Mouse Gastric Epithelial Cells Resist CagA Delivery by the Helicobacter pylori Type IV Secretion System

The initial step in bacterial infection is adherence of the bacterium to the target cell surface. Helicobacter pylori exploits the interaction of bacterial adhesin protein HopQ with human epithelial CEACAMs (CEACAM1, 5, and 6) to stably adhere to gastric epithelial cells, which is necessary for delivery of the H. pylori CagA oncoprotein into the epithelial cells via a type IV secretion system. In contrast to human CEACAMs, however, HopQ does not interact with Ceacam1 (mouse CEACAM1) in vitro or in CHO cells ectopically expressing Ceacam1. Since the mouse genome lacks Ceacam5 and Ceacam6, no significant HopQ–Ceacam interaction may occur in mouse gastric epithelial cells. Here, we found that the mouse stomach has a much lower expression level of Ceacam1 than the expression level of CEACAM1 in the human stomach. Consistently, mouse gastric epithelial cells resist CagA delivery by cagA-positive H. pylori, and the delivery is restored by ectopic expression of human CEACAM1 or CEACAM5 in mouse gastric epithelial cells. Thus, despite the fact that mice are routinely used for H. pylori infection studies, a low expression level of Ceacam1 in the mouse stomach together with the loss or greatly reduced interaction of HopQ with Ceacams make the mouse an inappropriate model for studying the role of H. pylori-delivered CagA in gastric pathogenesis, including the development of gastric cancer.


Introduction
Chronic infection with Helicobacter pylori plays an etiologic role in most if not all human gastric cancers, the third leading cause of cancer-related deaths worldwide, making the gastric pathogen the strongest risk factor for the development of gastric cancer [1,2]. H. pylori is classified into two major subgroups depending on the presence or absence of the cagA gene, which is located on the cag pathogenicity island (cagPAI), an approximately 40 kbp DNA segment that encodes the type IV secretion system (T4SS) [3], the bacterial microsyringe through which the cagA-encoded CagA protein is delivered into the attached gastric epithelial cells [4][5][6]. Infection with cagA-positive strains has been considered to be responsible for the development of gastric cancer at least in part by injecting CagA into gastric epithelial cells [7][8][9]. Once delivered inside the host cells, CagA is tethered to the inner plasma membrane where it acts as a protein scaffold that promiscuously interacts with multiple host proteins such as the oncogenic phosphatase SHP2 and polarityregulating kinase PAR1b and perturbs their physiological functions, thereby contributing to Figure 1. Relative expression levels of human and mouse mRNAs encoding membrane CEACAMs (human) and membrane Ceacams (mouse) in the stomach. Relative levels of human CEACAM1, CEACAM5, CEACAM6, CEACAM7, CEACAM19, CEACAM20, and CEACAM21 mRNA expression compared to ACTB (β-ACTIN mRNA) in human stomach (A). Relative levels of mouse Ceacam1, Ceacam2, and Ceacam10 mRNA compared to Actb (β-Actin) mRNA in mouse stomach (B). It should be noted that the mouse genome does not contain human CEACAM5, CEACAM6, and CEACAM7 orthologues, whereas the human genome does not contain a mouse Ceacam2 orthologue. Relative expression levels of human and mouse mRNAs encoding membrane CEACAMs (human) and membrane Ceacams (mouse) in the stomach. Relative levels of human CEACAM1, CEACAM5, CEACAM6, CEACAM7, CEACAM19, CEACAM20, and CEACAM21 mRNA expression compared to ACTB (β-ACTIN mRNA) in human stomach (A). Relative levels of mouse Ceacam1, Ceacam2, and Ceacam10 mRNA compared to Actb (β-Actin) mRNA in mouse stomach (B). It should be noted that the mouse genome does not contain human CEACAM5, CEACAM6, and CEACAM7 orthologues, whereas the human genome does not contain a mouse Ceacam2 orthologue.
To test the above-described possibility, we used four mouse gastric epithelial cell lines-YTN2, YTN3, YTN5 and YTN16-that were established from gastric carcinoma induced by treating C57BL/6 mice with N-methyl-N-nitrosourea (MNU) [30]. Consistent with the mRNA expression profile in the stomach (Figure 1), immunoblotting analysis using an anti-CEACAM1 antibody, which recognizes both human CEACAM1 and mouse Ceacam1, revealed that the expression level of Ceacam1 in mouse gastric epithelial cells was much lower than the expression level of CEACAM1 in AGS human gastric cancerderived epithelial cells (Supplementary Figure S1). We next conducted an in vitro infection experiment using gastric epithelial cells infected with the H. pylori cagA-positive Western standard strain NCTC11637 strain. (Hereafter "H. pylori" denotes the NCTC11637 cagA positive strain unless otherwise stated.) Since delivered CagA undergoes tyrosine phosphorylation by host cell kinases [31,32], the level of tyrosine-phosphorylated CagA represents the amounts of CagA delivered into gastric epithelial cells by H. pylori. As in many other studies, AGS cells were used as positive control cells for the infection experiment. Since a CagA delivery experiment is usually performed by incubation of gastric cells with H. pylori for 3-7 h, during which the amount of delivered CagA is proportional to the duration of incubation, we used 7 h incubation to see maximal CagA delivery by the bacterial infection. At 7 h after the onset of H. pylori infection, cell lysates were prepared from infected gastric epithelial cells and subjected to immunoblotting with an anti-CagA antibody that detects both CagA derived from host cell surface-attached H. pylori as well as CagA delivered into host cells and with an anti-phosphotyrosine (pTyr) antibody that specifically detects host cell-delivered CagA. The results of the experiment showed that a substantial amount of H. pylori was attached to the surface of human gastric epithelial cells, to which CagA was efficiently delivered into the cells, as determined by the tyrosine-phosphorylated CagA band. In striking contrast, only a small amount of H. pylori was attached to and only a small amount of CagA was delivered into mouse gastric epithelial cells ( Figure 2). The same results were reproducibly obtained when another H. pylori cagA-positive strain (G27) was used for the infection experiment (Supplementary Figure S2). The results indicated that, in contrast to the case of human gastric epithelial cells, H. pylori neither associated stably with mouse gastric epithelial cells nor efficiently delivered CagA into mouse gastric epithelial cells.

Figure 1.
Relative expression levels of human and mouse mRNAs encoding membrane CEACAMs (human) and membrane Ceacams (mouse) in the stomach. Relative levels of human CEACAM1, CEACAM5, CEACAM6, CEACAM7, CEACAM19, CEACAM20, and CEACAM21 mRNA expression compared to ACTB (β-ACTIN mRNA) in human stomach (A). Relative levels of mouse Ceacam1, Ceacam2, and Ceacam10 mRNA compared to Actb (β-Actin) mRNA in mouse stomach (B). It should be noted that the mouse genome does not contain human CEACAM5, CEACAM6, and CEACAM7 orthologues, whereas the human genome does not contain a mouse Ceacam2 orthologue.

Figure 2.
Infection with cagA-positive H. pylori is not able to deliver CagA into mouse gastric epithelial cells. Mouse gastric epithelial cells (YTN2, YTN3, YTN5, and YTN16) and AGS human gastric epithelial cells were infected with H. pylori NCTC11637 strain for 7 h at a MOI of 100. The cells were harvested and analyzed for CagA and tyrosine phosphorylation (pTyr) of CagA proteins. CagA is detectable at 135 kDa as indicated by the arrow in the anti-CagA and anti-pTyr immunoblot. Asterisks (★) denote the non-specific band observed for cell lysates of mouse origins. Representative Figure 2. Infection with cagA-positive H. pylori is not able to deliver CagA into mouse gastric epithelial cells. Mouse gastric epithelial cells (YTN2, YTN3, YTN5, and YTN16) and AGS human gastric epithelial cells were infected with H. pylori NCTC11637 strain for 7 h at a MOI of 100. The cells were harvested and analyzed for CagA and tyrosine phosphorylation (pTyr) of CagA proteins. CagA is detectable at 135 kDa as indicated by the arrow in the anti-CagA and anti-pTyr immunoblot. Asterisks (*) denote the non-specific band observed for cell lysates of mouse origins. Representative image of two independent experiments. Experiments were carried out in biological duplicates (n = 2) and the representative images are shown.

The CagA Protein Is Efficiently Translocated and Phosphorylated in Human CEACAM-Expressing Mouse Gastric Epithelial Cells
To examine whether impairment of CagA delivery into mouse gastric epithelial cells by H. pylori was due to the weak interaction of H. pylori HopQ with mouse Ceacams both qualitatively and quantitatively, we established transfectant cells that stably express hemagglutinin (HA)-tagged human CEACAM1 (the CEACAM1-4s isoform), human CEACAM5, or human CEACAM6 using two mouse gastric epithelial cell lines-YTN2 and YTN16which were arbitrarily chosen from the four mouse gastric epithelial cell lines as they displayed similar morphology and growth rates ( Figure 3). The broad bands corresponding to human CEACAM1, 5, and 6 shown in immunoblotting were consistent with the notion that they were heavily glycosylated by post-translational modification.

The CagA protein Is Efficiently Translocated and Phosphorylated in Human CEACAM-Expressing Mouse Gastric Epithelial Cells
To examine whether impairment of CagA delivery into mouse gastric epithelial cells by H. pylori was due to the weak interaction of H. pylori HopQ with mouse Ceacams both qualitatively and quantitatively, we established transfectant cells that stably express hemagglutinin (HA)-tagged human CEACAM1 (the CEACAM1-4s isoform), human CEA-CAM5, or human CEACAM6 using two mouse gastric epithelial cell lines-YTN2 and YTN16-which were arbitrarily chosen from the four mouse gastric epithelial cell lines as they displayed similar morphology and growth rates ( Figure 3). The broad bands corresponding to human CEACAM1, 5, and 6 shown in immunoblotting were consistent with the notion that they were heavily glycosylated by post-translational modification.  Figure S4). In the infection study, we detected a larger amount of CagA proteins in lysates prepared from YTN2-derived cells stably expressing CEACAM1 (YTN2/hCEACAM1) or CEACAM5 (YTN2/hCEACAM5), possibly due to increased binding of H. pylori to the surface of mouse gastric epithelial cells expressing these human CEACAM proteins ( Figure 4A Figure S4). In the infection study, we detected a larger amount of CagA proteins in lysates prepared from YTN2-derived cells stably expressing CEACAM1 (YTN2/hCEACAM1) or CEACAM5 (YTN2/hCEACAM5), possibly due to increased binding of H. pylori to the surface of mouse gastric epithelial cells expressing these human CEACAM proteins ( Figure 4A, Supplementary Figures S3 and S4), providing additional evidence that both human CEACAM1 and human CEACAM5 efficiently bind to H. pylori HopQ. Since H. pylori HopQ also interacts with human CEACAM6 in vitro [22,23], we also generated stable human CEACAM6 transfectants from YTN2 or YTN16 cells as parental cells and infected them with H. pylori for 7 h. Despite sufficient expression of human CEACAM6 on the stable transfectants, there was no evidence for CagA delivery ( Figure 4A, Supplementary Figure S5). From these observations, we concluded that H. pylori CagA can be translocated and tyrosine-phosphorylated into mouse gastric epithelial cells stably expressing human CEACAM1 or human CEACAM5 but not into cells stably expressing human CEACAM6. In this regard, it should be noted that the AGS cell lysates used for positive controls (third panel from the top) did not display bands corresponding to human CEACAM1, 5, and 6. This was simply because the immunoblotting experiment was carried out by using an anti-HA antibody that detects ectopically expressed human CEACAMs but not endogenous human CEACAMs. Indeed, stripping and reprobing of the same filter with an antihuman CEACAM1 antibody revealed the presence of endogenous CEACAM1 in AGS cells ( Figure 4B).
in vitro [22,23], we also generated stable human CEACAM6 transfectants from YTN2 or YTN16 cells as parental cells and infected them with H. pylori for 7 h. Despite sufficient expression of human CEACAM6 on the stable transfectants, there was no evidence for CagA delivery ( Figure 4A, Supplementary Figure S5). From these observations, we concluded that H. pylori CagA can be translocated and tyrosine-phosphorylated into mouse gastric epithelial cells stably expressing human CEACAM1 or human CEACAM5 but not into cells stably expressing human CEACAM6. In this regard, it should be noted that the AGS cell lysates used for positive controls (third panel from the top) did not display bands corresponding to human CEACAM1, 5, and 6. This was simply because the immunoblotting experiment was carried out by using an anti-HA antibody that detects ectopically expressed human CEACAMs but not endogenous human CEACAMs. Indeed, stripping and reprobing of the same filter with an antihuman CEACAM1 antibody revealed the presence of endogenous CEACAM1 in AGS cells ( Figure 4B).

HopQ Interaction Transiently Reduces the Cellular Levels of Cellular CEACAM Protein
During the course of the H. pylori infection experiment, we noticed that the level of ectopically expressed human CEACAM1 or CEACAM5 was substantially diminished after 7 h of infection with wild-type H. pylori ( Figure  . Hence, the reduction in human CEACAM was correlated with its ability to bind to H. pylori HopQ and subsequent delivery of CagA into host cells. Given this, we next investigated the relationship between the functional T4SS and reduction in human CEACAM1 expression. For this purpose, we infected AGS cells with wildtype H. pylori and its isogenic ΔvirD4 strain that lack the functional T4SS ( Figure 5). The   5 and lane 6). Hence, the reduction in human CEACAM was correlated with its ability to bind to H. pylori HopQ and subsequent delivery of CagA into host cells. Given this, we next investigated the relationship between the functional T4SS and reduction in human CEACAM1 expression. For this purpose, we infected AGS cells with wild-type H. pylori and its isogenic ∆virD4 strain that lack the functional T4SS ( Figure 5). The results of the experiment revealed that tyrosine phosphorylation of CagA was detectable within 1 h after wild-type H. pylori infection. Reduced CEACAM1 expression was also detected within 1 h after infection and the magnitude of CEACAM1 reduction was correlated with elevated levels of CagA tyrosine phosphorylation in a time-dependent manner ( Figure 5). In AGS cells infected with the ∆virD4 isogenic strain, the level of CEACAM1 expression was stable and constitutive during 7 h infection. Thus, a reduction in human CEACAM1 expression was substantially dependent on the presence of the functional T4SS ( Figure 5). within 1 h after wild-type H. pylori infection. Reduced CEACAM1 expression was also detected within 1 h after infection and the magnitude of CEACAM1 reduction was correlated with elevated levels of CagA tyrosine phosphorylation in a time-dependent manner ( Figure 5). In AGS cells infected with the ΔvirD4 isogenic strain, the level of CEACAM1 expression was stable and constitutive during 7 h infection. Thus, a reduction in human CEACAM1 expression was substantially dependent on the presence of the functional T4SS ( Figure 5). We then investigated the possibility that CagA delivered into gastric epithelial cells via the functional T4SS generates a signal that downregulates CEACAM1 or CEACAM5 expression. To test this idea, we infected AGS cells with wild-type H. pylori or its isogenic strain lacking the cagA gene (∆cagA). The results of the experiment revealed a slight increase in the level of human CEACAM1 expression upon infection with the ∆cagA strain ( Figure 6). Similar results were observed in Kato-III cells (Supplementary Figure S6). Reciprocally, ectopic expression of cDNA-delivered CagA in AGS cells gave a slight reduction in the level of human CEACAM1 (Figure 7). These results indicated that the functional T4SS plays a major role in the reduction in CEACAM1 and CEACAM5 by H. pylori infection, although the contribution of delivered CagA was marginal if any. In addition, to evaluate the time span of reduction in CEACAM1 expression upon H. pylori infection, AGS cells and Kato-III cells were infected with cagA-positive H. pylori for time frames of 0, 7, and 24 h, and CEACAM1 expression was evaluated in the above-mentioned time frames of infection. It was observed that the expression level of CEACAM1 decreased at 7 h as observed previously, which was restored by 24 h of infection (Supplementary Figure S7), suggesting that the expression of CEACAM1 oscillates throughout the infection period.  Quantification of the intensity of CEACAM1 relative to β-actin (right). Error bars indicate mean ± SD, n = 3. Data were analyzed by one-way ANOVA and the Bonferroni post hoc test. *** p < 0.001, ** p < 0.01, ns; non-significant.  Quantification of the intensity of CEACAM1 relative to β-actin (right). Error bars indicate mean ± SD, n = 3. Data were analyzed by one-way ANOVA and the Bonferroni post hoc test. *** p < 0.001, ** p < 0.01, ns; non-significant.

Discussion
We found in the present study that, in stark contrast to human gastric epithelial cells, mouse gastric epithelial cells resist T4SS-mediated CagA injection by H. pylori cagA-positive strains. Though not natural hosts of H. pylori, mice have been routinely used for in vivo studies of H. pylori infection. However, in most cases, H. pylori-infected mouse only develop lymphocytic gastritis without progression to severe gastric lesions including gastric cancer [33][34][35]. Only a few cagPAI-positive strains have been successfully adapted for long-term colonization in the mice stomach and infection of mice with these strains can lead to the development of chronic active gastritis with neutrophil infiltration, progressing to atrophy and metaplasia [36,37]. Nevertheless, even long-term infection of mice with such cagA-positive strains does not spontaneously induce gastric cancer, indicating that mice do not faithfully phenocopy gastric lesions induced by chronic cagA-positive H. pylori infection in human. Since transgenic expression of CagA in mouse gastric epithelial cells gave rise to the induction of gastric cancer [17,38], the failure of gastric cancer development in the stomach of mice infected with cagA-positive H. pylori does not seem to be due to the insensitivity of mouse gastric epithelial cells to the oncogenic action of CagA.

Discussion
We found in the present study that, in stark contrast to human gastric epithelial cells, mouse gastric epithelial cells resist T4SS-mediated CagA injection by H. pylori cagApositive strains. Though not natural hosts of H. pylori, mice have been routinely used for in vivo studies of H. pylori infection. However, in most cases, H. pylori-infected mouse only develop lymphocytic gastritis without progression to severe gastric lesions including gastric cancer [33][34][35]. Only a few cagPAI-positive strains have been successfully adapted for long-term colonization in the mice stomach and infection of mice with these strains can lead to the development of chronic active gastritis with neutrophil infiltration, progressing to atrophy and metaplasia [36,37]. Nevertheless, even long-term infection of mice with such cagA-positive strains does not spontaneously induce gastric cancer, indicating that mice do not faithfully phenocopy gastric lesions induced by chronic cagA-positive H. pylori infection in human. Since transgenic expression of CagA in mouse gastric epithelial cells gave rise to the induction of gastric cancer [17,38], the failure of gastric cancer development in the stomach of mice infected with cagA-positive H. pylori does not seem to be due to the insensitivity of mouse gastric epithelial cells to the oncogenic action of CagA. Instead, the results of the present study revealed that cagA-positive H. pylori failed and is incapable of delivering the CagA oncoprotein into mouse gastric epithelial cells.
It has been shown that the H. pylori adhesin HopQ is capable of strongly interacting with several CEACAM proteins including CEACAM1, CEACAM3, CEACAM5, and CEA-CAM6 and that the HopQ-CEACAM interaction is crucial for T4SS-mediated delivery of CagA by H. pylori. Since human CEACAM3 is specifically expressed on granulocytes [26], it does not seem to play a substantial role in the delivery of CagA into gastric epithelial cells. Previous studies also showed that the HopQ-binding ability of mouse Ceacam1 is much less than that of human CEACAM1 (if any) [22,23]. In the present study, we found that the mouse stomach expresses only a small amount of Ceacam1. Although mice possess Ceacam2, a Ceacam1 homologue (80% similarity) that is absent in humans, Ceacam2 is expressed at a very low level in the mouse stomach [29]. Together with the fact that the mouse genome does not possess Ceacam5 and Ceacam6 genes [24,25], the present study revealed that epithelial cells in the mouse stomach do not express Ceacams that are capable of strongly binding with H. pylori HopQ both qualitatively and quantitatively. As a consequence, while cagA-positive H. pylori binds to the surface of mouse gastric epithelial cells via adhesins such as BabA, SabA, and LabA, their weak interactions do not allow substantial CagA delivery into mouse gastric epithelial cells via the T4SS in the absence of HopQ-Ceacam interaction (Figure 8). mouse genome does not possess Ceacam5 and Ceacam6 genes [24,25], the present study revealed that epithelial cells in the mouse stomach do not express Ceacams that are capable of strongly binding with H. pylori HopQ both qualitatively and quantitatively. As a consequence, while cagA-positive H. pylori binds to the surface of mouse gastric epithelial cells via adhesins such as BabA, SabA, and LabA, their weak interactions do not allow substantial CagA delivery into mouse gastric epithelial cells via the T4SS in the absence of HopQ-Ceacam interaction (Figure 8). A previous study showed that H. pylori binds to the membrane surface of Chinese hamster ovary (CHO) cells ectopically expressing human CEACAM1 in a HopQ-dependent manner [23]. However, it was not examined in that study if CagA is delivered into human CEACAM1-expressing CHO cells. Ectopic expression of human CEACAM1, CEA-CAM5 or CEACAM6 in human embryonic kidney 293 (HEK293) cells reconstituted the HopQ/CEACAM interaction, which enabled translocation of H. pylori CagA into HEK293 cells [22,23]. The results of the present study also revealed that, although cagA-positive H. pylori failed to deliver CagA into mouse gastric epithelial cells, ectopic expression of a single human epithelial CEACAM such as CEACAM1 or CEACAM5 enables H. pylorimediated CagA delivery in mouse gastric epithelial cells, the bona fide targets of H. pylori. The results indicated that both membrane-spanning CEACAMs and GPI-anchored membrane CEACAMs are capable of provoking CagA delivery upon binding with HopQ. On A previous study showed that H. pylori binds to the membrane surface of Chinese hamster ovary (CHO) cells ectopically expressing human CEACAM1 in a HopQ-dependent manner [23]. However, it was not examined in that study if CagA is delivered into human CEACAM1-expressing CHO cells. Ectopic expression of human CEACAM1, CEA-CAM5 or CEACAM6 in human embryonic kidney 293 (HEK293) cells reconstituted the HopQ/CEACAM interaction, which enabled translocation of H. pylori CagA into HEK293 cells [22,23]. The results of the present study also revealed that, although cagA-positive H. pylori failed to deliver CagA into mouse gastric epithelial cells, ectopic expression of a single human epithelial CEACAM such as CEACAM1 or CEACAM5 enables H. pylorimediated CagA delivery in mouse gastric epithelial cells, the bona fide targets of H. pylori. The results indicated that both membrane-spanning CEACAMs and GPI-anchored membrane CEACAMs are capable of provoking CagA delivery upon binding with HopQ. On the other hand, ectopic expression of another GPI-anchored CEACAM, CEACAM6, did not support H. pylori-mediated CagA delivery into mouse gastric epithelial cells. Since HopQ binds to recombinant CEACAM6 in vitro with high affinity [22,23], the results of the present study indicate that GPI-anchored CEACAM6 may cause an allosteric change in the IgV-like N-terminal domain that dampens the HopQ-CEACAM6 interaction. Collectively, these observations indicated that the lack of stable interaction between H. pylori HopQ and mouse Ceacams, which dampens CagA delivery into gastric epithelial cells, is responsible for the observed differences in pathological changes in the stomach between mice and humans infected with cagA-positive H. pylori [33][34][35]39].
The level of tyrosine-phosphorylated CagA, a surrogate marker of host cell-delivered CagA, in mouse gastric epithelial cells that stably express human CEACAM1 or CEACAM5 was comparable to that of CagA in AGS cells upon infection with cagA-positive H. pylori. Consistently, a substantially larger amount of CagA was detected in cell lysates prepared from gastric epithelial cells ectopically expressing human CEACAM1 or CEACAM5 than in cell lysates prepared from parental mouse gastric cells, indicating enhanced H. pylori adhesion to the mouse cells via ectopically expressed human CEACAM1 or CEACAM5.
Aside from the role of HopQ-CEACAM interaction in CagA delivery, we also found a time-dependent reduction in the cellular level of human CEACAM1 following cagApositive H. pylori infection ( Figure 5). The presence of the functional T4SS but not CagA delivery is critical for CEACAM reduction. A reduction in human CEACAM1 in human gastric epithelial cells infected with H. pylori was observed for both endogenous and exogenous CEACAMs, suggesting that the downregulation occurs at the post-transcriptional level. Since the infection-associated CEACAM reduction, shown at 7 h after infection, was restored at 24 h after infection, the HopQ-CEACAM interaction provokes proteolytic cleavage of the extracellular domains of CEACAMs, which causes shedding or degradation of HopQ-bound CEACAMs. Whether the decreased HopQ-CEACAM interaction is beneficial to the host or bacterium is an important issue to be considered. The CEACAM degradation/cleavage might be advantageous for H. pylori as it would enable detachment of H. pylori from CagA-injected cells and movement to neighborhood epithelial cells as new targets. Alternatively, decreases in membrane-associated CEACAMs in H pylori-infected gastric epithelial cells may generate a negative feedback loop for the CagA pathogenic action of CagA, which prevents excess CagA delivery into a single epithelial cell that may provoke premature cell senescence or genomic instability depending on the functional status of p53 [16,40]. Molecular identification of the CEACAM proteases, which can be of human or bacterium origin, may provide a clue to understand why the H. pylori HopQhuman CEACAM1/5 interaction undergoes transient reduction during infection of H. pylori carrying the functional T4SS.
The repetitive expansion of CEACAM genes in primates and rodents has been assumed to be associated with positive selection [41,42]. Previous analyses have reported CEACAM genes encompass a large number of nonsynonymous Single Nucleotide Polymorphisms (SNPs) and Copy Number Variations (CNVs) with these variations having high differentiation in the population [43]. The genetic diversity in CEACAM has been reported as a probable factor of human susceptibility to meningococcal disease [44]. The diversity in the CNVs and SNPs may also influence H. pylori binding affinity and the T4SSmediated delivery of CagA. Future investigation on the relationship between variations in CEACAMs in humans and the ability of CagA injection could explain the variability in H. pylori infection and development of gastric cancer in the infected populations.
In summary, our study has revealed the mechanisms underlying species specificity in the pathogenicity of H. pylori. To investigate the role of CagA in gastric carcinogenesis using the mouse as an in vivo model of H. pylori infection, it should be important to specifically express human epithelial CEACAMs in mouse gastric epithelial cells because ectopic expression of CEACAMs in multiple distinct cell types, including immune cells, might artificially modify gastric lesions induced by chronic cagA-positive H. pylori infection.

Cell Culture and Transfection
AGS and Kato-III human gastric cancer cells were purchased from American Type Culture Collection (ATCC). YTN mouse gastric cancer cells were described previously [30]. AGS and Kato-III cells were cultured in RPMI 1640 containing 10% fetal bovine serum (FBS). YTN cells (YTN2, YTN3, YTN5, and YTN16) were cultured in DMEM containing 10% FBS and glucose 4500 mg/L using collagen type I-coated culture dishes. AGS cells and YTN16 cells were transfected with DNA using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA). YTN2 cells were transfected with DNA using Lipofectamine and PLUS reagent (Invitrogen, Carlsbad, CA, USA).

Helicobacter Strains and Culture
Helicobacter pylori standard strain NCTC11637 and its isogenic strains lacking the cagA gene (∆cagA) and lacking the virD4 gene (∆virD4) have been described previously [45,46]. H. pylori strains (NCTC11637 and G27) were grown in Brucella Broth supplemented with 10% FBS and incubated in a microaerobic atmosphere by using AnaeroPack helico (Mitsubishi Gas Chemical, Tokyo, Japan) at 37 • C.

Establishment of Stable Cell Lines
YTN2 and YTN16 cells were transfected with linearized hemagglutinin (HA)-tagged CEACAM expression vector (pcDNA3-CEACAM) together with linearized pBABE-puro vector. At 24 h post transfection, cells were incubated in culture medium containing puromycin, 8 µg/mL for YTN2 cells, and 12 µg/mL for YTN16 cells, respectively. Puromycinresistant colonies were picked up with cloning rings and then each colony was evaluated for the expression of CEACAM by immunoblotting.