Mechanisms Underlying Conduction Slowing and Arrhythmogenesis in Nonischemic Dilated Cardiomyopathy

Transcription

1 Mechanisms Underlying Conduction Slowing and Arrhythmogenesis in Nonischemic Dilated Cardiomyopathy Fadi G. Akar, David D. Spragg, Richard S. Tunin, David A. Kass, Gordon F. Tomaselli Downloaded from by guest on July 17, 2018 Abstract Heart Failure (HF) is associated with an increased risk of sudden death caused by ventricular tachyarrhythmias. Recent studies have implicated repolarization abnormalities and, in particular, exaggerated heterogeneity of transmural repolarization in the genesis of polymorphic ventricular tachycardia in a canine model of nonischemic dilated cardiomyopathy. The presence and degree to which conduction abnormalities play a role in arrhythmogenesis in this model are uncertain. HF was produced in dogs by rapid RV-pacing for 3 to 4 weeks. High-resolution optical action potentials were recorded from epicardial and endocardial surfaces of arterially perfused canine wedge preparations isolated from LV and RV of normal and failing dogs. Cellular and molecular determinants of conduction were investigated using patch-clamp recordings, Western blot analysis, and immunocytochemistry. HF was associated with marked prolongation (by 33%) of the QRS duration of the volume conducted electrocardiogram and significant ( 20%) slowing of epicardial and endocardial conduction velocities (CV) in both LV and RV. Cx43 expression was reduced by 40% in epicardial and endocardial layers of the LV, but was unchanged in the RV of failing hearts. Despite greater epicardial than endocardial Cx43 expression, epicardial CV was consistently slower (P 0.01). Immunocytochemical analysis revealed predominant colocalization of Cx43 with N-cadherin in normal versus failing samples, because Cx43 was redistributed from the intercalated disk to lateral cell borders in failing tissue. Moreover, a significant (P 0.05) increase in hypophosphorylated Cx43 was detected in the LV and RV of failing hearts. Action potential upstroke velocities in isolated ventricular myocytes from normal and failing hearts were not different (P 0.8, not significant), and Masson trichrome staining revealed no significant change in fibrosis content in HF. Nonischemic dilated cardiomyopathy is associated with significant slowing of CV that was not directly related to reduced Cx43 expression. Changes in phosphorylation and localization of Cx43 may contribute to gap-junction dysfunction, CV slowing, and arrhythmias in HF. (Circ Res. 2004;95: ) Key Words: heart failure arrhythmias optical mapping connexin gap junctions Sudden cardiac death (SCD), presumably because of ventricular tachyarrhythmias, accounts for 50% of the mortality in patients with congestive heart failure (HF). 1,2 Recent studies have highlighted the importance of repolarization abnormalities, including nonuniform prolongation of action potential durations (APD) across the ventricular wall in a canine model of nonischemic dilated cardiomyopathy. 3 Such repolarization changes may lead to the development of reentrant arrhythmias in HF. 3 When considering the prerequisites for reentry, however, two conditions must be met: (1) the excitation wavefront must undergo unidirectional conduction block; and (2) the path of the reentrant circuit must be sufficiently long or conduction of the wavefront sufficiently slow, such that each site along the circuit has ample opportunity to regain excitability before the return of the circulating wave to avoid collision and extinction (ie, reentrant circuit wavelength must be shorter than the path length). Despite our increased understanding of repolarization changes in HF, the extent to which ventricular conduction abnormalities are present in this model, and their underlying cellular and molecular mechanisms, remains unknown. Given the prominence of conduction slowing in the development of reentry, 4 we hypothesized that conduction velocity (CV) is altered in the pacing-tachycardia HF model in a way that promotes the development of potentially lethal reentrant ventricular arrhythmias. In the heart, impulse propagation is dependent on several factors, including intrinsic membrane excitability, intra- and extracellular resistivities, and cell-to-cell coupling. 5 Cell-tocell coupling is mediated by cardiac gap junction channels, which are comprised of a family of connexins (Cx), of which Cx43 is the predominant ventricular isoform. 6,7 Altered expression and distribution of Cx proteins is a recurring theme in the remodeling associated with a number of structural heart diseases Recently, reduced Cx43 expression was directly linked to changes in CV and repolarization gradients in the dog, 11 whereas other studies have demonstrated a lack of Original received February 23, 2004; revision received August 25, 2004; accepted August 26, From the Division of Cardiology, Johns Hopkins University School of Medicine, Baltimore, Md. This manuscript was sent to Michael Rosen, Consulting Editor, for review by expert referees, editorial decision, and final disposition. Correspondence to Gordon F. Tomaselli, MD, Professor and Vice-Chair of Research, Department of Medicine Division of Cardiology Johns Hopkins University, 720 Rutland Ave, Ross844, Baltimore, MD gtomasel@jhmi.edu 2004 American Heart Association, Inc. Circulation Research is available at DOI: /01.RES c 717

2 718 Circulation Research October 1, 2004 effect of even major changes in Cx43 expression on CV in the intact mouse 12 and monolayers of neonatal mouse ventricular myocytes. 13 Altered conduction is prominent and regionally heterogeneous in ischemic heart disease 14 and is caused by changes in both active 15 and passive membrane properties The mechanistic understanding of changes in conduction properties in models of ischemic cardiomyopathy are complicated by model-dependent regional heterogeneity, the presence of acute ischemia, and ischemic injury, as well as LV dysfunction. We chose to avoid the confounding influences of ischemia by studying the effects of LV dysfunction alone on conduction in a model of nonischemic dilated cardiomyopathy. 18 Our data demonstrate important changes in impulse propagation in HF that are not directly related to changes in the expression level of Cx43, but rather to the phosphorylation state and the localization of the protein. These changes likely contribute to increased risk of ventricular arrhythmias and SCD in patients with HF. Materials and Methods Experimental Model HF was produced by 3 to 4 weeks of rapid pacing as described previously. 18 LV wedges from 9 normal and 11 failing dogs and RV wedges from 3 normal and 3 failing dogs were prepared as previously described. 3,19 Adjacent myocardial tissue sections from Figure 1. A, Canine wedge preparation in custom designed imaging chamber. This preparation is mounted with the cut transmural surface facing the optical window. The duration of the QRS complex of the volume conducted ECG was prolonged significantly in HF wedges. B, Depolarization isochrone maps of the epicardial and endocardial surfaces of representative canine wedge preparations from normal (NL) and failing (HF) ventricles paced from the sites indicated by the square wave. C, Representative action potential upstrokes from sites along the slow axis of propagation in each layer. D, Summary bar plots of left and right ventricular epicardial and endocardial conduction velocities along the fast and slow axes of propagation. Fast and slow components of CV are significantly reduced in wedges from failing hearts. *P epicardial ( 1 mm) and endocardial ( 1 mm) layers were processed for Western blotting and immunohistochemical staining. Myocytes were enzymatically dissociated from epicardial and endocardial layers, as previously described, and action potentials were measured using standard patch-clamp recording in current-clamp mode, as described previously. 18,20 Optical Action Potential Mapping Wedges of myocardium (Figure 1A), dissected from the lateral LV free-wall and the RV, were perfused with normal Tyrodes solution at 36 1 C. The arterially perfused wedge preparations were placed in a custom-designed imaging chamber equipped to record volumeconducted electrocardiograms (ECGs) (Figure 1A). After staining with the voltage-sensitive dye, di-4-anepps, the imaged surface (epicardium or endocardium) of wedges was stabilized against a flat optical window. Preparations were excited by filtered (515 5 nm) light from a 192-W Tungsten-Halogen lamp. Emitted fluorescence was long-pass filtered at 630 nm and focused onto the photodiode array with a custom-designed optical macroscope. Western Blotting Epicardial ( 1 mm) and endocardial ( 1 mm) tissue layers were dissected from LV and RV myocardium adjacent to optically mapped wedges. Proteins were prepared as previously described. 18 Primary antibody incubations were performed overnight at 4 C as previously described 21 using antibodies to measure total Cx43 (1:1000 dilution; Chemicon Inc, Temeculda, Calif), the nonphosphorylated component of Cx43 (1:1000; Zymed Inc, San Francisco, Calif), and Cx45 expression (1:1000; gift from Dr Jeffery E. Saffitz, Washington University, St. Louis, Mo).

3 Akar et al Mechanisms of Conduction Slowing in Heart Failure 719 Figure 2. A, Epicardial (EPI) and endocardial (ENDO) transverse wavelength ( ) calculated during premature stimulation at relatively long, intermediate, and short coupling intervals in normal and failing wedges. Dotted line indicates mean length of preparations. B, Isochrone maps depicting spread of depolarization of premature impulses delivered at long, intermediate, and short coupling intervals in representative wedges. Microscopy The cellular distribution of Cx43 in tissue isolated from normal and failing hearts was determined using confocal immunofluorescence, and regional fibrosis was examined using standard Masson trichrome staining, as detailed in the expanded Materials and Methods section in the online data supplement available at Data Analysis All comparisons between normal and failing hearts were made using the Student t test. Summary data are presented as mean SD. Differences were considered significant at P Results Conduction Slowing and Arrhythmogenesis in HF Conduction Slowing The pacing tachycardia HF model exhibits several distinct electrocardiographic changes that are evident on the volumeconducted wedge ECG. These include a significant (P 0.01) prolongation of the QT-interval (control, ms; HF, ms), as well as the time between the peak and end of the T-wave (control, 33 7 ms; HF, 77 9 ms), both of which highlight the prolonged and heterogeneous nature of repolarization in the wedge. Importantly, the duration of the QRS complex in HF wedges was 136% that of control (Figure 1, P 0.02), consistent with slowed conduction. To further define the basis for prolonged QRS duration in HF, high-resolution optical action potential mapping was performed on epicardial and endocardial surfaces of the canine wedge preparation during steady-state pacing. As shown in Figure 1, conduction across the epicardial and endocardial surfaces of the wedge preparation is anisotropic with a fast-to-slow CV ratio of and , respectively. Depolarization isochrone maps were constructed to quantify CV, with sites of early activation shown in blue and those of late activation in red (Figure 1B). Superimposed upstrokes of representative action potentials recorded from adjacent sites along the slow axis of impulse propagation are displayed in Figure 1C. In LV wedges from normal hearts, endocardial CV was significantly (P 0.01) faster than epicardial CV (Figure 1B and 1D) in both the fast and slow axes of propagation. In HF CV is significantly decreased in both the fast and slow directions on both surfaces, as evidenced by the crowding of isochrone lines (Figure 1B) and the greater delay between action potential upstrokes (Figure 1C). It is notable that conduction slowing in epicardial and endocardial layers of HF was uniform across the mapping fields and not localized to specific regions. The sequential spread of membrane depolarization across the epicardial surface of representative normal and failing canine wedges is also shown in the animations (Video Supplement 1), which demonstrate markedly slower propagation in HF. Reduced Wavelength To assess the functional consequence of reduced CV on arrhythmia development in HF, we measured the cardiac wavelength ( ) along the transverse direction of propagation (product of CV and average APD) in both normal and HF preparations during baseline pacing and after delivery of premature stimuli from the same electrode at relatively long (S1S2 400 ms), intermediate (S1S2 200 ms), and short (S1S2 140 ms) coupling intervals. During steady-state pacing, was relatively long ( 10 cm) in both normal and HF wedges. Interestingly, because of faster CV and longer baseline APD, was significantly longer in endocardium

4 720 Circulation Research October 1, 2004 Figure 3. Action potentials recorded from representative wedges isolated from a normal (left) and failing (right) dog during premature stimulation at a comparable degree of refractoriness, as indexed by the percentage of the action potential amplitude. The premature impulse exhibited decremental conduction and block in the normal wedge but was successfully conducted in the failing wedge, which exhibited a run of polymorphic VT. than epicardium in both normal (by 34%, P 0.01) and failing (by 24%, P 0.01) hearts. During premature stimulation, progressively shortened in both epicardium and endocardium (Figure 2A). Importantly, however, stimulation at intermediate coupling intervals caused more pronounced shortening of in wedges from failing compared with normal hearts (Figure 2), attributable, in part, to disproportionate slowing of CV in HF (Figure 2B). Finally, premature stimulation at a relatively short coupling interval resulted in conduction block in the normal but not the failing preparations, which exhibited marked slowing of CV and a dramatic shortening of (Figure 2B). Importantly, whereas in normal wedges was always longer than the average length of the preparations (Figure 2A, dotted line), in HF, was decreased to less than the mean preparation length during premature stimulation at short coupling intervals. Shown in Figure 3 are representative epicardial action potentials recorded before and after delivery of a closely coupled premature stimulus at a comparable degree of refractoriness in wedges from normal (left) and failing (right) canine hearts. Although in the normal wedge, the premature beat conducted in a decremental fashion and blocked in all directions within 2 mm of the stimulating electrode, it was able to successfully conduct across the entire mapped surface of the failing wedge. Safe conduction of premature impulses at short coupling intervals was associated with the induction of polymorphic ventricular tachycardia (VT) in preparations from failing (6 of 14) but not normal (0 of 12) hearts. Mechanisms Underlying Conduction Slowing in Heart Failure Cellular Excitability and the Interstitium To investigate the cellular and molecular mechanisms underlying conduction slowing in HF, we assessed the roles of reduced excitability, fibrosis in the extracellular matrix, and Figure 4. A, Representative action potentials recorded using the patch clamp technique in current-clamp mode in isolated epicardial (EPI) and endocardial (ENDO) myocytes from a normal (NL) and failing (HF) LV. B, A bar plot of the average action potential upstroke velocity in epicardial and endocardial cells of normal and failing hearts. C, Representative Masson trichrome stains of epicardial and endocardial tissue slices from a normal and failing heart. altered gap junction protein expression, distribution, and phosphorylation. We have previously shown that Na -current density in ventricular myocytes is unchanged in this model. 18 The role of reduced excitability in conduction slowing was further examined by measuring the action potential upstroke velocity in myocytes isolated from normal and failing hearts (Figure 4). The maximum upstroke velocity was not significantly different in either endocardial (P 0.8, not significant [NS]) or epicardial (Figure 4; P 0.8, NS) myocytes of normal and failing hearts, suggesting that reduced Na - channel availability was not responsible for CV slowing. Changes in the amount and type of fibrous tissue in the interstitium can alter tissue resistivity and CV. Visual examination of Masson trichrome stained sections from the lateral LV wall of normal and failing hearts by an independent member of the pathology division in a blinded fashion indicated no discernable difference in fibrosis content. Similarly, detailed quantitative image analysis of multiple (n 40) myocardial sections (Figure 4) exhibited no major change in the percentage of epicardial ( % HF; % control; P 0.07) and endocardial ( % HF; % control; P 0.08) fibrous tissue content per total crosssectional surface area, which remained relatively minor. Connexin Expression The role of gap junction channels in HF-induced CV slowing was assessed by determining the level of expression of Cx43

5 Akar et al Mechanisms of Conduction Slowing in Heart Failure 721 Figure 5. Epicardial and endocardial Cx43 expression in left (A) and right (B) ventricles from normal (NL) and failing (HF) hearts. In HF, Cx43 expression is downregulated in both epicardial and endocardial layers of the LV but not RV. C, Correlation between total Cx43 expression (circles) and CV (triangles) in epicardial (EPI) and endocardial (ENDO) layers of normal hearts. D, Correlation between Cx43 (circles) and CV (triangles) in RV and LV of failing hearts. No clear correlation exists, indicating that CV is not modulated directly by the level of expression of Cx43. E, Cx45 expression in epicardial and endocardial layers of normal and failing hearts. There is no significant change in the level of immunoreactive Cx45. in epicardial and endocardial layers of normal and failing hearts. Shown in Figure 5A is a bar plot of Cx43 expression normalized to calsequestrin (top) and a representative Western blot (bottom). In normal myocardium, the amount of epicardial Cx43 is significantly greater (by 29%, P 0.001) than endocardial Cx43. In the failing heart, Cx43 expression decreases in both layers (epicardial 51%, P 0.01; endocardial 45%, P 0.01) of the LV. In contrast, Cx43 expression was not altered (P NS) in either epicardial or endocardial layers of the RV in HF compared with normal hearts (Figure 5B). Interestingly, the expression of Cx43 did not explain differences in CV between epicardial and endocardial layers of the dog. As shown in Figure 5C, the epicardial layer, which exhibited greater immunoreactive Cx43 expression, had a reduced CV compared with the endocardium. Moreover, CV slowing in RV preparations from failing hearts was not associated with reduced Cx43 expression in that chamber (Figure 5D). Finally, it did not appear that Cx43 was replaced by another connexin isoform, because neither epicardial nor endocardial Cx45 expression was altered in HF (Figure 5E). Figure 6. A, Representative epicardial sections from a normal and failing heart showing lateralization of Cx43 in HF. B, Colocalization of Cx43 and N-cadherin fluorescence in representative epicardial and endocardial tissue slices from a failing heart. C, Bar plots quantifying the degree of lateralization (absence of colocalization with N-cadherin) of Cx43 in epicardial and endocardial layers of normal (NL) and failing (HF) hearts. Cx43 Distribution Because CV slowing could not be explained by the level of expression of Cx43 (Figure 5), we used immunohistochemical staining and confocal microscopy of epicardial and endocardial sections to examine the distribution of Cx43 in normal and failing ventricular myocardium. In normal hearts, Cx43 is primarily localized at the end-to-end junctions of adjacent myocytes, or the intercalated disk. In contrast, Cx43 in HF sections was frequently observed along the length of myocytes (Figure 6A). The extent of Cx43 lateralization was quantified as the percentage of Cx43 fluorescence that is not colocalized with N-cadherin fluorescence versus total Cx43 signal. Shown in Figure 6B are representative epicardial and endocardial tissue slices from a failing heart that demonstrate the presence of Cx43 signal along the lateral margins of myocytes where N-cadherin is not present (arrows). On average, the percentage of Cx43 signal that was not colocalized with N-cadherin increased by 2-fold in epicardial and endocardial layers of failing hearts relative to controls (P 0.01) (Figure 6C).

6 722 Circulation Research October 1, 2004 Figure 7. A, Western blot of Cx43 in a tissue lysate treated with phosphatase inhibitors from normal canine myocardium. Multiple bands are present, including a faint band at a molecular mass of 41 kda representing a hypophosphorylated component of Cx43 (arrow). B, Western blots of samples from normal (NL) and failing (HF) myocardium indicating relative change in the density of the lower band. C, Western blot of Cx43 using antibody that is specific to the hypophosphorylated component of Cx43 (see under Materials and Methods ). D, Bar plots indicating increased expression of the hypophosphorylated component of Cx43 in HF in both RV and LV. Cx43 Phosphorylation Phosphorylation alters the function of Cx43. We investigated whether there were changes in the phosphorylation state of Cx43, which might reduce gap junction function and therefore result in CV slowing in this model of HF. This was performed by measuring the phosphorylated and hypophosphorylated components of Cx43 in Western blots of samples that were isolated in the presence of phosphatase inhibitors. As shown in Figure 7, in addition to reduced expression of total Cx43 in HF, there was a significant change in the amount of the hypophosphorylated component. In normal hearts, Cx43 is predominantly phosphorylated, migrating primarily ( 95% of total expression) at higher (44 to 46 kda) molecular weight bands (Figure 7A). Interestingly, in HF, a band at a lower molecular mass ( 41 kda) emerges, indicating an overall increase in apparently hypophosphorylated Cx43 (Figure 7B). A similar pattern of Cx43 dephosphorylation, but with no change in total band intensity (phosphorylated and hypophosphorylated components), was found in Western blots of tissue lysates extracted from the RV of failing compared with normal hearts (not shown). Moreover, quantitative measurement of the hypophosphorylated form of Cx43 (Figure 7) using a commercially available antibody that is specific for Cx43 dephosphorylated at serine-368, indicated a marked increase (P 0.01) in the level of expression of the hypophosphorylated form in failing compared with normal hearts in both LV (by 123%) and RV (by 88%). Structural Changes: Myocyte Dimensions Cardiac impulse propagation is dependent on intercellular resistivity, which, in turn, is a function of the shape and size of cardiac myocytes. Therefore, relative changes in myocyte Figure 8. Representative endocardial myocytes from a normal (NL) and failing (HF) heart and average cell dimensions including length, width, perimeter, and area of epicardial (EPI) and endocardial (ENDO) myocytes from normal and failing ventricles. dimensions, including length, width, perimeter, and area were quantified in myocytes isolated from epicardial and endocardial layers of normal and failing hearts. As shown in Figure 8, the tachycardia pacing HF model is associated with a significant increase in the length of cells isolated from the epicardium but not the endocardium. Moreover, the width of both epicardial and endocardial myocytes is decreased in HF. Despite significant changes in length and width, there were no significant changes in the overall area or perimeter of myocytes isolated from failing compared with control hearts. Discussion SCD attributable to ventricular tachyarrhythmias is a major cause of mortality in patients with HF. 1,2 Recent studies have appreciably advanced our understanding of repolarization abnormalities, including enhanced transmural repolarization gradients in the genesis of polymorphic VT in HF. 2,3 In the present study, high-resolution optical action potential mapping was performed in the canine wedge preparation for the purpose of measuring the velocity and pattern of epicardial and endocardial conduction in normal and failing hearts. Slowed impulse propagation was correlated with cellular and molecular determinants of myocardial conduction. Conduction Slowing in HF Although conduction slowing in models of myocardial infarction-induced HF is generally accepted, less is known regarding CV in nonischemic, noninfarcted myopathic hearts. A major finding of this study is that, in HF, both epicardial and endocardial CV are significantly reduced in both LV and

7 Akar et al Mechanisms of Conduction Slowing in Heart Failure 723 RV. Our report is consistent with previous data in a transgenic mouse model of dilated cardiomyopathy, which exhibits similar degrees of epicardial CV slowing. 22 To our knowledge, this is the first report of the measurement of epicardial and endocardial CV in both chambers of the failing heart. Myocardial conduction is dependent on membrane excitability and passive tissue resistivity. Membrane excitability is dictated by the availability of I Na, whereas resistivity is a function of intra-, extra-, and intercellular resistances. A major advantage of our experimental approach is the ability to accurately measure both epicardial and endocardial CV in LV and RV and directly correlate these metrics to cellular and molecular alterations in the same hearts. Our findings indicate that CV slowing in HF was not related to reduced excitability because the action potential upstroke velocity was not altered in HF. The absence of a change in action potential upstroke velocity is consistent with our previous data that demonstrated no difference in I Na density in myocytes isolated from the same model of HF. 18 In this pacing-tachycardia model of dilated cardiomyopathy, we did not observe major changes in the extracellular matrix that could explain slowing of CV. In fact, tissue sections isolated from epicardial and endocardial layers of the lateral LV free walls of normal and failing hearts exhibited no significant change in fibrosis content, which often contributes to CV slowing in human HF. This is most likely attributable to the relatively rapid induction of HF in this model and is consistent with the findings of a recent report that demonstrates markedly greater fibrosis in the atrium compared with the ventricles in a similar model. 23 Therefore, our results indicate that severe CV slowing can occur in the absence of macroscopic disruption of the extracellular network. Additionally, the overall pattern and degree of the anisotropy of conduction were not disrupted regionally, but instead a diffuse slowing throughout the wedge occurred, further arguing against localized disruption of tissue resistivity by fibrosis. In the heart, intracellular resistance is a function of the dimensions of the cardiac myocyte. Therefore, severe changes in myocyte dimensions could produce measurable changes in CV. We observed significant changes in myocyte length-to-width ratios in HF. However, these changes were relatively subtle ( 25%) compared with those reported in a model of ventricular hypertrophy, which exhibit marked ( 200%) increases in myocyte lengths. 24 An increase in myocyte length and a decrease in width are expected to decrease longitudinal and increase transverse tissue resistivity, the net effect of which would be a tendency toward greater anisotropic propagation. Because the anisotropy of conduction was not altered, either the structural changes in myocytes were minor or other alterations such as the lateralization of Cx43 offset these changes. The influence of cell-to-cell communication on CV slowing in HF was assessed by measuring the expression, distribution, and posttranslational modification of the major ventricular gap junctional protein, Cx43, in normal and failing hearts. We observed a significant downregulation of Cx43 in canine pacing tachycardia cardiomyopathy, consistent with previous data regarding gap junction expression 9 and tissue resistance 25 in other models of HF. Our data demonstrate that in normal hearts, epicardial CV is significantly slower than endocardial CV, consistent with previous reports. 26,27 Interestingly, the difference in CV does not correlate with the level of expression of Cx43 in the respective layers (Figure 5), because epicardial Cx43 expression is 30% greater than endocardial Cx43 expression in normal canine myocardium. Therefore, these data support the notion that changes in Cx43 expression (of up to 30%) do not translate directly into conduction slowing, as has been argued in some previous studies 12,13 but not in others. 28 Furthermore, a recent report has shown that Cx43 expression is greater in endocardium than epicardium, 27 and another study has showed no transmural differences in Cx43 expression in dog. 29 Such conflicting data may be reconciled if important regional differences in Cx43 are present in the canine LV. In fact, we have recently discovered major differences in the expression of key molecules, including Cx43 between anterior and lateral walls of the LV. 30 To eliminate the potential confounding variable of regional heterogeneity of Cx43 expression, we limited our study to the effect of LV failure on the high-stress lateral myocardial wall of the LV and the RV. Further studies are required to assess the expression of Cx43 in other regions of the heart in HF. Importantly, despite the absence of a change in Cx43 expression in the RV of the failing dog, CV was significantly reduced. As such, the level of expression of Cx43 was not a good predictor of CV slowing in this model of HF, in contrast to the conclusion of another recent report. 11 Recent evidence supports the hypothesis that Cx43 and Cx45 can form heterotypic and heteromeric gap junction channels in the heart, with lower conductance than homomeric (Cx43 or Cx45 alone) channels Because Cx45 levels are unchanged in this model in the wake of downregulated Cx43, the effective Cx45:Cx43 ratio is increased, potentially promoting the formation of more heterotypic gap junctions with lower conductance than in normal hearts. Recently, it has been shown that Cx45 expression is increased in human HF, further contributing to enhanced Cx45:Cx43 ratios 35 and possibly favoring the formation of gap junction channels with reduced conductance. In addition to downregulation of Cx43 expression, its distribution in the epicardial and endocardial layers of the failing heart is also disordered. In normal hearts, Cx43 resides predominately at end-to-end junctions of cells or the intercalated disk. Interestingly, in HF, this distribution is dramatically altered, because immunoreactive Cx43 is detected along the cell length as well (Figure 6). Lateralization of Cx43 may be expected to reduce the anisotropy of impulse propagation in the heart. Because no changes in anisotropy were observed, the redistributed Cx43 may not be functional, raising the possibility that even Cx43 located at the intercalated disk may also be non- or hypofunctional in HF. Therefore, our data promote the notion that altered distribution and decreased phosphorylation of Cx43 promote conduction slowing in this model of HF.

8 724 Circulation Research October 1, 2004 Arrhythmogenesis in HF We evaluated the impact of conduction slowing in HF on arrhythmogenesis by measuring the transverse cardiac wavelength ( ) during steady state pacing and after delivery of premature stimuli over a range of coupling intervals. These measurements revealed preferential reduction of epicardial and endocardial in failing compared with normal hearts during premature electrical stimulation, facilitating the induction of sustained ( 3 second) episodes of polymorphic VT (Figure 3) after a critical wavelength was reached (Figure 2). Underlying the presence of shorter in the failing heart is its ability to support slower propagation more safely, which is consistent with previous data from computer simulations and myocyte monolayers that showed slower but safer conduction because of a reduced nonexcitatory current sink to neighboring myocytes when cell-to-cell coupling was reduced. 36,37 Such ability to support ultraslow conduction in HF allows in preparations from failing but not normal hearts to shorten to critical levels that are smaller than the physical dimensions of these preparations, a requirement for reentry to persist. Taken together, our results indicate that CV slowing in HF is a critical component of the arrhythmia substrate in this pathophysiological state and that CV slowing is caused by changes in cell-to-cell communication not related to the expression level of the major ventricular gap junction protein, Cx43. Limitations There are several limitations to this study. First, we have recently described important differences in Cx43 expression between anterior and lateral walls of the dyssynchronously contracting canine heart. 30 The present analysis focuses exclusively on the high-stress, late-activated lateral wall. Cx43 expression in the early-activated anterior wall and septum were not measured. Second, action potential upstroke velocities measured in isolated cell preparations may not translate directly to excitability in intact tissue, where cell-to-cell coupling is present. However, to completely eliminate the influence of cell-to-cell coupling on propagation, we assessed dv/dt max (maximum action potential upstroke velocity) in isolated myocytes, to complement our previous data regarding lack of I Na change in this model of nonischemic dilated cardiomyopathy. Third, in addition to Cx43 and Cx45, Cx40 is also present in the heart, but its expression was not measured. Although Cx40 is known to play a critical role in the formation of gap junctions in the cardiac conduction system, its role in the normal ventricle is thought to be negligible. Finally, although changes in the density of fibrous tissue in the lateral LV wall were minor compared with those found in the atria in earlier studies in the same model, 23 we cannot exclude the possibility that the constituents of the fibrous tissue were not altered in HF. Furthermore, subtle changes in the distribution of fibrous tissue in the interstitium could have eluded detection. Acknowledgments This study was supported by NIH grant P50 HL We are grateful for excellent technical support from Yanli Tian and Deborah DiSilvestre. References 1. Stevenson WG, Sweeney MO. Arrhythmias and sudden death in heart failure. Jpn Circ J. 1997;61: Tomaselli GF, Beuckelmann DJ, Calkins HG, Berger RD, Kessler PD, Lawrence JH, Kass D, Feldman AM, Marban E. Sudden cardiac death in heart failure. The role of abnormal repolarization. Circulation. 1994;90: Akar FG, Rosenbaum DS. Transmural electrophysiological heterogeneities underlying arrhythmogenesis in heart failure. Circ Res. 2003;93: Wit AL, Rosen MR. Pathophysiologic mechanisms of cardiac arrhythmias. Am Heart J. 1983;106: Spach MS, Kootsey JM. The nature of electrical propagation in cardiac muscle. Am J Physiol. 1983;244:H3 H Saffitz JE, Laing JG, Yamada KA. Connexin expression and turnover: implications for cardiac excitability. Circ Res. 2000;86: Peters NS. New insights into myocardial arrhythmogenesis: distribution of gap-junctional coupling in normal, ischaemic and hypertrophied human hearts. Clin Sci (Lond). 1996;90: Peters NS, Coromilas J, Severs NJ, Wit AL. Disturbed connexin43 gap junction distribution correlates with the location of reentrant circuits in the epicardial border zone of healing canine infarcts that cause ventricular tachycardia. Circulation. 1997;95: Peters NS, Green CR, Poole-Wilson PA, Severs NJ. Cardiac arrhythmogenesis and the gap junction. J Mol Cell Cardiol. 1995;27: De Mello WC. Cell coupling and impulse propagation in the failing heart. J Cardiovasc Electrophysiol. 1999;10: Poelzing S, Rosenbaum DS. Altered connexin43 expression produces arrhythmia substrate in heart failure. Am J Physiol Heart Circ Physiol In press. 12. Morley GE, Vaidya D, Jalife J. Characterization of conduction in the ventricles of normal and heterozygous Cx43 knockout mice using optical mapping. J Cardiovasc Electrophysiol. 2000;11: Thomas SP, Kucera JP, Bircher-Lehmann L, Rudy Y, Saffitz JE, Kleber AG. Impulse propagation in synthetic strands of neonatal cardiac myocytes with genetically reduced levels of connexin43. Circ Res. 2003; 92: Kleber AG. Conduction of the impulse in the ischemic myocardium implications for malignant ventricular arrhythmias. Experientia. 1987;43: Shaw RM, Rudy Y. Electrophysiologic effects of acute myocardial ischemia. A mechanistic investigation of action potential conduction and conduction failure. Circ Res. 1997;80: Yao JA, Hussain W, Patel P, Peters NS, Boyden PA, Wit AL. Remodeling of gap junctional channel function in epicardial border zone of healing canine infarcts. Circ Res. 2003;92: Weber KT. Extracellular matrix remodeling in heart failure: a role for de novo angiotensin II generation. Circulation. 1997;96: Kaab S, Nuss HB, Chiamvimonvat N, O Rourke B, Pak PH, Kass DA, Marban E, Tomaselli GF. Ionic mechanism of action potential prolongation in ventricular myocytes from dogs with pacing-induced heart failure. Circ Res. 1996;78: Akar FG, Yan GX, Antzelevitch C, Rosenbaum DS. Unique topographical distribution of M cells underlies reentrant mechanism of torsade de pointes in the long-qt syndrome. Circulation. 2002;105: Akar FG, Wu RC, Deschenes I, Armoundas AA, Piacentino V 3rd, Houser SR, Tomaselli GF. Phenotypic differences in transient outward K current of human and canine ventricular myocytes: insights into molecular composition of ventricular Ito. Am J Physiol Heart Circ Physiol. 2004;286:H602 H Beardslee MA, Lerner DL, Tadros PN, Laing JG, Beyer EC, Yamada KA, Kleber AG, Schuessler RB, Saffitz JE. Dephosphorylation and intracellular redistribution of ventricular connexin43 during electrical uncoupling induced by ischemia. Circ Res. 2000;87: Hall DG, Morley GE, Vaidya D, Ard M, Kimball TR, Witt SA, Colbert MC. Early onset heart failure in transgenic mice with dilated cardiomyopathy. Pediatr Res. 2000;48: Hanna N, Cardin S, Leung TK, Nattel S. Differences in atrial versus ventricular remodeling in dogs with ventricular tachypacing-induced congestive heart failure. Cardiovasc Res. 2004;63: Qin D, Zhang ZH, Caref EB, Boutjdir M, Jain P, el-sherif N. Cellular and ionic basis of arrhythmias in postinfarction remodeled ventricular myocardium. Circ Res. 1996;79:

9 Akar et al Mechanisms of Conduction Slowing in Heart Failure Cooklin M, Wallis WR, Sheridan DJ, Fry CH. Changes in cell-to-cell electrical coupling associated with left ventricular hypertrophy. Circ Res. 1997;80: Wu J, Zipes DP. Transmural reentry during acute global ischemia and reperfusion in canine ventricular muscle. Am J Physiol Heart Circ Physiol. 2001;280:H2717 H Poelzing S, Akar FG, Baron E, Rosenbaum DS. Heterogeneous connexin43 expression produces electrophysiological heterogeneities across ventricular wall. Am J Physiol Heart Circ Physiol. 2004;286: H2001 H Eloff BC, Lerner DL, Yamada KA, Schuessler RB, Saffitz JE, Rosenbaum DS. High resolution optical mapping reveals conduction slowing in connexin43 deficient mice. Cardiovasc Res. 2001;51: Patel PM, Plotnikov A, Kanagaratnam P, Shvilkin A, Sheehan CT, Xiong W, Danilo P Jr, Rosen MR, Peters NS. Altering ventricular activation remodels gap junction distribution in canine heart. J Cardiovasc Electrophysiol. 2001;12: Spragg DD, Leclercq C, Loghmani M, Faris OP, Tunin RS, DiSilvestre D, McVeigh ER, Tomaselli GF, Kass DA. Regional alterations in protein expression in the dyssynchronous failing heart. Circulation. 2003;108: Desplantez T, Halliday D, Dupont E, Weingart R. Cardiac connexins Cx43 and Cx45: formation of diverse gap junction channels with diverse electrical properties. Pflugers Arch. 2004;448: Beyer EC, Gemel J, Martinez A, Berthoud VM, Valiunas V, Moreno AP, Brink PR. Heteromeric mixing of connexins: compatibility of partners and functional consequences. Cell Commun Adhes. 2001;8: Martinez AD, Hayrapetyan V, Moreno AP, Beyer EC. Connexin43 and connexin45 form heteromeric gap junction channels in which individual components determine permeability and regulation. Circ Res. 2002;90: Cottrell GT, Wu Y, Burt JM. Cx40 and Cx43 expression ratio influences heteromeric/ heterotypic gap junction channel properties. Am J Physiol Cell Physiol. 2002;282:C1469 C Yamada KA, Rogers JG, Sundset R, Steinberg TH, Saffitz JE. Up-regulation of connexin45 in heart failure. J Cardiovasc Electrophysiol. 2003;14: Shaw RM, Rudy Y. Ionic mechanisms of propagation in cardiac tissue. Roles of the sodium and L-type calcium currents during reduced excitability and decreased gap junction coupling. Circ Res. 1997;81: Rohr S, Kucera JP, Kleber AG. Slow conduction in cardiac tissue, I: effects of a reduction of excitability versus a reduction of electrical coupling on microconduction. Circ Res. 1998;83:

10 Mechanisms Underlying Conduction Slowing and Arrhythmogenesis in Nonischemic Dilated Cardiomyopathy Fadi G. Akar, David D. Spragg, Richard S. Tunin, David A. Kass and Gordon F. Tomaselli Circ Res. 2004;95: ; originally published online September 2, 2004; doi: /01.RES c Circulation Research is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 2004 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Data Supplement (unedited) at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation Research can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Circulation Research is online at:

11 ONLINE METHODS Experimental Model Wedges ( 3.5x1.5x1.2 cm) of ventricular myocardium were isolated from control (n=12) and terminally failing (n=14) canine hearts. HF was produced by 3-4 weeks of rapid RV apical pacing as described previously. 1 Briefly, adult male mongrel dogs underwent insertion of a permanent transvenous pacemaker and were paced at 250PPM until clinical signs of HF, such as anorexia, lethargy, ascites, tachypnea, and muscle wasting were evident ( 3-4 weeks). LV end diastolic pressures (LVEDP) were measured in all dogs, and only those exhibiting an elevated LVEDP ranging from 30-40mmHg were used in this study. After the dogs were sacrificed and their hearts rapidly excised, LV wedges were isolated from 9 normal and 11 failing dogs and RV wedges from 3 normal and 3 failing dogs and prepared according to the technique of Yan and Antzelevitch, and as previously reported. 2,3 Adjacent myocardial tissue sections from epicardial (<1mm) and endocardial (<1mm) layers were quickly (within 5min of cardiac arrest) processed for subsequent western blotting and immunohistochemical staining. Myocytes were enzymatically dissociated from epicardial and endocardial layers as previously described. 1 Action potential upstroke velocities were measured in isolated cells from normal and failing hearts using standard patch clamp recording in currentclamp mode, as described previously. 1,4 Briefly, ventricular myocytes from epicardial and endocardial layers were isolated by perfusing a diagonal branch of the circumflex coronary artery with a nominally Ca 2+ -free solution, containing collagenase and protease. 1,4

12 Akar et. al. Conduction Slowing and Arrhythmogenesis in Heart Failure 2 Glass pipettes were prepared to have a final tip resistance of 2-3MΩ when filled with internal solution containing (in mmol/l): 120 K-glutamate, 10 KCl, 10 N-2- hydroxyethylpiperazine-n 2-ethanesulphonic acid (HEPES), 5 ethylene glycolbis(aminoethyl ether)-n,n,n,n -tetraacetic acid (EGTA), 5 MgATP. ph was adjusted to 7.2 with KOH, yielding a final K-concentration of 140mmol/L. Cells w ere perfused with Tyrode s solution containing (in mmol/l): 140 NaCl, 5 KCl, 1 MgCl 2, 2 CaCl 2, 10 HEPES, 10 glucose, adjusted to ph of 7.4. Epicardial and Endocardial Optical Action Potential Mapping We designed a system capable of recording 464 optical action potentials with high spatial (365µm), temporal (0.6ms), and voltage (0.5mV) resolutions from the epicardial and endocardial surfaces of arterially-perfused canine wedge preparations isolated from the LV and RV during epicardial and endocardial pacing, respectively. Wedges of myocardium (Figure 1, Panel A), dissected from the lateral LV freewall and the RV were perfused with normal Tyrodes solution at 36±1 C. The arteriallyperfused wedge preparations were placed in a custom-designed, temperature controlled, imaging chamber equipped to record volume conducted ECGs (Figure 1A). Following staining with the voltage-sensitive dye, di-4-anepps (15µmol/L for 10 minutes), the imaged surface (epicardium or endocardium) of the wedges was stabilized against a flat optical window. The preparation was excited by filtered (515±5nm) light from a 192-W Tungsten-Halogen lamp (Oriel Corp, Stratford, CT). Emitted fluorescence was long-pass filtered at 630nm (Chroma Technology, Rockingham, VT) and focused onto the photodiode array with a custom-designed optical macroscope containing a high numerical aperture lens and a dichroic mirror, allowing the delivery of the excitation light and collection of the emitted light from the preparation along the same optical path;

13 Akar et. al. Conduction Slowing and Arrhythmogenesis in Heart Failure 3 thereby, maximizing signal-to-noise characteristics of recorded optical signals, and minimizing heterogeneities of light intensity across the mapping field. Conduction Velocity Measurements Epicardial and endocardial CV were measured by recording optical action potentials from the epicardial then endocardial surfaces of the wedge preparation during steady-state pacing (S1 1000ms) and premature stimulation (S2) delivered from the same electrode over a wide range of coupling intervals (S1S2 800ms to refractoriness or VT). Activation times were measured directly from all optical action potentials using previously validated algorithms. 3 Membrane depolarization was defined as the time of maximum change in fluorescence (df/dt max ) at each site and activation isochrone maps were constructed. To minimize artificial increases in CV resulting from simultaneous capture of tissue near the site of stimulation, the stimulus strength was set to just above the pacing threshold (at 1ms pulse duration), and was periodically verified throughout the experiment. To assess changes in the anisotropy of myocardial conduction, the unipolar pacing electrode was placed in the imaging window, and CV was calculated along the fast and slow axes of propagation using validated algorithms as previously described. 3,5 Western Blotting Epicardial (<1mm) and endocardial (<1mm) tissue layers were dissected from LV and RV myocardium adjacent to the optically mapped wedges. Proteins were prepared as previously described. 1 All samples were run in duplicate or triplicate on 12% Tris- HCl precast gels (Bio-Rad, Hercules, CA) in 25mM Tris, 192mM glycine and 0.1% (wt) SDS running buffer. A standard control sample was run on all gels to allow for comparisons across gels. Primary antibody incubations were performed overnight at 4 C as previously described 6 using commercially available antibodies to measure total

14 Akar et. al. Conduction Slowing and Arrhythmogenesis in Heart Failure 4 Cx43 (Chemicon Inc, Temeculda, CA; 1:1000 dilution), the non-phosphorylated component of Cx43 (Zymed Inc, San Francisco, CA 1:1000), and Cx45 expression (gift from Dr. Jeffery E. Saffitz, MD, PhD, Washington University, St. Louis, MO; 1:1000 dilution). Secondary horseradish peroxidase-conjugated antibodies were obtained from Jackson ImmunoResearch (West Grove, PA). Membranes were exposed and developed using ECL chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ) according to manufacturer s instructions. Audoradiograms were scanned and band densities quantified using the ImageQuant software package (Molecular Dynamics, Sunnyvale, CA). Cx43 band densities were normalized to the density of Calsequestrin (CSQ) to correct for variations in protein loading. Band intensities were expressed quantitatively as arbitrary optical density units, which correspond to the laser densitometric Cx protein band intensity after background subtraction divided by the CSQ signal intensity for the same sample. Microscopy The cellular distribution of Cx43 in tissue isolated from normal and failing hearts was determined using confocal immunofluorescence microscopy. Frozen tissue sections were thawed, fixed in a 4% paraformaldehyde solution for 5min, and incubated in blocking solution (10% BSA with 0.075% Saponin in PBS) overnight at 4 C. Sections were then probed with primary antibodies against Cx43 (Chemicon Intl., Temeculda, CA; 1:50,000 dilution) and cadherin (Zymed; 1:200 dilution) overnight at 4 C, rinsed with PBS, and probed with FITC- and rhodamine-conjugated secondary antibodies for 2h at room temperature. Imaging was performed on a Zeiss Axiovert confocal microscope with excitation and emission filtering at 488±10nm and 525±50nm, respectively (FITC) and 568±10nm and 600±45nm, respectively (rhodamine), and processed using the Metamorph software package (Universal Imaging Corp., Downingtown, PA). Cx43