Helicobacter pylori CagA induces Ras-independent morphogenetic response through SHP-2 recruitment and activation.

The CagA protein of Helicobacter pylori, which is injected from the bacteria into bacteria-attached gastric epithelial cells, is associated with gastric carcinoma. CagA is tyrosine-phosphorylated by Src family kinases, binds the SH2 domain-containing SHP-2 phosphatase in a tyrosine phosphorylation-dependent manner, and deregulates its enzymatic activity. We established AGS human gastric epithelial cells that inducibly express wild-type or a phosphorylation-resistant CagA, in which tyrosine residues constituting the EPIYA motifs were substituted with alanines. Upon induction, wild-type CagA, but not the mutant CagA, elicited strong elongation of cell shape, termed the "hummingbird" phenotype. Time-lapse video microscopic analysis revealed that the CagA-expressing cells exhibited a marked increase in cell motility with successive rounds of elongation-contraction processes. Inhibition of CagA phosphorylation by an Src kinase inhibitor, PP2, or knockdown of SHP-2 expression by small interference RNA (siRNA) abolished the CagA-mediated hummingbird phenotype. The morphogenetic activity of CagA also required Erk MAPK but was independent of Ras or Grb2. In AGS cells, CagA prolonged duration of Erk activation in response to serum stimulation. Conversely, inhibition of SHP-2 expression by siRNA abolished the sustained Erk activation. Thus, SHP-2 acts as a positive regulator of Erk activity in AGS cells. These results indicate that SHP-2 is involved in the Ras-independent modification of Erk signals that is necessary for the morphogenetic activity of CagA. Our work therefore suggests a key role of SHP-2 in the pathological activity of H. pylori virulence factor CagA.

H. pylori isolates have an EPIYA-containing 34-amino acid sequence that variably duplicates. As a result, that the number of EPIYA motifs in CagA varies among different Western H. pylori isolates (8,25,26). Intriguingly, Western CagA having a greater number of repeatable EPIYA motifs exhibits stronger activity to bind SHP-2 and to induce the hummingbird phenotype (24). In contrast, prevalent East Asian CagA proteins possess a significantly diverged EPIYA-containing sequence (26,27), which binds SHP-2 with higher affinity than those of most if not all Western CagA species (24).
SHP-2 and its orthologues, Drosophila Corkscrew and Caenorhabditis elegans PTP-2, play a positive role in transduction of signals relayed from receptor-type protein-tyrosine kinases (28 -32). The function of SHP-2 in signal transduction requires its phosphatase activity. Also, SHP-2 plays an important role in the regulation of cell morphology and motility (33)(34)(35)(36). Recently, Grb2 and c-Met have been reported to form physical complexes with CagA, although their interactions are independent of CagA phosphorylation (37,38). These observations collectively suggest that the bacterial protein functions as a docking/scaffolding protein that recruits and deregulates multiple signaling molecules in the injected host cells (39).
We established human gastric epithelial cells in which wildtype or a phosphorylation-resistant CagA was inducibly expressed under the control of a tetracycline-regulated (tet-on) system. Using the CagA-inducible cells, we found that tyrosine phosphorylation of CagA is an essential prerequisite for induction of the hummingbird phenotype as well as the cell-scattering phenotype. The morphogenetic activity of CagA is dependent on the activity of Erk MAPK but not on that of Ras or phosphatidylinositol 3-kinase (PI3K). We obtained evidence that SHP-2 is involved in the regulation of Erk MAPK activity by CagA.
Expression Vectors-Mammalian expression vectors for HA-tagged wild-type CagA and its phosphorylation-resistant (PR) mutant have been described previously (22). PR CagA-YF was generated from HAtagged wild-type CagA by substituting tyrosine residues that constitute 5ϫ EPIYA motifs with phenylalanine residues using a Chameleon site-directed mutagenesis kit (Stratagene), and the DNA fragment encoding the mutant was cloned into a pSP65SR␣ mammalian expression vector. Myc-tagged SHP-2 and its constitutively active mutant, myr-SHP-2⌬SH2, have been described previously (22). To generate siRNAinsensitive SHP-2, SHP-2RR-Myc, from Myc-tagged SHP-2, a nucleotide stretch corresponding to the SHP-2 siRNA was replaced with the nucleotides 5Ј-GGAGTATGGT-3Ј in the cDNA, and the resulting cDNA was cloned into pSP65SR␣. Myc-tagged WT Grb2 was cloned into pSP65SR␣. To generate SH3 domain-inactivated Grb2 mutant, Grb2 SH3mut-Myc, Trp-36 (W36) and Trp-193 (W193) were replaced by lysine residues. The mutant cDNA was cloned into the pSP65SR␣. H-Ras expression vectors were constructed by inserting cDNA encoding a dominant-negative H-Ras Asn17 or a constitutively active H-Ras Val12 into pSP65SR␣. To construct SHP-2-specific siRNA expression vector, two oligonucleotides, which, respectively, contained SHP-2-specific targeting sequence (5Ј-GAT CTC CGA ATA TGG CGT CAT GCG TGT TCA AGA GAC ACG CAT GAC GCC ATA TTC-3Ј) and reverse complementary sequence of the former oligonucleotide (5Ј-GAA TAT GGC GTC ATG CGT GTC TCT TGA ACA CGC ATG ACG CCA TAT TCG GAG ATC-3Ј), were synthesized. After an annealing step, the oligonucleotide was inserted into pSUPER vector (40), and the cloned vector was used in the following experiments as pSUPER-SHP-2.
Establishment of Stable Transfectants-Human AGS gastric epithelial cells were cultured in RPMI 1640/10% fetal calf serum (FCS). To establish the tet-on system, AGS cells (1 ϫ 10 6 cells/100-mm dish) were first transfected with 20 g of pTet-On (Clontech), which expresses reverse tetracycline-dependent transactivator (rtTA) regulator protein, by using the calcium phosphate method. At 12 h after transfection, cells were cultured in RPMI 1640/10% FCS supplemented with 500 g/ml G418 (Clontech). Cells resistant to G418 were selected and single-cellcloned by limiting dilution. Clone T71-2 showing the highest rtTA activity was subjected to a second round of transfection with 20 g of the expression plasmid, in which the gene encoding HA-tagged wildtype CagA or HA-tagged phosphorylation-resistant CagA were driven by the tetracycline-responsive elements (22). The transfected cells were selected by 150 g/ml hygromycin B (Sigma). Drug-resistant colonies were then picked up and were examined for induction of CagA. Positive cells were single-cell cloned by limiting dilution. To induce CagA, cells were incubated with RPMI 1640/10% FCS containing doxycycline (Dox) at a final concentration of 5 g/ml. In experiments using kinase inhibitors, cells were incubated with 5 M PP2 for 2 h or 20 M U0126 for 5 h before harvest. Time-dependent change of cell morphology was monitored by time-lapse video microscopy (Olympus IX71 Bioimaging system). After 48-h induction with Dox, cells were subjected to live cell time-lapse video microscopic analysis. To isolate stable SHP-2-knockdown cells, AGS cells were co-transfected with pSUPER-SHP-2 and pBabePuro. At 12-h post-transfection, transfected cells were primarily selected in RPMI 1640/10% FCS containing 0.4 g/ml puromycin (Sigma). The cloned cells were then examined for their SHP-2 expression by anti-SHP-2 immunoblotting. One of the clones, G11, in which expression of SHP-2 was strongly repressed, was used for experiment as a stable SHP-2-knockdown cell.
Cell Scatter Assay-Cells were seeded at 1 ϫ 10 3 cells in 60-mm dishes and were incubated in RPMI 1640/10% FCS for 48 h in the absence of Dox. After the additional incubation with or without 5 g/ml Dox for 60 h, colonies were observed by microscopy.
Transient Transfection and Infection-AGS cells and monkey COS-7 cells were cultured in RPMI 1640 medium and Dulbecco's modified Eagle's medium, respectively, supplemented with 10% FCS. For transient transfection, 30 g of plasmids was transfected into AGS cells (1.8 ϫ 10 6 cells/100-mm dish) by using 30 l of LipofectAMINE 2000 reagent (Invitrogen) according to the manufacturer's protocol. COS-7 cells (1.5 ϫ 10 6 cells/100-mm dish) were transfected with 20 g of plasmids by using the calcium phosphate method as described previously (22). Morphology of the AGS cells was examined at 17 h after transfection. Cells showing the hummingbird phenotype were counted in five different fields in three dishes (the area of one field ϭ 0.25 mm 2 ). In experiments using U0126 or U0124, cells were incubated with the reagent at various concentrations for 1 h before microscopic analysis. In experiments using other kinase inhibitors, cells were incubated with 5 M PP2 for 2 h, 20 M SB203580 for 1 h, 20 M SP600125 for 1 h, or 40 M LY294002 for 2 h before analysis. For H. pylori infection with AGS cells, standard strain NCTC11637 or its isogenic strain lacking cagA gene was passaged on 5% sheep blood agar (Nippon BD Biosciences) by incubation in an atmosphere consisting of 5% O 2 , 15% CO 2 , and 80% N 2 for 2-4 days at 37°C. Bacteria were then cultured in Brucella broth (Difco) supplemented with 5% FCS under the same conditions for 12-24 h at 37°C with agitation. AGS cells (2 ϫ 10 6 cells/100-mm dish) cultured at 12 h post seeding were infected with H. pylori at a multiplicity of infection of 50 for 5 h.
Immunoprecipitation and Immunoblotting-AGS cells and COS-7 cells were harvested at 36 h after transfection and lysed in lysis buffer as described previously (22). Cell lysates were treated with the appropriate antibody, and immune complexes were trapped on protein A-or protein G-Sepharose beads. Total cell lysates and immunoprecipitates were subjected to SDS-8% PAGE except experiments to detect Erk1 (12.5% PAGE), Grb2 (13.5% PAGE), and Ras mutants (13.5% PAGE). Proteins transferred to a polyvinylidene difluoride membrane filter (Millipore) were soaked in solutions of primary antibodies and then visualized using Western blot chemiluminescence reagent (PerkinElmer Life Sciences).
Analysis of Erk Phosphorylation-AGS or G11 cells (2 ϫ 10 5 ) were seeded in 60-mm dishes and were incubated in the medium containing 10% FCS for 24 h. In experiments using W1804 cells, the cells were seeded under the same condition as described above and were incubated in the medium with or without 5 g/ml Dox. The cells were subsequently serum-starved for 48 h and then released from serum starvation by adding medium with 10% FCS. The cells were lysed with SDS sample buffer at the indicated times. The samples were subjected to Western blotting analysis with anti-ERK1 and anti-phospho-ERK antibodies. Intensities of chemiluminescence on the immunoblotted filter were quantitated using a luminescent image analyzer LAS1000 (Fujifilm).

Establishment of Gastric Epithelial Cells That Inducibly Ex-
press CagA-Infection of cagA-positive H. pylori in human AGS gastric epithelial cells induces elongation of cell shape, referred to as the hummingbird phenotype (15). We previously demonstrated that transient expression of the cagA gene in AGS human gastric epithelial cells also elicits the hummingbird phenotype (22). To elucidate the mechanism by which CagA induces morphological change, we generated a series of AGSderived, stable transfectant clones that express HA-tagged, wild-type CagA, derived from H. pylori strain NCTC11637, or HA-tagged, phosphorylation-resistant CagA (PR CagA), in which the tyrosine residues constituting the five EPIYA motifs were substituted with alanines, under the control of a tetracycline-inducible (tet-on) gene expression system (Fig. 1A). Among these clones, wild-type cagA clone W1804 and PR cagA clone C065 were arbitrarily chosen for further experiments. In this tet-on system, wild-type and PR CagA proteins were detectable by Western blotting as early as 6 h after induction with a tetracycline derivative, doxycycline (Dox). In W1804 cells, the level of wild-type CagA oscillated after induction, showing peaks at 36 and 60 h. We reproducibly obtained the same results in three independent experiments. In the case of C065 cells expressing PR CagA, the protein level peaked at ϳ36 h after induction and was maintained at the peak level thereafter (Fig. 1B). Hence, a specific decrease in wild-type CagA level at 48 h after induction may be a phenomenon related to tyrosine phosphorylation of CagA. The induced wild-type CagA underwent tyrosine phosphorylation and formed a physical complex with SHP-2 in AGS cells (Fig. 1, A-C). In contrast, no tyrosine phosphorylation of PR CagA was detectable in the clone C065 ( Fig. 1, A and B). As a control experiment, AGS cells were infected with H. pylori strain NCTC11637. At 5 h after infection, cells were harvested and lysates prepared were subjected to immunoprecipitation with anti-CagA. As was the case with transfected CagA, CagA injected by the bacteria formed a physical complex with SHP-2 in AGS cells, excluding the possibility that SHP-2 binds transfected CagA but not injected CagA (Fig. 1D).
CagA Provokes Morphological Change-Induction of wildtype CagA but not PR CagA was followed by a change in cell morphology that was indistinguishable from the hummingbird phenotype induced by transient CagA expression in AGS cells (Fig. 1E). Here we designate "hummingbird cells" as those having one or more protrusions, the length of the protrusion being more than 2-fold of cell diameter. Hummingbird cells were observed as early as 24 h after induction of wild-type CagA. At 48 h of wild-type CagA induction, ϳ7-8% of the cells exhibited the hummingbird phenotype. Further incubation of cells with Dox, however, did not increase the number of hum-mingbird cells. The weak phenotype may be primarily due to the level of CagA induced in the AGS stable transfectant, which was significantly lower than that expressed in AGS cells by transient transfection of cagA gene or cagA-positive H. pylori infection. 2 Establishment of AGS cells that conditionally express CagA enabled us to study time-dependent changes in cell morphology following CagA induction. Using live cell time-lapse microscopy, we found that induction of CagA caused a marked increase in cell migration: a fraction of cells initiated to move from one place to another while showing successive elongation (i.e. the hummingbird phenotype) and contraction of cell shape (Fig. 2, arrows). After several rounds of the elongation-contraction processes, some cells became detached from the dish. The observed dynamic feature of cell shape may also explain why only a fraction of the cells exhibited the hummingbird phenotype despite continuous expression of CagA.
Essential Role of Tyrosine Phosphorylation of CagA in Both the Morphogenetic Change and Cell Scattering-The results of previous studies have indicated that the activity of CagA to induce the hummingbird phenotype is dependent on its tyrosine phosphorylation (18, 20 -22, 41). On the other hand, a recent report indicated that CagA is capable of inducing cell scattering regardless of its phosphorylation status (37). This prompted us to examine whether CagA phosphorylation is indeed an essential requirement for cell elongation by CagA.
CagA is phosphorylated by the Src family of protein-tyrosine kinases (20,21). Treatment of cagA-positive H. pylori-infected AGS cells with PP2, a specific inhibitor of Src family kinases, suppressed cell scattering (20). Accordingly, we examined the effect of PP2 on the induction of hummingbird cells by CagA per se. Treatment of the wild-type CagA-expressing AGS cells with PP2, which did not affect levels of induced CagA, potently inhibited tyrosine phosphorylation of CagA and strongly suppressed induction of hummingbird cells (Fig. 3A). The same results were also obtained by treatment of AGS cells transiently expressing cagA gene with PP2 (Fig. 3B). As a control, we expressed the membrane-targeting, constitutively active SHP-2, which induces a hummingbird-like morphology in AGS cells (22) and found that PP2 did not block the morphological change induced by the mutant SHP-2. The result excludes the possibility that PP2 inhibited the action of cell machinery generally involved in cell elongation (Fig. 3C). Thus, tyrosine phosphorylation of CagA is an essential prerequisite for the morphological transformation of gastric cells by CagA. Furthermore, the result reinforces the idea that SHP-2 is a downstream effector of CagA in induction of the hummingbird phenotype (see also below).
It was recently reported that a phosphorylation-resistant CagA mutant in which the tyrosine residues constituting the five EPIYA motifs were substituted with the conserved amino acid phenylalanine was biologically as active as wild-type CagA in inducing cell elongation and scattering of AGS cells (37). We therefore generated the same CagA mutant, PR CagA-YF, and expressed it in AGS cells. We also expressed another phosphorylation-resistant CagA mutant, PR CagA, which was made by replacing the tyrosine residues with alanines, as previously described. Neither of the CagA mutants underwent tyrosine phosphorylation nor bound SHP-2 in AGS cells. Furthermore, these CagA mutants were incapable of inducing elongated cell shape in the cells (Fig. 3D). The observations support the conclusion that the morphogenetic activity of CagA depends on its tyrosine phosphorylation.
Next, to investigate the relationship between CagA phosphorylation and cell scattering, we performed a cell scatter assay using CagA-inducible AGS cells (Fig. 3E). Upon induction of wild-type CagA in the colony made of W1804 cells, dissociation and scattering of individual cells from colonies were observed. Furthermore, a fraction of the scattered cells exhibited the hummingbird phenotype. On the other hand, the parental T71-2 cells as well as the PR CagA-inducible C065 cells were incapable of inducing cell scattering in the presence of Dox. Thus, CagA elicits not only cell elongation but also cell migration in a tyrosine phosphorylation-dependent manner.
Requirement of SHP-2 for CagA Activity-We previously demonstrated that tyrosine-phosphorylated CagA specifically binds to and thereby activates SHP-2 phosphatase, whereas the phosphorylation-resistant CagA mutant was incapable of doing so (Fig. 3D) (22).
To investigate whether SHP-2 is an essential CagA target involved in induction of the morphological transformation, we decided to inhibit SHP-2 expression in AGS cells. To do so, we used SHP-2-specific small interference RNA (siRNA) to silence the expression of SHP-2 by RNA interference. Treatment of AGS cells with siRNA for 24 h resulted in strong reduction of SHP-2 levels to as low as 20% of that of non-treated cells, without showing any sign of cell damage. CagA was subsequently expressed in the siRNA-treated AGS cells by either induction (Fig. 4A) or transient transfection (Fig. 4B). In both cases, SHP-2 silencing potently inhibited the activity of CagA to induce hummingbird cells (Fig. 4, A and B). This inhibitory effect of siRNA on CagA was cancelled by ectopic re-expression of SHP-2 (SHP-2RR-Myc), which was insensitive to siRNA as a result of a silent mutation introduced into the SHP-2 cDNA (Fig. 4C). As a control, treatment of the SHP-2-knockdown AGS cells with phorbol 12-myristate 13-acetate (PMA) induced an elongated cell morphology as previously reported (42) (Fig. 4D). Hence, reduced SHP-2 activity does not inhibit the action of cell machinery generally involved in cell elongation. From these observations, we concluded that SHP-2 is a cellular component essentially required for the morphological transformation of gastric epithelial cells by CagA. We also note here that treatment of cells with siRNA neither affected expression levels nor tyrosine phosphorylation levels of CagA (Fig. 4, A and B). The finding excludes a possibility that CagA is a substrate of SHP-2.
CagA-SHP-2 Complex Does Not Contain Grb2-CagA was recently reported to bind Grb2, although the interaction was independent of CagA tyrosine phosphorylation (37). Because Grb2 can also bind to SHP-2 (43, 44), we wondered if CagA, SHP-2, and Grb2 are capable of forming a heterotrimeric complex in mammalian cells. To investigate this possibility, we constructed SR␣ promoter-driven mammalian expression vectors for Myc-tagged SHP-2 and Myc-tagged Grb2, and expressed one or both of them together with the CagA expression vector in COS-7 cells (Fig. 5). As was the case with AGS cells, CagA was efficiently tyrosine-phosphorylated in COS-7 cells and formed a physical complex with Myc-tagged SHP-2 (Fig. 5,  lanes 4 and 11). On the other hand, a heterodimeric complex between CagA and Grb2 was not detected when the two proteins were co-expressed (Fig. 5, lanes 5 and 12). In a triple transfection experiment, in which Myc-Grb2 and Myc-SHP-2 were expressed in COS-7 cells in comparable amounts as examined by anti-Myc immunoblotting (Fig. 5, lane 9), CagA immunoprecipitation co-precipitated SHP-2 but not Grb2 (Fig.  5, lane 2). From these observations, we concluded that CagA does not bind Grb2 either directly or indirectly through SHP-2 under the condition in which CagA-SHP-2 interaction is easily detectable. Thus, the interaction between CagA and Grb2 is much weaker than that between CagA and SHP-2, if existed.
Requirement of Ras-independent Erk Activity in the Morphogenetic Activity of CagA-SHP-2 has been suggested to function upstream, downstream or at a parallel level of Ras in the mitogen-activated protein kinase (MAPK) cascade (45,46). Accordingly, we next investigated whether extracellular signalregulated kinase (Erk), a growth-related MAPK whose activity is known to be regulated by SHP-2 as well as Ras (30,31), is involved in cell elongation induced by CagA-SHP-2 interaction. To this end, we treated CagA-expressing AGS cells with U0126, a specific MEK inhibitor, and examined its effect on induction of hummingbird cells by CagA (Fig. 6, A-C). This MEK inhibitor did not have any effect on CagA expression or CagA phosphorylation in AGS cells but potently inhibited Erk activation as revealed by anti-phospho-ERK antibody. As shown in Fig. 6  (A and B), treatment of cells with U0126, but not treatment of cells with U0124, which is an inactive derivative of U0126, inhibited induction of hummingbird cells by CagA in a dosedependent manner. This finding indicates that Erk activity is also necessary for the morphological change induced by CagA. In contrast to U0126, SB203580 and SP600125, which are specific inhibitors of p38 MAPK and c-Jun N-terminal kinase (JNK), respectively, did not have any effect on the CagA activity (Fig. 6C). Furthermore, treatment of the CagA-expressing AGS cells with the PI3K inhibitor LY294002, which specifically inhibited PI3K activity as determined by Akt activation, had no effect on the morphological change induced by CagA. The find- ing excluded the possibility of a role of PI3K in the morphological transformation by CagA (Fig. 6D).
Because Erk could be activated by either a Ras-dependent or Ras-independent mechanism (47-50), we investigated whether Ras was involved in the hummingbird phenotype by expressing the constitutively active c-H-Ras Val12 (Ras V12) or the dominant-negative c-H-Ras Asn17 (Ras N17). Whereas the amounts of the mutant Ras proteins expressed were sufficient to interfere with the endogenous Ras activity as measured by status of Erk1 or Raf1 phosphorylation, neither of the Ras mutants had a significant effect on the hummingbird phenotype induced by CagA in AGS cells (Fig. 7, A and B). The results argue against the role of Ras in the morphogenetic activity of CagA. Consistently, ectopic expression of wild-type or dominant-negative Grb2 (51) had no effect on the induction of hummingbird cells by CagA (Fig. 7C). From these observations, we concluded that the hummingbird phenotype requires Ras-independent Erk activity. Sustained Activation of Erk by CagA-The kinetic patterns of Erk activation are related to specific cellular events (52). In particular, sustained Erk activation is thought to be associated with cell morphogenesis (53,54). We therefore decided to determine whether CagA modulates the duration of Erk activation. To this end, we investigated time-dependent changes in Erk activity after serum stimulation in AGS cells in the absence or presence of CagA. As shown in Fig. 8A, CagA did not modify the transient Erk activation but significantly prolonged duration of Erk activation. On the other hand, neither transient nor sustained Erk activations were affected by the presence of 5 g/ml Dox in the parental T71-2 cells (data not shown). This observation links CagA to an intracellular signaling system regulating duration of Erk activation.
It has been reported that the cellular activity of SHP-2 is required for the sustained Erk activation and thereby for cell morphological change induced by growth factors such as hep-atocyte growth factor (HGF). Hence, we next investigated the role of SHP-2 in the regulation of Erk activity in gastric epithelial cells. To do so, we established a stable SHP-2-knockdown cell line, G11, by transfecting an expression vector for SHP-2-specific siRNA into AGS cells (Fig. 8B). In the G11 cells, the level of SHP-2 was reduced to as low as 4% of that in parental AGS cells. Using the parental AGS and G11 cells, we examined the effect of SHP-2 on Erk activation. As shown in Fig. 8C, a reduced SHP-2 level abolished the sustained activation of Erk. The findings indicate that SHP-2 is required for maintaining Erk in its active form in gastric epithelial cells and therefore suggest that CagA positively modulates Erk activity through forming a complex with SHP-2.

DISCUSSION
In this study, we identified an important role of SHP-2 in the morphogenetic activity of H. pylori CagA in gastric epithelial cells. CagA, which is translocated from H. pylori into cells, is thought to play a crucial role in the pathogenesis of cagApositive H. pylori infection (2, 3, 14). However, the molecular mechanism by which CagA deregulates intracellular signaling is not fully understood. The translocated CagA induces an elongated cell shape termed the hummingbird phenotype, which is concomitantly associated with an increase in cell motility recognized as the "scattering" phenotype. We found that the morphogenetic activity of CagA requires SHP-2, which binds specifically to a tyrosine-phosphorylated form of CagA. Ras-independent Erk activity is also necessary for the CagAinduced hummingbird phenotype. We also found that CagA prolongs the duration of Erk activation and that this sustained Erk activation is mediated by SHP-2 in AGS cells. These findings indicate a pivotal role of SHP-2 in the CagA-dependent morphological transformation of gastric epithelial cells.
The translocated CagA protein is localized to the plasma membrane and undergoes tyrosine phosphorylation by Src family kinases in gastric epithelial cells (20,21). The CagA phosphorylation sites are characterized by the presence of EPIYA motifs, the number of which varies from one strain to the next (24,26). Tyrosine phosphorylation plays an important role in transmission of intracellular signals by regulating physical and functional interaction between upstream and downstream signal transducers (55)(56)(57)(58). It is therefore reasonable to assume that CagA, upon tyrosine phosphorylation, deregulates one or more of the intracellular signaling pathways and thereby causes pathogenesis of cagA-positive H. pylori infection. This notion is strongly supported by the results of the present work demonstrating that inhibition of CagA tyrosine phosphorylation inactivates the morphogenetic activity of CagA. Furthermore, two distinct CagA mutants, both of which are totally resistant to tyrosine phosphorylation, were incapable of inducing the hummingbird phenotype in AGS cells. Taken together, the results of the present work provide compelling evidence for the requirement of CagA tyrosine phosphorylation in its morphology-transforming activity.
CagA has been reported to interact with a number of cellular proteins, including SHP-2, C-terminal Src kinase (Csk), Grb2, and c-Met hepatocyte growth factor receptor (22,37,38,59). Among these, the results of our work indicate the importance of SHP-2 and Csk as relevant CagA targets, because they are the only known molecules that selectively bind the tyrosine-phosphorylated form of CagA. Upon complex formation, CagA activates Csk kinase activity, which in turn inhibits the activity of Src family kinases. Hence, CagA-Csk interaction is thought to be involved in a negative feedback regulation of CagA activity (59). As is the case of Csk, SHP-2 interacts with CagA via the typical SH2 domain-phosphotyrosine interaction (24). Through complex formation, SHP-2 is converted from the catalytically inactive form to the active form. Although the CagA-SHP-2 complex was originally demonstrated by transient expression of the cagA gene in AGS cells (22), it has also been detected in AGS cells infected with cagA-positive H. pylori (Fig. 1D) as well as in in vivo gastric mucosa from a patient with cagA-positive H. pylori infection (23). These findings exclude the possibility that the CagA-SHP-2 complex is detectable only when CagA is expressed by gene transfection. In this regard, Selbach et al. (60) reported that injected CagA failed to bind SHP-2 but that transfected CagA bound SHP-2 in AGS cells. In their study, bacteria-injected CagA possessed 3ϫ EPIYA (Tigr strain-derived CagA; "A-B-C" type according to the classification by Higashi et al. (24)), whereas transfected CagA had 5ϫ EPIYA (NCTC11637 strain-derived CagA; "A-B-C-C-C" type). We have demonstrated that the strength of activity of an individual CagA to bind SHP-2 is proportional to the number of EPIYA-C motifs (24). Accordingly, in their experimental setting, transfected CagA was much more potent than injected CagA in forming a complex with SHP-2. We have also shown that A-B-C type CagA is significantly less active in inducing the hummingbird phenotype than is A-B-C-C-C type CagA. Consistent with this, Backert et al. (41) reported that translocation and phosphorylation of the Tigr strain-derived CagA is necessary but not sufficient for induction of the morphological change in AGS cells. Possibly, a weaker CagA species may require additional signals to induce an elongated cell shape.
Our work provides compelling evidence that SHP-2 is an indispensable cellular target of CagA in induction of the hummingbird phenotype, reinforcing the notion of a critical role for the CagA-SHP-2 complex in the pathological activity of CagA. Notably, CagA recruits SHP-2 from the cytoplasm to the plasma membrane, where it activates SHP-2 phosphatase activity. Activated SHP-2 may then dephosphorylate substrates that are also located in close proximity to the membrane and thereby generate signals that lead to morphological changes. In this context, CagA is regarded as a bacterial protein that mimics mammalian docking/scaffolding molecules such as Gab or insulin receptor substrate family proteins, although it does not have any significant sequence homology with known mammalian proteins (39).
Signaling components lying downstream of the CagA-SHP-2 complex remain to be elucidated. Results of biochemical analyses of Gab-SHP-2 interaction as well as those of genetic studies on the Drosophila SHP-2 homologue Corkscrew indicate that SHP-2 is capable of activating Ras (61,62). Consistently, SHP-2 has been reported to bind the Grb2-Sos complex (43,44), the membrane localization of which results in the activation of the Ras/Raf/Mek/Erk pathway. However, based on the results of a series of experiments using dominant-negative and/or dominant-positive mutants, we conclude that Ras is not required for the hummingbird phenotype induced by the CagA-SHP-2 interaction. Meanwhile, Mimuro et al. (37) has recently reported that CagA binds Grb2, activates Ras, and induces the cell scattering phenotype in a manner independent of CagA tyrosine phosphorylation. However, we were not able to detect a physical interaction between CagA and Grb2 under the condition in which CagA-SHP-2 complex was easily detectable (Fig. 5). Furthermore, we found that tyrosine phosphorylation of CagA is an essential prerequisite for both the hummingbird phenotype and the cell scattering phenotype induced by CagA. At present, we cannot explain the discrepancies between our results and those reported by Mimuro et al. (37).
In epithelial cells treated with HGF, induction of morphological change and scattering requires SHP-2 recruitment to c-Met receptors through the scaffolding protein Gab. Maroun and co-workers (53) reported that a sustained activation of Erk by SHP-2 is necessary for cell morphological change in Madin-Darby canine kidney epithelial cells treated with HGF. In AGS cells, translocated CagA binds c-Met receptors upon treatment of cells with HGF, and this interaction enhances cell scattering by HGF (38). However, CagA is capable of inducing the hummingbird phenotype without HGF treatment. Accordingly, CagA appears to hijack the c-Met-dependent signaling pathway by deregulating its intracellular component, SHP-2. Upon complex formation with CagA, SHP-2 may provoke a series of biochemical events that are otherwise activated in response to HGF. Consistent with this notion, induction of morphological change by CagA requires Erk activity. In gastric epithelial cells, CagA prolongs the duration of Erk activation and SHP-2 is required for the sustained Erk activation. The results indicate that SHP-2 is involved in the positive regulation of Erk activity, which is necessary for induction of the hummingbird phenotype by CagA.
Consistent with our observation that the CagA activity is independent of Ras, sustained Erk activation has been reported to be regulated by a Ras-independent pathway, including Rap1, one of Ras-related proteins (49). Rap1 activates B-Raf and subsequently induces sustained Erk activation by a growth factor. Also, Rap1 regulates cell morphogenetic processes (63). Recently, Wessler et al. (64) reported that infection of cagA-positive H. pylori activated the Rap1/B-Raf pathway but not the Ras/Raf-1 pathway in AGS cells. Therefore, the CagA-SHP-2 complex might activate the Rap1/B-Raf pathway and induce sustained Erk activation, which is associated with the morphological change. Notably, however, the conclusion that Ras is not involved in the morphogenetic activity of CagA does not necessarily mean that CagA has no relation to the Ras pathway. Rather, activation of Ras by the CagA-SHP-2 complex may provoke a dysregulated mitogenic response of gastric epithelial cells. In any case, identification of bona fide substrates for SHP-2 should enable elucidation of the molecular link between SHP-2 and Erk as well as other molecules that collectively deregulate intracellular signaling and thereby cause cell dysfunction. In this regard, Amieva and co-workers (65) recently reported that CagA recruits SHP-2 in close proximity to the tight junction and thus modifies the apical-junctional complex functions. The work suggests that molecules constituting the tight junction are potential candidates for SHP-2 substrates.
It is most likely that CagA exerts a variety of biological activities once delivered into gastric epithelial cells. Some of these activities may depend on tyrosine phosphorylation, whereas others may not. Deregulation of signals controlled by tyrosine phosphorylation is a hallmark of cancer (66). Also, morphological abnormality is closely associated with malignant transformation (67). It is therefore reasonable to assume that deregulation of intracellular signaling by CagA, which most likely mimics the actions of mammalian docking/scaffolding proteins, substantially contributes to the multistep carcinogenesis of gastric epithelial cells.