Eur. J . Biochem. 192,263-273 (1990) 6 FEBS 1990

Phosphorylation of connexin 32, a hepatocyte gap-junction protein, by CAMP-dependent protein kinase, protein kinase C and Ca2 + /calmodulin-dependent protein kinase I1 Juan C. SAEZ', Angus C. NAIRN', Andrew J. CZERNIK', David C. SPRAY ', Elliot L. HERTZBERG', Paul GREENGARD' and Michael V. L. BENNETT' ' Department of Neuroscience, Albert Einstein College of Medicine, Bronx, USA ' Laboratory of Molecular and Cellular Neuroscience, The Rockefeller University, New York, USA

(Received April 18, 1990) - EJB 90 0443

Phosphorylation of connexin 32, the major liver gap-junction protein, was studied in purified liver gap junction and in hepatocytes. In isolated gap junctions, connexin 32 was phosphorylated by CAMP-dependent protein kinasc (CAMP-PK), by protein kinase C (PKC) and by Ca2+/calmodulin-dependent protein kinase I1 (Ca2+/CaM-PK 11) Connexin 26 was not phosphorylated by these three protein kinases. Phosphopeptide mapping of connexin 3; demonstrated that CAMP-PK and PKC primarily phosphorylated a seryl residue in a peptide termed peptide 1 PKC also phosphorylated seryl residues in additional peptides. Ca2+/CaM-PK I1 phosphorylated serine and tc a lesser extent, threonine, at sites different from those phosphorylated by the other two protein kinases. A synthetic peptide PSRKGSGFGHRL-amine (residues 228 - 239 based on the deduced amino acid sequence of rat connexiI 32) was phosphorylated by CAMP-PK and by PKC, with kinetic properties being similar to those for othei physiological substrates phosphorylated by these enzymes. Ca2+/CaM-PK I1 did not phosphorylate the peptide Phosphopeptide mapping and amino acid sequencing of the phosphorylated synthetic peptide indicated thai Ser233 of connexin 32 was present in peptide 1 and was phosphorylated by CAMP-PK or by PKC. In hepatocyte: labeled with [32P]~rthopho~phoric acid, treatment with forskolin or 20-deoxy-20-oxophorbol 12,13-dibutyratc (PDBt) resulted in increased 32P-incorporation into connexin 32. Phosphopeptide mapping and phosphoamino acid analysis showed that a seryl residue in peptide 1 was most prominently phosphorylated under basal conditions. Treatment with forskolin or PDBt stimulated the phosphorylation of peptide 1. PDBt treatment also increased the phosphorylation of seryl residues in several other peptides. PDBt did not affect the CAMP-PK activity in hepatocytes. It has previously been shown that phorbol ester reduces dye coupling in several cell types, however in rat hepatocytes, dye coupling was not reduced by treatment with PDBt. Thus, activation of PKC may have differential effects on junctional permeability in different cell types; one source of this variability may be differences in the sites of phosphorylation in different gap-junction proteins.

Gap junctions provide a pathway for cytoplasmic com- munication between adjacent cells [I, 21. These channels are permeable to ions and small molecules such as second messen- gers [3, 41. Hepatocyte gap junctions can be isolated by sub- cellular fractionation following detergent or alkaline extrac- tion [5 -91. These preparations are highly enriched in proteins that have mobilities in the range of 26-28 kDa as deter- mined by SDSjPAGE [7 - 1 I]. Rat and human liver cDNAs have been isolated [lo, 111 which encode proteins having molecular masses of approximately 32 kDa. The rat or human liver proteins have been termed connexin 32 [12]. Rat liver also expresses a homologous protein of apparent molecular mass 21 kDa for which a cDNA has also been sequenced. Its predicted molecular mass is 26 kDa and it has been termed connexin 26 [13]. The exact function of connexin

Correspondence to J. C. Saez, Department ofNeuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USA

Abbreviations. PDBt, 20-deoxy-20-oxophorbol 12,13-butyrate; CAMP-PK, CAMP-dependent protein kinase; PKC, protein kinase C; Ca2+/CaM-PK, Ca'+/calmodulin-dependent protein kinase.

Enzymes. Protein kinases (EC 2.7.1.37); trypsin (EC 3.4.21.4); Clostridium histolyticum collagenase (EC 3.4.24.3).

26 is not known, although it shares substantial sequence simi- larity with connexin 32, particularly in the proposed channel- forming region [13]. Connexin 26 is present in relatively low amounts in isolated rat liver gap junctions (z 10%) and in higher amounts in mouse liver gap junctions (as much as 30-50%0 of the total junctional protein) [14- 161. In mouse hepatocytes the two proteins have been found in the same gap junctions [14, 161.

Liver gap junctions are thought to be comprised of a dodecameric complex [17]. When purified connexin 32 or iso- lated rat liver gap junctions are incorporated into artificial lipid bilayers, unitary and macroscopic conductances are recovered that have properties similar to those exhibited be- tween hepatocytes [18,19]. Thus connexin 32, in the absence of other components, is believed to form functional gap-junction channels.

The second messengers Ca2+, CAMP, and diacylglycerol affect gap-junctional conductance or permeability [I, 21. Ele- vation of the intracellular free Ca2 + concentration well above the physiological range results in closure of gap junc- tions [20 - 221. The multifunctional Ca2 +-binding protein, calmodulin, may also be involved in regulating junctional

264

communication [23 - 251. In liver, heart and some cell lines, elevation of cAMP increases junctional conductance [26 - 281 but in retinal horizontal cells, Sertoli cells and myometrial smooth muscle cells, junctional conductance is reduced [29 - 321. Tumor-promoting phorbol esters or diacylglycerol reduce cell coupling in many cell lines, including some derived from liver [33]. In pancreatic acinar cells the microinjection of PKC caused a decrease in junctional conductance 1341. In contrast, a slight increase has also been observed in this cell preparation following the addition of phorbol esters [35]. In primary cul- tures of Sertoli cells [31] and cardiac myocytes (Burt, J. M. and Spray, D. C., unpublished results) phorbol esters increased coupling. In lacrimal acinar cells, diacylglycerol reduced coup- ling, whereas variable results were obtained with phorbol es- ters [36].

Connexin 32 is a phosphoprotein in adult rat hepatocytes 126, 371 and in embryonic mouse hepatocytes [14, 381. In hepatocytes, phosphorylation of the protein can be increased by raising the intracellular cAMP concentration [26, 37, 381. In purified rat liver gap junctions, the protein can be phos- phorylated by CAMP-PK and PKC [37 - 401. In addition, the effect of CaZt and calmodulin on gap-junctional communi- cation suggests a role for Ca2 +-dependent protein phosphor- ylation, possibly mediated by CaZ+/CaM-PKs. It is possible that the phosphorylation of different sites on the connexins by CAMP-PK and PKC might explain the opposite functional effects of activators of these enzymes observed in other cell types. To elucidate the roles of CAMP-PK, PKC and CaZ+/ CaM-PK we have carried out a detailed analysis of the sites phosphorylated in connexin 32 in intact hepatocytes and in isolated gap junctions. We have found that the Ser233 of connexin 32 is the primary site of phosphorylation by CAMP- PK and PKC suggesting that, at least in freshly isolated rat hepatocytes, these two kinases may have similar functional roles. Preliminary results of parts of this work have already been presented [39, 41, 421.

EXPERIMENTAL PROCEDURES

Materials

Collagenase type IV and type 11, forskolin, ATP, DL- phosphoserine, DL-phosphothreonine, DL-phosphotyrosine, glutamine, Triton X-100, dithiothreitol, EDTA, Tris, 20- deoxy-20-oxophorbol 12,13-dibutyrate, 4 a-phorbol, Coo- massie brilliant blue R-250, SDS, histone type 111-s, Nonidet P-40, diolein, phosphatidylserine, bovine serum albumin and Lucifer yellow were from Sigma. Hepes, formalin-fixed Stuphjhcoccus uureus Cowan cells (Pansorbin) and phenyl- methylsulfonyl fluoride were from Calbiochem. [y-32P]ATP and [32P]orthophosphoric acid were from New England Nu- clear. Tosylphenylalaninechloromethyl ketone-treated trypsin and histone f2b were from Worthington. Waymouth’s me- dium, minimal essential medium, fetal calf serum and Dulbec- CO’S phosphate-buffered saline were from Gibco. Phosphocel- lulose P-81 paper was from Whatman. Trifluroracetic acid and Extracti-gel D were from Pierce. Acetonitrile was from Burdick-Jackson. XAR-5 X-ray film and cellulose TLC sheets were from Eastman Kodak. The catalytic subunit of CAMP- PK was purified from bovine heart as described [43]. PKC was purified from bovine or mouse brain essentially as de- scribed [44]. Ca2’/CaM-PK I1 was purified from rat brain as described [45], calmodulin was purified as described [46] and synapsin I was purified as described [47]. Gap junctions were purified from the livers of female rats or mice as described [8].

In one preparation from rat the following cocktail of protease inhibitors was added to all buffers: phenylmethylsulfonyl flu- oride, 0.1 mM; leupeptin, 10 pg/nil; chymostatin, 10 pg/ml; pepstatin, 4 pg/ml; antipain, 10 pg/ml (from Chemicon). The myelin basic protein peptide [Alalos]myelin (residues 104 - 118 of the myelin basic protein) [48] was a gift of Dr Bruce Kemp (St Vincent’s Institute of Medical Research, Melbourne, Australia).

Phosphorylution of purijied gup junctions by CAMP-PK, Ca2‘ICuM-PK II or PKC

The phosphorylation of purified gap junctions was carried out in a reaction mixture (final volume 50 - 100 pl) containing 50 mM Hepes (pH 7.4), 10 mM MgC12, 1 mM EGTA, 1 mM dithiothreitol, 1 mg/ml bovine serum albumin, 50 pM [y- 32P]ATP (lo6 cpm/nmol) and 1 pg isolated gap-junction pro- tein. The catalytic subunit of CAMP-PK (2 - 4 pg) was added alone. Ca2+/CaM-PK 11 (0.5 - 5.0 pg) was added together with 1.5 mM CaC1, and 15 pg/ml calmodulin. PKC (1 - 2 pg) was added along with 1.5 mM CaCI2, 50 pg/ml phosphati- dylserine and 2 pg/nil diolein in the absence or presence of 0.025% Triton X-100. In some assays, synapsin I (5 pg) or histone type 111 (2 pg) replaced the gap junction protein. Reac- tions were carried out for 5-30 min. All reactions were ter- minated by the addition of a SDS-containing termination solution and samples were analyzed by SDSjPAGE and autoradiography. The incorporation of 32P into connexin 32 was measured by liquid scintillation counting of an excised gel piece containing the protein.

In some experiments we attempted to improve the interac- tion of the gap-junction protein with the lipid vesicle-associ- ated PKC. A solution of phosphatidylserine and diolein in chloroform (in the various amounts indicated in the legend to Fig. 1) was dried at 5 “C using a Nz stream. The PKC reaction mixture described above, in the absence or presence of the enzyme, was added to the dried lipids in the presence of 0.025% Triton X-100, and the samples were sonicated for 1 niin (5 ”C) using a sonifier cell disrupter (Model W 140, Heat System Ultrasonic Inc. Plainview NJ). Triton X-100 was then removed by passing the samples twice through a column con- taining 50 p1 of Extracti-gel D equilibrated with 50 mM Tris/ HCI (pH 7.6). PKC was added if necessary and the reaction mixtures were incubated as described above. Reactions were stopped by the addition of 40 p1 SDS termination solution and the samples analyzed by SDSjPAGE and autoradiography.

Synthesis andphosphorylation ojpeptides

Peptides corresponding to residues 228 - 239 of connexin 32 (PSRKGSGFGHRL-amide) and residues 263 - 280 of connexin 32 (LRRSPGTGAGLAEKSDRC-amide) of the de- duced amino acid sequence of rat connexin 32 [lo, l l ] and a synapsin I peptide (YLRRRLSDSN F-amide) corresponding to the residues surrounding phosphorylation site 1 [47] were synthesized by standard solid-phase techniques [49]. Peptides were purified by Sephadex G-15 chromatography and/or CIS reverse-phase HPLC. All peptides were > 95% pure as analyzed by HPLC and had the expected amino acid compo- sitions and mass spectra (data not shown).

Using a purified 32P-labeled peptide it was established that phosphorylated connexin 32 (residues 228 - 239) was quanti- tatively absorbed by P-81 phosphocellulose paper and was not released using the washing procedures in the standard histone kinase assay. Unless otherwise noted the phosphorylation of

265

synthetic peptides (10-200 pM) was carried out using the phosphocellulose paper assay described above except that 0.2 - 0.5 mM [ Y - ~ ~ P ~ A T P was used. Peptide phosphorylation was linear with respect to time and enzyme concentration. Reactions were carried out for 0.5 - 5.0 rnin and stopped by the addition of 50 p1 of 30% (by vol.) acetic acid. Kinetic parameters were derived from a linear regression analysis of the double-reciprocal plots.

Peptide purification and amino acid sequencing

Synthetic peptides, tryptic digests of synthetic peptides or tryptic digests of connexin 32 were purified by reverse-phase HPLC using a Vydac C18-column (0.45 cm x 25 cm) equilib- rated with 0.1 % trifluoroacetic acid. The peptides were eluted with a gradient of 70% acetonitrile in 0.085% trifluoroacetic acid (buffer A) as indicated: 0-3 min, 0% buffer A; 3- 8 min,0-15% bufferA; 8-28 min, 15-26% buffer A. Frac- tions were collected and monitored for radioactivity by Cerenkov counting. Peptide peaks were monitored by ultra- violet detection at 214 nm and 258 nm.

The amino acid sequence of 32P-labeled tryptic phospho- peptides was determined as described [50] using an Applied Biosystems (Foster City, CA) AB-470 gas-phase sequencer. Identification of the phenylthiohydantoin-amino-acid deriva- tives was accomplished by on-line CI8 HPLC analysis and was semi-quantitative.

Preparation and radiolabeling of hepatocytes and irnnzunoprecipitation of connexin 32

Hepatocytes were prepared from the livers of female Sprague Dawley rats (200 - 225 g, Charles River Breeding Laboratories) using a collagenase perfusion technique [51]. Hepatocytes were prepared from mouse liver by the same procedure used for rat except that collagenase type I1 (100 U/ ml) was used instead of type IV. Prior to labeling, cells were maintained in Leaffer's saline for 1-2 h in an ice bath. Hepatocytes in suspension were labeled for 1-2 h with [32P]orthophosphoric acid (0.5 mCi/106 cells/0.25 ml) essen- tially as described [52]. In control experiments there was no difference in the results obtained when the cells were labeled from 1 - 6 h. Immunoprecipitation of connexin 32 was carried out as described [26]. Samples were analyzed by SDSjPAGE and autoradiography performed. The incorporation of 32P into connexin 32 was measured by liquid scintillation counting on an excised gel piece containing the protein.

Antibodies to rat liver gap-junction proteins were raised in rabbits as described previously for other antibodies [8] and purified by affinity-chromatography. To purify connexin 32 for the affinity resin, rat liver gap junctions were solubilized in 10 rnM sodium phosphate, 1% SDS (pH 6.0), diluted 10- fold in 10 mM sodium phosphate (pH 6.0) and applied to a column of hydroxylapatite (Biogel HTP) equilibrated in the same buffer. Proteins were eluted using a linear gradient of 0.01 -0.5 M sodium phosphate (pH 6.0), containing 0.1% SDS. Connexin 32 eluted at about 0.4 M sodium phosphate. The pooled peak fractions were dialyzed and coupled to CNBr-activated Sepharose 4B (Pharmacia) according to the manufacturer's directions. By imniunoblotting, the affinity- purified antibodies recognized only connexin 32 in rat liver homogenates. By indirect immunofluorescence, punctate labeling was observed at cell interfaces on frozen sections of the rat liver.

Measurement of the CAMP-PK activity ratio

The CAMP-PK activity ratio was measured in cytosol pre- pared from rat hepatocytes essentially as described [53]. The activity ratio is defined as the ratio of CAMP-PK activity with and without added CAMP. If the activity ratio becomes close to unity after treatment then it can be inferred that CAMP levels are increased by the treatment. Cells (lo6 cells/O.S ml) were incubated at room temperature in minimal essential me- dium containing 10mM glutamine plus 10mM Hepes (pH 7.4) under control conditions or in the presence of either 20 pM forskolin or 0.3 pM PDBt for 5, 15 or 30 min. Cells were washed once with 2 ml of a solution containing 20 mM Tris (pH 7.6), 50 mM NaCI, 1 mM EGTA, 1 mM EDTA, 15 mM 2-mercaptoethanol and 50 pM 3-isobutyl-1-methyl- xanthine, then resuspended in the same buffer and sonicated for 20 s with a micro-ultrasonic cell disrupter (Kontos, pos- ition 2). Lysates were then centrifuged at 18500 x g for 15 min and the protein concentration in the supernatant fractions determined as described [54].

CAMP-PK activity in the supernatant fraction was assayed in a reaction mixture (final volume of 0.1 ml) containing 50 mM Hepes (pH 7.4), 10 mM MgC12, 1 mM EGTA, 5 mM dithiothreitol, 0.5 mg/ml histone f2b, 50 pM [y-32P]ATP (lo6 cpm/nmol), and 2-5 pg of supernatant protein, in the absence or presence of 1 pM CAMP. Samples were incubated at 30°C for 1 min. Reactions were initiated by the addition of ATP and carried out for 2 - 5 min. Reactions were terminated by the addition of 50 p1 of 30% (by vol.) acetic acid. The P-81 phosphocellulose paper method was used to quantitate the incorporation of 32P into the histone [45]. Briefly, an aliquot (25 - 50 pl) was spotted onto P-81 phosphocellulose paper (1 cm x 3 cm). The strips of paper were washed in H 2 0 and 32P-incorporation into the histone was measured by liquid scintillation counting. Experiments were performed five times.

Dye coupling

Cells were plated (lo5 cells/35 mm culture dish) in Waymouth's medium containing 10% fetal calf serum for 1 h at 3 7 T in a 95:5 air/C02 atmosphere. To prevent aikaliniza- tion, the culture medium was then replaced with Dulbecco's phosphate-buffered saline (pH 7.2). The incidence of dye coupling was evaluated between pairs of cells (n = 15 for each dish) by injecting one cell of a pair with 5% (mass/vol.) Lucifer yellow prepared in 0.15 M LiCl. The dye was injected by repeated 0.1 s, 0.1 nA current pulses or by brief over-compen- sation of the negative capacitance control on a WPI M707 electrometer until the impaled cell was brightly fluorescent [55]. The incidence of coupling was measured at 15, 30, 60, 120, 180 and 360 min after the addition of PDBt (0.1 - 1 .O pM). The cells were then observed for 2 - 3 rnin using a Nikon Diaphot microscope with Xenon arc lamp illumination in order to determine whether dye transfer had occurred.

Miscellaneous methods

Total protein content of the isolated gap junctions was determined as described by Bradford [54] and the amount of connexin 32 in the isolated gap junctions was estimated to be 60%' or 90% of the total junctional protein from mouse or rat, respectiveiy [I4 - 161. Proteins were separated using 12.5% SDSjPAGE [56] and were stained using Coomassie brilliant blue. Gels were dried under vacuum and autoradi- ography was performed using XAR-5 X-ray film (Eastman Kodak) and intensifying screens (Lightning Plus, Dupont).

266

4 B

--CONN 32

PS CAMP CaM PKC PKC 125 250 500 1000 PSer -PK -PK +Triton 1.25 2.5 5 10 0

Origin - 97 m 2 68

45 v) v) a

a 30

[r

; 20 0 w 6 14

5

Fig. 1. Phosphorylution of connexin 32 by CAMP-PK, CuZ+/CuM-PK II and PKC. (A) Coomassie-brilliant-blue staining (PS) of gap-junction membranes purified from mouse liver showed the presence of connexin 32 (CONN32) migrating at 27 kDa and connexin 26 migrating at 21 kDa. Gap junctions were incubated with [y-32P]ATP and different protein kinases, as indicated, for 15 min and the samples analyzed by SDSjPAGE and autoradiography. Ca*+/CaM-PK IT (CaM-PK), PKC plus 0.025% Triton X-100 (PKC + Triton). Phosphorylation by PKC was negligible in the absence of Triton X-300. Incubation of each of the protein kinases in the absence of gap-junction membranes (not shown) indicated that connexin 32 was phosphoryhted by each enzyme and that autophosphorylation of the catalytic subunit of CAMP-PK and the subunits of Ca2+/CaM-PK I1 accounted for 32P-labeling of bands of higher molecular mass. The autoradiograms shown were obtained from gels which were exposed for different periods of time. Using the autoradiograms as guides, the band containing phosphorylated connexin 32 was cut from the dried gel and the 3ZP-content determined by liquid scintillation counting. The region of the gel present in the lanes in which connexin 32 was ommitted from the assay mixture for each kinase, was also counted and used as a blank. The 32P incorporated into connexin 32 by each protein kinase was calculated by subtracting the blank and the phosphate content determined from the specific activity of the [Y-~'P]ATP. The stoichiometry of phosphorylation was calculated by dividing the phosphate content values by the amount of connexin 32 used. Results are representative of those obtained from at least three different experiments. (B) The effect of lipids on the phosphorylation of gap junctions by PKC was studied. Rat liver gap junctions and 0.2% Triton X-100 were added to increasing concentrations of phosphatidylserine (PSer) and diolein (DG) (present in a constant ratio of 100 : 1). The following concentrations of phosphatidylserinejdiolein were used (pg/ml): 125:1.25, 250:2.5, 500:5 and 1OOO:lO. The mixtures were sonicdted for 1 min and the detergent removed using extracti- gel D. Samples were incubated with [y-3zP]ATP and PKC at 30°C for 30 min. The reactions were analyzed by SDSjPAGE and antoradiography. Results are representative of those obtained from three different experiments

Gel pieces containing 32P-labeled connexin 32 were cut from the dried gels using the autoradiogram as a guide and the 32P radioactivity was measured by liquid-scintillation counting. Two-dimensional tryptic phosphopeptide mapping was performed as described [57]. Briefly, aliquots of the digests were spotted in the middle of the plate, 5 cm from the bottom and initially separated by horizontal electrophoresis at pH 3.5 in 10% acetic acid/l% pyridine until the dye front migrated 5 cm. Chromatography was performed in the vertical dimen- sion in 1-butanollacetic acid/water/pyridine (15: 10: 3: 12). Phosphopeptides were visualized by autoradiography and where indicated, densitometry was performed using a Cam- bridge Instruments Quantimet 970 to quantitate the degree of phosphorylation of the peptides. The tryptic digests were hydrolyzed with 6 M HC1 and phosphoamino acid analysis performed as described [58]. Briefly, phosphoamino acids were separated by electrophoresis at pH 1.9 in 8.7% acetic acid/ 2.5% formic acid followed by electrophoresis in the same direction at pH 3.5 in 10% acetic acid11 YO pyridine.

RESULTS

Phosphorylation of connexin 32 by CAMP-PK, PKC and Ca2+/CaM-PK I1

Phosphorylated connexin 32 was analyzed by using purified gap-junction preparations from mouse liver (Fig. 1) and rat liver (not shown) and purified protein kinases. Phos- phorylation of the other components of the purified gap, junctions, particularly connexin 26 which is abundant in prep- arations from mouse liver, was also examined. Gap junctions

from mouse liver were incubated with the catalytic subunit of CAMP-PK, PKC or the multifunctional Ca2 +/CaM-PK 11. Connexin 32 was phosphorylated by all three protein kinases (Fig. 1 A). As previously shown [26] the maximum stoichi- ometry of phosphorylation was 0.025 mol/mol using the cata- lytic subunit of CAMP-PK. Similar levels of phosphorylation of connexin 32 by CAMP-PK have also been obtained recently in a separate study [37]. Ca2+/CaM-PK I1 phosphorylated connexin 32 with a stoichiometry of 0.33 moljmol. The stoi- chiometry of phosphorylation was similar when rat gap junc- tions prepared in the presence of protease inhibitors were used as substrate. Connexin 26 from mouse (Fig. 1) or rat (not shown) was not phosphorylated by any of the protein kinases.

The stoichiometry of the phosphorylation of connexin 32 by PKC was very low and addition of 0.025% Triton X-100 significantly increased the phosphorylation of this protein (Fig. 1 A). In vivo, PKC is activated following its association with a membrane in the presence of diacylglycerol and Ca2+. In the in vitro assay for PKC, phosphatidylserine/ diacylglycerol vesicles serve as an artificial membrane. How- ever, substrates for PKC which are membrane-associated, may be present in separate vesicles or lipid structures thereby cre- ating a physical limitation for the accessibility of the kinase for the substrate. This is known to occur for the nicotinic acetylcholine receptor (R. L. Huganir, personal communi- cation) and may explain the results obtained with the dihydropyridine receptor of the voltage-sensitive Ca2 + chan- nel [59]. An attempt was made to improve the interaction between PKC and the gap-junction proteins by first sonicating the substrate together with various concentrations of phosphatidylserine and diolein in the presence of Triton X- 100. Triton X-100 was then extracted prior to incubation

267

Fig. 2. Two-dimensional phosphopeptide mupping and phosphoamino acid analysis of connexin 32 phosphorylated by CAMP-PK, PKC or Cu2 '1 CUM-PK ZI. Gel pieces containing 32P-labeled connexin 32, prepared as described in Fig. 1, were cut from the dried gel, washed, and incubated with trypsin. Phosphopeptides were separated by electrophoresis and chromatography as described in Experimental Procedures. Phosphopeptides were applied to cellulose thin-layer plates (origin at bottom middle, 0). The phosphopeptides were separated by electrophoresis in the horizontal dimension (negative pole, left; positive pole, right), followed by ascending chromatography in the vertical dimension and analyzed by autoradiography. Phosphopeptide maps of tryptic digests of connexin 32 phosphorylated by the following: (A) CAMP-PK; (B) PKC plus Triton X-100; (C) Ca2 +/CaM-PK I1 (CaM-PK); (D) PKC plus optimal amounts of phosphatidylserine/diolein (PKC and lipid); (E) mixture ofequal amounts of tryptic digests of connexin 32 phosphorylated by CaZ+/CaM-PK I1 (C) and PKC (D). Peptides were numbered 1-9. Peptide 2' was defined based on a comparison of several peptide maps generated from connexin 32 phosphorylated by PKC that suggested that peptides 2 and 2' were most likely to have been produced by an alternative tryptic digestion of the same phosphorylation site. (F) Phosphoamino acid analysis of connexin 32 phosphorylated by CAMP-PK, PKC or Ca*+/CaM-PK 11 (CaM-PK). Aliquots of the tryptic digests were hydrolyzed in 6 M HCI for 2 h. One-dimensional electrophoresis and autoradiography were performed. The position of standards are indicated; phosphoserine (P-Ser); phosphothreonine (P-Thr); phosphotyrosine (P-Tyr). Results are representative of those obtained from at least two different experiments

with PKC and [32P]ATP (Fig. 1 B). The phosphorylation of connexin 32 by PKC was significantly increased by concen- trations of phosphatidylserine and diolein at a ratio of 100: 1 up to 250 pg/ml and 2.5 pg/ml, respectively. Maximal phos- phorylation of 0.1 mol/mol was obtained. Higher concen- trations of these two lipids, at the same ratio, were less effec- tive. Similar increases in phosphorylation were obtained when PKC was added prior to sonication. Phosphorylation of connexin 32 by CAMP-PK or Ca2+/CaM-PK I1 was not affec- ted when the gap-junction preparation was first sonicated with the two lipids in the concentration ranges used with PKC (not shown). Similar levels of phosphorylation of connexin 32 by PKC have also been obtained in a separate study, although the effects of detergent or phospholipid were not investigated [37, 401.

The relative rates of phosphorylation of connexin 32 were compared with those of known substrates for the different protein kinases. Synapsin I was used as a substrate for CAMP- PK and Ca2+/CaM-PK 11, and histone type 111 was used as a substrate for PKC. The rates of phosphorylation of connexin 32 were approximately 1-5% of those for synapsin I and histone type I11 for the respective protein kinases.

Two-dimensional phosphopeptide mapping and phosphoamino acid analysis of connexin 32 phosphorylated in purified gap junctions

As previously shown [26] two-dimensional phosphopep- tide mapping demonstrated that CAMP-PK phosphorylated a major phosphopeptide, termed peptide 1 (Fig. 2A). A

268

peptide with the same migration as peptide 1 was also phos- phorylated by PKC in the presence of 0.025% Triton X- 100 (Fig. 2B). PKC also phosphorylated several additional peptides, termed peptides 2 - 5. When the gap-junction preparation was sonicated in optimal concentrations of pho- phatidylserine and diolein, and phosphorylated with PKC, peptides 2- 5 were phosphorylated to a relatively higher level than peptide 1 (compare Fig. 2 B and D). CAMP-PK and PKC phosphorylated only seryl residues in connexin 32 (Fig. 2F, lanes 1 and 2).

Ca2+/CaM-PK I1 phosphorylated most prominently a phosphopeptide termed peptide 6, phosphorylated several other peptides (peptides 7 - 9) to a lesser extent and phosphor- ylated peptide 1 to a minor extent. Mixing experiments showed that peptides 6 and 7 were distinct from peptides 2 and 2', that were phosphorylated by PKC (compare Fig. 2C and D with E). Ca2 +/CaM-PK 11 phosphorylated connexin 32 mainly on seryl residues but also on threonyl residues to a minor extent (Fig. 2 F. lane 3).

Phosplzoiy lat ion qf syn the tic pep tides containing p o ten tiul phosphor~.lutiolz~I~ sites f b r CAMP-PK and PKC

The minimum consensus amino acid sequence for phos- phorylation by CAMP-PK has been established [60 - 621. The consensus sequence includes at least two basic amino acids and a spacer amino acid residue followed by the phosphorylatable seryl or threonyl residue. The COOH-terminal region of the deduced amino acid sequences of rat and human liver connexin 32 [lo, 111 contains several potential phosphoryla- tion sites, particularly the sequence RKGS (residues 230 - 233) which conforms very well to the CAMP-PK consensus sequence, with Ser233 as the phosphorylated residue. The primary structural requirements for phosphorylation by PKC are less well defined than for CAMP-PK. PKC requires basic amino acid residues in close proximity to the phosphorylatable residue, although these may be located on the NH2-terminal or COOH-terminal side, or on both sides [48, 63, 641. The COOH-terminal region of connexin 32 contains several poten- tial phosphorylation sites for PKC. For example, serine-resi- dues at positions 229, 233, 240, 258, 266, 277 and 281 are all potential phosphorylation sites for this enzyme.

Two synthetic peptides which contained some, but not all of these seryl residues, were studied in an attempt to identify the phosphorylation sites for CAMP-PK and PKC. The peptides, PSRKGSGFGHRL-amide and LRRSPCTGAC- LAEKSDRC-amide corresponded to residues 228 - 239 and 263 - 280 of connexin 32, respectively. In the initial exper- iments, each peptide was incubated with CAMP-PK, PKC or Ca2 + /CaM-PK I1 and the unphosphorylated or phosphory- lated peptides purified by C1 ,-reverse-phase HPLC. Both CAMP-PK and PKC phosphorylated connexin 32 (residues 228 -239). Ca2+/CaM-PK I1 did not phosphorylate this peptide nor was any phosphorylation of connexin (residues 263 - 280) observed with any of the kinases.

The time course of phosphorylation of connexin 32 (resi- dues 228 - 239) by CAMP-PK (Fig. 3 A) or PKC (Fig. 3 B) was dctermined. At intervals an aliquot of the reaction mixture was analyzed by HPLC (results not shown) and the levels of the unphosphorylated and phosphorylated peptides measured by ultraviolet spectroscopy and by the 32P content of the eluted fractions. The unphosphorylated peptide (peak 1) was eluted with a retention time of 22.5 min. For phosphorylation by CAMP-PK, a single phosphopeptide (peak 2) was eluted

TIME (rnin)

+ I W I P

TIME (rnin)

Fig. 3. Phosplzorylation of' a synthetic peptide, connexin 32 (residues 228-239), by (AMP-PK and PKC. The synthetic peptide PSRKGSGFGHRL-amidc was incubated with [y-32P]ATP and either (A) CAMP-PK, or (B) PKC. An additional aliquot ofcAMP-PK or PKC was added to the respective incubation mixture after 30 min. At intervals an aliquot from each reaction mixture was analyzed by HPLC. Peptide peaks wcrc monitored by ultraviolet detection at 214 nm and 258 nm, fractions were collected and Ihe "P-content measured by Cercnkov counting. The amount of unphosphorylated peptide (0---0, peak I), pcptide phosphorylatcd at Ser233 (O-- -O, peak 2), peptide phosphorylated at Scr22Y (A- --A, peak 3) and peptide phosphorylatcd at both Ser229 and Ser233 (A---A, peak 4) were estimated by measuring the absorbance peak heights at 214 nm. Results are representative of those obtained from two experiments

with a retention time of 21.5 min. The phosphorylation of peptide was rapid with maximal levels of product being obtained after 5- 10 min (Fig. 3A). The peptide was stoichio- metrically phosphorylated as demonstrated in the presence of only the phosphorylated form in the samples separated by HPLC. The addition of more CAMP-PK after 30 min did not increase the phosphorylation of the peptide any further. The results obtained with PKC were more complex (Fig. 3B). At the beginning of the phosphorylation a major phosphopeptide (peak 2) was eluted with a retention time of 21.5 min. In ad- dition, a minor phosphopeptide (peak 3) was eluted with a retention time of 22.0 min. As the reaction progressed, the level of peak 2 decreased and was essentially zero after 60 min. The level of peak 3 decreased only slightly even though additional PKC was added after 30 min. As the level of peak 2 decreased, the level of a third phosphopeptide (peak 4), which was eluted with a retention time of 20.7 min, increased. Based on a com-

269

Fig. 4. Two-diniensional phosphopeptide mapping of tryptic. digests of connexin 32 (residues 228 - 239) and connexin 32. Phosphopeptides werc separated by electrophoresis and chromatography as described in Fig. 2. (A) Connexin 32 (residues 228 -239) was incubated with [y-"PIATP and CAMP-PK and the 32P-labeled peptide purified by HPLC. The purified peptide was then digested with trypsin, (ratio 5: 1 proteasejpeptide). (B) Gap junctions werc incubated with [y-32P]ATP and CAMP-PK, the samples analyzed by SDSjPAGE and autoradiography and thc gel piece containing 32P-labeled connexin 32 incubated with trypsin. (C) A mixture comprising of equal amounts of the tryptic digests of connexin 32 (rcsiducs 228- 239) and connexin 32. The major peptide detected was labeled as peptide I based on its identical migration properties of peptide 1 derived from connexin 32 phosphorylated in purified gap junctions (see Fig. 2A). It is not known if the minor peptide obscrved to the right of peptide 1 is the same as the minor peptide detected to the right of peptide 1 in Fig. 2A

Table 3 Amino mid sequences of peptides Dehydroserine (dS)-identified as an adduct with dithiothreitol resulting form the /I-elimination of phosphoserine during the Edman degradation. Complete digestion with trypsin requircd high ratios of protease to peptide ( 5 : 1) probably bccause of the proximity of the phosphory- lated seryl residues to the argininyl and lysyl residues. The sequence of the parent peptide, connexin 32 (residues 228-239) was PSRKGSGFGHRL-amide. Tryptic digest of peak 4 resulted in the appearance of two phosphopeptides. The proposed phosphorylation sites are underlined

Peak Kinasc Phosphopcptide Phosphorylation site scqucnce

2 CAMP-PK KG(dS)GFGHR PSRKGSGFGHRL-amide 2 PKC KC(dS)GFGHR PSRKGSGFGHRL-amide 3 PK C P(dS)RK PSRKGSGFGHRL-amide 4 PK C P(dS)R PSRKGSGFGHRL-amidc

KG(dS)GFGHR

Two-dimensional peptide nzapping of tryptic digests ofconnexin 32 and connexin 32 (residues 228-239) phosphoryluted by CAMP-PK

Connexin 32 in isolated gap junctions was phosphorylated by CAMP-PK, the samples separated by SDSjPAGE and gel pieces containing the 32P-labeled protein incubated with tryp- sin. Connexin 32 (residues 228 - 239) was phosphorylated by CAMP-PK and the phosphopeptide purified by HPLC and then digested with trypsin under conditions which generated the phosphopeptide KGS(P)GFGHR. Two-dimensional peptide mapping showed that the phosphopeptides resulting from the tryptic digests of connexin 32 and the synthetic peptide had identical migration properties (Fig. 4A and B), and comigrated when mixed together (Fig. 4C). In addition, the 32P-labeled peptides from each digest had identical reten- tion times after purification by HPLC (not shown).

parison of the radioactivity present in the three 32P-labeled peptides with their absorption at 214 nm, it appeared that the peptide present in peak 4 had twice the specific activity of the peptides in peaks 2 and 3 (not shown).

Anzino acid sequencing of tryptic digests of connexin 32 (residues 228 - 239) phosphorylated by CAMP-PK and PKC

The phosphopeptides present in peak 2 which had been phosphorylated by CAMP-PK, and in peaks 2, 3 and 4 which had been phosphorylated by PKC, were and digested with tryp- sin at a protease/peptide ratio of 5 : l (by mass). The tryptic fragments were repurified using HPLC and the amino acid se- quences of the purified phosphopeptides determined and com- pared with that of the parent peptide (Table 1). The results from the amino acid sequencing, together with the results illustrated in Fig. 3, indicated that CAMP-PK phosphorylated the peptide only on Ser233. When the peptide was treated with PKC, Ser233 was predominantly phosphorylated while Ser229 was phos- phorylated to a lesser extent. However, at longer incubation times, the peptide, initially phosphorylated on Ser233, was also slowly phosphorylated on Ser229. Based on the fact that the levels of peak 3 decreased slightly it is possible that the peptide initially phosphorylated on Ser229 may also have been sub- sequently phosphorylated on Ser233, albeit even more slowly.

Kinetics of the phosphorylation of synthetic peptides by CAMP-PK and PKC

The rate of phosphorylation of connexin 32 (residues 228 - 239) by CAMP-PK was compared to the rate of phos- phorylation of a synthetic peptide based on the amino acid sequence of synapsin 1 (site I), which is a very efficient substrate for CAMP-PK [47]. CAMP-PK phosphorylated connexin 32 (residues 228-239) with a V,,, of 16.7 pmol min- ' mg- ' and a K, of 82 pM (average of four experiments) and phosphorylated synapsin 1 (site I) with a V,,, of 23.4 pmol . min-' . mg-' and a K, of 18 pM (average of three experiments). The rate of phosphorylation of connexin 32 (resi- dues 228 - 239) by PKC was compared to the phosphorylation of a synthetic peptide based on the amino acid sequence of myelin basic protein, also a very efficient substrate for PKC [48]. PKC phosphorylated connexin 32 (residues 228 ~ 239) with a V,,, of 2.8 pmol . min-' . mg-' and a K, of 112 pM (werage of two experiments) and the myelin basic protein peptide, [Ah' 05]myelin basic protein (residues 104 - I1 X), with a V,,, of 1.2 pmol . min-' . mg-l and a K,, of 58 pM (one experiment). Under the experimental conditions used Ser233 would be expect- ed to be phosphorylated by PKC at a rate 2 - 3-fold faster than Ser229.

Phosphorylation of connexin 32 in heputocytes

Previous studies of partially dissociated rat hepatocytes have shown that connexin 32 is phosphorylated and that the

270

97 - 2 68-

45- (I) v)

2 30- U

-I 3

W -I

h

Y v

a

a

0 20-

0 = 14-

Originl phosphorylation of connexin 32, although the effects were less reproducible (not shown).

c C O N N 3 2

C F PDBt 4a PI Fig. 5. Forskolin and PDBt stirnulution of thephosphorylation qf connexin 32 in hcpfocy tc~s . Partially dissociated hepatocytes in suspension (lo6 cells/’250 pl) were incubated for 1 h in phosphate-free Krebs buffer con- taining 10 mM glutaminc and [3’P]orthopbosphoric acid (0.5 mCi/250 pl of cell suspension). Cells were treated as described. (C) No addition, (F) 20 pM forskolin, (PDBt) 0.1 pM PDBt or (4%) 0.1 pM 4%-phorbol. After incubation for 15 min cells were disruptcd with SDS (1% final concen- tration) and connexin 32 (CONN32) was immunoprecipitated and the samples were analyzed by SDSjPAGE and autoradiography. (PI) shows an example of the background obtained when preimmune serum was used without treatment. The band of approximately 20 kDa was present in immunoprecipitates using preimmune serum and showed variable intensity irrespective of treatment. Using the autoradiograms as guides, the band containing phosphorylated connexin 32 was cut from the dried gel and the 32P content dctcrmined by liquid scintillation counting. The 27-kDa region of the gel present in the preimmune lane for each treatment was also counted and subtracted as a blank. Forskolin or PDBt increased phosphorylation of connexin 32 by approximately 30%. Average incor- poration obtained from three different experiments was as follows: con- trol. 929 cpm; forskolin, 1183 cpm; PDBt, 1200 cpm

phosphorylation of connexin 32 is increased by agents that elevate CAMP levels [26,37,38]. To analyze the sites phosphor- ylated in connexin 32 in intact cells, either partially dissociated rat or mouse hepatocyte suspensions were labeled with [32P]orthophosphoric acid. The phosphorylation of connexin 32 was measured in control conditions and in response to treatment with forskolin, an activator of adenylyl cyclase, with phorbol esters, which activate PKC; or in a CaZf-containing buffer with ionomycin, a Ca2 + ionophore, which may activate both PKC and Ca2+/CaM-PK. Hepatocytes were treated for 5 - 15 min and then were solubilized with SDS. Connexin 32 was immunoprecipitated, the proteins separated by SDS/ PAGE, and autoradiography performed (Fig. 5). Phosphor- ylated connexin 32 was immunoprecipitated from control cells when a specific rabbit anti-(connexin 32) antibody was used but not when preimmune serum was used. As previously de- scribed for 8-bromo-CAMP, treatment of the cells with forskolin increased the phosphorylation of connexin 32 1.3- fold compared to the control cells, as did treatment of cells with the phorbol esters, PDBt (Fig. 5 ) or phorbol12-myristate 13-acetate (not shown). Treatment with 4a-phorbol, a phorbol ester analog that does not activate PKC, had no effect on the phosphorylation of the protein compared to the control cells. Similar increases in phosphorylation of connexin 32 in re- sponse to activation of PKC have recently been obtained [37]. Treatment with ionomycin showed a small increase in the

Two-dimensional phosphopeptide mapping and phosphoamino acid analysis of 32P-labeled connexitz 32 immunoprecipitated from hepatocytes

Two-dimensional phosphopeptide mapping of connexin 32 from control cells (Fig. 2A) revealed the presence of a major phosphopeptide. This peptide appeared to be peptide 1, phosphorylated in connexin 32 in purified gap junctions by CAMP-PK or PKC (cf. Figs 2 and 6). Several other minor phosphopeptides were also detected in basal conditions which comigrated with peptides 2 - 5, phosphorylated in connexin 32 in purified gap junctions by PKC. 32P-incorporation into peptide 1 was increased in samples obtained from cells treated with either forskolin (Fig. 6B) or PDBt (Fig. 6C). Treatment with PDBt also resulted in an increase in the phosphorylation of peptides 2-5, particularly peptide 2. In control cells and cells treated with forskolin or PDBt, connexin 32 was phos- phorylated only on serine (Fig. 6 D). Peptides 6 - 9, phos- phorylated in connexin 32 in purified gap junctions by Ca2 ’ / CaM-PK 11, were not detected in control cells. In some exper- iments ionomycin treatment resulted in a small increase in the phosphorylation of peptides 1 and 3-5 but the phospho- peptides 6-9 were never observed (not shown).

Activity ratio of CAMP-PK in cells treated with PDBt

Although the effect of PDBt was likely to be mediated by PKC, the fact that the phosphorylation of peptide 1 was increased in response to both forskolin and PDBt suggested that the phorbol ester might be acting indirectly by increasing cAMP levels. To test this possibility we compared the levels of CAMP in the control cells and cells treated with forskolin or phorbol ester by measuing the CAMP-PK activity ratio [51]. In the control cells, after incubation for 5 - 30 min, the activity ratio was 2.8 (average of five measurements). Forskolin treat- ment changed the activity ratio to 1.1, indicating that cAMP levels were stimulated. Treatment with PDBt had no effect on cAMP levels as indicated by an activity ratio of 3.2.

Tumor-promoting phorhol ester does not block dye transfer in rut hepatocytes

Several studies have demonstrated that activation of pro- tein kinase C was associated with the closure of the gap- junction channel [33, 341. In some cases this effect could be prevented by activation of CAMP-PK [65]. Thus, differential phosphorylation of the gap-junction protein by CAMP-PK and PKC could explain the opposing effects of cAMP and phorbol esters. However, the present study suggests that in rat hepatocytes activation of CAMP-PK or PKC results in the phosphorylation of a common residue in connexin 32 (Ser233). It was important, therefore, to study directly whether activation of PKC decreased coupling in rat hepatocytes. In this study the dye coupling was measured. In well-coupled cells, such as rat hepatocytes, this method is only appropriate for determining decreases in junctional permeability. The incidence of Lucifer yellow dye coupling between cell pairs of freshly dissociated rat hepatocytes was virtually 100% (see [55]). The incidence of coupling measured after PDBt treat- ment (10 nM or 100 nM) at 15 min, 1 h, 3 h or 5 h was still 100%.

271

Fig. 6. Two-dimensional phosphopeptide mapping and phosphoamino acid analysis of connexin 32 immunoprecipitated from hepatocytes. Hepatocytes were labeled with [32P]orthophosphoric acid and treated with forskolin or PDBt, and connexin 32 was immunoprecipitated as described in Fig. 5. Gel pieces containing phosphorylated connexin 32 were cut from the dried gel and incubated with trypsin as described in Experimental Procedures. Samples were (A) control cells or (B) cells treated with forskolin or (C) PDBt. Phosphopeptides were separated by electrophoresis and chromatography, and autoradiography was performed as described in the legend to Fig. 2. The relative levels of peptides 1 - 5 were measured by densitometry. The values for the density of each spot (corrected for background) were as follows: control, peptide 1, 6024; peptide 2, 581; peptide 3, 1831; peptidc 4; 1087; peptide 5, 1163. Forskolin, peptide 1, 7827; peptide 2, 551; peptide 3, 2918; peptide 4, 859; peptide 5, 1058. PDBt, peptide I , 10837; peptide 2, 3153; pcptide 3, 3861 ; peptide 4, 2751; pcptide 5, 1782. (D) Aliquots of the tryptic digests hydrolyzed in 6 M HCI for 2 h. One-dimensional electrophoresis and autoradiography was performed. Samples were from control cells (C), or cells treated with forskolin (F) or PDBt. The position of standards are indicated; phosphoserine (P-Ser), phosphothreonine (P-Thr), phosphotyrosine (P-Tyr). Only seryl residues were phosphorylated. Results are representative of those obtained from three different experiments

DISCUSSION

In the present study we have carried out a detailed analysis of the phosphorylation of connexin 32 in purified gap junc- tions and in intact hepatocytes. A comparison of two-dimen- sional tryptic phosphopeptide maps generated from the pro- tein under the various conditions enabled several conclusions to be made concerning the kinase systems involved in phos- phorylating connexin 32. In isolated gap junctions, CAMP- PK phosphorylated only one site present in connexin 32 in a peptide, termed peptide 1. (Although sites are referred to as being single, the possibility that peptides, apart from peptide 1, may contain multiple phosphorylation sites cannot be ruled out.) In intact hepatocytes, under basal conditions, connexin 32 was phosphorylated at several sites (peptides 1 - 5 ) , with the most abundant phosphorylation being found in peptide 1. Treatment of hepatocytes with forskolin increased the phos- phorylation of peptide 1 only. In isolated gap junctions, PKC phosphorylated peptides I - 5 in connexin 32. Treatment of hepatocytes with PDBt increased the phosphorylation of peptides 1 - 5 that were phosphorylated under basal con- ditions. In isolated gap junctions, Ca2+/CaM-PK I1 phos- phorylated several peptides in connexin 32, although none of these peptides were detected in the protein labeled under any

conditions in intact cells, including studies in which the cal- cium ionophore, ionomycin, was used. Under any conditions with intact cells or when the protein was phosphorylated by purified CAMP-PK or PKC only phosphoserine was detected in connexin 32.

These results suggest that CAMP-PK phosphorylates one site (peptide 1) in connexin 32. PKC phosphorylates the same site as well as several other sites. The fact that PKC phosphor- ylates peptide 1 in connexin 32, present in isolated gap junc- tions, suggests that the phosphorylation of peptide 1 in intact hepatocytes is catalyzed directly by PKC and not because PDBt treatment increases CAMP levels as seen in some other systems [66]. In support of this suggestion, PDBt did not increase the CAMP-PK activity ratio in intact cells. The simi- larity of the phosphopeptide map of connexin 32, phosphor- ylated in isolated gap junctions by PKC, to that of the protein phosphorylated in intact cells under basal conditions, raises the possibility that the basal phosphorylation of the protein observed in intact cells is largely catalyzed by PKC.

Compared to good substrates for CAMP-PK, PKC or Ca2+/CaM-PK I1 [61], the rate and stoichiometry of the phos- phorylation of connexin 32 in isolated gap junctions were low. Similar results for CAMP-PK and PKC have been obtained by Takeda et al. [37]. Addition of detergent to the purified

272

gap-junction preparation itnproved the extent of phosphoryla- tion by PKC significantly. When the protein was incorporated into phosphatidylserine/diolein vesicles the phosphorylation by PKC could be increased even further. A possible interpre- tation of the results indicating low stoichiometry is that connexin 32 is phosphorylated to high levels in intact cells and that the protein is not dephosphorylated during isolation of the gap junctions. A high level of phosphorylation of the protein in intact cells would explain the small increases ob- served following the activation of the various second mes- senger pathways. However, a high level of phosphorylation of connexin 32 in purified gap junctions is unlikely since we have found in previous studies that prior treatment of purified gap junctions with protein phosphatases had little effect on the ability of connexin 32 to serve as a substrate for CAMP- PK [26]. In addition, substrates for CAMP-PK and PKC are normally isolated in their dephosphorylated forms due to the activity of endogenous phosphatases during the purification procedures. In the present studies we also found that the addition of a cocktail of protease inhibitors during the purifi- cation procedure had no effect on the phosphorylation of connexin 32 by any of thc kinases tested. It is more likely that the low stoichiometry of phosphorylation of connexin 32 in purified gap junctions is due to denaturation of the protein during its isolation. In support of this possibility recent results show that the secondary structure of connexin in gap-junction preparations depends on the method used for purification 1671. Circular dichroism spectroscopy indicated that alkalinc extraction, the method used in this study, resulted in a lower a-helical content compared to that obtained if detergent ex- traction was used. It will be of interest to examine the phos- phorylation of connexin 32 in gap junctions prepared by deter- gent extraction. The correct lipid environment also appears to be important for phosphorylation of connexin 32. Particular concentrations of phosphatidylserine and diacylglycerol sig- nificantly increased the phosphorylation of connexin 32 by PKC. Optimal phosphorylation of connexin 32 by PKC and by CAMP-PK, may require other particular lipid environments.

A synthetic peptide, PSRKGSGFGHRL-amide (connexin 32, residues 228 - 239) was phosphorylated stoichiometrically on Ser233 by CAMP-PK with kinetic parameters comparable to those of peptides derived from other physiological sub- strates for this enzyme. Ser233 is contained within the se- quence RKGS which conforms with the consensus sequence for phosphorylation by CAMP-PK [60 - 621. PKC phosphor- ylatcd connexin 32 (residues 228 -239) with Ser233 being phosphorylated at a rate 2 - 3-fold faster than Ser229. PKC phosphorylated the peptide with kinetic parameters compa- rable to those for peptides derived from other physiological substrates for this enzyme. Both Ser229 and Ser233 are contained within sequences which conform to the consensus sequences for PKC [48. 63, 641. A second synthetic peptide, LKRSPGTGAGLAEKSDRC-amide (connexin 32, residues 263-280), was not phosphorylated by PKC. There are ad- ditional seryl residues in the COOH-terminal region of connexin 32 (residues 240, 253,258 and 281) which we did not study using synthetic peptides. With the exception of Ser253, each of these are in close proximity to basic amino acids suggesting that they might be phosphorylatcd by PKC.

The two phosphorylated seryl residues determined in these studies are located close together in the COOH-terminal re- gion of connexin 32, a region shown to reside in the cytoplas- mic domain of the protein [68-701. Although the deduced amino acid sequence of the heart gap-junction protein, con- nexin 43 [lo]. exhibits a large amount of sequence identity with

connexin 32 in the putative membrane spanning domains of the two proteins, much of the cytoplasmic COOH-terminal region, including seryl residues 229 and 233, is not conserved. There are no apparent consensus sequences for CAMP-PK phosphorylation in connexin 43, although seryl residues in the region between positions 364 and 373 appear to be potential phosphorylation sites for PKC. The phosphorylation sites for many protein kinases appear to be highly conserved in homologous proteins from different tissues or species. The fact that the CAMP-PK phosphorylation site is not conserved between the heart and liver gap-junction proteins suggests important divergence in the regulation of the connexins by phosphorylation in these two tissues. A high degree of amino acid sequence identity also exists between connexin 32 and connexin 26 but the latter has a shorter cytoplasmic COOH- terminal region lacking potentially phosphorylatable residues [13]. The present results show that connexin 26 is not phos- phorylated in purified gap-junction preparations, in agree- ment with recent studies by Traub ct al. [14].

We have previously reported that the stimulation of phos- phorylation of connexin 32 by a CAMP-dependent mechanism is temporally correlated with an increase in junctional conduc- tance between pairs of adult rat hepatocytes [26]. In the present study the addition of phorbol estcr to primary cultures of hepatocytes did not reduce the incidence of dye coupling within the time course of the increase in phosphorylation of connexin 32 or over longcr periods of incubation (up to 5 h). In well-coupled cells, such as rat hepatocytes, this method is appropriate for determining only decreases in junctional permeability. Additional measurements of junctional conduc- tance by electrophysiological methods will be necessary to detect any increascs.

Other studies of the functional effects of second-messenger pathways on junctional conductance are complicated by the fact that conflicting data have been reported. Activation of PKC had either no effect or caused a slight increase in junc- tional conductance in rat pancreatic acinar cells [35] which express connexin 32 and connexin 26 [14]. In contrast, injec- tion of PKC or 1 -oleoy1-2-acetyl-sn-glycerol into mouse pan- creatic acinar cells decreased the junctional conductance [34]. Comparison of our results with those using other tissues are also complicated because different connexins are expressed in different tissues. CAMP increased junctional conductance in heart cells [27, 281 which express connexin 43 [lo]. However, CAMP was also found to decrease junctional conductance in Sertoli cells [31] which also express connexin 43 [71]. In con- trast, phorbol esters increased junctional conductance in heart cells (Burt, J. M. and Spray, D. C., unpublished results) and in Sertoli cells [31]. Phorbol esters, acetyl and diacylglycerol, all of which activate PKC, caused uncoupling in a variety of other cell lines [33] where the type of connexin expressed remains unknown.

The observation that activation of CAMP-PK or PKC can either increase or decrease cell coupling raises the possibility that CAMP-PK or PKC phosphorylate different sites in the relevant connexin in different cell types. It also possible that other mechanisms not involving phosphorylation of con- nexin(s) might explain the differential regulation in response to activation of CAMP-PK or PKC 121. Studies of the phos- phorylation of connexins in other cell types, coupled with measurements of junctional conductance, are necessary to establish if phosphorylation of the different forms of gap- junction proteins is responsible for the differential effects on gap-junctional conductance or permeability observed in dif- ferent cell types.

273

1.

2.

3. 4.

5.

6. 7.

8. 9.

10. 11. 12.

13.

14.

15.

16.

17. 18.

19.

20. 21.

22.

23.

24. 25.

26.

27.

28.

29.

30. 31.

32.

33.

The protein sequencing was performed by the Rockefcller Univer-

Health Services Grant MH 40899 (ACN) GM 30667 and a Research Career development award, HD00713 (ELH) NS-1624 and HL-34479 (DCS) and NS-07512 (MVLB).

34. Somogyi, R., Batzer, A. & Kolb, H.-A. (1989) J . Memhr. Bid .

35. Chanson, M., Bruzzone, R., Spray, D. C., Rcgazzi, R. & Meda,

36. Randriamampita, C., Giaume, C., Neyton, J . & Trautmann, A.

37. Takeda, A., Saheki, S., Shimazu, T. & Takeuchi, N . (1989) .I.

38. Traub, 0.. Look, J., Paul, D. & Willecke, K . (1987) Eur. J . Cell

sity Protein Facility. Research supported by United States Public 108, 273 -282.

P. (1988) Am. J . Physiol. 255, C699-C704.

(1988) Pflugers Arch. 412, 462-468.

Biochem. (Tokyo) 106, 723 -737. REFERENCES Bennett, M. V. L. & Spray, D. C. (1985) in Cup junctions(Bennett,

M. V. L. & Spray, D. C., eds) pp. 1-3, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

Saez. J . C., Spray, D. C. & Hertzberg, E. L. (1990) In Vitro

Pitts, J. D. & Sims, J . W. (1977) Exp. Cc~ll Res. 104, 153-163. Saez, J. C., Connor, J. A., Spray, D. C. & Bennett, M. V. L. (1989)

Goodenough, D. A. & Stoeckenius, W. (1972) J . Cell B id . 54,

Evans, W. H. & Gurd, J . W. (1972) Biochem. J . 128,691 -700. Hertzberg, E. L. & Gilula, N. B. (1979) J . Biol. Chem. 254,2138 -

Hertzberg, E. L. (1984) J . Biol. Chem. 259,9936-9943. Henderson, D., Eibl, H. & Weber, K. (1979) J . Mol. B id . 132,

Toxicol. 3, 69 - 86.

Proc. Natl Acud. Sci. USA 86,2708 - 2712.

646 - 656.

2147.

193 -218. P a d , D. (2986) J . Cell Biol. 103, 123-134. Kumar, N. & Gilula, N. B. (1986) J . Cell Biol. 103, 767-776. Beycr, E. C., Paul, D. L. & Goodenough, D. D. (1987) J . Cell

Zhang, J.-T. & Nicholson, B. J. (1989) J . Cell Bid . 109, 3391 -

Traub, O., Look, J., Dermietzel, R., Briimmer, F., Hiilser, D. &

Hertzberg, E. L. & Gilula, N. B. (1 982) Cold Spring Harbor Symp.

Nicholson, B., Dermictzel, R., Teplow, D., Traub, O., Willecke,

Unwin, P. N. T. & Zampighi, G. (1980) Nature 283, 545 - 550. Young, J . D.-E., Cohn, Z. A. & Gilula, N. 13. (1987) C d l 48,

733 - 743. Spray, D. C., SBez, J. C., Brosius, D., Bennett, M. V. L. &

Hertzberg, E. L. (1986) Proc. Nut1 A u d . Sci. USA 83, 5494- 5497.

Rose, B. & Loewenstein, W. R. (1975) Science 190, 1204-1206. Rose, B., Simpson, I. & Loewenstein, W. R. (1977) Nature 267,

Spray, D. C., Stern, J. H., Harris, A. L. & Bennett, M. V. L. (1982) Proc. Nut1 Acad. Sci. U S A 79,441 -445.

Van Eldick, L. J., Hertzberg, E. L., Berdan, R. C. & Gilula, N. B. (1985) Biochem. Biophys. Res. Commun. 126, 825-832.

Peracchia, C. (1987) Pflugers Arch. 408. 379-385. Arellano, R. O., Ramon, F., Rivera, A. & Zampighi, G. A. (1988)

J . Menzhr. Biol. 101, 119-132. Saez, J. C., Spray, D. C., Nairn, A. C., Hertzberg, E. L.,

Greengard, P. & Bennett, M. V. L. (1986) Proc. Natl Acud. Sci.

DeMello, W. C. (1 988) Biophys. Biochem. Res. Commun. 154,

Burt, J. M. & Spray, D. C. (1988) Am. .J. Physiol. 254, H1206-

Piccolino, M., Gerschenfeld, H. M. & Neyton, J. (1984) J .

Lasater, E. M. (1987) Proc. Natl Acad. Sci. USA 84,7319-7323. Grassi, F., Monaco, L., Fratamico, G., D o h , S., Ianini, E.,

Conti, M., Eusebi, F. & Stefanini, M. (1986) Cell Biol. Int. Rep.

Cole, W. C. &Garfield, R. E. (1986) Am. J . Physiol. 251, C411 - C420.

Trosko, J. E., Chang, C.-C., Madhukar, B. V., Oh, S. Y., Bom- bick, D. & El-Fouly, M. H. (1988) in Gap junctions (Hertzberg, E. L. & Johnson, R. G., eds) pp. 435-448, Alan R. Liss Inc, NY.

Biol. 105, 2621 -2629.

3401.

Willecke, K . (1989) J . Cell B i d . 108, 1039 - 1051.

Quant. Biol. 46, 639 - 645.

K. & Revel, J.-P. (1987) Nature 329, 732-734.

625 - 627.

USA 83,2473 -2477.

509-514.

H1210.

Neurosci. 4, 2477-2488.

10,631 -639.

Biol. 43, 48 - 54. 39. S k z , J . C., Spray, D. C., Bennett, M. V. L. & Hertzberg, E. L.

40. Takcda, A,, Hashimoto. E., Yamamura, H. & Shimazu, T. (1987)

41. SBez, J . C., Nairn, A. C., Spray, D. C., Hertzberg, E. L., Greengard, P. & Bennett, M. V. L. (1986) J . Cell Biol. 103,732.

42. Saez, J. C., Nairn. A. C., Spray, D. C., I-Iertzberg, E. L., Greengard, P. & Bennett, M. V. L. (1987) Soc. Neurosci. 13, 1133.

43. Kaczmarek, L. Y., Jennings, U. R., Strumwasser, F., Nairn, A. C., Walter, U., Wilson, F. D. & Greengard, P. (1980) Proc. Nut1 Acad. Sci. U S A 77, 7487 - 7491.

44. Kitano, T., Go, M., Kikkawa, U. & Nishizuka, Y. (1986) Methods Enzymol. 124, 349 -352.

45. McGuinness, T. L., Lai, Y. & Greengard, P. (1985) J . Biol. Chem.

46. Grand, R. J. A., Perry, S. V. & Weeks, R. A. (1979) Biochem. J .

47. Czernik, A. J., Pang, D. T. & Grcengrad, P. (1987) Proc. Nut1

48. Turner, R. S., Kemp, B. E., Su, H. D. & Kuo, J. F. (1985) J . Biol.

49. Hodgcs, R. S. & Merrifield, R. B. (1975) Anal. Biochmz. 65,241 -

SO. Hunkapillar, M. W.. Hewick. R. M., Dreyer, W. J. & Hood. L.

51. Berry, M. N. & Fricnd, D. S. (1969) J . Cell B id . 43, 506-520. 52. Connelly, P. A.. Sisk, R. B., Schulman, H . & Garrison. J. C.

53. Palmer, W. R., McPherson, J . M. & Walsh, D. A. (1980) J . Biol.

54. Bradford. M. M. (1976) Anal. Biochenz. 72, 248-254. 55. Saez, J. C., Gregory, W. A,, Watanabe, T., Dermietzel, R.,

Hertzberg, E. L., Ried, L., Bennett, M. V. L. & Spray, D. C. (1989) Am. J . Plzysiol. 257, C1-CII .

56. Laemmli, U. K. (1970) Nature 227, 680-685. 57. Huttner, W. B., DeGennaro, L. J . & Greengard, P. (1981) J . B id .

58. Nairn, A. C. & Greengard, P. (1987) J . Biol. Chem. 262, 7272-

59. O’Callahan, C. M., Ptasienski, J. & Hosey, M. M. (1988) J . B id .

60. Williams, R. E. (1986) Science 192, 473-474. 61. Glass, D. B. & Krebs, E. G. (1980) Annu. Rev. Pharmacol. To.xico1.

62. Cohen, P. (1988) Proc. R. Soc. 1,onu’. B. Bid . Sci. 234, 115- 144. 63. Woodgett, J. R., Could, K. L. & Hunter, T. (1986) Eur. J . Bio-

64. Haure, C., Wettenhall, R. E. & Kemp, B. E. (1987) J . Biol. Chem.

65. Enamoto, T., Martel, N., Kanno, Y. & Yamwaki, H. (1984) J .

66. Yoshimasa, T., Sibley, D. R., Bouvier, M., Lefkowitz, R. J . &

67. Cascio, M., Gogol, E. &Wallace, B. A. (1990) J . Biol. Chem. 265,

68. Zimmer, D. B., Green, C. R., Evans, W. H. & Gilula, N. B. (1987)

69. Hertzberg, E. L., Disher, R. M., Tiller, A. A,, Zhou, Y. & Cook,

70. Milks, L. C., Kumar, N. M., Houghtcn, R., Unwin, N. & Gilula,

71. Che, M., SBez, J. C. & Risley, M. S. (1989) Bid . Reprod. 40, 143.

(1985) Biophys. J . 47, 504a.

FEBS Lett. 210,169 - 172.

260, 1696 - 1704.

177, 521 - 529.

Acad. Sci. U S A 8 4 , 7518-7522.

Clze~?~. 260, 11 503 - 11 507.

272.

E. (1983) Method Enzynzol. 91. 399 -413.

(1987) J . Biol. Chem. 262, 30154-10163.

Chem. 255,2663 - 2666.

Chem. 256, 1482-1488.

7281.

Chem. 263, 17342-27349.

20, 363 - 388.

chem. 161, 177-184.

262,772 - 777.

Cell. Physiol. 121, 323-333.

Caron, M. G. (1987) Nature 327, 67-70.

2358 - 2364.

J . Biol. Chem. 262, 7751 -7763.

R. G. (1989) J . Biol. Chetlz.263, 19105-19111.

N. B. (1988) EMBO J . 7,2967-2975.