Abstract

Structure-based methods use 3D information of the receptor as a mold to design molecules that, by placing complementary chemical groups at the right locations, can become efficient ligands. Pre-calculated potentials are of great help in ligand design because they inform about which moieties are more adequate at each spot. However, complex aspects of molecular recognition, such as protein flexibility and the effect of aqueous solvation degrade the quality of the predictions. This is particularly true for polar contacts, as they can have completely different behavior depending on the local environment and the existence of distinct hydration patterns. I will present MDmix (Molecular Dynamics simulations with mixed solvents) as a method to obtain atom-specific interaction free energy maps, taking explicit account of aqueous solvation and protein flexibility. MDmix not only reproduces experimentally-derived pharmacophores but also informs about hydration patterns and provides a direct evidence of water displaceability. Hence, the method tackles two problems that are usually dealt with separately, providing more accurate interaction maps than ordinary interaction potentials (e.g. GRID ) and more quantitative water displacement measures than inhomogeneous fluid solvation theory (WaterMap). This tool is being used in my group to exploit novel binding sites that give access to non-standard molecular mechanisms of action, such as alteration of protein-protein complexes, pharmacological chaperoning or allosterism.

http://pubs.acs.org/doi/abs/10.1021/jm5010418