Inflammasomes are protein complexes that promote caspase activation, resulting in processing of IL-1β and cell death, in response to infection and cellular stresses. Inflammasomes have been anticipated to contribute to autoimmunity. The New Zealand Black (NZB) mouse develops anti-erythrocyte Abs and is a model of autoimmune hemolytic anemia. These mice also develop anti-nuclear Abs typical of lupus. In this article, we show that NZB macrophages have deficient inflammasome responses to a DNA virus and fungal infection. Absent in melanoma 2 (AIM2) inflammasome responses are compromised in NZB by high expression of the AIM 2 antagonist protein p202, and consequently NZB cells had low IL-1β output in response to both transfected DNA and mouse CMV infection. Surprisingly, we also found that a second inflammasome system, mediated by the NLR family, pyrin domain containing 3 (NLRP3) initiating protein, was completely lacking in NZB cells. This was due to a point mutation in an intron of the Nlrp3 gene in NZB mice, which generates a novel splice acceptor site. This leads to incorporation of a pseudoexon with a premature stop codon. The lack of full-length NLRP3 protein results in NZB being effectively null for Nlrp3, with no production of bioactive IL-1β in response to NLRP3 stimuli, including infection with Candida albicans. Thus, this autoimmune strain harbors two inflammasome deficiencies, mediated through quite distinct mechanisms. We hypothesize that the inflammasome deficiencies in NZB alter the interaction of the host with both microflora and pathogens, promoting prolonged production of cytokines that contribute to development of autoantibodies.
Inflammasomes are large multiprotein complexes assembled in the cell in response to environmental or infectious stress. Canonical inflammasomes act as platforms for the activation of caspase-1 and caspase-8 (1, 2). Subsequent to inflammasome activation, caspase-1 cleaves the precursors of the inflammatory cytokines IL-1β and IL-18 and also initiates pyroptotic cell death. Inflammasome-activated caspase-8 simultaneously initiates apoptotic death (2). Inflammasome complex formation depends on the oligomerization of an initiator protein, either a member of the Nod-like receptor family (such as NLR family, pyrin domain containing 3 [NLRP3] or NLR family, CARD domain containing 4 [NLRC4]) or absent in melanoma 2 (AIM2), which is a member of the pyrin and HIN domain containing (PYHIN) protein family (3). These clustered proteins recruit the adapter molecule apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) and then procaspases. Stimuli initiating NLRP3 inflammasome formation include extracellular ATP, the potassium ionophore nigericin from Streptomyces hygroscopicus, and a wide range of particulates such as monosodium urate crystals, cholesterol crystals, β-amyloid, and asbestos (1). These are not direct ligands for NLRP3, and their mode of action is still debated but may involve ion flux, lysosomal destabilization, and reactive oxygen species. NLRP3 appears to play a role in detection of metabolic disturbance, and there is great recent interest in this pathway due to its apparent role in diverse diseases such as atherosclerosis, Alzheimer’s disease, and type II diabetes (4–6). However, NLRP3 is also activated in infection with Staphylococcus aureus and Candida albicans (7), and it may be involved in gut homeostasis (8, 9). NLRC4 mediates responses to cytosolic bacterial flagellin and type III secretion system components and is important in combatting a number of bacterial infections including Salmonella (7). AIM2 binds to cytosolic DNA and initiates inflammasome responses in response to viruses such as mouse CMV (MCMV) and vaccinia, and the cytosolic bacterium Francisella tularensis (10). In addition, there is a noncanonical inflammasome pathway elicited by cytosolic LPS, leading to caspase-11–dependent pyroptosis (11, 12). Recent work showed that mouse caspase-11 and human caspase-4 and -5 are activated as a result of direct interaction with LPS (12) and are thus pattern recognition receptors with effector function.
Inflammasome function has been proposed to contribute to autoimmunity (13, 14). Circulating levels of IL-1β were elevated in Sjögren’s syndrome (15), but not in autoimmune hemolytic anemia (16). Reported levels of IL-1β in systemic lupus erythematosus (SLE) patient serum range from undetected (17, 18), decreased in SLE relative to controls (19, 20), to increased in a small percentage of patients (21). IL-1β levels may be higher in subsets of patients such as those with neuropsychiatric SLE (22) or vascular damage (20). The other cytokine produced by the inflammasome, IL-18, is elevated in the serum and glomeruli of SLE patients with lupus nephritis and is associated with localization of plasmacytoid dendritic cells to glomeruli (23, 24). Because both IL-1β and IL-18 can be produced by inflammasome-independent means (25, 26), inflammasome involvement in autoimmunity needs confirmation. Human genome-wide association studies have suggested the inflammasome initiator NLRP3 may play a role in type I diabetes in a Brazilian population (27), and one study showed lowered expression of NLRP3 was associated with Crohn’s disease (28), although this has not been replicated (29). Experiments in mice have suggested that the NLRP3 inflammasome is activated in kidney during lupus nephritis in two mouse models of SLE: NZB×NZW F1 (first-generation cross between New Zealand Black [NZB] and New Zealand White [NZW] mice) and MRL/Faslpr (Murphy Roths Large with lymphoproliferation lpr mutation of Fas) (30, 31) and may contribute to kidney damage. Lupus-like disease can be induced in nonsusceptible mice by injection of pristane, a mineral oil that induces cell death and elevated IL-1β for at least a month after injection (32). Caspase-1 was required for maximal autoantibody production in pristane-induced lupus (33). Thus, inflammasome function promoted disease in this model, but because elevated IL-1β is not typical of human lupus, the relevance of this to the initiation of the majority of SLE is uncertain. In contrast, deletion of NLRP3 or ASC from a mild spontaneous model of SLE (Faslpr mutation on C57BL/6 background) causes exacerbated disease (34). Thus, the role of inflammasomes in lupus-like autoimmunity remains uncertain.
We have investigated inflammasome activity in the NZB mouse. Female NZB mice develop anti-nuclear Abs similar to SLE patients and also anti-erythrocyte Abs (35). They die around 1 y of age, due to hemolytic anemia, and are used as a model for human autoimmune hemolytic anemia as well as SLE (36). A more aggressive model of SLE is provided by the first-generation offspring of the cross between NZB and NZW mice. This is a genetically complex model with contributions of multiple loci from both strains (37, 38). Although NZW mice themselves do not have overt disease, they contribute a number of loci that exacerbate disease, and the NZB×NZW F1 animals die at 10–12 mo with severe lupus nephritis (39, 40). In comparison, the NZB mice display mild kidney pathology. Our previous work showed that NZB mouse macrophages are deficient in cell death mediated by AIM2 inflammasome recognition of cytosolic DNA (41, 42). AIM2-dependent cell death was limited by high expression in NZB cells of the related PYHIN family member p202 (42), which has been suggested as a lupus-susceptibility factor (43). p202 binds directly to AIM2 and prevents downstream ASC oligomerization (42). In this article, we extend this observation to show defects in the AIM2 response to a viral pathogen, and surprisingly, a complete lack of NLRP3 inflammasome function in the NZB strain. Thus, in contrast with the concept of an important role for the inflammasome in autoimmunity, here we show substantial inflammasome deficiencies in cells from NZB mice. This shows that effective inflammasome function is not required for loss of tolerance and suggests that innate immune imbalance could be one factor that contributes to development of autoantibodies.
Materials and Methods
Recombinant human CSF1 was a gift from Chiron. LPS from Salmonella minnesota Re595 (Sigma Aldrich) was dissolved in PBS/0.1% triethylamine at 10 mg/ml. Calf thymus DNA (CT DNA; Sigma Aldrich) was further purified as previously described (44).
Mice and cell culture
C57BL/6, Asc−/− (45), Casp1−/−Casp11null (11, 46), Nlrp3−/− (47), Nlrc4−/− (45), and NZB mice were housed in specific pathogen-free facilities at the University of Queensland. Aim2−/− bone marrow was obtained from animals gene-targeted on a C57BL/6 background and housed at The University of Bonn (V. Hornung, unpublished data). Mice were used under approval from The University of Queensland and University of Bonn Animal Ethics Committees. Bone marrow–derived macrophages (BMMs) were cultivated from female bone marrow in “complete RPMI 1640” (RPMI 1640, 25 mM HEPES, 1X GlutaMAX, 50 U/ml penicillin and 50 μg/ml streptomycin, 10% heat-inactivated FCS; all Life Technologies) supplemented with 104 U/ml CSF1. BMMs were used between days 7 and 11 of culture.
Small interfering RNA gene knockdown
Gene knockdown was conducted using Stealth small interfering RNAs (siRNAs; Life Technologies) targeting p202 or TLR9 as a control, as previously reported (41). BMMs (days 6–7) were electroporated with 600 nM siRNA duplexes in 400 μl complete RPMI 1640 at 260 V, 1000 μF using a Gene Puler MX (Bio-Rad), and then cultured for a further 3–4 d in complete RPMI 1640 supplemented with 104 U/ml CSF1.
Electroporation with CT DNA
A total of 500,000 BMMs were electroporated in 400 μl complete RPMI 1640 with 10 μg CT DNA at 240 V, 1000 μF. Cells were immediately washed at room temperature in 10 ml additive-free RPMI 1640 and then resuspended in 350 μl complete RPMI 1640 lacking heat-inactivated FCS, and incubated for a further 35 min at 37°C before cell culture supernatants and cell lysates were processed for immunoblot analysis.
MCMV BMM infection assays
BMMs were plated at 350,000 cells/well (24-well plate) and cultured overnight in 500 μl antibiotic-free complete RPMI 1640 supplemented with 104 U/ml CSF1. The following day, cells were primed for 1 h with 10 ng/ml LPS, after which time medium was replaced with 180 μl serum- and antibiotic-free complete RPMI 1640 supplemented with 10 ng/ml LPS and 0.5 × 104 U/ml CSF1. MCMV (K181) (48) was used to infect BMM monolayers at the indicated multiplicity of infection (MOI), and after 1-h culture, volumes were increased to 350 μl with serum- and antibiotic-free complete RPMI 1640 supplemented with 10 ng/ml LPS and 0.5 × 104 U/ml CSF1. After a further 5 h of incubation at 37°C, cell culture supernatants and cell lysates were processed for immunoblot analysis or ELISA.
Candida albicans BMM infection assays
BMMs were plated at 350,000 cells/well (24-well plate) and cultured overnight in 500 μl antibiotic-free complete RPMI 1640 supplemented with 104 U/ml CSF1. The following day, cells were primed for 4 h in antibiotic-free complete RPMI 1640 with 10 ng/ml LPS, at which time medium was replaced with 300 μl serum- and antibiotic-free complete RPMI 1640 supplemented with 10 ng/ml LPS and 0.5 × 104 U/ml CSF1. Candida albicans strain SC5314 prepared from overnight growth at 30°C on yeast extract peptone dextrose agar was used to infect BMM monolayers at the indicated MOI making culture volumes to 350 μl. After a further 6 h of incubation at 37°C, cell culture supernatants and cell lysates were processed for immunoblot analysis or cell culture supernatants were collected and processed for ELISA analysis.
Cell supernatants were collected, cell debris was pelleted at 500 × g for 5 min, and 300 μl supernatant was collected and proteins were precipitated by addition of 4 vol 100% acetone with centrifugation at 17,000 × g. Cell monolayers were directly lysed in 66 mM Tris pH 7.4, 2% SDS and combined with any debris pelleted in the initial 500 × g centrifugation. Immunoblot analysis for caspase processing was conducted using NuPAGE 4–12% gradient Bis-Tris precast gels (Invitrogen), whereas expression-level analysis used 5–20% gradient mini-PROTEAN TGX gels (Bio-Rad). Proteins were transferred to polyvinylidene difluoride (Millipore) using a mini-trans blot system (Bio-Rad) with Tris-Glycine transfer buffer containing 10–20% methanol. Immunoblot analysis was conducted as previously described (42). Primary Abs used included rabbit polyclonals against caspase-1 p10 (M-20; Santa Cruz Biotechnology), ASC (N-15; Santa Cruz Biotechnology), full-length caspase-3 (9662; Cell Signaling Technologies), cleaved caspase-3 (9661; Cell Signaling Technologies) and AIM2 (made to full-length recombinant mouse AIM2 and tested on Aim2−/− extracts), rabbit monoclonal against S6 ribosomal protein (5G10; Cell Signaling Technologies), goat polyclonals against mouse IL-1β (AF-401-NA; R&D Systems) and p202 (S-19; Santa Cruz Biotechnology), mouse monoclonals against NLRP3 (Cryo-2; Adipogen), caspase-1 p20 (Casper-1; Adipogen), S6 ribosomal protein (54D2; Cell Signaling Technologies) and α-Tubulin (B-5-1-2; Sigma-Aldrich). S6 ribosomal protein and tubulin were used as interchangeable loading controls depending on both the size of previously detected proteins on membranes and the species of primary Abs used to detect them. Secondary Abs used included anti-rabbit IgG-HRP Ab (7074; Cell Signaling Technologies), anti-goat IgG-HRP Ab (HAF019; R&D Systems), and anti-mouse IgG-HRP Ab (7074; Cell Signaling Technologies). Sequential reprobing of blots was performed after inactivation of HRP with 0.1% NaN3 where possible, followed by extensive washing. When not possible, membranes were “stripped” by two 15-min washes with agitation at 50°C in 63 mM Tris pH 6.7, 2% SDS, 0.8% 2-ME, followed by extensive washing.
ELISA analysis of IL-1β
Cell culture supernatants were collected, cell debris was pelleted at 500 × g for 5 min, and 300 μl supernatant was collected and further cleared by centrifugation at 10,000 × g for 2 min at which time supernatants were aliquoted and stored at −80°C for further analysis. IL-1β levels in cell culture supernatants were determined using the Mouse IL-1β/IL1F2 DuoSet ELISA kit (DY401; R&D Systems).
mRNA and microRNA quantitative RT-PCR analysis
RNA preparation, cDNA synthesis, and quantitative RT-PCR were conducted as outlined previously (3). Primers targeting murine genes included Nlrp3-forward 5′-GCT CCA ACC ATT CTC TGA CC-3′, Nlrp3-reverse 5′-AAG TAA GGC CGG AAT TCA CC-3′, Hprt-forward 5′-CAG TCC CAG CGT CGT GAT TAG-3′ and Hprt-reverse 5′-AAA CAC TTT TTC CAA ATC CTC GG-3′. Although the Nlrp3 primers are in exons 7 and 8 and flank the pseudoexon, control reactions showed that they amplified cloned C57BL/6 and NZB Nlrp3 cDNA equivalently. Primers targeting human genes included TATA binding protein (TBP)-forward 5′-GCT GGC CCA TAG TGA TCT TT-3′ and TBP-reverse 5′-CTT CAC ACG CCA AGA AAC AGT-3′. Preparation of RNA for microRNA (miR) analysis was conducted using an miRNeasy Mini Kit (Qiagen), and miR-223 and RNU6B levels were determined as previously described (49).
Analysis of nlrp3 mRNA variants
PCR was conducted on cDNA generated from C57BL/6 and NZB BMMs using the forward primer 2706 (5′-AGAAACTGTGGTTGGTGAG-3′) and reverse primer 3125 (5′-TGTGGTTGTGGGTCAGAA-3′) that after an initial denaturation of 95°C for 3 min used 35 cycles of 94°C/30 s, 60°C/30 s, 72°C/60 s and a final extension of 10 min at 72°C. Products were visualized after electrophoresis on a 2% agarose gel.
Construction of NLRP3 and NLRP3′ expression constructs
Expression of NLRP3 variants was achieved using HEK293 cells and Lipofectamine 2000 (Invitrogen) for transfection. C57BL/6 N-terminal Flag-tagged coding sequence (CDS) was engineered to include the exon 7b sequences, to direct expression of the truncated form of NLRP3 predominantly present in NZB (NLRP3′).
Assessment of statistical significance of data generated from ELISA analysis was conducted using a one-tailed paired ratio t test in Prism 6.0.
NZB macrophages are deficient in AIM2 inflammasome production of IL-1β
In previous work we showed that NZB macrophages have low AIM2-initiated caspase cleavage and resulting cell death, because of high expression of p202, an antagonist of AIM2 (42). In addition to cell death, another output of inflammasome function is pro–IL-1β cleavage, which requires priming of cells with a TLR signal to induce pro–IL-1β. Our published work defining the mechanism of p202 inhibition of AIM2 inflammasome function did not examine responses in cells primed with TLR stimuli (42). It cannot be assumed that under conditions of TLR-mediated priming, NZB cells will maintain deficient AIM2 function, because the ratio between AIM2 and its antagonist p202 may change. Comparison of the expression of components of the AIM2 inflammasome pathway in NZB and C57BL/6 cells with and without LPS treatment (Fig. 1A) revealed that AIM2, ASC, and procaspase-1 were expressed identically in the two strains, and pro–IL-1β was similarly induced. The only difference between the strains was in p202 expression, which was undetectable in C57BL/6 but readily observed in NZB and induced further by LPS. In some experiments, a minor induction of AIM2 by LPS was observed, but this was always less than the effect on p202. Thus, if anything, under the conditions of LPS priming in this study, the inhibitory role of p202 would be expected to be reinforced. Indeed, electroporation of LPS-primed NZB and C57BL/6 BMMs with CT DNA revealed that NZB cells are deficient in processing and release of IL-1β (Fig. 1B). Similarly, NZB BMMs showed minimal release of processed caspase-1 under conditions where the C57BL/6 BMM response was robust (Fig. 1B). These data confirm that AIM2 inflammasome activity remains low in NZB cells after LPS priming. To confirm the role of p202 in limiting AIM2-induced IL-1β cleavage, we knocked down p202 with two independent siRNAs (41, 42), which restored IL-1β processing toward levels observed in C57BL/6 cells (Fig. 1C, 1D). Cleavage of caspase-1 to p10 and p20 fragments was also increased by p202 knockdown, as demonstrated previously in BMMs without LPS priming (41, 42). We have earlier established that caspase-8 and caspase-3 are activated by the inflammasome in parallel with caspase-1 (2), and increased cleavage of caspase-3 further confirms inflammasome restoration with p202 knockdown (Fig. 1C). Thus, similar to its role in dampening the AIM2 inflammasome-mediated processing of caspases and cell death (41, 42), p202 also limits AIM2-dependent release of IL-1β from LPS-primed NZB macrophages.
Deficient AIM2 response to MCMV in NZB macrophages
AIM2 responses are important in combatting cytosolic bacteria such as Francisella tularensis (50) and DNA viruses such as MCMV and vaccinia virus (10). To determine whether NZB cells are deficient in responses to MCMV, we primed BMMs with LPS to induce pro–IL-1β and then infected them with MCMV. MCMV caused far less cleavage of caspases in NZB than C57BL/6 BMMs (Fig. 2A). Processing and release of IL-1β was also much lower in NZB BMMs (Fig. 2A, 2B). Interestingly, both strains displayed a lower level of pro–IL-1β protein in cell lysates postinfection, which at least for NZB was not explained by IL-1β cleavage and release, suggesting that the viral infection either hinders pro–IL-1β induction by LPS priming or promotes its degradation. To confirm the AIM2 dependence of inflammasome activation by MCMV, we infected BMMs from a number of knockout mouse strains. This showed that MCMV detection was dependent on the dsDNA sensor AIM2 and inflammasome adapter ASC, and not on NLRP3 or NLRC4 (Fig. 2C, 2D), as expected (10).
No NLRP3 inflammasome response to Candida albicans in NZB BMMs
Because the basal elements of inflammasome function (ASC, caspase-1, pro–IL-1β) were expressed similarly in C57BL/6 and NZB mice (Fig. 1A), we thought inflammasomes other than AIM2 were likely to work efficiently in NZB cells. We investigated the activation of the NLRP3 inflammasome as a control inflammasome, to demonstrate the specificity of the defect in AIM2 function. NLRP3 was activated by treatment of cells with LPS followed by addition of ATP or nigericin. To our surprise, no NLRP3 inflammasome activity was evident in NZB cells, whereas caspase and pro–IL-1β cleavage in C57BL/6 cells was robust (Fig. 3A). To investigate this further, we infected cells with the opportunistic fungal pathogen Candida albicans, which activates the NLRP3 inflammasome (51). Exposure of BMMs from C57BL/6 mice to C. albicans resulted in NLRP3-dependent cleavage of caspases and IL-1β, with a complete absence of activity in NZB BMMs (Fig. 3B). Failure of release of IL-1β by cells from both NZB and Nlrp3−/− mice in response to C. albicans infection was confirmed by ELISA (Fig. 3C).
NLRP3 protein is absent from NZB cells
In investigating the absence of NLRP3 responses, we found a lack of NLRP3 protein in NZB cells, whereas C57BL/6 BMMs contained readily detected levels of full-length protein, which was further induced with LPS priming (Fig. 4A). Overexposure of this blot did reveal a very low level of full-length NLRP3 protein in NZB and a smaller species we have termed NLRP3′ (Fig. 4A). Evidently this low-level expression was not sufficient to allow responses, and indeed TLR-mediated induction of NLRP3 to a critical level is thought to be essential for its activity (52). Examination of lysates prepared from erythrocyte-depleted bone marrow confirmed that the lack of NLRP3 expression in the NZB mouse strain was not restricted to in vitro–differentiated BMMs (Fig. 4B). Members of the PYHIN family to which p202 and AIM2 belong have been implicated in regulation of gene expression, so we wished to exclude that p202 expression in NZB BMMs was responsible for the lack of NLRP3. p202 knockdown did not restore expression of NLRP3 (Fig. 4C). We next examined Nlrp3 mRNA levels. The mRNA was reduced ∼6-fold in NZB compared with C57BL/6 (Fig. 4D), a ratio maintained after LPS treatment that induced Nlrp3 mRNA ∼9-fold over the basal level in both strains. Despite levels of Nlrp3 mRNA in LPS-primed NZB BMMs being similar to levels observed in unprimed C57BL/6 BMMs (Fig. 4D), the protein levels were not comparable (Fig. 4A). Hence the reduced levels of mRNA did not fully account for the lack of NLRP3 protein in NZB BMMs. We considered whether microRNA miR-223, which has previously been shown to be a posttranscriptional regulator of NLRP3 levels (49, 53), could be involved. However, miR-223 was not differentially expressed in NZB and C57BL/6 (Fig. 4E), suggesting it was not contributing to the divergent NLRP3 expression between these two mouse strains.
A point mutation causes aberrant splicing of the NZB NLRP3 mRNA
Sequencing of the Nlrp3 cDNAs from C57BL/6 and NZB mice identified that the NZB mRNA contained an aberrantly spliced exon, denoted in this article as 7b, between exons 7 and 8 of the C57BL/6 reference sequence (NM_145827.3; Fig. 5A). Exon 7b introduces four new amino acids followed by a stop codon. This encodes a protein product of 103 kDa consistent with the smaller NLRP3′ species observed in the overexposed blot (Fig. 4A). Apart from this, there were no other differences in the NLRP3 amino acid sequence between NZB and C57BL/6. Premature termination codons upstream of exon splice boundaries are known to trigger nonsense-mediated decay of transcripts (54), which is congruent with the reduced levels of NZB Nlrp3 mRNA observed in Fig. 4D. PCR primers were designed across the exon 6/7 boundary and within exon 9 of the C57BL/6 mRNA, expected to amplify a 420-bp product in C57BL/6 and a 516-bp product with inclusion of exon 7b in NZB, to assess the prevalence of the variant within the total population of NZB Nlrp3 mRNAs (Fig. 5A). Indeed, although a 420-bp product was observed in C57BL/6, the majority of product observed from NZB mice was the larger 516-bp product (Fig. 5B). A minor amount of 420-bp product observed in NZB represents correctly spliced mRNA that creates the small amount of full-length NLRP3 protein (Fig. 4A). However, as noted earlier, the level of the truncated NZB NLRP3′ protein (Fig. 4A) was much lower than expected from a 6-fold lowering of Nlrp3 mRNA in NZB mice (Fig. 4D). Also, despite the high ratio of abnormal Nlrp3 mRNA to normally spliced mRNA in NZB mice (Fig. 5B), there were similar levels of the two protein forms (NLRP3 and NLRP3′) within the LPS-primed NZB sample (Fig. 4A). Consequently we investigated whether the truncated protein is unstable. Normal and truncated proteins were transiently expressed in HEK293 cells under control of the same promoter. Given that these were encoded by cDNA with no introns, the mRNA encoding the truncated NLRP3′ protein was not subject to nonsense-mediated decay (54), and there was similar mRNA expression from the two constructs (Fig. 5C). Despite this, the truncated NLRP3′ protein was present at somewhat lower levels in transfected cells than the normal protein, and this is particularly evident comparing transfections with 200 ng plasmid (Fig. 5D). This is consistent with reduced stability of the truncated form contributing to its low expression in NZB BMMs, in addition to the lowered mRNA level. To determine why NZB Nlrp3 mRNA was inappropriately spliced, we sequenced the genomic DNA at the exon–intron boundaries of exon 7b in NZB. This identified a G-to-A transition in NZB, creating a novel splice acceptor site immediately before exon 7b (Fig. 5E), confirming exon 7b to be a pseudoexon. In summary, both nonsense-mediated decay and an unstable form of NLRP3 result from the introduction of a pseudoexon in NZB mice, leading to a state of NLRP3 inflammasome unresponsiveness.
This article defines a point mutation in the Nlrp3 gene of NZB mice leading to complete lack of NLRP3 inflammasome function and also confirms the low activity of the AIM2 inflammasome in these mice. Although inflammasomes have been proposed to contribute to autoimmunity (13, 14), effective AIM2 or NLRP3 activity is clearly not necessary for the autoantibodies and autoimmune hemolytic anemia that develop in the NZB mouse. The AIM2 and NLRP3 defects led, respectively, to a low inflammasome response to MCMV and complete lack of response to C. albicans. This may contribute to the published susceptibility of NZB mice to C. albicans and MCMV (55, 56). The two inflammasome deficiencies in NZB mice arise from completely unrelated mechanisms. Our prior work demonstrated that p202, which is highly expressed in NZB, is an antagonist of AIM2 (41, 42). p202 and AIM2 are both members of the PYHIN family with DNA-binding HIN domains, but p202 lacks the pyrin domain involved in recruitment of ASC (3). p202 forms a tetramer that directly binds two molecules of AIM2, and prevents AIM2 from initiating ASC oligomerization (42). In this study, we confirmed that p202 is responsible for the low inflammasome response to MCMV in NZB cells, but has no role in the NLRP3 inflammasome deficiency. Instead, an intronic point mutation in NZB Nlrp3 creates a novel splice acceptor site, and subsequent incorporation of a pseudoexon with a premature termination codon. The observed lowered level of NZB Nlrp3 mRNA is likely due to nonsense-mediated decay, which is elicited by a stop codon upstream of an exon–exon boundary, marked by an exon junction complex (54). However, the truncated protein also apparently lacks stability, and the normal protein is not expressed at a high enough level in NZB to provide an NLRP3 response. To our knowledge, this is the first identification of a naturally occurring mutation of NLRP3 resulting in a loss of function, in either mice or humans.
Whether the low inflammasome function actually plays a role in promoting or modifying NZB autoimmunity is an open question. Interestingly, recent work on another lupus model demonstrated that NLRP3 protected against disease development (34). Contrary to their hypothesis that NLRP3 would contribute to disease in C57BL/6 mice with a Faslpr mutation, the authors found exacerbated pathology upon deletion of the genes for either NLRP3 or ASC. In this model, NLRP3 or ASC deficiency increased lymphoproliferation, spleen size, and kidney immune infiltrate, and this was associated with elevated expression of cytokines such as IL-6 and IFN-γ, as well as IFN-responsive genes. Notably, loss of NLRP3 or ASC alone on the C57BL/6 background was not sufficient to induce kidney pathology at 12 mo of age or any autoantibodies by 6 mo (34). The suppressive role of NLRP3 was not explained by IL-1β or IL-18 production, and instead C57BL/6 Faslpr Nlrp3−/− dendritic cells were hyporesponsive to TGF-β (34), as previously described in kidney epithelial cells (57). The means by which NLRP3 contributes to TGF-β signaling has not been established.
The role of NLRP3 in autoimmunity may be dependent on the genetic context, and results from the C57BL/6 Faslpr model may not reflect all possible outcomes of inflammasome deficiency. Inflammasomes are involved in host defense against a range of pathogens and possibly homeostatic interactions with microflora. A consequence of inflammasome deficiency would be the maintained viability of cells harboring intracellular infections, when they would normally undergo pyroptotic death (58). Increased cell viability would allow prolonged production of proinflammatory cytokines and IFN. We have shown this effect in vitro, where NZB macrophages release between 4- and 10-fold more IFN-β in response to cytosolic DNA than C57BL/6, correlating with macrophage survival (42). In addition, deficient inflammasome IL-1β output at sites of infection may contribute to ineffective pathogen control. Inflammasome-deficient mice may also have an altered interaction with their microflora. Inflammasome components are abundantly expressed in the gut, and NLRP6 and NLRP3 are suggested to play a role in homeostasis of the mucosal environment (9, 59). Alteration in the intestinal microbiome has been observed in NLRP3 knockout mice (8) and would be anticipated in NZB mice, which are functionally null for Nlrp3. Consistent with a role for the microbiota in NZB autoimmunity, germ-free NZB mice had delayed and greatly reduced splenomegaly (60), a feature reportedly related to autoantibody-dependent red cell sequestration in the NZB spleen (36).
Despite having anti-nuclear Abs typical of SLE, the NZB mice develop only very mild glomerulonephritis compared with the NZB×NZW F1 mice (39, 40). We have examined inflammasome status in the NZB×NZW F1 macrophages; p202 levels are high enough to maintain relatively low AIM2 function, and NLRP3 protein levels are ∼50% of NZW. An effective NLRP3 response occurs if the priming stimulus is strong enough, but with lower levels of priming for induction of NLRP3, the inflammasome response is suboptimal (S.J. Thygesen and D.P. Sester, manuscript in preparation). Earlier published work implicated NLRP3 in kidney damage in various scenarios, including lupus (30, 31, 61), and from this work it would be reasonable to propose that the lack of NLRP3 in NZB protects them from kidney pathology seen in the NZB×NZW F1 mice. The contradictory results of Lech et al. (34) suggest that such assumptions on the role of NLRP3 require testing in this model. Interestingly, type I IFN, which is commonly chronically elevated in SLE, antagonizes NLRP3 inflammasome function (62). Thus, even where there is no intrinsic inflammasome deficiency, disease progression and IFN production may lead to a functional deficiency.
Direct evidence for a role for inflammasome deficiency in NZB mouse disease will require genetic manipulation to restore inflammasome function. p202 has been suggested to be a mouse lupus susceptibility factor (43, 63), although this remains to be proved. The gene encoding p202, Ifi202, has variable copy number in different strains (3), all located within the Nba2 lupus-susceptibility locus at 1q22, and p202 is highly expressed in NZB and other lupus-prone mouse strains (43, 63, 64). The Nba2 congenic strain has this region of chromosome 1 from NZB backcrossed onto C57BL/6. Dissection of Nba2 into smaller subregions suggested a major role for other loci such as genes for signaling lymphocytic activation molecule factors and FcγRIIb, with little or no role for Ifi202 (65). However, it appears that the high expression of p202 seen in NZB may not be fully maintained in the Nba2 congenic (43), and if there are epistatic effects of other NZB loci on Ifi202 expression, the Nba2 congenic is not a suitable system to investigate the role of p202. Regarding the Nlrp3 gene, it is notable that this is located within the NZB lbw8 lupus susceptibility locus defined in NZB×NZW F2 intercrosses (66). Whether Nlrp3 contributes to the lbw8 locus remains to be established.
Data on inflammasome function in human SLE are currently limited. One recent article suggested that neutrophil DNA NETs and the antimicrobial peptide LL37 activate the NLRP3 inflammasome, and this response is elevated in SLE patient macrophages (67), although the observed difference in IL-1β production between SLE and healthy control was seen with LPS priming alone. In contrast, two reports investigating IL-1β production by monocytes from SLE patients showed a subset of SLE patients with very limited NLRP3 inflammasome response to extracellular ATP (68, 69).
There is precedent for proposing that appropriate innate immune responses are necessary for preventing SLE. One genetically identified SLE susceptibility factor is neutrophil cytosolic factor 2 (also known as p67phox), a component of NADPH oxidase that generates reactive oxygen species essential for microbial killing by neutrophils and macrophages (70). The SLE-associated variant showed 2-fold decrease in reactive oxygen species production, and it is possible that a deficient bacterial killing could predispose to SLE. In this case, genetic and environmental factors leading to autoimmunity would not be acting entirely independently. Infections are frequently invoked as part of the environmental risk for autoimmune diseases, and genetic defects in innate immune pathways in patients could mean that it is not just an infection per se that triggers autoimmunity, but an infection that is inappropriately dealt with. In summary, we find a profound deficiency in inflammasome systems in the autoimmune NZB mouse. Establishing the relevance of this to disease will require both genetic manipulation of the mouse and further exploration of these responses in patients.
We thank Larissa Labzin for preparation of Aim2−/− bone marrow.
This work was supported by National Health and Medical Research Council (NHMRC) Grants 1010887 and 1050651. K.J.S. was supported by Australian Research Council (ARC) and NHMRC Fellowships FT0991576 and 1059729. K.S. and M.J.S. were supported by ARC Fellowships FT130100361 and FT100100657. M.A.D.-E. was supported by a NHMRC research fellowship and NHMRC Program Grant 569938. V.H. was supported by the European Research Council (Grant 243046).
Abbreviations used in this article:
absent in melanoma 2
apoptosis-associated speck-like protein containing a caspase recruitment domain
bone marrow–derived macrophage
- CT DNA
calf thymus DNA
multiplicity of infection
NLR family, CARD domain containing 4
NLR family, pyrin domain containing 3
New Zealand Black
New Zealand White
pyrin and HIN domain containing
small interfering RNA
systemic lupus erythematosus
TATA binding protein.
The authors have no financial conflicts of interest.