Reduced intercellular coupling leads to paradoxical propagation across the Purkinje-ventricular junction and aberrant myocardial activation

Abstract

Ventricular tachycardia is a common heart rhythm disorder and a frequent cause of sudden cardiac death. Aberrant cell–cell coupling through gap junction channels, a process termed gap junction remodeling, is observed in many of the major forms of human heart disease and is associated with increased arrhythmic risk in both humans and in animal models. Genetically engineered mice with cardiac-restricted knockout of Connexin43, the major cardiac gap junctional protein, uniformly develop sudden cardiac death, although a detailed electrophysiological understanding of their profound arrhythmic propensity is unclear. Using voltage-sensitive dyes and high resolution optical mapping techniques, we found that uncoupling of the ventricular myocardium results in ectopic sites of ventricular activation. Our data indicate that this behavior reflects alterations in source-sink relationships and paradoxical conduction across normally quiescent Purkinje-ventricular muscle junctions. The aberrant activation profiles are associated with wavefront collisions, which in the setting of slow conduction may account for the highly arrhythmogenic behavior of Connexin43-deficient hearts. Thus, the extent of gap junction remodeling in diseased myocardium is a critical determinant of cardiac excitation patterns and arrhythmia susceptibility.

Electrical uncoupling of the murine heart by cardiac-restricted inactivation of the Connexin43 (Cx43) gap junction channel gene results in slowing of ventricular conduction velocity (CV), spontaneous ventricular arrhythmias, and sudden cardiac death (1). However, the early mortality in these conditional knockout (CKO) mice complicated mechanistic studies of arrhythmia initiation and maintenance. Therefore, by selectively breeding longer-term survivors, we recently established a subline of conditional knockout mice, termed OCKO mice, in which the Cre-dependent inactivation of Cx43 in the myocardium was temporally delayed and death from ventricular arrhythmias, although still inevitable, was correspondingly delayed on average by about a month (2). ECGs in OCKO mice during the weeks preceding their demise demonstrated that all were in normal sinus rhythm (NSR), but, surprisingly, unlike control mice, they displayed a progressive reduction in the amplitude of the QRS complex, a time- and voltage-dependent signal indicative of myocardial depolarization (2). The aim of the current study was to investigate the mechanisms through which reductions in cell–cell coupling influence cardiac electrophysiological behavior and ultimately lead to the development of ventricular tachyarrhythmias.

Methods

Mice. Cardiac-restricted Cx43-mutant mice have been described, as have the derivative OCKO subline, obtained by selectively breeding longer-term survivors. The OCKO mice have progressive loss of Cx43 in the ventricular myocardium without compensatory changes in the abundance of either Cx45 or Cx40 (1, 2). Control mice did not carry the Cre recombinase transgene. All procedures were approved by the New York University School of Medicine Institutional Animal Care and Use Committee.

Patch-Clamp Recordings. Action potentials were obtained from myocytes from 6- to 12-week-old mice according to published techniques (3–5).

Optical Mapping. Procedures for optical mapping of ventricular activation and the calculation of conduction velocities have been described at length elsewhere (1, 6, 7). Epicardial pacing was performed with a unipolar electrode (stimuli at 1.5× diastolic threshold) at a cycle length of 120 ms. Endocardial pacing was performed with an octapolar electrode (CIBer mouse-EP, NuMed, Hopkinton, NY) introduced into the right ventricle through the tricuspid valve, which was visualized after unroofing the right atrium. A basic cycle length of 100 ms was used with stimuli at 2× diastolic threshold. No pharmacological or mechanical manipulations were used to limit motion. For pharmacological uncoupling, palmitoleic acid (Sigma–Aldrich) was added to the perfusate.

Statistical Analysis. Data are expressed as mean ± SEM. P values <0.05 were considered statistically significant.

Results

We previously reported that electrocardiograms from OCKO showed progressive diminution in the amplitude of the QRS complex, indicative of altered ventricular depolarization. An example of this phenomenon is shown in Fig. 1. To explore the basis for this behavior, we first investigated the electrophysiological properties of individual myocytes isolated from OCKO mice and compared these with myocytes from control mice. Current-clamp studies demonstrated that action potential amplitudes (125.3 ± 2.0 mV vs. 122.2 ± 2.3 mV; OCKO vs. control; P = NS), and maximum upstroke velocities (264.4 ± 14.4 V/s vs. 244.1 ± 15.4 V/s; P = NS) were not significantly different in adult OCKO (n = 33) and control (n = 28) myocytes, suggesting that cell autonomous properties were not primarily responsible for the electrocardiographic abnormalities observed in intact mice.

Fig. 1.

Diminished QRS amplitude in OCKO mice. Representative three lead surface ECG (leads I, II, and III) of control (A) and OCKO (B) mice are shown.

Slowing of CV has been suggested as a possible mechanism for diminution of QRS amplitude (8). Indeed, in hearts from 8- to 16-week-old OCKO mice, CV_min was diminished by ≈50% compared with age-matched controls (0.206 ± 0.017 m/s (OCKO) vs. 0.402 ± 0.019 m/s (control), P < 0.001) (Fig. 2 A and B). However, treatment of control hearts with the gap junction uncoupler palmitoleic acid (PA), at doses (5–10 mM) that diminished CV_min to a similar extent as did genetic uncoupling [0.238 ± 0.005 m/s (control + PA) vs. 0.206 ± 0.017 m/s (OCKO), P = NS] did not result in any measurable diminution of QRS amplitude [1.15 ± 0.30 mV (control + PA) vs. 1.05 ± 0.16 mV (control), P = NS] (Fig. 2C).

Fig. 2.

Analysis of isolated-perfused hearts. (A–C) Representative optical maps (Left), volume conducted ECGs in sinus rhythm (Center), and volume conducted ECGS during epicardial ventricular pacing (Right) in isolated-perfused control hearts (A), OCKO hearts (B), and control hearts treated with PA at doses that decreased CV_min by 50% (C). Shown are optical maps obtained during sinus rhythm in a series of control (D) and OCKO (E) hearts.

To determine whether the pattern, rather than the rapidity, of myocardial activation was a key determinant of the diminished QRS amplitude, we examined the effects of bypassing the His-Purkinje system with direct ventricular pacing on QRS amplitude in isolated-perfused OCKO and control hearts. In fact, ventricular pacing of OCKO hearts substantially normalized the diminished QRS amplitude seen during sinus rhythm (Fig. 2). We quantified the extent of augmentation by calculating the ratio of QRS amplitude with pacing compared with that in NSR (QRS_VP/QRS_SR). This value was significantly increased in OCKO hearts compared with controls (5.14 ± 0.96 vs. 1.71 ± 0.18, P < 0.05). In contrast, treatment of control hearts with PA had no effect on the QRS_VP/QRS_SR ratio (2.58 ± 0.81 vs. 1.71 ± 0.18, P = NS). Thus, moderate slowing of impulse propagation throughout the myocardium seemed not to be a major cause of the diminished QRS amplitude. Rather, the diminution of QRS amplitude observed in OCKO mice during NSR seemed related to the pattern of ventricular excitation, implicating aberrant activation of the myocardium by the specialized cardiac conduction system.

To test this hypothesis directly, we recorded voltage-dependent activity in isolated-perfused control and OCKO hearts in NSR (1). As expected, control hearts showed epicardial breakthroughs over each ventricular apex (Fig. 2D), typical of mature mammalian hearts, including humans and mice (9). In contrast, OCKO hearts showed a range of aberrant sinus activation patterns (Fig. 2E). The abnormalities ranged from only one epicardial breakthrough, which always occurred on the right ventricle, to multiple sites of activation throughout the heart, resulting in wavefront collisions within the ventricular myocardium.

We explored several potential mechanisms that might account for these aberrant activation patterns during sinus rhythm in the OCKO hearts. First, it is well known that gap junction channels play critical roles during cardiovascular morphogenesis (10–14) and loss of Cx43 in the developing ventricular myocardium could conceivably influence the patterning of the underlying Purkinje fiber network, thereby providing an anatomic basis for the aberrant activation patterns. To explore this possibility, we crossed the OCKO mice with the CCS-lacZ reporter strain, in which a β-galactosidase transgene is specifically expressed throughout the murine specialized cardiac conduction system, including the distal Purkinje fiber network (15). As shown in Fig. 3, the appearance of the ventricular portion of the conduction system in OCKO hearts was similar to control hearts expressing normal levels of Cx43. Thus, the aberrant sinus activation patterns in OCKO mice seemed functional in nature, rather than a manifestation of any underlying structural conduction system abnormalities.

Fig. 3.

Normal patterning of the specialized conduction system in OCKO hearts. Shown is representative LacZ staining of a CCS-lacZ heart (A) and compound transgenic CCS-lacZ/OCKO hearts (B and C).

We also considered that the aberrant epicardial activation patterns we recorded might reflect the spatially heterogeneous nature of the knockout induced in the OCKO model, in which residual foci of cells that continue to express Cx43 are intermingled with those in which Cre-mediated inactivation has taken place (2). Conceivably, excitation of the endocardial surface of the heart by the specialized cardiac conduction system at Purkinje-ventricular muscle junctions may be normal, but the excitatory wavefronts may fractionate as they propagate transmurally through heterogeneously excitable myocardium, manifest as multiple breakthroughs on the mapped epicardial surface. If wavefront fractionation were occurring during transmural propagation, an aberrant epicardial pattern should be seen not only during NSR, but with endocardial pacing as well. However, although mapping of an OCKO heart during NSR revealed, as expected, multiple epicardial breakthroughs on both the right and left free walls of the heart (Fig. 4A), pacing of the endocardial surface of the RV of this same heart resulted in smooth activation of the myocardium (Fig. 4B). The epicardial breakthrough was first visible overlying the site of stimulation in the right ventricle and propagated without evidence of fractionation across both the right and left ventricular free walls. These data excluded the possibility that spatially heterogeneous uncoupling was responsible for the aberrant activation patterns we recorded.

Fig. 4.

Optical maps during endocardial pacing and ventricular tachyarrhythmias. Shown are sequential optical maps obtained from an OCKO heart obtained during sinus rhythm (A) and then during pacing of the endocardial surface of the right ventricle (B). Paired images of the right (RV) and left (LV) ventricular free walls are shown. In each image, the apex is at the bottom left and the base is at the top right. (C and D) Maps showing reentry during sustained ventricular tachycardia in two OCKO hearts.

In mammalian hearts, including the mouse, Purkinje-ventricular junctions are arranged in a complex three-dimensional pattern throughout both ventricular chambers (9, 16). Evidence from canine studies, where studies of impulse propagation across Purkinje-ventricular junctions are technically feasible, indicates that propagation across only a subset of such junctions is normally successful (17, 18). Our results showing multiple ectopic sites of activation of the ventricular myocardium in the setting of a structurally normal cardiac conduction system indicate that uncoupling must increase the proportion of Purkinje-ventricular junctions across which conduction succeeds. Indeed, computer simulation models and direct experimentation in cell culture systems have predicted that partial uncoupling of focal areas of myocardial tissue should allow successful propagation of impulses at these regions due to improvements in current source and current load matching (19–21). In fact, in studies modeling a narrow strand emerging into a large tissue area, decreases in coupling in the large medium facilitated conduction and decreased the diameter of the strand required for successful propagation (22). This situation is quite analogous to the OCKO model, where coupling is specifically reduced in the Cx43-expressing area of abrupt tissue expansion (ventricular myocardium), but preserved in the Cx40-expressing strands (Purkinje fibers). Thus, the critical strand diameter (h_c) is reduced, and impulses carried by thin fibers that normally block may now successfully propagate into the ventricular myocardium.

As a consequence of the multiple breakthroughs we observed in many of the OCKO mice, the likelihood of wavefront collisions in the ventricular myocardium was enhanced. Such collisions, in concert with CV slowing, provide a substrate highly conducive to the initiation and maintenance of reentrant arrhythmias (23). Indeed, the majority of OCKO hearts studied developed incessant ventricular tachycardia, either spontaneously or with programmed stimulation. We successfully mapped the epicardial electrical activation patterns during sustained arrhythmias in Langendorff-perfused hearts from nine OCKO mice. In all such mice, repetitive, self-sustained activation sequences were observed, including five with circulating waves indicative of reentry, as shown in Fig. 4 C and D.

Taken together, these data demonstrate complex interactions at the Purkinje-ventricular junction that likely reflect both the anatomic characteristics of the Purkinje fibers as well as the passive and active electrical properties of the Purkinjes and the downstream ventricular tissue. Uncoupling of the myocardium, as might occur acutely with stressors such as ischemia or chronically as a manifestation of gap junction remodeling, may enhance propagation across normally quiescent Purkinje-ventricular junctions, promoting wavefront collisions that serve to initiate the reentrant process. Thus, strategies to inhibit or reverse gap junction remodeling may be of value in the prevention of lethal cardiac arrhythmias.