Docking and 3-D QSAR studies on the binding of tetrahydropyrimid-2-one HIV-1 protease inhibitors

• Surflex-dock poses show greater observed propensity of hydrogen bonding with NH of Ile 150 compared to Glide poses.
• The docking scores of Surflex-dock poses correlate better with experimental binding affinities than Glide docked poses.
• Pose prediction accuracy is improved when urea core constraints are imposed during the docking of the THP ligands.
• 3-D QSAR CoMFA and CoMSIA models developed using Surflex-SIM (Sybyl-X) and Phase_Shape (Maestro) alignments have similar predictive powers.

We present molecular docking and 3-D QSAR studies on a series of tetrahydropyrimid-2-one HIV-1 protease inhibitors whose binding affinities to the enzyme span nearly 6 orders of magnitude. The docking investigations have been carried out with Surflex (GEOM, GEOMX) and Glide (SP and XP) methodologies available through Tripos and Schrodinger suite of tools in the context of Sybyl-X and Maestro interfaces, respectively. The alignments for 3-D QSAR studies were obtained by using the automated Surflex-SIM methodology in Sybyl-X and the analyses were performed using the CoMFA and CoMSIA methods. Additionally, the top-ranked poses obtained from various docking protocols were also employed to generate CoMFA and CoMSIA models to evaluate the qualitative consistency of the docked models with experimental data. Our studies demonstrate that while there are a number of common features in the docked models obtained from Surflex-dock and Glide methodologies, the former sets of models are generally better correlated with deduced experimental binding modes based on the X-ray structures of known HIV-1 protease complexes with cyclic ureas. The urea moiety common to all the ligands are much more tightly aligned in Surflex docked structures than in the models obtained from Glide SP and XP dockings. The 3-D QSAR models are qualitatively and quantitatively similar to those previously reported, suggesting the utility of automatically generated alignments from Surflex-SIM methodology.