Historically, the most successful strategy for the discovery of novel cancer therapeutics has been the use of phenotypic screens in cell- or organism-based assays. A major advantage of phenotypic screens is that they can target any enzyme, receptor or other macromolecule in its biological context, without a priori knowledge of a target, thus allowing discovery of compounds that act on catalytic domains of enzymes, but also those that act through allosteric mechanisms or interrupt protein-protein interactions. Although the rate of discovery of interesting new small molecules is high using phenotypic approaches, the difficulty of target identification and verification is a bottleneck that is difficult to overcome.
Working with the lab of Michael White, Ph.D., we have developed a method for the broad-scale classification of the biological activity of natural products in human cells, by employing an information-rich, high throughput, endogenous reporter gene expression platform that allows quantitative discrimination of concordant cellular responses to genetic and chemical perturbations. In a proof-of-concept, gene expression-driven functional signatures were employed as cross-platform phenotypic discriminators to link concordant cellular responses to 1124 genetic (siRNA, miRNA) and 1186 natural product perturbations, providing Functional Signature of Ontology (FUSION) maps, allowing us to predict the mechanism of action of natural products.