Characterization of Hepatic UDP-Glucuronosyl Transferase Enzyme Abundance - Activity Correlations and Population Variability using a Proteomics Approach and Comparison with Cytochrome P450 Enzymes

The expression of ten major drug-metabolizing UDP-glucuronosyl transferase (UGT) enzymes in a panel of 130 human hepatic microsomal samples was measured using a LC-MS/MS-based approach. Simultaneously, ten cytochrome P450s and P450 reductase were also measured and activity-expression relationships assessed for comparison. The resulting data sets demonstrated that, with the exception of UGT2B17, 10th-90th percentiles of UGT expression spanned 3- to 8-fold ranges. These ranges were small relative to ranges of reported mean UGT enzyme expression across different laboratories. We tested correlation of UGT expression with enzymatic activities using selective probe substrates. A high degree of abundance-activity correlation (rs > 0.6) was observed for UGT1As (1A1, 3, 4, 6) and CYPs. In contrast, protein abundance and activity did not correlate strongly for UGT1A9 and UGT2B enzymes (2B4, 7, 10, 15 and 17). Protein abundance was strongly correlated for UGTs 2B7, 2B10 and 2B15. We suggest a number of factors may contribute to these differences including incomplete selectivity of probe substrates, correlated expression of these UGT2B isoforms, and the impact of splice and polymorphic variants on the peptides used in proteomics analysis and exemplify this in the case of UGT2B10. Extensive correlation analyses identified important criteria for validating the fidelity of proteomics and enzymatic activity approaches for assessing UGT variability, population differences and ontogenetic changes. Significance Statement Protein expression data allow detailed assessment of inter-individual variability and enzyme ontogeny. We observed that expression and enzyme activity are well correlated for hepatic UGT1A enzymes and P450s. However, for the UGT2B family, caution is advised when assuming correlation of expression and activity as is often done in physiologically based pharmacokinetic modeling. This can be due to incomplete probe substrate specificities but may also be related to presence of inactive UGT protein materials and the effect of splicing variations.