punal and Roth [2] analyzed the
mechanical bidomain equations by assuming that σ is smaller
than the size of the tissue sheet L and then using perturbation
theory to solve the equations, where the small parameter for
the perturbation analysis was σ/L.
17. The solution in Eqs. 18–19 has the form of a monodomain
term containing y2 for which the intracellular and extracellular
displacements are equal and a bidomain term containing
cosh(y/σ) for which the intracellular and extracellular displace-
ments are different. The membrane force arises from differ-
ences between the intracellular and extracellular displacements,
and therefore only the bidomain term contributes to the mem-
brane force.
18. In Eqs. 20 and 21, the monodomain terms are proportional to y,
whereas the bidomain terms are proportional to σ. If the intra-
cellular and extracellular spaces are tightly coupled, σ L, and
the bidomain term is small compared to the monodomain term.
This implies that the intracellular and extracellular displace-
ments are nearly equal. This is the displacement that would be
predicted by a monodomain model and is shown in Fig. 2a.
102 Bradley J. Roth
19. In Eqs. 20 and 21 the monodomain terms are large but are
exactly the same in the intracellular and extracellular spaces, so
they do not contribute to the difference ux wx. The mem-
brane force is therefore entirely due to the bidomain term
(Fig. 2b). The main point of the mechanical bidomain model
is to predict these small differences between the two
displacements.
References
1. Puwal S, Roth BJ (2010) Mechanical bidomain 6. Chiquet M, Gelman L, Lutz R, Maier S (2009)
model of cardiac tissue. Phys Rev E 82:041904 From mechanotransduction to extracellular
2. Punal VM, Roth BJ (2012) A perturbation solu- matrix gene expression in fibroblasts. Biochim
tion of the mechanical bidomain model. Bio- Biophys Acta 1793:911–920
mech Model Mechanobiol 11:995–1000 7. Dabiri BE, Lee H, Parker KK (2012) A
3. Roth BJ (2013) The mechanical bidomain potential role for integrin signaling in mechan-
model: a review. ISRN Tiss Eng 2013:863689 oelectrical feedback. Prog Biophys Mol Biol
4. Roth BJ (2013) Boundary layers and the distri- 110:196–203
bution of membrane forces predicted by the 8. Ohayon J, Chadwick RS (1988) Effects of colla-
mechanical bidomain model. Mech Res Com- gen microstructure on the mechanics of the left
mun 50:12–16 ventricle. Biophys J 54:1077–1088
5. Henriquez CS (1993) Simulating the electrical 9. Love AEH (1944) A treatise on the mathemati-
behavior of cardiac tissue using the bidomain cal theory of elasticity. Dover, New York
model. Crit Rev Biomed Eng 21:1–77
Chapter 8
Fabrication of a Myocardial Patch with Cells Differentiated

from Human-Induced Pluripotent Stem Cells
Lei Ye, Joydeep Basu, and Jianyi Zhang
Abstract
The incidence of cardiovascular disease represents a significant and growing health-care challenge to the
developed and developing world. The ability of native heart muscle to regenerate in response to myocardial
infarct is minimal. Tissue engineering and regenerative medicine approaches represent one promising
response to this difficulty. Here, we present methods for the construction of a cell-seeded cardiac patch
with the potential to promote regenerative outcomes in heart muscle with damage secondary to myocardial
infarct. This method leverages iPS cells and a fibrin-based scaffold to create a simple and commercially viable
tissue-engineered cardiac patch. Human-induced pluripotent stem cells (hiPSCs) can, in principle, be
differentiated into cells of any lineage. However, most of the protocols used to generate hiPSC-derived
endothelial cells (ECs) and cardiomyocytes (CMs) are unsatisfactory because the yield and phenotypic
stability of the hiPSC-ECs are low, and the hiPSC-CMs are often purified via selection for expression of a
promoter-reporter construct. In this chapter, we describe an hiPSC-EC differentiation protocol that
generates large numbers of stable ECs and an hiPSC-CM differentiation protocol that does not require
genetic manipulation, single-cell selection, or sorting with fluorescent dyes or other reagents. We also
provide a simple but effective method that can be used to combine hiPSC-ECs and hiPSC-CMs with
hiPSC-derived smooth muscle cells to engineer a contracting patch of cardiac cells.
Key words Stem cells, Tissue engineering, Heart, Myocardial infarction
1 Introduction
Damage to myocardial tissue secondary to myocardial infarct (MI)

is a typical and clinically challenging outcome associated with this
variant of ischemic heart disease. Although certain species including
neonatal mammals retain the ability to actively regenerate damaged
myocardial tissue [1], adult mammals including humans will nor-
mally respond to ischemic damage by induction of reparative
responses that results ultimately in the creation of nonfunctional,
fibrotic scar tissue. Regenerative medicine and tissue engineering
methodologies may be potentially applicable towards transitioning
the adult body’s response to ischemic damage from a principally
reparative to a regenerative outcome, associated with reconstitution
103
104 Lei Ye et al.
of structurally organized cardiac tissue and concomitant

functionality [2]. To this end, both cell-based therapies and tissue
engineering approaches are being actively explored as catalysts in
mediating the innate regenerative responses latent in adult human
heart. Such cell-based methodologies typically leverage mesenchy-
mal stem cell (MSC) populations derived from autologously
sourced adipose tissue or bone marrow that are microinjected
directly into the infarct site [3]. In addition, bioactive cell popula-
tions may be delivered as cell-based sheets, organoids, or cardio-
spheres [4, 5].
Alternatively, populations of therapeutically bioactive or regen-
erative cells may be targeted to the infarct site through application
of a biomaterial scaffold which facilitates the localized engraftment
and (where relevant) directed differentiation of the regenerative cell
population specifically at the site of injury. The presence of a bioac-
tive cell population is critical for initiation of a regenerative
response; although a variety of acellular pericardial patches based
on bovine or porcine pericardium or synthetic polymers such as
polyglycolic acid (PGA) are available, such materials serve princi-
pally to stabilize the injury site without promoting regeneration of
myocardium and associated cardiac functionality [3]. Regenerative
outcomes are specifically secondary to the presence of bioactive cell
populations; the mechanism of action (MOA) is presumed to be
principally based on the recreation of a localized regenerative
microenvironment based on paracrine signaling at the infarct site
[6].
From a commercial perspective focused on ease of manufac-
ture, quality control, and regulatory defensibility, any candidate for
a commercially viable cardiac patch (or other tissue-engineered
product prototype) is best likely to succeed if the cellular and
biomaterial components present the following characteristics:
1. Cellular: The cell represents the active biological ingredient
(ABI) and should be readily isolatable at high frequency with
readily defined characteristics and potency [7]. Genetic modifi-
cation should be avoided as its application will trigger increased
scrutiny by the regulatory agencies. Cells should be easily and
rapidly expandable without loss of defining character or
potency.
2. Biomaterial: Although biomaterials may be of synthetic or
natural origin, use of biomaterials with a proven history of
application in the human body will greatly facilitate acceptance
of the product prototype.
Human-induced pluripotent stem cells (hiPSCs) are among the
most provocative new developments in medical science because
they can be used to generate a potentially unlimited range of
patient-matched cell types for autologous regenerative therapy.
Fabrication of a Myocardial Patch with Cells Differentiated from Human-Induced. . . 105
Effective protocols for differentiating hiPSCs into smooth muscle

cells (SMCs) are well established, but previous attempts to obtain
sufficiently large and pure populations of hiPSC-derived endothe-
lial cells (ECs) and cardiomyocytes (CMs) have been only moder-
ately successful. The protocols most commonly used to generate
hiPSC-ECs are limited by low yields (~15 % or less) [8–10] and by
poor stability of the EC phenotype (~2 weeks or less) [8, 9], while
many hiPSC-CM differentiation protocols can only be performed
with genetically modified cells, because the differentiated cells are
purified by selecting for the expression of a CM-specific promoter-
reporter construct [10–13] and, consequently, are less desirable for
clinical applications. Here, we introduce an hiPSC-EC differentia-
tion protocol with markedly higher yield (~45 %) and phenotypic
stability (~4 weeks) and an hiPSC-CM differentiation protocol that
does not require reporter-gene expression, because the differen-
tiated cells are purified via microdissection and preplating. We also
describe how the hiPSC-ECs and hiPSC-CMs can be combined
with hiPSC-SMCs to form a fibrin-based patch of contracting
cardiac cells that can subsequently be used for investigations of
cell therapy. This protocol may form the basis of a future
manufacturing platform for a contractile fibrin-based cardiac
patch. All protocols were developed and tested with hiPSCs that
had been engineered from human dermal fibroblasts or blood
mononuclear cells.
2 Materials
2.1 Supplies/ 1. Dulbecco’s phosphate-buffered saline (DPBS) (catalog num-

Reagents Needed for ber: 14190-136; Life Technologies Corporation, Grand Island,
Culturing hiPSCs NY, USA).
2. Fetal bovine serum (FBS) (SH30394.03; Thermo Scientific,
West Palm Beach, FL, USA).
3. Irradiated mouse embryonic fibroblasts (MEF) (catalog num-
ber: GSC-6001G; GlobalStem, Rockville, MD, USA).
4. Dulbecco’s Modified Eagle Medium, high glucose (11965-
118; Life Technologies Corporation, Grand Island, NY, USA).
5. Sodium pyruvate 100 mM solution (11360-070; Life Technol-
ogies Corporation, Grand Island, NY, USA).
6. Dulbecco’s Modified Eagle Medium/Mixture F-12 (DMEM/
F12) (11330-032; Life Technologies Corporation, Grand
Island, NY, USA).
7. Knockout serum (10828-028; Life Technologies
Corporation).
8. Basic fibroblast growth factor (bFGF) (PHG0021; Life Tech-
nologies Corporation).
106 Lei Ye et al.
9. Penicillin-streptomycin (15140-122; Life Technologies

Corporation).
10. Nonessential amino acid solution (NEAA) (11140-050;
Gibco, Langley, OK, USA).
11. Glutamine (25030-081; Life Technologies Corporation).
12. 55 micromolar (μM) mercaptoethanol (21985-023, Gibco).
13. Collagenase type IV, powder (17104-019; Life Technologies
Corporation).
14. 6-well plate (356721; Corning Life Sciences, Tewksbury, MA,
USA).
2.2 Supplies/ 1. Dulbecco’s phosphate-buffered saline (DPBS) (14190-136;

Reagents Needed for Life Technologies Corporation).
Differentiating hiPSCs 2. Versene (15040-066; Life Technologies Corporation).
into ECs
3. 55 μmol mercaptoethanol (21985-023, Gibco).
4. Nonessential amino acid solution (NEAA) (11140-050; Gibco,
Langley, OK, USA).
5. hiPSC culture medium (85 % DMEM/F12 supplemented with
15 % knockout serum, 8 ng/mL bFGF, 0.5 penicillin-
streptomycin, 1 NEAA, 1 mM glutamine, and 55 μM
mercaptoethanol).
6. Knockout serum (10828-028; Life Technologies
Corporation).
7. mTeSR1 medium (5850; STEMCELL Technologies Inc,
Vancouver, BC, Canada).
8. Rho-associated, coiled-coil-containing protein kinase (ROCK)
inhibitor Y-27632 (688000; EMD Millipore, Billerica, MA,
USA).
9. Fibronectin (F0895; Sigma-Aldrich Corporation, St. Louis,
MO, USA).
10. Thrombin (T7009; Sigma-Aldrich Corporation).
11. Aprotinin (A3428; Sigma-Aldrich Corporation).
12. Endothelial Cell Growth Medium-2 MV (EGM2-MV)
(CC-4147; Lonza, Basel, Switzerland).
13. Activin-A (338-AC-010; R&D Systems, Minneapolis, MN,
USA).
14. BMP4 (314-BP-010; R&D Systems).
15. Vascular endothelial growth factor (VEGF), human (CYT-241;
ProSpec-Tany Technogene Ltd., East Brunswick, NJ, USA).
16. Transforming Growth Factor 1 (TGF-β1) (100-21; Pepro-
Tech, Rocky Hill, NJ, USA).
17. Erythropoietin (EPO) (PHC9431; Life Technologies

Corporation).
18. Collagenase type IV, powder.
19. Growth factor-reduced Matrigel (356231; BD, Franklin Lakes,
NJ, USA).
20. B27 supplement, complete (17504044; Life Technologies
Corporation).
21. B27 supplement minus insulin (0050129SA; Life Technologies
Corporation).
22. SB431542 (S1067; Selleckchem, Houston, TX, USA).
23. 0.05 % trypsin-EDTA (1), phenol red (25300-054; Life
Technologies Corporation, Grand Island, NY, USA).
24. Fetal bovine serum (FBS) (SH30394.03; Thermo Scientific,
West Palm Beach, FL, USA).
25. UltraV block (TA-060-UB; Thermo Scientific, West Palm
Beach, FL, USA).
26. R-Phycoerythrin (PE)-conjugated anti-CD144 antibody
(560410; BD Biosciences, USA).
27. Allophycocyanin (APC)-conjugated anti-CD31 antibody
28. PE-conjugated IgG1, κ isotype control (for CD144) (551436;
BD Biosciences, USA).
29. APC-conjugated IgG2a, κ isotype control (for CD31)
30. Propidium iodide (81845; Sigma-Aldrich Corporation).
31. NUNC™ 15 mL Graduated Centrifuge Tubes (366052;
Thermo Scientific).
32. Bright-Line™ hemacytometer (Z359629-1EA; Sigma-Aldrich
Corporation).
USA).
USA).
35. 70 μm strainer (352350; BD Biosciences, USA) flow cytometer
instrument (Aria instrument, BD Biosciences, USA).
2.3 Supplies/ 1. DPBS.

Reagents Needed for 2. Versene.
Differentiating hiPSCs
3. hiPSC culture medium.
into CMs
4. mTeSR1 medium.
5. ROCK inhibitor Y-27632.
108 Lei Ye et al.
6. Growth factor-reduced Matrigel (356231; BD).

7. RPMI1640 medium (11875-093; Life Technologies
Corporation).
8. B27 supplement minus insulin (0050129SA; Life Technologies
Corporation).
9. Activin-A.
10. BMP4.
11. VEGF, human.
12. B27 supplement, complete.
13. Collagenase type IV, powder.
14. Trypsin.
15. FBS.
16. Centrifuge tubes.
17. Bright-Line™ hemacytometer.
18. 6-well plate.
2.4 Supplies/ 1. 2 million hiPSC-CMs.

Reagents Needed for 2. 1 million hiPSC-ECs.
Manufacturing a
3. 1 million hiPSC-SMCs.
Contracting, Fibrin-
Based, Cardiac-Cell 4. Fibronectin.
Patch 5. Hydroxyethyl piperazineethanesulfonic acid (HEPES)
(15630080; Life Technologies Corporation).
6. Thrombin.
7. Calcium chloride (CaCl2) (223506; Sigma-Aldrich
Corporation).
8. RPMI1640 medium.
9. FBS.
10. Penicillin-streptomycin.
11. Hydroxyethyl piperazineethanesulfonic acid (HEPES)
(15630080; Life Technologies Corporation).
12. ε-Aminocaproic acid (EAC) (A2504; Sigma-Aldrich
Corporation).
13. 6-well plate.
14. Penicillin-streptomycin (15140-122; Life Technologies
Corporation).
15. ε-Aminocaproic acid (A2504; Sigma-Aldrich Corporation).
16. RPMI medium (11875-093; Life Technologies Corporation)
these were included.
3 Methods
3.1 hiPSC Culture Irradiated MEFs were cultured in 6-well plate with MEF growth
medium (90 % DMEM supplemented with 10 % FBS, 1 penicillin-
streptomycin, 1 NEAA, 1 mM glutamine, and 1) for 24 h.
On the day of passaging hiPSCs, MEFs are washed with DPBS
followed twice with hiPSC growth medium (85 % DMEM/F12
supplemented with 15 % knockout serum, 8 ng/mL bFGF, 0.5
penicillin-streptomycin, 1 NEAA, 1 mM glutamine, and 55 μM
mercaptoethanol) before used for hiPSC cell culture.
hiPSC would be added with 100 U/mL collagenase type IV in
DMEM/F12 for 5–7 min. Then, the collagenase medium would
be removed. hiPSC would be washed with DPBS (once) and hiPSC
medium (twice). Then, hiPSC colonies would be scratched into
small clusters and split into 6-well plates pre-cultured with irra-
diated MEF at 1:3 ratio.
3.2 Differentiation This hiPSC-EC differentiation protocol is performed with cells that
of hiPSCs into ECs have been suspended in a three-dimensional fibrin scaffold. The
hiPSC-containing scaffold is created 2 days before (i.e., on Day 2)
differentiation is induced on Day 0; then, the differentiation media
is modified over the ensuing 5 days, and the hiPSC-ECs are purified
on Day 14 (Fig. 1).
3.2.1 Day 2 1. Wash the hiPSCs with DPBS (calcium and magnesium free),
and then incubate them with 2 mL Versene for 5 min at 37 C.
2. Aspirate the Versene solution, wash the hiPSCs once with
DPBS and once with hiPSC culture medium, and then dissoci-
ate the hiPSC population into single cells by gently pipetting
them with hiPSC culture medium.
3. Add the dissociated cells to a 15 mL centrifuge tube containing
fresh hiPSC culture medium.
4. Centrifuge the cells for 5 min at 200 g (Model 5430, Eppen-
dorf, USA). Then, resuspend the cells in hiPSC culture
medium and centrifuge them for another 5 min at 200 g.
5. Resuspend the hiPSCs in mTeSR1 medium supplemented with
10 μM ROCK inhibitor, and then measure the cell density with
a hemacytometer (see Note 2).
Day-2Day 0 Day 1 Day 3 Day 5 Day 14
Scaffold Activin-A VEGF VEGF EGM-2MV Purification

Creation BMP4 TGF TGF VEGF
EPO EPO SB
Fig. 1 Chronological summary of the hiPSC-EC differentiation protocol

110 Lei Ye et al.
6. Add 1.5 106 hiPSCs to 250 μL of a 12.5 mg/mL fibronectin

solution (see Note 1).
7. Add 250 μL of 20 U/mL thrombin solution to one well of a
24-well plate.
8. Add the cell-containing fibronectin solution (prepared in
step 7) to the thrombin-loaded well (prepared in step 8); the
mixture will solidify to form a hiPSC-containing fibrin scaffold
within 1 min.
9. Transfer the scaffold to a 6-well plate, and culture the scaffold
in 2 mL of mTeSR1 medium containing 300 U/mL aprotinin
and 10 μM ROCK inhibitor (see Note 3).
3.2.2 Day 1 1. Replace the medium with 2 mL of fresh mTeSR1 medium only.
3.2.3 Day 0 1. Remove the mTeSR1 medium; wash the scaffold twice with
DPBS, and then culture the scaffold in EGM2-MV medium
containing 1 B27 minus insulin 50 ng/mL Activin-A and
25 ng/mL BMP4 for 24 h (see Note 4).
3.2.4 Day 1 Through 1. On Day 1, replace the supernatant with EGM2-MV medium
Day 3 containing B27 minus insulin, 50 ng/mL VEGF, 10 ng/mL
TGF-β1, and 100 ng/mL erythropoietin; culture the scaffold
for 2 days.
2. On Day 3, replace the supernatant with fresh EGM2-MV
medium containing the same concentrations of B27 minus
insulin, VEGF, TGF-β1, and erythropoietin; culture the scaf-
fold for 2 days.
3.2.5 Day 5 1. Digest the scaffold with collagenase type IV solution

(100 U/mL) for 5 min; then, add the digested scaffold and
cells to a 15 mL centrifuge tube and centrifuge for 5 min at
200 x g.
2. Collect the differentiated cells (i.e., the sediment layer) from
the tube, and coat a 6-well plate with growth factor-reduced
Matrigel (0.5 mg per 6-well plate); then, add the cells to the
coated well, and culture the cells in EGM2-MV medium sup-
plemented with B27, 50 ng/mL VEGF, and 10 μM SB431542
(see Note 4).
3. Replace the culture medium with fresh EGM2-MV medium
supplemented with the same concentrations of B27, VEGF,
and SB431542 every 2 days (i.e., on Day 7, Day 9, Day 11, and
Day 13).
3.2.6 Day 14 1. Wash the cells with DPBS and detach them from the plate by
(Purification) adding 0.25 % trypsin for 5 min; then, neutralize the trypsin
with FBS, and collect the cells via centrifugation at 200 g

for 5 min.
2. Resuspend the cells in DPBS, and then filter the suspension
through a 70 μm strainer.
3. Collect the filtered cells via centrifugation at 200 g for
5 min; then, incubate the cells with UltraV block for 7 min at
room temperature.
4. Collect the cells via centrifugation (200 g for 5 min), and
resuspend them in 0.5 mL DPBS containing 2 % FBS and either
(1) PE-conjugated anti-CD144 antibodies and PE- or APC-
conjugated anti-CD31 antibodies or (2) isotype-control anti-
bodies for 30 min at 4 C.
5. Wash the cells with 2 % FBS/DPBS, resuspend them in 0.3 mL
of 2 % FBS/DPBS containing 5 μL of propidium iodide
(10 μg/mL), and collect cells of adequate size and granularity
that are positive for both CD144 and CD31 expression via flow
cytometry [14].
6. After purification, the hiPSC-ECs can be maintained in the
same EC culture medium on a fibrinogen-coated surface; the
culture medium should be changed every 2 days (see Note 4).
3.3 Differentiation of Cardiomyocyte differentiation is induced after the hiPSCs have

hiPSCs into CMs grown to form a single, completely confluent layer of cells, which
typically requires ~4 days of culture. Thus, the hiPSCs are seeded
into culture plates on Day 4, and differentiation is initiated on
Day 0 (Fig. 2).
3.3.1 Day 4 1. Prepare the hiPSCs as described under Day 2, steps 1–6 of
the hiPSC-EC differentiation protocol.
2. Coat the wells of a 6-well plate with growth factor-reduced
Matrigel; then, add 1 106 of the prepared hiPSCs and
2.5 mL of mTeSR1 medium supplemented with 10 μM
ROCK inhibitor to each well (see Notes 1, 2 and 5).
3. Change the mTeSR1/ROCK inhibitor medium daily until the
cells reach 100 % confluence.
3.3.2 Day 0 1. Remove the mTeSR1 medium, wash the cells twice with
RPMI1640 medium, and then culture the cells in RPMI1640
medium supplemented with B27 minus insulin, 50 ng/mL
Activin-A, and 25 ng/mL BMP4 (see Note 4).
Day-4 Day 0 Day 1 Day 4 Day 10 ~Day 15 ~Day 22
hiPSC Activin-A VEGF RPMI+B27 Contracting cells Purification Collection of

seeding BMP4 typically appear purified cells
Fig. 2 Chronological summary of the hiPSC-CM differentiation protocol

112 Lei Ye et al.
3.3.3 Day 1 1. Replace the medium with RPMI1640 medium supplemented

with B27 minus insulin and 5–10 ng/mL VEGF, and then
culture the cells for 3 days.
3.3.4 Day 4 Through 1. Wash the cells with RPMI1640 medium, and then culture
~Day 15 them with RPMI1640 medium supplemented with B27
complete.
2. Replace the medium with fresh RPMI1640/B27 complete
medium every 3 days; contracting cells typically begin to appear
on Day 10 after differentiation is initiated.
3. Continue culturing the cells for 10–11 more days.
3.3.5 ~Day 15 1. Collect clusters of contracting cells via microdissection; then,

(Purification) wash the collected cells with HBSS and incubate them in HBSS
containing 100 U/mL collagenase type IV for 10 min at 37 C
with gentle shaking.
2. Add 2 mL 0.25 % trypsin-EDTA for 5 min, and then neutralize
the trypsin solution with 2 mL FBS.
3. Resuspend the cells in 3 mL RPMI1640/B27 complete
medium, and culture them on cell culture dishes for at least 3 h.
4. Collect the nonattached cells (or clusters of cells), and culture
them for 7 days on cell culture dishes that have been coated
with growth factor-reduced Matrigel (see Note 4).
3.4 Manufacture of a The benefit of engineered cardiac patches for the treatment of
Contracting, Fibrin- injured myocardium has been convincingly demonstrated in
Based, Cardiac-Cell small-animal studies. Here, we present a simple and effective
Patch method for combining hiPSC-ECs and hiPSC-CMs with hiPSC-
SMCs to create a patch of contracting cardiac cells that may be used
for subsequent large-animal and clinical investigations of cell ther-
apy. We suggest that the hiPSC-ECs and hiPSC-CMs be generated
via the protocols described above [15].
1. Add two million hiPSC-CMs, one million hiPSC-ECs, and
one million hiPSC-SMCs to a solution containing 0.12 mL of
25 mg/mL fibrinogen and 0.56 mL of 20 mM HEPES
[16, 17].
2. Mix the cell-containing fibrinogen solution with a solution
containing 0.017 mL thrombin (20 U/mL), 0.0013 mL
CaCl2 (2 M), and 0.3 mL RPMI1640 medium in the wells of
a 6-well plate to a final volume of 1 mL/well; the mixture will
solidify within a few minutes.
3. Add culture medium consisting of RPMI1640 medium, 10 %
FBS, 1 penicillin-streptomycin, and 2 mg/mL ε-aminocaproic
acid to the wells.
4. Change the culture medium every 2 days; isolated areas of

contracting cells will typically appear on the third day after
patch manufacture.
4 Notes
1. The efficiency of the hiPSC-EC and hiPSC-CM differentiation

protocols varies with cell density (1–1.5 million each).
2. ROCK inhibitor will be used for 24 h when hiPSCs are dis-
sociated into single cells that are prepared for differentiation.
3. Aprotinin is used in step 9 on Day 2 of the hiPSC-EC
differentiation protocol to prevent the fibrin scaffold from
degrading during differentiation.
4. B27 without insulin is used during differentiation, and B27
complete is used to maintain the hiPSC-ECs and hiPSC-CMs
after differentiation is complete.
5. When applying growth factor-reduced Matrigel to the culture
surface, the Matrigel-coated plates must be incubated for 1 h at
37 C before use.
Acknowledgments
This work was supported by US Public Health Service grants NIH

RO1 HL67828, HL95077, HL114120, and UO1HL100407.
The authors would like to thank Mr. W. Kevin Meisner for his
editorial assistance.
References
1. Gamba L, Harrison M, Lien CL (2014) Car- regeneration in the final 1-year results of the
diac regeneration in model organisms. Curr CADUCEUS trial (CArdiosphere-Derived
Treat Opt Cardiovasc Med 16:288 aUtologous stem CElls to reverse ventricUlar
2. Basu J, Ludlow JW (2010) Platform technolo- dySfunction). J Am Coll Cardiol 63:110–122
gies for tubular organ regeneration. Trends 6. Caplan AI, Correa D (2011) The MSC: an
Biotechnol 28:526–533 injury drugstore. Cell Stem Cell 9:11–15
3. Lakshmanan R, Krishnan UM, Sethuraman S 7. Guthrie K, Bruce A, Sangha N et al (2013)
(2012) Living cardiac patch, the elixir for car- Potency evaluation of tissue engineered and
diac regeneration. Exp Opin Biol Ther regenerative medicine products. Trends Bio-
12:1623–1640 technol 31:505–514
4. Freytes DO, Santambrogio L, Vunjak- 8. Choi KD, Yu J, Smuga-Otto K, Salvagiotto G,
Novakovic G (2012) Optimizing dynamic Rehrauer W, Vodyanik M, Thomson J, Slukvin
interactions between cardiac patch and inflam- I (2009) Hematopoietic and endothelial differ-
matory host cells. Cells Tissues Organs entiation of human induced pluripotent stem
195:171–182 cells. Stem Cells 27:559–567
5. Malliaras K, Makkar RR, Smith RR et al (2014) 9. Li Z, Hu S, Ghosh Z, Han Z, Wu JC (2011)
Intracoronary cardiosphere-derived cells after Functional characterization and expression
myocardial infarction: evidence of therapeutic profiling of human induced pluripotent stem
114 Lei Ye et al.
cell- and embryonic stem cell-derived endothe- JH, Conklin BR, Chen HS, Mercola M (2009)
lial cells. Stem Cells Dev 20:1701–1710 Lentiviral vectors and protocols for creation of
10. Rufaihah AJ, Huang NF, Jame S, Lee JC, stable hesc lines for fluorescent tracking and
Nguyen HN, Byers B, De A, Okogbaa J, Roll- drug resistance selection of cardiomyocytes.
ins M, Reijo-Pera R, Gambhir SS, Cooke JP PLoS One 4:e5046
(2011) Endothelial cells derived from human 14. Ye L, Haider H, Esa WB, Law PK, Zhang W, Su
ipscs increase capillary density and improve per- L, Zhang Y, Sim EK (2007) Nonviral vector-
fusion in a mouse model of peripheral arterial based gene transfection of primary human skel-
disease. Arterioscler Thromb Vasc Biol 31: etal myoblasts. Exp Biol Med (Maywood)
e72–e79 232:1477–1487
11. Anderson D, Self T, Mellor IR, Goh G, Hill SJ, 15. Hill KL, Obrtlikova P, Alvarez DF, King JA,
Denning C (2007) Transgenic enrichment of Keirstead SA, Allred JR, Kaufman DS (2010)
cardiomyocytes from human embryonic stem Human embryonic stem cell-derived vascular
cells. Mol Ther 15:2027–2036 progenitor cells capable of endothelial and
12. Huber I, Itzhaki I, Caspi O, Arbel G, Tzuker- smooth muscle cell function. Exp Hematol
man M, Gepstein A, Habib M, Yankelson L, 38:246–257
Kehat I, Gepstein L (2007) Identification and 16. Zhang G, Nakamura Y, Wang X, Hu Q, Suggs
selection of cardiomyocytes during human LJ, Zhang J (2007) Controlled release of stro-
embryonic stem cell differentiation. FASEB J mal cell-derived factor-1 alpha in situ increases
21:2551–2563 c-kit + cell homing to the infarcted heart. Tis-
13. Kita-Matsuo H, Barcova M, Prigozhina N, Sal- sue Eng 13:2063–2071
omonis N, Wei K, Jacot JG, Nelson B, Spiering 17. Zhang G, Wang X, Wang Z, Zhang J, Suggs L
S, Haverslag R, Kim C, Talantova M, Bajpai R, (2006) A pegylated fibrin patch for mesenchy-
Calzolari D, Terskikh A, McCulloch AD, Price mal stem cell delivery. Tissue Eng 12:9–19
Chapter 9
Efficient Differentiation of Cardiomyocytes from Human

Pluripotent Stem Cells with Growth Factors
Rajneesh Jha, Ren-He Xu, and Chunhui Xu
Abstract
Human pluripotent stem cells have tremendous replicative capacity and demonstrated potential to generate
functional cardiomyocytes. These cardiomyocytes represent a promising source for cell replacement therapy
to treat heart disease and may serve as a useful tool for drug discovery and disease modeling. Efficient
cardiomyocyte differentiation, a prerequisite for the application of stem cell-derived cardiomyocytes, can be
achieved with a growth factor-guided method. Undifferentiated cells are sequentially treated with activin
A and BMP4 in a serum-free and insulin-free medium and then maintained in a serum-free medium with
insulin. This method yields as much as >75 % cardiomyocytes in the differentiation culture within 2 weeks,
and the beating cardiomyocytes have expected molecular, cellular, and electrophysiological characteristics.
In this chapter, we describe in detail the differentiation protocol and follow-up characterization focusing on
immunocytochemistry, quantitative RT-PCR, and flow cytometry analysis.
Key words Cardiomyocytes, Differentiation, Flow cytometry analysis, Growth factors, Immunocyto-
chemical analysis, Pluripotent stem cells, qRT-PCR, Serum-free medium
1 Introduction
Human pluripotent stem cells (hPSCs) are a potential cell source

for tissue engineering and cellular therapy to treat heart disease, as
they have extensive proliferative capacity and can differentiate into
functional cardiomyocytes. In addition, hPSC-derived cardiomyo-
cytes can be an excellent system for evaluating cardiotoxicity in
drug discovery—the ability to generate large amounts of cardio-
myocytes with relevant physiological phenotypes offers consider-
able advantage over primary or immortalized cell models and could
translate to accurate drug evaluation in a cost-effective manner. For
these potential applications, controlled lineage-specific differentia-
tion is one of the critical steps. In earlier studies, cardiomyocytes are
generated from hPSCs by either embryoid body formation in
serum-containing medium [1–3] or coculture with mouse END2
cells [4]. In later and recent studies, more efficient cardiomyocyte
115
116 Rajneesh Jha et al.
differentiation has been achieved with growth factors, small mole-

cules, and other specific culture conditions (for reviews, see refs.
5–7). Members of the TGFβ family (e.g., activin A and/or BMP2
or BMP4) and the Wnt regulators (e.g., Wnt3a added to early stage
of differentiation and Dkk1 to late stage of differentiation) have
been found to promote cardiomyocyte differentiation from hPSCs
[8–14]. In particular, sequential treatment of hPSCs with two
growth factors actvin A and BMP4 [8] or with small molecules
targeting the Wnt pathway is sufficient to induce efficient cardio-
myocyte differentiation [15, 16]. In addition, extracellular matrix
plays an important role in improving efficiency of cardiomyocyte
differentiation [17], and glucose depletion from culture medium
containing abundant lactate can selectively enrich cardiomyocyte
populations due to differential metabolic requirements for cardio-
myocytes and noncardiac cells [18].
Cardiomyocyte differentiation from hPSCs can be assessed by
several methods. Observation of spontaneously beating cells is the
first indication of successful cardiomyocyte differentiation. For fur-
ther confirmation, examination of cardiomyocyte-associated gene
expression is essential. Immunocytochemical analysis is useful not
only to evaluate the expression of certain markers but also to obtain
information on their subcellular localization. Alternatively, marker
expression can be examined by qRT-PCR or Western blotting if
antibodies are unavailable or incompatible for immunocytochemi-
cal analysis. Flow cytometry detecting cardiac-specific markers can
be used to quantitatively analyze cardiac purity in differentiation
cultures. Other in vitro functional assays such as pharmacological
and electrophysiological analyses are also critical to confirm the
cardiac phenotype, as described elsewhere [3, 19–23].
In this chapter, we provide a detailed protocol for activin A- and
BMP4-directed cardiomyocyte differentiation from hPSCs in a
serum-free medium as previously described [8] with minor mod-
ifications. We also describe several in vitro assays for the characteri-
zation of hPSC-derived cardiomyocytes. These methods have been
developed using human embryonic stem cells (hESCs) but are also
applicable to human induced pluripotent stem (iPS) cells.
2 Materials
2.1 Growth Media 1. Knockout DMEM (Life Technologies, Catalog #10829-018)

or DMEM/F12 (Life Technologies, Catalog #11330-057).
2. Conditioned medium from mouse embryonic fibroblasts
(MEF-CM): prepare MEF-CM as previously described [24].
MEF-CM can be collected every day for 6 days, pooled, ali-
quoted, and stored at 20 C. After thaw, use within a week
when stored at 4 C (see Note 1).
Efficient Differentiation of Cardiomyocytes from Human Pluripotent Stem Cells. . . 117
3. RPMI/B27 insulin-free medium (500 ml): mix 490 ml RPMI

1640 (Life Technologies, Catalog #11875) with 10 ml B27
insulin-free supplement (Life Technologies, Catalog
#0050129SA) and pour into a 500 ml filter unit (0.22 μM,
Corning, cellulose acetate, low protein binding) and filter.
Store the medium at 4 C and use within 2 weeks. Warm the
desired aliquot of the medium at 37 C before use (see Note 2).
4. RPMI/B27 medium (500 ml): similarly to RPMI/B27
insulin-free medium, replace B27 insulin-free supplement
with B27 supplement (Life Technologies, Catalog #17504-
044) (see Note 2).
5. 10 % FBS medium: DMEM (Life Technologies, Catalog
#11965118) supplemented with 10 % FBS.
2.2 Matrigel-Coated 1. 1:4 Matrigel stock: growth factor-reduced Matrigel (Becton

Plates Dickinson, Catalog #356231) is used for coating plates. To
prepare Matrigel stock aliquots, slowly thaw Matrigel at 4 C
overnight to avoid the formation of gel. Add 10 ml of cold
knockout DMEM to the bottle containing 10 ml Matrigel, and
transfer Matrigel to a 50-ml tube. Wash the Matrigel bottle
with 20 ml cold knockout DMEM and transfer the medium to
the Matrigel tube (total volume is now 40 ml; 1:4 diluted). Mix
well and aliquot 1 or 2 ml into each prechilled tube; store at
20 C immediately.
2. Tissue culture plates and flasks: 6-well plates (Falcon, Catalog
#3046) and T75 flasks (Corning, Catalog #430641) are used
for hPSC culture. T225 flasks (Corning, Catalog #431082) are
used for MEF culture.
3. To coat plates/flasks with Matrigel, slowly thaw 1:4 Matrigel
aliquots at 4 C for at least 2 h to avoid the formation of a gel.
Alternatively, add 28 ml cold knockout DMEM to each of the
2-ml Matrigel aliquot and pipet several times until the Matrigel
dissolves into the solution. Knockout DMEM can be replaced
with DMEM/F12.
4. Dilute the thawed Matrigel aliquots 1:15 in cold knockout
DMEM (for a final dilution of 1:60).
5. Add diluted Matrigel to plates or flasks (0.5 ml/well for 24-
well plates, 1 ml/well for 6-well plates, and 7.5 ml/T75).
6. Incubate the plates or flasks at room temperature for at least
1 h. The Matrigel-coated plates or flasks can be stored in a
sealed container at 4 C for up to 1 week. Remove Matrigel
solution immediately before use.
2.3 Cardiomyocyte 1. Cells: H7 hESCs, iPS(IMR90)-1 cells (WiCell Research

Differentiation Institute).
2. Collagenase IV solution (200 units/ml): dissolve 20,000 units
of collagenase IV (Life Technologies, Catalog #17104-019) in
100 ml knockout DMEM. Add all components to a 250-ml

filter unit (0.22 μM, Corning, cellulose acetate, low protein
binding) and filter. Aliquot and store at 20 C until use.
3. Versene (EDTA) (Life Technologies, Catalog #15040-066).
4. Trypan blue (Life Technologies, Catalog #15250061).
5. Recombinant human basic fibroblast growth factor (bFGF)
(10 μg/ml): dissolve 10 μg bFGF (Life Technologies, Catalog
#13256-029) in 1 ml D-PBS with 0.2 % bovine serum albumin
(BSA, Sigma, Catalog #A2153). Filter the solution using a
0.22 μM Corning cellulose acetate, low protein-binding filter.
When handling bFGF, prewet all pipette tips, tubes, and the
filter with D-PBS + 0.2 % BSA (bFGF can bind to pipettes,
tubes, and filters, and this will prevent some loss of the factor).
Store stocks at 20 C or 80 C for long-term storage. Store
thawed aliquots at 4 C for up to 1 month.
6. D-PBS without Ca2+Mg2+ (Life Technologies, Catalog
#14190-144).
7. 0.25 % trypsin with 0.53 mM EDTA (Life Technologies, Cata-
log #25200-056). Store aliquots at 20 C.
8. Recombinant human activin A (R&D Systems, Catalog #338-
AC): reconstitute in sterile D-PBS containing 0.2 % BSA to
prepare a stock solution of 100 μg/ml. Store aliquots at
20 C.
9. Recombinant human bone morphogenetic protein-4 (BMP4,
R&D Systems, Catalog #314-BP): reconstitute in sterile 4 mM
HCl containing 0.2 % BSA to prepare a stock solution of
10 μg/ml. Store aliquots at 20 C.
10. Defined trypsin inhibitor (Cascade Biologics, Catalog
#R-007-100).
11. Incubators: all cell culture conditions are performed in humi-
dified incubators in a 5 % CO2–95 % air atmosphere at 37 C.
2.4 In Vitro 1. Paraformaldehyde (PFA): prepare fresh 2 % or 4 % PFA solution

Characterization by diluting 16 % PFA (Electron Microscopy Science, Catalog
#15710) in D-PBS. The solution can be stored at 4 C in a tube
covered with foil and used within a week. Perform under a
chemical hood when using PFA solution.
2. Ethanol, 200 proof (Sigma, Catalog # E7023).
3. Normal goat serum (NGS) (Life Technologies, Catalog
#16210): heat inactivate NGS by incubating the serum in a
56 C water bath for 30 min and gently swirl the bottle every
10 min during incubation. Store the heat-inactivated serum in
small aliquots at 20 C. Prepare a 5 % or 1 % NGS solution in
D-PBS. Store at 4 C and use within 2 weeks.
4. Primary antibody for immunocytochemical analysis: mouse

IgG1 against α-actinin (1:200, Sigma) and rabbit antibodies
against NKX2-5 (1:200, Santa Cruz Biotech). For each new lot
of primary antibody, it is highly recommended to titrate the
antibody.
5. Secondary antibodies for immunocytochemical analysis: FITC-
conjugated goat anti-mouse IgG (Sigma, Catalog #F2012),
goat anti-mouse IgG1 conjugated with Alexa 488 (Life Tech-
nologies, Catalog # A-21121), or goat anti-rabbit IgG conju-
gated with Alexa 594 (Life Technologies, Catalog #A-11012).
6. Vectashield® mounting media containing DAPI (40 ,6-diami-
dino-2-phenylindole) (Vector Laboratories, Catalog #H-
1200).
7. Qiagen RNeasy kit (Qiagen, Catalog #74104) or Aurum total
RNA mini kit (Bio-Rad, Catalog #732-6820).
8. RNaseZap (Ambion, Catalog #AM9780).
9. Nuclease-free water (Ambion, Catalog #AM9939).
10. QIAshredder column (Qiagen, Catalog #79656).
11. Benchtop centrifuge (Eppendorf centrifuge 5424).
12. NanoDrop spectrophotometer (Thermo Scientific).
13. DNase I (Ambion, Catalog # 18047-019).
14. SuperScript® VILO™ cDNA synthesis kit (Life Technologies,
Catalog #11754-250).
15. Thermal cycler (Bio-Rad, C1000 touch).
16. TaqMan gene expression assays (Applied Biosystems).
17. TaqMan master mix (Applied Biosystems, Catalog #4369016).
18. iTaq SYBR Green master mix (Bio-Rad, Catalog #172-5121).
19. Forward and reverse primers (100 μM, Integrated DNA
Technology).
20. Optical 96-well thermal cycling plates (GeneMate, Catalog #T-
3107-1).
21. Polyolefin sealing film (GeneMate, Catalog #T-2450-1).
22. 7500 or 7700 Sequence Detection System (Applied
Biosystems).
23. 10 % FBS medium or defined trypsin inhibitor (Cascade Bio-
logics, Catalog # R-007).
24. Staining buffer: D-PBS with 2 % fetal bovine serum (Life
Technologies, Catalog #10439-024).
25. Methanol (Sigma, Catalog #34860-IL-R): prechill aliquots by
storing at 20 C.
26. Blocking buffer: staining buffer supplemented with 20 % heat-

inactivated normal goat serum.
27. Primary antibodies for flow cytometry analysis: mouse IgG1
against α-actinin (Sigma, Catalog #A7811; use at 0.5 μg/
5 105 cells/100 μl), mouse IgG2b against cardiac troponin
I (cTnI) (Millipore, Catalog #MAB1691; use at 0.05 μg/
5 105 cells/100 μl). Isotype controls: mouse IgG1 (Becton
Dickinson Biosciences, Catalog #554121), mouse IgG2b (BD
Bioscience, Catalog #557351).
28. Secondary antibodies for flow cytometry analysis: Alexa 488
goat anti-mouse IgG1 (Life Technologies, Catalog #A-21121)
or Alexa 647 goat anti-mouse IgG2b (Life Technologies, Cat-
alog #A-21242).
29. Ethidium bromide monoazide (EMA, Sigma, Catalog #E2028
or Life Technologies, Catalog #E1374): prepare a stock solu-
tion as 5 mg/ml (5,000) in DMSO under a chemical hood
and store as single-use aliquots at 20 C. Minimize exposure
to light when making the stock since EMA is extremely light
sensitive.
30. FACS tubes (Becton Dickinson Biosciences, Catalog #
352052).
31. FACSCanto™ II Flow Cytometer (Becton Dickinson
Biosciences).
3 Methods
3.1 Growth Factor- Stock cultures of undifferentiated hPSCs are maintained in feeder-
Guided Cardiomyocyte free culture conditions and passaged every 5–7 days using collage-
Differentiation nase IV or Versene. Examples are given using cells maintained on
Matrigel in MEF-CM [24]. Similar method can be used for cells
3.1.1 Culture of maintained in serum-free medium supplemented with growth fac-
Undifferentiated hPSCs tors [14]. Detailed methods for culture and characterization of
undifferentiated hPSCs are described elsewhere [25]. Note that
successful cardiomyocyte differentiation is highly dependent upon
the quality of undifferentiated cells (see Note 3).
Cardiomyocyte differentiation can be achieved through
sequential treatment of activin A and BMP4 in RPMI/B27
medium [8]. As illustrated in Fig. 1a, to induce cardiomyocyte
differentiation, undifferentiated cells are first cultured on Matrigel
in MEF-CM for a few days and then treated with activin A for 1 day
followed by BMP4 for 4 days in RPMI/B27 insulin-free medium.
Insulin-free B27 is expected to improve differentiation efficiency
because insulin negatively affects cardiomyocyte differentiation
[26, 27]. Subsequently, the growth factors are removed, and the
Fig. 1 Cardiomyocyte differentiation and characterization. Differentiation procedure is shown in (a). When
undifferentiated cells become fully confluent after cultured on Matrigel in MEF-CM as shown in (b), cells were
induced to differentiate by treatment with activin A (AA) for 1 day followed by BMP4 for 4 days in RPMI/B27
insulin-free medium. Cells were maintained in RPMI/B27 for 10–15 days after the treatment of growth factors
and were harvested for in vitro characterization, such as immunocytochemical analysis, qRT-PCR, and flow
cytometry, as shown in (c), (d), and (e), respectively. The day of adding activin A is designated as day 0
cells are maintained in RPMI/27 for 10–20 days. Cells are har-
vested for in vitro characterization at the end of differentiation or
earlier during differentiation when characterizing progenitors.
3.1.2 Setup 1. After stock culture of undifferentiated cells has been main-
Differentiation Cultures tained for 4–6 days or until colonies occupy approximately
80 % of the well surface area, cells are ready to be passaged
and set up for cardiac differentiation. Example here is stock
culture maintained in 6-well plates.
2. Warm up required amount of MEF-CM for setting up differ-
entiation cultures and supplement the MEF-CM with 8 ng/ml
bFGF (see Note 4).
3. Remove medium from each well of the stock culture and rinse
cells with 2 ml D-PBS/well.
4. Aspirate D-PBS and add 2 ml Versene to each well and incubate
at 37 C for 5 min (see Note 5).
5. Remove Versene and add 1 ml MEF-CM into each well.
6. Dislodge cells by gently adding 1 ml MEF-CM to the wells and
then triturating (approximately ten times) with 1-ml pipet tip
to make it single-cell suspension (see Note 6). Transfer and pool
the cells into a 50-ml conical tube.
7. Further, take 1 ml MEF-CM to wash each well by transferring
MEF-CM from one well to another and finally pool into the
conical tube.
8. Count cells using trypan blue with a hemacytometer and make
dilution of required cells in a 50-ml tube (which works better
than a 15-ml tube for properly mixing and evenly distributing
cells into wells).
9. Seed 2 105 to 4 105 cells in 1 ml of MEF-CM for each well
of a 24-well Matrigel-coated plate. For other culture formats,
seed cells at 1 105 to 2 105 cells/cm2. Feed cells daily by
replacing MEF-CM supplemented with bFGF (8 ng/ml) until
cells compactly cover the wells (~100 % confluence) (see Note 7).
10. Passage the rest of the culture as stock culture using
200 units/ml collagenase IV or Versene. Detach the cells from
the surface using a cell scraper and triturate the cells less than ten
times using a 5-ml pipet and culture the cells on Matrigel-coated
6-well plates in MEF-CM supplemented with 8 ng/ml bFGF.
3.1.3 Growth Factor- When undifferentiated cells reach full confluence, typically 2–4 days
Induced Differentiation after the seeding, the cultures are sequentially treated with activin A
for 1 day followed by BMP4 for 4 days (see Note 8). The day when
activin A treatment starts is designated as differentiation day 0.
1. Aspirate medium and add 1 ml/well of RPMI/B27 insulin-free
medium supplemented with 100 ng/ml activin A onto a 24-
well plate. Adjust medium volume based on culture vessel
surface areas, if other culture formats are used (we have
obtained successful differentiation in 96-well plates and T75
or T225 flasks).
2. 24 h later (on differentiation day 1), aspirate the medium to
remove activin A.
3. Add RPMI/B27 insulin-free supplemented with BMP4 at
10 ng/ml (1 ml/well for a 24-well plate).
4. Cells are maintained in the BMP4-containing medium for 4

days. No medium exchanges are performed until day 5. During
this treatment, some cell death is observed, and surviving cells
will continue to differentiate and proliferate.
5. On differentiation day 5, aspirate the medium to remove
BMP4 and add RPMI/B27 (1 ml/well for a 24-well plate).
6. Change medium every other day. Beating cells usually appear
from day 9 onwards.
7. Harvest cells at days 14–20 for flow cytometry analysis to
determine the percentage of cells expressing cardiac markers,
such as α-actinin and/or cardiac troponin I, or for further
enrichment and characterization (see following section). The
cells can also be cryopreservated for later use in cell transplan-
tation and other applications. The cryopreservation procedure
is described in detail elsewhere [14].
3.1.4 Cell Harvesting 1. Feed cells a day before harvesting.

2. Remove medium and add D-PBS (0.5 ml/well of a 24-well
plate) to wash the cells.
3. Remove D-PBS, add 0.25 % trypsin/EDTA (0.5 ml/well of a
24-well plate) and incubate the cells at 37 C for 5–7 min.
4. During incubation, add 10 % FBS medium or trypsin inhibitor
to 15-ml centrifuge tubes (1.5 ml/tube for a well of cells in a
24-well plate).
5. Observe cells under a microscope. When cells are rounded up,
gently pipet up and down a few times with a 5-ml pipet to
dislodge cells from the surface.
6. Incubate partially dissociated cells at 37 C for additional
1–2 min. Skip this step if dissociation is complete.
7. Pipet up and down 5–7 times with a 5-ml pipet. Observe cells
under a microscope to ensure complete dissociation.
8. Transfer the cell suspension to a tube containing 10 % FBS
medium to neutralize trypsin. If >15 samples are processed,
add 10 % FBS medium (0.5 ml/well of a 24-well plate) to each
well before performing the dissociation and transfer. 10 % FBS
can be replaced with defined trypsin inhibitor if serum-free
medium is required for the procedure.
9. Wash the plates with 10 % FBS medium (1 ml/well of a 24-well
plate). Observe the plate under a microscope to make sure the
harvesting is complete. Mix the cell suspension well.
10. Centrifuge at 300 g for 5 min and resuspend cells in 10 %
FBS medium. Skip this step if cell counting can be done within
30 min.
11. Perform cell counting using trypan blue and aliquot cells for
flow cytometry assays.
12. For immunocytochemistry or other assays, centrifuge, resus-
pend the cells in RPMI/B27 medium, and seed the cells onto
Matrigel or gelatin-coated 96-well plates or chamber slides.
3.2 In Vitro Immunocytochemistry analysis allows detection of expression and

Characterization cellular location of proteins/antigens of interest.
3.2.1 Immuno- 1. Plate dissociated differentiation cultures in RPMI/B27 onto
cytochemistry Analysis Matrigel-coated chamber slides (see Note 9) and culture for
2–7 days. Change medium every 2–3 days.
2. Fix cells with 2 % PFA in D-PBS at room temperature for
15 min.
3. After washing with D-PBS, permeabilize cells in 100 % ethanol
for 2 min. After another wash, incubate cells in 5 % NGS in
D-PBS at room temperature for 2 h or at 4 C overnight.
4. Incubate the cells at room temperature for 2 h with primary
antibodies, e.g., antibodies against cTnI, α-actinin, or NKX2-
5, diluted appropriately in 1 % NGS in D-PBS (see Notes 10
and 11).
5. After washing, incubate the cells with corresponding secondary
antibodies diluted in D-PBS containing 1 % NGS at room
temperature for 30 min to 1 h in the dark. For single staining
of cTnI, use FITC-conjugated goat anti-mouse IgG (1:120).
For double staining of α-actinin and NKX2-5, use goat anti-
mouse IgG1 conjugated with Alexa 488 (1:1,000) together
with goat anti-rabbit IgG conjugated with Alexa 594
(1:1,000).
6. Wash the cells three times with D-PBS (5–10 min/washing)
and mount the slides with Vectashield® mounting media con-
taining DAPI for examination with a UV microscope. Images
can be merged as in Fig. 1c.
3.2.2 Quantitative RT- qRT-PCR analysis is an alternative method to determine relative

PCR Analysis gene expression, particularly when specific antibodies are lacking.
1. Remove culture medium from cells maintained in 24- or 6-well
plates and wash with PBS. Harvest the cells by adding
350–700 μl lysis buffer to each well of cell culture.
2. Isolate RNA using a Qiagen RNeasy kit or Bio-Rad Aurum
total RNA mini kit as per manufacturer’s recommendations
following the tissue isolation procedure recommended for the
QIAshredder (see Note 12).
3. Prior to RT-PCR analysis, treat RNA samples with DNase I to
remove contaminating genomic DNA.
4. Convert 1 μg RNA into cDNA in a 20-μl reaction using Super-

Script® VILO™ cDNA synthesis kit (Life Technologies). Set up
the RT reaction in Bio-Rad C1000 Touch thermal cycler as
follows—25 C for 10 min, 42 C for 120 min, and 85 C for
5 min—and then keep in 4 C.
5. Dilute cDNA 15 times by adding 280 μl of nuclease-free water
and then take 2 μl for each reaction real-time PCR run.
6. Examine relative gene expression by real-time PCR using Taq-
Man primer probe or SYBR Green reaction.
For TaqMan PCR reaction, use specific primers and probes for
cardiac markers (examples listed in Table 1). Each reaction
mixture contains 1 RT Master Mix, 1 TaqMan gene-specific
primer probe (300 nM of each primer and 80 nM of probe),
and 2 μl diluted cDNA in a final volume of 20 μl. As a control,
the samples are also subjected to the analysis of 18S ribosomal
RNA by real-time RT-PCR.
For SYBR Green reaction master mix, each reaction mixture
contains 1 iTaq SYBR Green master mix, 4 nM forward
primer, 4 nM reverse primer, and 2 μl diluted cDNA in a final
volume of 20 μl. As a control, the samples are also subjected to
the analysis of housekeeping genes such as GAPDH.
7. Set up PCR reaction master mix without cDNA, mix well, and
distribute 18 μl of the master mix to each well of a 96-well PCR
plate.
8. Add 2 μl of diluted cDNA to each well. Seal plate from all sides
with RT-grade sealing film. Centrifuge plate at 1,500 g for
2 min.
9. Perform real-time RT-PCR on the ABI PRISM 7700
Sequence Detection System using the following conditions:
denaturation and AmpliTaq Gold activation at 95 C for
10 min and amplification for 40 cycles at 95 C for 15 s and
60 C for 1 min.
10. Analyze the reactions using the software from the ABI
PRISM 7700 Sequence Detection System. The relative
quantitation of gene expression can be obtained by normali-
zation against endogenous 18S ribosomal RNA using the
ΔΔCT method described in the ABI User Bulletin, Guide
to Performing Relative Quantitation of Gene Expression
Using Real-Time Quantitative PCR, 2008 (Fig. 1d, see
Note 13).
3.2.3 Flow Cytometry Flow cytometry analysis permits quantitative analysis of purity of
Analysis cardiomyocytes in differentiation cultures. Here we provide a
procedure for intercellular staining of cardiomyocyte-associated
protein α-actinin. Other intercellular proteins can be detected
Table 1
Primers and probes for real-time RT-PCR assays
Genes Sequences
TaqMan assays
NKX2-5 Primers and probe Purchased from Applied Biosystems
(assay number Hs00231763_m1)
TNNT2 Primers and Purchased from Applied Biosystems
probe (assay number Hs00165960_m1)
18S Primers and Purchased from Applied Biosystems
probe (assay number Hs03003631_g1)
SYBR Green
reactions
NKX2-5 Forward CTGTCTTCTCCAGCTCCACC
Reverse TTCGACCTGCAGGAGAAGTT
(http://primerdepot.nci.nih.gov/)
TNNT2 Forward CTGTCTTCTCCAGCTCCACC
Reverse TTCTATCCACGTGCCTACAGC
(http://primerdepot.nci.nih.gov/)
GAPDH Forward GCGGGTCTTGGAGACTTTCT
Reverse TTCGACCTGCAGGAGAAGTT
(http://pga.mgh.harvard.edu/primerbank/)
using similar method. Typically, dissociated cells are first labeled

with EMA to allow distinguishing live and dead cells, fixed, and
permeabilized before detection of proteins/antigens with
antibodies.
1. Harvest cells as described in Subheading 3.1. After cells are
dissociated and counted, aliquot cells (0.5 106 to 1 106
trypan blue negative cells/test) into 15-ml tubes. For each
culture, prepare two tests in one 15-ml tube, which will be
separated into one test for isotype control/EMA staining and
one test for α-actinin/EMA staining after or during the block-
ing step.
2. Prepare one extra tube (0.5 106 cells/test) for unstained cell
control and two extra tubes (0.5 106 cells/test) for single-
color staining compensation controls: one for EMA only and
another for α-actinin staining only using the culture containing
high amount of beating cells.
3. Wash cells with D-PBS (5 ml/tube).
4. Prepare EMA working solution: 1 μg/ml (dilute EMA stock

solution 1:5 and add 1 μl of the diluted EMA per 1 ml staining
buffer), used within the day of staining (see Note 14).
5. Centrifuge, remove D-PBS, resuspend the cells in staining
buffer containing EMA at 1 μg/ml (0.5 ml/test), and incubate
the cells on ice in the dark for 15 min. All centrifuge steps for
flow cytometry analysis are performed at 300 g for 5 min.
6. Centrifuge, remove EMA, and resuspend the cells in 1 ml
D-PBS.
7. Place the tubes horizontally on ice and expose them to bright
light (a reading lamp with a fluorescent light bulb) at a distance
of 3–4 in. for 10 min.
8. Add D-PBS (2 ml/test), spin, remove EMA, and resuspend the
cells in 500 μl of D-PBS/sample.
9. Fix the cells by adding equal volume (0.5 ml/test) of 4 % PFA
under a chemical hood. Incubate the cells at room temperature
for 15 min in the dark.
10. Centrifuge, remove PFA into a PFA waste container, and resus-
pend the cells in 2 ml D-PBS/test to wash the cells under a
chemical hood.
11. Centrifuge and resuspend the cells in the staining buffer. The
fixed cells can be stored at 4 C overnight before continuing
the following steps.
12. Centrifuge, remove staining buffer, and resuspend the cells in
D-PBS (100 μl/sample).
13. Permeabilize the cells by adding nine volumes (900 μl/sample)
of cold methanol.
14. Mix well and incubate the cells on ice for 30 min in the dark.
15. Add 1 ml D-PBS/sample, mix, and centrifuge.
16. Remove the supernatant and resuspend the cells in blocking
buffer (100 μl/test). Incubate cells with the blocking buffer at
room temperature for 30 min.
17. Separate each sample into two FACS tubes—one for isotype
control and one for α-actinin staining.
18. Prepare the primary antibodies and corresponding isotypes in
50–100 μl blocking buffer/test.
19. After blocking, incubate the cells for 20–30 min at room
temperature with primary antibodies in the blocking solution.
To titrate a new lot of antibody, for example, use sarcomeric α-
actinin at 0.5–5 μg/test in 100 μl.
20. Wash the cells twice with staining buffer (2 ml/test).
21. Prepare the secondary antibody, for example, Alexa 488-

conjugated goat anti-mouse IgG1 (0.5 μg in 100 μl blocking
buffer/test).
22. Remove the supernatant and add 100 μl of the secondary
antibody to each tube, except for the EMA only tube, add
blocking buffer instead.
23. Incubate the cells for 15 min at room temperature in the dark.
24. Wash the cells three times with staining buffer (2 ml/test).
Resuspend the cells in 200 μl staining buffer/tube.
25. Use a FACSCanto™ II Flow Cytometer to acquire the data. Set
PMT voltage using unstained controls and compensation using
single-color staining controls. Set quadrant markers based on
isotype controls using the appropriate excitation and detection
channels (FITC and PerCP Cy5.5 for α-actinin and EMA,
respectively). Acquire at least 10,000 EMA negative events.
26. Dot plots or histograms are generated upon data analysis using
FlowJo software to display the frequency of α-actinin or isotype
positive cells versus forward scatter in differentiated culture
samples (Fig. 1e).
4 Notes
1. Materials such as medium components and growth factors

require appropriate storage. Store aliquots at 20 C or
80 C in a manual defrost freezer. Avoid repeated freeze-thaw
cycles. To prevent contamination, it is optional to supplement
with 1 % penicillin-streptomycin (Life Science Tech, Catalog #
15140-122) to media for growth of undifferentiated cells and
differentiation.
2. For small-scale experiments, prepare less amount of the differ-
entiation medium that can be used within 2 weeks. B27 supple-
ments or B27 without insulin supplements can be aliquoted
and stored at 20 C.
3. In early-passage cultures, spontaneously differentiated cells in
between colonies of undifferentiated cells appear fibroblast-
like. When the cells reach higher passages, the amount of
fibroblast-like cells typically is reduced, while the percentage
of undifferentiated cells increases in cultures [14]. To passage
stock cultures using collagenase, incubation time varies among
different passages of the cells and different batches of collage-
nase; therefore, it is advised to determine the appropriate incu-
bation time by examining the colonies. Stop incubation when
the edges of the colonies start to pull away from the plate.
Typically, cells at higher passages require shorter collagenase

incubation time.
4. Before feeding or passaging cells, aliquot the amount of
medium needed each time and warm up only the aliquot rather
than the entire bottle of medium. Avoid repeated warming and
cooling or overheating the medium. Add the appropriate
amount of growth factors right before feeding.
5. Observe cells under a microscope during treatment of Versene.
Stop incubation with Versene if the edge of colonies curls up
and cells become rounded up. If not, incubate cells for another
3 min, but not more than 10 min in total. Typically, cells at
early passages require longer time incubation with Versene,
while cells at late passages need shorter treatment with Versene.
6. Do not over triturate the cells; some of the cells may still stay as
small clumps of cells after the dissociation.
7. The doses of growth factors are critical for achieving efficient
differentiation of cardiomyocytes. For example, H7 cells
respond to activin A in a dose-dependent manner—a high
dose of activin A (100 ng/ml) induces differentiation reaching
higher purity of cardiomyocytes compared with lower doses of
activin A (25 or 12 ng/ml) (data not shown). Therefore, it is
important to make sure growth factor aliquots are stored and
handled appropriately to maintain their biological activity.
8. Cell density at the time of growth factor induction plays an
important role in cardiomyocyte differentiation. It is important
to make sure that cells are seeded evenly and reach full conflu-
ence before induction (Fig. 1b). It is advisable to establish the
most efficacious cell seeding density and determine the optimal
timing of induction in pilot experiments.
9. Chamber slides can be coated with 1:30 Matrigel at room
temperature for 2 h since overnight Matrigel-coated surface
tends to produce nonspecific background staining.
10. To detect multiple markers at the same time, incubate cells with
a mixture of the first antibodies, wash, and then incubate with a
mixture of the corresponding secondary antibodies conjugated
with dyes that can be detected through different filters.
11. Due to lot-to-lot variations, it is recommended to work out the
optimal concentrations for each specific lot of antibodies when
performing immunocytochemical analysis or flow cytometry
analysis.
12. Great care should be taken to prevent RNase contamination
during or after the isolation of RNA; therefore wipe RNA work
area with RNAseZap and use only RNase-free materials.
13. It is recommended for each sample to run PCR reactions in
triplicate along with minus RT and non-template controls.
Triplicate biological cultures are used to derive relative levels of

gene expression in mean standard deviation.
14. For safety, EMA liquid waste and all tips and tubes used for
EMA need to be collected into separate liquid and solid con-
tainers, respectively, and handled appropriately according to
safety guidelines.
Acknowledgments
The C. Xu laboratory gratefully acknowledges the funding from the

Children’s Pediatric Research Trust from Emory Children’s Pedi-
atric Research Center; the funding from the National Heart, Lung,
and Blood Institute; National Institutes of Health, under Contract
No. HHSN268201000043C; grants from the National Institutes
of Health (R21HL118454 and R21HL123928); and a grant from
CASIS (GA-2014-126).
References
1. Kehat I, Kenyagin-Karsenti D, Snir M, Segev H, pluripotent stem cells to cardiomyocytes: a
Amit M, Gepstein A, Livne E, Binah O, methods overview. Circ Res 111:344–358
Itskovitz-Eldor J, Gepstein L (2001) Human 8. Laflamme MA, Chen KY, Naumova AV, Musk-
embryonic stem cells can differentiate into myo- heli V, Fugate JA, Dupras SK, Reinecke H, Xu C,
cytes with structural and functional properties of Hassanipour M, Police S, O’Sullivan C, Collins
cardiomyocytes. J Clin Invest 108:407–414 L, Chen Y, Minami E, Gill EA, Ueno S, Yuan C,
2. Xu C, Police S, Rao N, Carpenter MK (2002) Gold J, Murry CE (2007) Cardiomyocytes
Characterization and enrichment of cardio- derived from human embryonic stem cells in
myocytes derived from human embryonic pro-survival factors enhance function of infarcted
stem cells. Circ Res 91:501–508 rat hearts. Nat Biotechnol 25:1015–1024
3. He JQ, Ma Y, Lee Y, Thomson JA, Kamp TJ 9. Burridge PW, Anderson D, Priddle H, Barba-
(2003) Human embryonic stem cells develop dillo Munoz MD, Chamberlain S, Allegrucci
into multiple types of cardiac myocytes: action C, Young LE, Denning C (2007) Improved
potential characterization. Circ Res 93:32–39 human embryonic stem cell embryoid body
4. Mummery C, Ward-van Oostwaard D, Doe- homogeneity and cardiomyocyte differentia-
vendans P, Spijker R, van den Brink S, Hassink tion from a novel V-96 plate aggregation sys-
R, van der Heyden M, Opthof T, Pera M, de la tem highlights interline variability. Stem Cells
Riviere AB, Passier R, Tertoolen L (2003) 25:929–938
Differentiation of human embryonic stem 10. Yao S, Chen S, Clark J, Hao E, Beattie GM,
cells to cardiomyocytes: role of coculture Hayek A, Ding S (2006) Long-term self-
with visceral endoderm-like cells. Circulation renewal and directed differentiation of human
107:2733–2740 embryonic stem cells in chemically defined
5. Burridge PW, Keller G, Gold JD, Wu JC conditions. Proc Natl Acad Sci U S A
(2012) Production of de novo cardiomyocytes: 103:6907–6912
human pluripotent stem cell differentiation 11. Yang L, Soonpaa MH, Adler ED, Roepke TK,
and direct reprogramming. Cell Stem Cell Kattman SJ, Kennedy M, Henckaerts E, Bon-
10:16–28 ham K, Abbott GW, Linden RM, Field LJ,
6. Xu C (2012) Differentiation and enrichment of Keller GM (2008) Human cardiovascular pro-
cardiomyocytes from human pluripotent stem genitor cells develop from a KDR+ embryonic-
cells. J Mol Cell Cardiol 52:1203–1212 stem-cell-derived population. Nature
7. Mummery CL, Zhang J, Ng ES, Elliott DA, 453:524–528
Elefanty AG, Kamp TJ (2012) Differentiation 12. Tomescot A, Leschik J, Bellamy V, Dubois G,
of human embryonic stem cells and induced Messas E, Bruneval P, Desnos M, Hagege AA,
Amit M, Itskovitz J, Menasche P, Puceat M electrophysiological assessment of human

(2007) Differentiation in vivo of cardiac com- embryonic stem cell-derived cardiomyocytes:
mitted human embryonic stem cells in post- a novel in vitro model for the study of conduc-
myocardial infarcted rats. Stem Cells tion. Circ Res 91:659–661
25:2200–2205 20. Zhu WZ, Xie Y, Moyes KW, Gold JD, Askari B,
13. Leschik J, Stefanovic S, Brinon B, Puceat M Laflamme MA (2010) Neuregulin/ErbB sig-
(2008) Cardiac commitment of primate naling regulates cardiac subtype specification
embryonic stem cells. Nat Protoc in differentiating human embryonic stem
3:1381–1387 cells. Circ Res 107:776–786
14. Xu C, Police S, Hassanipour M, Li Y, Chen Y, 21. Liang P, Lan F, Lee AS, Gong T, Sanchez-
Priest C, O’Sullivan C, Laflamme MA, Zhu Freire V, Wang Y, Diecke S, Sallam K, Knowles
WZ, Van Biber B, Hegerova L, Yang J, JW, Wang PJ, Nguyen PK, Bers DM, Robbins
Delavan-Boorsma K, Davies A, Lebkowski J, RC, Wu JC (2013) Drug screening using a
Gold JD (2011) Efficient generation and cryo- library of human induced pluripotent stem
preservation of cardiomyocytes derived from cell-derived cardiomyocytes reveals disease-
human embryonic stem cells. Regen Med specific patterns of cardiotoxicity. Circulation
6:53–66 127:1677–1691
15. Lian X, Hsiao C, Wilson G, Zhu K, Hazeltine 22. Nunes SS, Miklas JW, Liu J, Aschar-Sobbi R,
LB, Azarin SM, Raval KK, Zhang J, Kamp TJ, Xiao Y, Zhang B, Jiang J, Masse S, Gagliardi M,
Palecek SP (2012) Robust cardiomyocyte dif- Hsieh A, Thavandiran N, Laflamme MA,
ferentiation from human pluripotent stem cells Nanthakumar K, Gross GJ, Backx PH, Keller
via temporal modulation of canonical Wnt sig- G, Radisic M (2013) Biowire: a platform for
naling. Proc Natl Acad Sci U S A 109: maturation of human pluripotent stem cell-
E1848–E1857 derived cardiomyocytes. Nat Methods
16. Minami I, Yamada K, Otsuji TG, Yamamoto T, 10:781–787
Shen Y, Otsuka S, Kadota S, Morone N, Barve 23. Zhang D, Shadrin IY, Lam J, Xian HQ, Snod-
M, Asai Y, Tenkova-Heuser T, Heuser JE, grass HR, Bursac N (2013) Tissue-engineered
Uesugi M, Aiba K, Nakatsuji N (2012) A cardiac patch for advanced functional matura-
small molecule that promotes cardiac differen- tion of human ESC-derived cardiomyocytes.
tiation of human pluripotent stem cells under Biomaterials 34:5813–5820
defined, cytokine- and xeno-free conditions. 24. Xu C, Inokuma MS, Denham J, Golds K,
Cell Rep 2:1448–1460 Kundu P, Gold JD, Carpenter MK (2001)
17. Zhang J, Klos M, Wilson GF, Herman AM, Feeder-free growth of undifferentiated human
Lian X, Raval KK, Barron MR, Hou L, Soerens embryonic stem cells. Nat Biotech 19:971–974
AG, Yu J, Palecek SP, Lyons GE, Thomson JA, 25. Xu C (2006) Characterization and evaluation
Herron TJ, Jalife J, Kamp TJ (2012) Extracel- of human embryonic stem cells. Methods
lular matrix promotes highly efficient cardiac Enzymol 420:18–37
differentiation of human pluripotent stem 26. Xu XQ, Graichen R, Soo SY, Balakrishnan T,
cells: the matrix sandwich method. Circ Res Rahmat SN, Sieh S, Tham SC, Freund C,
111:1125–1136 Moore J, Mummery C, Colman A, Zweigerdt
18. Tohyama S, Hattori F, Sano M, Hishiki T, R, Davidson BP (2008) Chemically defined
Nagahata Y, Matsuura T, Hashimoto H, medium supporting cardiomyocyte differentia-
Suzuki T, Yamashita H, Satoh Y, Egashira T, tion of human embryonic stem cells. Differen-
Seki T, Muraoka N, Yamakawa H, Ohgino Y, tiation 76:958–970
Tanaka T, Yoichi M, Yuasa S, Murata M, Sue- 27. Lian X, Zhang J, Zhu K, Kamp TJ, Palecek SP
matsu M, Fukuda K (2013) Distinct metabolic (2013) Insulin inhibits cardiac mesoderm, not
flow enables large-scale purification of mouse mesendoderm, formation during cardiac differ-
and human pluripotent stem cell-derived cardi- entiation of human pluripotent stem cells and
omyocytes. Cell Stem Cell 12:127–137 modulation of canonical Wnt signaling can res-
19. Kehat I, Gepstein A, Spira A, Itskovitz-Eldor J, cue this inhibition. Stem Cells 31:447–457
Gepstein L (2002) High-resolution
Chapter 10
Isolation, Culturing, and Characterization of Cardiac Muscle

Cells from Nonhuman Primate Heart Tissue
Steven M. Hoynowski and John W. Ludlow
Abstract
Cardiac safety pharmacology requires in vitro testing of all drug candidates before clinical trials in order to
ensure they are screened for cardiotoxic effects which may result in severe arrhythmias and, ultimately,
cardiomyopathy (Chi, Nat Rev Drug Discov 12:565–567, 2013). Given the physiological similarities
between nonhuman primates and humans, isolated primate cardiac muscle cells are an ideal animal model
for such in vitro testing. The aims of this chapter are to describe two methods for isolating and culturing
primate cardiac muscle cells. One method uses mechanical dissociation of the tissue followed by placing the
small pieces onto a Petri dish and culturing these tissue explants. The other method also uses mechanical
dissociation but is then followed by enzymatic digestion and culturing of the cell suspension. Methods are
also described for phenotypically characterizing cardiac muscle cells by flow cytometry. Based on the
location within the heart tissue chosen for cell isolation, a dividing population of cardiac muscle cells
expressing cardiomyocyte cell markers was obtained.
Key words Nonhuman primate, Heart, Cardiac muscle, Cardiomyocyte, Explant culture, Enzymatic
digestion, Flow cytometry
1 Introduction
Nonhuman primates are important animal models for translational

research based in large part on their close resemblance to humans.
Nonhuman primates and humans have what may be considered a
“shared biology,” which includes a majority of their genes, simila-
rities in anatomy, and very comparable physiologies. The use of
nonhuman primate models in research has enabled discoveries with
direct application to human studies, bridging the gap between basic
science and human medicine. Discoveries using this type of animal
model are enabling test treatments for human conditions such as
drug addiction, obesity, malaria, and neurodegenerative diseases,
accelerating the pace at which these research advances can be trans-
lated into patient treatments. Perhaps most importantly, primates
more closely resemble humans than any other animal model with
133
134 Steven M. Hoynowski and John W. Ludlow
regard to manifestations of chronic diseases. For example,

nonhuman primates have naturally occurring atherosclerosis,
osteoporosis, and hypertension, making them ideal models for
research treatments for these conditions.
As mentioned in the abstract above, nonhuman primate cardiac
muscle cells have a proven utility for toxicologic screens of potential
new pharmaceutical compounds for treating heart disease [1]. In
addition, these cells are also important for investigating tissue
engineering and regenerative medicine approaches toward heart
disease treatment [2, 3]. For example, progressive necrotic, apo-
ptotic, and/or oncotic cardiac muscle cell death underlies many
forms of cardiovascular disease. Indeed, pathophysiologic cardiac
muscle cell loss is often accompanied by fibrosis. Therefore, gener-
ation of functional cardiac muscle cells to treat or prevent heart
failure has become a much needed approach toward treating cardiac
disease.
2 Materials
2.1 Nonhuman 1. Rhesus monkey (Macaca mulatta).

Primate Heart ( See 2. Cynomolgus monkey (Macaca fascicularis).
Note 1)
2.2 Culture 1. Sterile tissue culture dishes, 100 mm.

Plasticware 2. 6-well tissue culture plate.
3. Pipettes, 5, 10, and 25 mL volumes.
4. Centrifuge tubes, 15 and 50 mL volumes.
5. 1.5 mL microcentrifuge tubes.
6. 1.5 mL cryopreservation vials.
7. Filter sterilization unit, 0.22 mm pore size.
8. 100 mm Steriflip™ (Millipore Cat # SCNY00100).
2.3 Dissection 1. Scissors.

Instruments 2. Scalpel.
3. Forceps.
4. Hemostats.
2.4 Equipment 1. Class II biosafety cabinet.

2. Hemocytometer with cover slips.
3. Centrifuge with swinging-bucket rotor.
4. Microcentrifuge.
5. Inverted microscope for cell culture analysis.
6. 80 C freezer.
Isolation, Culturing, and Characterization of Cardiac Muscle Cells from Nonhuman. . . 135
7. Liquid nitrogen dewar.

8. Variable speed test tube mixer or rocking platform.
9. Cell culture incubator—humidified, 37 C, 5 % CO2/95 % air.
10. 37 C water bath.
11. Accuri flow cytometer or equivalent.
12. Mr. Frosty™ Freezing Container (Thermo Fischer Scientific,
Waltham, MA, USA, Cat # 5100-0001).
2.5 Immunoreagents 1. Anti-heavy chain cardiac myosin antibody (Abcam Inc.,

Eugene OR, USA. Cat # ab15).
2. Anti-sarcomeric alpha actinin antibody (Abcam, Cat #
ab68167).
3. Anticardiac troponin T antibody (Abcam, Cat # ab8295).
4. Goat anti-mouse IgG Alexa Fluor 488 (Invitrogen, Molecular
Pobes, Grand Island, NY, USA Cat # A11029).
5. Allophycocyanin Goat anti-mouse IgG (Invitrogen, Cat #
A-865).
6. Cytofix/Cytoperm fixation and permeabilization solution (BD
Biosciences, Franklin lakes, NJ, USA, Cat # 554722).
7. Perm/wash buffer (BD Biosciences, Cat # 554723).
2.6 Culture Media Storage conditions, shelf life, and expiration dates for all media,
Components and components, and supplements are provided by the manufacturer.
Supplements 1. Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12
(DMEM/F-12) (Invitrogen, LifeTechnologies Corp., Grand
Island, NY, USA).
2. Dulbecco’s phosphate-buffered saline (DPBS) (Invitrogen).
3. 0.25 % Trypsin EDTA, 1 (Invitrogen).
4. Fetal bovine serum (FBS) (Invitrogen).
5. Antibiotic/antimycotic, 100 (Invitrogen).
6. Gentamicin, 50 mg/mL concentration (Invitrogen).
7. Collagenase Type IV (Worthington).
8. Sterile 500 mM CaCl2 solution.
9. Dimethyl sulfoxide (DMSO), tissue culture grade (Sigma-
Aldrich, St Louis, MO, USA, Cat # D2650).
10. 0.4 % trypan blue.
2.7 Formulations All manipulations take place in a biosafety cabinet to reduce the risk
of microbial contamination.
1. Digestion solution—0.25 % trypsin EDTA (1) containing
0.1 % (w/v) collagenase type IV, 5 mM CaCl2. Sterilize
solution through a 0.22 mm filter. Make fresh immediately

before use. Excess may be stored at 20 C for up to
1 month, thawed, and used immediately.
2. Culture medium—DMEM/F-12 supplemented with 10 %
FBS and 0.1 % antibiotic/antimycotic. Store at 4 C, shelf life
of 1 month.
3. Wash medium—DMEM/F-12 containing 50 μg/mL genta-
micin and 0.1 % antibiotic/antimycotic. Store at 4 C, shelf life
of 1 month.
4. Neutralization medium—DMEM/F-12 containing 2 % FBS
and 0.1 % antibiotic/antimycotic. Store at 4 C, shelf life of
1 month.
5. Cryopreservation medium—90 % FBS, 10 % DMSO. Store at
4 C, shelf life of 1 month.
3 Methods
All manipulations take place in a biosafety cabinet to reduce the risk

of microbial contamination.
3.1 Tissue Explant 1. Remove the heart from its shipping container and place in a
Cultures sterile 100 mm tissue culture dish.
3.1.1 Heart Tissue 2. Dissect away any extraneous tissue and remove the pericardium
Handling and Preparation membrane (see Fig. 1).
3. Flush the outside of the organ with 10 mL of Wash medium
(see Note 2).
Fig. 1 Rhesus monkey (Macaca mulatta) heart with pericardium membrane

removed
Fig. 2 Bisected heart showing locations of tissue excision for cell isolation
4. Aspirate the Wash medium from the culture dish.

5. Repeat steps 3 and 4 of Subheading 3.1.1 a total of three times.
6. Bisect the organ down the middle using scissors and a scalpel to
expose the inside of the ventricle and atrial chambers (see
Fig. 2).
7. Flush the inside of the organ with 10 mL of Wash medium (see
Note 3).
8. Aspirate the Wash medium from the culture dish.
9. Repeat steps 7 and 8 of Subheading 3.1.1 a total of three times.
3.1.2 Tissue Explant 1. Excise tissue from the inside of the ventricles.
Method 2. Pre-fill the wells of a sterile 6-well plate with 10 mL of pre-
warmed Wash medium in each well (see Note 4).
3. Using sterile forceps, gently place the biopsy tissue into the first
well of the 6-well washing plate.
4. Gently agitate the tissue in the well using the sterile forceps for
5–10 s.
5. Carefully lift the tissue from the first well and place in the
second well and repeat the agitation procedure in step 4 of
Subheading 3.1.2 above.
6. Continue the successive washing of the tissue through each
until all six wells have been used.
7. After washing, carefully move the tissue into a sterile 100 mm
tissue culture dish for dissection.
8. Using sterile forceps and either small sterile scissors or a sterile
scalpel, cut cardiac muscle tissue into small pieces approxi-
mately 1 mm in diameter.
Fig. 3 A 100 mm tissue culture dish showing tissue explants adhered to the
surface
9. Using forceps, carefully place each explant onto a 100 mm

tissue culture dish.
10. Repeat steps 8 and 9 of Subheading 3.1.2 until the dish is
evenly covered with approximately 20–25 tissue explants.
11. Allow the plate to sit open inside the biological safety cabinet
for 10–15 min to allow the tissue explants time to adhere to the
dish (see Fig. 3).
12. Gently add 10 mL culture medium to moisten and submerge
tissue fragments without dislodging them from the dish.
13. Place dish into humidified 37 C incubator in 5 % CO2 undis-
turbed for 3 days.
14. Gently remove dishes from the incubator, aspirate medium,
and gently feed with 10 mL of fresh culture medium without
dislodging adhered explants from the dish.
15. Return dishes to incubator.
16. Repeat steps 14 and 15 of Subheading 3.1.2 until dish is
70–90 % confluent with cells (see Fig. 4 and Note 5).
3.1.3 Enzymatic 1. Excise tissue from the inside of the ventricles.

Digestion Method of 2. Pre-fill the wells of a sterile 6-well plate with 10 mL of room
Cardiac Muscle Cell temperature Wash medium in each well.
Isolation
3. Using sterile forceps, gently place the biopsy tissue into the first
well of the 6-well washing plate.
4. Gently agitate the tissue in the well using the sterile forceps for
5–10 s.
Fig. 4 (a) Phase contrast image of cardiac muscle cells migrating away from tissue explant (Day 6). (b) Phase
contrast image of cardiac muscle cells surrounding a tissue explant (Day 9). (c) Proliferation of cardiac muscle
cells subsequent to migration from tissue explant resulting in confluency (Day 12) (d) Space left on the dish
after explant spontaneously lifted off
5. Carefully lift the tissue from the first well and place in the
second well and repeat the agitation procedure in step 4 of
Subheading 3.1.3 above.
6. Continue the successive washing of the tissue through each
until all six wells have been used.
7. After washing, carefully move the tissue into a sterile 100 mm
tissue culture dish for dissection.
8. Using sterile forceps and either small sterile scissors or a sterile
scalpel, mince tissue and transfer approximately 1 g to a 50 mL
tube.
9. Repeat step 8 of Subheading 3.1.3 for the remainder of the
tissue.
10. Add 40 mL of digestion solution, cap tightly, and rock vigor-
ously at 37 C for 20 min (see Note 6).
11. Add 10 mL of neutralization medium to deactivate the
Enzyme.
12. Connect the digestion tube to a 100 mm Steriflip™ assembly
and apply a gentle vacuum to filter out any remaining undi-
gested material.
Fig. 5 Cardiac muscle cell cultures at Day 2 (a), Day 7 (b), and Day 18 (c) following enzymatic digestion of
tissue, cell isolation, and plating
13. Disconnect the receiving 50 mL tube from the Steriflip™, cap

tightly, and centrifuge at 300 g for 5 min.
14. Aspirate the supernatant and resuspend the cell pellet in 40 mL
culture medium.
15. Pipet the cell suspension into four T-75 flasks or two T-175
flasks.
16. Place flasks into humidified 37 C incubator in humidified, 5 %
CO2-containing atmosphere.
17. Leave the flasks undisturbed for 2–3 days (see Fig. 5 and
Note 7).
3.1.4 Cardiac Muscle 1. For explant culture method, carefully aspirate any remaining
Cell Passaging tissue and conditioned medium from the plate without disturb-
ing the surrounding cell colonies.
2. For cultures from enzymatic digestion method, aspirate
conditioned medium from the flask.
3. Wash cell surface with 10 mL of DPBS.
4. Aspirate the DPBS, and then add 5 mL of 0.25 % trypsin
EDTA.
5. Monitor the cell detachment visually using the inverted
microscope.
6. When most of the cells have detached, add 5 mL of culture
medium to neutralize the trypsin and detach any lightly
attached cells.
7. Transfer the trypsin/medium mixture to a 15 mL sterile cen-
trifuge tube and centrifuge at 300 g for 5 min to pellet the
cells.
8. After centrifugation, carefully aspirate the supernatant making
sure not to disturb the cell pellet.
9. Resuspend the cell pellet in 5 mL of culture medium and

quantitate the number of cells by staining with trypan blue
and counting on a hemocytometer.
10. Passage the cells by seeding the appropriate number of culture
dishes at 2,000–4,000 cells/cm2.
3.1.5 Phenotypic Marker 1. Pellet at least 50,000 cells in a 1.5 mL microcentrifuge tube.
Analysis 2. Fix in 100 μL Cytofix/Cytoperm fixation and permeabilization
solution for 20 min at 4 C.
3. Spin in microcentrifuge at 500 g for 3 min to pellet fixed
cells.
4. Wash fixed cells once with 500 μL of perm/wash buffer.
5. Resuspend pellet in 20 μL of perm/wash buffer.
6. Add 1 μg primary antibody.
7. Incubate for 1 h at 4 C.
8. Wash once with 500 μL of perm/wash buffer.
9. Resuspend pellet in 20 μL of perm/wash buffer.
10. Add 1 μg secondary antibody.
11. Incubate for 30 min at 4 C.
12. Wash once with 500 μL of perm/wash buffer.
13. Resuspend pellet in 150 μL DPBS.
14. Analyze by flow cytometry (see Fig. 6).
3.1.6 Cardiac Muscle 1. Count cells and pellet at 300 g for 5 min in a 50 mL sterile
Cell Cryopreservation centrifuge tube.
2. Resuspend cells in cryopreservation medium at desired concen-
tration (see Note 8).
3. Aliquot 1 mL into cryopreservation vials.
4. Place vials into Mr. Frosty™.
5. Place Mr. Frosty™ into 80 C freezer for 24–72 h.
6. Transfer vials to liquid nitrogen dewar for long-term storage.
4 Notes
1. Working with nonhuman primate tissue requires a biosafety

level 2 facility. Excised heart should be aseptically flushed with
sterile tissue culture media or sterile saline before being sus-
pended in tissue culture media or saline for shipping. Best
results are obtained when the organ is received for processing
within 24 h after collection.
Fig. 6 Flow cytometric analysis of nonhuman primate cardiac muscle cell cultures. (a) Cells stained with goat
anti-mouse IgG Alexa Fluor 488, revealing nonspecific binding of this secondary antibody (control), and
cardiac myosin heavy chain staining with goat anti-mouse IgG secondary, showing a shift in the population,
indicating a positive staining for this marker (MH). (b) Cardiac troponin T staining with goat anti-mouse IgG
secondary, showing a very slight shift in the population, indicating weakly positive staining for this marker
(troponin). (c) Cells stained with allophycocyanin goat anti-mouse IgG, revealing nonspecific binding of this
secondary antibody (control), and cells stained with sarcomeric alpha actinin antibody followed by allophy-
cocyanin secondary, showing a further shift in the population, indicating a positive staining for this marker
(actinin)
2. Bathing the organ in antibiotic-/antimycotic-containing

medium prior to dissection will help to further reduce the
risk of microbial contamination.
3. Additional bathing of the exposed muscle surface in antibi-
otic-/antimycotic-containing medium prior to tissue removal
is optional, since the excised tissue pieces are also bathed in
antibiotic-/antimycotic-containing medium prior to explant
culturing and mincing for enzymatic digestion.
4. This step is the most critical to reducing the risk of microbial
contamination of the cultures.
5. Explanted tissue may spontaneously lift off, even with gentle
handling, and should be removed from the culture by aspira-
tion. Failure to do so may result in culture becoming contami-
nated or disruption of newly formed cell colonies.
6. Tissue may not be completely digested, but as long as the
digestion mixture appears cloudy after the 20 min, at least
some cardiac muscle cells are in suspension having been
liberated from the tissue pieces.
7. Cells and small cell colonies may be very sparse at this time, but
they should be readily visible under the microscope.
8. Cardiac muscle cells may be cryopreserved at any useful con-
centration between 0.5 106 and 30 106 cells/mL.
References
1. Chi KR (2013) Revolution dawning in cardio- 3. Hirt MN, Hansen A, Eschenhagen T (2014)
toxicity testing. Nat Rev Drug Discov 12 Cardiac tissue engineering: state of the art. Circ
(8):565–567 Res 114(2):354–367
2. Gu Y, Yi F, Liu G-H, Izpisua Belmonte JC
(2013) Beating in a dish: new hopes for cardio-
myocyte regeneration. Cell Res (3):314–316
Chapter 11
Mouse Embryonic Stem Cell-Derived Cardiac Myocytes

in a Cell Culture Dish
Carley Glass, Reetu Singla, Anshu Arora, and Dinender K. Singla
Abstract
Embryonic stem (ES) cells are pluripotent stem cells capable of self-renewal and have broad differentiation
potential yielding cell types from all three germ layers. In the absence of differentiation inhibitory factors,
when cultured in suspension, ES cells spontaneously differentiate and form three-dimensional cell aggre-
gates termed embryoid bodies (EBs). Although various methods exist for the generation of EBs, the
hanging drop method offers reproducibility and homogeneity from a predetermined number of ES cells.
Herein, we describe the in vitro differentiation of mouse embryonic stem cells into cardiac myocytes using
the hanging drop method and immunocytochemistry to identify cardiomyogenic differentiation. In brief,
ES cells, placed in droplets on the lid of culture dishes following a 2-day incubation, yield embryoid bodies,
which are resuspended and plated. 1–2 weeks following plating of the EBs, spontaneous beating areas can
be observed and staining for specific cardiac markers can be achieved.
Key words Cardiac myocytes, Embryonic stem cells, Embryoid bodies, Hanging drop, Cell culture,
Culture medium, Differentiation medium, Cell differentiation
1 Introduction
Embryonic stem (ES) cells possess unsurpassed differentiation

potential yielding lineage-specific cell types from all three germ
layers including the ectoderm, endoderm, and mesoderm. Specifi-
cally, the differentiation of ES cells into cardiac myocytes has
become an essential tool for the investigation of cardiac develop-
ment and been widely used in regenerative and reparative medicine
for the treatment of cardiac anomalies [1–4]. With the aforemen-
tioned capacity of ES cells in mind, methods to facilitate the
differentiation of ES cells into cardiac lineage cell types are impera-
tive to continued translational research and the generation of
in vitro disease-specific models.
ES cells cultured in the presence of feeder layers or differentia-
tion inhibitory factors, including leukemia inhibitory factor (LIF),
retain their pluripotency and self-renewal characteristics [5].
145
146 Carley Glass et al.
However, when cultured in the absence of such factors, ES cells

spontaneously differentiate forming three-dimensional cell aggre-
gates termed embryoid bodies (EBs). Under time-dependent
conditions, EBs increase in cell complexity, yielding specialized cell
types including hepatocytes, pancreatic beta cells, renal proximal
tubular progenitor cells, neurons, and cardiac myocytes, to name a
few [6–9]. Human and mouse ES cell-derived EBs are generated
through various culture methods, including liquid suspension,
methylcellulose, and hanging drop [10]. Notably, the hanging
drop method is widely used for the differentiation of mouse ES
cells into a variety of cell types, including cardiac myocytes, as
reproducible, homogenous EBs can be generated [6, 11, 12].
Within the current protocol, differentiation of ES cells into
cardiac myocytes using the hanging drop method is first achieved
by disassociating cultured ES cells on gelatin-coated plates using
trypsin and obtaining a predetermined concentration of enriched
ES cells. Next, cells are seeded as droplets onto the lid of a 100 mm
dish and incubated at 37 C upside down for 2 days. Following
incubation, formed EBs are harvested and transferred to suspension
culture for 3 days. On day 6, EBs are placed onto gelatin-coated
plates with media replacements occurring every other day. 1–2
weeks post-day 6, physiological visual examination can be per-
formed to identify beating EBs. Cardiomyogenic differentiation
can be more specifically characterized by various methods including
western blot with specific cardiac myocyte markers, RT-PCR, and
immunocytochemistry as described within the current protocol.
2 Materials
2.1 0.1 % Gelatin- 1. 500 ml beaker.

Coated Plates 2. 100 mm tissue culture dishes.
3. 1 PBS: 1 part 10 PBS (Fisher, #BP399-1), 9 parts DI water
(see Note 1).
4. Microwave oven.
5. 0.22 μm 500 ml filter bottle.
2.2 Cell Culture 1. Cell culture media: Dulbecco’s modified Eagle medium
Media (DMEM) was accompanied with 2 mM L-glutamine, 0.1 mM
β-mercaptoethanol, 1 nonessential amino acids (see Note 2),
1 mM sodium pyruvate, 10–15 % fetal calf or bovine serum
(FBS), leukemia inhibitory factor (LIF, 1,000-2,000 U/ml),
50 U/ml penicillin/streptomycin.
2. Differentiation media: DMEM along with 2 mM glutamine,
0.1 mM β-mercaptoethanol, 1 nonessential amino acids
(see Note 2), 1 mM sodium pyruvate, 15 % FBS, 50 U/ml
penicillin/streptomycin.
Mouse Embryonic Stem Cell-Derived Cardiac Myocytes in a Cell Culture Dish 147
2.3 Hanging Drop 1. CGR8 ES cells.

Method Components 2. Cell culture media (preparation detailed in Subheading 2.2,
item 1).
3. 0.1–0.2 % gelatin-coated 100 mm tissue culture dishes.
4. 1 PBS (preparation detailed in Subheading 2.1, item 3).
5. 0.25 % Trypsin/EDTA.
6. Differentiation medium (preparation detailed in Subheading 2.2,
item 2).
7. Centrifuge.
8. Bright-Line Hemacytometer.
9. Reagent reservoir.
10. Multichannel pipette.
2.4 Immunocyto- 1. Cell scraper.

chemistry 2. 0.1 % gelatin-coated 8-well chamber slide.
Components
3. Differentiation medium (preparation detailed in Subhead-
ing 2.2, item 2).
4. 1 PBS (preparation detailed in Subheading 2.1, item 3).
5. 4 % paraformaldehyde.
6. M.O.M. kit (Vector Laboratories, USA): Mouse Ig Blocking
Reagent, M.O.M. Diluent provided in the kit, M.O.M.™
Biotinylated Anti-Mouse IgG, and fluorescein avidin DCS
was also present among the kit components.
7. Monoclonal anti-α-actin (α-sarcomeric) mouse antibody.
8. Vectashield mounting medium with DAPI (Vector
Laboratories).
9. Coverslips.
3 Methods
Carry out all procedures at room temperature underneath a sterile

certified biosafety cabinet unless otherwise specified. All solutions
and equipment used in the below protocols must be sterile and
appropriate aseptic techniques are mandatory.
3.1 Preparation 1. Place 100 ml of 1 PBS into a beaker.

of 0.1 % Gelatin- 2. Add 0.5 g of tissue culture grade gelatin (Sigma, #G1890) into
Coated Plates PBS and stir to dissolve.
3. Heat solution on high in a microwave oven for 2 min (see Note 3).
4. Bring solution up to 500 ml with 1 PBS.
5. Filter sterilize the 0.1 % gelatin solution using a 0.22 μm

bottle-top filter into a sterile 500 ml bottle (see Note 4).
6. Add 2 ml of 0.1 % gelatin to the bottom of a 100 mm plate and
ensure that the entire surface is covered.
7. Keep prepared plates under the hood for 20–25 min.
8. Aspirate the remaining gelatin (see Note 5).
3.2 Hanging Drop 1. Culture ES cells in cell culture medium for 2–3 passages on
Method for the 0.1 % gelatin-coated 100 mm plates after thawing.
Formation of Embryoid 2. Aspirate culture medium and wash ES cells once with 5 ml PBS.
Bodies (Day 1) 3. Remove PBS and place 3 ml of trypsin/EDTA into the plate for
3–5 min.
4. Pipette the solution so that the cells are disassociated (see Note 6).
After isolation of the cells, no clusters should be visible in the
solution.
5. Place solution into a 15 ml tube containing 3 ml of differentia-
tion medium.
6. Spin down for 5 min at 1,000 rpm.
7. Remove supernatant and resuspend cells in 5–10 ml of differ-
entiation medium.
8. Calculate cells/ml using a hemacytometer.
9. Dilute the cells using differentiation medium so that the final
concentration is 2.5 104 cells/ml (see Note 7).
10. Resuspend cell suspension thoroughly and transfer into a sterile
reagent reservoir.
11. Remove the lid from a new, sterile 100 mm plate, placing it
upside down with the inside facing upright.
12. Fill the bottom of the 100 mm dish with 15 ml of 1 PBS to
prevent evaporation.
13. Using a multichannel pipette, place 30 μl drops onto the lid of
the petri dish.
14. Repeat this step 5–6 times until the lid is full (50–60 drops per
lid) (see Note 8).
15. Place the lid containing the drops back onto the 100 mm plate
(see Note 9).
16. Incubate the plate at 37 C for 2 days (see Note 10).
3.3 Resuspension 1. Keeping in mind the number of 100 mm dishes made in the
of Embryoid Bodies hanging drop step, place 5 ml of differentiation medium into
(Day 3) the same number of bacterial plates.
2. Remove the lid from the 100 mm plate containing the hanging
drops. EBs can be visualized at this time as a white point at the
end of each drop.
3. Add 5 ml of differentiation medium to the lid and resuspend all

the drops (see Note 11).
4. Aspirate the solution with EBs from the lid with a pipette and
place it into the new 100 mm dish containing 5 ml of differen-
tiation medium.
5. Incubate at 37 C for 3 days
3.4 Plating Embryoid 1. Prepare the same number of 0.1 % gelatinized 100 mm plates as
Bodies (Day 6) the number of plates incubated in step 5 of Subheading 3.3.
Resuspension of embryoid bodies.
2. Place 10 ml of differentiation medium into each new plate.
3. Very gently, pick up all EBs with a 1 ml pipette and place into a
new gelatin-coated 100 mm plate.
4. Incubate at 37 C.
3.5 Media Change 1. Change the media in the plates every other day using differen-
(Day 7 and Every Other tiation medium.
Day Thereafter) 2. Aspirate and discard the culture medium in the 100 mm dishes.
3. Place 10 ml of fresh differentiation medium into each plate.
4. Incubate at 37 C.
3.6 EB Observation 1. At day 7 to day 21, microscopically evaluate EBs for beating
areas (see Notes 12 and 13).
2. When >50 % of the EB is beating, cardiomyogenic markers can
be evaluated.
3.7 Immunocyto- 1. Aspirate culture medium and gently pick up EBs using a
chemistry to Detect scraper.
Cardiomyogenic 2. Place 1 EB per well into a prepared 0.1 % gelatin-coated 8-well
Markers on EBs chamber slide.
(See Note 14) 3. Place 500 μl of differentiation medium into each well.
4. Incubate at 37 C for 48–72 h.
5. Remove differentiation medium and wash 1 with 1 PBS.
6. Fix EBs with 4–5 % paraformaldehyde for 15–20 min at room
temperature (see Note 15).
7. Wash 3 with 1 PBS for 5 min each.
8. Add 0.3 % Triton X-100 for 15 min at room temperature
(see Note 16).
9. Wash 3 in 1 PBS for 5 min each.
10. Incubate chamber slide in working solution of M.O.M. Mouse
Ig Blocking Reagent for 1 h at room temperature (see Note 17).
11. Wash 2 in 1 PBS for 2 min each.
12. Incubate chamber slide for 5 min in protein working solution

of M.O.M. Diluent at room temperature (see Note 18).
13. Dilute monoclonal anti-α-actin (α-sarcomeric) mouse antibody
to appropriate concentration in protein working solution
(prepared in step 12) and incubate for 30 min (see Note 19).
14. Wash each well in the chamber slide 2 for 2 min each in
1 PBS.
15. Use working solution of M.O.M.™ Biotinylated Anti-Mouse
IgG Reagent on slides and incubate chamber slide for 10 min
(see Note 20).
16. Wash chamber slide 2 for 2 min each in 1 PBS.
17. Apply fluorescein avidin DCS and incubate chamber slide for
5 min (see Note 21).
18. Wash slide 2 for 5 min each in 1 PBS.
19. Mount with Vectashield containing DAPI and gently place
coverslip.
4 Notes
1. For any prepared PBS used in cell culture, it is highly recom-

mended that the PBS be filter sterilized using a 0.22 μm bottle-
top filter.
2. Nonessential amino acid stock solution is 100. If preparing
500 ml, use 5 ml (1) of nonessential amino acid solution for
medium preparation.
3. Solution will boil. Be sure to place solution in beaker large
enough so that the solution does not spill over when heated
in the microwave.
4. The gelatin may be stored at 4 C and can be used for up to
2 months following preparation.
5. Prepared 0.1 % gelatinized plates can be stored at 4 C for up to
1 month for future use.
6. ES cells are sticky, so dissociation requires vigorous washing
and pipetting to break clumps. This is VERY important to
ensure proper formation of EBs.
7. This translates to ~750 cells per 30 μl drop. The dilution should
be made in a new 50 ml tube. Use a 1 or 2 ml pipette to make
the dilution. Do not use an automatic pipette. You will need at
least 15 ml of cell suspension to use the multichannel pipette.
8. Make sure that you do not overlap any of the drops on the lid or
overcrowd the lid of the petri dish as this will impede EB
formation.
9. When flipping the lid onto the petri dish, do so carefully, with
grace and swiftness, to ensure that placement of the drops is not
disturbed.
10. For proper EB formation, the dishes with the drops must not
be disturbed for the 2-day duration.
11. This must be done with gentle flushes avoiding bubbles in
order not to disturb the delicate architecture of the EBs. Also,
be careful to avoid solution coming into contact with the
border of the lid to avoid contamination.
12. Although this protocol details up to day 21, EBs can be
cultured for months with medium changes occurring every
other day as previously specified.
13. A straightforward way to evaluate cardiomyogenic differentia-
tion is to count the number of beating EBs, count the beats/
min of each beating EB, and calculate the beating area/total
EB area.
14. Although here we detail immunocytochemisty for the detec-
tion of cardiomyogenic differentiation, other techniques
including western blot and RT-PCR can be used.
15. All procedures from this step forward do not have to be per-
formed under a biosafety cabinet.
16. This step is light sensitive, so make sure to protect the slides
from light.
17. The working solution for this blocking reagent can be made by
adding 1 drop of the stock solution to 1.25 ml of 1 PBS.
18. The working solution for the M.O.M. Diluent can be made by
adding 200 μl of protein concentrate to 2.5 ml 1 PBS.
19. As per our experience we suggest that the accurate times
explained in steps 12–15 will give you better staining results.
Additional longer incubation of slides may result in an increase
in background staining. If longer incubation times are
required, then additional appropriate negative controls are
recommended to ensure the efficacy of the M.O.M.™ kit.
20. To prepare the working solution of M.O.M.™ Biotinylated
Anti-Mouse IgG Reagent, add 4 μl of stock solution to 1 ml
of M.O.M. Diluent previously prepared.
21. To prepare fluorescein avidin DCS, add 16 μl of reagent to 1 ml
1 PBS.
References
1. Doetschman T, Shull M, Kier A et al (1993) embryonic stem cells in vitro. Cell Transplant
Embryonic stem cell model systems for vascu- 11:429–434
lar morphogenesis and cardiac disorders. 8. Kobayashi T, Tanaka H, Kuwana H et al
Hypertension 22:618–629 (2005) Wnt4-transformed mouse embryonic
2. Kehat I, Gepstein L (2003) Human embryonic stem cells differentiate into renal tubular cells.
stem cells for myocardial regeneration. Heart Biochem Biophys Res Commun 336:585–595
Fail Rev 8:229–236 9. Talavera-Adame D, Wu G, He Y et al (2011)
3. Kumar D, Kamp TJ, LeWinter MM (2005) Endothelial cells in co-culture enhance embry-
Embryonic stem cells: differentiation into car- onic stem cell differentiation to pancreatic pro-
diomyocytes and potential for heart repair and genitors and insulin-producing cells through
regeneration. Coron Artery Dis 16:111–116 BMP signaling. Stem Cell Rev 7:532–543
4. Singla DK (2010) Stem cells in the infarcted 10. Kurosawa H (2007) Methods for inducing
heart. J Cardiovasc Transl Res 3:73–78 embryoid body formation: in vitro differentia-
5. Williams RL, Hilton DJ, Pease S et al (1988) tion system of embryonic stem cells. J Biosci
Myeloid leukaemia inhibitory factor maintains Bioeng 103:389–398
the developmental potential of embryonic stem 11. Boheler KR, Czyz J, Tweedie D et al (2002)
cells. Nature 336:684–687 Differentiation of pluripotent embryonic stem
6. Kehat I, Kenyagin-Karsenti D, Snir M et al cells into cardiomyocytes. Circ Res
(2001) Human embryonic stem cells can dif- 91:189–201
ferentiate into myocytes with structural and 12. Kawai T, Takahashi T, Esaki M et al (2004)
functional properties of cardiomyocytes. J Efficient cardiomyogenic differentiation of
Clin Invest 108:407–414 embryonic stem cell by fibroblast growth factor
7. Miyashita H, Suzuki A, Fukao K et al (2002) 2 and bone morphogenetic protein 2. Circ J
Evidence for hepatocyte differentiation from 68:691–702
Chapter 12
Cryopreservation of Neonatal Cardiomyocytes

Adam C. Vandergriff, M. Taylor Hensley, and Ke Cheng
Abstract
Cardiomyocytes are frequently used for in vitro models for cardiac research. The isolation of cells is time-
consuming and, due to the cells limited proliferative abilities, must be performed frequently. To reduce the
time requirements and the impact on research animals, we describe a method for cryopreserving neonatal
rat cardiomyocytes (NRCMs), and subsequently thawing them for use in assays.
Key words Cardiomyocytes, Cryopreservation, Cardiac research
1 Introduction
In vitro models are crucial for cardiac research as there is greater

control and they afford the ability to perform measurements that
are not possible in vivo. Cardiomyocytes are frequently isolated
from mice or rats and are classified by the age of the animal:
adult, neonatal, or fetal. Neonatal rat cardiomyocytes (NRCMs)
are commonly used since their isolation and culture is easier than
that of adult rat cardiomyocytes [1]. NRCMs lack the ability to
proliferate extensively; therefore, isolations must be performed
frequently. There are numerous protocols for the isolation of
NRCMs with most generally requiring 6–48 h of work [2–4].
The cells work best when harvested from 2-day-old rat pups, so
performing the procedure at the correct time is crucial. The com-
bination of a long procedure that must be done within a small
window of time, coupled with the unpredictable nature of the rat
births can lead to inefficient work schedules. Other tasks may be
delayed due to the isolation of NRCMs taking precedence. Isola-
tions from a single litter can contain anywhere from 35 to 60
million cells, yet very few assays require this amount of cells.
Many of the cells isolated may be discarded because of the inability
to use all of the cells at the time of isolation. Many lab groups and
companies have created protocols or kits to improve the length of
153
154 Adam C. Vandergriff et al.
time required for the isolation procedure, but this does not address
the issue of timing the birth of the rats.
Many labs opt for cells that can be cryopreserved such as
cardiomyocytes derived from embryonic stem cells (ESCs) or
induced pluripotent stem cells (iPSCs), but the differentiation
process can be difficult and these cells have been shown to exhibit
differences from primary cardiomyocytes [5–7]. Dissociated
NRCMs are capable of being stored for several days using refriger-
ation [8], yet this does not allow for long term storage. For long
term storage in liquid nitrogen a cryoprotectant such as dimethyl
sulfoxide (DMSO) is necessary. Previous research has shown that
the ideal concentration between 5 and 10 % DMSO in the freezing
media allows for cryopreservation of NRCMs, yet even then their
viability remains low [9]. Although DMSO helps protect the cells
during freezing, it can be toxic to cells at concentrations above
1.5 % [10]. Previous studies have shown that slowly removing
DMSO from the cells may improve cell viability [11].
In order to improve upon cardiac in vitro models, we have
created a protocol for the isolation and cryopreservation of
NRCMs. By cryopreserving the cells lab efficiency can be increased
and the impact on research animals reduced. Using this method, we
show that it is possible to cryopreserve NRCMs and thaw them for
use at a later time.
2 Materials
2.1 Solutions 1. 20 % NRCM Culture Stock Media: In 400 mL Iscove’s Mod-

ified Dulbecco’s Medium (IMDM, with added 25 mM HEPES
and L-glutamine) add 100 mL fetal bovine serum (FBS),
5 mL L-glutamine, 2.5 mL gentamicin, and 0.9 mL 2-
mercaptoethanol. Sterilize using vacuum filter. Dilute 20 %
FBS stock media 1:1 with IMDM to produce 10 % FBS
NRCM media and 1:5 to produce 2 % FBS NRCM Media.
Only make in small batches to prevent repeated heating and
cooling of media (see Note 1).
2. Freezing media: Various freezing media with 5 % DMSO may
work, but good results have been achieved with Cryostor CS5
(BioLife Solutions).
3. Hank’s Balance Salts Solution (HBSS): pH 7.1–7.4 and
with added calcium and magnesium. Refrigerate and keep
cold during use.
4. 0.1 % Trypsin: 0.25 and 0.05 % trypsin mixtures can be pur-
chased for use in cell culture. Make 10 mL aliquots of stock to
prevent repeated and potentially damaging freeze–thaw cycles.
0.1 % trypsin can be made by mixing 0.05 and 0.25 % trypsin in
a 1:3 ratio to produce 40 mL 0.1 % trypsin. This can be done
Cryopreservation of Neonatal Cardiomyocytes 155
right before placing the cells in the trypsin. Begin thawing at

4 C the day before the procedure.
5. Collagenase (Corning Life Sciences).
2.2 Tools 1. Surgical tools: Large scissors, small scissors, large forceps, and
small sharp forceps. Autoclave prior to use.
2. 40 μm cell strainer.
3. Plastics: 15 mL conical tubes, 50 mL conical tubes, 150 mm
culture dishes, cryogenic vials.
4. Trypan blue.
5. Cryo-freezing container (such as Mr. Frosty™, Nalgene) or
controlled rate freezer to keep rate of cooling to 1 C/min.
6. Small sterile gauze pads.
7. Bench pads.
3 Methods
3.1 Cell Isolation Perform the following work in a culture hood to prevent contami-
Day 1 nation of the samples. This portion is best started late in the
afternoon since it ends with an overnight incubation.
1. Prepare workspace inside a sterile culture hood. Place bench
pad in culture hood. Fill two 250 mL beakers with 70 % ethanol
and place surgical tools in them. Place small bucket of ice in
hood. The bucket must be large enough to hold several tubes
and a 150 mm dish. UV-sterilize the hood for 10 min.
2. Retrieve neonatal rat pups. Place them on a bench pad in an
insulated container to keep them warm during transport and
during the operation.
3. Fill a 150 mm culture dish with 25 mL HBSS and place on ice.
4. Unwrap 2–3 gauze pads and place in work area.
5. Holding pup by the skin on the back of the neck, spray it with
70 % ethanol and place in hood (see Note 2).
6. Euthanize the pup by holding the pup by the skin on the back
of the neck. Then, using the large scissors, in one quick, strong
cut remove the head. Blot the blood with gauze.
7. Continue cutting down anterior side of chest through the
ribcage. Squeeze the skin on the neck to help open the ribcage
until the heart is visible. Using the small scissors, remove the
heart and place in the HBSS containing culture dish on ice.
8. Repeat for the other pups.
9. Remove any non-heart tissue from the samples and swirl the
HBSS gently to wash the hearts.
10. Transfer the heart to tissue to another culture dish containing

fresh, HBSS and further wash the tissue. Cut the tissue further
until the pieces are 1–3 mm3, and then transfer to the trypsin
solution. Place trypsin solution on rocker in 4 C refrigerator
overnight.
3.2 Cell Isolation 1. While aspirating in the following steps, use a pipette rather than
Day 2 a vacuum line. The tissue moves easily, and a vacuum line will
clog or remove tissue still containing cells. Therefore, it is best
to use a pipette for aspiration instead. If tissue enters the
pipette, pipette it back out to avoid losing it.
2. Make three aliquots of 10 % NRCM Media: one with 25 mL
and two with 15 mL. Place one 25 mL tube of media in 37 C
water bath.
3. Add 4 mL HBSS to five different 15 mL conical tubes and place
all on ice. Label the tubes 1 through 5.
4. Add 40 mL HBSS to a 50 mL conical tube. Add 10 mL of
HBSS to 50 mg lyophilized collagenase and pipette up and
down repeatedly to suspend. Add this solution to 40 mL
HBSS to create a 1 mg mL1 collagenase solution. Sterilize
by passing the solution through a 0.22 μm vacuum filter and
store at room temperature.
5. Retrieve heart tissue from each trypsin tube. The trypsin solu-
tion should be clear, and the tissue should look fluffy. Allow the
tissue to settle to a corner of tube and remove as much liquid as
possible by aspiration.
6. Add 25 mL warm 10 % serum culture media to the tube and
swirl. Replace the cap and rotate in 37 C water bath for 2 min.
7. Aspirate liquid again. The tissue may float at this step; if so,
aspirate from the bottom of the tube.
8. Add 10 mL of collagenase. Rotate in water bath for 2 min or
until the solution is cloudy.
9. Quickly aspirate the liquid from the tube.
10. Add 10 mL fresh collagenase and rotate in water bath for
2 min. Pipette to mix, transfer solution to the tube labeled
#1, and place on ice. Note that it may not be possible to remove
all of the liquid.
11. Repeat prior step for the other three 15 mL conical tubes.
12. Any remaining liquid and tissue can be transferred to the tube
labeled #5.
13. Centrifuge the 15 mL tubes at 410 rcf for 8 min. While
centrifuging, place a 15 mL tube containing 10 % media in
water bath at 37 C.
14. Place 40 μm cell strainer on top of a 50 mL conical tube and
wet with 2 mL cold HBSS.
15. Aspirate supernatant from spun down cells. Add 6 mL cold

HBSS to each tube and resuspend the cell pellets in each.
Pipette cell suspensions through the strainer into a 50 mL
conical tube to remove any remaining undissociated tissue.
Rinse the sides of each 15 mL tube with 3 mL cold HBSS
and add to strainer.
16. Centrifuge cells at 410 rcf for 10 min. Remove the supernatant
by aspiration and resuspend the cell pellet in 10 mL 10 %
NRCM media. Transfer to a T-75 flask. Rinse the conical
tube with 5 mL media and add to flask.
17. Place T-75 containing cells in cell culture incubator for 1 h and
place last 15 mL 10 % media tube into water bath.
18. Transfer the contents of the T-75 flask to a T-175 flask and
rinse the original flask with 15 mL media to remove any
remaining cells. Add the rinse to the new T-175 and incubate
for 1 h at 37 C.
3.3 Cryopreservation 1. Place freezing medium on ice.

2. Transfer cell media from the T-175 flask to a 50 mL conical
tube. Collect 10 μL of cell suspension for cell counting using
Trypan blue (see Note 3).
3. Centrifuge at 410 rcf for 5 min. While centrifuging, count cells
that have been exposed to Trypan blue.
4. Aspirate culture media and resuspend cells in freezing medium
at a concentration of 2–4 million cells per mL.
5. Place 1 mL of cell suspension in cryogenic vial then place vials
in a controlled cryo-freezing container, and place the container
in a 80 C freezer overnight.
6. Within 1–2 days transfer vials to a liquid nitrogen freezer.
3.4 Thawing Cells 1. Coat a culture dish with fibronectin by dissolving 250 μL of
fibronectin into 9.75 mL water. Add enough solution to cover
culture plate and incubate at 37 C for at least 30 min.
2. In four 15 mL conical tubes, make the DMSO/media mixtures
as follows by adding DMSO to 20 % Media:
(a) 5 %—9.5 mL media with 500 μL DMSO.
(b) 2.5 %—9.75 mL media with 250 μL DMSO.
(c) 0 %—10 mL media (make two of these).
3. Warm media in 37 C water bath.
4. Remove cells from liquid nitrogen and place on dry ice.
5. When the media mixtures are warm, thaw cells in 37 C water bath.
6. Transfer contents of cryogenic vial to 5 % DMSO/media tube.
Centrifuge at 410 rcf for 8 min. Aspirate supernatant and leave
the cell pellet.
7. Add contents of 2.5 % DMSO/media tube to cell pellet and

gently resuspend the cells. Centrifuge at 410 rcf for 8 min.
Aspirate supernatant and leave the cell pellet.
8. Add contents of one 0 % DMSO/media tube to cell pellet and
resuspend. Centrifuge at 410 rcf for 8 min. Aspirate superna-
tant and leave the cell pellet.
9. Add contents of second 0 % DMSO/media to the cell pellet and
gently resuspend. Take a small sample for cell counting. Centrifuge
at 410 rcf for 8 min. While centrifuging count cells in a hemocy-
tometer using Trypan. Record cell amount as well as viability.
10. Aspirate supernatant and leave the cell pellet. Resuspend in 10 %
NRCM media and plate cells on fibronectin-coated dishes.
Optional: add 100 μM BrdU (bromodeoxyuridine) (see Note 4).
3.5 Cell Culture 1. Cells are plated at about 100 k cells per cm2. The plating
density can be adjusted based on the intended use of the
culture. The following is a timeline of a typical culture.
2. Day 1—rinse dead cells away with warm IMDM add 10 %
NRCM (optionally containing 100 μM BrdU).
3. Day 2—rinse dead cells away with warm IMDM add 10 %
NRCM.
4. Day 3—add 2 % NRCM.
5. Day 4—add 2 % NRCM.
6. Day 5—cells should be confluent and beating (see Note 5)
(Fig. 1).
Fig. 1 Bright-field image of NRCMs at day 5 of culture

4 Notes
1. Media can be stored at 4 C for up to 3 weeks. This is scalable

for smaller or larger volumes.
2. To best maintain sterility, this is can be done by another person
who then places the cleaned pup in the hood.
3. Trypan blue is commonly used to label dead cells. Mix the 10 μL
sample of cell suspension with 10 μL of trypan blue, and then
add 10 μL of this mixture to a hemocytometer for counting.
4. Cardiac fibroblasts can become problematic by proliferating
faster than the NRCMs. BrdU can be added to reduce the
proliferation of the fibroblasts [12]. The point at which the
media switches from 10 to 2 % FBS can be done earlier to help
with this problem as well. Percoll gradients may also be used to
remove fibroblasts [13, 14]. See Fig. 2.
5. NRCM purity can be assayed by immunocytochemistry. A com-
mon marker used to identify NRCMs is α-sarcomeric actinin.
Fig. 2 Fluorescence staining of α-sarcomeric actinin (green) and nuclei (blue) shows numerous NRCMs
present with minimal fibroblasts
Acknowledgement
This work was supported by funding from American Heart Associ-

ation 12BGIA12040477, NC State University Chancellor’s Fac-
ulty Excellence Program, and National Natural Science Foundation
of China H020381370216.
References
1. Louch WE, Sheehan KA, Wolska BM (2011) artid¼3667201&tool¼pmcentrez&rendertype¼
Methods in cardiomyocyte isolation, culture, and abstract. Accessed 8 Aug 2014
gene transfer. J Mol Cell Cardiol 51:288–298, 8. Uchida T et al (2011) Optimal temperature
Available at: http://www.pubmedcentral.nih. range for low-temperature preservation of dis-
gov/articlerender.fcgi?artid¼3164875&tool¼ sociated neonatal rat cardiomyocytes. Cryobiol-
pmcentrez&rendertype¼abstract. Accessed 21 ogy 63:279–84, Available at: http://www.ncbi.
Oct 2013 nlm.nih.gov/pubmed/22005593. Accessed 12
2. Miragoli M, Salvarani N, Rohr S (2007) Myo- Feb 2014
fibroblasts induce ectopic activity in cardiac 9. Miyamura K et al (2010) Evaluation of viability
tissue. Circ Res 101:755–758 of cryopreserved rat cardiac myocyte. Cryobi-
3. Simpson P, McGrath A, Savion S (1982) Myo- ology 56:111–117
cyte hypertrophy in neonatal rat heart cultures 10. Rana P, Anson B, Engle S, Will Y (2012) Char-
and its regulation by serum and by catechola- acterization of human-induced pluripotent stem
mines. Circ Res 51:787–801, Available at: cell-derived cardiomyocytes: bioenergetics and
http://circres.ahajournals.org/cgi/doi/10. utilization in safety screening. Toxicol Sci
1161/01.RES.51.6.787. Accessed 6 Mar 130:117–31, Available at: http://www.ncbi.
2014 nlm.nih.gov/pubmed/22843568. Accessed 15
4. Rohr S, Fl€ uckiger-Labrada R, Kucera JP July 2014
(2003) Photolithographically defined deposi- 11. Yokomuro H, Mickle DA, Weisel RD, Li RK
tion of attachment factors as a versatile method (2003) Optimal conditions for heart cell cryo-
for patterning the growth of different cell types preservation for transplantation. Mol Cell Bio-
in culture. Pfl€ ugers Arch 446:125–32, chem 242:109–14, Available at: http://www.
Available at: http://www.ncbi.nlm.nih.gov/ ncbi.nlm.nih.gov/pubmed/12619872
pubmed/12690471. Accessed 26 Aug 2014 12. Simpson P, Savion S (1982) Differentiation of
5. Fu J-D et al (2006) Crucial role of the sarco- rat myocytes in single cell cultures with and
plasmic reticulum in the developmental regula- without proliferating nonmyocardial cells.
tion of Ca2+ transients and contraction in Cross-striations, ultrastructure, and chronotro-
cardiomyocytes derived from embryonic stem pic response to isoproterenol. Circ Res
cells. FASEB J 20:181–3, Available at: http:// 50:101–116, Available at: http://circres.
www.ncbi.nlm.nih.gov/pubmed/16249315 ahajournals.org/cgi/doi/10.1161/01.RES.
6. Liu J, Fu JD, Siu CW, Li RA (2007) Functional 50.1.101. Accessed 20 Mar 2014
sarcoplasmic reticulum for calcium handling of 13. Evans HJ, Goodwin RL (2007) Western array
human embryonic stem cell-derived cardio- analysis of cell cycle protein changes during the
myocytes: insights for driven maturation. hyperplastic to hypertrophic transition in heart
Stem Cells 25:3038–44, Available at: http:// development. Mol Cell Biochem 303:189–99,
www.ncbi.nlm.nih.gov/pubmed/17872499. Available at: http://www.ncbi.nlm.nih.gov/
Accessed 31 July 2014 pubmed/17457520. Accessed 18 Aug 2014
7. Knollmann BC (2013) Induced pluripotent 14. Golden HB et al. (2012) Isolation of Cardiac
stem cell-derived cardiomyocytes: boutique Myocytes and Fibroblasts from Neonatal Rat
science or valuable arrhythmia model? Circ Pups. In: Peng X, Antonyak M (eds) Cardiovas-
Res 112:969–976, Available at: http://www. cular Development. Methods in Molecular Biol-
pubmedcentral.nih.gov/articlerender.fcgi? ogy, vol 843. Springer, New York, pp 205–214
Chapter 13
Evaluation of Sarcomeric Organization in Human

Pluripotent Stem Cell-Derived Cardiomyocytes
Chrishan J.A. Ramachandra and Winston Shim
Abstract
High-resolution optical imaging provides valuable window in examining transitional and systemic changes in
cellular processes. The relative spatial relationship of structural, transport, and signaling proteins, surface
antigens, and transcription factors may reveal developmental state of the cellular system. Here, we describe the
use of confocal microscopy to evaluate the organization of sarcomeric structural proteins, sarcoplasmic channel
proteins, and cardiac transcription factors in human pluripotent stem cell (PSC)-derived cardiomyocytes.
Key words Cardiomyocytes, Sarcomeres, Cardiac striations, Cardiac progenitors, Confocal

microscopy
1 Introduction
Study of interaction of microstructures within a cell or tissue is

important to the understanding of cellular processes. Laser
confocal-based microscopy supports high-resolution optical imag-
ing that affords exceptional insights into the spatial relationship of
such interactive processes. The ability to obtain optical slices from
the studied specimen further provides a three-dimensional appreci-
ation of cellular physiology.
Human pluripotent stem cell (PSC) derived cardiomyocytes
have tremendous potential in regenerative medicine, disease model-
ing, and drug discovery [1, 2]. However, major question surround-
ing their cellular maturity required for such applications has not been
addressed adequately. While electrophysiological studies involving
whole-cell patch-clamp and multi-electrode arrays (MEA) reveal
cellular maturation based on membrane potentials and electrical
properties, confocal imaging may provide further confirmation of
the structural maturity of the studied cardiomyocytes [3].
Sarcomeres and sarcomeric organization—the ability of cardiac
structural proteins to stoichiometrically align into interdigitating
thick and thin myofilaments in an organized cross-striated pattern
161
162 Chrishan J.A. Ramachandra and Winston Shim
of sarcomeres is one of the key characteristics of maturation in

cardiomyocytes. However, such sarcomeric architecture may trans-
verse across multiple viewing planes of a specimen, which rendering
optical slicing technique well suited for such studies.
Unlike other cell types, cardiomyocytes are more susceptible to
environmental changes and physical manipulation, and therefore,
preparation of specimen is crucial in obtaining high-quality images
depicting sarcomeric organizations. In this protocol, we describe
how to prepare a PSC-derived cardiomyocyte specimen for confo-
cal imaging, with emphasis on maintaining sarcomeric integrity as
well as discerning maturity of cardiomyocytes based on their sarco-
meric organization.
2 Materials
2.1 Buffer 1. Wash solution: 1 phosphate buffered saline (PBS). Add

Compositions 100 ml 10 PBS (Gibco, Life Technologies) to 900 ml auto-
claved DI water. Store at 4 C.
2. Fixative solution: 4 % paraformaldehyde (PFA), 1 PBS, pH 7.4.
Dissolve 4 g PFA in 100 ml PBS (see Note 1). Filter using a
sterile low-protein binding 0.45 μM filter. Aliquot and store at
20 C. Working solution can be stored at 4 C in the dark.
3. Permeabilizing solution: 0.1 % Triton X-100, 1 PBS. Add
0.1 ml Triton X-100–100 ml PBS (see Note 2). Store at 4 C.
4. Blocking solution: 5 % bovine serum albumin (BSA), 1 PBS.
Dissolve 5 g BSA in 100 ml PBS (see Note 3). Filter using
a sterile low-protein binding 0.45 μM filter. Store at 4 C.
5. Antibody wash solution: 0.1 % Tween 20, 1 PBS. Add 0.1 ml
Tween 20–100 ml PBS (see Note 2). Store at room temperature.
6. Antibody diluent: 1 % BSA, 1 PBS. Dilute 1 ml blocking
solution (5 % BSA, PBS) in 4 ml PBS. Store at 4 C (see Note 3).
2.2 Imaging 1. Chamber slides: 8-well (Lab-Tek, Nunc).

Components 2. Mounting solution: 70 % glycerol. Add 7 ml glycerol to 3 ml
autoclaved DI water. Store at 4 C in the dark.
3. Coverslips: 22 60 mm, #1.5 thickness (Biomedia).
4. Imaging system: LSM 710 confocal microscope (Zeiss), Argon
laser, HeNe laser, NLO laser.
5. Immersion oil: Immersol 518 F fluorescence free (Zeiss).
2.3 Antigens 1. Primary antibodies: anti-mouse α-actinin (Sigma); anti-mouse

and Conjugates titin (Sigma); anti-mouse SERCA2A (Sigma); anti-mouse
MLC2v (Synaptic Systems); anti-mouse MLC2a (Synaptic
Systems); anti-mouse cardiac troponin T (US Biological);
anti-rabbit Nkx2.5 (Novus Biologicals).
Evaluation of Sarcomeric Organization in Human Pluripotent Stem. . . 163
2. Secondary antibodies: Alexa Fluor® 488 anti-mouse (Life

Technologies); Alexa Fluor® 555 anti-mouse (Life Technolo-
gies); Alexa Fluor® 555 anti-rabbit (Life Technologies).
3. Nuclear stain: 0.5 μg/ ml DAPI, 1 PBS. Dissolve 1 mg DAPI
(Sigma) in 1 ml autoclaved DI water to obtain a stock solution
of 1 mg/ml. Dilute 5 μl DAPI in 10 ml PBS. Working solution
can be stored at 4 C in the dark.
3 Methods
3.1 Specimen 1. Coat the chamber slide overnight with 0.1 % gelatin and
Preparation remove the following day (see Note 4).
2. Dissociate approximately 5–7 contracting cardiomyocyte
clusters (per chamber) and seed cells in the chambers
(see Note 5).
3. When cells have attached, remove the culture media and rinse
the chambers twice with 0.3 ml wash solution to remove non-
viable cells and debris (see Note 6).
4. Add 0.2 ml fixative solution to each chamber and incubate the
slide at 4 C for 20 min.
5. Remove fixative solution and rinse the chambers twice with
0.3 ml wash solution.
6. Add 0.2 ml permeabilizing solution to each chamber and incu-
bate the slide at 4 C for 10 min (see Note 7).
7. Remove permeabilizing solution and rinse the chambers twice
with 0.3 ml wash solution.
8. Add 0.2 ml blocking solution to each chamber and incubate the
slide at 4 C for 1 h.
9. Dilute primary antibodies in antibody diluent. All primary
antibodies can be used at a 1:200 dilution (see Note 8).
10. Remove blocking solution and add 0.2 ml antibody diluent
containing the respective primary antibodies to the chambers.
Incubate the slide at 4 C overnight.
11. Remove antibody diluent and rinse the chambers twice with
0.3 ml antibody wash solution (see Note 9).
12. Dilute secondary antibodies in antibody diluent. All secondary
antibodies can be used at a 1:400 dilution (see Note 8).
13. Add 0.2 ml antibody diluent containing the respective second-
ary antibodies to the chambers. Incubate the slide at 4 C in the
dark for 1 h.
14. Remove antibody diluent and rinse the chambers thrice with
0.3 ml antibody wash solution (see Note 9).
15. To counterstain cells, add 0.1 ml nuclear stain to each chamber

and incubate the slide at room temperature in the dark for 5 min.
16. Remove nuclear stain and rinse the chambers thrice with 0.3 ml
wash solution (see Note 9).
3.2 Specimen 1. Break off the chambers and add mounting solution on to the
Mounting and Imaging slide in a dropwise manner (see Note 10).
2. Carefully lay down a coverslip (see Note 11).
3. Once the coverslip is set firmly on the slide (approximately 1 h at
room temperature), seal the edges of the coverslip with nail varnish.
4. Mount slide on the confocal microscope stage and capture
images using a 10, 20, 40, or 63 objective lens (see
Note 12).
5. Blue, green, and red fluorescence can be detected at wave-
lengths of 350, 488, and 543, respectively.
6. For high-resolution images, immersion oil should be applied
when using 40 and 63 objective lenses (Fig. 1).
7. Store the slide at 4 C in the dark.
Fig. 1 Sarcomeric organization of human pluripotent stem cell-derived cardiomyocytes. High-resolution

images depicting sarcomeric structural proteins (a–e), sarcoplasmic channel proteins (f) and cardiac tran-
scription factors (a). Note that immature cardiomyocytes (yellow arrows) are positive for Nkx2.5, but negative
for cTnT (a) and as a result lack any sort of sarcomeric organization. Immature cardiomyocytes also express
MLC2a in the nucleus (b) as it has yet to be incorporated into a functional sarcomere. Immature cardiomyo-
cytes (yellow arrows) display poor polymerization of α-actinin (c) (Scale bar, 50 μM)
Evaluation of Sarcomeric Organization in Human Pluripotent Stem. . . 165
4 Notes
1. As PFA is toxic, the fixative solution should be prepared inside a

fume hood. Once PFA has been added to PBS, heat the solu-
tion to 60 C and stir. Care should be taken to avoid boiling.
1 N NaOH can be added dropwise to help dissolve PFA.
The solution should lose its cloudy appearance and turn color-
less. If the pH has risen following the addition of NaOH, dilute
HCL can be added dropwise to bring the pH down to 7.4.
2. Both Triton X-100 and Tween 20 are viscous solutions. Gently
pipette out the respective volumes to limit pipetting error and
eliminate air bubbles. Pipette tips could be cut at the ends to
increase the bore size which would also help in pipetting these
solutions. Once added to PBS, both solutions settle at the base
of the container. Gently invert the container several times to
ensure thorough mixing whilst avoiding frothing.
3. BSA dissolves in PBS slowly. Using a vortex to speed up the
process would result in frothing and should be avoided. Gently
invert the container several times to ensure thorough mixing.
Also, solutions containing BSA (blocking solution and anti-
body diluent) are prone to contamination. Use of aseptic tech-
niques when preparing these solutions is therefore mandatory.
4. Cardiomyocytes can be seeded on various substrates including
Matrigel, fibronectin, and gelatin (determined by end user).
Following the removal of gelatin, it is necessary to allow the
chambers to completely dry prior to seeding. The process could
be hastened by incubating the slide at 37 C for 1 h.
5. Contracting cardiomyocyte clusters comprise varying numbers
of cells. Following their dissociation, seeding cells at low den-
sity (<3 104 cells/ chamber) is ideal as over-confluent would
yield poor images due to the stacking of contracting regions.
6. Following dissociation, cardiomyocytes take >4 days to attach
to the surface of the chambers. Since dissociation disrupts the
sarcomeric organization, ensure that the seeded cells have
regained their contractile property prior to fixing. Fixing non-
contracting cells would yield poor images due to disintegrated
sarcomeric proteins.
7. Prolonging the permeabilization step should be avoided as this
would disintegrate sarcolemmal proteins (cardiac surface
proteins).
8. The dilutions indicated are for antibodies by the listed manu-
facturers. For other antibodies the end user would need to
determine the optimum dilutions. Antibodies from different
hosts could be incubated simultaneously (e.g., anti-mouse
cardiac troponin T/ anti-rabbit Nkx2.5).
9. Following incubation with primary antibodies, all downstream

wash steps should be performed with different pipette tips to
ensure no carryover of antibodies between the chambers.
10. When mounting, care must be taken to not overload the slide
with glycerol. An excess of glycerol would cause the coverslip to
slide, disrupting the cells and their antigens in the process.
11. When laying down a coverslip, if it falls in a crooked manner
refrain from straightening it as this too would disrupt the cells
and their antigens. Once a coverslip is placed on the slide, the
excess glycerol can be wicked using Kimwipes.
12. Whilst transcription factors (Nkx2.5) and sarcoplasmic channel
proteins (SERCA2A) are clearly visible at lower magnification
(10, 20), sarcomeric organization is harder to discern.
Higher magnification (40, 63) is more apt when evaluating
sarcomeric structural proteins (α-actinin, troponin T) (Fig. 1).
Acknowledgements
This work was supported by grants from the Singapore National

Research Foundation (NRF-CRP-2008-002), the Goh Founda-
tion Gift/Duke-NUS Graduate Medical School (GCR/2013/
008 and GCR/2013/001), and the Singapore Biomedical
Research Council (BMRC 13/1/96/686) to W.S.
References
1. Hoekstra M, Mummery CL, Wilde AA et al derived from virus-free induced pluripotent
(2012) Induced pluripotent stem cell derived stem cells. Cardiovasc Res 91:577–586
cardiomyocytes as models for cardiac arrhyth- 3. Mehta A, Chung Y, Sequiera GL et al (2013)
mias. Front Physiol 3:346 Pharmacoelectrophysiology of viral-free induced
2. Mehta A, Chung YY, Ng A et al (2011) Pharma- pluripotent stem cell-derived human cardiomyo-
cological response of human cardiomyocytes cytes. Toxicol Sci 131:458–469
Chapter 14
Electrotonic Coupled Metabolic Purification of Chick

Cardiomyocytes
Winston Shim, Haiyang Yu, K.P. Myu Mai Ja, Muhammad Parasuram,
Kee Pah Lim, and Philip Wong
Abstract
Cardiomyocytes isolated from chick and rodent are widely used in studying cardiac physiology. However,
contaminating non-cardiomyocytes are an inherent problem that hinders downstream analysis. Here, we
report a novel electrical stimulation coupled with metabolic selection method using cytosine arabinoside
(AraC) to efficiently eliminate contaminating cells in isolating chick embryonic cardiomyocytes. Compared
with conventional methods of pre-plating or AraC alone, electrical stimulation coupled with AraC increased
the percentage purity of cardiomyocytes by 2–6-fold with added effect of improved contractile function and
maturation. This simple method could be useful in isolating and maintaining purified cardiomyocytes for
long-term studies of cardiac physiology.
Key words Electrical stimulation, Cytosine arabinoside, Lactate, Purification, Cardiomyocytes
1 Introduction
Cardiomyocyte models established from chick embryonic cardio-

myocytes, neonatal rat cardiomyocytes, and induced pluripotent
stem (iPS) cells-derived cardiomyocytes have been explored to
study cardiac physiology [1–3]. These cardiomyocytes have proven
invaluable in studying cardiac repair and screening cardioactive
drugs [4]. However, purity of the derived cardiomyocytes has
been a perennial issue. Contaminating interstitial cells such as
fibroblasts, smooth muscle cells, and endothelial cells persist and
eliminating them has been technically challenging. One of the most
common purification methods, pre-plating technique, relies on
differential cellular sedimentary rates and disparate affinity of
cardiomyocytes and fibroblasts/endothelial cells towards culture
surfaces [5]. Another method that is based on cellular size is
density-gradient centrifugation whereby cells were segregated in
Kruftbr€uhe solution atop of 20–70 % Percoll gradient [6]. How-
ever, efficacy of these methods is invariably low. Contaminating
167
168 Winston Shim et al.
cells, even though minute, persist to proliferate rapidly and may

overwhelm the non-dividing cardiomyocytes in culture [7]. Alter-
natively, genetic methods are relied on to increase yield and purity
of cardiomyocytes by transfecting aminoglycoside phosphotrans-
ferase cDNA into embryonic stem cell line [8]. However, this
method is time-consuming, costly and potentially genotoxic.
Recently, non-genetic methods have been developed using mito-
chondrial dyes or antibodies that preferentially target cardiomyo-
cytes [9, 10]. Unfortunately lack of specific surface markers and
reliance of fluorescent activated cell sorting restrict its wide
application.
Metabolic selection methods have been developed to eliminate
proliferating cells from cardiomyocyte cultures. Cytosine arabino-
side (AraC), a drug used in the treatment of acute leukemia, has
been used to purify cardiomyocytes [11]. In addition, lactate can be
added to cultures to enrich for cardiomyocytes, based on the
marked biochemical differences in glucose and lactate metabolism
between cardiomyocytes and non-cardiomyocytes [12]. Large-
scale purification of mouse and human iPS-derived cardiomyocytes
using such metabolic requirements has been demonstrated [13].
We further improve this by synergizing metabolic selection with
programmed electrical pacing to eliminate contaminating intersti-
tial cells from crude cardiomyocyte preparations.
2 Materials
2.1 Chick Embryos 1. Fertilized chicken eggs were obtained from the Chew’s Farm
Singapore.
2.2 Culture Medium 1. Cell isolation medium of DMEM (Life Technologies) is

supplemented with 1 penicillin–streptomycin and 20 % fetal
bovine serum (Hyclone).
2. Standard culture medium of DMEM (Life Technologies) is
supplemented with 1 penicillin–streptomycin and 3 % horse
serum (Life Technologies).
3. Dissociation solution of trypsin–EDTA is obtained from Life
Technologies.
2.3 Metabolic 1. Make up 4 mM lactate (Sigma-Aldrich) in DMEM medium

Selection Medium without glucose (Life Technologies) (see Note 1).
2. Make up 10 μM of cytosine arabinoside (AraC) (Sigma-
Aldrich) in standard culture medium of DMEM.
2.4 Electrical 1. Apply pulsed electrical stimulation to the culture through

Stimulation Apparatus C-Pace EP Culture Pacer (IonOptix) with a 6-well C-Dish
electrode (see Note 2).
Electrotonic Coupled Metabolic Purification of Chick Cardiomyocytes 169
2. Electric pulse is generated by a multi-channel stimulus genera-

tor (IonOptix) with adjustable frequency (0.01–99 Hz), pulse
duration (0.4–24 ms) and voltage (up to 40 V).
2.5 Antibody, Dyes 1. Primary antibody from mouse anti-sarcomeric α-actinin

and Solution (Sigma-Aldrich) was used to identify cardiomyocytes.
2. Secondary antibody from goat anti-mouse IgG conjugated
with Alexa Fluor® 488 (Life Technologies) was used to visua-
lize signal.
3. Nuclear counterstain is prepared by dissolving 1 mg 40 ,
6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich) in 1 mL
autoclaved deionized (DI) water to obtain a stock solution of
1 mg/mL. Working solution is made by diluting 5 μL in 10 mL
PBS and store at 4 C in the dark.
4. Propidium iodide solution for live/dead assay is acquired from
Sigma-Aldrich.
5. Fixative of paraformaldehyde solution is made up by diluting
16 % stock solution to 4 % working concentration in
phosphate-buffered saline (PBS) and store at 4 C.
6. Permeabilization solution of Triton X-100 is made up of dilut-
ing 5 % stock solution to 0.05 % working concentration in
phosphate-buffered saline (PBS) and store at 4 C.
7. Coverslip mounting solution: 70 % glycerol. Add 7 mL glycerol
(Sigma-Aldrich) to 3 mL autoclaved DI water. Store at 4 C in
the dark.
3 Methods
3.1 Chick Embryo 1. Carefully isolate and finely mince embryonic chick hearts from
Cardiomyocytes 17-day-old fertilized eggs.
Isolation and Culture 2. Minced heart tissues are digested in 0.1 % trypsin in PBS in
37 C water bath for 10 min (see Note 3) as depicted in a
flowchart in Fig. 1a.
3. Remove supernatant and repeat step 2 twice to remove cardiac
fibroblasts and other interstitial cells.
4. The protease-treated heart tissue is further incubated in 0.05 %
trypsin in 37 C water bath for 8 min to isolate cardiomyocytes.
5. Transfer the supernatant containing cells into ice cold DMEM
high glucose medium containing 20 % fetal bovine serum
(FBS). This step is repeated 5–6 times until the tissue has
been fully digested and obtained a homogenous cell suspension
before passing through a 40 μm nylon mesh cell strainer (BD
Falcon) into a new 50 mL tube.
Fig. 1 Schematic chart of experimental (a) and bright-field images of cell morphology on day 7 after
purification treatment initiation. Scale bar in (b) showed 100 μm
6. Pellet the collected cells by centrifugation for 10 min at

100 g.
7. Resuspend the cells in standard DMEM culture medium con-
taining 3 % horse serum.
8. Plate the isolated cells onto 10-cm tissue culture dish for 2 h
(see Note 4).
9. Transfer the supernatant containing unattached cells and pellet
the cells by centrifugation for 10 min at 100 g.
10. Resuspend the cell pellet in DMEM containing 3 % horse
serum gently.
11. Seed cells in 6-well culture plate with FBS coated coverslips, at
a density of 80 103 cells/cm2.
12. Incubate the cells in a humidified 5 % CO2/95 % air tissue
culture incubator at 37 C.
13. At day 2, when late attaching cells are adhered to the culture
surface, fresh medium is exchanged for cardiomyocytes purifi-
cation step (see Notes 5 and 6).
3.2 Electrical and 1. Exchange standard culture medium with 4 mM lactate or

Metabolic Purification 10 μM AraC supplemented medium to the cells for metabolic
Setup selection (see Note 7).
2. Lower the C-Dish electrodes into the 6-well culture plate
containing isolated cardiomyocyte preparations.
3. Apply an electrical pulse of 0.5 Hz with a 5 V/cm gradient and
12 ms pulse duration continuously for 7 days (see Note 8).
4. Observe the culture daily and exchange with freshly supple-
mented medium every other day until day 7.
3.3 Propidium Iodide 1. At day 7, fresh medium supplemented with 0.5 μM of propi-
(PI) Staining dium iodine dye is added to the electrically paced culture for
live/dead cell assay (see Note 9).
2. After a 30-min incubation at 37 C, cells are washed by PBS for
three times, followed by fixation with 4 % paraformaldehyde
and 0.05 % Triton X-100 in PBS for 10 min.
3. Visualize and quantify the numbers of cells that have taken up
the propidium iodide dye, indicating their compromised cellu-
lar membrane.
4. Capture images using Zeiss Axiovert 200 M fluorescent
microscope.
3.4 Immuno- 1. Add anti-sarcomeric α-actinin antibody at a dilution of 1:100

fluorescent Staining to the paraformaldehyde fixed cells at the end of day 7 for 3 h at
room temperature (see Note 10).
2. Rinse the coverslips with PBS for three times of 5 min each,
with care not to dislodge the cells.
3. Add Alexa Fluor® 488 goat anti-mouse IgG at a dilution of
1:400 for 1 h at room temperature.
4. Rinse the coverslips with PBS for three times of 5 min each,
with care not to dislodge the cells.
5. Add 40 ,6-diamidino-2-phenylindole (DAPI) to the coverslips
to counterstain nuclei.
6. Mount the coverslip with mounting solution onto standard
microscope glass slide for visualization of signals.
7. Capture images using Zeiss LSM 710 confocal microscope.
3.5 Image 1. Captured images are analyzed by ImageJ 1.46r (see Note 11).
Processing and 2. Approximately 100 cells from each group are analyzed and
Statistical Study triplicate experiments are performed.
3. Comparisons at each time point are conducted using analysis of

variance (ANOVA) followed by Tukey’s test, and all data are
presented as mean standard deviation. Differences are con-
sidered statistically significant at p 0.05 (see Note 12).
4 Notes
1. Make up lactate solution in medium without glucose (Life

Technologies) as glucose would affect function of lactate.
2. C-Dish electrode assemblies are kept cleaned and disinfected by
soaking in absolute ethanol for overnight before experiment.
3. A modified method of trypsin digestion is adopted from a
previously described protocol [14].
4. Pre-plating method only eliminates minimal numbers of fibro-
blasts from the crude myocyte preparations. Cells exhibit dis-
parate morphologies from stellate, polygonal, to fibroblastic
after the treatment regimes (Fig. 1b).
5. Day 3 post attachment is optimal in our hand to initiate electrical
stimuli (Fig. 1a), which coincided with previous reports [15, 16].
However, premature electrical stimulation may result in poor
attachment of cardiomyocytes and inhibit expression of sarco-
meric proteins and weaken contractility.
6. Conversely, late application may show no functional benefits in
the culture as was reported previously [16].
7. Optimal concentrations of lactate (4 mM) and AraC (10 μM)
are adopted from previous reports [13, 17].
8. Frequency of electric pulse (0.5 Hz) is selected to mimic the
natural beating frequency of isolated cardiomyocytes observed
in culture. Voltage gradient (10–20 V/cm), which represents
minimal stimulation threshold to elicit pacing-induced contrac-
tions is determined empirically and coincided with the range
reported previously [15].
9. Electrical stimulation in combination with AraC significantly
induces cell death as shown by higher percentage of propidium
iodide positive staining (Fig. 2a). In comparison to control
group, higher percentage of propidium iodide positive staining
indicated that AraC resulted in drastically more cell death
(2.7-fold) that could benefit enrichment of cardiomyocytes,
and the effect was further enhanced by electrical stimulation
(80-fold) (Fig. 2b).
10. Cardiomyocytes in control and electrical stimulation groups
are healthy and have clear sarcomeres (Fig. 3a). In lactate
treated group, cell numbers are reduced, but the cardiomyo-
cytes form disorganized sarcomeric structures as shown in
Fig. 2 Cell viability analysis. (a) Cell viability by propidium iodide (PI) staining on day 7 after purification
treatment initiation. Pink color showed positive staining of dead cells. Scale bar showed 50 μm. (b) Graphic
representation of statistical result of cell viability. *p < 0.01 compared with other groups
Fig. 3a (middle column). Combining AraC with electrical

stimulation suppresses proliferation of other intersititial cells
with few α-actinin negative cells remaining close to healthy
cardiomyocytes with well-developed sarcomeres 7 days after
treatment (Fig. 3a).
11. Image analysis is carried out as described [18].
12. Electrical coupled metabolic selection and AraC augments total
cardiomyocytes recovery from the crude preparation (Fig. 3b),
but only electric+AraC regime significantly enhances purity of
the culture. Purity of cardiomyocytes increased by 6.1-fold in
electric+AraC group and by 2.6-fold compared to control
group and AraC only group respectively (both of p < 0.01)
(Fig. 3c).
Fig. 3 Cardiomyocytes characterization. (a) Showed immunostaining images of sarcomeric α-actinin. Scale
bar showed 50 μm while 20 μm in inserts. Green is α-actinin and blue is DAPI staining. (b) Quantification of
cardiomyocyte yield following treatment regimes by counting α-actinin stained cardiomyocytes/mm2 field of
view. (c) Showed the statistical result of the percentage of cardiomyocytes in culture on day 7 after purification
treatment. *p < 0.01 compared with other groups
Acknowledgements
This study was supported by grants from the National Research

Foundation Singapore (NRF2008-CRP003-02), the Goh Founda-
tion (Duke-NUS GCR/2013/0008, Duke-NUS GCR/2013/
0011) and the Biomedical Research Council Singapore (BMRC
13/1/96/686) to W.S.
References
1. Gregorio CC, Fowler VM (1995) Mechanisms T, Onitsuka T, Shimoji K, Ohno Y, Egashira T,
of thin filament assembly in embryonic chick Kaneda R, Murata M, Hidaka K, Morisaki T,
cardiac myocytes: tropomodulin requires Sasaki E, Suzuki T, Sano M, Makino S, Oikawa
tropomyosin for assembly. J Cell Biol S, Fukuda K (2010) Nongenetic method for
129:683–695 purifying stem cell-derived cardiomyocytes.
2. Muller-Ehmsen J, Whittaker P, Kloner RA, Nat Methods 7:61–66
Dow JS, Sakoda T, Long TI, Laird PW, Kedes 11. Sreenivasan Y, Sarkar A, Manna SK (2003)
L (2002) Survival and development of neonatal Mechanism of cytosine arabinoside-mediated
rat cardiomyocytes transplanted into adult apoptosis: role of Rel A (p65) dephosphoryla-
myocardium. J Mol Cell Cardiol 34:107–116 tion. Oncogene 22:4356–4369
3. Zhang J, Wilson GF, Soerens AG, Koonce CH, 12. Fisher DJ, Heymann MA, Rudolph AM (1981)
Yu J, Palecek SP, Thomson JA, Kamp TJ Myocardial consumption of oxygen and carbo-
(2009) Functional cardiomyocytes derived hydrates in newborn sheep. Pediatr Res
from human induced pluripotent stem cells. 15:843–846
Circ Res 104:e30–e41 13. Tohyama S, Hattori F, Sano M, Hishiki T,
4. Laflamme MA, Murry CE (2011) Heart regen- Nagahata Y, Matsuura T, Hashimoto H,
eration. Nature 473:326–335 Suzuki T, Yamashita H, Satoh Y, Egashira T,
5. Sanger JW, Kang S, Siebrands CC, Freeman N, Seki T, Muraoka N, Yamakawa H, Ohgino Y,
Du A, Wang J, Stout AL, Sanger JM (2005) Tanaka T, Yoichi M, Yuasa S, Murata M, Sue-
How to build a myofibril. J Muscle Res Cell matsu M, Fukuda K (2012) Distinct metabolic
Motil 26:343–354 flow enables large-scale purification of mouse
6. Zhang Y, Li TS, Lee ST, Wawrowsky KA, and human pluripotent stem cell-derived cardi-
Cheng K, Galang G, Malliaras K, Abraham omyocytes. Cell Stem Cell 12:127–137
MR, Wang C, Marban E (2010) Dedifferentia- 14. Meyer T, Boven KH, Gunther E, Fejtl M
tion and proliferation of mammalian cardio- (2004) Micro-electrode arrays in cardiac safety
myocytes. PLoS One 5:e12559 pharmacology: a novel tool to study QT inter-
7. Haddad J, Decker ML, Hsieh LC, Lesch M, val prolongation. Drug Saf 27:763–772
Samarel AM, Decker RS (1988) Attachment 15. Radisic M, Park H, Shing H, Consi T, Schoen
and maintenance of adult rabbit cardiac myo- FJ, Langer R, Freed LE, Vunjak-Novakovic G
cytes in primary cell culture. Am J Physiol 255: (2004) Functional assembly of engineered
C19–C27 myocardium by electrical stimulation of cardiac
8. Fijnvandraat AC, van Ginneken AC, Schuma- myocytes cultured on scaffolds. Proc Natl Acad
cher CA, Boheler KR, Lekanne Deprez RH, Sci U S A 101:18129–18134
Christoffels VM, Moorman AF (2003) Cardi- 16. Tandon N, Cannizzaro C, Chao PH, Maidhof
omyocytes purified from differentiated embry- R, Marsano A, Au HT, Radisic M, Vunjak-
onic stem cells exhibit characteristics of early Novakovic G (2009) Electrical stimulation sys-
chamber myocardium. J Mol Cell Cardiol tems for cardiac tissue engineering. Nat Protoc
35:1461–1472 4:155–173
9. Dubois NC, Craft AM, Sharma P, Elliott DA, 17. Jones SP, Kennedy SW (2009) Chicken
Stanley EG, Elefanty AG, Gramolini A, Keller embryo cardiomyocyte cultures–a new
G (2011) SIRPA is a specific cell-surface approach for studying effects of halogenated
marker for isolating cardiomyocytes derived aromatic hydrocarbons in the avian heart. Tox-
from human pluripotent stem cells. Nat Bio- icol Sci 109:66–74
technol 29:1011–1018 18. Abramoff MD, Magalhaes PJ, Ram SJ (2004)
10. Hattori F, Chen H, Yamashita H, Tohyama S, Image processing with ImageJ. Biophoton Int
Satoh YS, Yuasa S, Li W, Yamakawa H, Tanaka 11:36–42
Chapter 15
Gene Transfer into Cardiac Myocytes

Sarah E. Lang and Margaret V. Westfall
Abstract
Traditional methods for DNA transfection are often inefficient and toxic for terminally differentiated cells,
such as cardiac myocytes. Vector-based gene transfer is an efficient approach for introducing exogenous
cDNA into these types of primary cell cultures. In this chapter, separate protocols for adult rat cardiac
myocyte isolation and gene transfer with recombinant adenovirus are provided and are routinely utilized for
studying the effects of sarcomeric proteins on myofilament function.
Key words Adenovirus, Gene transfer, Cardiomyocytes, Homologous recombination, Cardiac,

Myocyte isolation
1 Introduction
The ability to introduce cDNA into primary cell culture is often

essential for identifying the structural and/or functional role of a
protein [1]. Genetic animal models are ultimately desirable for
gaining insight into the role one or more proteins play in modulat-
ing or regulating cardiac pump performance. However, studies in
primary adult myocytes can effectively bridge the gap between
biochemical/molecular experiments and studies in more complex
in vivo models [1–3]. In many organ systems, immortalized cells
are available to help bridge these gaps. Cardiac cell lines, such as
HL-1 cells, are available and are usually amenable to typical trans-
fection approaches for gene transfer, such as lipofectamine [4–6].
However, currently available cardiac cell lines are unable to
completely recapitulate the terminally differentiated, rod-shaped
morphology and contractile function produced by adult myocytes.
As a result, isolated adult myocytes continue to be the most utilized
model for cellular studies, particularly for work related to cardiac
contractile performance [7, 8].
Single cell work in adult cardiac myocytes has provided signifi-
cant insights into Ca2+ cycling and the role played by individual
sarcomeric proteins in modulating contractile function [2, 7, 9].
177
178 Sarah E. Lang and Margaret V. Westfall
Protein expression and the subsequent functional response can be

measured over several hours to days in cells, making isolated myo-
cytes a cost-effective approach for studying cardiac function [1, 7, 9].
In contrast to cell lines, transfection of terminally differentiated adult
cardiac myocyte is often toxic [9–11]. Additional approaches, such as
poly-L-ornithine, have proven to be less toxic but are inefficient and
fail to produce adequate protein expression [9]. Recombinant virus
is an alternative approach for efficient and nontoxic gene transfer into
many cells, including adult myocytes.
Multiple types of virus are available for gene transfer, including
both integrative and non-integrative vectors [9, 10, 12–16].
Single-stranded, RNA virus, such as the popular lentiviral vector,
integrate into the host genome and result in germ-line transmission
[12, 13, 17, 18]. Lentiviral constructs are capable of gene transfer
into many cell types and can produce persistent global or tissue-
specific transgene expression [19–22], making lentivirus a desirable
tool for long-term and/or in vivo studies. Lentiviral integration is
stochastic and can result in insertional mutagenesis if positioned
adjacent to an oncogene, which requires consideration when using
this vector for in vivo gene delivery [12, 13, 18]. Efficient and
sustained long-term expression also is achieved with other integrat-
ing vectors, such as adeno-associated virus (AAV). Rather than
RNA, AAV contains single stranded DNA and it is increasingly
utilized for gene therapy [23]. A portion of the AAV genome
integrates site-specifically on chromosome 19 and as episomal gen-
omes. Recombinant AAV vectors also can be constructed to pro-
duce gene targeting via homologous recombination [24, 25].
However, protein production for both lentiviral and AAV vector-
mediated gene transfer is preceded by a latent or lag phase due to
viral integration [23, 26]. This latency can be a disadvantage for
shorter-term cell culture studies in isolated cells. Isolated adult
myocyte studies are typically performed within a week of isolation,
as these cells are cultured in serum-free media to maintain the adult,
rod-shaped morphology. Clearly, the specific characteristics of each
vector are important to consider for optimal in vivo and in vitro
experiments.
The linear, non-integrative and double-stranded recombinant
adenovirus is an alternative vector used for gene transfer. This
vector is ideal for cellular studies because it produces rapid and
efficient gene expression in a large number of cell types due to the
ubiquitous expression of the receptor required for viral entry
[27–29]. In primary cultures of cardiac myocytes, adenoviral-
mediated gene transfer produces rapid and efficient gene transfer
and expression. Thus, gene transfer using recombinant adenoviral
constructs has produced substantial insights into cardiac cellular
function, including myofilament function. Most importantly,
insights into contractile performance can usually be achieved in
intact myocytes without vector-dependent disruption of sarcomere
Gene Transfer into Cardiac Myocytes 179
stoichiometry, myofilament integrity, or contractile function [2, 9,

30–36]. A limiting factor for in vivo gene transfer is the immune
response to adenovirus in animal models, which generally prevents
transgene expression beyond 2 weeks post-gene transfer [11, 37].
Our laboratory routinely utilizes recombinant adenoviral-
mediated gene transfer into isolated adult rat cardiac myocytes.
The two plasmid Cre-loxP system available from Microbix (Tor-
onto, CA) can generate the high titer recombinant adenoviral con-
structs used for studies in myocytes, and is described elsewhere [2,
11, 38, 39]. This chapter outlines our current protocol for the
isolation of Ca2+-tolerant, rod-shaped adult rat myocytes and the
use of high titer recombinant adenovirus for gene transfer into
these cells. A brief section also provides information about cultur-
ing these cells in serum-free media for 5–7 days [2, 9]. The protocol
is adapted from the method developed by Haworth et al. to pro-
duce 1–2 106 cardiac myocytes [40]. These protocols are opti-
mized for adult rat hearts and require adaptation for myocytes
isolated from other species [35, 41].
2 Materials
All solutions should be prepared with purified, deionized water

(dH2O; 18 Ω).
2.1 Reagents and 1. Perfusion setup: A Baker apparatus (Harvard “Baker” perfu-
Supplies for Isolation sion set #50-8382) is used for this protocol and has a double-
of Adult Rat Cardiac barrel warming coil with an upper, changeover stopcock to
Myocyte regulate flow from syringes containing perfusion buffers
(syringe #1: KHB + Ca2+, Subheading 2.1, item 17; syringe
#2: KHB-Ca2+, Subheading 2.1, item 16) (see Note 1). Two
50 mL glass syringes are attached to the warming coil by tubing
and are positioned to achieve a 70 mmHg gravimetric pressure
difference between the syringe and the heart. Unless otherwise
noted, all tubing used for the Baker apparatus is Tygon® tub-
ing. Two pieces of tubing also are connected to the top of the
warming coil with the opened ends positioned at the same
height as the syringes to serve as bubble traps. Two additional
pieces of tubing are attached to the sides of the warming coil to
adjust and drain buffers. Stopcocks are attached to the end of
each set of drain tubing and these stopcocks are closed in
between cell isolations. A final piece of tubing is inserted into
the bottom of the warming chamber and connected to a sec-
ond, smaller stopcock (i.e., the lower stopcock) to hold the
heart mounted to a tubing adapter. The Baker warming cham-
ber temperature is kept at 37 C by constant water flow from a
circulating water bath through a water inlet and outlet within
the warming chamber. A small piece of PE-20 tubing
connected to Tygon® tubing is used to collect effluent from the

heart, returning it to syringe #2 using a peristaltic pump as a
means of continuously reperfusing the heart during the diges-
tion phase of the perfusion [2].
2. The surgical tools for this procedure include: one pair Mayo-
Stille scissors, one standard pattern forceps, one pair of fine
dissection scissors, two pairs of Dumont #4 forceps, a Bellco
Cellector tissue sieve (mesh size: 230 μm), and a 16-G tubing
adapter (i.e., cannula). Surgical tools should be autoclaved
prior to the myocyte isolation and handled to maintain sterility
during the cell isolation procedure.
3. Sterile glassware used for myocyte isolations include: 3–50 mL
glass beakers, 1–25 mL glass beaker, 1–100 mL glass beaker.
4. Silanized trituration pipets (two pipets—one large bore, one
smaller bore). These pipets are prepared by marking a Pasteur
pipet with a diamond knife, cleanly breaking off the stem, and
fire polishing the glass prior to their first use. Trituration pipets
are washed in 7 soap, rinsed thoroughly with dH2O, silanized
in Sigmacote®, and then allowed to dry. Each pipet is auto-
claved after drying the silanized coating.
5. Sterile dH2O: Filter-sterilize 1 L dH2O. Unless otherwise
noted, filter-sterilized solutions described below are sterilized
using a 0.22 μm bottle-top filter and stored at room tempera-
ture for up to 1 month (see Note 2).
6. Heparin: Heparin sodium injection (1,000 USP units/mL).
7. Nembutal: Pentobarbital sodium (50 mg/mL).
8. 0.99 % saline (NaCl) solution: Dissolve 0.1485 g NaCl in
15 mL dH2O and filter-sterilize.
9. 70 % ethyl alcohol: Dilute 700 mL ethyl alcohol (200 proof)
with dH2O to a final volume of 1 L.
10. 50 U/mL Penicillin–50 μg/mL Streptomycin (P/S) + sterile
dH2O: Add 10 mL P/S into dH2O, bring to a final volume of
1 L, and filter-sterilize.
11. 1 M NaCl: Dissolve 58.44 g NaCl in a final volume of 1 L
dH2O and filter-sterilize.
12. 0.5 M KCl: Dissolve 37.27 g KCl in a final volume of 1 L dH2O
and filter-sterilize.
13. 0.5 M KH2PO4: Dissolve 68.05 g KH2PO4 to a final volume of
1 L with dH2O and filter-sterilize.
14. 100 mM MgSO4 7H2O: Dissolve 24.65 g MgSO4 7H2O
to a final volume of 1 L with dH2O and filter-sterilize.
15. 100 mM CaCl2 2H2O: Dissolve 14.70 g CaCl2 2H2O in
a final volume of 1 L with dH2O, and filter-sterilize.
16. Ca2+-Free Krebs–Henseleit Buffer (KHB-Ca2+): Mix

118.0 mL 1 M NaCl, 9.60 mL 0.5 M KCl, 2.40 mL 0.5 M
KH2PO4, and 12.0 mL 100 mM MgSO4 7 dH2O in
~600 mL dH20. Add 6.50 g HEPES buffer and 1.98 g Glu-
cose. Adjust the pH to 7.4 with 1 M HCl and a final volume of
1 L with dH2O. The solution is filter-sterilized and stored at
4 C for a maximum of 2 weeks (see Note 3).
17. KHB with Ca2+ (KHB + Ca2+): Prepare KHB as described in
Subheading 2.1, item 16 and add 10 mL 100 mM CaCl2 2
H2O to the solution prior to the addition of 6.50 g HEPES
buffer and 1.98 g Glucose. Adjust the pH to 7.4, bring the final
volume of the solution to 1 L, filter-sterilize, and store the
solution at 4 C for a maximum of 2 weeks.
18. Hyaluronidase (Type IV-S; bovine testes, embryo tested):
Thaw hyaluronidase on ice and resuspend to 10 mg/mL with
autoclaved dH2O. Aliquot hyaluronidase into prechilled
0.5 mL microfuge tubes and store at 20 C. Avoid freeze-
thawing the hyaluronidase aliquots.
19. Collagenase, Type 2: Store lyophilized collagenase in a sealed
and desiccated container at 4 C. Calculate the amount of
collagenase needed for cardiac myocyte isolation based on the
specific lot of collagenase, type 2 (see Note 4).
20. Digestion solution: Add hyaluronidase (300 μg/mL; Subhead-
ing 2.1, item 18) and fresh collagenase (75–90 U/mL; Sub-
heading 2.1, item 19) to 25 mL of KHB-Ca2+ (Subheading 2.1,
item 16) in a sterile 50 mL beaker. The digestion solution
should be made with fresh collagenase for each cell isolation
procedure. Store solution in a 37 C humidified, 5 % CO2
incubator for a maximum of 3 h prior to use.
2.2 Reagents for 1. Laminin (mouse): Store laminin at 80 C and thaw on ice just
Gene Transfer and Cell prior to dilution in sterile phosphate buffered saline (1 PBS).
Culture Laminin is diluted to a concentration of 40 μg/mL, aliquoted
into one-use prechilled microfuge tubes, and stored at 20 C
for up to 6 months.
2. Cell equilibration media (KHB + Ca2+ and 2 % Bovine Serum
Albumin [BSA]): Dissolve 2 g of BSA (Fraction V, heat-shocked,
fatty acid free) in 100 mL KHB + Ca2+ (Subheading 2.1, item
17). Filter-sterilize media and store at 4 C for up to 2 weeks.
3. DMEM + P/S: Add 5 mL P/S (Subheading 2.1, item 10) to
500 mL Dulbecco’s modified Eagle’s medium (DMEM; high
glucose with L-glutamine), filter-sterilize, and store at 4 C for
up to 2 weeks.
4. DMEM + 10%FBS + P/S: Add 10 mL fetal bovine serum
(FBS; premium) to 90 mL DMEM + P/S (Subheading 2.2,
item 3). Filter-sterilize media and store at 4 C for up to
2 weeks.
5. M199 + P/S: Mix M199 (1; HEPES modification with Ear-

le’s salts and L-glutamine) with 3.073 g L-glutathione, 200 mg
BSA (Fraction V, heat shock, fatty acid free), 2.2 g NaH2CO2,
and 2.6 g HEPES buffer in dH2O. Adjust the pH to 7.4 with
1 M NaOH, add 10 mL of P/S and then bring the volume to
1 L with dH2O. Filter-sterilize the media into two, sterile
500 mL bottles, and store at 4 C for up to 2 weeks.
3 Methods
3.1 Isolation of Adult 1. KHB-Ca2+ (Subheading 2.1, item 16) and KHB + Ca2+
Cardiac Myocytes from (Subheading 2.1, item 17) should be pre-warmed at 37 C
Rat Hearts prior to use.
3.1.1 Preparing for 2. All “open” tubes should be covered tightly with Parafilm
Myocyte Isolation between cell isolations. Just prior to each cell isolation, remove
Parafilm from all openings, place Pyrex gas dispersion tube
(e.g., oxygenator) in syringe #1 (KHB + Ca2+ syringe) and
adjust delivery of gas (95 % O2, 5 % CO2) to achieve a gentle
release of O2.
3. Turn on the water bath to begin heating the Baker perfusion
manifold to 37 C. Wash the perfusion apparatus, syringes #1
and #2 and all tubing with 70 % ethyl alcohol (Subheading 2.1,
item 9), then sterile dH2O (x2) (Subheading 2.1, item 5), and
finally sterile dH2O with P/S (Subheading 2.1, item 10).
Ensure both the primary tubing and the bubble traps and
drains are treated with ethyl alcohol and then thoroughly
rinsed with dH2O to remove all traces of ethyl alcohol prior
to the next step.
4. Turn the changeover stopcock to closed position while keeping
the lower stopcock in an open position. Fill syringe #1 and the
attached tubing with pre-warmed KHB + Ca2+ (Subhead-
ing 2.1, item 17). Fill syringe #2 with pre-warmed
KHB-Ca2+ (Subheading 2.1 item 16) (see Note 1). Remove
all air bubbles in the tubing while flushing the setup with KHB
from each syringe (see Note 3). Once air bubbles are removed,
perform a final flush with KHB + Ca2+. Then, keep the upper
stopcock in the syringe #1 open position. Close the lower
stopcock until the heart is ready to mount. Do not allow the
syringes and tubing to empty once the flushing process is
complete. Oxygenate the KHB + Ca2+ buffer for 10–15 min
prior to heart perfusion.
5. Fill a large bucket with ice loaded to be even with the rim of the
bucket. Place both the top and bottom portions of a cell
culture dish on the ice. Fill each portion with 50 mL KHB +
Ca2+ (Subheading 2.1 item 17) and keep one on ice. Fill a
10 mL syringe with ice-cold KHB + Ca2+, attach the cannula,

and remove air bubbles from the syringe. Place cannula (i.e.
tubing adapter) at a 90o angle above the ice bucket and posi-
tion the syringe at a ~45o angle toward to ice bucket. Angle the
syringe using the tube holder so that the cannula tip is just
below the surface of the KHB + Ca2+ solution in one of the cell
culture dishes. Place two loops of 4–0 surgical silk dipped in
70 % ethyl alcohol (Subheading 2.1, item 9) around the
cannula.
3.1.2 Removing 1. Administer an intraperitoneal (IP) injection of heparin

the Heart (1500U/kg; Subheading 2.1, item 6) to an adult rat
(200–250 g) (see Note 5). Wait 15 min for adequate heparini-
zation, and then inject 150–200 mg/kg Nembutal (Subhead-
ing 2.1, item 7). Prepare the digestion solution
(Subheading 2.1, items 18–20) during this heparinization
period.
2. Sterile gloves should be worn to remove the heart and along
with sterile surgical instruments. Sterilize the abdomen with
70 % ethyl alcohol solution (Subheading 2.1, item 9). Use the
Mayo-Stille scissors to cut open the thorax on either side of the
sternum and move the sternum with the standard pattern
forceps to expose the heart. Holding the heart gently between
the thumb and forefinger, excise the heart with a single cut
above the aortic arch (see Note 6).
3. Immerse the heart in ice-cold KHB + Ca2+ (50 mL beaker)
(Subheading 2.1, item 17) and rinse away blood by gently
massaging the heart.
4. Place the heart in the KHB + Ca2+-containing cell culture dish
(Subheading 2.1, item 17) without the syringe. Gently remove
the fascia around the aorta with Dumont #4 forceps until the
majority of the fascia is removed from the aorta. Carefully
remove any aortic branches from the top of the aorta with a
single cut using fine dissection scissors. Keep the heart
immersed in KHB + Ca2+ during this process. Gently place
aorta on the tip of the cannula without introducing bubbles
into the aorta or heart. Secure the heart to the cannula with
surgical silk (see Note 7). Test the attachment with a gentle
flush from the syringe. The heart should enlarge slightly and
stay attached to the cannula.
5. Turn the lower stopcock to allow the KHB + Ca2+ buffer
(Subheading 2.1, item 17; syringe #1) to flow in a rapid,
dropwise manner. While transferring the heart to the perfusion
apparatus, gently and continuously deliver KHB + Ca2+ via the
syringe. Top off the cannula with KHB + Ca2+ prior to remov-
ing the syringe and attaching the cannula/heart to the lower
stopcock on the perfusion apparatus.
3.1.3 Retrograde 1. Perfuse the heart for 5 min with KHB + Ca2+ (Subheading 2.1,
Perfusion of the Heart item 17; syringe #1). The perfusion rate during this time
should be 6–10 mL/min (see Note 8).
2. After 3 min, transfer the oxygenator to syringe #2 containing
KHB-Ca2+ (Subheading 2.1, item 16), then wait an additional
2 min to turn the changeover stopcock to the open position for
syringe #2 (KHB-Ca2+). Perfuse the heart with KHB-Ca2+ for
5 min (see Note 9). After 3 min of perfusion with KHB-Ca2+,
adjust the syringe volume to 35 mL with KHB-Ca2+ and add all
of the digestion solution (Subheading 2.1, item 20) using the
drain tubing (final volume 60 mL). Place the heart in a 25 mL
beaker and allow the perfusate to rise to level where the heart is
partially or fully submerged in buffer. Then, position the tip of
the reperfusion tubing (a small piece of PE-20 tubing) near the
top of the perfusate against the glass interface and away from
the heart. Perfuse the heart in digestion solution for 15–18 min
at a rate of up to 10 mL/min (see Note 9).
3. Add 3 150 μL aliquots of 100 mM CaCl2 (Subheading 2.1,
item 15) to syringe containing the KHB-Ca2+ (Subhead-
ing 2.1, item 16) with digestion solution (syringe #2) every
30 s for 1.5 min. Add 50 μL of 100 mM CaCl2 directly to the
solution surrounding the heart at the end of 1.5 min. Continue
perfusion for an additional 15–18 min (see Note 9).
4. At the end of this interval, turn the changeover stopcock to
stop perfusion, remove the cannula containing the heart from
the apparatus and place the heart in a sterile cell culture dish.
Remove the aorta, any remaining fascia, and the atria from the
heart. Drain all of the digestion solution (Subheading 2.1, item
20) from syringe #2 and the associated tubing into a sterile
100 mL beaker. Then, mince the ventricles into 5–6 pieces with
fine dissection scissors, and place into a sterile 50 mL beaker
containing 15–20 mL of warm digestion solution. Cover the
remaining digestion solution with Parafilm and store in a
37 C, 5 % CO2 incubator. Cover the beaker containing the
minced heart pieces with Parafilm to minimize contamination.
3.1.4 Myocyte Isolation 1. A small piece of PE-20 tubing connected via larger Tygon®
tubing is used to oxygenate the minced ventricle with O2 gas
(95 % O2–5 % CO2). Treat the PE-20 tubing tip with 70 % ethyl
alcohol (Subheading 2.1, item 9), allow tip to air-dry, and then
place tip just inside the beaker adjacent to the Parafilm. Gently
swirl the minced ventricular tissue in a 37 C water bath.
2. After 5 min, collect the supernatant into a sterile Blue Max tube
(15 mL; DB Falcon™) using a silanized triturator to transfer
the solution. Centrifuge in a tabletop clinical centrifuge for 10 s
at 45 g. This supernatant is generally discarded because there
are few rod-shaped cells, but this first digestion is highly vari-
able and should be determined by each lab.
3. Add fresh, warm digestion solution (10–15 mL; Subheading 2.1

item 20) to the remaining ventricular tissue. Repeat the swirling
incubation in water bath at 37 C for 5–10 min (Subheading
3.1.4 # 1), then gently repeat the trituration process. During this
second digestion, the trituration pipet can be used to gently
dissociate tissue (see Note 10). Briefly triturate (<30 s) if tissue
is soft and transfer supernatant to sterile Blue Max tube (15 mL;
DB Falcon™). Add fresh warm digestion solution to remaining
ventricular tissue and store in the 37 C humidified, 5 % CO2
incubator until the next swirling incubation. Collect the super-
natant and centrifuge in a tabletop clinical centrifuge at 45 g
for 10 s to pellet the myocytes.
4. Gently aspirate off the supernatant from the cell pellet in the
blue max tube and gently add 5 mL of warm cell equilibration
media (2 % BSA + KHB + Ca2+; Subheading 2.2 item 2) down
the side of the tube (see Note 11). Resuspend cells in media by
gently inverting the tube and place tube in 37 C incubator.
5. Repeat the previous water bath incubation and trituration steps
(Subheading 3.1.4 #3–4). Collect supernatant and centrifuge,
remove supernatant from the tube and resuspend the cell pellet
in 5 mL cell equilibration media (Subheading 2.2 item 2;
Subheading 3.1.4 #3–4). This digestion process can be
repeated for another 1–2 rounds, until the heart has largely
digested and/or all the digestion solution has been used. Trit-
urate the tissue and begin using a smaller bore triturator as the
tissue is progressively digested.
6. Place the Cellector in a new cell culture dish and pipette the
contents of the beaker through the Cellector. Then, pipette the
solution into a sterile 15 mL Blue Max tube (DB Falcon™) and
pellet the myocytes by centrifugation as described (Subheading
3.1.4 #3).
7. Aspirate supernatant from both tubes. Gently resuspend cell
pellets in warm, fresh cell equilibration media (2 % BSA +
KHB + Ca2+; Subheading 2.2 item 2) and pool aliquots into
a single tube with a final volume of 10 mL or less.
8. Over a 15 min period, add 25 μL of 100 mM CaCl2 (Subhead-
ing 2.1 item 15) every 5 min. Resuspend cells with a gentle
rocking motion after each addition of Ca2+ to bring the final
Ca2+ concentration to 1.8 mM. Store the tube in the 37 C
incubator between Ca2+ additions. After the final Ca2+ addition,
incubate the myocytes at 37 C for 5 min. Then, pellet the cells by
centrifugation (Subheading 3.1.4 #3), and resuspend the pellet
in 3 mL of a 1:1 ratio of DMEM + P/S (Subheading 2.2 item 3)
and DMEM + 10 % FBS + P/S (Subheading 2.2 item 4).
9. During the cell incubation in equilibration buffer, prepare

laminin-coated coverslips. A 22 mm2 glass coverslip is placed
in each well of a sterile flat-bottom, 6-well tissue culture plate,
and UV treated for 10 min in a biosafety cabinet. Add 100 μL
laminin (40 mg/mL; Subheading 2.2 item 1) to the center of
each coverslip. A small drop of laminin is initially added under
the coverslip to affix it to the 6-well plate and the remainder is
pipetted in the middle of each coverslip. Treat laminin coated
coverslips with UV light for an additional 10 min, and cover
until cells are ready to plate.
10. Transfer an aliquot of resuspended myocytes to a cleaned
hemocytometer then count the number of rod-shaped myo-
cytes and total number of myocytes in all nine fields. Calculate
the percentage of live cells:
Rods=Total Cells 100 % ¼ %Viability
Calculate the total amount of rod-shaped myocytes in the
preparation:
Total rod-shaped cells/9 total resuspension volume ¼
number of cells 104
Adjust the volume so the myocyte concentration is
1 105 rod shaped cells/mL using media containing 5 %
FBS (i.e., maintain the 1:1 ratio of DMEM + P/S and
DMEM + 10 % FBS + P/S when adjusting the total volume)
(see Note 12). Store the myocytes in the incubator until ready
for plating.
3.2 Gene Transfer Work with recombinant adenovirus requires prior BL2 approval.
into Rat Cardiac Viral vectors should be handled in a biosafety cabinet and the
Myocytes surface should be treated with 60 % Lysol followed by 70 % ethyl
alcohol with each entry and exit from the cabinet. Personnel
handling vectors should wear safety glasses, gloves, and a lab coat.
The biosafety cabinet should be equipped with a vacuum system
containing tandem Erlenmeyer flasks attached to a 0.1 μm
approved vacuum filter. The Erlenmeyer used for direct collection
of media and a 2 L beaker should have adequate amounts of 10 %
bleach to treat all virus for at least 20 min prior to discarding excess
media. All plasticware and glassware should be treated with 10 %
bleach for 20 min and then disposed of in a biocontainment bag.
1. Aspirate excess laminin from each coverslip (Subheading 3.1.4
#9) and pipette 200 μL of resuspended cells onto the laminin-
coated surface (2 104 cells/ coverslip) (see Note 13).
2. Carefully place the plates into the incubator (37 C; 5 % CO2)
and incubate the cells for 2 h.
3. During the incubation period, dilute the adenoviral stocks in
sterile microfuge tubes to the desired multiplicity of infection
(MOI) with DMEM + P/S (Subheading 2.2 item 3)
(see Notes 14 and 15). Store diluted viruses in the 37 C

incubator until needed.
4. At the end of the 2 h incubation, aspirate the excess media from
each coverslip. Pipette 200 μL of the diluted adenovirus onto
the cells (see Note 14). Return the plates to the incubator for
an additional 1 h.
3.3 Culturing Rat 1. One hour after the addition of virus, add 2 mL M199 + P/S
Cardiac Myocytes (Subheading 2.2 item 5) to each well and return the plates to
the incubator (see Note 16).
2. Replace the media in each well with 2 mL pre-warmed
M199 + P/S (Subheading 2.2 item 5) 24 h after gene transfer.
3. Change the media every other day after this initial media change.
4 Notes
1. Be sure to use the same KHB buffer (+ or Ca2+) for the

duration of the experiment. One suggestion is to mark the
syringe and the corresponding KHB bottle with the same
color tape to avoid confusing the buffers.
2. Stock solutions can be stored at room temperature for up to
1 month. If the solutions become cloudy, make fresh solutions
prior to the myocyte isolation.
3. Older KHB + Ca2+ (Subheading 2.1 item 17) and KHB-Ca2+
(Subheading 2.1 item 16) can be used to rinse the apparatus
after the wash steps (Subheading 3.1.1 #3–4). Fresh KHB
should be used for each myocyte isolation.
4. This protocol is optimized for collagenase Type 2 containing
5750 U collagenase, 12,800 U caseinase, 80.5 U clostripain,
and 4.6 U trypsin per milligram of lyophilized powder.
5. A small volume of Nembutal (Subheading 2.1 item 7; <0.1 cc)
can be added to the heparin (Subheading 2.1 item 6) syringe.
For both injections, bring the volume up to 6–7 cc using 0.9 %
saline solution (Subheading 2.1 item 7).
6. When removing the heart, cut upward to obtain as much of the
aorta as possible.
7. When positioning the heart on the cannula, ensure that the tip
of the cannula is visible through the wall of the aorta. The
cannula should not extend into the ventricle.
8. The heart often beats during perfusion with KHB + Ca2+ (Sub-
heading 2.1 item 17) and this is a sign of good perfusion. A
flow rate greater than 15 mL/min indicates the cannula posi-
tion and/or aortic vessel are compromised during the heart
perfusion.
9. Hearts should no longer beat during the Ca2+-free KHB (Sub-

heading 2.1 item 16) perfusion. As described in note 8, the
perfusion rate should not exceed 15 mL/min. If perfusion
exceeds this rate after addition of the digestion solution, the
heart should be immediately removed and minced as described
in Subheading 3.1.3 #4.
10. When triturating, tissue should not be forced up into the pipet.
Tissue should only be dislodged with a gentle and continuous
up and down motion. If tissue has not digested enough to be
softened, continue to incubate in digestion solution and refrain
from forcing tissue through the pipet tip.
11. Preheat the cell equilibration media (2 % BSA + KHB + Ca2+;
Subheading 2.2 item 2), DMEM + P/S (Subheading 2.2 item
3) and DMEM + 10 % FBS + P/S (Subheading 2.2 item 4)
prior to use with the myocytes.
12. Adjust the volume of media used to resuspend the myocytes
according to the size of the pellet. If the pellet is small, use
1.5 mL DMEM + P/S (Subheading 2.2 item 3) and 1.5 mL
DMEM + 10 % FBS + P/S (Subheading 2.2 item 4) (total
volume ¼ 3 mL). If a large pellet forms, increase the total
volume as necessary. This will aid in myocyte counting.
13. Resuspend the cells by gentle inversion prior to each plate. This
prevents the myocytes from pelleting, allowing for even myo-
cyte application to each coverslip.
14. The gene transfer protocol described is optimized for high-titer
recombinant adenovirus. High titer preparations contain ~1012
viral particles/mL and 1010 plaque forming units (pfu)/mL of
virus. Prior to gene transfer, the virus is diluted to a specific
multiplicity of infection (MOI), the number of plaque forming
units (pfu)/myocyte. For sarcomeric proteins, 200 MOI is
sufficient for adenoviral-mediated gene transfer with a viral
vector driven by a mouse cytomegalovirus immediately early
promoter (CMV) [2, 9, 42]. For recombinant adenoviral-
mediated gene transfer of non-sarcomeric proteins, 10 MOI
achieves 80 % myocyte transduction [41]. Control cells are
incubated with the same volume of DMEM + P/S (Subhead-
ing 2.2 item 3) without virus during the gene transfer step.
15. Viral aliquots are stored at 80 C and thawed on ice prior to
dilution. Multiple freeze–thaw cycles should be avoided by
making single-use aliquots of the virus.
16. Isolated myocytes are not cultured in serum-containing media
for contractile function studies, as serum promotes myocyte
dedifferentiation of the sarcomere [2, 9].
Acknowledgements
Funding from the National Institutes of Health HL067254

(MVW) supported this work, NIH T-32GM007315 (SEL), and
an American Heart Association Predoctoral Award
AHA12PRE8830022 (SEL).
References
1. Davis J et al (2008) Designing heart perfor- Handbook of Molecular and Cellular Methods
mance by gene transfer. Physiol Rev 88 in Biology and Medicine. CRC Press, Boca
(4):1567–1651 Raton, Florida, pp 557–578
2. Westfall MV et al (1997) Adenovirus-mediated 12. Jaenisch R (1976) Germ line integration and
myofilament gene transfer into adult cardiac Mendelian transmission of the exogenous
myocytes. Methods Cell Biol 52:307–322 Moloney leukemia virus. Proc Natl Acad Sci
3. Panchal RG et al (2001) Gene transfer: manip- U S A 73(4):1260–1264
ulating and monitoring function in cells and 13. Pfeifer A (2004) Lentiviral transgenesis. Trans-
tissues. Clin Exp Pharmacol Physiol 28 genic Res 13(6):513–522
(8):687–691 14. Fassler R (2004) Lentiviral transgene vectors.
4. McWhinney CD, Hansen C, Robishaw JD EMBO Rep 5(1):28–29
(2000) Alpha-1 adrenergic signaling in a car- 15. Wells K, Moore K, Wall R (1999) Transgene
diac murine atrial myocyte (HL-1) cell line. vectors go retro. Nat Biotechnol 17(1):25–26
Mol Cell Biochem 214(1–2):111–119 16. Kass-Eisler A et al (1993) Quantitative determi-
5. Beaumont J et al (2008) Overexpression of nation of adenovirus-mediated gene delivery to
human truncated peroxisome proliferator- rat cardiac myocytes in vitro and in vivo. Proc
activated receptor alpha induces apoptosis in Natl Acad Sci U S A 90(24):11498–11502
HL-1 cardiomyocytes. Cardiovasc Res 79 17. Hofmann A et al (2003) Efficient transgenesis
(3):458–463 in farm animals by lentiviral vectors. EMBO
6. Ting YK et al (2011) Transcriptional activation Rep 4(11):1054–1060
of the anchoring protein SAP97 by heat shock 18. Izsvak Z et al (2009) Efficient stable gene
factor (HSF)-1 stabilizes K(v) 1.5 channels in transfer into human cells by the Sleeping
HL-1 cells. Br J Pharmacol 162(8):1832–1842 Beauty transposon vectors. Methods 49
7. Louch WE, Sheehan KA, Wolska BM (2011) (3):287–297
Methods in cardiomyocyte isolation, culture, 19. Merkulov S et al (2012) In vivo cardiac myosin
and gene transfer. J Mol Cell Cardiol 51 binding protein C gene transfer rescues myofil-
(3):288–298 ament contractile dysfunction in cardiac myo-
8. Monge C et al (2009) Comparative analysis of sin binding protein C null mice. Circ Heart Fail
the bioenergetics of adult cardiomyocytes and 5(5):635–644
nonbeating HL-1 cells: respiratory chain activ- 20. Lee CJ et al (2011) Promoter-specific lentivec-
ities, glycolytic enzyme profiles, and metabolic tors for long-term, cardiac-directed therapy of
fluxes. Can J Physiol Pharmacol 87 Fabry disease. J Cardiol 57(1):115–122
(4):318–326
21. Fassler M et al (2013) Preferential lentiviral
9. Rust EM, Westfall MV, Metzger JM (1998) targeting of astrocytes in the central nervous
Stability of the contractile assembly and system. PLoS One 8(10):e76092
Ca2 + -activated tension in adenovirus
infected adult cardiac myocytes. Mol Cell Bio- 22. Wang L et al (2009) Long-term effect of neu-
chem 181(1–2):143–155 ronal nitric oxide synthase over-expression on
cardiac neurotransmission mediated by a lenti-
10. Kirshenbaum LA et al (1993) Highly efficient viral vector. J Physiol 587(Pt 14):3629–3637
gene transfer into adult ventricular myocytes by
recombinant adenovirus. J Clin Invest 92 23. Pacak CA, Byrne BJ (2011) AAV vectors for
(1):381–387 cardiac gene transfer: experimental tools and clin-
ical opportunities. Mol Ther 19(9):1582–1590
11. Kampert SE, Devaney E, Westfall MV (2011)
Gene transfer and expression in animal cells. In: 24. Deyle DR, Russell DW (2009) Adeno-
Cseke AKLJ, Kaufman PB, Westfall MV (eds) associated virus vector integration. Curr Opin
Mol Ther 11(4):442–447
25. Huang S et al (2012) Adeno-associated virus hypertrophic cardiomyopathy mutant alpha-

Rep-mediated targeting of integrase-defective tropomyosins in adult cardiac myocytes. Nat
retroviral vector DNA circles into human chro- Med 5(12):1413–1417
mosome 19. Biochem Biophys Res Commun 34. Herron TJ et al (2010) Ca2 + -independent
417(1):78–83 positive molecular inotropy for failing rabbit
26. Park F (2007) Lentiviral vectors: are they the and human cardiac muscle by alpha-myosin
future of animal transgenesis? Physiol Geno- motor gene transfer. FASEB J 24(2):415–424
mics 31(2):159–173 35. Day SM, Westfall MV, Metzger JM (2007)
27. Coyne CB, Bergelson JM (2005) CAR: a virus Tuning cardiac performance in ischemic
receptor within the tight junction. Adv Drug heart disease and failure by modulating myofil-
Deliv Rev 57(6):869–882 ament function. J Mol Med (Berl) 85
28. Bergelson JM et al (1997) Isolation of a common (9):911–921
receptor for Coxsackie B viruses and adenoviruses 36. del Monte F et al (2002) Targeting phospho-
2 and 5. Science 275(5304):1320–1323 lamban by gene transfer in human heart failure.
29. Fechner H et al (2003) Induction of coxsack- Circulation 105(8):904–907
ievirus-adenovirus-receptor expression during 37. Amalfitano A, Parks RJ (2002) Separating fact
myocardial tissue formation and remodeling: from fiction: assessing the potential of modified
identification of a cell-to-cell contact- adenovirus vectors for use in human gene ther-
dependent regulatory mechanism. Circulation apy. Curr Gene Ther 2(2):111–133
107(6):876–882 38. Ng P et al (1999) A high-efficiency Cre/loxP-
30. Westfall MV, Lee AM, Robinson DA (2005) based system for construction of adenoviral
Differential contribution of troponin I phos- vectors. Hum Gene Ther 10(16):2667–2672
phorylation sites to the endothelin-modulated 39. Ng P et al (2000) An enhanced system for
contractile response. J Biol Chem 280 construction of adenoviral vectors by the two-
(50):41324–41331 plasmid rescue method. Hum Gene Ther 11
31. Westfall MV, Rust EM, Metzger JM (1997) (5):693–699
Slow skeletal troponin I gene transfer, expres- 40. Haworth RA, Hunter DR, Berkoff HA (1980)
sion, and myofilament incorporation enhances The isolation of Ca2 + -resistant myocytes
adult cardiac myocyte contractile function. from the adult rat. J Mol Cell Cardiol 12
Proc Natl Acad Sci U S A 94(10):5444–5449 (7):715–723
32. Rust EM, Albayya FP, Metzger JM (1999) 41. Hwang H et al (2012) PKCbetaII modulation
Identification of a contractile deficit in adult of myocyte contractile performance. J Mol Cell
cardiac myocytes expressing hypertrophic Cardiol 53(2):176–186
cardiomyopathy-associated mutant troponin T 42. Lang SE et al (2013) Myofilament incorpora-
proteins. J Clin Invest 103(10):1459–1467 tion and contractile function after gene transfer
33. Michele DE, Albayya FP, Metzger JM (1999) of cardiac troponin I Ser43/45Ala. Arch Bio-
Direct, convergent hypersensitivity of calcium- chem Biophys 535(1):49–55
activated force generation produced by
Chapter 16
Analysis of 4D Myocardial Wall Motion During Early

Stages of Chick Heart Development
Madeline Midgett and Sandra Rugonyi
Abstract
4D myocardial wall motion analysis (3D structure over time) during early embryonic stages of chick heart
development provides a comprehensive view to characterize the biomechanical environment of cardiac
growth. Myocardial wall strains, velocity, and area shortening over the cardiac cycle are common wall
motion assessments and can be accurately measured from 4D datasets. Here, we describe how to employ a
variety of image modalities (optical, ultrasound, and optical coherence tomography imaging) and analysis
techniques to extract quantitative measures of myocardial wall motion.
Key words Myocardial wall motion, Chick heart development, Optical imaging, Ultrasound imaging,
Optical coherence tomography (OCT) imaging, 4D synchronization, 4D reconstruction, Cardiac
segmentation
1 Introduction
Embryonic cardiac development depends on the complex interac-

tion of genetic and environmental factors. While genes control the
generation and assembly of basic heart components, the local bio-
mechanical environment of the developing cardiac system influ-
ences cell behavior and future growth [1–4]. Myocardial wall
motion in the early stages of embryonic heart development is
studied to quantitatively characterize the biomechanical environ-
ment during cardiac development. Chick embryos are frequently
used as a biological model of cardiac development because of
developmental similarities with human embryos, and the accessibil-
ity and ease of manipulation of the chick heart within the egg. The
positioning of the chick heart within the egg in early developmental
stages also allows for the implementation of a variety of imaging
modalities to study cardiac motion and follow the progression of
cardiac development.
The imaging strategies used to measure the motion of the heart
myocardial walls include optical imaging, ultrasound imaging, and
191
192 Madeline Midgett and Sandra Rugonyi
most recently optical coherence tomography (OCT). Optical imag-

ing can be used to measure the ventricle size throughout the cardiac
cycle from the top view of the embryo, but lacks any image pene-
tration depth into the heart tissue. Ultrasound biomicroscopy pro-
vides enough tissue penetration and spatial resolution to observe
cardiac wall features and motion through tomographic views. The
disadvantages of ultrasound imaging are that it currently only
allows for 2D image analysis (see Note 1), the structure and velocity
data cannot be linked, and the resolution nears the limit for fine
wall border analysis at early stages of embryonic cardiac develop-
ment. OCT can achieve higher temporal and axial resolution than
ultrasound biomicroscopy and acquire simultaneous structure and
velocity data. Since OCT acquisition is noncontact and noninvasive,
2D datasets can be synchronized and reconstructed to form 3D
structural images and 4D (3D + time) heart motion datasets over
the cardiac cycle. 4D images and 4D wall motion analysis are used
to accurately track the dynamics of myocardial tissue layer motion
over time, even when motion and deformation are large. 4D ana-
lyses of the heart also allow for accurate 2D cross-sectional image
generation, perpendicular to the direction of blood flow, enabling
more accurate image analysis and measures of relative structural
position changes. Further, 4D imaging enables 4D computational
modeling of cardiac tissue mechanics.
These three imaging strategies employ a range of image display
views to assess cardiac wall motion. Top view optical images display
2D surface structure of the beating heart, while B-mode images are
2D line-scan tomographic images (of a view plane) collected over
time. M-mode images are used to display cardiac wall motion over
time, by collecting scans from a single line position in a
corresponding top view or B-mode, resulting in cardiac structure
(gray scale) along the selected line in the vertical direction versus
time in the horizontal direction. Doppler images display velocity
data and are used to analyze blood or tissue velocities.
In this chapter, we will review basic methods of optical, ultra-
sound, and OCT imaging in the analysis of chick embryonic myo-
cardial wall motion. While methods are general and can be
employed at different cardiac developmental stages, for specificity
purposes we will focus on imaging cardiac structures of chicken
embryos at days 3 and 4 of incubation, which correspond to the
Hamburger-Hamilton stages HH18 and HH24, respectively [5].
Further, without lack of generality, 4D OCT image acquisition and
analysis descriptions will focus on the outflow tract (OFT) portion
of HH18 embryonic heart that connects the ventricle and arterial
system. The OFT is frequently studied because it is very sensitive to
hemodynamic changes [6, 7] and is often the origin of congenital
heart defects during development [7].
Analysis of 4D Myocardial Wall Motion During Early Stages of Chick Heart Development 193
2 Materials
2.1 Embryo 1. Fertilized chicken eggs (typically white Leghorn).

Preparation 2. Heating incubator: positive flow incubator with filling reser-
voir, temperature and humidity control, and automatically
turning racks.
3. Heating platform, designed to keep embryos at 37.5 C during
imaging (see Note 2).
4. Tweezers.
2.2 Optical Imaging 1. Stereomicroscope, with at least 10 magnification for proper
image resolution at early cardiac development stages (3–4 days
of incubation).
2. CCD camera, with a frame rate of at least 100 frames per second
(fps) to achieve adequate time resolution within the cardiac cycle
(>40 frames per cycle with typical cardiac periods ~400 ms).
3. Egg positioning platform, with a cup to hold egg in place
during imaging.
2.3 Ultrasound 1. Ultrasound biomicroscopy system.

Imaging 2. Transducer probe (see Note 3).
3. Hanks’ balanced salt solution (HBSS) or chick Ringer’s solution.
4. Beaker, with egg stabilizers to keep the egg in place during
imaging.
5. Micromanipulator stage (XYZ platform), for fine positioning of
the egg during imaging.
2.4 Optical 1. OCT system.

Coherence 2. Micromanipulator stage (XYZ platform), for fine positioning of
Tomography Imaging the egg during imaging.
2.5 Imaging Analysis 1. Computer, with enough memory and processing capacity to
analyze acquired datasets.
2. Image analysis software, e.g., imaging system platform, Matlab,
Amira, etc.
3 Methods
3.1 Embryo 1. Incubate fertilized white Leghorn chick eggs blunt end up at
Preparation 37.5 C and 80 % humidity until they reach the desired imaging
embryonic developmental stage. Following Hamburger-
Hamilton (HH) staging, eggs reach stage HH18 and HH24
after approximately 3 and 4 days of incubation, respectively [5]
(see Note 4).
2. Remove a small section of the blunt-end shell with tweezers

and clear away any shell pieces on the inner shell membrane
(ISM).
3. Remove the ISM directly over the embryo to expose the beat-
ing heart for imaging.
4. Discard any embryos that bleed upon membrane removal, have
obvious structural defects, or are not at the correct develop-
mental stage (see Note 5). Normal embryos face right-side up
with the heart on the top surface, with strong circulation in
surrounding (vitelline) vessels and through the heart.
3.2 Imaging the Image embryos in vivo at development stages up to HH26 nonin-
Embryo Using Optical vasively with optical imaging (see Note 6), to acquire top view
Imaging (and a CCD images of the beating heart. Perform the following steps for optical
Camera) image acquisition:
1. Attach a CCD camera to a lens adapter and mount on a
stereomicroscope.
2. Position the embryo in the holding cup to obtain a clear view of
the heart, as shown in Fig. 1.
3. Record at least 3–4 cardiac cycles of video.
4. Record a scribed standard in the view field to convert pixel
dimensions to length measurements during image analysis.
Fig. 1 HH18/HH24 staging with optical imaging and corresponding heart

schematics. (a) Optical image of the chick embryo at HH18. (b) Heart
schematic at HH18. (c) Optical image of the chick embryo at HH24. (d) Heart
schematic at HH24. OFT outflow tract, A atrium, AVC atrial ventricular cushion, V
ventricle
3.3 Analysis of Optical images can be used to measure ventricular area and volume
Optical Images over the cardiac cycle to allow characterization of embryonic car-
diac mechanics, including stroke volume and cardiac output, and
3.3.1 Optical Image thus of myocardial function. Keller et al. among others have used
Geometrical Analysis planimetry software to measure ventricular epicardial perimeter and
ventricular area from optical images [8–11], and have also charac-
terized the relationship between ventricular area and pressure over
the cardiac cycle [12] (shown in Fig. 2). Further, ventricular vol-
ume has been estimated from optical images using both prolate
spheroid and spherical geometry [13]. These analyses of optical
images are outlined in this section.
1. Verify the heart region of interest was imaged, and that the
image contrast provides distinguishable heart structures from
surrounding tissue and fluid. Also verify that an adequate heart
rate was obtained throughout the entire acquired dataset.
2. Use the imaged scribed standard to calibrate image pixel count
to an appropriate length measurement (e.g., mm).
3. Select acquired image frames (from at least three cardiac cycles)
where maximum or end-diastole (ED) and minimum or end-
systole (ES) ventricular area occurs, and use planimetry soft-
ware to calculate ventricle perimeter and area in the imaging
plane from an outline of the ventricle.
4. Calculate shortening fractions (SF) for each variable measured,
using
x ED x ES
SF ¼ ð1Þ
x ED
Fig. 2 Pressure-area loops for HH stage 16–24, with the progression of the
cardiac cycle moving in a counterclockwise direction around each loop.
Ventricular diastole (relaxation and filling) occurs at the bottom edge of the
loop until the ventricle reaches maximum expansion. Ventricular systole (active
contraction) then causes the pressure to rapidly increase to the top edge of
the loop, where the outflow tract cushions open and blood flows from the
ventricle into the aorta. This is followed by ventricle relaxation to form the
pressure-area loop. Figure reproduced with permission from [12]
where xED and xES are the average measure values (ventricle
diameter, perimeter, or area) at ED and ES, respectively.
5. Calculate ventricle volume assuming either prolate spheroid or
spherical geometry [13]. With prolate spheroid geometry,
measure the long axis (2a) and short axis (2b) of the ventricle,
and calculate ventricle volumes (V) at each phase of the cardiac
cycle using
V ¼ ð4=3Þ π ab 2 : ð2Þ
Assuming the spherical model, calculate ventricle volumes

using,
V ¼ ð4=3Þ π 0:5 A 1:5 ð3Þ
where A is the measured surface area of the ventricle [13].
3.3.2 Optical Image Embryonic cardiac wall motion can also be quantified from optical
Centerline Point-Intensity images using centerline point-intensity analysis. This analysis uses
Analysis image intensities to estimate heart size over the cardiac cycle, and is
most effective to compare myocardial wall motion and determine the
phase of contraction events along the embryonic tubular heart (ven-
tricle and OFT) (see Note 7). Centerline point-intensity analysis of
optical images is summarized in Fig. 3 and outlined in this section.
1. Use image software to invert the intensity of monochrome
videos to create a direct correlation of pixel intensity with
lumen diameter, so that cardiac expansion and contraction
correspond to an increase and decrease of inverse image inten-
sity, respectively.
Fig. 3 Centerline point-intensity analysis overview. (a) Inverted intensity frame from a recorded video of the
embryonic chick heart, with the heart ventricle and outflow tract identified. (b) Frame from (a) showing
possible M-mode extraction locations, and M-mode image from location V2 over approximately eight cardiac
cycles. (c) Extracted pixel intensity versus time data from dashed centerline of the M-mode from (b)
2. Choose regions of interest to extract M-mode images, and

extract inverted gray-scale intensity data along lines perpendic-
ular to flow to monitor cardiac motion over time at different
regions of the heart. Confirm image quality with clear regions
of expansion and contraction in M-modes.
3. Select points along the centerline in each M-mode and plot
intensity over time.
4. Estimate relative diameters and wall contraction/expansion
rates from the intensity versus time curve.
3.4 Imaging the Use embryos at stages HH24 and older for ultrasound imaging (see
Embryo Using Note 8). B-mode images, M-mode images, and Doppler velocity
Ultrasound can be separately acquired with ultrasound. Doppler velocity within
Biomicroscopy the myocardium walls can be acquired with ultrasound, but it is
difficult to confirm that the measurement location (sampling area)
stays within the wall boundaries, so blood flow velocities through
the heart are more commonly measured. While the different imag-
ing modalities can be acquired independently, B-mode images are
first acquired to locate the heart and cardiac structures of interest.
3.4.1 Acquisition 1. During embryo preparation, window the egg as much as possi-
of B-Mode Ultrasound ble so that the probe can make direct contact with the fluid
Images surface. Wait to remove the ISM until the egg is immersed in
HBSS (see Note 9).
2. Place the egg in a beaker filled with warmed HBSS (37.5 C) so
that the level of HBSS just covers the embryo, and stabilize the
egg to ensure it stays upright.
3. Center the egg on an XYZ platform to allow maximal range of
motion during positioning, and lower the probe until it
touches the HBSS surface. Ensure good contact between the
ultrasound probe and the HBSS solution (which acts as ultra-
sound gel for embryonic cardiac image acquisition), as air gaps
at the probe surface degrade ultrasound signals.
4. Begin acquiring ultrasound images in B-mode. Gently move
the probe while observing acquired images to locate the
embryonic heart, which is easily distinguishable due to its
beating motion. A B-mode ultrasound biomicroscopy frame
of a HH24 chick embryo is shown in Fig. 4a.
5. Position the probe so that the desired cardiac imaging plane is
achieved and the image focal zone is in line with the heart
structure. To collect cross-sectional images of the OFT, for
example, maneuver the probe until the OFT is circular with
the lumen completely filling the OFT at maximum expansion
and having a slit-like shape at maximum contraction.
6. Decrease the image width as much as possible around the heart,
and set the line density to achieve optimal resolution and frame
capture rate (see Note 10). Adjust the persistence setting to
Fig. 4 Typical ultrasound longitudinal OFT image. (a) B-mode image of HH24 chick embryo. (b) Pulse wave
velocity of blood flowing through the OFT during approximately four cardiac cycles. L lumen, M myocardium,
C cardiac jelly, A atrium, V ventricle
optimize the system’s pixel averaging algorithm to produce the

clearest image.
7. Acquire B-mode data for at least 3–4 cardiac cycles.
3.4.2 Acquisition of 1. Start imaging in B-mode and position the probe to show a
Doppler Ultrasound Images longitudinal section of the heart.
2. Change the system setting to Doppler mode.
3. Select a sampling area in the center of the lumen, and make it as
small and close to the center of the lumen as possible to keep
the sampling area within the lumen over the cardiac cycle.
4. Use the ultrasound system’s software to estimate the angle
between the probe beam and the direction of flow (this is
typically done using the system software to manually draw a
line along the presumed direction of blood flow in the sampling
area). The ultrasound system will use the measured flow angle
(θ) to convert the measured velocity projection (Vz) to the
magnitude of flow velocity (V) (see Note 11) using
Vz
V ¼ : ð4Þ
cos θ
5. Acquire Doppler velocity data for at least 3–4 cardiac cycles. A
Doppler velocity trace of blood flow over time through the
HH24 heart OFT is shown in Fig. 4b.
6. After flow velocity is recorded, switch back to B-mode to make

sure the embryo stayed in position during acquisition. If the
velocity suddenly changes during recording, the embryo posi-
tioning may have shifted, and measurements need to be
repeated.
3.4.3 Acquisition 1. Start imaging in B-mode and position the probe to show the
of M-Mode Ultrasound heart region of interest.
Images 2. Change the system setting to M-mode.
3. Choose a line spanning a region of interest in the heart from a
B-mode image frame.
4. Acquire M-mode data for at least 3–4 cardiac cycles.
3.5 Analysis Ultrasound images can be used to measure heart dimension, strain,
of Ultrasound and blood flow velocity over the cardiac cycle to allow characteri-
Biomicroscopy Images zation of embryonic cardiac mechanics. Because ultrasound pro-
duces tomographic images, ultrasound cardiac views are not
restricted to top view images (as is the case for the optical images).
Thus ultrasound biomicroscopy enables characterization of ventric-
ular mechanics (in ways similar as those described in Subheading 3.3
but using 2D section views of the ventricle) and also atrio-
ventricular and OFT motion and function. Quantification of ultra-
sound biomicroscopy images is facilitated by the system’s software
(frequently manually done image by image), although image
sequences can also be exported and analyzed with external image
post-processing tools. The advantage of obtaining ultrasound
images is that different section views can be obtained, greatly
enhancing the analysis of embryonic myocardial motion. Ultra-
sound biomicroscopy analysis options are outlined in this section.
1. Verify the heart region of interest was imaged, and that an
adequate heart rate was obtained throughout the entire
acquired dataset.
2. Use the ultrasound system’s structural tracking features to
measure distances in selected cross-sectional B-mode frames
at several points in the cardiac cycle. Use the relative positions
to calculate strains (wall thickening and thinning) relative to
the maximum expansion position of the myocardium. Calcu-
late radial strain (εr) using
εr ¼ ðT T max Þ=T max ð5Þ
where T and Tmax are the myocardium thicknesses at the chosen

cross section and maximum expansion, respectively. Calculate
circumferential strain (εθ) using
εθ ¼ ðC C max Þ=C max ð6Þ
where C and Cmax are the myocardium perimeters at the chosen

cross section and maximum expansion, respectively.
3. Use the M-mode software tools to measure average cardiac wall
thicknesses, wall excursions, and expansion and contraction
times over several cardiac cycles from M-mode images.
4. Use the Doppler ultrasound software tools to measure average
maximum and minimum velocities in the cardiac cycle, time of
flow (width of the velocity peak), and heart rate (time between
peaks) from Doppler velocity data over several cardiac cycles
(see Note 12).
5. Use Doppler blood velocity measurements, together with car-
diac dimensions, to estimate volume flow rates and stroke
volumes (assuming, for example, Poiseuille parabolic flows) to
evaluate myocardial function.
3.6 Imaging the Image embryonic hearts at stage HH18 or younger using OCT (see
Embryo Using Optical Note 13). Typically, only B-mode images are collected during
Coherence OCT image acquisition, and M-mode images are later extracted
Tomography (OCT) from the B-mode image set. Further, raw OCT data allows simul-
taneous extraction of structural and Doppler data with the same
resolution.
4D images can be acquired with nongated image acquisition,
where collection of B-mode images starts at random points in the
cardiac cycle and requires post-acquisition synchronization; or with
gated image acquisition where collection is triggered by a cardiac
signal. Jenkins et al. performed prospective-gated OCT imaging by
triggering image acquisition with signals obtained by a laser Dopp-
ler velocimeter from a vitelline vessel [14]. Gated imaging can also
be triggered with ECG signals. The gating signals need to be strong
and clear enough to adequately trigger acquisition, because error
can be introduced in 4D reconstruction if the cardiac signals are
weak or triggering algorithms are not robust. Unlike gated acquisi-
tion, which relies heavily on the gated signal for image synchroni-
zation and reconstruction, nongated acquisition requires image
post-processing. At early stages of chick cardiac development suc-
cessful synchronization and reconstruction of 4D OCT images is
possible using nongated acquisition because the embryo is not yet
moving within the egg, and also due to the periodic motion of the
cardiac cycle. Acquisition of 2D and 4D OCT image sets, focusing
on OCT imaging of the HH18 heart OFT, is outlined in this
section.
1. Set up system acquisition parameters. To adequately analyze
heart wall motion data over the cardiac cycle, the OCT system
should achieve an axial resolution of at least 10 μm in tissue, a
penetration depth resolution of ~2 mm (in air), and acquire

B-scan images at least 100 fps (the cardiac cycle is approximately
400 ms at HH18 and slightly slower at earlier stages of develop-
ment). Each B-scan should cover ~1.1 mm to capture the whole
length of the OFT in a longitudinal frame. Set the number of
A-lines (vertical lines) in each B-scan to achieve an adequate
combination of image resolution and B-scan frame rate. OCT
systems with at least 256 A-lines per B-scan image and an acqui-
sition frame rate of 140 B-modes per second have been shown to
capture the dynamic movement of the embryonic chick heart
with adequate spatial resolution [15] (see Note 14).
2. Calibrate the OCT system. This is typically done prior to
imaging, e.g., using a glass slide of known dimensions.
3. Place the embryo on the XYZ platform, and use the OCT
display system to visualize the heart and localize the desired
imaging planes during positioning. For OFT imaging at
HH18, position the embryo so that the laser source perpendic-
ularly crosses the OFT near the inlet, and raise the embryo level
up until the membrane surface is at the top of the image view.
Further position the embryo so that the cross section of the
OFT is circular with the lumen completely filling the OFT at
maximum expansion and having a slit-like shape at maximum
contraction.
4. Collect B-mode cross-sectional images over time (2D image
sequences) at the OFT inlet position, as shown in Fig. 5.
Acquire approximately 200 B-scan frames (depending on the
Fig. 5 Representative OCT images of the embryonic chick OFT. (a) Cross-sectional 2D image at maximal
expansion. (b) Cross-sectional 2D image at maximal contraction. (c) Longitudinal 2D image at maximal
expansion. (d) Longitudinal 2D image at maximal contraction. (e) 2D scan strategy schematic. M myocardium,
C cardiac jelly, L lumen
acquisition frame rate) to image 3–4 cardiac cycles of the

beating heart.
5. Collect 2D image sequences (e.g., 200 B-scan frames) on
consecutive planes spanning the length of the OFT for post-
acquisition 4D reconstruction. To this end, program the OCT
system to move the scanner along the length of the OFT with a
step function so that the chosen number of positions covers the
entire OFT (~1.1 mm) (see Note 15).
6. Additionally, collect one B-scan image sequence of a longitudi-
nal OFT section to adjust phase lags during image analysis (see
Note 16).
3.7 Analysis of 2D Like ultrasound biomicroscopy, OCT acquires tomographic sec-

OCT Images tions of the beating heart over the cardiac cycle that can be similarly
analyzed, as previously described in Subheading 3.5. The advan-
tages of OCT images are (1) better image resolution (<10 μm); (2)
motion and Doppler velocities are simultaneously acquired with
the same resolution and thus can be analyzed together; and (3)
OCT imaging is noncontact (thus avoids dragging of the embryo
with probe motion enabling subsequent image reconstructions).
These characteristics enable further characterization of myocardial
function from OCT images, including 4D reconstruction of cardiac
motion.
Because of high OCT image resolution, clear M-mode images
can be extracted from B-mode images, which allow more control
and precision on where the M-mode image comes from than if
directly acquiring M-mode images with the OCT system (see Note
17). High resolution OCT M-mode images also allow for effective
calculation of myocardial wall thicknesses (radial strains) and radial
wall velocities over early stages of cardiac development
(HH12–HH18). Since structural and Doppler data are acquired
simultaneously, wall strains can also be more accurately obtained
directly from Doppler velocity data [16]. While OCT 2D image
analysis depends on what quantification is of interest, a general
overall procedure is shown in Fig. 6 and outlined in this section.
1. Verify the heart region of interest was imaged, and that the
image contrast provides distinguishable heart structures from
surrounding tissue and fluid. Also verify that an adequate heart
rate was obtained throughout the entire acquired dataset.
2. Use B-mode images to analyze cardiac motion from tomo-
graphic cardiac views as previously described in Subheading 3.5.
Similar to ultrasound analysis, radial and circumferential strain
can be calculated between specific 2D cross-sectional images
using structural feature tracking from either available or custom
imaging software and Eqs. 5 and 6.
Fig. 6 Flow chart of OCT image analysis. (a) Select a centerline in the middle OFT of the B-mode structural
image at maximal expansion. (b) Extract an M-mode image from the B-scan structural dataset along the
centerline, select points along the top and bottom borders of the upper myocardium wall, and interpolate
between the points and give the wall trace (red borders). (c) Extract the position data and convert to relative
wall depth, where z0 and zi are the outer and inner myocardial wall positions, respectively. (d) Select the B-
mode phase image at time tn, and select the centerline. (e) Extract an M-phase image from the B-scan phase
dataset along the centerline. (f) Extract the velocity profile along the center (dashed) line, import the
myocardium wall z-positions at time tn on the velocity profile, and perform a linear fit to the Doppler data
over the myocardial wall thickness to obtain the velocity gradient, i.e., strain rate. M myocardium
3. Use custom software to extract an M-mode image from an

acquired 2D image sequence (see Note 18). On the resulting
M-mode image, select points along the outer and inner borders
of the upper myocardial wall (the lower myocardial wall might
be harder to distinguish because of depth resolution). Use a
tracing algorithm to interpolate the selected points to complete
the border traces, or other image border tracking procedures to
more precisely delineate the wall.
4. Calculate wall thickness over time by subtracting the relative
positions of the outer and inner myocardial walls.
5. Calculate radial strain using the thickness of the myocardial wall
extracted from the M-mode image with Eq. 5 relative to the
maximum expansion position, or using the wall thickness
change between infinitesimal time intervals in the cardiac
cycle [15].
6. Calculate myocardial wall radial velocity (vr). This can be done
either by calculating the velocities from the motion of the wall
or using Doppler data. In both cases, it is more effective to first
extract an M-mode image of the wall to facilitate wall motion
detection and Doppler data identification. To this end, choose a
line perpendicular to the heart centerline from a 2D B-mode
image, such that the line coincides with an A-scan line (vertical
line) and thus vertical myocardial wall motion corresponds to
the tubular radial expansion and Doppler velocity (Vz) corre-

sponds to the wall radial velocity. Trace the myocardium wall
motion from the M-mode image as described before in Sub-
heading 3.7, step 3.
7. To calculate myocardial wall radial velocity directly from the M-
mode image, sum the radial wall position gradients of the outer
and inner borders using

1 z o ðt þ dt Þ z o ðt Þ z i ðt þ dt Þ z i ðt Þ
dvr ðt Þ ¼ þ ð7Þ
2 dt dt
where zo and zi are the radial positions of the outer and inner
borders of the upper myocardial wall (obtained from the
M-mode tracing algorithm), respectively [15].
8. To calculate myocardial wall radial velocity using Doppler OCT
data, first obtain an M-phase image that corresponds to the
extracted M-mode image. Phase images contain data on phase
shifts (caused by the movement of tissue and blood cells
between two adjacent A-lines in a B-scan image), and are
obtained together with structural images (and with the same
resolution) after processing raw OCT signals. M-phase images
are then obtained the same way M-mode images are obtained,
but data is extracted along a line from phase images (rather than
structural images). Then convert the phase shift (Δϕ) at each
pixel in the M-phase image to the vertical component of veloc-
ity (Vz) using
λ0 Δϕ
Vz ¼ ; ð8Þ
4π n τ
where λ0 is the central wavelength, n is the refractive index of
the tissue (~1.35), and τ is the time difference between two
adjacent A-scans. Vz of the myocardial wall corresponds to the
wall radial velocity. To identify the regions corresponding to
the myocardial wall in M-phase images, use the myocardial wall
traces extracted from M-mode images.
9. Measure myocardial wall strain rates (in the radial direction)
with Doppler OCT, using the velocity gradient in the myocar-
dium wall [16]. Use the delineated myocardial wall borders to
determine the portion of the extracted velocity profile that
corresponds to the myocardial wall. Since the velocity distribu-
tion is linear along the myocardial wall, find the radial strain rate
by calculating the slope of a linear fit to the velocity gradient.
10. The radial strain can be calculated (see also the alternative calcu-
lation presented in Subheading 3.7, step 5) by integrating the
strain rate velocity gradient over the time period of interest [16].
3.8 Reconstruction Synchronization of B-mode images acquired with nongated

of 4D OCT Images imaging is required for successful reconstruction of 3D structures
over time. Based on previous works [17, 18], Liu et al. developed a
3.8.1 OCT Image procedure that uses structural feature similarities across M-mode
Synchronization and images extracted from corresponding lines on cross-sectional
Reconstruction Algorithmic B-mode images to synchronize image sequences by time, combined
Implementation with a phase-lag synchronization adjustment based on longitudinal
B-mode images to recover the peristaltic-like motion of the heart
[19]. While accurate, the procedure is computationally efficient
because it only uses data along one line of every B-mode image
acquired. The procedure, while general, has been applied to the
synchronization and reconstruction of images from the heart OFT
of HH18 embryos. The overall imaging, synchronization, and
reconstruction process is shown in Fig. 7, where acquired 2D
cross-sectional images from each location along the OFT are
pooled, interpolated, and normalized to one cardiac cycle so that
each sequence starts at the same cardiac cycle time point and has the
same number of frames. 3D structural images are then generated
over one cardiac cycle to visualize the dynamic motion of the OFT.
4D analyses of the heart allow for accurate 2D cross-sectional image
generation, perpendicular to the direction of blood flow, which can
then be analyzed as described in Subheading 3.7. To focus the
description, but without losing generality, we will describe the
detailed procedure employed for 4D OFT image reconstruction
in this section.
1. Determine, from each cross-sectional B-mode sequence,
appropriate lines from which to extract M-mode images.
Y X OFT
t/T=0 t/T=1
Z X Z
Position 1
1st Scan
Position
1st Scan 2nd Scan 200th Scan 1st image 2nd image 196th image
2nd Scan Position 2

Position
1st Scan 2nd Scan 200th Scan Synchronization 1st image 2nd image 196th image
Image Interpolation
Position 110
120th Scan 1st image 2nd image 196th image

Position 1st Scan 2nd Scan 200th Scan
Original images obtained using OCT
196 sets of reconstructed 3D images
Fig. 7 Schematic of OFT synchronization and reconstruction. 200 cross-sectional images were acquired at
120 different positions along the OFT, images from each sequence are then pooled and synchronized to one
cardiac cycle, and then 2D images are pooled at each time point to reconstruct 196 3D images over one
cardiac cycle. Figure adapted from [20]
These lines should contain the OFT upper and lower walls, and
be centered with respect to the OFT cross section to achieve
optimal synchronization.
2. Extract M-mode images from each cross-sectional B-mode
sequence along the selected lines. The resulting M-mode
image can be represented by its intensity I (zm, tn), which varies
with position along the vertical axis (zm) and time point (tn).
3. Compute the cardiac period of each M-mode image (and thus
each B-mode sequence) using the string-length method (SLM)
or a similar algorithm [19] (see Note 19).
4. Re-arrange M-mode lines by pooling them into one cardiac
cycle and arranging them by phase (see Note 20). If tn is the
time corresponding to a structural line in the M-mode image,
the phase (pn: time position during the first cardiac cycle) for
that line is
ht i
n
pn ¼ t n T; ½: represents integer part: ð9Þ
T
5. Resample the phase-organized M-mode images into a normal-
ized cardiac cycle using cubic spline interpolation among line
gray-scale intensities (see Note 21), so that each M-mode image
has the same number of vertical lines (K), equally spaced in
time spanning exactly one cardiac cycle.
6. Compute the correlation (a measure of similarity) of M-mode
images (Ii and Ii+1) corresponding to image sequences acquired
at adjacent positions i and i + 1 along the OFT. Correlation is
computed using a correlation coefficient (Ci,i+1) defined as
M K
C i, iþ1 ðs Þ ¼ ∑ ∑ I i z m ; pn I iþ1 z m , pn s ð10Þ
m¼1n¼1
where M is the number of pixels in the vertical direction, K is

the number of lines, pn is now the normalized phase, and s is a
phase shift.
7. Find the phase shift (Si,i+1) between adjacent M-mode images
(Ii and Ii+1) by maximizing Ci,i+1 with
S i, iþ1 ¼ max C i, iþ1 ðs Þ, s ½0; 1: ð11Þ

s
8. Determine the absolute phase shift (Si) of M-mode images
along the OFT with respect to the first M-mode (e.g.,
corresponding to the OFT inlet B-mode sequence). The abso-
lute phase shift of each image sequence relative to the first
sequence is then the accumulated sum of relative phase shifts,
i1
Si ¼ ∑ S j , j þ1 with i ¼ 2 . . . L; S1 ¼ 0 ð12Þ
j ¼1
where L is the number of M-mode images (and B-mode image

sequences, e.g., 120 as in Fig. 7) along the OFT.
9. Calculate phase lag introduced by peristaltic-like motions of the
tubular heart (that cannot be captured using just image similar-
ity), by extracting M-modes from the OFT inlet and outlet of
the acquired longitudinal image sequence. (The M-mode loca-
tions should approximately correspond to the locations of B-
mode series 1 and L, respectively.) Calculate the average phase
lag (Δp) for maximal cardiac contraction between the two
locations from the corresponding M-modes.
10. Calculate the phase lag (Δp*) for maximal cardiac contraction
between the M-mode images 1 and L (corresponding to the
cross-sectional B-mode image series) after they have been shifted
in time using SL from Subheading 3.8, step 8. Δp* is the
peristaltic phase lag already captured by the similarity approach.
11. Adjust the absolute phase shift of image sequences using a
correction phase (plag) that assumes the phase shift between
adjacent image sequences is proportional to the motion phase
lag between the OFT inlet and outlet (see Note 22). Therefore,
the absolute adjusted phase shift (Si0 ) is
Δ p Δ p* 0
plag ¼ ; S i ¼ S i þ ði 1Þ plag : ð13Þ
L1
12. Using the absolute phase shifts calculated in Subheading 3.8,
step 11, synchronize all acquired image sequences (cross-
sectional B-mode images) along the OFT. Shift time points
for each image sequence by Si0 and pool the images to a normal-
ized cardiac cycle across all sequences to fully synchronize the
dataset. Use image interpolation to resample images at equally
spaced phases over the cardiac cycle, so that 2D images can
form 3D structural images spanning the cardiac cycle, and
reconstruct the 4D motion of the heart.
3.9 Analysis of 4D To fully analyze the reconstructed 4D motion of the heart, struc-
OCT Images tural cardiac features might be segmented from the rest of the
image. Segmented geometries are used to better visualize and
quantify cardiac wall motion over the cardiac cycle. Motions from
different control and treated embryos can be compared to assess
the effects of interventions on myocardial motion dynamics and
function. Further, segmented geometries can be used in computa-
tional models of the beating heart. Groups have used manual or
semi-manual segmentation to analyze heart wall motion by picking
specific time points within the cardiac cycle to analyze the 3D
structural data. Garita et al. chose 15 points in the cardiac cycle to
segment the myocardium, lumen, and cardiac jelly at six locations
spanning the OFT to produce 90 volumes of structural data [21].
Manual segmentation uses image processing software programs

(e.g., Amira) to manually select and then delineate (using the
computer mouse and software tools) cardiac features, typically on
cross-sectional 2D planes (see Note 23). By tracing the structure of
desired cardiac wall features, such as the lumen or myocardium,
over several 2D cross-sectional planes, 3D cardiac segmentations
start to emerge, from which volumes or areas can be computed.
While manual segmentation analysis provides 3D information at
specific points in the cardiac cycle, the movement of the heart is not
smooth over time. Manual segmentation analysis can also be time
consuming and prone to operator errors.
Yin et al. have developed a semiautomatic and reliable proce-
dure to extract 4D OFT myocardial layer surfaces using a double-
line model (DLM) [20]. DLMs are 2D piece-wise deformable
(matching) models, designed to detect and track high-intensity
tissue layers, such as the myocardium, and accurately capture the
thickness of specific layers. The DLM segmentation algorithm is
applied on 2D cross sections over space and time to render 4D
segmentations smoothed over the entire cardiac cycle. The overall
process is shown in Fig. 8, and outlined in this section.
1. Select an initial phase in the cardiac cycle to start segmenta-
tions. This is typically a phase in which the OFT is either fully
open or fully closed. Chose the corresponding reconstructed
3D OFT image, and select a plane within that image that is
approximately perpendicular to the direction of flow and
located at the inlet of the OFT.
2. Place several layer DLMs around the myocardium on the
selected image plane so that the algorithm, guided by a robust
maximum-likelihood estimator, can detect myocardial structures
Fig. 8 Sketch showing the OFT shape extraction procedure from OCT images. (a) Cross-sectional plane
depicting the myocardium and endocardium layers, the myocardium mid-layer contour, a layer DLM placed on
the myocardium, and an edge DLM placed on the endocardium. (b) Initial cross-sectional plane, with ps as the
centroid position vector in each plane, and ns as the normal vector in the direction of flow. (c) Extraction of the
cardiac shape is performed by sweeping cross-sectional planes along the OFT. Image reproduced from [20]
via intensity differences in the 2D images and adjust the DLM

position and thickness to match the myocardium layer. Use an
active-contour model to link layer DLMs and facilitate
smoothing.
3. Identify cross sections perpendicular to the direction of blood
flow by rotating the initial 2D plane until a minimum mid-layer
myocardium contour perimeter is achieved.
4. Copy the DLM configuration to successive planes along the
OFT to detect the myocardial layer on each plane throughout
the 3D dataset. Extract the centroids of the mid-layer contour
and smooth the resulting OFT centerline (using averaged data
from adjacent planes).
5. Adjust the position of cross-sectional planes, DLM location,
and parameters based on the smoothed centerline.
6. Copy the initial configuration of the layer DLMs to each time
point until the entire 4D cardiac cycle dataset is segmented (see
Note 24).
7. Smooth the segmented data in time and space by averaging
values from adjacent points (in time and space).
8. Calculate myocardial wall motions with radial and circumferential
strains using the DLM segmented myocardium contour perime-
ter and maximum perimeter reference using Eqs. 5 and 6 [9].
In summary, myocardial wall motion measures during early
stages of chick heart development can be extracted from acquired
optical, ultrasound, and OCT images. While the surface motion of
the ventricle and OFT can be analyzed with optical imaging, ultra-
sound and OCT systems image penetrated depths into the heart
structure, with the high resolution of OCT images allowing for
accurate 4D reconstruction. 4D datasets capture the dynamic
motion of the developing heart, and allow for feature segmentation
and analysis in any 2D plane.
4 Notes
1. Although only 2D ultrasound machines have been used for

previous chick embryo research, ultrasound machines that
acquire 3D structure data are becoming more available.
2. Embryonic temperature during imaging needs to be tightly
controlled since temperature greatly affects cardiac perfor-
mance, most noticeably heart rate. The heating setup will vary
for each imaging system, but needs to keep the embryo at
approximately 37.5 C to maintain a physiologically natural
heart rate at each stage of development. Use combinations of
heat lamps, heating pads, and plastic enclosures, with embryo
temperature controlled by automatic feedback thermocouples
placed near the embryo, or ECG heart rate tracking that work
with the employed imaging system.
3. Use a 40 MHz probe to fully penetrate the fluid and heart tissue
at stage HH24–HH26. Higher frequency probes can achieve
deeper penetration depths for subsequent embryo stages.
4. At these early stages of development, the chick heart is tubular
with no valves. The OFT connects the ventricle with the arterial
system and acts as a primitive valve by closing to limit blood flow
[7]. The ventricle and OFT are readily visible on the top of the
embryo.
5. Embryos at stage HH18 of development have a curved upper
body and straight tail, only slight limb buds, and un-pigmented
eyes, while embryos at developmental stage HH24 have a
curved upper body and tail, protruding limb buds, dark-
pigmented eyes, and more surrounding vasculature than at
stage HH18.
6. After HH26, the embryo sinks in the egg and the heart becomes
difficult to visualize using noninvasive optical methods.
7. Point-intensity analysis is based on the same principles as video
densitometry used by clinicians to track changes in x-ray and
ultrasound images. In the vasculature, image intensity is mainly
determined by the amount of blood present at different points in
the cardiac cycle. Red blood cells absorb a fraction of the light,
reflecting a less intense (darker) region in a digital image. There-
fore, decreasing image intensities are representative of increasing
blood vessel diameter in which more blood is present.
8. Ultrasound spatial resolution of heart structures in embryos
younger than HH24 is poor.
9. The ISM is easier to cleanly remove (less blood vessels tear),
when immersed in HBSS.
10. The motion of the beating heart is better resolved with higher
acquisition frame rates. Smaller image widths and lower line
densities produce higher system acquisition frame rates, but
decrease image resolution. Proper image acquisition should
balance a high frame rate with adequate image resolution.
11. The angle between the probe beam and flow for Doppler
velocity correction calculations introduces more error with
larger angles, as indicated by Eq. 4.
12. Velocity measurement variations will occur from ultrasound
probe placement angle measurement and biologic variations
between embryos.
13. From the onset of cardiac beating to about stage HH16, the
whole embryonic heart can be imaged with OCT. Between
HH17 and HH18, only the heart OFT can be completely
visualized on OCT images due to penetration limits. At stages
older than HH18, the heart becomes bigger (so that the lower
myocardium structures can no longer be distinguished) and
then it sinks down into the egg until it exceeds the maximum
depth penetration of the system.
14. Increasing the number of A-lines in each B-scan increases axial
resolution but decreases B-scan frame rate.
15. It is important to keep the distance between adjacent imaging
planes small so that structural features overlap in 4D
reconstruction.
16. Previous 4D acquisition of the OFT motion has used 200 time
frames at 140 fps, and 196 positions along the OFT spaced
7.5 μm apart, which takes about 20 min to image at HH18 [22].
17. M-mode images can also be extracted from B-mode ultrasound
images, except that the image resolution does not usually allow
for crisp M-mode images, and thus direct M-mode acquisition
is usually preferred to increase M-mode image resolution.
18. The line must be in similar orientation to the lumen slit/jelly
across datasets so as not to introduce any differences in radial
strains due to cardiac jelly influence on wall deformation.
19. Differences in periods across the dataset occur mainly because
of temperature fluctuations during imaging. Thus, temperature
needs to be controlled and only allowed to fluctuate minimally.
20. The M-mode A-lines are pooled to one cardiac cycle step to
improve the accuracy of the algorithm.
21. Cubic spline interpolation offers a good balance between algo-
rithm cost and performance. Other interpolation algorithms,
however, can also be used.
22. Constant contraction velocity is assumed. This is only true on
relatively small portions of the heart and is the case for the
length of the OFT. Phase lags should be evaluated in separate
regions of the heart to achieve accurate reconstruction.
23. 2D cross-sectional planes are used since the tubular tissue layer
forms a closed contour for segmentation. Image processing is
also much simpler on 2D frames than with 3D or 4D datasets.
24. Layer detection can sometimes fail and will have to be manually
corrected, especially in cases of close proximity between target
and adjacent tissues, incorrect initial DLM orientation, and
overlap of adjacent DLMs.
Acknowledgement
This work was supported by NIH grant R01 HL094570 and NSF
grant DBI 1052688.
References
1. Clark EB, Hu N, Frommelt P et al (1989) 13. Faber JJ, Green TJ, Thornburg KL (1974)
Effect of increased pressure on ventricular Embryonic stroke volume and cardiac output
growth in stage 21 chick embryos. Am J in the chick. Dev Biol 41:14–21
Physiol 257:H55–H61 14. Jenkins MW, Chughtai OQ, Basavanhally AN
2. Fishman MC, Stainier DY (1994) Cardiovas- et al (2007) In vivo gated 4-D imaging of the
cular development. Prospects for a genetic embryonic heart using optical coherence
approach. Circ Res 74:757–763 tomography. J Biomed Opt 12(3):030505
3. Groenendijk BC, Hierck BP, Vrolijk J et al (2005) 15. Li P, Yin X, Shi L et al (2011) Measurement of
Changes in shear stress-related gene expression strain and strain rate in embryonic chick heart
after experimentally altered venous return in the in vivo using spectral domain optical coherence
chicken embryo. Circ Res 96:1291–1298 tomography. IEEE Trans Biomed Eng
4. Hove JR, Koster RW, Forouhar AS et al (2003) 58:2333–2338
Intracardiac fluid forces are an essential epige- 16. Li P, Liu A, Shi L et al (2011) Assessment
netic factor for embryonic cardiogenesis. of strain and strain rate in embryonic chick
Nature 421:172–177 heart in vivo using tissue Doppler optical
5. Hamburger V, Hamilton HL (1951) A series of coherence tomography. Phys Med Biol
normal stages in the development of the chick 56:7081–7092
embryo. J Morphol 88:49–92 17. Liebling M, Forouhar AS, Gharib M et al
6. Hogers B, DeRuiter MC, Gittenberger-de (2005) Four-dimensional cardiac imaging in
Groot AC et al (1997) Unilateral vitelline vein living embryos via post-acquisition synchroni-
ligation alters intracardiac blood flow patterns zation of nongated slice sequences. J Biomed
and morphogenesis in the chick embryo. Circ Opt 10(5):054001
Res 80:473–481 18. Liebling M, Forouhar AS, Gharib M et al
7. Gittenberger-de Groot AC, Bartelings MM, (2005) Wavelet-based synchronization of non-
Deruiter MC et al (2005) Basics of cardiac devel- gated confocal microscopy data for 4-D imag-
opment for the understanding of congenital ing of the embryonic heart. Proc SPIE
heart malformations. Pediatr Res 57:169–176 5914:591409
8. Keller BB, Hu N, Clark EB (1990) Correlation 19. Liu A, Wang R, Thornburg KL et al (2009)
of ventricular area, perimeter, and conotruncal Efficient postacquisition synchronization of 4-
diameter with ventricular mass and function in D nongated cardiac images obtained from opti-
the chick embryo from stages 12 to 24. Circ cal coherence tomography: application to 4-D
Res 66:109–114 reconstruction of the chick embryonic heart.
9. Lin IE, Taber LA (1994) Mechanical effects of J Biomed Opt 14:044020
looping in the embryonic chick heart. J Bio- 20. Yin X, Liu A, Thornburg K et al (2012)
mech 27:311–312 Extracting cardiac shapes and motion of the
10. Keller BB, Yoshigi M, Tinney JP (1997) chick embryo heart outflow tract from four-
Ventricular-vascular uncoupling by acute con- dimensional optical coherence tomography
otruncal occlusion in the stage 21 chick images. J Biomed Opt 17(9):096005
embryo. Am J Physiol Heart Circ Physiol 21. Garita B, Jenkins MW, Han M et al (2011)
273:H2861–H2866 Blood flow dynamics of one cardiac cycle and
11. Stekelenburg-De Vos S, Steendijik P, Ursem relationship to mechanotransduction and tra-
NTC et al (2005) Systolic and diastolic ventric- beculation during heart looping. Am J Physiol
ular function assessed by pressure-volume Heart Circ Physiol 300:H879–H891
loops in the stage 21 venous clipped chick 22. Liu A, Yin X, Shi L et al (2012) Biomechanics
embryo. Pediatr Res 57:16–21 of the chick embryonic heart outflow tract at
12. Keller BB, Hu N, Serrino PJ et al (1991) Ventric- HH18 using 4D optical coherence tomogra-
ular pressure-area loop characteristics in the stage phy imaging and computational modeling.
16 to 24 chick embryo. Circ Res 68:226–231 PLoS One 7(7):e40869
INDEX
A Cell differentiation ..................................... 104, 109, 112,

113, 115–130, 145
Actin........................ 11, 56, 75, 76, 79, 86, 88, 147, 150 Chick embryonic cardiomyocytes ................................ 167
actin stress fibers........................................................ 76 Chick heart development .................................... 191–211
Adenovirus...................................... 1, 178, 179, 186–188
Collagenase
Ago2 ............................................... 28, 29, 33–35, 38, 46 Type 2 ............................................................. 181, 187
Aminoglycoside phosphotransferase cDNA ................ 168 Type IV .......................................... 106–110, 112, 135
Anisotropic Young’s modulus ........................................ 94
Computational methods................................................. 69
9-β-D-arabinofuranosyl cytosine.................................... 18 Computer simulation................................................65–73
AraC ..............................................................168, 171–173 Conditioned medium .......................................... 116, 140
Argonaute ........................................................................ 28
Confocal microscopy ...................................................... 11
Cre-loxP......................................................................... 179
B
Cryopreservation
B27 supplement ..................................107, 108, 117, 128 neonatal rat cardiomyocytes .......................... 153, 167
β mercaptoethanol ........................................................ 146 Cytofix/Cytoperm ............................................... 135, 141
Bioinformatics ...........................................................53, 55 Cytosine arabinoside ..................................................... 168
Bovine serum albumin (BSA)................11, 18, 118, 162,
165, 181, 182, 185, 188 D
Bowtie...........................................................31, 42–44, 47 DAPI (4’,6-diamidino-2-phenylindole) ................ 11, 25,
Bromodeoxyuridine (BrdU)................................ 158, 159
79, 88, 119, 124, 147, 150,
163, 169, 171, 174
C
Deep sequencing .......................................................27–48
Cardiac action potential............................................65–73 DESeq................................................................. 31, 44, 45
Cardiac cell line Diethylpyrocarbonate (DEPC) .................. 30, 32, 35, 36
HL-1 ........................................................................ 177 Differential equations ...................................... 65–72, 101
Cardiac cycle.......................192–196, 198–203, 205–211 Differential gene expression .....................................44–45
Cardiac electrophysiology.........................................65–73 Dimethylsulfoxide (DMSO)............................... 120, 135,
Cardiac fibroblast ...................................... 1–16, 159, 169 136, 154, 157, 158
Cardiac muscle ..................................................... 133–142 Displacements
Cardiac muscle cell death extracellular ......................................... 94, 96, 99, 101
apoptotic death ....................................................... 134 intracellular ..........................................................94, 96
oncotic death........................................................... 134 DNA
progressive necrotic death ...................................... 134 cDNA.............................................. 31, 34–37, 39, 40,
Cardiac progenitor ........................................................ 161 46, 52, 119, 125, 168, 177
Cardiac research ............................................................ 153 genomic DNA ......................................52, 59–61, 124
Cardiac safety pharmacology ........................................ 133 DNA fragmentation
Cardiac segmentation ................................................... 208 enzymatic digestion ................. 18, 52, 138, 140, 142
Cardiac striation ............................................................ 161 nebulization............................................................... 52
Cardiac structure .................................................. 192, 197 sonication................................................................... 52
Cardiomyocytes...........................................17–25, 65–73, DNA sequencing
76, 78, 81, 87, 88, 92–102, 105, 115–130, Bridge amplification ............................................53, 54
153–161, 164, 165, 167–169, 172–174 coverage depth ....................................................55, 60
Cardiomyogenic differentiation .......................... 146, 151 ion torrent .................................................... 40, 54, 55
Cardiotoxicity................................................................ 115 pyrosequencing ......................................................... 54
Cardiovascular disease...................................... 51–62, 134 DNase I................................................................. 119, 124
vol. 1299, DOI 10.1007/978-1-4939-2572-8, © Springer Science+Business Media New York 2015
213
214 C ARDIOMYOCYTES: METHODS
Index
AND PROTOCOLS
Double line model (DLM)......................... 208, 209, 211 Growth factors

4D reconstruction ...............................200, 202, 209, 211 actvin A .................................................................... 116
Drug discovery ..................................................... 115, 161 BMP4............................................106, 108–111, 116,
4D synchronization.............................................. 200, 205 118, 120–123
Dulbecco’s Medium
Dulbecco’s Modified Eagle Medium (DMEM)....... 3, H
12, 19, 105, 106, 109, 116–118, 135, 136, Hamburger-Hamilton stages........................................ 192
146, 168–170, 181, 185, 186, 188 Hanging drop....................................................... 146–148
F-12 (DMEM/F12) .........................19, 105, 106,
Hanks balanced salt solution (HBSS) .................. 78, 112,
109, 116, 117 154–157, 193, 197, 210
Dynabeads ................................................... 29, 33, 34, 46 Heart............................... 1–6, 12–14, 17–25, 28, 29, 31,
Dynal magnet ............................................................ 29
33, 45–47, 52, 65–73, 75, 88, 99, 103, 115,
130, 133–142, 155, 156, 160, 169, 179, 180,
E
182–185, 187–189, 191–211
ε-Aminocaproic acid ............................................ 108, 112 Hemacytometer.......................... 107–109, 122, 147, 148
E-miR ................................................................. 31, 42, 47 Hemocytometer ......................3, 7, 9, 21, 107–109, 122,
Electrical stimulation .................................. 168, 172, 173 134, 141, 147, 148, 158, 159, 186
Electrophysiology Heparin ........................................................ 180, 183, 187
cell patch................................................ 108, 112, 161 Homologous recombination ........................................ 178
clamp........................................................................ 161 HTSeq ................................................................ 31, 43–45
multi-electrode arrays (MEA) ................................ 161 Human basic fibroblast growth factor (bFGF) ..........105,
Embryoid bodies (EB) 106, 109, 118, 121, 122
human cell-derived.................................................. 146 Hyaluronidase ............................................................... 181
mouse cell-derived ......................................... 145–151 Hydroxyethyl piperazine ethane sulfonic acid
Embryonic stem cells (ESC) ............................... 116, 154 (HEPES) ............... 18, 78, 79, 108, 112, 154,
Enzymatic digestion....................... 18, 52, 138, 140, 142 181, 182
Ethidium bromide monoazide (EMA) .......................120, Hypertrophic cardiomyopathy .................................52, 57
126–130
Experimental cardiomyopathies ..................................... 28 I
Explant culture ..................................................... 136, 140 Imaging
Extracellular matrix (ECM)..................75, 78, 80, 86, 88
optical ................................... 161, 191–196, 199, 209
optical coherence tomography (OCT) ................. 192,
F
200–203, 205, 207–210
FastQC................................................................ 31, 41, 43 ultrasound..................................... 191–193, 197–199,
Fetal bovine serum (FBS) ..................................... 78, 105, 210, 211
107–109, 111, 112, 117, 119, 123, 135, 136, Immunocytochemistry...........................11, 21, 124, 146,
146, 154, 159, 169, 170, 181, 185, 186, 188 147, 149, 159
Fibrillogenesis............................................................75–90 Immunohistochemistry ...................11, 21–25, 124, 146,
Fibrin .......................................... 105, 108–110, 112, 113 147, 149, 159
Fibronectin .................................... 78, 88, 106, 108, 110, Immunoprecipitation............................28, 29, 33, 46, 47
157, 158, 165 Integrin ..................................... 75–77, 80–85, 88, 94, 99
Flow cytometry ................................. 111, 116, 120, 121, Interferon β ....................................................................... 2
123–125, 127, 129, 141 Intracellular cytoskeleton....................................... 94, 101
Intracellular hydrostatic pressure ................................... 94
G Isotropic shear stress ....................................................... 94
Gelatin .....................................19, 21, 24, 124, 146–150,
K
163, 165
Gene annotation ............................................................. 44 Krebs Henselait buffer (KHB) .................. 179, 181–185,
Gene transfer ........................................................ 177–189 187, 188
GenElute.......................................................................... 59 Kruftbr€
uhe solution ...................................................... 167
Genetics .....................................................................51–62
Gentamicin ....................................... 3, 12, 135, 136, 154 L
Glutamine ..................................12, 18, 19, 78, 106, 109, Lactate .................................................116, 168, 171, 172
146, 154, 181, 182 Laminin................................................................. 181, 186
CARDIOMYOCYTES: METHODS AND PROTOCOLS
Index 215
Leukemia inhibitory factor (LIF)........................ 145, 146 Penicillin/streptomycin ................................................ 146
Linux................................................................... 41, 42, 67 Percoll gradient .................................................... 159, 167
Perfusion.............................................179, 180, 182–184,
M 187, 188
Phosphate buffered saline (PBS)................2–7, 9–14, 18,
Markov models..........................................................66, 68
Mathematical models ................................................65–73 78, 79, 118, 122–124, 126, 127, 135, 140,
Matlab..................................................67, 72, 77, 88, 193 141, 146–151, 162, 163, 165, 169, 171, 181
Matrigel ..................................... 107, 108, 110–113, 117, Phosphoinositides ........................................................... 76
Pluripotent stem cells
120–122, 124, 129, 165
Mechanoelectric feedback............................................... 99 human induced pluripotent stem cells
Mechanotransduction ........................................ 76, 94, 99 (hiPSC) ............................................... 104–113
human pluripotent stem cells (hPSC).......... 115–117,
Membrane potential..................... 65, 68, 70, 71, 73, 161
Methylcellulose ............................................................. 146 120
Mice induced pluripotent stem cells (iPSC) ......... 103–113,
154
BALB/c ..................................................................... 20
Swiss Albino .............................................................. 20 Pluronic F-127 ................................................................ 78
Migration................................................... 76, 77, 87, 139 Poly-L-ornithine ........................................................... 178
Polyacrylamide gel ....................................................32, 79
MiRdeep ....................................................................31, 42
Modeling Polydimethylsiloxane (PDMS) .................................78, 88
chemo-mechanical model ......................................... 77 Polymerase chain reaction (PCR)
electrical bidomain model ........................................ 99 emPCR ................................................................52, 54
qPCR ...................................................................12, 28
mathematical modeling ......................................93, 99
mechanical bidomain model.............................93–102 qRT-PCR.........................................12, 116, 121, 124
Mouse embryonic fibroblasts (MEF)................. 105, 109, RT-PCR ......................................... 124–126, 146, 151
Primary cell culture ...................................................2, 176
116, 117, 120–122
Mr. Frosty freezing container....................................... 135 Propidium iodide ....................................... 107, 111, 169,
Myocardial infarction ........................................... 103–113 171–173
Purkinje network............................................................. 65
Myocardial wall motion ....................................... 191–211
Myocarditis .................................................................. 1–16
Q
Myocyte isolation ................................180–182, 184, 187
Myofibrillogenesis .....................................................75–90 QIAquick ......................................................30, 36–38, 40
Myofibrils Qubit fluorimeter............................................................ 60
myofibril bundles ...................................................... 75
pre-myofibrils .................................. 76, 81, 83, 84, 88 R
Read alignment ............................................................... 41
N
Reentrant wave..........................................................66, 69
Nanodrop .........................................................32, 33, 119 Regenerative medicine ................................ 103, 134, 161
Nembutal..................................................... 180, 183, 187 Reovirus ......................................................................... 1, 2
Neonatal mice ...........................................................20, 24 Rho GTPases ................................................................... 76
Neonatal rat cardiomyocytes (NRCM).......................153, Ribo-Zero........................................................................ 34
154, 156, 158, 159 RiboMinus ....................................................................... 34
Next generation sequencing (NGS) ...................... 51, 52, RNA
55–57, 60, 61, 116, 118, 124 microRNA ........................................................... 27–48
Non-essential amino acids ...............................19, 78, 146 mRNA.................................................................. 27–48
Non-human primate mRNA binding proteins ........................................... 28
Cynomolgus monkey .............................................. 134 RNA-induced silencing complex (RISC) ...............28–30,
Rhesus monkey .............................................. 134, 136 32–34, 36, 38–40, 42, 43
Numerical integration........................................ 66–68, 72 RNase H ....................................................................35, 36
RNeasy .................................................................. 119, 124
P Roswell Park Memorial Institute (RPMI) growth
Paraformaldehyde (PFA) .......................19, 21, 118, 124, medium ..................................... 108, 111, 112,
127, 162, 165 117, 120–124
216 C ARDIOMYOCYTES: METHODS
Index
AND PROTOCOLS
S TruSeq ............................................ 29, 32, 57, 59, 60, 62

Trypan blue ....................................... 19, 21, 24, 25, 118,
Sarcolemmal proteins.................................................... 165 122, 124, 126, 135, 141, 155, 157, 159
Sarcomere ..................................57, 75, 76, 86, 161, 162, Trypsin ..................................................3, 6, 9, 12, 13, 19,
164, 172, 173, 178, 188 20, 24, 78, 88, 107, 108, 110, 112, 118, 119,
Sarcoplasmic channel proteins............................. 164, 166 135, 140, 146, 148, 154–156, 168, 169,
Serum free medium..................................... 116, 120, 123 172, 187
Shear modulus...........................................................92, 94 Trypsin inhibitor ................... 3, 6, 12, 13, 118, 119, 123
Sprague Dawley rats..................................................78, 88 Trypsin/EDTA .....................................12, 123, 147, 148
SuperScript III.................................................... 30, 35, 36 Tween 20.............................................................. 162, 165
T V
Taqman ........................................................ 119, 125, 126 Vascular endothelial growth factor (VEGF) ...............106,
Thrombin ........................................................... 106, 108, 108, 110, 112
110, 112 Ventricular area.............................................................. 195
Tissue displacement Ventricular pressure....................................................... 195
bidomain.................................................................... 98 Ventricular volume........................................................ 195
extracellular ......................................... 96, 98, 99, 101 Versene............................................... 106, 107, 109, 118,
intracellular ...........................................94–96, 99, 101 120, 122, 129
monodomain .............................................. 67, 98–102 Viral myocarditis
Tissue engineering ..............................103, 104, 115, 134 adenovirus .................................1, 178, 179, 186–188
TopHat ............................................................... 31, 43, 44 coxsackievirus B3 ........................................................ 1
Traction force microscopy ........................................79, 88 enterovirus ................................................................... 1
Transcription factors Viral vectors
GATA-4 ................................................. 19, 21, 23, 25 adeno-associated virus (AAV)................................. 178
Nkx2.5 ............................................. 25, 162, 164–166 adenovirus ................................ 1, 178, 179, 186, 188
Transforming growth factor (TGF) ............................106, lentivirus .................................................................. 178
110, 116
Triton X-100 ........................................10, 149, 162, 165, W
169, 171
Trizol .....................................................29, 31, 34, 45, 46 Whole genome sequencing ......................................51, 55

(Methods in Molecular Biology 1299) Gary R. Skuse, Maureen C. Ferran (Eds.) - Cardiomyocytes - Methods and Protocols-Humana Press (2015)

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

(Methods in Molecular Biology 1299) Gary R. Skuse, Maureen C. Ferran (Eds.) - Cardiomyocytes - Methods and Protocols-Humana Press (2015)

Diunggah oleh

Hak Cipta:

Format Tersedia

Methods in

Molecular Biology 1299

For further volumes:

Methods and Protocols

Gary R. Skuse and Maureen C. Ferran

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Springer New York Heidelberg Dordrecht London

Printed on acid-free paper

Humana Press is a brand of Springer

Rochester, NY, USA Gary R. Skuse

1 Generating Primary Cultures of Murine Cardiac Myocytes

14 Electrotonic Coupled Metabolic Purification of Chick Cardiomyocytes . . . . . . 167

ANSHU ARORA  Biomolecular Science Center, Burnett School of Biomedical Sciences,

CHRISHAN J.A. RAMACHANDRA  Stem Cell Laboratory, National Heart Research

Generating Primary Cultures of Murine Cardiac Myocytes

Viral myocarditis (cardiac damage) [1–3] is the second leading

myocytes are largely nondividing and unable to self-renew after

2.2 On Lab Bench 1. Bench paper.

6. Two sterile small “flea” (7 2 mm) spin bars (VWR #58948-

2.4 Additional 1. Refrigerated tabletop centrifuge set at 4 C.

2.5 Reagents 1. Ethanol for sterilizations.

3.1 Euthanizing Mice 1. Euthanize nonpregnant dams by IACUC-approved protocol.

9. Use the same sterile forceps to transfer the hearts ~equally to

Final volume including

14. Use 5–10 ml cDMEM to resuspend the pellet.

3.6 Infecting 1. After incubating cultures overnight, assess by light microscopy

3.7 Assessing 1. Plate cultures on Poly-D-Lysine-coated 8-well chamber slides

10. Wash with ice-cold PBS as for step 8.

virus DNA or RNA (e.g., qPCR or qRT-PCR) or protein (e.g.,

1. All procedures, including housing mice, mating mice, and

next neonate or decapitate a group of five to ten neonates

20. A significant fraction of cells do not pellet in the first centrifu-

29. Cardiac fibroblast cultures are difficult to visualize by light

34. Paraformaldehyde should be disposed of properly as hazardous

This work was supported by NIH grant R01AI083333. I thank

Enrichment of Cardiomyocytes in Primary Cultures

Neonatal murine cardiomyocyte cultures have been used as a model

Here, we demonstrate a protocol for the primary culture of

Prepare all solutions using UPW (ultrapure water—prepared by

2.1 Buffer 1. Dulbecco’s Phosphate Buffered Saline (DPBSA): Place a clear

2.2 Medium Commercial media are available as powdered media, 1 working

1. MAINTENANCE MEDIUM—DMEM/F12 (1:1) with fetal

2.3 Enzymes, 1. 0.05 and 0.5 % Trypsin-EDTA; (Invitrogen, Carlsbad, CA).

2.5 Instruments 1. Class II A Bio-safety cabinet.

Carry out all procedures at room temperature unless specified

5. Spin down at 2,205 g (RCF) for 10 min at 4 C and resus-

3.3 Characterization 1. Fix the cells by incubating them at room temperature in an

Fig. 1 Phase contrast photomicrograph of cultures showing different morpholo-

5. Add an adequate volume of Hoechst 33342 (5 μg/mL)

Fig. 2 Immunocytochemical analysis of GATA-4 and Nkx-2.5 in cardiomyocytes.

1. Handle the animals gently in order to minimize stress that may

This work is supported by grants to R.S.V. by the Ministry of

Deep Sequencing of Cardiac MicroRNA-mRNA Interactomes

Cellular homeostatic maintenance and responses to stress stimuli

transcripts into their active forms [1–4]. The mechanism of action of

by Karginov et al. [24], who evaluated microRNA targets in 293T

2.1 miR-Seq Library 1. Trizol (Invitrogen).

2.2 PolyA+ 1. Invitrogen oligo(dT)20 Dynabeads (Invitrogen #610-06) for

2.3 Ago2 1. Anti-mouse Ago2 monoclonal antibody (Wako Pure, clone

Name Sequence 50 –30

8. Enzymes and reagents required for adapter ligation to cDNA

2.5 Suggested 1. FastQC (http://www.bioinformatics.babraham.ac.uk/projects/

fraction, increase the length of the isopropanol precipitation

ANSHU ARORA Biomolecular Science Center, Burnett School of Biomedical Sciences,

CHRISHAN J.A. RAMACHANDRA Stem Cell Laboratory, National Heart Research