NEW YORK (GenomeWeb News) – A systems biology study is revealing how epistatic interactions between mutations in two HIV-1 enzymes influence the virus' ability to grow alone or in the presence of more than a dozen drug treatments.
The Swiss and American research team, which included investigators from Monogram Biosciences, used fitness assays, sequence information, and mathematical modeling to assess epistatic interactions involving protease and reverse transcriptase enzymes in more than 70,000 HIV-1 samples being tested for drug resistance. The study, which appeared online yesterday in Nature Genetics, highlights the additional HIV fitness information that can be gained by considering epistatic interactions.
"Our approach allows the reconstruction of an approximate fitness landscape of HIV protease and reverse transcriptase and thus offers the first quantitative description of a large, realistic, and biologically relevant fitness landscape," co-corresponding author Sebastian Bonhoeffer, a research with the Swiss Federal Institute of Technology Zurich's Institute of Integrative Biology, and co-authors wrote.
Although there are already more than 20 drugs licensed to treat individuals with HIV, the team explained, resistance still presents hurdles to treating these infections. For instance, they added, past studies have uncovered hundreds of viral mutations that render HIV resistant to drug treatment, though the consequences of these genetic changes — and their effects on viral fitness — varies dramatically.
"The quantitative dissection of the fitness effects of resistance mutations in the presence or absence of drugs and, in particular, the determination [of] how the effect of mutations depends on the presence or absence of other mutations … represents a major challenge," they wrote.
In an effort to better understand how various mutations in HIV influence one another and affect the virus' ability to replicate when exposed to different drug treatments, the team used a strategy called generalized kernel ridge regression, or GKRR, to assess 70,081 HIV samples from individuals infected with the HIV-1 subtype B.
Amino acid sequence information on the HIV protease enzyme and reverse transcriptase was available for each of the isolates, they explained.
The approach allowed them to look at how 1,859 individual amino acid changes in protease and reverse transcriptase proteins influenced the fitness of these HIV samples — specifically, their ability to replicate when untreated or exposed to 15 different HIV drugs.
In contrast to models looking at the fitness consequences of individual mutations alone, the researchers explored the way viral replicative capacity was influenced by both individual amino acid alterations in the proteins as well as pairs of these changes.
By bringing together information on mutations and epistatic interactions, the team came up with a model that could predict more than half — nearly 55 percent — of the variation in HIV replication capacity in the drug and drug-free conditions tested. That represents enhanced predictive power of more than 18 percent compared to models that don't include epistatic information, they noted.
"Our approach provides us with a predictive model for realistic fitness landscapes, opening up new avenues to study evolutionary adaptation on complex fitness landscapes and to simulate the evolution of drug resistance," the team concluded.