A frequent CYP2D6 variant promotes skipping of exon 3 and reduces CYP2D6 protein expression in human liver samples

CYP2D6 is one of the most polymorphic drug-metabolizing enzymes in the liver. While genetic CYP2D6 variants serve as clinical biomarkers to predict CYP2D6 activity, large inter-person variability in CYP2D6 expression remains unaccounted for. Previous results suggest that there is variable expression of a CYP2D6 splice isoform with an in-frame deletion of exon 3 (CYP2D6 & UDelta;E3) encoding a protein lacking numerous active site residues. Here, using fragment analysis and RT-qPCR, we revealed that rs1058164 G (MAF = 27%-43%) is associated with increased formation of CYP2D6 increment E3 in human liver samples (1.4-2.5-fold) and transfected cells. Furthermore, western blots showed that rs1058164 G was associated with a 50% decrease in full-length hepatic CYP2D6 protein expression. In addition, by studying a larger liver cohort, we confirmed our previous results that rs16947 (CYP2D6*2) reduces full-length CYP2D6 mRNA by increasing the production of an unstable splice isoform lacking exon 6 (CYP2D6 & UDelta;E6) and that the impact of CYP2D6 & UDelta;E6 is offset in carriers of the downstream enhancer variant rs5758550. The three frequent SNPs (rs1058164, rs16947, and rs5758550) form various 3-SNP-haplotypes, each with distinct CYP2D6 expression characteristics. Using an expression score (ES) system, we tested the impact of the 3-SNP-haplotype on improving the standard model to predict hepatic CYP2D6 protein expression based on genotype. A model that incorporates the 3-SNP-haplotype provided the best fit for CYP2D6 expression and also accounted for more variability in CYP2D6 protein levels (59%) than a model based on the accepted standard (36%) or one that only adds rs16947 and rs5758550 (42%). Clinical studies are needed to determine whether including the 3-SNP-haplotype alongside current standard CYP2D6 models improves the predictive value of CYP2D6 panels.