Site-Specific Fragment Identification Guided by Single-Step Free Energy Perturbation Calculations

The in silico Site Identification by Ligand Competitive Saturation (SILCS) approach identifies the binding sites of representative chemical entities on the entire protein surface, information that can be applied for computational fragment-based drug design. In this study, we report an efficient computational protocol that uses sampling of the protein-fragment conformational space obtained from the SILCS simulations and performs single-step free energy perturbation (SSFEP) calculations to identify site-specific favorable chemical modifications of benzene involving substitutions of ring hydrogens with individual non-hydrogen atoms. The SSFEP method is able to capture the experimental trends in relative hydration free energies of benzene analogues and for two data sets of experimental relative binding free energies of congeneric series of ligands of the proteins a-thrombin and P38 MAP kinase. The approach includes a protocol in which data obtained from SILCS simulations of the proteins are first analyzed to identify favorable benzene binding sites, following which an ensemble of benzene protein conformations for that site is obtained. The SSFEP protocol applied to that ensemble results in good reproduction of experimental free energies of the a-thrombin ligands, but not for P38 MAP kinase ligands. Comparison with results from a P38 full-ligand simulation and analysis of conformations reveals the reason for the poor agreement being the connectivity with the remainder of the ligand, a limitation inherent in fragment-based methods. Since the SSFEP approach can identify favorable benzene modifications as well as identify the most favorable fragment conformations, the obtained information can be of value for fragment linking or structure-based optimization.