INTRODUCTION

The Rous sarcoma virus transforming protein, v-Src, and its cellular homologue, c-Src, are protein tyrosine kinases associated with the plasma membrane (1). Expression of v-Src leads to neoplastic cell transformation, accompanied by increased tyrosine phosphorylation of cellular proteins (2-4). Signaling through Src is due, in part, to protein-protein interactions mediated by SH2¹ and SH3 domains (5-7). Mutations in either the SH2 or SH3 domain of c-Src can lead to increased tyrosine kinase activity and oncogenic potential (5, 8, 9). SH2 domains bind Tyr(P) residues in target proteins. The specificity of binding is determined by residues immediately C-terminal to the Tyr(P) (6, 10, 11). SH3 domains bind short proline-rich peptide motifs (6, 12-14). Studies using phage display libraries to identify peptides that bind SH3 domains showed that a minimal PXXP consensus sequence is required for binding (15, 16).

In addition to playing a role in cell growth and transformation, the Src protein tyrosine kinase has been implicated in regulating GJC (17-20). Gap junctions are membrane channels which mediate the intercellular passage of ions, second messengers, and small molecules (21). It has been proposed that growth regulatory molecules pass between cells through gap junctions (22-25).

Gap junctions are formed by specialized proteins termed connexins, arranged in the cell membrane so that each connexin has four membrane spanning regions, a cytoplasmic loop, two extracellular loops, and cytoplasmic N- and C-terminal ends. Among the 13 connexin family members identified to date, the C-terminal tail is the most divergent region. In some cases this region contains consensus protein kinase phosphorylation sites (26-28). Connexins 32 and 43 are phosphoproteins (29-32), however, Cx43 is phosphorylated on tyrosine by Src, but Cx32 is not (32). Comparison of the amino acid sequences of connexins 32 and 43 revealed that putative SH3-binding regions and tyrosine phosphorylation sites in Cx43 were absent in connexin 32 (26).

Reduced GJC is a characteristic of cells transformed by several oncogenes (24), including SV-40 (33), polyomavirus middle T antigen (34), v-ras (35, 36), v-mos (37), and v-fps (38), as well as v-src. The effects of Src on GJC have been studied extensively (17-20, 30, 32, 39-41). Several lines of evidence suggest that tyrosine phosphorylation of Cx43 is important in the regulation of GJC by v-Src. First, cells infected with temperature-sensitive Rous sarcoma virus show reduced levels of GJC (17, 19), correlated with a rapid increase in tyrosine phosphorylation of Cx43 (40), at the permissive temperature. Second, expression of v-Src in communication-competent paired Xenopus oocytes expressing Cx43, but not in oocytes expressing Cx32, leads to reduced GJC, which depends on phosphorylation of Cx43 on tyrosine 265 (32). Finally, purified Src phosphorylates Cx43 in vitro at sites that are phosphorylated in vivo in v-src transformed Rat-1 fibroblasts (41). These results support the hypothesis that inhibition of GJC in v-src-transformed cells is a consequence of tyrosine phosphorylation of Cx43 by v-Src.

Regulation of GJC by v-Src is of particular interest because of the possible role of GJC in cell growth and transformation. To further clarify the role of v-Src in regulating Cx43 phosphorylation and GJC, we investigated direct interactions between Cx43 and v-Src. Cx43 associated with the SH3 and SH2 domains of v-Src, but not c-Src, both in vitro and in vivo. Mutation of the proline-rich region of Cx43 greatly reduced the association between Cx43 and v-Src. Mutation of Cx43 at Tyr-265 resulted in reduced association between Cx43 and v-Src. Mutations in the SH3 or SH2 domains of Src, which disrupt interactions with Src-associated proteins (42-48), also reduced interactions between Cx43 and v-Src. The reduced association between Cx43 and v-Src resulted in decreased tyrosine phosphorylation of Cx43. We conclude that tyrosine phosphorylation of Cx43 by v-Src is mediated by SH2 and SH3 domain interactions.

EXPERIMENTAL PROCEDURES

Rat-1 fibroblasts, v-src-transformed Rat-1 fibroblasts (49), and human embryonic kidney 293 cells (50) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% normal calf serum (Intergen Co., Purchase, NY) at 37 °C. 293 cells were transiently cotransfected with (2 µg) pRK5 wild-type or mutant Cx43 and (2 µg) pCDNA3 v-src or mutant v-src expression vectors by calcium phosphate precipitation.

The SH3 domains of mouse n-Src (residues 81-155) and c-Src (residues 79-147) were amplified by polymerase chain reaction from vectors containing full-length n-Src and c-Src. The polymerase chain reaction products were subcloned into pGEX2TK at the BamHI and EcoRI restriction sites. Constructs encoding the v-Src SH3, SH2, and mutant W118R and P133L SH3 domain GST fusion proteins were kindly provided by Dr. Xing Quan Liu (Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto). The R95W and T96I c-Src SH3 domain mutant fusion proteins were kindly provided by Dr. David Schlaepfer (The Scripps Research Institute, La Jolla, CA). The GST fusion proteins were prepared as described (51).

Proline to alanine and tyrosine to phenylalanine substitution mutations in Cx43 cDNA, and mutations in the v-Src SH2 and SH3 domains, were introduced by oligonucleotide-directed mutagenesis or polymerase chain reaction-based mutagenesis (52, 53). The Cx43 mutants were subcloned into the eukaryotic expression vector pRK5 (54). The v-Src mutants were subcloned into the eukaryotic expression vector pCDNA3 (Invitrogen, San Diego, CA). Mutations and polymerase chain reaction products were verified by DNA sequencing.

Lysates of transiently transfected 293 cells, Rat-1 fibroblasts, or v-src transformed Rat-1 fibroblasts were immunoprecipitated with a monoclonal peptide antibody corresponding to amino acids 2-17 of Src, which recognizes both c-Src and v-Src (Src2-17, Microbiological Associates) or rabbit antiserum against a peptide corresponding to amino acids 368-382 of rat heart Cx43 (Cx43) as described (30). The immunoprecipitated proteins were resolved by SDS-PAGE on 12% polyacrylamide gels (55). For metabolic radiolabeling, cells were incubated in phosphate-free Dulbecco's modified Eagle's medium containing 4% dialyzed calf serum for 1 h at 37 °C, followed by incubation with 1 mCi/ml [³²P_i] (ICN, Irvine, CA) for 5 h at 37 °C. Phosphoamino acid composition was determined as described (56).

Proteins resolved by SDS-PAGE were electrotransferred to Immobilon P membrane (Millipore, Bedford, MA) and blocked in 3% Blocko (3% bovine serum albumin, radioimmunoassay grade, 50 mM Tris (pH 7.5), 150 mM NaCl, and 0.1% Tween 20) overnight at room temperature on a rocker. The membrane filter was probed with Cx43, Src2-17, or a monoclonal phosphotyrosine antibody (4G10pTyr, Upstate Biotechnology Inc., Lake Placid, NY) followed by a horseradish peroxidase-coupled secondary antibody. The immunoreactive proteins were detected by enhanced chemiluminescence (Amersham Corp.).

10 µg of purified GST fusion proteins bound to glutathione-agarose beads were incubated with 2 µg of a membrane fraction of uninfected Sf9 cells, 2 µg of partially purified baculovirus Cx43 (41) or cellular lysates of v-src transformed Rat-1 fibroblasts containing 1 mg of protein prepared in 1% Nonidet P-40 lysis buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 1 mM dithiothreitol, 20 µg/ml leupeptin, 1 µg/ml aprotinin) for 2 h at 4 °C on a rocker. The protein complexes were separated by SDS-PAGE and immunoblotted as described above.

Increasing concentrations of the Cx43 synthetic peptide CSSPTAPLSPMSPPGYK corresponding to amino acids 271-287 of the proline-rich region of Cx43, or a nonspecific synthetic peptide CHQELNALNVFKISF, were preincubated with immobilized GST v-Src SH3 domain fusion proteins (10 µg) for 30 min at 4 °C on a rocker. Following the preincubation period, partially purified baculovirus Cx43 (2 µg) was added and the samples were incubated for an additional 60 min at 4 °C. The samples were resolved by SDS-PAGE and electrotransferred to Immobilon P membrane. Cx43 was detected by immunoblot analysis as described above.

RESULTS

Putative SH3 domain-binding sites are located in the C-terminal region of Cx43 at residues 253-256 (PLSP) and 274-284 (PTAPLSPMSPP). To test whether Cx43 can bind the Src SH3 domain, immobilized GST fusion proteins containing the SH3 domains of c-Src, n-Src, or v-Src were incubated with uninfected Sf9 cell lysates (Fig. 1A, odd numbers) or with partially purified baculovirus Cx43 (Fig. 1A, even numbers). The bound proteins were immunoblotted with Cx43 antiserum. As shown in Fig. 1A, multiple forms of baculovirus Cx43 associated with the SH3 domain of v-Src. The slower migrating forms of Cx43 are phosphorylated forms of Cx43 (30, 41, 57). No specific binding was observed with GST alone, n-Src, or c-Src SH3 fusion proteins (Fig. 1A, compare lane 8 with lanes 2, 4, and 6). Cx43 was not detectable in samples incubated with uninfected Sf9 cells (Fig. 1A, odd numbered lanes). The association between baculovirus Cx43 and the SH3 domain of v-Src was greatly reduced by mutation of two highly conserved residues in the SH3 domain of v-Src, W118R and P133L (Fig. 1A, lanes 10 and 12). These mutations also abolish interactions between the v-Src SH3 domain and the p85 subunit of PI 3-kinase (44) or Sam68 (58). Cx43 proteins from lysates of v-src transformed Rat-1 fibroblasts behaved similarly to baculovirus Cx43: they associated with GST v-Src SH3, but not GST alone, n-Src, or c-Src SH3, and mutations in the SH3 domain of v-Src abolished the interaction (Fig. 1B).

The highly conserved RT loop of the SH3 domain of c-Src is exposed and potentially involved in binding to other proteins (8, 9, 58-60). We tested SH3 domains of c-Src with mutations in the RT loop that converted arginine 95 and threonine 96 to the corresponding amino acids in v-Src. Immobilized GST fusion proteins were incubated with uninfected Sf9 cell lysates (Fig. 1C, odd numbers) or partially purified baculovirus Cx43 (Fig. 1C, even numbers). Mutation of arginine 95 to tryptophan (R95W) increased the association with baculovirus Cx43 (Fig. 1C, compare lanes 4 and 10). However, mutation of threonine 96 to isoleucine (T96I) had no effect (Fig. 1C, lane 6). The double mutant containing both mutations (W/I) also showed increased association with Cx43 (Fig. 1C, lane 8).

The specificity of the interaction between the SH3 domain of v-Src and Cx43 was further examined by peptide competition. A synthetic peptide corresponding to the proline-rich region of Cx43 (residues 271-287) was used to compete with baculovirus Cx43 for binding to GST v-Src SH3. As shown in Fig. 1D, the Cx43 synthetic peptide inhibited the association between baculovirus Cx43 and GST v-Src SH3. Half-maximal inhibition occurred at a concentration of 100 µM (Fig. 1D). A nonspecific synthetic peptide had no effect on the binding (data not shown). Taken together, these results show that Cx43 interacts with the SH3 domain of v-Src, but not the SH3 domain of c-Src or n-Src, in vitro.

The v-Src SH2 domain associates with several tyrosine-phosphorylated proteins in v-src transformed cells (42, 61-66). To test whether the v-Src SH2 domain can bind tyrosine-phosphorylated Cx43 from v-src transformed cells in vitro, lysates of normal or v-src transformed Rat-1 fibroblasts were incubated with GST v-Src SH2 domain fusion proteins bound to beads (Fig. 2). Comparable amounts of Cx43 proteins from normal (R) and v-src transformed (V) Rat-1 fibroblasts were used in the binding experiments (Fig. 2, top panel, lanes 1 and 2). Cx43 from v-src-transformed Rat-1 cells, but not from normal cells, associated with the SH2 domain of v-Src (Fig. 2, top panel, compare lanes 3 and 4). GST v-Src SH2 bound to the slower migrating forms of Cx43 from v-src transformed Rat-1 cells (Fig. 2, top panel, lane 4). To determine whether the associated Cx43 proteins were tyrosine phosphorylated, the membrane was stripped and reprobed with anti-Tyr(P) antibody. The slower migrating forms of Cx43 from v-src-transformed Rat-1 cells (Fig. 2, bottom panel, lane 2) as well as the Cx43 proteins which associated with GST v-Src SH2 (Fig. 2, bottom panel, lane 4) were tyrosine phosphorylated. These results demonstrate that the SH2 domain of v-Src interacts with tyrosine-phosphorylated Cx43 from v-src transformed Rat-1 cells.

To test whether Cx43 associates with v-Src SH3 and SH2 domains in vivo, as well as in vitro, coprecipitation experiments were performed. Cx43 was immunoprecipitated from lysates of normal or v-src transformed Rat-1 fibroblasts (Fig. 3A). The immunoprecipitates were immunoblotted with Src2-17 antiserum. As shown in Fig. 3A, v-Src was present in Cx43 immunoprecipitates from v-src transformed Rat-1 fibroblasts but not c-Src from control Rat-1 fibroblasts. The reverse experiment was conducted by immunoprecipitating Src from normal and v-src transformed Rat-1 fibroblasts and blotting for Cx43. Cx43 was present in Src immunoprecipitates from v-src transformed fibroblasts, but not normal Rat-1 fibroblasts.²

To further examine in vivo interactions between Cx43 and v-Src, coprecipitation experiments were conducted using transient cDNA expression in human embryonic kidney 293 cells. Src was immunoprecipitated from 293 cells transiently cotransfected with either vector alone, c-src or v-src, and wild type Cx43. Transfected 293 cells expressed comparable levels of Cx43 (Fig. 3B, lanes 1, 2, 5, and 6). The mobility shift of Cx43 in the presence of v-Src was due to increased phosphorylation of the protein (Fig. 3B, compare lanes 1 and 2). This change in mobility of Cx43 was not observed in the presence of c-Src (Fig. 3B, compare lanes 5 and 6). Cx43 was detected in Src immunoprecipitates from 293 cells cotransfected with v-src and Cx43 (Fig. 3B, lane 4), but not in Src immunoprecipitates from 293 cells cotransfected with Cx43 and either vector (Fig. 3B, lanes 3 and 7) or c-src (Fig. 3B, lane 8). Therefore, Cx43 interacts with v-Src, but not c-Src, in vivo as well as in vitro.

Previous studies have identified residues in Src important for SH3- and SH2-mediated interactions (44, 45, 58, 67-69). Tryptophan 118 (Trp-118), which is highly conserved among different SH3 domains, contributes to the hydrophobic surface important for SH3 binding (67, 69). Arginine 175 (Arg-175), an invariant residue in the highly conserved FLVRES motif of SH2 domains, forms an ion pair with the phosphotyrosine residue of interacting proteins (68). Mutation of Trp-118 and Arg-175 of v-Src reduces the binding of v-Src to SH3- and SH2-domain interacting proteins, respectively (44, 45, 58).

To determine whether Trp-118 and Arg-175 of v-Src are important for interactions between Cx43 and v-Src, Trp-118 and Arg-175 were mutated to arginine and lysine, respectively. Variant v-Src proteins were coexpressed with Cx43 in transiently transfected 293 cells. The lysates were immunoprecipitated with Src2-17 antiserum and analyzed by immunoblotting with Cx43 antiserum. Fig. 4A shows that association of Cx43 with the three v-Src mutants (W118R, R175K, W118R/R175K) was greatly reduced compared with wild type v-Src (top panel). A small amount of residual association was observed between Cx43 and mutant v-Src proteins (Fig. 4A, top panel).

Substrates that depend on interactions with SH3 and SH2 domains of Src for tyrosine phosphorylation by Src include Sam68 (46, 70), GAP-associated p62 (47, 48), and AFAP-110 (42, 43). These substrates are not tyrosine phosphorylated by Src SH3 or SH2 mutants with reduced binding ability (42, 46-48). We examined how mutations in the SH3 and SH2 domains of v-Src affected tyrosine phosphorylation of Cx43. First, we examined whether Cx43 that coprecipitated with wild type v-Src, or with v-Src mutants, was tyrosine phosphorylated. As shown in Fig. 4A, bottom panel, tyrosine phosphorylation was readily observed in Cx43 which coprecipitated with wild type v-Src, whereas little or no tyrosine phosphorylation was detected in Cx43 coprecipitated with v-Src mutants.

We further examined tyrosine phosphorylation of Cx43 immunoprecipitated from 293 cells expressing mutant SH3 and SH2 v-Src proteins. Cx43 immunoprecipitates were analyzed by anti-Tyr(P) immunoblotting. Comparable levels of Cx43 were expressed in all samples (Fig. 4B, top panel). Mutations in either the SH3 or SH2 domain of v-Src abolished tyrosine phosphorylation of Cx43 (Fig. 4B, bottom panel).

293 cells cotransfected with wild type Cx43 and variant v-Src constructs were metabolically radiolabeled with [³²P]orthophosphate and Cx43 was analyzed for phosphoamino acid content. In the absence of v-Src, Cx43 was phosphorylated only on serine (Fig. 4C, pCDNA3). Coexpression of v-Src resulted in the appearance of phosphotyrosine and trace amounts of phosphothreonine (Fig. 4C, v-Src). Coexpression of the mutant v-Src proteins resulted in little or no detectable phosphotyrosine and phosphothreonine in Cx43 (Fig. 4C, W118R, R175K, and W118R/R175K). These results show that mutations in the SH3 and SH2 domains of v-Src decrease both binding to Cx43 and tyrosine phosphorylation of Cx43, suggesting that tyrosine phosphorylation of Cx43 requires association between v-Src and Cx43.

To identify regions of Cx43 important for SH2- and SH3-mediated interactions between Cx43 and v-Src, proline-rich regions and possible tyrosine phosphorylation sites of Cx43 were mutated. Fig. 5 shows the Cx43 mutants used.

Wild type Cx43 (WTCx43) or various proline to alanine Cx43 mutants were coexpressed with the control vector (Fig. 6A, odd numbered lanes), or v-Src (Fig. 6A, even numbered lanes) in transiently transfected 293 cells and coprecipitation experiments were performed. Comparable amounts of Cx43 proteins were expressed in all samples (Fig. 6A, top panel). The proline to alanine mutations altered the banding profile of Cx43 proteins (Fig. 6A, top panel, compare lanes 3, 5, 7, and 9). Slower migrating forms of Cx43 were observed in the presence of v-Src compared with vector control (Fig. 6A, top panel, compare odd numbered lanes with even numbered lanes). WTCx43 and the N-terminal proline to alanine substitution mutant Cx43 protein (N PA) coprecipitated with v-Src (Fig. 6A, bottom panel, lanes 4 and 6). However, the C-terminal proline to alanine mutant Cx43 protein (C PA) and the mutant containing both N- and C-terminal proline to alanine substitutions (NC PA), did not coprecipitate with v-Src (Fig. 6A, bottom panel, lanes 8 and 10). Therefore the C-terminal proline-rich motif of Cx43 is important for association with v-Src.

To identify the tyrosine residue(s) in Cx43 required for binding to v-Src, tyrosine residues 247, 265, and 267 in Cx43 were mutated to phenylalanine (Fig. 5) and tested in coprecipitation experiments (Fig. 6B). The mutant Cx43 proteins were expressed at similar levels (Fig. 6B, top panel). Slower migrating forms of Cx43 were observed in the presence of v-Src compared with vector control (Fig. 6B, top panel, compare odd numbered lanes with even numbered lanes). WTCx43, and the Y247F and Y267F mutant Cx43 proteins, coprecipitated equally well with v-Src (Fig. 6B, bottom panel, lanes 14, 16, and 20). In contrast, mutation of tyrosine 265 greatly reduced coprecipitation with v-Src (Fig. 6B, bottom panel, lane 18). Mutation of tyrosine 265, but not tyrosines 247 or 267, also abolished the ability of the GST-v-Src SH2 domain fusion protein to bind Cx43 in vitro (data not shown). Therefore, tyrosine 265 of Cx43 is important for binding with v-Src.

To determine whether mutations in Cx43 which reduced association with v-Src also affected tyrosine phosphorylation of Cx43, Cx43 was immunoprecipitated from lysates of 293 cells coexpressing v-Src and variant Cx43 proteins and analyzed by anti-Tyr(P) immunoblotting. The variant Cx43 proteins were expressed at similar levels (Fig. 6C, top panel). The NC PA mutant migrated faster than WTCx43 or Y265F mutant Cx43 proteins (Fig. 6C, top panel). Tyrosine phosphorylation of the mutant Cx43 proteins was greatly reduced compared with WTCx43 (Fig. 6C, bottom panel). The immunoblotting results were confirmed by immunoprecipitation and phosphoamino acid analysis of ³²P-radiolabeled Cx43 proteins, as shown in Fig. 6D. Tyrosine phosphorylation of the Cx43 proteins mutated in the proline-rich regions or at tyrosine 265 was again greatly reduced compared with WTCx43. Tyrosine phosphorylation of Cx43 mutated at tyrosines 247 or 267 was not affected (data not shown). Therefore phosphorylation of Cx43 on tyrosine requires association of Cx43 with v-Src, mediated by interactions with the C-terminal proline-rich region and tyrosine 265 of Cx43.

DISCUSSION

The results described here show that Cx43 binds to v-Src, but not to c-Src, in vitro and in vivo. The association of Cx43 with v-Src is mediated by interactions between the SH3 domain of v-Src and a proline-rich region of Cx43, and by the SH2 domain of v-Src and tyrosine 265 of Cx43. The association leads to tyrosine phosphorylation of Cx43, which is correlated with down-regulation of GJC and cell transformation.

Azarnia et al. (20) demonstrated that overexpression of c-Src had little or no effect on GJC, whereas expression of v-Src, or kinase-activated mutant c-Src (Y527F c-Src), reduced GJC. The extent of the reduction of GJC correlated with the level of tyrosine phosphorylation of Cx43. Overexpression of c-Src induces low levels of tyrosine phosphorylation of cellular proteins, and does not cause transformation, whereas expression of Y527F c-Src, or v-Src, induces elevated levels of tyrosine phosphorylation and causes transformation (49, 71-76). The transforming activity of Y527F c-Src is lower than that of v-Src (75, 76). Differences in the transforming activity of c-Src, Y527F c-Src, and v-Src may reflect differences in binding and phosphorylation of cellular substrates, including Cx43.

Previous studies demonstrated differential association of c-Src and v-Src to cellular substrates (45, 49, 61, 77-79). The amino acid sequences of c-Src and v-Src show differences throughout the protein, including changes in the SH3 and SH2 domains. Mutations in the SH3 and SH2 domains can alter the transforming activity of Src (5, 8, 9, 80-88), possibly by altering kinase activity or interactions with cellular substrates. A number of substrates that interact with Src through its SH3 and SH2 domains have been identified, including proteins that affect cytoskeletal architecture and cell morphology, and signaling proteins associated with cell growth and transformation (reviewed in Ref. 89). Cx43 likewise interacts with v-Src through its SH3 and SH2 domains, leading to tyrosine phosphorylation of Cx43, loss of GJC, and cell transformation.

Structural analyses of Src SH3 and SH2 domains have identified amino acid residues important for interactions with substrates (60, 68, 69, 90, 91). In particular, arginine 95 and threonine 96 within the SH3 domain of c-Src, termed the RT loop, are involved in substrate binding (8, 9, 58-60). Mutation of arginine 95 of c-Src to the corresponding amino acid of v-Src (R95W) increases Src transforming potential, as well as binding of substrate proteins, including the p85 subunit of PI 3-kinase (8, 9, 58). Similarly, mutation of the RT loop of the Src family member, Fyn, increases binding to the HIV-1 Nef protein (92). We observed similar increases in Cx43 binding to R95W, but not T96I, c-Src mutants in vitro (Fig. 1C). Various strains of v-Src, including Schmidt-Ruppin A, Prague A, and Prague C, contain mutations at arginine 95 but not threonine 96 (93, 94). Therefore, the arginine residue of the RT loop appears to play an important role in substrate binding, and mutation of this residue may increase the binding affinity of v-Src for cellular substrates, including Cx43.

Mutation of tryptophan 118 or proline 133 in the SH3 domain of v-Src, also important for SH3 domain interactions (91), abolished binding to the p85 subunit of PI 3-kinase, in vitro (44). Binding of PI 3-kinase to v-Src, which results in tyrosine phosphorylation of p85 and enhanced PI 3-kinase activity, has been implicated in transformation (65, 95, 96). Cells expressing v-Src SH3 domain mutants defective in PI 3-kinase binding show reduced PI 3-kinase activity and normal or fusiform morphology (58, 88). Mutation in the SH3 domain of v-Src greatly reduced the association with Cx43, both in vitro (Fig. 1, A and B) and in vivo (Fig. 4A). Tyrosine phosphorylation of Cx43 was dependent on interactions with v-Src (Fig. 4B).

The highly conserved FLVRES motif (residues 172-177) in the SH2 domain of Src contributes to the phosphotyrosine binding pocket (68). Arginine 175 forms an ion pair with the phosphotyrosyl residue of the SH2-binding protein, and is conserved in SH2 domain-containing proteins (68). Mutation of arginine 175 of v-Src greatly reduced binding to and tyrosine phosphorylation of Cx43, in vivo (Fig. 4). Similar effects of SH2 domain mutations on binding and tyrosine phosphorylation have been observed with other Src substrates, including AFAP-110 (42), GAP-associated p62 (47), and RNA-binding protein Sam68 (46).

Fig. 7 summarizes a model that could account for the interactions and phosphorylation events involving Cx43 and v-Src. Binding of v-Src to Cx43 is initiated by an SH3-mediated interaction, bringing the kinase domain of v-Src in close proximity to the tyrosine 265 phosphorylation site of Cx43. Following phosphorylation of Tyr-265 by v-Src, the association is stabilized by an SH2-Tyr(P)-265 interaction. (The amino acid sequence on the C-terminal side of tyrosine 265, pYAYF, suggests that tyrosine 265 is a good candidate for binding by group II SH2 domains (6).) Cx43 may then be further phosphorylated on other tyrosines by v-Src, or on serines by v-Src-associated kinases (v-Src immunoprecipitated from v-src-transformed Rat-1 fibroblasts phosphorylated Cx43 on serine and tyrosine residues, in vitro, suggesting that a serine kinase is complexed with v-Src; data not shown). Tyrosine phosphorylation of Cx43 by v-Src may also recruit other SH2-containing signaling molecules that could affect Cx43 phosphorylation and gap junction channel function. A similar sequence of binding reactions, SH3 domain followed by SH2 domain, has been suggested for the interaction of Src with its substrates, AFAP-110 (42) and Sam68 (46).

Recently, Holmes et al. (97) demonstrated that the human potassium channel, hKv1.5, associates with v-Src, in vivo. The interaction was mediated by the proline-rich region of hKv1.5 and the SH3 domain of Src. Src SH2-domain-phosphotyrosine interactions were not examined. However, in the presence of v-Src, hKv1.5 was tyrosine-phosphorylated and the potassium channel current was blocked. In addition, the N-methyl-D-aspartate channel is regulated by tyrosine phosphorylation (98, 99). Src was found to coprecipitate with the N-methyl-D-aspartate channel, in vivo and the activity of the channel was regulated by Src (100). Therefore, the Src protein tyrosine kinase can bind to and regulate other pore forming channels, in addition to gap junctions, suggesting that Src may exert widespread effects on intercellular communication.