Kinetic characterization and cross resistance pattern studies of HIV-1 aspartic protease (PR) inhibitors have shown that some mutations cause considerable reduction in inhibition efficiency. We have performed a computational study of the binding of ABT-538 (ritonavir) with wild type (wt) PR and 12 model mutant structures (R8Q, V32I, M46I, V82A, V82F, V82I, I84V, M46I/V82F, M46I/I84V, V32I/I84V, V82F/I84V and V32I/K45I/F53L/A71V/I84V/L89M (6X)) for which inhibition data are available. Our computational studies indicate a significant correlation between computed complexation energies of ABT-538 with the modeled mutant enzyme structures and the corresponding experimental inhibition constants. By evaluating non-bonding interaction energies between the inhibitor and the mutant enzymes, we have carried out a mechanistic analysis to ascertain the reasons underlying the decrease in binding affinities. This analysis indicated that several residues in addition to the mutated residues contribute to the loss of binding. Taking these considerations into account, a number of new derivatives of ABT-538 were designed, so as to increase van der Waal's and hydrogen bonding interactions with selected mutants. A significant improvement in calculated complexation energies towards both mutant and wt PR structures was obtained for several of the redesigned analogues.
