Pairwise GB/SA scoring function for structure-based drug design

An accurate and efficient energy model to compute free energies of ligand binding to proteins is essential to Understanding protein-ligand interactions, particularly to structure-based drug design. A bottleneck to derive such ail energy model is how to account for the solvent effect. A grid-based free energy model was proposed (Zou et al. J. Am. Chem. Soc. 1999, 121, 8033-43),(1) based on the generalized-Born (GB/SA) model of solvation for small molecules (Still et al. J. Am. Chem. Soc. 1990, 112, 6127 -9).(2) Although the grid-based free energy model shows much improved results than the current DOCK force field energy model (Meng et al. J. Comput. Chem. 1992, 13, 505-24)(3) by taking into account the solvent effect, the computational speed for the free energy model restricts the model from a direct drug screen of a huge compound database. Rather, the model is used for a second-step screening after a pre-screening by the force field method. Here. we present a fast algorithm, the pairwise free energy model, for ligand bindin g affinity calculations. Specifically, a pairwise descreening approximation (Hawkins et al. Chem. Phys. Lett. 1995, 246, 122-9)(4) is used in calculations of the electrostatic energy contribution. A procedure is also developed to account for the low-dielectric region that might form between the ligand and the receptor during docking processes. The method has been tested on database screening for dihydrofolate reductase, trypsin, and a fatty acid-binding protein. We obtain similar results compared with the grid-based free energy model but with much less computation efforts. Our pairwise algorithm takes approximately 0.5 s per orientation (with minimization) on a Silicon Graphics Octane R12000 workstation, rapid enough to be used for direct screening of a database or combinatorial library.