DNMT 3 a rs 1550117 Polymorphism Association with Increased Risk of Helicobacter pylori Infection

Gastric cancer (GC) is a major health problem. In 2011, 989,600 new cases were predicted and more than 738,000 deaths worldwide. Half of these cases and deaths were estimated to occur in China (Jemal et al., 2011; Bray et al., 2012). Helicobacter pylori (H.pylori) infection is the main cause of gastric atrophy (GA), gastroduodenal ulcer disease, gastric adenocarcinoma and mucosa-associated lymphoid tissue (MALT) lymphoma (Lehours et al., 2003; Chuah et al., 2011). H.pylori infection was found in half of the population of the world, and the risk for non-cardiac GC in H.pylori infected person was estimated to be 3-20 fold or even higher (Kamangar et al., 2006; Malfertheiner et al., 2012). However, not all infected subjects eventually develop GA and only a small numbers develop to GC. Factors that influence the risks for atrophy and cancer in the presence of infection may be related to the characteristics of the host. Various studies show that inherited cancer susceptibility genes and SNPs are correlated with individual’s risk of developing GC (Saeki et al., 2013).


Introduction
Gastric cancer (GC) is a major health problem.In 2011, 989,600 new cases were predicted and more than 738,000 deaths worldwide.Half of these cases and deaths were estimated to occur in China (Jemal et al., 2011;Bray et al., 2012).Helicobacter pylori (H.pylori) infection is the main cause of gastric atrophy (GA), gastroduodenal ulcer disease, gastric adenocarcinoma and mucosa-associated lymphoid tissue (MALT) lymphoma (Lehours et al., 2003;Chuah et al., 2011).H.pylori infection was found in half of the population of the world, and the risk for non-cardiac GC in H.pylori infected person was estimated to be 3-20 fold or even higher (Kamangar et al., 2006;Malfertheiner et al., 2012).However, not all infected subjects eventually develop GA and only a small numbers develop to GC. Factors that influence the risks for atrophy and cancer in the presence of infection may be related to the characteristics of the host.Various studies show that inherited cancer susceptibility genes and SNPs are correlated with individual's risk of developing GC (Saeki et al., 2013).

DNMT3a rs1550117 Polymorphism Association with Increased Risk of Helicobacter pylori Infection
Xue-Yuan Cao1 , Zhi-Fang Jia2 , Dong-Hui Cao 2 , Fei Kong 2 , Mei-Shan Jin3 , Jian Suo 1 , Jing Jiang 2 * Changes in epigenetic marks such as DNA methylation is associated with several diseases, including cancer, metabolic disorders, and various reproductive conditions (Cortessis et al., 2012).CpG island hypermethylation and genomic DNA hypomethylation are found in gastric tumors and premalignant lesions (Park et al., 2009), and H.pylori infection can induces aberrant CpG island hypermethylation in gastric mucosa (Nakajima et al., 2009;Park et al., 2009).Several reports indicated that accumulation of aberrant methylation in gastric mucosa produces a field for cancerization, and methylation levels correlate with gastric cancer risk (Guo et al., 2011;Wang et al., 2012).However, few studies have investigated the host factors, such as SNPs of DNA methyltransferases (DNMTs) gene, which may affect epigenetic regulation in gastric mucosa (Fan et al., 2010;Wu et al., 2012;Yang et al., 2012).
In our previous study among Chinese, subjects with rs10420321 GG and rs8111085 CC genotype of the DNMT1 gene were associated with reduced risks for H.pylori infection and higher risks for GA, supporting the hypothesis that polymorphisms of DNMTs may 5714 influence the enzyme activities to affect the H.pylori infection and carcinogenesis (Jiang et al., 2012).DNA methyltransferase-3a (DNMT3a) also plays a significant part in the development of embryogenesis and in the generation of aberrant methylation in carcinogenesis.The DNMT3a gene is located on chromosome 2p23 with a total size of 109.6 kb and is constituted of 26 exons (Weisenberger et al., 2002).Mutations in coding regions of the DNMT3a have been reported in various malignant hematological disorders, such as a missense mutation in codon 882 leading to an amino acid exchange (Kim et al., 2013).Also, allelic loss of DNMT3a has been frequently observed in many common solid cancers, including GC, lymphomas and lung cancers (Kim et al., 2013).Several studies have found that DNMT3a is over-expressed in human cancers, suggesting that it may be involved in tumorgenesis and tumor progression (Ding et al., 2008;He et al., 2011;Yang et al., 2011).Recently, DNMT3a SNPs rs1550117 and rs13420827 were also found to be significantly associated with GC (Fan et al., 2010;Wu et al., 2012;Yang et al., 2012).The aim of this study is to confirm the formerly reported association between the DNMT3a polymorphisms and GC in a large number of Chinese subjects, as well as to examine the association with H.pylori infection and GA.

Study populations
Four hundred and forty seven GC (non-cardiac GC) cases were selected from the First Hospital of Jilin University, between 2008 and 2010.All patients underwent tumor resections with histological confirmed gastric adenocarcinoma.Individuals with GA and healthy controls were recruited from health examinees in the health check-up centre of the same hospital from 2009 to 2010.In brief, a total of 1111 individuals without cancer history (654 males and 457 females, ages of 35 to 80 years old) participated in the study.The examinees were Han descent from the area of Changchun.150 subjects were found to have GA by serum PG examination and 111 of them were confirmed by biopsy and histopathological examinations; 39 subjects were excluded from the study as they were either rejected endoscopic examination (22 cases) or diagnosed negative for GA by endoscopic examination and pathological examinations (17 cases).The remaining health examinees (961) were included in the control group.The inclusion criteria for the control group were: negative by serum PG screening.Subjects with a history of gastric disease, ulcer, and history of cancers were not considered candidates for the control.Informed consent was obtained from all patients and the study protocol was approved by the ethics committee of the First Hospital of Jilin University.

Tests for H.pylori infection and diagnosis of GA
Serum immunoglobulin (Ig) G antibodies to H.pylori were detected using a kit for H.pylori -Ig G enzyme-linked immunosorbent assay (ELISA) (Biohit, Helsink, Finland).The antibody titers were quantified by optical density (OD) readings according to the manufacturer's protocol and titres higher than the threshold value of 30EIU were considered as positive for H.pylori infection.Levels of PepsinogenⅠ(PG Ι) and II in serum were measured using ELISA kits (Biohit, Helsink, Finland).For gastric atrophy screening, cut-off points used in this study were < 82.3 ng/ml for PG Ι and <6.05 for PGΙ/PG II ratio, as they had previously been validated against histological confirmatory studies for GA in Chinese population (Cao et al., 2007).The quality control samples in kits showed coefficients of variation of 4.5%, 4.3% and 4.7% for H.pylori, PG Ι and PG II, respectively.111subject of GA suspected cases, based on the serum screening, were confirmed by endoscopy, biopsy and histological examinations for final diagnosis.

Genotyping
Genomic DNA from whole blood sample was extracted with an AxyPrep blood Genomic DNA extraction Kit (AP-MN-BL-GDNA-250, Axygen Biosciences, Union City, USA).Polymerase chain reactions (PCR) were carried out in a 5 ul per reaction on the genomic DNA (10 ng) using a TaqMan universal PCR master mix (Applied Biosystems).Forward, reverse primers, FAM and VIC labeled probes were designed by Applied Biosystems (ABI Assay-by-Designs).Sequences of primers and probes are available on request.Amplification conditions on BIO-RAD S1000TM thermal cyclers (Bio-Rad Laboratories, Hercules, California) were as follows: 1 cycle of 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min.The PCR products were genotyped on an ABI PRISM 7900HT Sequence Detector in end-point mode using the Allelic Discrimination Sequence Detector Software V2.3 (Applied Biosystems).For the software to recognize genotypes, two non-template controls were included in each of 384-well plates.All samples were arrayed together in four 384-well plates, and the fifth plate contained eight duplicate samples from each of four plates to ensure the quality of genotyping (the concordance was >99% for all SNPs).

Immunohistochemistry
Tissue microarray (TMA) recipient blocks were constructed containing paraffin-embedded cancer tissues from 89 archival patient specimens by a tissue Array (Minicore; Alphelys Impasse Paul Langevin, Plaisir, France).The TMA blocks were cut into 4 µm thick continuous sections and mounted on poly-L-lysinecoated glass slides.Slides were deparaffinized in xylene, rehydrated in graded alcohols and washed in tap water.Prior to a primary antibody staining, the slide was pretreated with ethylenediaminetetraacetic acid buffer in a pressure cooker for antigen retrieval.Endogenous peroxidase activity was quenched by 3% H 2 O 2 blocking reagent for 10 min.The slide was incubated with DNMT3a antibody (H-295, sc-20703, Santa Cruz, CA, USA; dilution 1:200) at 4 °C overnight, and then immunostained with the avidin-biotin peroxidase complex (DAKO, CA, USA).Finally, the slide was stained with diaminobenzidine according to the manufacturer's protocol (DAKO, CA, USA).The slide was rinsed three times with phosphate buffered saline after each step of staining.The stained slides were evaluated by two independent pathologists who were blinded for clinical data.The widely accepted HSCORE system that considered both the staining intensity and the percentage of cells stained at the specific range of intensity was adapted.The HSCORE was calculated following the equation: HSCORE=∑Pi(i) i=0, 1, 2, 3, Pi= 0~100%, where i is the intensity of the stained tumor cells(no staining=0, weak staining=1, moderate staining=2 and strong staining=3) and Pi is the percentage of stained tumor cells for each intensity varying from 0 to 100%.The HSCORE ranged from 0 to 300.

Statistics analysis
Deviations of genotype frequencies in controls from those expected under the Hardy-Weinberg equilibrium were assessed by a goodness-of-fit x 2 -test.Linkage disequilibrium (LD) between pairs of biallelic loci was determined using two measures, D´ and r 2 .Either Chi-square test or Fisher's exact test was performed by comparing distributions of genotype frequencies between patients and controls.Risks associated with rare genotypes were estimated as odds ratios (ORs).Corresponding 95% confidence intervals (CIs) were adjusted by age (scale variable), sex (nominal variable) and H.pylori antibody (nominal variable) by unconditional logistic regression.The association of DNMT3a expression with polymorphisms was analyzed by the Kruskal-Wallis H test and Fisher's exact test.All statistical tests were twotailed and P values less than 0.05 were considered to be statistically significant.All analyses were performed using statistical software for windows SAS version 9.2 (SAS Institute, Cray, NC, USA).The power of the statistical tests was calculated using the QUANTO Version 1.2.3 software program (Jim Gauderman and John Morrison, University of Southern California, CA, USA).

Allele frequencies of the SNPs
A total of 447 patients with GC (322 males and 125 females, aged between 35 to 80 years old), 111 subjects with GA and 961 healthy subjects who passed health checks were recruited in this study.The characteristics of subjects are summarized in Table 1.The mean age was older in gastric cancer patients compared to the control group (61.6 vs. 50.6years; P < 0.001).There were more females in the control group (P <0.001).Prevalence rates of H.pylori seropositivity were significantly higher in the GC and GA groups compared to the control group (69.1%, 75.7% vs. 49.7%,P<0.001).Genotype distributions for rs1550117 and rs13420827 in the control group were consistent with the Hardy-Weinberg equilibrium (P = 0.66 and P = 0.51, respectively).No statistically significant differences were found between groups for the SNPs (P =0.59 and P =0.65, respectively).The linkage disequilibrium structure of these two polymorphic loci suggested that rs1550117 had very lower D' and r 2 with rs13420827 (0.07 and 0.14, respectively), therefore it was not necessary to assess the DNMT3a haplotypes.

Associations of SNPs with H.pylori seropositivity, GA and GC
Rates of H.pylori seropositivity were statistically significantly increased in subjects bearing with rs1550117 AA genotype compared to GG/AG genotypes (66.7% vs. 49.1%,P value=0.048).After adjusted by age and sex, the ORs for H.pylori seropositivity of AA genotype were 2.08 (95%CI: 1.02-4.32)relative to the GG/AG genotypes, showing a recessive effect of the rare allele.No significant association was found between rs13420827 polymorphism and the H.pylori seropositivity (Table 2).
The age-, sex-and H.pylori antibody-adjusted ORs for GA were calculated, however, no association between two and the risk of GA was identified using unconditional logistic analysis (Table 3).Relationships of the SNPs with GC were also examined, but no significant correlation was found either (Table 3).The results of subgroup analysis did not suggest that sex, age and H.pylori status were effect modifier of the association between GA or GC and DNMT3a polymorphisms (data not shown).No relation was found between the SNPs and Lauren's classification, differentiation or TNM staging (data not shown).

Association between SNPs and expression of DNMT3a
The results of immunohistochemstry staining are shown in Table The strong nuclear expression of DNMT3a was shown in most of slides and a faint cytoplasm staining was only seen in a few slides.The DNMT3a expression is graded into four categories, high expression (HSCORE>200), moderate expression (100<HSCORE≤200), low expression (HSCORE≤100) and negative (HSCORE=0).The grades of high, moderate, low and negative expressions were detected in 53 (59.5%), 18 (20.2%),6 (6.7%) and 12 (13.5%)gastric cancer tissues respectively.No differences of DNMT3a expression were seen between genotypes of rs1550117 and rs13420827 (P values were 0.09 and 0.16 respectively).To gain an insight of aberrant DNMT3a expression induced by H.pylori infection, we analyzed the expression level of DNMT3a in patients with and without H.pylori infection (62 vs. 27patients).No increase in the DNMT3a expression was observed in the patients with H.pylori infection (P value = 0.44).

Discussion
H.pylori infection has been generally accepted as an important factor in gastric carcinogenisis.Previous studies have shown that H.pylori infection could induce aberrant DNA methylation at the CpG islands, resulting in inactivations of tumor suppressor genes in gastric mucosa, and creating predisposed fields for cancerization (Maekita et al., 2006;Nakajima et al., 2006;Tsang et al, 2011).In contrast, a study reported that H.pylori infection did not induce either mRNA or protein expression of DNMT1, DNMT3a and DNMT3B in gastric mucosa directly (Nakajima et al., 2009).It has been hypothesized that SNPs of DNMTs may affect the methyltransferases levels of expression, involving in tumorgenesis and tumor progression.
Several SNPs of the DNMT3a gene have been identified, which may affect catalytic activity of the DNMT3a enzyme.For example, the rare A allele of the rs1550117 polymorphism in promoter was found to significantly increase DNTM3a transcriptional activity and was associated with an increased risk for GC and colorectal cancer (Fan et al., 2010;Zhao et al., 2012).Moreover, subjects carrying the CG heterozygosity of rs13420827 in 3'UTR was reported with a decreased risk for GC (Yang et al., 2012).However, we did not find any associations between DNMT3a polymorphisms and risk for GA or GC, but the subjects bearing rs1550117 AA genotype had significantly increased a risk of H.pylori infection in our study.
So far the relation between the variants of DNMT3a and H.pylori infection has not been investigated.A previous work has shown that MUC1, a membrane bound mucin expressed on the surface of gastric epithelial cells, provides a protective barrier against H.pylori infection.MUC1 gene expression is regulated by DNA methylation, for instance, MUC1 mRNA expression in MUC1-negative cells was restored by the treatment with  (Yamada et al., 2008).The luciferase assay demonstrated that the promoter activity of rs1550117 A allele of DNMT3a was significantly higher (more than fold) than the G allele in Fan's study (Fan et al., 2010).Therefore, we can propose that DNMT3a rs1550117 GgA transition might increase the expression of DNMT3a and induce transcriptional repression of MUC1, resulting in the attenuation of MUC1 protective function.Further studies of DNMT3a sequence variants and their biologic functions may shed light on biological importance of the DNMT3a polymorphism in H.pylori infection.Fan et al. (2010) recently found that rs36012910 polymorphism, which located at -2720 bp from the transcription start site in promoter associated with genetic susceptibility to GC (Fan et al., 2010;Wu et al., 2012).However they did not directly confirm that the rs36012910 polymorphism affects the activity of the DNMT3a promoter.The relationship between rs36012910 A>G polymorphism and risk of GC, may be due to the linkage disequilibrium with functional polymorphism rs1550117.In Yang's and our studies, no any relationships between rs1550117 with susceptibilities to GC were observed (Yang et al., 2012).Other studies also failed to find significant associations between tag SNPs of DNMT3a and breast cancer risk, or ovarian carcinoma (Mavaddat et al., 2009;Kelemen et al., 2010).Although Yang et al examined the associations of 5 SNPs from DNMT3a (rs1550117, rs11887120, rs13420827, rs13428812, rs6733301) with GC in the Southern Chinese population, there was a significant protective effect between GC heterozygosity in rs13420827 and GC in the overdominant model, with OR = 0.66 (0.45-0.97) (Yang et al., 2012).We were unable to find the similar observed association between rs13420827 with risk of GC.The inconsistent results from these studies may be caused by different selection of study subjects and environmental backgrounds, such as age, smoke, and H.pylori infection group.In our study, GC cases were non-cardiac GC patients, and in Yang's studies, GC cases include non-cardiac GC and cardiac GC patients.It has been long thought that differences of etiology between non-cardiac GC and cardiac GC.In addition, we have selected cancer-free healthy examinees as healthy controls, and Yang et al selected cancer-free outpatients and inpatients as hospital controls, thus, the discordance of results may result from sampling.Furthermore, in our study, H.pylori infection as an important influencing factor for GC was adjusted and stratified in the analysis, which may have led to preferably explore the association between the DNMT3a polymorphisms and risk of GC.The statistical power is sufficient (power > 0.99) to examine the OR > 2.0 association between SNPs and risk of GC in the present study.
In addition, the over-expression of DNMT3a mRNA and protein were detected in various cancers including GC, these results revealed that increased DNMT3a expressiom may play a role in tumorgenesis of GC (Oh et al., 2007;Ding et al., 2008;Rahman et al., 2010;He et al., 2011;Yang et al., 2011).However, no differences of DNMT3a expression was detected between different genotype carriers in GC patients in our study.DNMT3a mutations and allelic losses have been reported in many solid cancers (Kim et al., 2013), which decreased the emzymatic activity of encoded protein, suggesting that the abnormal expression or accumulation of DNMT3a in cancer tissues may be due to defects in the degradation of mutant products and no relation with the polymorphisms.
In summary, we did not find significant associations for DNMT3a rs1550117 and rs13420827 polymorphisms with GA or GC, but our finding provides the first evidence of association between rs1550117 and increased risks for H.pylori infection in the Chinese population.Further investigations are required to fully understand the biological importance of these polymorphisms.