PROJECT SUMMARY Quantitatively predicting drug-resistant mutations to improve precision oncology My work builds towards a mechanistically informed approach to model and predict drug-resistant kinase mutations that will enhance patient treatment regimens. Protein kinases are important signaling enzymes often dysregulated in cancer; their pharmacological value as drug targets exemplified by the clinical use of over 75 FDA-approved inhibitors. Unfortunately, multiple clinically observed kinase mutant resist inhibitors and drastically reduce patient survival rates. Precision oncology approaches, matching specific tumor profiles to optimally therapies, have proven useful thanks to tumor sequencing and mutation profiling. However, it remains challenging to identify drug-resistant mutants prior to treatment and develop regimens to circumvent them. A lack of mechanistic information describing clinically observed kinase mutants makes it difficult to predict whether a mutation will resist canonically used kinase inhibitors. Kinase mutations may decrease drug-binding affinity, increase kinase activity, tune inhibitor sensitivity profiles, or any combination of these mechanisms. Structure-based methods promise to help predict the impact of kinase mutations. I hypothesize that a kinase inhibitor's utility against drug-resistant mutants is expressed using physical, quantitative properties like structural state populations and binding affinities. My work quantitatively assesses the impact of clinical kinase mutations on inhibitor resistance, sensitivity, and susceptibility. Specifically, I will develop models that predict whether clinical kinase mutations perturb inhibitor-binding, increase kinase activity by stabilizing active configurations, or sensitize kinases to alternative inhibitors. In this proposal, I draw upon clinical mutation databases to study mutation-inhibitor pairs of c-Met kinase, the target in Non-Small-Cell lung cancers (NSCLCs), building upon previous studies of resistance mutations in Abl kinase. As a mentee (K99), I will use binding free-energy calculations to predict how clinical mutations reduce c-Met inhibitor affinity (Aim 1). As I transition to independence (K99/R00), I will use molecular simulations to biophysically evaluate whether clinical mutations increase kinase activity by shifting kinase populations to catalytically active conformations (Aim 2). Upon independence (R00), I will study whether clinical mutations sensitize kinases to rarely used alternative inhibitors (Aim 3). These computationally intensive calculations can often take years to collect sufficient data on a normal computer. Instead, I will run calculations on the planetary-scale Folding@home distributed computing platform in collaboration with high-throughput biophysical experiments that measure kinase activity and inhibitor binding affinity. Overall, my proposal, and future lab, will build towards a precision-oncology platform that helps clinicians plan treatment regimens.

PROJECT NARRATIVE My long-term goal is to build quantitative models that predict and circumvent cancer therapy resistance, specifically focusing on protein kinases currently targeted by over 75 FDA-approved inhibitors. Unfortunately, multiple kinase mutations have been found to resist known inhibitors and drastically reduce patient survival rates, and it remains unclear why these mutations induce such dramatic drug-resistance. Through a combination of high-throughput experiments, physics-based simulation, and machine learning based data-science methods, I will identify the mechanistic impact of clinically observed kinase mutants and build models that predict if a drug- resistant mutant is newly sensitive to alternative clinical therapeutics.