Reconstructing early HIV

The idea is to create an immunogen that will cover the genetic diversity of thousands of strains of circulating virus. These synthetic immunogens, which are genetically closer to the earliest strains of HIV-1, are generating immune responses in guinea pigs and mice. The key is getting the approach to work in humans. The idea could save millions of dollars in human vaccine development, by reducing the costs associated with making and testing separately each strain of a polyvalent vaccine, says Barton Haynes, from the Duke Human Vaccine Institute in Durham, NC, who led the testing of the synthetic vaccines in animals and is now applying for National Institutes of Health funding for a Phase I human trial.

Haynes and others are picking that single gene by comparing genetic sequences from thousands of human HIV-1 strains within group M, the main family of genetically related isolates that is responsible for the AIDS pandemic. With more than 180,000 HIV sequences housed in a database run by the Los Alamos National Laboratory Database, the combinations of approaches are endless. The researchers are coming up with a consensus sequence for the strains, an assembly of the most commonly occurring base pairs at each position. The researchers make a group M consensus gene in silico, synthesize it, inject it into mice, and watch the immune responses.

The synthetic gene spurs a greater number of T-cell responses, and these responses are greater in magnitude and breadth, compared with responses from both single wild-type immunogens and polyvalent immunogens (J Virol, 80:6745-56, 2006). The consensus sequences tested also induced broader levels of neutralizing antibody responses and higher titers of antibodies, compared with wild-type immunogens (Virology, 353:268-82, 2006). "We're still trying to figure out why it is better" at inducing broader antibody responses, Haynes says. Still, the improvement of antibody responses generated by the consensus approach is not ready for clinical utility, he points out.

The approach that Haynes and colleagues are taking happens to yield similar sequences (about 95% similar) as phylogenetic approaches taken to design immunogens (Expert Rev Vaccines, 3:1-8, 2004). A phylogenetic approach to immunogens, taken by Learn and others, has been used to synthesize proteins from group M strains, eliciting T-cell responses in mice (J Virol, published online May 30, 2007).

The results are so similar because of the way HIV-1 has diverged - out from a central point, forming a star-like pattern of various families (clades) of virus, says Bette Korber, a researcher from the Los Alamos National Laboratory. She has traced back the hypothetical common ancestor for the M group to originate between 1911 and 1945 (Science, 288:1789-96, 2000). This makes all the circulating sequences different from each other and less than ideal for vaccine immunogens.

Merck, in collaboration with the NIH, is taking the opposite (polyvalent) approach for vaccine development, though they are still using phylogenetic data to come up with conserved isolates of HIV that are most likely to generate cross-clade immune responses. They are testing these isolates in Phase II human clinical trials. Without the large amount of viral sequencing, "there would certainly be insufficient guidance," Merck's John Shiver says. "We still don't know if it will still be enough, even with the information we have."

Even with all this data, "we're a bit in the dark about exactly what information we need to get to develop a good vaccine," says Richard Harrigan, virologist from the British Columbia Center for Excellence in HIV/AIDS. "That next step of what a successful vaccine will look like, I don't know."