OECD-compliant QSAR modeling and molecular simulations prioritize plant-derived potential COX-2 inhibitors with experimentally observed anti-inflammatory activity through NO inhibition

Cyclooxygenase-2 (COX-2) is a key mediator of inflammation and a significant therapeutic target for numerous inflammatory disorders. We aimed to prioritize phytochemicals with potential COX-2 inhibitory activity using an integrated computational-experimental approach. Several OECD-compliant 2D- and 3D-QSAR models based on an assay-homogenized dataset of 50 known COX-2 inhibitors were developed and applied to screen 54 phytocompounds. The prioritized compounds were subsequently evaluated through computational ADMET, molecular docking, molecular dynamics (MD), and cellular assays. The QSAR models returned internal cross-validated Q 2 values of 0.63–0.70, with external predictivity assessed on small test and validation sets and interpreted as provisional. Applicability-domain analysis was extended to every screened compound, and predictions falling outside the modeled chemical space were identified as extrapolative. ADMET and predicted-toxicity screening narrowed the prioritized compounds to apigenin-7-sulfate and genistein, both within the AD. Molecular docking indicated favorable interactions within the COX-2 active site, and 50 ns molecular dynamics simulations showed both ligands remaining bound over the period sampled. Because apigenin-7-sulfate was commercially unavailable, its non-sulfated parent apigenin was evaluated in its place; apigenin and genistein inhibited nitric oxide production in LPS-stimulated RAW 264.7 macrophages in a concentration-dependent manner while maintaining > 90% cell viability. This endpoint reports suppression of the macrophage inflammatory response and does not establish COX-2 inhibition, which in this study remains a computational prediction. Collectively, the findings highlight apigenin-derived compounds and genistein as potential natural anti-inflammatory leads with likely involvement of multiple molecular pathways, including COX-2 modulation and NO inhibition. This also demonstrates the utility of integrating QSAR modeling, molecular simulations, ADMET profiling, and cellular validation for natural-product-based drug discovery.