The Pharmacogenomics Journal (2008) 8, 152–161 & 2008 Nature Publishing Group All rights reserved 1470-269X/08 $30.00 www.nature.com/tpj ORIGINAL ARTICLE

Hepatocyte nuclear factor-1 alpha is associated with UGT1A1, UGT1A9 and UGT2B7 mRNA expression in human

J Ramı´rez1, S Mirkov1, Experimental evidence suggests HNF1a regulates UGT expression. This study 1 2 2,3 investigates (1) whether the variability in HNF1a expression is associated with W Zhang , P Chen , S Das , the variability in UGT1A1, UGT1A9 and UGT2B7 expression in human 1 1,3,4 W Liu , MJ Ratain and (2) the functionality of 12 HNF1a variants using mRNA expression as and F Innocenti1,3,4 phenotype. Controlling for known UGT variation in cis-acting elements known to affect UGT expression, we demonstrate that a combination of 1Department of Medicine, University of Chicago, HNF1a mRNA levels and UGT genotype predicts variance in UGT expression 2 Chicago, IL, USA; Department of Human to a higher extent than UGT genotype alone. None of the HNF1a Genetics, University of Chicago, Chicago, IL, USA; 3Committee on Clinical Pharmacology and polymorphisms studied, however, seem to have an effect on HNF1a, Pharmacogenomics, University of Chicago, UGT1A1, UGT1A9 and UGT2B7 expression, ruling out their functional role. Chicago, IL, USA and 4Cancer Research Center, Our data provide evidence for HNF1a being a determinant of UGT1A1, University of Chicago, Chicago, IL, USA UGT1A9 and UGT2B7 mRNA expression. However, the amount of UGT a Correspondence: intergenotype variability explained by HNF1 expression appears to be Dr F Innocenti, Section of Hematology- modest, and further studies should investigate the role of multiple Oncology, Department of Medicine, University factors. of Chicago, 5841 S, Maryland Avenue, The Pharmacogenomics Journal (2008) 8, 152–161; doi:10.1038/sj.tpj.6500454; MC2115, Chicago, IL 60637, USA; E-mail: [email protected] published online 17 April 2007

Keywords: HNF1a; UGT1A1; UGT1A9; UGT2B7; regulation

Introduction

Glucuronidation by the uridine diphosphate- (UGT) is a major metabolic pathway that facilitates the elimination of a large variety of molecules into urine and bile by increasing their water solubility. The UGT1 gene codes for nine functional UGT1A enzymes (UGT1A1, UGT1A3, UGT1A4, UGT1A5, UGT1A6, UGT1A7, UGT1A8, UGT1A9 and UGT1A10) generated through a process of RNA splicing of one of the multiple first to the set of common exons 2–5. The UGT2 family is encoded by separate homologous and is subdivided into two subfamilies: UGT2A (encoding UGT2A1) and UGT2B (encoding UGT2B4, UGT2B7, UGT2B10, UGT2B11, UGT2B15, UGT2B17 and UGT2B28). Most UGTs are expressed in the liver but they can also be found in extrahepatic tissues. Genetic polymorphisms in UGTs have been associated with interindividual variability in glucuronidation. The genetic variation in UGT1A1 is the most extensively studied. UGT1A1 gene transcriptional activity is affected by the length of the TATA box in the promoter, where a larger number of TA repeats Received 26 November 2006; revised 18 February 2007; accepted 15 March 2007; results in reduced transcriptional activity. Homozygosity for the (TA)7 allele published online 17 April 2007 (UGT1A1*28) is associated with Gilbert’s syndrome (a mild form of conjugated HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 153

hyperbilirubinemia)1,2 and predisposition to severe toxicity might influence the mRNA expression of HNF1a and the of the anticancer drug irinotecan.3,4 Even though the effect UGTs of interest. of the UGT1A1*28 allele on the UGT1A1 glucuronidation rates in humans is well known, the phenotypic variability Results within each genotypic group remains high.4,5 Similar to UGT1A1, the UGT1A9 and UGT2B7 isoforms are highly Effect of HNF1a genotypes on HNF1a expression expressed in the liver and play an important role in The median (not log-transformed) hepatic mRNA level of the hepatic glucuronidation of numerous drugs. Multiple HNF1a was 254 (range 30–944, n ¼ 53). The median studies have been conducted to evaluate the functional mRNA level of 18S was 1.18 (range 0.18–1.84, n ¼ 54). The consequences of common polymorphisms in UGT1A9 and frequencies of the HNF1a variants are listed in Table 1. All UGT2B7, although the results to date have been incon- alleles were in Hardy–Weinberg equilibrium, and their sistent.5–10 It can be hypothesized that the phenotypic frequencies were comparable to the data in the HapMap variability in UGTs that remains unaccounted for by for the CEPH Caucasians. None of the HNF1a genotypes had cis-acting genetic polymorphisms resides in the variability a significant effect (P40.1) on the expression of HNF1a. in transcription factors. Initiation of transcription is an integrated and complex mechanism involving both Effect of UGT genotypes on UGT expression cis-acting elements and trans-acting factors. Transcription The median (not log-transformed) mRNA levels of UGT1A1, factors involved in the regulation of UGT constitutive UGT1A9 and UGT2B7 were 759 (range 71–15,712, n ¼ 54), expression include the hepatocyte nuclear factor 1 (HNF1), 520 (range 6–14,303, n ¼ 44) and 1263 (range 17–17.245, the octamer 1 (Oct-1), the pre-B-cell n ¼ 54), respectively. Analysis of the linear relationship factor 2 (Pbx2), the caudal-related homeo- between the mRNA levels of the UGTs showed that the domain 2 (Cdx2, an intestine-specific transcription UGT1A1 mRNA levels were significantly and moderately factor) and Prep1.11–16 UGT regulation by hormones and related to those of UGT1A9 (r2 ¼ 0.49, Po0.0001, n ¼ 44; xenobiotics is mediated through the aryl hydrocarbon Figure 1a) and UGT2B7 (r2 ¼ 0.39, Po0.0001, n ¼ 54; Figure (AhR), the constitutive (CAR), 1b). Similarly, the mRNA content of UGT1A9 and UGT2B7 the (PXR), the , were significantly correlated (r2 ¼ 0.54, Po0.0001, n ¼ 44; the peroxisome proliferated-activated receptors and trans- Figure 1c). Similar results were obtained in partial correla- cription factors that respond to stress.17 tions performed to adjust for a potential confounding effect The HNF1 family is comprised of HNF1a and HNF1b, of 18S mRNA levels (data not shown). which increase gene transcription by binding as homo- The genotype frequencies of the UGT variants are shown dimers or heterodimers to DNA sequences in target genes. in Table 1. The alleles were in Hardy–Weinberg equilibrium. HNF1a is well conserved among species and is the pre- The association between UGT1A1*28 and UGT1A1 expres- dominant form expressed in the human liver.18 Experimen- sion was best described by an additive model (r2 ¼ 0.17, tal evidence suggests a prominent role of HNF1a as a P ¼ 0.003, n ¼ 51; 6/6 ¼ 0, 6/7 ¼ 1, 7/7 ¼ 2; Figure 2). A trans-acting factor regulating UGT expression. HNF1a binds recessive model (r2 ¼ 0.08, P ¼ 0.07, n ¼ 43) described the and activates the proximal promoters of several human relationship between UGT1A9*1b genotype and UGT1A9 UGTs, including UGT1A1, UGT1A9 and UGT2B7.11,13,15,16,19 expression (10/10: mean7s.d. ¼ 1.8870.92, n ¼ 6; 9/9 HNF1a is polymorphic. The I27L (79A4C) variant, a and 9/10: mean7s.d. ¼ 2.5870.83, n ¼ 37). The 985A4G common missense polymorphism in HNF1a, is located in the dimerization domain of the protein.20 79A4Cis associated with insulin resistance and glucose intolerance Table 1 Genotype frequencies as well as with differences in high-density lipoprotein cholesterol among individuals.20–22 The molecular function Gene Variant Frequency of this variant has not been established but preliminary studies showed that luciferase UGT2B17 activity was reduced HNF1a 79A4C 0.25 after cells were co-transfected with HNF1a 79A4C compared rs1169286 A/G 0.33 to constructs without this variant.14 Additionally, HNF1a rs11065385 A/G 0.74 rare variants have been associated with maturity onset rs2393791 G/A 0.63 rs1169293 G/A 0.88 of the young (MODY3).23 The effect of HNF1a rs12427353 G/C 0.22 genetic variation on its mRNA expression in human liver, as rs2071190 T/A 0.26 well as its association with the expression of UGTs known to rs1169302 T/G 0.43 be regulated by HNF1a, has never been studied. rs1169303 A/C 0.54 This study had two aims: (1) to investigate whether HNF1a rs3999413 C/T 0.14 mRNA expression levels predict mRNA expression level rs1169306 C/T 0.37 variability in UGT1A1, UGT2B7 and UGT1A9 while rs1169307 T/C 0.62 taking into account the known effect of cis-acting variants UGT1A1 À53(TA)647 0.38 on UGT expression and (2) to determine whether any of UGT1A9 À118T9410 0.38 the 12 HNF1a polymorphisms typed in normal livers UGT2B7 985A4G 0.13

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 154

ab c r2 = 0.49, p<0.0001, n = 44 r2 = 0.39, p<0.0001, n = 54 r2 = 0.54, p<0.0001, n = 44 4 4 4 mRNA) mRNA) mRNA) 18S 18S 18S 3 3 3 mRNA / mRNA mRNA / mRNA mRNA / mRNA 2

2 2 UGT1A1 UGT1A9 UGT1A1 1 log ( log ( log ( 0.5 1.5 2.5 3.5 4.5 12345 12345 log (UGT1A9 mRNA /18S mRNA) log (UGT2B7 mRNA /18S mRNA) log (UGT2B7 mRNA /18S mRNA)

Figure 1 Association between the hepatic mRNA levels of (a) UGT1A1 and UGT1A9,(b) UGT1A1 and UGT2B7, and (c) UGT1A9 and UGT2B7. Data points represent means of triplicate determinations from a single experiment.

4.5 associated with UGT1A1 expression in a recessive model r2 = 0.17, p = 0.003, n = 51 (r2 ¼ 0.06, P ¼ 0.09, n ¼ 52; A/A and A/G: mean7s.d. ¼ mRNA) 2.9070.56, n ¼ 46; G/G: mean7s.d. ¼ 3.0170.78, n ¼ 6).

18S 3.5 Linear regression of UGT expression versus gender and history of intake of UGT inducers mRNA / mRNA 2.5 Higher mean mRNA levels were observed in females (n ¼ 13) compared with males (n ¼ 41) for both UGT2B7 (r2 ¼ 0.10, 2

UGT1A1 P ¼ 0.02) and HNF1a (r ¼ 0.07, P ¼ 0.05). No significant 1.5 gender differences were observed for UGT1A1 and UGT1A9. log ( 6/6 6/7 7/7 History of medications and alcohol intake was available for UGT1A1*28 31 donors. Ten of these individuals received the UGT inducers phenobarbital and dexamethasone; nine donors Figure 2 Association between UGT1A1*28 and UGT1A1 mRNA levels. were alcohol consumers. Smoking history was available The model is UGT1A1 mRNA ¼À0.35 * UGT1A*28 (additive) þ 3.20. for 43 individuals, 15 of which were smokers. No association Data points represent means of triplicate determinations from a single experiment. was found between intake of any of these potential UGT inducers and the mRNA levels of any of the four genes (P40.05). UGT2B7 variant had a dominant effect (r2 ¼ 0.09, P ¼ 0.03, 7 7 n ¼ 52) on UGT2B7 expression (A/A: mean s.d. ¼ 2.97 Multivariate analysis of UGT1A1, UGT1A9 and UGT2B7 mRNA 7 7 0.61, n ¼ 39; A/G and G/G: mean s.d. ¼ 3.37 0.42, n ¼ 13). expression Multivariate models were designed to identify covariates Association between HNF1a expression/genotypes and UGT associated with the variability in UGT mRNA expression. expression The best models with the highest r2 and with all indepen- A modest and significant degree of correlation was observed dent variables having a Po0.05 are described on Table 2. between the mRNA levels of HNF1a and those of UGT1A1 UGT1A1 expression was best predicted by a combination (r2 ¼ 0.14, P ¼ 0.006, n ¼ 53; Figure 3a), UGT1A9 (r2 ¼ 0.25, of HNF1a expression and UGT1A1*28 genotype (additive P ¼ 0.0006, n ¼ 44; Figure 3b) and UGT2B7 (r2 ¼ 0.18, model: 6/6 ¼ 0, 6/7 ¼ 1, 7/7 ¼ 2). UGT1A9 expression could P ¼ 0.001, n ¼ 53; Figure 3c). Partial correlation analysis be best predicted by a combination of HNF1a expression and adjusting for 18S mRNA levels showed similar results (data UGT1A9*1b genotype (recessive model: 9/9 and 9/10 ¼ 0, not shown). 10/10 ¼ 1). UGT2B7 expression can be best explained by a We tested whether the HNF1a single nucleotide poly- combination of HNF1a expression and the UGT2B7 morphisms (SNPs) have an effect on UGT expression. 985A4G polymorphism (dominant model: A/A ¼ 0, A/G Significant although very weak associations were observed and G/G ¼ 1). The percent of variance in UGT mRNA between two HNF1a variants and UGT1A1 expression. The expression predicted by a combination of HNF1a mRNA HNF1a 79A4C variant was associated with UGT1A1 expres- expression and UGT genotype increased from 17% to 30%, sion in a dominant model (r2 ¼ 0.08, P ¼ 0.04, n ¼ 51; A/A: 8% to 33%, and 9% to 26% for UGT1A1, UGT1A9 and mean7s.d. ¼ 2.7770.59, n ¼ 29; A/C and C/C: mean7s.d. ¼ UGT2B7, respectively, when compared with the variance 3.1070.54, n ¼ 22). The rs1169286A4G HNF1a SNP was predicted by UGT genotype alone. None of the HNF1a

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 155

abc 4.5 5 r2 = 0.14, p = 0.006, n = 53 r2 = 0.25, p = 0.0006, n = 44 r2 = 0.18, p = 0.001, n = 53 4.0 4 mRNA) mRNA) mRNA) 4

18S 3.5 18S 18S 3 3.0 3 mRNA / mRNA mRNA / mRNA mRNA / mRNA 2 2.5 2 2.0 UGT1A1 UGT2B7

UGT1A9 1

log ( 1.5

log ( 1 log ( 1.5 2.0 2.5 3.0 1.5 2.0 2.5 3.0 1.5 2.0 2.5 3.0 Iog (HNF1α mRNA /18S mRNA) Iog (HNF1α mRNA /18S mRNA) Iog (HNF1α mRNA /18S mRNA)

Figure 3 Association between the hepatic mRNA levels of HNF1a and (a) UGT1A1,(b) UGT1A9 and (c) UGT2B7. Data points represent means of triplicate determinations from a single experiment.

Table 2 Multivariate models describing the combined effect of HNF1a mRNA expression and UGT genotype on UGT mRNA expression

UGT mRNA Model r2 Overall P-value

UGT1A1 0.80*HNF1a mRNA (P ¼ 0.003)À0.34*UGT1A1 genotype (additive) (P ¼ 0.002) 0.30 0.0002 UGT1A9 1.74*HNF1a mRNA (P ¼ 0.0004)+0.71*UGT1A9 genotype (recessive) (P ¼ 0.03) 0.33 0.0003 UGT2B7 0.90*HNF1a mRNA (P ¼ 0.001)+0.37*UGT2B7 genotype (dominant) (P ¼ 0.03) 0.26 0.0007

variants were entered into the multivariate models as they measure protein expression and we cannot assume that a were not significant at the P ¼ 0.10 level in the univariate high level of correlation exists between mRNA and protein analyses. Multivariate models including gender and UGT levels, the data from the literature are in support of a good genotypes (without HNF1a expression) were also built, as we correlation in UGT genes. For example, a strong correlation hypothesized that the gender-related differences observed between UGT1A1 protein level and both UGT1A1 activity in UGT expression might be due to the effect of gender on and mRNA level (r2 ¼ 0.82 and 0.72, respectively) has been HNF1a expression. The only significant model was UGT2B7 observed.25 Similar correlations between mRNA and activity mRNA ¼À0.42*gender (male) (P ¼ 0.02) þ 0.37*UGT2B7 were also reported in another study for both UGT1A1 genotype (dominant) (P ¼ 0.04) (overall model: r2 ¼ 0.18, (r2 ¼ 0.73) and UGT1A9 (r2 ¼ 0.59).26 This also holds true for P ¼ 0.008). UGT1A6 (r2 ¼ 0.53 between mRNA and activity).27 Besides clinical association studies, there are no data on the molecular function of the human variation of HNF1a. Discussion Our study is the first one to investigate whether several HNF1a variants might have functional consequences on We present evidence showing that the interindividual . The HNF1a 79A4C variant does not seem variability in UGT1A1, UGT1A9 and UGT2B7 is partly to be associated with the phenotypic variability of HNF1a, related to variation in HNF1a expression. Controlling for UGT1A1, UGT1A9 and UGT2B7 mRNA expression although known UGT variation in cis-acting elements known to affect our sample set was limited to only four samples homo- expression, we demonstrate that a combination of UGT zygous for the 79C allele. The weak correlation observed genotype and HNF1a mRNA expression predicts variance in between HNF1a 79A4C and UGT1A1 expression might be UGT expression to a higher extent than UGT genotype spurious, as it is not consistent with previous association alone. The correlation observed between the mRNA expres- studies showing decreased pancreatic function20–22 and sion of HNF1a and UGT2B7 is in agreement with a previous reduced luciferase activity.14 The tagging SNPs investigated study showing a high correlation between the expression did not have a functional effect on the gene expression levels of these two genes in 12 human livers of Caucasian of any of the four genes. It is unknown whether these origin.24 Although the strength of the correlation is not polymorphisms have a functional effect. A study investigat- high, our results, obtained in a liver collection larger than ing whether tagging SNPs of HNF1a (including rs2071190 that of Toide et al.,24 are in agreement with literature and rs3999413) were associated with type II diabetes showed indicating that HNF1a can bind and activate promoters of negative results.28 In addition, our study does not rule out UGT1A1, UGT1A9 and UGT2B7.11,13,19 Although we did not the functional significance of the tagging SNPs of the HNF1a

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 156

gene, as their effect on gene splicing and protein stability expression is in turn regulated by several other trans-acting has not been investigated. This is of particular importance elements. It has been demonstrated that HNF1a-mediated due to the existence of splicing variants of the gene with activation of the UGT2B7 promoter is enhanced by the modified protein activity compared with the reference ubiquitous transcription factor Oct-1.11 The transcription sequence protein of HNF1a.29 Further studies are needed to factors Pbx2-Pbx1 can downregulate activation of UGT2B17 elucidate the effects of these variants on protein expression, by HNF1a by restricting access of HNF1a dimers to the DNA. activity, dimerization, DNA binding properties as well as on Coactivators of HNF1a include CBP/p300, P/CAF, SRC-1 and constitutive/ of the HNF1a mRNA. .45,46 The GATA family of proteins is expressed in a In our study, common polymorphisms in UGT1A1, wide range of tissues and has been demonstrated to interact UGT1A9 and UGT2B7 affected their mRNA expression to with HNF1a in vitro and in vivo.47 HNF4a has also been different extents. There are numerous reports describing the shown to regulate transcription of UGT1A9 and HNF1a in functionality of UGT1A1*28, which reduces glucuronidation the liver, and both HNF4a and HNF1a seem to be co- of , SN-38 and benzo(a)pyrene-trans-7R,8R-dihydro- regulated.48–51 It is likely that the action of HNF1a is finely diol.1,4,30–33 However, a correlation with UGT1A1 mRNA coordinated with that of other factors, and that UGT expression in the liver had not been reported previously. expression is the ultimate result of the interaction among In luciferase studies, a variable promoter region repeat multiple regulatory elements. Network-based approaches

(À118T9410, UGT1A9*1b)inUGT1A9 was reported to and more comprehensive studies of other transcription correlate with modest increases (1.4- to 2.6-fold) in protein factors may clarify the contribution of each transcription expression,10,34 although no significant effect have been factor in the regulation of UGT expression. observed at the protein level.9,10 We found decreased UGT1A9 mRNA expression in individuals with the Materials and methods À118T10/10 genotype. This suggests that the variant might negatively affect UGT1A9 transcription, although the ob- Human liver samples served effect did not reach statistical significance. These DNA and RNA samples from 54 Caucasian donors were results are in agreement with the decreased glucuronidation processed through Dr Mary Relling’s Laboratory at St Jude activity previously observed in individuals with the Children’s Research Hospital (Memphis, TN, USA). The 5 À118(T)10 allele observed in this same set of livers. In hepatic tissue was provided by the Liver Tissue Procurement addition to the effect of UGT genetic variants on gene and Distribution System (funded by #NO1-DK-9–2310) and expression, we detected a correlation among UGT1A1, by the Cooperative Human Tissue Network. Donor livers UGT1A9 and UGT2B7 mRNA levels, which was expected as were procured for possible transplant with approval of they are likely to share common regulatory pathways in the human subjects. Organs were flushed, in situ, with a cold liver. storage solution such as University of Wisconsin solution. As phenobarbital, dexamethasone, alcohol and nicotine Livers were placed on wet ice and sent to isolation lab. Total are inducers of glucuronidation,35–40 and gender may play a cold ischemic time was usually less than 24 h. Pieces of liver role in interindividual differences in drug metabolism,41 we tissue not used for cell isolation were snap frozen in liquid tested whether these variables were significant predictors of nitrogen and stored at À80 1 until used for DNA and RNA mRNA expression. A detailed drug history was not available isolation. To insure that the liver tissue was viable, we relied for all samples, as this information is not always adequately on the viability of the cells isolated from a different portion collected from liver donors. Although gender differences of the same organ and the integrity of the RNA. All livers were observed in the expression of UGT2B7 and HNF1a, were treated identically. replacement of HNF1a expression by gender in the UGT2B7 multivariate model decreased the level of correlation, Real-time PCR indicating that HNF1a is a better predictor of UGT mRNA UGT and HNF1a mRNA levels were measured by two-step expression than gender. real-time PCR using the Mx3000P system (Stratagene, Cedar Our results suggest that HNF1a is a determinant of Creek, TX, USA). Integrity of the RNA samples was verified UGT1A1, UGT1A9 and UGT2B7 hepatic expression, and by agarose gel electrophoresis. cDNA was synthesized in contributes to their phenotypic variability. Nevertheless, 20 ml reaction volumes using the iScript cDNA Synthesis Kit there is still residual phenotypic variability in UGT expres- (Bio-Rad Laboratories, Hercules, CA, USA) and 2 mg of total sion that is unaccounted for by any of the factors RNA. The thermal profile was as follows: 251C for 5 min, investigated in this study, and further studies should 421C for 30 min, and 851C for 5 min. 18S was used as the investigate the role of multiple transcription factors. Other control gene. 18S cDNA was prepared using the high- transcription factors known to regulate UGT expression capacity cDNA Archive Kit (Applied Biosystems, Foster City, independently from HNF1a should be tested in a similar CA, USA) and following the manufacturer’s instructions. model. For UGT1A1, these proteins include AP1, AP3, SP1, The thermal profile consisted of 251C for 10 min and 371C CEBPA, TITF1, TBP and CREB1 in the 600 bp 50-flanking for 2 h. All 54 cDNA samples were synthesized in a single 1 area,19,42 and CAR, PXR and AhR in the PBREM experiment, and stored at –801C until further analysis. cluster.43,44 To add to the complexity, intricate regulatory Real-time PCR reactions were performed using IQSYBR interactions exist among several transcription factors, whose Green Supermix (Bio-Rad Laboratories). Briefly, cDNA was

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 157

amplified in 25 ml of reaction buffer containing IQSYBR was mixed with 1 U of shrimp alkaline phosphatase and 1 U Green Supermix, 0.5 mM of specific primers and nuclease-free of exonuclease I in 1 Â of shrimp alkaline phosphatase water. The oligonucleotide sequences of the primers used are buffer, and incubated at 371C for 45 min. Inactivation of the shown in Table 3. Reactions were performed in triplicate and enzymes was done by heating at 951C for 15 min. The SBE included standard curves for both the target and control reaction (10 ml) contained 1 mM of extension primer (50- genes. After preheating (hot start reaction) at 951C for GCTCACCCAGTGCCTGGA-30), 250 mM of each ddNTP, 6 ml 10 min, real-time PCR amplifications of the target genes of purified PCR product and 1.25 U of ThermoSequenase. were performed as follows: 40 cycles with melting at 951C The reaction was run under the following conditions: 961C for 30 s, annealing at 551C for 1 min and extension at 721C for 2 min, followed by 60 cycles at 961C for 30 s, 551C for 30 s for 30 s. Data were collected as end points at 551C. Real-time and 601C for 30 s. Separation of the SBE products was PCR amplifications for 18S were performed as follows: performed after denaturation of samples (at 961C for 4 min) preheating (hot start reaction) at 951C for 10 min, 40 cycles on a WAVE 3500HT DHPLC system (Transgenomic Inc., with melting at 951C for 15 s, and annealing and extension Omaha, NE, USA) at 701C. The gradient used for elution of at 601C for 1 min. Data were collected as end points at 601C. the SBE products from the DNASep column (Transgenomic A disassociation curve was used to confirm the specificity of Inc.) was created by the software based on the length of the the PCR product. The thermal profile for the disassociation extended product. The mobile phase consisted of 0.1 M curve was 951C for 1 min to denature the DNA, followed by triethylammonium acetate (A) and 25/75 acetonitrile/0.1 M 551C for 30 s. A ramp up to 951C was then applied. Data triethylammonium acetate (B). The gradient changed from collection was performed during the ramp. Initial template 74.9% A and 25.1% B to 64.9% A and 35.1% B over 2 min. quantities (ng/reaction) were calculated using threshold The four bases at the 30-end of the extended products were cycle (Ct) values and a standard curve. eluted according to hydrophobicity differences of the order CoGoToA. Genotyping The HNF1a tagging SNPs are listed in Table 4. They were Genotyping of HNF1a 79A4C was performed by single base selected using HapMap data publicly available (CEPH extension (SBE) and denaturing high-performance liquid Caucasian population, r2 ¼ 0.8, minor allele frequency of chromatography (DHPLC). Hepatic genomic DNA was 0.1) and the algorithm Tagger-pairwise tagging. Samples amplified by PCR using a GeneAmp PCR System 9600 were genotyped by PCR and SBE-DHPLC. Some PCR Thermal Cycler (Applied Biosystems). The amplification amplicons contained two or three tagging SNPs while primer sequences were 50-TTTCTAAACTGAGCCAGC-30 (for- other SNPs were amplified in a duplex PCR reaction. The SBE ward) and 50-GTCTCCCCCAGCCCATTG-30 (reverse). PCRs reactions were done either by individual or duplex reaction. were set up in a 12.5 ml volume containing 125 nM of each Some SBE products were mixed after the SBE reaction and primer, 25 ng of genomic DNA, 2.5 mM of MgCl2,50mmof run together in a duplex format on DHPLC (see Table 5). The each dNTP and 0.3 U of AmpliTag Gold polymerase (Applied primer sequences used for PCR and SBE in each genotyping Biosystems) in the buffer provided by the manufacturer. assay are listed in Table 5. Each primer was designed using Reactions were denatured initially at 951C for 15 min then the Oligo 4.0 software with the genomic sequence as- followed by 40 cycles of 951C for 15 s, 581C for 15 s and 721C sembled from the RefSeq Gene NM_000545.3 in the UCSC for 45 s, and a final extension at 721C for 10 min. To remove genome browser. The primers were checked for specificity unincorporated dNTPs and primers, the PCR product (10 ml) using BLAT in the UCSC genome browser and BLAST in the NCBI database. Each assay was also verified with the known genotype controls selected from HapMap samples. PCRs Table 3 The oligonucleotide sequences of primers used for were performed in a 15 ml volume containing 125 nM of each real-time PCR

Gene Primer Sequence for real-time PCR Table 4 HNF1a tagging SNPs

UGT1A1 Forward 50-AACAAGGAGCTCATGGCCTCC-30 Tag SNP rs# Alleles Exon/intron Reverse 50-CCACAATTCCATGTTCTCCAG-30 location

UGT1A9 Forward 50-GAGGAACATTTATTATGCCACCG-30 (1) rs1169286 A/G Intron 1 Reverse 50-TGCCCAAAGCATCAGCAATT-30 (2) rs11065385 A/G Intron 1 (3) rs2393791 G/A Intron 1 UGT2B7 Forward 50-CAGCTTCTCTCCTGGCTACACTT-30 (4) rs1169293 G/A Intron 1 Reverse 50-CAGGAGTTTCGAATAAGCCATAC-30 (5) rs12427353 G/C Intron 2 (6) rs2071190 T/A Intron 2 HNF1a Forward 50-TACACCTGGTACGTCCGCAA-30 (7) rs1169302 T/G Intron 4 Reverse 50-CACTTGAAACGGTTCCTCCG-30 (8) rs1169303 A/C Intron 7 (9) rs3999413 C/T Intron 9 18S Forward 50-CGATGCTCTTAGCTGAGTGT-30 (10) rs1169306 C/T Intron 9 Reverse 50-GGTCCAAGAATTTCACCTCT-30 (11) rs1169307 T/C Intron 9

The Pharmacogenomics Journal 158 h hraoeoisJournal Pharmacogenomics The

Table 5 PCR and SBE primers for genotyping the HNF1a tagging SNPs

PCR primer Sequence Amplicon PCR Ta SBE primer Sequence a Notes size (1C)

0 0 rs1169286 PCR-F 5 -AGGAAAGGAGCCAGGGAGAG 342 bp 60 rs1169286 ext-F22 5 -CAGAATGTCTAGTAGAAGCTGT Duplex PCR and individual SBE HNF1 for rs1169286 and rs1169303 a rs1169286 PCR-R 50-GGAGGGGAACGAAACCAAGG of regulation HNF1a_892 bp PCR-F 50-ACCATACCCAGTTTACAATTCAG 892 bp 62 rs11065385 ext-R 50-AATCTCCTGGGCTGAATTGGT The amplicon-892 bp contains rs11065385 and rs2393791 SNPs; the two SBEs were mixed on DHPLC UGT1A1 HNF1a_892 bp PCR-R 50-CCAGGTGGAGCAGAGTTTATC rs2393791 ext-F 50-CAGTTCCTTGCGTGCCCTC

0 0 ,

HNF1a_528 bp PCR-F 5 -TCCCTGTGTCCTGGCATAAA 528 bp 58 rs1169293 ext-R 5 -GTGCTGGAACCTTGTCAAC The amplicon-528 bp contains UGT1A9 rs1169293 and rs12427353 SNPs; the SBE was done Ramı J and individually UGT2B7 ´ HNF1a_528 bp PCR-R 50-AGGTTGAATCCCACTGACTT rs12427353 ext-F 50-ACTCTCCACTAACTGATAGGT rez (Ta 501C for SBE) al et rs2071190 PCR-F 50-GATGATTTCTAAGTTCTAGCTG 189 bp 58 rs2071190 ext-R 50-CGCCACTCTCAGCTTCCC Duplex PCR for rs2071190 and rs1169302, and mixed two SBEs on DHPLC rs2071190 PCR-R 50-ATCACCTGTGGGCTCTTCAA rs1169302 PCR-F 50-CAGTAAGGTCCACGGTAAGT 250 bp 58 rs1169302 ext-R 50-AATGGAATGGAACCAAACTGA Duplex PCR for rs2071190 and rs1169302, and mixed two SBEs on DHPLC rs1169302 PCR-R 50-TGGAATGGGGTTAATTGTGGT rs1169303 PCR-F 50-AAACACTCATAACCTCATACCA 273 bp 60 rs1169303 ext-R 50-ACACGCATACACTCTCAACCA Duplex PCR and individual SBE for rs1169286 and rs1169303 rs1169303 PCR-R 50-CCTCAGGAGACACATAGACC HNF1a_761 bp PCR-F 50-ATTCATATTCCCATATAACAGTG 716 bp 58 (2 mM rs3999413 ext-R 50-CTGAGCTGACCGCCCCAC The amplicon-761 bp contains MgCl2) rs3999413, rs1169306 and rs1169307; duplex SBE for rs3999413 and rs1169306; individual SBE for rs1169307 HNF1a_761 bp PCR-R 50-GGTGAGATGCCGAGGTAGGT rs1169306 ext-F 50-TCTCCAGGAACACAGAACCAA rs1169307 ext-F 50-AAATCTGTTGCTGGCAGCTC

Abbreviations: DHPLC, denaturing high-performance liquid chromatography; SBE, single base extension. aUnderlined base refers to the modified base to minimize 30-end primer–dimers and hairpins. HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 159

primer, 2.5 mM MgCl2 (except one amplicon which was Thermo Sequenase (GE Healthcare Bio-Sciences Corp.) in a adjusted to 2 mM MgCl2), 50 or 100 mm of each dNTP, 10 ml volume, and denatured initially at 961C for 2 min 0.375 U of AmpliTaq Gold polymerase (Applied Biosystems) followed by cycling at 961C for 30 s, 501C for 30 s and 601C and 30 ng of genomic DNA. Amplification was performed in for 30 s for 60 cycles. Samples were denatured at 961C for a GeneAmp PCR System 9600 thermal cycler (Applied 4 min before separation of the extension products. Exten- Biosystems) with an initial denaturing step at 951C for sion products were separated using the WAVE 3500HT 15 min followed by 40 cycles of 951C for 15 s, appropriate DHPLC system (Transgenomic Inc.). The mobile phase annealing temperature (Ta) for 15 s and 721C for 30 s or consisted of 0.1 M triethylammonium acetate (A) and 25/ 1 min, and a final extension step at 721C for 10 min. The Ta 75 acetonitrile/0.1 M triethylammonium acetate (B). Separa- for the different amplicons is listed in Table 5. Some tion on the column was performed at 701C using a gradient conditions in the PCR reaction were adjusted due to the of 76% A and 24%, to 62.5% A and 37.5% B over 2.7 min. amplicon size or a duplex reaction. PCR products were The samples had been previously genotyped for UGT1A1 purified by treatment with shrimp alkaline phosphatase and À53(TA)647 (UGT1A1*28) and UGT1A9 À118T9410 (UG- exonuclease I at 371C for 45 min before SBE reactions. SBE T1A9*1b).5,31 reactions were performed in a 12 ml volume containing 1 mM of SBE primer, 250 mM each of four ddNTPs, 1.5 U Statistics of ThermoSequenase (GE Healthcare Bio-Sciences Corp., The mRNA levels were expressed as the ratio of the target Piscataway, NJ, USA) and 7.2 ml of purified PCR products. For gene to the control gene, 18S. The coefficients of variation the duplex SBE reaction, 1 mM of each extension primer was in the mRNA levels derived from triplicate determinations added to the total volume. Reactions were run in a 9600 were within 21%. Ten out of 54 samples were excluded from Thermal Cycler under the following conditions: 91C for analysis of UGT1A9 mRNA levels due to insufficient amount 2 min, followed by 60 cycles of 961C for 30 s, 551C for 30 s of mRNA. Quantification of HNF1a mRNA levels failed in and 601C for 30 s (except one SBE was at Ta 501C for one sample. Frequency distribution analyses and the rs12427353). Samples were denatured at 961C for 4 min and Kolmogorov–Smirnov (KS) normality test showed that the held at 41C before separation of the SBE products on the distribution of the mRNA levels of the UGTs was not normal WAVE 3500HT DHPLC system (Transgenomic Inc.). Nine (UGT1A1 KS distance ¼ 0.23, P ¼ 0.006, n ¼ 54; UGT1A9 KS microliters of SBE products of each sample or 18 ml of two distance ¼ 0.29, P ¼ 0.001, n ¼ 44; UGT2B7 KS dis- mixed SBE products per sample were injected onto the tance ¼ 0.25, P ¼ 0.003, n ¼ 54). The HNF1a mRNA levels DHPLC for analysis. The mobile phase consisted of 0.1 M were normally distributed (KS distance ¼ 0.11, P40.10, triethylammonium acetate (A) and 25/75 acetonitrile/0.1 M n ¼ 53). To normalize the data, the mRNA levels (including triethylammonium acetate (B). Samples were run on a high HNF1a, for consistency) were log-transformed before statis- throughput column (Transgenomic Inc.) with an oven tical analysis. Log-transformed data passed the normality temperature of 701C using a start gradient of 21.8 %B or test (P40.10). 24.1 %B for 3 min (slope-5%B per min). The start gradient Univariate linear regressions were used to select indepen- (%B) was adjusted depending on the length and GC content dent variables that could be possible predictors of UGT of the extension primer. The extended products were eluted mRNA expression levels. We first investigated whether there of the order of CoGoToA dependent on the hydrophobi- was a correlation between the hepatic mRNA levels of city differences of the four bases. In the duplex reaction, SBE HNF1a and those of each UGT. We also evaluated whether products and unextended primers were designed to elute at the mRNA levels of the UGTs correlated among each other. different times and could be distinguished from each other. To explore whether the pair-wise correlations observed Controls of known genotype were included in each run. between the mRNA levels were related to 18S (present in The UGT2B7 intron 1 985A4G variant was genotyped also the denominator), partial correlations were also performed. by SBE-DHPLC. The PCR and extension primers were 50- We investigated next the relationship between genotype GATTTCAGCCATACTCTCAACA-30 (forward), 50-GAGTCCT and gene expression levels. Three univariate genetic models CCAACAAAATCAAC-30 (reverse) and 50-CTACTATTGAAG (additive, dominant and recessive) were tested, as the modes GTTTAAG-30 (reverse extension primer with one base of inheritance were either unknown or not clearly identifi- modified underlined). PCRs were set up in a 15 ml volume able from the plotted data. For example, for the hypothetical containing 2.5 mM MgCl2, 100 mM each dNTP, 125 nM variant A4G, we have genotypes A/A, A/G and G/G. The forward and reverse primers, 0.375 U AmpliTaq Gold additive model would be: A/A ¼ 0, A/G ¼ 1, G/G ¼ 2. The (Applied Biosystems) and 25 ng of DNA. Reactions were dominant model would be: A/A ¼ 0, A/G ¼ 1, G/G ¼ 1. denatured initially at 951C for 15 min then cycled at 951C The recessive model would be: A/A ¼ 0, A/G ¼ 0, G/G ¼ 1. for 15 s, 611C for 15 s, and 721C for 45 s for 40 cycles. For the Linear regression analysis was also used to compare gene SBE reaction, the PCR amplified product was treated with expression levels between females and males, and between shrimp alkaline phosphatase (Roche Diagnostics Corpora- donors who had ingested UGT inducers (phenobarbital and/ tion, Indianapolis, IN, USA) and exonuclease I (USB or dexamethasone, alcohol) or had smoked cigarettes, and Corporation, Cleveland, OH, USA) to remove excess dNTPs those who did not. The variables with significant P-values and primers. Purified PCR product (6 ml) was combined with (Pp0.1, based on the t-tests with the null hypothesis that 1 mM of extension primer, 250 mM each ddNTP and 1.25 U the slope or r2 ¼ 0) were then selected for multivariate

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 160

analysis to predict UGT mRNA levels. The final multivariate 13 Gregory PA, Lewinsky RH, Gardner-Stephen DA, Mackenzie PI. models included all variables that were significant with Coordinate regulation of the human UDP-glucuronosyltransferase P 0.05. Linear regressions were performed using the 1A8, 1A9, and 1A10 genes by hepatocyte nuclear factor 1alpha and o the caudal-related homeodomain protein 2. Mol Pharmacol 2004; 65: 52 stats:lm library in the R Statistical Package. 953–963. 14 Mackenzie PI, Gardner-Stephen DA, Gregory PA, Ishii Y, Lewinsky RH. Acknowledgments Mechanisms of UDP-Glucuronosyltransferase Gene Regulation. Inter- national Workshop on Glucuronidation: University of Dundee, Scotland, 2004. We thank James Chandler for assistance in genotyping HNF1a 15 Odom DT, Zizlsperger N, Gordon DB, Bell GW, Rinaldi NJ, Murray HL 79A4C. This work was supported by the Pharmacogenetics of et al. Control of and liver gene expression by HNF Anticancer Agents Research (PAAR) Group (http://pharmacogenetics. transcription factors. Science 2004; 303: 1378–1381. org) (NIH/NIGMS grant U01GM61393). Data will be deposited into 16 Gardner-Stephen DA, Gregory PA, Mackenzie PI. Identification and PharmGKB (supported by NIH/NIGMS U01GM61374, http:// characterization of functional hepatocyte nuclear factor 1-binding pharmgkb.org/). sites in UDP-glucuronosyltransferase genes. Methods Enzymol 2005; 400: 22–46. 17 Zhou J, Zhang J, Xie We. Xenobiotic -mediated regulation of UDP-glucuronosyl-. Curr Drug Metab 2005; Duality of Interest 6: 289–298. 18 Cereghini S. Liver-enriched transcription factors and hepatocyte None declared. differentiation. FASEB J 1996; 10: 267–282. 19 Bernard P, Goudonnet H, Artur Y, Desvergne B, Wahli W. Activation of the mouse TATA-less and human TATA-containing UDP-glucuronosyl- References 1A1 promoters by hepatocyte nuclear factor 1. Mol Pharmacol 1999; 56: 526–536. 1 Bosma PJ, Chowdhury JR, Bakker C, Gantla S, de Boer A, Oostra BA et al. 20 Chiu KC, Chuang LM, Ryu JM, Tsai GP, Saad MF. The I27L The genetic basis of the reduced expression of bilirubin UDP- polymorphism of hepatic nuclear factor-1alpha is associated with insulin glucuronosyltransferase 1 in Gilbert’s syndrome. N Engl J Med 1995; resistance. J Clin Endocrinol Metab 2000; 85: 2178–2183. 333: 1171–1175. 21 Urhammer SA, Moller AM, Nyholm B, Ekstrom CT, Eiberg H, Clausen JO 2 Monaghan G, Ryan M, Seddon R, Hume R, Burchell B. Genetic variation et al. The effect of two frequent amino acid variants of the hepatocyte in bilirubin UDP-glucuronosyltransferase gene promoter and Gilbert’s nuclear factor-1alpha gene on estimates of the pancreatic beta-cell syndrome. Lancet 1996; 347: 578–581. function in Caucasian glucose-tolerant first-degree relatives of type 2 3 Ando Y, Saka H, Ando M, Sawa T, Muro K, Ueoka H et al. diabetic patients. J Clin Endocrinol Metab 1998; 83: 3992–3995. Polymorphisms of UDP-glucuronosyltransferase gene and irinotecan 22 Babaya N, Ikegami H, Fujisawa T, Nojima K, Itoi-Babaya M, Inoue K toxicity: a pharmacogenetic analysis. Cancer Res 2000; 60: 6921–6926. et al. Association of I27L polymorphism of hepatocyte nuclear factor-1 4 Innocenti F, Undevia SD, Iyer L, Chen PX, Das S, Kocherginsky M et al. alpha gene with high-density lipoprotein cholesterol level. J Clin Genetic variants in the UDP-glucuronosyltransferase 1A1 gene Endocrinol Metab 2003; 88: 2548–2551. predict the risk of severe neutropenia of irinotecan. J Clin Oncol 2004; 23 Winter WE. Molecular and biochemical analysis of the MODY 22: 1382–1388. syndromes. Pediatr Diabetes 2000; 1: 88–117. 5 Innocenti F, Liu W, Chen P, Desai AA, Das S, Ratain MJ. Haplotypes of 24 Toide K, Takahashi Y, Yamazaki H, Terauchi Y, Fujii T, Parkinson A et al. variants in the UDP-glucuronosyltransferase1A9 and 1A1 genes. Hepatocyte nuclear factor-1alpha is a causal factor responsible for Pharmacogenet Genomics 2005; 15: 295–301. interindividual differences in the expression of UDP-glucuronosyl- 6 Bhasker R, McKinnon W, Stone A, Lo A, Kubota T, Ishizaki T et al. transferase 2B7 mRNA in human livers. Drug Metab Dispos 2002; 30: Genetic polymorphism of UDP-glucuronosyltransferase 2B7 (UGT2B7) 613–615. at amino acid 268: ethnic diversity of alleles and potential clinical 25 Ritter JK, Kessler FK, Thompson MT, Grove AD, Auyeung DJ, Fisher RA. significance. Pharmacogenetics 2000; 10: 679–685. Expression and inducibility of the human bilirubin UDP-glucuronosyl- 7 Holthe M, Rakvag TN, Klepstad P, Idle JR, Kaasa S, Krokan HE et al. transferase UGT1A1 in liver and cultured primary hepatocytes: Evidence Sequence variations in the UDP-glucuronosyltransferase 2B7 (UGT2B7) for both genetic and environmental influences. Hepatology 1999; 30: gene: identification of 10 novel single nucleotide polymorphisms (SNPs) 476–484. and analysis of their relevance to morphine glucuronidation in cancer 26 Sutherland L, Ebner T, Burchell B. The expression of UDP-glucuronosyl- patients. Pharmacogenomics J 2003; 3: 17–26. transferases of the UGT1 family in human liver and and in 8 Sawyer MB, Innocenti F, Das S, Cheng C, Ramirez J, Pantle-Fisher FH response to drugs. Biochem Pharmacol 1993; 45: 295–301. et al. A pharmacogenetic study of uridine diphosphate-glucuronosyl- 27 Krishnaswamy S, Hao Q, Al-Rohaimi A, Hesse LM, von Moltke LL, transferase 2B7 in patients receiving morphine. Clin Pharmacol Ther Greenblatt DJ et al. UDP-glucuronosyltransferase (UGT) 1A6 pharma- 0 2003; 73: 566–574. cogenetics: I. Identification of polymorphisms in the 5 -regulatory 9 Girard H, Court MH, Bernard O, Fortier LC, Villeneuve L, Hao Q et al. and exon 1 regions, and association with human liver UGT1A6 Identification of common polymorphisms in the promoter of the gene expression and glucuronidation. J Pharmacol Ex Ther 2005; 313: UGT1A9 gene: evidence that UGT1A9 protein and activity levels are 1331–1339. strongly genetically controlled in the liver. Pharmacogenetics 2004; 14: 28 Winckler W, Burtt NP, Holmkvist J, Cervin C, de Bakker PI, Sun M et al. 501–515. Association of common variation in the HNF1alpha gene region with 10 Girard H, Villeneuve L, Court M, Fortier LC, Caron P, Hao Q et al. The risk of . Diabetes 2005; 54: 2336–2342. novel UGT1A9 intronic polymorphism I399 appears as a predictor of 29 Bach I, Yaniv M. More potent transcriptional activators or a transdo- SN-38 glucuronidation levels in liver microsomes. Drug Metab Dispos minant inhibitor of the HNF1 homeoprotein family are generated by 2006; 34: 1220–1228. alternative RNA processing. EMBO J 1993; 13: 4229–4242. 11 Ishii Y, Hansen AJ, Mackenzie PI. Octamer transcription factor-1 30 Fang JL, Lazarus P. Correlation between the UDP-glucuronosyltransfer- enhances hepatic nuclear factor-1alpha-mediated activation of the ase (UGT1A1) TATAA box polymorphism and carcinogen detoxification human UDP glucuronosyltransferase 2B7 promoter. Mol Pharmacol phenotype: significantly decreased glucuronidating activity against 2000; 57: 940–947. benzo(a)pyrene-7, 8-dihydrodiol(À) in liver microsomes from subjects 12 Gregory PA, Mackenzie PI. The homeodomain Pbx2–Pbx1 complex with the UGT1A1*28 variant. Cancer Epidemiol Biomarkers Prev 2004; modulates hepatocyte nuclear factor 1alpha-mediated activation of the 13: 102–109. UDP-glucuronosyltransferase 2B17 gene. Mol Pharmacol 2002; 62: 31 Innocenti F, Grimsley C, Das S, Ramirez J, Cheng C, Kuttab-Boulos H 154–161. et al. Haplotype structure of the UDP-glucuronosyltransferase 1A1

The Pharmacogenomics Journal HNF1a regulation of UGT1A1, UGT1A9 and UGT2B7 J Ramı´rez et al 161

promoter in different ethnic groups. Pharmacogenetics 2002; 12: 42 Brierley CH, Senafi SB, Clarke D, Hsu MH, Johnson EF, Burchell B. 725–733. Regulation of the human bilirubin UDP-glucuronosyltransferase gene. 32 Iyer L, Hall D, Das S, Mortell MA, Ramirez J, Kim S et al. Phenotype– Adv Enzyme Regul 1996; 36: 85–97. genotype correlation of in vitro SN-38 (active metabolite of irinotecan) 43 Sugatani J, Kojima H, Ueda A, Kakizaki S, Yoshinari K, Gong QH et al. and bilirubin glucuronidation in human liver tissue with UGT1A1 The phenobarbital response enhancer module in the human bilirubin promoter polymorphism. Clin Pharmacol Ther 1999; 65: 576–582. UDP glucuronosyltransferase UGT1A1 gene and regulation by the 33 Iyer L, Das S, Janisch L, Wen M, Ramirez J, Karrison T et al. UGT1A1*28 nuclear receptor CAR. Hepatology 2001; 33: 1232–1238. polymorphism as a determinant of irinotecan disposition and toxicity. 44 Sugatani J, Yamakawa K, Tonda E, Nishitani S, Yoshinari K, Degawa M et Pharmacogenomics J 2002; 2: 43–47. al. The induction of human UDP-glucuronosyltransferase 1A1 mediated 34 Yamanaka H, Nakajima M, Katoh M, Hara Y, Tachibana O, Yamashita J. through a distal enhancer module by flavonoids and xenobiotics. A novel polymorphism in the promoter region of human UGT1A9 gene Biochem Pharmacol 2004; 67: 989–1000. (UGT1A9*22) and its effects on the transcriptional activity. Pharmaco- 45 Ban N, Yamada Y, Someya Y, Miyawaki K, Ihara Y, Hosokawa M et al. genetics 2004; 14: 329–332. Hepatocyte nuclear factor-1alpha recruits the transcriptional co- 35 Fleischmann R, Remmer H, Starz U. Induction of cytochrome P-488 iso- activator p300 on the GLUT2 gene promoter. Diabetes 2002; 51: enzymes and related glucuronyltransferases in the human liver by 1409–1418. cigarette smoking. Eur J Clin Pharmacol 1986; 30: 475–480. 46 Soutoglou E, Papafotiou G, Katrakili N, Talianidis I. Transcriptional 36 Walle T, Walle UK, Coward TD, Conradi EC, Gaffney TE. Selective activation by hepatocyte nuclear factor-1 requires synergism induction of propranolol metabolism by smoking: additional effects between multiple coactivator proteins. J Biol Chem 2000; 275: on renal clearance of metabolites. J Pharmacol Exp Ther 1987; 241: 12515–12520. 928–933. 47 Boudreau F, Rings EH, van Wering HM, Kim RK, Swain GP, Krasinski SD 37 Doodstar H, Grant MH, Melvin WT, Wolf CR, Burke MD. The effects of et al. Hepatocyte nuclear factor-1 alpha, GATA-4, and caudal inducing agents on cytochrome P450 and UDP-glucuronyltransferase related homeodomain protein Cdx2 interact functionally to modulate activities in human HEPG2 hepatoma cells. Biochem Pharmacol 1993; intestinal gene transcription. Implication for the developmental 46: 629–635. regulation of the sucrase-isomaltase gene. J Biol Chem 2002; 277: 38 Ritter JK, Kessler FK, Thompson MT, Groove AD, Auyeung DJ, Fisher RA. 31909–31917. Expression and inducibility of the human bilirubin UDP-glucuronosyl- 48 Kritis AA, Ktistaki E, Barda D, Zannis VI, Talianidis I. An indirect negative transferase UGT1A1 in liver and cultured primary hepatocytes: evidence autoregulatory mechanism involved in hepatocyte nuclear factor-1 for both genetic and environmental influences. Hepatology 1999; 30: gene expression. Nucleic Acid Res 1993; 21: 5882–5889. 476–484. 49 Ktistaki E, Talianidis I. Modulation of hepatic gene expression by 39 Kardon T, Coffey MT, Banhegyi G, Conley AA, Burchell B, Mandl J et al. hepatocyte nuclear factor 1. Science 1997; 277: 109–112. Transcriptional induction of bilirubin UDP-glucuronosyltransferase by 50 Eeckhoute J, Formstecher P, Laine B. Hepatocyte nuclear factor 4alpha ethanol in rat liver. Alcohol 2000; 21: 251–257. enhances the hepatocyte nuclear factor 1alpha-mediated activation of 40 Ramirez J, Komoroski BJ, Mirkov S, Graber AY, Fackenthal DL, Schuetz transcription. Nucleic Acids Res 2004; 32: 2586–2593. EG et al. Study of the genetic determinants of UGT1A1 inducibility by 51 Barbier O, Girard H, Inoue Y, Duez H, Villeneuve L, Kamika A et al. phenobarbital in cultured human hepatocytes. Pharmacogenet Genomics Hepatic expression of the UGT1A9 gene is governed by hepatocyte 2006; 16: 79–86. nuclear factor 4alpha. Mol Pharmacol 2005; 67: 241–249. 41 Parkinson A, Mudra DR, Johnson C, Dwyer A, Carroll KM. The effects of 52 The R package. R Development Core Team. R: A Language and gender, age, ethnicity, and liver cirrhosis on cytochrome P450 enzyme Environment for Statistical Computing. R Foundation for Statistical activity in human liver microsomes and inducibility in cultured human Computing: Vienna, Austria, 2005. ISBN 3-900051-07-0, URL http:// hepatocytes. Toxicol Appl Pharmacol 2004; 199: 193–209. www.R-project.org.

The Pharmacogenomics Journal