EZH2 genetic variants affect risk of gastric cancer in the Chinese Han population

MOLECULAR CARCINOGENESIS

EZH2 Genetic Variants Affect Risk of GastricCancer in the Chinese Han Population

Yuan Zhou,1,2 Wei-Dong Du,1,3,4,5* Qiang Wu,3 Yi Liu,6 Gang Chen,4 Jian Ruan,1,3 Song Xu,1,3

Feng Yang,3 Fu-Sheng Zhou,4 Xian-Fa Tang,4 Hua-Yang Tang,4 Xian-Bo Zuo,4 Feng-Yu Zhang,4

Liang-Dan Sun,4 and Xue-Jun Zhang4

1Key Lab of Genome Research of Anhui Province, Anhui Medical University, Hefei, China2Department of Biology, Anhui Medical University, Hefei, China3Department of Pathology, Anhui Medical University, Hefei, China4State Key Laboratory Incubation Base of Dermatology, Ministry of National Science and Technology, Anhui MedicalUniversity, Hefei, China5Sektion Experimentelle Anaesthesiologie, Universitaetsklinikum Ulm, Ulm, Germany6Department of Surgery, The First Affiliated Hospital of Anhui Medical University, Hefei, China

Enhancer of zeste 2 (EZH2) gene encodes a histone methyltransferase that constitutes the catalytic component ofthe polycomb repressive complex-2 (PRC2) to initiate epigenetic silencing of genes. It is reported that the expression

level of EZH2 in gastric cancer tissue was highly correlated with tumor progression, however, whether EZH2 geneticvariants were associated with the risk of gastric cancer remains yet unknown. In this study, we conducted a genotyp-ing analysis for EZH2 in 311 cases of gastric cancer and 425 controls from the Chinese Han population. We found fivesingle nucleotide polymorphisms (SNP; rs12670401, rs6464926, rs2072407, rs734005, and rs734004) of EZH2 gene

were significantly associated with the risk of gastric cancer. Of which, the rs12670401 with the minor allele C andrs6464926 with the minor allele T revealed strong associations with increased gastric cancer risk [P ¼ 0.009, adjustedodds ratio (aOR) ¼ 1.327, 95% CI ¼ 1.075–1.683 and P ¼ 0.012, aOR ¼ 1.310, 95% CI ¼ 1.059–1.619]. The other

three SNPs, rs2072407, rs734005, and rs734004 contributed to significantly reduced risk of gastric cancer (P ¼ 0.033,aOR ¼ 0.787, 95% CI ¼ 0.633–0.981, P ¼ 0.045, aOR ¼ 0.799, 95% CI ¼ 0.642–0.995 and P ¼ 0.048, aOR ¼0.803, 95% CI ¼ 0.645–0.999), respectively. We further found that rs12670401 and rs6464926 were in a strong LD

while rs2072407, rs734005, and rs734004 were in another. Haplotype analysis of the five SNPs showed that haplo-type CCTCT reduced the risk of gastric cancer (P ¼ 0.031 and aOR ¼ 0.784), while haplotype GTCTC significantlyelevated the risk of gastric cancer (P ¼ 0.011 and aOR ¼ 1.310). We concluded that EZH2 variants were significantly

associated with gastric cancer risk. Our results for the first time provided new insight into susceptibility factors ofEZH2 gene variants in carcinogenesis of gastric cancer of the Chinese Han population. � 2012 Wiley Periodicals, Inc.

Key words: EZH2; gastric cancer; single nucleotide polymorphism

INTRODUCTION

Gastric cancer is a major health condition withlimited treatment options and poor prognosis. Itcauses about 800,000 deaths worldwide per year[1]. Despite the complexity of cancer development,both gene mutation and environment factor bearequally influential forces in carcinogenesis of thisentity.Polycomb group (PcG) proteins are epigenetic

chromatin modifiers that contribute to cancerdevelopment. Three PcG proteins, enhancer ofzeste 2 (EZH2), embryonic ectoderm development(EED), and suppressor of zeste 12 homolog(SUZ12), comprise the core of polycomb repressivecomplex-2 (PRC2) [2,3]. EZH2, also called histone–lysine N-methyltransferase (HKMT), is an enzymethat plays a crucial role in initiating silencing oftarget genes [4,5]. EZH2 contains a SET domainthat has histone methyltransferase (HMT) activity

and acts mainly as a gene silencer, adding threemethyl groups to lysine 27 of histone 3 (H3K27).The methylation of H3K27 by EZH2 results in therecruitment of PRC1. PCR1 and PCR2 cooperatewith each other in gene silencing [6–8]. Overex-pression of EZH2 is associated with poor prognosis

Abbreviations: PcG, Polycomb group; EZH2, enhancer of zeste 2;EED, embryonic ectoderm development; SUZ12, suppressor ofzeste 12 homolog; PRC2, polycomb repressive complex-2; HKMT,histone–lysine N-methyltransferase; HMT, histone methyltransfer-ase; H3K27, lysine 27 of histone 3; SNP, single nucleotide polymor-phism; HWE, Hardy–Weinberg equilibrium; OR, odds ratio; aOR,adjusted odds ratio; DNMT, DNA methyltransferase.

*Correspondence to: Key Lab of Genome Research of AnhuiProvince, Anhui Medical University, Meishan Road 81, Hefei230032, China.

Received 1 June 2011; Revised 9 December 2011; Accepted 12December 2011

DOI 10.1002/mc.21871

Published online in Wiley Online Library(wileyonlinelibrary.com).

� 2012 WILEY PERIODICALS, INC.

or increased cancer risk in a variety of epithelialcell-derived digestive tumors, such as oesophaguscancer [9], gastric cancer [2,10,11], colorectalcancer [12], hepatocellular carcinoma liver [13],pancreatic ductal adenocarcinoma [14], and blad-der cancer [15] in the both preclinical and clinicalmodels [16,17]. In gastric cancer, EZH2 acts as anoncogene. High EZH2 expression is correlated withKi-67 and p53 expressions and is significantlyassociated with distant metastases and non-signetring cells [2]. The single nucleotide polymorphisms(SNPs) associated with worse prognosis wouldbecome predictive of higher EZH2 expression.

The domain structure of EZH2 and mutationssites have been displayed in the previous investiga-tions [18,19]. Both the C-terminal SET domainand the adjacent cysteine-rich CXC domain arerequired for HMT activity [6,20,21]. Missense, in-frame deletion and splice mutation of the EZH2domain attribute to loss function of the gene.The EZH2 mutations might occur in monoallelicor biallelic states [22]. The truncating mutationsdistributed throughout the gene, whereas the mis-sense mutations mostly targeted evolutionarilyhighly conserved residues that were located indomain II and the CXC-SET domains. Thus, mostof the frame-shift and truncating mutations wereinactivating. Furthermore, Western blotting datarevealed an almost complete absence of H3K27me3in the EZH2-mutant ELF-153 and SKM-1 cell lines[18,22,23].

A number of SNPs within the EZH2 gene are as-sociated with the susceptibility or outcomes tocancers. A recurring monoallelic EZH2 point muta-tion (Tyr641) in the SET domain was described inlymphoma [23]. The altered tyrosine correspondedto a key residue in the active site of the EZH2 pro-tein, leading to reduced enzymatic activity. Thisevidence was in consistence with the previousfunctional observation of a comparable mutationin the Drosophila E (z) ortholog [24]. PRC2 com-plexes containing mutated EZH2 protein reducedH3K27 trimethylation activity in vitro [23]. Thewild-type copy of EZH2 was present in all the sam-ples with Tyr641-altering mutations and presenceof both alleles was also detected in the mutantsamples profiled by transcriptome sequencing [23].Mutant and wild-type gene products cooperate toincrease H3K27trimeth in neoplastic cells [25].Increasing impression was in that variants of EZH2gene influenced development and consequenceof cancers. Breyer et al. [26] found rs2302427(D185H) of EZH2 had a minor allele frequency of3.7% in cases and 5.2% in controls, conferringsignificantly reduced risk of prostate cancer in aheterozygote manner (P ¼ 0.0085, OR ¼ 0.63).Bachmann et al. [27] conducted a mutation screenthrough EZH2 gene in hereditary prostate cancer.They observed 8 haplotypes with frequency of

>0.01, 1 of the haplotypes was significantly lessfrequent in familial cases than in controls[27].In addition, Yoon et al. [19] demonstratedEZH2 genetic polymorphisms rs6950683 andrs3757441 contribute to significant associationswith reduced lung cancer risk, indicating protec-tive effects of the variants on lung cancer risk. Sim-ilarly, EZH2 variants were associated with a poorprognosis of myeloid disorders [22] and myelodys-plastic syndromes [18]. More recently, one allelicgenotype of rs3757441 (C/C vs. C/T or T/T) ofEZH2 gene was observed being significantly associ-ated with shorter PFS and OS in colorectal cancer[28]. However, association study of genetic variantsin EZH2 and gastric cancer susceptibility remainsyet absent.We addressed the hypothesis that genetic varia-

tions in EZH2 might contribute to the susceptibili-ty to gastric cancer. Thus, we performed a geneticanalysis for the EZH2 variants in a case–control studyand investigated potential associations of thesevariants with biological behaviors of gastric cancer.

MATERIALS AND METHODS

Study Population

This study was a hospital-based case–controlstudy. Patients with gastric cancers (n ¼ 311) werecollected from the First Affiliated Hospital ofAnhui Medical University between March 2008and July 2009 and received no preoperative che-motherapy or radiotherapy before surgical gastrec-tomy. The patients were comprised 246 men and65 women with an average age of 60.4 � 10.4 yr(range, 24–83 yr). In addition, a total of 336 menand 89 women control subjects were matched withan average age of 60.6 � 8.4 yr (range, 30–86 yr).These control subjects without a history of cancerswere recruited from patients who visited hospitalfor a conventional cancer-screening program. In-formation on demographic characteristics, such asgender, age, smoking habits, alcohol consumption,and family history of cancer were obtained from apersonal interview administered by trained person-nel. Smoking habit was defined as non-smoker andsmoker. Individuals who smoked one cigarette perday for over 1 yr were defined as smokers. Alcoholconsumption was defined as non-drinker and drinker.Individuals who consumed more than 200 mL al-cohol per day were defined as drinker. There wasno significant difference in the distribution of age,gender, smoking, and drinking (P > 0.05, data notshown). All the subjects tested were Chinese Hanfrom the same geographic regions of China. In-formed consents were obtained from all the partic-ipants. This study was approved by the ethicscommittee for genome research of the Anhui Med-ical University of China and was conductedaccording to the Declaration of Helsinki Principles.

2 ZHOU ET AL.

Molecular Carcinogenesis

METHODS

Extraction of Peripheral Blood DNA

Blood sample was collected from each subject inEDTA and stored at �808C.Genomic DNA wasextracted from peripheral blood by using QIAampDNA Blood Midi Kit (Qiagen Inc., Hilden, Germany)according to the manufacturer’s protocol. The con-centration of DNA was measured by the NanodropSpectrophotometer (ND-1000, Wilmington, DE) offull wavelength and standardized to 50 ng/mL.

SNP Selection and Sequenom Assay

We first checked the SNP genotype data of EZH2gene in the Chinese Han from InternationalHapMap Project website (http://hapmap.ncbi.nlm.nih.gov/). The data were then formatted by Haplo-view software. The SNPs rs6950683 and rs2303427of EZH2 with regard to the association with risk oflung [19] and prostate cancer [26] were excludedfor further analysis because the SNP rs6950683 wascompletely absent in the Chinese population,while the MAF of SNP rs2303427 was only up to0.011 in the Chinese population HapMap database.Thus, a total of 25 SNPs of the EZH2 gene werefinally selected in this study with the MAF > 0.05and were in Hardy–Weinberg equilibrium (HWE).

Fifteen nanograms of genomic DNA was used togenotype each sample. Locus-specific PCR and de-tection primers were designed using the MassAR-RAY Assay Design 3.0 software (Sequenom, SanDiego, CA) at the Key Laboratory of Dermatologyat Anhui Medical University, Ministry of Educa-tion, China, following the manufacturer’s instruc-tions. The DNA samples were amplified bymultiplex PCR reactions, and the PCR productswere then used for locus-specific single-base exten-sion reactions. The resulting products weredesalted and transferred to a 384-element Spectro-CHIP array. Genotyping analysis of the SNPs forfast-track validation analysis was performed usingthe Sequenom MassArray system with MALDI-TOFMS for allele detection. Genotyping quality wasexamined by a detailed QC procedure consistingof >95% successful call rate, duplicate calling ofgenotypes, internal positive control samples,and HWE testing. The mass spectrograms were an-alyzed by the MassARRAY Typer software (Seque-nom, San Diego, CA).

Data Analysis

We used a logistic regression model while con-trolling for age and gender as covariates to analyze(1) association of EZH2 polymorphisms with the

Table 1. Analysis of EZH2 Polymorphisms Associated With the Risk of Gastric Cancer

SNPNo. SNP rs# Alleles Position

HWE MAF

P-value OR (95% CI)Controls Cases Controls Cases

22 rs12670401 T ! C 148193868 1.000 0.134 0.392 0.461 0.009 1.327 (1.075,1.683)14 rs6464926 C ! T 148149944 0.836 0.246 0.383 0.449 0.012 1.310 (1.059–1.619)6 rs2072407 C ! T 148139766 0.677 0.433 0.372 0.318 0.033 0.787 (0.633–0.981)2 rs734005 T ! C 148136797 0.678 0.303 0.376 0.325 0.045 0.799 (0.642–0.995)1 rs734004 G ! C 148136393 0.918 0.199 0.380 0.330 0.048 0.803 (0.645–0.999)7 rs2072408 C ! T 148139130 0.677 0.362 0.372 0.323 0.053 0.805 (0.646–1.003)3 rs740949 G ! A 148136997 0.755 0.436 0.370 0.323 0.063 0.812 (0.653–1.012)4 rs1544571 A ! G 148172594 0.605 0.159 0.377 0.330 0.065 0.815 (0.655–1.014)19 rs10274535 G ! A 148174458 0.680 0.158 0.378 0.332 0.069 0.817 (0.657–1.016)11 rs3779038 T ! C 148173755 0.605 0.159 0.377 0.331 0.072 0.819 (0.659–1.019)10 rs3779037 C ! T 148170256 0.757 0.126 0.376 0.331 0.074 0.820 (0.659–1.020)15 rs6959647 C ! T 148161651 0.919 0.129 0.379 0.334 0.076 0.822 (0.661–1.021)16 rs6975291 G ! C 148179326 0.678 0.202 0.376 0.331 0.076 0.821 (0.660–1.022)12 rs3807450 A ! C 148153202 0.677 0.199 0.372 0.329 0.083 0.824 (0.663–1.026)18 rs9691534 C ! T 148175852 0.757 0.126 0.376 0.332 0.085 0.826 (0.665–1.027)13 rs3807451 C ! T 148169555 0.678 0.159 0.376 0.332 0.088 0.827 (0.665–1.029)20 rs10274701 T ! C 148183389 0.759 0.127 0.377 0.342 0.162 0.856 (0.688–1.065)8 rs3735219 G ! A 148144400 0.785 0.021 0.232 0.202 0.167 0.837 (0.649–1.078)21 rs10488070 T ! C 148159803 0.505 0.091 0.241 0.217 0.284 0.873 (0.681–1.119)25 rs17171127 C ! T 148172919 0.684 0.060 0.235 0.212 0.301 0.876 (0.682–1.126)23 rs17171118 C ! T 148148237 0.356 0.191 0.244 0.222 0.346 0.888 (0.694–1.136)9 rs3757441 T ! C 148155685 0.126 0.348 0.257 0.238 0.399 0.902 (0.708–1.148)5 rs1880357 G ! C 148208543 0.681 0.059 0.231 0.212 0.400 0.898 (0.699–1.154)24 rs17171119 T ! G 148148389 0.295 0.109 0.246 0.229 0.454 0.911 (0.714–1.163)17 rs7785022 T ! C 148164317 0.534 0.011 0.379 0.370 0.730 0.961 (0.765–1.207)

The ORs (95% CI) and corresponding P-values were derived from an age and gender adjusted logistic regression. P-values <0.05were indicated in bold font.

EZH2 GENETIC VARIANTS AFFECT GASTRIC CANCER RISK 3


Table

2.EZH2GenotypeDistributionsin

theControlandin

theGastricCancer

SNP

Genotype

Frequency

CodominantModel

DominantModel

RecessiveModel

Controls

GC

OR(95%

CI)

P-value

OR(95%

CI)

P-value

OR(95%

CI)

P-value

rs12670401

TT

157

96

11

1TC

202

138

1.117(0.800–1.560)

0.515

1.101(0.817–1.479)

0.121

CC

65

71

1.786(1.172–2

.723)

0.007

1.676(1.152–2

.437)

0.007

rs6464926

CC

161

98

11

1CT

196

140

1.173(0.842–1.635)

0.345

1.307(0.958–1.783)

0.092

TT

63

66

1.721(1.123–2

.638)

0.013

1.571(1.072–2

.303)

0.020

rs2072407

CC

164

146

11

1TC

202

128

0.712(0.520–0

.975)

0.034

0.710(0.527–0

.955)

0.024

TT

56

35

0.702(0.435–1.132)

0.147

0.835(0.532–1.310)

0.432

rs734005

TT

162

144

11

1TC

203

128

0.709(0.518–0

.972)

0.033

0.714(0.530–0

.961)

0.026

CC

57

37

0.730(0.456–1.169)

0.191

0.871(0.560–1.356)

0.541

rs734004

GG

161

142

11

1CG

201

126

0.711(0.518–0

.976)

0.035

0.717(0.532–0

.966)

0.029

CC

60

39

0.737(0.464–1.170)

0.195

0.978(0.569–1.353)

0.556

TheORs(95%

CI)andP-valueswere

derivedfrom

anageandgenderadjustedlogisticregression.P-values<0.05were

indicatedin

bold

font.DominantModel:(DD,Dd)versusdd.Recessive

Model:DDversus(Dd,dd).Dwastheminorallele

anddwasthemajorallele.

4 ZHOU ET AL.


risk of gastric cancer; (2) comparison of EZH2genotypes; and (3) haplotype distributions in 311gastric cancers and 425 normal controls. Thosewith the frequency of <0.05 were ignored in theanalysis. Hardy–Weinberg proportion was tested incontrol samples to ensure genotyping quality. Forthe stratified association analysis, chi-squared testwas used through case-only logistic regressionanalyses, while if there was a number of genotype�5, Fisher’s exact test was used. All analyses forP-values were two-sided. The statistical softwareplink v1.07 (http://pngu.mgh.harvard.edu/purcell/plink/) and Stata/SE version 10 were used for statis-tical analyses (StataCorp LP, College Station, TX).The Haploview 4.2 was used to calculate the link-age disequilibrium.

RESULTS

Association Analysis of 25 SNPs of EZH2

The minor allele, P-values, and the estimatedodds ratios (OR) of the 25 SNPs of EZH2 gene inpatients with gastric cancer and in controls wereshown in Table 1. Five SNP polymorphisms(rs12670401, rs6464926, rs2072407, rs734005, andrs734004) of EZH2 were identified to be signifi-cantly associated with the risk of gastric cancer. Ofwhich, the rs12670401 with the minor allele Cand rs6464926 with the minor allele T showed

strong associations with increased gastric cancerrisk [P ¼ 0.009, adjusted OR (aOR) ¼ 1.327, 95%CI ¼ 1.075–1.683; and P ¼ 0.012, aOR ¼ 1.310,95% CI ¼ 1.059–1.619], while the rs2072407,rs734005 and rs734004 polymorphisms presentedsignificantly reduced risk of gastric cancer(P ¼ 0.033, aOR ¼ 0.787, 95% CI ¼ 0.633–0.981;P ¼ 0.045, aOR ¼ 0.799, 95% CI ¼ 0.642–0.995;and P ¼ 0.048, aOR ¼ 0.803, 95% CI ¼ 0.645–0.999, respectively). The effects of the five SNPswere further analyzed using alternative models,such as codominant, dominant, and recessivemodels (Table 2). The homozygous genotype CCof rs12670401 and homozygous genotype TT ofrs6464926 accounted for higher proportion incases and significantly elevated the risk of gastriccancer (P ¼ 0.007, aOR ¼ 1.786, 95% CI ¼ 1.172–2.723; and P ¼ 0.013, aOR ¼ 1.721, 95% CI ¼1.123–2.638), respectively. Conversely, the frequen-cies of heterozygous genotypes TC (rs2072407), TC(rs734005), and CG (rs734004) were significantlyhigher in controls, leading to reduced risk of gas-tric cancer (P < 0.05).

The Linkage Disequilibrium and Haplotypes Analysis

We performed linkage disequilibrium analysisand showed the LD plot in Figure 1. The SNPsrs12670401 and rs6464926 were in a strong LDand the SNPs rs2072407, rs734005, and rs734004

Figure 1. Pairwise linkage disequilibrium (r2) in the study population.



were within another. A haplotype analysis wasperformed further (Table 3). We found that thehaplotypes of EZH2 genes affected susceptibility togastric cancer. The haplotypes with either minoralleles of rs12670401 and rs6464926 or majoralleles of rs2072407, rs734005 and rs734004 signif-icantly elevated risk of gastric cancer, while thehaplotypes with either major alleles of rs12670401and rs6464926 or minor alleles of rs2072407,rs734005, and rs734004 conferred reduced risk ofgastric cancer. In combination with the 25 SNPs,we found Ht3 and Ht4 exhibited statistically signif-icant differences. Ht3 represented elevated riskof gastric cancer (P ¼ 0.049, aOR ¼ 1.230). Ht4reduced the risk of gastric cancer (P ¼ 0.050,aOR ¼ 0.701, Table 3), instead.

Association Analysis of the Five SNPs by ClinicalCharacteristics

We performed the stratified association analysiswith regard to the five SNPs in different clinicalvariables by gender, age, tumor location, tumorsize, histologic subtype, depth of invasion, lymphnode metastasis, TNM stage, smoking, alcoholdrinking, family history, and salted food. Theresults were summarized in Table 4. However,there were no associations of the five SNPs withthe clinical characteristics.

DISCUSSION

In this study, we performed a genetic associationanalysis of EZH2 variants with gastric cancer.We demonstrated that the five SNPs, rs12670401,rs6464926, rs2072407, rs34005, and rs734004,were significantly associated with gastric cancersusceptibility. To the best of our knowledge, thiswas the first study addressing EZH2 variants mightbe novel susceptibility markers linked to gastriccancer.

In this study, we observed that the rs12670401with the minor allele C and rs6464926 withthe minor allele T showed strong associations withincreased gastric cancer risk (P ¼ 0.009, aOR ¼1.327, 95% CI ¼ 1.075–1.683; P ¼ 0.012, aOR ¼1.310, 95% CI ¼ 1.059–1.619). The SNP rs12670401located near the transcriptional start position ofEZH2, according to the web-based SNP analysistool, TF search 1.3 [29]. The T to C change ofrs12670401 might affect predicted GATA-X tran-scription factors binding, resulting in directly acti-vating or repressing the expression of target genes.GATA family includes essential regulators, whichshare two highly conserved zinc fingers of theC2H2 type that mediate not only DNA bindingbut also the great majority of protein interactions.GATA factors typically bind to the elementA/TGATA A/G [30]. T allele may create a bindingsite for the GATA-X factors. Patients carrying theC/C genotype are lack of binding sites for this fac-tor. In patients carrying one T allele, EZH2 expres-sion may be inhibited by GATA-X factors. In C/Chomozygotes, EZH2 expression could be deregu-lated. We supposed that C allele of rs12670401failed to be targeted by GATA factors and attribut-ed to the development of gastric cancer. The SNPrs6464926 located in the intron region betweenexons 8 and 9, C to T change of the SNPrs6464926 may affect the binding activity of TFIID,which serves as the scaffold for assembly of the re-mainder of the transcription complex, and acts asa channel for regulatory signals [31]. In this study,we found that the minor allele T of rs6464926 con-ferred susceptibility with gastric cancer. We sup-posed that the effect of T allele might contributeto higher EZH2 expression. EZH2 and DNA meth-yltransferases (DNMTs) are physiologically andfunctionally linked to each other. EZH2 is neces-sary for DNMT binding and CpG methylation of

Table 3. Haplotype Distributions in 311 Gastric Cancers and 425 Controls

SNPs

Frequency

Haplotype OR P-valueControls Cases

snp14,22 0.385 0.450 TC 1.290 0.018snp14,22 0.614 0.549 CT 0.770 0.015snp1,2,6 0.375 0.321 CCT 0.788 0.034snp1,2,6 0.624 0.678 GTC 1.250 0.042snp1,2,6,14,22 0.387 0.458 GTCTC 1.310 0.011snp1,2,6,14,22 0.375 0.321 CCTCT 0.784 0.031snp1,2,6,14,22 0.238 0.220 GTCCT 0.897 0.404snp1-25 (Ht1) 0.258 0.242 CCAGGTTGCTCCTCTCCTACTTCTC 0.895 0.360snp1-25 (Ht2) 0.228 0.205 GTGACCCATCTACCCGTCGTCTTGT 0.840 0.194snp1-25 (Ht3) 0.391 0.459 GTGAGCCGTCTACTCGTCGTTCCTC 1.230 0.049snp1-25 (Ht4) 0.122 0.092 CCAGGTTGTTCCTCTCCTACTTCTC 0.701 0.050

The ORs and corresponding P-values were derived from an age and gender adjusted logistic regression. P-values <0.05 were indicat-ed in bold font.

6 ZHOU ET AL.


Table

4.StratifiedAssociationAnalysisoftheFive

SNPsbyClinicalSubgroups

Parameter

rs734004

rs734005

rs2072407

rs6464926

rs12670401

CC

CG

GG

P-value

CC

TC

TT

P-value

CC

TC

TT

P-value

CC

CT

TT

P-value

CC

TC

TT

P-value

Gender

0.266

0.289

0.149

0.152

0.660

Male

29

98

116

28

99

118

119

99

27

73

113

55

60

110

72

Female

10

28

26

929

26

27

29

825

27

11

11

28

24

Age

0.178

0.521

0.403

0.305

0.986

�55

29

95

110

27

97

112

114

97

25

74

109

48

54

106

73

<55

10

31

32

10

31

32

32

31

10

24

31

18

17

32

23

Tumorlocation

0.530

0.414

0.599

0.423

0.478

Cardia

923

22

17

47

63

63

47

17

42

54

29

32

52

41

Antrum

627

35

627

35

35

27

621

32

14

15

33

20

Corpus

18

46

61

923

22

23

23

819

25

10

10

25

19

Tumorsize

(cm)

0.453

0.503

0.435

0.949

0.982

�521

51

61

20

52

62

62

53

19

47

55

30

32

58

45

<5

12

54

57

12

54

58

59

54

11

37

63

22

25

60

37

Histologictype

0.195

0.185

0.213

0.081

0.142

Tubularor

papillary

34

96

110

33

97

112

113

98

31

79

114

45

51

111

77

Others

213

18

212

18

18

12

28

13

11

11

13

8Depth

ofinvasion

0.828

0.559

0.667

0.801

0.797

T1

311

14

322

14

15

10

37

14

67

11

8T2

719

21

719

22

22

20

613

29

57

28

12

T3

26

74

89

25

75

90

90

76

24

65

80

43

45

83

63

Lymphnode

metastasis

0.372

0.302

0.229

0.202

0.202

N0

13

37

55

12

38

56

57

38

11

31

52

22

24

49

31

N1

17

50

50

17

50

51

51

51

16

38

48

29

32

49

36

N2

615

19

615

19

19

15

616

20

44

20

16

TNM

stage

0.447

0.388

0.328

0.867

0.661

Iþ

II15

49

62

14

50

63

64

50

13

38

63

24

28

60

37

IIIþ

IV21

57

65

21

57

66

66

58

20

49

62

31

32

64

48

Smoking

0.425

0.317

0.309

0.658

0.717

Nooroccasionally

14

62

67

13

63

69

70

63

12

42

69

31

33

68

41

Yes

22

60

68

22

60

68

69

60

21

48

69

31

34

68

47

Alcoholdrinking

0.854

0.956

0.794

0.751

0.674

Nooroccasionally

20

66

73

19

67

74

76

67

17

51

71

34

36

72

50

Yes

16

56

62

16

56

63

63

56

16

39

67

28

31

64

38

Family

history

0.103

0.114

0.138

0.621

0.762

No

36

111

119

35

112

121

123

112

33

82

127

54

59

125

80 (Continued)



target genes. Hyper-methylation of the GATA fam-ily promoter made the GATA gene expression si-lenced [32]. Candidate target genes of these GATAfactors were commensurately decreased in their ex-pression and also displayed methylated promoters[30]. On the other hand, we found that EZH2 SNPsrs2072407, rs734005, and rs734004 presented pro-tective effort on the gastric cancer progression(P ¼ 0.033, aOR ¼ 0.787; P ¼ 0.045, aOR ¼ 0.799;P ¼ 0.048, aOR ¼ 0.803), respectively. The SNPsrs2072407, rs734005, and rs734004 showed astrong LD located in the intron region betweenexons 15 and 19. The G to C change of SNPrs734004 might affect the binding of poly A down-stream elements and further the transcription ofgene terminated [33]. Furthermore, we found thathaplotypes of the EZH2 SNPs conferred variablesusceptibility to gastric cancer. The mechanismwith regard to effect of EZH2 SNPs on biologicalphenotypes of gastric cancer needs to be investi-gated further.In conclusion, we found the five SNPs of EZH2

gene were significantly associated with gastric can-cer susceptibility. Our findings might provide in-sight into the genetic basis of gastric cancer. Theeffect of EZH2 variants on recruitment of PRC1and PRC2 components and state of methylation ofgenes would warrant further investigation of thepotential mechanism in gastric cancer of the Chi-nese Han population.

REFERENCES

1. Compare D, Rocco A, Nardone G. Risk factors ingastric cancer. Eur Rev Med Pharmacol Sci 2010;14:302–308.

2. Choi JH, Song YS, Yoon JS, Song KW, Lee YY. Enhancerof zeste homolog 2 expression is associated with tumorcell proliferation and metastasis in gastric cancer. APMIS2010;118:196–202.

3. Cho KS, Oh HY, Lee EJ, Hong SJ. Identification of enhanc-er of zeste homolog 2 expression in peripheral circulatingtumor cells in metastatic prostate cancer patients: A pre-liminary study. Yonsei Med J 2007;48:1009–1014.

4. Chen H, Rossier C, Antonarakis SE. Cloning of a humanhomolog of the Drosophila enhancer of zeste gene (EZH2)that maps to chromosome 21q22. Genomics 1996;38:30–37.

5. Fiskus W, Pranpat M, Balasis M, et al. Histone deacetylaseinhibitors deplete enhancer of zeste 2 and associatedpolycomb repressive complex 2 proteins in human acuteleukemia cells. Mol Cancer Ther 2006;5:3096–3104.

6. Cao R, Wang L, Wang H, et al. Role of histone H3 lysine27 methylation in Polycomb-group silencing. Science2002;298:1039–1043.

7. Martinez-Garcia E, Licht JD. Deregulation of H3K27 meth-ylation in cancer. Nat Genet 2010;42:100–101.

8. Plath K, Fang J, Mlynarczyk-Evans SK, et al. Role of his-tone H3 lysine 27 methylation in X inactivation. Science2003;300:131–135.

9. Yamada A, Fujii S, Daiko H, Nishimura M, Chiba T, OchiaiA. Aberrant expression of EZH2 is associated with a pooroutcome and P53 alteration in squamous cell carcinomaof the esophagus. Int J Oncol 2011;38:345–353.T

able4.(Continued)

Parameter

rs734004

rs734005

rs2072407

rs6464926

rs12670401

CC

CG

GG

P-value

CC

TC

TT

P-value

CC

TC

TT

P-value

CC

CT

TT

P-value

CC

TC

TT

P-value

Yes

111

16

111

16

16

11

19

11

88

11

9Saltedfood

0.516

0.460

0.315

0.975

0.795

Nooroccasionally

20

63

65

20

63

67

67

65

18

45

71

31

35

68

44

Yes

17

58

70

16

59

70

72

57

14

45

67

31

32

68

44

Thenumberof1genotypewas�5

,Fisher’sexact

test

wasused.Forothers,chi-squaredtest

wasused.

8 ZHOU ET AL.


10. Cai GH, Wang K, Miao Q, Peng YS, Chen XY. Expressionof polycomb protein EZH2 in multi-stage tissues of gastriccarcinogenesis. J Dig Dis 2010;11:88–93.

11. Mattioli E, Vogiatzi P, Sun A, et al. Immunohistochemicalanalysis of pRb2/p130, VEGF, EZH2, p53, p16(INK4A),p27(KIP1), p21(WAF1), Ki-67 expression patterns in gastriccancer. J Cell Physiol 2007;210:183–191.

12. Kodach LL, Jacobs RJ, Heijmans J, et al. The role of EZH2and DNA methylation in the silencing of the tumour sup-pressor RUNX3 in colorectal cancer. Carcinogenesis2010;31:1567–1575.

13. Sudo T, Utsunomiya T, Mimori K, et al. Clinicopathologicalsignificance of EZH2 mRNA expression in patients with he-patocellular carcinoma. Br J Cancer 2005;92:1754–1758.

14. Toll AD, Dasgupta A, Potoczek M, et al. Implications ofenhancer of zeste homologue 2 expression in pancreaticductal adenocarcinoma. Hum Pathol 2010;41:1205–1209.

15. Raman JD, Mongan NP, Tickoo SK, Boorjian SA, ScherrDS, Gudas LJ. Increased expression of the polycomb groupgene, EZH2, in transitional cell carcinoma of the bladder.Clin Cancer Res 2005;11:8570–8576.

16. Chang CJ, Yang JY, Xia W, et al. EZH2 promotes expan-sion of breast tumor initiating cells through activation ofRAF1-beta-catenin signaling. Cancer Cell 2011;19:86–100.

17. Chase A, Cross NC. Aberrations of EZH2 in cancer. ClinCancer Res 2011;17:2613–2618.

18. Nikoloski G, Langemeijer SM, Kuiper RP, et al. Somaticmutations of the histone methyltransferase gene EZH2 inmyelodysplastic syndromes. Nat Genet 2010;42:665–667.

19. Yoon KA, Gil HJ, Han J, Park J, Lee JS. Genetic polymor-phisms in the polycomb group gene EZH2 and the risk oflung cancer. J Thorac Oncol 2010;5:10–16.

20. Kuzmichev A, Nishioka K, Erdjument-Bromage H, TempstP, Reinberg D. Histone methyltransferase activity associat-ed with a human multiprotein complex containing the en-hancer of Zeste protein. Genes Dev 2002;16:2893–2905.

21. Muller J, Hart CM, Francis NJ, et al. Histone methyltrans-ferase activity of a Drosophila Polycomb group repressorcomplex. Cell 2002;111:197–208.

22. Ernst T, Chase AJ, Score J, et al. Inactivating mutations ofthe histone methyltransferase gene EZH2 in myeloid disor-ders. Nat Genet 2010;42:722–726.

23. Morin RD, Johnson NA, Severson TM, et al. Somatic muta-tions altering EZH2 (Tyr641) in follicular and diffuse largeB-cell lymphomas of germinal-center origin. Nat Genet2010;42:181–185.

24. Joshi P, Carrington EA, Wang L, et al. Dominant allelesidentify SET domain residues required for histone methyl-transferase of Polycomb repressive complex 2. J BiolChem 2008;283:27757–27766.

25. Sneeringer CJ, Scott MP, Kuntz KW, et al. Coordinatedactivities of wild-type plus mutant EZH2 drive tumor-associated hypertrimethylation of lysine 27 on histone H3(H3K27) in human B-cell lymphomas. Proc Natl Acad SciUSA 2010;107:20980–20985.

26. Breyer JP, McReynolds KM, Yaspan BL, Bradley KM,Dupont WD, Smith JR. Genetic variants and prostatecancer risk: Candidate replication and exploration of viralrestriction genes. Cancer Epidemiol Biomarkers Prev 2009;18:2137–2144.

27. Bachmann N, Hoegel J, Haeusler J, et al. Mutation screenand association study of EZH2 as a susceptibility genefor aggressive prostate cancer. Prostate 2005;65:252–259.

28. Crea F, Fornaro L, Paolicchi E, et al. An EZH2 polymor-phism is associated with clinical outcome in metastatic co-lorectal cancer patients. Ann Oncol 2011; [Epub ahead ofprint].

29. Heinemeyer T, Wingender E, Reuter I, et al. Databases ontranscriptional regulation: TRANSFAC, TRRD and COMPEL.Nucleic Acids Res 1998;26:362–367.

30. Zheng R, Blobel GA. GATA transcription factors and can-cer. Genes Cancer 2010;1:1178–1188.

31. Lee TI, Young RA. Transcription of eukaryotic protein-coding genes. Annu Rev Genet 2000;34:77–137.

32. Akiyama Y, Watkins N, Suzuki H, et al. GATA-4 andGATA-5 transcription factor genes and potential down-stream antitumor target genes are epigenetically silencedin colorectal and gastric cancer. Mol Cell Biol 2003;23:8429–8439.

33. Chu H, Cao W, Chen W, et al. Potentially functionalpolymorphisms in IL-23R and risk of esophageal cancerin a Chinese population. Int J Cancer 2011;130:1093–1097.



Documents

EZH2 genetic variants affect risk of gastric cancer in the Chinese Han population