Changes in Connexin Expression and the Atrial Fibrillation Substrate in Congestive Heart Failure
Rationale: Although connexin changes are important for the ventricular arrhythmic substrate in congestive heart failure (CHF), connexin alterations during CHF-related atrial arrhythmogenic remodeling have received limited attention.
Objective: To analyze connexin changes and their potential contribution to the atrial fibrillation (AF) substrate during the development and reversal of CHF.
Methods and Results: Three groups of dogs were studied: CHF induced by 2-week ventricular tachypacing (240 bpm, n=15); CHF dogs allowed a 4-week nonpaced recovery interval after 2-week tachypacing (n=16); and nonpaced sham controls (n=19). Left ventricular (LV) end-diastolic pressure and atrial refractory periods increased with CHF and normalized on CHF recovery. CHF caused abnormalities in atrial conduction indexes and increased the duration of burst pacing-induced AF (DAF, from 22±7 seconds in control to 1100±171 seconds, P<0.001). CHF did not significantly alter overall atrial connexin (Cx)40 and Cx43 mRNA and protein expression levels, but produced Cx43 dephosphorylation, increased Cx40/Cx43 protein expression ratio and caused Cx43 redistribution toward transverse cell-boundaries. All of the connexin-alterations reversed on CHF recovery, but CHF-induced conduction abnormalities and increased DAF (884±220 seconds, P<0.001 versus control) remained. The atrial fibrous tissue content increased from 3.6±0.7% in control to 14.7±1.5% and 13.3±2.3% in CHF and CHF recovery, respectively (both P<0.01 versus control), with transversely running zones of fibrosis physically separating longitudinally directed muscle bundles. In an ionically based action potential/tissue model, fibrosis was able to account for conduction abnormalities associated with CHF and recovery.
Conclusions: CHF causes atrial connexin changes, but these are not essential for CHF-related conduction disturbances and AF promotion, which are rather related primarily to fibrotic interruption of muscle bundle continuity.
Congestive heart failure (CHF) predisposes to atrial fibrillation (AF), although the underlying mechanisms remain incompletely understood.1 Ventricular tachypacing produces a clinically relevant animal model of CHF.2 The atria of dogs with ventricular tachypacing-induced CHF are characterized by structural remodeling, conduction abnormalities and the ability to sustain AF.3
Myocardial electric continuity is assured by gap junctions, cell-to-cell connections that maintain low-resistance intercellular coupling via specialized hemichannel subunit proteins called connexins. Connexin (Cx)43 and Cx40 are the principle atrial gap junctional subunits4; abnormalities in their expression and localization are commonly observed in patients and experimental animals with AF.5 Phosphorylation of Cx43 can regulate channel assembly,6 degradation,7 and conductance.8 In the ventricles, CHF produces hypophosphorylation of Cx43 and redistribution to lateral cell membranes, associated with proarrhythmic conduction slowing.9,10 Little is known, however, about connexin changes during CHF-related atrial remodeling and their role in AF maintenance. The present study was designed to assess the changes in atrial connexin expression caused by tachypacing-induced CHF in the dog.
In initial experiments, we noted significant changes in connexin phosphorylation and sought to understand their role in AF. We previously noted that the cessation of ventricular tachypacing, which is followed by the reversal of CHF, dissociates atrial size and function changes from structural remodeling and AF sustainability.11,12 Both structural (particularly tissue fibrosis) and connexin remodeling could contribute to CHF-associated AF. We therefore exploited CHF reversal to assess the reversibility of atrial connexin alterations and evaluate their contribution to CHF-related conduction disturbances and AF maintenance.
An expanded Methods section is available in the Online Data Supplement at http://circres.ahajournals.org.
Animal-handling procedures followed National Institutes of Health guidelines. Animals were prepared and studied as described previously.11,12 Forty-nine mongrel dogs (18 to 34 kg) were instrumented with a right ventricular tachypacemaker. Dogs were assigned to 3 groups: (1) pacemaker inactive sham controls (CTL group, n=19); (2) 2-week ventricular tachypacing at 240 bpm to induce CHF (CHF group, n=15); (3) 2-week ventricular tachypacing followed by 4-week recovery (REC group, n=16).
Atrial effective refractory period (ERP) was measured with 10 basic (S1) stimuli. ERP was measured at multiple basic cycle lengths (BCLs) in the left atrial (LA) appendage of all dogs and at 7 additional sites (BCL, 300 ms) in 5 dogs per group. AF was induced by burst pacing and mean AF duration estimated based on 10 inductions. AF ≥30 minutes was considered sustained and was cardioverted. Five plastic arrays containing a total of 240 bipolar electrodes were sewn to the atrial epicardial surface.3 Phase-delay and conduction velocity analyses were performed in 5 dogs per group as previously reported.3,11,12 LA tissue samples were taken from these dogs for histological analysis. Other tissue samples were snap-frozen and stored at −80°C for biochemistry and confocal microscopy.
Blocks were sectioned along longitudinal and transverse planes. Sections were cut at room temperature and stained with Masson’s trichrome (n=5 dogs per group). Microscopic images were captured with a Zeiss Axioplan-2 Imaging microscope. Tiled micrographs were obtained with a ×20 objective and mechanical stage using the MosaicX tile feature of Zeiss Axiovision 4.5 software.13 Fibrous tissue content was quantified as percentage surface area, excluding blood vessel–containing regions. Images were analyzed blinded to group assignment.
Total RNA was isolated from CTL, CHF, and REC atrial tissues (n=6 to 10 dogs per group), then DNase-treated, quantified, and quality-controlled as described previously.14 Real-time RT-PCR was performed with 6-carboxy-fluorescein–labeled fluorogenic TaqMan primers, probes, and universal master mix. Fluorescence signals were detected in duplicate, normalized to 18S ribosomal RNA and quantified. The expression of 18S ribosomal RNA was similar among experimental groups (CTL 0.83±0.09; CHF 1.02±0.08; REC 0.77±0.19; P=NS).
Membrane proteins were extracted and processed as described previously.14,15 Phosphorylated and dephosphorylated forms of Cx43 (p-Cx43 and np-Cx43, respectively) were detected with a mouse monoclonal antipan-Cx43 antibody (1/1000, Chemicon) based on molecular masses. Serine 368–phosphorylated Cx43 (368p-Cx43) was specifically detected with a rabbit polyclonal antibody (1/1000, Cell Signaling). Cx40 and GAPDH were detected with rabbit polyclonal anti-Cx40 (1/1000, Chemicon) and mouse monoclonal anti-GAPDH (1/10 000, RDI) primary antibodies. Following application of primary antibodies, membranes were incubated with either antimouse (1/10 000, Santa Cruz Biotechnology) or anti-rabbit (1/10 000, Jackson ImmunoLabs) horseradish peroxidase–conjugated secondary antibodies. Signals were detected by chemiluminescence and quantified by video densitometry. Band intensities are expressed relative to GAPDH. GAPDH expression was consistent among groups (CTL, 0.86±0.04; CHF, 0.81±0.04; REC, 0.90±0.02; P=NS).
Immunofluorescence and Confocal Imaging
Serial 14-μm cryosections prepared from CTL, CHF and REC were fixed as previously described.14 Slides were incubated with mouse anti–pan-Cx43, rabbit anti–phospho368-Cx43, rabbit anti-Cx40, or rabbit anti–pan-cadherin primary antibodies (all 1/200), followed by AF555-conjugated phalloidin (1/200), donkey anti-mouse AF488, and goat anti-rabbit AF647 secondary antibodies (both 1/600), with parallel negative-control studies omitting primary antibodies. Slides were imaged in Z-series every 0.25 μm with a Zeiss LSM-510 inverted confocal microscope. Deconvolved Z-series maximum projections were used to create 3D reconstructions. Tissue analysis was performed at equal magnifications over equivalent tissue areas and thickness, excluding vessel-containing regions, blinded to group assignment.
Connexin lateralization was analyzed using a method based on quantification of the angle formed between the local longitudinal cell axis and the main axis of individual connexin clusters. The ratio between transverse and cell end connexin clusters was used to indicate lateralization (for details, see the Online Data Supplement).
A novel method to quantify fibrosis in the transverse direction relative to fiber orientation was developed based on the probability density function of angles between the fibrosis cluster major axis and fiber orientation. For details, see the Online Data Supplement.
Following a left lateral thoracotomy and heparin administration (5000 U, IM), atria were excised and perfused via the circumflex artery with Krebs solution (in mmol/L: 120 NaCl, 4 KCl, 1.2 MgSO4 7H2O, 1.2 KH2PO4, 25 NaHCO3, and 1.2 CaCl2, saturated with 95% O2-5% CO2) at 20 mL/min. Optical recordings were obtained in the LA roof area in the presence of 2,3-butanedione monoxime (15 mmol/L) and di-4-ANEPPS. A charge-coupled device camera (80×80 pixels, RedShirt Imaging) recorded fluorescence at 1 kHz. Optical signals were recorded during 2-Hz electric stimulation. Experiments were performed in CTL and REC hearts to assess the effects of fibrosis on conduction.
After optical mapping, the imaged zones were dissected and stored in formaldehyde for longitudinal tissue sectioning, Masson trichrome staining, and image analysis for fibrous tissue mapping. Maximum |dF/dt| was used to define activation time. Phase analysis was performed with a grid of 20×20 points (every 4×4 points of the charge-coupled device matrix).
Simulations of electric propagation were initially performed on 2D rectangular tissue section reconstructions to compare with extracellular mapping results obtained in vivo. The fibrous tissue distribution patterns corresponding to control, CHF, and recovery cases were derived from longitudinally oriented histological sections obtained in individual canine LA images with a color-based segmentation of Masson trichrome–stained images. The simulated tissue measured ≈9.5×11.5 mm. Propagation was initiated by applying 2-ms square pulses of 180 μA at a cycle length of 300 ms on a 1-mm2 surface at the bottom right corner of the tissue. Phase analysis on a square 56-pseudoelectrode lattice was obtained with activation times at pseudoelectrode points separated by ≈1.5 mm to approach the experimental conditions.
Additional simulations were based on fibrous tissue distribution patterns corresponding to control and recovery cases derived from longitudinally oriented histological sections obtained following optical mapping experiments.
The mathematical model was based on the 2D reaction–diffusion equation: equation
where V is the transmembrane potential, Iion the total ionic current, a the cell radius (a=5 μm), ri the tissue resistivity (ri=75 Ohm-cm), and Cm the cell capacitance (Cm=100 μF).
where INa, IK1, Ito, IKur,d, IKr, IKs, ICa, IClCa, IpCa, INaCa, INaK, Ib,Na, Ib,Ca, Ib,Cl indicate Na+, inward-rectifier K+, transient-outward K+, ultrarapid, rapid, and slow delayed-rectifier K+, L-type Ca2+, Ca2+-dependent Cl−, Ca2+ pump, Na+/Ca2+ exchange, Na+/K+-ATPase, and background Na+, Ca2+, and Cl− currents, respectively. Fibrosis was modeled by replacing active cells of the discretized 2D substrate by holes with no-flux boundary conditions. Simulations were performed with an operator-splitting and finite element method17 and 12×12-μm2 spatial discretization. Numeric integration was obtained by forward-Euler difference with a 5-μs time step. Phase analysis18 on a square pseudoelectrode lattice was obtained with activation times at pseudoelectrode points separated by ≈1.0 mm to approach the optical mapping experimental conditions.
Atrial conduction properties were analyzed as previously described.3,11,12,18 Activation time delays between each electrode site and neighboring sites were normalized to interelectrode distance and the largest values taken to reflect the activation phase relation at that site. Values were binned to create phase-delay histograms. The phase-delay range between the 5% lowest and 5% largest values (P5–95) represents the difference between fastest- and slowest-conducting zones and is increased by regions of slow conduction. The phase-delay range divided by the median value (P5–95/P50) is a conduction heterogeneity index independent of conduction velocity.
Data are presented as means±SEM. Multiple group comparisons were obtained with 1-way ANOVA or 2-way repeated-measures ANOVA as appropriate. When 1-way ANOVAs revealed significant effects, Bonferroni adjusted pairwise comparisons were performed by multiplying probability values by 3. For 2-way repeated-measures ANOVAs, a mixed model using one repeated main factor was applied. In the case of a significant interaction between factors, contrasts based on the global model were used to compare groups (CTL, CHF, and REC) within the other main factor. Normality of distribution was verified. A 2-tailed probability value of <0.05 was considered statistically significant.
At open-chest study, pulmonary congestion and pericardial effusions were evident in all CHF dogs but no control or REC dogs. Overall group characteristics and hemodynamic data are presented in Online Table I. Ventricular and arterial systolic and arterial diastolic pressures were reduced in CHF dogs, whereas LV end-diastolic, LA, and RA pressures were increased. Hemodynamic indexes among REC dogs were significantly different from CHF dog values and were statistically indistinguishable from control animals.
In Vivo Electrophysiology
Online Figures I and II show electrophysiological properties at open-chest study. CHF increased LA ERP at all BCLs (Online Figure I, A), a change that reversed with recovery. Group was also a significant determinant of regional atrial ERP, and ERP lengthening caused by CHF was regionally variable (Online Figure I, B). Regional ERPs in recovery dogs were significantly shorter than CHF dogs and not significantly different from control.
Mean AF duration (Figure 1A) increased in CHF dogs (1100±171 versus 22±7 seconds in CTL) and remained prolonged in REC dogs (884±220 seconds), despite full hemodynamic and ERP recovery. Phase-delay analyses revealed no significant change in the shortest activation delays (P5; Online Figure II, A) or the median phase-delay (P50; Online Figure II, B), reflecting overall conduction speed. The phase-delay range reflecting slow conduction zones (P5–95; Figure 1B), and the conduction heterogeneity index (P5–95/P50; Figure 1C) were both significantly greater in CHF and REC dogs at all BCLs compared to controls, with no difference between CHF and REC groups. Overall conduction velocities per se were not affected by CHF (Online Figure III), consistent with previous observations.3
Connexin Gene Expression
Cx40 and Cx43 mRNA expression data are presented in Figure 2A and 2B, respectively. Neither Cx40 nor Cx43 mRNA expression differed significantly among groups.
Connexin Protein Expression and Phosphorylation
Figure 3 illustrates the results of Western blot analysis of connexin protein expression. Cx43 was detected by an antibody that reacts with both the phosphorylated (p-Cx43, identified with more slowly migrating, higher-molecular-mass bands9,19,20) and nonphosphorylated (np-Cx43, faster-migrating bands) isoforms and also by an antibody specific to Ser368-phosphorylated Cx43 (368p-Cx43). Figure 3A shows representative immunoblots for Cx40, total Cx43, 368p-Cx43, and GAPDH (from top to bottom, respectively), with mean band intensity data shown in Figure 3B through 3G. Total Cx40 (Figure 3B) and Cx43 (Figure 3C, sum of lower and higher-molecular-mass bands) expression were not significantly different among groups. Expression of the higher-molecular-mass band was significantly reduced and the lower-molecular-mass nonphosphorylated form significantly increased in CHF, with full reversal in REC dogs (Figure 3D). Consistent with the changes in different molecular-mass bands, directly detected 368p-Cx43 expression was reduced by ≈73% in CHF and returned to control levels in REC dogs (Figure 3E).
Phosphorylation state and connexin subunit stoichiometry may alter gap junctional communication.5 Comparison of upper (p-Cx43) to lower (np-Cx43) band intensities within individual samples allowed for determination of Cx43 phosphorylation state (Figure 3F). The ratio of p-Cx43/np-Cx43 was reduced ≈80% by CHF and returned to control values in REC. Although absolute expression values cannot be accurately compared between connexin-isoforms detected by different antibodies (because their respective antibodies probe at different, protein-specific epitopes, and the relative affinities of different antibodies differ), relative expression changes can be determined by comparing Cx40 with total Cx43 within each sample. Cx40/total Cx43 protein ratio was increased ≈35% by CHF (Figure 3G), likely because of small but consistent concomitant increases in Cx40 and decreases in total Cx43 expression, which were not sufficient to achieve statistical significance over background noise but which emerged as statistically significant when analyzed as a relative expression ratio. CHF-related changes in both phosphorylation state and Cx40/total Cx43 protein ratio were completely reversed in REC dogs.
Immunofluorescence and confocal imaging of control, CHF and REC atria allowed visualization of Cx40, total Cx43, and 368p-Cx43 expression and distribution (Figure 4). Costaining with the F-actin intracellular marker phalloidin permitted identification of cell borders. Shown in Figure 4A through 4I are representative 3D reconstructions of the end-to-end intercalated disc region between paired adjoining cardiomyocytes (bottom of each image) and front face views of the gap junction complexes connecting them (top of image). A decrease in 368p-Cx43 with CHF that reverses following recovery from CHF is clearly apparent, with no obvious alterations in total connexins. Quantitative analysis of the mean tissue area data confirms the lack of change in Cx40 and T-Cx43, along with significantly reduced 368p-Cx43 expression (Figure 4J through 4L) in CHF that returned to values not significantly different from control with recovery. Additional images of connexin staining in front face maximum projection views of several individual gap junctions are provided in Online Figure IV.
In normal atrial and ventricular myocardium, Cx43 is primarily localized at end-to-end junctions of adjacent cardiomyocytes. Previous studies have reported redistribution of connexins from cell ends to lateral margins in CHF.9,10,19,20 Figure 5 shows longitudinally oriented sections with detection by antibodies to Cx40 (Figure 5A through 5C) and total Cx43 (Figure 5D through 5F) and actin myofilaments to assess potential connexin redistribution. Laterally oriented connexins were rare in control conditions (Figure 5A and 5D), consistent with a well-recognized prominent atrial anisotropy ratio of ≈10:1, reflecting a paucity of transverse connections.21 Linear staining of Cx43 at the lateral cell margins greatly increased with CHF (Figure 5E), whereas no apparent lateralization is seen for Cx40 (Figure 5B). With recovery from CHF, the predominantly cell end Cx43 distribution pattern seen in control returned (Figure 5F). Quantitative analysis showed no change in Cx40 lateralization under the conditions studied (Figure 5G) but a statistically significant increase in Cx43 lateralization with CHF that reversed with REC (Figure 5H). Online Figure V illustrates Cx43 lateralization with CHF in images coimmunostained for the gap junction marker cadherin.
Figure 6 shows representative examples of atrial histopathology in control (Figure 6A), CHF (Figure 6B), and REC (Figure 6C) dogs. Control dogs displayed grossly normal atria with small amounts of interstitial fibrous tissue (Figure 6D), in contrast to the substantial interstitial fibrosis in CHF and REC dogs. To assess the potential role of fibrosis in CHF-related conduction abnormalities, we examined the relationship between fibrous tissue deposition and tissue orientation. If interstitial fibrosis simply proceeded along lateral muscle bundle and cell boundaries, typical of reactive fibrosis,22 fibrotic changes would not be expected to interfere importantly with longitudinal conduction. If, in contrast, there is reparative fibrosis that replaces zones of dead cardiomyocytes within muscle bundles by collagen, interruption of longitudinal cell–cell communication and important disruptions in conduction might be expected. We closely examined tissue sections containing longitudinally oriented muscle bundles to identify the distribution of fibrosis relative to cardiomyocyte-strands. Both CHF and REC dog samples showed transversely oriented fibrosis interrupting longitudinally running fibers (Figure 6B and 6C, arrows). Quantification of transversely distributed fibrosis confirmed a statistically significant, ≈10-fold increase of transversely oriented fibrous tissue in CHF and REC versus CTL (Figure 6E).
Mathematical Modeling and Optical Mapping
Connexin-remodeling results indicate that connexin changes cannot explain conduction alterations in REC dogs, because they completely reversed in the recovered condition. This finding points to tissue fibrosis, which did not change with recovery, as a strong candidate to explain abnormal conduction. We used mathematical simulation to assess whether the speculated role of fibrosis is plausible. The results of conduction simulations are shown in Figure 7. Fibrous tissue is represented by blue dots on the 2D atrial grid. Under control conditions (Figure 7A), conduction was smooth, with smooth lines of wave-front propagation. In contrast, with CHF and REC conditions, conduction became much more heterogeneous, with wave-fronts having less smooth frontal boundaries and less discrete isochrones. The results of phase-delay analysis are provided in Online Table II. Consistent with experimental data, P5–95 and the heterogeneity index (P5–95/P50) were increased relative to control in both CHF and REC conditions. These results suggest that tissue fibrosis is sufficient to account for CHF-related changes in conduction indices.
As a final test of the role of fibrosis in conduction slowing, we performed optical mapping experiments in additional CTL and REC dog coronary artery–perfused atrial preparations, followed by mathematical simulations of electric propagation with fibrosis distributions derived obtained from the same preparations. Examples of atrial activation maps from one preparation of each type are shown in Figure 8A and 8B. Corresponding simulations are shown in Figure 8C and 8D, with fibrous tissue represented by white pixels on the atrial grid. There was generally good agreement between optical mapping results and mathematical model simulations. The results of phase-delay analysis are presented in Online Figure VI. Both P5–95 and the heterogeneity index (P5–95/P50) were significantly increased in REC compared to CTL, in both experimental results and mathematical simulations.
We found that CHF resulting from 2 weeks of ventricular tachypacing causes Cx43 hypophosphorylation and lateralization, along with increased Cx40/Cx43 expression ratios. These alterations are accompanied by disturbances in local atrial conduction and a substrate for AF maintenance. We then examined corresponding features after recovery from CHF and noted that whereas connexin abnormalities reverse completely on resolution of CHF, atrial conduction abnormalities and AF-maintaining substrates show no significant improvement. These findings suggest that connexin alterations are not essential for atrial conduction disturbances and AF promotion associated with CHF, rather implicating other factors like atrial fibrosis that show no recovery with CHF reversal. In both CHF and REC dogs, atrial fibrosis interrupts cardiac muscle bundles in their longitudinal orientation, and a mathematical model of the effects of fibrosis on impulse propagation provided results consistent with experimentally measured conduction abnormalities.
Relationship to Previous Studies of Gap Junctional Remodeling in CHF
Several studies have described abnormalities in the expression, distribution, and regulation of ventricular connexins in CHF. Absolute Cx43 expression is generally reduced in CHF ventricles,9,20,23–25 likely related to the activation of the mitogen-activated protein kinase c-Jun N-terminal kinase.26 Recent work suggests an important role for defects in Cx43 phosphorylation in CHF-induced ventricular cardiomyocyte uncoupling,9,10,20 thought to be attributable to increased dephosphorylating activity of protein phosphatase-2A colocalized with Cx43.20 Connexin dephosphorylation plays significant roles in targeting connexins to intercalated disks and in regulating connexin conductance.8,10 One study showed ventricular Cx40 upregulation in CHF,23 possibly as a compensation for Cx43 downregulation; however, the functional importance of this alteration is uncertain in view of low level ventricular Cx40 expression.
Clinical and experimental studies of gap junctional remodeling in the atria have produced highly discrepant results.5,27 The variable findings may relate to technical issues, differences in models, and species studied in experimental work, as well as population-related factors such as underlying heart disease, duration and type of AF, and concomitant drug therapy in clinical studies. The most consistent findings are hypophosphorylation and increased heterogeneity in connexin-distribution.5,19,27 In the present work, we noted atrial Cx43 dephosphorylation and lateralization in CHF. In most previous studies of subjects with AF, the relative participation of AF versus underlying heart disease was unclear. Our dogs had CHF but not AF, so we were able to analyze the effects of CHF on the AF substrate without contamination from AF-induced remodeling. Like Rucker-Martin et al, who studied atrial remodeling in a postmyocardial infarction rats,19 we observed clear connexin dephosphorylation. Rucker-Martin et al did not measure conduction indices in their rats but reported complete atrial-cardiomyocyte uncoupling (as indicated by Lucifer-yellow dye transfer) in CHF,19 which is difficult to reconcile with maintained intraatrial conduction (albeit with increased PR intervals). There is evidence for dissociation between alterations in large-solute transmission and electric communication through connexins.28
We found no change in conduction abnormalities with recovery from CHF, despite full return of connexin phosphorylation. This observation suggests that connexin dephosphorylation did not contribute significantly to CHF-related atrial conduction-abnormalities and agrees with previous findings indicating that AF-induced atrial connexin remodeling does not produce detectable conduction changes.29 Ausma et al observed that AF promotion persists beyond the period required for reversal of connexin expression changes in the goat-AF model,30 consistent with the dissociation we noted between AF promotion and connexin changes during recovery. Like our findings, a recent study reported that AF associated with severe CHF showed an increased Cx40/Cx43 ratio.31 The same study found elevated Cx43 expression in AF associated with mild CHF but decreased Cx43 in AF with severe CHF, reiterating the notion that changes may depend on the degree and/or type of underlying pathology.32
Despite many studies of connexin changes in animal and human AF paradigms, their role in the arrhythmic substrate remains unclear.5,27 One of the most consistent findings in AF patients is increased connexin lateralization.33,34 Few studies have examined connexin phosphorylation in AF. Cx43 phosphorylation is an important regulatory mechanism for channel assembly,6 degradation,7 and conductance.8 Our finding of reduced Cx43 phosphorylation associated with increased connexin lateralization agrees with observations at the ventricular level in CHF9,10,20 and at the atrial level in rats with atrial dilation postmyocardial infarction.19 Our study, the first to our knowledge to evaluate the relationship between atrial connexin phosphorylation changes and functional conduction abnormalities/AF substrates, suggests that connexin changes are not required to produce CHF-related conduction disturbances and AF maintenance. This conclusion is consistent with recent studies in which the gap junction conductance-enhancing peptide rotigaptide failed to improve atrial conduction disturbances or AF maintenance in CHF dogs.35,36 The lack of a demonstrable contribution of connexin dephosphorylation/lateralization to conduction abnormalities in our CHF/REC atria contrasts with the evidence for a significant role in CHF-induced ventricular conduction abnormalities.9,10,20 This discrepancy may be related to greater fibrosis at the atrial level,37 which may obscure the contribution of connexin43 dephosphorylation/lateralization. These findings underscore the need for further studies of the interactions between fibrosis and connexin changes in the control of conduction, as well as analyses of the functional consequences of connexin changes in various AF paradigms.
There is extensive evidence for a role of atrial fibrosis in the AF substrate.3,11,12,38 A recent study demonstrated the importance of posterior LA fibrosis in fibrillation wave dynamics of CHF-related AF.39 In the present work, we add to the evolving information regarding the role of atrial fibrosis in AF by providing time course evidence, structural observations and mathematical modeling findings to support the primacy of fibrosis in AF-related conduction disturbances in CHF. Our results lend support to the targeting of fibrosis development for AF prevention, an approach that is attracting considerable interest.22,38
CHF can result from many etiologies and can show various forms of pathophysiological evolution. We used a specific animal model that mimics clinical tachycardiomyopathies, but caution is necessary relating these findings to other forms of CHF. In addition, fibrosis likely plays a varying role in different pathological forms of AF and is completely absent in some models.3 The connexin changes in the present model do not appear to contribute markedly to the AF substrate; however, connexin abnormalities may play a greater role in other AF-promoting pathologies. Furthermore, any animal model of AF is clearly oversimplified compared to complex clinical pathophysiology.40
We studied Cx43 phosphorylation state by examining the ratio of higher- to lower-molecular-weight bands believed to represent phosphorylated and nonphosphorylated-Cx43 respectively, as well as with a Ser368-phosphorylated Cx43–specific antibody. However, in addition to this important site, there are other potentially significant phosphorylation sites on Cx43,41 and we cannot be certain that similar changes would have occurred at other sites.
The quantitative relationship between fibrosis and AF remains to be elucidated, as does the effect of the spatial distribution of fibrosis on atrial conduction and AF-susceptibility. The findings of this study relate to the substrate for AF maintenance and do not bear on other determinants of AF occurrence, such as neurohormonal tone and atrial ectopic activity.
We thank Mariève Cossette, Jacynthe Laliberté, Nathalie L’Heureux, and Chantal St-Cyr for technical assistance; Gernot Planck for help with mathematical modeling; and France Thériault for secretarial support.
Sources of Funding
This work was supported by the Canadian Institutes of Health Research (CIHR; award MGP 6957), the Quebec Heart and Stroke Foundation, the Fondation Leducq (European-North American Atrial Fibrillation Research Alliance, ENAFRA; award 07/CVD/03), and the Mathematics of Information Technology and Complex Systems (MITACS) network of centers of excellence. B.B. received a CIHR MD/PhD studentship. P.C. holds a Young Investigator Award from the “Fonds de Recherche en Santé du Quebec.” G.M. holds a research fellowship from the Heart and Stroke Foundation of Canada and K.N. held research fellowships from Nihon Kohden/St. Jude Medical and Japan Heart Foundation/The Japanese Society of Electrocardiology.
↵*These authors contributed equally to this work and should be considered to share first authorship.
Original received July 16, 2008; revision received October 14, 2009; accepted October 15, 2009.
Moe GW, Armstrong P. Pacing-induced heart failure: a model to study the mechanism of disease progression and novel therapy in heart failure. Cardiovasc Res. 1999; 42: 591–599.
Li D, Fareh S, Leung TK, Nattel S. Promotion of atrial fibrillation by heart failure in dogs: atrial remodeling of a different sort. Circulation. 1999; 100: 87–95.
Gros D, Jarry-Guichard T, Ten V, I, de Maziere A, van Kempen MJ, Davoust J, Briand JP, Moorman AF, Jongsma HJ. Restricted distribution of connexin40, a gap junctional protein, in mammalian heart. Circ Res. 1994; 74: 839–851.
Akar FG, Spragg DD, Tunin RS, Kass DA, Tomaselli GF. Mechanisms underlying conduction slowing and arrhythmogenesis in nonischemic dilated cardiomyopathy. Circ Res. 2004; 95: 717–725.
Akar FG, Nass RD, Hahn S, Cingolani E, Shah M, Hesketh GG, DiSilvestre D, Tunin RS, Kass DA, Tomaselli GF. Dynamic changes in conduction velocity and gap junction properties during development of pacing-induced heart failure. Am J Physiol Heart Circ Physiol. 2007; 293: H1223–H1230.
Shinagawa K, Shi YF, Tardif JC, Leung TK, Nattel S. Dynamic nature of atrial fibrillation substrate during development and reversal of heart failure in dogs. Circulation. 2002; 105: 2672–2678.
Cha TJ, Ehrlich JR, Zhang L, Shi YF, Tardif JC, Leung TK, Nattel S. Dissociation between ionic remodeling and ability to sustain atrial fibrillation during recovery from experimental congestive heart failure. Circulation. 2004; 109: 412–418.
Burstein B, Qi XY, Yeh YH, Calderone A, Nattel S. Atrial cardiomyocyte tachycardia alters cardiac fibroblast function: a novel consideration in atrial remodeling. Cardiovasc Res. 2007; 76: 442–452.
Godreau D, Vranckx R, Maguy A, Rucker-Martin C, Goyenvalle C, Abdelshafy S, Tessier S, Couetil JP, Hatem SN. Expression, regulation and role of the MAGUK protein SAP-97 in human atrial myocardium. Cardiovasc Res. 2002; 56: 433–442.
Ramirez RJ, Nattel S, Courtemanche M. Mathematical analysis of canine atrial action potentials: rate, regional factors, and electrical remodeling. Am J Physiol Heart Circ Physiol. 2000; 279: H1767–H1785.
Rucker-Martin C, Milliez P, Tan S, Decrouy X, Recouvreur M, Vranckx R, Delcayre C, Renaud JF, Dunia I, Segretain D, Hatem SN. Chronic hemodynamic overload of the atria is an important factor for gap junction remodeling in human and rat hearts. Cardiovasc Res. 2006; 72: 69–79.
Ai X, Pogwizd SM. Connexin 43 downregulation and dephosphorylation in nonischemic heart failure is associated with enhanced colocalized protein phosphatase type 2A. Circ Res. 2005; 96: 54–63.
Kostin S, Dammer S, Hein S, Klovekorn WP, Bauer EP, Schaper J. Connexin 43 expression and distribution in compensated and decompensated cardiac hypertrophy in patients with aortic stenosis. Cardiovasc Res. 2004; 62: 426–436.
Poelzing S, Rosenbaum DS. Altered connexin43 expression produces arrhythmia substrate in heart failure. Am J Physiol Heart Circ Physiol. 2004; 287: H1762–H1770.
Petrich BG, Gong X, Lerner DL, Wang X, Brown JH, Saffitz JE, Wang Y. c-Jun N-terminal kinase activation mediates downregulation of connexin43 in cardiomyocytes. Circ Res. 2002; 91: 640–647.
Nattel S, Maguy A, Le Bouter S, Yeh YH. Arrhythmogenic ion-channel remodeling in the heart: heart failure, myocardial infarction, and atrial fibrillation. Physiol Rev. 2007; 87: 425–456.
Bao X, Lee SC, Reuss L, Altenberg GA. Change in permeant size selectivity by phosphorylation of connexin 43 gap-junctional hemichannels by PKC. Proc Natl Acad Sci U S A. 2007; 104: 4919–4924.
van der Velden HM, Ausma J, Rook MB, Hellemons AJ, van Veen TA, Allessie MA, Jongsma HJ. Gap junctional remodeling in relation to stabilization of atrial fibrillation in the goat. Cardiovasc Res. 2000; 46: 476–486.
Ausma J, van der Velden HM, Lenders MH, van Ankeren EP, Jongsma HJ, Ramaekers FC, Borgers M, Allessie MA. Reverse structural and gap-junctional remodeling after prolonged atrial fibrillation in the goat. Circulation. 2003; 107: 2051–2058.
Wetzel U, Boldt A, Lauschke J, Weigl J, Schirdewahn P, Dorszewski A, Doll N, Hindricks G, Dhein S, Kottkamp H. Expression of connexins 40 and 43 in human left atrium in atrial fibrillation of different aetiologies. Heart. 2005; 91: 166–170.
Kostin S, Klein G, Szalay Z, Hein S, Bauer EP, Schaper J. Structural correlate of atrial fibrillation in human patients. Cardiovasc Res. 2002; 54: 361–379.
Guerra JM, Everett TH IV, Lee KW, Wilson E, Olgin JE. Effects of the gap junction modifier rotigaptide (ZP123) on atrial conduction and vulnerability to atrial fibrillation. Circulation. 2006; 114: 110–118.
Shiroshita-Takeshita A, Sakabe M, Haugan K, Hennan JK, Nattel S. Model-dependent effects of the gap junction conduction-enhancing antiarrhythmic peptide rotigaptide (ZP123) on experimental atrial fibrillation in dogs. Circulation. 2007; 115: 310–318.
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: 236–244.
Tanaka K, Zlochiver S, Vikstrom KL, Yamazaki M, Moreno J, Klos M, Zaitsev AV, Vaidyanathan R, Auerbach DS, Landas S, Guiraudon G, Jalife J, Berenfeld O, Kalifa J. Spatial distribution of fibrosis governs fibrillation wave dynamics in the posterior left atrium during heart failure. Circ Res. 2007; 101: 839–847.