Evaluation and application of MD-PB/SA in structure-based hierarchical virtual screening

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Molecular dynamics (MD) based molecular mechanics Poisson-Boltzmann and surface area (MM-PB/SA) calculations are standard for estimating receptor-ligand binding free energies.
• Existing research primarily assesses MD-PB/SA accuracy and efficiency, with less focus on its role in lead discovery.

Purpose of the Study:

• To critically evaluate the performance of MD-PB/SA in hierarchical virtual screening (HVS) for lead discovery.
• To assess the method's theoretical and practical applicability across different protein targets.

Main Methods:

• Utilized MD-PB/SA calculations to predict relative binding energies for congeneric and diverse ligands.
• Combined physics-based scoring with a knowledge-based structural filter to enhance predictability.
• Validated the improved method in identifying novel inhibitors for p38 MAP kinase, HIV-1 RT, and NA.

Main Results:

• MD-PB/SA accurately predicts relative binding energies when using native ligand poses.
• A limitation was observed in distinguishing native from artificial poses for ligands with significant structural differences.
• The integrated approach successfully identified novel inhibitors for p38 MAP kinase, demonstrating practical utility.

Conclusions:

• MD-PB/SA is a valuable tool for predicting relative binding energies in virtual screening, particularly with native poses.
• Combining MD-PB/SA with structural filters enhances its predictive power and applicability in lead discovery.
• The validated approach shows general validity across multiple protein targets, supporting its use in identifying novel drug candidates.