Connexin 43 Downregulation and Dephosphorylation in Nonischemic Heart Failure Is Associated With Enhanced Colocalized Protein Phosphatase Ty
http://www.100md.com
《循环研究杂志》
From the Department of Medicine, University of Illinois at Chicago.
In nonischemic heart failure (HF), ventricular tachycardia initiates by a nonreentrant mechanism, but there is altered conduction (that could lead to re-entry) that could arise from changes in gap junctional proteins, especially connexin43 (Cx43). We studied Cx43 expression and phosphorylation state in the left ventricle (LV) from an arrhythmogenic rabbit model of nonischemic HF and from patients with HF attributable to idiopathic dilated cardiomyopathy. We also investigated the role of protein phosphatases that dephosphorylate Cx43—PP1 and PP2A. In HF rabbit LV, Cx43 mRNA and total protein were decreased by 29% and 34%, respectively (P<0.05 and P<0.001). In controls, Cx43 was primarily in the phosphorylated state, but with HF there was a 64% increase in nonphosphorylated Cx43 (Cx43-NP, normalized to total Cx43; P<0.05). Similar results were noted in HF rabbit myocytes (P<0.05) and in human idiopathic dilated cardiomyopathy LV (P<0.05). We found that PP1 and PP2A colocalized with Cx43 in rabbit LV. With HF, the level of colocalized PP2A increased >2.5-fold (P<0.002), whereas colocalized PP1 was unchanged. We also found intercellular coupling (assessed by Lucifer Yellow dye transfer) was markedly reduced in HF. However, okadaic acid (10 nmol/L) reduced the amount of Cx43-NP and significantly improved cell coupling in HF. Thus, in nonischemic HF in rabbits and humans, there is a decrease in both Cx43 expression and phosphorylation that contributes to uncoupling. Increased levels of PP2A that colocalize with Cx43 can underlie enhanced levels of Cx43-NP in HF. Modulation of Cx43 phosphorylation may be a potential therapeutic target to improve conduction in HF.
Key Words: gap junctions , phosphorylation , phosphatases , heart failure , arrhythmia
Heart failure (HF), whether nonischemic or ischemic, is associated with a nearly 50% incidence of sudden death, primarily from ventricular tachycardia (VT) degenerating to ventricular fibrillation (VF).1 Whereas VT in nonischemic HF initiates primarily by a nonreentrant mechanism,2 myocardium from patients with idiopathic dilated cardiomyopathy (IDCM) exhibits nonuniform anisotropy, slow conduction, and conduction block3 that could underlie reentry during the transition from VT to VF. Conduction slowing could arise from decreased depolarizing currents and/or decreased gap junctional coupling.4 However, the degree of slow conduction and block in failing myocardium appears to be out of proportion to the changes in active membrane properties.5 Moreover, LV myocytes from an animal model of nonischemic HF exhibit markedly decreased gap junctional conductance.6 Thus, alterations in intercellular coupling involving cardiac gap junctions may underlie slow conduction in nonischemic HF.
Gap junctions are specialized membrane structures consisting of arrays of intercellular channels that directly connect adjacent cells by providing chemical and electrical communication. They are composed of connexins, a multigene family of conserved proteins. The relative amounts, composition, and distribution of these connexins appear to influence the conduction properties of cells.7,8 Connexin43 (Cx43) is the major connexin protein in ventricular myocardium, and downregulation of Cx43 (mRNA and protein) has been demonstrated in some experimental HF models and in the failing human heart.9 Studies in Cx43 heterozygous knockout mice, and in the canine rapid pacing HF model, support the fact that decreased Cx43 expression per se can lead to a slowing of conduction, and that arrhythmogenesis can be further enhanced under pathophysiologic conditions (eg, myocardial ischemia) and when gap junctional alterations are heterogeneous.10–12
Cx43 is a phosphoprotein that can be phosphorylated by a number of kinases13 and dephosphorylated by protein phosphatases such as PP1 and PP2A.14 Most phosphorylation sites are serine residues on the C-terminus, although threonine and tyrosine phosphorylation sites are present to a much lesser extent.13 Whereas the Cx43 phosphorylation states mediated by kinases can have variable effects on gap junctional communication,15,16 dephosphorylation of Cx43 (by phosphatases) has been shown to decrease gap junctional communication—whether assessed in neonatal rat ventricular cell pairs with activation of endogenous phosphatases14 or in perfused whole rat hearts during myocardial ischemia.17 Little is known about whether there is altered Cx43 phosphorylation in HF; however, if so, it could contribute to altered conduction and arrhythmogenesis.18
Total cellular PP1 and PP2A activity is increased in the failing heart.19 Recent studies have shown that Cx43 interacts with a number of proteins (eg, ZO-1 and protein kinase C) to form a protein complex.15,20 It is conceivable that protein phosphatases may also directly associate with Cx43 at gap junctions to modulate its phosphorylation state, as has recently been shown for other macromolecular complexes,21 and that the level of protein phosphatases in direct proximity to Cx43 may be enhanced in HF.
As such, the goals of these studies were to investigate: (1) alteration of Cx43 protein and mRNA expression; (2) change in Cx43 phosphorylation state; (3) possible colocalization of PP1 and PP2A with Cx43; (4) alteration of intercellular coupling; and (5) the contribution of PP1 and PP2A to Cx43 dephosphorylation and altered cell coupling in HF. Studies were performed in LV tissue and myocytes from an arrhythmogenic rabbit model of nonischemic HF in which 90% of HF rabbits exhibit spontaneous VT and 10% die suddenly.2,22 These were validated by studies in LV tissue from patients with end-stage HF attributable to IDCM and from nonfailing human hearts.
An expanded Materials and Methods section is in the online data supplement available at http://circres.ahajournals.org.
Arrhythmogenic Rabbit Nonischemic HF Model
New Zealand White rabbits underwent induction of HF by aortic insufficiency followed by thoracic aortic constriction, as previously described, until LV end-systolic dimension (LVESD) exceeded 1.20 cm (30% increase).2,22 HF and age-matched control rabbit hearts were rapidly excised and the LV free wall was flash-frozen in liquid nitrogen. LV myocytes were isolated as previously described.22 The protocol was approved by the University of Illinois at Chicago Animal Studies Committee.
Human Heart Tissue
LV tissue from failing human hearts (end-stage IDCM, n=8, mean ejection fraction 13±2%) was obtained at the time of cardiac transplantation performed at University of Illinois at Chicago or Loyola University Hospital. LV tissue from nonfailing human hearts that were not used for transplantation for technical reasons (but with normal LV function, n=8) was obtained through the Regional Organ Bank of Illinois. Immediately after explantation, LV tissue was flash-frozen in liquid nitrogen. The study was approved by the Human Studies Committees of University of Illinois at Chicago, Loyola University, and Regional Organ Bank of Illinois.
Northern and Western Blot Analysis
High-yield undegraded mRNA and protein were used in all studies. Synthesis of RNA probes and Northern hybridization were based on methods of Srivastava et al,23 with modifications.
Western blotting was performed as previously described.22 Total Cx43 protein (Cx43-T) was assessed using a polyclonal Cx43-T antibody (Zymed). The nonphosphorylated isoform of Cx43 (Cx43-NP) was detected by a specific monoclonal Cx43-NP antibody (Zymed).17
Immunoconfocal Microscopy and Quantitative Analysis
Immunolabeling was performed with polyclonal Cx43-T and monoclonal Cx43-NP antibodies described, monoclonal Cx43-T antibody (Chemicon), and monoclonal PP1 and PP2A catalytic subunit (PP2Ac) antibodies (BD Sciences). Images were collected from a laser scanning confocal microscope (Carl Zeiss) and quantitatively analyzed using National Institutes of Health image software.
Phosphatase Digestion and Phosphatase Inhibition
LV homogenates were incubated with calf intestine alkaline phosphatase (Roche) for phosphatase digestion, and isolated myocytes were incubated with 10 nmol/L okadaic acid (OA; Sigma) for phosphatase inhibition. Changes in substrate phosphorylation status were analyzed by immunoblotting with Cx43-T and Cx43-NP antibodies.
Cosedimentation and Coimmunoprecipitation
Sedimentation of PP1 and PP2A from rabbit LV homogenates was performed using microcystin-Sepharose beads (Upstate). Immunoprecipitation of Cx43-T using a specific monoclonal Cx43-T antibody was based on the method of Ausubel et al,24 with modifications. Immunoblotting was then performed with polyclonal and monoclonal Cx43-T antibodies and monoclonal PP1 and PP2Ac antibodies as described.
Microinjection of Lucifer Yellow
Lucifer Yellow 4% combined with Rhodamine B dextran 1% was microinjected into one cell of each cell pair in M199 cell culture medium in the absence or presence of OA (10 nmol/L), according to the method of Ruiz-Meana et al.25 Microinjection and image recording were performed under a confocal microscope (Carl Zeiss).
Statistical Analysis
All data were presented as means±SEM. Differences between two groups were evaluated using Student t test, and P<0.05 was considered to be significant.
Nonischemic HF Rabbit Echocardiographic Analysis
HF rabbit hearts (n=15) exhibited marked LV dilatation and systolic dysfunction compared with their baseline condition (as well as compared to age-matched controls, n=8; Table, see online supplement). With HF, LV end-diastolic dimension (LVEDD) increased by 46% (P<0.001), LV end-systolic dimension (LVESD) increased by 63% (P<0.001), and mean fractional shortening (FS) decreased by 23% (P<0.001). Heart weight and heart-to-body weight ratio were increased by 66% (P<0.001) and 80% (P<0.001), respectively, in HF versus controls, whereas body weight did not differ.
Cx43 mRNA and Protein Expression in Control and HF Rabbit LV
Cx43 mRNA was detected by Northern blotting as a single3-kb band (Figure 1A, top), and its abundance was normalized to GAPDH mRNA (Figure 1A, bottom). There was a 29% decrease in normalized Cx43 mRNA in HF versus controls (n=7 and n=6, respectively; P<0.05; Figure 1B). Similar results were found when Cx43 mRNA was normalized to 18S ribosomal mRNA (data not shown).
Figure 1. A, Northern blot images of Cx43 (top) and GAPDH mRNA (bottom) from control and HF rabbit LV. B, Summarized Cx43 mRNA levels (*P<0.05). C, Western blots of Cx43-T (top), Cx43-NP isoform (middle), and GAPDH (bottom) from control and HF rabbits in the absence (left, –AP) and presence (right, +AP) of alkaline phosphatase. D, Summarized data for Cx43 expression in control and HF rabbit LV (**P<0.001, *P<0.05). At left are the changes in Cx43-T protein expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody).
Western blot analysis was performed in LV homogenates from HF and age-matched control hearts, and the signal increased linearly with respect to the amount of protein loaded (Figure I in the online data supplement). The polyclonal Cx43-T antibody recognizes both phosphorylated and Cx43-NP isoforms. As such, multiple bands (corresponding to Cx43-NP and higher-molecular-weight phosphorylated isoforms of Cx43) (Figure 1C, left) migrated at 43 kDa (42 to 46 kDa). Cx43 total protein (normalized to GAPDH) was decreased 34% in HF versus controls (n=14 and n=14, respectively; P<0.001; Figure 1D, left).
Cx43 Phosphorylation in Control and HF Rabbit LV Tissue
Under physiological conditions, cardiac Cx43 is primarily phosphorylated, but phosphorylation and dephosphorylation can be an important pathway to regulate cell-cell communication.14–16 The appearance of the Cx43-T bands in Western blots in Figure 1C suggests that with HF, there is a shift from the higher-molecular-weight phosphorylated form of Cx43 to the lower molecular weight Cx43-NP. With the Cx43-T Western blotting data in rabbit LV tissue (as in Figure 1C), we were consistently able to distinguish the lower-molecular-weight (42kD) band, which represents Cx43-NP, from the higher-molecular-weight (43 to 46 kD) bands that represent phosphorylated isoforms of Cx43 (Cx43-P). Based on these data, we determined that the proportion of phosphorylated to nonphosphorylated Cx43 changed from 88:12 (or 7.5:1) in control to 81:19 (or 4:1) with HF (Figure 1D, left), and the ratio of Cx43-NP to Cx43-T increased 64% (P<0.05; Figure 1D, middle). We also assessed changes in Cx43 phosphorylation state directly by using an antibody specific for Cx43-NP that has been extensively validated.17 Immunoblotting revealed that Cx43-NP (normalized to total Cx43) increased by 66% in HF versus control rabbit LV (P<0.05; Figure 1C and Figure 1D, right). This result was comparable to the data described.
To further characterize the accumulation of Cx43-NP isoform in HF, phosphatase digestion and immunoblotting with the Cx43-T and Cx43-NP antibodies were performed. Figure 1C shows a greater amount of Cx43-NP isoform in HF rabbit LV at baseline. With alkaline phosphatase treatment, there is a dramatic increase in the Cx43-NP signal (which now reflects the total amount of Cx43), indicating that the phosphorylated form of Cx43 predominates in both control and HF, but in HF there is both less Cx43-T and a greater proportion of Cx43-NP.
Cx43 Phosphorylation in Isolated Rabbit Myocytes
Immunoblot data from isolated rabbit LV myocytes revealed that Cx43-T protein expression (measured as the 42 to 46 kDa band) was decreased 30% in HF versus controls (P<0.05; n=7 and n=5, respectively; Figure 2A and 2B, left), with a similar shift from phosphorylated Cx43 to the lower-molecular-weight Cx43-NP. As mentioned, we also determined the proportion of phosphorylated to nonphosphorylated Cx43 and found that it changed from 86:14 (or 6:1) in control to 75:25 (or 3:1) with HF (Figure 2B, left), and the ratio of Cx43-NP to Cx43-T increased 72% (P<0.05; Figure 2B, middle). When we used the antibody specific for Cx43-NP, we found that with HF, the level of Cx43-NP (normalized to total Cx43) increased by 69% (P<0.05; Figure 2A and Figure 2B, right), results that were comparable to those noted in LV tissue (Figure 1D).
Figure 2. A, Immunoblot images of Cx43-T (top), Cx43-NP isoform (middle), and GAPDH (bottom) bands from isolated LV myocytes of control and HF rabbits. B, Summarized data for Cx43 expression in control and HF rabbit LV myocytes (*P<0.05). At left are the changes in Cx43-T protein expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody). C, Double immuno-fluorescence staining with Cx43 total (green, top) and Cx43-NP isoform (red, bottom) antibodies in rabbit LV myocytes from control and HF (scale bar=20 μm). D, Bar graph of immunofluorescence density results for Cx43-T (Total) and the proportion of Cx43-NP to Cx43-T (NP/Total) (*P<0.05).
We confirmed these changes in Cx43 phosphorylation states with HF by immunofluorescent staining using confocal microscopy. Double staining was performed in isolated rabbit LV myocytes with Cx43-T and Cx43-NP antibodies. There was no fluorescence signal found by staining with the secondary antibody only (negative control, data not shown). Figure 2C shows that Cx43 protein was localized throughout the myocyte membrane in distinctive punctuate patterns, primarily at cell ends (rather than sides). There was similar localization between immunolabeled Cx43-T (green) and Cx43-NP isoform (red) signals. Cx43-T was reduced by 28% and Cx43-NP (normalized to Cx43-T protein) was enhanced by 71% in HF myocytes versus controls (P<0.05; n=3 and n=3, respectively; Figure 2D), results that were consistent with the immunoblotting data in rabbit LV tissue and myocytes above.
Cx43 Expression and Phosphorylation States in LV From Patients With IDCM
To validate the results from HF rabbit LV, similar experiments were performed in LV tissue from nonfailing and failing (IDCM) human hearts. The Northern blot data revealed that Cx43 mRNA (normalized to GAPDH) was reduced by 28% in IDCM versus nonfailing human LV (n=8 and n=8, respectively; P<0.05; Figure 3A and 3B).
Figure 3. A, Northern blot of Cx43 mRNA (upper) and GAPDH mRNA (lower) from LV samples of nonfailing (NF) and IDCM human hearts. B, Changes in Cx43 mRNA expression with IDCM versus NF (P<0.05). C, Immunoblot images of Cx43-T (top), Cx43-NP (middle), and GAPDH (bottom) in human NF and IDCM LV. D, Summarized data for Cx43 expression in control and IDCM human LV (*P<0.05, **P<0.01). At left are the changes in Cx43-T expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody). E, Alkaline phosphatase digestion of samples from human nonfailing and IDCM hearts.
Immunoblotting of human LV homogenates was performed using both Cx43-T and Cx43-NP antibodies, and the results were normalized to GAPDH (Figure 3C and 3D). Cx43-T was decreased by 25% in IDCM versus nonfailing LV (n=8 and n=8, respectively; P<0.05, left). From band analysis of the Cx43 total blot (as described), we calculated that the proportion of phosphorylated to nonphosphorylated Cx43 changed from 88:12 (or 7:1) in control to 80:20 (or 4:1) with IDCM (Figure 3D, left), and the ratio of Cx43-NP to Cx43-T increased 65% (P<0.01; Figure 3D, middle). There was a 60% increase (P<0.05; Figure 3C and Figure 3D, right) in Cx43-NP (normalized to total Cx43) in HF by using the specific Cx43-NP antibody. All these results, along with phosphatase digestion data shown in Figure 3E, were comparable to the data from HF rabbit LV described.
Colocalization of Protein Phosphatases With Cx43 in LV and Enhanced Colocalized Protein Phosphatases With HF
To determine whether enhanced Cx43-NP, which was observed in HF LV, is related to the interaction of protein phosphatases (PP1 and PP2A) with Cx43, coimmunoprecipitation, cosedimentation, and immunoconfocal analysis were performed in control and HF rabbit LV tissue.
To explore for a possible direct interaction of PP1 and PP2A with Cx43, PP1 and PP2A proteins were first isolated from homogenates by binding to microcystin (a specific inhibitor of PP1 and PP2A), followed by immunoblotting analysis. We observed a single cosedimented Cx43 reactive band at 42 to 46 kDa (Figure 4A, top), along with PP1 and PP2Ac bands (Figure 4A, middle and bottom).
Figure 4. A, Western blot images of Cx43-T (top), PP1 (middle), and PP2Ac (bottom) from control and HF rabbit LV homogenates after sedimentation of PP1 and PP2A with microcystin-Sepharose beads. B, Similar Western blot images after immunoprecipitating Cx43 with monoclonal Cx43-T antibody. C, Quantitation of the levels of PP1 and PP2Ac that coimmunoprecipitate with Cx43-T antibody (normalized to immunoprecipitated Cx43; **P<0.01).
We confirmed these findings by immunoprecipitation using a monoclonal Cx43-T antibody. Along with immunoprecipitated Cx43 bands (which were detected by both polyclonal and monoclonal Cx43-T antibodies, and which were distinct from IgG bands), immunoblotting images revealed coimmunoprecipitated PP1 and PP2Ac reactive bands (Figure 4B). IgG alone was the negative (data not shown). Quantitation of the coimmunoprecipitated PP1 and PP2Ac (normalized to immunoprecipitated Cx43 in each individual sample) revealed that the level of PP1 was unchanged, but the level of coimmunoprecipitated PP2Ac was increased 2.5-fold in HF rabbit LV tissue versus controls (n=4 and n=4, respectively; P<0.002; Figure 4C).
We further confirmed the increased colocalization of PP2Ac with Cx43 in HF by using immunofluorescence staining with Cx43-NP and PP2Ac antibodies in control and HF LV myocytes. Figure 5 shows representative confocal images where Cx43-NP isoform (red) is present in distinctive punctate patterns throughout the cell membrane, and it colocalizes with PP2Ac (green, also distributed in cytosol) that is increased 2-fold in HF myocytes (consistent with the immunoprecipitation and cosedimentation data). Similar results were found for double staining using Cx43-T and PP2Ac antibodies (data not shown). PP1 also colocalized with Cx43 at distinct membrane sites, but the amount of colocalized PP1 appeared unchanged in HF (data not shown).
Figure 5. Double immunofluorescence staining of control (left) and HF rabbit (right) isolated LV myocytes with Cx43-NP and PP2Ac antibodies alone (top, red; middle, green), and the overlap (bottom, yellow). Scale bar is 20 μm.
Alteration of Cell-Cell Communication and the Role of PP2A in Accumulation of Dephosphorylated Cx43 in HF
To define alterations of intercellular communication in HF and it modulation by protein phosphatases, we performed Lucifer Yellow dye transfer (microinjection) in the absence or presence of the protein phosphatase inhibitor OA, at a concentration (10 nmol/L) that has been shown to inhibit PP2A, but not PP1 (Figure 6).26,27 To delineate the injected cell of the cell pairs, we also injected the high-molecular-weight (10 000) Rhodamine B dextran, which is retained in the injected myocyte without passing gap junction channels to the recipient cell (Figure 6, bottom left). Figure 6 shows images of the microelectrode and control, HF, and OA-treated HF cell pairs (injected and recipient cells), and sequential confocal images at times 1, 38, and 214 seconds. Figure 7A shows representative time course data for fluorescence (normalized to plateau) in the recipient cell (30x80 μm window; Figure 6, bottom right panel), and Figure 7B shows summarized rate constant (, time to reach 63% peak) data. HF myocytes had decreased cellular coupling, reflected by slower dye transfer with an increased of dye transfer of 53±12 seconds versus 19±5 seconds for control (P<0.05; n=6 and n=5, respectively). However, OA-treated HF cell pairs showed improved dye transfer with a reduced of dye transfer (=24±5; P<0.05 versus HF and P=NS versus control; n=4). Protein phosphatase inhibition by OA (10 nmol/L) was verified by immunoblotting, with OA-treated HF myocytes exhibiting a 42±9.3% decrease in Cx43-NP (Figure 7C; n=3). OA-treated control cell pairs showed no additional improvement in cell coupling (=18±7; n=4; P=NS versus untreated control cell pairs) (Figure 7A).
Figure 6. Lucifer Yellow (LY) microinjection images of end-to-end cell pairs (left) and subsequent LY dye transfer (right, yellow) from a single injected cell to a recipient cell in control, HF, and OA-treated HF myocytes at 1, 38, and 214 seconds (right). A representative 30x80 μm window, which was used to quantitate dye transfer (fluorescence intensity in the recipient myocyte), is shown in the bottom right panel. At bottom left is an example of Rhodamine B dextran (Rh, red) retained in the injected myocyte. Scale bar at lower left=20 μm.
Figure 7. A, Representative fitted single time courses of LY transfer in recipient cell from end-to-end cell pairs of control, HF, and 10 nmol/L OA-treated control and HF myocytes, respectively. B, Summarized data of LY transfer in recipient cell from control, HF, and OA-treated control and HF myocytes (*P<0.05). C, Representative Western blot images of Cx43-NP and GAPDH from myocytes in the absence or presence of OA (10 nmol/L).
In the present study, we studied LV tissue and myocytes from an arrhythmogenic rabbit model of nonischemic HF and from age-matched control rabbits. We found that Cx43 expression was downregulated on both a protein and mRNA level. HF rabbits exhibited a significant enhancement of the nonphosphorylated isoform of Cx43 (and decreased phosphorylated Cx43). All these results were validated in LV tissue from hearts of patients with end-stage IDCM (at the time of transplantation) compared with nonfailing hearts. We also observed that protein phosphatases PP1 and PP2Ac both colocalized with Cx43 in rabbit LV, and there was a 2.5-fold increase in the level of colocalized PP2Ac in HF. Moreover, we found that intercellular coupling between LV myocytes was reduced markedly in HF. However, OA reduced the amount of Cx43-NP and significantly improved cell coupling in HF. Our observations suggest that both reduced Cx43 expression and enhanced Cx43-NP within gap junctions underlie impaired cellular coupling and ultimately altered conduction in nonischemic HF. The increased levels of colocalized PP2Ac (with Cx43) contribute to enhanced Cx43-NP in HF.
Reduced Cx43 Expression and Increased Cx43-NP Is Associated With Altered Cell Coupling and Conduction in Nonischemic HF
Ventricular arrhythmias (VT and VF) account for nearly half of deaths in patients with HF (both nonischemic and ischemic).1 In 3-dimensional cardiac mapping studies, we have previously shown that spontaneously occurring VT in nonischemic HF (both in our HF rabbit model and in IDCM patients) is initiated and maintained by a nonreentrant mechanism such as delayed or early afterdepolarizations.2 However, mapping in human IDCM has revealed altered anisotropic conduction and conduction block that could underlie the development of reentry (especially during the transition from VT to VF).3
In the present study, we found that HF myocytes exhibit reduced cellular coupling. This was associated with decreased Cx43 expression on both an mRNA and protein level, similar to that reported by others.11 Cx43 downregulation may be a common feature of failing heart,9 although the underlying biochemical basis remains to be determined. There has been considerable debate as to the role of Cx43 in modulating conduction.10,29 However, studies in heterozygous Cx43 knockout mice,15 along with those in other transgenic models,30 and in dogs with pacing-induced HF,11 support the contention that reduced Cx43 expression per se could contribute to altered cell-coupling and conduction in nonischemic HF.
Cx43 is a phosphoprotein that is predominantly phosphorylated in the control state. We confirmed this in control rabbit and nonfailing human LV. With HF, we found a significant increase in the Cx43-NP on Western blots using an extensively validated antibody specific for the Cx43-NP isoform.17 Under physiological conditions, the C-terminus of Cx43 is extensively phosphorylated, primarily at serine sites, although recent studies suggest there may be changes in tyrosine phosphorylation in HF.31 The specific Cx43-NP antibody used in the present study recognizes the nonphosphorylated serine-rich segment of the Cx43 C-terminus that includes putative phosphorylation sites for multiple kinases including protein kinase A, protein kinase C, and MAPK.15–17 Our results of Cx43-NP in HF rabbits were consistent with both LV tissue and isolated myocytes. Moreover, findings in human IDCM were nearly identical. Despite the consistency of this finding, interpretation is limited because the extent to which Cx43 is phosphorylated at baseline in rabbit and human LV is not known. However, we took advantage of the Cx43 expression pattern on Western blots and determined the extent of Cx43 phosphorylation in controls and HF. Our results in controls were consistent between rabbits and nonfailing humans (and similar to that noted by others17). The changes in Cx43 phosphorylation with HF were substantive and confirmed the relative changes in Cx43-NP using a specific Cx43-NP antibody.
The Role of Phosphatases in Cx43 Dephosphorylation in Nonischemic HF
Reduced Cx43 phosphorylation in HF could be attributable to reduced phosphorylation by kinases (eg, protein kinase A, protein kinase C) and/or to increased dephosphorylation by protein phosphatases. Here, we focused on the role of phosphatases PP1 and PP2A, the primary phosphatases that dephosphorylate Cx43,14 because levels of total cellular PP1 and total cellular PP1 and PP2A activity are increased in HF.19 Moreover, activation of phosphatases that dephosphorylate Cx43 can reduce communication between cells.14
We reasoned that local levels of an enzyme (rather than total cellular levels) may best reflect its actions, and that PP1 and PP2A might interact directly with Cx43. Cx43 is a part of a protein complex, and it has been reported that kinases and structural proteins have direct interactions with cardiac Cx43.15,20 The present study provides the first evidence of an interaction between Cx43 and PP1 and PP2Ac in rabbit LV. Cx43, PP1, and PP2A form a close complex that could be immunoprecipitated by a Cx43 antibody or sedimented by microcystin-Sepharose beads (that bind PP1 and PP2A). We confirmed this with immunofluorescence staining showing colocalization at gap junctional sites in the membrane. With HF, we found that the level of coimmunoprecipitated PP2Ac with Cx43 was dramatically increased >2.5-fold, whereas colocalized PP1 was unchanged. Moreover, OA, at a concentration that inhibits PP2A, but not PP1,26,27 and that increased the levels of phosphorylated Cx43, significantly improved cellular coupling between HF rabbit myocytes. The rapidity and near-complete recovery of coupling (even in the setting of reduced Cx43 protein levels) suggest that phosphorylation provides a dynamic means of regulation and that there may be a steep relationship between Cx43 phosphorylation state and cellular coupling (that may plateau, as reflected by the unchanged coupling in OA-treated control cell pairs). Whether dephosphorylated Cx43 per se reduces coupling remains to be determined.17
Animal Model of HF
Our HF rabbit model exhibits contractile dysfunction, pathologic alterations, and arrhythmogenesis similar to nonischemic human HF,2,22,28 and we have consistently validated biochemical and electrophysiologic findings in this model with those in nonischemic HF in humans2,22 (including the results of the present study). This suggests that this arrhythmogenic HF rabbit model is ideally suited for studies of altered Cx43 expression, phosphorylation, and function.
Limitations
Our observations strongly suggest that increased levels of colocalized PP2Ac (with Cx43) contribute to enhanced Cx43-NP and reduced cellular coupling in HF LV. Whether increased activity of other protein phosphatases, such as PP1 and calcineurin, or decreased activity of kinases contribute to Cx43 dephosphorylation in HF remains unknown.14
In addition to Cx43 downregulation and altered Cx43 phosphorylation, there could be changes in gap junction distribution with HF that may modulate conduction. Although this was not assessed in the present study, recent studies have demonstrated altered gap junction distribution in human IDCM.33 We focused on alterations in Cx43, by far the predominant ventricular gap junction proteins, so we cannot rule out alterations in the expression of other connexins (eg, Cx45 or Cx40) in HF.
Implications
Our findings in the present study suggest that the combination of Cx43 downregulation and altered phosphorylation contribute to decreased cellular coupling and, ultimately, slow conduction in nonischemic HF (Figure 8). The colocalization of PP1 and PP2Ac with Cx43 indicates that Cx43 represents a macromolecular complex with numerous components, including both kinases and phosphatases. Enhanced levels of colocalized PP2Ac in HF indicate an additional regulation of gap junctional function in HF that is locally controlled.
Figure 8. Schematic diagram showing changes in Cx43 expression, phosphorylation state, and colocalized phosphatases with HF. With HF, the combination of decreased Cx43 expression and enhanced nonphosphorylated Cx43 (arising from increased associated PP2A) contribute to reduced cell-coupling and arrhythmogenesis.
Alterations in Cx43 phosphorylation could modulate connexin channel gating, as well as connexin trafficking and degradation.32 Whereas effects of Cx43 phosphorylation on channel gating are complex (dependent on specific kinases and phosphorylation sites), decreased Cx43 phosphorylation has been shown to be associated with decreased gap junctional conductance, whether assessed in vitro in cell pairs with activation of protein phosphatases14 or in vivo in the ischemic heart.17 Heterogeneity of these changes in HF would be expected to further enhance the effects on conduction. Moreover, if Cx43 dephosphorylation enhances gap junction degradation,32 the degree of Cx43 dephosphorylation in HF could be underestimated, and may even contribute in part to the decreased Cx43 protein levels.
Our findings, including the improved cellular coupling with OA in HF, suggest that modulation of Cx43 phosphorylation state and Cx43-associated proteins (eg, phosphatases) could represent novel therapeutic approaches to improve conduction in the failing heart.
References
Packer M. Sudden unexpected death in patients with congestive heart failure: a second frontier. Circulation. 1985; 72: 681–685.
Pogwizd SM. Nonreentrant mechanisms underlying spontaneous ventricular arrhythmias in a model of nonischemic heart failure in rabbits. Circulation. 1995; 92: 1034–1048.
Anderson KP, Walker R, Urie P, Ershler PR, Lux RL, Karwandee SV. Myocardial electrical propagation in patients with idiopathic dilated cardiomyopathy. J Clin Invest. 1993; 92: 122–140.
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: 727–741.
Vermeulen JT, McGuire MA, Opthof T, Coronel R, de Bakker JMT, Klopping C, Janse MJ. Triggered activity and automaticity in ventricular trabeculae of failing human and rabbit hearts. Cardiovasc Res. 1994; 28: 1547–1554.
De Mello WC. Renin-angiotensin system and cell communication in the failing heart. Hypertension. 1996; 27: 1267–1272.
Kumar NM, Gilula NB. The gap junction communication channel. Cell. 1996; 84: 381–388.
Saffitz JE, Davis LM, Darrow BJ, Kanter HL, Laing JG, Beyer EC. The molecular basis of anisotropy: role of gap junctions. J Cardiovasc Electrophysiol. 1995; 6: 498–510.
Wang XJ, Gerdes AM. Chronic pressure overload cardiac hypertrophy and failure in guinea pigs: III. Intercalated disc remodeling. J Mol Cell Cardiol. 1999; 31: 333–343.
Guerrerro PA, Schuessler RB, Davis LM, Beyer EC, Johnson CM, Yamada KA, Saffitz JE. Slow ventricular conduction in mice heterozygous for a connexin43 null mutation. J Clin Invest. 1997; 99: 1991–1998.
Poelzing S, Rosenbaum DS. Altered connexin43 expression produces arrhythmia substrate in heart failure. Am J Physiol Heart Circ Physiol. 2004; 287: H1762–H1770.
Lerner DL, Yamada KA, Schuessler RB, Saffitz JE. Accelerated onset and increased incidence of ventricular arrhythmias induced by ischemia in Cx43-deficient mice. Circulation. 2000; 101: 547–552.
Kwak BR, Jongsma H. Regulation of cardiac gap junction channel permeability and conductance by several phosphorylating conditions. Mol Cell Biochem. 1996; 157: 93–99.
Duthe F, Plaisance I, Sarrouilhe D, Herve JC. Endogenous protein phosphatase 1 runs down gap junctional communication of rat ventricular myocytes. Am J Physiol Cell Physiol. 2001; 281: C1648–C1656.
Bowling N, Huang X, Sandusky GE, Fouts RL, Mintze K, Esterman M, Allen PD, Maddi R, McCall E, Vlahos CJ. Protein kinase C- and - modulate connexin43 phosphorylation in human heart. J Mol Cell Cardiol. 2001; 33: 789–798.
Darrow BJ, Fast VG, Kleber AG, Beyer EC, Saffitz JE. Functional and structural assessment of intercellular communication. Increased conduction velocity and enhanced connexin expression in dibutyryl cAMP-treated cultured cardiac myocytes. Circ Res. 1996; 79: 174–183.
Beardslee MA, Lerner DL, Tadros PN, Laing JG, Beyer EC, Yamada KA, Kleber AG, Schuessler RB, Saffitz JE. Dephosphorylation and intracellular redistribution of ventricular connexins 43 during electrical uncoupling induced by ischemia. Circ Res. 2000; 87: 656–662.
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.
Neumann J, Eschenhagen T, Jones LR, Linck B, Schmitz W, Scholz H, Zimmermann N. Increased expression of cardiac phosphatases in patients with end-stage heart failure. J Mol Cell Cardiol. 1997; 29: 265–272.
Barker RJ, Price RL, Gourdie RG. Increased association of ZO-1 with connexin43 during remodeling of cardiac gap junctions. Circ Res. 2002; 90: 317–324.
Marx SO, Reiken S, Hisamatsu Y, Jayaraman T, Burkhoff D, Rosemblit N, Marks AR. PKA phosphorylation dissociates FKBP12.6 from the calcium release channel: defective regulation in failing hearts. Cell. 2000; 101: 365–376.
Pogwizd SM, Qi M, Yuan W, Samarel AM, Bers DM. Upregulation of Na+/Ca2+ exchanger expression and function in an arrhythmogenic rabbit model of heart failure. Circ Res. 1999; 85: 1009–1019.
Srivastava RA, Schonfeld G. Use of riboprobes for Northern blotting analysis. BioTechniques. 1991; 11: 584–588.
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K. Short protocol in molecular biology. 4th ed. New York: John Wiley and Sons; 1999: 90–91.
Ruiz-Meana M, Garcia-Dorado D, Hofstaetter B, Piper HM, Soler-Soler J. Propagation of cardiomyocyte hypercontracture by passage of Na+ through gap junctions. Circ Res. 1999; 85: 280–287.
Herzig S, Neuman J. Effects of serine/threonine protein phosphatases on ion channels in excitable membranes. Physiol Rev. 2000; 80: 173–210.
Bialojan C, Takai A. Inhibitory effect of a marine-sponge toxin, okadaic acid, on protein phosphatases. Specificity and kinetics. Biochem J. 1988; 256: 283–290.
Pogwizd SM, Schlotthauer K, Li L, Yuan W, Bers DM. Arrhythmogenesis and contractile dysfunction in heart failure: Roles of sodium-calcium exchange, inward rectifier potassium current, and residual beta-adrenergic responsiveness. Circ Res. 2001; 88: 1159–1167.
Morley GE, Vaidya D, Samie FH, Lo C, Delmar M, Jalife J. Characterization of conduction in the ventricles of normal and heterozygous Cx43 knockout mice using optical mapping. J Cardiovasc Electrophysiol. 1999; 10: 1361–1375.
Gutstein DE, Morley GE, Tamaddon H, Vaiday D, Schneider MD, Chen J, Chien KR, Stuhlmann H, Fishman GI. Conduction slowing and sudden arrhythmic death in mice with cardiac-restricted inactivation of connexin43. Circ Res. 2001; 88: 333–339.
Toyofuku T, Yabuki M, Otsu K, Kuzuya T, Tada M, Hori M. Functional role of c-Src in gap junctions of the cardiomyopathic heart. Circ Res. 1999; 85: 672–681.
Musil LS, Goodenough DA. Biochemical analysis of connexin43 intracellular transport, phosphorylation, and assembly into gap junctional plaques. J Cell Biol. 1991; 115: 1357–1374.
Kostin S, Rieger M, Dammer S, Hein S, Richter M, Klovekorn WP, Bauer ER, Schaper J. Gap junction remodeling and altered connexin43 expression in the failing human heart. Mol Cell Biochem. 2003; 242: 135–144.(Xun Ai, Steven M. Pogwizd)
In nonischemic heart failure (HF), ventricular tachycardia initiates by a nonreentrant mechanism, but there is altered conduction (that could lead to re-entry) that could arise from changes in gap junctional proteins, especially connexin43 (Cx43). We studied Cx43 expression and phosphorylation state in the left ventricle (LV) from an arrhythmogenic rabbit model of nonischemic HF and from patients with HF attributable to idiopathic dilated cardiomyopathy. We also investigated the role of protein phosphatases that dephosphorylate Cx43—PP1 and PP2A. In HF rabbit LV, Cx43 mRNA and total protein were decreased by 29% and 34%, respectively (P<0.05 and P<0.001). In controls, Cx43 was primarily in the phosphorylated state, but with HF there was a 64% increase in nonphosphorylated Cx43 (Cx43-NP, normalized to total Cx43; P<0.05). Similar results were noted in HF rabbit myocytes (P<0.05) and in human idiopathic dilated cardiomyopathy LV (P<0.05). We found that PP1 and PP2A colocalized with Cx43 in rabbit LV. With HF, the level of colocalized PP2A increased >2.5-fold (P<0.002), whereas colocalized PP1 was unchanged. We also found intercellular coupling (assessed by Lucifer Yellow dye transfer) was markedly reduced in HF. However, okadaic acid (10 nmol/L) reduced the amount of Cx43-NP and significantly improved cell coupling in HF. Thus, in nonischemic HF in rabbits and humans, there is a decrease in both Cx43 expression and phosphorylation that contributes to uncoupling. Increased levels of PP2A that colocalize with Cx43 can underlie enhanced levels of Cx43-NP in HF. Modulation of Cx43 phosphorylation may be a potential therapeutic target to improve conduction in HF.
Key Words: gap junctions , phosphorylation , phosphatases , heart failure , arrhythmia
Heart failure (HF), whether nonischemic or ischemic, is associated with a nearly 50% incidence of sudden death, primarily from ventricular tachycardia (VT) degenerating to ventricular fibrillation (VF).1 Whereas VT in nonischemic HF initiates primarily by a nonreentrant mechanism,2 myocardium from patients with idiopathic dilated cardiomyopathy (IDCM) exhibits nonuniform anisotropy, slow conduction, and conduction block3 that could underlie reentry during the transition from VT to VF. Conduction slowing could arise from decreased depolarizing currents and/or decreased gap junctional coupling.4 However, the degree of slow conduction and block in failing myocardium appears to be out of proportion to the changes in active membrane properties.5 Moreover, LV myocytes from an animal model of nonischemic HF exhibit markedly decreased gap junctional conductance.6 Thus, alterations in intercellular coupling involving cardiac gap junctions may underlie slow conduction in nonischemic HF.
Gap junctions are specialized membrane structures consisting of arrays of intercellular channels that directly connect adjacent cells by providing chemical and electrical communication. They are composed of connexins, a multigene family of conserved proteins. The relative amounts, composition, and distribution of these connexins appear to influence the conduction properties of cells.7,8 Connexin43 (Cx43) is the major connexin protein in ventricular myocardium, and downregulation of Cx43 (mRNA and protein) has been demonstrated in some experimental HF models and in the failing human heart.9 Studies in Cx43 heterozygous knockout mice, and in the canine rapid pacing HF model, support the fact that decreased Cx43 expression per se can lead to a slowing of conduction, and that arrhythmogenesis can be further enhanced under pathophysiologic conditions (eg, myocardial ischemia) and when gap junctional alterations are heterogeneous.10–12
Cx43 is a phosphoprotein that can be phosphorylated by a number of kinases13 and dephosphorylated by protein phosphatases such as PP1 and PP2A.14 Most phosphorylation sites are serine residues on the C-terminus, although threonine and tyrosine phosphorylation sites are present to a much lesser extent.13 Whereas the Cx43 phosphorylation states mediated by kinases can have variable effects on gap junctional communication,15,16 dephosphorylation of Cx43 (by phosphatases) has been shown to decrease gap junctional communication—whether assessed in neonatal rat ventricular cell pairs with activation of endogenous phosphatases14 or in perfused whole rat hearts during myocardial ischemia.17 Little is known about whether there is altered Cx43 phosphorylation in HF; however, if so, it could contribute to altered conduction and arrhythmogenesis.18
Total cellular PP1 and PP2A activity is increased in the failing heart.19 Recent studies have shown that Cx43 interacts with a number of proteins (eg, ZO-1 and protein kinase C) to form a protein complex.15,20 It is conceivable that protein phosphatases may also directly associate with Cx43 at gap junctions to modulate its phosphorylation state, as has recently been shown for other macromolecular complexes,21 and that the level of protein phosphatases in direct proximity to Cx43 may be enhanced in HF.
As such, the goals of these studies were to investigate: (1) alteration of Cx43 protein and mRNA expression; (2) change in Cx43 phosphorylation state; (3) possible colocalization of PP1 and PP2A with Cx43; (4) alteration of intercellular coupling; and (5) the contribution of PP1 and PP2A to Cx43 dephosphorylation and altered cell coupling in HF. Studies were performed in LV tissue and myocytes from an arrhythmogenic rabbit model of nonischemic HF in which 90% of HF rabbits exhibit spontaneous VT and 10% die suddenly.2,22 These were validated by studies in LV tissue from patients with end-stage HF attributable to IDCM and from nonfailing human hearts.
An expanded Materials and Methods section is in the online data supplement available at http://circres.ahajournals.org.
Arrhythmogenic Rabbit Nonischemic HF Model
New Zealand White rabbits underwent induction of HF by aortic insufficiency followed by thoracic aortic constriction, as previously described, until LV end-systolic dimension (LVESD) exceeded 1.20 cm (30% increase).2,22 HF and age-matched control rabbit hearts were rapidly excised and the LV free wall was flash-frozen in liquid nitrogen. LV myocytes were isolated as previously described.22 The protocol was approved by the University of Illinois at Chicago Animal Studies Committee.
Human Heart Tissue
LV tissue from failing human hearts (end-stage IDCM, n=8, mean ejection fraction 13±2%) was obtained at the time of cardiac transplantation performed at University of Illinois at Chicago or Loyola University Hospital. LV tissue from nonfailing human hearts that were not used for transplantation for technical reasons (but with normal LV function, n=8) was obtained through the Regional Organ Bank of Illinois. Immediately after explantation, LV tissue was flash-frozen in liquid nitrogen. The study was approved by the Human Studies Committees of University of Illinois at Chicago, Loyola University, and Regional Organ Bank of Illinois.
Northern and Western Blot Analysis
High-yield undegraded mRNA and protein were used in all studies. Synthesis of RNA probes and Northern hybridization were based on methods of Srivastava et al,23 with modifications.
Western blotting was performed as previously described.22 Total Cx43 protein (Cx43-T) was assessed using a polyclonal Cx43-T antibody (Zymed). The nonphosphorylated isoform of Cx43 (Cx43-NP) was detected by a specific monoclonal Cx43-NP antibody (Zymed).17
Immunoconfocal Microscopy and Quantitative Analysis
Immunolabeling was performed with polyclonal Cx43-T and monoclonal Cx43-NP antibodies described, monoclonal Cx43-T antibody (Chemicon), and monoclonal PP1 and PP2A catalytic subunit (PP2Ac) antibodies (BD Sciences). Images were collected from a laser scanning confocal microscope (Carl Zeiss) and quantitatively analyzed using National Institutes of Health image software.
Phosphatase Digestion and Phosphatase Inhibition
LV homogenates were incubated with calf intestine alkaline phosphatase (Roche) for phosphatase digestion, and isolated myocytes were incubated with 10 nmol/L okadaic acid (OA; Sigma) for phosphatase inhibition. Changes in substrate phosphorylation status were analyzed by immunoblotting with Cx43-T and Cx43-NP antibodies.
Cosedimentation and Coimmunoprecipitation
Sedimentation of PP1 and PP2A from rabbit LV homogenates was performed using microcystin-Sepharose beads (Upstate). Immunoprecipitation of Cx43-T using a specific monoclonal Cx43-T antibody was based on the method of Ausubel et al,24 with modifications. Immunoblotting was then performed with polyclonal and monoclonal Cx43-T antibodies and monoclonal PP1 and PP2Ac antibodies as described.
Microinjection of Lucifer Yellow
Lucifer Yellow 4% combined with Rhodamine B dextran 1% was microinjected into one cell of each cell pair in M199 cell culture medium in the absence or presence of OA (10 nmol/L), according to the method of Ruiz-Meana et al.25 Microinjection and image recording were performed under a confocal microscope (Carl Zeiss).
Statistical Analysis
All data were presented as means±SEM. Differences between two groups were evaluated using Student t test, and P<0.05 was considered to be significant.
Nonischemic HF Rabbit Echocardiographic Analysis
HF rabbit hearts (n=15) exhibited marked LV dilatation and systolic dysfunction compared with their baseline condition (as well as compared to age-matched controls, n=8; Table, see online supplement). With HF, LV end-diastolic dimension (LVEDD) increased by 46% (P<0.001), LV end-systolic dimension (LVESD) increased by 63% (P<0.001), and mean fractional shortening (FS) decreased by 23% (P<0.001). Heart weight and heart-to-body weight ratio were increased by 66% (P<0.001) and 80% (P<0.001), respectively, in HF versus controls, whereas body weight did not differ.
Cx43 mRNA and Protein Expression in Control and HF Rabbit LV
Cx43 mRNA was detected by Northern blotting as a single3-kb band (Figure 1A, top), and its abundance was normalized to GAPDH mRNA (Figure 1A, bottom). There was a 29% decrease in normalized Cx43 mRNA in HF versus controls (n=7 and n=6, respectively; P<0.05; Figure 1B). Similar results were found when Cx43 mRNA was normalized to 18S ribosomal mRNA (data not shown).
Figure 1. A, Northern blot images of Cx43 (top) and GAPDH mRNA (bottom) from control and HF rabbit LV. B, Summarized Cx43 mRNA levels (*P<0.05). C, Western blots of Cx43-T (top), Cx43-NP isoform (middle), and GAPDH (bottom) from control and HF rabbits in the absence (left, –AP) and presence (right, +AP) of alkaline phosphatase. D, Summarized data for Cx43 expression in control and HF rabbit LV (**P<0.001, *P<0.05). At left are the changes in Cx43-T protein expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody).
Western blot analysis was performed in LV homogenates from HF and age-matched control hearts, and the signal increased linearly with respect to the amount of protein loaded (Figure I in the online data supplement). The polyclonal Cx43-T antibody recognizes both phosphorylated and Cx43-NP isoforms. As such, multiple bands (corresponding to Cx43-NP and higher-molecular-weight phosphorylated isoforms of Cx43) (Figure 1C, left) migrated at 43 kDa (42 to 46 kDa). Cx43 total protein (normalized to GAPDH) was decreased 34% in HF versus controls (n=14 and n=14, respectively; P<0.001; Figure 1D, left).
Cx43 Phosphorylation in Control and HF Rabbit LV Tissue
Under physiological conditions, cardiac Cx43 is primarily phosphorylated, but phosphorylation and dephosphorylation can be an important pathway to regulate cell-cell communication.14–16 The appearance of the Cx43-T bands in Western blots in Figure 1C suggests that with HF, there is a shift from the higher-molecular-weight phosphorylated form of Cx43 to the lower molecular weight Cx43-NP. With the Cx43-T Western blotting data in rabbit LV tissue (as in Figure 1C), we were consistently able to distinguish the lower-molecular-weight (42kD) band, which represents Cx43-NP, from the higher-molecular-weight (43 to 46 kD) bands that represent phosphorylated isoforms of Cx43 (Cx43-P). Based on these data, we determined that the proportion of phosphorylated to nonphosphorylated Cx43 changed from 88:12 (or 7.5:1) in control to 81:19 (or 4:1) with HF (Figure 1D, left), and the ratio of Cx43-NP to Cx43-T increased 64% (P<0.05; Figure 1D, middle). We also assessed changes in Cx43 phosphorylation state directly by using an antibody specific for Cx43-NP that has been extensively validated.17 Immunoblotting revealed that Cx43-NP (normalized to total Cx43) increased by 66% in HF versus control rabbit LV (P<0.05; Figure 1C and Figure 1D, right). This result was comparable to the data described.
To further characterize the accumulation of Cx43-NP isoform in HF, phosphatase digestion and immunoblotting with the Cx43-T and Cx43-NP antibodies were performed. Figure 1C shows a greater amount of Cx43-NP isoform in HF rabbit LV at baseline. With alkaline phosphatase treatment, there is a dramatic increase in the Cx43-NP signal (which now reflects the total amount of Cx43), indicating that the phosphorylated form of Cx43 predominates in both control and HF, but in HF there is both less Cx43-T and a greater proportion of Cx43-NP.
Cx43 Phosphorylation in Isolated Rabbit Myocytes
Immunoblot data from isolated rabbit LV myocytes revealed that Cx43-T protein expression (measured as the 42 to 46 kDa band) was decreased 30% in HF versus controls (P<0.05; n=7 and n=5, respectively; Figure 2A and 2B, left), with a similar shift from phosphorylated Cx43 to the lower-molecular-weight Cx43-NP. As mentioned, we also determined the proportion of phosphorylated to nonphosphorylated Cx43 and found that it changed from 86:14 (or 6:1) in control to 75:25 (or 3:1) with HF (Figure 2B, left), and the ratio of Cx43-NP to Cx43-T increased 72% (P<0.05; Figure 2B, middle). When we used the antibody specific for Cx43-NP, we found that with HF, the level of Cx43-NP (normalized to total Cx43) increased by 69% (P<0.05; Figure 2A and Figure 2B, right), results that were comparable to those noted in LV tissue (Figure 1D).
Figure 2. A, Immunoblot images of Cx43-T (top), Cx43-NP isoform (middle), and GAPDH (bottom) bands from isolated LV myocytes of control and HF rabbits. B, Summarized data for Cx43 expression in control and HF rabbit LV myocytes (*P<0.05). At left are the changes in Cx43-T protein expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody). C, Double immuno-fluorescence staining with Cx43 total (green, top) and Cx43-NP isoform (red, bottom) antibodies in rabbit LV myocytes from control and HF (scale bar=20 μm). D, Bar graph of immunofluorescence density results for Cx43-T (Total) and the proportion of Cx43-NP to Cx43-T (NP/Total) (*P<0.05).
We confirmed these changes in Cx43 phosphorylation states with HF by immunofluorescent staining using confocal microscopy. Double staining was performed in isolated rabbit LV myocytes with Cx43-T and Cx43-NP antibodies. There was no fluorescence signal found by staining with the secondary antibody only (negative control, data not shown). Figure 2C shows that Cx43 protein was localized throughout the myocyte membrane in distinctive punctuate patterns, primarily at cell ends (rather than sides). There was similar localization between immunolabeled Cx43-T (green) and Cx43-NP isoform (red) signals. Cx43-T was reduced by 28% and Cx43-NP (normalized to Cx43-T protein) was enhanced by 71% in HF myocytes versus controls (P<0.05; n=3 and n=3, respectively; Figure 2D), results that were consistent with the immunoblotting data in rabbit LV tissue and myocytes above.
Cx43 Expression and Phosphorylation States in LV From Patients With IDCM
To validate the results from HF rabbit LV, similar experiments were performed in LV tissue from nonfailing and failing (IDCM) human hearts. The Northern blot data revealed that Cx43 mRNA (normalized to GAPDH) was reduced by 28% in IDCM versus nonfailing human LV (n=8 and n=8, respectively; P<0.05; Figure 3A and 3B).
Figure 3. A, Northern blot of Cx43 mRNA (upper) and GAPDH mRNA (lower) from LV samples of nonfailing (NF) and IDCM human hearts. B, Changes in Cx43 mRNA expression with IDCM versus NF (P<0.05). C, Immunoblot images of Cx43-T (top), Cx43-NP (middle), and GAPDH (bottom) in human NF and IDCM LV. D, Summarized data for Cx43 expression in control and IDCM human LV (*P<0.05, **P<0.01). At left are the changes in Cx43-T expression with the relative contributions of Cx43-NP and Cx43-P indicated. In the middle are data for the ratio of Cx43-NP isoform to Cx43-T protein (NP/Total) using the Cx43-T antibody. At right is the ratio of Cx43-NP (using a specific Cx43 phospho-antibody) to total Cx43 (using Cx43-T antibody). E, Alkaline phosphatase digestion of samples from human nonfailing and IDCM hearts.
Immunoblotting of human LV homogenates was performed using both Cx43-T and Cx43-NP antibodies, and the results were normalized to GAPDH (Figure 3C and 3D). Cx43-T was decreased by 25% in IDCM versus nonfailing LV (n=8 and n=8, respectively; P<0.05, left). From band analysis of the Cx43 total blot (as described), we calculated that the proportion of phosphorylated to nonphosphorylated Cx43 changed from 88:12 (or 7:1) in control to 80:20 (or 4:1) with IDCM (Figure 3D, left), and the ratio of Cx43-NP to Cx43-T increased 65% (P<0.01; Figure 3D, middle). There was a 60% increase (P<0.05; Figure 3C and Figure 3D, right) in Cx43-NP (normalized to total Cx43) in HF by using the specific Cx43-NP antibody. All these results, along with phosphatase digestion data shown in Figure 3E, were comparable to the data from HF rabbit LV described.
Colocalization of Protein Phosphatases With Cx43 in LV and Enhanced Colocalized Protein Phosphatases With HF
To determine whether enhanced Cx43-NP, which was observed in HF LV, is related to the interaction of protein phosphatases (PP1 and PP2A) with Cx43, coimmunoprecipitation, cosedimentation, and immunoconfocal analysis were performed in control and HF rabbit LV tissue.
To explore for a possible direct interaction of PP1 and PP2A with Cx43, PP1 and PP2A proteins were first isolated from homogenates by binding to microcystin (a specific inhibitor of PP1 and PP2A), followed by immunoblotting analysis. We observed a single cosedimented Cx43 reactive band at 42 to 46 kDa (Figure 4A, top), along with PP1 and PP2Ac bands (Figure 4A, middle and bottom).
Figure 4. A, Western blot images of Cx43-T (top), PP1 (middle), and PP2Ac (bottom) from control and HF rabbit LV homogenates after sedimentation of PP1 and PP2A with microcystin-Sepharose beads. B, Similar Western blot images after immunoprecipitating Cx43 with monoclonal Cx43-T antibody. C, Quantitation of the levels of PP1 and PP2Ac that coimmunoprecipitate with Cx43-T antibody (normalized to immunoprecipitated Cx43; **P<0.01).
We confirmed these findings by immunoprecipitation using a monoclonal Cx43-T antibody. Along with immunoprecipitated Cx43 bands (which were detected by both polyclonal and monoclonal Cx43-T antibodies, and which were distinct from IgG bands), immunoblotting images revealed coimmunoprecipitated PP1 and PP2Ac reactive bands (Figure 4B). IgG alone was the negative (data not shown). Quantitation of the coimmunoprecipitated PP1 and PP2Ac (normalized to immunoprecipitated Cx43 in each individual sample) revealed that the level of PP1 was unchanged, but the level of coimmunoprecipitated PP2Ac was increased 2.5-fold in HF rabbit LV tissue versus controls (n=4 and n=4, respectively; P<0.002; Figure 4C).
We further confirmed the increased colocalization of PP2Ac with Cx43 in HF by using immunofluorescence staining with Cx43-NP and PP2Ac antibodies in control and HF LV myocytes. Figure 5 shows representative confocal images where Cx43-NP isoform (red) is present in distinctive punctate patterns throughout the cell membrane, and it colocalizes with PP2Ac (green, also distributed in cytosol) that is increased 2-fold in HF myocytes (consistent with the immunoprecipitation and cosedimentation data). Similar results were found for double staining using Cx43-T and PP2Ac antibodies (data not shown). PP1 also colocalized with Cx43 at distinct membrane sites, but the amount of colocalized PP1 appeared unchanged in HF (data not shown).
Figure 5. Double immunofluorescence staining of control (left) and HF rabbit (right) isolated LV myocytes with Cx43-NP and PP2Ac antibodies alone (top, red; middle, green), and the overlap (bottom, yellow). Scale bar is 20 μm.
Alteration of Cell-Cell Communication and the Role of PP2A in Accumulation of Dephosphorylated Cx43 in HF
To define alterations of intercellular communication in HF and it modulation by protein phosphatases, we performed Lucifer Yellow dye transfer (microinjection) in the absence or presence of the protein phosphatase inhibitor OA, at a concentration (10 nmol/L) that has been shown to inhibit PP2A, but not PP1 (Figure 6).26,27 To delineate the injected cell of the cell pairs, we also injected the high-molecular-weight (10 000) Rhodamine B dextran, which is retained in the injected myocyte without passing gap junction channels to the recipient cell (Figure 6, bottom left). Figure 6 shows images of the microelectrode and control, HF, and OA-treated HF cell pairs (injected and recipient cells), and sequential confocal images at times 1, 38, and 214 seconds. Figure 7A shows representative time course data for fluorescence (normalized to plateau) in the recipient cell (30x80 μm window; Figure 6, bottom right panel), and Figure 7B shows summarized rate constant (, time to reach 63% peak) data. HF myocytes had decreased cellular coupling, reflected by slower dye transfer with an increased of dye transfer of 53±12 seconds versus 19±5 seconds for control (P<0.05; n=6 and n=5, respectively). However, OA-treated HF cell pairs showed improved dye transfer with a reduced of dye transfer (=24±5; P<0.05 versus HF and P=NS versus control; n=4). Protein phosphatase inhibition by OA (10 nmol/L) was verified by immunoblotting, with OA-treated HF myocytes exhibiting a 42±9.3% decrease in Cx43-NP (Figure 7C; n=3). OA-treated control cell pairs showed no additional improvement in cell coupling (=18±7; n=4; P=NS versus untreated control cell pairs) (Figure 7A).
Figure 6. Lucifer Yellow (LY) microinjection images of end-to-end cell pairs (left) and subsequent LY dye transfer (right, yellow) from a single injected cell to a recipient cell in control, HF, and OA-treated HF myocytes at 1, 38, and 214 seconds (right). A representative 30x80 μm window, which was used to quantitate dye transfer (fluorescence intensity in the recipient myocyte), is shown in the bottom right panel. At bottom left is an example of Rhodamine B dextran (Rh, red) retained in the injected myocyte. Scale bar at lower left=20 μm.
Figure 7. A, Representative fitted single time courses of LY transfer in recipient cell from end-to-end cell pairs of control, HF, and 10 nmol/L OA-treated control and HF myocytes, respectively. B, Summarized data of LY transfer in recipient cell from control, HF, and OA-treated control and HF myocytes (*P<0.05). C, Representative Western blot images of Cx43-NP and GAPDH from myocytes in the absence or presence of OA (10 nmol/L).
In the present study, we studied LV tissue and myocytes from an arrhythmogenic rabbit model of nonischemic HF and from age-matched control rabbits. We found that Cx43 expression was downregulated on both a protein and mRNA level. HF rabbits exhibited a significant enhancement of the nonphosphorylated isoform of Cx43 (and decreased phosphorylated Cx43). All these results were validated in LV tissue from hearts of patients with end-stage IDCM (at the time of transplantation) compared with nonfailing hearts. We also observed that protein phosphatases PP1 and PP2Ac both colocalized with Cx43 in rabbit LV, and there was a 2.5-fold increase in the level of colocalized PP2Ac in HF. Moreover, we found that intercellular coupling between LV myocytes was reduced markedly in HF. However, OA reduced the amount of Cx43-NP and significantly improved cell coupling in HF. Our observations suggest that both reduced Cx43 expression and enhanced Cx43-NP within gap junctions underlie impaired cellular coupling and ultimately altered conduction in nonischemic HF. The increased levels of colocalized PP2Ac (with Cx43) contribute to enhanced Cx43-NP in HF.
Reduced Cx43 Expression and Increased Cx43-NP Is Associated With Altered Cell Coupling and Conduction in Nonischemic HF
Ventricular arrhythmias (VT and VF) account for nearly half of deaths in patients with HF (both nonischemic and ischemic).1 In 3-dimensional cardiac mapping studies, we have previously shown that spontaneously occurring VT in nonischemic HF (both in our HF rabbit model and in IDCM patients) is initiated and maintained by a nonreentrant mechanism such as delayed or early afterdepolarizations.2 However, mapping in human IDCM has revealed altered anisotropic conduction and conduction block that could underlie the development of reentry (especially during the transition from VT to VF).3
In the present study, we found that HF myocytes exhibit reduced cellular coupling. This was associated with decreased Cx43 expression on both an mRNA and protein level, similar to that reported by others.11 Cx43 downregulation may be a common feature of failing heart,9 although the underlying biochemical basis remains to be determined. There has been considerable debate as to the role of Cx43 in modulating conduction.10,29 However, studies in heterozygous Cx43 knockout mice,15 along with those in other transgenic models,30 and in dogs with pacing-induced HF,11 support the contention that reduced Cx43 expression per se could contribute to altered cell-coupling and conduction in nonischemic HF.
Cx43 is a phosphoprotein that is predominantly phosphorylated in the control state. We confirmed this in control rabbit and nonfailing human LV. With HF, we found a significant increase in the Cx43-NP on Western blots using an extensively validated antibody specific for the Cx43-NP isoform.17 Under physiological conditions, the C-terminus of Cx43 is extensively phosphorylated, primarily at serine sites, although recent studies suggest there may be changes in tyrosine phosphorylation in HF.31 The specific Cx43-NP antibody used in the present study recognizes the nonphosphorylated serine-rich segment of the Cx43 C-terminus that includes putative phosphorylation sites for multiple kinases including protein kinase A, protein kinase C, and MAPK.15–17 Our results of Cx43-NP in HF rabbits were consistent with both LV tissue and isolated myocytes. Moreover, findings in human IDCM were nearly identical. Despite the consistency of this finding, interpretation is limited because the extent to which Cx43 is phosphorylated at baseline in rabbit and human LV is not known. However, we took advantage of the Cx43 expression pattern on Western blots and determined the extent of Cx43 phosphorylation in controls and HF. Our results in controls were consistent between rabbits and nonfailing humans (and similar to that noted by others17). The changes in Cx43 phosphorylation with HF were substantive and confirmed the relative changes in Cx43-NP using a specific Cx43-NP antibody.
The Role of Phosphatases in Cx43 Dephosphorylation in Nonischemic HF
Reduced Cx43 phosphorylation in HF could be attributable to reduced phosphorylation by kinases (eg, protein kinase A, protein kinase C) and/or to increased dephosphorylation by protein phosphatases. Here, we focused on the role of phosphatases PP1 and PP2A, the primary phosphatases that dephosphorylate Cx43,14 because levels of total cellular PP1 and total cellular PP1 and PP2A activity are increased in HF.19 Moreover, activation of phosphatases that dephosphorylate Cx43 can reduce communication between cells.14
We reasoned that local levels of an enzyme (rather than total cellular levels) may best reflect its actions, and that PP1 and PP2A might interact directly with Cx43. Cx43 is a part of a protein complex, and it has been reported that kinases and structural proteins have direct interactions with cardiac Cx43.15,20 The present study provides the first evidence of an interaction between Cx43 and PP1 and PP2Ac in rabbit LV. Cx43, PP1, and PP2A form a close complex that could be immunoprecipitated by a Cx43 antibody or sedimented by microcystin-Sepharose beads (that bind PP1 and PP2A). We confirmed this with immunofluorescence staining showing colocalization at gap junctional sites in the membrane. With HF, we found that the level of coimmunoprecipitated PP2Ac with Cx43 was dramatically increased >2.5-fold, whereas colocalized PP1 was unchanged. Moreover, OA, at a concentration that inhibits PP2A, but not PP1,26,27 and that increased the levels of phosphorylated Cx43, significantly improved cellular coupling between HF rabbit myocytes. The rapidity and near-complete recovery of coupling (even in the setting of reduced Cx43 protein levels) suggest that phosphorylation provides a dynamic means of regulation and that there may be a steep relationship between Cx43 phosphorylation state and cellular coupling (that may plateau, as reflected by the unchanged coupling in OA-treated control cell pairs). Whether dephosphorylated Cx43 per se reduces coupling remains to be determined.17
Animal Model of HF
Our HF rabbit model exhibits contractile dysfunction, pathologic alterations, and arrhythmogenesis similar to nonischemic human HF,2,22,28 and we have consistently validated biochemical and electrophysiologic findings in this model with those in nonischemic HF in humans2,22 (including the results of the present study). This suggests that this arrhythmogenic HF rabbit model is ideally suited for studies of altered Cx43 expression, phosphorylation, and function.
Limitations
Our observations strongly suggest that increased levels of colocalized PP2Ac (with Cx43) contribute to enhanced Cx43-NP and reduced cellular coupling in HF LV. Whether increased activity of other protein phosphatases, such as PP1 and calcineurin, or decreased activity of kinases contribute to Cx43 dephosphorylation in HF remains unknown.14
In addition to Cx43 downregulation and altered Cx43 phosphorylation, there could be changes in gap junction distribution with HF that may modulate conduction. Although this was not assessed in the present study, recent studies have demonstrated altered gap junction distribution in human IDCM.33 We focused on alterations in Cx43, by far the predominant ventricular gap junction proteins, so we cannot rule out alterations in the expression of other connexins (eg, Cx45 or Cx40) in HF.
Implications
Our findings in the present study suggest that the combination of Cx43 downregulation and altered phosphorylation contribute to decreased cellular coupling and, ultimately, slow conduction in nonischemic HF (Figure 8). The colocalization of PP1 and PP2Ac with Cx43 indicates that Cx43 represents a macromolecular complex with numerous components, including both kinases and phosphatases. Enhanced levels of colocalized PP2Ac in HF indicate an additional regulation of gap junctional function in HF that is locally controlled.
Figure 8. Schematic diagram showing changes in Cx43 expression, phosphorylation state, and colocalized phosphatases with HF. With HF, the combination of decreased Cx43 expression and enhanced nonphosphorylated Cx43 (arising from increased associated PP2A) contribute to reduced cell-coupling and arrhythmogenesis.
Alterations in Cx43 phosphorylation could modulate connexin channel gating, as well as connexin trafficking and degradation.32 Whereas effects of Cx43 phosphorylation on channel gating are complex (dependent on specific kinases and phosphorylation sites), decreased Cx43 phosphorylation has been shown to be associated with decreased gap junctional conductance, whether assessed in vitro in cell pairs with activation of protein phosphatases14 or in vivo in the ischemic heart.17 Heterogeneity of these changes in HF would be expected to further enhance the effects on conduction. Moreover, if Cx43 dephosphorylation enhances gap junction degradation,32 the degree of Cx43 dephosphorylation in HF could be underestimated, and may even contribute in part to the decreased Cx43 protein levels.
Our findings, including the improved cellular coupling with OA in HF, suggest that modulation of Cx43 phosphorylation state and Cx43-associated proteins (eg, phosphatases) could represent novel therapeutic approaches to improve conduction in the failing heart.
References
Packer M. Sudden unexpected death in patients with congestive heart failure: a second frontier. Circulation. 1985; 72: 681–685.
Pogwizd SM. Nonreentrant mechanisms underlying spontaneous ventricular arrhythmias in a model of nonischemic heart failure in rabbits. Circulation. 1995; 92: 1034–1048.
Anderson KP, Walker R, Urie P, Ershler PR, Lux RL, Karwandee SV. Myocardial electrical propagation in patients with idiopathic dilated cardiomyopathy. J Clin Invest. 1993; 92: 122–140.
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: 727–741.
Vermeulen JT, McGuire MA, Opthof T, Coronel R, de Bakker JMT, Klopping C, Janse MJ. Triggered activity and automaticity in ventricular trabeculae of failing human and rabbit hearts. Cardiovasc Res. 1994; 28: 1547–1554.
De Mello WC. Renin-angiotensin system and cell communication in the failing heart. Hypertension. 1996; 27: 1267–1272.
Kumar NM, Gilula NB. The gap junction communication channel. Cell. 1996; 84: 381–388.
Saffitz JE, Davis LM, Darrow BJ, Kanter HL, Laing JG, Beyer EC. The molecular basis of anisotropy: role of gap junctions. J Cardiovasc Electrophysiol. 1995; 6: 498–510.
Wang XJ, Gerdes AM. Chronic pressure overload cardiac hypertrophy and failure in guinea pigs: III. Intercalated disc remodeling. J Mol Cell Cardiol. 1999; 31: 333–343.
Guerrerro PA, Schuessler RB, Davis LM, Beyer EC, Johnson CM, Yamada KA, Saffitz JE. Slow ventricular conduction in mice heterozygous for a connexin43 null mutation. J Clin Invest. 1997; 99: 1991–1998.
Poelzing S, Rosenbaum DS. Altered connexin43 expression produces arrhythmia substrate in heart failure. Am J Physiol Heart Circ Physiol. 2004; 287: H1762–H1770.
Lerner DL, Yamada KA, Schuessler RB, Saffitz JE. Accelerated onset and increased incidence of ventricular arrhythmias induced by ischemia in Cx43-deficient mice. Circulation. 2000; 101: 547–552.
Kwak BR, Jongsma H. Regulation of cardiac gap junction channel permeability and conductance by several phosphorylating conditions. Mol Cell Biochem. 1996; 157: 93–99.
Duthe F, Plaisance I, Sarrouilhe D, Herve JC. Endogenous protein phosphatase 1 runs down gap junctional communication of rat ventricular myocytes. Am J Physiol Cell Physiol. 2001; 281: C1648–C1656.
Bowling N, Huang X, Sandusky GE, Fouts RL, Mintze K, Esterman M, Allen PD, Maddi R, McCall E, Vlahos CJ. Protein kinase C- and - modulate connexin43 phosphorylation in human heart. J Mol Cell Cardiol. 2001; 33: 789–798.
Darrow BJ, Fast VG, Kleber AG, Beyer EC, Saffitz JE. Functional and structural assessment of intercellular communication. Increased conduction velocity and enhanced connexin expression in dibutyryl cAMP-treated cultured cardiac myocytes. Circ Res. 1996; 79: 174–183.
Beardslee MA, Lerner DL, Tadros PN, Laing JG, Beyer EC, Yamada KA, Kleber AG, Schuessler RB, Saffitz JE. Dephosphorylation and intracellular redistribution of ventricular connexins 43 during electrical uncoupling induced by ischemia. Circ Res. 2000; 87: 656–662.
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.
Neumann J, Eschenhagen T, Jones LR, Linck B, Schmitz W, Scholz H, Zimmermann N. Increased expression of cardiac phosphatases in patients with end-stage heart failure. J Mol Cell Cardiol. 1997; 29: 265–272.
Barker RJ, Price RL, Gourdie RG. Increased association of ZO-1 with connexin43 during remodeling of cardiac gap junctions. Circ Res. 2002; 90: 317–324.
Marx SO, Reiken S, Hisamatsu Y, Jayaraman T, Burkhoff D, Rosemblit N, Marks AR. PKA phosphorylation dissociates FKBP12.6 from the calcium release channel: defective regulation in failing hearts. Cell. 2000; 101: 365–376.
Pogwizd SM, Qi M, Yuan W, Samarel AM, Bers DM. Upregulation of Na+/Ca2+ exchanger expression and function in an arrhythmogenic rabbit model of heart failure. Circ Res. 1999; 85: 1009–1019.
Srivastava RA, Schonfeld G. Use of riboprobes for Northern blotting analysis. BioTechniques. 1991; 11: 584–588.
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K. Short protocol in molecular biology. 4th ed. New York: John Wiley and Sons; 1999: 90–91.
Ruiz-Meana M, Garcia-Dorado D, Hofstaetter B, Piper HM, Soler-Soler J. Propagation of cardiomyocyte hypercontracture by passage of Na+ through gap junctions. Circ Res. 1999; 85: 280–287.
Herzig S, Neuman J. Effects of serine/threonine protein phosphatases on ion channels in excitable membranes. Physiol Rev. 2000; 80: 173–210.
Bialojan C, Takai A. Inhibitory effect of a marine-sponge toxin, okadaic acid, on protein phosphatases. Specificity and kinetics. Biochem J. 1988; 256: 283–290.
Pogwizd SM, Schlotthauer K, Li L, Yuan W, Bers DM. Arrhythmogenesis and contractile dysfunction in heart failure: Roles of sodium-calcium exchange, inward rectifier potassium current, and residual beta-adrenergic responsiveness. Circ Res. 2001; 88: 1159–1167.
Morley GE, Vaidya D, Samie FH, Lo C, Delmar M, Jalife J. Characterization of conduction in the ventricles of normal and heterozygous Cx43 knockout mice using optical mapping. J Cardiovasc Electrophysiol. 1999; 10: 1361–1375.
Gutstein DE, Morley GE, Tamaddon H, Vaiday D, Schneider MD, Chen J, Chien KR, Stuhlmann H, Fishman GI. Conduction slowing and sudden arrhythmic death in mice with cardiac-restricted inactivation of connexin43. Circ Res. 2001; 88: 333–339.
Toyofuku T, Yabuki M, Otsu K, Kuzuya T, Tada M, Hori M. Functional role of c-Src in gap junctions of the cardiomyopathic heart. Circ Res. 1999; 85: 672–681.
Musil LS, Goodenough DA. Biochemical analysis of connexin43 intracellular transport, phosphorylation, and assembly into gap junctional plaques. J Cell Biol. 1991; 115: 1357–1374.
Kostin S, Rieger M, Dammer S, Hein S, Richter M, Klovekorn WP, Bauer ER, Schaper J. Gap junction remodeling and altered connexin43 expression in the failing human heart. Mol Cell Biochem. 2003; 242: 135–144.(Xun Ai, Steven M. Pogwizd)