Although protein dynamics plays an essential role in function, it is rarely considered explicitly in current structure-based approaches to drug design. Here I propose the computer-aided design of ligands by modulation of protein dynamics, or equivalently, protein structural ensembles. The detailed understanding of ligand-induced perturbations of protein dynamics that will result from this study is crucial not just to accurately predicting binding affinities and tackling "undruggable" targets, but also to understanding protein allostery.
Three major aims will be pursued during this project.
First, I will combine concepts from chemoinformatics and non-equilibrium thermodynamics to detect cryptic "druggable" small molecule binding sites in computed structural ensembles. New computational methods will be developed to predict how binding at these putative sites is likely to influence protein function. This will enable rational approaches to allosteric control of protein function.
Second, new classes of non-equilibrium sampling algorithms will be developed to improve by 2-3 orders of magnitude the speed of computation of protein/ligand structural ensembles by molecular simulations. This will enable routine consideration of protein flexibility in ligand optimisation problems.
Third, I will address with the above methods a frontier problem in molecular recognition: the rational design of protein isoform-specific ligands. To achieve this goal, I will integrate computation with experiments and focus efforts on the therapeutically relevant cyclophilin protein family. Experimental work will involve the use of purchased or custom-synthesized competitive and allosteric ligands in enzymatic assays, calorimetry and crystal structure analyses.
Overall, this project proposes fundamental advances in our ability to quantify and engineer protein-ligand interactions, therefore expanding opportunities for the development of future small molecule therapeutics.