by Bernice Wuethrich
April 16, 2004

Unlike Sumo wrestlers—who can bind their opponents tightly using a number of moves and positions—proteins rely on the precise shape of their binding sites to attach to target molecules. Only through such specificity, for instance, can an antibody capture a virus. The ability of proteins to bind other molecules underlies innumerable life processes and medical treatments. For example, protein binding allows cell-to-cell communication, gives hormones their punch, and delivers chemotherapy to cancer cells.

Scientists study protein binding with the goal of finding ways to control it. An important step in this direction comes from computational biologist Homme Hellinga, Ph.D., of the Duke University Medical Center. His approach involves designing proteins that bind to new targets, which has potential applications across vast areas of medical science, toxic waste clean-up, and drug development.

Hellinga started with a known bacterial protein that binds to nutrients. He chose to modify this protein to bind molecules it would never encounter in nature, including the explosive TNT and the brain chemical serotonin. Hellinga used a cluster of 20 linked computers to explore all of the ways atoms could be arranged in the protein's binding site—an astounding 10⁷⁶ possibilities (that's more than a quadrillion multiplied by itself five times). With a sophisticated computer algorithm, he pared down these virtually countless possibilities to 17 promising arrangements that could be tested directly. Hellinga then constructed these 17 altered proteins in the laboratory. To determine the ability of the synthetic proteins to attach to their new targets, he engineered the proteins to glow when binding took place. The experiments lit up the lab. When inserted back into living bacteria, the designer proteins continued to carry out their new functions, taking the research closer to real-world applications.

The TNT-grabbing protein could serve as a biosensor to detect land mines or undetonated underwater explosives. Similar redesigned proteins could sniff out pollutants or chemical warfare agents. In addition, the ability to bind serotonin suggests possible diagnostic uses, since conditions such as depression and anxiety cause fluctuations in serotonin levels in the brain. The newly designed proteins could prove a boon to the pharmaceutical industry as well, since they can differentiate between mirror-image forms of molecular compounds. One form may be biologically active, while the other is inactive or even harmful. The ability to distinguish one from the other could lead to safer drugs in less time and at lower cost.