TUCSON, ARIZONA--If drug companies are good at one thing, it's making tons of new molecules. In recent years, they've embraced a high-speed drug-producing technique known as combinatorial chemistry to churn out novel compounds by the millions. The problem is that few ever work against diseases. But at a meeting on the future of combinatorial chemistry here yesterday, a chemist proposed a new way to boost the odds: pair the most promising building blocks to produce compounds whose activity is far greater than the sum of their parts.
One popular technique in combinatorial chemistry is to start with a handful of molecular building blocks and then link them in every possible combination. As you might expect, the number of compounds quickly becomes unwieldy. Using all the currently available building blocks, researchers could produce a mind-boggling 1060 small, druglike molecule combinations.
To make things more manageable, Jonathan Ellman of the University of California, Berkeley, and his colleagues focused their efforts on the components with the best chance of success. They turned to a class of molecular building blocks called oximes that are common components of pharmaceuticals. The researchers screened a library of just over 300 oximes for their ability to bind to a cell-signaling protein called Src, which is thought to be overly active in cancer and osteoporosis. When they tested the oximes alone, about 10% showed a moderate ability to inhibit Src in the test tube. However, when these mediocre binders were linked, the story changed dramatically. Ellman's team synthesized about 4000 oxime pairs and found a handful of powerful Src inhibitors, a hit rate that typically requires many more compounds.
The new approach is powerful, because it has the potential to produce drugs capable of binding to more than one part of a target molecule, says Sheila DeWitt, a combinatorial chemist at Arqule, a company in Woburn, Massachusetts. That could lead to a new way to make ultrapotent drugs. "If it works, it will have tremendous value," she says.