BIOCHEMICAL
Vol. 178, No. 3, 1991 August
15, 1991
AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 1312-1318
CONNEXON INTEGRITY IS MAINTAINED INTRAMOLECULAR
BY NON-COVALENT BONDS:
DISULFIDE BONDS LINK THE EXTRACELLULAR RAT CONNEXIN-43
DOMAINS IN
Scott A. John and Jean-PaulRevel Division of Biology 156-29 California Institute of Technology, Pasadena,California 91125
Received
June
14,
1991
Summary. Rat heart connexin-43 (RCx43) has been isolated using a modified procedure that is rapid and can be usedwith fresh or frozen hearts. When RCx43 is isolated in the presence of the alkylating reagent iodoacetamide,no intermolecular disulfide bonds are found. However, the alkylated RCx43 does have at least one intramolecular disulfide bond. By using site directed antibodies and proteolytic cleavage the location of the intramolecular disulfide bonding is shown between the two extracellular loops of RCx43.
All metazoan animals examined to date are linked by intercellular connections called gap junctions. These structures provide a pathway for the diffusion driven passageof small (-c Mr 1000) cytoplasmic moleculesand ions betweencontacting cells (1,2). The universal occurrence of gap junctions in organs,tissuesand speciessuggestsan essentialrole or roles.(For reviews see(3 and 4)). Gap junctions are composedof pairs of hemi-channels,called connexons. Connexonsin one cell’s plasmamembranecontact connexons in a neighboring cell’s plasmamembraneto form an aqueousconduit linking their cytoplasms. X-ray and electron diffraction studies show that each connexon is composedof six subunits(5,6,7,8). There is widespreadagreementin the literature that the structural proteins of the connexon are membersof a family of proteins called connexins. (However see(9 and 10)). The topology of connexin-43 (RCx43)l isolated from rat heart has been previously determined. The amino- and carboxy- termini and a Mr 4000 loop are on the cytoplasmic side. There are four transmembranespans(11) and two extracellular loops (12). The results of alignment and comparisonof nine different connexins (13), someof which have only been describedby DNA sequencing,suggeststhat all connexins have a similar topology. The multiple alignment showsa high degreeof conservation, particularly in the The abbreviations usedare: all connexins are abbreviated Cx followed by their predicted molecular massfrom translatedcDNA sequence.MOPS, PMSF, phenylmethylsulfonyl fluoride;DOC, sodiumdeoxycholate; PAGE, polyacrylamide gel electrophoresis. 0006-291X/91 Copyright All rights
$1.50
0 1991 by Academic Press, Inc. of reproduction in any form reserved.
1312
Vol. 178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL
RESEARCH COMMUNICATIONS
regions of the predicted extracellular loops. Each loop in RCx43 has three cysteines, with the consensusfor the first extracellular loop being CX6CX3C (Cys 54,61,65 ) and for the second extracellular loop CX4CX5C (Cys 187,192, 198 ). RCx31 is the exception (13), within the secondpredicted extracellular loop the cysteine pattern is CX5CX5C. Other cysteine residuesthat are not as well conserved between the various connexins are located either within the predicted fourth transmembranespanor within the cytoplasm (14). DuPont et al. (15) have shown that cysteine residuesof RCx43, found in the cytoplasmic Mr 17 000 carboxy terminal tail( 16), form artifactual intermolecular disulfide bonds during isolation. We now present evidence that the RCx43 molecule only has intramolecular disulfide bonds between the two extracellular loops:-non covalent bonds are therefore responsiblefor maintaining connexon integrity. Materials and Methods. Marerials Proteolytic enzymes were obtained from Boehringer Mannheim Biochemicals (Indianapolis, IN). Chemicalswere from SIGMA Chemical Co. (St Louis MO). Molecular weight standardswere obtained from Bio-Rad Laboratories (Cambridge, MA). Sprague-Dawley rats were suppliedby SimonsenLaboratories (Gilroy, CA). Isolation of RCx43 Enriched Fractions Hearts were removed from Sprague-Dawley (250300g) rats after cervical dislocation. Both freshly isolated and frozen hearts can be used in this protocol. Hearts were frozen by plunging in liquid nitrogen and stored at -80 C. To prepare a RCx43 enriched fraction the rat hearts were treated in a procedure modified from Lucas-Heron et al. (17) as follows. Approximately 3g of heart tissue (three rat hearts) were homogenizedin ice cold buffer I (12ml)l. All subsequentstepswere carried out at O-4 oC unlessotherwise specified. The homogenatewas filtered through six layers of cheesecloth, centrifuged in a Sorvall SS34 rotor at 3,000xg for 10 min. The pellet was resuspendedin buffer I and centrifuged in a Sorvall SS34rotor at 7,500xg for 10 min. The pellet was suspendedin buffer II (dilution l/6 v/v) and incubated with agitation for 30 mins at 4 oC. The resulting suspensionwas centrifuged in a Soivall SS34 rotor at 70,OOOxgfor 40 mins. The pellet was resuspendedin 5 mM Tris pH 10 (- 2.5ml) and an equal volume of 0.75% N-lauryl sarcosinein 5 mM Tris (pH 10) added and incubated with stirring for 15 mins at room temperature. The preparation of rat heart connexin-43 then followed the procedure as describedby (11). Antibodies Two previously characterized polyclonal site directed antisera were used in this study without prior purification. One was raised against a synthetic peptide corresponding to amino acids 2-20 of the N-terminal region of the RCx43 protein (AT-2)(11). The secondwas raised against a synthetic peptide corresponding to amino acids 186-206, the second extracellular loop of the RCx43 protein (EL-186)( 12). Miscellaneous Procedures Proteolysis of RCx43 using endoproteinaseGlu-C was carried out aspreviously described(11). SDS PAGE was carried out in 0.75mm slab gels with 3% stacking gels and 15% separatinggels as describedin (11). The samplebuffer (1X) had a final concentration of 2% SDS, 5% beta mercaptoethanol,5% glycerol, 0.004% bromophenolblue and 30mM Tris-HCl, pH 6.8. Sampleswere first solubilized in 4X samplebuffer. Transfer of proteins from gel to nitrocellulose was carried out using a semi-dry technique and immunoblotting was carried out asdescribedin (11). Results.Isolation of RCx43- The procedure usedfor isolating RCx43, detailed in the materials and methods,differs mainly from the published protocols in that the contractile 1Buffers I (20 mM MOPS, 100 mM KCl, ph 7.2) and buffer II (120 mM sodium pyrophosphate,20 mM MOPS, 100 mM KCl, pH 7.2) had 100 iodoacetamideand an equivalent, if it had all dissolved, of 1 mM PMSF. All subsequentstepshad 10 mM iodoacetamideup to and including the sucrosegradients. 1313
Vol.
178,
No.
BIOCHEMICAL
3, 1991
AND
BIOPHYSICAL
RESEARCH
COMMUNICATIONS
proteins are solubilized with sodium pyrophosphate, instead of high concentrations of potassium iodide. The ensuing treatment generally follows other methods used for the isolation of gap junctions. Analysis of the preparation by SDS PAGE and subsequent immunoblotting show that the AT-2 antiserum (recognizing amino acids 2-20 of the Nterminal sequence of RCx43) labels protein bands at 43 kDa , 35 kDa, 33 kDa, and 31 kDa (Fig 1A). These labelled bands are characteristic of the “native” protein and its three major breakdown products (18). Using this modified procedure we routinely isolate microgram amounts of RCx43 from three rat hearts (- 3 g). Immunoblot
Analysis of RCx43 after Alkylation-Alkylating
homogenization
the protein at the initial
step (see Methods) prevents the formation of artifactual disulfide bonds (15).
SDS PAGE of alkylated RCx43 protein under reducing or non reducing conditions, (Fig 1A and 1B respectively) followed by immunoblotting with the AT-2 antiserum shows labelling of protein bands at 43 kDa, 35 kDa, 34 kDa and 31 kDa. The electrophoretic
profiles are the same for both the reduced and non reduced samples,
demonstrating that allcylated RCx43 and its breakdown products are unaffected by reduction. This result indicates that there are no intermolecular disulfide bonds present between RCx43 molecules. (Evidence for intermolecular multimers with molecular weights
disultide bonds would have been the generation of
at least twice those detected in the reduced sample.)
Based upon the demonstration disulfide bonds in RCx43, we analyzed for the presence of
Immunoblot Analyses of Proteolysed Rex43 after Alkylation-
that there are no intermolecular
Figure
2 0 1. Immunoblotof Alkylated Preparationof RCx43. Rat heartRCx43 fractionswere
isolatedin the presenceof tbe alkylating reagentiodoacetamide, electrophoresed underSDS denaturingconditions,in the presence(A) or absence(3) of betamercaptoethanol, and transferredto nitrocellulose.Tbe blot wasprobedwith AT-2 antiserum.Numbersrefer to Mr x10-3. 2. Immunoblotof ProteolysedRCx43 after Alkylation. Rat heartRCx43 enriched fractionsweredigestedwith endoproteinase Glu-C, electrophoresed underSDS denaturing conditions,samples were solubilizedin the presence(lanes1 and3) or absence(lanes2 and 4) of betamercaptoethanol andtransferedto nitrocellulose.The blotswereprobedwith AT-2 antiserum(lanes1 and2) andEL-186antiserum(lanes3 and4). Numberscorrespondto molecularweightstandards(Bio-Rad)Mr x10-3 : phosphorylase B 97 400, bovine serum albumin,66 200ovalbumin42 700,carbonicanhydrase31 000, soybeantrypsin inhibitor 21 500, andlysozyme14400. Figure
1314
Vol.
178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL RESEARCH COMMUNICATIONS
intramolecular bonds. To this end, we treated the gap junction plaques, which had been alkylated with iodoacetamide, with proteolytic enzymes. Exogenously added enzymes cannot access those parts of the molecule buried within the membrane or present within the extracellular gap (19) they cleave protein only on the cytoplasmic side of the membrane. The exclusion of the gap will presumably not affect the access of iodoacetamide (154 Da) and prevent it reacting with any free sulfhydryl groups. The effect of such proteolytic cleavage of RCx43 has been well documented (11,12). Endoproteinase Glu-C generates two fragments -14 kDa and -18 kDa, each containing two transmembrane spans and one of the two extracellular loops. The 14 kDa fragment has the binding region for AT-2 but not for EL-186 and is thus composed of transmembrane spans one and two and the first extracellular loop. Conversely, the 18 kDa fragment only has the binding region for EL-186 and is thus composed of transmembrane spans three and four and the second extracellular loop. The products of endoproteinase Glu-C digestion of alkylated RCx43 were examined by immunoblotting. Proteolyzed plaques were either solubilized under reducing or non reducing conditions. Probing of the reduced proteolytic fragments with the AT-2 antiserum labelled a major band at 14 kDa. When the non reduced proteolytic fragments were probed with the AT-2 antiserum a band at -30 kDa was labelled. Examination of the reduced proteolytic fragments by immunoblotting with the EL-186 antiserum showed binding to a major band at - 18 kDa. This band is more diffuse than the 14 kDa band, a result previously described (11). When the non reduced proteolytic fragments were analyzed with the EL-186 antiserum, a major band at -30 kDa was labelled. Higher molecular weight bands which are labelled by the two antisera are presumbed to be multimers, a common characteristic of connexins (e.g. see (11 and 12)) (Fig 2). Since both AT-2 and EL-186 antisera label the same -30 kDa band in the non reduced, alkylated protein, but to two separate bands, 14 kDa and 18 kDa respectively, under reducing conditions, there must be at least one intramolecular disulfide bond linking the two membrane protected domains. Discussion.Our results show that the cytoplasmic cysteine residues do not form intermolecular disulfide bonds, because the electrophoretic profile of alkylated RCx43 and its characteristic breakdown pattern is the same, whether the protein is reduced or not. A similar result has been reported earlier (15). However Manjunath and Page (14), using 0.1 mM p-hydroxymercuribenzoate (PHMB) to block free sulfhydryl groups in the initial homogenization step, had earlier concluded that the cytoplasmic cysteine residues did form intermolecular disulfide bonds. A possible explanation for this apparent discrepancy may be that 0.1 mM PHMB was insufficient to prevent oxidization of free cysteine residues in the cytoplasmic C-tail. Both we and DuPont et al. (15) used much higher concentrations, 100 mM iodoacetamide, of sulfhydryl reagent in the initial homogenization step. Analysis of fragments of alkylated RCx43 resulting from the proteolysis of isolated plaques by reducing and non reducing SDS PAGE followed by immunoblotting showsthere is at least one disulfide bond linking the two membraneembeddeddomains; fragments 14 kDa
1315
Vol. 178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL
RESEARCH COMMUNICATIONS
and 18 kDa . Since alkylation prevents the formation of artifactual intermolecular disulfide bonds we assumeany intramolecular disulfide bonds present are in the native protein and not generatedduring the isolation procedure. The RCx43 amino acid sequenceas deduced from cDNA (20) showsthat as many as six cysteine residuescould be involved in disulfide bond formation in the extracellular region, as there are three cysteine residuesin each extracellular loop (loop I: cys 54,617 65 and loop 2: cys lg7,192,19g). The intramolecular disulfide bond(s) linking fragment 14 kDa with 18 kDa must occur in the extracellular region. Whether more than one disulfide bond is present has not been determined. So how many are there? Between the six cysteine residuesthere are a maximum of three potential disulfide bondslinking the two loops. Our data indicate at least one linkage between the two extracellular loops, an “inter-loop” disulfide bond. The two remaining cysteinesin each loop could fotm “intra-loop” disulfide bonds. A second “inter-loop” disulfide linkage would leave a single unpaired cysteine residue in each loop. However, an unpaired cysteine residue is under evolutionary pressureto be removed Thus, one would predict that the number of “inter-loop” disulfide bondsis likely to be either one or three. The more bonds linking the extracellular loops the more structural constraint is placed on the molecule, particularly in the extracellular region and the arrangementof the transmembranehelices. The extracellular region of RCx43 is responsiblefor two functions in viva: self recognition i.e connexon to connexon pairing acrossthe extracellular spaceallowing the gap junction to form; and, as a corollary the extension of the aqueousconduit acrossthe extracellular gap. The highly conserved nature of the extracellular loops and particularly the conservation of the cysteine residues(Fig 3) suggeststhat theseresiduesand the bond(s) they form may play an essentialrole in these two functions. The extracellular disulfide bonding we show here may also be involved with the spatial alignment of the transmembranespans;enabling the correct formation or maintenanceof the channel lining residues.Precedentfor such a role comesfrom studieson bovine rhodopsin.
EL1 RCX26
xcx30 RCX31 HCX32 RCX32
XCX38 CCx42 xx43 CCX43 HCx43 RCX43 xcx43 ccx45
DFVCNTLQIPG AFTCNTQQPG DFDCNTRQPG SFICNTLQPG SFICNTLQPG DFICNTQQPG DFMCDTQQPG AFRCNTQQPG AFRCNTQQPG AFRCNTQQPG AFRCNTQQPG AFVCNTQQPG KFVCNTEQPG
CKNVCYDHYF CNSVCYDHFF CTNVCYDNFF CNSVCYDQFF CNSVCYDHFF CTNVCYDQAF CENVCYDKAF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDRFA
EL2 PISHIRLWAL PISHIRLWAL PISNIRLWAL PISHVRLWSL PISHVRLWSL PIYHVRYWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PLSHVRFWVF
FF"QRLVKCN YSMIRLLKCD FTMPRLWXA YAMVRLVKCD YA"VRLVKCE FVMSPIFVCE IFLETLYICQ FSLSAVYTCK FSLSAIYTCE FSLSAVYTCK FSLSAVYTCK FSLSAIYXK FEVSPVFVCS
.AWPCPNTVD .AYPCPNTVD SVVFCPNTVD .VYPCPNTVD .AFPCPNTVD .RIPCKHKVS .RAPCPHPVN .RDPCFHQVD .RDPCPHRVD .RDPCPHQVD .RDPCPHQVD .RDPCPHQVD .RKPCPHKID
CFISRPTEXT CFVSRPTEKT CYIARPTEKK CFVSRPTEKT CFVSRPTEKT CFVSRPMEKT CWSRPTEKN CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFISRPTEKT
Figure 3.Extracellular Cysteine Rich Alignment of Connexins. The two putative extracellular loops for 13 connexins are shown with the following ptefixes:B, bovine; C, chicken; H, human, R, rat; X, Xenopur. The numbers refer to their predicted molecular weight determined by the translation of DNA into protein sequence. RCx26 from (25), XCx30 from(26). RCx31 from (13), HCx32 from (27), ,RCx32 from (28), XCx38 from (29), CCx42 from (30). BCx43 from (31), CCx43 from (32), HCx43 from (33),RCx43 from (20), XCx43 from (34) CCx45 from (30).
1316
Vol.
178,
No.
BIOCHEMICAL
3, 1991
AND
BIOPHYSICAL
Site directed mutation analysis has shown that the intradiscal
RESEARCH
COMMUNICATIONS
(=extracellular)
disulfide
between cysteines 110 and 187 in rhodopsin is necessary for the correct alignment of the membrane helices and assembly of the functionally
critical cytoplasmic domain (21).
Additional evidence for extracellular disultide bonds playing important roles in the maintenance of transmembrane structure comes from studies on the beta adrenergic receptor where two disulfide bonds am necessary for ligand binding, which is presumed to occur within the hydrophobic transmembrane domains of the protein (22). The demonstrated lack of intermolecular disulfide bonds makes it likely that individual subunits of the connexon are held together by non-covalent bonds. Covalent intermolecular bonds such as desmosine and their analogues, which cross link polypeptides through lysyl or 3-3’methylbityrosyl proteolysis
residues (23) would not yield the electrophoretic
profile observed after
and reduction. Similar reasoning applies to connexon-connexon
interactions,
whether between connexons of the same cell or those from the apposing cell. Thus the integrity of gap junction plaques is maintained by bonds of a non-covalent nature. The role of protein-lipid
interactions
in the maintenance of connexon and plaque integrity has not been
established (24). In conclusion, we have shown that RCx43 is internally disulfide bonded between the two extracellular loops. This linkage may play important roles in the self recognition of connexons across the gap and /or may provide stability for the aqueous conduit that spans the extracellular gap. The homology that exists between the various connexins in the extracellular regions suggests that the demonstrated disulfide bond formations for RCx43 are a common feature of the connexins. In agreement with Dupont et al. (15) however, there are no intermolecular
disulfide bonds between RCx43 molecules.
Acknowledmnents We thank Drs Jan H.Hoh, S.Barbara Yancey and Dale W. Laird for their critical review of the manuscript. Initial studies into the disulfide bonds present in RCx43 were carried out in collaboration with Drs S.Barbara Yancey and Dale W. Laird.
References 1. 2. 3. 4.
Flagg-Newton, J., Simpson, I. and Loewenstein, W.R.(1979) Science 205,404-407. Barr, L., Dewey, F.F.and Berger, W.(1965) J. Gen. Physiol. 48,797-823. Beyer, E.C., Paul, D.L.and Goodenough, D.A.(1990) J. Membr. Biol. 116, 187-194. Bennett, M.V.L., Barrio, L.C., Bargiello, T.A., Spray, D.C., Hertzberg, E. and Saez, J.C.(1991) Neuron 6,305-320. 5. Makowski, L., &spar, D.L.D., Philips, W.C.and Goodenough, D.A.( 1977) J. Cell Biol. 74, 629-645. 6. Caspar, D.L.D., D. A. Goodenough, L. Makowski and Phillips, W.C.(1977) J. Cell Biol. 74, 605-628. 7. Unwin, P.N.T.and Zampighi, G.(1980) Nature 283,545-549. 8. Unwin, P.N.T.and Ennis, P.D.(1984) Nature 307, 609-613. 9. Finbow, M.E., J. Shuttleworth, A. E. Hamilton and Pitts, J.D.(1983) EMBO J. 2, 14791486. 10. Leitch, B., and Finbow, M.E.(1990) Exp. Cell Res. 190,218-226. 11. Yancey, S.B., John, S.A., Lal, R., Austin, B.J.and Revel, J.-P.(1989) J. Cell Biol. 108, 2241-2254. 12. Laird, D.W.and Revel, J.-P.(1990) J. Cell Sci. 97, 109-117. 1317
Vol.
178,
No.
3, 1991
BIOCHEMICAL
AND
BIOPHYSICAL
RESEARCH
COMMUNICATIONS
13. Hoh, J.H., John, S.A.and Revel, J.-P.(1991) J. B. C. 266,6524-6531. 14. Manjunath, C., K. and Page, E.(1986) J. Membrane Biol. 90,43-57. 15. DuPont, E., El Aoumari, A., Briand, J.P., Fromaget, Cand Gros, D.(1989) J. Membrane Biol. 108,247-252. 16. Manjunath, C.K., Nicholson, B.J., Teplow, D., Hood, L., Page, E., and Revel, J.-P.( 1987) B&hem. Biophys. Res. Commun. 142,228-234. 17. Lucas-Heron, B., Loirat, M.J., and Ollivier, B.(1987) Comp. B&hem. Physiol. 88B, 421427. 18. Manjunath, C.K., Goings, G.E.and Page, E.(1984) Am. J. Physiol. 246, H865-H875. 19. Goodenough, D.A., and Revel, J.-P.(1971) J.Cell Biol. 50,81-91. 20. Beyer, E.C., Paul, D.L. and Goodenough, D.A.(1987) J. Cell Biol. 105,2621-2629. 21. Karnik, S.S.and Khorana, H.G.(1990) J. Biol. Chem. 265, 17520-17524. 22. Dohlman, H.G., Caron, M.G., DeBlasi, A., Frielle, T.and Lefkowitz, R.J.(1990) B&hem. 29,2335-2342. 23. Kikuchi, Y., Tsuchikura, O., Hirama, M., and Tamiya, N.( 1987) Eur. J. Biochem 164,397402. 24. Malewicz, B., Kumar, V.V., Johnson, R.G.and Baumann, W., J.(1990) Lipids 25,419-427. 25. Zhang, J.-T. and Nicholson, B.J.(1989) J. Cell Biol 109, 3391-3401. 26. Gimlich, R.L., Kumar, N.andGilula, N.B.(1988) J. Cell Biol. 107, 1065-1073. 27. Kumar, N.M.and Gilula, N.B.(1986) J. Cell Biol. 103,767-776. 28. Paul, D.L.(1986) JCell Biol 103, 123-134. 29. Ebihara, L., Beyer, E.C., Swenson, K.I., Paul, D.L.and Goodenough, D.A.(1989) Science 243, 1194-l 195. 30. Beyer, E.C.(1990) J. Biol. Chem. 265, 14439-14443. 31. Lash, J., A.,, Crister,E., S., andPressler, M.,L.(1990) J. Biol. Chem. 265, 13113-13117. 32. Musil, L.S., Beyer, E.C. and Goodenough, D.A.(1990) J. Membrane Biol. 116, 163-175. 33. Fishman, G.I., Spray, DC. and Leinwand, L.A.(1990) J. Cell Biol. 111,589-598. 34. Gimlich, R.L., Kumar, N. and Gilula, N.B.(1990) J. Cell Biol. 110,597~605.
1318
Vol. 178, No. 3, 1991 August
15, 1991
AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 1312-1318
CONNEXON INTEGRITY IS MAINTAINED INTRAMOLECULAR
BY NON-COVALENT BONDS:
DISULFIDE BONDS LINK THE EXTRACELLULAR RAT CONNEXIN-43
DOMAINS IN
Scott A. John and Jean-PaulRevel Division of Biology 156-29 California Institute of Technology, Pasadena,California 91125
Received
June
14,
1991
Summary. Rat heart connexin-43 (RCx43) has been isolated using a modified procedure that is rapid and can be usedwith fresh or frozen hearts. When RCx43 is isolated in the presence of the alkylating reagent iodoacetamide,no intermolecular disulfide bonds are found. However, the alkylated RCx43 does have at least one intramolecular disulfide bond. By using site directed antibodies and proteolytic cleavage the location of the intramolecular disulfide bonding is shown between the two extracellular loops of RCx43.
All metazoan animals examined to date are linked by intercellular connections called gap junctions. These structures provide a pathway for the diffusion driven passageof small (-c Mr 1000) cytoplasmic moleculesand ions betweencontacting cells (1,2). The universal occurrence of gap junctions in organs,tissuesand speciessuggestsan essentialrole or roles.(For reviews see(3 and 4)). Gap junctions are composedof pairs of hemi-channels,called connexons. Connexonsin one cell’s plasmamembranecontact connexons in a neighboring cell’s plasmamembraneto form an aqueousconduit linking their cytoplasms. X-ray and electron diffraction studies show that each connexon is composedof six subunits(5,6,7,8). There is widespreadagreementin the literature that the structural proteins of the connexon are membersof a family of proteins called connexins. (However see(9 and 10)). The topology of connexin-43 (RCx43)l isolated from rat heart has been previously determined. The amino- and carboxy- termini and a Mr 4000 loop are on the cytoplasmic side. There are four transmembranespans(11) and two extracellular loops (12). The results of alignment and comparisonof nine different connexins (13), someof which have only been describedby DNA sequencing,suggeststhat all connexins have a similar topology. The multiple alignment showsa high degreeof conservation, particularly in the The abbreviations usedare: all connexins are abbreviated Cx followed by their predicted molecular massfrom translatedcDNA sequence.MOPS, PMSF, phenylmethylsulfonyl fluoride;DOC, sodiumdeoxycholate; PAGE, polyacrylamide gel electrophoresis. 0006-291X/91 Copyright All rights
$1.50
0 1991 by Academic Press, Inc. of reproduction in any form reserved.
1312
Vol. 178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL
RESEARCH COMMUNICATIONS
regions of the predicted extracellular loops. Each loop in RCx43 has three cysteines, with the consensusfor the first extracellular loop being CX6CX3C (Cys 54,61,65 ) and for the second extracellular loop CX4CX5C (Cys 187,192, 198 ). RCx31 is the exception (13), within the secondpredicted extracellular loop the cysteine pattern is CX5CX5C. Other cysteine residuesthat are not as well conserved between the various connexins are located either within the predicted fourth transmembranespanor within the cytoplasm (14). DuPont et al. (15) have shown that cysteine residuesof RCx43, found in the cytoplasmic Mr 17 000 carboxy terminal tail( 16), form artifactual intermolecular disulfide bonds during isolation. We now present evidence that the RCx43 molecule only has intramolecular disulfide bonds between the two extracellular loops:-non covalent bonds are therefore responsiblefor maintaining connexon integrity. Materials and Methods. Marerials Proteolytic enzymes were obtained from Boehringer Mannheim Biochemicals (Indianapolis, IN). Chemicalswere from SIGMA Chemical Co. (St Louis MO). Molecular weight standardswere obtained from Bio-Rad Laboratories (Cambridge, MA). Sprague-Dawley rats were suppliedby SimonsenLaboratories (Gilroy, CA). Isolation of RCx43 Enriched Fractions Hearts were removed from Sprague-Dawley (250300g) rats after cervical dislocation. Both freshly isolated and frozen hearts can be used in this protocol. Hearts were frozen by plunging in liquid nitrogen and stored at -80 C. To prepare a RCx43 enriched fraction the rat hearts were treated in a procedure modified from Lucas-Heron et al. (17) as follows. Approximately 3g of heart tissue (three rat hearts) were homogenizedin ice cold buffer I (12ml)l. All subsequentstepswere carried out at O-4 oC unlessotherwise specified. The homogenatewas filtered through six layers of cheesecloth, centrifuged in a Sorvall SS34 rotor at 3,000xg for 10 min. The pellet was resuspendedin buffer I and centrifuged in a Sorvall SS34rotor at 7,500xg for 10 min. The pellet was suspendedin buffer II (dilution l/6 v/v) and incubated with agitation for 30 mins at 4 oC. The resulting suspensionwas centrifuged in a Soivall SS34 rotor at 70,OOOxgfor 40 mins. The pellet was resuspendedin 5 mM Tris pH 10 (- 2.5ml) and an equal volume of 0.75% N-lauryl sarcosinein 5 mM Tris (pH 10) added and incubated with stirring for 15 mins at room temperature. The preparation of rat heart connexin-43 then followed the procedure as describedby (11). Antibodies Two previously characterized polyclonal site directed antisera were used in this study without prior purification. One was raised against a synthetic peptide corresponding to amino acids 2-20 of the N-terminal region of the RCx43 protein (AT-2)(11). The secondwas raised against a synthetic peptide corresponding to amino acids 186-206, the second extracellular loop of the RCx43 protein (EL-186)( 12). Miscellaneous Procedures Proteolysis of RCx43 using endoproteinaseGlu-C was carried out aspreviously described(11). SDS PAGE was carried out in 0.75mm slab gels with 3% stacking gels and 15% separatinggels as describedin (11). The samplebuffer (1X) had a final concentration of 2% SDS, 5% beta mercaptoethanol,5% glycerol, 0.004% bromophenolblue and 30mM Tris-HCl, pH 6.8. Sampleswere first solubilized in 4X samplebuffer. Transfer of proteins from gel to nitrocellulose was carried out using a semi-dry technique and immunoblotting was carried out asdescribedin (11). Results.Isolation of RCx43- The procedure usedfor isolating RCx43, detailed in the materials and methods,differs mainly from the published protocols in that the contractile 1Buffers I (20 mM MOPS, 100 mM KCl, ph 7.2) and buffer II (120 mM sodium pyrophosphate,20 mM MOPS, 100 mM KCl, pH 7.2) had 100 iodoacetamideand an equivalent, if it had all dissolved, of 1 mM PMSF. All subsequentstepshad 10 mM iodoacetamideup to and including the sucrosegradients. 1313
Vol.
178,
No.
BIOCHEMICAL
3, 1991
AND
BIOPHYSICAL
RESEARCH
COMMUNICATIONS
proteins are solubilized with sodium pyrophosphate, instead of high concentrations of potassium iodide. The ensuing treatment generally follows other methods used for the isolation of gap junctions. Analysis of the preparation by SDS PAGE and subsequent immunoblotting show that the AT-2 antiserum (recognizing amino acids 2-20 of the Nterminal sequence of RCx43) labels protein bands at 43 kDa , 35 kDa, 33 kDa, and 31 kDa (Fig 1A). These labelled bands are characteristic of the “native” protein and its three major breakdown products (18). Using this modified procedure we routinely isolate microgram amounts of RCx43 from three rat hearts (- 3 g). Immunoblot
Analysis of RCx43 after Alkylation-Alkylating
homogenization
the protein at the initial
step (see Methods) prevents the formation of artifactual disulfide bonds (15).
SDS PAGE of alkylated RCx43 protein under reducing or non reducing conditions, (Fig 1A and 1B respectively) followed by immunoblotting with the AT-2 antiserum shows labelling of protein bands at 43 kDa, 35 kDa, 34 kDa and 31 kDa. The electrophoretic
profiles are the same for both the reduced and non reduced samples,
demonstrating that allcylated RCx43 and its breakdown products are unaffected by reduction. This result indicates that there are no intermolecular disulfide bonds present between RCx43 molecules. (Evidence for intermolecular multimers with molecular weights
disultide bonds would have been the generation of
at least twice those detected in the reduced sample.)
Based upon the demonstration disulfide bonds in RCx43, we analyzed for the presence of
Immunoblot Analyses of Proteolysed Rex43 after Alkylation-
that there are no intermolecular
Figure
2 0 1. Immunoblotof Alkylated Preparationof RCx43. Rat heartRCx43 fractionswere
isolatedin the presenceof tbe alkylating reagentiodoacetamide, electrophoresed underSDS denaturingconditions,in the presence(A) or absence(3) of betamercaptoethanol, and transferredto nitrocellulose.Tbe blot wasprobedwith AT-2 antiserum.Numbersrefer to Mr x10-3. 2. Immunoblotof ProteolysedRCx43 after Alkylation. Rat heartRCx43 enriched fractionsweredigestedwith endoproteinase Glu-C, electrophoresed underSDS denaturing conditions,samples were solubilizedin the presence(lanes1 and3) or absence(lanes2 and 4) of betamercaptoethanol andtransferedto nitrocellulose.The blotswereprobedwith AT-2 antiserum(lanes1 and2) andEL-186antiserum(lanes3 and4). Numberscorrespondto molecularweightstandards(Bio-Rad)Mr x10-3 : phosphorylase B 97 400, bovine serum albumin,66 200ovalbumin42 700,carbonicanhydrase31 000, soybeantrypsin inhibitor 21 500, andlysozyme14400. Figure
1314
Vol.
178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL RESEARCH COMMUNICATIONS
intramolecular bonds. To this end, we treated the gap junction plaques, which had been alkylated with iodoacetamide, with proteolytic enzymes. Exogenously added enzymes cannot access those parts of the molecule buried within the membrane or present within the extracellular gap (19) they cleave protein only on the cytoplasmic side of the membrane. The exclusion of the gap will presumably not affect the access of iodoacetamide (154 Da) and prevent it reacting with any free sulfhydryl groups. The effect of such proteolytic cleavage of RCx43 has been well documented (11,12). Endoproteinase Glu-C generates two fragments -14 kDa and -18 kDa, each containing two transmembrane spans and one of the two extracellular loops. The 14 kDa fragment has the binding region for AT-2 but not for EL-186 and is thus composed of transmembrane spans one and two and the first extracellular loop. Conversely, the 18 kDa fragment only has the binding region for EL-186 and is thus composed of transmembrane spans three and four and the second extracellular loop. The products of endoproteinase Glu-C digestion of alkylated RCx43 were examined by immunoblotting. Proteolyzed plaques were either solubilized under reducing or non reducing conditions. Probing of the reduced proteolytic fragments with the AT-2 antiserum labelled a major band at 14 kDa. When the non reduced proteolytic fragments were probed with the AT-2 antiserum a band at -30 kDa was labelled. Examination of the reduced proteolytic fragments by immunoblotting with the EL-186 antiserum showed binding to a major band at - 18 kDa. This band is more diffuse than the 14 kDa band, a result previously described (11). When the non reduced proteolytic fragments were analyzed with the EL-186 antiserum, a major band at -30 kDa was labelled. Higher molecular weight bands which are labelled by the two antisera are presumbed to be multimers, a common characteristic of connexins (e.g. see (11 and 12)) (Fig 2). Since both AT-2 and EL-186 antisera label the same -30 kDa band in the non reduced, alkylated protein, but to two separate bands, 14 kDa and 18 kDa respectively, under reducing conditions, there must be at least one intramolecular disulfide bond linking the two membrane protected domains. Discussion.Our results show that the cytoplasmic cysteine residues do not form intermolecular disulfide bonds, because the electrophoretic profile of alkylated RCx43 and its characteristic breakdown pattern is the same, whether the protein is reduced or not. A similar result has been reported earlier (15). However Manjunath and Page (14), using 0.1 mM p-hydroxymercuribenzoate (PHMB) to block free sulfhydryl groups in the initial homogenization step, had earlier concluded that the cytoplasmic cysteine residues did form intermolecular disulfide bonds. A possible explanation for this apparent discrepancy may be that 0.1 mM PHMB was insufficient to prevent oxidization of free cysteine residues in the cytoplasmic C-tail. Both we and DuPont et al. (15) used much higher concentrations, 100 mM iodoacetamide, of sulfhydryl reagent in the initial homogenization step. Analysis of fragments of alkylated RCx43 resulting from the proteolysis of isolated plaques by reducing and non reducing SDS PAGE followed by immunoblotting showsthere is at least one disulfide bond linking the two membraneembeddeddomains; fragments 14 kDa
1315
Vol. 178, No. 3, 1991
BIOCHEMICAL
AND BIOPHYSICAL
RESEARCH COMMUNICATIONS
and 18 kDa . Since alkylation prevents the formation of artifactual intermolecular disulfide bonds we assumeany intramolecular disulfide bonds present are in the native protein and not generatedduring the isolation procedure. The RCx43 amino acid sequenceas deduced from cDNA (20) showsthat as many as six cysteine residuescould be involved in disulfide bond formation in the extracellular region, as there are three cysteine residuesin each extracellular loop (loop I: cys 54,617 65 and loop 2: cys lg7,192,19g). The intramolecular disulfide bond(s) linking fragment 14 kDa with 18 kDa must occur in the extracellular region. Whether more than one disulfide bond is present has not been determined. So how many are there? Between the six cysteine residuesthere are a maximum of three potential disulfide bondslinking the two loops. Our data indicate at least one linkage between the two extracellular loops, an “inter-loop” disulfide bond. The two remaining cysteinesin each loop could fotm “intra-loop” disulfide bonds. A second “inter-loop” disulfide linkage would leave a single unpaired cysteine residue in each loop. However, an unpaired cysteine residue is under evolutionary pressureto be removed Thus, one would predict that the number of “inter-loop” disulfide bondsis likely to be either one or three. The more bonds linking the extracellular loops the more structural constraint is placed on the molecule, particularly in the extracellular region and the arrangementof the transmembranehelices. The extracellular region of RCx43 is responsiblefor two functions in viva: self recognition i.e connexon to connexon pairing acrossthe extracellular spaceallowing the gap junction to form; and, as a corollary the extension of the aqueousconduit acrossthe extracellular gap. The highly conserved nature of the extracellular loops and particularly the conservation of the cysteine residues(Fig 3) suggeststhat theseresiduesand the bond(s) they form may play an essentialrole in these two functions. The extracellular disulfide bonding we show here may also be involved with the spatial alignment of the transmembranespans;enabling the correct formation or maintenanceof the channel lining residues.Precedentfor such a role comesfrom studieson bovine rhodopsin.
EL1 RCX26
xcx30 RCX31 HCX32 RCX32
XCX38 CCx42 xx43 CCX43 HCx43 RCX43 xcx43 ccx45
DFVCNTLQIPG AFTCNTQQPG DFDCNTRQPG SFICNTLQPG SFICNTLQPG DFICNTQQPG DFMCDTQQPG AFRCNTQQPG AFRCNTQQPG AFRCNTQQPG AFRCNTQQPG AFVCNTQQPG KFVCNTEQPG
CKNVCYDHYF CNSVCYDHFF CTNVCYDNFF CNSVCYDQFF CNSVCYDHFF CTNVCYDQAF CENVCYDKAF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDKSF CENVCYDRFA
EL2 PISHIRLWAL PISHIRLWAL PISNIRLWAL PISHVRLWSL PISHVRLWSL PIYHVRYWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PISHVRFWVL PLSHVRFWVF
FF"QRLVKCN YSMIRLLKCD FTMPRLWXA YAMVRLVKCD YA"VRLVKCE FVMSPIFVCE IFLETLYICQ FSLSAVYTCK FSLSAIYTCE FSLSAVYTCK FSLSAVYTCK FSLSAIYXK FEVSPVFVCS
.AWPCPNTVD .AYPCPNTVD SVVFCPNTVD .VYPCPNTVD .AFPCPNTVD .RIPCKHKVS .RAPCPHPVN .RDPCFHQVD .RDPCPHRVD .RDPCPHQVD .RDPCPHQVD .RDPCPHQVD .RKPCPHKID
CFISRPTEXT CFVSRPTEKT CYIARPTEKK CFVSRPTEKT CFVSRPTEKT CFVSRPMEKT CWSRPTEKN CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFLSRPTEKT CFISRPTEKT
Figure 3.Extracellular Cysteine Rich Alignment of Connexins. The two putative extracellular loops for 13 connexins are shown with the following ptefixes:B, bovine; C, chicken; H, human, R, rat; X, Xenopur. The numbers refer to their predicted molecular weight determined by the translation of DNA into protein sequence. RCx26 from (25), XCx30 from(26). RCx31 from (13), HCx32 from (27), ,RCx32 from (28), XCx38 from (29), CCx42 from (30). BCx43 from (31), CCx43 from (32), HCx43 from (33),RCx43 from (20), XCx43 from (34) CCx45 from (30).
1316
Vol.
178,
No.
BIOCHEMICAL
3, 1991
AND
BIOPHYSICAL
Site directed mutation analysis has shown that the intradiscal
RESEARCH
COMMUNICATIONS
(=extracellular)
disulfide
between cysteines 110 and 187 in rhodopsin is necessary for the correct alignment of the membrane helices and assembly of the functionally
critical cytoplasmic domain (21).
Additional evidence for extracellular disultide bonds playing important roles in the maintenance of transmembrane structure comes from studies on the beta adrenergic receptor where two disulfide bonds am necessary for ligand binding, which is presumed to occur within the hydrophobic transmembrane domains of the protein (22). The demonstrated lack of intermolecular disulfide bonds makes it likely that individual subunits of the connexon are held together by non-covalent bonds. Covalent intermolecular bonds such as desmosine and their analogues, which cross link polypeptides through lysyl or 3-3’methylbityrosyl proteolysis
residues (23) would not yield the electrophoretic
profile observed after
and reduction. Similar reasoning applies to connexon-connexon
interactions,
whether between connexons of the same cell or those from the apposing cell. Thus the integrity of gap junction plaques is maintained by bonds of a non-covalent nature. The role of protein-lipid
interactions
in the maintenance of connexon and plaque integrity has not been
established (24). In conclusion, we have shown that RCx43 is internally disulfide bonded between the two extracellular loops. This linkage may play important roles in the self recognition of connexons across the gap and /or may provide stability for the aqueous conduit that spans the extracellular gap. The homology that exists between the various connexins in the extracellular regions suggests that the demonstrated disulfide bond formations for RCx43 are a common feature of the connexins. In agreement with Dupont et al. (15) however, there are no intermolecular
disulfide bonds between RCx43 molecules.
Acknowledmnents We thank Drs Jan H.Hoh, S.Barbara Yancey and Dale W. Laird for their critical review of the manuscript. Initial studies into the disulfide bonds present in RCx43 were carried out in collaboration with Drs S.Barbara Yancey and Dale W. Laird.
References 1. 2. 3. 4.
Flagg-Newton, J., Simpson, I. and Loewenstein, W.R.(1979) Science 205,404-407. Barr, L., Dewey, F.F.and Berger, W.(1965) J. Gen. Physiol. 48,797-823. Beyer, E.C., Paul, D.L.and Goodenough, D.A.(1990) J. Membr. Biol. 116, 187-194. Bennett, M.V.L., Barrio, L.C., Bargiello, T.A., Spray, D.C., Hertzberg, E. and Saez, J.C.(1991) Neuron 6,305-320. 5. Makowski, L., &spar, D.L.D., Philips, W.C.and Goodenough, D.A.( 1977) J. Cell Biol. 74, 629-645. 6. Caspar, D.L.D., D. A. Goodenough, L. Makowski and Phillips, W.C.(1977) J. Cell Biol. 74, 605-628. 7. Unwin, P.N.T.and Zampighi, G.(1980) Nature 283,545-549. 8. Unwin, P.N.T.and Ennis, P.D.(1984) Nature 307, 609-613. 9. Finbow, M.E., J. Shuttleworth, A. E. Hamilton and Pitts, J.D.(1983) EMBO J. 2, 14791486. 10. Leitch, B., and Finbow, M.E.(1990) Exp. Cell Res. 190,218-226. 11. Yancey, S.B., John, S.A., Lal, R., Austin, B.J.and Revel, J.-P.(1989) J. Cell Biol. 108, 2241-2254. 12. Laird, D.W.and Revel, J.-P.(1990) J. Cell Sci. 97, 109-117. 1317
Vol.
178,
No.
3, 1991
BIOCHEMICAL
AND
BIOPHYSICAL
RESEARCH
COMMUNICATIONS
13. Hoh, J.H., John, S.A.and Revel, J.-P.(1991) J. B. C. 266,6524-6531. 14. Manjunath, C., K. and Page, E.(1986) J. Membrane Biol. 90,43-57. 15. DuPont, E., El Aoumari, A., Briand, J.P., Fromaget, Cand Gros, D.(1989) J. Membrane Biol. 108,247-252. 16. Manjunath, C.K., Nicholson, B.J., Teplow, D., Hood, L., Page, E., and Revel, J.-P.( 1987) B&hem. Biophys. Res. Commun. 142,228-234. 17. Lucas-Heron, B., Loirat, M.J., and Ollivier, B.(1987) Comp. B&hem. Physiol. 88B, 421427. 18. Manjunath, C.K., Goings, G.E.and Page, E.(1984) Am. J. Physiol. 246, H865-H875. 19. Goodenough, D.A., and Revel, J.-P.(1971) J.Cell Biol. 50,81-91. 20. Beyer, E.C., Paul, D.L. and Goodenough, D.A.(1987) J. Cell Biol. 105,2621-2629. 21. Karnik, S.S.and Khorana, H.G.(1990) J. Biol. Chem. 265, 17520-17524. 22. Dohlman, H.G., Caron, M.G., DeBlasi, A., Frielle, T.and Lefkowitz, R.J.(1990) B&hem. 29,2335-2342. 23. Kikuchi, Y., Tsuchikura, O., Hirama, M., and Tamiya, N.( 1987) Eur. J. Biochem 164,397402. 24. Malewicz, B., Kumar, V.V., Johnson, R.G.and Baumann, W., J.(1990) Lipids 25,419-427. 25. Zhang, J.-T. and Nicholson, B.J.(1989) J. Cell Biol 109, 3391-3401. 26. Gimlich, R.L., Kumar, N.andGilula, N.B.(1988) J. Cell Biol. 107, 1065-1073. 27. Kumar, N.M.and Gilula, N.B.(1986) J. Cell Biol. 103,767-776. 28. Paul, D.L.(1986) JCell Biol 103, 123-134. 29. Ebihara, L., Beyer, E.C., Swenson, K.I., Paul, D.L.and Goodenough, D.A.(1989) Science 243, 1194-l 195. 30. Beyer, E.C.(1990) J. Biol. Chem. 265, 14439-14443. 31. Lash, J., A.,, Crister,E., S., andPressler, M.,L.(1990) J. Biol. Chem. 265, 13113-13117. 32. Musil, L.S., Beyer, E.C. and Goodenough, D.A.(1990) J. Membrane Biol. 116, 163-175. 33. Fishman, G.I., Spray, DC. and Leinwand, L.A.(1990) J. Cell Biol. 111,589-598. 34. Gimlich, R.L., Kumar, N. and Gilula, N.B.(1990) J. Cell Biol. 110,597~605.
1318
