New Compound Library to Speed Drug Discovery

CAMBRIDGE, Mass.  — Making drugs is a difficult and costly business. Even before companies spend exorbitant amounts on clinical trials (most of which fail), they already have spent significant time and money identifying the best drug candidates for those trials. Whitehead Institute scientist Brent Stockwell has developed a possible shortcut for this early drug-development stage, a process that could make it easier and faster for companies and research labs to find the compounds most suitable for drug trials.

Stockwell leads a research team that has assembled a library of approximately 2,000 tiny molecules (or compounds) that have demonstrated some sort of biological activity. What sets this library apart from similar collections is that Stockwell’s group also devised an algorithm that annotates everything each compound can do, incorporating up to 12,000 different functions per molecule. Essentially, says Stockwell, “it’s an annotation strategy that captures all of the information associated with the compound,” which in turn gives scientists greater and more immediate insight into the molecules they study.

In the September issue of the journal Chemistry and Biology, Stockwell and three other associates from his lab published research demonstrating the power of this annotated compound library. Taking a line of engineered cells originally derived from lung cancer tumor samples, Stockwell and his group tested the entire library for each compound’s ability to either prohibit or prevent tumor cell growth. Out of the 2,000 compounds, 85 demonstrated this capability, a surprisingly high number considering that none of these compounds had ever been tested for these types of properties. Stockwell then queried and cross referenced the annotated information for each compound and found common properties shared by most of these 85 compounds.

“Through this assay, we discovered 12 new mechanisms that could be of interest in cancer research,” Stockwell says.

For typical compound libraries that lack an annotation system, discovering these mechanisms would have taken years. Stockwell’s group located them in a little over a week.

Thinking Big

Imagine a library of Alexandria for biologically active compounds, and this collection’s potential becomes clear. Although 2,000 compounds is only a fraction of all biologically active molecules (which in turn is only a subset of all compounds), Stockwell’s platform can be expanded indefinitely. Says David Root, project scientist in Stockwell’s lab and co-author of the paper, “Our annotation system automatically culls the literature. We can continually update the system. It’s flexible. As more information is published, the profile of that compound will be adjusted.”

The current library of 2,000 compounds includes about 90 percent of all controlled substances and 50 percent of all drugs approved for use in the United States.

“My guess,” says Stockwell, “is that there are probably 5,000 to 10,000 commercially available compounds that we could, in principle, access.” To find them, researchers would have to comb through published research about individual compounds and then decide which ones can be purchased and which need to be synthesized.

All information in the library is available on the Stockwell lab Web site. Anyone can log on and access all annotated information for each compound, although the actual physical compounds are not publicly available. And Stockwell continues to collaborate with other labs interested in using his library for experiments similar to the cancer cell research.

Gathering the Fallen Angels

Stockwell uses as his model other similar publicly available databases for biologists. For example, the National Institutes of Health houses both a library of DNA sequences and one of microarray data. Jill Heemskerk, program director at NIH’s Neurological Disorders and Stroke Division, is overseeing a project to develop an annotated library similar to Stockwell’s.

“It’s a different starting philosophy from Brent’s,” says Heemskerk. “Our focus is to obtain compounds that had made it partway through a clinical trial and then were dropped for one reason or another.”

Such compounds, often referred to as “fallen angels” by the pharmaceutical industry, are usually remaindered in industry collections and ultimately neglected if they show no promise for the short list of diseases the company is interested in. If, however, Heemskerk can obtain them and incorporate them into her library, any laboratory could discover a new application for the molecule which may yield a fiscal benefit for the company. “Basically,” says Heemskerk, “they would have all of academia doing free testing for them, finding new ways to use these old drugs.”

While the NIH library is still in its earliest planning stages, Heemskerk observes that the value of Stockwell’s collection “is something that’s widely recognized in industry. Brent is forging new territory in academia.”

And although this new territory could one day include a completely comprehensive library of biologically active compounds, Stockwell’s sights are more focused. “I don’t envision that in the end the ‘Brent Stockwell lab’ must be the one that maintains the annotated compound library. This is a public resource and tool for the community. And for now, my main goal is to use it to do interesting biology.”

This research was supported by the National Cancer Institute, the Burroughs Wellcome Fund and Whitehead Institute for Biomedical Research.

Citation

Root, D. E., Flaherty, S. P., Kelley, B. P., & Stockwell, B. R. (2003). Biological mechanism profiling using an annotated compound library. Chemistry & biology, 10(9), 881-892.