Network toxicology-driven repurposing of metformin for hepatocellular carcinoma: insights from molecular docking and dynamics simulations.

contextHepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality with limited therapeutic options and poor prognosis, particularly in advanced stages. This study integrates computational and experimental approaches to evaluate metformin as a candidate for repurposing in HCC. Key oncogenic targets EGFR, MAPK3, MMP9, and PRKACA were prioritized via protein-protein interaction network analysis. Molecular docking predicted favorable metformin binding poses for EGFR (ΔG = -7.42 kcal/mol), MAPK3 (ΔG = -7.45 kcal/mol), MMP9 (ΔG = -7.84 kcal/mol), and PRKACA (ΔG = -8.33 kcal/mol). Using AmberTools/Antechamber-derived GAFF2 parameterization and AM1-BCC partial charges for ligand topology, repeated 100 ns molecular dynamics simulations showed ligand retention with target-specific receptor and pocket RMSD adaptation and stable radius-of-gyration profiles. Network toxicology predicted low hepatotoxicity risk using ProTox-II and ADMETlab 2.0 endpoints. ITC measurements produced detectable metformin-protein binding responses and apparent dissociation constants under the reverse-orientation assay conditions (20 µM protein in the cell and 200 µM metformin in the syringe). Western blot analysis showed reduced total EGFR, MAPK3, and MMP9 expression after metformin treatment, while PRKACA changed only slightly. Triplicate MTT assays showed concentration-dependent inhibition of HepG2 cell viability, with an IC50 of 16.02 µM.

methodsMetformin and HCC-associated targets were collected from SwissTargetPrediction, DrugBank, GeneCards, DisGeNET, BindingDB, and OMIM. PPI networks were constructed using STRING (confidence >  = 0.7, Homo sapiens) and visualized with Cytoscape v3.10.1; core targets were identified via CytoHubba. Hepatotoxicity and ADMET risk were evaluated using ProTox-II and ADMETlab 2.0. Molecular docking was performed using MOE 2022 with Amber99 force field and MOPAC7.0, generating 30 poses per target. MD simulations (100 ns) were conducted in GROMACS using AMBER99SB-ILDN protein topologies and metformin parameters generated with AmberTools/Antechamber, GAFF2, AM1-BCC partial charges, parmchk2, and ACPYPE, followed by NVT/NPT equilibration and production runs at 298 K/1 bar with particle mesh Ewald electrostatics. ITC analysis was performed using MicroCal VP-ITC at 25 °C in PBS (pH 7.4), using 20 µM protein in the cell and 200 µM metformin in the syringe. Western blot (HepG2, 50 µM metformin, 24 h) and MTT assay (0.01-500 µM, 48 h) were performed in triplicate as in vitro validation.