Gap junctions are critical to maintaining synchronized impulse propagation and repolarization. Heterogeneous expression of the principal ventricular gap junction protein connexin43 (Cx43) is associated with action potential duration (APD) dispersion across the anterior ventricular wall. Little is known about Cx43 expression patterns and their disparate impact on regional electrophysiology throughout the heart. We aimed to determine whether the anterior and posterior regions of the heart are electrophysiologically distinct. Multisegment, high-resolution optical mapping was performed in canine wedge preparations harvested separately from the anterior left ventricle (aLV; n = 8) and posterior left ventricle (pLV; n = 8). Transmural APD dispersion was significantly greater on the aLV than the pLV (45 ± 13 vs. 26 ± 8.0 ms; P < 0.05). Conduction velocity dispersion was also significantly higher (P < 0.05) across the aLV (39 ± 7%) than the pLV (16 ± 3%). Carbenoxolone perfusion significantly enhanced APD and conduction velocity dispersion on the aLV (by 1.53-fold and 1.36-fold, respectively), but not the pLV (by 1.27-fold and 1.2-fold, respectively), and produced a 4.2-fold increase in susceptibility to inducible arrhythmias in the aLV. Confocal immunofluorescence microscopy revealed significantly (P < 0.05) greater transmural dispersion of Cx43 expression on the aLV (44 ± 10%) compared with the pLV wall (8.3 ± 0.7%), suggesting that regional expression of Cx43 expression patterns may account for regional electrophysiological differences. Computer simulations affirmed that localized uncoupling at the epicardial-midmyocardial interface is sufficient to produce APD gradients observed on the aLV. These data demonstrate that the aLV and pLV differ importantly with respect to their electrophysiological properties and Cx43 expression patterns. Furthermore, local underexpression of Cx43 is closely associated with transmural electrophysiological heterogeneity on the aLV. Therefore, regional and transmural heterogeneous Cx43 expression patterns may be an important mechanism underlying arrhythmia susceptibility, particularly in disease states where gap junction expression is altered.
- action potentials
- optical mapping
there is emerging awareness of the importance of transmural electrophysiological heterogeneities in disease states such as long QT syndrome (37), Brugada syndrome (5), and heart failure (30). Differences in ion channel composition spanning the anterior left ventricle (aLV) distinguish the electrophysiological phenotype of epicardial (Epi), midmyocardial (Mid), and endocardial (Endo) myocytes (25). Previously, we have shown that the presence of steep action potential duration (APD) gradients at the Epi-Mid interface can be amplified by various disease states to sufficient magnitude to form a substrate for conduction block and ventricular arrhythmias (1).
The development and maintenance of transmural electrophysiological gradients depends on not only ion channel heterogeneities intrinsic to cellular layers, which span the transmural wall, but also the degree of intercellular coupling between them (20, 43). Mediated by gap junctions, intercellular coupling serves to attenuate APD gradients to allow for rapid and synchronous communication between cell layers. Comprised principally of connexin 43 (Cx43) in the ventricle (40), gap junction expression patterns have been implicated in the maintenance of electrophysiological heterogeneity in the intact heart. For instance, reduced intercellular coupling during ischemia (8, 14, 35) or heart failure (16, 28) is associated with enhanced electrophysiological heterogeneity (30). Importantly, reduced Cx43 expression at the Epi-Mid interface is associated with transmural electrophysiological heterogeneity (29). Moreover, we have shown that restoring Cx43 in heart failure restores conduction velocity (10), reaffirming a role of gap junctions in the mechanism of conduction slowing in heart failure.
To date, the investigation of transmural electrophysiological heterogeneity has focused on the transmural wall of the aLV. However, mounting evidence suggests that the anterior and posterior left ventricle (pLV) walls are distinct. For example, differences in wall thickening (7), and stress-strain relationship (11, 42), between the aLV and pLV are known to exist. Since gap junction expression is importantly influenced by stretch (34, 45, 47), gap junction distribution may vary between different regions of the heart. Interestingly, there is evidence that infarct location influences clinical phenotype and outcomes in patients. For instance, anterior wall myocardial infarction (MI) is associated with greater likelihood of ventricular fibrillation and higher mortality than posterior wall MI (13). Although the mechanisms for these differences are unknown, these data suggest that inherent electrophysiological differences between the aLV and pLV may account for differences in electrical stability between different regions of the myocardium. In the present study, we hypothesized that conduction velocity, repolarization heterogeneities, and arrhythmia susceptibility in different left ventricle (LV) regions can be attributable to regional differences in gap junction expression patterns.
Multisegment transmural optical mapping.
To assess the functional consequence of Cx43 expression patterns across the transmural wall in different regions of the LV, we optically mapped action potentials from cells spanning the entire transmural wall of the arterially perfused canine wedge preparation as described previously (4, 46). Briefly, hearts were excised from eight male mongrel dogs (body wt, 20–25 kg). The procedures used were approved by the Institutional Animal Care and Use Committee and were in accordance with their guidelines. Wedges of myocardium were dissected separately from either the anterior or posterior walls of the LV in proximity to secondary branches of the left anterior descending and circumflex arteries, respectively. Wedge preparations were first allowed to equilibrate for 20 min while being perfused with normal Tyrode's solution and were then perfused with 100 μmol/l carbenoxolone for 10 min.
The optical mapping system used to measure transmural action potentials has been described previously (2, 4, 18, 32). Briefly, electrophysiological heterogeneity across the transmural wall was assessed by recording 256 optical action potentials simultaneously from across all myocardial layers of the transmural LV wall during steady-state Endo pacing [basic cycle length (BCL), 2,000 ms at 2× diastolic threshold]. Optical action potentials were recorded with spatial and temporal resolutions of 0.89 mm and 0.5 ms, respectively, corresponding to a total mapping area of 14 × 14 mm. APD was measured as the difference between depolarization and repolarization times as previously described (12, 18). APD was measured from 16 sites per muscle layer (each site was equidistant from the epicardial surface) for each of 16 equally spaced transmural muscle layers. The average APD within each layer was calculated to generate a transmural APD profile from epicardium to endocardium. APD dispersion was calculated as the maximal difference in APD between all layers, which, in every case, was given by the maximum APD difference between the Epi and Mid layers. Conduction velocity was measured selectively (±15°) in the transmural direction of propagation (θTM) as a function of distance from the epicardium of each of the 16 transmural muscle layers as described (30). Conduction velocity dispersion was calculated as the difference between the Epi and Mid layers as a percentage of conduction velocities in the midmyocardium. Functional measurements were then compared with Cx43 expression within each muscle layer of wedges from the anterior or posterior LV wall. (See expanded materials and methods section in suppplemental material for further details; all supplemental material can be found with the online version of this article.)
Programmed electrical stimulation was performed on all wedge preparations harvested separately from the anterior and posterior walls of the LV before and after administration of carbenoxolone. After an Endo 20-beat drive-train at a BCL of 2,000 ms, an Epi premature stimulus (S2) was delivered at an S1S2 coupling interval of 500 ms. The S1S2 interval was sequentially decremented by 10 ms until refractoriness was reached or an arrhythmia was induced. An arrhythmia was defined as three or more unpaced beats.
Transmural Cx43 quantification by immunofluorescence.
Immunofluorescence analysis of Cx43 was performed using a polyclonal Cx43 total antibody and a monoclonal Cx43-nonphosphorylated antibody (both Zymed) (29). Images were collected from a laser scanning confocal microscope and analyzed as previously described (6, 29). (See expanded materials and methods section in suppplemental material for further details.)
Theoretical multicellular fiber model.
To determine whether regional alterations in intercellular coupling could explain electrophysiological differences observed between the aLV and pLV, we constructed a theoretical multicellular fiber composing 180 ventricular cells each of Luo Rudy (LRd) formulation (26, 43) interconnected through resistive gap junctions. The one-dimensional theoretical multicellular fiber contained an Endo region (cells 1–50), a Mid cell region (cells 51–140), and an Epi region (cells 141–190), as previously described (43). To determine whether localized diminishment of coupling could enhance APD dispersion, gap junction coupling was systematically varied from 2.5 to 0.125 μS across the theoretical fiber until 100% of the fiber was homogeneously uncoupled. To assess whether the observed gap junction protein expression could produce APD gradients, experimentally measured conduction velocities across the aLV and pLV at baseline and carbenoxolone infusion were used to determine conductance values from the theoretical relationship of discontinuous conduction (36). As such, gap junction conductance across the fiber was varied between 0.5 and 0.05 μS to simulate intercellular coupling at baseline and carbenoxolone conditions. APD was calculated using APD at 90% repolarization in each cell, and APD dispersion was defined as the difference between the longest and shortest APD. Finally, the derivative of the APD profile across the fiber was calculated by averaging the slopes of two adjacent data points.
Myocyte isolation and patch clamp recording.
Since important differences in ion channel current composition between cells constitute electrophysiological heterogeneities across the aLV (25), it was important to confirm that similar ionic current differences exist across the pLV. To do so, Epi, Mid, and Endo myocytes were isolated from LV posterior wedge of six additional canines using a standard enzymatic dispersion technique (23). Single-cell action potentials as well as slowly activating component of the delayed rectifier (IKs), transient outward potassium (Ito), and late calcium (ICa,L) currents were recorded. Currents are reported at peak current potentials. We validated that 100 μM of the selective intercellular uncoupler carbenoxolone (9, 17) had no effect on APD morphology or duration of isolated cells. For detailed solutions, see expanded materials and methods section in suppplemental material.
Statistical analysis of the data was performed using Student's t-test for paired data or a single-factor ANOVA. Differences in susceptibility to ventricular tachycardia between control and carbenoxolone wedges were compared using the Fisher Exact test. Summary data are presented as means ± SE. Differences were considered significant at P < 0.05.
Electrophysiological properties of the aLV versus pLV.
Representative transmural APD profiles measured across the aLV (Fig. 1A) and pLV (Fig. 1C) are plotted in Fig. 1 with their corresponding APD gradient profiles (Fig. 1, B and D). At baseline conditions (squares), there was a substantial difference in APDs across the transmural wall of the aLV, with the sharpest transition in transmural APD occurring between the Epi and deeper layers of tissue (Fig. 1A). Correspondingly, the maximal APD gradient was consistently localized between the Epi and deeper layers of tissue (Fig. 1B). In contrast, the APD profile across the transmural wall of the pLV was relatively flat (Fig. 1C), resulting in a negligible APD gradient (Fig. 1D). When all of the experiments are considered (Fig. 1E), transmural APD dispersion was significantly higher in the aLV compared with the pLV. Moreover, the maximal APD gradient tended to be larger across the aLV than the pLV at baseline although this did not achieve statistical significance (Fig. 1F). Furthermore, the largest APD gradient occurred at a depth ∼30% from the Epi surface (Fig. 1B). Importantly, this location has been shown to be precisely the region of reduced Cx43 expression (29).
To determine whether intercellular coupling between muscle layers could explain the formation and maintenance of transmural APD gradients, wedges were perfused with 100 μM of the selective intercellular uncoupler carbenoxolone (9, 17). A subset of experiments confirmed that 100 μM of carbenoxolone did not alter morphology or duration [P = not significant (NS)] of the action potential (305 ± 30 ms at baseline vs. 314 ± 32 ms postcarbenoxolone infusion) recorded from isolated myocytes of normal hearts. In wedges isolated from the aLV, the administration of carbenoxolone shortened Epi APD while prolonging Mid and Endo APD (Fig. 1A, circles), resulting in marked amplification of the APD gradient (Fig. 1B). In contrast, in the pLV (Fig. 1, C and D), carbenoxolone had no effect on the APD profile (Fig. 1C) or APD gradient (Fig. 1D). In all experiments, carbenoxolone significantly increased transmural APD dispersion in the aLV (Fig. 1E), and APD gradient (Fig. 1F), but failed to alter APD gradients in the pLV. These data demonstrate significant differences in susceptibility to uncoupling of different regions of the heart, which implies that these regions may have distinct gap junction protein expression patterns.
In addition to repolarization properties, conduction velocity was measured in each transmural muscle layer of the aLV and compared with the pLV. The representative transmural conduction velocity profile shown in Fig. 2 reveals that under baseline conditions (Fig. 2A, squares), there is significant and selective conduction velocity slowing in the Epi of the aLV compared with deeper muscle layers. However, the transmural conduction velocity profile across the pLV was relatively flat (Fig. 2B). Summary data revealed that conduction velocity dispersion, which serves as an index of heterogeneity of conduction, was more than 2.3-fold higher across the aLV compared with the pLV (Fig. 2C) at baseline. As expected, pharmacological uncoupling with carbenoxolone significantly reduced conduction velocity throughout the transmural walls of both the aLV and pLV. However, carbenoxolone significantly enhanced conduction velocity dispersion in the aLV, but not the pLV (Fig. 2C). Importantly, following carbenoxolone administration, conduction velocity dispersion remained more than 2.5-fold higher across the aLV than the pLV (Fig. 2C).
Susceptibility to inducible ventricular tachycardia was compared between the aLV and pLV. Under baseline conditions, no arrhythmias were induced in wedges harvested from either the anterior or posterior wall (Fig. 3B). In contrast, under conditions of reduced coupling with carbenoxolone, the aLV exhibited a 4.2-fold increase in susceptibility to inducible ventricular tachycardia compared with the pLV. These arrhythmias typically self-terminated, lasted for ∼5 s, and were generally polymorphic in appearance (Fig. 3A).
Cx43 expression patterns of aLV versus pLV.
Expression patterns of Cx43 across the transmural wall of the aLV and pLV are compared in representative immunofluorescent images shown in Fig. 4A. In each case, Cx43 localized to the longitudinal ends of individual myocytes as reported previously (22). A representative transmural profile of Cx43 quantity in one animal shown in Fig. 4B demonstrates that across the aLV, Cx43 signal was selectively reduced in the Epi compared with deeper muscle layers. In contrast with the aLV, in the pLV, there were no significant differences in Cx43 expression between the Epi and either Mid or Endo layers. Averaged data from all experiments revealed that localized reduction of Epi Cx43 in the aLV produced transmural dispersion of Cx43 that was significantly greater across the aLV compared with the pLV (Fig. 4B, left inset). In other words, the average Epi expression of Cx43 across the aLV in all animals was significantly smaller than deeper muscle layers [0.87 ± 0.34 vs. 1.1 ± 0.3 arbitrary units (AU), respectively; P < 0.05], whereas across the pLV, Epi expression of Cx43 did not significantly differ from deeper muscle layers (0.97 ± 0.3 vs. 0.97 ± 0.3 AU, respectively; P = NS). Interestingly, there were no significant differences in the total amount of Cx43 expression between the two regions of the ventricle (Fig. 4B, right inset). Since phosphorylation of Cx43 has been linked to proper function of gap junctions, we investigated whether the aLV or pLV exhibited differences in the relative proportions of phosphorylated to nonphosphorylated Cx43 signal. Across the aLV, the nonphosphorylated percentage of total Cx43 was similar across Epi, Mid, and Endo (1.5 ± 1.2% vs. 3.0 ± 1.0% vs. 1.1 ± 1.6%; P = NS). Across the pLV, the nonphosphorylated percentage of total Cx43 was also similar (1.7 ± 1.2% in Epi vs. 1.4 ± 0.4% in Mid vs. 2.0 ± 0.9% in Endo), suggesting that the electrophysiological differences observed between aLV and pLV were not influenced by nonphosphorylated Cx43.
Effect of coupling on APD heterogeneity.
The aforementioned data suggested that Cx43 expression patterns differ between the aLV and pLV. Moreover, unlike the pLV, the aLV was characterized by diminished Cx43 in regions that exhibited conduction velocity slowing and maximal APD gradients. Taken together, these data suggest that Cx43 expression patterns may be responsible for electrophysiological differences between the aLV and pLV. To further explore this hypothesis, computer simulations were performed using a one-dimensional single-cell chain containing Endo, Mid, and Epi cells each of LRd single-cell formulation and interconnected by gap junctions. As shown in Fig. 5, when intercellular coupling was reduced homogenously across the entire fiber from 2.5 μS (solid blue line) to 0.125 μS (solid red line), the APD profile exhibited a markedly increased transmural APD gradient, located at the Epi-Mid interface.
Since our experimental data showed that Cx43 expression patterns of aLV and pLV differed as a result of reduced Cx43 expression localized to the subepicardial region, computer simulations were repeated to determine whether uncoupling localized to subepicardium was sufficient to explain transmural dispersion of APD measured in aLV. The degree of coupling was systematically stepped from 2.5 to 0.125 μS at different depths from the Epi surface (Fig. 5, dashed lines at bottom). Uncoupling within the Epi (dashed blue line) failed to amplify transmural APD dispersion. However, uncoupling to a depth that spanned the Epi-Mid interface (Fig. 5, black dashed line) did produce an APD profile that was essentially identical to the one caused by homogeneous uncoupling of the entire fiber (solid red line). The simulation results are summarized in the Fig. 5 inset showing that transmural APD dispersion is amplified so long as the depth of uncoupling spanned the Epi-Mid interface, and uncoupling to deeper layers had little incremental effect on APD dispersion.
To further assess whether the observed gap junction protein expression patterns (Fig. 4) could produce the measured electrophysiological changes (Figs. 1 and 2), gap junction coupling was varied to mimic the transmural Cx43 profiles of the aLV and pLV at baseline and carbenoxolone conditions. Briefly, experimentally measured conduction velocities across the aLV and pLV at baseline and carbenoxolone infusion were used to determine conductance values from the theoretical relationship of discontinuous conduction (36). The effect of transmural gap junction expression patterns on the APD gradient is illustrated in Fig. 6. Under conditions simulating gap junction coupling across the aLV, the maximal APD gradient, which occurred at the Epi-Mid interface (data not shown) was larger on the aLV compared with the pLV. As expected, the APD gradient was further augmented during conditions simulating functional uncoupling by carbenoxolone. Interestingly, the difference in APD gradient between aLV and pLV was further enhanced when coupling was diminished to simulate carbenoxolone infusion. Taken together, these data indicate that local uncoupling across the Epi-Mid interface is an important factor in the maintenance of APD gradients. Moreover, these data may help explain the occurrence of unidirectional block leading to reentry at the Epi-Mid interface in models of long QT (4) and heart failure (1).
Ionic properties are not associated with electrophysiological differences in aLV versus pLV.
Since transmural APD differences arise from differences in transmural ionic channel compositions across the aLV, it was important to confirm that such heterogeneities were also present in the pLV. First, we verified the presence of gradient of Ito between the Epi and Endo cells in the pLV, which has been previously reported across the aLV (21, 25). In the pLV, Ito current density was significantly (P < 0.05) larger in Epi cells (11.8 ± 1.1 pA/pF) compared with deeper muscle layers (4.7 ± 1.6 pA/pF) (see supplemental Fig. 1 for the current voltage graph). Importantly, current density of IKs was significantly (P < 0.05) smaller (0.18 ± 0.02 pA/pF) in cells isolated from the midmyocardium than in Epi (0.4 ± 0.05 pA/pF) or Endo (0.44 ± 0.08 pA/pF) myocytes amounting to a near 57% reduction in IKs current density in the Mid compared with the Epi. We measured a similar reduction of IKs current density in the Mid (50%) compared with the Epi in the aLV. Since the duration of the action potential can also be influenced by L-type calcium current, we verified that there were no differences in ICa,L current density between the Epi, Mid, or Endo myocytes in the pLV and confirmed this same lack of difference in current density in the aLV. (A full description of current voltage relationships may be found in supplemental Fig. 2.) These findings strongly suggest that the differences we report in APD gradients between aLV and pLV are attributable to differences in gap junction expression patterns and not spatial gradients of ion channel density.
In this study, we identify significant differences in electrophysiological function between different regions of the LV myocardium. In contrast with the pLV, the aLV exhibited enhanced transmural APD dispersion, heterogeneous conduction, greater susceptibility to functional uncoupling, and enhanced vulnerability to inducible ventricular arrhythmias. Although these results have relevance to mechanisms of electrical instability in the heart, to our knowledge this is the first study to offer a potential explanation for regional variations in electrophysiological properties across the LV. Furthermore, these findings highlight the potential importance of differences in regional electrophysiological function across the LV when considering arrhythmia mechanisms at the level of the whole heart.
The technique of transmural optical mapping from separate myocardial segments permitted detailed and independent electrophysiological phenotyping from aLV and pLV, including the potential contribution of each segment separately to electrical stability of the LV. We found that the aLV possess significantly different electrophysiological properties than the pLV. Specifically, the aLV exhibited substantially greater transmural APD gradients than pLV (Fig. 1). Our findings from the aLV are consistent with previous reports by our laboratory (29) and others (46) of sharp transitions in APD within a few millimeters of the Epi, at the interface between the Epi and Mid layers. Additionally, the aLV and pLV also exhibited different transmural conduction velocity profiles. Unlike the pLV, in which conduction velocity was uniform, conduction velocity slowing was consistently observed in the Epi of the aLV compared with deeper muscle layers consistent with the location of increased tissue resistivity (46) and reduced Cx43 expression (29).
Anterior LV segments were also more susceptible to functional uncoupling. Although across the pLV the APD gradient remained unperturbed, uncoupling by carbenoxolone infusion in the aLV significantly amplified the magnitude of transmural APD gradient to >10 ms/mm, a value previously shown to be critical for the development of conduction block and reentry (1, 19). It is important to note that carbenoxolone had no effect on action potential morphology or duration in isolated cells, reaffirming that our observation of transmural APD gradients amplified by carbenoxolone were most likely attributable to uncoupling. Furthermore, ventricular arrhythmias were more easily induced in the aLV after carbenoxolone perfusion, whereas arrhythmia susceptibility in the pLV was unchanged by carbenoxolone. These data are consistent with earlier observations that large APD gradients between Epi and Mid and conduction slowing (4, 30) can form substrates for conduction block and reentry. Such findings may provide potential insight to clinical outcomes that are dependent on infarct location. Specifically, the association between anterior wall MI and greater likelihood of ventricular fibrillation, and higher mortality rates (13), may be due to variable electrophysiological properties across different regions of the LV.
A major focus of the present work was to ascertain potential mechanisms responsible for differing electrophysiological properties of aLV versus pLV. To that end, we probed for major difference in the functional expression of ion channels spanning the ventricular wall and investigated the degree of intercellular coupling interconnecting cells in maintaining transmural electrophysiological heterogeneity. Previous reports in the aLV have identified important differences in the expression of the Ito current (38), Iks current (24), the late sodium current (INa,L) (48), and sodium-calcium exchange current (INa-Ca) (49) across cells spanning the ventricular wall. Such ionic distinctions result in a longer APD in cells of the Mid wall and an increased sensitivity of these cells to APD prolonging interventions such as IKr blocking drugs and bradycardia (3, 38). Differences in repolarizing currents between right and left ventricles have been reported, which potentially influence clinical expression of Brugada syndrome (5) and susceptibility to drug related proarrhythmia (41). However, to this point, the existence of electrophysiological heterogeneity in different regions of the left ventricle has not been addressed.
Our data indicate that ion channel heterogeneities, which underlie transmural heterogeneity of APD in the aLV, were also present in pLV. Transmural gradient in IKs current density and ICa,L current density were compared between the aLV and pLV. These currents were examined since they are thought to play a major role in controlling APD in the dog. As expected, ICa,L current density did not differ between the Epi, Mid, or Endo cells isolated from either the aLV or pLV. In contrast, IKs current density was significantly lower in the Mid compared with Epi myocytes in both aLV and pLV. Taken together, these findings strongly suggest that ion channel heterogeneities are not likely to account for the electrophysiological differences between the aLV and pLV.
Since functional uncoupling by carbenoxolone in the ALV enhanced APD gradients (by preferentially shortening APD in the Epi while prolonging APD in the Mid) but failed to affect the transmural APD profile of the pLV, we hypothesized that differences in gap junction expression patterns may explain electrophysiological differences between aLV and pLV. To test this hypothesis, we analyzed Cx43 expression patterns across transmural layers in wedges harvested separately from the anterior and posterior free walls. Across the aLV, Epi Cx43 expression was reduced compared with deeper myocardial tissue layers. Across the pLV, however, Cx43 was evenly distributed (Fig. 4). These expression patterns mirrored the conduction velocity profiles across the aLV, which exhibited slowed conduction in the Epi, and pLV, which exhibited uniform conduction (Fig. 2). The relative proportion of nonphosphorylated Cx43 to total Cx43 was similar across all muscle layers of the aLV and pLV, indicating that nonphosphorylated Cx43 was unlikely playing a role in regional electrophysiological differences.
The precise mechanisms for the observed expression patterns of gap junctions remain poorly understood. It is widely accepted that gap junctions are importantly mediated by pulsatile stretch (34, 45). Mid and Endo layers undergo greater segment shortening (33) and strain (7) than Epi fibers, which may explain the relative paucity of Cx43 expression in Epi layers compared with deeper counterparts across the aLV. Based on this, it is conceivable that different regions of the ventricular myocardium experience variable stretch forces during the cardiac cycle. Tagged MRI studies, however, reveal that during steady state conditions, the strain experienced throughout the LV myocardium is homogenous (15). Therefore, the observed heterogeneous Cx43 expression patterns across the aLV and homogeneous expression patterns of Cx43 across the pLV cannot be explained by stretch alone. An alternative explanation for the observed expression patterns may lie in the dependence of Cx43 expression on the pattern of conduction. Patel et al. (27) reported enhanced Cx43 content distal to the site of ventricular pacing, suggesting that gap junction protein expression patterns are dependent on the pattern of activation. If this were true, one might expect to see more, not less, Cx43 content in the Epi of the aLV exposed to normal sinus rhythm, during which the Epi activates last. Hence, activation patterns alone cannot explain the observed relative reduction of Cx43 and overall heterogeneous expression pattern observed across the aLV. Therefore, the mechanism responsible for heterogeneous expression patterns of Cx43 in some regions of the heart and not in others will require further study.
Computer simulations were performed to determine whether the differences in gap junction protein expression patterns between the aLV and pLV could explain the observed transmural APD gradient differences. Previously, it was shown that intercellular coupling attenuates transmural APD heterogeneities and that uniform uncoupling can unmask electrophysiological heterogeneity (20, 43). In this study, we extended this work to demonstrate that localized reduction of coupling can amplify transmural APD heterogeneity provided that the region of reduced coupling spans the Epi-Mid interface (Fig. 5). Moreover, to determine whether the observed degree of Cx43 heterogeneity could produce the magnitude of transmural APD gradients in the aLV but not the pLV, we simulated the transmural coupling profiles of the aLV and pLV to show that local uncoupling across the Epi-Mid interface was indeed an important factor in the maintenance of APD gradients (Fig. 6). These data implicate localized underexpression of Cx43 as an important mechanism of transmural heterogeneity.
There are important limitations to our modeling approach that should be considered before extrapolating the results to whole heart function. First, we made an important simplifying assumption of proportionality between total Cx43 expression and the degree of intercellular coupling without regard for stoichiometric composition, gap junction distribution around the myocyte, or phosphorylation status. Second, in a one-dimensional model, such as the one employed in this study, small changes in gap junction coupling can have a great impact on transmural heterogeneity, but in a three-dimensional structure, higher connectivity between cells might act to minimize APD heterogeneity. Notwithstanding these limitations, the model allowed us to selectively alter intercellular coupling (a task unfeasible in intact preparations) to establish the principle that APD heterogeneity is profoundly dependent on the degree of cell-to-cell coupling and, just as importantly, the transmural profile of coupling proteins. Finally, it is conceivable that variations of fiber orientation between transmural muscle layers (i.e., rotational anisotropy) could influence tissue resistivity in the transmural direction and, therefore, conduction along the transmural axis. However, we have previously investigated this possibility in detail using a three-dimensional computer simulation where transmural conduction was compared with and without rotational anisotropy. We found conduction along the transmural axis was not dependent on fiber rotation orthogonal to that axis. The data suggested that transmural conduction most closely mimicked transverse conduction within a single plane of myocardium since conduction velocity and estimated space constant were similar in the transmural direction and the axis transverse to fibers within the Epi layer. Therefore, it is unlikely that localized subepicardial conduction slowing was attributable to rotational anisotropy (31).
There have been some discrepancies between APD gradients estimated from activation recovery intervals in situ versus those measured in the wedge preparation (44). Therefore, it is important to consider the limitations of using the wedge preparation as an indicator of intact ventricular function. The wedge cannot fully recapitulate in vivo conditions because it lacks autonomic intervention and metabolic and hormonal activity. Perhaps for these reasons alone, the observed APD and Cx43 gradients have not been convincingly reproduced in situ. For example, Soltysinska et al. (39) recently investigated the transmural profile of Cx43 expression in human hearts but failed to show a transmural gradient, presumably due to either the low number of sample points, the environmental and genetic variance of human subjects, or the sensitivity of markers used. Clearly, further investigation is required to address such apparent discrepancies.
In summary, the present study demonstrates that a heterogeneous expression pattern of Cx43 has important implications to normal function of the heart. Taken together, our data suggest that regional differences in electrophysiological function may dictate susceptibility to a variety of arrhythmias. Therefore, regional Cx43 underexpression patterns may be an important mechanism underlying arrhythmia susceptibility, particularly in disease states where gap junction expression is altered.
This study was supported by National Institutes of Health Grants RO1-HL054807 (to D. S. Rosenbaum) and T32-HL-07653 (to M. Strom).
No conflicts of interest are declared by the author(s).
We thank Dr. Isabelle Deschênes for helpful discussions in preparation of this article. This work was presented in part at the 26th annual scientific sessions of the Heart Rhythm Society (New Orleans) 2005.
- Copyright © 2010 the American Physiological Society