A structural model for the retroviral proteases.

In many retroviruses the 5' end of the pol gene codes for a protease vital for the processing of the gag polyprotein into the separate core proteins. In some viruses this protease is encoded at the 3' end of the gag gene, or between the gag and pol genes in a different reading frame to either. A sequence, Asp-Thr-Gly, which is conserved in retroviral proteases is also conserved in the active sites of aspartic proteases, an observation which has led to the suggestion that the retroviral proteases could belong to this family. We have examined the sequences of the aspartic and retroviral protease families, using pattern-recognition, structure prediction and molecular modelling techniques, and conclude that the viral protease sequences probably correspond to a single domain of an aspartic protease and may function in a dimeric form. We have constructed a model of the pol-protease of human immunodeficiency virus 1 (HIV-1) to test this hypothesis.