Experimental and theoretical analyses of the domain architecture of mammalian protein disulphide-isomerase

The high resolution structure of full-length protein disulphide-isomerase (PDI) has not been determined, but the polypeptide is generally assumed to comprise a series of consecutive domains. Models of its domain organisation have been proposed on the basis of various sequence-based criteria and, more recently, from structural studies on recombinant fragments corresponding to putative domains. We here describe direct studies of the domain architecture of full-length mammalian PDI based on limited proteolysis of the native enzyme. The results are consistent with an emerging model based on the existence of 4 consecutive domains each with the thioredoxin fold. The model was further tested by expressing recombinant fragments corresponding to alternative domain models and to truncated domains; the observed properties of these purified fragments supported the 4-domain model. A multiple alignment of many PDI-like sequences was generated to test whether domain boundaries could be predicted from any features of the alignment, such as sequence variability or hydrophilicity; neither of these parameters reliably predicted the domain boundaries determined by experiment.