Gap junction channels are unique. No other channel in vertebrates provides an enclosed conduit for direct diffusional exchange of ions and small molecules between cells, and few other membrane channels have pore diameters large enough to accommodate passage of metabolites and signaling molecules with molecular weights as high as 1000 Da. Moreover, as addressed in two articles in this issue of Circulation Research, gap junctions are formed by proteins with unusually rapid turnover times1 and extremely flexible expression patterns. 2

The connexin proteins that form gap junction channels are encoded by a gene family with at least 14 members in rodents. Each connexin protein has four transmembrane domains, one intracellular and two extracellular loops, and cytoplasmically located carboxyl and amino termini. Six connexin molecules, most likely arranged so that their third transmembrane domains line the channel lumen, comprise the hemichannels or connexons that are contributed by each cell of the coupled pair. Complete gap junction channels, with connexons docked across the gap of extracellular space by interactions of the extracellular loops, are commonly found clustered together, forming islands of particles or pits in freeze-fractured preparations, linearly apposed but slightly separated membranes in thin-section electron micrographs and macular regions of intercellular immunostaining with gap junction antibodies. Remarkably, in studies first performed on rat liver in vivo3, 4 and subsequently in cardiac myocytes and hepatocytes and cell lines in culture, 5–8 the turnover times for connexin molecules have been found to be very short. The article by Beardslee et al1 provides direct evidence for rapid turnover dynamics of connexin43 (Cx43) in cardiac tissue, using [35S] methionine added to Langendorff-perfused rat hearts. In these experiments, the measured decay of radioactivity in immunoprecipitated Cx43 was monoexponential, which was best fit by a half-life of only 1.3 hours. This finding is significant in several respects. First, it extends in vitro observations on cultured myocytes and cell lines, in which intracellular distribution and processing of connexins might not be exactly the same as in vivo, to cardiac tissue and shows that half-life measurements on Cx43 from both preparations are similar. Second, the monoexponential decay of incorporated radioactivity in immunoprecipitated Cx43 implies the absence of a significant pool of longer-lived proteins. Finally, this finding indicates that at every interface between