Interaction-profile cheminformatic read-across identifies the UV filter benzophenone-4 as a PPARγ agonist and...

Obesogens are chemicals, often encountered as environmental contaminants, that disrupt metabolic regulation and promote obesity. Here, we present a cheminformatics framework that integrates interaction-profile docking simulations with cluster-level enrichment analysis to enhance read-across and prioritize candidate environmental metabolic disruptors. Protein-ligand contact features from docking to obesity-related nuclear receptors were summarized at the pose level and combined into a 327-dimensional interaction-profile descriptor. Dimensionality-reduced descriptors from 6022 Tox21 compounds were clustered, and enrichment analysis against Tox21 assay results identified clusters associated with specific nuclear receptor activities. One cluster was selectively enriched for peroxisome proliferator-activated receptor γ (PPARγ) agonists. Although benzophenone-4 (BP-4, sulisobenzone), a sunscreen UV filter in this cluster, is labeled as inactive in Tox21, experimental validation confirmed selective PPARγ binding and recruitment of SRC-2 and PGC-1α coactivators. In human bone marrow-derived mesenchymal stem cells, BP-4 promoted adipogenic differentiation, lipid accumulation, and adiponectin production, establishing its potential as an environmental obesogen. This study demonstrates the power of combining interaction-profile read-across with functional assays to predict environmental metabolic disruptors and provides a mechanistic template for systematic chemical safety evaluation.