Gating Connexin 43 Channels Reconstituted in Lipid Vesicles by Mitogen-activated Protein Kinase Phosphorylation*

The regulation of gap junctional permeability by phosphorylation was examined in a model system in which connexin 43 (Cx43) gap junction hemichannels were reconstituted in lipid vesicles. Cx43 was immunoaffinity-purified from rat brain, and Cx43 channels were reconstituted into unilamellar phospholipid liposomes. The activities of the reconstituted channels were measured by monitoring liposome permeability. Liposomes containing the Cx43 protein were fractionated on the basis of permeability to sucrose using sedimentation in an iso-osmolar density gradient. The gradient allowed separation of the sucrose-permeable and -impermeable liposomes. Liposomes that were permeable to sucrose were also permeable to the communicating dye molecule lucifer yellow. Permeability, and therefore activity of the reconstituted Cx43 channels, were directly dependent on the state of Cx43 phosphorylation. The permeability of liposomes containing Cx43 channels was increased by treatment of liposomes with calf intestinal phosphatase. Moreover, liposomes formed with Cx43 that had been dephosphorylated by calf intestinal phosphatase treatment showed increased permeability to sucrose. The role of phosphorylation in the gating mechanism of Cx43 channels was supported further by the observation that phosphorylation of Cx43 by mitogen-activated protein kinase reversibly reduced the permeability of liposomes containing dephosphorylated Cx43. Our results show a direct correlation between gap junctional permeability and the phosphorylation state of Cx43.

Intercellular communication through the gap junction is believed to be modulated during cellular processes such as cell cycle progression (10 -12), embryogenesis, and development (13,14). A correlation exists between decreases in gap junctional communication (GJC) 1 and the cellular events that lead to cell cycle progression; for example, GJC is reduced during the G 1 to S phase transition of the cell cycle (11,12). In addition, factors that affect cell proliferation were found to inhibit GJC, including tumor promoters (15,16), growth factors (10,17), carcinogens (18,19), and oncogene products (20 -22). The temporal loss of GJC in proliferating cells was further supported by the observed growth retardation in cells transfected with overexpressing connexin genes (23)(24)(25).
It has been proposed that GJC is modulated by a mechanism that involves the posttranslational phosphorylation of gap junction proteins. Connexin 43 (Cx43) is a major gap junction protein found in animal heart and brain (26,27). Studies have implicated phosphorylation of Cx43 on tyrosine and/or serine residues as a regulatory mechanism for channel gating. Two types of Cx43 phosphorylation events have been described. One type is the tyrosine phosphorylation of Cx43 by oncogene-induced tyrosine kinases (21,22,28). The other type involves the rapid phosphorylation of Cx43 on serine residues in cells stimulated to grow (10 -12, 29, 30). However, the biochemical mechanism by which posttranslational modification is associated with the modulation of gap junction channels remains obscure.
It has been suggested that mitogen-activated protein (MAP) kinase plays an important role in the loss of GJC during mitotic division (31). Therefore, we sought to determine whether MAP kinase is directly involved in the gating mechanism of gap junction channels. Detailed studies of GJC have been severely constrained by the fact that gap junction channels are not exposed to the extracellular space, and access to the channels is only available via the cytoplasm or dialyzed cytoplasm in a whole-cell patch clamp configuration. Thus, it is difficult to correlate unequivocally the effects of agents that alter channel phosphorylation with channel permeability, as any of these agents might elicit secondary metabolic effects. One approach is to make gap junction channels accessible to study by splitting the isolated gap junctions and reconstituting them in an artificial lipid bilayer. Several studies report the induction of channel activities in a lipid bilayer by incorporating gap junction proteins into lipid vesicles (32)(33)(34)(35)(36).
In the present study, we reconstituted Cx43 gap junction channels in liposomes and measured the permeability with the use of iso-osmolar sucrose/urea density gradient sedimentation (34). Our results indicate that MAP kinase phosphorylation of Cx43 directly modulates channel gating.
Preparation of anti-Cx43 Antibody-The sequence of synthetic peptide SSRASSRPRPDDLEI corresponded to amino acid residues 368 to 382 of the C-terminal region of Cx43 (26). The peptide was conjugated to keyhole limpet hemocyanin, and antisera were then prepared by the method of Beyer et al. (38).
Immunoaffinity Purification of Cx43-Cx43 was immunoaffinity-purified by the modified method of Rhee et al. (36). Brain tissue from Sprague-Dawley rats (ϳ1.3 g) was minced in 25 ml of 5 mM NaHCO 3 buffer (pH 8.2) containing 5 mM EDTA and 1 mM phenylmethylsulfonyl fluoride, and the homogenate was centrifuged at 48,000 ϫ g for 10 min. The crude membrane fraction pellet was solubilized in buffer A (50 mM sodium phosphate (pH 7.4), 50 mM NaCl, 5 mM EDTA, 50 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 mM leupeptin, and 0.75% CHAPS). After centrifugation at 100,000 ϫ g for 1 h, the supernatant was loaded onto an immunoaffinity column in which anti-Cx43 antibody was crosslinked to CNBr-activated Sepharose 4B. After successive washes with buffer A and buffer A plus 0.5 M NaCl, the column was equilibrated with buffer A in which CHAPS was replaced with 80 mM octylglucoside. Bound Cx43 was eluted from the column by brief exposure to 50 mM sodium acetate buffer (pH 2.4) containing 10 mM KCl, 459 mM urea, 0.1 mM EDTA, 80 mM octylglucoside, 1 mM phenylmethylsulfonyl fluoride, and 1 mM leupeptin. The affinity column eluent was rapidly neutralized by adding it directly to 0.3 volumes of neutralization buffer (1 M HEPES (pH 7.4), 10 mM KCl, 459 mM urea, 0.1 mM EDTA, and 80 mM octylglucoside). To prepare a denatured form of Cx43, a portion of the affinity column eluent was incubated at 30°C for 3 h before it was mixed with the neutralization buffer. All the procedures were carried out at 4°C.
Gel Electrophoresis and Immunoblot Analysis-Immunoaffinity-purified Cx43 protein was analyzed by an 8% SDS-polyacrylamide gel. For staining the gel-bound protein with silver nitrate, the gel was preincubated in a mixture of ethanol, acetic acid, and water (4:1:5, v/v/v) for 1 h followed by an additional incubation in ethanol, acetic acid, and water (5:5:90, v/v/v) for 2 h. Protein bands were cross-linked by soaking the gel in a solution containing 1% glutaraldehyde and 0.5 M sodium acetate for 30 min. The gel was rinsed extensively with deionized water, and the protein bands were stained in a freshly made 0.8% ammoniacal silver nitrate solution containing 0.02 N NaOH. After washing, the images were developed by incubating the gel in a solution containing 0.05% citric acid and 0.1% formaldehyde for 5 to 10 min. The reaction was stopped by transferring the gel into 5% Tris buffer containing 2% acetic acid. For immunoblot analysis, the gel-bound protein bands were electrotransferred to a nitrocellulose membrane, and the membrane was blocked by incubation for 1 h at 37°C in 3% gelatin in Tris-buffered saline (25 mM Tris-HCl (pH 8.3), 50 mM NaCl). The blocked membrane was then incubated in Tris-buffered saline containing 0.05% (v/v) Tween 20 and 10 g/ml anti-Cx43 antibody with gentle shaking for 1 h at 37°C. The membrane was transferred to Tris-buffered saline containing 0.05% (v/v) Tween 20 and incubated for 30 min with buffer changes every 10 min. The washed membrane was then incubated for 1 h at 37°C with 0.5 g/ml alkaline phosphatase-conjugated anti-rabbit IgG. The Cx43 bands were developed in 0.1 mg/ml nitro blue tetrazolium and 0.05 mg/ml 5-bromo-4-chloro-3-indolyl phosphate in 10 mM Tris buffer (pH 9.5) containing 10 mM MgCl 2 and 5 mM NaCl.
Sucrose Gradient Analysis of Cx43 Assembly-The oligomeric state of immunoaffinity-purified Cx43 was examined by sucrose density sedimentation as described by Falk et al. (39). Immunoaffinity-purified Cx43 protein was loaded onto a 5-ml linear gradient of 5 to 20% (w/v) sucrose containing 10 mM Hepes (pH 7.4), 459 mM urea, 0.1 mM EDTA, and 80 mM octylglucoside. After centrifugation for 16 h at 200,000 ϫ g, the gradients were fractionated, and 0.1 ml fractions were collected. Aliquots (50 l) from the gradient fractions were transferred to microwell plates, and the amount of Cx43 in each fraction was quantified by enzyme-linked immunosorbent assay using anti-Cx43 antibody. Standard proteins with known sedimentation coefficients (ovalbumin, 3.5 S; bovine serum albumin, 4.3 S) were subjected to sedimentation on a separate gradient and used to determine the S value of Cx43 oligomers.
Cross-linking Analysis of Cx43 Assembly-The quaternary structure of immunoaffinity-purified Cx43 protein was examined by chemical cross-linking using DMS as a coupling agent (40,41). Aliquots of immunoaffinity-purified Cx43 protein (ϳ10 g in 180 l) in 10 mM triethanolamine buffer (pH 8.3) containing 80 mM octylglucoside, 10 mM KCl, and 0.1 mM EDTA were mixed with 20 l of 10 mM triethanolamine buffer (pH 8.3) containing 100 mM DMS. After incubation for 8 h at 4°C, the cross-linking reaction was stopped by adding 1 l of 1 M glycine buffer (pH 7.4) to each sample and incubating the reaction mixtures at room temperature for 30 min. The cross-linked protein samples were then denatured by adding an equal volume of SDS sample buffer containing 125 mM Tris-Cl (pH 6.8), 20% SDS, 10% ␤-mercaptoethanol, and 20% glycerol. The cross-linked complexes were analyzed by electrophoresis on a 4 to 15% SDS-polyacrylamide gradient gel followed by immunoblot analysis using anti-Cx43 antibody as described above.
Reconstitution of Cx43 Channels into Liposomes-Cx43 channels were reconstituted in unilamellar phospholipid vesicles. A procedure for the incorporation of gap junction protein in unilamellar liposomes has been described (36). The lipids phosphatidylcholine, phosphatidylserine, and lissamine rhodamine B-labeled PE were dissolved in chloroform at a molar ratio of 2:1:0.05. Fluorescently labeled PE was used as a marker for liposomes in the gradient fractions. The lipid mixture was dried to a thin film under a stream of nitrogen gas, and the lipid film was suspended in 10 mM HEPES buffer (pH 7.4) containing 459 mM urea, 10 mM KCl, 0.1 mM EDTA, and 80 mM octylglucoside. Immunoaffinity-purified Cx43 was then added to the mixture, and the protein: lipid ratio was adjusted to 1:45 (w/w) for most procedures employed in this study. The protein-lipid mixture was applied to a Sephadex G-50 (Amersham Pharmacia Biotech) column (2 ϫ 100 cm), and the column was eluted with a buffer containing 10 mM HEPES (pH 7.4), 459 mM urea, 10 mM KCl, and 0.1 mM EDTA at the flow rate of 0.1 ml/min. The liposomes were collected in the void volume of the column, whereas octylglucoside was retained by the column. The liposomes were concentrated by centrifugation at 240,000 ϫ g for 18 h at 4°C.
Channel Activity of Cx43 in Liposomes-The permeability of the Cx43 channels reconstituted in liposomes was examined by the technique of transport-specific density shift as described (34). The liposomes were layered onto linear, 0 to 400 mM, iso-osmolar sucrose density gradients in which a reverse linear gradient of 459 to 0 mM urea in the same buffer compensated the sucrose gradient osmotically. The gradient was then centrifuged at 250,000 ϫ g for 12 h at 4°C. The distribution of fluorescently labeled liposomes in the gradient was photographed under UV illumination in the dark. The gradient was eluted through a hole formed at the bottom of a centrifuge tube, and the distribution of the lipid vesicles was determined by measuring the intensity of rhodamine fluorescence in each fraction with a spectrofluorometer (LS-3B, Perkin-Elmer) at 590 nm excited at 560 nm. To measure channel activity of Cx43 in the liposomes, the loss of lucifer yellow fluorescent communication dye from the liposomes was monitored. Liposomes formed with Cx43 in the presence of 3 mM lucifer yellow were layered on a gradient. After centrifugation at 250,000 ϫ g for 12 h, gradients were eluted from the bottom. Each fraction was monitored for the fluorescence of lissamine rhodamine B-labeled PE in the liposome membrane. The amount of lucifer yellow retained in the liposomes was quantified by measuring its fluorescence intensity at 530 nm excited at 428 nm.
Dephosphorylation of Cx43-Cx43 (ϳ30 g) which was either solubilized in detergent or incorporated in liposomes was mixed with 0.1 volume of 50 mM Tris buffer (pH 8.0) containing 1 mM EDTA, 80 mM octylglucoside, and 50 units/ml agarose-conjugated CIP. The reaction mixtures were incubated for 3 h at 30°C with gentle agitation. In the control experiment, an equal amount of agarose beads was added to the mixture to replace the agarose-conjugated CIP. The mixtures were then centrifuged for 5 min at 1,000 ϫ g to remove the agarose beads and conjugated enzyme. Dephosphorylated protein samples were resolved by SDS-polyacrylamide (8%) gel electrophoresis (PAGE) followed by immunoblot analysis with anti-Cx43 antibody. The permeability of liposomes containing dephosphorylated Cx43 was examined by sedimentation in an iso-osmolar sucrose density gradient.
Phosphorylation of Cx43 by MAP Kinase Treatment-Dephosphorylated Cx43 (ϳ30 g) was mixed with 0.1 volume of a buffer containing 100 mM Tris-Cl (pH 7.4), 10 mM MgCl 2 , 1 mM ATP, and 50 units/ml MAP kinase. The reaction mixture was incubated for 3 h at 30°C in the presence or absence of [␥-32 P]ATP (0. 1 mCi/ml). Protein samples phosphorylated in the presence of [␥-32 P]ATP were resolved by 8% SDS-PAGE followed by autoradiography. The permeability of liposomes formed with the MAP kinase-treated Cx43 was compared with that of untreated control Cx43 by sedimentation analysis in an iso-osmolar density gradient.

RESULTS
Purification of Cx43-Cx43, immunoaffinity-purified from rat brain using anti-Cx43 antibody, was resolved on an SDSpolyacrylamide (8%) gel, and a protein band with a molecular size of ϳ41 kDa was detected by silver staining. Immunoblot analysis identified the band as Cx43 (Fig. 1). The anti-Cx43 antibody used in this study was directed against a nonconserved cytoplasmic epitope and may not cross-react with other types of connexins (27,38). Other connexin family proteins expressed in rat brain (27) were also probed using antibodies against connexin 32 (Cx32) and connexin 26 (Cx26), but no corresponding protein bands were detected (data not shown).
Oligomeric State of Immunoaffinity-purified Cx43-The assembly of immunoaffinity-purified Cx43 was examined by velocity sedimentation in a sucrose density gradient, and the composition of the purified Cx43 complex was deciphered by chemical cross-linking with DMS. In previous studies, connexins were isolated from insoluble lipid membranes under denaturing conditions with the use of ionic detergents. It is possible that these isolation conditions alter the secondary structure of connexin proteins, influencing the ability of connexin subunits to assemble into gap junctional connexons, which are composed of six connexin subunits. Therefore we sought to determine whether Cx43 solubilized with a nonionic detergent (octylglucoside) under nondenaturing conditions maintained its oligomeric state after purification. To accomplish this, we subjected immunoaffinity-purified Cx43 to sucrose density gradient sedimentation. The gradient was fractionated, and the amount of Cx43 in each fraction was determined by enzyme-linked immunosorbent assay using an anti-Cx43 antibody. Two distinct Cx43 protein peaks corresponding to 5 S and 9 S particles were detected. Previous studies (39) have demonstrated that the 5 S Cx43 population contributes the monomeric form of the protein, whereas the 9 S species represents the hexameric form. Our data showed that only one-half of the purified Cx43 protein was assembled into connexons. When the Cx43 samples were denatured with a low pH treatment (pH 2.2), the 9 S Cx43 population was dissociated to yield the 5 S form (Fig. 2A).
The quaternary structure of immunoaffinity-purified Cx43 was also examined by chemical cross-linking with DMS. DMS cross-linking of Cx43 monomers (ϳ41 kDa; Fig. 2B, lane 1) produced a protein complex that migrated at ϳ210 kDa (Fig.  2B, lane 2). This 210-kDa estimate may not represent an accurate measurement of the molecular size of a connexon complex, as chemically cross-linked proteins often migrate farther than expected on SDS-polyacrylamide gels (42).
Permeability of Cx43 Channels Formed in Unilamellar Liposomes-We next incorporated immunoaffinity-purified Cx43 protein into unilamellar phospholipid liposomes. The size of the liposomes is strongly influenced by the lipid:detergent ratio and by the chain length of the detergent molecules. In Rhee et al. (36), phospholipid unilamellar vesicles were constructed under the same experimental conditions as ours. The size distribution of the vesicles in that study was examined by filtra- Solubilized membrane proteins from rat brain (lane 1) and Cx43 immunoaffinity-purified with an antibody specific to the C-terminal region of Cx43 (amino acids 368 to 382) (lane 3) were subjected to electrophoresis on an SDS-polyacrylamide (8%) gel and visualized by silver staining. Immunoblot analysis was carried out using the same antibody (lane 2).

FIG. 2. Analysis of connexon assembly.
A, sucrose density gradient analysis of Cx43 assembly. The denatured (E) or nondenatured Cx43 protein (q) was loaded onto a linear gradient of 5 to 20% (w/v) sucrose (total 5 ml) containing 10 mM Hepes (pH 7.4), 459 mM urea, and 80 mM octylglucoside. After centrifugation for 16 h at 200,000 ϫ g, gradients were fractionated, and the amount of Cx43 in each fraction was quantified by enzyme-linked immunosorbent assay using anti-Cx43 antibody. The Cx43 protein peaks corresponding to 5 S (monomeric connexins) and 9 S particles (assembled gap junction connexons) are marked. B, chemical cross-linking analysis of Cx43 assembly. The quaternary structure of immunoaffinity-purified Cx43 was examined by chemical cross-linking procedures using DMS as a coupling agent. Cx43 was incubated either in the presence or absence of 10 mM DMS for 8 h at 4°C, and the treated proteins were resolved on a 4 to 15% SDS-polyacrylamide gradient gel followed by immunoblot analysis using anti-Cx43 antibody. DMS cross-linked the ϳ41 kDa Cx43 monomers (lane 1) to form complexes that migrated at ϳ210 kDa (lane 2). tion over a size exclusion high performance liquid chromatography column, and the average vesicle diameter was estimated to be 45 nm. Assuming a bilayer thickness of 4.0 nm and an average area of 0.7 nm 2 /phospholipid molecule (43), the protein:lipid ratio used in this study should give rise to contain an average of one gap junction channel per vesicle (36). The Poisson distribution predicts that 67% of the vesicles will contain at least one connexon.
The permeability of reconstituted Cx43 channels was examined by sedimentation velocity analysis of liposomes in an iso-osmolar sucrose density gradient (34). Liposomes that were formed in the absence of Cx43 or with denatured Cx43 were sucrose-impermeable; thus they migrated only a short distance into the gradient and banded near the top of the gradient (Fig.  3A, tubes 1 and 3). In contrast, liposomes formed with intact Cx43 were fractionated into two populations within the gradient, a sucrose-permeable fraction that migrated near the bottom of the gradient and a sucrose-impermeable fraction that migrated near the top (Fig. 3A, tube 2). It has been suggested that sucrose-permeable liposomes that contain open Cx43 channels continuously equilibrate their internal solution with the surrounding external solution and thus migrate to a more dense position in the gradient. Liposomes that are impermeable to sucrose, either because they lack Cx43 channels or contain a closed form of the channels, are buoyed by the internal urea buffer, which is lighter than the external sucrose buffer. Impermeable liposomes are therefore retained at the top of the gradient (36).
We next monitored the transfer of lucifer yellow, a well known communicating dye molecule, through reconstituted Cx43 channels. Liposomes were formed with or without Cx43 in the presence of 3 mM lucifer yellow, and the loss of the entrapped lucifer yellow from the liposomes was monitored. After velocity sedimentation, the liposomes in the iso-osmolar density gradient were eluted from the bottom of the tube. The amount of lucifer yellow retained in liposomes was quantified by measuring its fluorescence intensity at 530 nm excited at 428 nm. Liposomes formed with Cx43 fractionated into two populations within the gradient, but lucifer yellow was detectable only in the sucrose-impermeable fraction. Sucrose-permeable liposomes, which contained open Cx43 channels, specifically released lucifer yellow, whereas sucrose-impermeable liposomes retained the dye. These results suggest that sucrosepermeable channels formed with Cx43 are also permeable to lucifer yellow (Fig. 3B). The respective permeability of liposomes formed with varying protein:lipid ratios is shown in Fig.   FIG. 4. Changes in the permeability of liposomes by the amount of Cx43 incorporation. A, liposomes were formed with varying amounts of Cx43. The lipid:protein (w/w) ratios were 100:1 (tube 1), 50:1 (tube 2), 25:1 (tube 3), and 5:1 (tube 4). Liposomes were then separated in an iso-osmolar sucrose density gradient on the basis of sucrose permeability. B, the gradients were then fractionated from the bottom, and the amount of liposomes in each fraction was quantified by measuring the fluorescence intensity of lissamine rhodamine B-labeled PE in the liposome membrane.
FIG. 3. Permeability of reconstituted Cx43 channels to sucrose and lucifer yellow. Unilamellar liposomes composed of phosphatidylcholine, phosphatidylserine, and lissamine rhodamine B-labeled PE at a molar ratio of 2:1:0.05 were formed with purified Cx43. The protein: lipid ratio was typically 1: 45 (w/w). A, the liposomes were fractionated in an iso-osmolar sucrose density gradient on the basis of sucrose permeability. Liposomes marked with the fluorescently labeled lipid were illuminated with UV light in the dark. Shown are liposomes formed without Cx43 (tube 1), liposomes formed with Cx43 (tube 2), liposomes formed with denatured Cx43 (tube 3). B, the distribution of liposomes formed with Cx43 in the presence of 3 mM lucifer yellow was also examined after sedimentation. The fractions were eluted from the bottom of the gradient tube, and the amount of liposomes in each fraction was quantified by measuring the fluorescent intensity of lissamine rhodamine B-labeled PE in the liposome membrane (panel 1). The distribution of lucifer yellow within the gradient was examined by measuring the fluorescence intensity at 530 nm excited at 428 nm (panel 2).

The data indicate that the fraction of sucrose-permeable liposomes increases with increasing concentrations of Cx43 incorporated.
Increased Channel Permeability by Dephosphorylation of Cx43-The effect of Cx43 phosphorylation on gap junctional permeability was presented (Fig. 5). Cx43 preparations from rat brain are known to contain both the phosphorylated and nonphosphorylated forms of the protein (27). Our immunoaffinity-purified Cx43 was treated with CIP either before or after incorporation into liposomes. Treatment of Cx43 with CIP converted the 43-kDa phosphorylated form (Fig. 5, lanes 1 and 3) into the dephosphorylated 41-kDa form (Fig. 5, lanes 2 and 4). The fraction of permeable liposomes containing Cx43 was in-creased by treating the liposome with CIP (Fig. 5, tube 2). The distribution analysis of liposomes after sedimentation revealed that 15% of the total liposome population was sucrose-permeable. However, 27% of the total liposome population became sucrose-permeable if the liposomes were treated with CIP (5 units for 3 h) before sedimentation. The hypothesis that dephosphorylation of Cx43 increases gap junctional permeability is further supported by our subsequent experiment, which showed an increase in the permeability of liposomes formed with CIP-treated Cx43 (Fig. 5, tube 4). The fraction of permeable liposomes formed with dephosphorylated Cx43 was about 2-fold greater than that observed with control liposomes incorporated with intact Cx43 (Fig. 5, tubes 3 and 4). FIG. 5. Gating mechanism of Cx43 channels by phosphorylation. Panel 1 (left), liposomes formed with solubilized Cx43; center, lane 1, Cx43 in the liposomes was resolved by 8% SDS-PAGE followed by immunoblot analysis; right, tube 1, the permeability of liposomes containing Cx43 was analyzed by sedimentation in an iso-osmolar sucrose density gradient. The relative amount of sucrose-permeable liposomes was represented by the densitometric scanning of the gradient after sedimentation using Tina 2.0 software (FujiFilm). Panel 2 (left), liposomes containing Cx43 were treated with 5 units of CIP for 3 h at 30°C; center, lane 2, the liposome was treated with CIP, and Cx43 was resolved by 8% SDS-PAGE followed by immunoblot analysis; right, tube 2, Cx43 in the liposomes was dephosphorylated, and the liposomes were separated by sedimentation. Reduced Channel Permeability by MAP Kinase Phosphorylation of Cx43-In an effort to further characterize the role of phosphorylation in the gating of gap junction channels, we used MAP kinase to phosphorylate solubilized Cx43 (Fig. 5, lane 5b). The fraction of permeable liposomes containing CIP-treated Cx43 was reduced by 79% if the dephosphorylated Cx43 became phosphorylated by MAP kinase (Fig. 5, tubes 5a and 5b). MAP kinase-dependent gating mechanism of Cx43 channels was further supported by the observation that MAP kinaseinduced inhibition of channel permeability was reversed by treatment of Cx43 with CIP (Fig. 5, tube 5c). These data suggest that phosphorylation of Cx43 by MAP kinase may be directly involved in the channel gating process. Because protein kinase C has also been implicated in the regulation of GJC, we tested whether protein kinase C could phosphorylate the immunoaffinity-purified Cx43. Only a trace amount of phosphorylated Cx43 was detected by autoradiography, whereas histone H1, a known substrate protein of protein kinase C, was readily phosphorylated under the same experimental conditions (data not shown).

DISCUSSION
Gap junctions are membrane channels that link end-to-end across the space between juxtaposed cells. In this study, we present evidence that the phosphorylation of connexin is directly involved in the gating mechanism of gap junction channels. We showed that the activity of Cx43 channels reconstituted in liposomes was dramatically increased by dephosphorylation of Cx43. The activity of the reconstitituted Cx43 channels was monitored by measuring the permeability of the liposomes to sucrose in an iso-osmolar sucrose density gradient. Several other studies have reported the functional reconstitution of connexons consisting of Cx26 (35) and Cx32 (36).
Information regarding the gating mechanism of gap junction channels can be obtained from the assessment of the permeability of hemichannels to small molecules. Structural, biochemical, and physiological data show the hemichannels have properties that are similar to those of intact gap junctions. For example, the activities of hemichannels are, like those of intact gap junction channels, voltage-dependent. Trexler et al. (44) demonstrate the voltage-dependent gating properties of Cx46 hemichannels in excised patches from Xenopus oocytes. Hemichannels have also been shown to exist in the plasma membrane (45,46). Several studies have proposed that the electrical coupling between horizontal cells of the fish retina occurs through hemichannels. Recently, Li et al. (47) presented data in support of the existence of nonjunctional plasma Cx43 hemichannels in Novicoff hepatoma cells. In this cell line, they measured properties of fluorescent dye transfer through nonjunctional hemichannels that were similar to those observed with gap junction channels. The hemichannels are believed to be typically in the closed state to prevent the loss of cytoplasmic solutes and the entry of extracellular ions. But some of the hemichannels formed in Xenopus oocyte membranes open by membrane depolarization (48) and induce cation fluxes (44). Open hemichannels were also shown to form in a system in which Cx32 channels were reconstituted in lipid vesicles that lacked the cytoplasmic or membrane factors necessary for maintaining the channels in a closed state (32,34,36). From the result of our study, we propose that Cx43 forms open hemichannels in lipid vesicles. The results herein showed that liposomes containing open Cx43 channels were permeable to sucrose and communicating dye (Fig. 3). Indeed we observed an increase in the fraction of permeable liposomes as more connexon protein was incorporated (Fig. 4).
Although the gating mechanism of gap junction channels remains obscure, the rapid and reversible changes in the GJC by protein kinase activators have led to the hypothesis that gap junction channels undergo a conformational change by phosphorylation to gate the channels. Cx43 has been identified as a substrate of MAP kinase (30), protein kinase C (49), and p34 cdc2 kinase (29), and it has been shown that GJC is rapidly reduced by the increased phosphorylation of Cx43 on serine residues (11,12,29). MAP kinase constitutes one of the most important protein kinase families for the progression of the cell cycle. Activated MAP kinase phosphorylates a variety of targets including other protein kinases and transcription factors (50). Recently it was shown that Cx43 is a substrate of MAP kinase and that the phosphorylation of Cx43 is mitosis-specific in many types of cells (11,12,29,30). Comparison of the amino acid sequences of MAP kinase substrates has identified PX 1-2(S/T)P as a consensus phosphorylation site (51)(52)(53). Sequence analysis of Cx43 revealed MAP kinase consensus phosphorylation sites at serine 255, 279, and 282. In fact Warn-Cramer et al. (37) confirm that MAP kinase phosphorylates Cx43 at Ser-255, -279, and -282. Data presented in this study suggest that phosphorylation of Cx43 by MAP kinase is directly involved in the gating mechanism of Cx43 channels.