Researchers at Stony Brook Universitys Center for Structural Biology wanted to understand how an essential HIV protein switched between two known conformations.
They used computer simulation to model the transition and identified a new conformation that helps explain HIVs vulnerability to a class of drugs known as protease inhibitors.
The research at the university, based in Stony Brook, N.Y., could eventually lead to more effective forms of AIDS drugs.
Small molecule drugs, typically taken as pills, tend to work by gumming up cellular machinery, usually proteins. This happens because both proteins and drug molecules have specific shapes. Such drugs are identified by finding molecules that fit into crevices or cavities in the protein.
Drug researchers often make crystals of proteins to examine the shape of these crevices and design drug molecules that fit the protein more snugly.
But in the case of HIV protease, crystal structures were little help. In the crystallized forms, the cavities and crevices in the protein were very small, too small to let drug molecules in.
Stony Brooks Carlos Simmerling used a supercomputer to model how HIV protease switched between two shapes, and saw a new conformation where the cavities open wider.
The work reveals how drugs like Kaletra and Viracept fit into the protein and stop it from working.
An "open" conformation of HIV protease was expected, but had not previously been described in detail.
Computer modelling is routinely used in drug design, but certain simulations are often not attempted. Simmerling said that access to Silicon Graphics Altix through the NCSA (National Center for Supercomputing Applications) made him go ahead with the project.
"I dont know that I would have tried this a couple of years ago," said Simmerling in a statement. "The resources were out there, but they were too precious."
Simmerlings team designed a software application to model the protease movements. The team typically used 64 processors of NCSAs SGI Altix 3700 Bx2 system for each of the simulations, leaving the remainder of the 1,024-processor resource available for other projects.
He said the HIV protease simulations took 20,000 CPU hours on NCSAs Altix system, or about three months in real time. He estimated that it would have taken more than a year to complete the work using his labs own Linux cluster.
"It would have been at least six to seven times slower than the Altix, and the cluster doesnt scale as well," he said. "I wouldnt have done it for such a risky project."
Still, Simmerling says that for a quickly mutating virus, drug designers need to consider more than just the shape of the protein.
"If we target its shape with a drug, then any change in shape can diminish the drugs effectiveness. But if we can target its job, the shape doesnt matter as much and it will have a harder time evading the drugs."