Regulation of Epidermal Growth Factor-induced Connexin 43 Gap Junction Communication by Big Mitogen-activated Protein Kinase 1/ERK5 but Not ERK1/2 Kinase Activation*

The gap junction protein, Cx43, plays a pivotal role in coupling cells electrically and metabolically, and the putative phosphorylation sites that modulate its function are reflected as changes in gap junction communication. Growth factor stimulation has been correlated with a decrease in gap junction communication and a parallel activation of ERK1/2; the inhibition of epidermal growth factor (EGF)-induced Cx43 gap junction uncoupling was observed by using the MEK1/2 inhibitor, PD98059. Because 1) BMK1/ERK5, another MAPK family member also activated by growth factors, possesses a phosphorylation motif similar to ERK1/2, and 2) it has been reported that PD98059 can inhibit not only MEK1/2-ERK1/2 but also MEK5-BMK1 activation, we investigated whether BMK1 can regulate EGF-induced Cx43 gap junction uncoupling and phosphorylation, comparing this to the role of ERK1/2 on Cx43 function and phosphorylation induced by EGF. Selective activation or inactivation of ERK1/2 by using a constitutively active form or a dominant negative form of MEK1 did not regulate Cx43 gap junction coupling. In contrast, we found that BMK1, selectively activated by constitutively active MEK5α, induced gap junction uncoupling, and the inhibition of BMK1 activation by transfection of dominant negative BMK1 prevented EGF-induced gap junction uncoupling. Activated BMK1 selectively phosphorylates Cx43 on Ser-255 in vitro and in vivo, but not on S279/S282, which are reported as the consensus phosphorylation sites for MAPK. Furthermore, by co-immunoprecipitation, we found that BMK1 directly associates with Cx43 in vivo. These data indicate that BMK1 is more important than ERK1/2 in EGF-mediated Cx43 gap junction uncoupling by association and Cx43 Ser- 255 phosphorylation.

The gap junction protein connexin 43 (Cx43) 1 is the main conductor of intercellular current in ventricular myocytes (1). Gap junctions also enable the cytoplasm of individual cells to communicate directly and allow exchange of nutrients, ions, metabolites, and small molecules up to 1000 Da. Diverse biological processes including cardiac function, cellular growth, propagation of calcium waves, and embryogenesis are regulated by gap junctional communication (GJC). It is proposed that regulatory signaling molecules for these processes are exchanged through gap junctions, and thus, GJC directly impacts intracellular events (2).
Gap junctional coupling can be regulated by post-translational phosphorylation of Cx43 (3). Phosphorylation of Cx43 by various protein kinases regulates the assembly of gap junctions to form functional connexons in the plasma membrane (2), connexin redirection from the plasma membrane (3), and also alters channel open probability (5). For example, EGF stimulates the rapid and transient disruption of GJC, and a marked increase in the phosphorylation of Cx43, not on tyrosine, but on serine255, serine279, and serine282 (4). EGF activation of ERK1/2, a member of the mitogen-activated protein kinase (MAPK) family, is thought to mediate Cx43 phosphorylation. This conclusion was supported by the observation that several copies of the ERK1/2 consensus phosphorylation motif are present in the cytoplasmic C-terminal tail of Cx43 and also because of the sensitivity of GJC uncoupling to the ERK1/2 inhibitor, PD98059 (4).
In the present study, we examined the specific role of BMK1 in EGF-induced phosphorylation of Cx43 and inhibition of GJC by using specific molecular reagents to manipulate the ERK1/2 or the BMK1 pathways independently in Cx43 heterologously expressed in human embryonic kidney (HEK) 293 cells. Our data demonstrate that selective activation of BMK1 but not ERK1/2 promotes gap junction uncoupling in the absence of EGF. Additionally, a dominant negative form of BMK1, but not a dominant negative form of MEK1, prevents EGF-induced inhibition of GJC. In support of this, we also found that BMK1 binds to Cx43 and phosphorylates preferentially on Ser-255 and a dominant negative form of BMK1 attenuates this phosphorylation in vivo. These data demonstrate that BMK1 kinase activity alone is both a necessary and sufficient component in the mediation of EGF-induced Cx43 Ser-255 phosphorylation and subsequent inhibition of GJC.
Cell Culture and Transfection-HEK293 cells (ATCC, Manassas, VA) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Cells seeded at a density of 5 ϫ 10 5 cells/35-mm dish were transiently transfected with the indicated plasmids using LipofectAMINE Plus (Invitrogen) following the manufacturer's recommended protocol. The total plasmid amount was adjusted to 1.5 g/dish for single-plasmid transfections or 3.0 g/dish (35 mm dishes) or 10 g/dish (100-mm dishes) for co-transfection with multiple plasmids. Blank pCI/neo vector (Promega, Madison, WI) was used where appropriate. Experiments were carried out 24 h post-transfection. A Cx43E stable cell line was selected with Geneticin (1 mg/ml, Invitrogen), and stable expression of the transgene was confirmed by immunoblotting for Cx43.
Fluorescence Recovery after Photobleaching (FRAP) Assay for Gap Junction Coupling-HEK293 cells were grown on glass coverslips to 70% confluence and loaded with 10 M calcein blue-AM (CBAM, Molecular Probes, Eugene, OR) in external solution (in mM: 140 NaCl, 3 CaCl 2 , 1 MgCl 2 , 2.8 KCl, 10 HEPES buffer, pH 7.4) by incubating the cells with the solution at 37°C for 10 min. Calcein blue-AM (molecular mass, 465 daltons; excitation 360 nm and emission 440 nm) was used as the intercellular tracer to minimize interference from the GFP and DsRed fluorescent tags. After loading, the cells were washed and maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum at room temperature. The FRAP assay, described elsewhere (13), was performed on cells visualized under a Nikon Diaphot inverted epifluorescent microscope equipped with a Noran OZ (Noran Instruments, Middleton, WI) confocal apparatus. The recovery phase of the normalized fluorescence intensity plots, taking into account background photobleaching, was fitted with a monoexponential function F(t) ϭ F 0 (1 Ϫexp Ϫt/rec ) using Origin, version 6.0 (Origin Laboratories, Northampton, MA). F(t) is the normalized fluorescence at time t, F 0 is the initial pre-bleach fluorescence intensity, and rec is the mono-exponential recovery time constant. The average rec is given as mean Ϯ S.E., and statistical analysis was performed with the StatView 4.0 package (Abacus Concepts, Berkeley, CA).

MALDI-TOF Mass Spectrometry Analysis of Cx43
Phosphorylation by BMK1-Tryptic digestion of pooled gel slices containing immunoprecipitated Cx43 after endogenous BMK1 was activated was subjected to enzymatic cleavage for the generation of peptide fragments. Pieces were washed with 100 mM ammonium bicarbonate, reduced (dithiothreitol) and alkylated (iodoacetamide), and then dehydrated via acetonitrile evaporation. The gel pieces were re-swollen with 25 mM ammonium bicarbonate containing ϳ 0.2 g of enzyme to achieve a substrate/ enzyme ratio of ϳ 10:1. ZipTip tippets (Millipore, Bedford, MA), packed with C18 matrix, were utilized to clean and concentrate peptide samples prior to analysis. Tips were washed with acetonitrile before peptides were bound and then eluted with either acetonitrile or matrix solution. ZipTip use affords a recovery of 50 -70% in a 1 l volume. Digested protein was mixed with the matrix ␣-cyano-4-hydroxycinnamic acid, and matrix-assisted laser desorption/ionization time-offlight (MALDI-TOF) mass spectrometric analysis was performed in the University of Rochester Protein/Peptide Core Facility as described previously (14). Mass fingerprinting analysis and determination of phosphorylation were performed initially by MS-FIT. 2 The data base search was considered significant if the protein was ranked as the best hit with a sequence coverage of more than 30%. Significance was defined as a molecular weight search (MOWSE) score of at least 1e ϩ003 (MS-FIT) or a difference in probability of 10 Ϫ3 from the first to the second protein candidate (ProFound).
Statistical Analysis-Statistical analysis was performed using Student's t test as appropriate. A p value of less than 0.05 was considered appropriate for accepting or rejecting the null hypothesis.

EGF-induced Cx43
Phosphorylation Is Inhibited by PD98059 -We examined whether extracellular EGF can activate signal transduction pathways leading to Cx43 phosphorylation. HEK293 cells do not express Cx43 but do express all of the necessary components for activation of MAP kinases by EGF (15,16) and thus are a good model system for the study of events mediated by this pathway. First we confirmed that HEK293 cells do not express detectable endogenous Cx43, which is demonstrated in an overexposed Western blot described previously (15) (Fig. 1A). Next, HEK293 cells were transiently transfected with Cx43 cDNA fused at the C terminus to EGFP cDNA; the C-terminal EGFP tag on Cx43 (Cx43E) allowed us to detect plasma membrane targeting of the fusion protein (15). EGF (100 ng/ml) stimulation of the transfected cells resulted in an electrophoretic mobility shift on Cx43 but not EGFP alone, demonstrating that any electrophoretic mobility shift on Cx43E is not an artifact of phosphorylation of the EGFP tag (Fig. 1B). Indeed, if the mobility shift was because of phosphorylation of Cx43E, it should be collapsed by treatment with alkaline phosphatase (2). We determined that EGF-induced Cx43E mobility shift was abolished by alkaline phosphatase treatment (Fig. 1C).
We examined EGF activation of the four best-characterized MAP kinase family members by immunoblotting for the activated forms of ERK1/2, JNK, and p38 and by in vitro kinase assay for BMK1. EGF activated all MAP kinases examined (not shown). Because EGF-induced GJC uncoupling and Cx43 phos- 2 On the Web at Prospector.ucsf.edu. phorylation is blocked by PD98059, we examined the sensitivity of MAPK activation to this antagonist. PD98059 inhibited EGF-induced Cx43E mobility shift and ERK1/2 and BMK1 activation (Fig. 1D) but not p38 or JNK activation (data not shown). These data suggest that both ERK1/2 and BMK1 are potential kinases responsible for the EGF-induced Cx43 phosphorylation and its functional uncoupling. Because PD98059 sensitivity alone is not sufficient to discriminate between responses evoked by ERK1/2 and BMK1, we developed more selective molecular reagents to manipulate the two pathways independently.
Effects of ERK1/2 and BMK1 Activation on EGF-induced Cx43 Functional Uncoupling of Gap Junctions-To circumvent the limitations of drug non-specificity, we created highly specific molecular reagents to selectively activate and inhibit the ERK1/2 and BMK1 pathways. DN-MEK1 and CA-MEK1 were created, and their specificity was detected. We did not detect BMK1 activation in CA-MEK1-transfected cells, although ERK1/2 was potently activated (Fig. 2A). DN-MEK1 blocked EGF-activated ERK1/2 but did not inhibit EGF-induced BMK1 activation, suggesting the specificity in inhibiting the ERK1/2 pathway (Fig. 2B). CA-MEK5␣, which potently activates endogenous BMK1, does not activate ERK1/2 (Fig. 2C), supporting its specificity. Finally, ERK5b, a splice variant of fulllength BMK1, was used as the selective inhibitor of BMK1 (8) because we discovered that a DN-MEK5 construct described previously (12) was not quite as efficient in inhibiting BMK1 in our cell type under our particular conditions. ERK5b is referred to as DN-BMK1. We found that unlike wild-type BMK1 (ERK5a), ERK5b has no kinase activity (Fig. 2D, panel 1), although it can specifically inhibit BMK1-induced MEF2C transcriptional activity (Fig. 2D, panel 2) without affecting EGF-induced ERK1/2 activation (Fig. 2D, panel 3). This gives us confidence that the most downstream component of the signaling pathway can be inhibited effectively.
We sought to determine the consequences of EGF-induced Cx43 phosphorylation through ERK1/2 or BMK1 activation in order to establish functional significance, because Cx43 phosphorylation has been reported to inhibit GJC (4). To further evaluate the role of ERK1/2 and BMK1 in EGF-induced GJC uncoupling, we used CA-MEK1, DN-MEK1, CA-MEK5␣, and DN-BMK1 to selectively activate or inhibit the two pathways.
With the exception of DN-BMK1, these constructs were Cterminal DsRed fusion proteins, allowing clear visual confirmation of successful co-expression with Cx43E. Typically, we obtained 80 -90% transfection efficiency, determined from colocalization of the GFP and DsRed reporter fluorescence (data not shown). The direct functional coupling between cells was studied using the FRAP assay. A typical cell cluster selected for FRAP assay exhibited Cx43E in punctate arrangements at the plasma membrane (Fig. 3A). Ultraviolet laser bleaching of the selected cell resulted in the loss of fluorescence with a subsequent time-dependent recovery of this fluorescence as the reporter molecule diffused back through gap junctions coupled to neighboring cells (Fig. 3B). A representative fluorescence recovery curve for control and EGF-stimulated Cx43E-expressing HEK293 cells is shown in Fig. 3C. Quantitation of cell-to-cell coupling revealed that EGF inhibited GJC (Fig. 3D, lane 2), which was apparent from the increase in the mean recovery time constant. Co-transfection of DN-BMK1 prevented this inhibition (lane 3), but in contrast, DN-MEK1 had no effect on the EGF-induced uncoupling (lane 4). CA-MEK5␣, which activated endogenous BMK1, blocked GJC in the absence of EGF stimulation, but CA-MEK1 failed to inhibit GJC (lanes 5 and 6, respectively). In conjunction with the biochemical data demonstrating the specificity of the reagents we used, these findings suggest a novel BMK1-mediated signaling pathway that regulates GJC. BMK1 activation by EGF is apparently a hitherto unrecognized critical component in Cx43 gap junction uncoupling. BMK1 Phosphorylates and Associates with Cx43-Having demonstrated that BMK1 activation promoted GJC uncou- pling, we hypothesized that BMK1 could possibly directly phosphorylate Cx43. We transfected CA-MEK5␣ to constitutively activate endogenous BMK1, and this was immunoprecipitated with an anti-BMK1 antibody. An in vitro kinase assay was performed between activated BMK1 and GST (negative control), GST-MEF2C (activation domain, positive control), and GST-Cx43 (C-terminal region) as substrates. Activated BMK1 can phosphorylate GST-MEF2C and GST-Cx43 but not GST (Fig. 4A).
To further determine the role of BMK1 as a possible ''Cx43 kinase'' in vivo, we investigated the possible association between BMK1 and Cx43. We transfected a Cx43E-expressing stable cell line with epitope-tagged BMK1. Immunoprecipitation of Cx43 by the epitope tag and immunoblotting for the epitope tag on BMK1 revealed robust Cx43 and BMK1 association (Fig. 4B) but not with GFP and BMK1 (Fig. 4C). This protein-protein association was decreased when BMK1 was activated by CA-MEK5␣ co-transfection. This Cx43 and BMK1 association was a reciprocal phenomenon and was detected by repeating the co-immunoprecipitation experiment using antibodies against the epitope tag on each protein in a reverse manner (Fig. 4D, IP, BMK1, and IB, Cx43). In HEK293 cells, endogenous inactive BMK1 also co-immunoprecipitated with Cx43, and this association was absent when BMK1 was activated (Fig. 4E, panel 1), but all attempts to co-immunoprecipitate either active or inactive ERK1/2 with Cx43 proved unsuccessful (Fig. 4E, panel 2). These data suggest a novel role for BMK1 in regulating Cx43 GJC via phosphorylation and possibly also suggest association with Cx43.
The BMK1 Phosphorylation Site on the C Terminus of Cx43 Is Ser-255 but Not Ser-279 or Ser-282 in Vitro and in Vivo-It has been reported that Ser-255, Ser-279, and Ser-282 are the preferred ERK1/2 phosphorylation sites on Cx43 (4). To evaluate the preferred phosphorylation site on Cx43 by BMK1, we performed tryptic peptide mapping using MALDI-TOF mass spectrometry after immunoprecipitated endogenous BMK1 was activated by CA-MEK5␣ transfection and incubated with GST-Cx43CT in an in vitro kinase reaction. Computer-assisted . Endogenous BMK1 was activated by CA-MEK5␣ transfection and then immunoprecipitated from HEK293 cells immediately before the in vitro kinase assay was performed. Ponceau S staining of the nitrocellulose membrane demonstrates the position of the proteins after separation by SDS-PAGE. IP, immunoprecipitation. B, Cx43E coimmunoprecipitates with basally active BMK1 but less so when BMK1 is activated by CA-MEK5␣ transfection. A stably expressing Cx43E HEK293 cell line was transfected with Xpress-tagged BMK1 without (Ϫ) or with CA-MEK5␣ (ϩ). Cx43E was immunoprecipitated with anti-GFP antibody and immunoblotted with anti-Xpress antibody. C, additional control (GFP alone co-expressed with Xpress-tagged BMK1) shows lack of association and confirms the Cx43-BMK1 association demonstrated. The total cell lysate (TCL) lanes demonstrate the presence of input Cx43E proteins for both conditions. D, a complementary experiment where BMK1 was immunoprecipitated with anti-Xpress antibody and Cx43E was immunoblotted with anti-GFP antibody. A loading control is shown at the bottom of each autoradiogram. These data are representative of four separate experiments. E, panel 1, Cx43E was immunoprecipitated with anti-GFP antibody, and endogenous BMK1 was immunoblotted with anti-BMK1 antibody. BMK1 is very potently activated by H 2 O 2 as we reported previously, and thus it was used to activate BMK1 (10). Panel 2, Cx43E was immunoprecipitated with anti-GFP antibody, and endogenous ERK1/2 was immunoblotted with anti-ERK1/2 antibody. Cx43 and ERK1/2 association was not apparent basally or when ERK1/2 or BMK1 activation status was altered by EGF stimulation Ϯ DN-MEK1 or DN-BMK1. The TCL lanes demonstrate the presence of input of total proteins for each condition. proteomic analysis revealed two tryptic peptide ions with mass/ charge ratios of 1877.7906 and 3155.3995, corresponding to the C-terminal fragmented peptides spanning amino acids 244 -258 and 265-293, respectively. Furthermore, we observed that only the amino acid region 244 -258 was modified by phosphorylation on a single serine residue. Threonine and tyrosine residues within that region were not modified, and there were no covalent modifications within the amino acid 265-293 tryptic-digested region. Because Ser-255 is the only potential BMK1 phosphorylation site within that region, we suspected that it might be the preferred phosphorylation site for BMK1, even though two other potential BMK1 phosphorylation sites were available for phosphorylation.
To support this result, we generated three GST-Cx43CT peptides with mutations in the proline-directed serines that are potential BMK1 phosphorylation sites (Ser-255, Ser-279, and Ser-282). We transfected CA-MEK5␣ to constitutively activate endogenous BMK1, which was immunoprecipitated with an anti-BMK1 antibody. As before, an in vitro kinase assay with mutant GST-Cx43 (C-terminal region, amino acids 230 -382) as a substrate was performed. As shown in Fig. 5, we found that the double mutant (S279A/S282A) was potently phosphorylated by BMK1. However, activated BMK1 failed to phosphorylate Cx43 single (S255A) or triple mutant (S255A/ S279A/S282A) peptides. Our in vitro kinase assay strongly suggests that Ser-255 is the only residue that is accessible for BMK1-mediated phosphorylation, which is consistent with our MALDI-TOF data.
To investigate the role of BMK1 on Cx43 Ser-255 phosphorylation in vivo, we studied full-length Cx43 point mutants (S255A, S279A/S282A, or S255A/S279A/S282A). Again, the Cterminal GFP tag on Cx43 demonstrated that all Cx43 mutants express equally and target to the plasma membrane. We then co-transfected mutants of Cx43E with CA-MEK1 or CA-MEK5␣ to activate endogenous ERK1/2 or BMK1, respectively, in HEK293 cells.
As shown in Fig. 6A, transfection of both CA-MEK5␣ and CA-MEK1 can phosphorylate Cx43E. When we co-transfected the Cx43 (S255A) mutant with CA-MEK1 or CA-MEK5␣, we observed that only CA-MEK1 induced a modest electrophoretic mobility shift (Fig. 6B). When Cx43 (S279A/S282A) was cotransfected with CA-MEK1 or CA-MEK5␣, we observed that both CA-MEK1 and CA-MEK5␣ induced an electrophoretic shift (Fig. 6C, panel 1). Moreover, co-transfection of Cx43 (S279A/S282A) with a null vector and stimulation with EGF yielded a reproducibly small electrophoretic mobility shift on Cx43. This shift was abolished when EGF stimulation was repeated in cells co-transfected with DN-BMK1 or DN-MEK1 (Fig. 6C, panel 1), and co-transfection of wild-type Cx43 with DN-BMK1 together with CA-MEK1 showed that Cx43E phosphorylation was still present (Fig. 6C, panel 2). Finally, cotransfection of Cx43 (S255A/S279A/S282A) with CA-MEK1 or CA-MEK5␣ demonstrated an electrophoretic mobility shift only when CA-MEK1 was transfected (Fig. 6D). These findings strongly suggest that activated BMK1 induces Cx43 phosphorylation preferentially at Ser-255 but not at Ser-279 or Ser-282 in vivo. In contrast, ERK1/2 may phosphorylate Cx43 at several sites in addition to Ser-255, Ser-279, and Ser-282. DISCUSSION In the present study, we investigated the role of the activated MEK5-BMK1 module in Cx43 phosphorylation and subsequent inhibition of GJC. The major finding of this study is that BMK1 associates with Cx43 and phosphorylates Ser-255 on Cx43, and BMK1 kinase activation is sufficient to inhibit GJC in an in vivo model. In contrast, CA-MEK1 did not inhibit GJC, and DN-MEK1 could not relieve EGF-induced gap junction inhibition. Our functional study suggests that activated ERK1/2 alone is not a sufficient step for growth factor-induced gap junction phosphorylation and uncoupling, which is in agree- ment with other studies (17)(18)(19)(20).
To our knowledge our work is the first to document Cx43 as a downstream target of BMK1. Earlier work has suggested that ERK1/2 can phosphorylate Cx43 at Ser-255, Ser-279, and Ser-282, all residues that are found within the MAP kinase phosphorylation consensus motif (21), and we confirmed the inhibition of Cx43 phosphorylation by the MEK inhibitor, PD98059. However, recent studies indicate that PD98059 quite effectively inhibits MEK5, a specific upstream MEK for BMK1 (12,22,23). As such, previously reported functional effects credited to EGF-induced phosphorylation of Cx43 by ERK1/2 (4, 24) may be, in part or entirely, through the MEK5/BMK1 pathway. Through the use of dominant negative and constitutively active molecular reagents with greater specificity than pharmacological reagents, we have demonstrated that activated BMK1 leads directly to Cx43 phosphorylation in vitro and in vivo and to the inhibition of GJC. Cx43 phosphorylation by selective BMK1 activation and consequent inhibition of GJC but insensitivity to MEK1 manipulations indicate that activated BMK1 but not ERK1/2 mediates the EGF-induced Cx43 gap junction uncoupling. This is supported by highly specific molecular manipulations showing a lack of cross-talk between the ERK1/2 and BMK1 signaling pathways and thus a delineation of responses elicited by these two pathways when activated by EGF (Fig. 2).
Our data also demonstrate that EGF-induced ERK1/2 activation and consequent Cx43 phosphorylation, in isolation, may not be a direct correlate for inhibition of GJC in vivo. ERK1/2 activation by CA-MEK1 co-transfection with the triple Cx43 mutant (S255A/S279A/S282A) could still induce electrophoretic mobility shift, and so the existence of alternative ERK1/2 phosphorylation sites on Cx43 in addition to Ser-255, Ser-279, and Ser-282 is apparent. The fact that we did not observe this same result with activated BMK1 may indicate a greater specificity for this protein kinase.
DN-MEK1 or DN-BMK1 can significantly inhibit phosphorylation of the double Cx43 mutant (S279A/S282A) after EGF stimulation, and thus Ser-255 is clearly targeted by each of these activated kinases. It remains unresolved as to why BMK1 but not ERK1/2 can regulate EGF-induced Cx43 gap junctional uncoupling through Ser-255 phosphorylation alone. Cx43 can be phosphorylated in vivo on multiple residues by ERK1/2, but because only activated BMK1 inhibits Cx43 function, ERK1/2 phosphorylation elsewhere may affect BMK1 regulation of Cx43. A coordinated effort between activated ERK2 and activated BMK1 in signaling pathways has been reported (25). However, BMK1 but not ERK1/2 association with Cx43 was demonstrated, and so it is quite possible that BMK1 binding coupled with BMK1-induced Cx43 phosphorylation on Ser-255 has an equally important effect in controlling GJC. Moorby and Patel (26) reported that single mutants of Ser-255, Ser-279, and Ser-282 independently prevented GJC blockade following treatment of the cells with platelet-derived growth factor, suggesting the important role for each independent phosphorylated serine of Cx43 in controlling its gating properties. It is noteworthy that the gating of Cx43 and gap junctions can indeed be achieved by phosphorylation-independent mechanisms, particularly via protein-protein interactions such as the interaction of Cx43 with the tight junction protein ZO-1 and the oncogenic product v-Src (15,27). Such protein-to-protein interaction with connexins, recently referred to as the ''nexus,'' is an emerging mechanism that could regulate GJC independently (28). Because the C-terminal region of BMK1 interacts with the transcription factor MEF2 to potently promote its transcrip-tional activation independently from phosphorylation (29), it is possible that the physical interaction of BMK1 with Cx43 may influence GJC. Peracchia et al. (30) have shown that the calcium-binding protein, calmodulin, can directly gate Cx32 in the so-called ''cork'' model. Cx43 not only co-localizes with the v-Src tyrosine kinase to plasma membrane regions of the cell, but it also appeared to interact directly with v-Src (31,32). This is particularly intriguing because we demonstrate here that basally active overexpressed BMK1 and endogenous BMK1 complex with Cx43 and that this association is decreased substantially when BMK1 is activated. Because we were unable to demonstrate Cx43 and ERK1/2 association, it is reasonable that the profound effect of BMK1 on GJC could be assigned in part to Cx43 association as well as to BMK1 phosphorylation. Additional studies will be required for the further investigation of the role of BMK1/Cx43 protein-protein interaction in regulating GJC.