Contents list

Abstract

To understand the determinants of membrane protein interactions, we have developed an in vivo genetic assay system for detecting homodimerization of transmembrane (TM) segments from integral membrane proteins. Our approach is to generate gene fusions between potentially dimerizing TM segments and a cytoplasmic DNA-binding protein that lacks its intrinsic dimerization domain. This genetic approach allows us to screen and distinguish among known dimerizing domains and weakly dimerizing mutants, as well as non-dimerizing TM segments. We replaced the bacteriophage λ cI repressor C-terminal dimerization domain with the human erythrocyte glycophorin A transmembrane segment (GpA TM). GpA TM forms SDS-resistant homodimers in vitro. Expression of this membrane-associated fusion in Escherichia coli conferred the same degree of immunity to λcI phages as the wild-type, intact λ repressor. Single amino acid substitutions that disrupt the GpA TM dimer interface were introduced into the λ-GpA TM fusion proteins. These mutations dramatically reduced immunity of E. coli to λcI, such that the efficiency of plating these phages increased by greater than 10,000-fold over that conferred by the wild-type λ-GpA TM fusion. Introduction of the putatively non-dimerizing first TM from E. coli MalF into the λ-TM fusion vector resulted in no immunity to λcI phages. Fusion of the homodimeric, periplasmically localized, mature alkaline phosphatase domain to the C terminus of the λ-TM fusion proteins containing weakly to non-dimerizing TM segments restored immunity to λcI phages. Results from this in vivo genetic assay system demonstrate that (1) dimerization of the λ cI DNA-binding domain can be promoted by dimerizing TM segments, (2) strongly, weakly, and non-dimerizing TM segments can be distinguished on the basis of their ability to confer immunity to λcI phages, and (3) introduction of a dimerizing periplasmic domain can provide functionality to λ-TM fusions containing weakly to non-dimerizing TM segments.