Toward a Global View of the Conformational Landscape of the Human Kinome

Abstract

The human genome contains about 500 protein kinases, which play a central role in the regulation of the majority of cellular pathways. Mutations in kinase genes - often resulting in dysregulation of their phosphotransferase activity - are a frequent cause of disease, including many types of cancer. Kinases are especially flexible proteins, and undergo significant conformational changes during their catalytic and regulatory cycles. This conformational heterogeneity is also of fundamental importance in determining the binding affinity and selectivity of inhibitory ligands, their susceptibility to resistance mutations, and the discovery of putative allosteric binding sites. Markov state models have recently emerged as a practical computational approach to the enumeration of protein conformational states, and can be constructed by aggregating the data from multiple, independent, short molecular dynamics trajectories in a statistical fashion. We aim to apply this technique to the entire human kinome, simulating each protein kinase catalytic domain using a range of high performance compute resources, including the distributed simulation framework, Folding@Home. In combination with recent developments in GPU-accelerated simulation algorithms, this approach allows us to obtain aggregate trajectory lengths on the order of milliseconds. An automated software pipeline provides the ability to quickly generate multiple starting configurations for each kinase, while a central database of publicly available kinase data has been set up and used for tasks such as the selection of catalytic domain sequences and the assignment of relative priorities to each kinase. In parallel with our computational approach, we are working towards expressing a diverse range of kinases in bacterial systems, and scaling up a fluorescence-based assay to plate format for direct measurement of kinase inhibitor binding affinities. Our poster will present preliminary results from these efforts.