Developing and applying systems-based approaches to kinase-centered biology

Abstract

Kinases play diverse and critical roles that span nearly the entire field of molecular biology. As the available methods and knowledge regarding kinase biology continues to grow in the high-throughput era of the biosciences, there is a great need for new tools to interpret the resulting data and connect it to known information. Here, I develop and apply three such tools for the study of kinase biology. The first tool, KInhibition, addresses the polypharmacology of small molecule kinase inhibitors (KIs) by connecting researchers to available data from large-scale drug screens. By leveraging these data in combination with a newly-developed metric for quantifying selectivity, I constructed a portal that enables data-driven decisions on the most selective kinase inhibitor to use for a specific application. I further exploit this polypharmacology in my application of Kinome Regularization (KiR) to the study of late stage prostate cancer. I performed a series of small-scale KI screens in a panel of prostate cancer cell lines and applied an updated and improved version of KiR to build predictive models of kinase signaling. These models predicted two compounds, PP121 and SC-1, that effectively inhibited growth of late-stage prostate cancer tumors in multiple in vitro and in vivo model systems. Finally, I extend the scope of KiR to a network-based modeling tool named KiRNet. KiRNet integrates functional hypotheses and additional large-scale, molecular data types with known protein-protein interactions to predict differentially regulated subnetworks within a model system. I demonstrate the potential of KiRNet by identifying known and new regulators of a mesenchymal cell state in hepatocellular carcinoma. All three of these tools bridge important gaps in the field of kinase biology and will accelerate future research progress in this area.