Elucidation of Artemisinin as a Potent GSK3β Inhibitor for Neurodegenerative Disorders via Machine Learning-Driven QSAR and Virtual Screening of Natural Compounds.

Research overview

The text below is the source article’s original abstract under a verified reuse license. cc by

Background/Objectives: Glycogen synthase kinase-3 beta (GSK3β) is a key enzyme involved in neurodegenerative diseases such as Alzheimer's and Parkinson's, contributing to tau hyperphosphorylation, amyloid-beta (Aβ) aggregation, and neuronal dysfunction. Methods: This study applied a machine learning-driven virtual screening approach to identify potent natural inhibitors of GSK3β. A dataset of 3092 natural compounds was analyzed using Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN), with feature selection focusing on key molecular descriptors, including lipophilicity (ALogP: -0.5 to 5.0), hydrogen bond acceptors (0-10), and McGowan volume (0.5-2.5). RF outperformed SVM and KNN, achieving the highest test accuracy (83.6%), specificity (87%), and lowest RMSE (0.3214). Results: Virtual screening using AutoDock Vina and molecular dynamics simulations (100 ns, GROMACS 2022) identified artemisinin as the top GSK3β inhibitor, with a binding affinity of -8.6 kcal/mol, interacting with key residues ASP200, CYS199, and LEU188. Dihydroartemisinin exhibited a binding affinity of -8.3 kcal/mol, reinforcing its neuroprotective potential. Pharmacokinetic predictions confirmed favorable drug-likeness (TPSA: 26.3-70.67 Å2) and non-toxicity. Conclusions: While these findings highlight artemisinin-based inhibitors as promising candidates, experimental validation and structural optimization are needed for clinical application. This study demonstrates the effectiveness of machine learning and computational screening in accelerating neurodegenerative drug discovery.

Related topics