The Pharmacogenomics Journal (2010) 10, 524–536 & 2010 Macmillan Publishers Limited. All rights reserved 1470-269X/10 www.nature.com/tpj ORIGINAL ARTICLE

Genetic variation in and susceptibility to isoniazid-induced hepatotoxicity

S Yamada1,2, K Richardson3, Treatment of latent tuberculosis infection (LTBI) generally includes isoniazid 4 1 (INH), a drug that can cause serious hepatotoxicity. (CES) M Tang , J Halaschek-Wiener , are important in the metabolism of a variety of substrates, including 4,5 4 VJ Cook , JM FitzGerald , xenobiotics. We hypothesized that genetic variation in CES genes expressed K Elwood4, F Marra4,6,8 in the liver could affect INH-induced hepatotoxicity. Three CES genes are and A Brooks-Wilson1,7,8 known to be expressed in human liver: CES1, CES2 and CES4. Our aim was to systematically characterize genetic variation in these novel candidate genes 1Cancer Genetics, Canada’s Michael Smith and test whether it is associated with this adverse drug reaction. As part of a Genome Sciences Centre, British Columbia pilot study, 170 subjects with LTBI who received only INH were recruited, Cancer Agency, Vancouver, Canada; including 23 cases with hepatotoxicity and 147 controls. All exons and the 2Department of Life Science, Ritsumeikan University, Kusatsu Shiga, Japan; 3Centre for promoters of CES1, CES2 and CES4 were bidirectionally sequenced. A large Clinical Epidemiology and Evaluation, Vancouver polymorphic deletion was found to encompass exons 2 to 6 of CES4.No Coastal Health Research Institute, Vancouver, significant association was found. Eleven single-nucleotide polymorphisms 4 Canada; British Columbia Centre for Disease (SNPs) in CES1 were in high linkage disequilibrium with each other. One of Control, Vancouver, Canada; 5Faculty of Medicine, University of British Columbia, these SNPs, C(À2)G, alters the translation initiation sequence of CES1 and Vancouver, Canada; 6Department of Family represents a candidate functional polymorphism. Replication of this possible Practice, University of British Columbia, association in a larger sample set and functional studies will be necessary to 7 Vancouver, Canada and Department of determine if this CES1 variant has a role in INH-induced hepatotoxicity. Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, The Pharmacogenomics Journal (2010) 10, 524–536; doi:10.1038/tpj.2010.5; Canada published online 2 March 2010

Correspondence: Keywords: pharmacogenetics; tuberculosis; genetic association; single-nucleotide polymorph- Dr A Brooks-Wilson, Cancer Genetics, Canada’s ism; haplotype; case–control study Michael Smith Genome Sciences Centre, British Columbia Cancer Agency, 675 W 10th Ave., Vancouver BC V5Z 4E6, Canada. E-mail: [email protected] Introduction

Tuberculosis (TB) is a major global health problem. In Canada, groups at increased risk include Aboriginal persons, the foreign-born and inner city populations.1–3 Isoniazid (INH) is recommended as the drug of choice to treat latent tuberculosis infection (LTBI). Several adverse drug reactions (ADRs) are associated with INH, including hepatitis. A better understanding of the basis of this potentially life-threatening ADR is needed to inform preventive measures.4–6 The occurrence of ADRs related to INH, especially hepatotoxicity, has been well characterized.7 The incidence of INH-induced hepatotoxicity ranges from 1 to 36%, depending on different regimens, the population being treated and the definition of hepatic injury used.8,9 Alcohol consumption, presence of HIV, advanced age and chronic liver disease have been reported to increase the risk of INH-induced hepatotoxicity.10–14 8These authors co-led the study. N-acetyltransferase 2 (NAT2) is directly involved in INH metabolism, and genetic variation in the NAT2 has been reported to be a risk factor for INH- Received 23 August 2009; revised 22 November 2009; accepted 29 December 2009; published induced hepatotoxicity. There is wide variability in reported associations of NAT2 online 2 March 2010 variants with INH-induced hepatotoxicity in different populations;15–18 however, Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 525

controversy remains regarding the importance of NAT2 ing exons of the three genes in all study subjects. This not variation for this ADR in different ethnic groups. Factors only addresses the effect of genetic variation in these genes that may contribute to heterogeneity of NAT2/hepatotoxi- on INH-induced hepatotoxicity in our population, but also city associations include genetic differences between popu- provides an extensive catalog of genetic variation to support lations17 and contributions of variation in other genes to other pharmacogenetic analyses of these genes. this ADR. It is likely that NAT2 variation accounts for only a portion of INH-induced hepatotoxicity.15–18 An amidase (s) catalyzes two steps in the metabo- Patients and methods lism of INH.19,20 There is strong evidence from an animal model that amidase activity levels influence hepatotoxicity; Study subjects rabbits treated with an amidase inhibitor, bis-p-nitrophenyl All individuals in British Columbia who are identified to phosphate, at the same time they are dosed with INH fail to have LTBI are eligible to receive preventative treatment develop the severe hepatotoxicity developed by animals through a publicly funded program. We enrolled subjects treated with INH alone.21 Modulation of hepatic amidase receiving treatment with INH (300 mg daily) for LTBI at the activity therefore affects the development of hepatotoxicity, Vancouver or Victoria TB Clinics from 2004 to 2006. likely by altering INH metabolism. By extension, naturally Inclusion criteria were as follows: subjects were included if occurring genetic variation in amidase genes may account they were 19 years of age or older, not receiving other anti- for at least some of the variation in susceptibility to INH- TB drugs concurrently with INH, nonreactive to hepatitis B induced hepatotoxicity. Given that INH metabolism occurs surface antigen and negative for antibody to hepatitis C by in the liver and that toxic metabolites released there are the serology, not having any liver or metabolic diseases, without cause of hepatotoxicity, amidase genes expressed in the liver an HIV þ test result, not consuming seven or more alcoholic are logical candidate genes for INH hepatotoxicity. Candi- beverages per day and had sufficient aspartate aminotrans- date genes for INH-induced hepatotoxicity were chosen ferase (AST) monitoring to detect an INH-induced hepato- based on three criteria: their ability to cleave amide bonds, toxicity event. their expression in the liver and their inhibition by bis-p- We selected serum AST at baseline and follow-up as a nitrophenyl phosphate. Cleavage of amide bonds is carried marker of hepatotoxicity. Although alanine transaminase is out by , which can also cleave ester bonds. There are more specific for liver dysfunction, our local BC Centre for many esterases, but only a subset of them is inhibited by bis- Disease Control TB Clinic uses AST alone and considers a rise p-nitrophenyl phosphate, a selective inhibitor of ‘type B’ in AST after drug initiation without other confounders esterases, or and carboxylesterases (CES). (heart disease, muscle disease and so on) to be attributable to Cholinesterases ( and pseudocholines- medication. In addition, our research was conducted to terase) break down the neurotransmitter acetylcholine and, impact policy within our local TB Clinic and, as such, we as such, would not be expected to be involved in INH have conducted this study using measures that are consis- metabolism. tent with the Clinic’s policies and procedures. Three CES genes are well characterized in the human Information was collected regarding subject age, sex, genome: CES1, CES2 and CES3. All three are expressed in the ethnicity of each grandparent, concurrent medical illnesses liver. CES3 is more highly expressed in brain endothelial and alcohol and cigarette consumption. Confirmation of cells than in liver and has been suggested to function at the medication use, duration of treatment, all AST test dates and blood–brain barrier.22 CES1 (OMIM 114835) and CES2 results and hepatitis serology were obtained from the TB (OMIM 605278) are both highly expressed in the liver and Control database at the BC Centre for Disease Control. represent good candidates for genes involved in INH Baseline AST was measured before the initiation of INH metabolism. Both are located on human 16; treatment or as the first value entered into the subject’s CES1 has 14 exons and CES2 has 12.23 CES423 is a transcribed medical record within the first 2 weeks of treatment pseudogene located adjacent to and 28 kb upstream from initiation. Values were measured monthly thereafter until CES1. It has six transcribed exons and spans approximately treatment discontinuation or whenever subjects had symp- 14 kb.24 CES1 and CES4 have very high sequence similarity toms of suspected hepatitis (anorexia, nausea, vomiting, to each other; CES4 appears to be an inverted duplication of malaise and tea-colored urine). Serum hepatitis B virus CES1.25,26 Because of the proximity of CES1 and CES4 to surface antigen, immunoglobulin M antibody to hepatitis A each other and their similarity, we also characterized CES4 as virus and antibody to hepatitis C virus were tested at part of this study. baseline. Drug-induced hepatitis was defined according to To assess the effect of genetic variation in these three CES the criteria of the International Consensus Meeting in genes on INH-induced hepatotoxicity, we have performed a Paris27 as an increase in serum AST level more than two case-control-based association study of 170 subjects in times the upper limit of normal value during the 9-month British Columbia (BC) using single-nucleotide polymorph- treatment with INH, normalization of serum AST level after isms (SNPs) and haplotypes of these genes. Because these discontinuation of INH and a causality assessment score genes are novel candidate genes for this ADR, we have greater than 8. carried out genotyping for this study by means of complete Signed informed consent was obtained from each subject. re-sequencing of the promoter and all coding and noncod- This study was approved by the joint research ethics board

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 526

of the University of British Columbia and the BC Cancer Expand Long Range dNTPack PCR master mix (Roche, Basel, Agency. Switzerland). Subsequently, primers specific to exons 12, 13 and 14 were used to amplify nested PCR products for Genotyping of CES1, CES2 and CES4 by re-sequencing sequencing. A 30 ml blood sample was collected from each study subject Genotypes of the CES2, CES1 and CES4 gene variants were in EDTA tubes, 6 ml was immediately used for hepB and determined for each of the 170 subjects from bidirectional hepC testing and the remainder was frozen for later DNA sequence reads. Detection of polymorphisms was carried out extraction. DNA was extracted from blood of 170 hepB- and using PolyPhred35 and Consed36 or Mutation Surveyor hepC-negative subjects using the PureGene DNA isolation (SoftGenetics, State College, PA, USA). Each genotype was kit following the manufacture’s instructions (Gentra Sys- confirmed by independent examination of sequence traces tems, Minneapolis, MN, USA). DNA samples were quantified by two individuals. by fluorometry using PicoGreen (Invitrogen, Carlsbad, CA, USA) and a Victor2 fluorescence plate reader (Perkin-Elmer, Statistical methods Waltham, MA, USA). SNPs with less than 5% minor allele frequency (MAF) were For each of CES1, CES2 and CES4, all exons and intron/ excluded from statistical analyses. For each common SNP, exon boundaries including the 50 and 30 untranslated logistic regression was used to assess whether the genotypic regions (UTRs), 50 upstream region including the promoter and allelic frequencies were independently associated with and conserved noncoding sequences were sequenced in hepatotoxicity after adjustment for age and sex. When the each subject. Supplementary Online Table A lists the number of rare homozygotes was less than five, the rare amplicons, primer pairs and PCR conditions used. For homozygotes and heterozygotes were combined for analysis. CES1, a total of 5916 bp were PCR amplified in each sample Tests for trend were performed when five or more rare using 13 primer pairs and 3 nested primer pairs. For CES2,a homozygous alleles were present. Hardy–Weinberg equili- total of 8825 bp were PCR amplified using 25 primer pairs brium was tested in the control groups by Fisher’s exact test. and for CES4, a total of 5738 bp were sequenced using 14 Adjustment was made for multiple testing by computing the 37 primer pairs. The VISTA browser28 was used to identify false discovery rate. 38 conserved noncoding sequences with 70% or greater The Haplo.stats package was used to test associations sequence similarity over at least 100 bp between human, between statistically inferred haplotypes and hepatotoxicity, mouse and rat orthologous gene sequences.28,29 Exons were and to calculate adjusted odds ratios (ORs) and 95% amplified using primers designed in the intronic sequences confidence intervals for each haplotype. Haplotypes with near the exon boundaries to allow re-sequencing across all frequencies 45% were tested. Rare haplotypes (with in- splice sites. The 50 and 30 UTRs were amplified in over- dividual frequencies of o5%) were combined in the lapping segments. Primer design was completes using CES2, association test. When no haplotypes showed 45% fre- CES1 and CES4 genomic sequences (NM_001266, quency, the threshold for grouping rare haplotypes was NM_003869 and NM_016280, respectively) retrieved from lowered to 3%. All analyses were performed with SAS version the UCSC genome browser23 and Primer3.30 In silico PCR31 9 (SAS, Cary, NC, USA) and R version 2.8 (The R Project for was used to check primer specificity. All PCR primers Statistical Computing; www.r-project.org). incorporated M13 forward or reverse sequencing tags to allow efficient sequencing of the PCR products. PCR, Results sequencing and sequence analysis were carried out as described previously,32 with the exception that we used A total of 170 subjects were enrolled (Table 1). The mean Big Dye Terminator Mix v3.1 (Applied Biosystems, Foster duration of treatment was 239.6 days (s.d. 96.8) and the City, CA, USA) at 0.33 ml of mix per reaction in a total mean age was 40.8 years (s.d. 12.1). More than half of the volume of 4 ml with 50 cycles amplification. Haploview33 participants were women (61.6%). Asian was the most was used to calculate and display inter-SNP linkage dis- common ethnic group (40.7%), followed by Caucasian equilibrium (LD) information derived from sequence data. (30.5%). The mean AST at baseline was 23.7 (s.d. 6.4). There CES1 and CES4 are located adjacent to each other and in were no statistically significant differences between cases opposite orientation (Figure 1a), and show high sequence and controls with respect to demographic variables and similarity to each other (Figure 1b). Exons 12, 13 and 14, baseline AST. Twenty-three subjects (13.5%) met our criteria and adjacent intron sequences of CES1 are nearly 100% for INH-induced hepatitis. The maximum AST for these identical to predicted but nontranscribed exons of CES4.To subjects was a mean of 125 (s.d. 131) compared to 31 (s.d. 9) specifically amplify exons 12, 13 and 14 of CES1, we used for the controls. The case group had significantly higher nested PCR34 with a first primer set positioned such that the maximum AST and during-treatment AST than controls. forward primer had one mismatch with the CES4 sequence Bidirectional sequencing was used to detect and genotype and the reverse primer was located outside the nearly genetic variants in each gene in each of the 170 subjects. Of identical region. Unique localization of this specific PCR the 57 amplicons used for CES1, CES2 and CES4, four had primer pair was verified using NCBI Blast.26 This primer set low-quality sequence reads in both directions. Three amplified a single 7959-bp-long PCR amplicon containing amplicons had mononucleotide stretches that prevented these three exons of CES1. Long PCR was carried out using good quality sequence reads in one direction but not in the

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 527

Table 1 Characteristics of the cases and controls approximately 10 kb in the CES4 gene in some samples, as revealed by 14 samples failing to amplify PCR products from All subjects (n ¼ 170) exon 2 to the 30 end of the gene (Figure 1). Sequencing of the samples that did not have a large deletion revealed 85 Case, n (%) Control, n (%) Total n (%) variants within this region, including 64 (59.3%) transi- tions, 40 (37.0%) transversions and 4 (3.7%) small insertions AST level (U lÀ1) Baseline, mean (s.d.) 19.5 (3.2) 23.0 (5.4) 23.7 (5.9) or deletions. Maximum, median (IQR) 142 (107–293) 36 (27–57) 38 (29–76) Of the 48 observed variants in CES2, 31 (64.6%) were singletons (observed only once in this data set). Six variants Gender (12.5%) had MAF of 5% or greater. In contrast, only 37 of the Female 20 (87.0) 84 (57.1) 104 (57.1) Male 3 (13.0) 63 (42.9) 66 (42.9) 109 variants observed in the CES1 and nondeletion region of CES4 (33.9%) were singletons. Forty-four of these (40.7%) Age group (years) had an MAF of 5% or greater. 20–29 5 (21.7) 35 (23.8) 40 (23.8) The large polymorphic deletion in CES4 spanning exons 2 30–39 6 (26.1) 38 (25.9) 44 (25.9) 40–49 5 (21.7) 39 (26.5) 44 (26.5) to 6 was characterized by both PCR and sequencing. 50–71 7 (30.4) 35 (23.8) 42 (23.8) Homozygous deletion was deduced for 14 DNA samples in which exons from 2 to 6 did not produce PCR products. PCR Ethnicity alone, however, could not distinguish homozygous nonde- Asian 8 (34.8) 64 (43.5) 72 (43.5) Caucasian 10 (43.5) 39 (26.5) 49 (26.5) leted samples from heterozygous samples. For that, sequence South Asian 0 (0.0) 22 (15.0) 22 (15.0) data from PCR products within the deleted region were used Hispanic 0 (0.0) 7 (4.8) 7 (4.8) to categorize samples. Ninety-one samples that did not Middle Eastern 0 (0.0) 8 (5.4) 8 (5.4) appear to be heterozygous at any of 85 polymorphic sites First Nations 2 (8.7) 3 (2.0) 5 (2.0) Other/Mixed/Unknown 3 (13.0) 4 (2.7) 7 (2.7) within the deleted region were interpreted to be hemizygous, that is, heterozygous for the large deletion. Fifty-seven Ever smoked samples that showed three or more heterozygous sites within Yes 9 (39.1) 41 (27.9) 50 (27.9) the deletion region were interpreted to be nondeleted on No 14 (60.9) 105 (71.4) 119 (71.4) Unknown 0 (0.0) 1 (0.7) 1 (0.7) both alleles. Sixteen samples with only one or two hetero- zygotes were conservatively considered ambiguous due to the Alcohol consumption potential for PCR error to create false heterozygotes. None 17 (73.9) 107 (72.8) 124 (72.8) Figures 2a and b show the inter-SNP LD, in terms of pair- 1–2 drinks per week 2 (8.7) 33 (22.4) 35 (22.4) 2 3–7 drinks per week 4 (17.4) 7 (4.8) 11 (4.8) wise r values, calculated in our data for common variants (MAF X5%). A block of LD encompasses SNP10, SNP11 and BMI (kg mÀ2) SNP13–21; LD among these 11 SNPs is high (r2 ¼ 0.83–1.00). Underweight, o18.5 2 (8.7) 16 (10.9) 18 (10.9) Overall, the CES1–CES4 region shows three LD blocks. LD Normal, 18.5–24.9 15 (65.2) 81 (55.1) 96 (55.1) Overweight, 25–29.9 3 (13.0) 36 (24.5) 39 (24.5) across CES2 is lower. Obese, 30+ 3 (13.0) 14 (9.5) 17 (9.5) Hardy–Weinberg equilibrium was tested in Caucasian controls and Asian controls separately, for each of the 6 Total 23 147 170 common SNPs in CES2 and 44 common SNPs in CES1 and Abbreviations: AST, aspartate aminotransferase; BMI, body mass index. CES4. Genotype frequencies of CES2 variants did not deviate P-values for differences in sociodemographic characteristics between cases and significantly from Hardy–Weinberg equilibrium. Seven controls were as follows: gender, 0.01; age group, 0.67; ethnicity, 0.52; smoking, 0.28; alcohol consumption, 0.39 and BMI, 0.90. variants in CES1 and CES4 deviated significantly from Hardy–Weinberg equilibrium (one in Asian controls, six in Caucasian controls; see Supplementary Online Table B). other. Six amplicons contained a small insertion or deletion These variants were excluded from all analyses. in more than 5% of individuals, causing those sequence Table 3 shows the results of association tests of each reads to appear superimposed, even though they were of common variant in CES1 with INH-induced hepatotoxicity. good quality. Good quality sequence reads were obtained in A small group of individuals with other ethnicity including both forward and reverse direction for 77.2% of sample/ mixed/unknown was excluded from these analyses, which amplicon combinations. On average 93.0% of sample/ were adjusted for age and sex. None of the 28 common amplicon combinations had good quality sequence reads variants was associated with a significantly increased risk of in at least one direction. INH-induced hepatotoxicity. We did observe initially a We observed 48 variants, including 38 novel variants, in significant result at Po0.05 in all subjects. However, this the CES2 gene and 194 variants in the CES1–CES4 region (40 P-value did not remain significant after correction for and 42 novel, respectively). Genetic variants are listed in multiple testing. There were no statistically significant SNPs Table 2. Of these 48 variants, 31 (64.6%) were transitions, 13 results associated with an increased risk of INH-induced (27.1%) were transversions and 4 (8.3%) were small inser- hepatotoxicity in CES2 or CES4 (data not shown). tions or deletions. Of the coding region changes, seven Table 4 summarizes the results of association tests of resulted in amino-acid changes. There was a large deletion, haplotypes predicted using the Haplo.stats package38 in the

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 528

Table 2 Variants observed by re-sequencing CES genes in 170 individuals

Gene Variant Name of variant dbSNP reference Flanking sequencea Nucleotide Codon Amino-acid Minor allele number (build 127) change change change frequency (%)

CES2 1 CNS2 (À8811) G/A NA gtgagg/atattt G/A NA NA 0.29 2 À8585 A/G NA cggtaa/ggatta A/G NA NA 0.29 3 CNS5 (À7182) T/C NA ccctgt/caccct T/C NA NA 1.17 4 CNS5 (À7080) C/T rs28531829 aggcac/ttgatg C/T NA NA 0.58 5 CNS6 (À5156) G/A rs17767961 agtctg/atccag G/A NA NA 3.51 6 À5019 G/T NA tgattg/tagagg G/T NA NA 0.88 7 À2343 T/G NA ctcctt/gcctca T/G NA NA 0.29 8 À2342 C/T NA tccttc/tctcag C/T NA NA 0.58 9 À2282 ins(T) NA ttttg*/ttattt ins(T) NA NA 0.29 10 À2263 A/G rs3759994 atgaga/gtttca A/G NA NA 21.93 11 À2259 C/A rs28382809 gatttc/aaccat C/A NA NA 0.29 12 CNS7 (À1993) C/T NA caggcc/tccact C/T NA NA 0.29 13 À1668 G/A NA ggagag/aatgtc G/A NA NA 2.05 14 À1578 C/A NA aggcac/aaacaa C/A NA NA 0.29 15 À1513 G/A NA ctgctg/acacag G/A NA NA 0.29 16 50 UTR (À936) G/A NA tgcgag/aagaag G/A NA NA 0.29 17 50 UTR (À630) A/G NA gcacca/gtaggc A/G NA NA 0.29 18 50 UTR (À594) C/A NA tggacc/aggcag C/A NA NA 0.29 19 50 UTR (À568) A/T NA tggaca/tggcag A/T NA NA 0.29 20 50 UTR (À503) A/G NA aggaaa/gagggg A/G NA NA 0.29 21 50 UTR (À319) G/A NA gcgccg/aacact G/A NA NA 0.29 22 50 UTR (À171) C/G rs11075646 tcgatc/gcccca C/G NA NA 6.43 23 X1 (+5) C/G NA tatgac/gtgctc C/G ACToAGT Thr 2 Ser 0.29 24 X2 (+407) G/A NA gctgcg/aatttg G/A CGAoCAA Arg 136 Gln 0.29 25 CNS9 (À152) T/C rs2303218 tggcct/cgtaac T/C NA NA 9.65 26 IVS3 (À49) G/T NA ctggtg/tgggtt G/T NA NA 0.29 27 X5 (+801) T/G NA gtggct/ggcact T/G GCToGCG Ala 267 Ala 0.58 28 X5 (+808) C/T NA cactac/tgctgg C/T CGCoTGC Arg 270 Cys 0.29 29 IVS6 (+54) G/A NA gggctg/agcagg G/A NA NA 0.29 30 IVS6 (+70) C/T rs11568310 acggcc/tgtcat C/T NA NA 0.29 31 IVS7 (+35) A/G NA tgggga/ggccca A/G NA NA 0.29 32 X8 (+1322) C/T NA gttaac/tgctgc C/T ACGoATG Thr 441 Met 0.29 33 X9 (+1366) G/A NA gggagg/aagtac G/A GAGoAAG Glu 456 Lys 0.29 34 IVS9 (+16) C/T NA ttggac/tcaaag C/T NA NA 0.29 35 IVS9 (À39) G/C NA ttcagg/cgggag G/C NA NA 0.29 36 X10 (+1499) A/G NA cttcta/gcgagt A/G TACoTGC Tyr 500 Cys 0.29 37 X10 (+1500) C/T NA ttctac/tgagtt C/T TACoTAT Tyr 500 Tyr 1.17 38 X10 (+1589) G/C NA tttcag/caagtt G/C AGAoACA Arg 530 Tyr 0.29 39 X12 (+1839) C/T rs28382827 gagctc/tgagga C/T CTCoCTT Leu 613 Leu 0.88 40 30 UTR (À69) A/G rs8192925 cacaca/gcccac A/G NA NA 12.57 41 30 UTR (À193) C/G NA aatccc/gagcta C/G NA NA 7.49 42 30 UTR (À232) C/T NA gaggcc/tagagg C/T NA NA 0.29 43 30 UTR (in poly A stretch) ins NA Unknown ins(?) NA NA 21.05 44 30 UTR (in poly A stretch) ins NA Unknown ins(?) NA NA 0.88 45 30 UTR (À372) T/C NA attttt/caaaa T/C NA NA 0.29 46 30 UTR (À749) del(AAG) NA agaatag/***agcta del(AAG) NA NA 0.29 47 30 UTR (À849) G/A rs28382829 accccg/agtgga G/A NA NA 1.46 48 30 UTR (À878) C/T NA ccccac/ttgagc C/T NA NA 0.29 CES1 1 À1155 C/T NA tacacc/tgcaca C/T NA NA 0.29 2 À1114 G/A rs34428341 acaacg/acatcc G/A NA NA 37.13 3 À1033 C/T NA cggtac/tcacag C/T NA NA 0.29 4 À780 T/C NA ctttgt/catcca T/C NA NA 2.34 5 À756 delC rs35346303 caggtgc/*aagca del C NA NA 17.25 6 À686 T/C NA gactat/cggggg T/C NA NA 0.88 7 À614 C/T NA gacacc/tgatgg C/T NA NA 5.26 8 À272 A/G rs7498748 tgggca/gagttt A/G NA NA 5.26 950 UTR (À75) T/G rs3815583 tgggct/gccagg T/G NA NA 31.29 10 50 UTR (À46) A/G rs28429139 tctgaa/gctgca A/G NA NA 20.76 11 50 UTR (À39) A/G rs28494177 tgcaca/ggagac A/G NA NA 22.22 12 50 UTR (À30) G/C NA acctcg/ccaggc G/C NA NA 0.88 13 50 UTR (À21) G/C rs28520463 gccccg/cagaa G/C NA NA 20.76 14 50 UTR (À20) A/G rs28499065 ccccga/ggaact A/G NA NA 21.05 15 50 UTR (À2) C/G rs28515828 ttccac/ggatgt C/G NA NA 20.47 16 X1 (+11) G/C NA gctccg/ctgcct G/C CGT4CCT Arg 4 Pro 19.88 17 X1 (+15) C/T NA cgtgcc/ttttat C/T GCC4GCT Ala 5 Ala 20.47 18 X1 (+17) T/C NA gtgcct/cttatc T/C TTT4CTT Phe 6 Leu 20.47

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 529

Table 2 Continued

Gene Variant Name of variant dbSNP reference Flanking sequencea Nucleotide Codon Amino-acid Minor allele number (build 127) change change change frequency (%)

19 X1 (+19) A/G NA ccttta/gtcctg A/G ATC4GTC Ile 7 Val 20.47 20 X1 (+34) T/C rs28563878 ctctct/cctgct T/C TCT4CCT Ser 12 Pro 20.47 21 IVS1 (+16) A/G rs12149359 ctgaaa/gtcaaa A/G NA NA 20.18 22 IVS1 (+22) del(T) NA tcaaat/*atgcgg del(T) NA NA 41.81 23 IVS1 (+30) G/T NA gcgggg/tcactt G/T NA NA 0.29 24 IVS1 (À15) A/C rs3826189 gattta/cttctc A/C NA NA 0.29 25 X2 (+56) G/T rs3826190 agcagg/tgcatc G/T GGG4GTG Gly 19 Val 0.58 26 X2 (+227) G/A rs2307240 atggag/actttg G/A AGC4AAC Ser 75 Asn 3.51 27 IVS2 (+62) C/A NA caaggc/aagtcc C/A NA NA 0.88 28 IVS2 (+74) T/C rs3848300 cctgat/cgggct T/C NA NA 30.70 29 IVS2 (+125) A/G NA taagaa/gcattc A/G NA NA 2.05 30 IVS2 (+280) A/T NA gggaga/tcagat A/T NA NA 0.29 31 IVS2 (+308) C/T NA tactcc/tgtgga C/T NA NA 0.29 32 IVS2 (+402) A/T rs4784574 catgaa/tttctg A/T NA NA 3.80 33 IVS3 (+17) C/T rs2307228 gacccc/tctggt C/T NA NA 0.29 34 IVS3 (À76) T/C NA gtaggt/ccacca T/C NA NA 0.29 35 X4 (+514) C/T NA aatatc/tgcctg C/T CGC4TGC Arg 172 Cys 0.29 36 IVS4 (+10) C/T rs2307244 gaaatc/tggact C/T NA NA 4.12 37 IVS4 (+66) T/G rs2307236 gctctt/ggtcat T/G NA NA 12.57 38 IVS4 (+127) A/G rs28415252 ccttca/gtaatt A/G NA NA 39.47 39 IVS4 (+245) T/A NA cttcat/acctgc T/A NA NA 1.46 40 IVS4 (+247) C/T NA tcatcc/ttgcaa C/T NA NA 0.29 41 IVS4 (+269) A/G rs34617642 cagtta/gcctcc A/G NA NA 0.29 42 IVS4 (+347) G/T rs8192938 agcagg/tcagc G/T NA NA 41.52 43 IVS4 (+382) C/G NA gacttc/gtcatg C/G NA NA 1.46 44 X5 (+560) G/A NA cagccg/aggga G/A CGG4CAG Arg 187 Gln 0.29 45 X5 (+599) G/A rs2307243 cctgcg/actggg G/A CGC4CAC Arg 200 His 0.58 46 X5 (+612) C/A rs2307227 caggac/aaacat C/A GAC4GAA Asp 204 Glu 1.46 47 IVS5 (+32) C/G rs2307234 ccccac/ggcttg C/G NA NA 1.46 48 IVS5 (+129) delC rs3217164 actcac/*agccca del C NA NA 38.01 49 IVS5 (+212) G/C NA ctgggg/cccata G/C NA NA 1.17 50 IVS5 (+246) G/A NA atggtg/agcatt G/A NA NA 0.29 51 IVS5 (À64) G/T rs2307235 tggaag/tagcga G/T NA NA 41.23 52 IVS5 (À61) T/C rs2307233 aatagt/cgagtg T/C NA NA 41.23 53 IVS5 (À21) A/T rs2307229 tctcca/tcccct A/T NA NA 41.23 54 X6 (+717) C/T rs17850910 aagaac/tctctt C/T AAC4AAT Asn 239 Asn 0.29 55 X6 (+747) C/T NA agtggc/tgtggc C/T GGC4GGT Gly 249 Gly 0.29 56 IVS6 (+15) G/A NA ggctgg/atacgt G/A NA NA 0.29 57 IVS6 (+25) C/T rs2307237 gtctcc/tggctg C/T NA NA 0.28 58 IVS6 (+28) C/A rs35856864 cacccc/acacct C/A NA NA 0.29 59 X7 (+808) G/T NA aaattg/tctatc G/T GCT4TCT Ala 270 Ser 1.46 60 X7 (+846) C/A NA gctgtc/aatggt C/A GTC4GTA Val 282 Val 0.29 61 X7 (+855) C/G NA gttcac/gtgcct C/G CAC4CAG His 284 Gln 0.29 62 IVS7 (+1) G/T rs4513095 aaatgg/ttaggt G/T NA NA 2.63 63 IVS7 (+6) T/C NA gtaggt/ctgcct T/C NA NA 0.88 64 IVS7 (À21) C/T NA cacatc/tctctg C/T NA NA 0.29 65 IVS7 (À5) C/T NA gtcttc/tacaga C/T NA NA 0.29 66 IVS8 (+18) C/T NA tgtttc/ttggatt C/T NA NA 0.29 67 IVS8 (+28) G/A NA ttacgg/agtttt G/A NA NA 0.29 68 IVS9 (+11) A/C NA aaggca/ccatgt A/C NA NA 0.29 69 IVS9 (À5) T/G rs28694729 aaggca/ccatgt T/G NA NA 43.86 70 IVS10 (À41) C/T rs2244614 tgcatc/tgctca C/T NA NA 44.44 71 IVS10 (À33) A/C rs2244613 tcacca/cggggc A/C NA NA 39.77 72 X11 (+1259) A/G NA cctgga/gcttga A/G GAC4GCG Asp 420 Gly 0.29 73 X12 (+1410) G/C NA cacggg/cgatga G/C GGG4GGC Gly 470 Gly 0.29 74 IVS13 (+19) A/G NA caaaga/gcagag A/G NA NA 2.05 75 30 UTR (+14) C/T NA ccagcc/tggcct C/T NA NA 0.29 CES4 1 À1085 C/T NA aataag/atagct G/A NA NA 2.08 2 À1068 T/C NA atcact/catttt T/C NA NA 0.58 3 À1028 C/T rs7198663 cggtac/tcacag C/T NA NA 30.65 4 À859 A/C rs3785161 atcaca/ccctac A/C NA NA 22.62 5 À812 C/T NA aacccc/tttgag C/T NA NA 0.30 6 À783 A/G NA aattga/gccttt A/G NA NA 0.30 7 À782 C/T NA attgac/tctttg C/T NA NA 0.30 8 À717 C/T rs1974708 atgggc/tacatg C/T NA NA 27.38 9 À682 T/C NA agctat/ctgaga T/C NA NA 1.79

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 530

Table 2 Continued

Gene Variant Name of variant dbSNP reference Flanking sequencea Nucleotide Codon Amino-acid Minor allele number (build 127) change change change frequency (%)

10 À585 G/A NA ggctag/attggc G/A NA NA 0.60 11 À470 C/A rs1974709 agctgc/agatat C/A NA NA 34.71 12 À443 C/T NA ctgagc/ttgtga C/T NA NA 0.29 13 À422 A/G NA gctcta/gacatt A/G NA NA 1.18 14 À258 T/C NA agctct/cctgta T/C NA NA 0.29 15 À253 A/G rs7199731 tctgta/gatctg A/G NA NA 6.76 16 À251 T/A NA tgtaat/actgag T/A NA NA 0.59 17 PROM (À160) G/A NA atctgg/aggaaa G/A NA NA 0.29 18 50 UTR (À107) T/C rs3815576 ggcagt/cgcagg T/C NA NA 21.35 19 50 UTR (À90) G/C rs3815577 taactg/cggggc G/C NA NA 21.35 20 50 UTR (À88) G/T rs3815578 aactgg/tgggcc G/T NA NA 21.35 21 50 UTR (À83) C/G rs3815579 ggggcc/gagggt C/G NA NA 21.35 22 50 UTR (À82) A/G rs3815580 gggcca/ggggtg A/G NA NA 21.35 23 50 UTR (À79) G/C rs3815581 ccaggg/ctggcg G/C NA NA 21.35 24 50 UTR (À75) ins(G) NA gggtgg*/gcgcca ins(G) NA NA 21.35 25 50 UTR (À74) G/T NA gtggcg/tccagg G/T NA NA 21.35 26 50 UTR (À38) A/G NA gctgca/gcggag A/G NA NA 0.29 27 50 UTR (À39) C/T rs7199449 ctgcac/tggaga C/T NA NA 9.65 28 50 UTR (À24) C/T NA caggcc/tcccgg C/T NA NA 0.58 29 X1 (+26) C/G NA cctggc/gcactc C/G NA NA 0.58 30 IVS1 (+30) G/T rs3815584 gcgggg/tcactt G/T NA NA 37.43 31 IVS1 (+60) A/G NA gccgaa/gctggg A/G NA NA 1.46 32 IVS1 (+61) C/G NA ccgaac/gtgggc C/G NA NA 1.46 33 X2-30 downstream NA NA large deletion NA NA 33.30 Deleted 1 IVS1 (À132) G/C ENSSNP1227839 tacatg/caagag G/C NA NA 39.24 region in CES4 2 IVS1 (À129) C/G ENSSNP1227840 atcaac/gagaaa C/G NA NA 41.28 3 IVS1 (À75) T/G rs35859463 ttagat/gagtgg T/G NA NA 39.10 4 IVS1 (À73) G/C ENSSNP6456146 agagag/ctggga G/C NA NA 47.45 5 IVS1 (À64) C/T ENSSNP6502153 gaaacc/tccacc C/T NA NA 47.77 6 IVS1 (À6) G/A rs34801523 tccatg/atccag G/A NA NA 39.10 7 X2 (+56) T/G ENSSNP1227843 agcagt/ggcatc T/G NA NA 41.72 8 X2 (+62) C/T ENSSNP1227844 gcatcc/tgtcct C/T NA NA 41.08 9 X2 (+73) C/A NA cgccac/actttg C/A NA NA 0.32 10 X2 (+76) G/T ENSSNP1227845 cacctg/ttggtg G/T NA NA 39.42 11 X2 (+88) G/C ENSSNP1227846 acaccg/ctgcat G/C NA NA 39.74 12 X2 (+115) A/G rs34649320 agttca/gtcagc A/G NA NA 39.42 13 X2 (+148) G/A NA tggccg/attttc G/A NA NA 0.32 14 X2 (+177) C/G rs35384853 agcccc/gcctct C/G NA NA 39.17 15 X2 (+203) C/T ENSSNP1227849 tactcc/taccac C/T NA NA 39.56 16 X2 (+207) G/A ENSSNP1227850 ctaccg/acagcc G/A NA NA 39.24 17 X2 (+222) G/A ENSSNP1227851 gagccg/atggaa G/A NA NA 38.92 18 X2 (+227) G/A ENSSNP1227852 atggag/actttg G/A NA NA 38.92 19 X2 (+248) T/C ENSSNP1227853 cacctt/cgtacc T/C NA NA 39.56 20 IVS2 (+6) ins(CAGGGGTGGC) NA gtaagc**********/cagg ins(CAGGG NA NA 38.92 GTGGC) 21 IVS2 (+19) C/T rs28408958 ggctgc/tggcat C/T NA NA 40.06 22 IVS2 (+43) A/G rs28415966 tgttca/gcctca A/G NA NA 39.74 23 IVS2 (+49) A/C rs34595220 cctcaa/cagtga A/C NA NA 40.58 24 X3 (+263) T/G NA caggtt/gcaccc T/G NA NA 1.27 25 X3 (+295) A/C rs28499390 agttaa/ctctca A/C NA NA 2.22 26 X3 (+331) C/T NA acattc/tctctc C/T NA NA 0.32 27 IVS3 (+466) T/A rs28685672 cttcat/aatcag T/A NA NA 51.27 28 IVS3 (+521) C/T rs28719073 tccagc/tacatc C/T NA NA 38.85 29 CNS1 (+584) G/A NA acatgg/agacaa G/A NA NA 1.59 30 CNS1 (+630) A/G rs28414633 tctgga/gatttt A/G NA NA 39.87 31 CNS1 (+671) A/C rs28520242 ggcaaa/cggctg A/C NA NA 40.82 32 CNS1 (+672) T/G rs28657846 gcaact/ggctga T/G NA NA 38.92 33 CNS1 (+717) A/G rs28376804 agggca/gggaga A/G NA NA 38.92 34 CNS1 (+745) G/A NA tgcagg/attgga G/A NA NA 0.32 35 CNS1 (+761) G/C NA gggagg/cgtctg G/C NA NA 0.32 36 IVS3 (À304) T/C rs28372062 acatct/cttagc T/C NA NA 57.28 37 IVS3 (À242) T/C rs34460446 aggcat/ctggag T/C NA NA 49.05 38 IVS3 (À187) C/T rs34460446 ggtggc/tttggg C/T NA NA 10.13 39 IVS3 (À111) A/G ENSSNP1227894 tttgga/gtagca A/G NA NA 38.92

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 531

Table 2 Continued

Gene Variant Name of variant dbSNP reference Flanking sequencea Nucleotide Codon Amino-acid Minor allele number (build 127) change change change frequency (%)

40 X4 (+462) T/C ENSSNP1227895 acctat/cgatgg T/C NA NA 38.92 41 X4 (+479) C/T NA ccttgc/ttgccc C/T NA NA 0.63 42 X4 (+524) G/A NA cctggg/acatct G/A NA NA 3.48 43 IVS4 (+10) T/C ENSSNP1227896 gaaatt/cggact T/C NA NA 50.32 44 IVS4 (+21) A/T NA tcctca/tctgca A/T NA NA 0.32 45 IVS4 (+66) G/T rs2307236 gctctg/tgtcat G/T NA NA 38.92 46 X5 (+573) T/G rs28538364 aactgt/gggtca T/G NA NA 46.28 47 X5 (+644) G/C NA cccagg/cctctg G/C NA NA 2.22 48 X5 (+671) C/T rs28366434 gtcagc/tgggag C/T NA NA 50.32 49 IVS5 (+17) C/G NA tggcac/gcgggc C/G NA NA 0.32 50 IVS5 (À64) T/G ENSSNP6744093 tggaat/gagcaa T/G NA NA 50.95 51 IVS5 (À60) G/A ENSSNP1227918 agagcg/aagtga G/A NA NA 40.19 52 IVS5 (À33) C/A NA ctgagc/aatgaa C/A NA NA 0.63 53 X6 (+829) T/C ENSSNP1227919 gtctct/cggctg T/C NA NA 38.92 54 X6 (+842) A/C rs28444662 caccca/ccacct A/C NA NA 43.99 55 30 UTR (+25) T/C rs35566026 gtggtt/cggttg T/C NA NA 51.27 56 30 UTR (+238) C/G NA ccacac/gtccac C/G NA NA 26.90 57 30 UTR (+268) G/A rs28444264 gccatg/attggg G/A NA NA 38.92 58 30 UTR (+353) C/A rs28547176 ccacac/atgagc C/A NA NA 12.03 59 30 UTR (+359) G/A NA tgagcg/acctga G/A NA NA 0.32 60 30 UTR (+431) C/T NA ggtgac/ttaagt C/T NA NA 0.63 61 30 UTR (+456) G/A NA gaaatg/aggcag G/A NA NA 0.63 62 30 UTR (+551) T/C rs3743789 agacat/cctaac T/C NA NA 38.92 63 30 UTR (+565) A/G NA tcccca/gagctt A/G NA NA 0.32 64 30 UTR (+609) C/A rs28647208 agaatc/aacaca C/A NA NA 0.32 65 30 UTR (+707) A/C rs3743790 ctggaa/cgaggt A/C NA NA 49.36 66 30 UTR (+878) A/G rs28628572 ggagca/gcggcg A/G NA NA 49.05 67 30 UTR (+979) del(AGG) NA ggaag/***aggtgat del(AGG) NA NA 0.32 68 30 UTR (+1053) A/C rs28393348 ggtaga/ctttga A/C NA NA 50.32 69 30 UTR (+1222) C/T ENSSNP11136630 attagc/tctggt C/T NA NA 40.91 70 30 UTR (+1268) G/A rs28576997 agaccg/aaatga G/A NA NA 43.48 71 30 UTR (+1319) G/A rs28456031 caacag/aagtga G/A NA NA 43.48 72 30 UTR (+1322) T/G ENSSNP11136632 cagagt/ggatac T/G NA NA 34.75 73 30 downstream (À3184) G/A rs28406579 acttcg/actctt G/A NA NA 38.29 74 CNS2 (À3304) G/C rs28434448 tcttgg/cagacg G/C NA NA 0.32 75 30 downstream (À3350) A/C ENSSNP11136651 caaaca/cctgta A/C NA NA 0.32 76 30 downstream (À3433) G/A rs28703854 tctcgg/agagtc G/A NA NA 0.32 77 CNS3 (À6066) T/C rs28666799 ccactt/caccca T/C NA NA 38.92 78 CNS3 (À6114) C/T rs28538101 tctgcc/ttttgt C/T NA NA 38.61 79 CNS3 (À6158) T/C rs28418425 gagact/ctcaaa T/C NA NA 38.61 80 CNS3 (À6172) G/A NA gtaagg/aacctt G/A NA NA 0.32 81 CNS3 (À6226) T/G rs28613114 ccaatt/gaactg T/G NA NA 38.61 82 30 downstream (À6261) A/G NA tggtta/gtagta A/G NA NA 0.32 83 30 downstream (À6264) A/G rs28594163 ttataa/gtaact A/G NA NA 38.29 84 30 downstream (À6298) G/A rs28715312 agagag/agggta G/A NA NA 2.53 85 30 downstream (À6356) T/C rs28615435 gaacct/catttt T/C NA NA 0.63

*Inserted nucleotides. aEach variant site is in bold.

high LD region of CES1, including SNP10, SNP11 and Phen.39,40 This variant, however, was not predicted to be SNP13–21. We compared haplotype frequencies in all cases functional by SIFT41 (see Supplementary online Table D). vs all controls, with adjustment for age and sex. Rare haplotypes (frequency o5%) were combined in the analysis. However, no significant association was observed in all Discussion subjects. Haplotype analysis was also performed for CES2 (data not CES can activate prodrugs containing ester linkage and shown) and for the entire CES1–CES4 region (Supplemen- increase the solubility and bioavailability of therapeutic tary online Table C). None of these haplotype-based agents.42 They are also involved in the metabolism of illegal association tests achieved significance. SNP16 was predicted drugs such as heroin and cocaine.43 Genetic variants in both to be ‘possibly damaging’ to protein function by Poly- CES1 and CES2 genes may contribute to ADRs and increased

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 532

Gene structure of CES1 and CES4 CES4 CES1 123456

14 13 12 11 10 9 876 5 4 3 2 1 large polymorphic deletion

Sequence similarity between CES1 and CES4 Long Range PCR

123 456 78910117891011121314 CES1

Nested PCRs

90% 95% 97% 99% 96% 100% 97%94% 92% 92% 91% 100% 100% 100% CES4

123 4 5 6 Those exons are predicted, but not transcribed.

Figure 1 The CES1–CES4 locus. (a) Genomic organization of the CES1 and CES4 genes. (b) Sequence similarity between the CES1 and CES4 genes. Red boxes indicate 499%, orange boxes 95–99% and yellow boxes indicate o95% sequence similarity, as estimated by NCBI blast.26 The diagrams are not to scale.

CES1 CES1 and CES4 CES4

CES2

Figure 2 Linkage disequilibrium across the candidate genes. (a) CES1 and CES4.(b) CES2. This plot was generated using sequence data for 44 common variants in CES1 and CES4 in 170 individuals. The variants are listed in order of their physical position. Blocks with no value indicate an r2 of 1.0. This figure was generated using Haploview33 with some modifications. The diagrams are not to scale. Numbering of SNPs is according to Table 2. sensitivity or resistance to drug treatment. For example, an irinotecan.46 CES are clearly important in a variety of SNP of the CES1 gene has been associated with responsive- pharmacogenetic phenotypes. No studies to date, however, ness to imidapril,44 an angiotensin-converting enzyme have addressed CES genes in drug-induced hepatotoxicity. inhibitor used to treat hypertension and congestive heart Other genes have been assessed for association with failure. Rare nonsynonymous variants in CES1 encode hepatotoxicity induced by INH alone and/or with other with impaired activities that dramatically alter drugs, including NAT2, CYP2E1, glutathione S- the pharmacokinetics and drug response to the psychosti- M1, glutathione S-transferase T1 and HLA variants.47–49 We mulant methylphenidate.45 Several CES2 variants have been have recently carried out a pharmacogenetics study of NAT2 shown to be functionally deficient, and some have and CYP2E1 on INH-induced hepatotoxicity using the same decreased activities toward the anticancer agent study population; although no single variant showed a

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 533

Table 3 ORs and 95% CI’s for association tests of CES1 SNPs Table 3 Continued with hepatotoxicity SNP All subjects (n ¼ 170) SNP All subjects (n ¼ 170) Cases, n Controls, n OR (95% CI) P Cases, n Controls, n OR (95% CI) P SNP19 SNP2 AA 16 93 1.00 GG 8 64 1.00 0.29* AG 7 47 0.97 (0.36–2.61) 0.96 GA 9 60 1.13 (0.39–3.25) 0.8186 GG 0 7 AA 6 23 2.06 (0.60–6.99) 0.2486 SNP20 SNP5 TT 16 93 1.00 wt/wt 17 102 1.00 TC 7 47 0.97 (0.36–2.61) 0.96 wt/del 6 39 0.80 (0.27–2.32) 0.6743 CC 0 7 del/del 0 6 SNP21 SNP7 AA 16 93 1.00 CC 23 130 NA AG 7 47 0.97 (0.36–2.61) 0.96 CT 0 16 GG 0 7 TT 0 1 SNP22 SNP8 wt/wt 9 51 1.00 0.27* AA 23 130 NA wt/del 12 66 1.04 (0.39–2.78) 0.9332 AG 0 16 del/del 2 30 0.34 (0.07–1.74) 0.19 GG 0 1 SNP28 SNP9 TT 14 73 1.00 0.40* TT 17 67 1.00 0.03* TC 7 55 0.72 (0.26–1.96) 0.5176 TG 4 63 0.19 (0.06–0.62) 0.01 CC 2 19 0.58 (0.12–2.88) 0.50 GG 2 17 0.54 (0.11–2.72) 0.4537 SNP38 SNP10 TT 17 116 1.00 0.29* AA 16 91 1.00 TG 4 28 0.98 (0.30–3.24) 0.9729 AG 7 49 0.78 (0.29–2.08) 0.62 GG 2 3 6.15 (0.71–53.08) 0.10 GG 0 7 SNP39 SNP11 AA 12 59 1.00 0.99* AA 16 82 1.00 AG 5 60 0.44 (0.14–1.38) 0.16 AG 7 57 0.53 (0.20–1.40) 0.20 GG 6 28 1.24 (0.40–3.80) 0.7132 GG 0 8 SNP43 SNP13 GG 12 54 1.00 0.57* GG 16 91 1.00 GT 6 64 0.45 (0.15–1.33) 0.15 GC 7 49 0.78 (0.29–2.08) 0.62 TT 5 29 0.88 (0.27–2.86) 0.8243 CC 0 7 SNP49 SNP14 wt/wt 9 59 1.00 0.91* AA 16 90 1.00 wt/del 10 65 0.95 (0.35–2.58) 0.9206 AG 7 50 0.74 (0.28–1.98) 0.55 del/del 4 23 1.13 (0.30–4.36) 0.855 GG 0 7 SNP52 SNP15 GG 11 54 1.00 0.80* CC 16 92 1.00 GT 6 61 0.53 (0.18–1.60) 0.2596 CG 7 48 0.79 (0.30–2.11) 0.64 TT 6 32 0.97 (0.32–3.00) 0.9619 GG 0 7 SNP53 SNP16 TT 11 54 1.00 0.89* GG 16 94 1.00 TC 6 63 0.50 (0.17–1.51) 0.2212 GC 7 47 0.83 (0.31–2.22) 0.72 CC 6 30 1.07 (0.35–3.33) 0.90 CC 0 6 SNP54 SNP17 AA 11 55 1.00 0.83* CC 16 93 1.00 AT 6 59 0.98 (0.32–3.00) 0.3138 CT 7 47 0.97 (0.36–2.61) 0.96 TT 6 33 0.57 (0.19–1.71) 0.9691 TT 0 7 SNP70 SNP18 TT 19 119 1.00 TT 16 93 1.00 TG 4 28 1.01 (0.30–3.39) 0.99 TC 7 47 0.97 (0.36–2.61) 0.96 GG 0 0 CC 0 7

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 534

Table 3 Continued arisen by duplication or retroposition might not be subject to natural selection, particularly if the original copy remains SNP All subjects (n ¼ 170) functional. Transcribed pseudogenes may also compete with their functional counterparts for transcription factors.53 It is Cases, n Controls, n OR (95% CI) P therefore not unreasonable to consider an effect of CES4 (and/or CES1) on INH metabolism and that genetic variation SNP71 CC 10 49 1.00 0.77* in CES4 could affect the risk of INH-induced hepatotoxicity. CT 6 65 0.42 (0.14–1.28) 0.13 We discovered a large polymorphic deletion in the CES4 TT 7 33 0.97 (0.31–3.03) 0.9577 gene. Sequence data from 85 polymorphisms within the deleted region were used to distinguish nondeleted samples SNP72 AA 11 58 1.00 0.80* from heterozygous samples. This large deletion, however, AC 5 59 0.48 (0.15–1.51) 0.2072 was not significantly associated with hepatotoxicity. We CC 7 30 1.35 (0.45–4.02) 0.5945 identified eight SNPs in CES4 (SNPs 85–92) that are in 2 Multivariate analyses, adjusted for age and sex. For analyses for which there were complete LD (r ¼ 1.00) with each other (Figure 2), in a less than five rare homozygotes, heterozygotes and rare homozygotes were region that contains putative Sp1 binding sites.54 These combined for the analysis. Numbers (n) represent the total number of samples SNPs, however, were not significantly associated with INH- that gave successful genotypes for each variant. No P trend values remained induced hepatotoxicity. significant after adjustment for multiple testing using the false discovery rate. *Test for trend. SNPs in CES1 were challenging to genotype due to the high similarity between CES1 and CES4. We were able to distinguish the two genes using long PCR and nested PCR. Table 4 Association test of haplotypes of CES1 SNPs in the SNP4 is located in a putative CCAAT/enhancer binding high LD with hepatotoxicity protein site in the CES1 promoter region. CCAAT/enhancer binding proteins are expressed in several organs and CES1 haplotypes in high All subjects (n ¼ 170) involved in controlling differentiation-dependent gene LD expression.54 It is possible that this SNP could have a Frequency in Frequency in OR (95% CI) P cases controls functional effect on the gene, although it did not appear to affect risk of hepatotoxicity in this study. Moreover, we 0 AAGACGCTATA 0.85 0.75 1.00 reported that several SNPs in the 5 UTR and exon 1 of CES1 GGCGGCTCGCG 0.15 0.20 0.68 (0.27–1.67)0.40 are in high LD with each other (Figure 2). Point mutations in Combined rare* 0.00 0.05 NA the translation initiation or ‘Kozak’ sequence, especially in Multivariate analyses, adjusted for age and sex. position from À1toÀ5 relative to the start codon, influence *Rare haplotypes (with individual frequencies of o5%) were combined in the translational efficiency.55 SNP15, C(À2)G, alters the Kozak association test. sequence of CES1 (Table 2). Previous studies suggested that a C allele in position À2 of a Kozak sequence enhances translation in vivo by twofold.55 SNP15 is a good candidate significant association, there was evidence of a trend of for a functional variant in CES1, though the high LD in this increasing hepatotoxicity across the rapid, intermediate and region makes it impossible to determine, by genetics alone, slow acetylator categories of NAT2 genotypes.50 which SNP could be exerting an effect. Of the coding region We have assessed the effect of genetic variation in three changes in this haplotype, only four were nonsynonymous. novel candidate genes for hepatotoxicity, CES1, CES2 and SNP16 was classified by PolyPhen39,40 as ‘possibly damaging’ CES4, by systematically cataloguing the variants present in (see Supplementary Table D) and may warrant further study. subjects with and without hepatotoxicity, and conducting Four other SNPs in this haplotype were in the 50 UTR or association tests of SNPs and haplotypes. We found no promoter region of CES1 and could affect gene regulation. evidence for association of genetic variation in CES2 with Thus, functional studies56 will be needed to investigate INH-induced hepatotoxicity. There were no common (MAF whether any of these SNPs affect CES1 activity. X5%) nonsynonymous variants in CES2. We observed In conclusion, though SNPs and haplotypes at CES2 and common SNPs in the upstream region including the 50 CES1/CES4 were not associated with hepatotoxicity in this UTR and promoter of CES2, but they were not significantly study, we identified genetic variants that could affect CES1 associated with INH-induced hepatotoxicity. function and be relevant to other pharmacogenetic pheno- Six human CES genes have been reported on chromosome types. One SNP alters the translation initiation ‘Kozak’ 16. CES1 and CES2 are the most extensively studied51; CES4 sequence of CES1. Tests for association in a larger sample set is less well characterized. We found very high sequence will be necessary to determine if genetic variation in CES1 similarity between CES1 and CES4; CES4 is likely to be an has a role in INH-induced hepatotoxicity. Our systematic inverted duplication of CES1 (Figure 1). Moreover, these characterization of genetic variation in these genes will be genes cluster with CES7 on and may have useful in other pharmacogenetic studies, including those arisen from an ancestral duplication event.41 CES4 is a that address metabolism of therapeutic drugs, such as transcribed pseudogene,26 and its mRNA expression level in imidapril, methylphenidate and irinotecan, as well as liver is lower than that of CES1.52 A pseudogene that has pharmacogenetics of illicit drugs.

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 535

Conflict of interest of hepatitis during treatment of tuberculosis with regiments containing isoniazid and rifampin. Am Rev Dis 1986; 133: 1072–1075. The authors declare no conflict of interest. 13 Mitchell JR, Thorgeisson UP, Black M, Timbrell JA, Snodgrass WR, Potter WZ et al. Increased incidence of isoniazid hepatitis in rapid acetylators: possible relation to hydranize metabolites. Clin Parmacol Ther 1975; 18: 70–79. Abbreviations 14 Black M, Mitchell JR, Zimmerman HJ, Ishak KG, Elper GR. Isoniazid- associated hepatitis in 114 patients. Gastroenterology 1975; 69: INH isoniazid 289–302. ADRs adverse drug reactions 15 Huang YS, Chern HD, Su WJ, Wu JC, Lai SL, Yang SY et al. Polymorphism CES carboxylesterases of the N-acetyltransferase 2 gene as a susceptibility risk factor of AST aspartate aminotransferase antituberculosis drug-induced hepatitis. Hepatology 2002; 35: 883–889. OR odds ratio 16 Ohno M, Yamaguchi I, Yamamoto I, Fukuda T, Yokota S, Maekura R UTR untranslated region et al. Slow N-acetyltransferase 2 genotype affects the incidence of LD linkage disequilibrium. isoniazid and rifampicin-induced hepatotoxicity. Int J Tuber Lung Dis 2000; 4: 256–261. 17 Vuilleumier N, Rossier MF, Chiappe A, Deqoumois F, Dayer P, Mermillod B et al. CYP2E1 genotype and isoniazid-induced hepato- toxicity in patients treated for latent tuberculosis. Eur J Clin Pharmacol Acknowledgments 2006; 62: 423–429. 18 Cho HJ, Koh WJ, Ryu YJ, Ki CS, Nam MH, Kim JW et al. Genetic polymorphisms of NAT2 and CYP2E1 associated with antituberculosis This study was funded by a grant from the BC Lung Association to drug-induced hepatotoxicity in Korean patients with pulmonary FM and AB-W. AB-W is a senior scholar of the Michael Smith tuberculosis. Tuberculosis 2007; 87: 551–556. Foundation for Health Research. JH-W was supported in part by an 19 Sunahara S, Urano M, Ogawa M, Yoshida S, Mukoyama H, Kawai K. Erwin Schroedinger Fellowship from the Austrian Science Founda- Genetical aspect of isoniazid metabolism. Jinrui Idengaku Zasshi 1963; tion (FWF). VC was supported in part by ‘in it for life’, Vancouver 187: 93–111. Coastal Health Research Institute. JMFG is a recipient of a Michel 20 Kita T, Tanigawara Y, Chikazawa S, Hatanaka H, Sakeda T, Komada F Smith Distinguished Scholar Award and a CIHR/BC Lung Scientist et al. N-acetyltransferase 2 genotype correlated with isoniazid acetylation in Japanese tuberculous patients. Biol Pharm Bull 2001; 24: 544–549. Award. SY was supported in part by a Ritsumeikan University 21 Dickinson DS, Bailey WC, Hischowitz BI, Soong SJ, Eidus L, Hodgkin International Research Fellowship. MM. Risk factors for isoniazid (INH)-induced liver dysfunction. J Clin Gastroenterol 1984; 118: 271–279. 22 Online Mendelian Inheritance in Man, OMIM:. Carboxylesterase-3 References (CES-3) John Hopkins University: Baltimore, MD, MIM Number: 605279 http://www.ncbi.nlm.gov/omim/. 1 FitzGerald JM, Wang L, Elwood RK. Tuberculosis: control of the disease 23 The UCSC Genome Bioinformatics Genome Browser: http://geno- among aboriginal people in Canada. CMAJ 2000; 162: 351–355. me.ucsc.edu/index.html/org ¼ Human. 2 FitzGerald JM. Optimizing tuberculosis control in the inner city. CMAJ 24 Ensembl Genome Browser: http://www.ensembl.org/index.html. 1999; 160: 821–822. 25 Satoh H, Hosokawa M. Structure, function and regulation of carbox- 3 Hernandez E, Kumimoto D, Wang L, Rogrigues M, Black W, Elwood RK ylesterases. Chem Biol Interact 2006; 162: 195–211. et al. Predictors for clustering among TB cases in Vancouver: a four year 26 NCBI Browser: http://www.ncbi.nlm.nih.gov/. molecular epidemiology study. CMAJ 2002; 167: 349–352. 27 EASL International Consensus Conference on Hepatitis C. Paris, 26–28 4 Durand F, Jebrak G, Pessayre D, Fournier M, Bernuau J. Hepatotoxicity February 1999, Consensus Statement. European Association for the of antitubercular treatments. Rationale for monitoring liver status. Drug Study of the Liver. J Hepatol 1999; 30: 956–961. Saf 1996; 15: 394–405. 28 VISTA Browser: http://pipeline.lbl.gov/cgi-bin/gateway2?bg ¼ hg1. 5 Dossing M, Wilcke JT, Askgaard DS, Nybo B. Liver injury during 29 Hena G, Stephen PM. Conserved noncoding sequences among antituberculosis treatment: an 11-year study. Tuber Lung Dis 1996; 77: cultivated cereal genomes identify candidate regulatory sequence 335–340. elements and patterns of promoter evolution. Plant Cell 2003; 15: 6 Nolan CM, Goldberg SV, Buskin SE. Hepatotoxicity associated with 1143–1158. isoniazid preventive therapy: a 7-survey from public health tuberculosis 30 Rozen S, Skaletsky H. Primer3 on WWW for general users and for clinic. JAMA 1999; 281: 1014–1018. biologist programmers. Methods Mol Biol 2000; 132: 365–386. 7 Schaberg T, Gialdroni-Grassi G, Huchon G, Leophonte P, Manresa F, 31 In Silico PCR http://qsnp.gen.kyushu-u.ac.jp/genome/InSilicoPCR.html. Woodhead M. An analysis of decisions by European general practi- 32 Brooks-Wilson AR, Kaurah P, Suriano G, Leach S, Snez J, Grehan N et al. tioners to admit to hospital patients with lower respiratory tract Germline E-cadherin mutations in hereditary diffuse gastric cancer: infections. The European Study Group of Community Acquired assessment of 42 new families and review of genetic screening criteria. Pneumonia (ESOCAP) of the European Respiratory Society. Throax J Med Gent 2004; 41: 508–517. 1996; 51: 1017–1022. 33 Barrett J, Fry B, Maller J, Daly M. Haploview: analysis and visualization of 8 Hwang SJ, Lee SD, Li CP, Lu RH, Chan CY, Wu JC. Clinical Study of LD and haplotype maps. 2005 Available at: http://www.broad.mit.edu/ cryoglobulinaemia in Chinese patients with chronic hepatitis C. mpg/haploview/). J Gastroenterol Hepatol 1997; 12: 513–517. 34 Shin NR, Yoon SY, Shin JH, Kim YJ, Rhie GE, Kim BS et al. Development 9 Pande JN, Singh SP, Khilnani GC, Khilnani S, Tandon RK. Risk factors for of enrichment semi-nested PCR for Clostridium botulinum types A, B, E hepatotoxicity from antituberculosis drugs: a case-control study. Thorax and F and its application to Korean environmental samples. Mol Cell 1996; 51: 132–136. 2007; 24: 329–337. 10 Yee D, Valiquette C, Pelletier M, Parisien I, Rocher I, Menzies D. 35 Nickerson DA, Tobe VO, Taylor SL. PolyPhred: automating the detection Incidence of serious side effects from first-line antituberculosis drugs and genotyping of single nucleotide substitutions using fluorescence- among patients treated for active tuberculosis. Am J Respir Crit Care Med based resequencing. Nucleic Acids Res 1997; 25: 2745–2751. 2003; 167: 1472–1477. 36 Gordon D, Abajian C, Green P. Consed: a graphical tool for sequence 11 Steele MA, Burk RF, DesPrez RM. Toxic hepatitis with isoniazid and finishing. Genome Res 1998; 8: 195–202. rifampin a meta-analysis. Chest 1991; 99: 465–471. 37 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a 12 Saram GR, Imanuel C, Kailasam S, Narayana ASL, Venkatesan P. practical and powerful approach to multiple testing. JR Stat Soc B Rifampin-induced release of hydrazine from isoniazid. A possible cause 1995; 57: 289–300.

The Pharmacogenomics Journal Genetic variation in CES genes and INH hepatotoxicity S Yamada et al 536

38 Lake SL, Lyon H, Tantisira K, Silverman EK, Weiss ST, Laird NM et al. 47 Hussain Z, Kar P, Husain SA. Antituberculosis drug-induced hepatitis: Estimation and tests of haplotype-environment interaction with linkage risk factors, prevention and management. Indian J Exp Biol 2003; 41: phase is ambiguous. Hum Hered 2003; 55: 56–65. 1226–1232. 39 Sunyaev S, Ramensky V, Bork P. Towards a structure basis of human 48 Roy PD, Majumder M, Roy B. Pharmacogenomics of anti-TB drugs- non-synonymous single nucleotide polymorphisms. Trends Genet 2000; related hepatotoxicity. Pharmacogenomics 2008; 9: 311–321. 16: 198–200. Available at: http://genetics.bwh.harvard.edu/cgibin/ 49 Sharma SK, Balamurugan A, Saha PK, Pandey RM, Mehra NK. pph/polyphen.cgi ).. Evaluation of clinical and immunogenetic risk factors for the develop- 40 Sunyaev S, Ramensky V, Koch I, Lathe III W, Kondrashov AS, Bork P. ment of hepatotoxicity during antituberculosis treatment. Am J Respir Prediction of deleterious human alleles. Hum Mol Genet 2001; 10: Crit Care Med 2002; 166: 916–919. 591–597. 50 Yamada S, Tang M, Richardson K, Halaschek-Wiener J, Chan M, Cook VJ 41 Sorting Intolerant From Tolerant http://blocks.fhcrc.org/sift//SIFT.html. et al. Genetic variation of NAT2 and CTP2E1 and isoniazid hepatotoxi- 42 Tanimoto K, Kaneyasu M, Shimokuni T, Hiyama K, Nishiyama M. city in a diverse population. Pharmacogenomics 2009; 10: 1433–1445. Human carboxylesterase 1A2 expressed from carboxylesterase 1A1 and 51 Holmes RS, Chan J, Cox LA, Murphy WJ, VandeBerg JL. Opossum 1A2 genes is a potent predictor of CPT-11 cytotoxicity in vitro. carboxylesterases: sequences, phylogeny and evidence for CES gene Pharmacogenet Genomics 2007; 17: 1–10. duplication events predating the marsupial-eutherian common ances- 43 Kamendulis LM, Brzenzinski MR, Pindel EV, Bosron WF, Dean RA. tor. BMC Evol Biol 2008; 8: 54. Metabolism of cocaine and heroin is catalyzed by the same human liver 52 Gene Card CES http://www.genecards.org/cgi-bin/carddisp.pl?gene carboxylesterases. J Pharmacol Exp Ther 1996; 279: 713–717. ¼ CES. 44 Geshi E, Kimura T, Yoshimura M, Suzuki H, Koba S, Sakai T et al. A single 53 Balakirev ES, Ayala FJ. Pseudogenes: are they ‘Junk’ or functional DNA? nucleotide polymorphism in the carboxylesterase gene is associated Annu Rev Genet 2003; 37: 123–151. with the responsiveness to imidapril medication and the promoter 54 Yoshimura M, Kimura T, Ishii M, Ishii K, Matsuura T, Geshi E et al. activity. Hypertens Res 2005; 28: 719–725. Functional polymorphisms in carboxylesterase 1A2 (CES1A2) gene 45 Zhu HJ, Patrick KS, Yuan HJ, Wang JS, Donovan JL, DeVane CL et al. Two involves specific protein 1 (Sp1) binding sites. Biochem Biophys Res CES1 gene mutations lead to dysfunctional carboxylesterase 1 activity Commun 2008; 369: 939–942. in man: clinical significance and molecular basis. Am J Hum Genet 2008; 55 Kozak M. Point mutations define a sequence flanking the AUG initiator 82: 1241–1248. codon that modulates translation by eukaryotic ribosomes. Cell 1986; 46 Kubo T, Kim SR, Sai K, Saito Y, Nakajima T, Matsumoto K et al. 44: 283–292. Functional characterization of three naturally occurring single nucleo- 56 Buckland PR. The importance and identification of regulatory poly- tide polymorphisms in the CES2 gene encoding carboxylesterase 2 morphisms and their mechanism of action. Biochim Biophys Acta 2006; (HCE-2). Drug Metab Dispos 2005; 33: 1482–1487. 1762: 17–28.

Supplementary Information accompanies the paper on The Pharmacogenomics Journal website (http://www.nature.com/tpj)

The Pharmacogenomics Journal