Virtual screening using a conformationally flexible target protein: models for ligand binding to p38α MAPK

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• p38α MAPK is a key target in inflammatory diseases.
• Developing effective inhibitors requires understanding binding site dynamics.
• Existing virtual screening methods face challenges with protein conformational flexibility.

Purpose of the Study:

• To develop robust virtual screening models for aryl substituted heterocycles targeting p38α MAPK.
• To evaluate the performance of single protein structures versus receptor ensembles.
• To assess the utility of this approach for other protein targets.

Main Methods:

• Virtual screening using a library of known p38α MAPK inhibitors and decoy compounds.
• Docking simulations on multiple p38α MAPK crystal structures, considering various protonation states and tautomers.
• Evaluation of docking models using Receiver Operating Characteristic (ROC) curves and enrichment factors.
• Generation and assessment of receptor ensembles.

Main Results:

• The single crystal structures 1BL7 and 2EWA showed good performance (enrichment factors 14.1 and 13.0).
• Receptor ensembles generally improved performance, with high ROC AUCs and good enrichment.
• The 1BL7-2EWA ensemble outperformed individual structures, achieving an ROC AUC of 0.94 and high enrichment factors.

Conclusions:

• Ensemble docking models can significantly enhance virtual screening accuracy for p38α MAPK inhibitors.
• This strategy is effective in addressing protein conformational diversity.
• The developed methodology holds potential for application to other drug discovery targets.