Helicobacter pylori is a highly successful human pathogen that colonizes approximately 50% of the world's population. It is typically transmitted orally within families during early childhood and can persist for decades in its preferred niche, the gastric mucosa, despite triggering vigorous innate and adaptive immune responses. H. pylori infection causes chronic gastritis, which is asymptomatic in the majority of carriers but is considered a major risk factor for the development of gastric and duodenal ulcers and the gastric malignancies mucosa-associated lymphoid tissue lymphoma and gastric adenocarcinoma1. In addition to its association with cancer, H. pylori stands out from other Gram-negative bacterial pathogens in its ability to persist and establish chronic infection.

Contrary to long-held dogma, the stomach is not a sterile organ and is estimated to support a community of up to 200 bacterial species2. However, when H. pylori is present it is usually numerically dominant and is readily visible in gastric biopsy tissue sections as helical rod-shaped organisms covering the gastric epithelial cells and surrounding mucus. Initial colonization depends on bacterial urease activity and helical cell-shape modulation to penetrate the gastric mucus. Constitutive DNA and protein repair pathways, combined with bacterial genome diversification and attenuation of chemical radical production by the host cell, are now recognized as essential for persistence of the bacterium in this niche. The two known H. pylori toxins, vacuolating cytotoxin (VacA) and cytotoxin-associated gene A (CagA), have been the focus of attempts to understand H. pylori virulence. Although work on VacA has recently been reviewed3, we highlight new insights into functional interactions between VacA and CagA and the modulation of immune responses by VacA and another secreted virulence factor, the γ-glutamyl transpeptidase (GGT).

Besides its arsenal of virulence factors, persistence of H. pylori is strongly influenced by the ability of the bacterium to evade, subvert and manipulate the host's immune system. This bacterium can evade detection by several innate immune receptors through target modification and it can subvert other innate recognition pathways through the suppression of downstream signal transduction, whereas evasion of adaptive immunity is achieved by the modulation of effector T cell functions. In this Review, we discuss the remarkable ability of H. pylori to colonize and persist in the hostile environment of the human stomach through the interplay of several secreted virulence factors and the sophisticated manipulation of both innate and adaptive immune responses. We also highlight progress on understanding the consequences of persistence for both the bacterium and the host.

Colonization of the gastric mucosa

Escape from the acidic lumen. The stomach is a particularly challenging niche for bacterial habitation. In the lower bowel, which has a neutral or a slightly alkaline pH, bacterial density is highest in the lumen; by contrast, the production of gastric acid in the stomach, which results in a pH of 1–2, severely limits luminal colonization. Indeed, H. pylori can only survive for minutes in the stomach lumen and must quickly migrate to the gastric epithelial surface4. Similar to the intestine, the mucous layer in the stomach forms a physical barrier to bacterial penetration and probably acts as a scaffold for binding of the host's antimicrobial compounds5. Bacterial urease production is required for acid resistance through the localized production of ammonium ions, and flagellar motility allows penetration of the mucus6 (Fig. 1). Furthermore, urease activity facilitates flagellar motility through the mucous layer by changing the viscoelasticity properties of gastric mucins. At low pH, gastric mucins form a gel that effectively traps the bacteria, but urease-catalysed production of ammonium ions raises the pH to near neutral and the mucous gel transitions to a viscoelastic solution through which H. pylori can swim7,8. Regulators of motility, including chemotaxis9,10,11,12 and cell shape13,14,15, have been probed to discover additional colonization factors and to better define the optimal niche for H. pylori. Helical cell shape is thought to enhance motility through viscous media by a corkscrew mechanism, and cell shape mutants that have lost helical twist and/or curvature exhibit attenuated colonization13,14,15. Chemotaxis mutants have an altered localization, including lower numbers of bacteria that are in close association with gastric epithelial cells16 and that are deeply penetrating the gastric glands9. In addition to promoting clearance, the altered localization of chemotaxis mutants correlates with lower inflammation, impaired recruitment of CD4+ T cells and the absence of a T helper 17 (TH17) response10,16. Thus, the intimate association with the gastric epithelium promotes stable infection while simultaneously provoking more inflammation. Higher inflammation correlates with lower bacterial loads17, which suggests that H. pylori must actively manage its interaction with the host epithelium to avoid clearance and to persist at this site (Fig. 1).

Figure 1: H. pylori colonization and persistence factors.
figure 1

During initial infection of the stomach lumen, urease-dependent ammonia production locally raises the pH, which promotes bacterial survival and solubilizes the mucous gel to facilitate bacterial motility. Chemotaxis (driven by pH and possibly other gradients) and helical rod shape promote flagellar motility away from the acidic lumen to the preferred niche of Helicobacter pylori, which is on and adjacent to gastric epithelial cells. SabA, BabA and other variably expressed adhesins might shift the balance from mucus-associated to cell-associated bacteria. Cell-associated bacteria alter gastric epithelial cell behaviour through vacuolating cytotoxin (VacA), cytotoxin-associated gene A (CagA) and CagL, which all have multiple cellular targets. CagL interactions with the α5β1 cell surface receptor are mediated through the RGD motif of the protein, whereas interactions with the other cell surface receptor (αvβ5 integrin) are RGD-independent. The combined action of these three effectors leads to a number of changes in the gastric epithelial cell, including CagA- and VacA-dependent disruption of cell polarity, which can promote iron acquisition and cell extrusion; CagA- and CagL-dependent induction of chemokines and/or the gastric hormone gastrin; CagL-dependent inhibition of acid secretion by the (H++K+)ATPase and cellular proliferation, apoptosis and differentiation, which are mediated by all three effectors. In addition to CagL, CagA and CagY (not shown) have been demonstrated to bind α5β1 integrins, although the precise interaction surface is unknown. PS, phosphatidylserine; T4SS, type IV secretion system.

PowerPoint slide

Persistent colonization of the gastric mucosa. H. pylori uses diverse strategies to promote its survival despite robust immune responses. All H. pylori strains encode proteins that are important for detoxifying reactive oxygen species (ROS) — for example, catalase and superoxide dismutase — and H. pylori arginase limits nitric oxide production by macrophage-, neutrophil- and epithelial cell-derived nitric oxide synthase18,19. Moreover, multiple DNA repair pathways contribute to efficient colonization20 even while the surrounding host tissue accumulates DNA lesions21,22. Recent work has shown that H. pylori strains constitutively express DNA repair proteins such as RecA and therefore lack a classic SOS response to DNA damage23,24. Following DNA damage, H. pylori instead upregulates natural competence, which promotes chronic persistence, probably through enhanced genetic diversification23,25.

The H. pylori genome contains multiple intragenic and extragenic repeat sequences26. CagY, which is expressed on the cell surface and is required for the type IV secretion system (T4SS)-mediated translocation of the effector CagA (see below), can undergo recombination between internal repeat motifs that generally preserve the reading frame27. During experimental infection of mice and rhesus macaques, there is an accumulation of H. pylori CagY variants that have gained or lost T4SS activity. These results suggest that CagA translocation and the associated biological responses, including inflammation, can have both beneficial and detrimental effects on bacterial persistence, leading to selection for both retention and loss of T4SS activity28.

Among the 60 predicted outer-membrane proteins, the HOP family shares highly similar or identical sequences at their amino and carboxyl termini and includes several known or predicted H. pylori adhesins that promote binding to the gastric epithelium29. These shared sequences could promote intragenomic or intergenomic recombination. Sequencing of H. pylori HOP loci from human clinical strain collections has revealed probable gene conversion of the Lewis B binding adhesin gene babA with babB or babC30,31,32,33 and of sabB with the sialyl-Lewis binding adhesin sabA or omp27 (Refs 34, 35, 36). During experimental infection of rhesus macaques or mice with H. pylori, replacement of babA with babB produces strains that have lost the ability to bind immobilized Lewis B antigens32,37. Additionally, replacement of sabB with sabA leads to strains expressing two copies of sabA, which results in increased binding to sialyl-Lewis antigens on murine gastric tissues36. Some alleles of babA and sabA can undergo phase variation by slipped strand mispairing at dinucleotide sequences in the coding sequences or the homopolymeric tracts in their promoters, again leading to either loss of or elevated gene expression38,39,40. The carbohydrate antigens that are bound by these adhesins are expressed on the cell surface and/or on secreted glycoproteins such as mucin. Furthermore, some of these antigens, such as sialyl-Lewis antigens, are induced during inflammation. Phase variation by gene conversion and slipped strand mispairing leads to the development of subpopulations with variable adherence properties that could allow the pathogen to evade immune responses or resist shedding. This ability to generate diverse subpopulations might also affect transmission to new hosts.

Secreted toxins of H. pylori

VacA and CagA effectors. H. pylori strains actively manipulate host tissues and promote their own persistence through the activity of several secreted toxins, some of which are discussed below. VacA is a pore-forming toxin that disrupts cell polarity, promotes apoptosis of epithelial cells and inhibits T cell proliferation and effector functions3. The vacA gene is carried by all H. pylori strains, and sequence variation in several domains of its encoded protein is linked to varying expression levels and cell type-specific toxicity, as well as disease severity3. Another important toxin is CagA. Originally characterized as an immunodominant antigen from patients that are infected with highly virulent vacA alleles41,42, CagA is translocated into host cells by the Cag T4SS, which is encoded on the cagpathogenicity island (PAI)1,43. Strains that express CagA are associated with an increased risk of cancer, and transgenic expression of CagA in mice induces gastric carcinoma and other malignancies, which has led to its designation as a bacterial oncoprotein44.

CagA–VacA interactions. To function as an oncoprotein, CagA must persist in cells or act in a 'hit and run' manner. CagA is not readily detected in gastric cancer tissues45 and was therefore suggested to have a causative role only early in cancer progression. It has now been shown that translocated CagA is degraded by autophagy when the infecting strain has the m1 allele of VacA, owing to the ability of this VacA isoform to bind the cell surface receptor low-density lipoprotein receptor-related protein 1 (LRP1)46. VacA binding of LRP1 leads to a loss of reduced glutathione (GSH) in the cell and increased production of ROS. This in turn activates AKT kinase-dependent degradation of the tumour suppressor p53 and results in the induction of autophagy, leading to CagA degradation. Interestingly, autophagy is not activated in cells that express a variant form of the CD44 adhesion molecule46. These cells have increased intracellular levels of GSH owing to activation of xCT, a glutamate-cysteine transporter47, and therefore do not induce ROS or autophagy on VacA binding. CD44 is a cell surface marker that is associated with epithelial cancer stem cells and CagA can be detected in cells expressing variant-CD44 from patients with gastric cancer46. Paradoxically, tissue changes that are associated with H. pylori-induced gastric carcinogenesis, including the development of intestinal metaplasia, were thought to render the stomach less hospitable for H. pylori colonization, leading to lower colonization loads. However, H. pylori was shown to intimately interact with gastric progenitor cells in a mouse infection model48. This ability of H. pylori to colonize cells that have stem cell-like properties, and the persistence of CagA protein in these cells due to the activation of xCT, could provide a mechanism to account for a sustained role of H. pylori colonization and CagA in oncogenesis.

Once translocated into host cells, CagA can be tyrosine phosphorylated on EPIYA motifs49 by SRC and ABL family kinases. These two types of kinase are activated sequentially and in a tightly regulated manner, with SRC kinases mediating the initial, preferential phosphorylation of EPIYA-C (and EPIYA-D) motifs and ABL kinases phosphorylating any EPIYA motif later during the infection50. Phosphorylated CagA interacts with SHP2 tyrosine phosphatase and CSK kinase, whereas unphosphorylated CagA is known to interact with CRK adaptor, MET, growth factor receptor-bound protein 2 (GRB2), PAR1 (also known as MARK) and E-cadherin43. Collectively, these interactions lead to altered cell signalling and changes in cell polarity, extrusion, motility, proliferation and pro-inflammatory cytokine secretion1,43. As discussed below, many of these phenotypes have now been linked to the acquisition of nutrients by the bacterium to promote persistence and/or host pathology.

Under standard conditions, CagA expression is not required for stomach colonization, but it does promote inflammation in the Mongolian gerbil model51. Cag T4SS activity is often lost during murine infection, which complicates efforts to elucidate the pathophysiological roles of CagA during chronic H. pylori infection28,52,53. However, in a polarized cell culture model, CagA promotes increased basolateral uptake and transcytosis of transferrin, and VacA drives mislocalization of the transferrin receptor to sites of bacterial attachment to facilitate iron acquisition by the bacterium54. In cagA mutants, the formation of microcolonies on the apical surface of the cell requires iron supplementation, whereas this is not a requirement for wild-type bacteria, suggesting that CagA- and VacA-dependent cell polarity perturbations confer a nutritional benefit. Consistent with this hypothesis, CagA is required for efficient colonization of Mongolian gerbils under iron-limiting conditions54. Thus, CagA and VacA collaborate to promote efficient colonization in the iron-limited environment of the stomach and to moderate the pathological effects of CagA.

Ultrastructural insights into CagA secretion. Given the importance of CagA in persistence and pathology, there has been much interest in the mechanisms governing CagA delivery into host cells. Translocation of CagA from the bacterium to the host cell cytosol is mediated by the Cag T4SS. This is a contact-dependent secretion system that forms a large complex spanning the inner and outer membranes, which contains a pilus and several ATPases that promote T4SS assembly, pilus formation and CagA translocation55. The H. pylori Cag PAI encodes homologues or paralogues of the prototypical Agrobacterium tumefaciens Vir T4SS56, including the putative VirB7 (CagT), VirB9 (CagX) and VirB10 (CagY); inner- and outer-membrane-spanning channel subunits57; the major, VirB2 (CagC), and the minor, VirB5 (CagL), pilus subunits; and several additional H. pylori-specific Cag proteins that are required for CagA translocation (for example, CagH and CagI)58,59. Many Cag T4SS components have domain structures that are distinct from their Vir counterparts. For example, the VirB10 homologue CagY is considerably larger (220 kDa) and contains additional domains that are composed of repeat regions27. Additionally, transmission electron microscopy studies suggest that the three core cell envelope-spanning channel subunit homologues (CagY, CagT and CagX) localize to the pilus surface or to the base of the pilus60,61. A later study localized CagL and CagA to the tip of the pilus62. CagL was suggested to function as a tip adhesin that binds to α5β1 integrin (a host cell receptor for CagL) through an RGD motif and neighbouring sequences62,63 (Fig. 1). CagL binding and α5β1 integrin signalling were found to be required for both pilus extension and CagA translocation. Soluble RGD peptide could partially rescue the CagA translocation defect of a cagLRGA mutant, but not a ΔcagL deletion strain, suggesting a two-step model in which surface exposed CagL binds and activates α5β1 integrin, partially activating focal adhesion kinase (FAK) and SRC kinase, promoting pilus extension. In a second step, pilus-associated CagL further stimulates α5β1 integrin, in addition to stimulating the activities of FAK and SRC, thereby inducing CagA translocation and ensuring its rapid tyrosine phosphorylation by SRC.

The relationship between pilus formation and CagA secretion was further explored by field emission scanning electron microscopy, which readily detects Cag T4SS-dependent pili64. This technique confirmed the requirement of CagL for pilus formation and also revealed a hyperpiliated phenotype for cagH mutants, which, like cagL mutants, fail to translocate CagA64. One study showed that a ΔcagY mutant produces pili28, which is surprising because another study found that CagL is unstable in a ΔcagY mutant59. Currently, the mechanism by which CagL (or CagA, CagT, CagX and CagY) becomes surface exposed or incorporated into pili has not been explored. Collectively, these data suggest that pilus formation is not sufficient for CagA translocation, that pilus formation can proceed in the absence of at least one core T4SS component, and that there might be CagL-independent mechanisms of integrin activation, of pilus assembly and of CagA translocation in some strains. In fact, CagA, CagI and CagY were shown to bind α5β1 integrin in vitro and in yeast two-hybrid studies65. CagA, in particular, shows a much higher integrin-binding affinity in vitro than CagL; unlike CagL, CagA binding is not inhibited by the Yersinia entercolitica RGD-containing invasin, which would indicate that CagA and CagL use different integrin interaction surfaces. Antibodies that prevent integrin switching between a bent and an open configuration block CagA translocation65, and the α5β1 integrin interaction domain of CagA was shown to inhibit CagA translocation when provided as a soluble peptide66, indicating that a complex series of molecular interactions is required for integrin activation and CagA secretion. Further insights into the precise nature of the interactions between CagA, CagL and host interaction partners are beginning to be revealed by structural and molecular evolution studies (Box 1).

CagL effector functions. Although the CagA translocation defect of cagL mutants suggests that CagL has a structural role as part of the T4SS, a number of studies suggest additional functions67,68,69,70. Studies using purified recombinant CagL revealed that the protein can induce cell spreading and focal adhesion formation in a similar manner to the host extracellular matrix RGD-containing protein fibronectin69. CagL activates epidermal growth factor receptor (EGFR) more efficiently than fibronectin, and this was shown to result from RGD-dependent displacement of ADAM17 (disintegrin and metalloproteinase domain-containing protein 17) from α5β1 integrin, thus activating ADAM17 protease activity68. ADAM17 cleaves and releases surface-bound heparin-binding EGF-like growth factor. The resulting activation of EGFR in gastric epithelial cells represses (H++K+)ATPase activity (diminishing acid secretion) via a repressive nuclear factor-κB (NF-κB) binding site in the (H++K+)ATPase promoter. CagL also binds αvβ5 integrin independently of its RGD motif, which mediates the induction of gastrin70. Gastrin is a potent inducer of acid secretion, so simultaneous activation of gastrin and repression of the (H++K+)ATPase could explain the observed hypergastrinaemia and hypochlorhydria during chronic H. pylori infection. Finally, CagL RGD-dependent activation of α5β1 integrin activates the pro-inflammatory cytokine interleukin-8 (IL-8) independently of CagA translocation and nucleotide-binding oligomerization domain-containing 1 (NOD1) signalling67, indicating that CagL induces inflammation. An increased risk of cancer and ulcers, which is associated with the carriage of the Cag PAI, has mostly been attributed to CagA but these studies indicate that CagL may be an equally important effector. Furthermore, studies on the evolution of the Cag PAI suggest that additional Cag proteins can directly interact with host proteins through exposure on the cell surface or as novel effectors (Box 1).

Evasion of innate immune recognition

In addition to the multiple virulence factors that H. pylori uses to manipulate the host and ensure its persistence, the bacterium has evolved elaborate strategies to evade and subvert host immune defences, and these strategies are key to the success of this pathogen. The first defence barrier against H. pylori is the mucus produced by the epithelial cells lining the gastric mucosa and the innate immune cells that either reside in the gastric lamina propria under steady state conditions or are recruited there during infection. The detection of conserved pathogen-derived molecular structures (pathogen-associated molecular patterns (PAMPs)) by epithelial cells and innate immune cells occurs via four distinct classes of innate immune receptors (pattern recognition receptors (PRRs)) that differ in their subcellular localization, their coupling to downstream signalling pathways and their specificity. H. pylori avoids detection by several types of PRR that are crucial for the recognition of other Gram-negative enteropathogens.

Evasion and manipulation of TLR and RLR recognition. The best-defined among the four classes of PRR are the Toll-like receptors (TLRs). TLRs are either exposed on the surface of the plasma membrane or localized to endosomes, and they bind diverse classes of PAMPs. Among these are the ligands for TLR4 (lipopolysaccharide (LPS)), TLR2 (lipoteichoic acid and lipoproteins), TLR3 (double-stranded RNA and polyinosinic:polycytidylic acid), TLR5 (flagellin) and TLR9 (unmethylated CpG). H. pylori largely avoids recognition by TLRs, the best understood example of this being the evasion of TLR4 detection of LPS. H. pylori LPS is predominantly tetra-acylated and is 1,000-fold less biologically active than the hexa-acylated LPS of Escherichia coli71. Furthermore, the reduced biological activity of H. pylori LPS was recently shown to result from the removal of phosphate groups from the 1′- and 4′-positions of the lipid A backbone, which generates LPS that has less negative charge, resists binding by antimicrobial peptides (such as polymyxin B) and escapes detection by TLRs72. The phosphatases responsible for lipid A modification in H. pylori have been identified and the respective gene deletion mutants fail to colonize experimentally infected mice72. The TLR (or TLRs) involved in the residual detection of H. pylori LPS remain a matter of debate; whereas several studies using purified LPS have implicated the classical LPS sensor TLR4 (Refs 73, 74), other studies suggest that TLR2 is the main sensor of H. pylori LPS75,76 (Fig. 2). A clear interpretation of the published studies is complicated by the fact that they rely on models in which the respective TLR is ectopically expressed, often in the absence of its co-receptor, and the fact that both TLR4 and TLR2 participate in the detection of other non-LPS-related PAMPs of H. pylori77,78, which may contaminate LPS preparations.

Figure 2: H. pylori subversion of innate immune recognition.
figure 2

Helicobacter pylori harbours pathogen-associated molecular patterns (PAMPs) that have evolved to evade detection by pro-inflammatory Toll-like receptors (TLRs). H. pylori expresses tetra-acylated lipopolysaccharide (LPS), which is less biologically active than the hexa-acylated form that is typical of other Gram-negative pathogens owing to specific lipid A modifications that prevent detection by TLR4. H. pylori flagella are not detected by TLR5 owing to mutations in the TLR5 binding site of flagellin. The DNA of the bacterium, as well as a currently uncharacterized PAMP (and possibly H. pylori LPS) are detected by TLR9 and TLR2, respectively; these TLRs predominantly activate anti-inflammatory signalling pathways and anti-inflammatory interleukin-10 (IL-10) expression. 5′ triphosphorylated RNA is detected by the RIG-like helicase receptor family (RLR) RIG-I, which activates the transcription factors IRF3 and IRF7 to induce type I interferon (IFN; IFNα and IFNβ) expression. Bacterial RNA is also potentially detected by TLR8 in endosomes. The fucosylated DC-SIGN ligands of H. pylori suppress activation of the signalling pathways downstream of this C-type lectin receptor (CLR) and activate anti-inflammatory genes. Please note that not all depicted TLRs, RLRs and CLRs are necessarily expressed by the same cell type; only one generic cell type is shown here for simplicity. CARD, caspase activation and recruitment domain; DC-SIGN, dendritic cell-specific intercellular adhesion molecule-3 grabbing non-integrin; DD, death domain; MYD88, myeloid differentiation primary response gene 88; NF-κB, nuclear factor-κB; TIR, Toll/interleukin-1 receptor domain.

PowerPoint slide

Another putative H. pylori PAMP, flagellin, escapes recognition by TLR5 owing to modifications in the N-terminal TLR5 recognition domain79 (Fig. 2). Mutating residues 89–96 of Salmonella enterica subsp. enterica serovar Typhimurium flagellin to the corresponding flaA sequence of H. pylori abolishes its recognition by TLR5 (Ref. 80). Experiments using dendritic cells lacking TLR2, TLR4, TLR7 and TLR9, or combinations thereof, revealed that the innate immune system recognizes H. pylori nucleic acids77. Intracellular delivery of H. pylori DNA to dendritic cells by lipofection efficiently activates endosomally localized TLR9 (Ref. 77); however, the net effect of this activation is anti-inflammatory rather than pro-inflammatory81,82,83 (Fig. 2). TLR9 signalling has anti-inflammatory consequences in the early stages of infection in a mouse model81, and H. pylori DNA can even be used therapeutically to treat experimentally induced inflammatory bowel disease in mice82,83. The biological activity of H. pylori DNA may account for the inverse correlation between H. pylori colonization and the risk of developing inflammatory bowel diseases84, which has been attributed to a specific immunoregulatory sequence (TTTAGGG) that seems to be unique to the H. pylori genome82,83. H. pylori RNA sensing by dendritic cells has been suggested to be mediated by endosomally localized TLR8 (Ref. 77), as well as by a cytoplasmic nucleic acid sensor, RIG-I, which belongs to the RIG-like helicase receptor family (RLR). RIG-I seems to be required for the detection of 5′ triphosphorylated H. pylori RNA and the ensuing IRF3- and IRF7-dependent induction of type I interferons (IFNs) by dendritic cells77 (Fig. 2). It is currently unknown whether the activation of RIG-I and the H. pylori-induced production of type I IFNs has predominantly pro-inflammatory or anti-inflammatory effects.

The detection of H. pylori non-LPS ligands by TLR2 is another example of how H. pylori exploits the immune system for the induction of anti-inflammatory responses. Activation of TLR2 triggers the myeloid differentiation primary response gene 88 (MYD88)-dependent expression of several anti-inflammatory genes, most notably IL-10 (Ref. 77) (Fig. 2). Furthermore, Tlr2−/− micethat are infected with Helicobacter felis, a close relative of H. pylori, are better able to control experimental infections than wild-type mice and develop stronger T cell responses and T cell-driven immunopathology78. The effects of TLR2 gene deletion are phenocopied by Myd88−/− mice, indicating that the absence of anti-inflammatory signals induced by Helicobacter spp. is phenotypically dominant over the simultaneous lack of MYD88-dependent pro-inflammatory signals that are induced by other TLRs78.

Suppression of CLR-mediated signalling. In addition to its TLR and RLR ligands, H. pylori also harbours ligands for a third class of PRR, the C-type lectin receptors (CLRs). The best characterized of these are fucosylated ligands that bind to the CLR family member DC-SIGN85. In contrast to pathogens such as Mycobacterium tuberculosis and HIV, which express mannosylated DC-SIGN ligands and which activate pro-inflammatory downstream signalling pathways, the fucose residues of the DC-SIGN ligands of H. pylori actively dissociate the signalling complex downstream of DC-SIGN (consisting of the scaffold proteins LSP1, KSR1 and CNK and the kinase RAF1) and suppress pro-inflammatory signalling85 (Fig. 2). The differential biological effects of mannosylated and fucosylated DC-SIGN ligands are consistent with the proposed role of this PRR in tailoring and fine-tuning adaptive immunity to specific pathogens through the DC-SIGN- and RAF1-mediated acetylation of TLR-activated NF-κB86. Acetylation of the NF-κB subunit p65 both prolongs and increases IL-10 transcription to enhance anti-inflammatory cytokine responses86.

In summary, most of the available data support the conclusion that H. pylori avoids the induction of a strong pro-inflammatory response, as well as subsequent adaptive immunity and clearance, through two main mechanisms: the evasion of innate immune detection by pro-inflammatory TLRs and the preferential activation and manipulation of anti-inflammatory TLRs and CLRs. Together, these strategies promote the persistence of the organism.

Activation of NLRs and the inflammasome

The heterogeneous cytoplasmic family of NOD-like receptors (NLRs) comprise the fourth and final family of PRRs. NLRs detect a wide range of PAMPs and are essential for sensing host-derived damage-associated molecular patterns that are released following perturbations of tissue homeostasis87. Broadly speaking, NLRs fall into two categories: NOD1 and NOD2 recognize metabolites and activate the transcription factor NF-κB to induce innate and adaptive immune response genes88, whereas most other NLRs promote the assembly of multiprotein complexes called inflammasomes, which activate the cysteine protease caspase 1 (Ref. 89).

Detection of H. pylori peptidoglycan by NOD1. NOD1-mediated detection of H. pylori peptidoglycan was one of the first PRR-mediated innate immune pathways found to become activated on H. pylori infection90. Although initial work indicated that only T4SS-proficient H. pylori strains (harbouring a functional Cag T4SS) could deliver peptidoglycan and its active metabolite (meso-diaminopimelate-containing N-acetylglucosamine-N-acetylmuramic acid) into the cytoplasm of host epithelial cells90, it is now clear that outer-membrane vesicles (OMVs) from Cag PAI-negative strains of H. pylori can also target peptidoglycan to NOD1 (Ref. 91) (Fig. 3). Intragastric delivery of OMVs in mice induces innate and adaptive immune responses through a NOD1-dependent but TLR-independent mechanism91. The delivery of peptidoglycan by both OMVs and the T4SS occurs at cholesterol-rich lipid rafts91,92 (Fig. 3). In addition to the initially reported NOD1 signalling pathway resulting in NF-κB translocation to the nucleus90, NOD1 also activates the transcription factor AP1 via ERK- and p38-dependent pathways93. A direct consequence of NOD1 signalling is efficient killing of H. pylori by β-defensin 2, an antimicrobial peptide produced by NOD1-activated gastric epithelial cells94. The idea that H. pylori-induced activation of NF-κB depends on NOD1 has recently been challenged by a report showing that the introduction of a small interfering RNA (siRNA) specific for NOD1 does not alter the nuclear translocation of the NF-κB subunit p65 (Ref. 95). This new study provides evidence for an alternative NOD1-dependent signalling pathway, which activates the IRF3 and IRF7 transcription factors to induce the production of type I IFNs that are required for H. pylori-specific cytokine and chemokine responses, and infection control95 (Fig. 3).

Figure 3: H. pylori activation of NLRs, NF-κB signalling and caspase 1.
figure 3

Helicobacter pylori peptidoglycan is delivered to the cytoplasmic NOD-like receptor (NLR) NOD1 (nucleotide-binding oligomerization domain-containing 1) through either the type IV secretion system (T4SS; via its interaction with α5β1 integrin at cholesterol-rich lipid rafts) or through outer-membrane vesicles (OMVs). Activated NOD1 induces the AP1- and nuclear factor-κB (NF-κB)-dependent expression of pro-inflammatory cytokines and defensins and the IRF3- and IRF7-dependent expression of type I interferons (IFNs; IFNα and IFNβ). Additional unidentified H. pylori NLR ligands activate the inflammasome to induce cleavage of autoproteolytic pro-caspase 1 and the subsequent processing and release of mature interleukin-1β (IL-1β) and IL-18. IL-18 binds to its receptor on naive T cells and promotes FOXP3-dependent CD4+CD25+ regulatory T (TReg) cell differentiation and immune tolerance, which in turn prevents clearance and ensures persistent colonization of H. pylori. By contrast, IL-1β binding to its receptor induces T-box transcription factor (TBET)- or retinoid-related orphan receptor γt (RORγt)-dependent T helper 1 (TH1) and TH17 differentiation and the expression of the respective signature cytokines IFNγ and IL-17. Note that the pictured innate immune cell is a dendritic cell, whereas peptidoglycan-induced NOD1 signalling has been demonstrated in gastric epithelial cells. ASC, apoptosis-associated speck-like protein containing a CARD; CARD, caspase activation and recruitment domain; LRR, leucine-rich repeat domain; MAPK, mitogen-activated protein kinase; PAI, pathogenicity island; RICK, receptor-interacting serine/threonine kinase.

PowerPoint slide

Inflammasome activation by H. pylori. H. pylori harbours one or more ligands that trigger activation of the inflammasome and of caspase 1, a cysteine protease that controls the processing and secretion of two cytokine precursors, pro-IL-1β and pro-IL-18 (Ref. 87). Like other caspases, caspase 1 is synthesized as an inactive precursor, which becomes auto-proteolytically activated only after inflammasome assembly. Inflammasome assembly in turn is regulated by ligand binding and subsequent hetero-oligomerization of inflammasome sensors in conjunction with an adaptor molecule, ASC (apoptosis-associated speck-like protein containing a caspase activation and recruitment domain (CARD)), and pro-caspase 1 (Refs 87, 89). Whereas the inflammasome ligands and NLR sensors that are involved in H. pylori detection remain obscure, in vitro and in vivo studies have demonstrated that caspase 1 becomes activated in dendritic cells following co-culture with H. pylori, and that IL-18 and IL-1β are processed and released into the infected gastric mucosa96,97 (Fig. 3).

There is no evidence to suggest that H. pylori actively avoids inflammasome or caspase 1 activation. In fact, mice lacking caspase 1 can clear an experimental infection with H. felis or H. pylori more efficiently than wild-type animals and have more pronounced pathogen-specific T cell responses and T cell-driven immunopathology96. The explanation for this unexpected observation was provided by mouse strains lacking either IL-18 or its receptor, IL-18R: these mice phenocopy the effects of caspase 1 gene deletion; that is, they clear the infection better than wild-type mice owing to enhanced T cell responses and, as a consequence, they develop more severe immunopathology96,98. Further analysis revealed that IL-18 is crucial for inducing CD4+CD25+FOXP3+regulatory T cell (TReg cell) responses to H. pylori (Fig. 3), which in turn restrict excessive effector T cell activation and promote persistence98. Interestingly, IL-1β (the other caspase 1 cytokine substrate) apparently opposes IL-18 function. Il1r−/− animals that lack the receptor for IL-1β fail to launch H. pylori-specific TH1 and TH17 responses, and cannot control an experimental infection (even when vaccinated against H. pylori before challenge) and, as a consequence, are protected against even the mildest forms of infection-associated immunopathology96. These data corroborate an earlier report showing that stomach-specific expression of human IL-1β is sufficient to induce gastric inflammation and gastric cancer in transgenic mice99, and they also explain why promoter polymorphisms that are associated with increased steady-state levels of IL-1β predispose carriers to a high risk of gastric cancer100. Furthermore, the effects of Il1r gene deletion seem to be phenocopied by H. pylori-infected mice lacking the inflammasome adaptor ASC101. In conclusion, detection of H. pylori by NLRs and subsequent activation of the inflammasome and downstream signalling pathways is crucial for efficient infection control (in the case of NOD1 signalling and inflammasome-mediated IL-1β secretion) and at the same time ensures the restriction of excessive T cell responses and immunopathological tissue damage (by inflammasome-mediated IL-18 secretion).

Modulation of effector T cell responses

Suppression of TH1- and TH17-mediated immunity. Experimental infection studies have highlighted the elements of the innate and adaptive immune systems that are required for the control of H. pylori infections, particularly for the generation of vaccine-induced protective immunity17,102,103,104,105,106,107. Whereas B cells and antibodies are dispensable for H. pylori control103,104,107 (at least for the suboptimal, non-sterilizing reduction in colonization by 1–2 orders of magnitude that is considered the gold standard in the H. pylori vaccinology field (Box 2)), it is now clear that CD4+ effector T cells (not to be confused with the CD4+ TReg cells mentioned above), and in particular TH1 and TH17-polarized effector T cell subsets and their signature cytokines, are crucial for the control of this infection17,102,106. The same T cell subtypes have been implicated in promoting the immunopathological changes of the chronically infected gastric mucosa that manifest histologically as atrophic gastritis, compensatory epithelial hyperplasia and intestinal metaplasia in experimentally infected animals and symptomatic human carriers108,109.

Two virulence factors have been specifically implicated in the manipulation and inhibition of human T cells (Fig. 4). VacA inhibits T cell proliferation by interfering with the T cell receptor–IL-2 signalling pathway at the level of the Ca2+/calmodulin-dependent phosphatase calcineurin110,111. VacA prevents nuclear translocation of the T cell transcription factor NFAT and its subsequent transactivation of T cell-specific immune response genes110,111 (Fig. 4). Further studies have since identified β2 integrin (CD18) as the receptor for VacA on human T cells112; β2 integrin associates with CD11a on T cells to form the heterodimeric transmembrane receptor LFA1 (lymphocyte function-associated antigen 1). H. pylori exploits the recycling of LFA1 to facilitate VacA uptake112 in a manner that depends on protein kinase C-mediated serine/threonine phosphorylation of the β2 integrin cytoplasmic tail113. The other H. pylori virulence determinant that is implicated in T cell inhibition is GGT114,115. Similar to VacA, GGT is a secreted factor that blocks the proliferation of T cells through a mechanism that involves the inhibition of cyclin-dependent kinase activity in the G1 phase of the cell cycle through the disruption of the RAS signalling pathway114,115 (Fig. 4).

Figure 4: H. pylori impairs T cell-mediated immunity by direct and indirect mechanisms.
figure 4

All strains of Helicobacter pylori express the secreted virulence factors vacuolating cytotoxin (VacA) and GGT to directly inhibit T cell activation, proliferation and effector functions. Hexameric VacA binds to the β2 integrin subunit of the heterodimeric transmembrane receptor lymphocyte function-associated antigen 1 (LFA1); the receptor complex is internalized following protein kinase C (PKC)-mediated serine/threonine phosphorylation (P) of the β2 integrin cytoplasmic tail. Cytoplasmic VacA prevents nuclear translocation of NFAT by inhibiting its dephosphorylation by the Ca2+/calmodulin-dependent phosphatase calcineurin, and thereby blocks interleukin-2 (IL-2) production and subsequent T cell activation and proliferation. GGT arrests T cells in the G1 phase of the cell cycle, preventing their proliferation. Both VacA and GGT also indirectly prevent T cell immunity by re-programming dendritic cells (DCs); VacA- and GGT-exposed dendritic cells produce IL-10, and induce the FOXP3- and contact-dependent differentiation of T cells into CD4+CD25+FOXP3+ regulatory T (TReg) cells while simultaneously preventing T helper 1 (TH1) and TH17 differentiation. TReg cell differentiation further depends on dendritic cell-derived IL-18, which is processed upon activation of caspase 1, and binds to its receptor on naive T cells. Depicted interactions at the T cell–dendritic cell synapse include major histocompatibility complex class II (MHCII) binding to the T cell receptor (TCR) and binding of co-stimulatory molecules CD80 and CD86 to CD28. Dendritic cell-derived and/or TReg cell-derived IL-10 further suppresses TH1 and TH17 effector functions. Note that the direct effects of VacA on T cells seem to be specific to humans, whereas indirect effects of VacA and GGT on T cells through dendritic cells have only been documented in the murine system. CaM, calmodulin; CnA, calcineurin A; GGT, γ-glutamyl-transpeptidase; RORγt, retinoid-related orphan receptor γt; TBET, T-box transcription factor.

PowerPoint slide

Skewing of T cell responses. Both VacA and GGT also affect T cell activity in an indirect manner by promoting the preferential differentiation of naive T cells into TReg cells116. TReg cell differentiation in response to H. pylori infection requires the direct interaction of naive T cells with 'tolerogenic' dendritic cells that have been exposed to H. pylori, either in the gastric mucosa or in the stomach-draining (gastric or mesenteric) lymph nodes98,117,118. Dendritic cells that have been exposed to H. pylori fail to induce effector T cell responses of the TH1 and TH17 type in vitro and in vivo; instead, such dendritic cells preferentially induce the expression of the TReg cell-specific transcription factor FOXP3, the surface marker CD25 and the anti-inflammatory cytokine IL-10 in naive T cells98,107,117 (Fig. 4). Such peripherally induced TReg cells profoundly affect the control of H. pylori, as shown in chronically infected patients119,120,121,122,123 and by animal experiments in which TReg cells are systemically depleted in infected hosts52,107. TReg cells accumulate in H. pylori-infected human gastric mucosa119,121, especially in children123 and in asymptomatic carriers122, and effectively suppress H. pylori-specific memory T cell responses120.

Experimental depletion of TReg cells facilitates the clearance of H. pylori in infected animals52 and enhances vaccine-induced protective immunity in vaccinated mice107. The TReg-facilitated persistence of H. pylori requires T cell-specific expression of IL-10; in fact, Il10−/− mice and a strain lacking IL-10 expression in the CD4+ T cell compartment are capable of spontaneously controlling experimental infections52,78,124. The efficient control or even clearance of H. pylori in animals invariably comes at the price of enhanced gastric immunopathology (gastritis and epithelial changes such as atrophy and intestinal metaplasia). Interestingly, an analogous observation has been reported for human carriers, which either accumulate large numbers of IL-10-producing, H. pylori-specific TReg cells and are colonized heavily (asymptomatic carriers), or develop gastric ulcers because their TReg cell response is inadequate122. The induction of H. pylori-specific tolerance to dendritic cells, which seems to be a prerequisite for the skewing of T cell responses (at least in experimental models98,117), requires the activity of both VacA and GGT116 (Fig. 4). Although the exact mechanism of VacA- and GGT-specific dendritic cell tolerance remains unclear, the newly assigned function of both factors in TReg cell induction and persistence is consistent with previous reports showing that gene deletion mutants lacking VacA or GGT have colonization defects relative to their parental VacA- or GGT-proficient wild-type isolates125,126.

Systemic consequences of immunosuppression

The active inhibition and manipulation of adaptive T cell-driven immune responses by H. pylori has various consequences for the host. The persistence mechanisms of H. pylori are dominant enough to override the protective effects conferred by H. pylori-specific vaccination; a challenge infection can only be cleared (or at least strongly reduced) by vaccinated mice if TReg cells or dendritic cells are depleted107. These observations partly explain the difficulties and obstacles faced in H. pylori vaccine development (Box 2). An interesting side effect of H. pylori-specific immunomodulation and manipulation is evident in Western societies from which H. pylori is gradually disappearing owing to reduced transmission rates, the frequent use of antibiotics in childhood and generally improved sanitation conditions127. In these populations, the incidence of allergic asthma, other allergic disease manifestations and chronic inflammatory diseases is steadily increasing, and an inverse association with H. pylori colonization has been documented for allergic asthma128,129,130,131,132 and inflammatory bowel diseases84 (Box 3). Although the exact mechanisms underlying this inverse association remain to be elucidated, the idea that H. pylori-induced immune regulation and manipulation are causally linked to protection from such immune disorders is compelling (Box 3). The fact that TReg cells that have been isolated from H. pylori-infected mice are sufficient to protect naive recipients against allergen-induced asthma in adoptive transfer models argues in favour of TReg-mediated cross-protection against allergen-specific immune responses98,133. Further work in this area is urgently needed to reveal the intricate interactions of this extraordinarily well-adapted persistent pathogen with the host adaptive immune system.

Conclusions and future perspectives

The work summarized in this Review outlines how H. pylori uses a combination of virulence factors and immune subversion and manipulation mechanisms to colonize and persist in the challenging environment of the gastric mucosa. Recent experimental work has elucidated exciting details on the structure and function of the T4SS, the pleiotropic effects of CagA delivery, the CagA-independent effects of the secretion system and newly discovered functions of the extracellular effector CagL. The role of cell shape and chemotaxis in persistent colonization is now well documented with respect to the genes involved. Progress in other areas, particularly H. pylori-specific vaccine development, has suffered from setbacks in Phase I clinical trials and from a lack of continuous industry support. The necessity of overriding the persistence strategies of the bacterium has been identified as a major challenge in H. pylori-specific vaccine development. Interest in the field has shifted towards gaining a better understanding of the benefits (suggested from epidemiological studies) that the infection may bestow on the large majority of asymptomatic carriers, and experimental evidence has been forthcoming to support such claims. In particular, it is now becoming increasingly clear that the virulence factors used by H. pylori and the mechanisms that are exploited to override T cell-driven immunity and to ensure persistent infection have systemic immunomodulatory effects that probably explain the benefits of the infection to asymptomatic carriers. The molecular mechanisms that allow H. pylori to suppress T cell activation through production of VacA and GGT, and to skew T cell responses towards regulatory T cells, are increasingly well understood.

Other aspects of the H. pylori–host interaction have received surprisingly little, if any, attention. These include the specifics of inflammasome activation by H. pylori and of innate immune activation by H. pylori in general, the molecular basis of host specificity, and the relative (or perhaps additive) contributions of its direct and indirect (inflammation-mediated) carcinogenic properties to gastric cancer development. Another important aspect of H. pylori biology that has been mostly ignored relates to its transmission. Although several independent, mostly older, studies indicate that mothers serve as the predominant source of their children's H. pylori infection, the transmission route remains unclear. Furthermore, little is currently known about the vast differences in the risk of gastric cancer development among human populations (often closely related and physically close), which is likely to be influenced by human genetic predisposition, population ecology and behaviour. In summary, many of the peculiarities that set H. pylori apart from other Gram-negative enteropathogens remain underexplored and deserve further work.