(continued)

FEMTOSECONDS TO MINUTES
J. Andrew McCammon, an HHMI investigator at the University of California, San Diego, is pushing mathematics to the extreme in his research on the infinitesimal and varied motions inside protein molecules—movements that can provide important clues about how proteins work and interact with other molecules, such as drug compounds. McCammon’s group uses supercomputers to model, in extraordinary detail, the quivers, jiggles, twists, and bends that proteins undergo, as governed by the chemical and physical forces of each of the molecule’s thousands of atoms. The time scales of these various motions range widely, moreover, from femtoseconds (quadrillionths of a second) to minutes.

Modeling protein movements at that level of detail takes a lot of computing power, says McCammon. “Just to get to the microsecond (millionth of a second) time scale in simulating a medium-sized protein requires on the order of a billion small steps in time, and each step might involve as many as a million septe calculations of forces between the pairs of atoms in the protein.”

His team recently simulated 50 nanoseconds (billionths of a second) of movements in a nerve-cell receptor as it binds to a neurotransmitter molecule. “The simulation model includes the receptor, a portion of the lipid bilayer it passes through, and water molecules on both sides of the bilayer, for a total of about 150,000 atoms,” McCammon says. “To simulate the dynamics for about 50 nanoseconds, where we can begin to see the response of the receptor, requires about 500 processors on the DataStar supercomputer at the San Diego Supercomputer Center, running each processor for 200 hours.”

Such intensely focused modeling may seem like much too much about much too little, but it can yield real-world payoffs—for instance, a new generation of drugs for treating HIV/AIDS. While inhibitors of one of HIV’s enzymes, called protease, have been effective anti-AIDS drugs, in recent years protease inhibitor-resistant HIV strains have emerged. Another of HIV’s enzymes, integrase, allows the retrovirus to stitch its genetic material into the genome of the human host. Although a static picture of the integrase structure was known, the enzyme hadn’t been successfully investigated as a drug target when McCammon’s lab started scrutinizing it a few years ago.

In modeling two nanoseconds’ worth of integrase’s movements, the team discovered that part of the protein chain moves in such a way that, for an instant, a “trench” opens up near the active site—the part of the protein that catalyzes the stitching reaction. McCammon and his colleagues then designed compounds that they predicted would fit into the trench, thereby jamming the enzyme so that it could no longer work. Scientists at Merck Research Laboratories have since confirmed this hypothesis by testing a number of the compounds as potential drugs to inhibit the HIV integrase, and clinical trials of one of them are expected to begin later this year.

—Paul Muhlrad

HHMI INVESTIGATOR

HHMI INVESTIGATOR