UGT1A1 polymorphisms are important determinants of dietary carcinogen detoxification in the liver

Authors

  • Hugo Girard,

    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    Search for more papers by this author
  • Jean Thibaudeau,

    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    Search for more papers by this author
  • Michael H. Court,

    1. Department of Pharmacology and Experimental Therapeutics, Tufts University, Boston, MA
    Search for more papers by this author
  • Louis-Charles Fortier,

    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    Search for more papers by this author
  • Lyne Villeneuve,

    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    Search for more papers by this author
  • Patrick Caron,

    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    Search for more papers by this author
  • Qin Hao,

    1. Department of Pharmacology and Experimental Therapeutics, Tufts University, Boston, MA
    Search for more papers by this author
  • Lisa L. von Moltke,

    1. Department of Pharmacology and Experimental Therapeutics, Tufts University, Boston, MA
    Search for more papers by this author
  • David J. Greenblatt,

    1. Department of Pharmacology and Experimental Therapeutics, Tufts University, Boston, MA
    Search for more papers by this author
  • Chantal Guillemette

    Corresponding author
    1. Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, Oncology and Molecular Endocrinology Research Center, CHUL Research Center and Faculty of Pharmacy, Laval University, Québec, Canada
    • Canada Research Chair in Pharmacogenomics, Laboratory of Pharmacogenomics, CHUL Research Center, T3-48, 2705 Boul. Laurier, Québec, Canada, G1V 4G2
    Search for more papers by this author
    • fax: 418-654-2761


  • Potential conflict of interest: Nothing to report.

  • Preliminary data from this study were presented at the 2004 International Society for Study of Xenobiotics; Vancouver, Canada; August 29–September 2, 2004 and at the 11th Workshop on Glucuronidation, Dundee, Scotland; September 5–September 8, 2004.

Abstract

PhIP (2-amino-1-methyl-6-phenylimidazo[4,5-f]pyridine), the most abundant heterocyclic amine in diet, is involved in the etiology of cancer. PhIP and its carcinogenic metabolite N-hydroxy-PhIP (N-OH-PhIP) are extensively conjugated by UDP-glucuronosyltransferase (UGTs) with wide variability. This study aimed to determine the genetic influence of UGTs on the hepatic detoxification of this carcinogen. The formation of N-OH-PhIP glucuronides was studied in 48 human liver samples by mass spectrometry. Liver samples were genotyped for common polymorphisms and correlated with UGT protein levels and N-OH-PhIP glucuronidation activities. The formation of four different N-OH-PhIP glucuronide metabolites was observed in all livers. The major metabolite was N-OH-PhIP-N2-glucuronide (N2G), which is the primary metabolite found in human urine, and showed a high interindividual variability (up to 28-fold). Using an heterologous expression system, the bilirubin-conjugating UGT1A1 enzyme was identified among all known UGTs (n = 16) as the predominant enzyme involved. The significant correlation between UGT1A1 protein content and formation of N2G (Rs = 0.87; P < .0001) suggests a critical role for UGT1A1 in the hepatic metabolism of this carcinogen. UGT1A1 expression was strongly determined by the presence of the common promoter polymorphisms, UGT1A1*28 (TATA box polymorphism) (P = .0031), −3156G/A (P = .0006) and −3279G/T (P = .0017), and rates of N2G were indeed correlated with these polymorphisms (P < .05), whether analyzed individually or in combination (haplotypes). In conclusion, UGT1A1 polymorphisms modulate the hepatic metabolism of the carcinogenic intermediate of PhIP and may determine the level of its exposure and potentially influence the risk of cancer through dietary exposure to HCAs. (HEPATOLOGY 2005.)

The Western diet is a main source of dietary carcinogens involved in neoplastic diseases, the evidence being strongest for colon cancer.1–3 Dietary carcinogens include two groups of chemicals, heterocyclic amines (HCAs) and polyaromatic hydrocarbons (PAHs). Consumption of pan-fried, grilled, or barbecued well-done meat fosters HCA and PAH exposure1, 4 and is positively associated with colon cancer.1

The enzymatic processes that transform both HCAs and PAHs comprise multiple phase I and phase II pathways involved in chemical activation and metabolic detoxification.5 Among the different food-borne carcinogens, 2-amino-1-methyl-6-phenylimidazo[4,5-b]-pyridine (PhIP) is the predominant HCA.6, 7 A large proportion of the ingested PhIP undergoes bioactivation to N-hydroxy-PhIP (N-OH-PhIP).8 This metabolite with mutagenic and carcinogenic properties is subsequently conjugated with glucuronic acid by UDP-glucuronosyltransferases (UGTs).9 The direct glucuronidation of PhIP and N-OH-PhIP represents a primary route of detoxification in humans, for which the liver is believed to be the predominant site.8 After PhIP exposure, the main metabolite in urine after PhIP exposure is N-OH-PhIP-N2glucuronide, followed by PhIP-N2-glucuronide, N-OH-PhIP-N3glucuronide, and 4′-PhIP-sulfate.10–13

Controlled feeding studies revealed a significant variation in the excretion of urinary PhIP and its polar metabolites, indicating that there are substantial interindividual differences in glucuronidation capacity.12, 14–16 Multiple genes encoding for glucuronidation enzymes display polymorphic distribution.17 UGT genetic factors might determine the host capacity to detoxify HCA carcinogens, thus modifying their exposure to HCAs and potentially their risk of developing cancer. However, the identity of the UGTs responsible for the glucuronic acid conjugation of PhIP and particularly N-OH-PhIP remains unclear, because literature reports are inconsistent.18, 19

Our initial goal was to study the glucuronidation profile of N-OH-PhIP in human liver microsomes and to test the 16 known human UGT1A and UGT2B20 for their capacity to form these metabolites in enzymatic assays using recombinant proteins. Four N-OH-PhIP metabolites were identified, and a limited number of UGTs (among them UGT1A1) expressed in the liver and other target sites of this carcinogen were efficient at forming these metabolites with varying capacity and regional selectivity.

We then investigated (1) the degree of variation in the hepatic glucuronidation of N-OH-PhIP and (2) the role of UGT polymorphisms in determining the variability of N-OH-PhIP glucuronidation between individuals. We chose to focus on the UGT1A1 gene because of its major involvement in the in vitro formation of N-OH-PhIP-N2G and because the liver is a predominant site of UGT1A1 expression.21 The dinucleotide TA repeat in the TATA box of the gene (UGT1A1*28), the −3156G/A and −3279G/T promoter variants22–24 were genotyped in 48 subjects. Then the correspondence between N-OH-PhIP glucuronidation activities, UGT1A1 genotypes, UGT1A1 phenotypic expression, and UGT1A1-probe substrate activities was assessed.

Results demonstrated the ability of the UGT1A1 genotypes to predict variable levels of UGT1A1 protein and rates of hepatic glucuronidation of N-OH-PhIP at positions N2 and N3. Our findings indicate that UGT1A1 promoter polymorphisms modulate the metabolism of N-OH-PhIP and are therefore likely to determine the levels of exposure to this carcinogenic agent.

Abbreviations

HCA, heterocyclic amines; PAH, polyaromatic hydrocarbons; PhIP, 2-amino-1-methyl-6-phenylimidazo[4,5-b]-pyridine; N-OH-PhIP, N-hydroxy-PhIP; UGT, UDP-glucuronosyltransferase; HPLC, high-performance liquid chromatography.

Materials and Methods

Chemicals.

N-OH-PhIP was purchased from Toronto Research Chemical (Downsview, ON, Canada). Methanol (HPLC grade) was obtained from VWR Canlab (Montréal, Canada). β-Glucuronidase type VII from Escherichia coli and all other reagents were from Sigma (St. Louis, MO). Commercial sources of UGT1A1, 1A4, 1A7, 1A9, and 1A10 were obtained from BD Biosciences (Woburn, MA). N-OH-PhIP glucuronide standards were generated by using liver microsomes and purified by high-performance liquid chromatography (HPLC). Metabolite fractions were collected, lyophilized, and the residue was dissolved in methanol and stored at −80°C. To determine the concentration of purified glucuronides, each one was treated individually with 5,000 U/mL β-glucuronidase in a 25 mmol/L phosphate buffer (pH 6.5), incubated at 37°C, and further quantified using an N-OH-PhIP calibration curve. The detection of compounds was linear from 0.5 ng/mL to 100 ng/mL for N2G, from 0.1 ng/mL to 25 ng/mL for N3G and from 0.5 ng/mL to 100 ng/mL for N-OH-PhIP. The uncharacterized N-OH-PhIP-glucuronide 1 (Glucu 1) and N-OH-PhIP-glucuronide 2 (Glucu 2) were quantified with a N-OH-PhIP-N2G calibration curve because they shared the same fragmentation profile.

Human Samples.

DNA samples and liver microsomes from 48 American subjects were obtained from the International Institute for the Advancement of Medicine (Exton, PA), the National Disease Research Interchange (Philadelphia, PA), and the University of Minnesota Liver Tissue Procurement and Distribution system (Minneapolis, MN), as previously described.25, 26 Population characteristics were previously described.27 All livers were either intended for transplantation but failed to match the tissue or were normal tissue adjacent to surgical biopsy specimens. All subjects provided written consent for experimental purposes, and the Tufts University and Laval University Institutional Review Boards approved the use of these samples in the current study.

LC-MS/MS Analytical Method.

Samples were analyzed using HPLC (Alliance 2690, Waters, Milford, MA) and eluted at a flow rate of 0.9 mL/min. The initial conditions were 60% A and 40% B, followed by a linear gradient up to 55% B in 3 minutes. This concentration was held for 4 minutes, after which the column was flushed with 95% B for 2 minutes and then re-equilibrated to initial conditions over 3 minutes. The effluent from the HPLC system was connected directly to an API 3000 triple quadrupole mass spectrometer (Sciex, Toronto, Canada) equipped with a turbo-ion spray source with a split of 1:4 in positive mode. The mass spectrometer was operated in the multiple reaction monitoring mode using the following conditions: spray probe temperature at 400°C, ionization voltage at 5,000 V, the orifice and the ring at 31 V and 190 V, respectively. Data were acquired with a dwell time of 400 milliseconds, a pause time of 5 milliseconds, and a scan time of 1.2 seconds. The multiple reaction monitoring transitions used for analysis were 241→223 (N-OH-PhIP), 417→241 (N-OH-PhIP N2G, Glucu 1, and Glucu 2), and 417→224 (N-OH-PhIP-N3G).

Enzymatic Assays and UGT1A1 Genotyping Analyses.

UGT microsomes and determination of UGT expression have been previously described.25 Enzymatic assays were performed in 50 mmol/L Tris-HCl PH 7.5, 10 mmol/L MgCl2, 2 mmol/L uridinediphosphoglucuronic acid, 5 μg/mL pepstatine, 5 μg/mL leupeptine, 10 μg/mL phosphatidylcholine, 0.5 mmol/L EDTA, 60 μg/mg protein alamethicin in a final volume of 100 μL and stopped with 100 μL methanol. A first series of experiments were performed using 25 μmol/L and 100 μmol/L N-OH-PhIP incubated with human liver or recombinant UGT microsomes for 2 hours, to determine the most reactive UGT enzymes. Incubations were then performed with at least 10 concentrations of N-OH-PhIP (1-200 μmol/L) for at least two independent experiments in duplicate, to determine the kinetic parameters (Km and Vmax estimates). Incubation times were 2 hours for UGT1A4, 1A8, 1A9, 2B10 and 30 minutes for 1A1 and HHOP2 (a pool of human liver microsomes). Enzymatic assays were also performed for 30 minutes with 25 μmol/L N-OH-PhIP with 48 human liver microsomes. An aliquot of each reaction was submitted to liquid chromatography–mass spectrometry analysis for quantification. The UGT1A1 protein from 48 subjects was quantified using a specific polyclonal antibody directed against the N-terminal portion of UGT1A1.28 The relative levels of UGT proteins were compared with the sample with the lowest expression. DNA samples from the 48 subjects were genotyped for the dinucleotide insertion/deletion present in the promoter region of the UGT1A1 gene using the previously described Genescan method.29 A portion of UGT1A1 PBREM (phenobarbital-responsive enhancer module) region was amplified by polymerase chain reaction with primers #652 forward (−3638 to −3656 relative to the ATG) 5′-CTGGGGATAAACATGGGATG-3′and #653 reverse (−3052 to −3071 relative to the ATG) 5′-CACCACCACTTCTGGAACCT-3′ and submitted to automated sequencing using primer #652 to genotype polymorphisms at positions −3156 and −3279.

Statistical Analysis.

Correlation studies of genotypes, protein levels and enzymatic assays were performed using the JMP V4.0.2 program (SAS Institute, Cary, NC) and the bivariate analysis of variances (nonparametric Spearman's test). The one-way ANOVA test for mean comparison was used for correlations involving nominal values (UGT1A1 genotypes). UGT1A1 activity and expression data were analyzed for normality of distribution using the Shapiro-Wilk W test (P > .05). In some instances, logarithmic transformation of the raw data was necessary to achieve normal distribution, which is a prerequisite for the statistical tests used. A P value of less than .05 and an Rs value higher than 0.50 were considered significant. Correlation between UGT1A1 genotypes, expression levels, and activities were performed using ANOVA and Tukey-Kramer HSD test with the program JMP V4.0.2.

Results

N-OH-PhIP Glucuronidation Profiles and Variability in Human Livers.

N-OH-PhIP glucuronide metabolites were initially identified in a pool of 8 human liver microsomes based on HPLC co-elution with authentic metabolite standards, mass spectral analysis, and susceptibility to enzymatic cleavage. Chromatographic separation of liver incubation reactions indicated four metabolite peaks eluting at 1.91 (peak 1), 3.55 (peak 2), 3.92 (peak 3), and 4.86 (peak 4) minutes (Fig. 1A). Based on the mass spectral analysis, metabolites were identified as glucuronide derivatives of N-OH-PhIP with primary molecular ions at 241.0 m/z [M+H-glucuronic acid]+ for peaks 1, 2, and 3 and 224.7 m/z [M+H-glucuronic acid-OH]+ for peak 4 (Fig. 1B-C). Based on our observations and previous reports,19, 30 peaks 3 and 4 were identified as N-OH-PhIP-N2G and N-OH-PhIP-N3G and peak 1 as an uncharacterized N-OH-PhIP-G 1 (Glucu 1). An additional glucuronide N-OH-PhIP-G 2 (Glucu 2) corresponding to peak 2 that has never been previously described demonstrated a fragmentation profile similar to N-OH-PhIP-N2G.

Figure 1.

High-pressure liquid chromatography (HPLC) mass spectrometer analysis of N-OH-PhIP glucuronides formation by human liver microsomes. Metabolites were identified based on HPLC co-elution with authentic N-OH-PhIP metabolite standards, mass spectral analysis, and susceptibility to enzymatic cleavage. (A) Chromatographic separation of N-OH-PhIP glucuronides formed in a pool of human liver microsomes. Microsomes (8 mg of protein) were incubated at 37°C for 30 minutes as described in Materials and Methods. (B) Mass spectral analysis of peaks corresponding to N2G, Glucu 1, and Glucu 2. (C) Mass spectral analysis of peaks corresponding to N3G. R.T, retention time; N2G, N-OH-PhIP-N2glucuronide; N3G, N-OH-PhIP-N3glucuronide; Glucu-1, undefined N-OH-PhIP-glucuronide 1; Glucu-2, undefined N-OH-PhIP-glucuronide 2.

To determine whether variability in N-OH-PhIP glucuronidation observed in vivo14, 30 can be observed at the hepatic level, enzymatic assays were performed with microsomes prepared from 48 liver specimens. The formation of N-OH-PhIP-N2G was predominant in all tested livers, representing 40% to 93% of total glucuronides (28-fold variation; 23 to 651 pmol/min/mg). Significant variability was also observed for the other metabolites, with absolute levels of 1.4 to 25.3 pmol/min/mg for N3G (18-fold variability), 2.1 to 29.6 pmol/min/mg for Glucu 1 (14-fold variability), and 1.1 to 9.6 pmol/min/mg for Glucu 2 (8-fold variability). Glucu 1 contributed for 1% to 43%, Glucu 2 for 1% to 11%, and N3G for 3% to 8% of total glucuronide formation.

Regioselective Formation of N-OH-PhIP Glucuronides by UGTs.

To identify which UGT proteins were capable of forming each of the 4 N-OH-PhIP glucuronides, microsome preparations from 16 human recombinant UGT1A- or UGT2B-overexpressing -HEK293 cells were incubated with 25 μmol/L and 100 μmol/L N-OH-PhIP, in parallel with commercial preparations of UGT1A1, 1A4, 1A7, 1A9, and 1A10 (data not shown).

The enzyme kinetics was then performed for the most reactive UGTs, namely, 1A1, 1A4, 1A8, 1A9, and 2B10. The formation of N-OH-PhIP glucuronides was undetectable with all the other tested UGT isoenzymes, namely UGT1A5, 1A6, 1A7, 1A10, 2B4, 2B7, 2B11, 2B15, 2B17, and 2B28. The commercial sources of yielded N-OH-PhIP conjugation profiles and rates of formation similar to those obtained with HEK293 UGT microsomes. Among liver-expressed UGTs, 1A1 presented the highest catalytic efficiency (Vmax/Km) for both the formation of the N-OH-PhIP-N2G and N3G with estimated values of 7.2 and 0.2 μL/min/UGT protein respectively, in a range closest to that observed for the pool of human liver microsomes. UGT1A8 was the second most reactive UGT with Clint values of 1.8 and 0.1 μL/min/UGT protein for N2G and N3G, respectively, but because this enzyme is not present in the liver, its role would be limited to extrahepatic tissues, especially the conjugation of N-OH-PhIP in the gastrointestinal tract. In contrast, the low capacities of the hepatic 1A4 and 1A9 suggest that these isoforms play a minor role in the formation of the main glucuronide N-OH-PhIP-N2G. UGT1A4 was the only UGT involved in the formation of Glucu 1, whereas Glucu 2 would most likely be formed predominantly by UGT2B10 and 1A9 (Table 1). UGT1A4,21 1A9,31 and 2B1032 are present in the liver.

Table 1. Kinetic Parameters for N-OH-PhIP Glucuronidation by Human UGTs
N-OH-PhIP GlucuronidesPool of Liver MicrosomesHuman UGTs
1A11A41A81A92B10
  • NOTE. Data are expressed as mean ± SD of at least 2 independent experiments.

  • *

    Units: Km (μmol/L); Vmax (pmol/min/mg protein); CLint (μL/min/mg); Relative Vmax (pmol/min/UGT protein); Relative CLint (μL/min/UGT protein).

  • n: Hill coefficient, sigmoid profile.

  • Relative Vmax was calculated by dividing the Vmax values by the expression level of the corresponding UGT protein. The relative Vmax was not calculated for UGT2B10 because 2 separate antibodies were used to quantify UGT1A and UGT2B proteins and therefore values can not be compared. Compared with all other UGT2B enzymes tested (2B4, 2B7, 2B11, 2B15, 2B17, and 2B28), UGT2B10 demonstrated the lowest expression level. All the kinetics followed hyperbolic profiles, except for UGT1A4, which demonstrated a sigmoid profile for all 4 N-OH-PhIP glucuronides. Relative levels of UGT expression were assessed by Western blot analysis using a specific anti-UGT1A antibody for UGT1A family members and a specific anti-UGT2B antibody for UGT2B10. Levels of expression were UGT1A1 (1.8), 1A4 (12.8), 1A8 (4.7), 1A9 (1.9) relative to 1A7 (1.0).

N2-G*      
 n  1.4 ± 0.2   
 Km22.5 ± 0.218.9 ± 3.3127.7 ± 46.271.0 ± 12.4218.5 ± 57.3 
 Vmax272.9 ± 65.0245.4 ± 5.264.8 ± 18.3589.3 ± 87.517.0 ± 3.0 
 CLint12.113.00.38.30.08 
 Relative Vmax 136.45.1125.48.9 
 Relative CLint 7.20.021.80.04 
N3-G*      
 n  1.3 ± 0.1   
 Km34.8 ± 4.123.9 ± 3.6163.0 ± 59.383.6 ± 19.951.2 ± 2.2 
 Vmax14.1 ± 3.610.3 ± 2.42.4 ± 0.340.0 ± 2.79.0 ± 3.1 
 CLint0.40.40.010.50.2 
 Relative Vmax 5.70.28.54.7 
 Relative CLint 0.2<0.010.10.1 
Glucu 1*      
 n  1.4 ± 0.1   
 Km2055.5 ± 49.4 104.6 ± 23.4   
 Vmax1219.2 ± 245.8 1144.7 ± 576.6   
 CLint0.6 6.0   
 Relative Vmax  89.4   
 Relative CLint  0.5   
Glucu 2*      
 n  2.6 ± 0.4   
 Km38.7 ± 17.9 104.3 ± 30.3 36.3 ± 5.157.8 ± 51.0
 Vmax7.0 ± 3.0 12.1 ± 4.2 1.3 ± 0.51.4 ± 0.4
 CLint0.2 0.06 0.040.02
 Relative Vmax  0.9 0.7 
 Relative CLint  <0.01 0.02 

Genotype and Phenotype Relationships.

Based on our in vitro enzymatic assays, UGT1A1, which is predominantly expressed in the liver, emerged as the most efficient enzyme at forming N2G and N3G. To determine whether the formation of N-OH-PhIP glucuronides was correlated to the presence of the UGT1A1 protein, its expression was assessed in the 48 liver samples (Fig. 2A) and varied widely (43-fold). The significant correlation between levels of UGT1A1 protein and the formation of N2G (Rs = 0.874; P < .0001) supports the critical role of UGT1A1. The formation of N3G was also highly determined by the level of UGT1A1 protein (Rs = 0.869; P < .0001). In turn, UGT1A1 would play a limited role in the formation of Glucu 2 (Rs = 0.516; P = .0002) and especially Glucu 1 (Rs = 0.217; P = .06) in agreement with results of the in vitro metabolic experiments with isolated UGT proteins.

Figure 2.

N-OH-PhIP glucuronidation levels in liver microsomes in 48 subjects stratified by UGT1A1 TATA box genotypes. The influence of the UGT1A1*28 allele on the level of UGT1A1 protein (B), N-OH-PhIP N2G (C), and N3G formation (D) is illustrated. Western blot analyses were performed using a specific anti-UGT1A1 antibody to determine the level of UGT1A1 protein in 48 human liver microsomes as described in Materials and Methods. A representative portion of the Western blot from subjects with the three UGT1A1 genotypes is shown in panel A. Also shown above each graph are the P values for comparison between groups. All groups were compared with the others with the Tukey-Kramer HSD statistical test (alpha value set to 0.05) and using one-way ANOVA on normalized data. n.s., nonsignificant. Bars indicate mean values.

The wide range of distribution of the UGT1A1 protein expression is linked to the presence of several promoter polymorphisms in this gene alone or combined (haplotypes).22, 24, 29, 33 To determine whether UGT1A1 polymorphisms could contribute to the variability in N-OH-PhIP glucuronide levels, DNA from the 48 liver specimens were first genotyped for the dinucleotide repeat polymorphisms in the atypical TATA-box region of the UGT1A1 promoter. Six TA repeats characterize the common allele (UGT1A1*1), whereas the variant allele consists of 7 TA repeats (UGT1A1*28) associated with a decrease in the UGT1A1 gene expression in vitro.22, 29, 33 Two less frequent alleles, A(TA)5TAA and A(TA)8TAA, are referred to as UGT1A1*36 and UGT1A1*37 (http://som.flinders.edu.au/FUSA/ClinPharm/UGT).

In the tested population, the allelic frequency of the UGT1A1 promoter variant alleles was in Hardy-Weinberg equilibrium with frequencies of 0.66, 0.31, 0.02, and 0.01 for the *1, *28, *36, and *37 alleles, respectively, similar to those reported (reviewed in Guillemette17). Two subjects were found to have the TA5 allele (*1 /*36), whereas 1 subject carried the TA8 allele (*1 /*37). For genotype-phenotype analyses, subjects with the *36 and the *37 alleles were grouped with individuals with the *1 and *28 alleles, respectively, based on their function.22, 29, 33 As expected, the presence of UGT1A1*28 was significantly correlated with a decrease in UGT1A1 expression (P = .0031) (Fig. 2B). A significant reduced amount of UGT1A1 (−58%) was observed for the UGT1A1*28/*28 genotype compared with UGT1A1*1/*1 (P < .05 Tukey-Kramer HSD test), whereas the heterozygous UGT1A1*1/*28 presented an intermediate level of expression (Fig. 2B).

We also found that the UGT1A1 promoter genotypes were predictive of the rates of glucuronidation of the probe substrate estradiol (E2) (P = .013),34 the N2G (P = .0074), and N3G (P = .016) formation (Fig. 2; Table 2).

Table 2. UGT1A1 Protein Expression and Glucuronosyltransferase Activities in Human Liver Genotyped for Polymorphisms in the UGT1A1 Promoter
Genotype UGT1A1UGT1A1 Protein Arbitrary UnitsN-OH-PhIP N2G pmol/min/mgN-OH-PhIP N3G pmol/min/mgN-OH-PhIP Glucu-1 pmol/min/mgN-OH-PhIP Glucu-2 pmol/min/mgE2-3G Probe Substrate pmol/min/mg
  • NOTE. Data are expressed as mean ± SD. NS, nonsignificant (P > .05).

  • ANOVA test.

  • Significantly different (Tukey-Kramer HSD) from the corresponding homozygote “wild-type” group, alpha value set at 0.05.

−53TA      
 *1/*19330 ± 3984203 ± 1439.6 ± 5.413.4 ± 7.05.3 ± 1.80.38 ± 0.28
 *1/*287113 ± 5152144 ± 1307.2 ± 5.110.9 ± 7.34.7 ± 2.10.33 ± 0.29
 *28/*283944 ± 4559107 ± 1475.6 ± 5.715.6 ± 9.65.3 ± 2.20.22 ± 0.42
 P value.0031.0074.016NSNS.013
−3156      
 G/G9408 ± 3994201 ± 1419.5 ± 5.313.3 ± 7.25.4 ± 1.80.38 ± 0.29
 G/A6001 ± 5166124 ± 1246.3 ± 4.910.0 ± 6.94.3 ± 2.30.31 ± 0.27
 A/A3944 ± 4559107 ± 1475.6 ± 5.715.6 ± 9.65.3 ± 2.20.22 ± 0.42
 P value.0006.0034.0061NSNS.014
−3279      
 G/G4681 ± 370092 ± 1075.3 ± 4.213.7 ± 7.95.2 ± 2.00.25 ± 0.32
 G/T8213 ± 5356184 ± 1408.8 ± 5.511.9 ± 7.74.5 ± 2.20.34 ± 0.30
 T/T10071 ± 3146218 ± 15110.0 ± 5.613.4 ± 7.45.9 ± 1.50.43 ± 0.29
 P value.0017.0014.01NSNS.025

Two single-nucleotide polymorphisms at positions −3156 G→A and −3279 G→T have recently been identified.23, 24 In the tested population, allelic frequencies were, respectively, 0.73 and 0.27 for the −3156G (reference sequence in Genbank) and −3156A alleles and 0.50 for both the −3279G allele and −3279T variant. The levels of UGT1A1 expression were significantly lower in subjects with at least one −3156A variant allele (P = .0006) or in the presence of the −3279G allele (P = .0017) (Fig. 3A,D; Table 2). The relationship between the UGT1A1 −3156 and −3279 variants and glucuronidation of N-OH-PhIP at position N2 and N3 was also significant. In the case of the −3156, homozygous A/A subjects presented significantly lower rates of formation of both N2G and N3G compared with homozygous −3156 G/G individuals (P < .05) (Fig. 3B-C; Table 2). However, the variant at position −3279 may be a better predictor of the rates of N2G and N3G formation because subjects with at least one −3279G allele presented significantly lower rates of formation compared with homozygous −3279 T/T (P = .0014) (Fig. 3E-F; Table 2).

Figure 3.

N-OH-PhIP glucuronidation levels in liver microsomes in 48 subjects stratified by UGT1A1 −3156 (A-C) and −3279 (D-F) genotypes. The influence of the UGT1A1 −3156 G/A and −3279 G/T polymorphism on the level of UGT1A1 protein, N-OH-PhIP N2G and N3G formation is illustrated. Western blot analyses were performed using a specific anti-UGT1A1 antibody to determine the level of UGT1A1 protein in 48 human liver microsomes as described in Materials and Methods. The P values for comparison between groups are shown above each graph. All groups were compared with each other with the Tukey-Kramer HSD statistical test (alpha value set to 0.05) and using one-way ANOVA on normalized data. n.s., nonsignificant. Bars indicate mean values.

Because the extended marker haplotypes may provide additional power in the detection of associations with phenotypes, we investigated the haplotype structure of the UGT1A1 promoter using the obtained polymorphism data at positions −3279, −3156, and the (TA)n repeat. Four haplotypes and 8 different diplotypes could be inferred (Fig. 4A). Haplotype I, at a frequency of 0.50, corresponds to the presence of the UGT1A1*1, −3156G and −3279T variations individually linked to the highest UGT1A1 expression and rates of glucuronidation. Haplotype II (0.27) includes variants (UGT1A1*28, −3156A, and −3279G) individually associated with a significant decrease in UGT1A1 expression and rates of glucuronidation. Haplotype III (0.18) corresponds to the presence of the UGT1A1*1, −3156G, and −3279G variations and haplotypes IV (0.05) to UGT1A1*28, −3156G, and −3279G.

Figure 4.

Haplotype frequencies of the UGT1A1 promoter (A), UGT1A1 diplotypes, based on PHASE program analysis, association with (B) UGT1A1 protein expression levels and (C) rates of N2G formation. The influence of diplotypes on UGT1A1 protein expression and N2G formation levels is illustrated. Only significant P values for comparison between groups are shown above each graph. All groups were compared with each other with the Tukey-Kramer HSD statistical test (alpha value set to 0.05). Bars indicate mean values. aThe UGT1A1*1 allele presents 6 TA repeats, whereas the UGT1A1*28 allele presents 7 TA repeats. bN corresponds to the number of subjects in this group. NS, nonsignificant (P > .05).

For each phenotype parameter, UGT1A1 expression, E2-G, N-OH-PhIP-N2G and N3G rates, subjects were stratified according to diplotypes. Compared with individuals with the I/I pairs (n = 13), subjects with diplotypes I/II (n = 9) and II/II (n = 7) demonstrated a significantly lower expression of the UGT1A1 protein (P = .034 and P = .012, respectively) (Fig. 4B), lower rates of formation of glucuronide of the control substrate E2-3G (P = .096 and P = .0025) (data not shown), lower rates of N-OH-PhIP glucuronides at positions N2 (P = .062 and P = .032) (Fig. 4C), and N3 (P = .072 and P = .013) (data not shown). Also, II/III pairs (n = 3) demonstrated significantly lower UGT1A1 expression (P = .003), rates of formation of N-OH-PhIP-N2G (P = .003), and N3G (P = .018), compared with individuals with I/I pairs.

Discussion

N-OH-PhIP is the carcinogen of PhIP, the most abundant food-borne heterocyclic amine present in the human diet, arising specifically from commonly eaten cooked meat.6, 7, 35 Human dietary epidemiologic studies suggest that several cancers may be related to HCA intake, the evidence being strongest for colorectal cancer,4, 36, 37 whereas HCA-induced adducts in the colon and other tissues occur at dietary relevant doses.37, 38 Glucuronidation by UGT enzymes is one of the few phase II drug metabolism pathways that significantly contribute to the detoxification of HCAs in liver and extrahepatic tissues.8, 10, 11, 19, 39, 40 Large interindividual differences in the rates of PhIP N-oxidation producing the carcinogenic intermediate N-OH-PhIP have previously been observed.30, 41 However, we report here for the first time that its hepatic glucuronidation is also highly variable. The current research led to important additional findings on the influence of specific genetic polymorphisms in the UGT-mediated hepatic detoxification pathway potentially influencing the exposure to HCAs and subsequent cancer risk.

A limited number of relevant UGT enzymes were shown to be involved in the formation of 4 glucuronide derivatives of N-OH-PhIP demonstrating regioselectivity for the N-OH-PhIP molecule. Formation of N-OH-PhIP-N2G, the primary metabolite found in human urine, was shown to be predominant in all livers (average, 82% of total glucuronides formed). Our results point to UGT1A1 as the key enzyme involved in the hepatic formation of this metabolite as well as in the formation of N-OH-PhIP-N3G also found in urine at significant levels.11, 12, 42, 43

It was critical to assess whether different UGT enzymes were involved particularly in the formation of N2G and N3G. N2-glucuronides are resistant to β-glucuronidases present in the intestinal flora and are readily excreted into the urine. In contrast, N3-glucuronides can be deconjugated by bacterial β-glucuronidases back to N-OH-PhIP,44 and re-conjugation with N-acetyltransferases or sulfotransferases can possibly occur in the gastrointestinal tract to form highly reactive compounds capable of binding DNA.45, 46 As a result, liver detoxification, a predominant site for PhIP metabolism8 via N3 glucuronidation, in addition to N2G formation, would be likely to determine the exposure to HCA in the intestine and the colon. UGT1A1 is largely expressed in the liver but also in a number of extrahepatic tissues, including the intestine and the colon. In turn, UGT1A8, highly efficient at producing N-OH-PhIP-N2G and N3G, is not expressed in the liver, but its role would be critical in the gastrointestinal tract, a predominant site of its expression. Another metabolite N-OH-PhIP-Glucu 1 (average, 10% of total glucuronides), previously identified by Malfatti and Felton,19 was formed selectively by UGT1A4. A fourth glucuronide named Glucu 2, which shares a similar fragmentation profile to N2G and Glucu 1, was significant in all studied livers (average, 4% of total glucuronides) and formed by UGT2B10 and 1A9. Further analyses are required to fully characterize Glucu 1 and Glucu 2 and to assess the significance of their in vivo formation in relation to HCA exposure.

The wide range of glucuronidation activities for the formation of the N-OH-PhIP glucuronide derivatives in the liver suggested that the expression or activity of multiple N-OH-PhIP-glucuronidating enzymes differ between individuals. On in vivo metabolism studies with radiolabeled 14C N-OH-PhIP, Lang and colleagues41 concluded that there was a greater variation in the N2- than in the N3- glucuronidation pathway. Our study is in agreement with these findings and shows that N-OH-PhIP-N2G was the most variable (28-fold variation) compared with N3G (18-fold variation). The strong correlation of the hepatic N2G formation to the UGT1A1 protein content exposed UGT1A1 as an important enzyme in the metabolism of N-OH-PhIP in vivo. Conversely, the strong correlation with UGT1A1 protein content is not in favor of the expression of a UGT1A1 protein with altered enzymatic activity. This finding is in agreement with the fact that functional variations in the coding region of the UGT1A1 gene are rather uncommon in the general population,17 especially in the Caucasian population, whereas in the Asian population the coding region polymorphism (G71R) is common.47 Conversely, our data demonstrate that the variable expression of the UGT1A1 gene is strongly associated with the presence of genetic variations in its promoter.

To date, 3 polymorphisms of the UGT1A1 promoter have been reported, the TA repeat (UGT1A1*28) and the variants at positions −3156 and −3279.22–24 These polymorphisms are important for the etiology of Gilbert syndrome and side effects of the anticancer agent irinotecan.48, 49 When analyzed individually, all 3 polymorphic sites were significantly associated with N-OH-PhIP-N2G and N3G rates of formation in liver microsomes, consistent with the main role of UGT1A1 in their formation. Analyses of the haplotypic structures of the UGT1A1 promoter revealed that the UGT1A1*1, −3156G, and −3279T variations, individually linked to the highest UGT1A1 expression and rates of glucuronidation, are in strong linkage and represent the most frequent haplotype (I). The second most frequent haplotype II comprises the combination of UGT1A1*28, −3156A, and −3279G individually associated with significant decreased UGT1A1 expression and rates of glucuronidation. Individuals with these combined polymorphisms would more likely be exposed to higher levels of N-OH-PhIP. However, because of the limited number of patients (n = 48), this investigation does not provide sufficient power to identify the causal polymorphism or to evaluate whether they act in synergy to produce the observed phenotype. In fact, functional in vitro studies only provide limited information on the role of each polymorphism on the expression of the UGT1A1 gene because only the UGT1A1*28 allele was studied. The *28 allele, associated with a significant decreased expression, is clearly a contributing factor to the UGT1A1 basal expression.22, 29, 33 The presence of the −3279 G>T polymorphism in the PBREM region of UGT1A1 would suggest an influence on inducible transcription activity, whereas the effect on gene expression is still unknown for the −3156G>A polymorphism. Finally, whether polymorphisms in UGT1A4, 1A8, 1A9, and 2B10 genes have any influence on the rates of formation of N-OH-PhIP-N3G, Glucu 1, and Glucu 2 remains to be determined.

In conclusion, this study demonstrates the formation of 4 glucuronide derivatives of the carcinogen N-OH-PhIP and shows the wide interindividual variability in its hepatic glucuronidation. A limited number of key UGT enzymes are involved and demonstrate regioselectivity. UGT1A1 plays a key role in the hepatic detoxification of N-OH-PhIP through N2 and N3 glucuronidation and most likely in other tissues such as the gastrointestinal tract along with other enzymes such as UGT1A8. A previous study suggested that the relatively high concentration of N-OH-PhIP-N2-glucuronide and the fact that it is an indicator of bioactivation make this metabolite a potential biomarker for PhIP exposure and activation.11 The genotype/phenotype association for the three most common polymorphisms of the UGT1A1 gene demonstrates that these genetic factors modulate the formation of this biomarker, N-OH-PhIP-N2G. Therefore, the UGT1A1 status is most likely to influence the level of exposure to N-OH-PhIP and potentially the risk of cancer through dietary exposure to HCAs. Consistent with this hypothesis, Peters and coworkers50 recently showed in a controlled feeding study, that the UGT1A1*28 allele modifies the effect of intake of meat cooked at high temperature, generating HCAs, on urinary mutagenicity (a biological measurement of exposure). Given the strong biological plausibility for the role of UGT1A1 in the etiology of cancer, epidemiologic studies in population-based samples with detailed exposure assessment are needed to fully explore this hypothesis and have recently been initiated.

Acknowledgements

The authors thank Virginie Bocher for critical reading of the manuscript.

Ancillary