Activation of β-catenin by carcinogenic Helicobacter pylori

Abstract

Persistent gastritis induced by Helicobacter pylori is the strongest known risk factor for adenocarcinoma of the distal stomach, yet only a fraction of colonized persons ever develop gastric cancer. The H. pylori cytotoxin-associated gene (cag) pathogenicity island encodes a type IV secretion system that delivers the bacterial effector CagA into host cells after bacterial attachment, and cag⁺ strains augment gastric cancer risk. A host effector that is aberrantly activated in gastric cancer precursor lesions is β-catenin, and activation of β-catenin leads to targeted transcriptional up-regulation of genes implicated in carcinogenesis. We report that in vivo adaptation endowed an H. pylori strain with the ability to rapidly and reproducibly induce gastric dysplasia and adenocarcinoma in a rodent model of gastritis. Compared with its parental noncarcinogenic isolate, the oncogenic H. pylori strain selectively activates β-catenin in model gastric epithelia, which is dependent on translocation of CagA into host epithelial cells. β-Catenin nuclear accumulation is increased in gastric epithelium harvested from gerbils infected with the H. pylori carcinogenic strain as well as from persons carrying cag⁺ vs. cag^- strains or uninfected persons. These results indicate that H. pylori-induced dysregulation of β-catenin-dependent pathways may explain in part the augmentation in the risk of gastric cancer conferred by this pathogen.

The pathogen Helicobacter pylori colonizes the human stomach for decades, and a biological consequence of long-term colonization is an increased risk of developing gastric adenocarcinoma (1, 2). One strain-specific H. pylori constituent that augments cancer risk is the cytotoxin-associated gene (cag) pathogenicity island (2), a genetic locus that encodes a type IV secretion system. Upon delivery into host cells by the cag secretion system, the product of the terminal gene in the island, CagA, undergoes Src-dependent tyrosine phosphorylation and activates a eukaryotic phosphatase (SHP-2), leading to dephosphorylation of host cell proteins and cellular morphological changes (3-5). Translocation, but not phosphorylation, of CagA also disrupts apical-junctional complexes, resulting in a loss of cellular polarity (6).

β-Catenin is a ubiquitously expressed molecule that performs at least two distinct functions within host cells. Membrane-bound β-catenin is a component of cadherin-based adherens junctions that link cadherin receptors with the actin cytoskeleton (7). Cytoplasmic β-catenin is a downstream component of the Wnt pathway; when Wnt signaling is inactive, β-catenin is bound within a multiprotein inhibitory complex that includes glycogen synthase kinase-3β, the adenomatous polyposis coli tumor suppressor protein, axin, protein phosphatase 2A, and dishevelled (Dsh) (7). Under basal conditions, β-catenin is constitutively phosphorylated by glycogen synthase kinase-3β, ubiquitinated by the E3-SCF^β-TrCP complex, and subjected to regulated degradation by the proteasome. Binding of Wnt to its receptor, Frizzled (Frz), inhibits β-catenin phosphorylation, leading to its nuclear accumulation and the formation of heterodimers with the transcription factor lymphocyte enhancer factor/T cell factor (LEF/TCF) that results in targeted transcriptional activation of genes implicated in carcinogenesis. β-Catenin also can be activated by bacterial contact, and coculture of intestinal epithelial cells with nonpathogenic Salmonella leads to β-catenin activation through blockade of ubiquitination (8, 9).

Nuclear accumulation of β-catenin is increased within gastric cancer precursor lesions such as gastric adenomas (10), suggesting that aberrant activation of β-catenin precedes the development of gastric adenocarcinoma. β-Catenin-responsive genes include c-myc, cyclin D, MMP-7, and COX-2, which influence apoptosis, proliferation, and carcinogenesis, and H. pylori increases the expression of each of these target genes within colonized gastric mucosa and during coculture with gastric epithelial cells in vitro (11-17). H. pylori also increases the expression of β-catenin in cultured T84 intestinal epithelial cells (18).

Determination of the contribution of the host microenvironment to H. pylori-induced gastric cancer necessitates the use of animal models, and such systems have provided valuable insights into the factors involved in gastric carcinogenesis. Long-term (>1 year) H. pylori infection of Mongolian gerbils can eventuate in gastric adenocarcinoma (19, 20), but this prolonged time course precludes large-scale analyses that evaluate the effects of both pathogen and host in the carcinogenic cascade. We previously identified bacterial loci within an H. pylori clinical isolate (B128) that are required for inflammatory responses within gerbil mucosa (21). However, adenocarcinoma did not develop among any of the infected gerbils, consistent with observations that H. pylori-induced cancer in this model has never been reported outside of Japan or China. Serial passage of H. pylori in rodents increases colonization efficiency; therefore, we investigated whether in vivo adaptation of H. pylori strain B128 would enhance its carcinogenic potential. Because β-catenin is aberrantly activated by bacterial contact, is overexpressed within H. pylori-associated premalignant lesions, and regulates the transcription of genes implicated in tumorigenesis, we used this adapted isolate to determine whether H. pylori alters β-catenin activation and to investigate the molecular pathways underpinning these events to define a potential tumor-promoting response toward this pathogen.

Materials and Methods

Animals and H. pylori Challenge. All procedures were approved by the institutional animal care committee of Vanderbilt University. A male gerbil infected with a minimally in vitro passaged single-colony isolate of the cag⁺ vacA s1a human H. pylori gastric ulcer strain B128 (21) was euthanized 3 weeks after challenge, and a single-colony derivative (7.13) was purified. H. pylori strains 7.13 and B128 will be made available to qualified investigators upon request. Male Mongolian gerbils (Harlan Labs, Indianapolis) 4-8 weeks of age were orogastrically challenged with either sterile Brucella broth or H. pylori strain B128 or 7.13 and were euthanized between 4 h and 16 weeks after inoculation (21). One half of the glandular stomach was fixed for histologic examination, and the other half was homogenized in sterile PBS, plated on selective Trypticase (BBL) soy agar plates, and incubated for 3-5 days for H. pylori culture (21). Linear strips extending from the squamocolumnar junction through proximal duodenum were fixed in neutral-buffered 10% formalin. Tissues were paraffin-embedded and stained with hematoxylin and eosin, and indices of inflammation and injury were scored on a 0-3 scale by a single pathologist blinded to treatment groups. Dysplasia and adenocarcinoma were diagnosed by using morphologic criteria previously established for gastrointestinal neoplasia in mouse models of disease (22).

Clinical Specimens. Patients (n = 25) were prospectively enrolled in this study after written informed consent. The study was approved by the Vanderbilt University and Nashville Department of Veterans Affairs institutional review boards. Exclusion criteria included the use of nonsteroidal anti-inflammatory medications, antibiotics, or bismuth compounds before endoscopy; history of malignancy; or detection of H. pylori by histology but not culture. To isolate H. pylori, antral biopsies were placed in normal saline, homogenized, plated onto Trypticase soy agar with 5% sheep blood, and incubated for 96 h under microaerobic conditions (21). cagA genotype was determined by PCR of H. pylori genomic DNA (11). Human serum IgG responses to H. pylori whole-cell antigens and CagA were measured by ELISAs (11).

Immunohistochemistry. A single observer blinded to H. pylori infection and cagA status evaluated β-catenin immunolabeling within gerbil and human gastric mucosa. Immunohistochemical analysis was performed on paraffin-embedded gastric tissue by using an anti-β-catenin antibody (1:200, Transduction Laboratories, Lexington, KY) (23). For each sample, 10 well oriented representative gastric glands were scored.

Cell Culture, Plasmids, and Reagents. AGS gastric epithelial cells were grown in RPMI medium 1640 (GIBCO/BRL) with 10% FBS (Sigma) and 20 μg/ml gentamycin under a 5% CO₂/95% air atmosphere at 37°C. Conditionally immortalized murine gastric epithelial cells were grown in RPMI medium 1640 with 5% FBS and 20 μg/ml gentamycin under a 5% CO₂/95% air atmosphere at 33°C. Topflash and Fopflash reporter plasmids and TCF were kind gifts from K. Kinzler and B. Vogelstein (Johns Hopkins University, Baltimore). MG-262 was purchased from Biomol (Plymouth Meeting, PA).

H. pylori Strains. The H. pylori cag⁺ isolate B128, rodent-adapted strain 7.13, cag⁺ clinical strain J166, or cag^- clinical isolate J68 was grown in Brucella broth with 5% FBS for 18 h, harvested by centrifugation, and added to gastric cells at a bacteria-to-cell ratio of 100:1. Isogenic cagA^-, cagE^-, and vacA^- null mutants were constructed within strains 7.13 and J166 by insertional mutagenesis using aphA (11) and were selected with kanamycin (25 μg/ml).

Western Blot Analysis. Total bacterial or gastric cell lysates were harvested, and protein concentrations were quantified by the Bradford assay (Pierce) (11). Proteins (20 μg) were separated by SDS/PAGE and transferred to poly(vinylidene difluoride) membranes (Pall). CagA was detected by using anti-CagA (1:2,000, Austral Biological) antibodies (11). Primary antibodies were detected by using goat anti-rabbit IgG (Sigma) horseradish peroxidase-conjugated secondary antibodies, visualized by using the enhanced chemiluminescence detection system (Cell Signaling Technology, Beverly, MA), and quantitated by using the Bio-Rad Quantity One system.

Transfections and Luciferase Assays. AGS cells (2 × 10⁵) were transfected with 3 μl of Lipofectamine 2000 (Life Technologies, Grand Island, NY), 1 μg/ml Topflash or Fopflash, 0.5 μg/ml TCF/LEF, and 0.5 μg/ml CMX-GAL reporter (kindly provided by J. Eid, Vanderbilt University) in Opti-MEM (Life Technologies) for 5 h. Transfection mixtures were then replaced with complete medium containing either H. pylori or medium alone. After 24 h, cells were harvested in 1× Reporter Lysis Buffer (Promega). Luciferase activity was determined by using a luminometer and normalized to β-galactosidase activity. For β-catenin and CagA immunofluorescence, AGS cells were transfected by using Fu-GENE 6 (Roche Diagnostics) per the manufacturer's instructions with either 6 μg of wild-type full-length CagA or empty vector (5).

Immunofluorescence. Gastric cells were cultured on glass cover slides, and cells were treated with or without H. pylori, washed twice with PBS, permeabilized, and fixed with ice-cold methanol at -20°C. Slides were incubated in 3% BSA (Sigma) for 10 min and incubated with rabbit anti-β-catenin antibody (1:100, Sigma) overnight at 4°C. For dual immunofluorescence, slides were stained with mouse anti-β-catenin antibody (1:100, BD Transduction Laboratories) and rabbit anti-CagA antibody (1:100, Austral Biological). Washed slides were then incubated with either goat anti-rabbit IgG-Cy2 (1:100; Molecular Probes) for single immunofluorescence or Alexa Fluor 488-conjugated anti-mouse IgG antibody and Alexa Fluor 546-conjugated anti-rabbit IgG antibody (1:250, Molecular Probes) for dual immunofluorescence at room temperature for 30 min. For each sample, at least 100 cells were evaluated by an independent observer unaware of experimental conditions.

Statistical Analysis. Patients were considered to be H. pylori-infected if the results from histological examination of biopsy tissue and culture and serology for H. pylori whole-cell antigens were positive. The Mann-Whitney U test was used for statistical analyses of intergroup comparisons. Significance was defined as P ≤ 0.05.

Results

Host Adaptation Engenders the Development of a Prototype Carcinogenic H. pylori Strain. To determine whether rodent adaptation modified the pathogenic potential of H. pylori, a gerbil infected with a human clinical H. pylori strain (B128) was euthanized 3 weeks after challenge, and a single-colony output derivative (7.13) was used to infect an independent population of gerbils. All animals (n = 16) challenged with strain 7.13 for up to 8 weeks were successfully colonized, and the kinetics and intensity of inflammation were similar to those induced by human parental strain B128. However, gastric dysplasia developed by 4 weeks in seven (88%) of eight gerbils infected with rodent-adapted 7.13 that were euthanized at this time point, which was accompanied by adenocarcinoma in two (25%) of eight animals. By 8 weeks, gastric adenocarcinoma was present in six (75%) of the remaining eight infected gerbils that were euthanized at this time point (Fig. 1). Similar to our previous results (21), gastric dysplasia and adenocarcinoma were not present in gerbils (n = 16) infected with the progenitor H. pylori strain B128. The genetic composition of H. pylori isolates can change over time within an individual human host (24); therefore, the DNA content of strains B128 and 7.13 was compared by using whole-genome microarray (see Supporting Materials and Methods, which is published as supporting information on the PNAS web site). No detectable differences in genetic content were identified by this analysis (data not shown), indicating that the malignant phenotype induced by strain 7.13 was not due to loss of known ORFs.

Fig. 1.

Development of premalignant and malignant lesions within gerbil gastric mucosa after infection with H. pylori strain 7.13. No significant pathologic abnormalities were present in gerbils challenged with broth alone (A). Dysplastic foci (B) were often accompanied by gastric carcinomas arising within a background of inflammation (C and D). Gastric carcinomas were characterized by marked cellular pleomorphism, cellular atypia, and euchromatic nuclei that exhibited bizarre morphologic features.

To more rigorously define the kinetics of carcinogenesis, a larger number of gerbils were infected with either H. pylori strain 7.13 or broth alone for 2-16 weeks in multiple independent experiments. There was no evidence of gastric inflammation or injury in control animals (n = 42) (Fig. 1). A total of 116 (93%) of 125 H. pylori-challenged animals were successfully colonized, and the incidence of gastric dysplasia increased steadily such that by 16 weeks, dysplasia was present in 76% of infected animals. Gastric adenocarcinoma developed in 17%, 59%, and 59% of H. pylori-infected gerbils by 4, 8, and 16 weeks after inoculation, respectively, and tumors were accompanied by severe lymphofollicular gastritis (Fig. 1). Coded specimens were evaluated by two independent pathologists, and absolute concordance regarding the presence of dysplasia and adenocarcinoma was achieved. These results indicate that rodent adaptation led to the identification of an H. pylori strain that rapidly and reproducibly induces gastric adenocarcinoma in a highly penetrant fashion.

H. pylori Strain 7.13 Selectively Alters β-Catenin Localization in Gastric Epithelial Cells. β-Catenin regulates the expression of genes implicated in carcinogenesis; therefore, we used oncogenic isolate 7.13 as a prototype to determine whether H. pylori alters β-catenin in gastric epithelial cells. Conditionally immortalized murine gastric epithelial cells that harbor a temperature-sensitive mutation of the simian virus 40 large T antigen (25) (Fig. 5, which is published as supporting information on the PNAS web site) were grown under primary conditions and infected with H. pylori strain 7.13, B128, or medium alone. β-Catenin membrane staining decreased 6 h after infection with strain 7.13 in conjunction with increased nuclear levels of β-catenin (Fig. 6A, which is published as supporting information on the PNAS web site). In contrast, β-catenin localization was similar in uninfected and H. pylori B128-infected cells (Fig. 6A).

H. pylori is a human pathogen; therefore, we extended these results into a human cell model by infecting AGS human gastric epithelial cells that possess wild-type adenomatous polyposis coli (26) with H. pylori strains 7.13 and B128. Similar to the pattern observed in conditionally immortalized cells, nuclear translocation of β-catenin occurred 6 h after infection with strain 7.13 but not B128 (Fig. 2 A and C). Western immunoblotting using AGS cytosolic and nuclear extracts confirmed increased levels of β-catenin in nuclear fractions of cells cocultured with H. pylori strain 7.13 vs. medium-treated controls (Fig. 6B). These results indicate that carcinogenic H. pylori strain 7.13 selectively induces nuclear accumulation of β-catenin in two independent gastric epithelial model systems.

Fig. 2.

H. pylori strain 7.13 induces nuclear translocation of β-catenin in a CagA-dependent manner and activates a β-catenin responsive transcriptional reporter. (A) AGS gastric epithelial cells incubated with medium alone, strain 7.13, or strain B128 for 6 h were immunostained with an anti-β-catenin antibody and visualized by fluorescence microscopy. Representative images are shown. H. pylori strain 7.13, but not B128, induced β-catenin translocation to the nucleus (arrow). (B) AGS cells were transfected with luciferase reporter constructs containing LEF/TCF-binding motifs (Topflash) or mutated LEF/TCF sites (Fopflash) in the absence or presence of strain 7.13. Luciferase activity was determined after 24 h of treatment. ^*, P < 0.05 vs. Topflash alone. (C) AGS cells were cultured in the absence or presence of H. pylori cag⁺ strain 7.13 or isogenic cagA, vacA, or cagE mutant derivatives, progenitor strain B128, cag⁺ strain J166 or isogenic cagA^- or vacA^- mutants, or a cag^- clinical strain (J68) for 6 h. The number of cells with nuclear β-catenin per 100 cells evaluated is shown for each sample. Error bars = SD. ^*, P < 0.05 vs. AGS cells alone. (D) Cellular distribution of β-catenin and CagA in AGS cells was detected by immunofluorescence (β-catenin, red; CagA, green) after transfection with reagent alone, vector alone, or wild-type CagA. (E) Lysates from H. pylori strains 7.13 or B128 grown in broth alone or after coculture with AGS or conditionally immortalized (Immorto) gastric epithelial cells were used for Western blot analysis using anti-CagA antibodies.

Coculture of intestinal epithelia with nonpathogenic Salmonella leads to activation of β-catenin through blockade of ubiquitination (8). To determine whether modulation of β-catenin by H. pylori was regulated by a similar mechanism, AGS cells were pretreated with the proteasomal inhibitor MG-262 to stabilize the ubiquitinated form of β-catenin. Immunoblotting showed an accumulation of the phosphorylated form of β-catenin, and ubiquitinated β-catenin adducts developed in MG-262-treated, but not untreated, gastric cells (Fig. 6C). Both of these posttranslational β-catenin modifications also occurred during coculture with H. pylori strain 7.13 (Fig. 6C), a condition that results in nuclear translocation of β-catenin (Fig. 2A), indicating that stabilization of β-catenin by H. pylori is not due to inhibition of β-catenin phosphorylation or ubiquitination.

H. pylori Induces β-Catenin-Dependent Transcriptional Activation. The functional consequences of H. pylori-induced alterations of endogenous β-catenin were next determined by transfecting AGS cells with a reporter containing three tandem LEF/TCF-binding motifs upstream of the luciferase gene (Topflash) or a control construct containing mutant LEF/TCF sites (Fopflash) and then infecting cells with H. pylori strain 7.13. Luciferase activity did not differ in cells transfected with the control construct with or without H. pylori; however, activity was significantly higher in H. pylori-infected than in uninfected cells harboring the β-catenin-responsive construct (Fig. 2B), indicating that β-catenin is functionally responsive to H. pylori strain 7.13 in gastric epithelial cells.

H. pylori-Induced Nuclear Translocation of β-Catenin Is Dependent on CagA. The H. pylori cag island induces epithelial responses that may lower the threshold for gastric cancer (3-6); therefore, we examined effects of cag genes on nuclear translocation of β-catenin. H. pylori strain 7.13 potently induced IL-8 release, and CagA was translocated into and phosphorylated within AGS cells (Fig. 7, which is published as supporting information on the PNAS web site), confirming the presence of a functional cag island. AGS cells were incubated with H. pylori strain 7.13, its isogenic cagA^- or cagE^- null mutant derivatives, an isogenic derivative that lacks vacA (an independent H. pylori locus associated with gastric cancer) (2), a clinical H. pylori isolate that lacks the cag island (J68), or strain B128. Inactivation of vacA had no effect on the nuclear accumulation of β-catenin, compared with the wild-type strain 7.13 (Fig. 2C); however, β-catenin nuclear translocation did not occur in gastric cells incubated with the 7.13 cagA^- or cagE^- mutants (Fig. 2C). These results were confirmed by using another H. pylori strain (J166) that contains a functional cag island, because loss of cagA, but not vacA, significantly attenuated β-catenin nuclear translocation (Fig. 2C). Similarly, the clinical cag^- strain J68 failed to induce nuclear translocation of β-catenin (Fig. 2C). To more rigorously determine the role of CagA in β-catenin activation, recombinant CagA containing ABCCC-type EPIYA motifs (5) was transfected into AGS cells (Fig. 2D). Compared with vector alone, transfection of cells with recombinant CagA not only induced the previously reported cellular elongation phenotype (5) but also strongly induced nuclear translocation of β-catenin (Fig. 2D), indicating that H. pylori-induced β-catenin nuclear accumulation is dependent on translocation of CagA.

H. pylori Strain 7.13 Translocates an Increased Amount of CagA into Host Cells. Based on the ability of H. pylori strain 7.13, but not strain B128, to induce nuclear localization of β-catenin, coupled with the finding that CagA is sufficient to induce β-catenin translocation (Fig. 2D), we compared CagA expression and translocation patterns between these two strains. Both strains expressed similar levels of CagA when grown in broth alone, and both strains translocated CagA into AGS and conditionally immortalized cells, indicating that each strain harbors a functional cag secretion system (Fig. 2E). However, the amount of CagA translocated by strain 7.13 was substantially greater than strain B128 (Fig. 2E). Because translocation of CagA is dependent on H. pylori adherence to host cells (3), we compared the binding avidity of strain 7.13 to B128 using both AGS and conditionally immortalized cells (Fig. 8, which is published as supporting information on the PNAS web site). Carcinogenic strain 7.13 bound significantly more avidly (>1.5 log-fold increase) to both epithelial models than did strain B128, suggesting that one mechanism through which H. pylori strain 7.13 can deliver more CagA into epithelial cells may be through its ability to adhere more avidly to host cells.

Nuclear Accumulation of β-Catenin Is Increased Within Gerbil Gastric Mucosa Early After Infection with H. pylori Carcinogenic Strain 7.13. Our in vitro data indicated that H. pylori strain 7.13 activates β-catenin rapidly and in a mouse model of infection, H. pylori up-regulates genes implicated in carcinogenesis such as COX-2 within gastric epithelial cells as early as 3 h after challenge (27). Therefore, we performed immunohistochemistry for β-catenin on gastric mucosa harvested from gerbils infected with H. pylori strain 7.13 or broth alone that were euthanized between 4 h and 1 week after inoculation. Compared with uninfected gerbils in which β-catenin was virtually exclusively localized to the gastric epithelial cell membrane, membrane β-catenin staining decreased in H. pylori-infected gerbils in conjunction with an increase in the number of cells containing cytoplasmic and nuclear β-catenin (Fig. 3), albeit to a lesser degree than observed in vitro. These changes occurred in the absence of mucosal inflammation. We also performed immunohistochemistry for β-catenin on gastric mucosa harvested from gerbils infected for longer periods of time (6 and 12 weeks) to determine whether nuclear β-catenin was present in specimens that contained dysplasia or adenocarcinoma. In these samples, the number of cells containing cytoplasmic or nuclear β-catenin was no different from that in uninfected animals (data not shown), indicating that in this model, H. pylori alters β-catenin signaling in the early stages of gastric carcinogenesis.

Fig. 3.

β-Catenin membrane localization is decreased within gerbil gastric mucosa early after infection with H. pylori strain 7.13. Immunohistochemistry for β-catenin was performed on gastric mucosa harvested from H. pylori strain 7.13-infected and uninfected gerbils euthanized 4 h, 24 h, and 1 week after inoculation. Representative staining for β-catenin is shown for uninfected (A, C, and E) and H. pylori-infected (B, D, and F) gerbils at the indicated time points. Arrows indicate cells with decreased β-catenin membrane staining and/or increased cytoplasmic and nuclear β-catenin.

Nuclear β-Catenin Is Increased Within Human Gastric Mucosa Colonized by H. pylori cag⁺ Strains. To extend these results into the natural niche of H. pylori, human gastric biopsies (n = 25) were immunostained to define β-catenin localization. Membrane-localized β-catenin was detected in gastric epithelial cells in infected and uninfected patients (Fig. 4), but epithelial nuclear β-catenin was detected significantly more frequently in specimens harvested from H. pylori cag⁺-colonized, compared with either cag^--infected (P = 0.05) or uninfected (P = 0.008) persons (Fig. 4A). Within cag⁺-infected mucosa, immunolabeling for nuclear β-catenin was focal and frequently localized to epithelial cells within the proliferative zone in the mid to deep antral glands and was not present in foveolar epithelium (Fig. 4D). Two of the 25 patients had intestinal metaplasia, a precursor lesion for intestinal-type gastric adenocarcinoma; both of these patients were colonized by H. pylori cag⁺ strains, and their biopsy specimens contained the highest number of cells with nuclear β-catenin (Fig. 4A).

Fig. 4.

Nuclear β-catenin is increased in gastric epithelium harvested from H. pylori cag⁺-colonized persons. (A) Gastric epithelial cells with nuclear β-catenin, as assessed by immunohistochemistry, were quantified by an observer unaware of H. pylori status. Results are expressed as the number of cells per sample with detectable nuclear β-catenin. Mean values (▴) are shown adjacent to data points (▪). Representative staining for β-catenin is shown for uninfected (B), H. pylori cag^--infected (C), or H. pylori cag⁺-infected (D) persons. Arrows indicate nuclear β-catenin.

Discussion

H. pylori has evolved numerous strategies to facilitate its persistence within the human stomach, and a biological consequence of long-term colonization is an increased risk of gastric adenocarcinoma. Our current results have identified a potential mechanism that may help to explain the increased cancer risk conferred by H. pylori by (i) capitalizing on a rodent model that recapitulates human disease to identify a pathogenic strain of H. pylori that rapidly induces gastric cancer, (ii) investigating the ability of this prototype H. pylori strain to activate β-catenin by using biologically relevant in vitro models of microbial-epithelial cell interactions, (iii) defining bacterial constituents required for these effects, and (iv) confirming these results in specimens harvested from the natural niche of this pathogen.

Our in vitro data indicate that activation of β-catenin by H. pylori requires cag-mediated translocation of CagA into host cells. Although these findings mirror signaling patterns that regulate other cellular responses to H. pylori (e.g., cell elongation) (3-5), they are distinct from mechanisms that regulate β-catenin activation by other bacteria. Coculture of intestinal epithelia with nonpathogenic Salmonella leads to activation of β-catenin signaling through AvrA-mediated blockade of β-catenin ubiquitination (8, 9). However, our results are not consistent with this model, because H. pylori failed to inhibit ubiquitination and database searches reveal no homologs of Salmonella AvrA within the genomes of the two sequenced H. pylori strains. Bacteroides fragilis toxin induces nuclear accumulation of β-catenin in human intestinal cells, but this induction does not require direct contact between bacteria and host cells (28). Bacterial LPS can stimulate nuclear localization of β-catenin in myeloid cells through toll-like receptor (TLR)-4-dependent activation of phosphatidylinositol 3-kinase, which subsequently inhibits glycogen synthase kinase-3β, thereby increasing steady-state levels of free β-catenin (29). However, H. pylori fails to activate TLR4-mediated pathways, in either epithelial or myeloid cells (30, 31). Our results have demonstrated that accumulation of phospho-β-catenin is independent of H. pylori-mediated effects, suggesting that H. pylori does not directly affect upstream signaling components of the Wnt pathway. Recent data have demonstrated that treatment of cells with growth factors, such as EGF, resulting in transactivation of the EGF receptor, culminates in increased transcriptional activity of β-catenin, but in a Wnt-independent manner (32). Of interest, H. pylori cag⁺ strains have been shown to selectively transactivate the EGF receptor through the release of HB-EGF (33), but this activation is a CagA-independent consequence of cag-island-mediated epithelial contact, which is not supported by our current findings. Although H. pylori can induce expression of β-catenin in T84 intestinal epithelial cells in a CagA-dependent manner (18), the precise mechanisms through which H. pylori induces nuclear accumulation and functional activation of β-catenin in gastric epithelial cells remain to be clarified.

The vast majority of H. pylori in colonized hosts are free-living; however, ≈20% bind specifically to gastric epithelial cells, and adherence is required for prolonged persistence in the stomach and for induction of injury and disease (34, 35). Binding of H. pylori to gastric epithelial cells also is required for translocation of CagA (3), and the presence of the cag island influences the topography of gastric colonization, because H. pylori cag^- strains predominate within the mucus gel layer, whereas cag⁺ strains are found immediately adjacent to epithelial cells (36). Having demonstrated that carcinogenic strain 7.13 binds more avidly to epithelial cells, the current findings lend support to the hypothesis that dynamic interactions between H. pylori cag⁺ strains and host receptors legislate responses with carcinogenic potential.

H. pylori strains that possess a functional cag island clearly increase disease risk, and we have identified one molecule within this locus (CagA) that is required for β-catenin activation. Our current studies using an H. pylori whole-genome microarray did not identify differences in the genetic composition of strains 7.13 and B128, although this assessment does not eliminate the possibility that genetic differences may exist at the level of point mutations or small deletions or among loci that are not present in the two sequenced strains (26695 and J99) used to generate the arrays. Both H. pylori strains also possess a functional cag island as evidenced by the fact that they induce similar levels of IL-8 when cocultured with gastric epithelial cells, express similar levels of CagA when grown in broth alone, and can successfully translocate CagA into AGS and conditionally immortalized gastric epithelial cells. We also have sequenced cagA from these two strains (data not shown), and the nucleotide sequences are identical. However, other H. pylori disease-associated virulence constituents also could potentially alter β-catenin signaling. BabA is an outer membrane protein encoded by the H. pylori strain-specific gene babA2 (37), and H. pylori babA2⁺ strains are associated with an increased risk for gastric adenocarcinoma (34). Another H. pylori outer membrane protein, SabA, also has been linked to gastric cancer (35). Based on the finding that carcinogenic strain 7.13 bound significantly more avidly to gastric epithelial cells than did strain B128, a high priority for the future will be to examine differences in expression of H. pylori adhesins, such as BabA and SabA. Importantly, we now have unique in vitro models of bacterial-epithelial interactions by using a carcinogenic H. pylori strain that is easily transformable in conjunction with a robust animal model of H. pylori-induced gastric cancer in which to evaluate the individual and collective effects of each of these factors, as well as additional microbial determinants related to pathogenesis as they are identified in the future.

In addition to H. pylori constituents, host factors also influence gastric carcinogenesis. IL-1β and TNF-α are proinflammatory cytokines with potent acid-suppressive properties and high-expression polymorphisms within the human IL-1β and TNF-α gene promoters heighten the risk for gastric adenocarcinoma among H. pylori-infected subjects (38, 39). Our data indicate that, independent of inflammation, activation of β-catenin by H. pylori also may be an important early event that precedes malignant transformation, and these results are similar to those in a previous publication that investigated β-catenin localization in Mongolian gerbils infected with H. pylori and treated with the carcinogen N-methyl-N-nitrosourea (40). In that study, nuclear β-catenin was identified in only 1 of 45 gastric adenocarcinomas, providing further evidence that H. pylori dysregulates signaling pathways with carcinogenic potential early after infection.

In conclusion, in vivo adaptation engendered an H. pylori strain with the ability to rapidly induce gastric neoplasia, providing a system that offers unique advantages for investigations focused on host-microbial interactions related to gastric carcinogenesis. That β-catenin is selectively activated in a CagA-dependent manner has not only provided insights into the functional significance of β-catenin activation in regulating H. pylori carcinogenic responses but also may provide mechanistic insights into other β-catenin-mediated epithelial responses that are activated by bacteria, such as growth and differentiation of gastrointestinal cells.