Regulation of Connexin-43 Gap Junctional Intercellular Communication by Mitogen-activated Protein Kinase*

Activation of the Ras/Raf/mitogen-activated protein kinase kinase/mitogen-activated protein (MAP) kinase signaling cascade is initiated by activation of growth factor receptors and regulates many cellular events, including cell cycle control. Our previous studies suggested that the connexin-43 gap junction protein may be a target of activated MAP kinase and that MAP kinase may regulate connexin-43 function. We identified the sites of MAP kinase phosphorylation in in vitro studies as the consensus MAP kinase recognition sites in the cytoplasmic carboxyl tail of connexin-43, Ser255, Ser279, and Ser282. In this study, we demonstrate that activation of MAP kinase by ligand-induced activation of the epidermal growth factor (EGF) or lysophosphatidic acid receptors or by pervanadate-induced inhibition of tyrosine phosphatases results in increased phosphorylation on connexin-43. EGF and lysophosphatidic acid-induced phosphorylation on connexin-43 and the down-regulation of gap junctional communication in EGF-treated cells were blocked by a specific mitogen-activated protein kinase kinase inhibitor (PD98059) that prevented activation of MAP kinase. These studies confirm that connexin-43 is a MAP kinase substrate in vivo and that phosphorylation on Ser255, Ser279, and/or Ser282 initiates the down-regulation of gap junctional communication. Studies with connexin-43 mutants suggest that MAP kinase phosphorylation at one or more of the tandem Ser279/Ser282 sites is sufficient to disrupt gap junctional intercellular communication.

Activation of the Ras/Raf/mitogen-activated protein kinase kinase/mitogen-activated protein (MAP) kinase signaling cascade is initiated by activation of growth factor receptors and regulates many cellular events, including cell cycle control. Our previous studies suggested that the connexin-43 gap junction protein may be a target of activated MAP kinase and that MAP kinase may regulate connexin-43 function. We identified the sites of MAP kinase phosphorylation in in vitro studies as the consensus MAP kinase recognition sites in the cytoplasmic carboxyl tail of connexin-43, Ser 255 , Ser 279 , and Ser 282 . In this study, we demonstrate that activation of MAP kinase by ligand-induced activation of the epidermal growth factor (EGF) or lysophosphatidic acid receptors or by pervanadate-induced inhibition of tyrosine phosphatases results in increased phosphorylation on connexin-43. EGF and lysophosphatidic acid-induced phosphorylation on connexin-43 and the down-regulation of gap junctional communication in EGF-treated cells were blocked by a specific mitogen-activated protein kinase kinase inhibitor (PD98059) that prevented activation of MAP kinase. These studies confirm that connexin-43 is a MAP kinase substrate in vivo and that phosphorylation on Ser 255 , Ser 279 , and/or Ser 282 initiates the down-regulation of gap junctional communication. Studies with connexin-43 mutants suggest that MAP kinase phosphorylation at one or more of the tandem Ser 279 /Ser 282 sites is sufficient to disrupt gap junctional intercellular communication.
Connexin-43 (Cx43) 1 is the 43-kDa member of a conserved family of membrane spanning gap junction proteins. The connexin proteins have intracellular amino and carboxyl termini and four membrane spanning regions that form two extracel-lular loops and one intracellular loop (2)(3)(4). There is considerable homology among the connexins with the greatest divergences in the intracellular loops and the carboxyl-terminal tails. Connexins are assembled into hexamers or connexons in the trans Golgi network prior to insertion into the plasma membrane (4). Connexons in one cell dock with connexons in adjacent cells through non-covalent interactions that involve the extracellular loops (5-7). An aqueous pore or channel (gap junction) forms and allows for the passive intercellular exchange of small molecular mass molecules, ions, and second messengers (Ͻ1000 Da).
The different connexin proteins form channels with unique functional properties including differences in the size and charge of the molecules and ions that can traverse the channel (2)(3)(4)8). Channel gating can be modified by physiologic parameters (such as cellular pH and Ca 2ϩ concentration (9 -13)) and by pharmacological agents (such as 12-O-tetradecanoylphorbol-13-acetate and retinoic acid (14 -17)). Cell-to-cell communication mediated through gap junctions is known to be essential to the synchronization of events such as contractions in the myocardium and uterus (18,19) and is thought to play a critical role in regulating cell growth and differentiation (20 -22). Down-regulation of gap junctional communication (GJC) has been noted in many tumor cells (20,23,24). Conversely, the up-regulation of GJC in cells deficient in communication has been associated with a decrease in cellular growth rates (25)(26)(27)(28). Recently, a study by Martyn et al. (29) has reported that communication-deficient fibroblast cell lines isolated from Cx43 knock-out (K/O) mouse embryos display a subset of the properties of transformed cells, including increased growth rates. These data further advance the hypothesis that the loss of GJC is associated with the progression toward neoplastic transformation.
Some posttranslational phosphorylation on serine is required for the proper assembly of Cx43 gap junctions (4). However, the sites of phosphorylation and the kinases involved have not as yet been identified. Additional posttranslational phosphorylation on serine and/or on tyrosine has been associated with a decrease in GJC. Down-regulation of GJC in epidermal growth factor (EGF)-treated cells and cells transformed with v-ras has been associated with increased serine phosphorylation on Cx43 (30 -33), whereas tyrosine phosphorylation on Cx43 has been associated with decreased GJC in cells transformed with the v-src or v-fps oncogene (34 -37).
EGF induces disruption of GJC and increased serine phosphorylation on Cx43 in a rapid and transient manner that is not mediated by the activation of 12-O-tetradecanoylphorbol-13-acetate-sensitive isoforms of protein kinase C (30,31). Mitogen-activated protein (MAP) kinase, another candidate serine/threonine kinase activated by the EGF receptor (38 -40), has been demonstrated to phosphorylate a recombinant Cx43 preparation in vitro on phosphotryptic peptides that co-mi-grated with phosphotryptic peptides obtained from EGFtreated cells, implicating a potential role for MAP kinase in Cx43 phosphorylation and the regulation of GJC (31).
We characterized MAP kinase-mediated phosphorylation on Cx43 utilizing a glutathione S-transferase (GST) fusion protein that contained the carboxyl-terminal cytoplasmic tail (CT) of Cx43 (aa Val 236 -Ile 382 ). Site-directed mutagenesis, phosphotryptic peptide analysis, and peptide sequencing were used to identify the specific sites of serine phosphorylation on Cx43 (41). MAP kinase phosphorylated Cx43 in vitro on two tryptic peptides that contained the three MAP kinase consensus recognition sequences (tryptic peptide Ser 244 -Lys 258 , phosphorylated at Ser 255 , and tryptic peptide Tyr 265 -Lys 287 , phosphorylated at Ser 279 and at Ser 282 ). When we deleted the three consensus MAP kinase phosphorylation sites by altering the serine residues to alanine, MAP kinase still phosphorylated Cx43 in vitro on the Tyr 265 -Lys 287 peptide at Ser 272 and at Ser 273 . Ser 273 fits the minimal recognition sequence for MAP kinase (serine residue followed by a carboxyl-terminal proline (42)). Phosphorylation at these alternate serine sites occurred only in the absence of the consensus MAP kinase phosphorylation sites. Taken together, our studies have supported the hypothesis that EGF-induced activation of the Ras/Raf/MEK (MAP kinase kinase)/MAP kinase signal transduction pathway leads to MAP kinase-mediated phosphorylation on Cx43 at Ser 255 , Ser 279 , and Ser 282 (Fig. 7). The data suggest that phosphorylation occurring at one or more of these MAP kinase phosphorylation sites may be sufficient to disrupt GJC.
Lysophosphatidic acid (LPA) has also been reported to downregulate GJC and to increase serine phosphorylation on Cx43 (43). LPA binds to an ϳ39-kDa membrane receptor and transmits a signal through interaction with a heterotrimeric G protein (Fig. 7), leading to the activation of Ras (44) and to signal transduction through the Ras/Raf/MEK/MAP kinase pathway (38 -40). Partially purified MAP kinase from LPA-stimulated WB rat cells was shown to phosphorylate a Cx43 peptide, aa 247-260, that contained the Ser 255 site (43). Zhao et al. (45) have reported that treating cells with pervanadate (a potent inhibitor of tyrosine phosphatases) was sufficient to trigger the activation of Raf, MEK, and MAP kinase (Fig. 7). Pervanadate dramatically increased the overall level of tyrosine phosphorylation on cellular proteins in the treated HeLa cells. Pervanadate has also been reported to down-regulate GJC and to increase phosphorylation on Cx43 in studies with hamster lung fibroblasts (46).
In the studies reported here, we have directly examined the role of MAP kinase in mediating the EGF-induced increased phosphorylation of Cx43 in vivo. We expressed wild type (wt) and mutant forms of Cx43 in communication-deficient cells and examined the effects of EGF, LPA, and pervanadate on Cx43 phosphorylation in the presence or absence of a specific MEK inhibitor. Compound PD98059 has been demonstrated to block MEK activation in many cells ((47, 48) Fig. 7) and thus prevent the MEK-mediated phosphorylation of MAP kinase at the Thr 183 and Tyr 185 sites that are essential for MAP kinase activity (49). Our studies have demonstrated that Cx43 is a target of the MAP kinase signaling pathway in vivo and that the activation of MAP kinase decreases Cx43-mediated junctional conductance and permeability. These studies demonstrate that phosphorylation on one or more of the MAP kinase consensus serine phosphorylation sites is responsible for the disruption of GJC.

MATERIALS AND METHODS
Cell Culture, Metabolic Labeling, and Cx43 Immunoprecipitation-The Cx43 knock-out (K/O) mouse cell line, Ϫ/Ϫ3 (29), human HeLa cells, and K/O or HeLa cells expressing exogenous Cx43 were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 10% fetal calf serum (Hyclone) in a humidified 5% CO 2 incubator at 37°C as described (29). Confluent monolayers were metabolically labeled for 2-3 h with [ 35 S]methionine (Amersham Pharmacia Biotech) at 100 Ci/ml in methionine-free basal minimum Eagle's medium supplemented with 4% calf serum. The MEK inhibitor (PD98059, New England Biolabs) or dimethyl sulfoxide (Me 2 SO) control was used to pretreat cells for 1 h prior to stimulation with EGF, LPA, or pervanadate as described below. Cells were treated with EGF, LPA, or pervanadate during the last 15-30 min of the labeling period. At the end of the labeling period, the monolayers were rinsed twice with cold phosphatebuffered saline (PBS) containing 10 mM NaF, 160 M Na 3 VO 4 , and 1 mM phenylmethylsulfonyl fluoride (PMSF) and then frozen at Ϫ20°C. Cells were lysed on ice with RIPA buffer (150 mM NaCl, 1% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 10 mM Tris (pH 7.2)) containing phosphatase and protease inhibitors (10 mM NaF, 160 M Na 3 VO 4 , and 1 mM PMSF). Cx43 was immunoprecipitated from clarified cell lysates with rabbit antiserum directed against the carboxyl-terminal of Cx43 (aa 368 -382) and activated protein A-Staphylococcus aureus (41). Controls for nonspecific immunoprecipitation were performed with half of each cell lysate using non-immune rabbit serum. Immunoprecipitates were resolved on SDS-polyacrylamide gradient gels (7.5-15% acrylamide), and radiolabeled Cx43 was visualized by radioautography of the dried fluorographed gels. For phosphoamino acid analysis, cell monolayers were labeled for 2-3 h with 32 P i (NEN Life Science Products, NEX-053) at 1-2 mCi/ml in phosphate-free basal minimum Eagle's medium with 4% calf serum. Cx43 was immunoprecipitated from clarified cell lysates, resolved on gradient gels, and transferred to Immobilon-P membranes (Millipore). Phosphorylated Cx43 was visualized by radioautography, excised from the membrane, acid-hydrolyzed, lyophilized, and mixed with non-radioactive internal phosphoamino acid standards. Phosphoamino acids were resolved by two-dimensional electrophoresis at pH 1.9 and pH 3.5 on thin layer cellulose plates as described (50). Migration of the phosphoamino acid standards was visualized with ninhydrin prior to radioautography at Ϫ70°C.
Expression of Exogenous Cx43-The EcoRI insert containing the G2 rat Cx43 gene with 5Ј and 3Ј non-coding sequences (51) was excised from Bluescript with BamHI and SalI and inserted into the pBABE retroviral vector (52). The pBABE-Cx43 DNA was transfected into the PE501 packaging cell line using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol. Virus from the PE501 cells was amplified by infecting a second packaging cell line, PA317. The PE501 and PA317 packing cell lines were generously provided by Dr. Dusty Miller (53). Virus harvested from the PA317 cells was used to infect Cx43 K/O cells or HeLa cells. Infected cells were selected using puromycin (6 g/ml for the K/O cells or 0.5 g/ml for the HeLa cells) and subcloned by limiting dilution.
Mutant forms of Cx43 were prepared from double-stranded DNA using the Chameleon site-directed mutagenesis kit (Stratagene) with primers designed to alter a serine residue(s) to alanine or to aspartate (41). Selection for mutants was provided by the simultaneous use of a selection primer designed to eliminate a restriction site in the plasmid DNA (54). The fidelity of the mutants was confirmed by sequencing the full-length Cx43 DNA.
The K/O cell clone expressing wt Cx43 used in this study (clone G3) has been characterized previously ((29), termed clone 3R2). A second wt clone was also used for the GJC and junctional conductance studies (clone C9). The K/O cell clone, G11, expressing the S255A,S279A,S282A Cx43 mutant was used for most of these studies. Unless otherwise noted all data reported are for this clone. However, some results were also obtained with a second S255A,S279A,S282A Cx43 mutant clone, H5. The HeLa cell clones expressing wt Cx43 or the S255D Cx43 mutant used in this study were clones A7-A12 and C2-A5, respectively.
Immunofluorescence Microscopy-Clones stably expressing wt Cx43 or mutant Cx43 were initially screened for membrane-localized protein in punctate gap junctions by immunofluorescence microscopy. Cells were grown overnight on glass coverslips, rinsed with PBS, and fixed in ice-cold 100% methanol for 30 min at Ϫ20°C. Cell membranes were permeabilized with 1% Triton X-100 in PBS for 10 min, rinsed twice with PBS, and blocked with 1% bovine serum albumin (BSA) in PBS. The coverslips were then incubated with rabbit anti-Cx43 at 1:1000 dilution in 1% BSA/PBS for 1 h at room temperature, rinsed twice with PBS and twice with 1% BSA/PBS, and then incubated with goat antirabbit fluorescein isothiocyanate conjugate (Sigma), 1:70 for 1 h in the dark. Coverslips were rinsed three times with PBS, mounted on glass slides with 50% glycerol, 50% PBS containing 0.1% phenylenediamine (Sigma), and visualized using a Zeiss Axioplan Universal microscope equipped with epifluorescence.

Measurement of GJC by Dye Transfer-
The membrane-impermeant fluorescent dye, Lucifer Yellow (Sigma), was used to characterize cell clones for GJC (transfer of dye from a microinjected cell to neighboring cells) and to monitor GJC in treated cells. Single cells were microinjected with 10% (w/v) Lucifer Yellow dye (457 Da) in 330 mM LiCl with a glass micropipette (Flaming-Brown P80/PC micropipette puller, Sutton Instrument Co.) using an Eppendorf micromanipulator and Eppendorf pneumatic injector. Dye transfer to neighboring cells was determined after ϳ3 min using a Zeiss Axiovert 10 phase contrast inverted microscope equipped with epifluorescence. Approximately 85% of the K/O cells expressing wt Cx43 or the S255A,S279A,S282A Cx43 mutant communicated to neighboring cells, and all injections were used to determine the mean value for GJC. Approximately 60% of the HeLa cells expressing wt Cx43 or the S255D Cx43 mutant communicated. Only those injections that resulted in dye transfer to at least one other cell were used to determine the mean value for GJC. Dye transfer was determined in the cells treated with EGF ϳ25-30 min after the addition of EGF in injections over ϳ15-30 min.
Measurement of Junctional Conductance-C9 cells expressing wt Cx43 or G11 cells expressing the S255A,S279A,S282A Cx43 mutant were trypsinized and replated in culture medium on glass coverslips. After 0.5 h the cells were removed from the incubator and maintained at room temperature in a 5% CO 2 in air environment until used for electrophysiological study. Coverslips were mounted in a perfusion chamber, pairs of cells visualized, and patch type microelectrodes lowered onto the surface of each cell. After establishment of the whole cell recording mode, macroscopic junctional conductance was continuously monitored, as described previously (29), during exposure of the cells to 0 or 100 ng/ml EGF.
Cell Treatment Protocols-Confluent 60-mm monolayers were treated with 100 ng/ml recombinant human EGF (U. S. Biochemical Corp. and Amersham Pharmacia Biotech) in conditioned medium (or phosphate-deficient or methionine-free media for metabolically labeled cells) for 30 min at 37°C. Cells were treated with 50 M LPA (Sigma, 5 mM stock in PBS/1% fatty acid-free BSA) in 1 ml of medium for 15 min at 37°C. Pervanadate was prepared freshly for each experiment as a 1:1 mix of 100 mM H 2 O 2 and 100 mM Na 3 VO 4 , incubated at room temperature for 10 -20 min prior to addition to the monolayers in 1 ml of medium (final concentration 100 M pervanadate). Cells were incubated with pervanadate for 20 min at 37°C. To block MEK activation, cells were treated with 50 M PD98059 (50 mM stock in Me 2 SO) for 1 h at 37°C prior to stimulation with EGF, LPA, or pervanadate. Controls for these experiments were treated with 0.1% Me 2 SO. Following treatment, cells were used for GJC measurements or the monolayers were rinsed twice with cold PBS containing 10 mM NaF, 160 M Na 3 VO 4 , 1 mM PMSF and stored frozen. Metabolically labeled cells were treated with the MEK inhibitor or Me 2 SO, and EGF, LPA, or pervanadate during the last part of the labeling period.
In Vitro MAP Kinase Assays-Confluent monolayers were pretreated with MEK inhibitor or Me 2 SO (control) for 1 h at 37°C and then stimulated with EGF. At the end of the treatment time, the monolayers were rinsed twice with cold PBS with phosphatase and protease inhibitors (10 mM NaF, 160 M Na 3 VO 4 , and 1 mM PMSF) and frozen at Ϫ70°C. In vitro MAP kinase assays were carried out with clarified cell lysates (in RIPA buffer with 10 mM NaF, 160 M Na 3 VO 4 , and 1 mM PMSF). Reactions contained 7.5 l of the cell lysate (2.5 ml of lysate for a 60-mm plate) and 15 l kinase reaction buffer. Final concentrations in the reaction were 20 mM HEPES (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 20 M unlabeled ATP, 0.5 mg/ml myelin basic protein (MBP, Sigma) as a MAP kinase substrate, and 1 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech). Reactions were incubated at 30°C for 20 min and stopped by addition of 7.5 l of 4ϫ SDS sample buffer and placing the samples in boiling water. Proteins were resolved on 12% acrylamide SDS-polyacrylamide gels and stained with Coomassie Blue. The gels were dried and radioautographed at Ϫ70°C.
Immunoblotting-Confluent monolayers in 35-mm dishes were treated with 0.1% Me 2 SO or MEK inhibitor (50 M) for 1 h and then treated with or without EGF (100 ng/ml) for 2 min. At the end of the treatment period, the monolayers were rinsed with cold PBS and frozen at Ϫ70°C. Frozen cells were lysed in 100 l of hot SDS sample buffer, heated in boiling water, and centrifuged for 30 min in an Eppendorf microcentrifuge. The supernatant proteins (30 l) were resolved on 12% acrylamide SDS-polyacrylamide gels and transferred to Immobilon (Millipore). The membrane was blocked with 5% BSA in PBS, 0.1% Tween and incubated with an antibody specific for the active, phosphorylated form of MAP kinase (Promega) or with an antibody to p42 MAP kinase (Santa Cruz). Immunoblots were developed with a peroxidase-conjugated secondary antibody using the ECL chemiluminescence system (Amersham Pharmacia Biotech).

RESULTS
Previous studies implicated MAP kinase in mediating the EGF-induced phosphorylation of Cx43 and the accompanying disruption of GJC (30,31,41). The goal of the present study was to determine whether MAP kinase is directly responsible for these EGF-induced effects on Cx43 in vivo. Our experimental approach to this problem was to analyze cells expressing wt or phosphorylation site mutants of Cx43 treated with different agonists to activate MAP kinase and/or with a specific MEK inhibitor to block the activation of MAP kinase.
Phosphorylation Site Mutants of Cx43 Are Localized in Punctate Gap Junction Plaques-We prepared a Cx43 mutant that lacked the three consensus MAP kinase phosphorylation sites (serine to alanine mutations, S255A, S279A, S282A, termed here as S255A,S279A,S282A). This triple phosphorylation site mutant and wt Cx43 were expressed in the non-communicating Cx43 K/O cells. We also expressed wt Cx43 and an aspartate mutant of Cx43 (S255D) in HeLa cells. Membrane-localized Cx43 was apparent in punctate gap junction plaques as demonstrated by immunofluorescence microscopy in cells expressing wt Cx43 (Fig. 1, panels B (34) and others (4,16) have previously shown that Cx43 exhibits a multiple banding pattern on SDS-PAGE where the non-phosphorylated Cx43 (NP) migrates most rapidly and the phosphorylated isoforms (P1 and P2) migrate more slowly. Compound PD98059 has been reported to specifically block the activation of MEK (MAP kinase kinase, Fig. 7) and thus prevent the activation of MAP kinase in many cells (47,48). The MEK inhibitor in Me 2 SO (or Me 2 SO alone) was used to pretreat some monolayers prior to the addition of EGF to examine the role of MAP kinase in the stimulated cells.
Cx43 obtained from unstimulated cells expressing wt Cx43 ( Fig. 2A, lane 2) migrated by SDS-PAGE primarily in the non-phosphorylated form (NP) with some Cx43 migrating more slowly as a phosphorylated species (P1). When the cells were treated with EGF, most of the Cx43 migrated as phosphorylated isoforms consisting primarily of the P1 form as well as the slowest migrating P2 form ( Fig. 2A, lane 4). In cells pretreated with the PD98059 MEK inhibitor the increase in the P1-and P2-phosphorylated isoforms was not observed after EGF stimulation, and Cx43 was primarily present in the non-phosphorylated form (Fig. 2A, lane 6). Control non-immune precipitations of cell lysates are shown for comparison for each treatment ( Fig. 2A, lanes 1, 3 and 5). Thus, the PD98059 MEK inhibitor that is reported to prevent the activation of MAP kinase appeared to block the EGF-induced increased phosphorylation of Cx43.
EGF treatment of the S255A,S279A,S282A Cx43 mutant also resulted in increased phosphorylation of Cx43 (see Fig. 2B, compare lane 4 with lane 2). The increase in the P1 and P2 isoforms of Cx43 was prevented in the cells pretreated with the MEK inhibitor (Fig. 2B, lane 6), indicating that increased phosphorylation of the S255A,S279A,S282A Cx43 mutant was also dependent upon the activation of MAP kinase. The EGF-induced increased phosphorylation of Cx43 in the S255A,S279A,S282A mutant was also demonstrated with Cx43 immunoprecipitated from 32 P i -labeled cells, on Western blots of whole cell lysates from unlabeled cells, and using a second clone (H5) expressing the S255A,S279A,S282A mutant in the K/O cells (data not shown).
The MEK Inhibitor Blocks Activation of MAP Kinase in EGFtreated Cells-MAP kinase is activated through a signal transduction pathway initiated by ligand-induced activation of the EGF receptor (38 -40). We examined the effect of the PD98059 MEK inhibitor on the activation of MAP kinase in EGF-treated cells in in vitro kinase assays using MBP as a MAP kinase substrate (Fig. 3). In unstimulated cells the basal level of MAP kinase activity, measured by phosphorylation of MBP, was very low (Fig. 3A, zero time points, K/O cells expressing the S255A,S279A,S282A Cx43 mutant (clone H5), upper panel or HeLa cells expressing wt Cx43, lower panel). When the cells were stimulated with EGF, there was a rapid and significant increase in MAP kinase activity as shown by increased phosphorylation of MBP at the 2-and 5-min time points. A similar significant increase in MAP kinase activity at 3 min was obtained with the S255A,S279A,S282A G11 clone (data not shown). When the cells were pretreated with the MEK inhibitor for 1 h and then stimulated with EGF, the increase in MAP kinase activity was blocked in the K/O cells expressing the S255A,S279A,S282A Cx43 mutant (Fig. 3B, upper panel) and was significantly decreased in the HeLa cells expressing wt Cx43 (Fig. 3B, lower panel).
To provide further evidence that the MEK inhibitor interfered with MAP kinase activation, whole cell lysates from K/O cells expressing wt Cx43 (clone G3) untreated or treated with EGF in the presence or absence of the MEK inhibitor were subjected to SDS-PAGE and transferred to Immobilon. Immunoblot analysis with an antibody that specifically recognizes the activated, phosphorylated form of MAP kinase clearly demonstrated MAP kinase activation at 2 min after EGF stimulation and that the MEK inhibitor interfered with MAP kinase activation (Fig. 3C, upper panel). The blot was reprobed with an antibody to the p42 form of MAP kinase (Fig. 3C, lower panel) to visualize total MAP kinase in the cell lysates.
LPA-and Pervanadate-induced Increased Phosphorylation of Cx43-LPA and pervanadate have also been reported to induce increased phosphorylation on Cx43 and down-regulate GJC (43,46). In addition, MAP kinase was activated by LPA treatment in WB rat liver cells (43) and by pervanadate in HeLa cells (45). We were interested in determining the effects of LPA and pervanadate on Cx43 phosphorylation in our cells (Fig. 4). Both LPA and pervanadate stimulated increased phosphorylation of Cx43 in cells expressing wt Cx43 as shown by increases in the more phosphorylated isoforms (P1 and P2) of Cx43 (see  1 and 2), treated with 100 ng/ml EGF for 30 min (lanes 3 and 4), or pretreated with 50 M MEK inhibitor for 1 h and then treated with EGF for 30 min (lanes 5 and 6). Cell lysates were divided in half and immunoprecipitated with non-immune rabbit serum as a control for nonspecific immunoprecipitation (lanes 1, 3, and 5) or with anti-Cx43 peptide antibody (lanes, 2, 4, and  6).  Table  I). The G3 clone expressing wt Cx43 in the K/O cells (see Fig.  1B) communicates to ϳ4 -5 neighboring cells as described previously (Ref. 29; termed clone 3R2). The C9 clone expressing wt Cx43 in the K/O cells communicates to ϳ10 cells (Fig. 1C). The G11 clone expressing the S255A,S279A,S282A Cx43 mutant in the K/O cells (Fig. 1D) communicates to ϳ4 neighboring cells and the H5 clone (Fig. 1E) to ϳ2 cells. EGF treatment induced an ϳ36 -40% decrease in GJC in the K/O cells expressing wt Cx43 for the G3 and C9 clones, respectively, and an ϳ22% decrease in GJC in the HeLa cells expressing wt Cx43 (see Fig.  5A and Table I). However, EGF appeared to have no effect on GJC in the G11 S255A,S279A,S282A Cx43 mutant that lacks the consensus MAP kinase phosphorylation sites. Pretreatment with the MEK inhibitor prevented the EGF-induced decrease in GJC in the K/O cells expressing wt Cx43, clone C9 (n ϭ 3), and in the HeLa cells expressing wt Cx43 (n ϭ 3, see Fig. 5B).

pervanadate-treated HeLa cells). LPA and pervanadate also induced increased phosphorylation of
Dual whole cell voltage clamp was used to obtain a more EGF induced a significant decrease in junctional conductance within minutes in the wt Cx43 expressing cells (Fig. 6A, closed circles) but had little effect on the S255A,S279A,S282A Cx43 mutant expressing cells (Fig. 6B, closed circles). These data demonstrated that despite ongoing phosphorylation of the mutated connexin protein, elimination of the main MAP kinase consensus sites interfered with EGF-stimulated down-regulation of junctional function.

A Negative Charge at Position 255 Did Not Prevent GJC or EGF-induced Disruption of GJC-Disruption of GJC in EGF-
treated cells could be due to the presence of a negatively charged phosphate group at one or more of the three MAP kinase serine phosphorylation sites, i.e. at Ser 255 (on the Ser 244 -Lys 258 tryptic peptide), and/or at Ser 279 , and/or Ser 282 (on the Tyr 265 -Lys 287 tryptic peptide). To mimic the effect of a negatively charged phosphate group at the Ser 255 position, we prepared a serine to aspartate Cx43 mutant (S255D) and expressed this mutant in HeLa cells. The S255D Cx43 protein was expressed in punctate gap junctions (see Fig. 1H) and the cells communicated to ϳ7 neighboring cells (Fig. 5A and Table  I). Thus, the presence of a negative charge at position 255 in Cx43 did not prevent GJC as determined by the transfer of Lucifer Yellow dye. EGF treatment induced an increase in serine phosphorylation of Cx43 (data not shown) and an ϳ31%  1 and 2), treated with 100 M pervanadate for 20 min (lanes 3 and 4), or pretreated with MEK inhibitor for 1 h prior to pervanadate treatment for 20 min (lanes 5 and 6). Cell lysates were divided in half and immunoprecipitated with non-immune rabbit serum as a control for nonspecific immunoprecipitation (lanes 1, 3, and 5) or with anti-Cx43 peptide antibody (lanes 2, 4, and 6).  Table I). DISCUSSION Our earlier studies demonstrated that MAP kinase phosphorylated full-length Cx43 and a GST fusion protein containing the cytoplasmic carboxyl tail (CT) of Cx43 (aa Val 236 -Ile 382 ) on tryptic peptides that co-migrated with a subset of the phosphotryptic peptides obtained from Cx43 in EGF-treated cells (31,41). Furthermore, MAP kinase phosphorylated the GST-Cx43-CT fusion protein at the consensus MAP kinase phosphorylation sites, Ser 255 , Ser 279 , and Ser 282 (41). These studies supported a role for MAP kinase in the regulation of Cx43 by serine phosphorylation and implicated one or more of the MAP kinase serine sites as functionally important for the passage of molecules through the Cx43 channel.
In the current study we examined the role of MAP kinase in mediating Cx43 phosphorylation in vivo in cells stimulated with EGF, LPA, or pervanadate to induce MAP kinase activation. In some experiments, compound PD98059 was used to prevent the activation of MEK and MAP kinase. We also studied a Cx43 mutant that lacked the 3 MAP kinase consensus phosphorylation sites to determine whether the absence of EGF-induced phosphorylation at these sites prevented the EGF-induced disruption of GJC.
Cx43 was localized in gap junctional plaques in cells expressing wt and mutant forms of Cx43 (Fig. 1). EGF induced an increase in phosphorylation on Cx43 in the K/O cells expressing wt Cx43, and increased phosphorylation was dependent on the activation of MAP kinase (Fig. 2A). We were surprised to find that EGF also induced increased phosphorylation on the S255A,S279A,S282A Cx43 mutant (Fig. 2B), since this mutant lacks the 3 consensus MAP kinase phosphorylation sites. In this instance, the increased phosphorylation was also dependent on the activation of MAP kinase. Phosphorylation most likely occurs on the Tyr 265 -Lys 287 tryptic peptide at the Ser 272 and Ser 273 alternate MAP kinase sites that were identified in our earlier in vitro study (41). It is important to note that phosphorylation at these alternate sites in vitro occurred only in the absence of the consensus MAP kinase phosphorylation sites.
Our efforts to obtain high resolution tryptic maps of the Cx43 isolated from the EGF-treated cells expressing the S255A,S279A,S282A Cx43 mutant were thwarted due to the relatively low levels of the Cx43 protein expressed in these cells. However, we were able to distinguish a phosphotryptic peptide that migrated with the characteristics of the doubly phosphorylated Tyr 265 -Lys 287 peptide (peptide b in our earlier studies) on tryptic maps (data not shown). This is consistent with phosphorylation occurring on the Tyr 265 -Lys 287 peptide at the Ser 272 and Ser 273 sites in the S255A,S279A,S282A Cx43 mutant lacking the MAP kinase consensus phosphorylation sites.
Although PD98059 prevents MEK activation in many cells, MEK activation is not completely blocked in some cells following a strong stimulus, allowing catalytic activation of significant amounts of MAP kinase (48). We were fortunate with the Ϫ/Ϫ3 K/O cells that the PD98059 MEK inhibitor successfully blocked MAP kinase activation (see Fig. 3B for the S255A,S279A,S282A Cx43 mutant in the K/O cells and Fig. 3C for wt Cx43 in the K/O cells) and thus blocked the increased phosphorylation on Cx43 induced by EGF treatment (see Fig.  2A, wt Cx43 and Fig. 2B, S255A,S279A,A282A Cx43 mutant in the K/O cells). Compound PD98059 also substantially reduced the activation of MAP kinase in the EGF-treated HeLa cells expressing wt Cx43 (Fig. 3B).
LPA also stimulated increased phosphorylation of wt Cx43 and the S255A,S279A,S282A Cx43 mutant expressed in the K/O cells (Fig. 4). Increased phosphorylation required the activation of MAP kinase, since it was blocked in cells pretreated with the MEK inhibitor. Hii et al. (43) suggested that LPAinduced increased phosphorylation on Cx43 might occur through the activation of MAP kinase. Our data support the role of MAP kinase in mediating the increased serine phosphorylation on Cx43 observed in LPA-treated cells. The LPA-induced, MAP kinase-dependent phosphorylation on the S255A,S279A,S282A Cx43 mutant presumably occurred at the Ser 272 and Ser 273 alternate MAP kinase sites.
Taken together, these data provide strong evidence that  MAP kinase is the serine/threonine kinase responsible for the EGF-and LPA-induced increased serine phosphorylation on Cx43. Phosphorylation appears to occur at the MAP kinase consensus phosphorylation sites, Ser 255 , Ser 279 , and Ser 282 in cells expressing wt Cx43.
The MEK inhibitor appeared to only partially block the pervanadate-induced increased phosphorylation on Cx43. Pervanadate initiated a greater increase in Cx43 phosphorylation, and most of the Cx43 migrated in the more phosphorylated P2 isoform on SDS-PAGE (see Fig. 4C, lane 6 for wt Cx43 in the HeLa cells and Fig. 4D, lane 6 for the S255A,S279A,S282A Cx43 mutant in the K/O cells). Pervanadate is a potent inhibitor of tyrosine phosphatases and has been reported to increase the overall level of tyrosine phosphorylation on cellular proteins ϳ200-fold in HeLa cells (100 M pervanadate) as compared with the phosphorylation that was observed in the HeLa cells treated with EGF (50 ng/ml (45)). Furthermore, pervanadate induced a sustained activation of MAP kinase in the HeLa cells compared with the more transient activation of MAP kinase induced by EGF treatment (45). We thought that pervanadate treatment might induce tyrosine phosphorylation on Cx43 as well as serine phosphorylation, and the MEK inhibitor would not prevent tyrosine phosphorylation. We subjected Cx43 isolated from 32 P i -labeled cells expressing the S255A,S279A,S282A Cx43 mutant to phosphoamino acid analysis. Phosphotyrosine was detected in Cx43 isolated from the pervanadate-treated cells but not in Cx43 isolated from untreated cells or EGF-treated cells (data not shown). Mikalsen et al. (55) have also recently confirmed that pervanadate induced tyrosine phosphorylation on Cx43 in hamster embryo fibroblasts as had been suggested in their earlier study (46). Some phosphothreonine was also detected in Cx43 isolated from our pervanadate-treated cells. Pretreating cells with the MEK inhibitor did not prevent the tyrosine or threonine phosphorylation on Cx43. MAP kinase activation and its blockade by the MEK inhibitor were demonstrated on a Western blot of cell lysates from the pervanadate-treated K/O cells expressing the S255A,S279A,S282A Cx43 mutant probed with an antibody to phosphotyrosine to detect activated MAP kinase (data not shown). These data support a role for MAP kinase in mediating increased serine phosphorylation on Cx43 and the participation of additional kinase(s) in mediating the tyrosine and threonine phosphorylation on Cx43 in the pervanadate-treated cells. The kinase(s) responsible for the pervanadate-induced tyrosine phosphorylation on Cx43 and the role that such a kinase(s) may have in the regulation of Cx43 function under basal conditions is unknown.
Earlier studies correlated the decrease in GJC with the increase in serine phosphorylation of Cx43 in EGF-treated T51B rat liver epithelial cells (30,31). However, there has been no direct evidence that the EGF-induced phosphorylation of Cx43 is responsible for the disruption of GJC. Our data demonstrated that activation of MAP kinase was essential for the EGF-induced decrease in GJC. EGF decreased GJC by ϳ36 -40% in the K/O cells expressing wt Cx43 and ϳ22% in the HeLa cells expressing wt Cx43 (Fig. 5A and Table I). Junctional conductance was decreased ϳ55% in the C9 wt Cx43 expressing K/O cells. The MEK inhibitor blocked the EGF-induced decrease in GJC in the HeLa cells and K/O cells expressing wt Cx43 compared with cells treated with the MEK inhibitor alone (Fig. 5B). These results indicated the likelihood that phosphorylation of Cx43 by MAP kinase not only regulates Cx43 function but, in addition, phosphorylation on one or more of the MAP kinase consensus serine phosphorylation sites is sufficient to diminish the passage of molecules through Cx43 gap junction channels. The slight stimulation of GJC induced by the MEK inhibitor in the HeLa cells may be due to inhibition of a basal level of MAP kinase activation in these cells which could contribute to the regulation of Cx43 in unstimulated cells.
EGF had no effect on GJC in the K/O cells expressing the S255A,S279A,S282A Cx43 mutant (Fig. 5A and Table I)  duce MAP kinase-dependent phosphorylation of the mutant connexin protein, presumably at the alternate Ser 272 and Ser 273 MAP kinase sites that are normally not phosphorylated by MAP kinase in vivo. This suggests that phosphorylation at these alternate sites has no effect on GJC. Furthermore, high resolution junctional conductance measurements confirmed that EGF has no effect on the function of gap junctions formed by the S255A,S279A,S282A Cx43 mutant (Fig. 6B).
The S255D Cx43 site-directed mutant established a level of GJC comparable to that observed in cells expressing wt Cx43 (see Fig. 5A and Table I). Because the aspartate residue introduces a negative charge at this position and thus mimics the phosphorylation of Ser 255 , the detection of control levels of GJC suggested that phosphorylation at this site alone is insufficient to disrupt GJC. EGF induced an ϳ31% decrease in GJC in the HeLa cells expressing the S255D mutant ( Fig. 5A and Table I) which correlated with phosphorylation of the S255D Cx43 mutant on serine (data not shown). Therefore, it was likely that the observed phosphorylation in the S255D mutant was occurring on the Tyr 265 -Lys 287 peptide containing the Ser 279 and Ser 282 MAP kinase phosphorylation sites. These data suggested that EGF-induced serine phosphorylation on the Tyr 265 -Lys 287 peptide, at Ser 279 and/or at Ser 282 , is sufficient to disrupt the passage of molecules through the Cx43 channel and that phosphorylation at the Ser 255 site is not required for the disruption of GJC.
The carboxyl-terminal tail of Cx43 is thought to be important in the regulation of GJC by phosphorylation and the regulation of channel function by pH gating (11). However, very little functional data are available on the specific sites in Cx43 that are essential for GJC. The studies of Swenson et al. (56) demonstrated down-regulation of GJC in Xenopus oocytes expressing wt Cx43 and v-Src but not in cells expressing a Y265F mutant of Cx43 and v-Src. These authors concluded that phosphorylation of Tyr 265 was sufficient to interfere with GJC. Our studies have indicated that phosphorylation at the Ser 279 and/or Ser 282 sites is sufficient to disrupt GJC. It is interesting to note that a proline-rich Cx43 peptide (Cys 271 -Lys 287 ) prevented pH gating of Cx43 expressed in Xenopus oocytes. 2 This peptide contains the tandem MAP kinase phosphorylation sites at Ser 279 and Ser 282 and is phosphorylated by MAP kinase in vitro (41). Furthermore, insulin and insulin-like growth factor induced uncoupling in Xenopus oocytes expressing wt Cx43, whereas uncoupling was not observed in oocytes expressing Cx43 that had a deletion of amino acids 261-280 which deleted the Ser 279 MAP kinase site. 3 These studies suggest that residues in this region of the carboxyl-terminal tail of Cx43 are critical to the regulation and function of the Cx43 channel. Interestingly, phosphorylation at Tyr 265 (56) or at Ser 279 and/or Ser 282 (present study) interferes with GJC, whereas phosphorylation of the intervening Ser 272 and Ser 273 sites does not interfere with GJC as demonstrated with the S255A,S279A,S282A Cx43 mutant. This suggests site-specific effects of phosphorylation on Cx43 function rather than more regional effects on protein conformation.
Although it is not known which serine sites in Cx43 undergo serine phosphorylation during the normal assembly of gap junctions, phosphorylation at the Ser 255 , Ser 279 , and Ser 282 sites is not required, since the S255A,S279A,S282A Cx43 mutant was membrane-localized in punctate gap junctions and established GJC. Furthermore, the presence of a negative charge at the Ser 255 position in the S255D Cx43 mutant did not prevent the normal assembly of the protein into gap junctions.
In summary, our studies have demonstrated that Cx43 GJC is regulated in vivo by the EGF-induced activation of a signal transduction pathway that leads to the activation of MAP kinase and that MAP kinase-mediated phosphorylation on Cx43 at the Ser 279 and/or Ser 282 site(s) is sufficient to disrupt GJC (Fig. 7). We are currently preparing additional Cx43 mutants to determine which of these serine phosphorylation sites is critical for the MAP kinase-mediated disruption of GJC.