To further confirm the epithelial cell origin of IL-18 and also assess its role in
H. pylori infection, we employed a bone marrow (BM) reconstitution mouse model. Consistent with the idea that epithelial cells are a major source of gastric IL-18, γ-irradiated
Il18−/− recipient mice that were reconstituted with BM cells either from
Il18+/+ or
Il18−/− donor mice and infected for 5 weeks with
H. pylori SS1 produced significantly lower levels of IL-18 than
Il18+/+ recipient mice, (Fig.
2a;
p = 0.006 and 0.004, respectively). Similar findings were observed at 18 weeks p.i. (data not shown). Comparable levels of IL-1β and bacterial loads were observed in both
Il18+/+ and
Il18−/− recipient mice, suggesting that epithelial cell-derived IL-18 does not affect IL-1β production nor bacterial colonisation (Fig.
2b,
2c). Other studies also reported that IL-18 was not required for protection against
H. pylori infection in mouse challenge
20 and prophylactic immunisation models
7, 20. Importantly, we observed that mice specifically lacking IL-18 in the non-haematopoietic compartment, and reconstituted with IL-18-producing BM, had increased stomach weights (Fig.
2d;
p = 0.002), mucosal thickness (gastric hyperplasia; Fig.
2d, e) and acid mucin production (Fig.
2f) when compared with wild-type (WT) animals. The
Il18−/− mice reconstituted with
Il18+/+ BM also displayed increased stomach weights (Fig.
2d;
p = 0.004) and mucosal thickness (Fig.
2e; p = 0.0001) when compared with
Il18−/− BM/
Il18−/− recipient mice. This finding suggests that IL-18-secreting haematopoietic cells may be pathogenic in mice lacking IL-18 production in the non-haematopoietic compartment.
Similar to the Il18−/− recipient mice reconstituted with Il18−/− BM (Fig. 2d), Il18−/− total knockout (KO) mice exhibited significantly increased stomach weights at 2 months p.i. with H. pylori, as compared with Il18+/+ animals (Supplementary Fig. 1a, 1b). Despite this difference, however, similar expression levels of the inflammatory cytokine genes, Il1b and Cxcl2, were observed in the Il18+/+ and Il18−/− mice (Supplementary Fig. 1c). Consistent with these data, Il18−/− mice that were infected for 13 months with Helicobacter felis, which is more pathogenic than H. pylori in mice 21, also displayed larger stomachs than WT animals (Supplementary Fig. 1d, 1e). Taken together, our data show that non-haematopoietic cells in the stomach produce IL-18 which plays a protective role against the development of the pre-neoplastic lesions typically observed in 10 during chronic Helicobacter infection.
IL-18 responses to H. pylori infection are independent of the canonical inflammasome
We next sought to determine the inflammasome signalling molecules required for epithelial cell production of IL-18 in response to H. pylori infection. For this, primary gastric epithelial cells were isolated from mice deficient in the canonical inflammasome molecules, Nlrp3, Asc (encoded by the Pycard gene) and caspase-1. Similar to primary human gastric epithelial cells 6, some IL-18 was produced constitutively by primary mouse gastric epithelial cells (Fig. 3a). This may be attributed to the presence of NOD1 ligand (peptidoglycan) in the serum used for tissue culture which is able to promote low level NOD1 activation 22. Primary epithelial cells from Casp1−/− mice, but not those from Nlrp3−/− nor Pycard−/− animals, produced significantly less IL-18 in response to stimulation with H. pylori SS1 bacteria, when compared with WT cells (Fig. 3a; p = 0.0004). H. pylori bacteria did not induce any IL-18 production in BM-derived macrophages (BMDMs) (Supplementary Fig. 2). As reported previously 23, the bacterium induced IL-1β responses in BMDMs (Supplementary Fig. 2). These findings underscore the differences in epithelial and myeloid cell responses to H. pylori infection.
Next, we performed H. pylori infection studies in WT, Nlrp3−/−, Pycard−/− and Casp1−/− mice. Although all animal groups showed similar levels of bacterial loads at 6 months post-infection (Fig. 3b), significantly increased inflammatory scores and stomach weights were only observed in Casp1−/− mice when compared with WT animals (Fig. 3c, 3d; p = 0.027 and p = 0.018, respectively). Moreover, we infected mice deficient in other known canonical inflammasome molecules, Nlrp1 and Nlrc4, the latter responding to bacterial flagellin and rod proteins 2, but observed no significant differences in colonisation levels nor stomach pathology when compared with WT animals (Supplementary Fig. 3). Taken together, the findings suggest that the regulation and functions of IL-18 in gastric epithelial cells in response to H. pylori infection is dependent on caspase-1, but independent of canonical inflammasome pathways, including the key inflammasome molecules NLRP3 and ASC, which have been implicated in responses to bacterial pathogens.
NOD1 in epithelial cells mediates IL-18 processing in response to H. pylori infection
To further characterise the inflammasome pathway involved in H. pylori-induced IL-18 processing, we used human AGS gastric epithelial cells, a standard cell line used in the H. pylori field and in which IL18 mRNA 6 and protein 24 are upregulated in response to H. pylori stimulation. Consistent with the in vivo data (Fig. 3), we were unable to detect the presence in AGS cells of either NLRP3 or ASC, nor other important inflammasome proteins, NLRC4 or AIM2 (Supplementary Fig. 4). These findings suggested that a non-classical inflammasome protein may be involved in H. pylori-induced IL-18 processing in human gastric epithelial cells.
We hypothesised that this protein may be NOD1 which plays important roles in the sensing of H. pylori bacteria 13, 14, 17 and is strongly expressed in AGS cells (Supplementary Fig. 4d). Indeed, interfering with NOD1 signalling in AGS cells, by either short hairpin RNA (shRNA) knockdown (shNOD1; 14, 25) or CRISPR/Cas9 knockout (NOD1 KO; 17), resulted in the loss of IL-18 processing in response to live H. pylori bacteria (Fig. 4a, b). A similar NOD1-dependent effect was observed for IL-18 processing in response to H. pylori MVs, which contain peptidoglycan 14, 15. The BHI broth, from which MVs were isolated, also induced IL-18 processing, most likely due to serum in the medium 22. As reported 26, we also did not detect any IL-1β production in the AGS cell line (data not shown).
We next studied the contribution of the H. pylori T4SS on IL-18 processing using strains lacking a functional T4SS i.e. ΔcagPAI, ΔcagM isogenic mutants and SS1 13. The levels of mature IL-18 production in cell culture supernatants of control (shEGFP) AGS cells were reduced in response to strains lacking a functional T4SS when compared with the respective WT (251) and parental human clinical (10700) isolates (Fig. 4b). The reduced levels of pro-IL-18 production induced in shNOD1 AGS cells by the T4SS-deficient strains, when compared with the WT isolates (Fig. 4a), may be attributed to the ability of the latter to activate T4SS-dependent but NOD1-independent pathways that converge on NF-κB 12, resulting in the upregulation of pro-IL-18 production. These data show that H. pylori T4SS activation of the NOD1 signalling pathway is required for maximal IL-18 production and processing in human gastric epithelial cells. This result is consistent with previous findings for NOD1-dependent IL-8 production 13, 14 and interleukin-33 (IL-33) processing 15 in human gastric epithelial cells. Importantly, no differences in cell death were observed between the NOD1 KD and WT AGS cells in response to H. pylori stimulation (Fig. 4d), suggesting that the presence of mature IL-18 in culture supernatants was not due to cell lysis.
To further investigate the role of NOD1 in IL-18 processing to H. pylori infection, we next used immortalised and primary mouse gastric epithelial cells. Total levels of IL-18 production were increased in the culture supernatants of the mouse GSM06 cell line 27 in response to H. pylori SS1 stimulation, whereas it was significantly reduced when NOD1 self-oligomerisation was inhibited by pre-treatment with the NOD1-specific inhibitor, ML130 28 (Fig. 4a; p = 0.0002). ML130 had an effect on the formation of both precursor and mature forms of IL-18 in response to H. pylori stimulation (Fig. 4b). Similarly, primary gastric epithelial cells from Nod1−/− mice stimulated with H. pylori bacteria produced significantly lower levels of total IL-18 (Fig. 4c; p = 0.0005), as well as pro- and mature IL-18 (Fig. 4c, 4d), when compared with Nod1+/+ cells. Although a functional T4SS was required for H. pylori bacteria to induce optimal IL-18 processing in human gastric epithelial cells (Fig. 4a, b), this did not appear to be the case in epithelial cells of murine origin. We also found that Nod1 was not required for IL-18 (nor IL-1β) production by BMDMs in response to either H. pylori or the canonical inflammasome activators, LPS and nigericin (Supplementary Fig. 5), again indicating that NOD1 mediates IL-18 processing in a cell-type-specific manner.
NOD1 in epithelial cells mediates IL-18 processing in response to P. aeruginosa infection
To investigate whether NOD1 may mediate IL-18 processing in response to other mucosal pathogens, we stimulated human A549 pulmonary epithelial cells with P. aeruginosa 18. Total IL-18 levels were significantly decreased in P. aeruginosa-stimulated cells that had been transfected with either of two validated NOD1 siRNAs, siRNA-1 and − 2 (Fig. 5a; p = 0.0004 versus control), when compared with cells transfected with a scramble siRNA (Fig. 5b; p = 0.002). IL-18 processing was also affected in cells that were transfected with NOD1 siRNA-2, but not in those transfected with siRNA-1 (Fig. 5c). This may be explained by the fact that NOD1 siRNA-2 targets exon 6, encoding the NOD, which is required for NOD1 oligomerisation, whereas NOD1 siRNA-1 targets exon 7, encoding a region of the LRR that is dispensable for NOD1 detection of its ligand 29. Taken together, these data suggest that NOD1 may play a broad role in regulating IL-18 processing in epithelial cells responding to various bacterial infections.
NOD1 mediates IL-18 processing independently of RIPK2–NF-kB signalling
Classically, NOD1 activation results in the recruitment of the adaptor molecule RIPK2, induction of the NF-κB signalling pathway and upregulation of pro-inflammatory cytokine production 13, 14, 16, 17. We investigated the role of RIPK2 in H. pylori-induced IL-18 processing in epithelial cells by small interfering RNA (siRNA) transfection or blocking its activity with the kinase inhibitor, WEHI-345 30. siRNA-mediated KD of RIPK2 was confirmed by qPCR detection of RIPK2 expression and by measuring IL-8 production as a read-out for classical NOD1 signalling 17. Blocking RIPK2 signalling had no effect on IL-18 processing in response to H. pylori stimulation in AGS cells (Fig. 6a, 6b). In agreement with this finding, IL-18 processing was unaffected in both primary gastric epithelial cells and BMDMs from Ripk2−/− mice, whereas epithelial cell production of the NOD1-regulated chemokines, Cxcl1/keratinocyte chemoattractant (KC) and Cxcl2/macrophage inflammatory protein 2 (MIP2) 13, 15, 31, 32, was reduced in response to H. pylori stimulation (Supplementary Fig. 6; p = 0.0005 and p = 0.0002, respectively).
To demonstrate that NF-κB activation by NOD1 is dispensable for IL-18 processing in epithelial cells, we restored NOD1 expression in shNOD1 AGS cells with plasmid constructs expressing either WT or a mutant form of NOD1 (K208R NOD1) that is unable to activate NF-κB signalling 33. As expected, NOD1 shNOD1 AGS cells transfected with plasmid encoding WT NOD1 exhibited significantly higher IL-8 responses to H. pylori stimulation than cells transfected with either the K208R NOD1 mutant NOD1 or empty control plasmids (Fig. 6c; p = 0.0004 and p = 0.0002, respectively). Conversely, WT and K208R mutant NOD1 plasmids were equally able to rescue IL-18 processing in shNOD1 AGS cells (Fig. 6d), suggesting that NOD1 mediates gastric epithelial IL-18 processing in an NF-κB independent manner. We propose that H. pylori induces IL-18 processing via a non-classical type of NOD1 signalling pathway.
It was recently reported that NOD1 plays a non-redundant role with the host adapter protein, tumour necrosis factor (TNF) receptor-associated factor-interacting protein with a forkhead-associated domain (TIFA), in regulating inflammatory responses by gastric epithelial cells to H. pylori 34. Consistent with this finding, we showed that transfection with TIFA siRNA significantly reduced IL-8 responses to H. pylori in AGS cells, when compared with scramble siRNA (Supplementary Fig. 7a, b; p < 0.0001). Moreover, TIFA siRNA transfection abrogated IL-18 processing in AGS cells stimulated with H. pylori bacteria (Supplementary Fig. 7c), thus suggesting that cross-talk between NOD1 and TIFA pathways may be important for H. pylori-induced IL-18 processing in epithelial cells.
It is well established that caspase-1 is a key protease responsible for the cleavage of pro-IL-18 to its mature form in innate immune cells 1, 19. Although early studies reported potential CARD–CARD interactions between NOD1 and pro-caspase-1 33, these findings have never been re-visited. Therefore, to examine whether NOD1 mediates IL-18 processing in epithelial cells via caspase-1 activation, shNOD1 AGS cells that were transiently expressing yellow fluorescent protein (YFP)-labelled NOD1 were incubated with anti-caspase-1 antibody. As can be observed (Fig. 7a), NOD1 is normally proximal to the cell membrane under basal conditions, but moved towards the cytosol and formed yellow punctate stained aggregates with endogenous caspase-1, in response to H. pylori stimulation (Fig. 7a). This suggested that NOD1 may associate with caspase-1 upon activation by H. pylori. To demonstrate NOD1-mediated activation of caspase-1, we used the fluorescently labelled inhibitor of caspase (FLICA) reagent that binds irreversibly to active caspase-1. WT (shEGFP) AGS cells exhibited enhanced staining with FLICA after stimulation with H. pylori, whereas staining was barely detected in shNOD1 AGS cells (Fig. 7b). Furthermore, we could only detect the mature, active form of caspase-1 (p20) in the culture supernatants of WT but not shNOD1 AGS cells in response to H. pylori stimulation (Fig. 7c). We also observed reduced levels of active caspase-1 in AGS cells stimulated with H. pylori mutant strains lacking either a functional T4SS (ΔcagPAI or ΔcagM) or defective in activating the NOD1 pathway (slt−; 13) (Fig. 7c). The data suggest that H. pylori activation of the NOD1 pathway contributes to caspase-1 activation in gastric epithelial cells, but that other pathway(s) may also be involved. Collectively, these data show that NOD1 sensing of H. pylori promotes IL-18 processing via activation of caspase-1.
We next sought to elucidate potential NOD1-caspase-1 molecular interactions and their involvement in IL-18 processing. It is well established that NLRP3 and AIM2 need the adaptor protein, ASC, to recruit and activate caspase-1. However, as reported above (Supplementary Fig. 4), the AGS cell line does not produce detectable levels of ASC protein, suggesting that NOD1 may directly interact with caspase-1 via homotypic CARD–CARD interactions. This hypothesis was addressed using the fluorescence lifetime imaging microscopy-Förster resonance energy transfer (FLIM-FRET) technique in live cells. For this, we transfected shNOD1 AGS cells with NOD1-mOrange plasmid (donor alone control) or both NOD1-mOrange and the acceptor plasmid, encoding caspase-1-GFP (Fig. 8a). We found that H. pylori stimulation led to a significant reduction in the lifetime of the donor fluorochrome, indicative of quenching of the donor by the acceptor caspase-1-GFP due to close proximity (< 10 nm) of NOD1 and caspase-1 proteins (Fig. 8a; p < 0.0001). In contrast, the lifetime of the donor remained unchanged in unstimulated cells. Reconstitution of NOD1 expression with full-length NOD1 in NOD1 KO AGS cells resulted in maximal IL-18 processing in response to H. pylori (Fig. 8b). In contrast, cells transfected with CARD-deficient NOD1 exhibited background levels of IL-18 processing (Fig. 8c), which was most likely due to the effect of H. pylori on pro-IL-18 synthesis. Collectively, these results suggest that NOD1 sensing of H. pylori in epithelial cells activates caspase-1 via direct CARD–CARD interactions, resulting in the processing of IL-18.
As IL-18 appeared to be important for protecting tissues against excessive pathology during chronic H. pylori infection in vivo (Fig. 2, Supplementary Fig. 1), we next asked whether NOD1-mediated IL-18 secretion may play a role in regulating gastric epithelial cell survival responses. For this, we performed in vitro assays including MTT assay and Annexin V/propidium iodide (PI) staining to assess cell proliferation and apoptosis, respectively. shNOD1 KD AGS cells exhibited significantly higher levels of both cell growth and apoptosis in response to H. pylori stimulation, when compared with shEGFP AGS cells (Fig. 9a, 9b; p = 0.005 and p = 0.01, respectively). Differences in apoptosis were not observed in cells stimulated with the apoptotic-inducing agent, etoposide (Fig. 8b). These data suggest that NOD1 is involved in regulating both epithelial cell growth and death in response to H. pylori stimulation. To confirm these findings, we established a 3-D gastric organoid model from Nod1+/+ and Nod1−/− mice. H. pylori bacteria were added to the lumen of the organoids by microinjection (Fig. 9c). Consistent with data for AGS cells (Fig. 2a, b), Nod1−/− organoids exhibit higher levels of cell proliferation and apoptosis, as compared with Nod1+/+ organoids (Fig. 9d, 9e; p = 0.02 and p = 0.003, respectively). Collectively, these results show that NOD1 is important for maintaining homeostasis in gastric epithelial cell turnover in response to H. pylori infection.