Can the calculation of ligand binding free energies be improved with continuum solvent electrostatics and an ideal-gas entropy correction?

The prediction of a ligand binding constant requires generating three-dimensional structures of the complex concerned and reliably scoring these structures. Here, the scoring problem is investigated by examining benzamidine-like inhibitors of trypsin, a system for which errors in the structures are small. Precise and consistent binding free energies for the inhibitors are determined experimentally for this test system. To examine possible improvement of scoring methods, we test the suitability of continuum electrostatics to account for solvation effects and use an ideal-gas entropy correction to account for the changes in the degrees of freedom of the ligand. The small observed root-mean-square deviation of 0.55 kcal/mol of the calculated relative to the experimental values indicates that the essentials of the binding process have been captured. Even though all six ligands make the same salt bridge and H-bonds to the protein, the electrostatic contribution varies among the ligands by as much as 2 kcal/mol, Moreover, although the ligands are rigid and similar in size, the entropic terms also significantly affect the relative binding affinities (by up to 2.7 kcal/mol). The present approach to solvation and entropy may allow the ranking of the ligands to be considerably improved at a cost that makes the method applicable to the optimization of lead compounds or to the screening of small collections of ligands.