Structure-guided discovery of ubiquitin-specific protease 7 inhibitors through integrative quantitative structure-activity relationship modeling, docking, and molecular dynamics.

Ubiquitin-specific protease 7 (USP7) plays a crucial role in cancer progression by regulating key oncogenic proteins, making it a promising therapeutic target. We developed robust QSAR models using 837 curated USP7 inhibitors, 12 fingerprint descriptors, and a random forest algorithm, achieving high predictive accuracy (R2 = 0.96 ± 0.01, Q2 = 0.92 ± 0.02). Virtual screening identified high-potential compounds from NPASS, TCM, and ZINC databases. Redocking of the co-crystallized ligand (GNE6640, PDB ID: 5UQV) validated the docking protocol (RMSD = 0.330 Å). Top hits NPC472846, TCM11676, TCM36723, ZINC18193314, and ZINC65536649 were evaluated via molecular docking, revealing strong interactions with key USP7 residues, including Asp163, His217, Arg115, and Gln111. Molecular dynamics simulations (200 ns) demonstrated that TCM36723 and ZINC65536649 exhibited the highest dynamic stability, while NPC472846 induced more dynamic yet well-maintained conformational states. Hydrogen bond analysis confirmed persistent anchoring interactions, and PCA clustering revealed ligand-specific conformational flexibility. MM-GBSA free energy calculations identified NPC472846 as the top binder (-45.7 kcal/mol), followed by ZINC65536649 (-40.4 kcal/mol) and TCM11676 (-39.9 kcal/mol), all outperforming the reference ligand GNE6640 (-31.6 kcal/mol). These integrated computational results demonstrate that the identified hits particularly NPC472846 and ZINC65536649 are potent USP7 inhibitors, offering a foundation for future experimental validation and therapeutic development.