Database Digs Up New Meds

Can an online genetic database turn drug discovery from a game of chance into a science?

By Dave Mosher|Tuesday, October 24, 2006

Using an online search engine, scientists can now discover new medical treatments based on existing drugs. The Connectivity Map, created by researchers at MIT and Harvard University's Broad Institute, is a free public database filled with information about how drugs turn human genes on or off—important activity considering that just one misbehaving gene can cause a life-threatening disease.

To function normally, the human body produces a delicate balance of around 100,000 different proteins—"building-block" molecules that do everything from dismantling poisonous chemicals to creating energy from food. Controlling protein building is a gene's primary task, so the activation or deactivation of genes plays a crucial role in keeping the body alive. If genes turn on or off at the wrong times, they can create the wrong amounts of proteins and thereby cause disease. Now Justin Lamb, lead architect of the Connectivity Map, is creating a master list of drugs that can knock malfunctioning genes back into line.

Building the Connectivity Map proved no small feat. Lamb and his team extracted sets of the body's 22,000 active genes from prostate, breast, blood, and skin cells, drawing from a wide variety of tissues because genes activate differently within different cells. The group applied 164 drugs to the genes one by one, measured how effectively each drug turned them on or off, and then stored the information in the database.

Although it represents a large collection of information, the Connectivity Map doesn't reveal new connections on its own. Scientists looking for new treatments need to pick genes they're curious about—such as those linked to diseases like Alzheimer's or cancer—and indicate what the genes do wrong in people who have the disease. Scientists call this information a gene expression signature. When a disease's signature is entered on the Web site, the Connectivity Map generates a ranked list of drugs that may help effect a healthy gene signature.

The database is still relatively small, but early results are promising. For years, doctors have been frustrated by a nasty form of leukemia that resists treatment by the drug dexamethasone. But when some of Lamb's colleagues at the Broad Institute plugged in a signature linked to that particular leukemia, they discovered that sirolimus—a drug used to prevent organ rejection—could reverse the resistance. Researchers also found that gedunin, a cedar-tree extract normally used to fight malaria, could also fight prostate cancer.

"We occasionally trip upon new uses for old drugs, but the Connectivity Map is a systematic, logical way of discovering new uses for them," says Todd Golub, director of the Broad Institute's cancer program. While the database is a powerful drug discovery tool, says Chris Austin, director of the NIH's Chemical Genomics Center, Golub and Lamb aren't the first to do it. "Every large pharmaceutical company out there has been doing similar kinds of things for years," Austin says—including GeneLogic and Merck—"but they don't make their data public." Because Golub and Lamb's Connectivity Map is public, medical researchers can better and more quickly find disease treatments.

Lamb hopes to circumvent the "extremely tortuous process" of turning hypothetical drugs into products on pharmacy shelves by adding all FDA-approved drugs to the database. "Only a fraction of drugs are ever FDA-approved, so it's a truly rarefied collection of drugs," Lamb says. "If we get enough support, we could add them all in under a year." Austin believes testing drugs on different cells will be just as important for the project's future because drugs can have "entirely different" effects on different cells. To make the Connectivity Map as effective as possible, he says, "you need to test the whole kahuna."

Comment on this article