Structural Basis and Functional Consequence of Helicobacter pylori CagA Multimerization in Cells*

Helicobacter pylori cagA-positive strains are associated with gastric adenocarcinoma. The cagA gene product CagA is delivered into gastric epithelial cells where it localizes to the plasma membrane and undergoes tyrosine phosphorylation at the EPIYA-repeat region, which contains the EPIYA-A segment, EPIYA-B segment, and Western CagA-specific EPIYA-C or East Asian CagA-specific EPIYA-D segment. In host cells, CagA specifically binds to and deregulates SHP-2 phosphatase via the tyrosine-phosphorylated EPIYA-C or EPIYA-D segment, thereby inducing an elongated cell shape known as the hummingbird phenotype. In this study, we found that CagA multimerizes in cells in a manner independent of its tyrosine phosphorylation. Using a series of CagA mutants, we identified a conserved amino acid sequence motif (FPLXRXXXVXDLSKVG), which mediates CagA multimerization, within the EPIYA-C segment as well as in a sequence that located immediately downstream of the EPIYA-C or EPIYA-D segment. We also found that a phosphorylation-resistant CagA, which multimerizes but cannot bind SHP-2, inhibits the wild-type CagA-SHP-2 complex formation and abolishes induction of the hummingbird phenotype. Thus, SHP-2 binds to a preformed and tyrosinephosphorylated CagA multimer via its two Src homology 2 domains. These results, in turn, indicate that CagA multimerization is a prerequisite for CagA-SHP-2 interaction and subsequent deregulation of SHP-2. The present work raises the possibility that inhibition of CagA multimerization abolishes pathophysiological activities of CagA that promote gastric carcinogenesis.

Helicobacter pylori is a micro-aerophilic spiral-shaped bacterium. It colonizes the human stomach and is estimated to inhabit at least half of the world's human population. Chronic infection in the stomach with H. pylori causes gastric diseases such as chronic gastritis and peptic ulceration later in life. Epidemiological and pathological studies have further indicated that infection with cagA-positive H. pylori strains is associated with a high risk of gastric cancer (1)(2)(3).
The cagA gene encodes an ϳ120 -145-kDa CagA protein that is delivered into gastric epithelial cells via the type IV secretion system and localizes to the plasma membrane (4 -10). Translocated CagA then undergoes tyrosine phosphorylation by members of the Src family of protein tyrosine kinases such as c-Src, Fyn, Lyn, and Yes (11)(12)(13). Tyrosine phosphorylation of CagA occurs at the EPIYA motif, a fiveamino-acid sequence (Glu-Pro-Ile-Tyr-Ala) that is present in variable numbers in the C-terminal EPIYA-repeat region of the protein (13). Because of homologous recombination that occurs within the cagA gene, four distinct EPIYA segments (EPIYA-A to -D), each of which possesses a single EPIYA motif, have been identified (14 -16). The EPIYA-repeat region of the CagA protein of H. pylori isolated in Western countries contains the EPIYA-A, EPIYA-B, and Western CagA-specific EPIYA-C segments. The EPIYA-C segment variably multiplies among distinct Western CagA proteins, mostly ranging from 1 to 3. On the other hand, the EPIYArepeat region of the CagA protein of H. pylori isolated in East Asian countries contains the EPIYA-A, EPIYA-B, and East Asian CagA-specific EPIYA-D segments. EPIYA motifs present in the EPIYA-C and EPIYA-D segments are major sites of CagA tyrosine phosphorylation, to which SHP-2 protein tyrosine phosphatase specifically binds in a phosphorylationdependent manner (14,16). Upon complex formation, CagA deregulates SHP-2 phosphatase activity, which, in turn, dephosphorylates and inactivates focal adhesion kinase, thereby inducing an elongated cell shape with elevated cell motility (17). CagA-deregulated SHP-2 also provokes sustained Erk mitogen-activated protein kinase activation, which plays a crucial role in cell cycle progression (18).
The CagA-SHP-2 interaction requires both the N-SH2 and C-SH2 domains of SHP-2 (14,16). In contrast, a single EPIYA-C or EPIYA-D segment is sufficient for CagA to form a physical complex with SHP-2. These observations indicate that a SHP-2 protein simultaneously interacts with two tyrosine-phosphorylated CagA proteins via the SH2 domains. Consistent with this idea, we previously reported that CagA multimerizes in mammalian cells (14). In this work, we examined the structural basis of CagA multimerization and identified the multimerization motif that is highly conserved between Western and East Asian CagA proteins. We also investigated the biological role of the CagA multimerization and found that SHP-2 specifically binds to a preformed CagA multimer in a tyrosine phosphorylation-dependent manner, and we discuss the pathophysiological relevance of CagA multimerization in cells.

EXPERIMENTAL PROCEDURES
Expression Vectors-Mammalian expression vectors for hemagglutinin (HA) 2 -tagged, wild-type CagA (CagA-ABCCC derived from H. pylori NCTC11637 Western strain) and its mutants, CagA-⌬ABCC, CagA-⌬ABCC-s, CagA-⌬ABCCC, CagA-⌬ABCCCϩEPIYA, CagA-⌬ABCCCϩ3EPIYA, CagA-⌬ABCC-⌬EPIYA, and CagA-ABD have been described previously (14,15). Genes encoding a series of mutant CagA-ABD molecules were generated from cagA-ABD, which was isolated from H. pylori F32 East Asian strain, by restriction enzyme digestions and/or site-directed mutagenesis as previously described (14). Structures of the mutant CagA molecules made are schematically represented in the figures in which they are used. The amino acid residue numbers shown in the schematics are from NCTC11637 CagA in the cases of CagA-ABCCC derivatives and are from F32 CagA in the cases of CagA-ABD derivatives. The mutant cagA genes were cloned into a pSP65SR␣ mammalian expression vector. Human SHP-2 cDNA and its derivatives were tagged with Myc epitope as previously described (14,15).
Cell Culture and Transfection-Monkey COS-7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum. AGS human gastric epithelial cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum. COS-7 cells (1.5 ϫ 10 6 cells) were transfected with 30 g of plasmids by using the calcium phosphate method as previously described (19). AGS cells were seeded into 35-mm 2 dishes (1.2 ϫ 10 5 cells/dish), and 8 g of plasmid was transfected into cells as previously described (13). Cell morphology was examined by light microscopy at 18 h after transfection.
Antibodies-Anti-HA monoclonal antibody 3F10 (Roche Applied Science) and anti-FLAG monoclonal antibody M2 (Sigma) were used for immunoblotting and immunoprecipitation of HA-tagged CagA and FLAG-tagged CagA. Anti-Myc monoclonal antibody 9E10 was used for immunoblotting and immunoprecipitation of Myc-tagged CagA. Anti-phosphotyrosine monoclonal antibody 4G10 (Upstate Biotechnology) and anti-SHP-2 polyclonal antibody C-18 (Santa Cruz Biotechnology) were used for immunoblotting.

RESULTS
Phosphorylation-independent Multimerization of CagA-In a previous study, we demonstrated that ABCCC-type Western CagA (CagA-ABCCC) can multimerize and thus exist as a dimer, oligomer, or a higher order of protein complex in cells (14). However, we also noticed that the recombinant CagA-ABCCC protein produced in Escherichia coli does not spontaneously multimerize in vitro, indicating that CagA multimerization occurs only in FIGURE 1. Phosphorylation-independent multimerization of East Asian CagA. A, schematic view of CagA-ABD, which possesses the EPIYA-repeat region derived from East Asian CagA. Boxes A-D indicate the EPIYA segments, each of which contains a single EPIYA motif. B, FLAG-tagged CagA-ABD and HA-tagged CagA-ABD were co-expressed in COS-7 cells. Cell lysates were subjected to immunoprecipitation (IP) with anti-HA antibody. Anti-HA immunoprecipitates and total cell lysates (TCL) were immunoblotted (IB) with anti-HA or -FLAG antibody. The data shown are representative of those from three separate experiments. C, FLAG-tagged and HA-tagged CagA-ABD or its phosphorylation-resistant derivative, CagA-abd-HA, were co-expressed in COS-7 cells. Cell lysates were subjected to immunoprecipitation with anti-HA antibody. Anti-HA immunoprecipitates and total cell lysates were immunoblotted with the indicated antibodies. In a control experiment, SHP-2 was found to form a physical complex with CagA-ABD but not CagA-abd. The data shown are representative of those from three separate experiments. Schematics of the CagA derivatives used are shown at the top. pY, phosphotyrosine. mammalian host cells. 3 To understand the biological role of CagA multimerization, we first investigated whether East Asian CagA, another major type of CagA, can also multimerize in cells. Because Western CagA and East Asian CagA are primarily characterized by the difference in the structure of the EPIYA-repeat region, we employed a CagA-ABD mutant that was made from CagA-ABCCC by replacing the entire EPIYA-repeat region with that derived from F32 East Asian CagA (Fig. 1A) (14). CagA-ABD was tagged with a FLAG or HA epitope, and the resulting FLAG-and HA-tagged CagA-ABD molecules were co-expressed in COS-7 cells. Sequential immunoprecipitation and immunoblotting of the cell lysates with anti-HA and -FLAG antibodies revealed that CagA-ABD, which possesses the EPIYA-repeat region from East Asian CagA, multimerizes in cells (Fig. 1B). Quantitation of each of the CagA bands indicated that only a small fraction (Ͻ5%) of FLAG-tagged CagA-ABD proteins expressed in COS-7 cells were associated with HAtagged CagA-ABD proteins. Thus, the CagA multimers may be labile and easily dissociate into monomers during protein extraction and immunoprecipitation in our experimental conditions. Alternatively, predominant CagA proteins may exist as monomers, with a small amount of CagA existing as multimers, in cells. In this case, the formation of CagA multimer may require an additional cellular component whose expression level determines the amount of CagA multimers.
Next, to address the role of tyrosine phosphorylation in CagA multimerization, we made a phosphorylation-resistant form of CagA-ABD, CagA-abd, by replacing tyrosine residues that are present in the EPIYA-A, -B, and -D segments with alanine residues. The CagA-abd mutant was HA-tagged and was co-expressed together with FLAG-tagged CagA-ABD in COS-7 cells. As was the case for CagA-ABD, CagA-abd, which did not undergo tyrosine phosphorylation, was capable of interacting with CagA-ABD (Fig. 1C). Based on this result, we concluded that CagA multimerizes independently of tyrosine phosphorylation. This result also excluded the possibility that CagA multimerization requires cellular proteins (such as SHP-2 and Csk) that interact with CagA in a tyrosine phosphorylation-dependent manner (11,19).
Delineation of the CagA Region Involved in Multimerization-To investigate the mechanism of CagA multimerization in more detail, we narrowed down the CagA region that is involved in multimerization by using a series of CagA deletion mutants. When CagA-ABD was co-expressed with CagA-⌬AB-CCC, which lacks the entire EPIYA-repeat region in COS-7 cells, the two CagA derivatives failed to interact with each other ( Fig. 2A). This result revealed a critical role of the EPIYA-repeat region in CagA multimerization. Restoration of a single EPIYA motif or triple EPIYA motifs in the CagA-⌬ABCCC mutant did not confer CagA multimerization, arguing against an active role of the EPIYA motif in the multimerization process ( Fig. 2A). We then generated two CagA-ABD derivatives, CagA-ABD-⌬D and CagA-ABD-⌬AB. CagA-ABD-⌬D contains EPIYA-A and EPIYA-B segments but lacks the EPIYA-D segment. Conversely, CagA-ABD-⌬AB lacks the EPIYA-A and EPIYA-B segments (Fig. 2B). Studies on co-expression in COS-7 cells show that CagA-ABD-⌬AB is capable of interacting with CagA-ABD, whereas CagA-ABD-⌬D failed to do so (Fig. 2B). Accordingly, the 91-amino-acid sequence (residues 922-1012) in the EPIYA-repeat region of CagA-ABD is responsible for CagA multimerization.
To further delineate the CagA sequence that is required for CagA multimerization, the above-identified 91-amino-acid residues were subdivided into D1 and D2 sequences (Fig. 3A), and a CagA-ABD-⌬AB mutant that also lacks the D1 or D2 sequence was generated. When co-expressed in COS-7 cells, FLAG-tagged CagA-ABD-⌬AB was capable of interacting with HA-tagged CagA-ABD-⌬ABD1 but not HA-tagged CagA-ABD-⌬ABD2 (Fig. 3B). Again, deletion of the EPIYA sequence from CagA-ABD-⌬AB did not influence CagA multimerization. A reciprocal immunoprecipitation experiment confirmed the interaction between CagA-ABD-⌬ABD1 and CagA-ABD-⌬AB (Fig. 3C). These observations indicated that the D2 sequence is responsible for CagA multimerization in ABD-type East Asian CagA species.
Identification of the CagA Multimerization Motif That Is Conserved between East Asian and Western CagA Species-We have previously shown that Western CagA multimerizes in mammalian cells (14). Identification of the D2 sequence in East Asian CagA therefore prompted us to investigate the sequence that is utilized for Western CagA multimerization. The observation that the EPIYA-A and EPIYA-B segments, which are highly conserved between Western and East Asian CagA species, are not involved in CagA multimerization suggested that the EPIYA-C segment is involved in the multimerization of Western CagA. Accordingly, we expressed a Western CagA mutant (CagA-⌬ABCC-s) that lacks the amino acid sequence between 869 and 1008 together with CagA-ABCCC and found that they interact with each other (Fig. 4A). Taken together with the previous finding that CagA-⌬ABCCC (which was made from CagA-ABCCC by deleting the entire EPIYA-repeat region) lost the ability to multimerize, we concluded that the sequence between residues 1009 and 1086, which contains a single repeat of the EPIYA-C segment, is responsible for Western CagA multimerization.
Assuming that CagA multimerization is mediated by the homophilic interaction of an integrated CagA structure, we wondered whether such a structure is conserved between Western CagA and East Asian CagA proteins. Accordingly, we expressed CagA-ABD-⌬AB together with CagA-⌬ABCC, which was made from CagA-⌬ABCC-s by deleting a 44-aminoacid stretch located immediately after the last repeat of the EPIYA-C segment in the EPIYA-repeat region of CagA-AB-CCC, in COS-7 cells (Fig. 4B). The co-expression study FIGURE 3. Determination of the sequence involved in CagA-ABD multimerization. A, definition of D1 and D2 sequences in the 91-amino-acid stretch that is present in CagA-ABD-⌬AB that binds CagA-ABD (but not in CagA-⌬ABCCC that does not bind CagA-ABD). B, HA-tagged CagA-ABD-⌬AB was co-expressed with each of the HA-tagged CagA-ABD-⌬AB derivatives that are schematically shown at the top. Cell lysates were subjected to immunoprecipitation (IP) with anti-HA antibody. Anti-HA immunoprecipitates and total cell lysates (TCL) were immunoblotted (IB) with the indicated antibodies. In a control experiment, SHP-2 formed a physical complex with CagA-ABD and CagA-ABD-⌬ABD1 (but not CagA-ABD-⌬ABD2 or CagA-ABD-⌬AB-⌬EPIYA). The result was totally consistent with our previous observation (14). The data shown are representative of those from three separate experiments. C, HA-tagged CagA-ABD was co-expressed with FLAG-tagged CagA-ABD-⌬AB or FLAGtagged CagA-⌬ABD1. Cell lysates were subjected to immunoprecipitation with anti-FLAG antibody. Anti-FLAG immunoprecipitates and total cell lysates were immunoblotted with anti-HA or -FLAG antibody. The data shown are representative of those from three separate experiments.

Multimerization of H. pylori CagA in Cells
OCTOBER 27, 2006 • VOLUME 281 • NUMBER 43 JOURNAL OF BIOLOGICAL CHEMISTRY 32347 revealed that the two CagA derivatives mutually interact with each other in cells, indicating that the mechanism of CagA multimerization is conserved between Western and East Asian CagA proteins. Furthermore, the result indicated that the sequence responsible for multimerization of Western CagA is present within the EPIYA-C segment. However, as we reported previously, the EPIYA-C and EPIYA-D segments exhibit only a limited homology around the EPIYA motif (Fig. 4C, upper panel) (14). As noted, the EPIYA motif itself does not seem to play a role in CagA multimerization (see Figs. 2A and 3B). Consistent with this notion, deletion of the EPIYA motif from CagA-⌬ABCC did not abolish CagA multimerization (Fig. 4B). Furthermore, deletion or substitution of the TID sequence, which is also conserved between the EPIYA-C and EPIYA-D segments, did not affect CagA multimerization (Fig. 4C). Meanwhile, we noticed that the first 16-amino-acid residues of the EPIYA-C segment are closely related to the residues that located immediately downstream of the EPIYA-D segment in the D2 sequence of East Asian CagA (Fig. 4D, upper panel, gray boxes). This observation raised the idea that the conserved 16-amino-acid sequence mediates CagA multimerization. To investigate this possibility, we deleted the 16-amino-acid stretch from CagA-ABD-⌬ABD1 to make CagA-ABD-⌬ABD1-16AA. When co-expressed in COS-7 cells, CagA-ABD-⌬ABD1-16AA lost the ability to interact with CagA-ABD-⌬AB (Fig. 4D, lower panel). Thus, the conserved 16-amino-acid sequence is responsible for East Asian CagA multimerization. The identified sequences between the EPIYA-C segment of Western CagA and the D2 sequence of East Asian CagA are characterized by the presence of a conserved FPLXRXXXVXDLSKVG motif, which we designated as the CagA multimerization (CM) motif. It should also be noted that the sequences that immediately follow the EPIYA-C and EPIYA-D segments are well conserved between Western and East Asian CagA species. Accordingly, Western CagA has an additional CM motif that follows the last repeat of the EPIYA-C segment (Fig. 4D, upper panel) (see also "Discussion").
Role of CagA Multimerization in CagA-SHP-2 Complex Formation-We previously reported that both the N-SH2 and C-SH2 domains of SHP-2 are indispensable for the complex formation of SHP-2 with Western CagA (14). We wished to know whether this is also the case with East Asian CagA. To this end, we co-expressed SHP-2 and a series of EPIYA mutants for CagA-ABD in COS-7 cells and confirmed that the CagA-SHP-2 interaction is strictly dependent on the phosphorylation of the EPIYA-D site in the case of East Asian CagA (Fig. 5A). We also co-expressed East Asian CagA and a series of SHP-2 mutants and confirmed that the CagA-SHP-2 interaction requires both of the SH2 domains of SHP-2 (Fig. 5B). Notably, each of the SH2 domains of SHP-2 independently binds to tyrosine-phosphorylated peptides with similar specificity (20). In contrast, major CagA species, such as ABD-type CagA and ABC-type CagA, possess only a single SHP-2-binding site, the EPIYA-C or EPIYA-D segment. These findings collectively indicated that a single SHP-2 molecule binds to a preformed CagA dimer or multimer. If this is the case, then the phosphorylation-resistant FLAG-tagged CagA-ABD was co-expressed with HA-tagged CagA-abd, HA-tagged CagA-abd-⌬TID that lacks the conserved TID sequence, or HA-tagged CagA-abd-mTID that has QVN at the conserved TID sequence in COS-7 cells. Cell lysates were subjected to immunoprecipitation with anti-HA antibody. Anti-HA-immunoprecipitates and total cell lysates were immunoblotted with the indicated antibodies. Experiments were done twice, and the results were reproducible. D, upper panel, alignment of the sequences between the EPIYA-C segment of Western CagA and the D2 sequence of East Asian CagA. Gray boxes are the 16-amino-acid sequences conserved between the EPIYA-C segment and the D2 sequence of CagA-ABD. Lower panel, FLAG-tagged CagA-ABD-⌬AB was co-expressed with HA-tagged CagA-ABD-⌬AB or its derivative, CagA-ABD-⌬ABD1-16AA that lacks the 16-amino-acid residues conserved between the EPIYA-C segment of Western CagA, and the D2 sequence of East Asian CagA. Cell lysates were immunoprecipitated with an anti-HA antibody. Anti-HA immunoprecipitates and total cell lysates were subjected to immunoblotting with the indicated antibodies. The data shown are representative of those from three separate experiments. CagA-abd, which multimerizes but cannot bind SHP-2, should act as a dominant-negative CagA that inhibits CagA-SHP-2 complex formation as a component of the CagA dimer (or multimer). To this end, we co-transfected CagA-ABD and CagAabd in COS-7 cells and compared the levels of endogenous SHP-2 that formed complexes with CagA-ABD. The results of the experiment showed that CagA-SHP-2 complex formation was abolished by co-expression of CagA-abd. Thus, the phosphorylation-resistant CagA-abd acted as a dominant-negative mutant in the CagA-SHP-2 complex formation (Fig. 6A). The result, in turn, indicated that CagA multimerization is a prerequisite for the interaction of SHP-2 with ABC-or ABD-type CagA. Furthermore, efficient inhibition of CagA-SHP-2 complex formation by phosphorylation-resistant CagA suggested that the molecular nature of the CagA multimer is an oligomer such as a dimer. Finally, to consolidate the role of CagA multimerization in the pathophysiological activity of CagA, we examined the effect of the dominant-negative CagA-abd mutant on the induction of cell elongation known as the hummingbird phenotype by CagA-ABD (6). The morphological change has been shown to be induced by the complex formation between CagA and SHP-2 in AGS cells (13,18). As shown in Fig. 6B, expression of CagA-ABD together with CagA-abd significantly inhibited induction of the hummingbird phenotype by CagA-ABD in AGS cells. The result indicated that CagA multimerization plays an important role in the morphogenetic activity of CagA in gastric epithelial cells.

DISCUSSION
In this work, we demonstrated that CagA, either Western or East Asian species, multimerizes in cells independently of its tyrosine phosphorylation status. We also identified a CM motif that is conserved between Western and East Asian CagA species. Finally, we showed that CagA multimerization is critically involved in the formation of the CagA-SHP-2 signaling complex, which plays an important role in cagA-positive H. pylori-mediated gastric pathogenesis.
Upon delivery into gastric epithelial cells, CagA functionally mimics mammalian scaffolding adaptor proteins, such as Gab family proteins, through its interaction with multiple cellular proteins by both tyrosine phosphorylation-dependent and -independent mechanisms (21). As a result, CagA perturbs intracellular signaling that regulates cell growth, cell motility, and cell polarity and thereby predisposes CagA-injected gastric epithelial cells to transformation. The best-characterized cellular target of CagA is the SH2 domain containing tyrosine phosphatase SHP-2 (13,21). CagA specifically interacts with SHP-2 in a tyrosine phosphorylation-dependent manner and deregulates the phosphatase activity. The results of our previous and present works provide compelling evidence that both the N-and C-SH2 domains of SHP-2 are required for the CagA-SHP-2 interaction. This finding, in turn, indicates that two phosphotyrosine-containing CagA sequences must interact with the two SH2 domains of a single SHP-2 molecule. In this regard, prevalent Western or East Asian CagA proteins possess only a single SHP-2-binding site, EPIYA-C in the case of the ABC-type Western CagA and EPIYA-D in the case of the ABD-type East Asian CagA. Given this notion, there are two possible mechanisms that explain the interaction of SHP-2 with CagA carrying a single SHP-2-binding site. One possible mechanism is that CagA multimerizes, or dimerizes, at the host cell plasma membrane, and the preformed CagA multimer (dimer) interacts with a single SHP-2 molecule. The other possible mechanism is that translocated CagA exists as a monomer and a single SHP-2 molecule bridges two tyrosine-phosphorylated CagA monomers to make a stable tripartite complex. The latter possibility is excluded by the results of our present work showing that CagA multimerizes regardless of its tyrosine phosphorylation status and that a phosphorylation-resistant CagA, which multimerizes but cannot bind SHP-2, inhibits the interaction between wild-type CagA and SHP-2. Hence, CagA-SHP-2 complex formation is impaired when the phosphorylation-resistant CagA is incorporated into the CagA multimer, such as the dimer. It should be also noted that there are a few CagA species that possess multiple EPIYA-C or EPIYA-D segments (4,14,16). Our present work does not exclude the possibility that such CagA proteins bind to SHP-2 via the multiple SHP-2-binding sites regardless of CagA multimerization.
Our study, employing a series of CagA mutants, uncovered a critical role of the 16-amino-acid sequence motif in CagA multimerization, which we termed as the CM motif. It is possible that two CagA proteins dimerize via the homophilic interaction of the structure that is created by the CM motif. Given that CagA multimerization occurs only in mammalian cells, however, it is also possible that the multimerization process requires additional cellular proteins. Such proteins, if they exist, may exist as a multimer (dimer) to which CagA binds via the CM motif. The CM motif is present in the N-terminal sequence of the 34-amino-acid EPIYA-C segment, which variably duplicates among distinct Western CagA species. The motif is also present in the EPIYA-repeat region that is located immediately downstream of the EPIYA-C or EPIYA-D segment (Figs. 4D  and 7). This finding provides additional insights into the structural composition of the EPIYA-repeat region, which was made via homologous recombination within the cagA gene (14 -16). In East Asian CagA, the EPIYA-D segment is immediately followed by an incomplete EPIYA-C-like sequence, which contains the CM motif but lacks the EPIYA motif. Similarly, the last repeat of the EPIYA-C segment of Western CagA is followed by an incomplete EPIYA-C-like sequence (Fig. 7). Accordingly, Western CagA possesses multiple CM motifs, whereas East Asian CagA possesses a single CM motif. The difference in the number of CM motifs might influence the potential of individual CagAs to multimerize in host cells.
SHP-2 is a cytoplasmic signal transducer that physiologically functions as a positive regulator of cell growth and cell movement (22). Furthermore, gain-of-function mutations in the PTPN11 gene, which encodes SHP-2, have been found in a variety of human malignancies (23). These observations indicate that deregulated activation of SHP-2 by CagA plays a key role in the development of gastric carcinogenesis (21). A critical role for CagA-SHP-2 interaction in the development of gastric cancer has also been implicated from the studies of cagA-positive H. pylori strains isolated in gastric cancer patients (24,25). The present work revealed that CagA multimerization plays an important role in the pathophysiological activity of CagA in disturbing host cell function via SHP-2 deregulation. This, in turn, indicates that inhibition of CagA multimerization, which abolishes CagA activity as a potential oncoprotein, in gastric epithelial cells is an effective way to prevent gastric cancer development in individuals infected with cagA-positive H. pylori. Obviously, the identified CM motif is an excellent structural target for the development of a specific inhibitor of CagA multimerization.