Molecular mechanisms of gastric epithelial cell adhesion and injection of CagA by Helicobacter pylori

H. pylori colonises the stomach of about half of the human world population, which is associated with chronic, often asymptomatic gastritis in all infected individuals. Depending on various criteria, more severe gastric diseases including peptic ulcer disease can occur in up to 10-15% of infected persons [1–3]. H. pylori infections are commonly diagnosed with a strong inflammatory response, but the bacteria evolved numerous mechanisms during evolution to avoid recognition and clearance by the host defence machineries, and if not treated with antibiotics, they can persist for life. H. pylori-induced gastritis is the strongest singular risk factor for developing cancers of the stomach; however, only a small proportion of infected individuals develop malignancy such as mucosa-associated lymphoid tissue (MALT) lymphoma and even gastric adenocarcinoma [1–3]. Gastric adenocarcinoma constitutes the second leading cause of cancer-associated death worldwide, and about 700,000 people die from this malignancy annually [3]. The clinical outcome of infection with H. pylori is dependent on a very complex scenario of host-pathogen crosstalk. Disease progression is determined by various factors including the genetic predisposition of the host, the bacterial genotype and environmental parameters [1–3]. The cellular and molecular mechanisms developed by H. pylori to undermine host defence strategies have been under intense investigation worldwide.

Clinical H. pylori strains are highly diverse both in their genetic information and potential to induce pathogenicity. Myriads of bacterial factors have been reported to influence the pathogenesis of H. pylori infections. There are two classical virulence determinants expressed by H. pylori, the CagA protein encoded by the cytotoxin-associated genes pathogenicity island (cag PAI) and the vacuolating cytotoxin (VacA). Secreted VacA can trigger various responses including pore formation in the host cell membrane, modification of endo-lysosomal trafficking, cellular vacuolation, immune cell inhibition and apoptosis. VacA's activities are highlighted in several review articles [1–4] and will not be discussed here. In the mid nineties, the cag PAI was entirely sequenced from various H. pylori strains and found to represent a 40-kb DNA insertion element in the chromosome, which is flanked by 31-bp direct repeats and carrying up to 32 genes [5, 6]. Large scientific interest concentrates on the CagA protein which is present in more virulent isolates, but is typically absent in less virulent H. pylori strains. Thus, CagA serves as a virulence marker for the cag PAI [7, 8]. Work in the last ten years has shown that the cag PAI encodes type-IV secretion system (T4SS) which injects CagA into target cells where it interferes with multiple host cell signaling cascades [9, 10]. Other well-described pathogenicity-associated mechanisms include flagella-driven bacterial motility, urease-mediated neutralization of pH, HtrA-mediated cleavage of E-cadherin, modification of host cell cholesterol, shedding of outer-membrane vesicles and peptidoglycan-dependent immune responses [1–3, 11–13]. In addition, H. pylori carries several classical surface adhesins permitting tight adherence of the bacteria to gastric epithelial cells. Here we review the various molecular adhesion strategies of H. pylori to gastric epithelial target cells which facilitate bacterial binding. We also discuss the structure and function of the T4SS, and how it makes contact with host cell surface factors to inject the CagA effector protein.

The OMP member BabA was the first H. pylori adhesin discovered. BabA mediates binding of the bacteria to Lewis B antigens, Leb [14] and related terminal fucose residues found on blood group O (H antigen), A and B antigens [15] that are expressed on the gastric mucosa. Multiple chromosomal loci and alleles for BabA have been reported to exist and Leb binding activity was shown to be facilitated by the BabA2 allele [14]. However, more recent work suggests that BabA1 alleles occur only very rarely and are difficult to detect [16] Substantial amino acid polymorphisms exist among BabA proteins expressed by different strains [17]. Diversity also appears with respect to the binding of strains to Leb and BabA expression, and binding affinity for A, B and O antigens correlates with blood group antigen expression of the host [15, 18]. A closely related gene babB, encodes for a translation product which has significant N- and C-terminal similarity to BabA. BabA and babB are nearly identical in their 5' and 3' regions but there is sequence divergence in their mid region [19] indicating that the central variable regions likely encode unique functions. Many Leb non-binding strains express silent babA gene sequences which may become activated by recombination into the babB locus forming chimeric babB/A genes [20]. BabA expression in vivo, however, appears to be highly dynamic. Experimental infection of Rhesus Macaques, mice and Mongolian gerbils resulted in loss of BabA expression and Leb binding [21, 22]. Post-infection strains isolated from gerbils contained a BabA2 protein that was modified by six amino acids from the strain used for inoculation. Complementation experiments confirmed these six amino acid residues are critical for binding to fucosylated blood group antigens [22]. In a recent study, experimental infection of Mongolian gerbils resulted in complete absence of expression of BabA at six months post-infection. Loss of BabA expression was attributable to nucleotide changes within the babA gene that resulted in a truncated BabA translation product [23]. BabA-mediated adherence of H. pylori to Leb on the surface of epithelial cells has been shown in vitro, using Leb transfected MDCK cells, and in vivo, using infection of Mongolian gerbils, to augment cag PAI-dependent H. pylori pathogenicity by triggering the production of proinflammatory cytokines and precancer-related factors [24] (Figure 1A). Thus, the expression of the BabA adhesin seems tightly connected to the onset of T4SS-related host cell responses in vivo. The presence of babA is associated with cagA and vacA s1 alleles, and strains that possess all three of these genes incur the highest risk of gastric cancer development [25].

Expression of sialyl-dimeric-Lewis × glycosphingolipid is upregulated upon infection with H. pylori and inflammation. This molecule also acts as a receptor for the pathogen and binding is mediated through the bacterial OMP member, SabA [26]. No binding to gangliosides was obtained with a SabA-negative mutant strain using a thin layer chromatography overlay assay [27]. Infection of the gastric epithelial cancer cell line MKN45 with H. pylori upregulated expression of the gene encoding β3 GlcNAc T5, a GlcNAc transferase essential for the biosynthesis of Lewis antigens. Overexpression of this gene in both the MKN45 and AGS gastric adenocarcinoma cell lines lead to expression of the SabA ligand, sialyl Lex, suggesting that H. pylori can modulate receptor expression [28]. SabA has been identified as a haemagglutinin that binds to sialylated structures found on the surface of red blood cells and there is a good correlation among strains between sialic acid dependent haemagglutination and sialyl Lex binding [29]. Like observations with BabA, a high level of polymorphism was reported in sialyl binding properties among clinical H. pylori isolates, which suggests that SabA adapts to its host depending on the mucosal sialylation pattern of the infected individual [29]. SabA has also been shown to mediate binding of H. pylori to sialylated moieties on the extracellular matrix protein laminin [30].

Two strongly homologous genes termed alpA and alpB were also characterized and shown to encode for integral OMPs. Adhesion experiments indicated that they are also involved in adherence of H. pylori to human gastric tissue biopsies [31]. OMP expression profiling of 200 strains from Germany revealed that virtually all clinical isolates produced the AlpA and AlpB proteins in contrast to many other OMPs that were produced at very variable rates [32]. Recently both AlpA and AlpB proteins have been shown to bind to mouse laminin in vitro and plasmid-borne alpA conferred laminin-binding ability on E. coli [33]. No other binding partners or receptors for AlpA and AlpB have yet been identified. The alpA/B l ocus has also been shown to influence host cell signaling and cytokine production upon infection. Deletion of alpA/B genes reduced IL-8 induction during infection with East Asian but not with Western H. pylori strains [34]. The alpA/B mutants poorly colonized the stomachs of C57BL/6 mice and were associated with lower mucosal levels of induced KC (the mouse name for human IL-8) and IL-6 [34]. In contrast to these results, in another recent study alpA and alpB gene mutants of H. pylori SS1 induced more severe inflammation than the parental strain in infected gerbils [33].

OipA (outer inflammatory protein A), encoded by the hopH gene, was initially identified as a surface protein that promoted IL-8 production in a T4SS-independent fashion [35]. OipA expression by H. pylori was shown to be significantly associated with the presence of duodenal ulcers and gastric cancer, high H. pylori density, and severe neutrophil infiltration [36]. Later studies identified that hopH knockout mutant strains adhered significantly less to gastric cancer cell lines, AGS and Kato-III, than wild-type strains, and complementation of the hopH gene restored the adherence properties of the hopH mutant [37]. The presence of oipA has also been shown to clearly enhance production of IL-8 in vitro but only in the presence of the cag PAI [32]. Further insights came from infection studies for up to 52 weeks in the Mongolian gerbil model system. All infected gerbils developed gastritis; however, inflammation was significantly attenuated in animals infected with the ΔcagA, but not the single ΔvacA or ΔoipA strains [38]. However, inactivation of oipA decreased nuclear localization of β-catenin, a factor involved in transcriptional up-regulation of genes implicated in carcinogenesis, and reduced the incidence of cancer in the gerbils. OipA expression was detected significantly more frequently among H. pylori strains isolated from human subjects with gastric cancer precursor lesions versus persons with gastritis alone [38]. The host receptor for OipA, however, remains unknown.

The hopZ gene encodes a protein which was shown by immunofluoresence to be located on the surface of the bacteria. A knockout mutant strain showed significantly reduced binding to the AGS cell line, compared to the corresponding wild-type strain [39]. Lack of production of HopZ did not affect the ability of the bacterium to colonize the stomachs of guinea pigs [40]. However, a role for HopZ in colonization in vivo has recently been proposed as deletion of hopZ reduced the ability of H. pylori to survive in a germ-free transgenic mouse model of chronic atrophic gastritis [41]. In addition, one of the few differences identified in H. pylori strains isolated from infected volunteers, was an OFF/ON switch in the phase-variable hopZ gene suggesting strong in vivo selection for HopZ during colonization [42]. Similar to OipA, the host receptor for HopZ has not yet been identified and will be a major challenging aim for future research.

Despite the above reports, it was assumed for many years that CagA can be randomly injected into gastric epithelial cells. However, this is obviously not the case because more recent studies showed that numerous host cell surface molecules are required for T4SS function, suggesting a sophisticated control mechanism through which H. pylori injects CagA [80]. The first identified host receptor for the T4SS was integrin β1 [61]. Based on a series of experiments including the use of integrin β1 knockout cell lines (GD25 and GD25β1), gene silencing RNAs, function-blocking antibodies and competition experiments with a well-known integrin β1 bacterial adhesin (InvA from Yersinia), compelling evidence was provided that integrin β1 plays a crucial role for injection of CagA during infection of several non-polarised AGS and mouse cell lines [61]. In line with these observations, various structural T4SS proteins have been demonstrated to bind to integrin β₁ in vitro, including CagL, CagA, CagI and CagY (Table 1). However, while very little is known about interactions of CagA and CagI with integrin, CagL has been investigated intensively. Like the human extracellular matrix protein fibronectin, CagL carries a RGD-motif shown to be important for interaction with integrin β₁ on host cells, as well as downstream signaling to activate tyrosine kinases including EGFR, FAK and Src [81]. However, mutation of the RGD-motif in CagL revealed no reduction of injected CagA^PY during infection in another study [82]. Another unsolved question is the structure of CagY with respect to which domain is exposed to the extracellular space. While the repeat domain in the middle of CagY on bacteria has been shown to be accessible to recognition by antibodies [64], in vitro binding studies and yeast-two hybrid screens revealed that the very carboxy-terminus interacted with integrin β₁ [82]. However, although it seems clear that each of the above factors exhibits an important functional role for injecting CagA, their interaction capabilities with integrin β1 during infection are unknown, and need to be investigated in future studies.

Another factor interacting with T4SS functions emerged from a recent study indicating that CagA binds directly to phosphatidylserine (PS) of the host cell and that this interaction is involved in CagA delivery into AGS cells based on saponin-fractionation experiments [83]. In vitro, the full-length protein binds to PS and this required a K-Xn-R-X-R motif present in the central region of CagA [83]. During H. pylori infection, the PS-binding protein annexinV and anti-PS antibodies both reduced CagA injection levels [83], suggesting that bacterial contact with PS is important for CagA delivery across the host membrane. Mutagenesis experiments showed that two arginine residues (R619 and R621) in the above motif are crucial for binding of purified CagA to PS and ensure membrane localization of transfected CagA in polarized MDCK cells [83]. Murata-Kamiya and co-workers proposed that the reported CagL-β1-integrin interaction may stabilize the CagA-PS interaction and may contribute to internalization of PS-bound CagA into host cells through activation of integrin signaling [83]. However, the injection mechanism of CagA into polarized MDCK cells is not completely clear because many other studies performed CagA transfection or biochemical fractionation experiments, but in most cases did not investigate phosphorylation of CagA [83–87]. The situation becomes even more puzzling because another group reported that MDCK cells are resistant to injection/phosphorylation of CagA upon infection [74]. Thus, the exact mechanism of CagA injection into polarized host cells requires further elucidation. Nevertheless, these findings point to the lipid bilayer in the host cell membrane as a second platform for T4SS-host cell interplay, rather then a "piercing"-like injection mechanism by the T4SS-pilus. Interestingly, recent findings from other groups indicated that cholesterol in lipid rafts, another component of the lipid bilayer, is also required for CagA translocation and pro-inflammatory signaling [88, 89]. Taken together, these studies indicate that there are at least three host cell factors being involved in proper T4SS functions of H. pylori.

H. pylori is one of the most successful human pathogens. Studies of host-bacterial interactions using their fundamental adhesins and the virulence factors CagA and T4SS have provided us with many detailed insights in processes ultimately connected to H. pylori colonisation and pathogenesis. The current opinion implies a model in which the major adhesins BabA, SabA and others make the initial and sustained host cell contact important for bacterial colonisation. Once intimate contact is established, the T4SS further interacts with specific host cell surface molecules including integrins and PS to facilitate injection of CagA, probably in cholesterol-rich microdomains on host cells, the lipid rafts. The above discussed studies indicate that at least three known host factors are involved in CagA injection (integrin, PS and cholesterol). However, it also seems clear that some cell lines are resistant to injection of CagA indicating that none of the reported host factors alone can fascililtate the injection process. We therefore assume that the T4SS injection mechanism is much more complicated than originally proposed and probably requires even more host factors, probably acting cooperatively. It should also be noted that almost all of the functional T4SS studies have been made in vitro using cultured cell lines, which are indeed very helpful. However, we are aware of only one in vivo study, in which phosphorylated CagA was isolated from biopsies of atrophic gastritis and in noncancerous tissues from H. pylori-positive patients using immunoprecipitation and Western blotting approaches [90]. Thus, more studies are clearly necessary to investigate under which circumstances and how CagA is injected during infection in vivo. In particular, it remains to be investigated if CagA injection occurs at the apical, basolateral and/or other sites of the gastric epithelium. In addition, it has been convincingly shown that CagA can be very efficiently translocated into certain immune cells in vitro (Table 2). Thus, future studies are necessary to investigate the importance of these findings in vivo. Finally, the evolutionary advantage of the T4SS for H. pylori is also not yet clear and needs to be investigated more thoroughly. For example, recent studies indicated that injected CagA enables H. pylori to grow as microcolonies adhered to the host cell surface even in conditions that do not support growth of free-swimming bacteria [91]. Thus, it appears that the H. pylori T4SS will continue to be a fascinating and rewarding research topic in future studies.

The authors declare that they have no competing interests.