Prediction of drug interactions, based on the induction of drug disposition, calls for the identification of chemicals, which activate xenosensing nuclear receptors. Constitutive androstane receptor (CAR) is one of the major human xenosensors; however, the constitutive activity of its reference variant CAR1 in immortalized cell lines complicates the identification of agonists. The exclusively ligand-dependent isoform CAR3 represents an obvious alternative for screening of CAR agonists. As CAR3 is even more abundant in human liver than CAR1, identification of its agonists is also of pharmacological value in its own right. We here established a cellular high-throughput screening assay for CAR3 to identify ligands of this isoform and to analyse its suitability for identifying CAR ligands in general. Proof-of-concept screening of 2054 drug-like compounds at 10 µM resulted in the identification of novel CAR3 agonists. The CAR3 assay proved to detect the previously described CAR1 ligands in the screened libraries. However, we failed to detect CAR3-selective compounds, as the four novel agonists, which were selected for further investigations, all proved to activate CAR1 in different cellular and in vitro assays. In primary human hepatocytes, the compounds preferentially induced the expression of the prototypical CAR target gene CYP2B6. Failure to identify CAR3-selective compounds was investigated by molecular modelling, which showed that the isoform-specific insertion of five amino acids did not impact on the ligand binding pocket but only on heterodimerization with retinoid X receptor. In conclusion, we demonstrate here the usability of CAR3 for screening compound libraries for the presence of CAR agonists.
Keywords: Agonist; Constitutive androstane receptor; High-throughput screening; Isoform; Molecular docking; Molecular dynamics simulations.