Characterization of protein-ligand interaction sites using experimental and computational methods.

The ability to identify the sites of a protein that can bind with high affinity to small, drug-like compounds has been an important goal in drug design. Accurate prediction of druggable sites and the identification of small compounds binding in those sites have provided the input for fragment-based combinatorial approaches that allow for a more thorough exploration of the chemical space, and that have the potential to yield molecules that are more lead-like than those found using traditional high-throughput screening. Current progress in experimental and computational methods for identifying and characterizing druggable ligand binding sites on protein targets is reviewed herein, including a discussion of successful nuclear magnetic resonance, X-ray crystallography and tethering technologies. Classical geometric and energy-based computational methods are also discussed, with particular focus on two powerful technologies, that is, computational solvent mapping and grand canonical Monte Carlo simulations (as used by Locus Pharmaceuticals Inc). Both methods can be used to reliably identify druggable sites on proteins and to facilitate the design of novel, low-nanomolar-affinity ligands.