Membrane insertion of gap junction connexins: polytopic channel forming membrane proteins

However, an inversion of the transmembrane orientation of connexins is not simple to explain, and several observations argue against an inverted membrane topology. First, an inversion of the membrane topology of connexins would suggest that their COOH-terminal domain would also be located in the vesicle lumen, and thus is protected from proteolytic degradation. No such connexin fragments were precipitated in the protease protection assays using the COOH terminus-specific antipeptide antibodies/31S (Fig. 7) , indicating that the COOH terminus was correctly located outside the microsomal vesicles. Second, protease digestions of isolated PM-derived GJ structures indicate that the intracellular loop domain of/31 GJ protein is accessible to proteases, but it seems to be relatively resistant to a complete proteolysis (Milks et al., 1988) . In digests peformed with endoproteinase Lys-C, the intracellular loop was only cleaved on its COOH-terminal side, while the NH2-terminal side resisted degradation. Therefore, it seems likely that the t31 GJ protein fragments that were precipitated in the protease protection assays with the ~lJ antibodies were generated by an incomplete degradation of the intracellular loop, rather than by its protection from proteolysis by the lipid bilayer of the membrane vesicles. Third, preliminary amino acid sequencing results obtained from radiolabeled NH2-terminally processed/31 connexin indicates that the potential signal peptidase cleavage site is located beyond the transmembrane region M1, suggesting that the extracellular loop E1 is correctly located in the lumen of the microsomes (unpublished results). Finally, in this context, it is also interesting to mention a study by Sipos and von Heijne (1993) that shows that the distribution of positively charged amino acid residues in the regions flanking the first transmembrane region of/31 connexin is consistent with its proposed transmembrane topology in the PM. Based on the distribution of positively and negatively charged amino acid residues flanking the transmembrane region, some models have been developed to predict the membrane orientation of bitopic membrane proteins (von Heijne and Gavel, 1988; Hartmann et al., 1989) . Recent data obtained in Escherichia coli suggests that these models could also apply for polytopic membrane proteins (Andersson and von Heijne, 1994; Gafvelin and von Heijne, 1994) to predict their transmembrane topology. Additional experiments and other approaches will be required to determine unequivocally the actual transmembrane orientation of the connexin proteins in the ER membrane. The results of this study demonstrate that the connexins contain a "cryptic" signal peptidase cleavage site that can be processed by this enzyme in association with their membrane insertion. What then prevents this proteolytic cleavage under normal conditions in vivo? One possibility is that an additional factor exists that permits the connexin polypeptides to obtain a proper localization within the membrane bilayer, thus preventing their cleavage. A factor binding to cytoplasmic domains of the connexins is possibly indicated by the results obtained with the proteolytic degradation of connexins. Although potentially accessible to proteolytic degradation, the NH2-terminal domain and the intracellular loop region of cormexins were found to be highly resistant to protease digestion (Zimmer et al., 1987; Milks et al., 1988; this study) . Recently, Musil and Goodenoug