A model reveals proteins bold fold.

To live, every organism looks to its genes for guidance. From its DNA, it receives instructions that tell it how to make proteins.

Those proteins--big, crumpled molecules--serve as an organism's biological workhorses. Starting out as long sequences of amino acids, proteins fold into unique shapes that enable them to perform their very specific tasks.

But exactly how they fold has remained mysterious. What folding rules do proteins follow to transform a certain sequence into a specific shape? Could one learn those rules and use them to predict how a protein actually folds?

Addressing this issue, chemist Peter G. Wolynes of the University of Illinois at Urbana-Champaign, physicist Jose N. Onuchic of the University of California, San Diego, and their colleagues describe a method to explain some of the intricacies of protein folding.

To do this, they merge a theoretical model with experimental data to show how an unfolded protein starts out with many possible shapes and, through a series of steps, narrows down its options to obtain a final natural structure.

"Folded proteins are marvels of molecular engineering," the scientists say in the March 17 Science. "A protein navigates with remarkable ease. . .as it explores many possible physical configurations. This feat is beginning to be quantitatively understood by means of statistical mechanics and simplified computer models."

An elaboration of their work will appear in the April 11 Proceedings of the National Academy of Sciences.

Before folding, a protein faces so many possible shapes--about 1060 options for a protein with 60 amino acids--that to evaluate every one is impossible. "An unguided search, like a drunk playing golf, would take practically forever," the scientists say.

So they looked at the problem from a different angle, asking how the molecule settles into its most stable shape.

In stabilizing itself, the protein moves through an energy "funnel" (diagram), which guides it through many potential shapes to the one that minimizes its energy, Wolynes says. While folding, the molecule runs the risk of getting stuck, so to speak, in a relatively stable (but not the most stable) shape. To prevent that, the team finds, the funnel sides must be fairly steep, helping the molecule to get through bottlenecks.

To hone the model, the scientists introduced laboratory data from real folded proteins into a computer program that represents unfolded molecules as "necklaces of beads." After cross-checking the model against natural proteins, they expect that under the right circumstances, a bead model will effectively show how and why a protein collapses into its natural structure.

"Obviously, a real protein isn't as simple as these little beads in the model," Wolynes says. "Real proteins have special structural features, like hydrogen bonds and side chains. But when you include all of those details in the model, it becomes too complicated computationally."

"Even though this model has relatively few parameters, we could still see a correspondence between what happens in the model and what happens in real proteins," he says. "The model allowed us to dissect the folding process."

Previous attempts to make such models focused more on qualitative than quantitative aspects of folding, he says.