Broad application of CYP3A4 LC-MS protein quantification in hepatocyte cytochrome P450 induction assays identifies nonuniformity in mRNA and protein induction responses.

Screening for cytochrome P450 (CYP) induction potential is routine in drug development. Induction results in a net increase in CYP protein and is assessed typically by measuring indirect endpoints, i.e., enzyme activity and mRNA in vitro. Recent methodological advancements have made CYP protein quantification by LC-MS in in vitro induction studies more accessible and amenable to routine testing. In this study, we evaluated CYP3A4 concentration dependence of induction response for 11 compounds (rifampin, rifabutin, carbamazepine, efavirenz, nitrendipine, flumazenil, pioglitazone, rosiglitazone, troglitazone, pazopanib, and ticagrelor) in plated hepatocytes from two or three donors incorporating in the assessment all three endpoints. In addition, the time-dependence of the induction was examined over 1, 2 or 3 days of treatment. For most compounds, mRNA, enzyme activity and protein endpoints exhibited similarity in induction responses. Pazopanib and ticagrelor were notable exceptions as neither protein nor enzyme activity were induced despite mRNA induction of a magnitude similar to efavirenz, pioglitazone or rosiglitazone, which clearly induced in all three endpoints. Static modeling of clinical induction responses supported a role for protein as a predictive endpoint. These data highlight the value of including CYP protein quantification as an induction assay endpoint to provide a more comprehensive assessment of induction liability. Significance Statement Direct, LC-MS-based quantification of CYP protein is a desirable induction assay endpoint, however the application of protein as an endpoint has been limited due to inefficient workflows. Here, we incorporate recent advancements in protein quantitation methods to efficiently quantify CYP3A4 protein in in vitro induction assays with 11 compounds in up to 3 donors. The data indicate induction responses from mRNA do not always align with those of protein suggesting assessment of induction liability is more complex than thought previously.