We have developed a procedure for the replacement of defined domains with specified domains from other proteins that we used to examine the molecular basis for the differences in voltage-dependent gating between connexins 26 (Cx26) and 32 (Cx32). This technique does not depend on sequence homology between the domains to be exchanged or the presence of restriction endonuclease sites. Rather, it makes use of a PCR strategy to create an adhesive "band-aid" that directs the annealing of the amplified sequence to the correct location in the recipient clone. With this technique we created a series of chimeras involving the replacement of topologically defined protein domains of Cx32 with the corresponding sequences of Cx26. We focused on domains that are predicted to line the gap junction channel as we expect that a component of the voltage-sensing mechanism resides there. Differences between Cx26 and Cx32 in the sequences of their first and second extracellular loops, the cytoplasmic loop, and the third transmembrane domain did not account for the difference in their calculated gating charges. Shifts along the voltage axis in the steady-state conductance-voltage relations of the chimeric connexins were produced by replacement of the first extracellular loop or the cytoplasmic loop and the amino-terminal half of the third transmembrane domain. These data suggest that the voltage-sensing mechanism arises from the interaction of domains lining the aqueous channel and domains deeper in the channel wall.