matrix metalloproteinase3 gene promoter polymorphisms and their haplotypes are associated with...

MOLECULAR CARCINOGENESIS

Matrix Metalloproteinase3 Gene PromoterPolymorphisms and Their Haplotypes AreAssociated with Gastric Cancer Risk in EasternIndian Population

Sanjib Dey,1 Sami Stalin,1 Arnab Gupta,2 Debjit Saha,1 Kousik Kesh,1 and Snehasikta Swarnakar1*1Department of Physiology, Drug Development Diagnostic and Biotechnology Division, Indian Institute of ChemicalBiology, Kolkata, India2Cancer Center Welfare Home and Research Institute, Kolkata, India

Single nucleotide polymorphisms (SNPs) of matrix metalloproteinase3 (MMP3) promoter in the development and

progression of gastric cancer of whole stomach has never been investigated in any population. We conducted ahospital-based case-control study to explore the MMP3 SNPs and their haplotypes with the risk of gastric cancer forthe first time in eastern Indian population. A total of 218 gastric cancer patients and 175 healthy controls were

genotyped for MMP3-1612 5A/6A (rs3025058) by PCR–RFLP and rechecked 10% by DNA sequencing. MMP3-707 A/G (rs522616) and MMP3-375 C/G (rs617819) were genotyped by DNA sequencing among 209 patients and 154controls. MMP3-1612 5A6A genotype (P ¼ 0.026, odds ratio (OR) ¼ 1.756, confidence interval (CI) ¼ 1.070–2.883),

combined 5A5A and 5A6A genotype (P ¼ 0.015, OR ¼ 1.791, CI ¼ 1.122–2.858) and 5A allele (P ¼ 0.002,OR ¼ 1.75, CI ¼ 1.21–2.53) and; MMP3-707 GG genotype (P ¼ < 0.0001; OR ¼ 9.612; 95% CI ¼ 3.403–27.147),combined GG and AG genotype (P ¼ 0.001, OR ¼ 2.201, CI ¼ 1.385–3.498) and G allele (P ¼ <0.0001, OR ¼2.189, CI ¼ 1.582–3.033) conferred significant risk for gastric cancer development. Also, tobacco addicted individualswith combined 5A5A and 5A6A genotype (P ¼ 0.005, OR ¼ 2.952, CI ¼ 1.377–6.327) at �1612 position of MMP3promoter displayed a higher risk to gastric cancer development. The genotypic combinations of all three MMP3 pro-moter polymorphisms and their haplotypes with increasing risk allele in a dose-dependent manner showed a potential

risk for developing gastric cancer. The analyses suggested that the MMP3-707 G/G and MMP3-1612 5A/6A polymor-phisms are potential independent predictors of gastric cancer risk development. � 2011 Wiley Periodicals, Inc.

Key words: Association study; DNA sequencing; PCR–RFLP; single nucleotide polymorphism; odds ratio

INTRODUCTION

Gastric cancer is the fourth most common ma-lignancies in the world and is the leading cause ofcancer related death in many Asian countries [1,2].Recent studies have demonstrated that degrada-tion of extracellular matrix (ECM) and basementmembrane barriers by matrix metalloproteinases(MMPs) plays an important role in tumorigenesisby modulating cell proliferation, apoptosis, hostimmune surveillance, tumor invasion, and metas-tasis [3–5]. MMP3, also called stromelysin-1, a keymember of the MMP family with broad substratespecificity is known to degrade basal membranecollagens, proteoglycans, gelatine, fibronectin,laminin, and elastin rendering MMP3 crucialin connective tissue remodeling and angiogenesis[6–8]. MMP3 gene is localized in a MMP cluster of400 kb at chromosome 11q21–23 that counts nineMMPs, and may activate other MMPs includingcollagenase, matrilysin, and gelatinase B [6–8].These activities of MMP3 are believed to be linkedto both cancer initation and progression. The ex-pression of MMP3 in carcinogenesis is regulated

primarily at the transcriptional level, but there isalso evidence of modulation of mRNA stability inresponse to growth factors and cytokines secretedby inflammatory cells as well as stromal cells[9–11]. Regulatory effects of such stimuli are medi-ated through a number of cis-elements located inthe stromelysin-1 promoter [12–14]. MMP3 tran-scription is higher in head and neck squamous cellcarcinoma (HNSCC) and several other types of

Additional Supporting Information may be found in the onlineversion of this article.

Abbreviations: SNP, single nucleotide polymorphism; MMP,matrix metalloproteinase; ECM, extracellular matrix; HWE, Hardy–Weinberg equilibrium; PCR–RFLP, polymerase chain reactionrestriction fragment length polymorphism; OR, odds ratio; CI, con-fidence interval.

S Dey, S Stalin, and A Gupta contributed equally to this work.

*Correspondence to: Department of Physiology, Indian Instituteof Chemical Biology, 4, Raja S.C. Mullik Road, Jadavpur, Kolkata700032, India.

Received 16 February 2011; Revised 18 October 2011; Accepted19 October 2011

DOI 10.1002/mc.21837

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

� 2011 WILEY PERIODICALS, INC.

malignancies including lung and breast carcino-mas [15–18]. Several single nucleotide polymor-phisms (SNPs) in the MMP3 promoter region havebeen identified and functional polymorphisms ofMMP3 promoter modulate transcriptional activityof MMP3 and may increase the risk of individualtumor incidence [7,18–20].

The promoter region of MMP3 is characterizedby a 5A/6A promoter polymorphism at position�1612 [21]. An in vitro promoter activity assayshowed that the MMP3 �1612 5A allele has a two-fold greater promoter activity than the �1612 6Aallele, suggesting a transcriptional repressor mightbind to the 6A allele [7,22]. The MMP3-1612 5A/6A polymorphism has been reported to be associat-ed with both susceptibility and invasiveness ofbreast cancer but not ovarian cancer [18,23]. In arecent study, the MMP3 polymorphism is reportedto be a predictive factor of response to neoadju-vant chemotherapy in HNSCC [24]. Association ofMMP3 SNPs with gastric cardiac adenocarcinomahas been reported in Chinese population. Thefunctional polymorphism of MMP3 promoter hasbeen associated with lymph node metastasis, vas-cular invasion and prognosis in esophageal squa-mous cell carcinoma but not with gastric cardiacadenocarcinoma [25]. However, the influence ofthis SNP on tumor development and invasion isnot consistent in different tumor types. Furthermolecular analyses of the MMP3 promoterrevealed a number of sequence variants in the pro-moter region of this gene [26]. Four known SNPs(�1487 T/G, �1340 A/C, �707 A/G, and �375 G/C) are present between the �1612 5A/6A polymor-phic site and the transcriptional start site of MMP3promoter. A combination of sequence variations inthe five polymorphic sites within the MMP-3 pro-moter region constitutes the MMP3 promoter hap-lotype. Patients homozygous for 5A show similarhaplotype (5A-G-A-A-G) while three different hap-lotypes (6A-G-C-A-C, 6A-G-C-G-C, and 6A-T-C-A-C) are found in patients homozygous for 6A [19].

So far, relationship between the MMP3 polymor-phisms and risk of development of gastric cancerof whole stomach as well as progression of gastriccancer has not been reported in any other includ-ing Indian population. In the present study, weconducted a hospital-based case-control study toexplore the role of three MMP3 promoter SNPs(�1612 5A/6A, �707 A/G, and �375 G/C) andtheir resultant haplotypes with risk of developinggastric cancer in eastern Indian population. Wealso explored the relationship between the poly-morphisms and the clinicopathological factorsamong gastric cancer patients. The study demon-strated for the first time the putative association ofMMP3-1612 5A/6A and MMP3-707 A/G polymor-phisms with risk of gastric cancer in eastern Indianpopulation.

METHODS AND MATERIALS

Study Population

The eastern Indian case–control cohort consistedof 218 gastric cancer cases and 175 control individ-uals. All patients with a clinical diagnosis of gastriccancer attending the Departments of Gastro-oncol-ogy of CCWHRI, Kolkata, Department of GastricSurgery, Medical College and Hospital, Kolkata,IPGMER, Kolkata during June 2008 to May 2010were identified from corresponding hospital regis-tries and contacted during follow-up investiga-tions. All participants gave written informedconsent for participation and the study protocolswere approved by the Ethical Review Boards of theconcerned hospitals and Human Ethics Committeeof Indian Institute of Chemical Biology, Kolkata,India. Patient’s demographics, symptoms, and tu-mor grading were recorded and blood samples col-lected after taking history and performing clinicalexaminations and endoscopic examination. Histo-logical tumor typing was determined on the basisof biopsies or resected specimens. The exclusioncriteria included previous cancer and other metas-tasized cancer. The diagnosis of gastric cancer andTNM staging was based on generally accepted clin-ical, histological, radiological, and immunofluores-cence findings and that of international AJCC-UICC criterion [27]. Individuals who formerly orcurrently addicted to tobacco for at least 2 yr weredefined as tobacco addicted. Persons visiting thesehospitals for routine check-up and not havinghistory or diagnosis of any cancer and geneticdiseases were also asked to participate and volun-teered to donate blood for the study and consid-ered as healthy controls. All of the cancer patientsand control subjects were unrelated Indian nation-ality from West Bengal or the surrounding easternIndian states and considered representative of anEastern Indian population.

DNA Extraction

From each subject, 5 ml of venous blood weredrawn aseptically in vacutainer tubes (QiagenGmbH, Feldbachstrasse, Switzerland) containingEDTA and stored at 48C prior to genomic DNAextraction. Genomic DNA was extracted within2 wk after sampling from whole-blood samplesusing QIAamp DNA Blood Midi Kit (Qiagen)according to the manufacturer’s protocol. Bloodvials, serum tubes, DNA tubes, and patient ques-tionnaire forms were equipped with correspondingsample identification numbers.

Primers and PCR Amplification

Primers were designed with FastPCR (www.biocenter.helsinki.fi) software so as to amplifypromoter regions of the MMP3 gene to analyzepolymorphisms by sequencing (Table 1). PCR was

2 DEY ET AL.

Molecular Carcinogenesis

performed in PCR SPRINT Thermal Cycler (ThermoElectron Corporation, Japan). The target sequencewas amplified in a 25-ml reaction volume contain-ing 10–20 ng of genomic DNA, 0.2 mM dNTP,10 mM Tris–HCl (pH 8.3), 50 mM KCl, 2 mMMgCl2, 0.3 mM of each primer, and 1.0 units ofTaq DNA polymerase (Fermentas Taq DNA poly-merase; Fermentas). The PCR amplification wascarried out with 35 cycles of denaturation at 948Cfor 30 s, annealing at 58–598C for 1 min (Table 1)followed by extension at 728C for 30 s after theinitial activation step of 948C for 5 min. PCR frag-ments were analysed comparing with 100 bp DNAladder (Fermentas, Vilnius, Lithuania) on anethidium bromide stained 1.5% agarose (Cat No.014011; Sisco Research Laboratories, Mumbai,Maharashtra, India) gel run for 60 min at 100 V.

Genotyping

For genotyping by sequencing, any unincorpo-rated dNTPs were dephosphorylated and unincor-porated primers were removed from PCR productby shrimp alkaline phosphatase (Fermentas) andexonuclease-I enzyme (Fermentas). PCR productswere used for sequencing with BigDye1 Termina-tor v3.1 Cycle Sequencing Kit (Applied Biosystems,Foster City, CA) and were sequenced by ABI PRISM3100 genetic analyzer (Perkin–Elmer ABI, FosterCity, CA) according to the manufacturer’s proto-col. The PCR amplicon was sequenced in bothdirections with forward and reverse PCR primerseliminating the possibility of compression arti-facts. Sequencing chromatograms were analysedusing Sequence Scanner Software v1.0 (AppliedBiosystems) to analyse alterations of the nucleoti-des and thus genotypes.The MMP3-1612 5A/6A polymorphism was gen-

otyped by polymerase chain reaction restrictionfragment length polymorphism (PCR–RFLP) meth-od as described previously [25].

Western Blotting

Equal volume of serum samples from controls(n ¼ 28) and patients (n ¼ 28) were passed

through gelatin agarose affinity matrix followed bydetection of MMP3 expression by Western blottingusing anti-MMP3 polyclonal antibody (Santa CruzBiotech, Santa Cruz, CA) as described previously[28]. Briefly, Serum samples were resolved by 10%reducing SDS–polyacrylamide gel electrophoresisand transferred to nitrocellulose membranes. Themembranes were blocked for 2 h at room tempera-ture in 3% bovine serum albumin (BSA) solutionin 20 mM Tris–HCl, pH 7.4 containing 150 mMNaCl and 0.02% Tween 20 (TBST) followed byovernight incubation at 48C in 1:200 dilution ofthe primary antibody in TBST containing 0.2%BSA. The membranes were washed five times withTBST and then incubated with alkaline phospha-tase-conjugated secondary antibody (1:2,000) for1.5 h. The bands were visualized using 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazoliumsubstrate solution.

Functional SNP Searching

For any putative functional findings, we furtherexamined SNPs function by using the relevant bio-informatics software or databases. SNPs that werepredicted to affect transcription factors bindingsites in the promoter region, we further evaluatedtheir effects on the transcription factor bindingin Matinspector software (http://genomatix.gsf.de/cgi-bin/matinspector/matinspector.pl), TFSEARCH(http://www.cbrc.jp/research/db/TFSEARCH) andTESS (http://www.cbil.upenn.edu/tess). All of themwere using the same TRANSFAC (http://thr.cit.nih.gov/molbio/signal/) database.

Statistical Analysis

Statistical analysis was performed using the SPSSsoftware (version 16.0J. SPSS, Inc., Chicago, IL).Significant differences between age at interviewfor controls and age at diagnosis for cases wereassessed using the Student’s t-test for comparisonof means using GraphPad InStat3 software (Graph-Pad Software, Inc., San Diego, CA). Hardy–Wein-berg equilibrium (HWE) analyses were performedto compare observed and expected allele

Table 1. PCR Primers used for Polymorphism Analysis

Polymorphisms Primer sequence (50–30) Ta (8C)Productsize (bp)

Screeningmethod used

MMP3-1,612 5A/6A TGACTCTTCCTGGGACTTGGGAA (F)CAATGCTTCCCTGATCACCACA (R)

58 974 Sequencing

MMP3-1,612 5A/6A GGTTCTCCATTCCTTTGATGGGGGGAAAGA (F)CTTCCTGGAATTCACATCATGCCACCACT (R)

58 129 PCR–RFLP

MMP3-707 A/G GTCATAGGGATCTTATTGCCACAG (F)GGATTTGCTGGTTCTTGAGGAGGA (R)

58.5 777 Sequencing

MMP3-375 C/G GGACAGAGAGAATTTCAGTCCGGT (F)CCTGAACAAGGTTCATGCTGGTGT (R)

58 920 Sequencing

Sequencing was performed with the same PCR primers.Primers for PCR–RFLP analysis of MMP3-1,612 5A/6A was taken from Ref. [25].

MMP3 PROMOTER POLYMORPHISMS AND GASTRIC CANCER RISK 3


frequencies using a chi-square test for controls toensure that each marker was in equilibrium(P > 0.05). The minor allele frequency (–modeloption) and HWE (–hardy option) for each SNP wasestimated from the control population using Plinkv0.99 [29]. Case–control data were analyzed usingtwo-sided 2-by-2 or 2-by-3 contingency tablesaccording to the genotype by the Pearsonchi-square test. The odds ratio (OR) and 95% confi-dence interval (CI) of the genotypes were calculat-ed from a multivariate logistic regression modeladjusted for age (continuous variable), sex andtobacco addiction. In this study, we defined thatthe 6A allele of the 5A-1612 6A SNP, the A allele ofthe A-707G SNP and C allele of the C-375G SNPwere reference alleles. The results were evaluatedwith above allele as a reference using the multivar-iate logistic regression model. In analyzing therelationship between the SNP genotypes anddisease status of GC, the stage of cancer and depthof tumor invasion were transformed to binary data(Stage I þ II vs. stage III þ IV and T1 þ T2 vs.T3 þ T4). The relationship between genotype dis-tributions and clinicopathological parameters wasalso examined using a multivariate logistic regres-sion model adjusted for age (continuous variable),sex and tobacco addiction as a potential confound-ing factor. Haplotype frequencies were analyzedusing the program Haploview (ver. 4.1; Broad Insti-tute, Chembridge, MA) [30]. All results were con-sidered statistically significant if the P-value was<0.05.

RESULTS

Description of the Study Population

The gastric cancer patients’ population consistedof 167 (76.6%) males and 51 (23.4%) females withmean age of 54.2 yr (�11.5), and control subjectshaving a mean age of 50.5 yr (�13.7) consisted of115 (65.7%) males and 60 (34.3%) females. Patientpopulation was consisting of 58.7% non-cardiacgastric cancers (middle and lower stomach cancer)and 33% cardiac associated gastric cancers (gastro-esophagial junction, fundus, and upper body ofstomach). There were statistically significant differ-ences in the distribution of sex (P ¼ 0.02) and age(P ¼ 0.004) between patients and controls. Also,significantly more tobacco-addicted individuals(P ¼ 0.0001) were present among patients (65.6%)compared to controls (28.6%). So, in the courseof risk estimations, an age, sex, and addictionadjusted ORs were calculated.

Polymorphism Detection in MMP3 Promoter

Naturally occurring, common sequence variantsof the MMP3 gene promoter in cases and controlswere searched. Sequencing of the promoter regionwas conducted with different sets of primers

(Table 1). Alignment of sequence chromatogramwith the MMP3 gene promoter contig sequence(Genebank accession no. AF405705) confirmed theSNP position at �1612 (5A/6A), �707 (A/G), and�375 (G/C) (Figure 1A and 1C). MMP3-1612 5A/6A (MMP3.1) genotyping by PCR–RFLP was thenperformed among the gastric cancer patients andcontrols. Figure 1B showed a typical PCR–RFLPpattern of MMP3 at �1612 5A/6A polymorphism.MMP3-707 A/G (MMP3.2) and MMP3-375 G/C(MMP3.3) genotypes were determined by PCR-se-quencing. The MMP3.1 genotype analysis by se-quencing was restricted to a pilot study as well asfor 10% of the population for rechecking. MMP3.1genotyping were performed on 218 patients and175 controls and for MMP3.2 and MMP3.3 geno-typing performed on 209 patients and 154controls.

Association Between Individual SNPs of MMP3

and Gastric Cancer Risk

The genotype distributions of MMP3 polymor-phisms were consistent with the HWE. MMP3.1polymorphism was accounted for a significant dif-ference in 5A6A genotype distribution (P ¼ 0.026,OR ¼ 1.756, CI ¼ 1.070–2.883), 5A6A and 5A5Acombined genotype distribution (P ¼ 0.015, OR ¼1.791, CI ¼ 1.122–2.858) and 5A allele distribution(P ¼ 0.002, OR ¼ 1.75, CI ¼ 1.21–2.53) betweenpatients and controls and associated with gastriccancer risk (Table 2).For the MMP3.2 polymorphism, GG genotype

(P ¼ <0.0001; OR ¼ 9.612; 95% CI ¼ 3.403–27.147),combined GG and AG genotype (P ¼ 0.001, OR ¼2.201, CI ¼ 1.385–3.498) and G allele (P ¼ <0.0001,OR ¼ 2.189, CI ¼ 1.582–3.033) conferred signifi-cant risk for gastric cancer development (Table 2).However, the MMP3.3 polymorphism showedno significant risk for gastric cancer occurrence(Table 2).

Association Between Individual SNPs of MMP3 andClinicopathological Features at the Time of Diagnosis of

Gastric Cancer

For MMP3.1 polymorphism, individuals of�50 yr age group with combination of 5A5A and5A6A genotype (P ¼ 0.008, OR ¼ 2.307, CI ¼1.245–4.276) conferred higher risk for gastric can-cer development (Table 3). Males with combina-tion of 5A5A and 5A6A (P ¼ 0.02, OR ¼ 1.927,CI ¼ 1.111–3.341) genotype also displayed ahigher risk to gastric cancer. In addition, tobaccoaddicted cases having combination of 5A5A and5A6A (P ¼ 0.005, OR ¼ 2.952, CI ¼ 1.377–6.327)genotype displayed a higher risk to gastric cancerdevelopment (Table 3).For MMP3.2 polymorphism, peoples above 50 yr

of age with GG (P ¼ 0.006, OR ¼ 18.47, CI ¼2.33–146.38) and combination of AG and GG

4 DEY ET AL.


(P ¼ 0.009, OR ¼ 2.228, CI ¼ 1.224–4.056) geno-type displayed a higher risk to gastric cancer(Table 3). Peoples belonging to below 50 yr of agegroup with GG (P ¼ 0.003, OR ¼ 7.38, CI ¼ 1.986–27.426) and combination of AG and GG(P ¼ 0.039, OR ¼ 2.161, CI ¼ 1.04–4.492) geno-type displayed a higher risk to gastric cancer(Table 3). Males with the GG (P ¼ 0.001,OR ¼ 8.321, CI ¼ 2.286–30.29) and combinationof AG and GG (P ¼ 0.005, OR ¼ 2.186, CI ¼ 1.27–3.762) genotype displayed a higher risk to gastriccancer. Females with the GG (P ¼ 0.005,OR ¼ 12.439, CI ¼ 2.16–71.68) genotype also con-ferred the risk to gastric cancer. Addicted peopleshaving combination of GG and AG (P ¼ 0.006,

OR ¼ 2.766, CI ¼ 1.333–5.736) genotype and non-addicted peoples with GG (P ¼ 0.001, OR ¼ 6.918,CI ¼ 2.183–21.92) genotype displayed a higher riskto gastric cancer (Table 3). Results suggested thatpeoples having homozygous G allele were in riskfor gastric cancer irrespective of their gender, ageand addiction status. Patients with GG genotype(P ¼ 0.026, OR ¼ 2.889, CI ¼ 1.134–7.361) showedsignificant risk for higher TNM staging (Table 3).Moreover, combination of AG and GG genotypewere found more frequently in gastric cancerpatients with 10 or more metastatic lymph nodes(P ¼ 0.05, OR ¼ 2.119, CI ¼ 0.994–4.518) and withhigher TNM staging (P ¼ 0.01, OR ¼ 2.197, CI ¼1.209–3.991). Also, patients with GG (P ¼ 0.045,

Figure 1. Genotyping of the MMP3 polymorphisms. (A) Sche-matic representation of MMP3 promoter polymorphisms. Locationof the polymorphic sites within MMP3 promoter and the natureof the polymorphisms are shown. The promoter region is not inscale. (B) Schematic representation of MMP3.1 RFLP analysis andrepresentative PCR–RFLP analysis of MMP3.1 polymorphism. Allpossible combinations of DNA fragments resulting from PCR–RFLPare shown. 129 bp target region of MMP3 gene promoter was

PCR-amplified and digested with Tth111I, which cleaved the 5Aallele to generate 97 bp and 32 bp fragments while 6A alleleremained undigested. Heterozygous allele shows three bands of129, 97, and 32 bp; (C) Representative chromatogram of DNAsequences for each genotype for the MMP3 polymorphisms withtheir flanking regions. Arrows indicate the position of polymor-phism in the chromatogram.



OR ¼ 2.383, CI ¼ 1.021–5.565) and combination ofAG and GG (P ¼ 0.037, OR ¼ 1.969, CI ¼ 1.042–3.721) genotype conferred significant risk for dis-tant metastasis. So, patient with GG genotype andcombination of AG and GG genotype showed sig-nificantly greater risk for progression of gastriccancer (Table 3). For MMP3-375 G/C polymor-phism, GG genotype showed a protective effect onpatients having large tumor size (�5 cm) (P ¼0.022, OR ¼ 0.306, CI ¼ 0.111–0.841) (Table 3).

Distribution of Genotype Combinations of MMP3 SNPsAmong Gastric Cancer Patients and Controls

Since MMP3.1 and MMP3.2 polymorphism indi-vidually accounted for the risk of gastric cancer,the distribution of linked paired loci frequencywere observed for evaluation of the effect ofincreasing order of variant alleles on the risk ofgastric cancer development. The potential com-bination between the MMP3.1 and MMP3.2polymorphisms and; MMP3.2 and MMP3.3 poly-morphisms showed gradual increase in risk for gas-tric cancer in a dose dependent manner (Table 4).The combination between the MMP3.1 andMMP3.3 showed no trend of increasing risk withincreasing order of variant allele (Table 4). Theall risk allele for all the paired loci combinations

(5A5A-GG, GG-GG and 5A5A-GG for MMP3.1-3.2,MMP3.2-3.3, and MMP3.1-3.3, respectively) failedto show any significant risk due to very small sam-ple numbers (Table 4).The potential combinations of all three MMP3

promoter polymorphisms linked loci were groupedinto four categories based on the presence of thenumber of variant alleles and risk were assessed.The difference in genotype distribution of thesethree polymorphisms among control and gastriccancer patient groups become significant whenpolymorphic alleles (5A-G-G) were present in allthree loci showing a trend in increase in risk(Table 5).

Association Between Haplotypes of MMP3 SNPsand Gastric Cancer Risk

The haplotype frequencies of all three MMP3polymorphisms showed a significant difference be-tween gastric cancer patients and healthy controls.Eight haplotypes were identified with a frequencyhigher than 1% and categorized by increasingorder of polymorphic variant alleles and analyzedfor gene-dosage effect (Table 5). The difference inthe haplotype frequency was increased in a dose-dependent manner between patients and controlsas the number of polymorphic variant alleles were

Table 2. Association Between MMP3 SNPs and the Risk of Gastric Cancer

Genotype/allele

GC Patient Controls

OR 95% CI P-valuen % n %

MMP3.1 (S1612 5A/6A) 218 1755A5A 16 7.3 7 4 1.962 0.742–5.189 0.1755A6A 70 32.1 38 21.7 1.756 1.070–2.883 0.0266A6A 132 60.6 130 74.3 1 (Ref)5A5A þ 5A6A 86 39.4 45 25.7 1.791 1.122–2.858 0.0155A allele 102 23.4 52 14.9 1.75 1.21–2.53 0.00276A allele 334 76.6 298 85.1 1 (Ref)MMP3.2 (S707 A/G) 209 154AA 75 35.9 85 55.2 1 (Ref)AG 97 46.4 64 41.6 1.657 1.023–2.682 0.04GG 37 17.7 5 3.2 9.612 3.403–27.147 <0.0001AG þ GG 134 64.1 69 44.8 2.201 1.385–3.498 0.001A allele 247 59.1 234 75.9 1 (Ref)G allele 171 40.9 74 24.1 2.189 1.58–3.033 <0.0001MMP3.3 (S375 G/C) 209 154GG 25 12 10 6.5 1.658 0.685–4.014 0.262GC 90 43 78 50.6 0.799 0.497–1.287 0.356CC 94 45 66 42.9 1 (Ref)GC þ GG 115 55 88 57.1 0.904 0.572–1.431 0.667G allele 140 33.5 98 31.8 1.079 0.788–1.478 0.634C allele 278 66.5 210 68.2 1 (Ref)

OR, odds ratio; CI, confidence interval; Ref, reference genotype or allele to calculate OR.Adjusted OR calculated for age, sex and addiction, by binary logistic regression model using SPSS v16.0 software.P-value is for x2 test showing the significance of difference in the distributions of the genotypes and alleles between patients andcontrols.Values in bold indicate positive significance (P < 0.05).

6 DEY ET AL.


Table

3.AssociationBetw

eentheGenotypesofMMP3SNPsandClinicopathologicalCharacteristicsofGastricCancerPatients

MMP3.1

(�1,6125A/6A),n(%

)

MMP3.2

(�707A/G),n(%

)

MMP3.3

(�375G/C),n(%

)

5A/5A

66A/6A

5A/6A

þ5A/5A

A/A

G/G

A/G

þG/G

G/G

C/C

G/C

þGG

Age(yr)

�50(Case/Cont)

13/4

(8.6/4.3)

89/71(58.9/76.3)

62/22(41.1/23.7)

53/45(36.6/56.2)

24/1

(16.6/1.2)

92/35(63.5/43.7)

15/4

(10.3/5.0)

63/41(43.4/51.2)

82/39(56.5/48.8)

OR(95%

CI)(a)

2.494(0.739–8.424)

1(Ref)

2.307(1.245–4

.276)

1(Ref)

18.47(2.33–1

46.38)

2.228(1.224–4

.056)

1.76(0.499–6.215)

1(Ref)

1.153(0.638–2.083)

<50(Case/Cont)

3/3

(4.5/3.7)

43/59(64.2/71.9)

24/23(35.8/28.1)

22/40(34.4/54.1)

13/4

(20.3/5.4)

42/34(65.6/45.9)

10/6

(15.6/8.1)

31/25(48.4/33.8)

33/49(51.5/66.2)

OR(95%

CI)(a)

1.206(0.210–6.936)

1(Ref)

1.205(0.573–2.531)

1(Ref)

7.38(1.986–2

7.426)

2.161(1.04–4

.492)

1.743(0.48–6.327)

1(Ref)

0.625(0.30–1.303)

Gender

Male(Case/Cont)

15/5

(9.0/4.3)

98/86(58.7/74.8)

69/29(41.3/25.2)

59/56(36.9/56.0)

25/3

(15.6/3.0)

101/44(63.1/44.0)

20/5

(12.5/5.0)

71/39(44.4/39.0)

89/61(55.6/61.0)

OR(95%

CI)(b)

2.44(0.818-7.282)

1(Ref)

1.927(1.111–3

.341)

1(Ref)

8.321(2.286–3

0.29)

2.186(1.27–3

.762)

1.677(0.547–5.139)

1(Ref)

0.777(0.451–1.338)

Female(Case/Cont)

1/2

(2.0/3.3)

34/44(66.6/73.3)

17/16(33.3/26.7)

16/29(32.7/53.7)

12/2

(24.5/3.7)

33/25(67.4/46.3)

5/5

(10.2/9.3)

23/27(46.9/50.0)

26/27(53.1/50.0)

OR(95%

CI)(b)

1.05(0.085–12.995)

1(Ref)

1.591(0.638–3.970)

1(Ref)

12.439(2.16–7

1.68)

2.121(0.851–5.286)

2.387(0.554–10.29)

1(Ref)

1.444(0.587–3.552)

Addictionto

tobacco

Yes(Case/Cont)

13/1

(9.1/2.0)

82/39(57.3/78.0)

61/11(42.7/22.0)

47/25(34.3/59.5)

23/0

(16.8/0)

90/17(65.7/40.5)

18/2

(13.1/4.8)

62/15(45.3/35.7)

75/27(54.7/64.3)

OR(95%

CI)(c)

7.226(0.909–57.477)

1(Ref)

2.952(1.377–6

.327)

1(Ref)

2.766(1.333–5

.736)

2.601(0.541–12.5)

1(Ref)

0.678(0.325–1.418)

No(Case/Cont)

3/6

(4.0/4.8)

50/91(66.7/72.8)

25/34(33.3/27.2)

28/60(38.9/53)

14/5

(19.4/4.5)

44/52(60.3/46.5)

7/8

(9.7/7.1)

32/51(44.4/45.5)

30/61(57.4/55.8)

OR(95%

CI)(c)

0.837(0.198–3.542)

1(Ref)

1.32(0.707–2.465)

1(Ref)

6.918(2.183–2

1.92)

1.826(0.996–3.35)

1.624(0.512–5.155)

1(Ref)

1.133(0.615–2.088)

Tumorsize

�5cm

/<5cm

5/11(5.0/9.3)

67/65(67.0/55.1)

33/53(33.0/44.9)

34/41(35.4/36.3)

16/21(16.7/18.6)

62/72(64.6/63.7)

6/19(6.2/16.8)

47/47(49.0/41.6)

49/66(51.0/58.4)

OR(95%

CI)(d)

0.415(0.134–1.287)

1(Ref)

0.609(0.349–1.063)

1(Ref)

0.913(0.408–2.039)

1.047(0.591–1.853)

0.306(0.111–0

.841)

1(Ref)

0.747(0.431–1.296)

Depth

ofinvasion(�)

T3þ

T4/T1þ

T2

8/8

(9.3/6.1)

53/79(61.6/59.8)

33/53(38.4/40.2)

30/45(36.1/35.7)

21/16(25.3/12.7)

53/81(63.9/64.3)

12/13(14.5/10.3)

35/59(42.2/46.8)

48/67(57.8/53.2)

OR(95%

CI)(d)

1.468(0.505–4.236)

1(Ref)

0.892(0.506–1.571)

1(Ref)

2.137(0.878–5.203)

0.934(0.519–1.682)

1.607(0.645–4.005)

1(Ref)

1.233(0.699–2.178)

Regionallymphnodemetastasis

Nþ

ve/N

�ve

12/4

(6.5/11.8)

14/18(62.0/52.9)

70/16(38.0/47.1)

58/17(33.0/51.5)

32/5

(18.2/15.2)

118/16(67.0/48.5)

21/4

(11.9/12.1)

80/14(45.5/42.4)

96/19(54.5/57.6)

OR(95%

CI)(d)

0.467(0.13–1.678)

1(Ref)

0.679(0.323–1.429)

1(Ref)

1.876(0.621–5.671)

2.119(0.994–4

.518)

0.885(0.26–3.014)

1(Ref)

0.889(0.417–1.894)

Distantmetastasis

Yes/no

4/12(5.3/8.4)

45/87(60.0/60.8)

30/56(40.0/39.2)

19/56(26.0/41.2)

17/20(23.3/14.7)

54/80(74/58.8)

7/18(9.6/13.2)

31/63(42.5/46.3)

28/53(57.5/53.7)

OR(95%

CI)(d)

0.802(0.237–2.714)

1(Ref)

1.101(0.614–1.973)

1(Ref)

2.383(1.021–5

.565)

1.969(1.042–3

.721)

0.779(0.28–2.166)

1(Ref)

1.213(0.675–2.181)

TNM

classification(y)

StageIIIþ

IV/Iþ

II10/6

(7.2/7.6)

86/46(61.9/58.2)

53/33(38.1/41.8)

39/36(29.1/48.0)

28/9

(20.9/12.0)

95/39(70.9/52.0)

16/9

(11.9/12.0)

58/36(43.3/48.0)

51/30(56.7/52.0)

OR(95%

CI)(d)

0.982(0.326–2.956)

1(Ref)

0.875(0.492–1.556)

1(Ref)

2.889(1.134–7

.361)

2.197(1.209–3

.991)

1.104(0.425–2.868)

1(Ref)

1.244(0.697–2.22)

P-value,ORsand95%

CIswere

calculatedbyunconditionalbinary

logisticregression(usingSPSSv16software)withrespectivereference

genotypes.

(a)adjustedwithsexandaddiction.(b)adjustedwith

ageandaddiction.(c)adjustedwithageandsex.

(d)adjustedwithage,sex,

andaddiction.

Valuesin

bold

indicate

positive

significance

(P<

0.05).

(�)Depth

oftumorwasdefinedaccordingto

thecriterionofAJCC.

(y)TNM

(Tumor,node,metastasis)

stagingswere

classifiedaccordingto

thecriterionoftheAmericanJointCommitteeonCancer(AJCC)andtheInternationalUnionAgainst

Cancer(UICC)TNM

stage

grouping.[27]


increased (Table 5) resulting in the increase of riskfor gastric cancer development.

In Silico Promoter Activity of MMP3 Promoter:Alteration Due to SNPs

Potential transcription binding sites in theMMP-3 promoter were predicted using Matinspec-tor software, TFSEARCH, and TESS. A database

search for transcription factor binding sites showedconsiderable changes in transcription factor con-sensus binding sites at the MMP3 promoter poly-morphisms at positions �707 bp and �375 bpbeside �1,612 bp (Figure 2). The expression ofMMP3 in serum of gastric cancer patients showed1.8 times higher compared to that of control asjudged by western blot analysis (Figure 2, inset).

Table 4. Distribution of MMP3 Paired Loci Polymorphisms With Increasing Order of Variant Alleles and AssociationWith the Risk of Gastric Cancer

Control, n (%) Patient, n (%) OR 95% CI P

MMP3.1-3.20 66 (42.9) 47 (22.5) 1 (Ref)1 60 (390) 80 (38.2) 1.872 1.133–3.093 0.0162 24 (15.6) 62 (29.7) 3.628 1.987–6.622 <0.00013 4 (2.5) 15 (7.2) 5.266 1.643–16.881 0.0024 0 5 (2.4)MMP3.2–3.30 39 (25.3) 32 (15.3) 1 (Ref)1 68 (44.2) 82 (39.2) 1.47 0.833–2.592 0.1972 38 (24.7) 65 (31.1) 2.085 1.26–3.858 0.023 8 (5.2) 21 (10.0) 3.199 1.251–8.183 0.0154 (�) 1 (0.6) 9 (4.3)MMP3.1–3.30 50 (32.5) 69 (33.0) 1 (Ref)1 74 (48.0) 65 (31.1) 0.636 0.388–1.043 0.0812 20 (13.0) 58 (27.7) 2.101 1.124–3.927 0.0223 10 (6.5) 11 (5.3) 0.797 0.314–2.022 0.6414 0 6 (2.9)

OR, odds ratio; CI, confidence interval; Ref, reference.0 ¼ no risk allele for both loci, 1 ¼ one risk allele for any one loci, 2 ¼ two risk allele for both the loci, 3 ¼ one risk allele for anyone locus and two risk allele for another locus, 4 ¼ All risk allele for both the loci.Adjusted OR calculated for age, sex, addiction, by binary logistic regression model using SPSS v16.0 software.P-value is for x2-test showing the significance of difference in the distributions of the variant alleles between patients and controls.Values in bold indicate positive significance (P < 0.05).No risk allele for both the gene was taken as reference to calculate OR.(�) OR not calculated because of very low frequency in control population.

Table 5. Distribution of MMP3 SNPs Linked Loci and Their Haplotypes With Increasing Order of Variant Alleles andAssociation with the Risk of Gastric Cancer

Control, (%) Patient, (%) OR 95% CI P

MMP3 linked loci0 (no variant at any locus) 20.1 12.4 1 (Ref)1 (polymorphic variant allele at any one locus) 40.3 33.5 1.346 0.722–2.511 0.4282 (polymorphic variant allele at any two loci) 31.2 37.3 1.938 1.028–3.65 0.053 (polymorphic variant allele at all three loci) 8.4 16.8 3.21 1.41–7.31 0.005MMP3 haplotype0 (No polymorphic variant allele) 48.3 35.8 1 (Ref)1 (One polymorphic variant allele) 34.1 34.9 1.381 1.128–1.691 0.0022 (Two polymorphic variant allele) 16.2 24.9 2.074 1.63–2.638 <0.00013 (All three polymorphic variant allele) 1.4 4.4 4.24 2.288–7.858 <0.0001

OR, odds ratio; CI, confidence interval; Ref, reference linked loci or haplotype to calculate OR.Adjusted OR was calculated for age, sex, addiction, by binary logistic regression model using SPSS v16.0 software.P-value is for x2-test showing the significance of difference in the distributions of the variant alleles between patients and controls.Bold values indicate positive significance (P < 0.05).Haplotype frequency was obtained by using Haploview 4.2 software. ncontrol ¼ 154, npatient ¼ 209.

8 DEY ET AL.


Loading of serum proteins were visualized throughPonceau S staining of the blot.

DISCUSSION

Many studies have provided considerable evi-dence implicating multifaceted role of MMPsin the degradation of the ECM during cancerdevelopment and progression. Regulation of ECMhomeostasis is partially dependent on transcrip-tion of different MMPs [31,32]. MMP3 may play arole in both local invasiveness and metastaticspread; the later involves the ability of neoplasticcells to cross the basal membrane of both the epi-thelium and the vascular endothelium. In support,MMP3 over expression by some tumor types isconsistent with the above hypothesis [18,33].Deletion of an A allele (5A) at �1,612 bp of

MMP3 promoter has been reported to be associatedwith susceptibility for oral, lung, esophageal, headand neck, ovarian, colorectal, as well as breast car-cinomas [15,18,23–25,34–36]. 5A allele appears tobe less frequently identified in Asian populations(frequency range 7–20%) [18,37,38] than in Euro-pean populations (frequency range 40–50%)[18,37,38] and this is also true in our study. Theassociation of MMP3 �1612 5A/6A polymorphismwith breast cancer is controversial [18,25,39], anda negative finding has been documented for gastric

cardiac adenocarcinoma [18,25,39]. There was nofinding available regarding the role of MMP3 poly-morphisms in risk development of gastric cancerof the whole stomach. Thus, this study is a first ofits kind in any population including eastern Indi-an population on susceptibility of gastric cancerand MMP3 promoter polymorphisms.Several polymorphic genes, such as methylene

tetrahydrofolate reductase, NADPH: quinone oxi-doreductase and Cyclin D1 have been reported tomodify the risk of gastric cancer progression andmetastasis [40–42]. The present study supportsthe hypothesis that an adenine insertion/deletionpolymorphism at �1,612 bp of MMP3 promotermay modify the risk of development but not pro-gression and metastasis of gastric cancer. Althoughsome other polymorphisms in the MMP3 promotermay play different roles in the development andprogression of gastric cancer. The present studydocuments that 5A allele (5A6A and 5A5A þ 5A6Agenotype) of MMP3.1 polymorphism is associatedwith the risk of gastric cancer development in cur-rent or ex-tobacco addicted peoples, suggestingpossible altered MMP3 expression with the meta-bolic changes induced by tobacco. Vairaktaris et al.[35] also reported a significant difference in the 5Aallele at �1,612 bp of MMP3 promoter in oral can-cer patients with smoking addiction. This explana-tion is supported by a report from Yu et al. [43] on

Figure 2. In silico promoter binding assay of MMP3 promoter.Transcription factor binding sites were determined usingTFSEARCH, TESS, and Matinspector software. (A) Schematic repre-sentation of MMP3 promoter region showing the most commontranscription elements that binds to the promoter and the positionshowing the SNPs of our study. The promoter region is not inscale. (B) Predicted transcription elements with their bindingsequences that bind to the polymorphic region. NF-kB (p65), nucle-ar factor-kB p65 homodimer; NMP4, nuclear matrix protein 4; CIZ,Cas-interacting zinc finger protein; Oct-6, Octamer factor 6; en-1,

Homeobox protein engrailed; Oct-1, octamer factor 1; MEF-2,myogenic MADS factor MEF-2/Myocyte-specific enhancer factor 2;C/EBPb, CCAAT/enhancer binding protein beta; AML-1a, acutemyeloid leukemia-1a, runt factor. (inset) Expression of MMP3 inserum of controls (n ¼ 28) and gastric cancer patients (n ¼ 28)that were estimated from Western blotting analysis. Midlineindicating the mean arbitrary expression in controls and patients.Arbitrary expressions were measured by Image QUANT designeddensitometry values from the corresponding Western blots of con-trols and patients.



an additive interaction between the MMP2 pro-moter polymorphism and smoking in elevation ofthe risk of lung cancer. It is also noticeable thattobacco addiction itself is significantly associatedwith gastric cancer development of whole stomachin our study population. Moreover, males andhigh age group peoples were more prone to gastriccancer and the activity of MMPs might be regulat-ed by sex hormones [44]. Furthermore, expressionof MMPs was negatively regulated by androgen,whereas the secretion and activation of MMPs wasinhibited by physiological concentrations of pro-gesterone [45]. Our study indicated a positive cor-relation of 5A allele (5A6A and 5A5A þ 5A6Agenotype) in both �50 yr age group and maleswith the risk of gastric cancer (Table 3) suggestingthat, 5A allele escalates the susceptibility towardsmale population and the late onset of gastric can-cer. Our study opens the avenue to find out riskfactors for gastric cancer occurrence in males ofhigher age group, using MMP3.1 SNP as a geneticmarker. In addition, both the MMP3.2 GG geno-type and combination of GG plus AG genotype aswell as the G allele displayed close correlation withgastric cancer risk in an age, sex, and addiction(to tobacco) independent manner suggesting as agenetic marker for gastric cancer risk assessment.

Invasion of surrounding structures and lymphat-ic metastasis are the main factors influencing theprognosis and survival of upper gastrointestinal tu-mor patients [25]. It is conceivable that the highertranscriptional activity associated with the 5A al-lele may enhance tumor invasiveness. Although,we did not detect any significant difference be-tween genotypes of MMP3.1 in subgroups assignedto metastasis and TNM stages. This might be be-cause of the lower 5A allele frequency (0.149) inour population and support the relevance of host-dependent factors in the limitation of gastric can-cer invasiveness and metastasis. Herein, genotypeswith at least one MMP3.2 G allele (combined AGand GG genotype) increased the risk of lymphaticmetastasis in gastric cancer of whole stomach sug-gesting that patients carrying the G allele mayneed to be treated more actively by lymph noderesection. The late stages (Stage III and IV) gastriccancer conferred risk of metastasis to G allele (bothAG and GG genotype) suggesting that the G alleleexhibits a dominant effect to the phenotype andthat the GG genotype has the detrimental effectson gastric cancer progression. This is probably be-cause the G allele frequency is high in the patientpopulation (0.409) in our study. In addition, asso-ciation between the MMP3.2 SNP and the risk ofdevelopment and metastasis of gastric cancerfound to be consistent with the report of MMP3�1612 5A/6A polymorphism by Ghilardi et al.[18], that showed the higher frequency of the 5Aallele in breast cancer patients compared to

controls and that 5A homozygote had a 2.4-foldhigher risk of metastasis. We hypothesize, MMP3.2polymorphism may be used as a stratificationmarker in predicting lymphatic metastasis, distantmetastasis and progression of gastric cancer. A neg-ative correlation is observed between MMP3.3 GGgenotype and tumor size in gastric cancer suggest-ing a direction to protection against progression ofgastric cancer with this genotype.As it is known that cancer is a multigenic disease

where one variant SNP may only have a modestindependent effect on the disease phenotype, yetin aggregate, multiple biologically relevant SNPsmay provide a more accurate representation of therisk. Studies in the interleukin-6 and cholesteryl es-ter transfer protein genes have shown functionalcooperation/interaction between promoter poly-morphisms [46,47]. It is possible that functionalcooperation/interaction also occurs between thepolymorphisms in the MMP3 gene promoter.Individuals with combinations of MMP3.1 andMMP3.2 polymorphisms and MMP3.2 and MMP3.3polymorphisms might be at risk of gastric cancerdue to the combined effect of 5A allele of MMP3.1and G allele of MMP3.2 to increase the chances ofMMP3 over expression. Cooperation between theMMP3 polymorphic alleles is presumed to enhancethe transcriptional activity, enhancing gastric can-cer risk. Moreover, the analysis of the additiveeffect of all three loci of MMP3 polymorphismsshowed an indirect role of MMP3.3 polymorphismin gastric cancer risk that is not identified in a sep-arate analysis of MMP3.1 genotypes and underlinesthe importance of using the combination test. Alsoa gene-dosage analysis with increasing order of riskallele in the haplotypes is consistent with the ge-notype analysis. We conclude that combined effectof risk conferred by all three SNPs is higher thansingle one, showing a complex nature of geneticassociation for gastric cancer. Further understand-ing of these relationships should lead to thedevelopment of better prevention and treatmentstrategies.This is the first report showing MMP3.1 and

MMP3.2 polymorphisms are important for gastriccancer development. In support, bioinformaticsanalysis revealed that sequence variations at posi-tions �1612 and �375 might also have some possi-ble significant contributions to modulate promoteractivity. Though, these in silico findings for tran-scription factors binding to the polymorphic siteshave not been confirmed by biological experi-ments, alteration of promoter activity was indirect-ly confirmed by Western blot analysis of MMP3expression in serum samples. We found thatpatients showed higher expression levels (1.8-fold)of MMP3 as compared to controls. Further studieswill be required to determine the size of the haplo-type block and to search for other functional SNPs

10 DEY ET AL.


within it. Evidence that the MMP3.3 G alleleexhibited a recessive effect to the phenotype sug-gesting other alleles of ‘‘more active promoter’’might be required to develop the risk for the gas-tric cancer. The present study demonstrates thatgastric cancer patients carrying the more activeMMP3 promoter haplotype (5A-G-G) with increas-ing number of risk allele have more risk factorresulting in alteration in matrix remodeling in gas-tric tissues in gastric cancer. Genetic variations inthe MMP3 gene promoter may be part of a com-plex genetic risk profile and further studies arewarranted to elucidate the pathophysiological roleof this gene in gastric cancer and the functionaleffects of many of the SNPs, their linked loci andhaplotypes.

ACKNOWLEDGMENTS

Department of Gastric Surgery, Medical Collegeand Hospital, Kolkata, India and Institute of PostGraduate Medical Education and Research,Kolkata, India were kind enough to provide bloodsamples and patient’s documents, Prof. SiddharthaRoy, Director, IICB, Kolkata, India constantly sup-ported the work and authors are grateful to themfor their contributions. SD was recipient of SeniorResearch Fellowship from Indian Council ofMedical Research, India. KK and SS are recipientsof Junior Research Fellowship from Council ofScientific and Industrial Research (CSIR) andNational Tea Research Foundation, India, respec-tively. The work was supported by grant IAP001of CSIR, India, NBA07 of Department of Biotech-nology, India and CLP261 of NTRF, India.

REFERENCES

1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statis-tics, 2002. CA Cancer J Clin 2005;55:74–108.

2. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, ParkinDM. Estimates of worldwide burden of cancer in 2008:GLOBOCAN 2008. Int J Cancer 2010:17.

3. Forget MA, Desrosiers RR, Beliveau R. Physiological rolesof matrix metalloproteinases: Implications for tumorgrowth and metastasis. Can J Physiol Pharmacol 1999;77:465–480.

4. Kohn EC, Liotta LA. Molecular insights into cancer inva-sion: Strategies for prevention and intervention. CancerRes 1995;55(9):1856–1862.

5. Liotta LA, Stetler-Stevenson WG. Tumor invasion and me-tastasis: An imbalance of positive and negative regulation.Cancer Res 1991;51:5054s–5059s.

6. Lievre A, Milet J, Carayol J, et al. Genetic polymorphismsof MMP1, MMP3 and MMP7 gene promoter and risk ofcolorectal adenoma. BMC Cancer 2006;6:270.

7. Ye S, Eriksson P, Hamsten A, Kurkinen M, Humphries SE,Henney AM. Progression of coronary atherosclerosis is as-sociated with a common genetic variant of the humanstromelysin-1 promoter which results in reduced geneexpression. J Biol Chem 1996;271:13055–13060.

8. Su L, Zhou W, Asomaning K, et al. Genotypes and haplo-types of matrix metalloproteinase 1, 3 and 12 genes andthe risk of lung cancer. Carcinogenesis 2006;27:1024–1029.

9. Impola U, Uitto VJ, Hietanen J, et al. Differential expres-sion of matrilysin-1 (MMP-7), 92 kD gelatinase (MMP-9),and metalloelastase (MMP-12) in oral verrucous and squa-mous cell cancer. J Pathol 2004;202:14–22.

10. Johansson N, Ahonen M, Kahari VM. Matrix metalloprotei-nases in tumor invasion. Cell Mol Life Sci 2000;57:5–15.

11. Kahari VM, Saarialho-Kere U. Matrix metalloproteinasesand their inhibitors in tumour growth and invasion. AnnMed 1999;31:34–45.

12. Sirum-Connolly K, Brinckerhoff CE. Interleukin-1 or phor-bol induction of the stromelysin promoter requires an ele-ment that cooperates with AP-1. Nucleic Acids Res 1991;19:335–341.

13. Buttice G, Quinones S, Kurkinen M. The AP-1 site is re-quired for basal expression but is not necessary for TPA-response of the human stromelysin gene. Nucleic AcidsRes 1991;19:3723–3731.

14. Quinones S, Buttice G, Kurkinen M. Promoter elements inthe transcriptional activation of the human stromelysin-1gene by the inflammatory cytokine, interleukin 1. BiochemJ 1994;302:471–477.

15. Fang S, Jin X, Wang R, et al. Polymorphisms in the MMP1and MMP3 promoter and non-small cell lung carcinomain North China. Carcinogenesis 2005;26:481–486.

16. Zinzindohoue F, Blons H, Hans S, et al. Single nucleotidepolymorphisms in MMP1 and MMP3 gene promotersas risk factor in head and neck squamous cell carcinoma.Anticancer Res 2004;24:2021–2026.

17. Hashimoto T, Uchida K, Okayama N, et al. Associationof matrix metalloproteinase (MMP)-1 promoter polymor-phism with head and neck squamous cell carcinoma. Can-cer Lett 2004;211:19–24.

18. Ghilardi G, Biondi ML, Caputo M, et al. A single nucleo-tide polymorphism in the matrix metalloproteinase-3 pro-moter enhances breast cancer susceptibility. Clin CancerRes 2002;8:3820–3823.

19. Oceandy D, Yusoff R, Baudoin FM, Neyses L, Ray SG.Promoter polymorphism of the matrix metalloproteinase 3gene is associated with regurgitation and left ventricularremodelling in mitral valve prolapse patients. Eur J HeartFail 2007;9:1010–1017.

20. Spurr NK, Gough AC, Gosden J, et al. Restriction frag-ment length polymorphism analysis and assignmentof the metalloproteinases stromelysin and collagenase tothe long arm of chromosome 11. Genomics 1988;2:119–127.

21. Ye S, Watts GF, Mandalia S, Humphries SE, Henney AM.,Preliminary report: Genetic variation in the human strome-lysin promoter is associated with progression of coronaryatherosclerosis. Br Heart J 1995;73:209–215.

22. Woo M, Park K, Nam J, Kim JC. Clinical implications ofmatrix metalloproteinase-1, -3, -7, -9, -12, and plasmino-gen activator inhibitor-1 gene polymorphisms in colorectalcancer. J Gastroenterol Hepatol 2007;22:1064–1070.

23. Smolarz B, Szyllo K, Romanowicz-Makowska H, Niewia-domski M, Kozlowska E, Kulig A. PCR analysis of matrixmetalloproteinase 3 (MMP-3) gene promoter polymor-phism in ovarian cancer. Pol J Pathol 2003;54:233–238.

24. Blons H, Gad S, Zinzindohoue F, et al. Matrix metallopro-teinase 3 polymorphism: A predictive factor of responseto neoadjuvant chemotherapy in head and neck squa-mous cell carcinoma. Clin Cancer Res 2004;10:2594–2599.

25. Zhang J, Jin X, Fang S, et al. The functional SNP in thematrix metalloproteinase-3 promoter modifies susceptibili-ty and lymphatic metastasis in esophageal squamous cellcarcinoma but not in gastric cardiac adenocarcinoma.Carcinogenesis 2004;25:2519–2524.

26. Beyzade S, Zhang S, Wong YK, Day IN, Eriksson P, Ye S.Influences of matrix metalloproteinase-3 gene variation onextent of coronary atherosclerosis and risk of myocardialinfarction. J Am Coll Cardiol 2003;41:2130–2137.



27. Green FL, Balch CM, Fleming ID, et al. Green FL, editor.AJCC cancer staging manual. 6th edition, Chapter10:Stomach. New York: Springer; 2002.

28. Paul S, Bhattacharya P, Das Mahapatra P, Swarnakar S.Melatonin protects against endometriosis via regulation ofmatrix metalloproteinase-3 and an apoptotic pathway.J Pineal Res 49:156–168.

29. Purcell S, Neale B, Todd-Brown K, et al. PLINK: A tool setfor whole-genome association and population-based link-age analyses. Am J Hum Genet 2007;81:559–575.

30. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: Analysisand visualization of LD and haplotype maps. Bioinformat-ics 2005;21:263–265.

31. Zucker S, Vacirca J. Role of matrix metalloproteinases(MMPs) in colorectal cancer. Cancer Metastasis Rev 2004;23:101–117.

32. Baruch RR, Melinscak H, Lo J, Liu Y, Yeung O, HurtaRA. Altered matrix metalloproteinase expression asso-ciated with oncogene-mediated cellular transformationand metastasis formation. Cell Biol Int 2001;25:411–420.

33. Chambers AF, Matrisian LM. Changing views of the roleof matrix metalloproteinases in metastasis. J Natl CancerInst 1997;89:1260–1270.

34. Tu HF, Liu CJ, Chang CS, et al. The functional (�1171 5A/6A) polymorphisms of matrix metalloproteinase 3 geneas a risk factor for oral submucous fibrosis among maleareca users. J Oral Pathol Med 2006;35:99–103.

35. Vairaktaris E, Yapijakis C, Vasiliou S, et al. Association of�1171 promoter polymorphism of matrix metalloprotei-nase-3 with increased risk for oral cancer. Anticancer Res2007;27:4095–4100.

36. Hinoda Y, Okayama N, Takano N, et al. Association offunctional polymorphisms of matrix metalloproteinase(MMP)-1 and MMP-3 genes with colorectal cancer. Int JCancer 2002;102:526–529.

37. Chaudhary AK, Singh M, Bharti AC, Shukla S, Singh AK,Mehrotra R. Synergistic effect of stromelysin-1 (matrixmetalloproteinase-3) promoter (�1171 5A->6A) polymor-phism in oral submucous fibrosis and head and necklesions. BMC Cancer 2010;10:369.

38. Tsukahara S, Shinozaki M, Ikari K, et al. Effect of matrixmetalloproteinase-3 functional SNP on serum matrix met-alloproteinase-3 level and outcome measures in JapaneseRA patients. Rheumatology (Oxford) 2008;47:41–444.

39. Lei H, Zaloudik J, Vorechovsky I. Lack of association of the�1171 (5A) allele of the MMP3 promoter with breast can-cer. Clin Chem 2002;48:798–799.

40. Miao X, Xing D, Tan W, Qi J, Lu W, Lin D. Susceptibility togastric cardia adenocarcinoma and genetic polymorphismsin methylenetetrahydrofolate reductase in an at-riskChinese population. Cancer Epidemiol Biomarkers Prev2002;11:1454–1458.

41. Zhang JH, Li Y, Wang R, et al. NQO1 C609T polymor-phism associated with esophageal cancer and gastric car-diac carcinoma in North China. World J Gastroenterol2003;9:1390–1393.

42. Zhang J, Li Y, Wang R, et al. Association of cyclin D1(G870A) polymorphism with susceptibility to esophagealand gastric cardiac carcinoma in a northern Chinese pop-ulation. Int J Cancer 2003;105:281–284.

43. Yu C, Pan K, Xing D, et al. Correlation between a singlenucleotide polymorphism in the matrix metalloproteinase-2 promoter and risk of lung cancer. Cancer Res 2002;62:6430–6433.

44. Marbaix E, Donnez J, Courtoy PJ, Eeckhout Y. Progester-one regulates the activity of collagenase and related gelat-inases A and B in human endometrial explants. Proc NatlAcad Sci USA 1992;89:11789–11793.

45. Schneikert J, Peterziel H, Defossez PA, Klocker H, LaunoitY, Cato AC., Androgen receptor-Ets protein interaction isa novel mechanism for steroid hormone-mediated down-modulation of matrix metalloproteinase expression. J BiolChem 1996;271(39):23907–23913.

46. Terry CF, Loukaci V, Green FR., Cooperative influence ofgenetic polymorphisms on interleukin 6 transcriptionalregulation. J Biol Chem 2000;275:18138–18144.

47. Frisdal E, Klerkx AH, Le Goff W, et al. Functional inter-action between �629C/A, �971G/A and �1337C/T poly-morphisms in the CETP gene is a major determinant ofpromoter activity and plasma CETP concentration in theREGRESS Study. Hum Mol Genet 2005;14:2607–2618.

12 DEY ET AL.


matrix metalloproteinase3 gene promoter polymorphisms and their haplotypes are associated with...

Documents