Computational discovery of ATP-competitive GSK3β inhibitors using database-driven virtual screening and deep learning

Area of Science:

• Biochemistry
• Medicinal Chemistry
• Computational Drug Discovery

Background:

• Glycogen synthase kinase 3 beta (GSK3β) is a key enzyme in various cellular processes.
• Dysregulation of GSK3β is linked to numerous pathological conditions, making it a significant therapeutic target.

Purpose of the Study:

• To identify novel ATP-competitive inhibitors of GSK3β using a structure-based drug discovery approach.
• To develop and validate a computational framework for inhibitor screening and prediction.

Main Methods:

• Systematic analysis of reported inhibitors to establish filters for candidate curation.
• Structure-based drug design, docking, toxicity assessment (Derek Nexus), and molecular dynamics simulations.
• Development of a deep learning model (GSK3BPred) for inhibitor classification.

Main Results:

• Seven lead compounds were identified with optimal docking scores and drug-likeness.
• Molecular dynamics simulations confirmed the stability of protein-ligand complexes.
• Key binding residues (Lys85, Asp133, Val135) were identified.
• A predictive model, GSK3BPred, was developed and made publicly available.

Conclusions:

• An integrative computational strategy successfully identified potential GSK3β inhibitors.
• The study provides a foundation for experimental validation and optimization of novel GSK3β-targeted therapeutics.
• The developed GSK3BPred model can aid in classifying potential GSK3β inhibitors.