NIH's Molecular Libraries Program is giving chemical genomics researchers the tools to develop molecules to probe cellular function. But will all pieces of this herculean effort be enough to bridge the chasm toward translational research?
High-throughput screening (HTS) was once the exclusive domain of large pharmaceutical companies. But with the NIH's Roadmap initiative, the Molecular Libraries Program (MLP), alongside publicly available libraries, today academic and government laboratories have become centers for looking at small molecules as probes of biological function, and are even taking steps toward developing lead molecules for drug discovery. In many ways this shift makes sense, this herculean effort feeds on an open cross collaboration among chemists, biologists, and informatics experts.
Initial work that married synthetic chemistry and biological assays with HTS is beginning to pay dividends, both in innovations that allow researchers to directly test the therapeutic benefit of molecules in physiologically relevant systems and in the development of small molecules that tweak activity within individual cells. “I think it's stunning. The progress of the MLP during this short period of time is remarkable,” says Stuart Schreiber of Harvard University and the Broad Institute.
Although these efforts are showing promise, real challenges lie ahead. With the vastness of chemical space, researchers are trying to increase the diversity of chemical structures available for screening. And as more HTS datasets emerge, further chemistry is needed verify and optimize those hits to produce biological probes or lead compounds for drug discovery.Mining chemical diversity
According to Schreiber, most compounds fail in the clinic not because they didn't engage and modulate a target of interest, but because that process either didn't lead to efficacy or it produced unexpected toxicity. “We're really bad at making predictions,” he acknowledges.
Many of those early predictions came from looking at nucleic acid targets and inferring the effect on proteins, which are the typical interaction partners for small molecules. But that correlation is imperfect. “The main shortcoming today is that we don't have a way to test therapeutic hypotheses in physiologically relevant conditions with small molecules,” Schreiber explains.
One challenge here is developing small molecule libraries that represent biological diversity. The chemical space of small molecules is nearly infinite, exceeding 1060, more molecules than could be synthesized or screened. Still, many biologically relevant classes of molecules are not well-represented in current libraries, especially complex natural products (See sidebar: Library bias and enhancing diversity). As a result, research efforts continue in the attempt to fill the gaps in biologically relevant chemical space. Schreiber's group at the Broad Institute, as well as other groups, are now using one approach called diversity oriented synthesis to construct new libraries, with the NIH supporting a variety of this research through their chemical methodology and libraries development (CMLD) initiative.
Part of the idea is to tap into the creativity and resources of synthetic organic chemists, who previously might not have been involved in biology, says Jeffrey Aubé of the University of Kansas, a principal investigator on one of the CMLD grants. Working initially with several PIs whose groups develop new synthetic methods and synthesize compounds with novel structures, Aubé and his colleagues have the resources to further advance these efforts into compound libraries that can be used for high throughput screening. After verifying that the molecules and structure represent nominally new chemical space, a postdoc or graduate student from the synthetic group might work with his staff to produce a library of 120-140 compounds which are prepared under quality controlled conditions for screening applications. The information is deposited in the public database, the Molecular Libraries Small Molecule Repository (MLSMR).Screening at multiple concentrations
While building library diversity is useful, researchers are also designing screens to elicit maximum information, and using cheminformatics to analyze and make predictions from the resulting data.
Early drug discovery screens typically tested small molecules at a single cutoff concentration as a way to screen a maximum number of compounds quickly and inexpensively. James Inglese of the NIH's Chemical Genomics Center questioned that approach.