Structure based prediction of binding affinity of human immunodeficiency virus-1 protease inhibitors.

A series of computations were performed to derive a strategy for the prediction of binding affinities of non-peptidic human immunodeficiency virus-1 (HIV-1) protease inhibitors. This paper describes the development of a 3D quantitative structure-activity relationship (3D-QSAR) methodology by using receptor information of HIV-1 protease. The docking and molecular dynamics simulations were performed on a model ligand/enzyme complex to optimize the variables involved in the generation of ligand/enzyme models. The protonation scheme of the active site aspartic acid residues of HIV-1 protease was derived from a computational study. The active site aspartate is monoprotonated with a proton placed on the OD1 atom of the ASP B25. This protocol of docking and molecular dynamics (MD) simulation was then used to derive the ligand-enzyme complexes of the molecules used in the present study. The molecular mechanics interaction descriptors were calculated from these ligand/enzyme models. A partial least squares (PLS) method was used to derive a linear correlation between the interaction descriptors and the biological activity. A good correlation was observed when the change in the energy of the ligand upon complex formation and the electrostatic contributions to the solvation energy of the ligand were included in the QSAR analysis. A highest cross-validated q2 value of 0.649 was observed. This model had a conventional r2 of 0.725, and when this model was used to predict the activity of the external test set, it produced a r2pred of 0.761. The total interaction energy was partitioned into interactions in different subsites and interactions with each of the amino acid residues of the enzyme. The PLS analysis using these descriptors helped to identify the important interactions which can be exploited for the design of HIV-1 protease inhibitors.