www.elsevier.com/locate/ygeno

Genomics 84 (20

Pharmacogenomic assessment of carboxylesterases 1 and 2

Sharon Marsha,b, Ming Xiaob,c, Jinsheng Yua,b, Ranjeet Ahluwaliaa,b, Matthew Mintonb,c,

Robert R. Freimutha,b, Pui-Yan Kwokb,c, Howard L. McLeoda,b,*

aDepartment of Medicine, Washington University School of Medicine and the Siteman Cancer Center, St Louis, MO, USAbCREATE Pharmacogenetic Research Network, St Louis, MO, USA

cCardiovascular Research Institute, Department of Dermatology, University of California, San Francisco, CA, USA

Received 9 July 2004; accepted 15 July 2004

Available online 14 August 2004

Abstract

Human carboxylesterases 1 and 2 (CES1 and CES2) catalyze the hydrolysis of many exogenous compounds. Alterations in

carboxylesterase sequences could lead to variability in both the inactivation of drugs and the activation of prodrugs. We resequenced CES1

and CES2 in multiple populations (n = 120) to identify single-nucleotide polymorphisms and confirmed the novel SNPs in healthy European

and African individuals (n = 190). Sixteen SNPs were found in CES1 (1 per 300 bp) and 11 in CES2 (1 per 630 bp) in at least one population.

Allele frequencies and estimated haplotype frequencies varied significantly between African and European populations. No association

between SNPs in CES1 or CES2 was found with respect to RNA expression in normal colonic mucosa; however, an intronic SNP (IVS10-

88) in CES2 was associated with reduced CES2 mRNA expression in colorectal tumors. Functional analysis of the novel polymorphisms

described in this study is now warranted to identify putative roles in drug metabolism.

D 2004 Elsevier Inc. All rights reserved.

Keywords: Carboxylesterase; Irinotecan; Pharmacogenomics; Pharmacogenetics; Single nucleotide polymorphism.

Introduction

Carboxylesterase enzymes are found across animal

species and play an important role in the hydrolysis of drugs

and xenobiotics. Drugs, including heroin, cocaine, salicy-

lates, steroids, and anticancer agents such as irinotecan, are

substrates of carboxylesterases [1,2]. Human tissues have a

range of carboxylesterases, including placenta, brain (CES3

and hBr1) and liver (CES1; OMIM 114835 and CES2;

OMIM 605278) forms [3].

CES1 and CES2 catalyze the hydrolysis of the

anticancer drug irinotecan (CPT-11) to its active form

SN-38 [4–7]. It has been demonstrated that CES2 has a

0888-7543/$ - see front matter D 2004 Elsevier Inc. All rights reserved.

doi:10.1016/j.ygeno.2004.07.008

* Corresponding author. Washington University School of Medicine,

Campus Box 8069, 660 S. Euclid Ave., St. Louis, MO 63110. Fax: +1 314

362 3764.

E-mail address: hmcleod@im.wustl.edu (H.L. McLeod).

12.5-to 26-fold higher affinity for CPT-11 compared to

CES1 [4,8]. Although originally identified as liver

carboxylesterases, both are expressed (CES2 at high

levels) in gastrointestinal tissue [9,10], a main target for

CPT-11 therapy. Carboxylesterase 1 is also active in the

metabolism of several clinical and elicit drugs including

cocaine, heroin, and lidocaine [1]. Variations in drug-

metabolizing enzyme genes can contribute to adverse

drug reaction and increased sensitivity/resistance to drug

treatment [11].

The majority of variations at the DNA level (over 90%)

are in the form of single-nucleotide polymorphisms (SNPs).

Alterations in carboxylesterase sequences could lead to

variability in drug metabolism between patients; yet there is

little information available on the genetic variation in these

genes. The genomic structure of the CES1 gene has been

ascertained; the gene lies on chromosome 16q13–q22.1,

contains 14 exons and spans about 30 kb [12]. The CES2

gene is also located on 16q13–q22.1, contains 12 exons, and

04) 661–668

S. Marsh et al. / Genomics 84 (2004) 661–668662

spans about 9 kb (http://www.ensembl.org/). CES2 lies

approximately 11.1 cM downstream from CES1. The close

proximity and homology (73% coding region homology) of

these genes imply an ancestral gene duplication event at this

chromosomal region.

To better characterize the extent and influence of genetic

variation in CES1 and CES2 on gene expression, single

nucleotide polymorphism discovery in these genes was

performed by resequencing of 120 individuals from three

populations (Caucasian American, African American, and

Asian American). In addition, SNP information for CES1

Table 1

Source and allele frequencies of SNPs in (a) CES1 and (b) CES2

SNP Location Source Validation

European African

p q p q

a. CES1

5VUTR�469C N Aa Resequencing 0.80 0.20 0.43 0.57

5VUTR�40G N A Resequencing 0.85 0.15 0.71 0.29

562G N C (G188R)a Resequencing 0.83 0.17 0.48 0.52

601G N A (A201T)a Resequencing/

dbSNP rs2307243

0.99 0.01 1 0

IVS2 + 72T N C Resequencing/

dbSNP rs3848300

0.86 0.14 0.83 0.17

IVS4 + 10C N Ga Resequencing/

dbSNP rs2307244

0.48 0.52 0.79 0.21

IVS4 + 66G N Ta Resequencing/

dbSNP rs2307236

0.48 0.52 0.72 0.28

IVS8 + 105C N T Resequencing 0.78 0.22 0.79 0.21

IVS8 + 128A N C Resequencing/

dbSNP rs2302719

0.91 0.09 0.83 0.17

IVS10 + 36C N Ta Resequencing 1 0 0.98 0.02

IVS10 + 99C N Ta Resequencing 1 0 0.99 0.01

IVS12�135A N G Resequencing 0.98 0.02 0.98 0.02

IVS12�159G N Aa Resequencing 0.70 0.30 0.91 0.09

3VUTR + 1970A N G Resequencing 0.96 0.04 0.99 0.01

3VUTR + 2480A N G Resequencing 0.64 0.36 0.61 0.39

3VUTR + 2845C N Ta Resequencing 0.63 0.37 0.22 0.78

b. CES2

5VUTR�363C N Ga Resequencing 0.88 0.12 0.67 0.33

1175A N T (E392V) CGAP ID 1518410 1 0 1 0

1647C N T (L459L)a Resequencing/

CGAP ID 887566

0.99 0.01 0.96 0.04

IVS1�57T N C Resequencing 1 0 1 0

IVS2�152T N C Resequencing/

dbSNP rs2303218

1 0 1 0

IVS5�146C N T Resequencing 1 0 1 0

IVS5�4A N T Resequencing 1 0 1 0

IVS7 + 48G N A Resequencing 1 0 1 0

IVS10�108G N A Resequencing/

dbSNP rs2241409

1 0 1 0

IVS10�88C N Ta Resequencing/

dbSNP rs3893757

0.81 0.19 0.55 0.45

IVS11 + 160G N Aa dbSNP rs3848286 0.98 0.02 1 0

3VUTR + 2068C N A dbSNP rs1058925 1 0 1 0

3VUTR + 3071G N Aa Resequencing 0.82 0.18 0.20 0.8

a statistically significant differences in allele frequencies between Europeans andb Predicted allele frequencies from resequencing.c NI, not identified.

and CES2 was mined from publicly available databases

(CGAP-GAI, dbSNP, and JSNP). These databases list SNPs

identified in human genes by resequencing or computational

mining efforts [13].

SNP validation and haplotype inference both of novel

resequencing and of existing database SNPs were per-

formed in 95 European and 95 African healthy unrelated

individuals. In addition, the genotype-expression relation-

ship was determined for CES1 and CES2 in paired normal

and tumor samples from 52 patients with colorectal

cancer.

Resequencing

European Americanb Asian Americanb African Americanb

p q p q p q

0.8 0.2 1 0 0.80 0.20

0.54 0.46 0.71 0.29 0.77 0.23

0.71 0.29 0.89 0.11 0.64 0.34

0.76 0.24 0.92 0.08 0.71 0.29

0.96 0.04 0.99 0.01 0.95 0.05

0.56 0.44 0.45 0.65 0.70 0.30

0.80 0.20 0.62 0.38 0.41 0.59

0.65 0.35 0.81 0.19 0.79 0.31

0.91 0.09 0.90 0.01 0.87 0.13

1 0 1 0 0.90 0.10

1 0 1 0 0.95 0.05

0.60 0.49 0.36 0.64 0.76 0.24

0.54 0.46 0.88 0.12 0.67 0.33

0.95 0.05 0.83 0.17 1 0

0.65 0.35 0.48 0.52 0.70 0.30

0.93 0.07 0.86 0.14 1 0

0.99 0.01 1 0 0.99 0.01

NIc NI NI NI NI NI

0.99 0.01 1 0 1 0

1 0 0.98 0.02 1 0

1 0 0.80 0.20 1 0

1 0 0.99 0.01 1 0

1 0 1 0 0.90 0.10

1 0 1 0 0.99 0.01

0.94 0.06 0.94 0.06 0.74 0.26

0.99 0.01 0.99 0.01 0.96 0.04

NI NI NI NI NI NI

NI NI NI NI NI NI

0.90 0.10 0.89 0.11 0.53 0.47

Africans ( pb0.001).

S. Marsh et al. / Genomics 84 (2004) 661–668 663

Results

CES1 SNP discovery and validation

SNPs in CES1 were identified by resequencing approx-

imately 10.5 kb in 120 individuals (average of 1 SNP per 330

bp). Two of the 16 SNPs observed were nonsynonymous

cSNPs, 2 SNPs were found in the 5V-untranslated region, and12 were intronic SNPs. Two SNPs (IVS10+36 and

IVS10+99) were unique to the African population in this

study and the nonsynonymous SNPA201Twas unique to the

European population (Table 1a). All 16 SNPs identified in

resequencing were validated in a distinct sample set from

European and African blood donors. Nine of the 16 SNPs had

statistically significant differences in allele frequency

between European and African populations (Table 1a).

Fig. 1. Frequencies of (A) CES1 and (B) CES2 rare alleles in European and Afric

flanking region; all other SNPs lie within the amino acid coding region.

CES2 SNP discovery and validation

Ten SNPs in CES2 were identified by resequencing

approximately 10.5 kb in 120 individuals (average of 1

SNP per 630 bp). In addition, 3 SNPs were mined from

publicly available SNP databases. Of the 13 SNPs

identified in CES2, 1 synonymous cSNP (L549L) was

confirmed in the validation populations at a low frequency

(European q =0.01; African q =0.04); no nonsynonymous

cSNPs were identified in the validation populations. One

SNP (-363 CNG) was located in the 5V-untranslated region

of CES2 and one (+3071 C N T) was located in the 3V-untranslated region. Four SNPs were unique to the

resequencing study and not found in any public SNP

database, and 8/13 SNPs were found to be monomorphic

in the validation (African and European) populations in

an populations. IVS, intronic region; –, 5VUTR/flanking region; +, 3VUTR/

S. Marsh et al. / Genomics 84 (2004) 661–668664

this study (Table 1b; Fig. 1B). Five SNPs identified by

resequencing which were monomorphic in the validation

populations in this study were originally found in

American Asians and/or African Americans but not

American Caucasians (Table 1b). One nonsynonymous

cSNP was mined from CGAP (ID 1518410) and one

3VUTR SNP from dbSNP (rs1058925) but these were

monomorphic in the validation populations in this study,

and were not found by resequencing. One intronic SNP

found by resequencing (6% in American Caucasians) and

dbSNP (rs2241409) was found to be monomorphic in the

validation populations (IVS10-108GNA). Of the five SNPs

that were confirmed in the validation populations, one

(IVS12+160 CNT) was unique to the European population

(q=0.02). All 5 confirmed SNPs had statistically signifi-

cantly different allele frequencies between European and

African populations (pb0.001) (Table 1b).

CES1 and CES2 haplotypes

Inferred haplotypes in both populations for CES1 and

CES2 were generated using the Polymorphism Haplotype

and Analysis Suite [14]. In CES1, 7 common haplotypes

(=5% in at least one population) were identified (Table 2a;

Fig. 2A). Haplotype frequencies differed between the

populations (Table 2b; Fig. 2A), which is expected due to

the significant differences in allele frequencies between

European and African populations (Table 1a).

Six unique haplotypes were identified in CES2, 1 unique

to the European and 2 unique to the African population (Fig.

2B). As with CES1, haplotype frequencies differed dramat-

ically between the populations. Significant (pb0.001) link-

age is present between CES2 SNPs –363 CNG and +3071

GNA (|DV| = 1; pb0.001) and IVS10-88 CNT and +3071

GNA (|DV| = 0.9; pb0.001) in the European population but

not the African population.

CES1 and CES2 in colorectal cancer

CES1 and CES2 SNPs were evaluated in normal and

tumor samples from 52 colorectal cancer patients. No

significant difference in allele frequencies was observed

between the colorectal cancer patients and the European

validation samples ( pN0.05; data not shown). Genotype

differences were not observed between the normal and the

tumor tissue for any SNP in either gene, supporting

previously published data that chromosome 16q is not a

genomic region frequently lost in colon cancer [15]. A range

of RNA expression for carboxylesterase 1 and 2 was seen in

both the normal [2.5–73.0 (median 20.3) units for CES1,

38.0–1781.6 (median 599.1) units for CES2] and tumor

[1.0–33.1 (median 4.0) units for CES1, 21.0–1319.0

(median 228.5) units for CES2] colon samples. Analysis

using the Wilcoxon matched-pairs test showed a statistically

significant difference between normal and tumor tissues for

both CES1 (median tumor:normal ratio 0.24) and CES2

(median tumor:normal ratio 0.34; pb0.001 in both cases).

No significant correlation between RNA expression and

polymorphisms in CES1 was observed in either normal or

tumor tissue. Similarly, no correlation was seen between any

polymorphism in CES2 and normal tissue RNA expression.

However, the intronic SNP IVS10-88 C N T in CES2 was

associated with reduced RNA expression in the colon tumor

tissue (p = 0.02; Fig. 3).

Discussion

Carboxylesterases 1 and 2 are responsible for the

metabolism of the chemotherapy drug CPT-11 to its active

form, SN-38. Interindividual variability of CES2 activity

may be particularly important, as this is the most active

human carboxylesterase for irinotecan metabolism. This

study identified and validated polymorphisms in both the

CES1 and the CES2 genes using a novel resequencing

method and in silico mining. In European samples,

resequencing of pooled DNA identified only 1 apparent

false positive out of 26 predicted SNPs (3%), demonstrating

its utility as a high-throughput SNP identification method.

This compares to a 30% false positive rate from in silico

mining for SNPs in CES2. The public databases also only

contained 10/27 (37%) SNPs identified by resequencing.

This can partly be a factor of limited database information

on the original population used to identify the polymor-

phism. The lack of overlap among publicly available SNP

databases [13], and the high false positive rate from in silico

mining, is still a cause of concern for pharmacogenetics

research.

CES2 appears to have very little genetic variation, with

the majority of SNPs occurring in intronic regions. There

was a SNP found every 630 bp, almost half the frequency

of CES1 (1 SNP per 330 bp), on average much lower than

that observed by most recent projects. For example, in

human membrane transporter genes, SNPs were found on

average one every 141 bp [16]. The ratio of nonsynon-

ymous to synonymous SNPs in a gene is considered to be a

useful measure of the fraction of deleterious alleles per gene

and consequently the level of conservation of a particular

gene. The ratio observed for CES2 (0 nonsynonymous:1

synonymous) is lower than that observed for CES1 (2

nonsynonymous:0 synonymous) and other drug metabolism

or transporter genes (0.2–0.5) [16–18]. However, the low

SNP frequency in CES2 makes this comparison of unclear

value.

Differences in haplotype structure and frequency were

observed between African and European populations for

both genes (Figs. 2A and B). The strong linkage between

SNPs in the European population within both CES1 and

CES2 and the significant differences in allele frequencies

between the populations explain the low degree of

haplotype variation seen in this study. The striking differ-

ence between African and European haplotypes in these

Table 2

Structure and frequency of haplotypes found in z5% of at least one population for (a) CES1 and (b) CES1 in European and African populations

(a) CES1

5VUTR�469C N A

5VUTR�40G N A

562G N C

(G188R)

601G N A

(A201T)

IVS2 +

72T N C

IVS4 +

10C N G

IVS4 +

66G N T

IVS8 +

105C N T

IVS8 +

128A N C

IVS10 +

36C N T

IVS10 +

99C N T

IVS12�135A N G

IVS12�159G N A

3VUTR +

1970A N G

3VUTR +

2480A N G

3VUTR +

2845C N T

%

European

%

African

Hap1 A A C G T C G C A C C A G A G T b5% 20%

Hap2 A A C G T C G C C C C A G A A T b5% 10%

Hap3 A A G G T C T C A C C A G A G T 2% 10%

Hap4 C G C G T C T C A C C A A A G T 15% b5%

Hap5 C G C G T C T C A C C A A A A C 52% b5%

Hap6 A G G G C G G C A C C A G A G T 8% b5%

Hap7 C A C G T C T C A C C A A A A C 8% b5%

(b) CES2

5VUTR�363C N G

1647C N T

(L459L)

IVS11 +

160G N A

IVS10�88C N T

3VUTR +

3V071G N A

%

European

%

African

Hap1 C C G C G 80% 24%

Hap2 C C G T A 8% 19%

Hap3 G C G T A 6% 0%

Hap4 C C G T G 2% 20%

Hap5 G C G T G 0% 18%

Hap6 G C G C G 0% 13%

S.Marsh

etal./Genomics

84(2004)661–668

665

Fig. 2. Estimated frequencies of common haplotypes in European and African populations in (A) CES1 and (B) CES2. Haplotype numbers are as listed in

Table 2.

S. Marsh et al. / Genomics 84 (2004) 661–668666

genes raises interesting issues for the field of pharmacoge-

netics. With emphasis being placed on resequencing whole

genes and identifying btag SNPsQ [19,20], care should be

taken that data for one population are not used to represent

another. For example, the –363 SNP in CES2 may be

informative for the CES2+3071 SNP in a European

population (|DV| = 1), but genotyping at this position alone

Fig. 3. Colon tumor CES2 RNA expression demonstrating significantly

lower expression in individuals IVS10-88 homozygous T/T.

would miss information in the African population where

these SNPs are not in linkage disequilibrium.

The 29 SNPs evaluated in this study did not have

predictive value for RNA expression in human tissues.

Recently three distinct promoter regions have been identi-

fied for CES2 [21]. Variable RNA expression is influenced

by promoter usage, which may explain some of the variation

in CES2 RNA expression seen in this study. In this study

SNPs were not found in these three promoter regions,

implying that polymorphisms in the upstream region of

CES2 may not be key players in promoter choice and

subsequent CES2 expression. Functional analysis of the

polymorphisms identified here in CES1 and CES2 would

help to elucidate any role in carboxylesterase protein

expression and catalytic activity.

Materials and methods

Patients and samples

Genomic DNA was extracted from whole blood of 95

European and 95 African healthy volunteers as previously

described [22,23]. DNA and RNA samples from tumor and

S. Marsh et al. / Genomics 84 (2004) 661–668 667

paired normal colon tissues from 52 Dukes’ C colorectal

cancer patients (29 male; 23 female, age range 33–102

years; median 68 years) were prepared by the Siteman

Cancer Center Tissue Precurement Core. Written informed

consent was obtained from all patients to bank tumor

tissue and to perform genomic analysis. This study was

approved by the Washington University Human Subjects

Committee.

In silico SNP mining

SNPs in CES1 and CES2 were identified from dbSNP

(http://www.ncbi.nlm.nih.gov/SNP/index.html) and CGAP-

GAI (http://lpgws.nci.nih.gov/perl/snpbr).

Resequencing

Human genomic sequence for CES1 and CES2 were

retrieved fromUCSCGolden path (http://genome.ucsc.edu/).

Repeat regions were masked with RepeatMasker (Smit, AFA

and Green, P; http://www.ftp.genome.washington.edu/RM/

RepeatMasker.html) and primers for each exonic region

selected using a modified Primer3 program [24] (see

supplemental material). PCR product sizes ranged from

0.3 to 1 kb and included exons plus 100 bp of flanking

intronic sequences. In addition, 1 kb of 5V-and 3V-untrans-lated regions were resequenced for both genes. Resequenc-

ing was carried out using DNA samples of TSC DNA panel

(http://snp.cshl.org/allele_frequency_project/panels.shtml),

which were obtained from the Coriell Institutes (http://

coriell.umdnj.edu/ccr/ccrsumm.html) of 40 African Ameri-

can, 40 Asian American, and 40 Caucasian American

samples. Samples were resequenced in 8 pools of 5 samples

per population; one sample was resequenced individually

and used a reference sequence. Dye terminator sequencing

was used and samples were analyzed on an ABI 3700

capillary sequencer (ABI, Foster City, CA), Electrochroma-

tograms were aligned and analyzed using the Sequencher

(GeneCode, Ann Arbor, MI) and Mutation Surveyor

software (SoftGenetics, LLC. State College, PA).

SNP validation

SNPs in the CES1 and CES2 genes from the resequenc-

ing study, and SNPs mined using the publicly available

SNP databases, were validated in human genomic DNA

from 95 European, 95 African individuals, and 52 color-

ectal cancer patients. PCR primers were designed within

intronic regions where possible to reduce coamplification

of homologous sequence, using Primer Express version 1.5

(ABI, Foster City, CA), and the Pyrosequencing primers

were designed using the Pyrosequencing SNP Primer

Design Version 1.01 software (http://www.pyrosequencing.

com). Unique localization of the PCR primers was verified

using NCBI Blast (http://www.ncbi.nlm.nih.gov/blast/) (for

PCR primers and conditions, see supplemental material).

PCR was carried out using Amplitaq Gold PCR master mix

(ABI, Foster City, CA), 5 pmol of each primer (IDT,

Coralville, IA), and 1 ng DNA. Pyrosequencing was carried

out as previously described [25] using the Pyrosequencing

PSQ hs96A instrument and software (Pyrosequencing,

Uppsala, Sweden). CES2 SNP L549L was assayed using

PCR-RFLP (for PCR primers and enzyme information, see

supplemental material). The genotype was called variant if it

differed from the Refseq consensus sequence at the single-

nucleotide polymorphism position (http://www.ncbi.nlm.

nih.gov/LocusLink/refseq.html).

Quantitative real-time RT-PCR

Tissue total RNA was isolated from the colon tumor or

adjacent normal colon mucosa with the TRIzol RNA

isolation kit (Invitrogen, Carlsbad, CA), and was reverse-

transcribed into cDNA using Superscript II reverse tran-

scriptase (Invitrogen). Real-time PCR was carried out in a

10-Al reactionmix containing 2 Al of cDNA (10 ng/Al), 5 Al of2�TaqMan universal PCR master mix (Applied Biosystem,

Foster City, CA), and 3 Al of primer and probe mix (400 nM

each forward and reverse primers and 200 nM TaqMan probe

(IDT, Caralville, IA)). Primer sequences are as follows: for

CES1, forward 5V-TGAGTTTCAGTACCGTCCAAGCT-3V,reverse 5V-CTCATCCCCGTGGTCTCCTA-3V, probe 5V-CTCATCAGACATGAAACCCAAGACGGTG-3V; for

CES2, forward 5V-AATCCCAGCTATTGGGAAGGA-3V,reverse 5V-CTGGCTGGTCGGTCTCAAAC-3V, probe 5V-TGGCCTCAAGCCATCCTCCCATCT-3V. All real-time

PCR assays were performed in triplicate on an ABI PRISM

7700 Sequence Detector System (ABI, Foster City, CA) with

the following program: 508C for 2 min to activate uracil

N-glycosylase enzyme, 958C for 10 min to denature uracil

N-glycosylase and activate DNA polymerase, and 40 cycles

at 958C for 20 s and at 608C for 1 min. The sequence

detection program calculates a threshold cycle number (CT)

at which the reporter fluorescence generated by cleavage of

the probe is statistically greater than that of the background

signal [26]. In this study, relative expression level was

calculated using the modified comparative CT method as

previously described [26,27]; i.e., the relative expression

level of an individual target gene was normalized to the

reference gene (amyloid g precursor protein) and to one of

the 104 colon tumor and normal RNA samples that had the

maximum CT value (i.e., the lowest expression level, called

calibrator sample or 1X sample) in any target gene [28]. The

formula for relative gene expression was utilized as

previously described [27,28].

Statistical analysis

Genotype-frequency analysis of Hardy-Weinberg equili-

brium and haplotype analysis was carried out using the

Polymorphism and Haplotype Analysis Suite (http://www.

ilya.wustl.edu/~pgrn/) [14]. Allele frequency differences

S. Marsh et al. / Genomics 84 (2004) 661–668668

between populations were analyzed using the m2 test. RNA

expression statistical analyses were performed using STA-

TISTICA from StatSoft, Inc. (Tulsa, OK). Significance of

the difference of relative expression level between paired

tumor and normal samples was evaluated by the Wilcoxon

matched-pairs test. The significance level was set at pb0.05.

Correlation between variables was observed with Spearman

rank correlation.

Acknowledgments

The authors thank Christi Ralph for technical support and

Derek Van Booven for informatics assistance. This work

was supported by the Siteman Cancer Center

(P30CA091842), NIH (R21 CA102461-01), and the NIH

Pharmacogenetics Research Network (U01 GM63340);

http://pharmacogenetics.wustl.edu. All data have been

deposited into http://pharmgkb.org.

Appendix A

Supplementary data associated with this article can be

found, in the online version, at doi:10.1016/j.ygeno.

2004.07.008.

References

[1] M.R. Redinbo, S. Bencharit, P.M. Potter, Human carboxylesterase 1:

from drug metabolism to drug discovery, Biochem. Soc. Trans. 31

(2003) 620–624.

[2] T. Satoh, M. Hosokawa, The mammalian carboxylesterases: from

molecules to functions, Annu. Rev. Pharmacol. Toxicol. 38 (1998)

257–288.

[3] E.V. Pindel, et al., Purification and cloning of a broad substrate

specificity human liver carboxylesterase that catalyzes the hydrolysis

of cocaine and heroin, J. Biol. Chem. 272 (1997) 14769–14775.

[4] R. Humerickhouse, K. Lohrbach, L. Li, W.F. Bosron, M.E. Dolan,

Characterization of CPT-11 hydrolysis by human liver carboxylester-

ase isoforms hCE-1 and hCE-2, Cancer Res. 60 (2000) 1189–1192.

[5] G. Xu, W. Zhang, M.K. Ma, H.L. McLeod, Human carboxyles-

terase 2 is commonly expressed in tumor tissue and is correlated

with activation of irinotecan, Clin. Cancer Res. 8 (2002) 2605–2611.

[6] R.H. Mathijssen, et al., Irinotecan pathway genotype analysis to

predict pharmacokinetics, Clin. Cancer Res. 9 (2003) 3246–3253.

[7] M.K. Ma, H.L. McLeod, Lessons learned from the irinotecan

metabolic pathway, Curr. Med. Chem. 10 (2003) 41–49.

[8] P.D. Senter, K.S. Beam, B. Mixan, A.F. Wahl, Identification and

activities of human carboxylesterases for the activation of CPT-11, a

clinically approved anticancer drug, Bioconj. Chem. 12 (2001)

1074–1080.

[9] S.P. Sanghani, et al., Carboxylesterases expressed in human colon

tumor tissue and their role in CPT-11 hydrolysis, Clin. Cancer Res. 9

(2003) 4983–4991.

[10] W. Zhang, G. Xu, H.L. McLeod, Comprehensive evaluation of

carboxylesterase-2 expression in normal human tissues using tissue

array analysis, Appl. Immunohistochem. Mol. Morphol. 10 (2002)

374–380.

[11] W.E. Evans, H.L. McLeod, Pharmacogenomics-drug disposition, drug

targets, and side effects, N. Engl. J. Med. 348 (2003) 538–549.

[12] T. Langmann, et al., Structural organization and characterization of the

promoter region of a human carboxylesterase gene, Biochim.

Biophys. Acta 1350 (1997) 65–74.

[13] S. Marsh, P. Kwok, H.L. McLeod, SNP databases and pharmaco-

genetics: great start, but a long way to go, Hum. Mutat. 20 (2002)

174–179.

[14] D.J. Schaid, S.K. McDonnell, L. Wang, J.M. Cunningham, S.N.

Thibodeau, Caution on pedigree haplotype inference with software

that assumes linkage equilibrium, Am. J. Hum. Genet. 71 (2002)

992–995.

[15] P.H. Rooney, et al., Comparative genomic hybridization and chromo-

somal instability in solid tumours, Br. J. Cancer 80 (1999) 862–873.

[16] M.K. Leabman, et al., Natural variation in human membrane

transporter genes reveals evolutionary and functional constraints,

Proc. Natl. Acad. Sci. USA 100 (2003) 5896–5901.

[17] J.K. Lamba, et al., Common allelic variants of cytochrome P4503A4

and their prevalence in different populations, Pharmacogenetics 12

(2002) 121–132.

[18] J. Zhang, et al., The human pregnane X receptor : genomic structure

and identification and functional characterization of natural allelic

variants, Pharmacogenetics 11 (2001) 555–572.

[19] X. Ke, L.R. Cardon, Efficient selective screening of haplotype tag

SNPs, Bioinformatics 19 (2003) 287–288.

[20] The International HapMap Project, Nature 426 (2003) 789–796.

[21] M.H. Wu, et al., Characterization of multiple promoters in the human

carboxylesterase 2 gene, Pharmacogenetics 13 (2003) 425–435.

[22] H.L. McLeod, et al., Ethnic differences in catechol O-methyltransfer-

ase pharmacogenetics: frequency of the codon 108/158 low activity

allele is lower in Kenyan than Caucasian or South-west Asian

individuals, Pharmacogenetics 8 (1998) 195–199.

[23] M.M. Ameyaw, E.S. Collie-Duguid, R.H. Powrie, D. Ofori-Adjei,

H.L. McLeod, Thiopurine methyltransferase alleles in British and

Ghanaian populations, Hum. Mol. Genet. 8 (1999) 367–370.

[24] S. Rozen, H. Skaletsky, Primer3 on the WWW for general users and

for biologist programmers, Methods Mol. Biol. 132 (2000) 365–386.

[25] C.M. Rose, S. Marsh, M.M. Ameyaw, H.L. McLeod, Pharmacoge-

netic analysis of clinically relevant genetic polymorphisms, Methods

Mol. Med. 85 (2003) 225–237.

[26] A.B. Incorporation (1997). Relative quantification of gene expression,

User Bulletin No 2.

[27] M.W. Pfaffl, A new mathematical model for relative quantification in

real-time RT-PCR, Nucleic Acids Res. 29 (2001) e45.

[28] J. Yu, S. Marsh, R. Ahluwalia, H.L. McLeod, Ferredoxin reductase:

pharmacogenomic assessment in colorectal cancer, Cancer Res. 63

(2003) 6170–6173.