Identification of New GSK3β Inhibitors through a Consensus Machine Learning-Based Virtual Screening

Area of Science:

• Medicinal Chemistry
• Computational Drug Discovery
• Neuroscience

Background:

• Glycogen synthase kinase-3 beta (GSK3β) is crucial in metabolic and signaling pathways.
• Aberrant GSK3β activity is implicated in neurodegenerative diseases like Alzheimer's and Parkinson's.
• GSK3β is a significant target for therapeutic intervention.

Purpose of the Study:

• To develop and evaluate machine learning models for identifying novel GSK3β inhibitors.
• To screen a large compound library using a consensus machine learning approach.
• To identify and validate new chemical entities with GSK3β inhibitory potential.

Main Methods:

• Development and assessment of multiple machine learning models.
• Consensus modeling approach for virtual screening of approximately 2 million compounds.
• In vitro assays to determine inhibitory activity of identified compounds.
• Molecular docking and dynamics simulations for binding mode prediction.

Main Results:

• A consensus machine learning model demonstrated high predictive reliability.
• Virtual screening identified two compounds, G1 and G4, with GSK3β inhibitory activity.
• Compound G1 exhibited low-micromolar inhibition, while G4 showed sub-micromolar inhibition.
• Docking and simulations provided insights into the binding interactions of G1 and G4.

Conclusions:

• The machine learning-based virtual screening approach is reliable for identifying drug candidates.
• Compounds G1 and G4 are promising starting points for developing new GSK3β inhibitors.
• Further hit-to-lead and lead optimization studies are warranted for G1 and G4.