and about 52 A
gap spanned by the channels subunits [Beyer et al., 1995]. The diameter
of the channel ranges between 20 and 30 A
and about 15 A
in the extracellular
half of the lipid bilayer. The inner pore appeared solvent lled, and is =2.5
nm at its widest point [Severs, 1994a, b]. Transmission electron microscopy of
positively stained cross-sectioned cardiac gap junctions of mammalian
ventricles or atria revealed a 7-layered structure: two 3-layered lipid membranes
and the gap between them. More recently Perkins et al. [1997] developed a
three-dimensional model of the connexon with 50 A
3
symport as well as other systems, e.g. the Na
+
/
Ca
2+
exchanger [Doering et al., 1996] and possibly the Na
+
/Ca
2+
-ATPase.
Thus, Yang et al. [1996] described that Na
+
/HCO
3
cotransport and Na
+
/H
+
exchange contribute to the rate of cell-to-cell electrical uncoupling in ischemic
myocardium potentially related to an attenuated Na-dependent calcium load-
ing. In addition, inhibition of proton extrusion with 1 mmol/l amiloride was
reported to enhance the eects of changes in pH on g
j
[Firek and Weingart,
1995].
As a decrease in pH
i
an elevation in pCO
2
can cause a dramatic decrease
in coupling in amphibian embryos [Turin and Warner, 1978] and other tissues
[Spray et al., 1985]. This has been used experimentally to uncouple prepara-
tions. The CO
2
-induced eect is reversible. The decrease in coupling after
exposure to CO
2
has been ascribed to the consecutive fall in pH
i
[Kolb and
Somogyi, 1991]. An eect of increasing CO
2
in the ventilation air of anesthe-
tized dogs on the cardiac activation pattern has been described and was
attributed to gap junctional uncoupling [Vorperian et al., 1994]. However, this
cannot be transferred easily on other pathophysiological conditions, since a
single increase in pCO
2
is seldom, and is often (in pathophysiological situa-
tions for e.g. myocardial infarction) accompanied by other changes includ-
ing depolarization, potassium eux and many others, which will also aect
the sodium channel availability, which will reduce longitudinal propaga-
tion velocity. That means that, under pathophysiological conditions, aec-
43 4 Function and Physiology of Gap Junction Channels
tion of the gap junctional channel will probably not be mono- but multi-
causal.
Increasing Mg
2+
has also been reported to cause a fall in junctional
conductance in pairs of adult guinea-pig cardiomyocytes [Noma and Tsuboi,
1987]. This eect was established for a pMg range from 2 to 3 corresponding
to 1 to 10 mmol/l at pH 7.4 in the absence of calcium. The obtained data
could be described by the equation:
g
j
>1(1/{1+(K
E
Mg
/[Mg])
n
}),
with K
E
Mg
>3.16 10
3
mol/l (pK
E
Mg
>2.5) and n>3. A Hill coecient of 3.0 was
calculated. Since the slope of the pCa-g
j
and the pMg-g
j
relationships were
similar, it was suggested that both divalent cations bind to the same receptor
site. The uncoupling eect of magnesium has been observed in other systems
as well as, for example insect salivary gland [Oliveira-Castro and Loewenstein,
1971]. The cardiac cytosolic free Mg
2+
has been measured in the range of
0.48 mmol/l [Murphy et al., 1989] which is below the concentrations in which
magnesium induces gap junctional uncoupling.
Another ion known to be involved in the regulation of gap junction
conductance is Na
+
. Na
+
withdrawal in adult rat cardiomyocytes induced
electrical uncoupling as indicated by a decrease in g
j
/g
j max
occurring within
3 min after exposure to 0 mmol/l Na
+
[Maurer and Weingart, 1987]. This has
been ascribed to the fact that lowering [Na
+
]
0
was reported to result in both
an increase in intracellular calcium and a decline in intracellular sodium
concentration, which have been interpreted as an impairment in the Na
+
/Ca
2+
-
exchange mechanism, since calcium extrusion via this mechanism requires the
transport of sodium [Weingart and Maurer, 1987]. Besides this, De Mello
[1976] described an increase in intracellular sodium concentration to cause
uncoupling within 500 ms in Purkinje bers as indicated by an increase in
input resistance. It is uncertain whether this was a direct eect of sodium or
may be secondary to a rise in intracellular calcium via the Na
+
/Ca
2+
-exchange
mechanism.
As described in the paragraphs above, conclusions on a direct eect of
any of these ions should be drawn with caution, since a change in the concentra-
tion of any of these may activate a regulatory mechanism to compensate for
the change, for example the Na
+
/H
+
exchanger, the Na
+
/HCO
3
symport or
the Na
+
/Ca
2+
-exchange mechanism.
In addition to ions, other small molecules have been described to play an
important physiological and pathophysiological role in the regulation of gap
junctional resistance. Thus, ATP acts as an important regulator. In 1979
Wojtcak described that hypoxia in glucose free solution resulted in a rise in
R
j
in cow ventricular trabeculae indicating that the intracellular ATP content
44 4 Function and Physiology of Gap Junction Channels
may participate in the regulation of gap junctional conduction. Lowering the
intracellular ATP concentration down to 0.5 mmol/l from the approximately
physiological concentration of 5.0 mmol/l led to a rapid decline in g
j
[Sugiura
et al., 1990] in pairs of adult guinea-pig cardiomyocytes. The investigators
kept intracellular calcium and magnesium concentrations at levels less than
10
9
and 0.3 10
3
mol/l, respectively. The decrease was reversible upon addition
of ATP. Similar to the method used by Noma and Tsuboi [1987], they deter-
mined the Hill coecient for ATP with 2.6 and the half maximum cytosolic
ATP concentration in the order of 0.68 mmol/l, suggesting a total uncoupling
(g
j
>0) if ATP is reduced below 0.1 mmol/l. The decrease in g
j
induced by
lowering [ATP] to 0.5 mmol/l was not reversible by adding ADP (10 mmol/l)
or 50 mol/l cAMP or 1 mol/l of the catalytic subunit of cAMP-dependent
protein kinase in these experiments. Thus, at Ca
2+
, Mg
2+
and H
+
concentra-
tions considered to be approximately in the physiological range, [ATP] acts
as a regulator of g
j
independent of cAMP-dependent phosphorylation. The
authors suggested a direct eect of ATP via a specic ligand-receptor inter-
action with the gap junctional proteins. Using metabolic inhibition by 2,4-
dinitrophenol (or decrease to 0.1 mmol/l ATP) in adult guinea-pig cardiomy-
ocytes Morley et al. [1992] also described an increase in gap junction resistance
6 min after addition of 2,4-dinitrophenol. However, the increase in R
j
was too
small to impair cell-to-cell propagation in this experimental system.
With regard to other possible regulators of g
j
, arachidonic acid, which
can be released from membrane phospholipids by activation of phospholipase
A
2
secondary to activation of a variety of receptors, has been investigated.
It became obvious from these experiments that in very high concentrations
(50100 mol/l) arachidonic acid can evoke cellular uncoupling within several
minutes in rat lacrimal gland cells [Giaume et al., 1989]. Since inhibitors of
arachidonic acid metabolism did not prevent the arachidonic acid eect, it
was suggested that, at least under certain conditions, arachidonic acid (as a
fatty acid) may interact directly with the gap junction proteins or their lipid
environment. It could be imagined that it is incorporated in the lipid bilayer
and alters the geometry of the lipid surrounding of the channels as was
suggested for the eect of other lipophilic agents, e.g. heptanol, octanol and
halothane. In neonatal rat heart cells, arachidonic acid has also been observed
to induce uncoupling [Schmilinsky-Fluri et al., 1990]. This was further investi-
gated by Massey et al. [1992], who described a concentration-dependent eect
of arachidonic acid on g
j
in neonatal rat heart cells in a physiologically more
relevant concentration range from 2 to 20 mol/l. This uncoupling eect could
be antagonized by inhibition of lipoxygenase with U70344A, but not with
indomethacine suggesting that the arachidonic acid eect at that concentration
is mediated by a lipoxygenase metabolite, e.g. a leukotriene, but not by a
45 4 Function and Physiology of Gap Junction Channels
cyclooxygenase metabolite. The incorporation of arachidonic acid in that
study was 0.695 mol min
1
. Thus, at least two eects of arachidonic acid have
to be considered: (1) an unspecic eect in high concentration probably due
to physicochemical interaction within the gap junction surrounding, and
(2) lipoxygenase metabolites inhibiting the channels in a yet unknown
mechanism.
Acetylcholine is involved in many aspects of the regulation of the cardio-
vascular system. Thus, it may also play a role in the control of intercellular
communication. Very early in gap junction research the eect of acetylcholine
as an important transmitter on gap junction conductance has been investigated.
First, Petersen and Ueda [1976] demonstrated an increase in junctional resis-
tance in pancreatic acinar cells following the application of acetylcholine.
Concomitantly, the release of amylase was stimulated. A minimum concentra-
tion of 1 mol/l acetycholine was required to evoke uncoupling. The next
question was, how is the acetylcholine eect mediated? Calcium has been
considered to contribute to the mechanism of action [Iwatsuki and Pertersen,
1978], but does not seem to be the sole mediator as Neyton and Trautmann
[1986] demonstrated an uncoupling eect in a double whole-cell technique
although calcium was strongly buered using 20 mmol/l EGTA in the pipette
solution. PKCstimulation has been discussed to participate in the transduction
of the acetylcholine eect [for a review see, Kolb and Somogyi, 1991]. In rat
submandibular gland cells Kanno et al. [1993] demonstrated a reduction in
dye coupling from 97.2% (percentage of dye-coupled cells) to 75% and nally
22.7% after application of 10
6
and 10
4
mol/l acetylcholine, respectively. The
eect occurred within 10 min. A similar result was found in rat pancreatic
acinar cells following the administration of 5 mol/l acetylcholine, which re-
sulted in cellular uncoupling (dye coupling method) and a 4- to 5-fold increase
in amylase release [Chanson and Meda, 1993]. This eect was independent
of cycloxygenase, calcium and PKC, but could be inhibited by 1 mol/l ocadaic
acid, an inhibitor of serine-threonine phosphatases, indicating the involvement
of a phosphatase in the acetylcholine action.
In neonatal heart cells Takens-Kwak and Jongsma [1992] investigated the
eect of acetylcholine. They reported a reduction in the intercellular current
(I
j
) in response to 100 mol/l of the parasympathomimetic drug, carbachol,
which could be mimicked by 8-Br-cGMP. The eect only occurred in the
whole cell patch conguration, but not in the perforated patch conguration,
suggesting that a cytosolic enzyme is necessary for the eect which is washed
out by the pipette in the whole cell patch. Since they found the carbachol
eect to be antagonized by alkaline phosphatase Takens-Kwak and Jongsma
[1992] concluded that a cytosolic phosphatase is involved in the action of
carbachol and, thus, probably of acetylcholine, too.
46 4 Function and Physiology of Gap Junction Channels
Fig. 13. Synopsis of the physiological regulators of gap junction channels. >PKC
changes the substate of conductance, *>not found by all investigators as dierences between
the various isoforms of connexins or species variabilities.
Another group of transmitters involved in the control of the cardiovascular
system by the autonomous nervous system includes the catecholamines, adren-
aline and noradrenaline. In acinar submandibular gland cells of the rat the ad-
ministration of 10
4
mol/l adrenaline elicits a reduction in dye coupling from 97
to75.3%dye-coupledcells [Kannoet al., 1993]. This couldnot be mimickedwith
isoprenaline, but was inhibited with phenoxybenzamine. Thus, the uncoupling
eect of adrenaline in this preparation is mediated by stimulation of the -adre-
noceptor, whereas a stimulation of the |-adrenoceptor has no eect.
In contrast, stimulation of the |-adrenoceptor in the heart increases inter-
cellular coupling [Veenstra, 1991b]. However, one has to be cautious with
generalizations because, following adrenergic stimulation in the intact heart,
intracellular calcium and heart rate will also be enhanced, so that a complex
eect will occur which is dicult to assess experimentally. Perhaps epicardial
mapping experiments measuring the anisotropic ratio, i.e. the ratio between
longitudinal and transverse conduction velocity with regard to the ber axis,
will give insight into the global eect of such a manipulation. De Mello [1986b]
reported that adrenaline increased the spread of electrotonic potentials during
47 4 Function and Physiology of Gap Junction Channels
diastolic depolarization in canine Purkinje bers probably due to a rise in
intracellular [cAMP].
Very recently an eect of the broblast growth factor-2 (FGF-2; a major
member of the heparin-binding family of growth factors) on Cx43 in cardiac
myocytes has been described [Doble et al., 1996] possibly involving PKC or
MAP kinase and tyrosine phosphorylation. The authors showed that incuba-
tion with 10 ng/ml FGF-2 for 30 min induces Cx43 phosphorylation on serine
residues, with a concomitant loss in intercellular dye coupling and masking
of Cx43 epitopes located in residues 261270. In a previous study it was already
shown that basic FGF exists in close association with cardiac gap junctions,
and it has been suggested that it, thus, may play a role in gap junctional
intercellular communication [Kardami et al., 1991]. FGF-2 can be released,
for example, from cardiomyocytes during contraction and after stimulation
with catecholamines. The factor is upregulated in response to myocardial
damage. In contrast to these ndings, FGF-2 induced an increase in Cx43
accumulation and, as a result, an enhancement of coupling between cardiac
broblasts and capillary endothelial cells, respectively [Doble and Kardami,
1995; Pepper and Meda, 1992].
The mechanisms described so far are synoptically summarized in gure 13.
An important point to mention is that, as already said, the aection of the
gap junction conductance is not mono- but multicausal under the most physio-
logical and pathophysiological conditions due to the interactions between the
intracellular mediators. Thus, most processes will aect intracellular calcium
and, on the other hand, a change in intracellular calcium will activate a variety
of intracellular mechanisms and aect the activity of many calcium-dependent
enzymes.
4.2 Functions in Heart and Vasculature
What arethefunctions of intercellular communicationchannels intheintact
organ? In spite of many details regarding single-channel conductance and the
overall conductivity or permeability of gap junctions, their role in the mature
and developing heart is currently under intensive investigation and we are prob-
ably only at the beginning of an understanding of the role of intercellular com-
munication. Experimentally, it is not possible to measure the gap junction
current in an intact heart. However, modern experimental setups will make it
possible to get a deeper understanding for the role of these channels in intact
tissue. Such setups include mapping experiments using voltage-sensitive dyes
like di8-ANEPPS and epicardial potential mapping. With these techniques it is
possible to visualize the spread of activation and to measure its velocity.
48 4 Function and Physiology of Gap Junction Channels
An essential role for the normal cardiac development has recently been
shown [Reaume et al., 1995] in mouse lacking Cx43 (see chapter 6). Intercellular
coupling obviously is a prerequisite for the correct development.
Furthermore, synchronization of contraction is facilitated by gap junc-
tional communication as well as synchronization of electrical activation. The
electrical coupling between cardiomyocytes mitigates dierences in the mem-
brane potential between these cells, for example in the course of an action
potential if both cells repolarize at dierent timepoints. This results in smaller
dierences in the repolarization times thereby causing a reduction in the
dispersion of the action potential duration. Since increased dispersion is known
to make the heart more prone to reentrant arrhythmia, sucient gap junctional
communication can be considered as an endogenous arrhythmia-preventing
mechanism. For a detailed discussion of the role of gap junctional communica-
tion in the biophysics of cardiac activation as related to anisotropy, nonuni-
formity and stochastic phenomena, see chapter 1; for a discussion of their role
in arrhythmia, see chapter 6, and for a possible pharmacological intervention at
the gap junctions for suppression of arrhythmia, refer to chapter 7.
As already pointed out (see chapter 3), in some specialized areas of the
heart, e.g. sinus node, there are special interdigitating patterns of gap junctions,
providing some form of isolation from the hyperpolarizing inuence of other
cells, for example the surrounding atrium. Interestingly, these gap junctions
are made from Cx40, which is more sensitive to the transjunctional voltage
than Cx43 channels, thereby providing better isolation if higher dierences in
the membrane potential occur. However, such a physiological function can be
imagined but has not been shown directly, yet.
The communication between cells via gap junctions can also provide
exchange of small molecules and has been shown to protect cells against
oxidative stress by exchange of glutathion [Nakamura et al., 1995]. Exchange
of small molecules may have more functions when considering signalling
molecules such as cAMP or Ca
2+
. However, there are no experiments available
at present on the role of gap junctions in intact organs.
Apart from providing communication, gap junctions can, on the contrary,
have the function to isolate cells from their surrounding. Such isolation by
closure of gap junctions in certain pathophysiologic conditions occurs, for
example, in the course of hypoxia [Wojtcak, 1979] and loss of ATP or during
myocardial ischemia. This may have the advantage that these cells no longer
communicate and participate in myocardial contraction, thereby saving energy,
although this has not yet been demonstrated.
In the vasculature release of local mediators such as endothelin, prosta-
cyclin and nitric oxide can only aect the local vasotone (at the site of release)
or regulate downstream constriction or dilation. Besides this, upstream regula-
49 4 Function and Physiology of Gap Junction Channels
tion has been observed, but not well understood. Often this was ascribed to
the inuence of the perivascular nerves. However, since Beny and Connat
[1992] showed that smooth muscle cells of the media of pig coronary arteries
are dye coupled via single hydrophilic channels, upstream regulation of vasoac-
tivity and transmural regulation (from the luminal side to the periphery and
in the opposite direction from the perivascular nerves to the lumen) has
been also considered to be inuenced or provided by gap junctional coupling
(g. 12).
4.3 Electrophysiology and Voltage-Dependent Gating
In general, with regard to the current-voltage relationship there are two
types of gap junction channels to distinguish: (a) channels with rectifying
behavior, and (b) nonrectifying channels. In addition, the gap junction channels
can be distinguished by their single-channel conductance. The single-channel
conductance of a given gap junction channel made of one connexin isoform,
however, may exhibit dierent substates of single-channel conductance. In
general, the following states can be distinguished: open states with (a) main
open state, (b) several substates, (c) residual state, and (d) the closed state.
Furthermore, it is important to dierentiate between channels exhibiting sensi-
tivity to transjunctional voltage and channels being more or less insensitive.
For investigation of the current-voltage relationship there are principally three
possibilities: (a) investigation of freshly dissociated cells with the advantage
that these cells are isolated from an intact tissue and should resemble the
properties of these postmitotic cells rather well, and with the disadvantage
that in the course of some hours many of the channels are internalized or
dissociate so that the investigator observes a so-called run-down with regard
to the coupling in such cell pairs; (b) investigation of cultured cells, e.g.
embryonic chick heart cells or neonatal rat cardiomyocytes or others, and (c)
investigation of gap junctional channels in transfected cells, e.g. SkHep1 cells,
tumor cell lines or xenopus oocytes. In such elegant transfection systems a
problem may occur if regulatory processes are to be investigated which involve
pathways not present in the transfected cell.
To investigate the current-voltage relationship in any of the given systems,
the standard method is the double-cell voltage-clamp technique [Spray et
al., 1981, 1985; Weingart, 1986] carried out as a double whole-cell patch as
described by Giaume [1991]. The principle is that a transjunctional voltage
dierence (by clamping the two cells to dierent potentials) is applied for a
short time to a pair of coupled cells and the current necessary for maintenance
of the voltage dierence is measured. In order to achieve such an experimental
50 4 Function and Physiology of Gap Junction Channels
Fig. 14. Experimental setup for double whole-cell patch measurement of the gap junction
current.
setup two voltage-clamp ampliers are connected via a patch-clamp pipette
to either cell (g. 14). While in one cell the membrane potential is kept for
example at 40 mV (in order to inactivate the sodium current), the membrane
potential of the other cell is set for example to 30 mV, thereby applying a
transcellular voltage of 10 mV. In this way transcellular voltages ranging
from 50 to +50 mV or from 100 to +100 mV are applied and symmetry
is controlled by alteration of the cell being kept at 40 mV. Since current
can ow across the cell membrane and across the junction between both
cells, the current in cell 1 I
1
can be described under these conditions by the
equation:
I
1
>(V
1
/r
m1
)+[(V
1
V
2
)/r
j
]
and accordingly the current in cell 2 by
I
2
>(V
2
/r
m2
)+[(V
2
V
1
)/r
j
],
with V
1
and V
2
being the voltage relative to the holding potential V
H
(e.g.
40 mVin order to inactivate the sodiumcurrent) and r
m1
and r
m2
the membrane
resistance in cell 1 and 2, respectively. If cell 2 is kept at V
H
, currents I
1
and
I
2
can be described as:
I
1
>(V
1
/r
m1
)+(V
1
/r
j
)
I
2
>V
1
/r
j
51 4 Function and Physiology of Gap Junction Channels
so that I
2
is a direct measure of the current owing across the junctional
membrane. Problems may arise from the series resistance of the pipettes and,
in some preparations, eventually from the cytosolic resistance or from the
ratio between these dierent resistances (for more details see chapter 8). The
current measured is then plotted against the transcellular voltage V
j
. Linear
regression reveals the total intercellular resistance. g
j
follows the equation:
g
j
>N
j
P
0
,
with N>number of channels capable of opening and closing, P
0
being the open
probability and
j
the single-channel conductance. If, under the experimental
conditions described above, cells are progressively uncoupled by application
of, e.g., heptanol or halothane, it is possible with some types of ampliers
to observe single-channel openings shortly before total uncoupling occurs.
Alternatively gap junction channels can be reconstituted in lipid bilayers for
observation of single-channel conductances. More details and protocols are
given in chapter 8.
Looking at the current measured using such a protocol with a pulse
duration of about 2 s, one can distinguish two components of the junctional
current: (a) the instantaneous component, and (b) the steady-state component.
Plotting the instantaneous current I
j
versus V
j
may reveal a linear relation as
shown by Veenstra et al. [1993], which means that the instantaneous gap
junctional conductance is constant, i.e. g
j
>I
j
/V
j
>constant. A linear current
voltage relation under such conditions means that the junctional channel
behaves like an ohmic resistor with a constant resistance which is insensitive
to the transjunctional voltage. However, not in all cases an ohmic behavior
will be seen. If with increasing transcellular voltage the current does not follow
a linear relation, some kind of rectication is present. Rectication means that
the channel resistance increases or decreases with increasing or decreasing
transjunctional voltage, i.e. the channel favors a current in one direction or
at a special range of transcellular voltage. In other words it behaves comparable
to some kinds of diodes.
Investigation of the steady-state current and the steady-state conductance
is normally carried out by tting the normalized g
ss
/V
j
relationships with the
two state Boltzmann distribution which follows the function:
g
ss
/g
inst
>{(g
max
g
min
)/(1+exp[A(V
j
V
0
)])}+g
min
according to Spray et al. [1981] and Veenstra et al. [1993] with g
max
being the
maximum conductance (>1, normalized to instantaneous g
j
) and g
min
the
minimum conductance. V
0
is the half-inactivation voltage where g
j
is between
g
min
and g
max
and A>zq/kT (z>number of equivalent electron charges, q>
voltage sensor, k>Boltzmann constant, T>temperature).
52 4 Function and Physiology of Gap Junction Channels
Fig. 15. Typical example of a measurement of junctional conductance. For details see
text. (Freshly isolated adult-guinea pig cardiomyocytes, holding potential 40 mV, series
resistance was overcome by using switch-clamp ampliers (SEC05).) For pipette solution,
etc., see chapter 8.
a
b
Using the technique of double whole-cell patch reveals data as shown
in gure 15. The holding potential of both cells is 40 mV. One cell is then
clamped to 50 mV while the other cell is kept at the holding potential
thereby applying a transcellular voltage of 10 mV. The other tracks show
the currents necessary to maintain these voltages. Transcellular voltages
53 4 Function and Physiology of Gap Junction Channels
ranging from 50 to +50 mV were applied. From these experiments the I/V
relationship in gure 15b could be constructed. A critical problem with such
measurements is the compensation of the series resistance since in the case of
the junctional resistance being in the order of the series resistance of one of
the patch pipettes, distortion of the measurement of the intercellular current
will result.
When considering the question whether gap junctional channels are regu-
lated by transcellular voltage or not, the reader may be confused by the various
ndings of dierent groups in various preparations. However, as a general
rule considering cardiac gap junction channels coupled by Cx43 channels, in
cell pairs of adult cardiac cells, which are coupled by large numbers of gap
junction channels, the instantaneous and the steady-state current-voltage rela-
tions have been demonstrated to be linear, i.e. g
j
is independent of transjunc-
tional voltage and rectication is not observed [Noma and Tsuboi, 1987;
Reverdin and Weingart, 1988; Weingart, 1986; White et al., 1985]. In contrast,
in embryonic cells (or with some limitations in neonatal cells) which communi-
cate by only a few gap junction channels, linear current-voltage relationships
for instantaneous and steady-state g
j
are observed in a discrete transcellular
voltage range: 50 mV (neonatal rat cardiomyocytes) [Rook et al., 1988];
30 mV (embryonic chick heart cells) [Veenstra, 1990], and 60 mV (neo-
natal hamster cardiomyocytes) [Veenstra, 1990]. Outside this range rectication
is observed and the slope of g
ss
declines progressively to values near zero [Rook
et al., 1988; Veenstra, 1990, 1991a], whereas the instantaneous g
j
remains
constant. This voltage-dependent behavior of g
ss
can be tted with the two-state
Boltzmann equation described above. What is the basis of this voltage-sensitive
gating? According to Rook et al. [1988] it is not the result of a change in the
single-channel conductance
j
but in the ratio t
open
/t
closed
, which is decreased
so that at a given time point more channels are in the closed state.
Spray et al. [1985] found g
j
to be unaected by the transjunctional potential
gradient and by the membrane potential in dispersed and reaggregated rat
ventricle cells. Similarly, Kameyama [1983] reported that R
j
is independent of
the transjunctional potential gradient. While the instantaneous g
j
is insensitive
to the transjunctional voltage, in most cases g
ss
can exhibit voltage dependence
in some preparations [Veenstra et al., 1993] (g. 16). Voltage dependence in
neonatal cardiac myocytes can especially be observed with large transjunc-
tional voltages in the order of 80 mV or more.
Rectifying behavior was observed in craysh axons with depolarization
at the presynaptic side increasing the junctional conductance [Furshpan and
Potter, 1959; Giaume et al., 1987] and in sh [Auerbach and Bennett, 1969].
It has been hypothesised by Bennett et al. [1991] that this rectifying behavior
may arise from a heterotypic composition of the channel.
54 4 Function and Physiology of Gap Junction Channels
Fig. 16. Voltage-dependent gating in pairs of rat Cx43 transfected RIN cells (a) and
pairs of mouse Cx40 transfected HeLa cells (b) [Banach and Weingart, 1996; Bukauskas
et al., 1995].
a
b
Do all connexins exhibit the same sensitivity to the transjunctional volt-
age? Trying to answer this question, Nicholson et al. [1993] have shown in an
xenopus oocyte expression system that g
ss
of gap junction channels constituted
of Cx37 are more voltage-sensitive than those made from Cx40. In comparison
to these, Cx32 was more insensitive and Cx26 channels exhibited the minimum
sensitivity. In addition, Veenstra et al. [1993] showed that the voltage sensitivity
of g
ss
of embryonic chick heart cells decreased with the age of embryos. In
Cx40-transfected neuroblastoma cells (N2A) the Boltzmann half-inactivation
voltage was determined with 54 and +47 mV at negative or positive V
j
,
respectively, indicating a sensitivity of g
ss
of the Cx40 channel to transjunctional
55 4 Function and Physiology of Gap Junction Channels
Table 3. Boltzmann equation parameters for various connexins
Connexin G
min
V
0
[mV] Slope factor References
37 0.27 28 0.08 Reed et al. [1993]
40 0.33/0.28 54/+47 0.13/0.11 Beblo et al. [1995]
0.32 35 0.225 Ebihara [1995]
43 0.37
a
60 0.106 Moreno et al. [1995]
45 0.072 13.4 0.115 Moreno et al. [1995]
0.17/0.16 16/+22 0.126/0.115 Ebihara [1995]
If two values are given, the data on the left are mean values at negative V
j
and on the
right mean values at positive V
j
. It becomes evident that Cx43 exhibits the lowest sensitivity
to V
j
since the half inactivation voltage V
0
is highest, whereas according to these data Cx45
and Cx37 exhibit the strongest V
j
dependence.
a
G
min
/G
max
.
voltage [Beblo et al., 1995]. The Cx43 channel has been shown to possess a
voltage-sensitive component as well with Boltzmann half-inactivation voltages
at 69 and +61 mV [Wang et al., 1992] (table 3).
In a recent study on mouse Cx40 transfected HeLa cells, Bukauskas et al.
[1995] determined V
0
45/49 mV and g
min
0.24/0.26. Valiunas et al. [1997]
investigated the dependence on transjunctional voltage in neonatal rat
cardiomyocytes, which are normally coupled via Cx43, and found, depend-
ing on the pipette solution used, V
0
51/51 mV, g
min
0.28/0.25 and z 3.1/2.9
(KCl solution), or V
0
59/59 mV, g
min
0.15/0.16 and z 2.0/2.3 (TEA-aspartate
solution). In another detailed study Banach and Weingart [1996] observed
asymmetrical-gating properties in rat Cx43-transfected RIN cells with V
0
73.7/65.1 mV and g
min
0.34/0.29, if an asymmetric protocol was used (i.e. one
cell is kept at the holding potential and the other is stepped to dierent voltages
thereby applying a transjunctional voltage), whereas if a symmetrical protocol
was used (both cells are clamped to the holding potential and then a certain
voltage clamp step is applied but of opposite polarity in both cells), the authors
observed symmetrical gating with V
0
60.5/59.5 mV, g
min
0.27/0.29.
The problems arising from the various ndings regarding the voltage
sensitivity of g
ss
initiated a very elegant study by Jongsma et al. [1993] answering
the question Are cardiac gap junction channels voltage sensitive?. In a com-
puter simulation they modelled two cardiomyocytes interconnected via a gap
junction and varied the number of gap junction channels within this intercon-
56 4 Function and Physiology of Gap Junction Channels
nection. The open probability of a single channel was assumed to follow the
equation:
p
0
>1/{1+exp([A
+A
|
][oV
j
V
0
])},
with > exp(A
[oV
j
V
0
]) and |> exp(A
|
[oV
j
V
0
]) and A
(>0.041 mV
1
) and A
|
(>0.021 mV
1
) being the voltage sensitivities of and |,
(>2.5 s
1
), the rate at which equals | and V
0
(>74 mV), the transjunctional
voltageat whichequals |(thedatawere obtainedfromsingle-channel measure-
ments carried out on neonatal rat heart cells by Rook et al. [1988]). They found
arelationshipbetweenthe pipette series resistance andthe appearance of voltage
insensitivity of g
ss
. It was shown that gap junctions with intermediate numbers
of channels (130or morechannels wereassumed) appear tobevoltageinsensitive
if pipettes with 60 MD are used and gap junctions with 300 or more channels if
20 MDpipettes are used. In real measurements the circuit is even more compli-
catedbythe fact that, part of the junctional current is shuntedtogroundthrough
the membrane resistance. The higher the series resistance the more dicult it is
to detect sensitivity. In addition, they demonstrated that, in cells well coupled
by large numbers of gap junction channels (as in real experiments often in adult
cells), voltage sensitivity cannot be detected or is only very moderately present,
whereas in cells weakly coupled by only a few channels sensitivity to transjunc-
tional voltage can be observed. Thus, Jongsma et al. [1993] concluded that the
cardiac gap junctions are moderately voltage sensitive and that the decrease
in voltage sensitivity in the course of embryonic development as described by
Veenstra [1991b] does not reect a change in the regulation of junctional resis-
tance but rather an increase in mean gap junction size. Finally, they state that,
in real experiments on well-coupled heart cell pairs, this voltage sensitivity of g
ss
cannot be observed mainly because of the presence of gap junction channel
access resistance and pipette series resistance.
Much has been speculated about the possible role of a sensitivity for
transjunctional voltage in the heart. It can be assumed that such a behavior
would protect a cell from the hyperpolarizing or depolarizing inuence of
other cells provided the transjunctional voltage is high enough, i.e. exceeding
for example 50 mV in rat heart cells (see above). This would be necessary for
sinusnodal cells which are in close vicinity to the atrium which is relatively
hyperpolarized with regard to the sinus node. Similarly, AV node cells adjacent
to non-nodal tissue may be subject to such an inuence. In these cells potential
gradients high enough to disturb for example the pacemaker function may
arise and, thus, such a disturbing inuence may be prevented by transjunctional
voltage-sensitive gating of the gap junctions. This ts with the general nding
that Cx40 and Cx45 are more sensitive to transjunctional voltage than Cx43.
However, there is at present no clear evidence for these hypotheses.
57 4 Function and Physiology of Gap Junction Channels
What is the real range of gap junction resistance measured in various
systems? Kameyama [1983] reported in reaggregated cell pairs of adult guinea-
pig hearts R
j
in the order of 1.42.1 MD (corresponding to 476714 nS) and
Noma and Tsuboi [1987] 0.2511 MD(903.900 nS) in adult guinea-pig cardio-
myocytes with a peak in the distribution around 1,000 nS (i.e. 1 MD). Weingart
[1986] calculated the gap junction conductance in adult cardiomyocytes with
204 nS (4.9 MD). In neonatal or embryonic heart cell pairs lower conductances
were reported: 030 nS (?33 MD) in embryonic chick heart cells from 7-day-
old embryos [Veenstra, 1990] and 0.0535 nS (?28 MD) in neonatal rat heart
cells [Burt and Spray, 1988a].
In cable preparations of various species resistance has been reported in
the order of 200600 Dcm: 523 Dcm in bullfrog trabeculae [Haas et al., 1983];
200250 Dcm in guinea-pig trabecular muscle [Daut, 1982]; 350530 Dcm in
rabbit Purkinje bers [Colatsky and Tsien, 1979] and 588 Dcm in frog ventricu-
lar trabeculae [Chapman and Fry, 1978].
An interesting phenomenon is that some of the voltage-insensitive gap
junctional channels are gated, i.e. the channels open and close rather than
remain open all the time. The mechanisms underlying this gating behavior of
voltage-insensitive channels found in avian and mammalian hearts and in
septate axons of earthworms are still unknown [Brink, 1991].
Since as pointed out above the overall gap junction conductance does
not only depend on the number and the open probability of the channels but
also on the single-channel conductance, single-channel conductance of various
connexins, including Cx37, Cx40, Cx43, Cx45, will be discussed.
From the channel geometry with a pore sink diameter of 1.5 nm, a pore
mouth diameter of 2.3 nm and a pore length of 15 nm (values according to
Makowski [1985] and Zampighi [1987]), Ru disu li and Weingart [1989] pre-
dicted the single-channel conductance. According to their considerations and
to Hille [1992], R
Channel
is given by the equation:
R
Channel
>R
Pore
+R
Access
with
R
Pore
>(1t(d/2)
2
)
and
R
Access
>2/td.
Assuming the inner pore to be lled with a 130-mmol/l salt solution equals
100 Dcm and R
Channel
equals 1020 GD. The single-channel conductance was
predicted with 50 to 100 pS. However, it has been suggested and shown by
various investigators [Loewenstein et al., 1978; Neyton and Trautmann, 1985;
58 4 Function and Physiology of Gap Junction Channels
Rook et al., 1988] that the regulation of gap junction conductance is not the
all or nothing and not quantal, but graduate so that dierent single-channel
conductances or substates have been postulated [Page, 1991]. This is probably
reected by various substates of single-channel conductance found in various
gap junctional channels which will be discussed as well.
The channel found most abundantly in cardiac tissue is the Cx43 channel.
In guinea-pig hearts the single-channel conductance of Cx43 channels has
been measured with 37 pS and it was characterized as insensitive to the non-
junctional membrane potential [Ru disu li and Weingart, 1989]. In contrast in
the neonatal rat heart Burt and Spray [1988a] found a single-channel conduc-
tance of about 60 pS. The voltage-insensitive component of g
j
of Cx43 channels
has been ascribed to a voltage-insensitive substate by Moreno et al. [1994a]
who observed a graded response in Cx43 transfected hepatoma cells. In another
study [Moreno et al., 1994b] these authors dened two substates of this channel,
i.e. 6070 pS and a higher conductance state with 90100 pS.
Kwak et al. [1995b] also characterized two substates in neonatal rat cardio-
myocytes with 20 and 4045 pS. However, when taking the regulation by
protein kinases into account Takens-Kwak and Jongsma [1992] discriminated
even three substates of single-channel conductance in neonatal rat heart cells:
21, 4045 and 70 pS. Similarly, in Cx43-transfected SKHep1 cells they found
three dierent substates: 30.59.1, 61.29.8 and 89.112 pS [Kwak et al.,
1995a]. From these dierent ndings one might conclude that at least three
dierent substates of single-channel conductance of Cx43 channels can be
distinguished. Depending on phosphorylation or dephosphorylation by vari-
ous protein kinases the dierent substates seem to be favored (see chapter
4.1). As investigated by Valiunas et al. [1997] gap junction channels of neonatal
rat heart cells formed by Cx43 possess several conductance states: a main
state, several substates, and a residual state as well as a closed state. Depending
on the pipette solution
j
(main state) was determined 96 pS (KCl), 61 pS
(Cs-aspartate) and 19 pS (TEA-aspartate) and
j
(residual state) was deter-
mined 23, 12 and 3 pS, respectively, revealing
j
(main state)-
j
(residual state),
ratios of 4.2, 5.1 and 6.3, indicating that the residual state restricts ion move-
ment more eciently than the main state. Transitions between the several
open states (main, substates and residual state) were fast (=2 ms) in contrast
to the transitions between open states and closed state (1565 ms). The dier-
ences in the results of various investigators may perhaps be due to dierent
experimental models and conditions. An example of single-channel recordings
is given in gure 17.
The next channel to be dealt with is the Cx40 channel. The unique
conductance and gating of gap junction channels formed by Cx40 has been
investigated in Cx40-transfected mouse neuroblastoma cells (N2A cells). In
59 4 Function and Physiology of Gap Junction Channels
Fig. 17. Single-channel conductance of neonatal rat heart cells. Note the residual con-
ductance [Valiunas et al., 1997]. The V
j
applied was 50 mV.
the presence of potassium glutamate (120 mmol/l) Beblo et al. [1995] measured
the slope conductance of single Cx40 channels in the order of 158 pS. The
macroscopic steady-state current exhibited dependence on transjunctional
voltage with a Boltzmann half-inactivation voltage of 50 mV. The authors
found a residual voltage-insensitive normalized junctional conductance in the
order of 35% of the maximum and a gating charge valence of 3. In these Cx40
channels substates have also been described [Beblo et al., 1995] equal to 21
and 48% of the main open-state conductance (which was in the range of
158 pS), although these substates were reported to occur only occasionally.
Bukauskas et al. [1995] investigated mouse Cx40-transfected HeLa cells and
measured single-channel conductance. They observed three open states: main
state (198 pS); several substates, and an residual state (36 pS) besides a closed
state. Transition between the open states were fast (12 ms) in contrast to the
transitions between open and closed states (1545 ms).
The third channel found in heart muscle is the Cx45 channel. This was
reported to exhibit an overall conductance of 3.10.4 nS in cell pairs [Kwak
et al., 1995a] with a mean single-channel conductance of 36.56.5 pS or
1.3 nS in human Cx45-transfected SKHep1 cells [Moreno et al., 1995] and
j
of 328 pS. At higher transjunctional voltages an additional conductance
state with 22.54.3 pS was observed. This channel strongly depends on trans-
junctional voltage [Moreno et al., 1995] comparable to Cx38 channels. How-
ever, single-channel conductance is not a function of transcellular voltage.
Besides Cx40, Cx43 and Cx45, Cx37 channels are expressed in the cardio-
vascular tissue. These channels are frequently found in endothelium. These
channels were expressed in human Cx37-transfected neuroblastoma cells (N2A
cells) and exhibited a pronounced voltage dependence and multiple conduct-
ance states [Reed et al., 1993]. Several single-channel conductances were found:
60 4 Function and Physiology of Gap Junction Channels
21922, 1656, 12313 and 531 pS at V
j
>3040 mV. At higher V
j
the
large conductance was no longer observed and
j
of 944, 695 and
3910 pS were detected at V
j
>8090 mV.
Finally, Cx26 channels were also investigated in an expression system
(SKHep1 cells) by Kwak et al. [1995a]. They observed a mean single-channel
conductance of 140 to 150 pS with substates of 70 and 110 pS.
Gating of the channel can be modulated either by an alteration in the
single-channel conductance or by a change in the open probability. The macro-
scopic channel conductance is then either inuenced by the single-channel
conductance, by the mean open time of a channel or by the frequency of
channel openings, this means by the number of channels in the open state at
the same time or at a given time interval.
Do the gap junctional channels exhibit some sort of selectivity? Do they
conduct cations and anions? Do they exhibit characteristics known from so
many ion-selective transmembrane channels or are they distinct from these?
As already said, the rst important dierence to other ionic channels is that
gap junction channels are permeable to small molecules up to a molecular
weight of about 1,000 Daltons. Another dierence is that they conduct both
anions and cations. In order to further elucidate these questions Brink [1991]
investigated the selectivity of gap junctional channels in septate earthworm
axons using the double whole-cell patch-clamp technique. According to this
study this channel exhibits a single-channel conductance of about 100 pS
and is sensitive to calcium and pH. There was no inhibition of the channel
conductance with various blockers commonly used in electrophysiology: tetra-
ethylammonium (TEA), 4-aminopyridine (4-AP), Zn
2+
, Co
2+
and Ni
2+
. There
was no selectivity for K
+
over Cs
+
, but the 100-pS channel seemed to be
somewhat selective for cations over anions with a chloride conductivity to
potassium conductivity ratio of 0.53. This channel is also voltage-insensitive.
Molecules not larger than 1 kD or not exceeding an ionic radius of 0.8 or
1.0 nm can be transported through the channel [Brink, 1991; Spray et al.,
1991].
The rat Cx40 channel exhibits a detectable chloride permeability of 0.29
relative to potassium [Beblo et al., 1995]. This indicates some selectivity for
cations over anions as well. These channels were also permeable to 2,7-
dichlorouorescein and to the more polar 6-carboxyuorescein dye. Interes-
tingly, the 2,7-dichlorouorescein permeability did not increase with increas-
ing junctional conductance in that study.
With regard to small molecules other than ions Tsien and Weingart [1976]
reported that
3
H-cAMP diuses across gap junctions in calf and cow ventricle.
Furthermore, Weingart [1974] has shown that
14
C-TEA (molecular weight 130)
diuses transjunctionally in sheep ventricular muscle and he found the channel
61 4 Function and Physiology of Gap Junction Channels
diameter to be in the order of 10 A