Ligand binding mode prediction by docking: mdm2/mdmx inhibitors as a case study.

Area of Science:

• Medicinal Chemistry
• Computational Biology
• Drug Discovery

Background:

• Mdm2 and Mdmx are key negative regulators of the tumor suppressor p53.
• Their p53-binding domains are validated targets for cancer therapeutics.
• Standard docking procedures struggle to accurately predict binding poses for certain inhibitors, notably nutlins.

Purpose of the Study:

• To evaluate alternative virtual screening pipelines for predicting binding poses of Mdm2/Mdmx inhibitors.
• To identify computational strategies that improve docking accuracy, especially for challenging compounds like nutlins.

Main Methods:

• Tested various conformation generation programs (LigPrep, ConfGen, MacroModel, Corina/Rotate) and docking programs (GOLD, Glide, MOE-dock, AutoDock Vina).
• Assessed the ability of different pipelines to reproduce known binding poses for nutlins and other Mdm2/Mdmx inhibitors.
• Investigated computationally intensive conformational exploration and "anchoring" procedures.

Main Results:

• Most standard docking pipelines failed to accurately predict nutlin binding poses.
• Successful prediction of nutlin poses required either extensive conformational searching or an "anchoring" approach.
• Other Mdm2/Mdmx inhibitors were generally docked successfully with standard methods.

Conclusions:

• Standard virtual screening pipelines are insufficient for accurately docking nutlin-class inhibitors of Mdm2/Mdmx.
• Specialized computational approaches, such as conformational anchoring, are necessary for successful drug design targeting these proteins.
• Improved docking accuracy is crucial for developing novel cancer therapeutics targeting the p53 pathway.