QSAR models in receptor-mediated effects: the nuclear receptor superfamily

The nuclear receptor (NR) superfamily is ligand-dependent transcriptional factors that mediate gene expression in humans and wildlife. These receptor-mediated effects are stimulated and/or inhibited by endogenous cognate ligands for each NR but also by exogenous substances including natural products and synthetic chemicals. The NRs and their ligands have thus attracted broad scientific interest, particularly in the pharmaceutical industry for drug discovery and in toxicology and environmental science for risk assessment as, for example, pertaining to endocrine disrupting chemicals. Besides advancing our fundamental knowledge of NR biology, these scientific efforts are generating relevant biological data on NR ligands particularly with respect to their binding affinities, receptor specificities, and agonist versus antagonist activities. These data from diverse sources serve as input for construction of quantitative structure-activity relationship (QSAR) models and related approaches that employ statistical regression techniques to correlate variations between the biological activities of NR ligands and their calculated structural and physicochemical properties. In this review, we attempt to summarize the substantial body of work in the published literature related to QSAR models for NR ligands, with special emphasis on different computational approaches and specific applications. Special attention is placed on the estrogen receptor, for which the greatest amount of relevant information is known at present. We also describe efforts to create 'benchmark' sets of high-quality biological data on NR ligands that may serve as resources for building statistically robust and predictive QSAR models. Published by Elsevier Science B.V.