In silico pipeline for GSK 3β inhibitor discovery in Alzheimer’s disease using pharmacophore screening, docking, ADME filtering, and MD validation

Mahmoud S. Elkotamy, Mohamed K. Elgohary, Ahmed S. Alkotami, Mohamed M. Eldesouki, Zainab M. Elsayed, Amr A. Mattar, Mahmoud F. Abo-Ashour, Haytham O. Tawfik, Wagdy M. Eldehna, Hatem A. Abdel-Aziz

Abstract Glycogen synthase kinase-3β (GSK-3β) is a key therapeutic target for Alzheimer’s disease, but identifying safe, brain-penetrant inhibitors remains difficult. This study aimed to discover novel CNS-active GSK-3β inhibitors using a rigorous multi-tier computational pipeline. The workflow combined ligand-based and structure-based pharmacophore modeling, virtual screening of the ZINCPharmer database, AutoDock Vina docking, ADME and blood-brain barrier (BBB) filtering with SwissADME, toxicity prediction using ProTox-3.0, and validation by 100-ns molecular dynamics simulations with MM/GBSA and MM/PBSA free energy calculations. Pharmacophore screening with a ≤ 1.0 Å RMSD cutoff identified 1,085 ligand-based and 36 structure-based hits. After docking and developability filtering, two BBB-permeant candidates were prioritized: SB1 , a structure-based hit (predicted LD 50 = 2500 mg/kg, toxicity class 5), and LB1 , a ligand-based hit (predicted LD 50 = 521 mg/kg, toxicity class 4). Molecular dynamics confirmed stable binding for both compounds. MM/GBSA analysis showed favorable binding free energies for SB1 (-27.68 kcal/mol) and LB1 (-25.74 kcal/mol), both surpassing the co-crystallized reference (-8.75 kcal/mol). These findings identify SB1 and LB1 as promising, safe, and brain-penetrant GSK-3β lead compounds for experimental validation in Alzheimer’s disease.