|Home | About | Journals | Submit | Contact Us | Français|
Recent evidence has indicated that innate immune sensing of cytosolic DNA in dendritic cells via the host STING pathway is a major mechanism leading to spontaneous T cell responses against tumors. However, the impact of the other major pathway triggered by intracellular DNA, the AIM2 inflammasome, on the functional output from the STING pathway is poorly understood. We found that dendritic cells and macrophages deficient in AIM2, ASC or caspase-1 produced markedly higher IFN-β in response to DNA. Biochemical analyses showed enhanced generation of cGAMP, STING aggregation, and TBK1 and IRF3 phosphorylation in inflammasome-deficient cells. Induction of pyroptosis by the AIM2 inflammasome was a major component of this effect, and inhibition of caspase-1 reduced cell death, augmenting phosphorylation of TBK1/IRF3 and production of IFN-β. Our data suggest that in vitro activation of the AIM2 inflammasome in murine macrophages and dendritic cells leads to reduced activation of the STING pathway, in part through promoting caspase-1-dependent cell death.
Spontaneous T cell responses against a growing tumor frequently occur, despite the absence of infectious agents; and the presence of activated CD8+ T cells in the tumor microenvironment correlates with improved prognosis (1). We and others have identified that production of host type I IFN has a critical role in the spontaneous activation of specific tumor CD8+ T cells (2, 3). Mechanistic studies using transplantable tumors in mice deficient in key innate sensing molecules showed that deficiency in the adaptor Stimulator of Interferon Genes (STING), or the downstream transcription factor IRF3, blunted T cell priming and impaired rejection of immunogenic tumors (4). Moreover, presence of DNA was found in the cytosol of intratumoral DCs, which correlated with IRF3 translocation to the nucleus and expression of IFN-β. These data suggest that activation of the STING pathway plays a critical role in the innate immune sensing of tumors in vivo, apparently through cytosolic sensing of DNA.
STING is an adaptor protein located in the endoplasmic reticulum (ER). In the presence of cytosolic DNA, the sensor GMP-AMP synthase (cGAS) produces cyclic-GMP-AMP (cGAMP), which binds to STING and triggers its activation (5, 6). Aggregated STING translocates in vesicles from the ER to perinuclear sites (7). This is accompanied by TBK1 recruitment, IRF3 phosphorylation and nuclear translocation (8), and transcription of type I IFN and other immune genes (9–11).
Activation of the STING pathway is critical in the host defense against pathogens (12) and in the generation of an anti-tumor T cell response. However, its inappropriate activation leads to generation of autoimmune diseases such as Aicardi-Goutières syndrome (13) or systemic lupus erythematosus (SLE) (14). Thus, several regulatory mechanisms coexist to keep production of type I IFN in check. Two levels of negative regulation of the STING pathway have been described: elimination of aberrant DNA by DNases (15) and posttranslational modification of STING following its activation (16). However, the fact that the type I IFN response to DNA stimulation is enhanced in Aim2-deficient cells (17–20) suggests that another potential level of negative regulation of the STING pathway may occur when additional innate immune pathways are simultaneously activated. The presence of cytosolic DNA also triggers formation of the AIM2 (absent in melanoma 2) inflammasome, a heterocomplex that contains AIM2, the adaptor protein ASC (apoptosis-associated speck-like protein) and caspase-1. This leads to activation of caspase-1 that generates matured forms of IL-1β and IL-18 (21), and pyroptosis, a form of cell death (22). Whether the AIM2 inflammasome influences STING pathway activation has not yet been described. In the current report, we investigated the regulatory role of the AIM2 inflammasome on STING pathway activation in vitro. We found increased cGAMP generation, STING aggregation, TBK1 and IRF3 phosphorylation, and IFN-β transcription in AIM2 inflammasome-deficient APCs upon cytosolic DNA exposure. Mechanistically, a major component of this effect could be the decreased cell death in inflammasome-deficient cells. Our results indicate an inhibitory effect of the AIM2 inflammasome on the STING pathway in response to cytosolic DNA, and suggest that targeted inflammasome inhibition could represent a strategy for prolonging APC survival in the context of cytosolic DNA sensing, leading to potentiation of the STING pathway.
Immortalized WT, Aim2−/−, ASC−/− and Caspase-1−/−/11−/− macrophages were obtained as described in Roberson et al. (23). Cells were maintained in DMEM supplemented with 10% heat-inactivated FCS, penicillin, streptomycin, L-arginine, L-glutamine, folic acid, and L-asparagine. Bone marrow-derived dendritic cells (BMDCs) were generated from the tibiae and femurs of WT, Aim2−/−, ASC−/−, Caspase-1−/−/11−/− female mice of 6 to 8 weeks age. Mice were housed in the SPF animal facility at the University of Chicago. Cells were cultured in DMEM supplemented with 10% heat-inactivated FCS, penicillin, streptomycin, L-arginine, L-glutamine, folic acid, and L-asparagine and the presence of rmGM-CSF (20 ng/ml; BioLegend) for 9 days. All cells were maintained at 37°C with 5% CO2. The macrophages used throughout the study were immortalized and the BM-DCs primary cells.
For stable overexpression of myc-ASC in ASC−/− macrophages, the full-length mASC-myc DNA sequence was generated by PCR. Sequence encoding full-length mASC was amplified from MGC Mouse Pycard cDNA (Dharmacon) using a 5′ primer containing a EcoR1 site, atg and myc sequence: GAATTCTCGAGATGGAGCAGAAGCTGATTTCCGAGGAGGACCTGGGGCGGGCACGAGATGCCATCCTGGACGCTCTTGAAAACTTGTCAGGGGATGAACTCAAAAAGTTCAAGATGAAGCTGCTGACAGTGCAACTGC; and a 3′ primer containing the NotI site: TATATGCGGCCGCTCAGCTCTGCTCCAGGTCCATCACCAAGT. For stable overexpression of myc-p204 in WT macrophages, the full-length p204-myc DNA sequence was generated by PCR. Sequence encoding full-length p204 was amplified from MGC Mouse p204 cDNA (Dharmacon) using a 5′ primer containing a EcoR1 site, atg and myc sequence: GTGAATTCATGGAGCAGAAGCTGATTTCCGAGGAGGACCTGGTGAATGAATACAAGAGAATTGTTCTGCTGAGAGGACTTGAATGTATC; and a 3′ primer containing the NotI site: atGCGGCCGCTCACTTTCTAGCATTGATGACCT. The mouse myc-ASC or myc-p204 PCR products were gel purified and double digested with EcoRI and NotI, then cloned into the multiple cloning site of pMXS-IRES-GFP with Quick ligation kit (NEB).
Tumor-derived DNA was used for all stimulations; for simplification, it is referred to as “DNA” throughout the manuscript. Genomic DNA from B16.F10 melanoma cells was purified using the QIAamp DNA Mini Kit according to the manufacturer’s instructions. To ensure DNA was free of other danger signals (i.e. LPS), immortalized STING−/− macrophages were stimulated with the purified DNA to assess for STING-dependence, and WT macrophages were stimulated with DNAse I-treated DNA. In both cases there was not expression of IFN-β.
Immortalized macrophages or BM-DCs were stimulated with different concentrations of DNA in the presence of Lipofectamine 2000 (Invitrogen). For some experiments, macrophages were stimulated with 50 μg/ml of DMXAA (Selleckchem).
Cells stimulated with 1 μg/ml DNA or 50 μg/ml DMXAA were stained with anti-CD11b-APC (M1/70; BioLegend), rabbit anti-HA-tag (C29F4; Cell Signaling) and anti-Rabbit IgG-PE (Invitrogen), and DAPI (Invitrogen). Single cell images were acquired in the ImageStreamxMark II (Amnis) and data were analyzed using IDEAS software.
Whole cell extracts were electrophoresed in 10% SDS-PAGE gels and transferred onto Immobilon-FL membranes (Millipore). Blots were incubated with antibodies specific for pTBK1 (Ser172), pIRF3 (Ser396), total TBK1, total IRF3, cGAS, STING, GAPDH or β-actin (Cell Signaling). For ASC-myc detection in ASC−/− macrophages, anti-ASC (Millipore) was used. For p204-myc detection in WT macrophages, anti-myc (Cell Signaling) was used. Anti-rabbit or anti-mouse IRDye 680RD label secondary antibodies were used for visualization of bands with the Odyssey Imaging system (LI-COR).
Conditioned media from cells stimulated with DNA were collected after 12 (macrophages) or 8 (BM-DCs) hours. IFN-β concentration was assessed using VeriKine™ Mouse Interferon Beta ELISA Kit (PBL interferon source) according to the manufacturer’s instructions.
Total RNA was isolated using the RNeasy® kit (Qiagen) and incubated with Deoxyribonuclease I, Amplification Grade (Invitrogen). cDNA was synthesized using High Capacity cDNA Reverse Transcription Kit (Applied Biosystem) and expression of IFN-β, CXCL10, and GAPDH as endogenous control was measured by real-time qRT-PCR with specific primers and universal probes from Roche for mouse INF-β (5′: GGAAAGATTGACGTGGGAGA, 3′: CCTTTGCACCCTCCAGTAAT, and universal probe number 108 CTGCTCTC), mouse CXCL10 (5′: GCTGCCGTCATTTTCTGC, 3′: TCTCACTGG CCCGTCATC, and universal probe number 3), and mouse GAPDH (5′: AGCTTGTCATCAACGGGAAG, 3′: TTTGATGTTAGTGGGGTCTCG, and universal probe number 9 CATCACCA); using a 7300 Real Time PCR system (Applied Biosystem). The results are expressed as relative expression to GAPDH. The expression of Aim2, p204 and p202b was analysed using the primers described in Brunette at al. (24).
cGAMP was quantified via LC/MS/MS at Climax Laboratories (San Jose, CA). Briefly, one million macrophages or BM-DCs were stimulated with DNA for one hour and then resuspended in PBS. cGAMP was extracted with 100% acetonitrile and analyzed by an LC/MS/MS system, Sciex API-4000Qtrap Mass Spectrometer and a Shimadzu HPLC/Autosampler with an ACE C18 column (2.1×100mm, 5μm). Positive Electronic Spray Ionization (ESI) and multiple reactions monitor (MRM) were used. The MRM transition of the test compound was 675/136 (m/z). The HPLC mobile phase A and B was 0.5% Formic acid in 5 mM NH4Ac solution and Acetonitrile/water (9/1) with 0.5% Formic acid. A related compound, Rp,Rp-c-diAMPSS, was used as an internal standard. The limit of quantification was 1.0 ng/mL and the dynamic range was 1.0–500 ng/mL.
Culture medium from cells stimulated with DNA was collected and incubated with the reaction mixture from the LDH Cytotoxicity Assay Kit (Pierce) according to the manufacturer’s instructions. The rates of cell death were calculated using the absorbance values from DNA-stimulated cells (Experimental cell death), Triton-X-treated cells (100% of cell death), and culture medium-treated (spontaneous cell death) and applying the formula: (Experimental cell death − spontaneous cell death)/(100% cell death − spontaneous cell death) × 100%.
Student’s paired t-test was used to calculate two-tailed p values to estimate statistical significance of differences between two treatment groups using Prism 6 software. Statistically significant P values are labelled in the figures and the legends with asterisks.
We previously had demonstrated that the introduction of DNA into the cytosol of DCs or macrophages resulted in IFN-β production by a mechanism dependent upon cGAS, STING, TBK1, and IRF3 (4). In order to study the potential impact of the AIM2 inflammasome on this process, we generated DCs from the bone marrow (BM-DCs) of WT mice versus Aim2-deficient mice, and stimulated them in vitro with DNA in the presence of lipofectamine. The induction of IFN-β mRNA was markedly enhanced with Aim2-deficient DCs as compared to DCs from WT mice (Figure 1A). This result was also confirmed at the protein level through analysis of IFN-β secretion (Figure 1B). To explore this phenomenon in a different cell type, we utilized immortalized bone marrow macrophages. Using a wide range of DNA doses, augmented induction of IFN-β expression was observed in AIM2 inflammasome-deficient macrophages (Figure 1C). The expression of CXCL10 was also increased in a dose-dependent manner in AIM2 inflammasome-deficient cells (Supplemental figure 1A). To test if expression of IFN-β occurred with different kinetics in Aim2-sufficient and -deficient cells, a time course was performed over 24 hours of stimulation with DNA. Aim2 deficiency resulted in higher IFN-β expression at all time points (Supplemental Figure 1B). In order to determine whether the entire AIM2 inflammasome complex was involved in this regulatory mechanism, we also analyzed BM-DCs from ASC- or caspase1/11-deficient mice. As observed with the Aim2−/− DCs, ASC−/− and caspase1−/−/11−/− BM-DCs showed significantly greater IFN-β gene expression compared to WT DCs (Figure 2A–B). Similarly, a DNA dose-response titration on macrophages showed increased expression of IFN-β in ASC−/− and caspase1−/−/11−/− compared to WT cells (Figure 2C). We also analyzed macrophages deficient in other inflammasomes, and observed that NLRP3- or NLRC4 (IPAF)- deficient macrophages showed only minimal changes in IFN-β production in response to DNA stimulation (Figure 2D). In the case of BM-DCs, caspase1−/−/11−/− showed a weaker phenotype than with macrophages, however it was still significantly different from WT BM-DCs (Figure 2A–2D). To be certain that the augmented IFN-β production seen in AIM2 inflammasome-deficient APCs was a direct result of the absence of this pathway and not a developmental alteration; we reintroduced ASC into ASC−/− macrophages (Figure 2E). Restoration of ASC expression completely eliminated the augmented IFN-β production seen in response to cytosolic DNA (Figure 2F). Together, these results demonstrate that induction of IFN-β production in response to cytosolic DNA stimulation is inhibited by all components of the AIM2 inflammasome, not only AIM2.
To explore further the mechanism by which the AIM2 inflammasome might antagonize activation of induction of IFN-β production via the STING pathway, we analyzed the stages of activation of the STING pathway in WT and AIM2 inflammasome-deficient cells. Phosphorylation of TBK1 and IRF3 was assessed by Western blot analysis at multiple time points following DNA stimulation. Early after DNA introduction, Aim2-deficient macrophages showed markedly augmented TBK1 and IRF3 phosphorylation compared to WT macrophages (Figure 3A). To confirm the higher activation of TBK1 and IRF3 in a different cell type, BM-DCs from WT and Aim2−/−, ASC−/− and caspase1−/−/11−/− mice were stimulated with two different concentrations of DNA. The phosphorylation of TBK1 and IRF3 correlated with the amount of DNA used for stimulation, and was higher in the AIM2 inflammasome-deficient cells (Figure 3B). The augmented activation of the STING pathway was not due to a higher expression of the upstream proteins, the DNA sensor cGAS or the adapter STING, as the level of expression of these proteins was similar in AIM2 inflammasome-sufficient and -deficient cells (Supplemental Figure 2A–B). To examine a potential regulatory effect of AIM2 upstream of phosphorylation of TBK1, we analyzed STING aggregation after treatment with DNA. WT or ASC−/− macrophages were stably transfected with STING-HA for evaluation of induction of aggregates by ImageStream analysis using an anti-HA antibody. In response to cytosolic DNA stimulation, we found that the percentage of cells with perinuclear STING aggregates was markedly higher in ASC−/− macrophages compared to WT macrophages (Figure 3C–D). To confirm STING activation in cells without exogenous expression of STING-HA tag, we assessed the degradation of total STING after DNA stimulation by Western blot analysis, as it has been demonstrated that STING undergoes ubiquitination and degradation after its activation (16). Total levels of STING were reduced over time after DNA stimulation in ASC−/− macrophages, but remained more constant in WT cells (Supplemental figure 2C).
Given that STING aggregation seemed to be augmented in AIM2 inflammasome-deficient cells, we moved upstream to evaluate activation of the DNA sensor cGAS in the absence of the AIM2 inflammasome. For this purpose, intracellular cGAMP produced enzymatically by endogenous cGAS was measured by liquid chromatography-tandem mass spectrometry (LC/MS/MS) in extracts from WT and ASC−/− macrophages stimulated with DNA. cGAMP levels were below the limit of detection in WT cells, whereas ASC−/− cells produced robust levels of cGAMP (Figure 3E). To confirm this result, extracts from DNA-stimulated BM-DCs from WT, Aim2−/−, ASC−/− and caspase1−/−/11−/− mice were also analyzed for cGAMP generation. Consistent with the results with macrophages, induced levels of cGAMP were markedly augmented in Aim2−/−, ASC−/− and caspase1−/−/11−/− BM-DCs compared with WT BM-DCs (Figure 3F).
It was important to exclude whether the activity of STING itself was affected by the absence of the AIM2 inflammasome. To this end, we utilized DMXAA, a direct agonist of mouse STING (25), to bypass cGAMP formation. WT and ASC−/− macrophages stimulated with DMXAA showed similar STING aggregation in perinuclear sites, STING degradation, and phosphorylation of TBK1/IRF3 in response to DMXAA stimulation (Figure 4A–F). Together, these data demonstrate that the inhibition of the STING pathway by the AIM2 inflammasome impacts upstream STING, thus affecting the entire STING pathway activation cascade.
Having shown that the STING pathway is antagonized by the AIM2 inflammasome, we sought to determine the mechanism of this inhibition. It has been shown that members of the AIM2-like receptor (ALR) gene family influence the STING or the inflammasome pathways (24). We compared the basal expression of p204 and p202b in APCs, two ALRs related to the activation of the STING pathway, and also Aim2 expression as control. The basal expression of these three ALRs was similar in WT and ASC−/− BM-DCs (Supplemental Figure 3A); however, the expression of p204 was higher in ASC-deficient macrophages (Supplemental Figure 3B). To evaluate whether the lower level of p204 in WT macrophages could be limiting the induction of IFN-β upon DNA stimulation, we overexpressed p204 to restore its expression (Supplemental Figure 3C–D). However, overexpression of p204 did not result in higher production of IFN-β in response to DNA stimulation (Supplemental Figure 3E), suggesting that differential levels of p204 between WT and ASC-deficient macrophages does not explain the lower production of IFN-β in WT cells stimulated with DNA.
We then tested if the enzymatic activity of caspase-1 was involved in this inhibitory mechanism. For this purpose, WT and ASC−/− macrophages were pre-incubated for one hour with the caspase-1 inhibitor Ac-Tyr-Val-Ala-Asp-chloromethylketone (Ac-YVAD-CMK), or with the caspase-3 inhibitor Z-Asp-Glu-Val-Asp-chloromethylketone (Z-DEVD-CMK) as a control, before stimulation with DNA. We found that WT macrophages pre-incubated with caspase-1 inhibitor showed increasedphosphorylation of TBK1 and IRF3, whereas ASC−/− macrophages failed to show any further increase in pTBK1 or pIRF3 (Figure 5A). Expression and secretion of IFN-β after DNA stimulation was also augmented upon caspase-1 inhibition in WT macrophages (Figure 5B–C). IFN-β production was also enhanced with WT BM-DCs pre-incubated with caspase-1 inhibitor (Figure 5D). Together, these data demonstrate that the enzymatic output from AIM2 inflammasome activation by DNA, namely caspase-1 activation, could explain most of the antagonistic effect on STING pathway activation.
Two potential mechanisms could explain how activation of the AIM2 inflammasome antagonize the activation of the STING pathway: the production of a factor processed by caspase-1 that could negatively regulate the activation of the STING pathway, or cell death via pyroptosis, resulting from the activation of the inflammasome that would limit the number of viable cells able to activate the STING pathway. In order to evaluate the presence of a soluble inhibitory factor, BM-DCs from STING−/− mice were stimulated with DNA, or only lipofectamine as a control, and the conditioned media was collected 24 hours later. STING−/− macrophages or BM-DCs do not produce IFN-β upon DNA stimulation (4) (Supplemental Figure 4A). Thus, the culture medium from stimulated STING−/− cells contains no IFN-β, but could contain a potential inhibitory factor. BM-DCs from Aim2−/− mice were stimulated with DNA in the presence or absence of conditioned media from STING−/− cells, and the amount of IFN-β was measured by ELISA. We found no difference in the production of IFN-β from Aim2−/− BM-DCs co-incubated with the different conditioned media (Figure 6A). To confirm this result, IRF3/7−/− macrophages, that do not produce IFN-β upon DNA stimulation (Supplemental figure 4B), were placed in the upper chamber of a 4 μm pore-size transwell; and Aim2−/− or ASC−/− macrophages were placed in the lower chamber of the transwell. The presence of IRF3/7−/− macrophages in the upper chamber did not reduce the production of IFN-β by the inflammasome-deficient cells in the lower chamber (Figure 6B). Together, these results argue against a mechanism mediated by an inflammasome-derived soluble factor.
Next, we evaluated the role of pyroptosis induced by inflammasome activation. Cell death was assessed by measuring the release of lactate dehydrogenase (LDH) 45 minutes after DNA stimulation in WT and inflammasome-deficient macrophages. DNA stimulation induced release of LDH in WT cells, however inflammasome-deficient macrophages showed minimal levels of LDH release (Figure 6C). Similar results were observed using BM-DCs from WT or inflammasome-deficient mice (Figure 6D). Pre-incubation of WT macrophages or BM-DCs with the caspase-1 inhibitor abolished the release of LDH after DNA stimulation (Figure 6E–F). Together with our observation that caspase-1 inhibition restored STING-dependent signaling and IFN-β production, these results support the notion that pyroptosis induced by activation of the Aim2 inflammasome upon DNA stimulation plays a significant role in controlling the activation of the STING pathway.
Our data provide clear evidence that concurrent activation of the AIM2 inflammasome and the STING pathway by cytosolic DNA in APCs leads to reduced IFN production. The STING pathway was hyper-activated in AIM2 inflammasome-deficient macrophages and DCs, and the activity of caspase 1 was found to contribute to this mechanism. Although it is well established that pro-IL1 and pro-IL-18 are major substrates for caspase-1, different bioinformatic approaches have been developed to predict caspases substrates (26). To date only are three substrates for caspase-1 proIL1β, proIL18 and gsdmd (27) and indeed, we found no evidence for any secreted soluble factor mediating inhibition of the STING pathway. We also questioned whether cGAS or STING could be substrates of caspase-1, but Western blot analysis showed no degradation products of cGAS or STING after DNA stimulation (data not shown). However, the other defined outcome of caspase-1 activation, cell death, revealed a clear difference between inflammasome-sufficient and -deficient cells. Thus, the AIM2 inflammasome activation by cytosolic DNA likely reduces the viability of cells, limiting the activation of other inflammatory pathways in the same APCs.
We have previously demonstrated that the activation of the STING pathway in the tumor microenvironment is critical for generating a spontaneous anti-tumor T cell response (4). Our current data suggest that a possible strategy for enhancing this response could be through targeted inhibition of the inflammasome. Indeed, it has been demonstrated that signaling though the IL1R inhibits expression of IFN-β (28). However, we need to consider other factors that could influence the generation of an efficient anti-tumor T cell response in vivo. First, signals from inflammasome-related cytokines, IL-1 and IL-18, might contribute to the generation of an anti-tumor immunity (29). Second, our studies focus on the consequences of concurrent activation of the AIM2 and cGAS/STING pathways within the same cell. It is conceivable that coordinated activation of these pathways in distinct cell types may occur in vivo, which could interact productively. In addition, it is also conceivable that this mechanism only occurs in some types of cells, such as DCs and macrophages that express both cGAS and AIM2, but may be absent in stromal cell in which AIM2 is not expressed or needs to be induced by type I IFN signaling (5, 20). Third, it was recently reported that Aim2-deficient animals may maintain a dysbiotic gut microbiota that accelerates tumorigenesis in an inflammatory model of colorectal cancer (30). This last consideration is important, as recent evidence has indicated that the composition of the gut microbiota modulates the systemic immune response against distant tumors (31, 32). These considerations imply that the in vivo effect of this regulatory mechanism might be more complex, and future work will be necessary to dissect the potentially more multifaceted interplay between the AIM2 inflammasome and the cGAS/STING pathway in vivo.
We thank Kate Fitzgerald for the inflammasome-deficient macrophages cells and Michael Leung and Ryan Duggan for technical assistance.
This work was supported by National Institutes of Health Grant R01CA181160. L.C. is supported by a Cancer Research Institute Irvington postdoctoral fellowship.
Competing financial interests
The authors have a patent application pending on the therapeutic potential for STING agonists in the cancer setting. S.M.M. and Thomas W.D. are paid employees of Aduro Biotech.