Computational tool predicts how drugs work in cells, advancing efforts to design better medicines

Boston University biomedical engineers, chemists collaborate on novel method

The ability to select and develop compounds that act on specific cellular targets has just gained a computational ally — a mathematical algorithm that predicts the precise effects a given compound will have on a cell’s molecular components or chemical processes. Using this tool, drug developers can design compounds that will act on only desired gene and protein targets, eliciting therapeutic responses free of unwanted side effects.

The research, which appears in the March 4 issue of Nature Biotechnology, reports on collaborative work by a team of biomedical engineers and chemists at Boston University. The team was led by Tim Gardner, an assistant professor in the College of Engineering’s Department of Biomedical Engineering (BME) and its Center for BioDynamics, and James Collins, a professor in BME and co-director of the Center for BioDynamics, and done in collaboration with Scott Schaus and Sean Elliott, assistant professors in BU’s Department of Chemistry and Center for Chemical Methodology and Library Development (CMLD).

Although drug development is an active field of research, there have been few ways to predict optimal drug design. The molecular targets of many drug candidates are unknown and are often difficult to tease out from among the thousands of gene products found in a typical organism. This “blindness” in the welter of potential cellular targets means that the process of designing therapeutic drugs is neither precise nor efficient.

The BU research team sought to bring precision and efficiency to this discovery process. The team used a combination of computational and experimental methods to build and verify their tool, first using a reverse-engineering approach to decipher the multitude of regulatory networks operating among genes in a simple organism, then testing the ability of the resulting network models to predict gene and pathway targets for a variety of drug treatments. Finally, they used the tool to predict the molecular targets of a potential new anticancer compound, PTSB, shown in CMLD studies to inhibit growth in the test organism (baker’s yeast) as well as in human small lung carcinoma cells.

Their algorithm predicted, and subsequent experiments verified, that PTSB acted on thioredoxin and thioredoxin reductase, findings that not only validate the tool’s capability but could also pave the way to investigations of a potentially new class of therapeutic compounds.