As an aid to structure-based drug design, we propose to develop an empirical model for calculating the free energy of binding for protein- ligand association in aqueous solution. The proposed research will utilize experimental data for both the development and parameterization of the proposed empirical model. Three types of experimental data will be exploited for this purpose: (1) three dimensional structures of protein-ligand complexes determined at high resolution by x-ray crystallography to provide the geometries of interactions; (2) microcalorimetry to obtain thermodynamic binding data; and (3) small molecule hydration free energy data to assess the influence of solvent on binding. In addition, input from molecular dynamics computer simulations will be used to augment experimental data where necessary and to provide a detailed atomic level description of processes not directly accessible by experiment. The proposed model will have two distinct components. The first is a generalized hydration shell model for computing solvent effects on the binding process. This will be developed with the help of computer simulations of small molecule solutes in water and experimental solvation free energy data. The second is a model for protein-ligand interaction energies, and will include hydrogen bond energy, which is dependent on both the atom types involved as well as their relative geometry, and dispersion energy, which is radially symmetric distance dependent term. Empirical energy parameters for the hydrogen bond and dispersion energies will be determined by combining data from calorimetric binding free energy measurements, three dimensional x-ray crystal structures, and solvation free energy measurements. The models for both solvation free energy and protein-ligand interaction energy will be refined in an iterative fashion. The models, once refined, will be used prospectively to prioritize potential lead compounds for synthesis and to screen experimental and artificial three-dimensional small molecule databases for leads.
