Rescoring docking hit lists for model cavity sites: predictions and experimental testing.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Molecular docking identifies potential drug candidates but often has low hit rates due to approximations.
• Rescoring with more accurate methods like molecular mechanics-generalized Born surface area (MM-GBSA) can improve prediction accuracy.

Purpose of the Study:

• To evaluate the effectiveness of MM-GBSA rescoring in improving the identification of true binders compared to docking alone.
• To assess MM-GBSA's ability to rescue false negatives and identify false positives in retrospective and prospective studies.

Main Methods:

• Retrospective analysis of docking hit lists for three model binding sites using MM-GBSA with binding-site minimization.
• Prospective experimental testing of 33 molecules that ranked poorly by docking but highly by MM-GBSA.
• X-ray crystallography to determine the binding modes of validated hits.

Main Results:

• MM-GBSA rescoring with minimization significantly improved the discrimination between known ligands and decoys compared to docking alone.
• Experimentally, 23 out of 33 MM-GBSA-selected molecules were confirmed binders, representing rescued docking false negatives.
• X-ray structures revealed that MM-GBSA often improved pose prediction, but sometimes failed to capture protein conformational changes.

Conclusions:

• MM-GBSA rescoring is a valuable strategy for rescuing docking false negatives and enhancing hit rates in drug discovery.
• While MM-GBSA improves accuracy, careful consideration of its limitations, such as handling protein conformational changes, is necessary.
• Further investigation is warranted for applying MM-GBSA rescoring to more complex, biologically relevant targets.