The topological structure of connexin 26 and its distribution compared to connexin 32 in hepatic gap junctions

Abstract

Of the gap junction proteins characterized to date, Cx26 is unique in that it is usually expressed in conjunction with other members of the family, typically Cx32 (liver [Nicholson et al., Nature 329:732-734, 1987], pancreas, kidney, and stomach [J.-T. Zhang, B.J. Nicholson, J. Cell Biol. 109:3391-3410, 1989]), or Cx43 (leptomeninges [D.C. Spray et al., Brain Res. 568:1-14, 1991] and pineal gland [J.C. Sáez et al., Brain Res. 568:265-275, 1991]). We have used specific antisera both to investigate the distribution of Cx32 and Cx26 in isolated liver gap junctions, and empirically establish the topological model of Cx26 suggested by its sequence and analogy to other connexins. Antipeptide antisera were prepared to four of the five hydrophilic domains which flank the four putative transmembrane spanning regions of Cx26. Antibodies to N-terminal residues 1-17 (alpha Cx26-N), to residues 101-119 in the putative cytoplasmic loop (alpha Cx26-CL), and to C-terminal residues 210-226 (alpha Cx26-C) were all specific for Cx26. An antibody to residues 166-185 between hydrophobic domains 3 and 4 of Cx32 had affinity for both Cx26 and Cx32 (alpha Cx32/26-E2). The antigenic sites Cx26-N, -CL and -C were each demonstrated to be cytoplasmically disposed, although the latter was conformationally hidden prior to partial proteolysis. The antigenic site for alpha Cx32/26-E2 was only accessible after exposure of the extracellular face by separation of the junctional membranes in 8 M urea, pH 12.3. This treatment also served to reveal the region between residues 45 and 66 to Asp-N protease. The topology thus demonstrated for Cx26 is consistent with that deduced for other connexins (i.e., Cx32 and Cx43). Comparison of immunogold decorated gap junctions reacted with antibodies specific to Cx26 (alpha Cx26-N and -CL), or to Cx32 [alpha Cx32-CL], indicates that these connexins do not aggregate in subdomains within a junction, at least within the resolution provided by the labeling density (one antibody per 15-22 connexons). Although the presence of both connexins within a single channel could not be distinguished, possible interactions between channels is discussed.