Improved docking, screening and selectivity prediction for small molecule nuclear receptor modulators using...

Area of Science:

• Biochemistry and Molecular Biology
• Pharmacology and Drug Discovery
• Computational Chemistry

Background:

• Nuclear receptors (NRs) are crucial transcriptional factors regulating key physiological processes.
• Dysregulation of NRs is implicated in diseases including cancer, inflammation, and diabetes.
• Targeting NRs offers therapeutic potential for various human diseases.

Purpose of the Study:

• To develop and validate a comprehensive library of conformational ensembles for thirteen human nuclear receptors.
• To assess the utility of these ensembles in virtual screening for ligand recognition, binding pose prediction, and affinity estimation.
• To evaluate the potential of ensemble models for predicting NR ligand selectivity profiles.

Main Methods:

• Generation of a pocket conformational ensemble library for thirteen human nuclear receptors.
• Virtual screening of known NR modulators using the ensemble library.
• Benchmarking against 157 known NR modulators and 66 experimental structures.
• Calculation of receptor-specific binding energy offsets for selectivity profiling.

Main Results:

• The pocket ensemble library achieved 94% accuracy in predicting ligand binding poses.
• Virtual screening models demonstrated high selectivity for active ligands, with AUCs ranging from 82% to 99%.
• Computationally derived binding energy offsets successfully predicted NR ligand selectivity profiles.

Conclusions:

• Receptor ensembles offer significant advantages for compound docking and virtual screening of nuclear receptors.
• The developed library provides a valuable tool for drug discovery targeting nuclear receptors.
• This approach enhances the prediction of binding affinity, pose, and selectivity, facilitating the development of novel NR modulators.