The constant mutation has made HIV infection difficult to treat, because drugs designed to interrupt the cycle of infection fail when their targets change. To better understand which mutations matter for drug resistance, the researchers compared sequences of HIV taken from patients treated with specific drugs to those from untreated patients. Using a novel statistical method, they identified clusters of mutations that seemed to be working together to help the virus escape treatment.
One drug targets a protein called protease, which the virus needs to assemble the capsule it uses to invade new cells. Substitutions in ten different places on protease occurred in patients who were taking the drug, but what combination of mutations would hinder the action of the drug was not clear before this analysis.
Chemists can determine how a drug fits to a particular protein using computer modelling, but those computations take considerable time. Evaluating all possible combinations of those ten substitutions is impractical. The statistical screen narrowed down the possibilities. "People never looked at this, because they did not know which mutation or which combination of mutations to study," Wei Wang, one of the two lead researchers, said. "That is the advantage of using the statistical method first to find the patterns. After the statisticians discovered the connections between mutations, then we focused on those combinations."
The researchers worked out how the substitutions would change the shape of protease and its affinity for the drug. One set of changes, for example, would tend to dislodge the drug from the pocket where it normally fits. The scientists also determined that the mutations must happen in a particular order for replicants to survive treatment.
Looking back into the database at samples taken from individual patients at several different times during the course of their treatment, the team found that mutations accumulated in the orders that they predicted would be possible during drug treatment. Sequential mutations that their models predicted would leave the virus vulnerable to drug treatment were not observed.
MEDICA.de; Source: University of California - San Diego