Scientists develop sophisticated new simulation method for protein folding

By Joy Ku

Stanford chemists have come up with a new way to study protein folding -- the process by which the proteins in every cell of your body assemble themselves into their functional 3-D shapes. Mistakes in this critical process could be the cause of diseases like Alzheimer's and Parkinson's. Stanford researchers have developed and validated one of the most sophisticated simulations to date to study this process. And with it, they gained new insight into water's role in protein folding.

"It's a new age for protein folding simulation," says Vijay Pande, assistant professor of chemistry and the principal investigator of this research. He emphasizes that this is the first time simulations that model how proteins reach their folded state have included individual water molecules and agreed quantitatively with experimental data.

Their results appeared April 16 on the advanced publication website of the Proceedings of the National Academy of Sciences (PNAS). A grant from the National Science Foundation supported this work, and Google Compute provided some computing resources.

The new approach uses computers to study how individual water molecules affect the protein folding process. "Without water, proteins would not be able to do anything, and so how we handle water [in a computer simulation] is a really important question," Pande explains.

Earlier efforts to simulate the dynamics of the protein folding process only included the general properties of water, like its electric charge. These simpler models could not provide an understanding of what individual water molecules are doing or how water's structure might influence a protein's folded configuration. The new model overcomes these limitations since it simulates each individual water molecule.

Until recently, the enormous computational power required for this type of simulation was not available. It would take a single fast computer processor approximately 300 years to simulate the average protein folding process, Pande estimates.

But a distributed computing system that launched in 2000, called Folding@Home (http://folding.stanford.edu), can run this new type of protein folding simulation in just a month. Folding@Home is one of a new breed of distributed supercomputers, consisting of computer time donated by some 140,000 individuals' computers from around the world.

Stanford graduate students Eric Sorin, Guha Jayachandran and Young Min Rhee, along with postdoctoral fellow Erik Lindahl, developed the software needed to run the new protein folding simulation using Folding@Home.

Rhee, the lead author of the PNAS paper, applied the approach to BBA5, one of the smallest proteins with a well-defined structure. He found that the simulation results agreed with experimental data, in terms of the protein's final 3-D structure and the rate at which it reached its folded configuration.

This new model may be more accurate than previous simpler models, but researchers won't know until more experimental data is available. It will be extremely challenging, though, to characterize the protein folding process more precisely through experiments.

Proteins are thousands of times smaller than the thickness of the human hair, and the fastest proteins fold in about a millionth of a second, Pande points out.

For now, simulations help fill the gaps in the experimental data, says Rhee. In fact, Rhee's simulation revealed a new mechanism in protein folding -- the "concurrent mechanism." This is a process whereby the protein squeezes out water molecules at the same time it collapses on itself. Current theories propose that only one of these actions occurs during protein folding. The new concurrent mechanism offers another explanation and may be the prevalent folding mechanism for smaller proteins, according to Rhee.

Rhee and Pande's groundbreaking work provides a new understanding of how individual water molecules affect protein folding. Perhaps, more importantly, it could signal a shift in the role of protein folding simulations in the biological community. Pande says, "We need a new generation of experimental data to go further, to learn what's going on."

Joy Ku is a postdoctoral researcher in the Pediatric Cardiology and Mechanical Engineering departments at Stanford University.