Have you ever contemplated matching the motor of a BMW with the gears of a Mercedes to make a better racing car? Researchers at the California Institute of Technology have developed an algorithm that helps them do exactly this with proteins that share a similar structure. The algorithm predicts which parts of a protein can be swapped — or recombined — without disturbing the overall shape of the proteins, making the hybrids more likely to be functional.
The algorithm, described online this month in Nature Structural Biology, calculates the interactions between the amino acid residues within a protein and determines which ones are disrupted when fragments, or schemas, are swapped. Fewer disruptions result in a more stable hybrid. “The remarkable aspect of our program is its simplicity,” said Christopher Voigt, a forth-year graduate student at Caltech and the lead author of the paper. Moreover, experimentally determined activities of hybrid proteins based on his calculations, he found, correlated well with the predictions.
Current methods for recombining fragments between proteins to engineer their function are purely experimental and restricted to proteins with high sequence identity. “Now we believe we can make combinatorial libraries based on this algorithm, so we could take three or more related proteins and shuffle them together, and screen that library for new functions,” said Frances Arnold, professor of chemical engineering and biochemistry at Caltech and the senior author of the paper. “They have to be similar in structure, but not in sequence.”
The authors are going to make the algorithm available to academic researchers on the web, probably by this fall. Several companies are interested in licensing the patent-pending technology from Caltech.
One of them is Xencor, a Monrovia, Calif.-based company focusing on developing protein therapeutics and improving industrial enzymes using structure-based computational approaches. Bassil Dahiyat, the company’s CEO, sees two main advantages in the new method. First, by allowing researchers to start with proteins more distantly related in sequence, it increases the diversity of the hybrids. Secondly, by helping choose the “right” fragments in advance, it lowers costs by reducing the number of hybrids to be tested experimentally. “You can look at enormous novelty with a lot less experimental effort,” he said.