All functions of a protein involve its physical interactions, however transient, with specific other molecules (ligands), large or small. Protein structure and its dynamics, determined by sequence, in turn determine which molecules are able to bind to the protein. Druggable proteins are defined as those whose function can be modulated with a small molecule. The contents of this dissertation focus on leveraging the knowledge about protein structure and changes in protein structure to identify small molecule modulators of protein function, thus expanding the druggable proteome. The studied proteins include members of the kinase and SLC transporter superfamilies, both important drug targets. First, we separately examine the impact of mutations on the structures of two kinases, FLT3 and BCR-ABL, and the resistance of these mutants to drugs. We provide a structure-based rationale for the cause of resistance and offer treatment alternatives. Second, we address the modulation of function of human organic cation transporters (OCTs), either through phosphorylation or structure-guided screens to discover novel small molecule inhibitors of these transporters. We establish that by combining docking and in vitro high-throughput screens, competitive and non-competitive ligands of OCTs can be predicted accurately. Third, we examine the quaternary structure of the human concentrative nucleoside transporters (CNTs) to gain new insight into their functions and use structure-guided screens to discover novel ligands to modulate them. We show that human concentrative nucleoside transporter 3 forms homo-oligomers, thus encouraging efforts on finding allosteric inhibitors. Finally, we present a large-scale study of the impact of cancer mutations on protein structure, with the hope of expanding the druggable proteome through the discovery of mutant-specific binding pockets that would allow for selective, functional inhibition or activation.

Membrane proteins serve as cellular gatekeepers, regulators, and sensors. Some examples of large membrane protein families with prominent roles in human health include ATP-binding cassette transporters (ABC) that protect cells from external toxins, G-protein coupled receptors that respond to extracellular signals, and voltage-sensing ion channels. Despite their importance for normal cellular function, less than 1% of the structures in the Protein Data Bank (PDB) are high-resolution structures of membrane-bound proteins, providing only sparse data for structure-based functional annotation. In the absence of structure, mechanistic insight into membrane protein function is lacking. In contrast, sequence databases are growing rapidly and include large numbers of uncharacterized genetic variants found in the human genome. Here, I use computational methods to improve functional annotation of point mutations in human membrane proteins, focusing on a large and diverse group of membrane transporters, the ABC transporter superfamily.

Models of protein structure at atomic resolution are fundamental to many problems in biology. Increasingly, we are interested in computing models of protein structure based on noisy, sparse, and heterogeneous data; demanding a rigorous approach to modeling. Bayesian inference is one such approach. In Chapter 1, we formalize modeling as a search for all models consistent with the input information. In Chapter 2, we develop a computational framework for computing models from heterogeneous biophysical software. In Chapter 3, we develop a modeling method to compute a model of mTORC2 in complex with a native substrate, Akt1. In Chapter 4, we develop a modeling method to compute multi-state models from X-ray crystallography.

Proteomics techniques have been used to generate comprehensive lists of protein interactions in a number of species. However, relatively little is known about how these interactions result in functional multiprotein complexes. This dissertation describes a computational procedure that bridges this gap by combining low-resolution data from affinity purification experiments with data from a heterogenous set of standard structural biology techniques, including electron microscopy (EM) and X-ray crystallography. One such assembly is the Nuclear Pore Complex (NPC), which serves as the sole mediator of nucleocytoplasmic exchange in eukaryotic cells. We used our method to determine the structure of the ~600 kDa heptameric Nup84 complex, an essential component of the NPC. This work demonstrates that integrative approaches based on low-resolution data can generate functionally informative structures at intermediate resolution.