Design of novel hydrazide–hydrazone derivatives targeting MGC803 gastric cancer cell line by integrating 3D-QSAR, ADMET, network pharmacology, docking, MD simulations and biological efficacy

Abstract The MGC803 cell line is a human gastric cancer model frequently used in cancer research. In this context, a combined in silico approach including 3D-QSAR modeling, ADMET analysis, network pharmacology, docking, molecular dynamics and ligand transport evaluations, was applied to design new antiproliferative molecules. A robust 3D-QSAR model with high predictive capacity ( R² and Q² ) was developed and used to design new compounds (PR1–PR4). After an ADMET screening, the putative biological targets of the non-toxic compounds were predicted using PharmMapper. A network pharmacology analysis identified several hub genes, of which HSP90AA1 had the highest degree value. Given its central role in stabilizing multiple oncogenic proteins involved in gastric cancer progression, as well as its suitability for structure-based studies, HSP90AA1 was selected for molecular docking and molecular dynamics simulations. In addition, molecular docking was performed on HSP90AA1 protein (1YET) in complex with the designed molecules (PR1-PR4), and their predicted binding behaviors were compared to both the most active molecule (M34) and the reference drug, geldanamycin. These results demonstrate a high predicted binding affinity and remarkable interaction profiles within the active site of the HSP90AA1. To further evaluate the dynamic stability of these complexes, we performed molecular dynamics simulations over a 100 ns period, thus confirming stable attachment modes and durable contact networks. The MM-PBSA approach demonstrated favorable binding free energies between the chosen PR4 ligand and the 1YET protein (-26.47 ± 2.89 kcal/mol). Finally, the ligand transport study showed that the PR4 ligand easily crosses tunnels 1 and 2 with optimal theoretical transport dynamics compared to the reference drug, geldanamycin (GA). This comprehensive computational method underlines the diverse potential of the examined molecules, identifying the most promising candidates for subsequent experimental validation against gastric cancer.