MODULATING PROTEIN FUNCTION WITH SMALL MOLECULES THROUGH COMPUTATIONAL AND EXPERIMENTAL DESIGN TECHNIQUES

Abstract

The ability to modulate protein function using exogenous small molecules is a longstanding goal in chemical biology. Selective activation or inhibition of a particular protein function can help elucidate crucial molecular mechanisms and enables important advances in cell biology. Small-molecule controlled molecular systems also possess tremendous value in bioengineering and biomedical applications: activation of protein function allows the construction of protein switches and biosensor proteins, whereas inhibition of protein function contributes to the development of novel therapeutic agents. The discovery of small-molecule modulators of function is greatly aided by computational modeling methodologies. By utilizing structural information obtained through X-ray crystallography or NMR spectroscopy, these tools allow efficient and affordable examination of large small-molecule databases and provide quantitative evaluation of the likelihood that a given protein-ligand interaction occurs. Advances in computer algorithms and hardware development continue to accelerate and scale up the computation and lower the cost of this discovery process. The primary focus of this thesis is the development of structure-based computer-aided methodologies for designing small-molecule modulators of protein function. To this end I explored two parallel paths, one to study activation and one to study inhibition of protein functions. Taken together, my work aims to not only apply rational design strategies to specific proteins, but also demonstrate their general applicability. The first project, focused on activation of protein function, is built on an approach developed by our laboratory that designs a de novo allosteric binding site directly into the catalytic domain of an enzyme. This approach achieves modulation of function by a novel "chemical rescue of structure approach": a tryptophan-to-glycine mutation disrupts local structure and induces conformational changes that distort the geometry at the active site; the subsequent binding of exogenous indole then reverts this conformational change and restores the native enzyme structure. The main challenge of generalizing this approach, however, is the difficulty of rationally designing analogous conformational changes in other proteins. It is therefore important to study the possible mechanisms that can be utilized by chemical rescue of structure. Through collaborative and multidisciplinary efforts, we find that the switchable proteins built via the chemical rescue of structure are frequently controlled indirectly by modulating protein stability, rather than discrete conformational changes. Since energetic evaluation of protein stability is far more tractable than designing and/or predicting allosteric conformational changes, this finding demonstrates how chemical rescue of structure can be applied to other systems for building a variety of new protein switches. To further generalize the applicability of chemical rescue of structure, I sought to extend it to include multiple amino acids, rather than just one. I chose ChxR, a homodimeric response regulator in Chlamydia, as the model protein to examine the feasibility of this strategy. I mutated a pair of tryptophans at the dimer interface to glycine in order to disrupt the dimerization of ChxR. To enable the subsequent functional rescue, I used the removed structural elements as a template for ligand-based virtual screening and discovered a set of candidate small molecules that mimic the three-dimensional geometry and chemical properties of the removed chemical moieties. Biophysical characterization of these compounds suggests that the majority of them selectively bind to the engineered ChxR variant. This observation shows promises in extending this generalized design strategy to allow alternate activating ligands. In parallel to these efforts I carried out studies aimed at inhibition of protein function, as exemplified by my project that uses small molecules to disrupt a protein-RNA interaction. Conventional methods of inhibitor design mostly target RNA-processing enzymes and cannot be generalized to the majority of RNA-binding proteins (RBPs). I contributed to the development of a general strategy of designing competitive inhibitors targeting RBPs. This method involves identifying "hotspot pharmacophores" from the protein-RNA interaction and using it as a template in ligand-based virtual screening. To evaluate the performance of this approach, my collaborators and I applied it to Musashi-1 (Msi1), a protein that upregulates Notch and Wnt signaling pathway and promotes cell cycle progression. Our "hotspot mimicry" approach led us to discover compounds that match the hotspot pharmacophore, and thus enabled the development of novel inhibitors to the Msi1/RNA interaction that we validated in both biochemical and cell-based assays. This approach extends the "hotspot" paradigm from protein-protein complexes to protein-RNA complexes, and helps establish the "druggability" of RNA-binding interfaces. It is the first example of a rationally-designed competitive inhibitor for a non-enzymatic RNA-binding protein. Owing to the simplicity and generality, I anticipate that the hotspot mimicry approach may lead to the identification of inhibitors of other protein-RNA complexes, which in future may serve as starting points for the development of a novel class of therapeutic agents.