Perth team tracks HIV's 'escape mutants'

The 'H' in human immunodeficiency virus (HIV) could stand just as well for Hydra, the multi-headed monster of Greek myth. By the time the human immune system has decapitated one strain of the virus, it has spawned a host of novel mutants. Mutants proliferate because natural selection endowed the virus with a dodgy reverse transcriptase enzyme that introduces random errors into the genetic blueprints of newly replicated virus particle.

But the survival and persistence of these mutants in a patient's body is anything but random, according to Prof Simon Mallal, executive director of Centre for Clinical Immunology and BioMedical Statistics at Royal Perth Hospital.

The centre is a collaborative venture between the hospital and Murdoch University. Murdoch statistician Prof Ian James is co-director.

In a paper published in Science in May, 2002, Mallal and his colleagues described their discovery of strong statistical associations between a patient's human leucocyte antigen (HLA) genes and specific epitopes of HIV proteins in a cohort of 472 AIDS patients in Perth.

The Science paper has had a major influence on research into HIV vaccine development, with research groups now adopting the approach pioneered by Mallal and his colleagues. In March, the Bill and Melinda Gates Foundation awarded the Perth research team a US$9.8 million grant to continue its work.

Mallal said the Perth team's original study related the enormous diversity of the virus to HLA type -- all 472 patients had different HLA types, and all carried different HIV strains. But within this great diversity of strains, they found that certain epitopes of the HIV proteins gag, pol and env remained essentially constant in the presence of specific alleles of HLA genes.

These 'escape mutants' had clearly found a way of evading the immune system's antibody and cytotoxic T-cell defences, where many other HIV strains had not.

The clear implication was that the HLA system constrains the virus' evolution: between them, the antibody and CTL responses winnow the mutant strains, eliminating those with easily recognised epitopes, and leaving a few escapees that, in Mallal's words, have accidentally "cracked the code".

Mallal said this information turns out to be critical for designing HIV vaccines. A person's combination of HLA alleles effectively forms a genetic 'barcode' that predicts the particular combination of epitopes required in a vaccine to protect that individual against locally extant HIV strains.

What happens at the cellular level is that that HLA molecules on the surface of B- and T-cells display peptide sequences cleaved from HIV proteins. Some epitopes successfully activate the immune system to eliminate virus particles -- or HIV-infected lymphocytes -- featuring those epitopes.

But epitopes that do not induce effective antibody or CTL responses allow the originating virus to persist and replicate, as if it were invisible to the immune system.

Mallal said the virus pays a cost in fitness -- the fact that the HLA genes impose functional constraints on the virtual proteins means that the 'escapee' viruses may be less virulent. There is some epidemiological evidence for this effect -- certain alleles, such as HLA B27 and HLA B57, are associated with longer median survival times in HIV/AIDS patients.

A vaccine featuring a variety of these epitopes should prime the immune system to recognise and 'remember' HIV strains that it would otherwise overlook, and mount a protective response should the person ever be exposed to those strains.

Mallal and his colleagues need two very large datasets to find associations between HLA alleles and viral epitopes. They have to sequence the entire HLA complex of genes from hundreds of individuals, then associate them with specific epitopes of the HIV's notoriously variable gag, pol and env genes.

Finding links demands substantial computing power, and advanced search algorithms. The research partners have spun out a new company, Epipop, to commercialise the software they have developed, also called Epipop.

Mallal said Microsoft has been helping to develop Epipop, by providing its expertise in anti-spam technology to help devise powerful learning algorithms.

The Perth researchers are using Epipop to identify HIV epitopes that could be used in a vaccine suited to the predominant HLA alleles in Perth's population.

The bad news is that individual variation in HLA alleles, combined with the virus's own high variability, means that 'generic' HIV vaccines are unlikely to be successful. But the good news, Mallal said, is that at a regional or population level, the presence of protective alleles at high frequency in a local gene pool means that vaccinating at-risk groups with custom designed vaccines could help prevent AIDS, and limit the virus' spread.

Such vaccines could also delay the emergence of virus strains resistant to front-line anti-viral drugs. "If you're trying to vaccinate 1000 people, each with a different HLA system, it becomes a probabilistic exercise," Mallal said.

"We can determine the probability that a vaccine containing particular antigens will protect a greater or lesser percentage of the population."

The Perth group is facilitating this regional approach to vaccine development by making Epipop available to other research agencies in South Africa, the US, Europe, Asia and the Caribbean.

"We're trying to facilitate, not compete," Mallal said. "We don't yet know which horses are going to win, but we're providing a firmer footing for all the runners."