Polymorphisms of Two Fucosyltransferase Genes (Lewis and Secretor Genes) Involving Type I Lewis Antigens Are Associated with the Presence of Anti-Helicobacter pylori IgG Antibody¹

Helicobacter pylori infection is known to be linked with duodenal and gastric ulcers (1), gastric adenocarcinomas of the distal stomach (2), and low-grade B-cell lymphoma of mucosa-associated lymphoid tissue (3), and H. pylori is classified as a definite carcinogen for gastric cancer by the International Agency for Research on Cancer (4). Using an animal experiment model, we have also proved an association between H. pylori infection and gastric cancer in that H. pylori infection enhanced glandular stomach carcinogenesis (5), and the eradication of H. pylori diminished these enhancing effects (6).

Recent progress in the molecular analysis of H. pylori infection has clearly revealed that the bacteria attach to the gastric mucosa with the blood group antigen-binding adhesin, BabA (7, 8). A clinical relevance has been shown for the babA2 gene encoding BabA adhesin with regard to H. pylori-related diseases (9). BabA binds to both Le^b 3 [Gal (α1.2Fuc)β1.3GlcNAc(α1.4Fuc)-R] and H type I blood group carbohydrate structures [H type I structures; Gal (α1.2Fuc) β1.3GlcNAc-R] expressed on the foveolar epithelium of the gastric mucosa (Fig. 1; Ref. 7).

We have been studying Le blood type antigens using biochemical and molecular biological methods and have obtained the following results concerning type I Le antigen synthesis from a series of previous studies. Individuals homozygous for nonfunctional alleles of Le gene (le/le) fail to express type I Le antigen in the entire body (Le negative; Refs. 10, 11). In the human fucosyltransferase family, only Le enzyme (FUT3, Fuc-T III) exhibits fucose transfer activity toward a type I precursor (Galβ1.3GlcNAc-R) or H type I structure with α1.4 linkage (Fig. 1; Refs. 10, 12, 13). Se enzyme (FUT2, Fuc-T II) exhibits fucose transfer activity toward the type I precursor with α1.2 linkage (Fig. 1) and is responsible for Le^b expression on erythrocytes, solely determines the secretor status, and makes a marked contribution to Le^b expression in colorectal tissues (10, 14, 15, 16). Individuals homozygous for nonfunctional alleles of the Se gene (se/se) fail to express ABH blood antigens in secreted fluids (nonsecretors; Refs. 14, 17), whereas those very rare individuals homozygous for nonfunctional alleles of the H gene (h/h) fail to express ABH blood antigens on erythrocytes (18, 19). The type I Le antigen synthetic pathway in secretor tissues is shown in Fig. 1.

Since the first report of nonfunctional H alleles in Caucasians (18), some ethnic-specific nonfunctional point mutations have been reported not only in the H gene but also in the Se and Le genes. In the Se gene in the Japanese population (14, 20), we found the sej (also called se2) allele with an A385T (Ile¹²⁹ to Phe) missense mutation with 40% frequency. This mutation results in much lower α1.2-fucosyltransferase activity (14). Koda et al. (20) reported other inactive alleles in the Japanese, i.e., se3 (C571T; Arg¹⁹¹ to terminal codon), se4 (C628T; Arg²¹⁰ to terminal codon), and se5, attributable to recombinant fusion between the Se gene and the Sec1 pseudogene. However, our recent population study indicated the frequencies of sej and se5 alleles to be 40.6 and 4.7%, respectively, whereas se1, se3, and se4 alleles were not found in any of 600 Japanese analyzed (21). Therefore, sej/sej, sej/se5, and se5/se5 individuals can be regarded essentially as the nonsecretors in Japan. On the other hand, we also found only three missense mutated Le alleles, le1 (T59G; Leu²⁰ to Arg and G580A; Gly¹⁷⁰ to Ser; Ref. 20), le2 (T59G and T1067A; Ile³⁵⁶ to Lys), and le3 (T59G only), in the Japanese population (10, 15). The le3 allele exhibited as much α1.3/4-fucosyltransferase activity as the wild-type allele in vitro, whereas both le1 and le2 showed very low activity compared with the wild-type Le allele (15). As a result, le1/le1, le1/le2, and le2/le2 individuals can be regarded essentially as the Le-negative individuals in Japan.

In a previous biochemical analysis, we demonstrated that the amount of Le^b antigen was affected by the number of Se alleles (16). In our population study, we proved that Se genotypes affected the serum level of sialyl-Le^a antigen (CA19.9; NeuAcα2.3 Galβ1.3GlcNAc-R) in individuals without malignancy attributable to the competition between Se enzyme and Galβ1.3GlcNAc α2.3-sialyltransferases (21). Considering the type I Le antigen synthetic pathway, it is possible that the type I precursor structure for acceptor substrate is used by Se and Le enzymes, with some competition between the two. Therefore, we expected that both Se and Le gene polymorphism would affect the susceptibility to H. pylori infection because of the impact of these polymorphisms on the synthesis of adhesin ligands. An increased number of H. pylori causes an inflammatory reaction in gastric mucosa through the up-regulation of IL-8 by nuclear factor-κB activation (22) via an intracellular signal pathway involving IKK, NIK, TRAF2, and TRAF6 (23). This may be reflected in the augmentation of anti-H. pylori antibody in patient serum. Therefore, we tested the presence of anti-H. pylori IgG antibody for the purpose of evaluating the association between Le blood genotype and H. pylori infection.

In 1999, we began an H. pylori eradication intervention study to assess H. pylori eradication therapy. In the present study, we report the correlation between the Le and Se genotypes and H. pylori infection, indicated by the augmentation of anti-H. pylori IgG antibody, in the Japanese population.

We began an intervention study in 1999 to eradicate H. pylori. Participants were from 40 to 69 years of age with no history of gastrectomy. All gave informed consent for the use of their clinical pathological specimens for this research prior to undergoing gastroscopy at Aichi Cancer Center Hospital. Participants found to have a malignancy were excluded from the present study. DNA and serum samples from 239 participants were examined for H. pylori infection and the determination of the genotypes. Of the 239 participants, 95 (39.7%) said they were taking medication for 105 diseases or conditions, including 23 cases of gastric/duodenal ulcer, 23 of so-called gastritis, 16 of hypertension, 7 of diabetes mellitus, 7 of pain including arthritis and lumbago, 6 of hyperlipidemia, and 23 other miscellaneous conditions. The remaining 60.3% were disease free. We determined the Se and Le genotypes of another 150 DNA samples from gastric cancer patients who had been diagnosed and had undergone total gastrectomy in the past at Aichi Cancer Center Hospital, after their written informed consent to genotyping was obtained. The patients with early gastric cancer and long-term survivors were included. Sixty surgically resected normal gastric mucosa specimens from these patients were available for evaluating immunohistochemically whether type I Le antigen expression was dependent on the Le and Se genotypes. Because the stomach cancer patients were prevalent cases, they were not appropriate for examining the genotypes for stomach cancer risk but were applicable for the study of association between phenotype and genotype.

Because the se3 and se4 alleles were not found in our previous analysis of >600 Japanese samples (21), only A385T mutations in the sej and the se5 alleles were assessed. The Le genotype was assigned based on detection of three missense mutations, T59G, G508A, and T1067A, in the Le gene (10). Aliquots of 7 ml of peripheral blood were obtained from the participants with 2Na-EDTA, and the buffy coat was separated to extract genomic DNA using a QIAamp DNA Blood Mini kit (Qiagen, Inc., Valencia, CA). The detection method for sej, le1, le2, and le3 alleles was based on the PCR-RFLP described in detail previously (21). In brief, the full-length open reading frame of the Se or Le gene was amplified with a specific primer set and subjected to the second PCR reaction for the PCR-RFLP. The second PCR products were digested with AluI to detect A385T mutations of the sej allele, with PvuII to G508A of le1, with HindIII to T1067A of le2, and with MspI to T59G of le1, le2, and le3. The se5 allele was detected by simple PCR with a specific primer set as reported previously (20, 21). The specifically amplified DNA fragment was subcloned in pCR II vector (TA Cloning kit; Invitrogen, Carlsbad, CA), and the nucleotide sequence was determined with an ABI 310 genetic analyzer.

Two-ml serum samples were subjected to testing for anti-H. pylori IgG antibodies with the high molecular weight campylobacter-associated protein (Enteric Products, Inc., Stony Brook, NY) ELISA conducted by SRL Co., Ltd. (Tokyo, Japan). Cases with ELISA values >2.2 were defined as positive for H. pylori infection, in accordance with global standards as in the suppliers’ manual. Many epidemiological studies in countries including Japan have adopted this method as a standard tool for detecting individuals infected with H. pylori (2, 24, 25, 26, 27, 28, 29). The causal relationship between stomach cancer and H. pylori has been established largely depending on epidemiological studies using the antibody rather than biopsy samples (2, 24, 25, 26, 28, 29).

We performed immunohistochemical analysis with monoclonal antibodies, anti-Le^a antigen (7LE; Seikagaku Co., Tokyo, Japan), and anti-Le^b antigen (BG-6; Signet Pathology Systems, Inc., Dedham, MA; Ref. 9). Sections were cut at 4 μm, deparaffinized with xylene, and rehydrated through ethanol. The sections were treated with 0.3% (v/v) H₂O₂ in methanol for 15 min to block endogenous peroxidase and washed three times in PBS. They were then incubated for 30 min with blocking agent [0.5% normal swine serum (Dako A/S, Copenhagen, Denmark), 0.1% NaN₃ in PBS] at room temperature to eliminate nonspecific staining. Sections were incubated overnight at 4°C with monoclonal antibodies. Control staining reactions included replacement of the primary antibodies with normal mouse serum (Nippon Bio-Supply Center, Tokyo, Japan). After washing out primary antibody, sections were rinsed three times with PBS for 5 min at room temperature. Binding was visualized with the streptavidin-biotin technique (Vectastain ABC kit; Vector Laboratories, Inc., Burlingame, CA), and nuclei were counterstained with Mayer’s hematoxylin.

For the purpose of estimating the H. pylori infection risk for the functional Se allele located in 19q13.3, we categorized the participants into three groups: Se/Se, Se/se, and se/se, homozygous for the functional Se alleles, heterozygous for the functional and nonfunctional allele (Se/sej Se/se5), and homozygous for the nonfunctional alleles (sej/sej, sej/se5 and se5/se5), respectively. We also grouped the participants depending on the Le alleles located in 19p13.3. As mentioned above, the Le genotypes were separated into three groups: group Le/Le, consisting of Le/Le and Le/le3; group Le/le, consisting of Le/le1 and Le/le2; and group le/le, consisting of le1/le1, le1/le2, and le2/le2. Statistical analysis was performed with the computer program STATA Version 6 (Stata Corp., College Station, TX). The categorical data were examined with a χ² test, and the ORs of the H. pylori infection were estimated using an unconditional logistic model. The adjustment for sex, age as a continuous variable, and genotypes as dummy variables were conducted using the logistic model, as well as a significance test for trends in ORs allocating consecutive values for genotypes. The 95% CIs for the percentage were calculated based on binomial distribution.

Se and Le gene polymorphisms were clearly determined by PCR-RFLP and specific PCR. Representative PCR products were subcloned and sequenced to confirm the specific PCR amplification. Participants were divided into three genotype groups according to the Se and Le alleles (for Se: Se/Se, Se/se, and se/se; for Le: Le/Le, Le/le, and le/le). As shown in Table 1, the allele frequency of Se and Le alleles among the participants was similar to our previous study [a 2 × 9 table for the independent test, χ² = 4.82 with d.f. = 8, P = 0.777; 2 × 3 tables for Se (χ² = 1.19, d.f. = 2, P = 0.553) and for Le (χ² = 1.19, d.f. = 2, P = 0.489); Ref. 21]. The genotype frequencies for Se and Le genes were independent of each other (χ² = 0.78, with d.f. = 4 for a 3 × 3 table; P = 0.941).

The prevalence of H. pylori infection defined by ELISA values was 73.3% in the Se/Se, 64.6% in the Se/se, and 45.1% in the se/se groups, whereas it was 55.6% in the Le/Le, 68.3% in the Le/le, and 76.4% in the le/le groups (Table 2).

The crude OR (0.32; 95% CI, 0.14–0.70) and the ORs adjusted by sex and age (OR, 0.35; 95% CI, 0.15–0.80), and sex and age genotype (OR, 0.34; 95% CI, 0.15–0.78) were significantly lower for se/se than Se/Se participants (Table 3). On the contrary, le/le individuals showed a 2.59 times higher crude OR than Le/Le individuals, although the difference was not statistically significant (Table 3). The estimated H. pylori infection rate decreased with the number of Se alleles (adjusted P, trend = 0.012), whereas it increased with the number of Le alleles (adjusted P, trend = 0.012; Table 3).

Fig. 2 shows the H. pylori infection rate for the individuals with different Se and Le genotypes. They were classified into three groups (low-, moderate-, and high-risk groups), based on the infection rate and number of alleles (Fig. 2). The combined rate was 33.3% (95% CI, 16.5–54.0%; n = 34) for the low-risk group, 62.3% (54.7–69.5%; n = 175) for the moderate-risk group, and 83.8% (68.0–93.8%; n = 37) for the high-risk group. The sex and age-adjusted OR relative to the low-risk group was 3.30 (1.40–7.78) for the moderate-risk group and 10.33 (3.16–33.80) for the high-risk group. As shown in Table 4, the trend test revealed a highly significant value (P trend <0.001). These observations and interpretations indicate the possibility that both Le and Se genotypes result in a greater number of H. pylori attaching effectively to the gastric mucosa, thereby augmenting the production of anti-H. pylori antibody.

In many studies, type I Le antigens were detected on the gastric foveolar epithelium, depending on the secretor and Le phenotype, with some exceptions (30, 31, 32). We performed immunohistochemical analysis with the specific monoclonal antibodies for the purpose of confirming whether type I Le antigen expression was dependent on Se and Le genotypes as a potential mechanism to explain the effects observed. The 60 gastric cancer patients selected consisted of 40 Se/− and 20 se/se cases, including 30 Le/− and 10 le/le patients, and 15 Le/− and 5 le/le patients, respectively. As indicated in Fig. 3,A by the strong positive staining, Le^b antigen was clearly detected on the foveolar epithelium of most (Se/−, Le/−) patients but rarely detected in (se/se, Le/−; Fig. 3,B) and (se/se, le/le) patients (Fig. 3,D). In (Se/−, le/le) patients, focal and small amounts of Le^b antigen expression were seen in foveolar epithelium (Fig. 3,C). In contrast, Le^a antigen was clearly detected on the foveolar epithelium of both Se/−, Le/− patients (Fig. 3,E) and se/se, Le/− patients (Fig. 3,F) but poorly detected in Se/−, le/le patients (Fig. 3,G) or se/se, le/le patients (Fig. 3 F). These studies indicate that type I Le antigens, expressed by gastric foveolar cells, are synthesized by Le and Se enzymes and raise the possibility that both Le and Se genotypes play a role in affecting H. pylori infection through BabA binding.

This study evaluated the association between H. pylori infection risk and both polymorphisms of the Se and Le genes involved in the BabA ligand carbohydrate antigen synthesis. We found that H. pylori infection, as determined by the presence of anti-H. pylori IgG antibody, was increased among individuals with a greater number of functional Se alleles and a lesser number of Le alleles, indicating that the polymorphisms of these genes are an inherited risk factor in the Japanese population. We also demonstrated that Le^b antigen expression was present in the gastric foveolar cells of Se/−, Le/− individuals but not in either se/se, Le/− or Se/−, le/le individuals.

A lower H. pylori infection frequency was associated with an increased number of nonfunctional se alleles (all P trends <0.05; Table 3; Fig. 2). According to the type I Le antigen synthetic pathway and our results from immunohistochemical analysis (Figs. 1 and 3), se/se individuals are unable to synthesize H type I structures, resulting in deficient Le^b antigen expression on gastric foveolar cells, as shown in Fig. 3,B. Therefore, these results could be interpreted as indicating that a decrease in both H type I structures and Le^b antigen lead to a state in which H. pylori does not attach to the gastric mucosa with BabA. Higher H. pylori infection frequency was also associated with an increased number of le alleles (all P trends <0.05; Table 3; Fig. 2). As shown in Table 3 and Fig. 2, Se and Le genes affected the infection rate independently. Any interaction terms between the genotypes were not statistically significant by the logistic model. These results may be explained as that an increased number of le alleles makes it easier for the Se enzyme to use the type I precursor synthesizing H type I antigen in individuals with at least one Se allele, through the type I Le antigen synthetic pathway (Fig. 1). This would be consistent with the significant correlation between the genotypes and the augmentation of anti-H. pylori antibody that is induced by the severe inflammatory reaction caused by nuclear factor-κB activation (22, 23).

Although our results clearly indicated a significant correlation between H. pylori infection and both Le and Se genotypes of the host, some clinical studies have cast doubt on the link between H. pylori infection and the host Le blood type (32, 33, 34). The failure to find a correlation in these studies may have resulted from either the determination methods for host Le phenotype and/or the definition of H. pylori infection. For example, the hemagglutination test for determination of Le blood phenotype often results in mistyping because of specific biological conditions such as pregnancy (35), alcoholic cirrhosis (36), pancreatitis (36), hydatid cysts (37), and intestinal cancers (38). With immunohistochemical techniques, the Le phenotype might not have been accurately determined in previous studies because of the aberrant Le^b antigen on foveolar epithelium, which might be caused by the H enzyme, caused by inflammation and intestinal metaplastic change (30, 32, 39). We obtained the same results with surgically resected gastric mucosa that had undergone these inflammatory changes. Furthermore, the different detection methods of H. pylori infection would make any correlation between Le blood type and H. pylori infection ambiguous. In one study, a significant correlation was found between the host Le blood type and H. pylori infection defined by the presence of anti-H. pylori antibody (33). No correlation was reported in other studies that used H. pylori culture or histological investigation to detect the H. pylori infection (32, 34). The IARC classification of H. pylori as a definite carcinogen for gastric cancer was based largely on epidemiological evidence in which anti-H. pylori IgG antibody tests were commonly used for detection of H. pylori infection.⁴ Therefore, an accurate Le blood type determination method using the genotypes and the anti-H. pylori IgG antibody test should be applied for the purpose of evaluating the association between Le blood type and H. pylori infection in patients with chronic gastritis, gastroduodenal ulcers, and gastric carcinogenesis. Taken together, our results indicate that host type I Le blood type is correlated with the augmentation of anti-H. pylori antibodies in the development of gastric cancer.

H. pylori infection was not only observed in the Se/− group but also in the se/se group with less than the expected expression of adhesin ligand on gastric foveolar cells, for which we can propose several possible explanations: (a) H. pylori-positive gastritis is caused not only by BabA2 gene-positive strains (9) but also by other adhesive factors, such as AlpA, AlpB adhesin, and the M_r 25,000 outer membrane protein of H. pylori, and by carbohydrate-carbohydrate interactions (40, 41); (b) Le antigens on lipopolysaccharides of H. pylori play a role in adhesion through the binding to lectin protein expressed on host gastric mucosa (40, 41); and (c) a slight fucose-transferring activity of the sej enzyme could synthesize small amounts of H type I structure in the gastric foveolar epithelial cells sufficient to retain H. pylori on the mucosa. In Fig. 3 B, slight Le^b antigen expression could be detected in some normal foveolar cells in sej/sej individuals using the highly sensitive immune-complex detection method. In our previous study, sej enzyme exhibited slight fucose transfer activity to the type I precursor structure (14).

The frequency of the se/se phenotype as a nonsecretor is identical in most ethnic groups (∼20%; Ref. 42), despite the difference in nonfunctional point mutations in the Se allele between Orientals and Caucasians. Henry et al. (43) and Yu et al. (44) proposed that individuals in the Orient homozygous for the sej allele could have a partial secretor status, whereas the nonfunctional se1 allele of Caucasians did not exhibit fucose-transferring activity by nonsense mutation in Se allele. The wide distribution of sej (14, 44, 45, 46, 47), the homozygote of which shows partial secretor status in Asian populations, may be somehow linked with the fact that H. pylori infection is more frequent in the Far East than in northern Europe or the United States (48). It may be true that improvement of environmental factors affects the H. pylori infection rate. The weak activity of the sej allele or other unidentified ethnic-specific se alleles might underline the high H. pylori infection frequency in Japanese, Blacks, and Hispanics, who still exhibit high incidences of infection, even after taking socioeconomic status into account (48). Because both le1 and le2 alleles are found out not only in Asians but also in Africans and Caucasians (49), there is a slight possibility that inactivated alleles of Le gene modify the risk of H. pylori infection in an ethnic group-specific manner similar to the sej allele. However, more population studies on Le alleles among different ethnic groups should be performed to determine whether the geographical variation of H. pylori infection and gastric cancer risk might be explained by further correlation analysis of these gene polymorphisms, the frequency of H. pylori infection in other countries, and incidence of gastric cancer.

In conclusion, our findings demonstrate that the Se and Le genotypes affect the risk of H. pylori infection defined by the presence of anti-H. pylori IgG antibody. We are investigating whether the determination of the fucosyltransferases polymorphisms is valuable in identifying patients at high risk of developing duodenal ulcer, gastric ulcer, adenocarcinomas of the distal stomach, and low-grade B-cell lymphoma of mucosa-associated lymphoid tissue.

We thank our many colleagues, including Drs. Hayao Nakanishi and Testuya Tsukamoto (Aichi Cancer Center), for helpful comments on the manuscript, as well as Yoko Nishikawa-Kurobe, Masami Yamamoto, and Nami Yamada for expert technical assistance. We also thank Kurt Magnuson for helpful comments to improve the manuscript.