Helicobacter pylori–Induced IL-1β Secretion in Innate Immune Cells Is Regulated by the NLRP3 Inflammasome and Requires the Cag Pathogenicity Island

Helicobacter pylori infection remains a highly prevalent infection worldwide (1) and the chronic inflammation elicited by the bacterium is one of the main causes of gastric cancer. Progression to gastric cancer has been shown to be related to the severity of the host’s inflammatory response, which is influenced by bacterial virulence factors. Among them, the presence of the cag pathogenicity island (cagPAI) and vacuolating cytotoxin A (VacA) has been associated with more severe gastritis and increased risk of gastric ulcer and stomach cancer (2–5).

The inflammatory response toward H. pylori is characterized by the recruitment of different immune cells, mainly dendritic cells (DCs), neutrophils, macrophages, and B and T lymphocytes to the site of infection (6). Cells of the innate immune system recognize pathogens via conserved structures known as pathogens associated molecular patterns and through different pattern recognition receptors, such as extracellular or endosomal membrane bound TLRs, cytosolic nucleotide-binding oligomerization domain (NOD)-like receptors (NLRs) and intracellular RIG-like receptors. The recognition of H. pylori by TLRs has been extensively studied, whereas the role of NLRs is mostly unknown. H. pylori was reported to be recognized via NOD1 in epithelial cells (7) and via NOD2 in bone marrow–derived DCs (BMDCs) (8). Many NLRs are involved in the assembly of a multiprotein complex known as inflammasome. The inflammasome consists of a cytoplasmic sensor protein (e.g., NLR family, pyrin domain–containing 1 [NLRP1], NLRP3, or NLR family, CARD domain–containing 4 [NLRC4] of the NLR family), the adaptor protein apoptosis-associated speck-like domain containing a caspase recruitment domain (ASC) and the effector protein procaspase-1. Pathogens as well as noninfectious stimuli can activate the inflammasome. Two signals are required for inflammasome oligomerization. A first signal, often referred to as the priming signal, leads to the activation of NF-κB and thereby transcription of pro–IL-1β and pro–IL-18. This signal can be induced by TLR agonists like LPS or peptidoglycan, NOD2 agonists like muramyldipeptide or by cytokines like TNF. Formation of the inflammasome complex is induced by a second, independent signal that is specific for the different types of inflammasomes. Assembly and thereby the activation of the inflammasome induces the cleavage and activation of caspase-1, which promotes the processing and secretion of the active IL-1β and IL-18 (9). Besides, caspase-1 also induces pyroptosis, an inflammation-associated form of cell death, in different cell types including DCs (10).

IL-1β and IL-18 have been extensively linked to gastric carcinogenesis. Both cytokines were also shown to be important in the context of H. pylori infection. Polymorphisms in the IL-1 and IL-18 genes inducing elevated levels of these inflammasome-regulated cytokines increase the predisposition to the development of gastric cancer (11–13). Stomach-specific expression of human IL-1β in a transgenic mouse model resulted in inflammation, dysplasia and gastric cancer (14) and more recently, IL-1β induced by H. pylori was shown to enhance gastric carcinogenesis in mice (15), whereas IL-18 has been linked to increased metastasis and immune escape of gastric tumor cells (16). IL-1β enables H. pylori to colonize the gastric corpus by inhibiting acid secretion (17) and IL-1 receptor knockout mice (IL-1R^−/−) were found to be protected against Helicobacter felis infection–induced gastritis. In contrast, IL-18^−/− mice showed enhanced immunopathology associated with lower H. felis colonization levels (18), suggesting a dual role for the inflammasome during Helicobacter infection.

In a recent report, H. pylori cagPAI and the interaction between TLR2/NOD2 and NLRP3 were reported to regulate IL-1β production in H. pylori–infected DCs (8). However, the cellular mechanisms implicated in the activation of the inflammasome, as well as the H. pylori factor acting as a second signal activating NLRP3 remain elusive.

In the current study, we show that H. pylori exhibits a complex intracellular pattern of NLRP3 inflammasome activation involving reactive oxygen species (ROS) production, lysosomal destabilization and potassium efflux (K⁺). Inflammasome activation depends on several bacterial virulence factors, namely an intact cagPAI and the presence of VacA. Finally, we show that NLRP3 is important for H. pylori colonization and inflammation-associated pathology in vivo.

The H. pylori strains G27 (19), PMSS1, and SS1 (20) were grown on Wilkins–Chalgren blood agar plates supplemented with 10% horse serum and Dent supplement (Oxoid) under microaerobic conditions (10% CO₂, 5% O₂, 85% N₂; 37°C).

The PMSS1 isogenic mutant strains PMSS1ΔgGT (21), PMSS1ΔVacA (21) and PMSS1ΔCagE (22), G27ΔBabA, G27ΔSabA, G27ΔBabASabA, G27ΔgGT (23), G27ΔVacA, G27ΔCagA, and G27ΔCagE were grown on Wilkins–Chalgren blood agar plates containing 20 μg/ml kanamycin or 15 μg/ml chloramphenicol, respectively, for selection under the same conditions.

Salmonella typhimurium was grown on Columbia blood agar plates (BD Biosciences).

For generation of murine BMDCs and bone marrow–derived macrophages (BMMs), bone marrow was obtained from femur and tibia of donor mice. Cells were cultured for 7 d at 37°C in RPMI 1640 medium or DMEM, respectively, containing 10% FCS (Sigma-Aldrich), 1% penicillin/streptomycin (Life Technologies), 50 μM 2-ME (Sigma-Aldrich), and 20 ng/ml GM-CSF (Miltenyi Biotec) to generate BMDCs or 15% M-CSF containing supernatant (v/v) from M-CSF–secreting L929 fibroblasts (Lcell conditioned medium) to generate BMMs.

Neutrophils were isolated from bone marrow cells by magnetic cell labeling (Miltenyi Biotec) with an anti-Ly6G Ab (Hycult Biotech).

Human PBMCs were obtained, after informed consent, from whole blood of H. pylori negative donors by biocoll (Biochrom) density gradient separation.

Monocytes, B cells, T cells, and NK cells were isolated from PBMCs with monocyte isolation kit II, B cell isolation kit II, T cell isolation kit II, and NK cell isolation kit (Miltenyi Biotec), according to manufacturer’s instructions.

Cells were plated in antibiotic free medium and stimulated with LPS (Sigma-Aldrich) 3 h prior to treatment unless otherwise indicated. LPS (10 ng/ml) was used for BMDCs and BMMs prestimulation, whereas neutrophils were stimulated with 50 ng/ml LPS. PBMCs, B cells, T cells, and NK cells were incubated with 5 ng/ml LPS.

Infections were carried out at the indicated time points with H. pylori at multiplicity of infection (MOI) 1, 5, 10, and 50. Stimulation with 5 mM ATP for 1 h and infection with S. typhimurium at MOI 10 for 6 h were used as controls.

Chemical inhibitors were added 30 min before cell infection or stimulation. The caspase-1 inhibitor Z-YVAD-FMK (Calbiochem) was used at 10 μM, KCl (Roth) at 50 mM, cytochalasin D (Sigma-Aldrich) at 2.5 and 5 μM, whereas the ROS inhibitors Ebselen (ALX-270-097; Enzo) and (2R, 4R)-4-aminopyrrolidine-2, 4-dicarboxylate (APDC) (A7361; Sigma-Aldrich) were used at 10 and 50 μM, and 50, 250, and 500 μM, respectively.

To discriminate effects induced by dead bacteria, H. pylori was either heat-inactivated for 5 min at 95°C or fixed in 4% paraformaldehyde (PFA) for 4 h at 4°C.

H. pylori uptake was evaluated by the gentamicin protection assay. Briefly, BMDCs were infected for 60 min and then incubated for 45 min at 37°C in medium containing gentamicin (100 μg/ml) to kill extracellular bacteria. Cells were then washed in PBS and lysed in 0.5% saponin/PBS, and the number of intracellular bacteria was determined by plating on Wilkins–Chalgren blood agar plates. Supernatants were plated in parallel as control for the antibiotic treatment.

Cell supernatants and stomach extracts were analyzed for cytokine (IL-1β, IL-18, IL-10, and TNF) secretion by ELISA (eBioscience), according to supplier's instructions.

Cell lysates were obtained by lysing cells in in 1× SDS buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS [w/v], 10% glycerine [v/v], 50 mM DTT, and 0.01% bromphenol blue [w/v]). Proteins from cell-free supernatants were extracted by TCA precipitation, resuspended in Laemmli buffer and 5× SDS buffer. Equal amounts of lysates were loaded on 12 or 15% SDS-PAGE gels. Separated proteins were transferred to polyvinylidene difluoride membranes that were incubated overnight with the specific Ab according to manufacturer’s instructions. Anti-mouse-IL-1β (R&D Systems), anti-mouse caspase-1 (Adipogene), anti-human IL-1β (Cell Signaling Technology), anti-human caspase-1 (Santa Cruz Biotechnology) and anti–β-actin (Sigma-Aldrich) were used as primary Abs. HRP-coupled secondary Abs were from Promega.

The release of lactate dehydrogenase (LDH) was measured in supernatants using the CytoTox 96 kit (Promega), according to manufacturer’s instructions.

Specific-pathogen-free C57BL/6 mice were originally purchased from Harlan Laboratories, Nlrp3-deficient mice (24), and Asc knockout mice (25) were provided by J. Tschopp (University of Lausanne, Lausanne, Switzerland) and V. Dixit (Genentech, South San Francisco, CA). Nlrc4 knockout mice (25) were provided by F. Greten (II. Medizinische Klinik und Poliklinik, Klinikum rechts der Isar der Technische Universität München, Munchen, Germany) and Caspase-1 deficient mice by Richard Flavell through The Jackson Laboratory (26). Wild-type littermates or C57BL/6 cohoused animals were used as controls to exclude an influence of difference in the gut flora of independently bred mice. Mice were bred in a pathogen-free animal facility at the Institute for Medical Microbiology, Immunology and Hygiene (Technische Universität München).

All animal experiments were approved by local authorities (Regierung von Oberbayern [55.2-1-54-2532-155-12]).

For in vivo experiments, 6- to 8-wk-old female C57BL/6 and C57BL/6 ΔNlrp3 were infected with an orogastric dose of 10⁹ CFUs of H. pylori strain PMSS1. Infection was repeated on days 3 and 5.

After mice were sacrificed, the stomach was removed, opened along the lesser curvature, washed with PBS, and dissected longitudinally into equal strips. Of every stomach, the same section was assigned to the same downstream processing to reduce sampling error.

For quantitative assessment of H. pylori colonization, one stomach section was homogenized in Brucella broth and serial dilutions were plated on WC dent agar plates and supplemented with bacitracin (200 μg/ml), nalidixic acid (10 μg/ml), and polymyxin B (3 μg/ml).

For assessment of gastric histopathology, paraffin-embedded stomach sections were H&E stained and scored for H. pylori colonization, gastritis degree of atrophy, and intestinal metaplasia, according to the updated Sydney score system (27).

Gastric IFN-γ and IL-17 expression was determined by RT-PCR. Total gastric RNA was isolated using RNA isolation kit (Quiagen). Total RNA (1–2 μg) was retrotranscribed (reverse transcriptase from Promega) and used for real-time PCR (CFX Bio-Rad). Absolute values of IFN-γ (IFN-γ forward 5′-TCAAGTGGCATAGATGTGGAAGAA-3′/IFN-γ reverse 5′-TGGCTCTGCAGGATTTTCATG-3′) and IL-17 (IL-17 forward 5′-GCT CCA GAA GGC CCT CAG A-3′/IL-17 reverse 5′-AGC TTT CCC TCC GCA TTG A-3′) expression were normalized to GAPDH expression (GAPDH forward 5′-GCA CAG TCA AGG CCG AGA AT-3′/ GAPDH reverse 5′-GCC TTC TCC ATG GTG GTG AA-3′).

To determine IL-1β, IL-18, IL-10, and TNF-α protein levels in stomach extracts, a piece of stomach was weighted and homogenized in PBS with proteinase inhibitor (Roche). After determining the whole protein content of the supernatant by Pierce BCA protein assay (Thermo), 5 μg protein were used for measuring cytokine levels by ELISA (eBioscience or R&D Systems, respectively).

For the isolation of gastric immune cells, a stomach section was digested in 1 mg/ml collagenase (Sigma-Aldrich) ± 200 μg/ml Dnase I (Roche) for 30 min at 37°C with shaking. Single-cell suspensions were filtered and stained for CD4 and CD45 (eBioscience). Flow cytometry was performed on a CyanADP instrument (DakoCytomation) and analyzed using FlowJo software (Tree Star).

Data are presented as mean ± SD unless otherwise stated. Data were analyzed for normal distribution before nonparametric tests (Mann–Whitney U test) or parametric tests (Student t test or ANOVA with Bonferroni correction for multiple comparisons) were used to determine statistical significance using GraphPad Prism software (GraphPad). Statistical significance was established when p ≤ 0.05.

Activation of the inflammasome by H. pylori has been shown in murine BMDCs, whereas human cells have not been investigated. Therefore, we initially infected human PBMCs with the H. pylori strain G27 and measured the release of IL-1β in the supernatants. We observed mature IL-1β secretion after infecting the cells with H. pylori (Fig. 1A, 1B), suggesting that H. pylori is able to provide the priming and, more importantly, the second signal required for inflammasome stimulation. By prestimulating the cells with LPS we aimed at discriminating the effect of H. pylori as second signal, since LPS alone lead to the production of pro–IL-1β but not to the cleavage to mature IL-1β (Supplemental Fig. 1A). Higher IL-1β secretion was observed when cells were prestimulated with LPS as priming signal, clearly showing that H. pylori can provide the second signal required for inflammasome activation.

Increased IL-1β secretion was also detected upon infection of murine BMDCs (Fig. 1C, 1D), BMDMs (Supplemental Fig. 1D) and bone marrow–derived neutrophils (Supplemental Fig. 1E), independently of the H. pylori strain used, indicating that H. pylori activates the inflammasome in different innate immune cells. To further characterize the effect of H. pylori infection on the release of IL-1β, time-course experiments were performed in BMDCs. Even though pro–IL-1β was produced after 3 h LPS prestimulation (Supplemental Fig. 1A, 1B), its processing was not detected until 3 h post-H. pylori infection, whereas H. pylori alone had no effect at this short interval (Supplemental Fig. 1C). At later time points, H. pylori induced the release of IL-1β but always at lower levels when compared with LPS-prestimulated cells, confirming that H. pylori provides the second signal for inflammasome activation.

For being biologically active, IL-1β is cleaved by caspase-1, the effector protease of the inflammasome. Thus, activation of the inflammasome after H. pylori infection was confirmed by the presence of the active caspase-1 subunits p20 or p10 (Fig. 1B, 1D). Furthermore, when PBMCs (Supplemental Fig. 1F) or BMDCs (Supplemental Fig. 1G) were incubated with a caspase-1–specific inhibitor, the secretion of IL-1β was significantly reduced, whereas no changes on the levels of the inflammasome-independent cytokine TNF-α or intracellular pro–IL-1β were detected (Supplemental Fig. 1F–H). These results demonstrate that IL-1β secretion during H. pylori infection is mediated by caspase-1.

Because the activated inflamamsome is also responsible for inducing pyroptosis, we analyzed cell viability by measuring the release of LDH. PBMCs (Fig. 1E) or BMDCs (Fig. 1F) infected with H. pylori did not secrete more LDH compared with uninfected cells, indicating that H. pylori does not induce pyroptosis at the time points investigated.

Inflammasomes are multiprotein complexes consisting of the adaptor protein ASC and a sensor protein, often of the NLR family (NLRP1, NLRP3, and NLRC4) in addition to caspase-1. To confirm that H. pylori induces inflammasome activation and to elucidate which sensor protein is engaged, BMDCs from Caspase-1^−/−, Asc^−/−, Nlrp3^−/−, and Nlrc4^−/− mice were cocultured with H. pylori. In the absence of Caspase-1, the release of IL-1β was significantly dampened, whereas no changes in TNF-α secretion was detected (Fig. 2A). Likewise, the lack of Asc or Nlrp3 was translated into a reduction of IL-1β levels (Fig. 2B), whereas no significant changes in the IL-1β release was detected in Nlrc4-deficient mice (Fig. 2C). As expected, TNF-α production was not affected in the absence of Asc, Nlrp3 (Fig. 2B), or Nlrc4 (Fig. 2C).

In addition, cleavage of caspase-1 and IL-1β was not detected when Nlrp3 and Asc-deficient BMDCs were infected with H. pylori (Fig. 2D). These results indicate that H. pylori primarily activates the NLRP3 inflammasome.

NLRP3 inflammasome has been described to be activated by K⁺ efflux in different infection models. To depict whether NLRP3 activation upon H. pylori infection involves K⁺ efflux, we blocked it by incubating BMDCs with 50 mM KCl 30 min prior infection. H. pylori–induced IL-1β production was inhibited by blocking K⁺ efflux (Fig. 3A), whereas the levels of TNF-α were not significantly altered (Supplemental Fig. 2A). Similar results were obtained when human PBMCs were treated with KCl prior to H. pylori infection (Supplemental Fig. 3A).

Phagocytosis as well as subsequent lysosomal damage with release and activation of cathepsin B are additional mechanisms associated with the activation of the NLRP3 inflammasome. Therefore, we next explored whether endocytic processes were necessary for H. pylori–induced IL-1β production. For this purpose, BMDCs were incubated with cytochalasin D to inhibit actin polymerization and therefore phagocytosis. Following inhibition of phagocytosis, H. pylori–induced IL-1β secretion was decreased (Fig. 3B). To rule out a deleterious effect of the inhibitor used, we assessed cellular toxicity. No adverse effects were detected when using cytochalasin D (Supplemental Fig. 2B). Comparable results were obtained for human PBMCs (Supplemental Fig. 3B).

It is well-known that H. pylori induces ROS; however, it is yet unclear whether ROS activate the inflammasome or rather act as priming signal leading to increased expression of pro–IL-1β and NLRP3 proteins. Nevertheless, we analyzed whether ROS could be involved in the H. pylori–mediated activation of the NLRP3 inflammasome. To minimize the effect on priming, the cells were prestimulated with LPS before inhibitor addition. To prevent ROS induction, the inhibitors ebselen and (2R, 4R)-4-aminopyrrolidine-2, 4-dicarboxylate (APDC) were used. Preincubation of BMDCs with ebselen and APDC resulted in significant lower levels of IL-1β secretion, showing that H. pylori–mediated ROS production is involved in inflammasome activation (Fig. 3C, 3D). The secretion of the inflammasome independent cytokines TNFα and IL-6, as well as the phagocytic bacterial uptake was analyzed to exclude off-target effects. No changes in TNFα or IL-6 levels (Supplemental Fig. 2C) were detected when using ebselen, whereas APDC induced a reduction in the levels of IL-6. This effect could be due to the agonistic activation of glutamate receptors expressed on DCs (28) by APDC, leading to inhibition of cAMP signaling and IL-6 production as reported previously (29). Importantly, no differences in bacterial uptake were observed after using the inhibitors (Supplemental Fig. 2E). When analyzing the effect of these inhibitors on the secretion of IL-1β by human PBMCs, we observed lower levels of IL-1β secretion after ebselen treatment, whereas APDC completely abolished IL-1β release (Supplemental Fig. 3C, 3D). Thus, our results show that H. pylori activates the NLRP3 inflammasome by several distinct mechanisms, involving K⁺ efflux, phagocytosis, and ROS production.

To elucidate whether activation of the inflammasome was actively induced by bacterial secreted factors, the bacterium was subjected to heat inactivation or PFA fixation. A slightly reduced IL-1β secretion was detected when cells were incubated with PFA-fixed H. pylori (Fig. 4A). Significantly lower levels of the cytokine were observed after heat-inactivating the bacterium, suggesting that heat-sensitive H. pylori virulence factors might actively regulate IL-1β secretion.

To identify the H. pylori virulence determinants involved in inflammasome activation, LPS-prestimulated BMDCs were infected with the H. pylori strain PMSS1 deficient for either of the virulence factors, CagA, CagE, VacA, or γ-Glutamyltranspeptisdase (gGT) and the SS1 strain, which expresses VacA (30) CagA and CagE (31) but has a nonfunctional cagPAI. Absence of CagE, a VirB4 ATPase of the type IV secretion system (T4SS) essential for both the assembly of the system and CagA transfer, resulted in lower levels of IL-1β (Fig. 4B). Similar results were detected when cells were stimulated with the SS1 strain. Interestingly, the H. pylori PMSS1 strain lacking CagA was still capable of inducing IL-1β, indicating that other components of the PAI are responsible for inflammasome activation. Also, a PMSS1 mutant lacking the pore-forming toxin VacA induced lower amounts of IL-1β secretion. Conversely, the absence of gGT did not influence the release of IL-1β. These results correlated with caspase-1 activation. Thus, active caspase-1 p20 was not detected after heat inactivation of the bacterium or when infected the cells with bacterial strains lacking a functional T4SS or VacA (Fig. 4C, 4D). Levels of TNF-α secretion were measured to exclude inflammasome independent effects. No changes in TNF-α release were observed (Supplemental Fig. 4A, 4B). In addition, no influence on bacterial uptake was detected (Supplemental Fig. 4C), indicating that all the H. pylori strains used were equally phagocytized (Supplemental Fig. 4C).

When corroborating these results in human PBMCs, we observed reduced levels of IL-1β upon infection with a CagE deficient strain, whereas only slightly reduced IL-1β levels were observed when infecting with VacA or gGT mutants (Fig. 4E). Importantly, CagA also seemed not important for induction of IL-1β in human PBMCs. Interestingly, we observed that adhesion of the bacterium to the cells is important for inflammasome activation, because infection with strains deficient in the adhesins blood group Ag–binding adhesin (BabA) or sialic acid–binding adhesin (SabA) led to lower IL-1β secretion. No changes in the levels of the inflammasome-independent cytokine TNF-α were found (Supplemental Fig. 4D). To identify the major source of IL-1β in human PBMCs, we performed negative selection to isolate monocytes/macrophages, NK cells, B cells, and T cells. After incubation of these cell subsets with H. pylori, we observed that inflammasome activation and IL-1β secretion occurred mainly in monocytes/macrophages (Fig. 4F).

These data indicate that in murine BMDCs and human PBMCs, the cagPAI, but not CagA, is the main H. pylori virulence factor involved in the activation of the inflammasome.

Our in vitro data indicated that IL-1β secretion upon H. pylori infection depended on the activation of the NLRP3 inflammasome. Therefore, we investigated whether the inflammasome plays a role in the immune response toward H. pylori infection in vivo. For this purpose we infected Nlrp3-deficient mice with H. pylori PMSS1. After 1 mo of infection, reduced levels of IL-1β and IL-18 were detected in Nlrp3-deficient mice corroborating our in vitro findings (Fig. 5A). We measured the levels of TNF-α and IL-10, both cytokines important in the context of H. pylori infection but independent on the inflammasome, as control. No differences were detected between wild type and Nlrp3-deficient mice as expected. Notably, the expression of pro–IL-1β was not affected (Supplemental Fig. 4E), indicating inflammasome-specific effects. Interestingly, higher colonization of the gastric mucosa was detected in Nlrp3-deficient mice when compared with wild-type animals (Fig. 5B), which was accompanied by a lower degree of inflammation (Fig. 5C). In addition, lower infiltration of CD45⁺ and CD4⁺ cells was observed in the stomach of Nlrp3 knockout mice compared with wild-type animals (Fig. 5D).

To assess the influence of the inflammasome in the adaptive immune response toward H. pylori, we measured the levels of IL-17 and IFN-γ in the gastric mucosa of infected mice. The expression of both cytokines was reduced in Nlrp3-deficient mice when compared with wild-type (Fig. 5E).

Altogether, these results indicate that the NLRP3 inflammasome influences H. pylori colonization of the gastric mucosa as well as the subsequent immune response and gastric pathology.

The proinflammatory cytokine IL-1β has been extensively reported to play an essential role in H. pylori infection and H. pylori–associated gastric pathology, but little is known about the mechanisms triggering its expression and release from immune cells upon infection. In the current study, we show that IL-1β secretion induced by H. pylori in human and murine innate immune cells depends on activation of the inflammasome, and we identify NLRP3 as the inflammasome mainly activated in response to the bacterium. Our results are in agreement with recent data showing that IL-1β secretion by H. pylori–infected murine DCs depended on the cooperative interaction between TLR2/NOD2 and NLRP3 inflammasome (8). However, we provide direct evidence that H. pylori provides the second signal inducing the assembly and activation of the inflammasome and identify the main virulence factors and cellular mechanisms involved.

Because of the broad spectrum of stimuli, which are able to activate the NLRP3 inflammasome, NLRP3 has been considered to be a sensor of the disruption of host physiology. H. pylori possesses several virulence factors able to profoundly change cellular physiology, and therefore, it is not surprising that we found a number of molecular mechanisms involved in the H. pylori–mediated activation of the NLRP3 inflammasome. Common NLRP3 inflammasome–activating mechanisms include potassium efflux, lysosomal destabilization, and generation of ROS. We observed that blocking of these three stimuli in immune cells substantially reduced IL-1β secretion upon H. pylori infection. The p58 subunit of H. pylori VacA was shown to induce potassium efflux from liposomes at acid pH values (32), whereas induction of ROS by VacA has been demonstrated in some studies (33–35). In addition, VacA was described to lower mitochondrial transmembrane potential (36), another mechanism by which the NLRP3 inflammasome can be activated (37). These observations suggest that VacA might be involved in inflammasome stimulation. Indeed, we observed that VacA-deficient bacteria were less efficient in inducing IL-1β release in murine DCs as well as in human PBMCs, confirming an involvement of VacA in inflammasome activation. This observation is in contrast to data from Kim et al. (8) who could not detect changes in IL-1β secretion after infecting cells with H. pylori strains deficient for VacA. One possible explanation for this discrepancy is our strategy of using LPS as priming signal to dissect the bacterial factor acting as second signal triggering the inflammasome, which was not included in the previous study.

Although other virulence factors such as gGT can induce ROS production, we did not observe significant effects on IL-1β secretion when using gGT-deficient bacteria. We speculate that in the absence of gGT, other virulence factors such as VacA can elicit the cellular mechanisms responsible of inflammasome activation. Our data further show that cagPAI, but not CagA, is important for IL-1β secretion. This observation is in line with the previous results by Kim et al. (8); however, we show that a functional T4SS, encoded by H. pylori cagPAI, is important to provide the second signal triggering assembly and activation of the inflammasome. Many bacterial species have been described to activate primarily the NLRC4 inflammasome via T3SS and T4SS (38). Interestingly, the delivery of effector molecules through the T4SS has been recently shown to be required for NLRP3 inflammasome activation as well (39), indicating that the presence of an active T4SS might represent an alternative mechanism of NLRP3 activation during bacterial infection. Apart from CagA, H. pylori peptidoglycan is also delivered into host cells through the T4SS. The particulate nature of peptidoglycan was found to be essential for activation of NLRP3 inflammasome and alteration of peptidoglycan by Staphylococcus aureus strongly reduces production of IL-1β in response to infection (40). Therefore, we speculate that H. pylori peptidoglycan injected into the host cell might be involved in the activation of the inflammasome. Further experiments are needed to substantiate this hypothesis.

In this context, it was interesting to note the markedly reduced IL-1β secretion when human PBMCs were infected with H. pylori strains lacking either of the adhesins BabA and SabA, suggesting that adhesion mediated T4SS-dependent secretion of a factor independent of CagA induces activation of the inflammasome. Thus, binding of the bacterium to the host cells represents a crucial event for inflammasome activation. This might also involve phagocytosis of H. pylori, which is delayed and reduced in adhesion-deficient strains.

Having observed that H. pylori can activate the inflammasome in different innate immune cells, we further analyzed the involvement of the inflammasome in the immune response to the bacterium. Because our in vitro data indicated that NLRP3 was mainly responsible of regulating IL-1β secretion during H. pylori infection, we infected Nlrp3-deficient mice with H. pylori strain PMSS1, which is proficient for all the virulence factors we previously showed to be important for inflammasome activation. In contrast to Kim et al., we found that mice lacking NLRP3 were colonized at higher levels compared with wild-type animals. We observed that higher bacterial colonization was accompanied with lower inflammation and T cell infiltration. Moreover, we also found reduced levels of IL-17 and IFN-γ in the gastric mucosa of Nlrp3-deficient mice upon infection with H. pylori. Our results suggest that NLRP3 activation triggered by H. pylori in innate immune cells represents an important mechanism contributing to the inflammatory response toward the bacterium and are in line with previous observations by Hitzler et al. (18) showing that IL-1 receptor deficient mice present less H. felis–associated pathology and show reduced Th1 and Th17 responses.

We propose that NLRP3 mediated inflammasome activation might be an early event in H. pylori infection, responsible for eliciting the strong adaptive immune response to control the infection. Because of the different mechanisms developed by the bacterium to evade the host’s immune system, the infection persists and the levels of IL-1β remain high. IL-1β has been shown to directly inhibit acid secretion from rat gastric parietal cells (41), which may contribute to hypoacidity and finally, together with TNF-α, to parietal cell loss (42). Furthermore, polymorphisms inducing elevated levels of IL-1β have been associated with increased risk of gastric cancer. It has to be noted that the risk was restricted to the non-cardia subsite, suggesting that the distribution of the inflammation is important in terms of gastric cancer risk (43). Taken together, these events pave the way to the development of gastric atrophy, which is considered the initial step of gastric carcinogenesis according to the Correa hypothesis, which can under conditions of continuous inflammation eventually lead to the development of gastric cancer.

We thank Sarah Franke for technical assistance and Stefanie Wüstner, Daniela Engler, and Mathias Oertli for valuable technical advice.

The authors have no financial conflicts of interest.