Persistent infection with Helicobacter pylori causes gastritis and peptic ulcer and is the strongest risk factor for the development of gastric adenocarcinoma.1, 2 Individual H. pylori isolates can be subdivided into cytotoxin-associated gene A (cagA)-positive and cagA-negative strains. Clinically, cagA-positive strains are much more potent in causing gastric mucosal damage and are intimately associated with the development of gastric adenocarcinoma.3, 4 The cagA gene is located within a 40-kb DNA fragment known as the cag pathogenicity island (cag PAI), which is suspected to have been acquired by horizontal transfer.5 The cag PAI contains 27–31 genes, which collectively encode a Type IV secretion apparatus that delivers the cagA-encoded CagA protein into host cells.6–8
Upon delivery into gastric epithelial cells, CagA undergoes tyrosine phosphorylation by Src family kinases or Abl kinase at multiple Glu-Pro-Ile-Tyr-Ala (EPIYA) motifs present in the C-terminal variable region.9–12 Tyrosine-phosphorylated CagA then binds and aberrantly activates SHP-2 tyrosine phosphatase, a signal-transducing molecule that positively regulates proliferation, motility and morphology of cells.13 Recent studies have revealed that SHP-2 is a bona fide oncoprotein whose gain-of-function mutations are associated with human malignancies.14–16 In addition to SHP-2, CagA binds to and inhibits the activity of partitioning-defective 1 (PAR1)/microtubule affinity-regulating kinase (MARK) in a tyrosine phosphorylation-independent manner,17 thereby causing disruption of tight junctions and inducing loss of apical-basolateral polarity in polarized epithelial cells. Furthermore, CagA has been reported to associate with Csk, c-Met, Grb-2, Crk and E-cadherin.18–22 Thus, perturbation of these molecules by CagA may collectively contribute to the development of gastric adenocarcinoma.
CagA proteins are noted for their structural diversity among H. pylori isolate and can be subdivided into 2 major isoforms—East Asian CagA and Western CagA. Interestingly, the geographic distribution of CagA isoforms correlates well with the geographical difference in incidence of gastric cancer.23–25 Prevalent H. pylori strains isolated in Japan, Korea and China, where the incidences of gastric adenocarcinoma are among the highest in the world, carry the East Asian CagA isoform, whereas those isolated in Western countries (Europe, North America and Australia), where gastric adenocarcinoma is relatively rare, carry the Western CagA isoform.
The main differences between the Western and East Asian CagA isoforms are in the C-terminal variable regions, which consist of various numbers and combinations of the 4 distinct EPIYA segments, EPIYA-A, -B, -C and -D. These EPIYA segments contain a single EPIYA motif, the site of CagA tyrosine phosphorylation, and are defined on the basis of sequences flanking the EPIYA motif. The C-terminal variable region of Western CagA consists of EPIYA-A, EPIYA-B and variable numbers of the Western CagA-specific EPIYA-C segments in the order of ABC, ABCC or ABCCC, whereas that of the East Asian CagA isoform carries EPIYA-A, EPIYA-B and the East Asian-specific EPIYA-D segments in the order of ABD or ABDD. Differences in the C-terminal variable regions of CagA correlate with functional differences in the CagA isoforms in vitro. SHP-2 binds specifically to the tyrosine-phosphorylated EPIYA-C and EPIYA-D segments in Western CagA and East Asian CagA, respectively, whereas EPIYA-D has a greater SHP-2-binding activity than that of EPIYA-C. Furthermore, CagA species carrying a larger number of EPIYA-C or EPIYA-D segments interact and deregulate SHP-2 more strongly than do those with a smaller number of EPIYA-C or EPIYA-D segments.26 Accordingly, strength of the CagA-SHP-2 interaction depends on the structure of CagA. Consistent with the important role of SHP-2 in human malignancy, CagA with greater ability to bind SHP-2 is clinically more closely associated with gastric carcinoma.27, 28 These observations indicate that the magnitude of the pathogenic/oncogenic activity of CagA is primarily determined by structural polymorphism in the C-terminal variable region of CagA.
We previously generated transgenic (Tg) mice systemically and constitutively expressing East Asian CagA-ABDD, which has the strongest SHP-2 binding activity among the East Asian CagA isoforms.29 By 12 weeks of age, approximately half of the East Asian CagA Tg mice developed a thickening of gastric mucosa due to epithelial hyperplasia of the glandular stomach and, by 72 weeks of age, several East Asian CagA Tg mice developed gastric hyperplastic polyps and adenocarcinomas in the stomach and intestine. In addition to gastrointestinal tumors, a fraction of East Asian CagA Tg mice developed hematopoietic malignancies such as myeloid leukemia, B cell lymphoma and T cell lymphoma by 72 weeks of age. The results indicate that CagA has the ability to induce neoplasms in vivo. We also generated and analyzed Tg mice systemically expressing phosphorylation-resistant (PR) CagA-ABDD.29 The PR CagA Tg mice neither showed gastric epithelial hyperplasia nor developed gastrointestinal or hematopoietic abnormalities, indicating that interaction of tyrosine-phosphorylated CagA with SHP-2 plays a pivotal role in abnormal cell proliferation and subsequent tumor formation in vivo.
In this work, we generated transgenic mice systemically expressing Western CagA (ABCCC-type) and found that the mice also develop gastrointestinal and hematopoietic lesions. However, the incidence of tumors in Western CagA-transgenic mice was significantly lower than that in East Asian CagA-transgenic mice. Our results indicate that Western CagA is qualitatively less oncogenic than East Asian CagA.
The humanized Western type cagA-ABCCC gene (cagAHs-ABCCC, ∼3.7-kb DNA) was chemically synthesized while converting bacterial cagA gene codons to those more commonly used in human genes without alteration of the amino acid sequence of the Western CagA protein (CagA-ABCCC) derived form H. pylori NCTC11637 strain. The synthesized DNA was 3′-tagged with a sequence encoding hemagglutinin (HA). The humanized cagAHs-ABCCC gene was subcloned into mammalian expression vector pCAGGS30 that has CAG (cytomegalovirus enhancer and a modified chicken β-actin/rabbit β-globin hybrid) promoter. The CAG-cagAHs-ABCCC fragment consisting of the CAG promoter, cagAHs-ABCCC and polyA cassettes derived from β-globin gene was excised from the pCAGGS-cagAHs-ABCCC by restriction enzyme digestion and was injected directly into fertilized eggs derived from C57BL/6 mice. The cagA transgene in the mouse genome was detected by genomic PCR. Mice were maintained under specific pathogen-free (SPF) conditions and were fed a sterile diet. Through random sampling, there was no evidence for the presence of Helicobacter spp. in the stomachs of transgenic mice established in this work. All of the animal experiments were carried out according to the protocol approved by the Ethics Committee for Animal Experiments at Hokkaido University.
Reverse transcription-polymerase chain reaction
Total RNA was extracted from frozen mouse tissues with the use of TRIzol Reagent (Invitrogen, Carlsbad, CA). RT-PCR was performed by using oligo dT primer (T18) and SuperScript II Reverse Transcriptase (Invitrogen). The following primer pairs were used for PCR amplification: cagAHs-ABCCC-forward: 5′-CCAAACTTGACAACTACGCGACGAATTCAC-3′, cagAHs-ABCCC-reverse: 5′- ACAGACCCTACATTGTG GGCTACAATCTTG-3′. The PCR condition was 94°C for 30 sec, 62°C for 30 sec, 72°C for 30 sec for 30 cycles. Quantitative RT-PCR was performed by using SYBR Green fluorescence detection system (Qiagen, Valencia, CA) with ABI PRISM 7700 sequencer (Applied Biosystems, Foster City, CA).
Anti-HA antibody (3F10; Roche Applied Science, Basel, Switzerland) and anti-actin antibody (C-11; Santa Cruz, Santa Cruz, CA) were used for immunoblottings of HA-tagged CagA and actin, respectively. Anti-PCNA antibody (PC10; DAKO Carpinteria, CA), anti-B220 antibody (RA3-6B2), anti-p53 antibody (CM5; Novocastra, Newcastle, U.K.) and anti-β-catenin antibody (BD Bioscience, San Jose, CA) were used for immunohistochemical staining.
Frozen stomachs prepared from mice were homogenized in lysis buffer T (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol, 2 mM Na3VO4, 10 mM NaF, 10 mM β-glycerophosphate, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 10 μg/ml trypsin inhibitor and 2 mM phenylmethylsulfonyl fluoride), and were then sonicated for 15 min. The lysates were subjected to immunoblotting as previously described.13
Tissue specimens were fixed in Masked Formalin (JAPAN TANNER CORPORATION, Osaka, Japan), embedded in paraffin, sectioned and stained with hematoxylin and eosin (H and E) for histopathological assessment. Alcian blue (pH 2.5) staining was applied for characterization of the gastric mucous. For immunohistochemical staining, sections were deparafinized, rehydrated and then incubated with primary antibodies, washed and incubated with appropriate secondary antibodies. Reacted antibodies were detected by the peroxidase reaction, using DAB (diaminobenzidine; DAKO) as substrate.
Statistical analysis was performed by using Student's t-test, Mann-Whitney U-test and χ2 test.
Generation of transgenic mice bearing humanized Western type cagA gene
To express Western CagA (CagA-ABCCC) systemically in mice, the humanized cagAHs-ABCCC gene was connected downstream of the chicken β-actin and rabbit β-globin fusion promoter (CAG promoter) in the same way as cagAHs-ABDD used to generate East Asian CagA-transgenic mice (cagAHs-ABDD mice; Fig. 1a). Upon injection of the gene construct into fertilized mouse eggs, 3 independent transgenic mouse lines, C-21, D-11 and F-02, were established (cagAHs-ABCCC mice). These cagAHs-ABCCC mice were indistinguishable from the wild-type littermates or cagAHs-ABDD mice in behavior and weight when they were born, and they developed normally. Through quantitative reverse transcription-polymerase chain reaction (RT-PCR) analysis of mRNAs prepared from the stomach, C-11 and D-11 strains showed the highest levels cagAHs-ABCCC mRNA expression (Fig. 1b). We hereafter show data obtained from Line C-21 mice unless otherwise noted (Fig. 1b). Upon non-quantitative RT-PCR analysis, cagAHs-ABCCC mRNAs were also detectable in various organs and tissues such as the stomach, ileum, colon, brain, thymus, heart and kidney in cagAHs-ABCCC mice (data not shown). To compare the expression levels of the CagA protein in the stomachs of the newly established cagAHs-ABCCC mice and the previously established cagAHs-ABDD mice (Line B-10), immunoblot analysis was performed using protein extracts from the stomachs of cagAHs-ABCCC mice and cagAHs-ABDD mice. The results of the experiment confirmed low levels of CagA protein expression in the stomach of the Tg mice. However, the results also showed that the level of the CagA-ABCCC protein in cagAHs-ABCCC mice was no less than that of the CagA-ABDD protein in cagAHs-ABDD mice (Fig. 1c).
Mucosal thickening in the glandular stomachs of cagAHs-ABCCC mice
We previously reported that 12-week-old cagAHs-ABDD mice developed gastric epithelial hyperplasia, primarily at the body of the stomach.29 To investigate whether epithelial hyperplasia also occurs in the stomachs of cagAHs-ABCCC mice, mice were sacrificed at 12 weeks of age and were subjected to histological analysis. In approximately half of the cagAHs-ABCCC mice (Line C-21) or cagAHs-ABDD mice (Line B-10) examined, there was thickening of the gastric mucosa (Fig. 2). The incomplete penetrance of the epithelial hyperplasia may be attributable to the differences in CagA levels among the transgenic mice, even though they are genetically identical. To peruse this possibility, we performed immunoblot analysis of CagA proteins expressed in the stomachs of individual transgenic mice. However, due to the low levels of CagA expression as well as high background of immunoblotting when whole tissue lysates were used as protein samples, we could not reproducibly generate quantitative data that can discriminate subtle differences in CagA protein levels among distinct transgenic mice. Because of this technical difficulty, we could not make any conclusion with regard to the relationship between the level of CagA protein expressed in the stomach and the degree of gastric mucosal thickening. Histologically, the proliferative zone in the gastric gland of cagAHs-ABCCC mice was markedly expanded as was found in cagAHs-ABDD mice (Fig. 2a). To quantitatively assess the degree of pathological change, we measured the thickness of gastric mucosae from C57BL/6, cagAHs-ABDD mice and cagAHs-ABCCC mice (15 mice each). It was found that thickening of gastric mucosa in cagAHs-ABCCC mice and that in cagAHs-ABDD mice were comparable (Fig. 2b). This result indicates that CagA-ABCCC is capable of inducing epithelial hyperproliferation to a level equivalent to that induced by CagA-ABDD.
Gastrointestinal tumors in cagAHs-ABCCC mice
Previously, we reported that cagAHs-ABDD mice developed hyperplastic polyps and adenoma in the stomach or intestine by 48 weeks of age and also developed adenocarcinoma by 72 weeks of age.29 To investigate gastrointestinal lesions in cagAHs-ABCCC mice, mice were sacrificed at 72 weeks of age and autopsy of the sacrificed mice was performed. As a result, 3 of 100 cagAHs-ABCCC mice (2 cases in Line C-21 and one in Line D-11) had developed hyperplastic polyps (Fig. 3a and Table I), 2 in the pyloric region of the stomach and one in the duodenum. These polyps consisted of surface epithelial cells with limited atypical cellularity without pseudostratification. One cagAHs-ABCCC mouse (Line F-02) had developed adenomatous polyp in the pyloric region (Fig. 3b). Histologically, the adenoma showed mild structural atypia of glands and pseudostratifications in apical ends of the glands. Furthermore, one cagAHs-ABCCC mouse (Line D-11) had developed adenocarcinoma in the jejunum (Fig. 3c). The adenocarcinoma consisted of irregular branching glands, fused glands and a sheet-like structure of glands lined by atypical columnar cells that showed plump nuclei with large nucleoli. Immunostaining of the adenocarcinoma cells showed intense nuclear positivity of p53 and β-catenin (Fig. 3c), suggesting mutations in genes involved in the Wnt-β-catenin system as well as the p53 gene. Though hypertrophic, gastric mucosae from 72-week-old cagAHs-ABCCC mice showed neither chronic inflammation nor atrophy. In addition, no Alcian blue staining at pH 2.5 was observed in the gastric mucosae of 72-week-old cagAHs-ABCCC mice, excluding the presence of intestinal metaplasia (Fig. 3d). From these observations, we concluded that CagA-ABCCC has the ability to induce gastrointestinal tumors in the absence of chronic inflammation or intestinal metaplasia.
Table I. Pathological Lesions Developed in CAGAHS Mice After Long-Term Observation
We previously reported that cagAHs-ABDD mice show mild granulocytosis in peripheral blood, followed by the development of hematopoietic malignancies, especially myeloid leukemia and B cell lymphoma.29 In the analysis of peripheral blood, however, cagAHs-ABCCC mice never displayed leukocytosis. Autopsy analysis of cagAHs-ABCCC mice revealed one case of lymphoma derived from mesenteric lymph nodes (Fig. 4). Histologically, the lymphoma cells were positive for the B cell-specific surface antigen B220 as determined by anti-B220 immunostaining and were characterized by enlarged and plump nuclei, with an increased mitotic index. We also found 2 cases of enlarged mesenteric lymph nodes, which were histopathologically suspected to be precancerous B cell lesions. Despite extensive autopsy investigation, however, no pathological lesions were found in organs/tissues other than the gastrointestinal tract and hematopoietic organs in cagAHs-ABDD mice.
Comparison of incidences of tumors in cagAHs-ABCCC mice and cagAHs-ABDD mice
At 72 weeks of age, 4 (4.0%) of the 100 cagAHs-ABCCC mice had developed benign tumors, hyperplastic polyps and adenoma, in the stomach and intestine. In addition, 2 (2.0%) of the 100 cagAHs-ABCCC mice had developed malignant tumors, one intestinal carcinoma and one B cell lymphoma. In a previous study, we found that 24 (11.0%) of the 219 cagAHs-ABDD mice had developed benign tumors (hyperplastic polyps and adenoma in the gastrointestinal tract) at 72 weeks of age.29 Furthermore, 20 (9.1%) of the 219 cagAHs-ABDD mice had developed malignant tumors in the gastrointestinal tract (adenocarcinoma) and hematopoietic cells (myeloid leukemias, B cell lymphomas and T cell lymphoma). No tumors were found in 107 C57BL/6 control mice at 72 weeks of age. The results indicate that the incidence of tumors in cagAHs-ABCCC mice is lower than that in cagAHs-ABDD mice.
In this work, we generated Tg mice that systemically express Western CagA (CagA-ABCCC) and found that Western CagA per se is capable of inducing gastric epithelial hypertrophy as well as gastrointestinal and hematological tumors. Establishment of Western CagA Tg mice enabled us to investigate in vivo pathogenic/oncogenic activity of geographically distinct CagA isoforms. Comparison of the pathological lesions developed in Western CagA Tg mice with those developed in East Asian CagA Tg mice indicated that Western CagA is less potent in inducing tumors in vivo than is East Asian CagA. Furthermore, our results provide the first direct evidence supporting the idea that CagA species with greater SHP-2 binding activity show stronger oncogenic activity in vivo than those with lesser SHP-2 binding activity.
Infection with cagA-positive H. pylori is associated with gastric cancer and B cell lymphoma termed mucosa-associated lymphoid tissue (MALT) lymphoma in humans.31, 32 Coincidentally, whereas the cagA transgene was systemically expressed in mice, both Western and East Asian CagA Tg mice displayed strong pathological tropisms towards the gastrointestinal and hematopoietic cells. The observed tissue tropism of CagA in Tg mice may be attributable to the differential activation of kinases such as Src family kinases and Abl kinase that mediate CagA tyrosine phosphorylation. Alternatively, elusive SHP-2 substrates that are critically involved in cell transformation might be expressed relatively abundantly in gastrointestinal cells and hematopoietic cells. Despite cagA mRNA expression, there was no colon carcinoma in CagA-transgenic mice. Interestingly, the colon is more resistant to carcinogenesis than is the small intestine in mice. This has been clearly demonstrated by studies using ApcMin mice as well as Apc knockout mice.33–35 In humans, deletion of APC causes colon carcinomas, whereas Apc deletion in mice primarily induces small intestinal carcinomas but not colon carcinomas, as was the case for CagA Tg mice. Currently, however, underlying mechanisms that make the colon more resistant than the small intestine in mice to carcinogenesis remain unknown.
Development of epithelial hyperplasia in the glandular stomach of cagAHs-ABCCC mice indicates that expression of Western CagA per se is growth stimulatory in gastric epithelial cells in vivo as was the case of East Asian CagA. Because SHP-2-binding ability of East Asian CagA (CagA-ABDD) is stronger than that of Western CagA (CagA-ABCCC),26, 36 it was expected that the growth-promoting activity of East Asian CagA is also greater than that of Western CagA in vivo. However, thickening of gastric mucosa in cagAHs-ABCCC mice was almost comparable to that observed in cagAHs-ABDD mice. These results, together with the fact that gastric mucosal thickening was not observed in all of the transgenic mice, indicate that induction of hyperplasia requires a certain threshold level of CagA-activated SHP-2 in gastric epithelial cells. Only when SHP-2 activity has exceeded that level can cells initiate hyperproliferation, although the magnitude of hyperproliferation is not simply determined by the level of SHP-2 activation.
At 72 weeks of age, a small fraction of cagAHs-ABCCC mice had developed hyperplastic polyps in the gastrointestinal tract. In humans, hyperplastic polyp is the most common polyp of the stomach, generally associated with H. pylori infection, and has been considered to be induced by infection-associated chronic inflammation/damage and subsequent regeneration of gastric mucosa.37, 38 It is, therefore, possible that development of hyperplastic polyps in cagAHs-ABCCC mice also involves inflammation, which could be repeatedly induced by physical/mechanical damages to the stomach mucosa. A 72-week-old cagAHs-ABCCC mouse had developed adenocarcinoma in the jejunum, and another 72-week-old cagAHs-ABCCC mouse had developed B cell lymphoma. Although the tumor incidence was extremely low, the results provide compelling evidence that Western CagA has an in vivo oncogenic potential. In the absence of associated inflammation, atrophy or intestinal metaplasia, CagA-mediated tumor development seems to be cell-autonomous, suggesting that CagA plays a key role in gastric carcinogenesis as a tumor initiator and also suggesting that chronic mucosal damage is not an essential prerequisite for the development of gastric carcinoma. The notion is supported by the finding that a H. pyloricagA-positive strain is not only associated with intestinal-type gastric carcinoma but also diffuse-type gastric carcinoma, which is not accompanied by inflammation or atrophy.39, 40 At the same time, however, the long latency and low penetrance of tumor formation in these mice suggest that the oncogenic activity of CagA on its own is weak and that additional host genetic changes such as inactivation of p53 and/or deregulation of the Wnt signaling pathway are required over time. The low incidence of tumors may also be due to the low level of CagA expression, most probably because deregulated hyperactivation of SHP-2 by CagA during embryogenesis leads to embryonic lethality.41 Generation of mice that conditionally express CagA could circumvent this problem. It would also be interesting to know whether chronic inflammation in the gastric mucosa potentiates the weak oncogenic activity of CagA in vivo. This idea is supported by the high incidence (∼80%) of gastric cancer development in mice infected with Helicobacter felis, which provokes strong inflammation in the stomach.42, 43
The incidences of benign tumors in the gastrointestinal tracts of cagAHs-ABCCC mice and cagAHs-ABDD mice were 4.0% and 11.0%, respectively. Approximately 2.0% of cagAHs-ABCCC mice developed gastrointestinal and hematopoietic malignancies, whereas 9.1% of cagAHs-ABDD mice developed gastrointestinal and hematopoietic malignancies. Given that the established Tg mice express similar levels of Western and East Asian CagA species, the present work provides in vivo evidence that the oncogenic activity of Western CagA (CagA-ABCCC) is weaker than that of East Asian CagA (CagA-ABDD). The in vivo observation, which is consistent with the in vitro observation that biological activity of CagA is correlated with the degree of SHP-2-binding activity,26, 33 supports the idea that endemic circulation of H. pylori carrying East Asian CagA substantially contributes to the high incidence of gastric carcinoma in East Asian countries. Furthermore, among Western CagA isoforms, those having more EPIYA-C segments bind SHP-2 more strongly than do those having less EPIYA-C segments.26 Consistently, a recent clinical study performed with European patients showed that the relative risk of gastric adenocarcinoma is directly correlated with the number of EPIYA-C segments of CagA.28 Thus, CagA-ABC, the most common Western CagA isoform, may be less virulent and less oncogenic than ABCC- or ABCCC-type CagA used in this work. It should also be noted here that CagA-ABCCC possesses 4 copies of the CM sequence, the binding site for PAR1, whereas CagA-ABDD possesses only a single CM sequence. Together with the results of our previous work showing that the presence of a single CM sequence is sufficient for functional inhibition of PAR1 by CagA,17 the results of the present work indicate that the number of CM sequences may not be a major determinant for the degree of oncogenic potential of individual CagA.
Notably, hematological abnormalities found in cagAHs-ABCCC mice were less prominent than those in cagAHs-ABDD mice. First, cagAHs-ABCCC mice did not develop leukocytosis, which was observed in cagAHs-ABDD mice. Second, the incidence of hematological malignancies in cagAHs-ABCCC mice was significantly lower than that in cagAHs-ABDD mice. The only developed hematological malignancy in cagAHs-ABCCC mice was B cell lymphoma. This is in sharp contrast to the fact that myeloid leukemias are prominent in cagAHs-ABDD mice. Although we cannot make any conclusion at this moment because of the low incidence of hematological malignancies, it is possible that CagA-ABCCC has less impact on myeloid lineage cells than does CagA-ABDD.
Our work provides evidence for the differential oncogenic potential of geographically distinct H. pylori CagA isoforms, which may contribute to differing degrees of risk for gastric cancer in geographically distinct regions of the world.
The authors thank Drs. Gen Yamada, Misao Suzuki and Hideaki Higashi for their help.