Identification of novel curcumin derivatives against pancreatic cancer: a comprehensive approach integrating 3D-QSAR pharmacophore modeling, virtual screening, and molecular dynamics simulations.

Pancreatic cancer, known as the "silent killer," poses a daunting challenge in cancer therapy. The dysregulation of the PI3Kα signaling pathway in pancreatic cancer has attracted considerable interest as a promising target for therapeutic intervention. In this regard, the use of curcumin derivatives as inhibitors of PI3Kα has emerged, providing a novel and promising avenue for developing effective treatments for this devastating disease. Computational approaches were employed to explore this potential and investigate 58 curcumin derivatives with cytotoxic activity against the Panc-1 cell line. Our approach involved ligand-based pharmacophore modeling and atom-based 3D-QSAR analysis. The resulting QSAR model derived from the best-fitted pharmacophore hypothesis (AAHRR_1) demonstrated remarkable performance with high correlation coefficients (R 2 ) of 0.990 for the training set and 0.977 for the test set. The cross-validation coefficient (Q 2 ) of 0.971 also validated the model's predictive power. Tropsha's recommended criteria, including the Y-randomization test, were employed to ensure its reliability. Furthermore, an enrichment study was conducted to evaluate the model's performance in identifying active compounds. AAHRR_1 was used to screen a curated PubChem database of curcumin-related compounds. Two molecules (CID156189304 and CID154728220) exhibited promising pharmacokinetic properties and higher docking scores than Alpelisib, warranting further investigation. Extensive molecular dynamics simulations provided crucial insights into the conformational dynamics within the binding site, validating their stability and behavior. These findings contribute to our understanding of the potential therapeutic effectiveness of these compounds as PI3Kα inhibitors in pancreatic cancer.Communicated by Ramaswamy H. Sarma.