AIM2 Drives Joint Inflammation in a Self-DNA Triggered Model of Chronic Polyarthritis

Mice lacking DNase II display a polyarthritis-like disease phenotype that is driven by translocation of self-DNA into the cytoplasm of phagocytic cells, where it is sensed by pattern recognition receptors. While pro-inflammatory gene expression is non-redundantly linked to the presence of STING in these mice, the contribution of the inflammasome pathway has not been explored. To this end, we studied the role of the DNA-sensing inflammasome receptor AIM2 in this self-DNA driven disease model. Arthritis-prone mice lacking AIM2 displayed strongly decreased signs of joint inflammation and associated histopathological findings. This was paralleled with a reduction of caspase-1 activation and pro-inflammatory cytokine production in diseased joints. Interestingly, systemic signs of inflammation that are associated with the lack of DNase II were not dependent on AIM2. Taken together, these data suggest a tissue-specific role for the AIM2 inflammasome as a sensor for endogenous DNA species in the course of a ligand-dependent autoinflammatory condition.


Introduction
The innate immune system has evolved a conserved set of so-called pattern recognition receptors (PRRs) to sense the presence of microbial pathogens. PRRs detect microbe-associated molecular patterns (MAMPs) as non-self and initiate signaling cascades geared at eliminating the microbial threat [1]. Under certain circumstances, these receptors can also respond to selfderived molecules. This, for example, includes endogenous nucleic acids that have gained access to compartments that are usually devoid of these [2]. Accidental activation of compartmentalized PRRs by endogenous nucleic acids is additionally prevented by the presence of nucleases that degrade and thereby deplete potential self-ligands under steady state conditions. These nucleases are expressed in a compartment and cell-type specific manner, creating a nonredundant system of nuclease activity, which prevents accidental activation of nucleic acid sensing PRRs. The importance of this safeguard system is impressively documented by the fact that defects in a number of these nucleases can result in severe sterile inflammatory conditions that are triggered by nucleic acid sensing PRRs [3]. Aicardi-Goutières syndrome (AGS), for example, is a rare genetic inflammatory disorder, in which defects in nucleic acid degrading or metabolizing enzymes can result in the spontaneous production of antiviral cytokines in cerebrospinal fluid and serum, manifesting as a sub-acute encephalopathy [4,5].
Detection of cytosolic DNA triggers at least two distinct core signaling cascades: On the one hand, cytosolic DNA leads to the activation of the nucleotidyltransferase cGAS, which upon DNA binding produces a 2'-5' linked cyclic dinucleotide second messenger molecule that in turn binds to and activates the ER-resident receptor STING. STING activation results in its translocation to a perinuclear Golgi compartment, where it achieves its signaling competent state, culminating in the activation of antiviral and pro-inflammatory gene expression [6,7]. While these two components form the non-redundant core of cytosolic DNA mediated antiviral gene expression, additional factors have been described to function in concert with cGAS and or STING [8]. In myeloid cells, cytosolic DNA is additionally sensed by the PYHIN protein AIM2, which forms an ASC-dependent inflammasome complex upon ligand binding. Inflammasome activation results in the processing and activation of pro-caspase-1, which itself leads to the processing of pro-cytokines such as IL-1β and IL-18 [9][10][11][12]. At the same time, inflammasome activation leads to the induction of a myeloid cell specific cell death, known as pyroptosis [13]. AIM2 has shown to be involved in the recognition of a number of microbial pathogens, such as DNA viruses or bacteria that release microbial DNA into the cytoplasm during their life cycle. However, a role for AIM2 in the recognition of endogenous self-DNA has not been established so far [14].
To study the possible role of AIM2 in the context of a self-DNA driven autoinflammatory disease, we made use of a DNase II-deficiency mouse model [15]. In this mouse model, absence of the lysosomal endonuclease DNase II (Dnase2) leads to a defect in disposing of DNA in lysosomal compartments [16]. This subsequently results in the translocation of undigested DNA into the cytoplasm, which in turn results in the activation of STING leading to unabated production of type I IFNs and pro-inflammatory cytokines [17]. Spontaneous cytokine production is already seen in early fetal development of Dnase2-deficient mice, due to the fact that macrophages play an important role in disposing of expelled nuclei from erythroid precursor cells in so-called erythroblastic islands in the fetal liver or bone marrow in the course of definitive erythropoiesis [16,18]. This macrophage-dependent type I IFN production in the context of Dnase2-deficiency results in lethal anemia, which can be fully rescued by ablating type I IFN production or its activity (e.g. Ifnar1 -/-) in these mice [19]. While Dnase2-Ifnar1 DKO mice are born healthy, macrophages in these mice are still responding to undigested DNA by the up regulation of pro-inflammatory genes. In fact, 3-6 months after birth Dnase2 -/-Ifnar1 -/mice develop a systemic auto-inflammatory disease, most prominently affecting the joints with chronic polyarthritis mimicking rheumatoid arthritis in humans [15]. Even though accompanied by the production of autoantibodies, this chronic polyarthritis is independent of the adaptive branch of the immune system, yet rather driven by the unabated production of proinflammatory cytokines, most prominently TNF, IL-6 and IL-1β [20]. In fact, in diseased joints of DNase2-deficient animals these cytokines appear to regulate each other's expression in a mutually dependent fashion. To this effect, blocking the activity of one of these cytokines blunts the expression of the others and thereby greatly ameliorates disease progression [20]. Il-18, another potential candidate cytokine that has been associated with the development and progression of arthritis, is not involved in this disease model. In fact, IL-18 -/mice still develop polyarthritis and express pro-inflammatory cytokines in the context of Dnase2-deficiency [20].
The inflammatory response in the context of defective disposal of erythroid precursors by macrophages in erythroblastic islands is an example of how exogenous self-DNA triggers the activation of innate sensing pathways in phagocytic cells. A recent study, however, has demonstrated that DNase2-deficiency also triggers cell-autonomous immune responses in cells that are not phagocytic [21]. In fact, these studies show that damaged DNA expelled from nuclei is normally subjected to DNase II mediated degradation, involving a process that depends on nuclear export and subsequent autophagy-dependent delivery to lysosomal structures. As such, next to its function in degrading apoptotic material from other cells, DNase II prevents the initiation of a cell autonomous innate immune response by disposing of nuclear-derived DNA.
Given the overlapping ligand spectrum of the cGAS-STING axis as well as the inflammasome sensor AIM2, we hypothesized that AIM2 might also be activated by the presence of cytosolic DNA in the course of Dnase2-deficiency. Moreover, we speculated that part of the disease activity in this mouse model could indeed be attributed to AIM2.

Clinical score
Swelling of the fore-and hind pads was analyzed in a blinded fashion, in monthly intervals and scored as follows: 0, no swelling; 1, mild swelling; 2, severe swelling. The scores were summed, and a total score (maximum 8) was assigned to each mouse [15].

RNA and protein isolation
Limbs of sacrificed mice were skinned and ground in liquid nitrogen and either transferred to TRIzol (Life technologies; Carlsbad, CA) for mRNA isolation or to tissue lysis buffer (T-PER, Tissue Protein Extraction Reagent, Pierce combined with Protease Inhibitor Tablets; Pierce/ Thermo Scientific; Pittsburgh, PA) for protein extraction.

Immunoblotting
Protein samples were diluted in 2x Laemmli buffer and boiled at 95°C for 5 min. Samples were separated by SDS gel electrophoresis for 2 h and further transferred to a 0.2 μM Nitrocellulose membrane [23]. As indicated, blots were incubated with rabbit polyclonal antibody to anti caspase-1 p10 (sc-514; Santa Cruz Biotechnology; Santa Cruz, LA), rabbit monoclonal anti MMP3 (ab52915; Abcam; Cambridge, UK) and mouse monoclonal anti Viperin (MABF106; Merck Millipore; Billerica, MA). Membranes were developed using ECL Western blotting substrate (Pierce/Thermo Scientific; Pittsburgh, PA)

Quantitative real time PCR (qPCR)
Relative gene expression is shown as a ratio of the expression level of the gene of interest to that of Hypoxanthine-guanine phosphoribosyltransferase (Hprt1). The following primer sets were used:

Multiplex protein analysis
Serum cytokines and cytokine expression in the joints were determined by using customized V-PLEX Proinflammatory Panel 1 (Meso Scale Discovery; Rockville, MD).

Statistical analysis
All data are presented as mean values, whereas error bars indicate SEM. Statistical tests were conducted as indicated using GraphPad Prism (GraphPad, La Jolla, CA).

AIM2-deficiency protects from self-DNA mediated polyarthritis
To study the contribution of AIM2 to Dnase2-deficiency triggered chronic polyarthritis, we crossed mice deficient in Aim2 with mice lacking Dnase2 and the common chain 1 of the type I IFN receptor (Aim2 -/-, Dnase2 -/-, Ifnar1 -/-). As controls, mice deficient in Dnase2 and Ifnar1 (Aim2 +/+ , Dnase2 -/-, Ifnar1 -/-), deficient in Aim2 and Ifnar1 (Aim2 -/-, Dnase2 +/+ , Ifnar1 -/-) or only deficient in Ifnar1 (Aim2 +/+ , Dnase2 +/+ , Ifnar1 -/-) were studied. For simplicity, the Ifnar1 genotype is not specifically referred to in the following, given the fact that all mice studied were deficient for Ifnar1. Mice of these genotypes were housed under specific pathogen free conditions and monitored for signs of polyarthritis. Consistent clinical signs of polyarthritis were observed around 8 months after birth. As such, Dnase2-deficient mice being wild type for Aim2 displayed a significant mean arthritis score of 1.8 compared to Dnase2-competent mice (Fig 1A). In the additional absence of Aim2, Dnase2-deficient mice (Aim2 -/x Dnase2 -/-) showed a significant reduction in their signs of arthritis with a clinical score that was comparable to Dnase2-competent mice. With increasing age, signs of chronic polyarthritis progressed to a score of up to 3.4 in the Aim2 +/+ x Dnase2 -/cohort at 15 months, whereas Aim2 -/x Dnase2 -/mice showed greatly reduced signs of arthritis (mean score of 0.5). Of note, in line with previous reports, arthritis was more pronounced in fore pads than in hind pads (Fig 1B). Altogether, these results indicated that AIM2 is required for the development of polyarthritis in the context of DNase2-deficiency.
AIM2-deficiency blunts pro-inflammatory as well as antiviral gene expression in diseased joints As mentioned above, it has been proposed that joint inflammation in this mouse model is driven by systemic inflammation. Therefore, to elucidate the mechanism of Aim2-deficiency ameliorating DNase2 -/--associated polyarthritis, we next assessed systemic signs of inflammation in these mice. Dnase2-deficient mice showed markedly increased levels of TNF and IL-10 in serum, however, this was observed irrespective of their Aim2 genotype (Fig 2A). Surprisingly, levels of the disease-relevant cytokines IL-6 and IL-1β were not elevated in serum of Dnase2 -/animals (Fig 2A) and cytokines associated with increased activation of the adaptive branch of the immune system (e.g. IL-2, IL-4 or IFNγ) were also not significantly increased above background level (data not shown). Dnase2-deficient mice developed marked splenomegaly, with spleen sizes being 4-5 fold increased compared to control animals ( Fig 2B). However, as observed for serum levels of TNF and IL-10, the absence of AIM2 had no impact on this phenomenon. Spleen enlargement was associated with an increased expression of IFNβ in splenic tissue (Fig 2C), again being independent of the Aim2 genotype. Moreover, anti-cyclic citrullinated peptide antibodies were also observed in Dnase2-deficient mice irrespective of their Aim2 genotype (Fig 2D).
While these data could not explain the positive impact of Aim2-deficiency in this mouse model, analyzing caspase-1 activation and cytokine expression in joints of Dnase2-deficient mice revealed a different picture. Compared to Dnase2-competent mice, joints of Aim2 +/+ x Dnase2 -/mice showed a marked increase in cleaved caspase-1, which was largely reduced in the absence of AIM2 (Fig 3A). Concomitant with this increased inflammasome activation, the expression of pro-inflammatory cytokines such as TNF, IL-6 and IL-1β expression was strongly increased both at protein, as well as at the mRNA level in joints of Dnase2-deficient mice ( Fig  3B and 3C). Of note, the expression of these pro-inflammatory cytokines was reduced in the absence of Aim2. At the same time, diseased joints also displayed a robust induction of IFNβ and also interferon stimulated gens such as Aim2, Bst2 or Viperin were upregulated at the mRNA or protein level (Fig 3D and S2 Fig). This phenomenon can be explained by the fact that these genes are also directly induced upon cytosolic PRR stimulation, independently of the IFNAR loop [24]. Interestingly, the expression of these ISGs was also Aim2-dependent, most prominently IFNβ.

AIM2 governs macrophage infiltration and local joint destruction in
Dnase2-deficient mice At 15 months of age, joints of Aim2 +/+ x Dnase2 -/mice showed severe signs of synovitis with hyperproliferation of synovial cells and massive immune cell infiltration, associated with pannus formation, cartilage destruction and bone erosion (Fig 4A and 4B). The immune cell infiltrate, most prominently pannus formation, was mainly dominated by the presence of macrophages, as revealed by F4/80 immunohistochemistry. Moreover, as a quantitative measure of local catabolic activity, expression of matrix metalloproteinase-3 (Mmp3) was strongly induced in joints of Aim2 +/+ x Dnase2 -/animals as observed by immunohistochemistry ( Fig  4A and 4B) and also by qPCR and immunoblotting of joint sections (Fig 4C and 4D). In line with the arthritis scores and the pro-inflammatory cytokine profile, histological signs of

Discussion
Altogether our data show that AIM2 plays an important role in the recognition of endogenous DNA species in the context of Dnase2 -/associated arthritis, thereby establishing AIM2 as a DAMP sensing PRR. Interestingly, the impact of AIM2 deficiency was primarily observed in diseased joints and not seen at the systemic level (e.g. splenomegaly or anti-CCP antibodies). These data imply that joint inflammation and subsequent tissue destruction are primary effects and mainly governed by local disease mechanisms and not just secondary to a systemic elevation of pro-inflammatory cytokines. Moreover these results suggest that AIM2 activation drives disease progression in a tissue specific manner in this model. This concept is in line with a previous study that has shown a predominant accumulation of self-DNA in joint tissue of Dnase2deficient mice, thereby arguing for an increase in ligand availability at the site of inflammation [21]. At the same time, it was also demonstrated that Dnase2-deficieny not only results in DNA accumulation resulting from undigested, phagocytized material, but also from cellautonomous sources [21]. As such, damaged DNA expelled from nuclei can gain access to the cytosol, yet is subject to autophagy-mediated transport into DNase II containing lysosomes under normal conditions. The fact that cell autonomous and phagocytized material can both serve as ligands in the context of Dnase2-deficiency might explain why ablating AIM2 does not ameliorate all aspects of this disease model. Of note, with AIM2 being largely confined to the myeloid lineage, it is expected that AIM2 has no impact on self-DNA recognition in non-myeloid cells.
As observed by caspase-1 immunoblotting, Dnase2-deficient mice displayed signs of inflammasome activation in diseased joints, and consistent with its role as an inflammasome receptor, Aim2-deficiency led to a decreased activation of caspase-1. In line with blunted inflammasome activation, IL-1β levels were decreased in joints of Aim2 -/x Dnase2 -/mice, which could explain part of the pathomechanism being operational in this disease model [20]. Interestingly, beyond its impact on the inflammasome, Aim2 deficiency also affected the expression of proinflammatory and antiviral genes in diseased joints, a phenomenon that had previously been ascribed to STING in this mouse model [17]. For example, expression of IFNβ, a cytokine that is completely STING-dependent in the context of cytosolic DNA recognition [25], was greatly reduced in diseased joints of AIM2-deficient mice. Whereas IFNβ itself does not play a role in this disease model (global Ifnar1 deficiency), these results indicate that the signaling cascade leading to IFNβ production is indeed operational. This is also documented by the fact that IFN-stimulated genes are upregualted in joints of Dnase2-deficient mice. While these results could imply an epistatic relationship in signaling, with AIM2 functioning upstream of STING, we hypothesize that this is not the case. Most importantly, in vitro AIM2-deficient macrophages display normal pro-inflammatory gene expression and even increased type I IFN production upon cytosolic DNA delivery, while caspase-1 dependent processing of pro-IL-1β and pro-IL-18 and pyroptosis are abrogated [9,26,27]. These observations clearly argue against a cell-autonomous role of AIM2 in this model functioning upstream or in conjunction with intermediate grade synovitis / infiltration; 3, high grade synovitis / infiltration. C: RNAs were isolated from the joints of 15 months old mice to quantify MMP3 expression normalized to HPRT1. Data are presented as mean values + SEM, whereas statistical significance was assessed using a two-tailed, unpaired ttest comparing the Dnase2 -/cohorts. D: Protein lysates were generated from the joints of 15 months old mice and immunoblotted for the presence of MMP3, whereas β-Actin served as a loading control. Three independent protein lysates were analyzed per cohort. Data are presented as mean values + SEM, whereas statistical significance was assessed using a two-tailed Mann-Whitney test (b) or using a two-tailed, unpaired t-test (c) comparing the Dnase2 -/cohorts.
doi:10.1371/journal.pone.0131702.g004 STING signaling. Moreover, the fact that Aim2 deficiency differentially impacts on pro-inflammatory cytokine expression in joints and in spleen furthermore disfavors a cell intrinsic role of AIM2 positively regulating STING-mediated signal transduction. In fact we speculate that the AIM2-dependent processing of IL-1 family cytokines and pyroptosis-dependent release of DAMPs leads to the priming of neighboring cells within joint tissue. A similar scenario has recently been proposed in the context of intramuscular DNA vaccination, where Aim2-deficient animals showed greatly reduced expression of pro-inflammatory and antiviral cytokines at the sight of DNA application [28]. In this context, it is furthermore noteworthy that proinflammatory cytokines can dramatically enhance the responsiveness of non-myeloid cells to DNA stimulation via the STING axis [29]. Extrapolated to the in vivo situation, this would imply that AIM2 functions in a non-cell autonomous fashion to trigger inflammation, with AIM2 not being active in the cells that express pro-inflammatory cytokines. Nevertheless, it has also been reported that AIM2, when stimulated at lower ligand concentrations, can engage signaling cascades other than the canonical inflammasome pathway. To this end, it was shown that AIM2 can recruit procaspase-8 in an ASC-dependent fashion at ligand concentrations below the threshold of caspase-1 activation or in the absence of caspase-1 respectively [30,31]. While not formally established, caspase-8 recruitment by AIM2 could drive NF-KB activation and as such impact on pro-inflammatory gene expression under these conditions. Altogether, the exact mechanism of AIM2 contributing to pro-inflammatory and antiviral gene expression in this model remains to be determined, while it should be informative to dissect the contribution of different cells types in the in vivo situation.
With the predominant involvement of the adaptive branch of the immune system unequivocally being documented in rheumatoid arthritis, the relevance of erroneous DNA sensing by PRRs has yet to be established in this disease entity. However, the Dnase2 -/model shares a number of prominent features with Systemic juvenile idiopathic arthritis (sJIA), which can be considered an autoinflammatory disease [20,32]. Having identified AIM2 as a critical proinflammatory driver in this disease model provides a molecular target that could be of interest for future therapeutic intervention.
Note added in proof: While this manuscript was under review, Ellen Gravallese and colleagues have published a likewise approach of studying the role of AIM2 in the context of Dnase2 -/x Ifnar1 -/induced polyarthritis [33]. They also found a critical role for AIM2 in driving disease pathology (clinical arthritis score, pro-inflammatory cytokine production in joints), yet with a slightly lower impact compared to our study. Moreover, in their analysis Aim2 deficiency led to a considerable decrease in IL-18 levels in serum and in joint tissue, a phenomenon that we did not observe in our cohorts (data not shown). It is possible that differences in mouse strains used or housing conditions account for these discrepancies. However, given the fact that IL-18 deficiency does not protect Dnase2-deficient animals from polyarthritis or proinflammatory gene expression [20], Aim2-dependent IL-18 production or maturation appears to be an epiphenomenon rather than a cause in this disease model. Indeed, our data suggest that Aim2 deficiency dampens caspase-1 activation and associated IL-1β secretion and that it also impacts on pro-inflammatory cytokine expression in joint tissues.