Genomic Approaches for Characterizing How Chemicals Act on Cells

Abstract

The realization of precision medicine demands new therapies to probe an increasingly diverse set of molecular targets. Unfortunately, existing drug development paradigms narrowly focus on single targets and are ill-equipped to efficiently discover collections of compounds that could address the need for individualized treatments. This challenge necessitates the development of new methods for systematically characterizing how chemicals act on cells. The completion of a comprehensive genetic interaction network in Saccharomyces cerevisiae presents a unique opportunity to develop new strategies for discovering new molecular probes. Genetic interactions have been used successfully to characterize the mode-of-action of compounds. However, to date, these methods have not been scaled to enable cost-effective screens of large compound libraries, which limits their practical use in a drug discovery context. This dissertation will describe my development of computational tools for integrating chemical-genetic and genetic interactions in S. cerevisiae to characterize compounds’ mode-of-action. With our experimental collaborators, we scaled this approach to generate and interpret chemical-genetic interaction profiles for more than 10,000 compounds. To facilitate broad use of these data and methods, I built a web portal, called MOSAIC, for visualization and analysis of our chemical genetic profiles. I also describe an application of this chemical genomics technology to discover new therapeutic leads for a rare human disease, called facioscapulohumeral muscular dystrophy (FSHD). I developed a cross-species approach that combines a murine phenotypic assay with yeast chemical-genetic profiling to characterize the mechanism of compounds protective against toxicity associated with this disease and discover new leads that target this mechanism. Finally, my dissertation extends beyond the study of single chemical-genetic perturbations to address the question of how pairwise combinations of genetic perturbations are affected across diverse chemical environments. We screened ~30,000 double mutants across 14 chemical stresses and characterized the extent to which genetic interactions are dependent on the chemical stress. We suggest a strategy for efficient selection of environments and genes that can be applied to study condition-dependent genetic interactions in other systems.