Evidence for regulation of UDP-glucuronosyltransferase (UGT) 1A1 protein expression and activity via DNA methylation in healthy human livers

Abstract

Objectives: Interindividual variability in glucuronidation of bilirubin and drugs by UDP-glucuronosyltransferase 1A1 (UGT1A1) is considerable and only partially explained by genetic polymorphisms and enzyme inducers. Here we determined whether a well-known epigenetic modification, cytosine methylation, explains a proportion of this variability in human liver.

Methods: UGT1A1 phenotypes, including UGT1A1 protein and bilirubin glucuronidation, and UGT1A1*28 genotype were determined using a human liver bank (n = 46). Methylation levels were quantified at 5 CpG sites associated with known transcription factor response elements in the UGT1A1 promoter and distal enhancer, as well as a CpG-rich island 1.5 kb further upstream.

Key findings: Individual CpG sites showed considerable methylation variability between livers, ranging from 10- to 29-fold variation with average methylation levels from 25 to 41%. Multivariate regression analysis identified *28/*28 genotype, -4 CpG site methylation and alcohol history as significant predictors of UGT1A1 protein content. Exclusion of livers with *28/*28 genotype or alcohol history revealed positive correlations of -4 CpG methylation with bilirubin glucuronidation (R = 0.73, P < 0.00001) and UGT1A1 protein content (R = 0.54, P = 0.008).

Conclusion: These results suggest that differential methylation of the -4 CpG site located within a known USF response element may explain a proportion of interindividual variability in hepatic glucuronidation by UGT1A1.

Figure 1

Examples of sequence chromatograms of PCR-amplified bisulfite-converted DNA samples used to quantitate methylation at the −4 CpG site located in the UGT1A1 promoter. The bisulfite reaction deaminates all unmethylated cytosines (C) to uracils, which are then replaced by thymines during subsequent PCR amplification. All methylated cytosines (methyl. C) remain unchanged after bisulfite treatment. Traces from a liver sample with (a, b) an estimated 76% methylation (LV17) and (c, d) 8% methylation (LV9). (a) and (c) The raw data chromatograms prior to sequencer software processing that adjusts average peak height and elution time adjustments to enable automated sequence calls. The results of this processing are shown in (b) and (d), respectively. Note the large increase in apparent height of the small methyl. C peak in (c) to that in (d) resulting from this processing. The relative amount of cytosine methylation was calculated from the processed chromatograms according to the method of Leakey et al.^[26] by expressing the height of the unmethylated cytosine (C) peak (converted to thymine) as a percentage of the average height of 10 surrounding non-CpG thymidine peaks (indicated in (b) and (d) by asterisks) and subtracting this value from 100.

Figure 2

(a) 5'-flanking region of UGT1A1 exon 1, including promoter and distal enhancer. The CpG sites were numbered according to their appearance upstream from the ATG start codon. Corresponding base pairs indicate the distance from the ATG codon. The positions of response elements for different transcriptional factors (PXR, Nrf-2, AhR, USF, HNF-1α and PBX) are marked as experimentally demonstrated previously.^[20,25] (b) Percentage methylation values derived for individual livers at the five different CpG sites and the CpG-rich island. The horizontal lines represent the median of the percentage methylation values derived for the individual CpG sites in 46 normal human liver samples and the CpG-rich island using 18 normal human liver samples with either high (n = 9) or low (n = 9) bilirubin glucuronidation and UGT1A1 protein content.

Figure 3

Correlation in methylation values determined for individual livers (n = 46) measured at (a) the −3 and −4 CpG sites and (b) the −28 and −29 CpG sites.

Figure 4

Correlation of (a) −4 CpG site methylation with (a) bilirubin glucuronidation and (b) UGT1A1 protein level measured in a bank of human livers (n = 46). The filled circles identify a subgroup that excludes donors with either a significant alcohol history or have the UGT1A1 *28/*28 genotype. Spearman correlation values (Rho coefficients and P values) are shown for this subgroup.