Computer charts path of diffusing molecules.

Computer simulations have given chemists a sneak preview of how small molecules worm their way into and out of larger ones. That preview, especially when combined with genetic engineering techniques, can guide researchers in attaining proteins for specific uses.

Two computational chemists have modeled carbon monoxide molecules exiting leghemoglobin, a plant protein that resembles human hemoglobin but which acts to trap rather than transport oxygen. Leghemoglobin has mystified researchers for years because the sites where oxygen or carbon monoxide bind to heme proteins lie deeply buried and seem inaccessible.

The simulations rely on new techniques that simplify the task of tracking the movements of the thousands of atoms that make up proteins, says Ron Elber of the University of Illinois at Chicago. He and Illinois colleague Gennady Verkhivker calculate that it takes just a few nanoseconds for two-atom molecules to escape from this protein. That's at least 10 times faster than those molecules can get out of similar proteins such as myoglobin, Elber says.

The molecules leave in two steps, the Illinois chemists and biochemist Quentin H. Gibson of Cornell University reported this week at an American Chemical Society meeting in Washington, D.C. First a carbon monoxide must push aside and move around a six-atom ring belonging to one of the protein's amino acids, a phenylalanine. That ring traps the carbon monoxide in an iron-rich cavity, called the heme pocket.

Then the carbon monoxide waits until the protein's ever-shifting backbone untwists. "[The protein] needs to change its shape very drastically," Elber told SCIENCE NEWS. "Hundreds of atoms move to the side." When the backbone's two helices nearest the heme pocket split apart, the small molecule can drift out.

Elber thinks molecules move into the protein along this same path.

"[The new work] is helping to clarify the mechanism of operation of this class of proteins," comments J. Andrew McCammon of the University of Houston.

In the past, scientists tended to simulate small molecules or much-simplified replicas of larger ones because computers lacked the computational power to keep track of all the atoms in a larger molecule. To complicate matters more, realistic simulations should deal with many molecules, each oriented differently in relation to the molecules with which they interact, Elber says.

"But with the steady increase in computing power and the development of more theoretical methods, people are beginning to look at very realistic biological molecules," says McCammon. "People are beginning to learn how these molecules actually work."

While at Harvard, Elber developed one method for avoiding the computational nightmare created by large molecules. Because the leghemoglobin dwarfs carbon monoxide, Elber's new program models just one leghemoglobin as hundreds of carbon monoxide molecules try to move around and through it.

These molecular pinballs then must maneuver through the protein pinball-machine. "The protein can move in so many different ways that there are going to be many ways that a [molecule] could go in," notes Elber. But only a few low-energy pathways exist, and his technique represents a promising way to determine those routes for carbon monoxide and other molecules, he adds.

"The molecular dynamics simulations offer a way of relating what is seen [in experiments] to the actual structure of the protein," says Gibson. "You can explain what is happening in terms of atoms." He and the Illinois duo have recently sought to harness simulations for designing new proteins.

As part of his research developing blood substitutes and understanding oxygen transport by heme proteins, biochemist John S. Olson at Rice University in Houston uses genetic engineering to make mutant forms of myoglobin. When Gibson used the Illinois model to study the dynamics of one mutant, the simulation indicated that the change would slow the movement of small molecules, which it did. This effort has prompted Elber to refine the model to make it more useful for biochemists seeking to redesign proteins, Olson says.

Another research group has already demonstrated the potential of this synergy between computational chemistry and molecular biology by making an enzyme that scavenges oxygen free radicals better than the natural version of that enzyme.

These enzymes work by attracting the negatively charged free radicals with a positively charged region.

To design the enzymes, researchers led by Elizabeth D. Getzoff, a structural biologist at the Scripps Research Institute in La Jolla, Calif., used computer modeling to examine how slight chnages in the natural enzyme's amino acid sequence might alter the rate at which the free radicals enter the enzyme. The simulations suggested a change that focused the positive force field and guided free radicals more directly to binding sites in the enzyme, the group reports in the July 23 NATURE.