Abstract

Many factors cause inter-person variability in the activity and expression of liver cytochrome P450 (CYP) drug-metabolizing enzymes, leading to variable drug exposure and treatment outcomes. Several liver-enriched transcription factors (TFs) are associated with CYP expression, with estrogen receptor alpha (ESR1) and constitutive androstane receptor (CAR or NR1I3) being the two top factors. ESR1 and NR1I3 undergo extensive alternative splicing that results in numerous splice isoforms, but how these splice isoforms associate with CYP expression is unknown. Here, we quantified 18 NR1I3 splice isoforms and the three most abundant ESR1 isoforms in 260 liver samples derived from African Americans (AA, n=125) and European Americans (EA, n=135). Our results showed variable splice isoform populations in the liver for both NR1I3 and ESR1. Multiple linear regression analyses revealed that, compared to gene-level NR1I3, isoform-level NR1I3 expression better predicted the mRNA expression of most CYPs and three UDP-glucuronosyltransferases (UGTs), while ESR1 isoforms improved predictive models for the UGTs and CYP2D6, but not for most CYPs. Also, different NR1I3 isoforms were associated with different CYPs, and the associations varied depending on sample ancestry. Surprisingly, non-canonical NR1I3 isoforms having retained introns (introns 2 or 6) were abundantly expressed and associated with the expression of most CYPs and UGTs, whereas the reference isoform (NR1I3-205) only associated with CYP2D6. Moreover, NR1I3 isoform diversity increased during the differentiation of induced pluripotent stem cells to hepatocytes, paralleling increasing CYP expression. These results suggest that isoform-level TF expression may help to explain variation in CYP or UGT expression between individuals.

Significance Statement We quantified 18 NR1I3 splice isoforms and three ESR1 splice isoforms in 260 liver samples derived from AA and EA donors and found variable NR1I3 and ESR1 splice isoform expression in the liver. Multiple linear regression analysis showed that, compared to gene-level expression, isoform-level expression of NR1I3 and ESR1 better predicted the mRNA expression of some CYPs and UGTs, highlighting the importance of isoform-level analyses to enhance our understanding of gene transcriptional regulatory networks controlling the expression of drug-metabolizing enzymes.