Gerald Learn¹, Allen Rodrigo², Fusheng Li¹, Matthew Rain¹ and James I. Mullins¹

1: Department of Microbiology, University of Washington, Seattle, WA 98195-7740 and 2: University of Auckland, School of Biological Sciences, Auckland, New Zealand

A primary concern in designing an AIDS vaccine to protect against infection by HIV-1 is which strains should be chosen to best provide protection against the myriad of evolving variants. Evolutionary trees of HIV-1 sequences sampled from infected individuals typically form a star-burst pattern with most of the variants roughly equidistant from the center of the tree and also approximately equidistant from any other circulating strain. To designate single variants that would be more closely related to any of the diverse circulating viruses we propose to choose ancestral viruses representative of the center of the evolutionary radiation; these would appear to have given rise to the vast range of viral diversity within an infected population. It is very unlikely that we could ever identify this ancestral virus even from stored specimens near the start regional AIDS epidemics. However, computational methods can be used to infer an ancestral viral sequence. -- We have recently constructed an ancestral sequence for the envelope gene (env) of HIV-1 subtype B which is on average more closely related to any given circulating virus than any other variant; nearly all known targets of the cellular immune response (99.2%) are represented within this ancestral sequence. Most recently, we have extended this procedure to include the Subtype E env gene sequence common in HIV-1 from Thailand and other countries in the epidemic in Southeast Asia.