Helicobacter pylori (H. pylori) is an etiologic agent of gastritis and peptic ulcers.1 Furthermore, H. pylori strains carrying the cagA gene elicit severe mucosal damage and are closely associated with severe atrophic gastritis and gastric adenocarcinoma.2, 3 The cagA gene encodes a 120–145-kilodalton (kDa) H. pylori virulence factor, CagA, that contains multiple Glu-Pro-Ile-Tyr-Ala (EPIYA) motifs in the C-terminal EPIYA-repeat region. CagA is characterized by structural diversity in the EPIYA-repeat region, which was generated through homologous recombination within the 3′ region of the cagA gene.4, 5, 6, 7 As a result, CagA is subdivided into 2 major species, one carried by H. pylori strains circulating in Europe, Africa, America and Australia (Western CagA) and the other carried by H. pylori strains circulating in East Asian countries, including Japan, Korea and China (East Asian CagA).
The CagA protein is directly delivered from H. pylori to the cytoplasm of the H. pylori-attached gastric epithelial cell via the bacterial type IV secretion system.8 The delivered CagA localizes to the inner surface of plasma membrane, where it undergoes tyrosine phosphorylation at the EPIYA motifs by Src family kinases.9, 10 Upon tyrosine phosphorylation, CagA acquires the ability to bind to the Src homology 2 (SH2) domain containing protein tyrosine phosphatase SHP-2 and deregulates the phosphatase activity, which in turn dephosphorylates and inhibits focal adhesion kinase (FAK), resulting in elevated cell motility with a highly elongated cell shape (hummingbird phenotype).11, 12 CagA-deregulated SHP-2 also provokes sustained Erk MAP kinase activation.13 Given that gain-of-function mutations in SHP-2 are associated with various human malignancies, CagA-deregulated SHP-2 has been suspected to play an important role in the development of gastric carcinoma.14, 15, 16
In addition to its phosphorylation-dependent activities, CagA perturbs cell functions in a tyrosine phosphorylation-independent manner. For instance, CagA has been shown to disrupt apical junctions and impair cell–cell contacts.17 CagA also activates the nuclear factor of activated T cells (NFAT) by stimulating calcineurin in a phosphorylation-independent manner.18 More recently, CagA has been reported to deregulate the β-catenin signal by destabilizing the E-cadherin/β-catenin complex.19 Aberrant activation of β-catenin has been shown to be associated with the oncogenic potential of cagA-positive H. pylori in Mongolian gerbils.20
Intestinal metaplasia, a transdifferentiation of gastric epithelial cells to an intestinal phenotype, has been recognized as a premalignant mucosal lesion from which intestinal-type gastric adenocarcinoma arises.21, 22 Since deregulated β-catenin is involved in transdifferentiation of various epithelial cell types,23, 24, 25 aberrant activation of the β-catenin signal by CagA is suspected to play a role in the development of intestinal metaplasia. In this study, we found that CagA deregulates β-catenin via the EPIYA-repeat region and we delineated the sequences that are involved in β-catenin deregulation in Western CagA and East Asian CagA species. The identified CagA sequences contain a conserved 16-amino-acid sequence, which we recently identified as the CagA-multimerization (CM) sequence.26 Furthermore, the CM sequence of CagA was found to be responsible for deregulation of the β-catenin signal. We discuss the role of CagA multimerization in β-catenin deregulation by CagA.
Csk, C-terminal Src kinase; DAPI, 4′,6-diamidino-2-phenylindole dihydrochloride n-hydrate; FAK, focal adhesion kinase; H. pylori, Helicobacter pylori; HA, hemagglutinin; NFAT, nuclear factor of activated T cells; SH2, Src homology 2; WT, wild-type.
Material and methods
Expression vector constructs
Western and East Asian cagA genes were isolated from H. pylori strains NCTC11637 and F32, respectively. The cagA mutants were generated from NCTC11637 or F32-derived cagA by site-directed mutagenesis and, after addition of a sequence that encodes the C-terminal hemagglutinin (HA) epitope, were cloned into the pSP65SRα mammalian expression vector. TOPtkLuciferase and FOPtkLuciferase reporter constructs carry 3 copies of wild-type Tcf/Lef-binding site and mutated Tcf/Lef-binding site, respectively.19 The pGL2-944 cyclin D1-luciferase reporter was made from the pGL2 luciferase reporter plasmid by introducing the human cyclin D1 promoter sequence (residues −944 to −23).27 Internal control pRL-TK Renilla luciferase reporter construct was purchased from Promega.
H. pylori strain and the production of isogenic cagA mutant strains
H. pylori NCTC11637 strain is a Western standard strain carrying ABCCC-type CagA. Isogenic cagA mutant strains were generated from the H. pylori standard strain NCTC11637 by natural transformation.28 Briefly, the mutant cagA gene, after addition of a sequence encoding C-terminal HA epitope, was cloned into pBluescript II SK (+). A kanamycin-resistance gene cassette from pENTR3C (Invitrogen) was inserted into the 3′ downstream of the mutant cagA gene. NCTC11637-derived HP0548 sequence and a portion of HP0549 sequence, which locate downstream of cagA, were amplified by polymerase chain reaction (PCR) using a primer pair 5′-AGCCAATGCATTTTCTACAGG-3′ (forward) and 5′-CACTAAAGACCCCACCAC-3′ (reverse), which was designed based on the sequence of strain 26695.29 The amplified 1.2-kb fragment was then inserted into the 3′ downstream of the kanamycin-resistance gene cassette. The recombinant plasmids were then used to introduce cagA mutations in the chromosome of wild-type H. pylori NCTC11637 strain according to a standard procedure.28 Correct integration of the mutant cagA genes was confirmed by genomic sequencing.
Cell culture and transfection
MKN28 human gastric epithelial cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS). Transfections were performed with Lipofectamine reagent and PLUS reagent (Invitrogen) according to the manufacturer's protocol.
H. pylori infection
MKN28 cells were infected with H. pylori NCTC11637 strain or its isogenic mutant strain for 8 hr at a multiplicity of infection of 100 as previously described.19
Anti-HA polyclonal antibody (Y-11) and anti-SHP-2 polyclonal antibody (C-18) were purchased from Santa Cruz Biotechnology. Anti-phosphotyrosine monoclonal antibody (4G10) was purchased from Upstate Biotechnology. Anti-β-catenin monoclonal antibody was purchased from Transduction Laboratories. Anti-CagA antibody was purchased from AUSTRAL Biologicals.
MKN28 cells were harvested at 24 hr after transfection and lysed in lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Brij-35) containing 2 mM Na3VO4, 2 mM PMSF, 10 μg/ml leupeptin, 10 μg/ml trypsin inhibitor, and 10 μg/ml aprotinin. Total cell lysates were subjected to sodium dodecyl sulfate (SDS)-8% polyacrylamide gel electrophoresis (PAGE). Proteins transferred to polyvinylidene difluoride (PVDF) membrane filters (Millipore) were incubated in primary antibodies in phosphate-buffered saline (PBS) containing 0.6% Tween-20, 1% skim milk and 0.1% bovine serum albumin (BSA) and then visualized by using Western Blot Chemiluminescence Reagent (PerkinElmer Life Sciences).
Luciferase reporter assays
MKN28 cells (2 × 105) were transfected with a total 1.25 μg of various combinations of plasmids: 0.2 μg of reporter plasmid (TOPtkLuciferase, FOPtkLuciferase, or pGL2-944 cyclinD1-luciferase), 0.05 μg of internal control pRL-TK (Promega) and 1 μg of expression vector. pRL-TK Renilla luciferase reporter construct was cotransfected in each sample to normalize for transfection efficiency. Luciferase assays were performed 24 hr after transfection using the Dual-Luciferase Reporter assay system (Promega) according to the manufacturer's protocol.
Cells fixed with Mildform 10 N (Wako) were treated with primary antibodies. The primary antibodies were then localized by Alexa Fluor 546-conjugated anti-rabbit IgG antibody or Alexa Fluor 488-conjugated anti-mouse IgG antibody (Invitrogen). The nuclei were stained with 4′,6-diamidino-2-phenylindole dihydrochloride n-hydrate (DAPI) (Wako). Images were acquired using a confocal microscope system (Fluoview, OLYMPUS).
Involvement of the EPIYA-repeat region in β-catenin deregulation by CagA
To elucidate the molecular mechanism underlying deregulation of the β-catenin signal by CagA, we carried out experiments to determine the CagA region that is responsible for translocalization of β-catenin from the membrane to the cytoplasm/nucleus. Since the EPIYA-repeat region plays a crucial role in a variety of CagA activities, we first examined a CagA mutant in which the entire EPIYA-repeat region (residues 869–1086) was deleted from ABCCC-type Western CagA (CagA-ΔABCCC) (Fig. 1a). When expressed in MKN28 human gastric epithelial cells, CagA-ΔABCCC neither associated with the plasma membrane nor activated the β-catenin signal as shown by the results of reporter luciferase assays (Figs. 1b and 1c). The results indicate that the EPIYA-repeat region plays a role in the CagA activity. To substantiate the results, we expressed another CagA mutant, CagA-N, consisting of the N-terminal 612 amino acids of CagA (Fig. 1a) and found that CagA-N does not associate with the plasma membrane and does not activate the β-catenin signal (Figs. 1b and 1c). On the basis of these results, we concluded that the EPIYA-repeat region of CagA is indispensable for the translocalization of β-catenin from the membrane to the cytoplasm/nucleus and subsequent activation of the β-catenin signal.
Deregulation of β-catenin by Western and East Asian CagA proteins
The EPIYA-repeat region of Western CagA contains EPIYA-A, EPIYA-B and variable numbers (usually 1–3) of Western CagA-specific EPIYA-C segments, each of which possesses a single EPIYA site,16, 30 whereas that of East Asian CagA contains EPIYA-A, EPIYA-B and East Asian CagA-specific EPIYA-D segments. To narrow down the sequence within the EPIYA-repeat region that is required for deregulation of the β-catenin signal by CagA, we generated a series of EPIYA-segment mutants from CagA as depicted in Figure 2a. We also generated a phosphorylation-resistant mutant of ABD-type East Asian CagA by replacing all of the EPIYA motifs with nonphosphorylatable EPIAA sequences. After C-terminal HA-epitope tagging, these CagA mutants were transiently expressed in MKN28 cells. As reported previously, the level of CagA tyrosine phosphorylation was found to be proportional to the number of EPIYA-C segments in Western CagA species.30 EPIYA-A and EPIYA-B segments were also tyrosine-phosphorylated but less efficiently (Fig. 2b). The level of tyrosine phosphorylation of ABD-type East Asian CagA (CagA-ABD) was almost the same as that of ABC-type Western CagA (CagA-ABC), whereas, as expected, the East Asian CagA-abd mutant did not undergo tyrosine phosphorylation as expected (Fig. 2b).
Using these mutants, we examined the ability of CagA to deregulate the β-catenin signal. To do so, MKN28 cells were transfected with an expression vector for each of the CagA mutants together with the TOPtkLuciferase reporter plasmid or control FOPtkLuciferase reporter plasmid. The results of the experiment showed that both CagA-ABC and CagA-ABD proteins were capable of activating the β-catenin reporter (Fig. 2c). Since the CagA proteins were expressed at the same levels in these transfected cells, the results indicate that CagA-ABC and CagA-ABD have comparable activities to deregulate the β-catenin signal. Furthermore, an increase in the number of EPIYA-C segments in Western CagA did not alter the ability of Western CagA to activate β-catenin. Therefore, as opposed to the CagA-SHP-2 interaction, the presence of multiple EPIYA-C segments in CagA does not influence the magnitude of β-catenin deregulation by Western CagA. In contrast to CagA-ABC or CagA-ABD, CagA-AB hardly activated the β-catenin reporter, ruling out the possibility of a role of EPIYA-A and EPIYA-B segments in the activation of β-catenin (Fig. 2c). Since CagA-AB was still capable of associating with the plasma membrane, the observations argue against the idea that destabilization of the E-cadherin/β-catenin complex is simply induced by the membrane tethering of CagA.31 Collectively, the results suggest that deregulation of the β-catenin signal is mediated through the C-terminal part of the EPIYA-repeat region that contains the EPIYA-C segment in the case of Western CagA (residues 1009–1086 of CagA-ABCCC) or the sequence that contains the EPIYA-D segment in the case of East Asian CagA (residues 908–1012 of CagA-ABD). We also noted that the phosphorylation-resistant CagA-abd mutant was still capable of deregulating the β-catenin signal (Fig. 2d). Thus, as in the case of Western CagA,19 the ability of East Asian CagA to activate the β-catenin signal is independent of EPIYA tyrosine phosphorylation.
We confirmed the ability of CagA to activate the β-catenin signal with the use of the promoter sequence of the cyclin D1 gene, a well-recognized β-catenin target gene.32 As shown in Figure 3, cotransfection of the cyclin D1 promoter luciferase reporter together with the CagA expression vector in MKN28 cells faithfully reproduced the results obtained using the TOPtkLuciferase reporter. On the basis of these observations, we concluded that deregulation of β-catenin is mediated by the CagA sequence that spans the Western EPIYA-C segment or East Asian EPIYA-D segment in a tyrosine phosphorylation-independent manner.
Deregulation of β-catenin by CagA correlates with β-catenin translocalization by CagA
CagA induces translocalization of membranous β-catenin to the cytoplasm/nucleus,19 thereby causing deregulation of β-catenin. We thus investigated whether activation of the β-catenin signal by the CagA mutants shown in Figure 2a was due to translocalization of β-catenin. To this end, MKN28 cells transfected with each of the CagA mutants were fixed and stained with an anti-β-catenin antibody (Fig. 4). As a positive control, MKN28 cells were transiently transfected with the expression vector for a constitutively active β-catenin (S33Y). In cells expressing AB-type CagA, which cannot activate the β-catenin signal (Fig. 2c), β-catenin was strictly localized to the membrane as was the case in control cells transfected with an empty vector. In contrast, β-catenin was detectable in the cytoplasm and nucleus in cells expressing CagA-ABC or CagA-ABD (Fig. 4). On the basis of these observations, we concluded that activation of the β-catenin signal by CagA is mediated by the translocalization of β-catenin from the membrane to the cytoplasm/nucleus.
The identified CagA sequence that spans EPIYA-C or EPIYA-D segment is sufficient for translocalization and deregulation of β-catenin
The earlier-described observations indicated that residues 1009–1086 of Western CagA-ABCCC or residues 908–1012 of East Asian CagA-ABD are necessary for deregulation of β-catenin by CagA. We therefore investigated whether the identified CagA sequence is sufficient to induce translocalization of β-catenin and subsequent activation of the β-catenin signal. To this end, EPIYA-A and EPIYA-B segments were deleted from CagA-ABCCC or CagA-ABD, and the resulting CagA mutants, CagA-C (deletion of residues 869–1008) and CagA-D (deletion of 856–907) (Fig. 5a), were expressed together with the luciferase reporter plasmid in MKN28 cells. The results of the reporter assay showed that both CagA-C and CagA-D mutants are capable of activating the β-catenin signal (Figs. 5b and 5c). Furthermore, immunocytochemical analysis of cells expressing CagA-C or CagA-D with an anti-β-catenin antibody revealed that their expression induces translocalization of β-catenin from the membrane to the cytoplasm/nucleus.
To consolidate the observations obtained by transfection of cagA genes in gastric epithelial cells, we next generated H. pylori NCTC11637-derived isogenic strains that carry mutant cagA genes directing CagA-C and CagA-D (Fig. 6a). Infection of MKN28 cells with these isogenic strains elicited cytoplasmic/nuclear localization of β-catenin whereas infection with NCTC11637-derived ΔcagA strain, which lacks the cagA gene, failed to induce translocalization of β-catenin (Fig. 6b).
From these observations, we concluded that residues 1009–1086 of ABCCC-type Western CagA or residues 908–1012 of ABD-type East Asian CagA are necessary and sufficient for translocalization and deregulation of β-catenin.
Involvement of the CM sequence in β-catenin deregulation by CagA
The identified residues 1009–1086 in ABCCC CagA and residues 908–1012 in ABD CagA are significantly diverged in their amino acid sequences. However, they contain a conserved 16-amino-acid stretch, which we recently identified as the CagA multimerization sequence (CM sequence) (Fig. 7a).26 The CagA CM sequence is required for multimerization, most probably dimerization, of CagA. Given this, we next examined the potential role of the CM sequence in deregulation of the β-catenin signal by CagA. CagA-ΔABCC and CagA-D, both of which contains a single CM sequence, still retained the ability to activate the β-catenin signal (Figs. 7b and 7d). Accordingly, we deleted the 16-amino-acid CM sequence from CagA-ΔABCC and CagA-D to make CagA-ΔABCC-16AA and CagA-D-16AA, respectively (Figs. 7b and 7c). As shown in Figure 7d, specific deletion of the CM sequence from CagA abolished the ability of CagA to activate the β-catenin signal. On the basis of these observations, we concluded that the CM sequence of CagA is necessary for deregulation of the β-catenin signal.
In the present study, we demonstrated that CagA-mediated translocalization of β-catenin from the membrane to the cytoplasm/nucleus, which causes aberrant activation of the β-catenin signal, requires the sequence that spans the EPIYA-C segment of Western CagA (residues 1009–1086 of ABCCC CagA) or the sequence that spans the EPIYA-D segment of East Asian CagA (residues 908–1012 of ABD CagA), and is mediated by the 16-amino-acid CM sequence that is conserved within the identified CagA sequences. Deregulation of β-catenin by CagA is independent of EPIYA tyrosine phosphorylation as well as the structural polymorphism between Western and East Asian CagA species.
Resent studies demonstrated that CagA activates the β-catenin signal in human epithelial cells.19, 20, 33, 34 Franco et al. showed that oncogenic H. pylori causes the activation of the β-catenin signal in the rodent infection model.20 Our present study further reveals the CagA region that is responsible for activation of the β-catenin signal. In contrast to our results as well as those of others, Bebb etal. reported that infection of HT29 colon carcinoma cells with H. pylori (either cagA-positive or cagA-negative strain) did not induce translocalization of the membranous β-catenin to the cytoplasm/nucleus.35 However, their work did not show actual translocation of CagA into HT29 cells, raising the possibility that H. pylori failed to deliver sufficient amounts of CagA into the cells in their infection experiment.
Aberrant activation of β-catenin plays a critical role in the development of gastrointestinal cancer. Mutation in the β-catenin, APC or Axin gene, which causes deregulation of the β-catenin signal, is frequently associated with colorectal carcinoma, pancreatic carcinoma and hepatocellular carcinoma.36 On the other hand, genetic changes that affect the β-catenin signal are not common in gastric cancer except hereditary gastric carcinoma, which is caused by E-cadherin mutation.37 At first glance, this fact indicates that the development of gastric cancer does not require an aberrant β-catenin signal. However, our finding showing that the β-catenin signal is functionally deregulated by H. pylori CagA provides an opportunity to reconsider the above idea.19 Indeed, the oncogenic potential of individual H. pylori has been shown to be correlated with its ability to deregulate β-catenin in an animal infection model.20 Thus, deregulation of β-catenin by CagA may be a functional equivalent of a mutation in a gene that encodes a component of the β-catenin signal. Of special interest is the fact that H. pylori-associated gastric adenocarcinoma generally arises from intestinal metaplasia, a transdifferentiation of gastric epithelial cells to an intestinal phenotype.38, 39, 40 Given that the CagA-activated β-catenin signal induces aberrant expression of intestinal-specific molecules such as MUC2,19 CagA-mediated β-catenin activation may play an important role in the development of precancerous intestinal metaplasia.
We previously reported that interaction of CagA with E-cadherin destabilizes the E-cadherin/β-catenin complex and thereby releases β-catenin from the membrane to the cytoplasm/nucleus.19 It is thus reasonable to assume that the CM sequence of CagA, which is required for β-catenin deregulation, mediates a physical association of CagA with E-cadherin. Despite much effort, however, we were not able to demonstrate a direct interaction between CagA and E-cadherin by an in vitro binding assay, which easily detected E-cadherin/β-catenin complexes (data not shown). The result argues against the idea that CagA and β-catenin competitively interact with the cytoplasmic domain of E-cadherin, while raising the possibility that the CagA-E-cadherin interaction is mediated by another cellular protein or protein complex that indirectly connects CagA and E-cadherin in mammalian cells and thereby destabilizes the E-cadherin/β-catenin complex. Intriguingly, we recently found that CagA interacts with PAR1/MARK2 polarity-regulating kinase via the CM sequence.41 It is therefore possible that the CagA-PAR1 interaction is involved in the destabilization of the E-cadherin-β-catenin complex by CagA. Consistent with this idea, PAR1 was found to be coprecipitated with CagA and E-cadherin in MKN28 cells (our unpublished observation). Potential role of CagA-PAR1 interaction in the aberrant activation of the β-catenin signal by CagA warrants further investigation.
Gain-of-function mutations of SHP-2 have recently been reported in human malignancies, indicating that SHP-2 is a bona fide oncoprotein.14, 15 SHP-2 binds to the EPIYA-C and EPIYA-D segments of Western CagA and East Asian CagA, respectively, in a manner that is dependent on EPIYA tyrosine phosphorylation.30 In the case of CagA-SHP-2 interaction, the EPIYA-D segment exhibits stronger activity to bind and deregulate SHP-2 than does the EPIYA-C segment.42 Consistently, East Asian CagA-positive H. pylori strains are more virulent and more closely associated with severe atrophic gastritis and gastric adenocarcinoma than are those carrying Western CagA.43 Among Western CagA-positive H. pylori strains, those having a greater number of EPIYA-C segments are more frequently isolated from patients with gastric carcinoma.44 Thus, the ability of individual CagA to perturb SHP-2 correlates with gastric carcinoma. On the other hand, the degree of β-catenin deregulation by CagA does not significantly differ between ABD-type East Asian CagA and ABCCC-type Western CagA despite the fact that ABD-type CagA possesses a single CM sequence while ABCCC-type CagA possesses multiple CM sequences. Also, the number of EPIYA-C segments, each of which contains a single CM sequence, does not affect the degree of β-catenin deregulation among distinct Western CagA species. Accordingly, in contrast to the case of SHP-2, structural diversity in the EPIYA-repeat region does not influence the degree of individual CagA to disturb the β-catenin signal. The observations also suggest that disruption of the E-cadherin/β-catenin complex is mediated by a single CM sequence even if CagA possesses multiple CM sequences. Hence, whereas the CM sequence is required for the activation of the β-catenin signal, the number of the CM sequences does not play a role in determining the magnitude of β-catenin deregulation by distinct CagA species.
Deregulation of β-catenin within a limited range of magnitude may be an inherent activity of CagA as an H. pylori virulence factor, which promotes the long-term colonization of H. pylori in the stomach while inducing intestinal transdifferentiation known as precancerous intestinal metaplasia. Thus, CagA-mediated deregulation of β-catenin and SHP-2, which simultaneously activates distinct intracellular signaling pathways, may play crucial roles in the multistep process of gastric carcinogenesis.