The ε Subtype of Protein Kinase C Is Required for Cardiomyocyte Connexin-43 Phosphorylation

Gap junctions (GJs) are protein channels found at the juxtaposition of the cell membranes of 2 adjacent cells.¹ They serve as conduits, permitting the exchange of small molecules (<1 kDa), such as ions and second messengers between cells. In excitable cells such as cardiomyocytes, GJs allow action potentials to pass unhindered between coupled cells and ensure the coordinated action of the cardiac pump. GJs also mediate metabolic coupling, allowing passage of signals that regulate growth and differentiation. GJs are composed of connexins that assemble into hexameric channels known as connexons. Connexons from adjacent cells dock together to form a continuous channel linking the cytoplasms of both cells. Connexins are members of a multigene family, showing cell- and tissue-specific distribution. The predominant connexin in cardiac ventricular working cardiomyocytes is connexin-43 (Cx43).¹²³

Cx43 is phosphorylated on serine and threonine residues, although tyrosine phosphorylation is also found in some instances.¹⁴ Phosphorylation of Cx43 is thought to be required for proper synthesis and assembly of connexins into GJs⁵⁶ and regulates electric conductance and permeability to small molecules.⁷ Although tyrosine phosphorylation of Cx43 has been linked to decreased gap junctional intercellular communication (GJIC),¹ phosphorylation of Cx43 on serine has been associated with both increased and decreased GJIC. Of the various signaling cascades involving serine/threonine kinases, increases in cAMP and subsequent activation of protein kinase A stimulate GJIC,⁸⁹¹⁰ whereas activation of mitogen-activated protein kinase (MAPK) or protein kinase C (PKC) has been associated with decreased GJIC.¹¹¹²¹³ The MAPKs are serine/threonine kinases activated downstream of the ras-raf pathway by many mitogens and other effectors.¹⁴ MAPK is proposed to be directly responsible for increasing the phosphorylation of Cx43 on serine, leading to inhibition of communication in epithelial cell lines.¹¹

PKC is a family of several (at least 12) phospholipid-dependent serine/threonine protein kinases.¹⁵ The PKC family has been implicated as an intracellular mediator of several neurotransmitters, hormones, tumor promoters, α1-adrenergic agonists, and phorbol esters, and it is important in the regulation of growth, differentiation, cell death, and neurotransmission.¹⁵ The following 3 PKC subfamilies have been described: conventional PKC (PKCα, -β1, -β2, and -γ; activated by Ca²⁺ and phorbol esters), novel PKC (PKC-δ, -ε, and -η; not regulated by Ca²⁺), and atypical PKC (not activated by Ca²⁺ or phorbol esters). Different isoforms may perform distinct functions, as suggested by their differential pattern of localization, differences in condition of activation, and some differences in substrate specificity.¹⁵¹⁶

We have previously shown that stimulation of neonatal rat cardiomyocytes by the mitogen fibroblast growth factor (FGF)-2 decreases metabolic coupling and increases the phosphorylation of Cx43 on serine.¹⁷ Binding of FGF-2 to its tyrosine kinase receptor(s) initiates a signal transduction cascade that has been shown to activate both the MAPK and PKC pathways in cardiomyocytes.¹⁸¹⁹ In this work, by using specific inhibitors, we have established that the PKC but not the MAPK pathway is involved in the FGF-2–induced effects on cardiomyocyte Cx43. In addition, we provide evidence consistent with the notion that the PKCε isoform is directly involved in mediating the FGF-induced cardiomyocyte Cx43 phosphorylation.

Materials and Methods

Materials

We obtained recombinant human FGF-2 from Upstate Biotechnology; chelerythrine from Research Biochemicals International; PD98059 from New England Biolabs; cell culture media DMEM/F12, F10, and trypsin from GIBCO/BRL; FBS from Hyclone; and all other chemicals from Sigma. Recombinant rat FGF-2 was produced as described.¹⁹

Primary Antibodies

Polyclonal rabbit antibodies to PKCε and PKCα and the corresponding immunizing peptides were obtained from Santa Cruz Biotechnology, monoclonal mouse antibodies to Cx43 and PKCε from Transduction Laboratories, and polyclonal rabbit antibody to dually phosphorylated MAPK (anti-ACTIVE MAPK) from Promega.

Rabbit Anti-Cx43 Antibodies

A keyhole limpet hemocyanin–conjugated peptide corresponding to residues 368 to 382 of the carboxyl terminus of Cx43 was synthesized by Quality Controlled Biochemicals and used to produce a rabbit antiserum to Cx43. It was used at 1:50 000 or 1:5000 dilution for Western blotting or immunofluorescence, respectively.

Culture

Neonatal rat cardiomyocyte cultures were obtained as described.¹⁷ Cells, plated at 6 to 8×10⁵ per 35-mm dish, were maintained for 6 days in maintenance medium (0.5% FBS–DMEM/F12, containing 20 nmol/L selenium, 10 μg/mL insulin, 10 μg/mL transferrin, and 2 mg/mL BSA), and the medium was replaced every 48 hours. Two hours before treatments, the medium was changed to fresh maintenance medium minus serum.

Cx43 Phosphorylation

Cx43 phosphorylation was performed as described.¹⁷ Chelerythrine and calphostin-C (both at 1 μmol/L) were added to the cells 15 minutes before the addition of FGF-2. PD98059 (50 μmol/L) was added to cells 1 hour before FGF-2. Myocytes were stimulated with FGF-2 (10 ng/mL) for 15 minutes at 37°C. Subsequently, cells were lysed as described.¹⁷ Protein content was determined using the bicinchoninic acid assay. Each sample (100 μg) in modified radioimmunoprecipitation buffer (1% NP-40, 0.1% SDS, 0.25% deoxycholate, and 150 mmol/L NaCl) was immunoprecipitated using 2 μL of rabbit anti-Cx43 serum; detection of ³²P-labeled Cx43 was as described.¹⁷

Scrape Loading and Immunofluorescence

Scrape loading and immunofluorescence were performed as described.¹⁷

Cx43/PKCε Coimmunoprecipitation

Myocytes kept in 0.5% FBS in DMEM/F12 medium for 6 days were treated with recombinant rat FGF-2 or vehicle. Myocyte lysates in cold coimmunoprecipitation buffer (containing 1% NP-40; 10% glycerol; in mmol/L, HEPES [pH 7.5] 50, NaCl 100, EDTA 1, β-glycerophosphate 20, NaF 10, sodium orthovanadate 1, and PMSF 1; and 2 μg/mL each of leupeptin, pepstatin, aprotinin, and E-64) were processed for immunoprecipitation with anti-Cx43 antibodies.¹⁷ Immunoprecipitated protein was analyzed by Western blotting for immunoreaction with anti-PKCε or anti-PKCα antibodies (1:2000 dilution). Antigen-antibody complexes were visualized by an enhanced chemiluminescence reaction (SuperSignal kit; Pierce).

Transfection With PKCε_(1-401)

We used a cDNA plasmid lacking the catalytic domain of murine PKCε_(1-401) in the pSVK3 vector, described by Cai et al,²⁰ and a modified calcium phosphate transfection method.²¹

Infection With Ad.εPKC(DN)

A dominant-negative mutant of εPKC(DN) was obtained through site-directed mutagenesis of the rabbit PKCε.²² Cardiomyocytes were infected with Ad.εPKC(DN) or with non-εPKC-DN–expressing virus at a multiplicity of infection of 50.

An expanded Materials and Methods section is available online at http://www.circresaha.org.

Results

Effects of FGF-2 on GJs Are Not Mediated by MAPK

To assess the activation status of MAPK in cardiac myocytes, we used antibodies that specifically recognize the dually phosphorylated (and thus activated) species of MAPK in Western blots of cell lysates. As shown in Figure 1A, untreated control cardiomyocytes had barely detectable levels of the activated MAPK. FGF-2 stimulated dual phosphorylation, and thus, activation, of the 42- and 44-kDa MAPK in cardiac myocytes, as shown before.¹⁸¹⁹ Pretreatment with the mitogen-activated protein kinase/extracellular signal–regulated kinase-1 (MEK1) inhibitor PD98059 (50 μmol/L) completely blocked the FGF-2 activation of MAPK.

Having established that FGF-2 does stimulate MAPK and that we can block this activation fully with the MEK1 inhibitor, we proceeded to examine whether MAPK activation mediated the FGF-2–induced changes in the phosphorylation of Cx43. As can be seen in Figure 1B, complete MAPK inhibition did not prevent the increase in Cx43 phosphorylation induced by FGF-2.

We then examined whether MAPK mediated the effects of FGF-2 on GJs. When confluent cardiomyocytes were loaded with the dye 6-CF under control conditions, dye movement from the primary-loaded cells through adjacent cells in the monolayer was clearly evident (Figure 1C-i; “maximal” dye coupling). As we showed before, FGF-2 induced a clear decrease in dye migration between myocytes (Figure 1C-iii; “minimal” coupling). PD98059 pretreatment did not affect maximal or minimal dye migration (Figure 1C-ii and 1C-iv), indicating that MAPK activation by FGF-2 was not affecting the GJ-mediated dye transfer. Combined semiquantitative data on dye migration from 3 separate experiments are shown in Figure 1D.

Effects of FGF-2 on GJs Are Mediated by PKC

Cardiac myocytes were pretreated with the specific PKC inhibitor chelerythrine (1 μmol/L) 15 minutes before stimulation with FGF-2 for 15 minutes. As seen in Figure 2A, this had no effect on basal Cx43 phosphorylation, but it did completely block the increase in Cx43 phosphorylation induced by FGF-2. We also used calphostin C as a specific PKC inhibitor²³ ; at 1 μmol/L, calphostin C pretreatment for 15 minutes reduced baseline levels, as well as FGF-2–induced levels of Cx43 phosphorylation (Figure 2B). Chelerythrine pretreatment did not prevent the MAPK activation induced by FGF-2 (Figure 2C). Chelerythrine and calphostin appeared to have different effects on the baseline levels of Cx43 phosphorylation, possibly as a result of multiple actions of these inhibitors.

In scrape-loading assays, chelerythrine did not affect the migration of 6-CF dye between myocytes compared with controls (Figure 2D). The FGF-2 decrease in dye migration (Figure 2D-iii) was completely prevented when FGF-2 was added in the presence of chelerythrine (Figure 2D-iv). Combined semiquantitative data of dye migration from 3 separate experiments are shown in Figure 2E.

Immunolocalization of PKCε and Cx43 Before and After Stimulation

In quiescent cardiomyocytes, PKCε produced an interrupted pattern of localization to areas of intercellular contact (Figure 3A). This localization pattern was very similar to that of cadherin (B.W.D., E.K., unpublished observations, 1999); cadherin is an intercellular attachment protein and a component of adherens junctions and desmosomes, all of which are present at intercalated disks.²⁴ In FGF-2–stimulated myocytes, the staining pattern of PKCε appeared more continuous in areas of cell-cell contact (Figure 3D).

Under control, unstimulated conditions, Cx43 staining with a monoclonal antibody produced the familiar discontinuous punctate pattern characteristic of GJs. Areas of apparent colocalization with PKCε were present, producing the yellow color seen in Figure 3C, as were areas of apparently exclusive localization of either PKCε (staining green) or Cx43 (staining red). A representative image is shown in Figure 3C. On stimulation with FGF-2, extensive colocalization of Cx43 and PKCε was observed, as might be predicted by the more extensive localization of PKCε to sites of cell-cell contact shown above (Figure 3C). A typical image of PKCε and Cx43 colocalization on FGF-2 treatment is shown in Figure 3F.

Treatment of cardiomyocytes with 10 nmol/L phorbol 12-myristate 13-acetate (PMA) for 15 minutes produced changes in PKCε localization that were identical to those induced by FGF-2 (Figure 4). Again, the discontinuous pattern of staining for anti-PKCε was prevalent in the controls (A), whereas PMA-treated myocytes displayed mostly an intense, continuous if irregular staining along cell contact sites (B). Incidence of areas staining only for Cx43 appeared higher in control, compared with PMA-treated, myocytes (C).

Coimmunoprecipitation of Cx43 and PKCε

To explore the possibility of a direct interaction between Cx43 and PKCε, we used anti-Cx43–specific antibodies to immunoprecipitate Cx43 and interacting protein(s) from control and stimulated cell lysates. Typical results are shown in Figure 5. Western blot analysis of immunoprecipitated samples revealed the presence of an anti-PKCε reactive band migrating at 90 kDa (Figure 5A); the only other bands detected by the anti-PKCε antibody were those of the rabbit IgG (as expected). In contrast, an anti-PKCα antibody preparation did not detect any protein band at ≈82 kDa (corresponding to the expected size for PKCα), even after very prolonged exposure; under these conditions the anti-PKCα antibody did eventually detect some bands, presumably nonspecifically, the intensity of which did not change (Figure 5B). Levels of the 90-kDa PKCε were significantly elevated in immunoprecipitates from FGF-2–treated samples compared with control samples (Figure 5C, n=4).

Effects of Inhibition of PKCε on Cx43 Phosphorylation

Cardiac myocytes were transfected with a truncated form of murine PKCε, PKCε_(1-401), which was shown previously to act as a dominant-negative inhibitor of PKCε.²⁰ Gene transfer was achieved using a modified calcium-phosphate protocol,²¹ resulting in nearly 20% transfection efficiency of myocytes (B.W.D., E.K., unpublished observations, 1998). Expression of the truncated PKCε in transfected cells was assessed by Western blotting with an antibody raised against the N terminus of PKCε and shown in Figure 6A. Native PKCε levels remain the same in cells transfected with vector or with PKCε_(1-401) (Figure 6A). Levels of Cx43 phosphorylation, assessed by immunoprecipitation of ³²P-labeled Cx43 and corrected for variations in total Cx43 present in the samples, were significantly decreased in cultures expressing the PKCε_(1-401) compared with vector-transfected controls (Figure 6B and 6C).

Cardiac myocytes were infected with a cDNA coding for a mutated, dominant-negative form of PKCε, Ad.εPKC(DN), using an adenoviral vector, as described.²² The mutated PKCε is still capable of binding to the membrane but cannot phosphorylate its target(s).²² Abundant expression of εPKC(DN) was detected by Western blotting, using conditions of enhanced chemiluminescence reagent development that do not detect the endogenous PKCε (Figure 7A). Longer exposure resulted in detection of the endogenous isoform (as in Figure 6A), whereas the lane containing lysates from Ad.εPKC(DN)–expressing myocytes appeared completely black. Levels of ³²P-labeled Cx43 decreased sharply in cultures expressing the Ad.εPKC(DN) compared with controls treated with the virus alone (Figure 7B and 7C).

The adenoviral vector allowed for high transfection efficiency (>95%) of cardiac myocytes, determined by staining transfected myocytes for PKCε with a monoclonal antibody preparation. As shown in Figure 8A, the antibody reacts very faintly with endogenous PKCε in vector-infected myocytes; however, it clearly stained the cytosol and intercellular contact sites of εPKC(DN)-infected cells (Figure 8B and 8C). Figure 8B was photographed under identical exposure conditions as in Figure 8A to illustrate differences in staining intensity. Figure 8C was underexposed to allow a better distinction between cytosolic and membrane-associated staining. Simultaneous staining for nuclei is also shown in Figure 8C.

Discussion

Our objective was to determine which signal transduction pathway(s) is involved in the FGF-2–mediated Cx43 phosphorylation in cardiomyocytes. Novel findings here are that (1) MAPK activation is apparently not required, (2) PKC mediates the FGF-2–induced Cx43 phosphorylation, (3) PKCε and Cx43 colocalize to areas of intermyocyte contact, (4) there is an association between PKCε and Cx43 that is enhanced by FGF-2, and (5) PKCε is required for the phosphorylation of Cx43.

Cx43 phosphorylation is an important modulator of GJIC and affects conductance, metabolic coupling, growth, and differentiation.¹²⁵ Multiple phosphorylation sites, potential targets of several groups of kinases, suggest that Cx43 may serve as a discriminating sensor of change¹²⁵ in the cellular environment. Identification of the kinase(s) and phosphatase(s) involved directly in altering the phosphorylation status of Cx43 may provide a way of interfering in situations such as arrhythmias,²⁶ uncontrolled growth²⁷ in the case of cancer cells, or inability to regenerate in terminally differentiated cells such as cardiomyocytes.

FGF-2 triggers signaling from tyrosine kinase plasma membrane receptors, which are also present in the cardiomyocyte and are known to activate the phosphoinositide pathway leading to PKC stimulation, as well as the ras-raf-MAPK pathway.²⁸ We have confirmed that FGF-2 activates both MAPK and PKC signaling in the cardiomyocyte.¹⁹ Both MAPK and PKC enzymes have been implicated in mediating Cx43 phosphorylation.¹¹¹²¹³ Thus, we used specific inhibitors of these pathways to determine their participation in the FGF-2 effects on cardiac GJs. The inhibitor PD98059, which completely blocked the FGF-2–induced dual phosphorylation, and thus, presumably the activation, of MAPK, failed to prevent the FGF-2–induced Cx43 phosphorylation and the effects of this factor on dye migration. Furthermore, MAPK remained active when the FGF-2 effects on Cx43 phosphorylation and dye coupling were blocked. Overall, our data did not indicate an obvious involvement of the MAPK pathway on FGF-2–induced effects on Cx43 phosphorylation and dye-transfer assessed by the technique of scrape loading. Our findings differ from those of Warn-Cramer et al¹¹ and Kanemitsu and Lau,²⁹ who have presented evidence that MAPK mediates the epidermal growth factor–triggered increases in Cx43 phosphorylation in a mouse fibroblast embryonic cell line. In yet another cell system, the TF1B rat liver epithelial cells, platelet-derived growth factor disruption of GJIC required both PKC and MAPK activation.²⁸ Thus, it is likely that Cx43 phosphorylation and regulation of GJIC by growth factors recruits different signal transduction pathways, in a growth factor–specific as well as cell type–specific manner. In agreement, we did not detect any effects of epidermal growth factor on cardiomyocyte Cx43 phosphorylation (B.W.D., E.K., unpublished observations, 1999).

We then investigated the involvement of the PKC pathway. Both chelerythrine and calphostin C prevented the FGF-2–induced changes in Cx43 phosphorylation, indicating that PKC is necessary for this event to occur. Chelerythrine, furthermore, abolished the effect of FGF-2 of dye migration, indicating that PKC mediated the FGF-2 decrease in cardiomyocyte dye coupling, and strengthening the link between effects on connexin phosphorylation and GJ permeability. Our data are in agreement with several previous reports that have shown that PKC decreases coupling and increases Cx43 phosphorylation.^12133031

Cx43 is an integral plasma membrane protein. We investigated the possibility that PKC, which translocates to membrane sites on activation, may be directly involved in the phosphorylation of Cx43. This notion is supported by the fact that the cytosolic carboxyl-terminal half of the Cx43 molecule has several serines that could serve as PKC substrates.³² The PKC family has, however, at least 12 members; it would thus be important to identify which isoform(s) is involved in Cx43 phosphorylation. Colocalization of the various PKC isoforms with their putative substrates is proposed to ensure their preferential and rapid phosphorylation on activation, and there is evidence suggesting that PKC binds to specific anchoring proteins, such as the RACKS (receptors for activated C kinase), located at various subcellular sites.¹⁶ Subcellular localization of the PKCs could help identify their physiologically relevant substrates; conversely, if Cx43 is a direct substrate for a particular PKC, one would expect colocalization of that isoform with Cx43. The calcium-independent PKCε has been reported to localize to intercalated disk–like sites on stimulation with PMA.³³ PKCε is stimulated by FGF-1³⁴ and FGF-2¹⁹ ; FGF-2 induced increased association of PKCε with the cardiac myocyte membrane fraction in neonatal and adult cardiomyocytes.¹⁹³⁵ We thus considered PKCε as a potential candidate for mediating Cx43 phosphorylation.

We detected PKCε in association with cardiomyocyte plasma membrane and cell-cell contact sites irrespectively of stimulation, but we also saw a change from an interrupted pattern of staining in control cells to a more continuous pattern on stimulation with FGF-2 or PMA. Our findings suggest that a fraction of PKCε is localized to the plasma membrane under all conditions, in contrast to Disatnik et al,³³ who detected PKCε at intercalated disks only in stimulated cells. It is possible that this reflects differences in culture conditions and in “baseline” levels of PKC activation. Our immunofluorescence findings have been confirmed by Western blotting analysis of cardiac sarcolemmal membranes from nonstimulated adult hearts and neonatal myocytes.¹⁹³⁵ Furthermore, as discussed below, PKCε coprecipitated with Cx43 irrespectively of stimulation, a finding reinforcing the validity of our immunofluorescence studies.

The continuous pattern of PKCε localization at intercellular contact sites on stimulation, in conjunction with the increased association of PKCε with membranes detected by Western blotting,³⁵ would suggest that additional PKCε is translocated to previously unoccupied plasma membrane sites. It is also possible that some redistribution of membrane-associated PKCε may occur on stimulation. FGF-2 and PMA induced the same qualitative changes in the distribution of membrane PKCε, in agreement with the notion that FGF-2 stimulates this PKC isoform in cardiac myocytes.

The colocalization and coimmunoprecipitation studies point to an interaction between PKCε and Cx43 at cell-cell contact sites, an interaction that becomes more extensive in stimulated cells. Interaction between a serine-threonine kinase and a potential substrate under activation conditions known to result in increased substrate phosphorylation offers strong support to the notion that PKCε can phosphorylate Cx43 on activation directly. It is also possible that the PKC-mediated decrease in the permeability of Cx43 GJs to dyes may be a consequence not only of a potential conformational change caused by phosphorylation, but also of the physical association between the 2 proteins, resulting in obstruction of the pore. Work from Calero et al,³⁶ Homma et al,³⁷ and Stergiopoulos et al³⁸ has strongly pointed to the possibility of such interactions mediating GJs permeability, although very few candidate proteins have been identified up until now.

To examine whether there is a cause-and-effect relationship between PKCε and Cx43 phosphorylation in the intact myocyte, we used expression of a dominant-negative truncated PKCε.²⁰ In cultures expressing the PKCε_(1-401), a significant reduction of Cx43 phosphorylation was achieved, despite the relatively low transfection efficiency of cardiomyocytes under these conditions. It is possible, however, that the truncated PKCε, lacking the catalytic carboxyl terminus, may not be very effective in targeting plasma membrane proteins such as Cx43.³⁹ We therefore used expression also of a mutated PKCε that retains its ability for membrane localization and has been shown to act very effectively in a dominant-negative fashion²² ; adenoviral infection ensured high levels of expression in virtually all myocytes. Under these conditions, we were able to show a dramatic decrease in Cx43 phosphorylation in cultures expressing the dominant-negative PKCε, demonstrating that active PKCε is required for Cx43 phosphorylation.

Our data offer additional support to the notion that individual PKC isoforms possess distinct biological functions. PKCα, another PKC isoform that (1) is activated by FGF-2, (2) is present in myocytes, and (3) has been localized to plasma membrane sites,³³ did not coprecipitate with Cx43, a finding indicating selectivity of interaction between the PKCε species and Cx43. In addition, expression of a truncated, dominant-negative form of PKCα³⁵ had no effect on Cx43 phosphorylation (B.W.D., E.K., unpublished observations, 1999).

PKCε has been implicated in contractility, cardioprotection, and preconditioning. There has, however, been no information up until now as to what the potential targets of PKCε may be. Evidence presented in this article points to Cx43, a plasma membrane GJ channel–forming protein, as a target for PKCε. It is possible that the above PKCε-mediated processes of cardioprotection and/or contractility may require alterations in Cx43 and channel pore permeability. It is of interest that anesthetics are cardioprotective and also cause decreased GJIC.⁴⁰

Finally, whereas PKCε is linked to stimulation of proliferation,⁴¹ Cx43 is regarded as a growth suppressor.⁴² FGF-2 is a well-characterized mitogen. It is logical, therefore, to speculate that the FGF-2–induced proliferation of cardiomyocytes requires involvement of PKCε, increased Cx43 phosphorylation, and decreased intercellular coupling.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Footnotes