Using Molecular Initiating Events to Develop a Structural Alert Based Screening Workflow for Nuclear Receptor Ligands Associated with Hepatic Steatosis

In silico models are essential to the development of integrated alternative methods to identify organ level toxicity and lead towards the replacement of animal testing. These models include (quantitative) structure-activity relationships ((Q)SARs) and, importantly, the identification of structural alerts associated with defined toxicological endpoints. Structural alerts are able both to predict toxicity directly and assist in the formation of categories to facilitate read-across. They are particularly important to decipher the myriad mechanisms of action that result in organ level toxicity. The aim of this study was to develop novel structural alerts for nuclear receptor (NR) ligands that are associated with inducing hepatic steatosis. Current knowledge on NR agonists was extended with data from the ChEMBL database of bioactive molecules and from studying NR ligand-binding interactions within the protein data base (PBD). A computational structural alerts based workflow was developed using KNIME from these data using molecular fragments and other relevant chemical features. In total 214 structural features were recorded computationally as SMARTS strings and, therefore, they can be used for grouping and screening during drug development and risk assessment and provide knowledge to anchor adverse outcome pathways (AOP)