Inflammasome Adaptor ASC Suppresses Apoptosis of Gastric Cancer Cells by an IL18-Mediated Inflammation-Independent Mechanism

Gastric cancer is the third most lethal cancer worldwide, and is among a growing number of cancers associated with precursory chronic inflammatory responses (1–3). The predominant histologic subtype of gastric cancer is intestinal-type, and despite the causal correlation between chronic gastric inflammation triggered by pathogenic microbes (i.e., Helicobacter pylori) and intestinal-type gastric cancer (2, 4), the identity of innate immune regulators within the host gastric mucosa that promote gastric cancer remains unclear. Consistent with an altered host immune response predisposing to gastric cancer, gene polymorphisms for the proinflammatory cytokine IL1β, which are associated with augmented gene expression, increase the risk of human gastric cancer (5, 6), and transgenic overexpression of IL1β in mice triggers gastric inflammation and tumors (7). In addition, clinical studies on the related IL1 cytokine family member IL18, which can display opposing anti- or protumorigenic effects dependent upon the tissue and cellular context in various cancers (8), demonstrate that IL18 levels are increased in gastric cancer patients and serve as a poor prognostic marker (9–11). Although experimental data from human gastric cancer cell lines also suggest that IL18 may contribute to the malignant progression of tumors (9, 12, 13), a definitive role for IL18 in gastric cancer remains unproven.

Inflammasomes have recently been identified as multiprotein complexes that are essential for the production of mature and bioactive IL1β and IL18 proteins. Accordingly, they have attracted much attention as key factors of the immune system with the potential to influence susceptibility to many autoimmune and inflammatory diseases, as well as cancers, where IL1β and IL18 are implicated (14–17). These cytokines are initially produced as inactive proIL1β and proIL18 forms, with proIL1β expression upregulated following ligand-mediated activation of pattern recognition receptors (PRR) such as Toll-like receptor (TLR) or nucleotide-binding oligomerization domain-containing (NOD) family members, whereas proIL18 is constitutively expressed. Subsequently, their inflammasome-mediated secretion as bioactive cytokines is controlled by members of the nucleotide-binding domain and leucine-rich repeat containing protein (NLR) family, namely NLR pyrin domain containing 1 (NLRP1), NLRP3, NLRP6, NLRP12, NLR CARD domain containing 4 (NLRC4) and NLR apoptosis inhibitory protein (NAIP), as well as the cytosolic DNA sensor absent in melanoma 2 (AIM2; refs. 15, 18). Specifically, each of these NLRs and AIM2 form the core of distinct inflammasome complexes, whereby upon sensing host- and/or microbial-derived ligands, they associate with the key adaptor protein apoptosis-related speck-like protein containing a CARD (ASC) to facilitate activation of caspase-1, which catalyzes the maturation of proIL1β and proIL18 precursors into bioactive secreted cytokines (15).

Although ASC is critical for inflammasome-mediated pathologies involving IL1β and/or IL18, investigations into the definitive role of ASC in tumorigenesis are still in their infancy and have been largely restricted to intestinal and skin carcinogenesis, with contrasting findings (19). Here, we reveal that elevated PYCARD (hereafter referred to as ASC) mRNA and mature IL18, but not IL1β, protein levels are a coincident feature of gastric tumors from both gastric cancer patients and the gp130^F/F intestinal-type gastric cancer mouse model (20). Furthermore, genetic ablation of either Asc or Il18 in gp130^F/F mice suppressed gastric tumor growth, independent of inflammation, which was associated with augmented neoplastic epithelial cell death. Interestingly, at the molecular level suppressed gastric tumorigenesis in both gp130^F/F:Asc^−/− and gp130^F/F:Il18^−/− mice was characterized by reduced activation levels of NF-κB. In further support of these findings, we also demonstrate in vitro that CRISPR/Cas9-mediated genetic ablation of ASC in human gastric cancer cells suppressed their colony-forming potential (i.e., growth), which was associated with reduced secretion of mature IL18, increased cellular apoptosis, and reduced activation levels of NF-κB. Collectively, these findings support the existence of a novel protumorigenic ASC inflammasome/IL18 axis in gastric cancer.

The gp130^F/F mice (20), along with mice homozygous null for Asc (Pycard) (21), Il1r (22), or Il18 (23) were used to generate gp130^F/F:Asc^−/−, gp130^F/F:Il1r^−/−, and gp130^F/F:Il18^−/− mice, respectively, on a mixed 129Sv × C57BL/6 background. Experiments comparing different mouse strains included genetically- and age-matched littermates, including where appropriate, wild-type (gp130^+/+) control mice. Mice were housed under specific pathogen-free conditions on a 12-hour light/dark cycle, and all animal studies were approved by the Monash University Monash Medical Centre “A” Animal Ethics Committee and the Walter and Eliza Hall Institute Animal Ethics Committee.

Gastric biopsies were collected at the Xin Hua Hospital (Shanghai, China) from patients, upon formal written informed consent, undergoing upper gastrointestinal endoscopy or surgical resection. Clinicopathologic features and demographics of gastric cancer patient cohorts are described in Supplementary Table S1. Biopsies were snap-frozen in liquid nitrogen. Studies were approved by the Xin Hua Hospital Human Research Ethics Committee and undertaken in accordance with the appropriate ethics guidelines. Patient studies were conducted in accordance with the World Medical Association Declaration of Helsinki statement on the ethical principles for medical research involving human subjects.

Tumor epithelial and stroma samples from gp130^F/F mice were collected from OCT-embedded frozen sections stained with toluidine blue using laser microdissection (Leica). Total RNA was extracted from microdissected samples using the miRNeasy microkit (Qiagen), and then reverse transcribed with the PrimeScript RT Reagent Kit (Takara). Quantitative RT-PCR (qPCR) was performed as described later.

Total RNA was isolated from snap-frozen mouse and human stomach tissues using TRI Reagent Solution (Sigma) followed by on-column RNeasy Mini Kit RNA clean-up and DNase treatment (Qiagen). qPCR was performed on cDNA with Taqman Gene Expression Assays (mouse Il1b, Il1r, Il18, Il18r1, Pycard; ThermoFisher Scientific) or SYBR Green chemistry (Life Technologies) using the Applied Biosystems 7300, 7900HT Fast, and Viia7 Real-Time PCR Systems (ThermoFisher Scientific). Data acquisition and analyses were undertaken using the Sequence Detection System Version 2.4 software (Applied Biosystems). Forward and reverse primer sequences for mouse 18S rRNA, Il1b, Tnfa, Cxcl1, Cxcl2, Ccnd1, Ccnd2, and c-myc have been previously published (24). Sequences for other mouse and human primers will be given upon request.

Gene expression data and clinical information from The Cancer Genome Atlas (TCGA) gastric cancer patients were obtained from the open access TCGA data portal (https://portal.gdc.cancer.gov/projects/TCGA-STAD). The alignment of sample identifiers yielded 18 primary gastric cancer cases for which there was available tumor and matched nontumor data. We used reads per kilobase of exon model per million mapped reads (RPKM) to quantify ASC expression levels from RNA sequencing (RNA-Seq) data generated from each gastric cancer patient within TCGA.

Total protein lysates from snap-frozen tissues were prepared at room temperature with two incubations of 20 minutes at 37°C before and after homogenization (25). ELISA for human total IL1β (R&D Systems), and mouse and human total IL18 (MBL International Corporation) were performed according to the manufacturer's instructions. Mature and free human IL18 was detected with a recently-developed sandwich ELISA (26), and mature mouse IL18 was detected by ELISA using equivalent methodology for the mature human IL18 ELISA. All ELISAs were performed using 50 μg of protein lysate per well of 96-well plates. Immunoblotting was performed with antibodies against mouse (AdipoGen) and human (Cell Signaling Technologies) caspase-1, IL18 (BioVision), phospho(p)NF-κB p65 (Ser536; Cell Signaling Technologies) and α-tubulin (Abcam), and protein bands were visualized using either the Odyssey Infrared Imaging System (LI-COR) for IL18 (p24/p18) and α-Tubulin, or enhanced chemiluminescence for caspase-1 (p45/p20). The bands were quantified using ImageJ software (NIH).

Formalin-fixed and paraffin-embedded tissue sections were stained with hematoxylin and eosin (H&E) for histologic evaluation. The terminal deoxynucleotidyl transferase (tdT)–mediated dUDP nick-end labeling (TUNEL) assay (Millipore), as well as immunohistochemistry to detect proliferating cell nuclear antigen (PCNA) and cleaved caspase-8 (Cell Signaling Technology), pNF-κB p65 (Ser536; Santa Cruz Biotechnology), and CD45, B220, CD3, and CD68 (BD BioSciences), were performed as before (24, 27). Positive-stained cells were counted manually [n = 20 high-power (×40) fields] with a random offset, or percentage area of positive staining was acquired using ImageJ software. Dual immunofluorescence was performed on paraffin-embedded stomach tissues using fluorescent-conjugated primary antibodies, and Alexa Fluor secondary antibody. Antigens were detected with antibodies against E-cadherin and active caspase-3 (Cell Signaling Technology), pNF-κB p65 (Ser536; Santa Cruz Biotechnology), and CD45 (BD Biosciences), as previously described (27). Slides were examined by confocal microscopy (Nikon) and analyzed for red and green fluorescence. Where appropriate, stereologic techniques were applied to enumerate dual-stained cells (27).

Single-cell suspensions of dissected mouse stomach tumor-bearing tissue were prepared by collagenase digestion as described previously (24). For cell sorting, single cells were collected and stained with fluorescence-conjugated antibodies against CD45 (Biolegend) and EpCAM (eBioscience), as well as propidium iodide, and then sorted using a FACSAria instrument (BD Biosciences). For analyses, single-cell populations were stained with fluorescence-conjugated antibodies against CD3, CD69, CD19, CD4, CD11c, B220, CD86, CD45 CD11b, CD8, CD11b (eBioscience), and Gr-1 and F4/80 (BD BioSciences). Apoptosis of human AGS cells was assessed with an Annexin V:FITC Apoptosis Detection Kit (BD Pharmingen). Stained cells were acquired on a FACSCanto instrument (BD Biosciences) and analyzed using FlowJo software (Tree Star) as described previously (24).

Human gastric cancer cell lines AGS (ATCC) and MKN1 (Japanese Collection of Research Bioresources Cell Bank) were maintained in RPMI supplemented with 10% heat-inactivated FCS, 1% penicillin—streptomycin, and 1% l-glutamine (GIBCO). Cell lines were authenticated via short tandem repeat profiling (PowerPlex HS16 System Kit; Promega) in our laboratory after receipt in 2013. For experiments, cell lines were passaged for under 3 months at a time between freeze–thaw cycles. Cell lines were routinely tested throughout the time of experiments for mycoplasma contamination (MycoAlert PLUS Mycoplasma Detection Kit; Lonza). For clonogenic assays, AGS and MKN1 cells were seeded in six-well plates (1 × 10⁴ cells/well) with RPMI/10% FCS media with or without either anti-hIL18-IgA antibody (1 μg/mL; InvivoGen) or anakinra (1 μg/mL; Swedish Orphan Biovitrum AB), the latter a specific IL1 receptor (IL1R) antagonist that blocks the activity of IL1β and IL1α. After 10 days of culture, colonies (containing ≥ 50 cells) were stained and fixed with a solution of 0.005% crystal violet (Sigma-Aldrich) in 10% methanol, and colonies counted.

Using published protocols (28), self-complementary oligonucleotides (Sigma-Aldrich) for human ASC (hASC-sgRNA1-O1-caccgCGAGGGTCACAAACGTTGAG; hASC-sgRNA1-O2-aaacCTCAACGTTTGTGACCCTCGc; hASC-sgRNA2-O1- caccgCATGTCGCGCAGCACGTTAG; hASC-sgRNA2-O2-aaacCTAACGTGCTGCGCGACATGc) and human caspase-1 (hCASP1-sgRNA1-O1-caccgAAAGCTGTTTATCCGTTCCA; hCASP1-sgRNA1-O2- aaacTGGAACGGATAAACAGCTTTc; hCASP1-sgRNA2-O1- caccgGCTCCCTAGAAGAAGCTCAA; hCASP1-sgRNA2-O2- aaacTTGAGCTTCTTCTAGGGAGCc) were ligated into the LentiCRISPRv2 construct (Addgene). The single-guided (sg) RNA sequences are designed and constructed for human ASC to allow for sgRNA targeting of constitutive exonic coding regions (Exon 1 and Exon 3). Lentivirus was produced by transfecting vectors into Lenti-X/H293T cells with LentiCRISPR:psPAX2:pMD2.G at a ratio of 4:3:1. Virus was harvested 48 hours after transfection, filtered and used to infect AGS cell cultures containing 5 μg/mL polybrene. Infected cells were selected with puromycin, and cells infected with nontarget control sgRNA vector were used as negative controls.

The gp130^F/F and gp130^F/F:Asc^−/− mice were lethally-irradiated (single 9.5 Gy dose) and reconstituted with 5 × 10⁶ unfractionated donor bone marrow cells from the indicated genotypes as described previously (24).

Statistical analyses were performed using GraphPad PrismV6.0 software. Data normality were assessed using the D'Agostino and Pearson omnibus K2 normality test, and the appropriate tests to identify statistical significance (P < 0.05) between the means of two or multiple groups are presented in the relevant figure legends. Data are expressed as the mean ± SEM.

In human gastric cancer, gene expression for the key inflammasome adaptor, ASC, was significantly elevated in tumors from two independent gastric cancer patient cohorts (Fig. 1A and B; Supplementary Table S1), thus implicating ASC in disease pathogenesis. To interrogate the role of ASC in gastric cancer, we generated gp130^F/F mice deficient for ASC (gp130^F/F:Asc^−/−). At 10 to 12 weeks of age, which is 4 to 6 weeks after the onset of antral gastric hyperplasia and tumor formation in gp130^F/F mice, the stomach size, tumor mass, and incidence were comparable between gp130^F/F and gp130^F/F:Asc^−/− mice (Supplementary Fig. S1A–S1E). However, compared to age- and sex-matched gp130^F/F littermates, 20- to 24-week-old and 30- to 34-week-old gp130^F/F:Asc^−/− mouse stomachs displayed a markedly reduced hyperplastic response and were visibly smaller and significantly reduced in mass by approximately 30% and 40%, respectively (Fig. 1C and D; Supplementary Fig. S1F and S1G), with no observable gender bias. Furthermore, although histologic assessment of gp130^F/F and gp130^F/F:Asc^−/− mouse stomachs revealed similar gastric adenomatous hyperplastic lesions with no evidence of low-grade dysplasia nor carcinoma in situ, the gastric antral tumor mass in gp130^F/F:Asc^−/− mice was significantly reduced by approximately 40% at 20–24 weeks and approximately 60% at 30–34 weeks compared with age-matched gp130^F/F mice (Fig. 1D; Supplementary Fig. S1H). In addition, the total incidence of gastric lesions in 20- to 24-week-old and 30- to 34-week-old gp130^F/F: Asc^−/− mice was also significantly reduced compared to gp130^F/F mice at the corresponding ages, which corresponded with smaller hyperplastic lesions in gp130^F/F:Asc^−/− mice (Fig. 1D and E; Supplementary Fig. S1I and S1J).

Protein levels of activated caspase-1 p20 subunit, the downstream effector of ASC inflammasomes, were significantly increased 1.7- to 2-fold in stomachs of 20- to 24-week-old gp130^F/F compared with gp130^+/+ wild-type mice (Supplementary Fig. S2A), and were significantly reduced by 7.1-fold (compared to 2.6-fold for procaspase-1) in gp130^F/F:Asc^−/− versus gp130^F/F tumors (Supplementary Fig. S2B). These data therefore confirm that the elevated inflammasome activity during gastric tumorigenesis in gp130^F/F mice is reduced in the absence of ASC.

Because ASC inflammasome activation can instigate potent inflammatory responses, we investigated whether suppressed tumorigenesis in gp130^F/F:Asc^−/− mice was associated with reduced gastric inflammation. Histologic assessment of H&E-stained gastric tissue sections from 20- to 24-week-old mice revealed that the presence of chronic inflammatory infiltrates comprising primarily plasma and lymphoid cells was comparable in the submucosa and mucosa regions of tumors from both genotypes (Fig. 1F). Immunohistochemical analyses also indicated comparable numbers of CD45⁺ leukocyte infiltrates, and CD68⁺ macrophage, B220⁺ B cell and CD3⁺ T cell immune subsets, in the gastric mucosa of gp130^F/F and gp130^F/F:Asc^−/− tumors (Fig. 2A and B; Supplementary Fig. S3A–S3D). Consistent with these observations, flow cytometry on tumor-bearing stomachs of 20- to 24-week-old gp130^F/F:Asc^−/− and gp130^F/F mice confirmed similar frequencies of gastric immune cell subsets, namely CD11b⁺Gr-1⁺ myeloid-derived suppressor cells (MDSC), CD11c⁺CD11b⁻ dendritic cells, CD11b⁺F4/80⁺ monocytes/macrophages, and B- and T-cell populations (Fig. 2C). Also, the activation status of B (B220⁺CD86⁺) and T (CD4⁺CD69⁺ and CD8⁺CD69⁺) cells, which are the predominant infiltrating immune cells in gp130^F/F and gp130^F/F:Asc^−/− gastric tumors, was comparable (Fig. 2C). The similar frequency and activation status of distinct immune cell subsets was also confirmed in perigastric lymph nodes of gp130^F/F and gp130^F/F:Asc^−/− mice (Supplementary Fig. S3E).

The unaltered inflammation was also coincident with similar mRNA levels of numerous inflammatory genes in gastric tumors from 20- to 24-week-old gp130^F/F and gp130^F/F:Asc^−/− mice (Fig. 3A). We also assessed whether the suppressed gastric tumorigenesis was associated with reduced expression of various genes encoding angiogenic factors implicated in gastric cancer, namely Cxcl1, Cxcl2, Vegf, Mmp2, and Mmp9. However, their mRNA levels were not significantly reduced in gp130^F/F:Asc^−/− tumors (Supplementary Fig. S3F), suggesting that ASC does not promote tumor angiogenesis in gp130^F/F mice.

Because ASC is expressed in both the gp130^F/F gastric tumor epithelium and immune cell-containing stroma, albeit significantly higher in the tumor epithelium (Fig. 3B), we assessed whether ASC-expressing myeloid cells contributed to gastric tumorigenesis by generating reciprocal bone marrow chimeras between irradiated 8-week-old gp130^F/F and gp130^F/F:Asc^−/− mice that were subsequently aged. The size of hyperplastic stomachs and tumor mass from 20- to 24-week-old gp130^F/F recipients reconstituted with gp130^F/F:Asc^−/− donor bone marrow was comparable to control gp130^F/F mice reconstituted with autologous gp130^F/F bone marrow (Fig. 3C and D). Similarly, the reciprocal reconstitution of gp130^F/F:Asc^−/− recipients with gp130^F/F bone marrow had no effect on gastric tumor growth (Fig. 3C and D). Collectively, these data indicate that ASC-expressing myeloid cells do not promote gastric tumorigenesis.

We have previously demonstrated that gastric tumorigenesis in gp130^F/F mice is associated with a high PCNA proliferative index compared to wild-type mice (24). However, immunostaining of gastric tumor sections from 20- to 24-week-old gp130^F/F and gp130^F/F:Asc^−/− mice revealed comparable PCNA⁺ cell numbers, indicating that suppressed tumorigenesis in gp130^F/F:Asc^−/− mice was not characterized by reduced tumor cell proliferation (Fig. 4A). Similarly, the expression of cell-cycle regulatory genes Ccnb1, Ccnd1, Ccnd2, c-myc, and Cdc42 was comparable in gp130^F/F:Asc^−/− and gp130^F/F tumors (Supplementary Fig. S3G).

We next investigated whether ASC contributed to increased survival of neoplastic gastric epithelial cells, a cellular process that is associated with gastric tumorigenesis in gp130^F/F mice (20). Indeed, TUNEL⁺ and cleaved caspase-8⁺ apoptotic cell numbers were significantly increased (∼3-fold) in gastric tumors of 20- to 24-week-old gp130^F/F:Asc^−/− compared with gp130^F/F mice (Fig. 4B and C). Dual immunofluorescence staining for cleaved caspase-3 and the epithelial marker E-cadherin further confirmed a significant increase in the number of apoptotic epithelial cells in gastric tumors of gp130^F/F:Asc^−/− versus gp130^F/F mice (Fig. 4D; Supplementary Fig. S4A).

Because activation of the protumorigenic transcription factor NF-κB within the intestinal epithelium has been linked with suppressing epithelial cell apoptosis in colitis-associated cancer (29), we investigated whether a similar function for NF-κB could be assigned to the gastric epithelium during tumorigenesis. Indeed, in gastric tumors of 20- to 24-week-old gp130^F/F mice, immunofluorescence indicated both nuclear and cytoplasmic staining for the phosphorylated (Ser536) p65 subunit of NF-κB primarily in the mucosal glandular epithelium, with little to no pNF-κB p65 staining in the immune/inflammatory cell-rich submucosal or lamina propria regions (Supplementary Fig. S4B). Furthermore, in the smaller gp130^F/F:Asc^−/− tumors displaying increased apoptosis, immunoblotting revealed significantly reduced pNF-κB p65 levels (Supplementary Fig. S4C). In addition, immunohistochemistry confirmed that the reduced phosphorylation (i.e., activation) of NF-κB associated with a significantly lower number of pNF-κB p65-positive cells in the mucosal epithelium, compared to gp130^F/F tumors (Fig. 4E). By contrast, immune/inflammatory cell aggregates found within the matching submucosal areas of gp130^F/F and gp130^F/F:Asc^−/− gastric tumors contained comparable low numbers of pNF-κB p65-positive cells (Supplementary Fig. S4D). Collectively, these data suggest that ASC deficiency in the gp130^F/F gastric tumor epithelium augments caspase-8-mediated cell death, which correlates with reduced NF-κB activation.

To elucidate a role for the ASC inflammasome effector cytokines IL1β and/or IL18 in gastric tumorigenesis, we initially measured IL1β and IL18 expression levels in gastric tumors of 10- to 12-week-old and 20- to 24-week-old gp130^F/F mice. Although Il1b mRNA levels were significantly increased by up to approximately 50-fold in tumor and approximately 6-fold in adjacent tumor-free gastric antrum tissues of gp130^F/F mice compared to normal gastric antrum tissue of gp130^+/+ mice (Fig. 5A), immunoblotting indicated that protein levels of pro (31 kD) and mature (17 kD) forms of IL1β were only increased (albeit significantly) by up to approximately 3-fold in gp130^F/F tumor and/or tumor-free lysates compared with gp130^+/+ gastric antrum lysates, with the highest increase in 20- to 24-week-old gp130^F/F tumors (Fig. 5B and C). By contrast, despite a <3-fold increase in Il18 mRNA levels in gp130^F/F gastric tumor and/or nontumor tissues compared with gp130^+/+ antrum tissue (Fig. 5A), immunoblots revealed that the mature 18 kDa form of IL18, but not the immature 24 kDa proIL18, was specifically upregulated by up to 27-fold in both 10- to 12-week-old and 20-to 24-week-old gp130^F/F tumor and tumor-free tissues (Fig. 5B and C).

In light of these findings suggesting increased processing of mature IL18 protein, we next assessed whether IL18 specifically acted downstream of ASC to promote gastric tumorigenesis. Indeed, immunoblotting revealed that mature IL18 protein levels were significantly reduced in tumor lysates from 20- to 24-week-old gp130^F/F:Asc^−/− versus gp130^F/F mice, which was not observed for mature IL1β (Fig. 5D). The reduced processing of IL18 in gp130^F/F:Asc^−/− gastic tumor lysates was also confirmed by an ELISA specific for mature IL18 (Supplementary Fig. S5A). Moreover, gastric hyperplasia and tumor burden were significantly reduced in 20- to 24-week-old, but not 10- to 12-week-old, gp130^F/F:Il18^−/− mice lacking IL18 compared with gp130^F/F mice (Fig. 6A–D; Supplementary Fig. S5B–S5E), which mimicked the suppressed gastric tumor phenotype of gp130^F/F:Asc^−/− mice. In contrast, in gp130^F/F mice, the genetic ablation of IL1R, leading to blockade of IL1β (and IL1α) activity, had no significant impact on tumor burden, as evidenced by comparable tumor mass (gp130^F/F, 0.130 ± 0.017 g vs. gp130^F/F:Il1r^−/−, 0.121 ± 0.064 g), tumor incidence and overall stomach size in 20- to 24-week-old gp130^F/F and gp130^F/F:Il1r^−/− mice (Supplementary Fig. S5F–S5K). Because the long-term, daily administration of tumor-bearing gp130^F/F mice with the IL1R antagonist anakinra (high dose, 100 mg/kg) also had no effect on tumorigenesis, these observations suggest that IL1β does not play a major role contributing to tumorigenesis in this model. In addition, treatment of human AGS and MKN1 gastric cancer cells with anakinra had no effect on anchorage-dependent and -independent colony formation (Supplementary Fig. S5L).

Similar to gp130^F/F:Asc^−/− mice, ameliorated gastric tumorigenesis in gp130^F/F:Il18^−/− mice was associated with significantly elevated numbers of apoptotic TUNEL⁺ (Fig. 6E) and caspase-8⁺ (Fig. 6F) cells, and reduced numbers of cells expressing activated NF-κB (Supplementary Fig. S6A), within the mucosal epithelium. In contrast, we observed no changes in the level of proliferation, infiltration of inflammatory cells, or expression of cell-cycle, angiogenic, and inflammatory genes (Supplementary Figs. S6B–S6D and S7A–S7C), in the tumor epithelium. Collectively, these findings support a protumorigenic role for ASC in gastric cancer that is mediated largely by IL18.

Consistent with elevated Asc gene expression in the gastric epithelium of gp130^F/F mice (Fig. 3B), mRNA levels for Il18 or its Il18r1 receptor gene were significantly increased in EpCam⁺ epithelial compared to CD45⁺ immune (or stroma) cells in gp130^F/F gastric nontumor or tumor tissues (Fig. 7A; Supplementary Fig. S7D). By contrast, although Il1r expression levels were generally unchanged in the gastric epithelial and immune compartments, Il1b was lowly-expressed in epithelial versus immune cells or stroma in gp130^F/F gastric tumors (Fig. 7A; Supplementary Fig. S7D). In support of these findings, ELISA confirmed that total and mature IL18 protein levels were significantly increased in tumor compared to matched nontumor tissue lysates from gastric cancer patients, with total IL18 levels higher than those for total IL1β (Fig. 7B). Importantly, the elevated levels of IL18, but not IL1β, positively correlated with increased ASC mRNA levels in gastric cancer patient tumors (Fig. 7C). Furthermore, secreted (i.e., mature) IL18 protein was detected by ELISA in human gastric cancer cell line supernatants, whereas as expected, secreted IL1β protein was undetectable due to the absence of steady-state cellular proIL1β levels (Fig. 7D).

The therapeutic utility of targeting IL18 in human gastric cancer was supported by the observation that treatment of AGS and MKN1 human gastric cancer cells with an IL18 neutralizing antibody suppressed anchorage-dependent and -independent colony formation, and augmented apoptosis (Fig. 7E–G). Furthermore, CRISPR/Cas9-mediated ASC ablation in AGS cells suppressed colony formation, which coincided with reduced secretion of mature IL18 (but not IL1β) and activation of caspase-1 and NF-κB, and conversely elevated apoptosis (Fig. 7H–K). Notably, these findings mimic those observed upon CRISPR/Cas9-mediated knockout of caspase-1 in AGS cells (Supplementary Fig. S8A–S8C), and therefore support the notion that elevated IL18 production (downstream of ASC and caspase-1) in human gastric cancer cells promotes cell-autonomous growth, which can be readily targeted therapeutically.

Over recent years, studies largely restricted to melanoma, skin and colon carcinogenesis have identified complex and contrasting tumor-promoter and tumor-suppressor roles for the inflammasome adaptor ASC, which reveal a functional dependency by ASC-associated inflammasomes on cell type (i.e., myeloid vs. epithelial), tissue specificity, and disease stage (19). For instance, in experimentally-induced skin carcinogenesis, myeloid-derived ASC expression favors tumorigenesis, whereas ASC expression in keratinocytes negates tumor formation (30, 31). In addition, modulating ASC expression in primary and metastatic human melanoma demonstrated opposing anti- and protumorigenic functions, respectively, for ASC, indicating that the role of ASC in cancer can depend on disease stage (32). With respect to the gut, ASC-deficient mice are more susceptible to azoxymethane (AOM)/dextran sodium sulfate (DSS)-induced colitis-associated carcinogenesis (CAC), thus supporting an antitumorigenic role for ASC (33). By contrast, a protumorigenic role for ASC in colorectal cancer is suggested in AhR^−/− and Apc^Min/+ spontaneous colorectal cancer models, whereby ASC deficiency ameliorates tumor formation (34).

Here, in gastric cancer, we reveal a protumorigenic role for ASC, whose elevated expression was observed in tumors of approximately 75% of intestinal-type gastric cancer patients, that is independent of gastric inflammation. Specifically, we demonstrate that suppressed gastric tumorigenesis in gp130^F/F:Asc^−/− mice is associated with reduced caspase-1 activation and elevated numbers of caspase-3- and caspase-8-expressing cells within the tumor epithelium, and thus uncover a hitherto unknown function for ASC inflammasomes in eliciting an antiapoptotic response in neoplastic gastric epithelial cells involving caspases-3 and -8. Furthermore, our discovery that ASC ablation in gastric epithelial (cancer) cells suppresses their growth potential, which correlates with reduced NF-κB activation (and IL18 processing/production), along with elevated apoptosis, suggests a role for NF-κB as a signaling facilitator of the antiapoptotic/prosurvival function for ASC in the gastric (tumor) epithelium. Indeed, this notion is consistent with the emergence that NF-κB promotes the initiation and/or progression of numerous epithelial cancers, as evidenced by the observation that during colorectal cancer, in which ASC has been assigned a protumorigenic role (34), NF-κB activation in the intestinal epithelium promotes tumorigenesis by suppressing apoptosis (29). Therefore, in the context of gastric cancer, our findings invoke the existence of potential signaling cross-talk between activation of the canonical ASC inflammasome (via NF-κB) and caspase-8 apoptotic machinery to dampen the latter, thus augmenting tumor growth. Furthermore, our observations revealing that the predominate NF-κB signal is localized to epithelial and not immune cells within gastric tumors is consistent with our proposed tumor (epithelial) cell-autonomous antiapoptotic role for the ASC inflammasome/IL18 axis, and thus provide an explanation (at least in part) for why inflammation is not affected in the gp130^F/F:Asc^−/− tumors.

Our current findings also expand upon the traditional role of ASC that has been linked to cell death facilitated by canonical inflammasome-mediated pyroptosis via caspase-1, or more recently the formation of a noncanonical, apoptosis-inducing ASC apoptosome (with AIM2 and NLRP3) via interaction with caspase-8, independent of inflammatory-related caspase-1 activation (35–37). Regarding the former, it has recently emerged that ASC inflammasome/caspase-1-dependent cleavage and activation of the pore-forming effector protein, gasdermin D, is a critical event in pyroptosis (38). Although the pathophysiological role of gasdermin D-mediated pyroptosis, including in the context of cancer, is ill-defined, our data presented here suggest that the novel antiapoptotic arm of the ASC inflammasome/caspase-1 axis in gastric epithelial cells, which protects against caspase-8–mediated cell death, occurs independent of gasdermin D processing by the ASC inflammasome. Considering that the mechanistic basis governing this dual mode of ASC-mediated cell death (i.e., apoptosis and pyroptosis) remains to be fully elucidated, our notion regarding the independent role of gasdermin D in gastric cancer (and potentially other cancers) warrants further investigation.

We also note that the above-mentioned pro-apoptotic and pro-pyroptotic functions previously assigned to ASC most likely explain its tumor suppressor actions in certain cancers, which have also been linked to methylation-induced silencing of ASC (39). In this respect, a recent study indicated that ASC is methylated in approximately one-third of gastric cancer cases, although the histologic subtype (i.e., intestinal or diffuse) involved and the effect on ASC expression in the tumors analyzed were not reported (40).

Another key finding of our study was the identification of IL18, but not IL1β, as a major downstream inflammasome effector cytokine, which propagates the growth potential of gastric cancer cells. Previous studies have shown that stomach-specific overexpression of human IL1β in transgenic mice induces gastric inflammation-associated carcinogenesis via the recruitment of MDSCs (7), and in human gastric cancer elevated gastric IL1β production is associated with inflammation (41). Notably, our demonstration that IL18, like ASC, promotes tumor growth independent of inflammation in gp130^F/F mice contrasts these findings for IL1β. Here, in gp130^F/F gastric tumors we show that gene expression for IL18 and its IL18R1 receptor, as well as for ASC, is elevated in epithelial compared to inflammatory/immune cells, which contrasts the low and high gene expression levels for IL1β in epithelial and inflammatory cells, respectively. Together with the high basal production of IL18 in supernatants of human gastric cancer cell lines, along with the growth inhibitory effect of a neutralizing IL18 monoclonal antibody on human gastric cancer cells, our findings support a direct, cell-autonomous effect of IL18 (downstream of ASC) on gastric cancer cells. We therefore propose that the functional dichotomy between IL18 and IL1β in gastric cancer can be explained by, at least in part, the preferential requirement of epithelial (tumor) cells for IL18, which directly acts to promote growth of the tumor epithelium, whereas immune cells predominantly utilize IL1β to provide a proinflammatory microenvironment that can support tumorigenesis.

This notion could also account for our previous observation in gp130^F/F mice that genetic ablation of the MyD88 adaptor protein, which is crucial for IL18R signaling, also suppresses growth of the gastric tumor epithelium independent of inflammation (42). Furthermore, in another gastric cancer mouse model, Gan, in which inflammation drives gastric tumorigenesis, the ablation of MyD88 in bone marrow–derived immune cells (thus disrupting IL1β/IL1R signaling, which also requires MyD88) suppressed the tumor-promoting inflammatory microenvironment (43). It is therefore perhaps not surprising that this latter observation contrasts our current bone marrow chimera data in gp130^F/F mice whereby myeloid cells expressing low levels of IL18 (and ASC) do not promote gastric tumorigenesis, suggesting that at least this cellular component of the tumor microenvironment does not influence apoptosis and tumor growth in the gp130^F/F gastric cancer model.

Although we reveal here a key role for IL18 as a downstream effector of ASC inflammasomes in gastric cancer, our study also raises the intriguing question concerning the identity of the specific PRR(s), as well as the source and nature of its agonist(s) (e.g., microbial- and/or host-derived), that comprises the protumorigenic ASC inflammasome? In the current absence of definitive in vivo evidence to answer this question, which for instance requires genetic ablation of specific ASC inflammasome-associated PRRs (e.g., AIM2, NLCR4, NLRP1, NLRP3) in our gp130^F/F gastric cancer model, we refer to our in vitro data demonstrating reduced levels of secreted (mature) IL18 and activated caspase-1 proteins upon CRISPR/Cas9-mediated knock-out of ASC in cultured human gastric cancer cells (Fig. 7J). These observations are suggestive of constitutive ASC inflammasome activation driven by the cell intrinsic production and release of inflammasome-activating agonists during culture. Interestingly, cultured human cancer cell lines constitutively release host-derived damage-associated molecular patterns (DAMP), for example DNA and HMGB1 (44, 45), which can activate the AIM2 and NLRP3 inflammasomes, respectively (46, 47). Therefore, it is tempting to speculate that cultured human gastric cancer cell lines, and by analogy the gastric tumor epithelium, also have the potential to constitutively release such host-derived DAMPs, which promote cellular growth via their specific PRR-associated inflammasome.

It is also worth considering our findings with previous investigations into the role of IL18 via inflammasome activation in gastrointestinal cancers, which have been restricted to the AOM/DSS-induced CAC model in concert with either Il18^−/− mice or the administration of IL18 to inflammasome-deficient (Casp1^−/−) mice. These studies indicate a protective role for IL18 in the initiating stages of CAC, attributed to high IL18 expression levels in the intestinal epithelium, which maintain barrier integrity through epithelial cell proliferation and survival (48, 49). Thus, although this growth-potentiating effect of IL18 on the intestinal epithelium protects against colonic tumorigenesis, our current findings indicate that, conversely, in the gastric epithelium IL18 suppresses apoptosis to promote gastric tumorigenesis. In this regard, our current study reveals tissue-specific activities of IL18, thus supporting the pleiotropic actions previously ascribed to IL18 on the migration and proliferation of human gastric cancer cell lines (9, 12, 13).

In summary, our current study suggests the existence of a novel protumorigenic ASC/IL18/NF-κB signaling axis that augments gastric epithelial cell survival in gastric cancer. Despite the recent development of numerous small-molecule inhibitors against individual ASC inflammasome components, namely NLRP3, NLRP1, NLRC4, and AIM2 (50), their specificity and efficacy as anticancer agents in vivo will require rigorous evaluation in numerous cancer models that take into account the tissue-specific and cell-type contexts that define the multifaceted activities of ASC in cancer. Furthermore, in gastric cancer, the potential application of such inhibitors will be influenced by the future identification of specific PRRs, which comprise the disease-associated ASC inflammasome. In light of our findings, and given the current lack of preclinical-validated inhibitors that can block the actions of ASC, therapies that neutralize the biological activity of IL18 may serve as effective weapons against gastric cancer characterized by dysregulated ASC inflammasome-driven IL18 production.

C. Girard reports receiving a commercial research grant from salary supported by an unrestricted grant from AB2 Bio S.A., Lausanne, Switzerland. C. Gabay reports receiving other commercial research support from AB2 Bio and is a consultant/advisory board member of AB2 Bio. No potential conflicts of interest were disclosed by the other authors.

Conception and design: V. Deswaerte, T. Putoczki, B.J. Jenkins

Development of methodology: V. Deswaerte, A. West, T. Putoczki, B.J. Jenkins

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Deswaerte, P. Nguyen, A. West, A.F. Browning, L. Yu, J. Balic, C. Girard, H. Oshima, K.Y. Fung, H. Tye, M. Najdovska, M. Oshima, C. Gabay, T. Putoczki, B.J. Jenkins

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Deswaerte, P. Nguyen, A. West, A.F. Browning, L. Yu, T. Livis, H. Tye, M. Najdovska, C. Gabay, T. Putoczki, B.J. Jenkins

Writing, review, and/or revision of the manuscript: P. Nguyen, A.F. Browning, C. Girard, M. Ernst, C. Gabay, T. Putoczki, B.J. Jenkins

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Livis, A. Preaudet, M. Najdovska, M. Ernst, T. Putoczki, B.J. Jenkins

Study supervision: T. Putoczki, B.J. Jenkins

Other (performing experiments and analyzing data): S.M. Ruwanpura

We thank R. Smith and P. Bouillet for comments, K. Fitzgerald (University of Massachusetts Medical School, Worcester, MA) for providing Asc^−/− mice, and C. Nold (Hudson Institute of Medical Research, Melbourne, Australia) for providing anakinra. This work was funded by the National Health and Medical Research Council (NHMRC) of Australia (to B. Jenkins), and the Operational Infrastructure Support Program by the Victorian Government of Australia. P. Nguyen, A. Browning, and J. Balic were supported by Australian Postgraduate Awards from the Australian Government, and A. West was supported by an NHMRC Early Career Fellowship. T. Putoczki was supported by a Victorian Cancer Agency Mid-Career Fellowship. B. Jenkins was supported by an NHMRC Senior Medical Research Fellowship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.