SHP-2 Tyrosine Phosphatase as an Intracellular Target of Helicobacter pylori CagA Protein

See allHide authors and affiliations

Science  25 Jan 2002:
Vol. 295, Issue 5555, pp. 683-686
DOI: 10.1126/science.1067147


Helicobacter pylori CagA protein is associated with severe gastritis and gastric carcinoma. CagA is injected from the attached Helicobacter pylori into host cells and undergoes tyrosine phosphorylation. Wild-type but not phosphorylation-resistant CagA induced a growth factor–like response in gastric epithelial cells. Furthermore, CagA formed a physical complex with the SRC homology 2 domain (SH2)–containing tyrosine phosphatase SHP-2 in a phosphorylation-dependent manner and stimulated the phosphatase activity. Disruption of the CagA–SHP-2 complex abolished the CagA-dependent cellular response. Conversely, the CagA effect on cells was reproduced by constitutively active SHP-2. Thus, upon translocation, CagA perturbs cellular functions by deregulating SHP-2.

Helicobacter pylori, which infects about 50% of the world's population, causes gastric diseases ranging from gastritis to cancer and has been classified as a group I carcinogen. CagA is the product of the cagA gene, which is carried in virulent type I strains of H. pylori. The correlation between expression of CagA and H. pylorivirulence has been well-documented (1–4). In particular, increased frequency of gastric carcinoma and MALT (mucosal-associated lymphoid tissue) lymphoma in patients infected with cagA+ H. pylori strains suggests that CagA is associated with an increased risk of gastric cancer. However, no direct role for CagA in pathogenesis or function has been described.

After attachment of cagA+ H. pylori to gastric epithelial cells, CagA is directly injected from the bacteria into the cells via the bacterial type IV secretion system and undergoes tyrosine phosphorylation in the host cells (5–9). The cagA+ H. pylori–host cell interaction also triggers morphological changes similar to those induced by growth factor (5). Translocated CagA may be involved in dysregulation of host cell functions, thereby contributing to pathogenesis.

To examine the role of the CagA protein in the host cell, thecagA gene isolated from the H. pylori standard strain NCTC11637 was COOH-terminal–tagged with the hemagglutinin (HA) and was cloned into pSP65SRα. The expression vector was transfected into AGS human gastric epithelial cells or monkey COS-7 cells, and CagA expression was confirmed by immunoblotting with antibody to HA (Fig. 1A) (10, 11). Immunoblotting with antibody to phosphotyrosine revealed that the expressed CagA underwent tyrosine phosphorylation (Fig. 1A). The NCTC11637-derived CagA protein possesses seven potential tyrosine phosphorylation sites (residues -117, -893, -912, -965, -999, -1033, and -1100). Of these, residues -893, -912, -965, -999, and -1033 constitute five copies of the EPIYA sequences. Notably, residues -965, -999, and -1033 are made by three repeats of the “D1+D2+D3” element (12). We generated a CagA mutant in which all of the tyrosine residues present in the five copies of the Glu-Pro-Ile-Tyr-Ala (EPIYA) sequences were replaced by alanine. The mutant CagA did not undergo any tyrosine phosphorylation in cells (Fig. 1A). Thus, one or several of these tyrosine residues were apparently in vivo tyrosine phosphorylation sites of CagA.

Figure 1

Ectopic expression of CagA. (A) AGS or COS-7 cells were transiently transfected with HA-tagged wild-type (WT) CagA, HA-tagged phosphorylation-resistant (PR) CagA or an empty vector (control). Cell lysates were immunoblotted (IB) with antibody to HA or phosphotyrosine. (B) The morphology of AGS cells transfected with CagA was examined under a microscope at 17 hours after transfection (left). The cells were stained with an antibody to HA (right).

AGS cells contacted with cagA+ H. pylorirespond by producing the hummingbird phenotype, characterized by elongation and spreading of cells (5). We examined morphological changes in AGS cells after transfection of the CagA expression vector. At 17 hours after transfection, 20 to 30% of the cells (transfection efficiency was approximately 40%) exhibited the hummingbird phenotype (Fig. 1B). In contrast, phosphorylation-resistant CagA failed to induce any morphological changes in AGS cells. Thus, CagA was the necessary and sufficient H. pylori component for the hummingbird phenotype, and tyrosine phosphorylation of CagA was required. CagA preferentially localizes in the plasma membrane (13), but this membrane localization was independent of tyrosine phosphorylation and was also observed in the case of phosphorylation-resistant CagA (Fig. 1B).

The hummingbird phenotype resembles the morphological changes caused by exposure to hepatocyte growth factor (HGF) (5). Other reports suggest that activation of SHP-2 (14–16), a cytoplasmic tyrosine phosphatase that contains two tandem SH2 domains, plays a major role in the HGF-induced cellular morphological changes (17). Indeed, SHP-2 positively regulates signal transduction events from a variety of activated receptor tyrosine kinases (14–16). We found that SHP-2 was expressed in AGS cells, whereas SHP-1, another SH2-containing phosphatase, was hardly detectable (18). Because SH2 domains are phosphotyrosine-binding modules (19), we investigated the capacity of CagA to bind SHP-2. In lysates from AGS cells transfected with the CagA expression vector, CagA co-immunoprecipitated endogenous SHP-2 and vice versa (Fig. 2A). In contrast, the phosphorylation-resistant CagA and SHP-2 did not co-immunoprecipitate each other. Furthermore, SHP-2 failed to co-immunoprecipitate unphosphorylated CagA (Fig. 2A). Thus, CagA binds SHP-2 in gastric epithelial cells in a tyrosine phosphorylation-dependent manner. Furthermore, immunodepletion of SHP-2 from the lysates of AGS cells expressing CagA simultaneously depleted almost all, if not all, of the tyrosine-phosphorylated CagA from the lysates (Fig. 2B). Thus, tyrosine-phosphorylated CagA binds SHP-2 stoichiometrically in these cells.

Figure 2

Interaction between CagA and SHP-2. (A) AGS cells were transfected with WT CagA-HA, PR CagA-HA, or a control empty vector. WT CagA-HA or SHP-2 was immunoprecipitated from lysates prepared in the presence or absence of Na3VO4, a tyrosine phosphatase inhibitor. The immunoprecipitates (IP) and total cell lysates (TCL) were immunoblotted (IB) with antibodies as described. Pre-immune rabbit IgG was used for a control antibody. (B) Amounts of tyrosine-phosphorylated CagA in the supernatants of AGS cells transfected with WT CagA-HA before and after immunodepletion with antibody to SHP-2 or the control antibody.

Physical interaction between tyrosine-phosphorylated CagA and SHP-2 was also demonstrated in COS-7 cells (Fig. 3A). Using the COS-7 cells, we examined regions in SHP-2 that are required for CagA binding. A SHP-2 mutant lacking the SH2 domains (SHP-2ΔSH2-Myc) had totally lost the ability to bind CagA, whereas another mutant lacking the phosphatase domain (SHP-2ΔPD-Myc) retained CagA-binding activity (Fig. 3B) (20). Thus, tyrosine-phosphorylated CagA appears to interact specifically with the SH2 domains of SHP-2.

Figure 3

Activation of SHP-2 by CagA. (A) COS-7 cells were co-transfected with CagA and SHP-2 expression vectors. Cell lysates were subjected to immunoprecipitation (IP) with an antibody to HA. Immunoprecipitates and total cell lysates (TCL) were immunoblotted (IB) with the indicated antibodies. (B) Co-expression of CagA-HA with Myc-tagged, wild-type (SHP-2-Myc) or mutant (ΔPD-Myc and ΔSH2-Myc) SHP-2. Samples were fractionated by 10% (lanes 1 to 9) or 13.5% (lanes 10 to 13) SDS-PAGE. NS, nonspecific band; Mr, relative molecular mass. (C) COS-7 cells were co-transfected with SHP-2-Myc and WT CagA-HA or a control vector. Cell lysates were subjected to anti-Myc immunoprecipitation. Phosphatase activities and amounts of SHP-2 protein in the precipitates were determined by pNPP assay and anti-SHP-2 immunoblotting, respectively. Intensities of chemiluminescence on the immunoblotted filter were quantitated using a luminescent image analyzer LAS1000 (FUJIFILM, Tokyo, Japan). The relative amount and phosphatase activity were calculated with the value for SHP-2 immunoprecipitates in the absence of CagA taken as a control.

Binding of phosphotyrosine-peptide to the SH2 domains of SHP-2 is considered to relieve the autoinhibitory mechanism and reveal its previously latent phosphatase activity (21–25). To determine whether this is also the case with the CagA-SHP-2 interaction, we expressed the Myc-tagged SHP-2 in the presence or absence of CagA in COS-7 cells and we immunoprecipitated SHP-2 with an antibody to Myc. The immune complex was then subjected to an in vitro phosphatase assay (26). Phosphatase activity of SHP-2 was potently stimulated when it formed a complex with CagA (Fig. 3C).

We next investigated whether the CagA-SHP-2 complex was involved in the induction of the hummingbird phenotype. Expression of CagA together with the phosphatase-defective SHP-2ΔPD-Myc, which still interacts physically with CagA via the SH2 domains (Fig. 3B), strongly reduced the induction of the hummingbird phenotype (Fig. 4A). To rule out the possibility that SHP-2ΔPD-Myc competitively inhibited the binding of CagA with SH2-containing molecules other than SHP-2 that are involved in the hummingbird phenotype, we ectopically expressed wild-type SHP-2, which, like SHP-2ΔPD-Myc, should act as a competitor for such molecules if they exist. Despite twofold higher levels than those of SHP-2ΔPD-Myc, wild-type SHP-2 did not inhibit the hummingbird phenotype by CagA (Fig. 4A), arguing against the existence of such molecules. Thus, complex formation of CagA and endogenous SHP-2 is an essential prerequisite for the induction of the hummingbird phenotype in AGS cells. Moreover, treatment of CagA-expressing AGS cells with an SHP-2-specific phosphatase inhibitor, calpeptin (27), inhibited the induction of the hummingbird phenotype (Fig. 4B), indicating that SHP-2 phosphatase activity is required for the morphological changes.

Figure 4

Role of SHP-2 in inducing hummingbird phenotype. (A) AGS cells were transiently co-transfected with WT CagA-HA and SHP-2-Myc or SHP-2ΔPD-Myc (n=3). Induction of the hummingbird phenotype and expressions of SHP-2-Myc and SHP2ΔPD-Myc proteins in the transfected AGS cells. Cells showing the hummingbird phenotype were counted in 10 different fields in each of three dishes (the area of one field being 0.25 mm2). (B) Cells transfected with CagA-HA were treated with either 100 μg/ml calpeptin or a vehicle for 1 hour before analysis (n=3). (C) Morphology of AGS cells transfected with SHP-2ΔSH2-Myc with or without the membrane-targeting sequence (Myr-) at 15 hours after transfection. (D) Cells expressing Myr-SHP-2ΔSH2-Myc were stained with an anti-Myc antibody.

Lastly, we investigated whether SHP-2 was capable of inducing the hummingbird-like morphological changes in the absence of CagA. Because CagA is cell membrane–associated (Fig. 1B), recruitment of SHP-2 by CagA may serve as a mechanism for relocalization of the cytoplasmic phosphatase to the cell membrane. Accordingly, we generated a membrane-targeted, constitutively active SHP-2 by adding the membrane-localization signal derived from v-Src (28) to the SHP-2 mutant lacking the SH2 domains (Myr-SHP-2ΔSH2-Myc) (20, 23, 24). Ectopic expression of Myr-SHP-2ΔSH2-Myc in AGS cells provoked cellular morphological changes indistinguishable from the hummingbird phenotype induced by CagA (Fig. 4, C and D) (13). On the other hand, constitutively active SHP-2 lacking the membrane-targeting signal (SHP-2ΔSH2-Myc) was incapable of inducing the morphological changes (Fig. 4C). Thus, membrane-tethering of activated SHP-2 was necessary and sufficient for the induction of the hummingbird phenotype.

We demonstrate here that H. pylori virulence factor CagA, which is translocated from the bacteria into gastric epithelial cells (5–9), can perturb mammalian signal transduction machineries and modify cellular functions by physically interacting with a host cell protein, SHP-2. SHP-2, like itsDrosophila homolog Corkscrew, is known to play an important positive role in the mitogenic signal transduction that connects receptor tyrosine kinases and ras (14, 15). Also, SHP-2 is actively involved in the regulation of spreading, migration, and adhesion of cells (29–32). Deregulation of SHP-2 by CagA may induce abnormal proliferation and movement of gastric epithelial cells, promoting the acquisition of a cellular transformed phenotype. Our results provide a molecular basis for the pathological actions of CagA on gastric epithelial cells.

CagA is noted for its amino acid sequence diversity among different H. pylori strains. The phosphorylation of the EPIYA motif is located in the repeat region of CagA and is expanded by duplication. Accordingly, the number and sequence polymorphism of the CagA phosphorylation sites, which collectively determine binding affinity of CagA to SHP-2, may be important variables in determining the clinical outcome of infection by different cagA+ H. pylori strains.

  • * To whom correspondence should be addressed. E-mail: mhata{at}


View Abstract

Navigate This Article