J Mol Graph Model. 2026 Oct 23. pii: S1093-3263(26)00354-2. [Epub ahead of print]150
109628
MDM2 is a principal negative regulator of the p53 tumor suppressor pathway, and pharmacological disruption of the p53-MDM2 interaction remains an important strategy for anticancer drug discovery. In this study, 176,407 drug-like compounds from the Otava Drug-Like Green Collection were evaluated through a multi-stage structure-based virtual screening workflow targeting the p53-binding pocket of MDM2. LigPrep preparation generated 256,757 ligand structures by enumerating possible protonation, tautomeric, stereochemical, and conformational states. Initial Glide/SP screening of these prepared states was followed by retention of the most favorable-scoring state for each Otava parent compound; the resulting Top 100 unique parent compounds were subsequently evaluated by Glide/XP, induced-fit docking, and quantum-polarized ligand docking. Cancer-QSAR scoring retained 32 candidates with predicted anticancer activity scores ≥ 0.50, after which their ADME and toxicity profiles were characterized for comparative assessment. From this 32-compound pool, five candidates (D1-D5) were advanced to 1 μs molecular dynamics simulations together with apo MDM2 and the positive control AMG232. Dynamic analyses showed that D5 maintained a stable binding-pocket position comparable to AMG232, whereas D3 exhibited substantially greater positional displacement despite showing the most favorable relative MM/GBSA binding-energy estimate among the candidate ligands; its trajectory-level binding geometry remained unresolved. Post-MD MM/GBSA calculations yielded the most favorable relative binding-energy estimate for D3, followed by D5, with average MM/GBSA estimates of -102.66 and -88.44 kcal/mol, respectively. Scaffold enrichment analysis further revealed marked overrepresentation of indole-containing and indole-coumarin-containing chemotypes among the final hits. Structural novelty analysis further indicated that D1-D32 did not show exact, parent-level, or Bemis-Murcko scaffold matches to the previous in-house/patented MDM2 comparator set. Collectively, these findings identify D3 and D5 as computationally prioritized virtual-screening candidates that warrant further biochemical and cellular validation.
Keywords: ADMET profiling; MDM2; MM/GBSA; Molecular dynamics simulation; Structure-based virtual screening; p53–MDM2 interaction