UGT1A7 Polymorphisms in Chronic Pancreatitis: an Example of Genotyping Pitfalls

UGT1A7 polymorphisms in chronic pancreatitis: an example of genotyping pitfalls

RHM te Morsche1, JPH Drenth1, UDP-glucuronosyltransferases (UGT) catalyze the glucuronidation of various 2 3 compounds and thus inactivate toxic substrates. Genetic variations reducing K Truninger , H-U Schulz , the activity of UGT1A7 have been associated with various gastrointestinal 4 5 1 A Kage , O Landt , M Verlaan , cancers. Most recently, the UGT1A7*3 allele has been reported as a J Rosendahl6, M Macek Jr7, significant risk factor for pancreatic disorders, but we could not confirm JBMJ Jansen1 and H Witt6 these data. This study focused on the possible causes for the noted discrepancy. UGT1A7 genotypes were assessed in 37 samples, which were 1Division of Gastroenterology and Hepatology, previously analyzed for UGT1A7 polymorphisms by others. We determined Department of Medicine, Radboud University genotypes by melting curve analysis and by DNA sequencing. Additionally, Medical Center Nijmegen, The Netherlands; we produced UGT1A7*1 and *3 constructs with or without a mutation at 2Department of Medicine, Division of Gastroenterology, Kantonspital, Aarau, position À 57 of UGT1A7 and analyzed various combinations of these Switzerland; 3Department of Surgery, Otto-von- constructs. In 14/37 samples UGT1A7 genotyping results differed. The Guericke University, Magdeburg, Germany; discrepancy could be explained by polymerase chain reaction bias owing 4 Department of Laboratory Medicine and to an unbalanced allelic amplification which was caused by a À57T4G Pathobiochemistry, Charite´, Virchow-Klinikum, Universita¨tsmedizin Berlin, Berlin, Germany; 5TIB variant located within the sequence of the chosen primer template in MOLBIOL, Berlin, Germany; 6Department of previous studies. Our findings indicate that most of the previously reported Gastroenterology, Charite´, Virchow-Klinikum, genetic associations between UGT1A7 and gastrointestinal cancers are based Universita¨tsmedizin Berlin, Berlin, Germany and on primer-dependent genotyping errors. 7Institute of Biology and Medical Genetics – University Hospital Motol and 2nd School of The Pharmacogenomics Journal (2008) 8, 34–41; doi:10.1038/sj.tpj.6500443; Medicine of Charles University, Prague, Czech published online 27 February 2007 Republic Keywords: PCR; UDP-glucuronosyltransferases; UGT1A7; polymorphism; genotyping error; Correspondence: sequencing Dr H Witt, Department of Gastroenterology, Charite´, Virchow-Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany. E-mail: [email protected]

Introduction

The UDP-glucuronosyltransferases (UGT) represent a superfamily of enzymes bound to the membrane of the endoplasmic reticulum.1,2 The human UGT family contains more than a dozen members from three subfamilies, UGT1A, 2A and 2B and are mainly of hepatic origin, but high UGT expression levels have been reported also in extrahepatic tissues such as intestine and kidney.3–8 UGTs catalyze the conjugation of exogenous and endogenous compounds with uridine diphosphoglucuronic acid. The resulting glucuronides are more soluble in water and facilitate biliary or renal excretion. UGTs play an important role in cellular defense not only by eliminating xenobiotics and endogenous toxins, but also by detoxifying human carcinogens, such as heterocyclic amines as well as heterocyclic and polycyclic hydrocarbons.9 The UGT1A family is located on a single locus on chromosome 2q37 and consists of at least nine functional proteins.10 Exons 2 to 5 are shared by all Received 8 September 2006; revised 26 11 November 2006; accepted 12 December UGT1A proteins, whereas exon 1 is specific for each isoform. One specific 2006; published online 27 February 2007 UGT1A protein, UGT1A7, is present in the esophagus, stomach, lung and PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 35

Table 1 UGT1A7 alleles UGT1A7*3 (K129-K131-R208). This base pair change is precisely located at the annealing site of the forward primer Allele Amino acid and nucleotide changes (at position-19 away from the 30-end of the primer) used in the previous analysis (Figure 1). PCR amplification of 129 131 208 UGT1A7*1 N (AAT) R (CGA) W (TGG) UGT1A7*1/*3 heterozygotes at an annealing temperature of 129 131 208 UGT1A7*2 K (AAG) K (AAA) W (TGG) 521C with forward primers that contained either a wild-type 129 131 208 UGT1A7*3 K (AAG) K (AAA) R (CGG) T (11F and 13F) or a mutant G (12F and 14F) at position À57 UGT1A7*4 N129 (AAT) R131 (CGA) R208 (CGG) resulted in a reduced amplification of the heterozygous allele with one mismatch. Higher annealing temperatures (56 and 601C) led to complete amplification of the pancreas and catalyzes the glucuronidation of simple and complementary allele in heterozygotes (data not shown). complex phenols, flavones, coumarins and benzo[a] This suggests that the use of a primer without the mutation pyrenes.9,12 UGT1A7 is highly polymorphic and so far 11 at position À57 in the sequence leads to uneven amplifica- different alleles have been identified. These polymorphisms tion of UGT1A7*3 (Figure 2). may have functional consequences as they reduce enzyme activity, resulting in a lower capacity for detoxification of Constructs human carcinogens.13–16 Codons 129, 131 and 208 are Next, we examined how the sequence of the primer affects highly variable among mammals and polymorphisms result the efficacy to detect UGT1A7*1 and UGT1A7*3 constructs in a range of separate UGT1A7 alleles: N129-R131-W208 by melting curve analysis using fluorescence resonance (UGT1A7*1), K129-K131-W208 (UGT1A7*2), K129-K131- energy transfer (FRET) probes. We tested two forward R208 (UGT1A7*3) and N129-R131-R208 (UGT1A7*4) (Table primers, one with a wild-type T (13F) and another with 1). On balance, this may have pathological consequences as, mutant G at position À57 (14F). We found that both primers for example, the UGT1A7*3 allele has been associated with detected UGT1A7*1 and UGT1A7*3 well at both annealing colorectal cancer, hepatocellular carcinoma, and orolaryn- temperatures, from a mixture of UGT1A7*1 and UGT1A7*3 geal cancer and in particular with pancreatic cancer and (with or without the mutation at position À57) (Figure 3a, b, alcoholic chronic pancreatitis (CP).16–20 e and f). We recently performed a case–control study and examined If we use a mixture of UGT1A7*1 with G at position À57 the relationship between the UGT1A7*3 allele and pancrea- (construct 3) together with UGT1A7*3 with a T at position tic disease in a population of 973 pancreatitis or pancreatic À57 (construct 4) as a template for the PCR, both primers cancer patients.15 In contrast to an earlier study,16 we failed will amplify both constructs at the annealing temperature of to confirm an enrichment of UGT1A7*3 in patients with 521C in a balanced manner. However, we found that upon pancreatic disorders. Apart from study size and the possibi- increasing the annealing temperature to 571C, the primer lity of non-random distribution of the alleles, we considered with a wild-type T at position À57 (13F) merely amplified the possibility that the differences may be owing to construct UGT1A7*3 with a T at position À57. (Figure 3c and polymerase chain reaction (PCR) bias. PCR bias leads to d). In contrast, at 571C the primer with the mutant G at unfair representation of the quantity of a template present position À57 (14F) preferably amplified UGT1A7*1 with G at in the reaction. This might be caused not only by differences position À57 (construct 3). Likewise, a mixture of UGT1A7*1 in template lengths, random variations in template number and UGT1A7*3 constructs (representative of a compound (especially with very small initial numbers) and random heterozygote UGT1A7*1/*3 carrier) yielded similar results. variations in PCR efficiency in each cycle, but also by the Both primers amplify UGT1A7*1 and UGT1A7*3 evenly at presence of a polymorphism at the primer-annealing site. 521C, but at 571C, the primer with T at position À57 (13F) We hypothesized that the detection of UGT1A7*3 may be preferentially amplifies UGT1A7*1 whereas the primer with hampered by PCR bias. To establish this, we used a range G at position À57 (14F) only amplifies UGT1A7*3 (Figure 3g of different primer sets and analyzed both human DNA and h). samples and mixtures of cloned DNA for UGT1A7*3 In summary, upon testing two different primers using two detection. UGT1A7 constructs with different nucleotides on position À57 as a template, we found that the primer with a matching sequence at that position prefers to amplify the Results analogous template. Spiking the PCR template with mouse genomic DNA did not affect the obtained results. Genotyping The results of the genetic analysis of 37 CP patients, which Discussion had previously been analyzed,16 differed in 14 cases (38%) (Table 2). Our analysis demonstrated that 11 out of 12 To examine whether the differences in genotyping results individuals carrying UGT1A7*1/*3 had been classified pre- from previous studies16–18 arose from PCR amplification viously as having UGT1A7*1/*4 alleles.16 Sequencing of bias, we analyzed 37 CP samples that had been part of a exon 1 revealed a T4G transversion at position À57 which previous case–control study focused on the role of UGT1A7 appears to be in complete linkage disequilibrium with polymorphisms in pancreatic disorders.16 In approximately

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 36

Table 2 UGT1A7 genotyping results obtained from 37 chronic pancreatitis patients analyzed with different primer combinations

Patient Results

Primer combination Result by others16 Allele status À57

3F+4R 5F+8R 6F+8R 7F+9R

D-L-0011 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0015 3/3 3/3 3/3 3/3 2/3 GG D-L-0021 1/2 1/2 1/2 1/2 1/2 TT D-L-0030 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0031 2/2 2/2 2/2 2/2 2/2 TT D-L-0047 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0074 2/3 2/3 2/3 2/3 2/3 TG D-L-0099 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0100 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/3 (2/4)* TG D-L-0103 3/3 3/3 3/3 3/3 3/3 GG D-L-0115 2/3 2/3 2/3 2/3 2/3 TG D-L-0116 1/2 1/2 1/2 1/2 1/2 TT D-L-0131 1/2 1/2 1/2 1/2 1/2 TT D-L-0134 3/3 3/3 3/3 3/3 2/3 GG D-L-0142 1/2 1/2 1/2 1/2 1/2 TT D-L-0148 2/3 2/3 2/3 2/3 2/3 TG D-L-0155 1/1 1/1 1/1 1/1 1/1 TT D-L-0166 2/3 2/3 2/3 2/3 1/3 (2/4)*TG D-L-0169 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0190 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0212 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0213 3/3 3/3 3/3 3/3 3/3 GG D-L-0219 1/2 1/2 1/2 1/2 1/2 TT D-L-0221 2/3 2/3 2/3 2/3 2/3 TG D-L-0226 3/3 3/3 3/3 3/3 3/3 GG D-L-0247 2/3 2/3 2/3 2/3 2/3 TG D-L-0251 1/1 1/1 1/1 1/1 1/1 TT D-L-0272 1/1 1/1 1/1 1/1 1/1 TT D-L-0374 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0404 3/3 3/3 3/3 3/3 3/3 GG D-L-0439 1/1 1/1 1/1 1/1 1/1 TT D-L-0454 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0475 3/3 3/3 3/3 3/3 3/3 GG D-L-0516 1/1 1/1 1/1 1/1 1/1 TT D-L-0528 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0535 1/3 (2/4)* 1/3 (2/4)* 1/3 1/3 (2/4)* 1/4 TG D-L-0537 1/2 1/2 1/2 1/2 1/2 TT

Comparison of genotyping results of 37 patients analyzed and published in a previous report. Genotypes obtained in the present study using different primer combinations (3F+4R, 5F+8R, 6F+8R, 7F+9R) are given in columns 2 to 5. Genotypes obtained by others are given in column 6 [16]. Column 7 shows the -57T4G status that is linked with UGT1A7*3. Differences in genotyping are shown in bold in column 6. * As N129K/R131K or N129R/R129K and W208R may occur in cis or trans, UGT1A7*1/*3 heterozygotes are indistinguishable from UGT1A7*2/*4 heterozygotes by most genotyping approaches.

40% of these samples, we obtained different genotyping from the 30-end of the chosen primer. In principle this results (Table 2). Upon sequencing, we first identified a T4G should not affect PCR amplification, but the region down- transversion at position À57 of the UGT1A7 gene, corrobor- stream of À57 is extremely AT-rich, making the 30-end of ating data from a recent report.21 This polymorphism the chosen oligonucleotide primer unstable. Moreover, a appears to be in complete linkage disequilibrium with GC-rich clamp, which does not align with the UGT1A7 UGT1A7*3 (K129-K131-R208) and was detected both in a sequence, was attached in previous studies at the 50-end of heterozygous or homozygous state in all UGT1A7*3 hetero- this primer, which further impairs the affinity to its target zygotes or homozygotes, respectively, but not in individuals (Figure 1). without a UGT1A7*3 allele. À57T4G is located within the We examined the possibility that À57T4G affects PCR annealing sequence of the forward primer used in previous amplification, and designed several primers either with a T studies,16–18,22 and is located in 19 nucleotides upstream or a G at position À57. Subsequent melting curve analysis

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 37

We found that the frequency of the UGT1A7*3 allele in our healthy controls is B40%.15 This is in line with frequencies of 32, 36 and 37% in control samples reported by others,13,14,26 but is appreciably higher than the 16–21% detected by early reports from the group that studied the Figure 1 UGT1A7 À57T4G polymorphism. The first line shows the role of UGT1A7 in pancreatic diseases.16–18,22 Our data forward primer used in a previous report;16 the second line shows suggest that the relatively low frequency of UGT1A7*3 the genomic sequence of UGT1A7. The GC-rich clamp at the 50-end of the primer that does not align with the UGT1A7 sequence is underlined. alleles in these studies can be explained by this PCR bias. The position of the mutated nucleotide is indicated in bold. On the other hand, PCR bias cannot explain the reported differences in the frequency of UGT1A7*3 alleles between controls and patients. One may speculate that the over- representation of the UGT1A7*3 allele in the patient group is due to the fact that samples from each group of subjects to be compared (e.g., cases and controls) have not been included in each batch analyzed as recommended as a quality measure in genetic association studies. As shown in Figure 3, the occurrence of uneven allelic amplification is dependent on the chosen primer-annealing temperature. Thus, a slight variation of annealing temperature (or of ion concentration of the PCR reaction mix) may drive the reaction to an even amplification of both alleles or to PCR bias with preferred amplification of the UGT1A7*1. Although the present study explains the basic principle of the genotyping error reported by previous studies, it does not answer why the À57T4G alteration led to *1/*4 instead of *1/*1 genotypes in UGT1A7*1/*3 heterozygotes. If the upstream primer does not appropriately anneal to Figure 2 Constructs. Construct 1 represents the normal wild-type UGT1A7*3, one would expect an exclusive amplification of allele present in humans cloned into the pGEM-T vector with a wild-type the second allele, UGT1A7*1, and that these individuals thymidine at position À57. The star indicates polymorphisms and would be classified as UGT1A7*1/*1 homozygotes. construct 2 represents the UGT1A7*3 allele in humans with a mutant A recent study from the same researchers who reported guanine at position À57. Construct 4 contains the mutant allele but the enrichment of UGT1A7*3 in pancreatic disorders with a wild-type thymidine at position À57 and construct 3 contains reported a À57G allele frequency of 0.39 among 427 control wild-type allele with a mutant guanine at position À57. subjects.21 Similar to our results, they found that À57G is in complete linkage disequilibrium with UGT1A7*3. As a corollary, it is more than likely that the ‘real’ UGT1A7*3 frequency in their control population is close to 0.39. showed that in À57T4G heterozygotes, primarily the À57T In most studies, UGT1A7 polymorphisms have been or À57G allele was amplified depending on the chosen determined by sequencing PCR fragments of the gene. primer. Although sequencing has been generally accepted as the As a next step, we used cloned DNA that allows normal- ‘gold standard’ for mutation detection, a recent study ization of a template input with a genomic reference demonstrated that in some instances PCR products that sequence for the most accurate quantification of PCR bias. were heterozygous in the restriction fragment length We prepared four constructs with different combinations of polymorphism (RFLP) assay came back as homozygous on nucleotides at position –57, 129, 131 and 208 of UGT1A7. sequencing.23 This effect was even more pronounced if We found that the nucleotide at À57 was critical as in primers were used that were not completely aligned heterozygous samples; an unmatched primer pair led to (mismatch) with the template sequence. This leads us back to unbalanced amplification with a clear preference for the UGT1A7 because we observed that most other studies16–18,22 template with the matching sequence. We found that used a ‘mismatched’ primer (because of the À57T4G reactions with À57 homozygous constructs led to successful polymorphism). This will cause uneven allele amplification amplifications regardless of the nature of the nucleotide at during PCR and erroneous results in the subsequent À57. This observation has immediate consequences as sequencing reaction. The observed differences in UGT1A7 À57T4G appears to be in complete linkage disequilibrium gene analysis of the 37 subjects can best be explained by a with K129-K131-R208 (UGT1A7*3). As a result, this poly- primer-dependent PCR bias for the various alleles. Accord- morphism reduces the amplification of the UGT1A7*3 allele ingly, as most analytical methods such as sequencing as the in the presence of UGT1A7*1 or UGT1A7*2 allele (contain- gold-standard, RFLP, TaqMan or hybridization probes de- ing the K129 and K131 polymorphisms), and confounds the pend on a PCR product that is generated by the mismatched interpretation of the results. primer, all diagnostic methods will always lead to incorrect

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 38

a 3 b 3

2 2

1 1

-d(RFU)/dT 0 -d(RFU)/dT 0 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70

-1 -1 Temperature Temperature c 3 d 5 3 4 2 3 1 2 1 -d(RFU)/dT 0 -d(RFU)/dT 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 0 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 -1 -1 Temperature Temperature e 4 f 3

3 2 2 1 1

-d(RFU)/dT -d(RFU)/dT 0 0 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 -1 -1 Temperature Temperature

g 5 h 5 4 4

3 3 2 2 1 1 -d(RFU)/dT -d(RFU)/dT

0 0 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 -1 -1 Temperature Temperature

Figure 3 Melting curves with FRET probes for codons 129 and 131 of the constructs. The blue line represents detection with primer 13F, the red line represents detection with primer 14F. Panels (a) and (b) represent melting curves for PCRs with constructs 1 and 4 (both À57T) as templates at annealing temperatures of 521C(a) and 571C(b). Panels (c) and (d) depict melting curves of PCRs with constructs 4 (À57T) and 3 (À57G) as input at annealing temperatures of 521C(c) and 571C(d). Panels (e) and (f) show a melting curve for constructs 2 and 3 (both À57G) at annealing temperatures of 521C (E) and 571C(f), and lastly, panels (g) and (h) are representative examples of melting curves with constructs 1 (À57T) and 2 (À57G) at annealing temperatures of 521C(g) and 571C(h). Construct 1: À57t, N129, R131, W208; construct 2: À57g, K129, K131, R208; construct 3: À57 g, N129, R131, W208; construct 4: À57t, K129, K131, R208 (see also Figure 2). results. In this respect, our report represents an important finding extend beyond analysis of UGT1A7 gene variants methodological example of how PCR amplification bias and may have an impact on other genetic association results in genotyping errors. As additional clinically sig- studies. Indeed, this is well illustrated by a study that nificant SNPs are being discovered, assessment of PCR bias initially indicated an association between a microsomal will be increasingly important. The implications of this epoxide hydrolase gene polymorphism and Crohn’s

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 39

Table 3 Primers used for PCR amplification of UGT1A7

Method Sequence

Constructs 1F 50-GAA TAA GTA CAC GCC TTC TT-30 2R 50-TTC AGA GGC TAT TTC TAA GA-30 Genotyping 3Fa 50-TGC CGA TGC TCG CTG GAC G-30 4Ra 50-ATC TCT ACA GCC ACA CAT CAA TT-30 5F 50-CAA TGG TAT TTT TGA CTT-30 6Fb 50-GGT GGA CTG GCC TCC TTg CA-30 7F 50-TGA ATG AAT AAG TAC ACG CCT TC-30 8R 50-CAG AGG CTA TTT CTA AGA-30 9R 50-ACT TAC ATA TCA ACA AGA GCT GC-30 10R 50-CCA TAG GCA CTG GCT TTC CCT GAT GAC A-30 11Fc 50-CCA CTT ACT ATA TTA TAG GAG CT-30 12Fc 50-CCA CGT ACT ATA TTA TAG GAG CT-30 13Fc 50-GCG GCT CGA GCC ACT TAC TAT ATT ATA GGA GCT-30 14Fc 50-GCG GCT CGA GCC ACG TAC TAT ATT ATA GGA GCT-30

Abbreviation: PCR, polymerase chain reaction. aPrimers published by Ko¨hle et al.26 bBold ‘g’ in the primer sequence denotes mutagenesis for more selective amplification of the UGT1A7*3 and UGT1A7*4 alleles. cBold ‘T’ and ‘G’ in the primer sequence denote the site of the À57 T4G variation linked with the UGT1A7*3 allele; the GC-rich clamp used in the previously described primer in other studies, which does not align with the UGT1A7 sequence, is underlined. disease.24 Subsequent testing demonstrated a genotyping forms a more stable duplex than the wild-type allele, error because of a silent polymorphism that interfered with resulting in an allele-specific melting curve (N129K/R131K: proper annealing of the primer and led to misinterpretation 58 vs 471C; W208R: 65 vs 60.51C). of the RFLP.25 The positive association completely disap- In addition, we used a set of 12 different primers to peared on proper reanalysis. Thus, assessment of assay genotype the various UGT1A7 polymorphisms (Table 3). quality by using a second independent set of primers or These primers were designed to flank the site of polymorph- sequencing of both primer regions to exclude genetic ism of interest at exon 1 of UGT1A7 and are based on the variants that affect primer annealing is strongly recom- published nucleotide sequence (Table 3; GenBank no. mended. U39570).

Cloning of UGT1A7 constructs Materials and methods To examine how the selection of a primer affects the detection of UGT1A7 codon 129 and 131 polymorphisms, Genotyping we designed four different constructs. (Figure 2) We used We investigated 37 CP patients with a N34S SPINK1 DNA isolated from homozygous UGT1A7*1/*1 and mutation (courtesy of Dr Volker Keim, Leipzig, Germany). UGT1A7*3/*3 carriers as a template for our constructs. These These participants were part of an earlier study and had been constructs were built in such a way that they contained four genotyped for UGT1A7 by others.16 We genotyped UGT1A7 different allelic versions of the UGT1A7*1 or *3 allele with or alleles by melting curve analysis with FRET probes in the without a mutation at position À57 of the UGT1A7 gene LightCycler (Roche Diagnostics, Mannheim, Germany). The (Figure 2). FRET probes for detection of the polymorphisms at codons To clone constructs 1 (representing UGT1A7*1) and 2 129 and 131 and the W208R mutation were designed and (representing UGT1A7*3), the fragment that contained part synthesized by TIB MOLBIOL, Berlin, Germany. For detec- of the UGT1A7 gene was amplified using PCR. The 50 ml tion of the polymorphisms at codons 129 and 131 the reaction mixture contained 200 ng of genomic DNA, 10 mM 0 sensor probe was 5 -LC 640-TTAAGTATTCTACTAA Tris–HCl (pH 9.0), 50 mM KCl, 0.1% TRITON, 2 mM MgCl2, TTTTTTGTCCTT-ph (LC: LightCycler Red attached to 50 0.25 mM dNTPs, 10 pmol of primers 1F and 2R (Table 3), and terminus; ph: phosphate) and the anchor probe 50-GGATC 3.0 U Taq-DNA-polymerase. The PCR product was subjected GAGAAACACTGCATCAAAACAACTCTCC-FL (FL: 5,6-car- to electrophoresis on a 1% agarose gel and we isolated the boxyfluorescein attached to 30-O-ribose) were used. For 808 bp fragments using the QIAEXII Gel Extraction Kit identification of W208R the sequence of the sensor probe (Qiagen, Hilden, Germany). The fragments were subse- was 50-TGATGTGGTTCCGTACTCTCTCCTT-FL and of the quently cloned into a pGEM-T Vector (Promega, Madison, anchor probe was 50-LC 705-AAAGTCATGGCGTCTGAG WI, USA). AACCCTAAG-ph. Both sensor probes were complementary Constructs 1 and 2 contain two NcoI restriction sites, one to the mutant sequences (129K/131K and 208R, respec- located in the multiple cloning site of the vector immedi- tively). During melting curve analysis, the mutant allele ately before the UGT1A7 fragment, whereas a second site is

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 40

at nucleotide 97 of the UGT1A7 cDNA sequence. To prepare 2 Radominska-Pandya A, Czernik PJ, Little JM, Battaglia E, Mackenzie PI. constructs 3 and 4 (representatives for UGT1A7*1 and Structural and functional studies of UDP-glucuronosyltransferases. Drug UGT1A7*3 but now both with a mutation at position Metab Rev 1999; 31: 817–899. 3 Mackenzie PI, Owens IS, Burchell B, Bock KW, Bairoch A, Belanger A et À57), we digested constructs 1 and 2 with NcoI. Subsequent al. The UDP glycosyltransferase gene superfamily: recommended electrophoresis on a 1% agarose gel separated a 219 bp nomenclature update based on evolutionary divergence. Pharmacoge- fragment. The gel-isolated fragment of construct 1 was then netics 1997; 7: 255–269. cloned in NcoI digested construct 2, creating a new construct 4 Tukey RH, Strassburg CP. Human UDP-glucuronosyltransferases: metabolism, expression, and disease. Annu Rev Pharmacol Toxicol 2000; 40: 3(UGT1A7*1 with a G at position À57). In the same way, we 581–616. cloned the fragment of construct 2 in NcoI digested 5 Pacifici GM, Franchi M, Bencini C, Repetti F, Di Lascio N, Muraro GB. construct 1, thus creating construct 4 (UGT1A7*3 with a T Tissue distribution of drug-metabolizing enzymes in humans. Xenobio- at position À57). (Figure 2) All constructs were sequenced to tica 1988; 18: 849–856. 6 Cappiello M, Giuliani L, Rane A, Pacifici GM. Uridine 50-diphosphoglu- verify the presence of the mutations. curonic acid (UDPGLcUA) in the human fetal liver, kidney and placenta. To mimic PCR conditions of human genomic DNA, the Eur J Drug Metab Pharmacokinet 2000; 25: 161–163. 25 ml reaction mixture contained 100 ng of mouse genomic 7 McDonnell WM, Hitomi E, Askari FK. Identification of bilirubin DNA spiked with eight combinations of 105 copies of 2 UDP-GTs in the human alimentary tract in accordance with the gut as a putative metabolic organ. Biochem Pharmacol 1996; 51: constructs representing two alleles, 10 mM Tris–HCl (pH 483–488. 9.0), 50 mM KCl, 0.1% TRITON, 2 mM MgCl2, 0.25 mM 8 Radominska-Pandya A, Little JM, Pandya JT, Tephly TR, King CD, Barone dNTPs, 0.2 mM of 14F or 15F, 0.2 mM of primer 2R, 100 nM GW et al. UDP-glucuronosyltransferases in human intestinal mucosa. Biochim Biophys Acta 1998; 1394: 199–208. of sensor probe A, 100 nM of anchor probe A, and 1.5 U 1 9 King CD, Rios GR, Green MD, Tephly TR. UDP-glucuronosyltransferases. AmpliTaq Gold. The reaction mix was denatured at 95 C for Curr Drug Metab 2000; 1: 143–161. 3 min followed by 40 cycles of denaturation at 951C for 30 s, 10 Owens IS, Ritter JK. Gene structure at the human UGT1 locus creates annealing at 52 or 571C for 30 s, elongation at 721C for 60 s diversity in isozyme structure, substrate specificity, and regulation. Prog and a final extension step for 2 min at 721C. The program for Nucleic Acid Res Mol Biol 1995; 51: 305–338. 1 1 11 Ritter JK, Chen F, Sheen YY, Tran HM, Kimura S, Yeatman MT et al. A analytical melting was 95 C for 60 s, 35 C for 40 s and an novel complex locus UGT1 encodes human bilirubin, phenol, and other increase to 751Cata11C/10 s ramp rate and was analyzed UDP-glucuronosyltransferase isozymes with identical carboxyl termini. using the i-Cycler (Biorad, Laboratories BV, Veenendaal, The J Biol Chem 1992; 267: 3257–3261. Netherlands). 12 Strassburg CP, Manns MP, Tukey RH. Expression of the UDP- glucuronosyltransferase 1A locus in human colon. Identification and characterization of the novel extrahepatic UGT1A8. J Biol Chem 1998; 273: 8719–8726. Abbreviations 13 Guillemette C, Ritter JK, Auyeung DJ, Kessler FK, Housman DE. Structural heterogeneity at the UDP-glucuronosyltransferase 1 locus: CP chronic pancreatitis functional consequences of three novel missense mutations in the FRET fluorescence resonance energy transfer human UGT1A7 gene. Pharmacogenetics 2000; 10: 629–644. PCR polymerase chain reaction 14 Villeneuve L, Girard H, Fortier LC, Gagne JF, Guillemette C. Novel RFLP restriction fragment length polymorphism functional polymorphisms in the UGT1A7 and UGT1A9 glucuronidating UGT UDP-glucuronosyltransferases enzymes in Caucasian and African-American subjects and their impact on the metabolism of 7-ethyl-10-hydroxycamptothecin and flavopiridol anticancer drugs. J Pharmacol Exp Ther 2003; 307: 117–128. 15 Verlaan M, Drenth JP, Truninger K, Koudova M, Schulz HU, Bargetzi M et al. Polymorphisms of UDP-glucuronosyltransferase 1A7 are not Acknowledgments involved in pancreatic diseases. J Med Genet 2005; 42: e62. 16 Ockenga J, Vogel A, Teich N, Keim V, Manns MP, Strassburg CP. UDP We thank Volker Keim (Leipzig, Germany) for providing DNA glucuronosyltransferase (UGT1A7) gene polymorphisms increase the samples and genotyping data of N34S SPINK1 chronic pancreatitis risk of chronic pancreatitis and pancreatic cancer. Gastroenterology patients. We also thank Mrs Christiane Jechorek (Magdeburg) and 2003; 124: 1802–1808. Claudia Gu¨ldner (Berlin) for their excellent technical assistance. 17 Vogel A, Kneip S, Barut A, Ehmer U, Tukey RH, Manns MP et al. Genetic link of hepatocellular carcinoma with polymorphisms of the UDP- Dr Joost PH Drenth is supported by an NWO-VIDI grant. This study glucuronosyltransferase UGT1A7 gene. Gastroenterology 2001; 121: was partly supported by the Deutsche Forschungsgemeinschaft 1136–1144. (Wi 2036/1-1 and Wi 2036/2-1), by grants from the Dutch 18 Strassburg CP, Vogel A, Kneip S, Tukey RH, Manns MP. Polymorphisms Foundation of Digestive Diseases (MLDS WS00-21) and Sonnen- of the human UDP-glucuronosyltransferase (UGT) 1A7 gene in color- feld-Stiftung, Berlin, Germany (Witt), and by IGA MZCR ectal cancer. Gut 2002; 50: 851–856. 00000064203/6112 to MM 19 Tseng CS, Tang KS, Lo HW, Ker CG, Teng HC, Huang CS. UDP- glucuronosyltransferase 1A7 genetic polymorphisms are associated with hepatocellular carcinoma risk and onset age. Am J Gastroenterol 2005; 100: 1758–1763. Duality of Interest 20 Zheng Z, Park JY, Guillemette C, Schantz SP, Lazarus P. Tobacco carcinogen-detoxifying enzyme UGT1A7 and its association with The authors declare no duality of interest. orolaryngeal cancer risk. J Natl Cancer Inst 2001; 93: 1411–1418. 21 Lankisch TO, Vogel A, Eilermann S, Fiebeler A, Krone B, Barut A et al. Identification and characterization of a functional TATA box poly- References morphism of the UDP glucuronosyltransferase 1A7 gene. Mol Pharmacol 2005; 67: 1732–1739. 1 Meech R, Mackenzie PI. Structure and function of uridine diphosphate 22 Vogel A, Ockenga J, Ehmer U, Barut A, Kramer FJ, Tukey RH et al. glucuronosyltransferases. Clin Exp Pharmacol Physiol 1997; 24: 907–915. Polymorphisms of the carcinogen detoxifying UDP-glucuronosyltrans-

The Pharmacogenomics Journal PCR bias leads to UGT1A7 genotyping errors RHM te Morsche et al 41

ferase UGT1A7 in proximal digestive tract cancer. Z Gastroenterol 2002; 25 Peters WH, van der Logt EM, te Morsche RH, Roelofs HM, de Jong DJ, 40: 497–502. Naber TH. No genetic association between EPHX1 and Crohn’s disease. 23 van der Heiden I, van der Werf M, Lindemans J, van Schaik RH. Gut 2005; 54: 1659–1660. Sequencing: not always the ‘gold standard’. Clin Chem 2004; 50: 26 Ko¨hle C, Mo¨hrle B, Mu¨nzel PA, Schwab M, Wernet D, Badary OA 248–249. et al. Frequent co-occurrence of the TATA box mutation associated 24 de Jong DJ, van der Logt EM, van Schaik A, Roelofs HM, Peters WH, with Gilbert’s syndrome (UGT1A1*28) with other polymorphisms Naber TH. Genetic polymorphisms in biotransformation enzymes in of the UDP-glucuronosyltransferase-1 locus (UGT1A6*2 and Crohn’s disease: association with microsomal epoxide hydrolase. Gut UGT1A7*3) in Caucasians and Egyptians. Biochem Pharmacol 2003; 2003; 52: 547–551. 65: 1521–1527.

The Pharmacogenomics Journal