Nod2 Deficiency Augments Th17 Responses and Exacerbates Autoimmune Arthritis

This article is featured in In This Issue, p.1811

Inflammatory arthritis is a chronic disorder of the joints involving inflammation of the synovial membrane, leukocyte recruitment, and progressive bone and cartilage destruction. Arthritis manifests in over 100 different rheumatic disorders, including rheumatoid arthritis, spondyloarthropathies (e.g., ankylosing spondylitis), psoriatic arthritis, systemic erythematosus lupus, and enteropathic arthritis (i.e., arthritis associated with inflammatory bowel disease). As a leading cause of disability in the United States, arthritis is a major public health concern because of its economic burden and impact on quality of life (1). Although the cause of arthritis is poorly understood, a current paradigm proposes that complex interactions between genetic and environmental factors result in a break in immune tolerance and the generation and propagation of autoreactive T cells. Compelling data from clinical studies and experimental models of arthritis implicate a central role for CD4⁺ Th cells in the pathogenesis of disease. In particular, a lineage of CD4⁺ T cells that produce the cytokine IL-17 (Th17 cells) is known to be pathogenic in arthritis (2, 3). Th17 cells promote the recruitment of neutrophils and activation of joint synoviocytes and chondrocytes, leading to inflammation and bone destruction (2, 3). Although Th17 immunity is understood to promote arthritis, anti–IL-17 biologics have had limited clinical benefit (4), thereby underscoring the complexity surrounding the role of Th17 cells in disease and that targeting IL-17 alone is not ideal. Thus, identification of additional mechanisms that engender a break in immune tolerance and control of autoreactive Th17 cells should be a critical focus of investigation.

Despite the wealth of knowledge about the role of the adaptive immune system in arthritis, its relationship with innate immunity is poorly defined. Intriguingly, NOD2, a member of the Nod-like receptor (NLR) family, appears to play a role in arthritis, as a point mutation in this molecule leads to ∼100% incidence of a heritable form of polyarticular arthritis as part of Blau syndrome (5–7). Thus, studies investigating the biological function(s) of Nod2 could offer important insights into how a single gene might function in arthritis. Nod2, a pattern-recognition receptor important for host defense against intracellular bacteria (8), was initially identified as being expressed predominantly by myeloid cells. However, Nod2 expression has since been detected across hematopoietic and nonhematopoietic cells, including T cells (9, 10), vascular endothelial cells (11), synoviocytes (12), and chondrocytes (13). Nod2 detects cytosolic bacteria by sensing peptidoglycan (PGN), of which muramyl dipeptide is the minimal moiety required for its activation. Recognition of PGN by Nod2 activates RIP2, MAPK, and NF-κB as part of a signaling cascade that results in transcription of genes responsible for antimicrobial immunity (14, 15). More recently, Nod2 has been reported to participate in signaling responses outside of its “canonical” PGN-sensing function, such as antiviral responses, autophagy, and endoplasmic reticulum stress responses (16–19). Previous studies suggest a role for Nod2 in local injurious responses of the synovium induced by intra-articular injection of PGN (20–23). However, our understanding of the role of Nod2 in the generation or function of autoreactive T cells remains limited.

Given the strong clinical link between NOD2 and rheumatic disease, we sought to investigate the role of Nod2 in a T cell–mediated model of arthritis. SKG mice are genetically prone to arthritis because of an inherent mutation in the T cell–signaling molecule Zap-70 (24), which diminishes the strength of TCR signaling initiated by TCRξ and CD3 chains (25). Thus, in SKG mice, compromised central tolerance results in the escape of autoreactive T cells from the thymus into the periphery where they can become Th17 cells that target the joint (26). Additional signals, such as those provided by fungal-derived β-glucan polymers such as zymosan, are required for the subsequent generation of pathogenic Th17 cells and development of arthritis in SKG mice (27).

These studies identify a previously unrecognized role for Nod2 as an essential protectant against development of arthritis. Absence of Nod2 resulted in a worsened form of arthritis in SKG mice, which was mediated by dysregulation of the Th17 response. An important finding from these studies is that Nod2-mediated protection was intrinsic to CD4⁺ T cells. In particular, this protection was conferred not through effects on regulatory T cell (Treg) development or function but rather through effects on CD4⁺ T cell production of IL-17. Reconstitution of lymphopenic recipients with CD4⁺ T cells purified from naive Nod2^−/−SKG mice versus SKG mice recapitulated the worsened phenotype observed in Nod2^−/−SKG mice, thereby indicating a T cell–intrinsic function for Nod2 in control over arthritis. Future studies aimed at further understanding an endogenous negative regulatory function of Nod2 within autoreactive T cells could inform us of potentially novel therapeutic strategies for arthritis.

Nod2^−/−SKG mice were generated by breeding SKG mice (24) with Nod2^−/− mice (The Jackson Laboratory) that we had backcrossed 10 generations onto the BALB/c background. Nod2 deficiency and the G489T mutation in Zap-70 of SKG progeny were confirmed by PCR as described (28). Athymic nude (nu/nu) and Rag1^−/− mice, both on BALB/c background (The Jackson Laboratory), were bred in our facility. Nod2^−/−Rag1^−/− mice were generated by breeding Nod2^−/− mice with Rag1^−/− mice, with the Nod2 deficiency confirmed by PCR. Animals were maintained under specific pathogen-free conditions at the Veterans Affairs (VA) Portland Health Care System. All studies were conducted with 6-wk-old female mice, and carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and institutional guidelines for animal welfare.

Arthritis was induced by a single i.p. injection of 1.5 mg zymosan (Sigma-Aldrich), a cell wall–derived preparation of Saccharomyces cerevisiae enriched in β-glucan polymers. Clinical arthritis for each paw was graded (0–4) in masked fashion using defined criteria (28), and scores for each paw were summed such that the total score per mouse ranged from 0 to 16. For calculation of disease incidence, a mouse was considered positive for arthritis when a total clinical score was ≥1 and maintained for 2 or more wk. Body weight loss was calculated as percentage change between body weights at the start and termination of study. For histological analysis of joints, dissected ankles were fixed, decalcified, and embedded in paraffin (28). Tissue sections (7 μm thick) were stained with H&E and examined using light microscopy (original magnification ×50).

Segments of the small intestine were coiled, fixed in 10% neutral-buffered formalin, and embedded in paraffin. Sections (7 μm) were stained with H&E, and 12 regions of the ileum of each mouse intestine were photographed (using 10× objective). To quantify the extent of disease, images were analyzed offline with ImageJ (National Institutes of Health) by a pathologist masked to experimental condition. Data are expressed as percentage lesion defined as the percentage length of discrete inflammatory lesions (i.e., accumulations of mononuclear and polymorphonuclear inflammatory cells within the mucosa, submucosa, and muscularis propria) relative to length of intestine within the photograph.

Upon termination of experiments, imaging was performed on legs and spines dissected from ProSense 680–injected (catalog number NEV10003; PerkinElmer) animals. Emitted intensities of the near-infrared (NIR) signal, indicating levels of protease activity, were quantified within defined regions of interest (ROI), as previously detailed in the SKG model of arthritis (28). Signal intensities in ROI were first normalized to those of corresponding ROI from healthy naive wild-type (WT) BALB/c mice using the small animal image analysis application of Image Studio Software (LI-COR Biosciences) before quantification.

SKG and Nod2^−/−SKG mice were housed in a 1:1 ratio to equally expose each genotype to a composite of fecal and bedding material, a strategy based on a prior recolonization study in SKG mice (29). Mice were cohoused starting at the time of weaning (3 wk of age) and throughout the duration of the study (i.e., out to 8 wk postzymosan).

For in vivo IL-17A neutralization studies, mice were i.p. injected with 100 μg of either anti–IL-17A mAb (Clone 50104; R&D Systems) or rat IgG2a isotype control (clone 54447; R&D Systems) on the day prior to zymosan challenge and weekly thereafter. This dosing regimen was based on a prior report of in vivo efficacy of IL-17 blockade in SKG mice (30).

Cytokine levels for IL-17A/F, IL-10, and IFN-γ were quantified by ELISA according to the manufacturer’s instructions (DuoSet ELISA kits; R&D Systems). To measure IL-17A levels in the intra-articular space of the joint, synovial fluid was aspirated from the synovial capsule of ankle joints using a 23-gauge needle and immediately frozen at −20°C. Equal volumes (1 μl) of synovial fluid were used to measure IL-17A by ELISA.

Synovial fluid was aspirated from the subcapsular space of ankle joints. Cells from ankle-draining (popliteal) lymph nodes and spleens were prepared by collagenase digestion (Clostridium histolyticum Collagenase D; Roche) and filtration through a 70-μm cell strainer. Erythrocytes were lysed using RBC lysis buffer (Sigma-Aldrich). Collected cells were blocked with mAb to FcγRIII/I (2.4G2; BD Biosciences) and subsequently incubated with mAb to the following surface molecules for T cells: Thy1.2 (53-2.1; BioLegend), CD8α (53-6.7; eBioscience), and CD4 (RM4-5; BD Biosciences) along with LIVE/DEAD Aqua stain (Life Technologies). They were then fixed in 4% paraformaldehyde (Sigma-Aldrich). For in vitro stimulation and intracellular cytokine staining, single-cell suspensions from popliteal lymph nodes were prepared as above and seeded at 5 × 10⁶ cells/ml in 96-well plates in RPMI 1640 supplemented with 10% FBS (endotoxin-free; Gemini Bio-Products). Cells were treated with 20 ng/ml PMA (LC Laboratories) and 1 μg/ml ionomycin (LKT Laboratories) versus media for 5 h in the presence of brefeldin A (1 μg/ml; BD Biosciences). Cells were blocked, stained for cell surface markers as above, and fixed and permeabilized using BD Cytofix/Cytoperm fixation/permeabilization kit (BD Biosciences), after which they were stained with mAbs to IL-17A (TC11-18H10), TNF-α (MP6-XT22), IFN-γ (XMG1.2), IL-22 (IL22JOP), and GM-CSF (MP1-22E9). Foxp3 expression was assessed using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Flow cytometry was performed on a BD LSRFortessa (BD Biosciences) and analyzed using FlowJo software (FlowJo).

T effector (Teff) cells (CD4⁺CD25⁻CD45.1⁺) were isolated (>95% purity) from spleens and lymph nodes of naive CD45.1 congenic BALB/c mice by magnetic bead separation using a CD4 Positive Selection Kit (STEMCELL Technologies) and subsequent depletion of CD25⁺ cells (CD25 positive selection kit; STEMCELL Technologies). Tregs (CD4⁺CD25⁺CD45.2⁺) were purified (>95%) from spleens and lymph nodes of naive BALB/c, SKG, or Nod2^−/−SKG mice using EasySep Mouse CD4⁺CD25⁺ Regulatory T Cell Isolation Kit II (STEMCELL Technologies). Teff cells (1 × 10⁵ per well) were pretreated with 5 μM CellTrace Violet cell proliferation dye (Thermo Fisher Scientific) and cultured with indicated numbers of Tregs, along with 2 × 10⁵ CD45.1⁺ autologous BALB/c APCs that had been irradiated (1000 rad). Cultures were stimulated with 0.5 μg/ml plate-bound anti-CD3 (145-2C11, eBioscience). Proliferation of Teff cells was assessed 72 h poststimulation by CellTrace Violet dilution on the BD LSRFortessa (BD Biosciences) and analyzed using FlowJo software (FlowJo).

CD4⁺ T cells isolated from spleens and lymph nodes of 6-wk-old naive SKG or Nod2^−/−SKG mice were purified (>95%) using EasySep Mouse CD4 Negative Selection Kit (STEMCELL Technologies). Cells were seeded at 8 × 10⁴ CD4⁺ T cells per well into 96-well round-bottom plates (BD Biosciences) and resuspended in RPMI medium containing 2 mM l-glutamine plus 10% FCS plus 50 μM 2-ME alone (unstimulated) or in media containing 20 ng/ml PMA and 1 μg/ml ionomycin. Supernatant was collected at 14 h for measuring cytokine levels by ELISA.

Single-cell suspensions were prepared from spleens and lymph nodes of 6-wk-old naive donor SKG or Nod2^−/−SKG mice. CD4⁺ T cells were purified (>95%) using the EasySep Mouse CD4 Positive Selection Kit (STEMCELL Technologies). Lymphopenic recipients (Nude or Rag1^−/− mice) were i.v. injected with 2 × 10⁶ donor CD4⁺ T cells and then were i.p. injected with 1.5 mg of zymosan 24 h later.

The nonparametric statistics test Mann–Whitney U test was used to compare two samples (e.g., zymosan-treated SKG to zymosan-treated Nod2^−/−SKG or BALB/c to Nod2^−/−BALB/c). Groups being compared are indicated within each panel. All data were analyzed using Prism (GraphPad Software) and considered significant when p < 0.05. Data are presented as box plots in which the horizontal lines represent the median, boxes indicate the interquartile range, and error bars indicate the minimum and maximum or as bar graphs with the mean and SEM. All experiments were independently performed two to three times.

To investigate whether Nod2 plays a role in regulation of autoreactive T cells and pathogenesis of arthritis, we used an established T cell–mediated model of arthritis in SKG mice and generated congenic SKG mice lacking Nod2 (Nod2^−/−SKG). SKG and Nod2^−/−SKG mice were injected with β-glucans (zymosan) and monitored for development of arthritis (27, 28). By week 3 after zymosan injection, Nod2^−/−SKG mice showed significantly worse clinical arthritis than SKG mice (Fig. 1A). Arthritis in Nod2-deficient animals worsened over time, whereas disease in SKG mice remained relatively stable. Consistent with a prior report (27), specific pathogen-free–housed SKG mice did not develop overt clinical arthritis in the absence of zymosan, a phenotype that was unchanged in Nod2-deficient SKG mice (Fig. 1A). Nod2^−/−SKG mice developed greater joint swelling, redness, and deformity within their forepaws, hind paws, and digits as compared with SKG mice (Fig. 1B; insets of clinical arthritis). Histopathological analysis of Nod2^−/−SKG mouse hind paws and ankle joints revealed more extensive pathological changes, including substantial synovial hyperplasia within the sublining region that was marked by massive accumulation of mononuclear cells and pannus-eroded cartilage (Fig. 1B, boxes). Exacerbated “end-stage” destruction observed in the ankles of Nod2^−/−SKG mice included fusion of subchondral bones and joint dislocations that are not typically present in SKG mice (Fig. 1B). In addition to their increased disease susceptibility, Nod2^−/−SKG mice experienced greater cumulative weight loss compared with their SKG counterparts (Fig. 1C).

To further evaluate how expression of Nod2 affects inflammation-associated molecular changes in cartilaginous and synovial joints, we applied NIR imaging technology in combination with a fluorophore-labeled probe that is activated upon cleavage by proteases. Thus, the intensity of signal positively correlates with protease-mediated inflammation (28), a critical molecular response of arthritis. At 8 wk after zymosan injection, the joints from both Nod2^−/−SKG and SKG mice had greater overall signal intensity than naive BALB/c WT (Fig. 1D). Inflammatory protease activity was localized primarily to the ankle joint but was also seen to extend from the knee to where the tibiotarsus is located to the ankle as well as within the region of the tarsometatarsal joints of the hind paws (Fig. 1D, inset). Protease activity was significantly greater in the ankle, wrist, and knees of Nod2^−/−SKG mice than in SKG mice. Although inflammation was detected within the lower vertebral joints of the spine as reported (28), it was not significantly altered by Nod2 deficiency (Fig. 1D). Taken together, these data indicate an endogenous protective function for Nod2 in mitigating joint inflammation caused by autoreactive T cells in SKG mice.

To determine whether Nod2 expression influences development of arthritis in the absence of the skg mutation, we injected WT (BALB/c) and Nod2^−/− (BALB/c) mice with zymosan. By 8 wk, neither genotype had developed signs of arthritis as evaluated by clinical scoring or joint pathology (Fig. 2A). The more sensitive NIR imaging detected a minimal but significant increase in inflammation within the ankle joints of WT mice compared with Nod2^−/− mice (Fig. 2B). Development of subclinical synovitis has been reported as being T cell independent (27), thereby underscoring the findings presented in this study suggesting that the protective role of Nod2 in arthritis (Fig. 1) occurs through a T cell–mediated mechanism conferred by the skg mutation rather than solely as a result of a response to zymosan.

In light of the sexual dimorphism of arthritis in SKG mice (28, 31), we next sought to rule out the possibility that the Nod2-associated arthritis phenotype in SKG mice was related to sex. Arthritis induced in male SKG and Nod2^−/−SKG mice (Supplemental Fig. 1) was found to be consistently worse in Nod2^−/−SKG mice compared with SKG counterparts, indicating that the role of Nod2 in controlling arthritis susceptibility is independent of sex.

In addition to the effect of sex, animal housing conditions (i.e., conventional versus germ-free) and the presence of particular species of microbiota can influence the development of arthritis in SKG mice (24, 32). These data coupled with numerous studies suggesting that Nod2 is a critical regulator of the gut microbial composition and intestinal inflammation (33), we sought to evaluate whether worsened arthritis in Nod2-deficient mice would be affected by normalizing their microbiota to that of SKG mice. To do this, SKG and Nod2^−/−SKG mice were cohoused starting at the time of weaning (4 wk of age), and after 4 wk of cohousing, arthritis was induced (Fig. 2C–H). Under cohousing conditions, SKG and Nod2^−/−SKG mice received equal exposure to a composite of fecal and bedding material. Nod2^−/−SKG mice developed substantially increased clinical arthritis regardless of housing status (Fig. 2C). Furthermore, those cohoused with SKG mice experienced greater cumulative weight loss (Fig. 2D), similar to our findings of noncohoused conditions (Fig. 1C). The worsened arthritis in cohoused Nod2^−/−SKG mice was corroborated by NIR imaging (Fig. 2E) and by histopathological evaluation (Fig. 2F), which revealed severe ankle inflammation, extensive synovial hyperplasia, and pannus-eroded cartilage similar to that observed in non-cohoused mice. Notably, clinical arthritis severity in SKG mice was the same regardless of cohousing history (Fig. 2C), thereby supporting a protective function for Nod2 in arthritis independent from microbiota composition. These data are consistent with prior reports of cohoused SKG and BALB/c mice, wherein arthritis was neither horizontally transmissible nor altered by microbiota transfer (29).

To further evaluate the potential influence of microbiota on disease state, we examined the extent to which Nod2 influences the SKG-mediated intestinal disease previously described as ileitis (34), which is mediated by dysbiosis in SKG mice. The severity of ileitis coinciding with arthritis in SKG and Nod2^−/−SKG mice was assessed (Fig. 2G, 2H). In response to zymosan, the small intestine of SKG mice developed considerable inflammation of the ileum, consisting of multiple discrete segments of tissue distortion caused by a mixed mononuclear and polymorphonuclear cell infiltrate within the mucosal epithelium that in many cases extended into the lamina propria (Fig. 2G). Notably, SKG-mediated ileitis was almost completely abrogated in the context of Nod2 deficiency, suggesting that in the ileum, expression of Nod2 is deleterious and contributes to ileitis in SKG mice. Collectively, data in Fig. 2 support a protective function of Nod2 in mitigation of arthritis that is not likely a direct influence of dysbiosis or strain-specific microbiota.

To gain insight into the role of Nod2 in arthritic T cell responses, we evaluated the immune cell composition of lymph nodes that drain the ankle (popliteal) in arthritic SKG and Nod2^−/−SKG mice. Popliteal lymph nodes from arthritic Nod2^−/−SKG mice tended to have increased mass (Fig. 3A) and cellularity (data not shown) compared with those of SKG mice, although the differences were NS in either case. However, we observed a significant increase in the total numbers of CD4⁺ T cells in the lymph nodes of arthritic Nod2^−/−SKG mice (Fig. 3B), suggesting expansion of a pathogenic T cell population. The expansion of CD4⁺ T cell responses was attributed to zymosan and the development of arthritis, as we have not observed differences in baseline T cell populations of untreated (naive) mice (Supplemental Fig. 2). The increase in T cell numbers was also observed in the joint, as synovial fluid obtained directly from inflamed joints of arthritic Nod2^−/−SKG mice had increased numbers of T cells, including CD4⁺, CD8⁺, and CD4⁻CD8⁻, in comparison with arthritic SKG mice (Fig. 3C). These data indicate that Nod2 deficiency results in increased numbers of T cells in response to zymosan, which could promote arthritis.

Because CD4⁺ T cells that produce IL-17 (Th17 cells) are necessary for arthritis in SKG mice (26), we next examined whether Nod2 deficiency altered Th17 immunity. Intracellular cytokine staining of T cells from the popliteal lymph nodes showed an increase in the percentage and total numbers of IL-17–producing CD4⁺ T cells in arthritic Nod2^−/−SKG compared with arthritic SKG mice (Fig. 3D–F). Of note, the propensity of Nod2-deficient cells to produce IL-17 was observed in arthritic CD4⁺ T cells even without stimulation in vitro (Fig. 3E). Further evaluation of how Nod2 deficiency affected polarization of Th populations revealed that arthritic Nod2^−/−SKG mice also had increased proportions of Th1 cells along with a greater proportion of CD4⁺ T cells producing additional cytokines including GM-CSF and TNF, but not IL-22 (Supplemental Fig. 3). Further analysis specifically of the gated Th17 population demonstrated that the majority (∼60%) of the Th17 cells coproduced TNF, along with smaller proportions coproducing GM-CSF, IL-22, and IFN-γ (Fig. 3G), a phenotype that was genotype independent. Consistent with the exacerbated Th17 response in Nod2^−/−SKG mice, levels of IL-17 (within ankle synovial fluid) were also significantly elevated in Nod2^−/−SKG versus SKG mice (Fig. 3H). These data indicate that Nod2 deficiency results in increased generation of Th17 cells that potentiate arthritis development.

Given the above data indicating the importance of Nod2 expression in control of the Th17 response, we evaluated the contribution of IL-17 to the pathogenesis of arthritis in Nod2^−/−SKG mice. Arthritis was induced in Nod2^−/−SKG and SKG mice in concert with administration of anti–IL-17A blocking Ab or isotype control (Fig. 4). In findings consistent with those of a prior report (26), we observed an important role for IL-17 in induction of arthritis in SKG mice, as evaluated by clinical scoring (Fig. 4A), NIR imaging (Fig. 4B), and histopathology (Fig. 4C) of ankle joints, which demonstrated reduced overall inflammation in the anti–IL-17 treated, compared with untreated, SKG mice. Importantly, Nod2^−/−SKG mice depleted of IL-17 exhibited arthritis significantly reduced to the level of arthritic SKG mice (Fig. 4A). This finding was confirmed by NIR imaging and histopathology that corroborated the reduction in joint inflammation in Nod2^−/−SKG mice (Fig. 4B, 4C). Collectively, these data indicate that increased IL-17 production is sufficient to cause worsened arthritis in Nod2-deficient SKG mice.

Tregs play an important role in suppression of autoimmunity, including negative regulation of inflammatory Th17 cells (35). Thus, we sought to determine whether Nod2 deficiency altered Treg (CD4⁺Foxp3⁺) responses in SKG mice. The frequencies of Tregs were indistinguishable between SKG and Nod2^−/−SKG mice, regardless of whether they were in a naive or arthritic state (Fig. 5A). Nod2^−/−SKG mice tended to have a slight, but NS, increase in Treg frequency during arthritis compared with their naive state. This could be a compensatory response to excessive inflammatory CD4⁺ T population in arthritic Nod2^−/−SKG mice. Nonetheless, there was no significant difference between genotypes, indicating that Nod2 expression does not overtly alter peripheral development of Tregs in SKG mice. We further evaluated whether Nod2 deficiency altered the suppressive function of Tregs in SKG mice. The ability of WT (BALB/c) Teff cells (CD4⁺CD25⁻) to proliferate in vitro upon activation by anti-CD3 and costimulation was evaluated in the presence of Tregs (CD4⁺CD25⁺) derived from naive WT (BALB/c), SKG, or Nod2^−/−SKG mice (Fig. 5B). In observations consistent with those of a prior report (25), we found that SKG Tregs had impaired suppressive function (Fig. 5B). However, Nod2 deficiency did not further alter the suppressive function of SKG Tregs (Fig. 5B). These data indicate that the cellular mechanism by which Nod2 mitigates the Th17 response and arthritis is independent of Tregs.

To gain further insight into the cellular mechanism by which the absence of Nod2 leads to more severe arthritis in SKG mice, we examined whether Nod2 influences inflammatory Th17 cell responses under homeostatic conditions (prior to treatment with zymosan). Indeed, we found that splenocytes from naive (untreated) Nod2^−/−SKG mice had an increased frequency of Th17 cells compared with SKG mice (Fig. 6A). We next tested whether Nod2 deficient CD4⁺ T cells produced more IL-17 on a per-cell basis. To do this, we stimulated equal numbers of CD4⁺ T cells purified from naive SKG or naive Nod2^−/−SKG mice with PMA/ionomycin (PMA/io). Indeed, CD4⁺ T cells from naive Nod2^−/−SKG mice produced greater amounts of IL-17 than SKG counterparts (Fig. 6B), indicating that Nod2-deficient SKG mice have increased Th17 responses under homeostatic conditions. In addition, we examined the ability of purified CD4⁺ T cells to produce other cytokines associated with Th1 and Th2 responses and found that in the absence of Nod2 production, IFN-γ was not altered and IL-10 production was reduced (Fig. 6C, 6D). Furthermore, zymosan did not have any direct effect on the ability of CD4⁺ T cells to produce IL-17, IFN-γ, or IL-10, as purified CD4⁺ T cells stimulated with zymosan (50 μg/ml) did not result in detectable amounts of cytokine (data not shown). Cumulatively, these data indicate a role for Nod2 as an endogenous suppressant of inflammatory Th17 responses.

We were next interested in determining if Nod2 affected the ability of pathogenic autoreactive T cells to cause arthritis on a single-cell level in vivo. To do this, we adoptively transferred equal numbers of purified CD4⁺ T cells from naive Nod2^−/−SKG or SKG mice into lymphopenic nude recipients that were injected with zymosan 24 h later and analyzed weekly for arthritis severity (Fig. 6E–G). Mice that received Nod2^−/−SKG CD4⁺ T cells developed significantly worse clinical arthritis than recipients of SKG CD4⁺ T cells (Fig. 6E). The exacerbated clinical arthritis was corroborated by a marked increase in NIR signal (Fig. 6F) and worsened histopathology that phenocopied data showed in arthritic Nod2^−/−SKG mice (Fig. 1B), including increased immune cell infiltrates and synovial hyperplasia (Fig. 6G). These data show that Nod2 is an intrinsic negative regulator of pathogenic autoreactive T cells.

To more thoroughly assess the T cell–intrinsic function of Nod2 in protection against arthritis in SKG mice, we adoptively transferred purified CD4⁺ T cells from naive SKG (Nod2-sufficient) mice into Rag1^−/− or Nod2^−/−Rag1^−/− recipients, who were then treated with zymosan and evaluated for arthritis (Fig. 6H). Severity of arthritis was similar regardless of the recipient’s genotype, indicating that lack of Nod2 expression by all cell types except T cells does not affect the pathogenicity of SKG CD4⁺ T cells. These data further support a protective function for Nod2 within CD4⁺ T cells as a suppressor of arthritogenic T cell responses and arthritis.

Nod2 is a critical regulator of immune homeostasis, as mutations in NOD2 cause the autoinflammatory disease Blau syndrome and confer significant risk for development of other inflammatory diseases, including Crohn disease, graft-versus-host disease, and sarcoidosis (36). Despite the definitive importance of Nod2 in joint homeostasis, as evident in patients with Blau syndrome, the cellular and molecular mechanisms by which Nod2 controls arthritis are unclear. Our studies provide insight into a novel role for Nod2 in regulation of T cell homeostasis as it relates to autoimmune forms of arthritis. We used the genetically susceptible strain of SKG mice that, upon exposure to β-glucans such as zymosan, develop joint inflammation and pathology that is mediated by autoreactive CD4⁺ T cells. Nod2-deficient SKG mice manifested an exacerbated form of arthritis that was accompanied by an increased magnitude of IL-17–producing autoreactive CD4⁺ T cells, which were responsible for promotion of joint pathology. Interestingly, the effects of Nod2 deficiency on Th17 responses and arthritis were not due to altered Treg development or function and occurred independently of sex or intestinal dysbiosis. Rather, Nod2-deficient CD4⁺ T cells produced more IL-17 and had increased ability to cause arthritis on a single-cell level. Mechanistically, Nod2 did not appear to increase production of autoreactive T cells, as Nod2-deficient BALB/c mice (with the WT allele of Zap-70) did not develop arthritis. Instead, naive Nod2^−/−SKG mice had increased Th17 responses, indicating that Nod2 is an endogenous negative regulator of Th17 responses in SKG mice. These data identify a pivotal role of Nod2 in natural protection against autoimmune arthritis in large part by homeostatic regulation of Th17 cell responses in SKG mice.

A key observation from this work is that the protective capacity of Nod2 occurred through control of autoreactive CD4⁺ T cells. CD4⁺ T cells purified from naive Nod2^−/−SKG mice produced more IL-17 and had enhanced capacity to cause arthritis on a single-cell level, indicating that endogenous Nod2 expression in T cells could directly suppress their arthritogenic capacity. A T cell–intrinsic role for Nod2 was previously established by Shaw et al. (10), wherein Nod2 deficiency led to increased susceptibility of mice to Toxoplasma gondii infection due to dysregulated c-Rel activation within the T cell that resulted in a delay in IFN-γ production that is essential to parasite clearance. Although IFN-γ is known to suppress Th17 responses and arthritis in SKG mice (26), we and others did not see a change in the ability of CD4⁺ T cells to produce IFN-γ in Nod2-deficient mice (37). This discrepancy in IFN-γ production could in part be explained by differences in the ways T cells were activated in vitro (Ag versus CD3 versus PMA/io) or, in the case of our study, skewed by the presence of the skg mutation. Nonetheless, we interpret these data to mean that diminished IFN-γ and/or Th1 response is not responsible for the enhanced Th17 responses of Nod2^−/−SKG mice.

The complexity surrounding Nod2 control over Th17 immunity is exemplified by the fact that Nod2 can either promote (38, 39) or repress (40, 41) the Th17 response, depending on specific factors within different scenarios (e.g., stimulus used). As with arthritis in SKG mice, there is evidence of negative regulation of Th17 immunity by Nod2 as a protective mechanism against intestinal disease; however, the cellular mechanisms appear to extend beyond T cell–mediated Nod2 function that includes intestinal epithelial cells (42). Our data are consistent with a prior report of a T cell–induced enteropathy model in which i.p. injection of anti-CD3 mAb was used to directly activate T cells in vivo and in which the ensuing intestinal inflammation and Th17 response was worsened by Nod2 deficiency (42). Interestingly, using a different dose of anti-CD3 mAb with the same model system, another group reported the opposite observation, in that Nod2 deficiency diminished both il-17a mRNA expression and intestinal disease (43). Collectively, we interpret this to mean that the role of Nod2 in Th17 immunity is likely influenced by the kinetics of T cell activation and/or strength of TCR signaling, which is relevant in the context of arthritis in SKG mice that arises from impaired TCR signaling and generation of autoreactive T cells.

Autoreactive T cells that escape central tolerance in the thymus are subject to negative regulation by CD4⁺ Tregs. We found that the expansion in Th17 cells was not likely because of diminished function of Tregs, as Nod2 deletion did not alter the peripheral development or suppressive function of Tregs in SKG mice. These data are supported by a previous study showing that Treg function is unchanged in Nod2-deficient C57BL/6 mice (44). Furthermore, we found that Nod2-deficient SKG CD4⁺ T cells had impaired production of IL-10, a cytokine that directly dampens deleterious Th17 responses during inflammation and autoimmunity (45, 46). Nod2 has previously been demonstrated as a positive regulator of IL-10 production by myeloid cells (47, 48) and CD8 T cells (43) as part of a protective mechanism in intestinal disease. However, to our knowledge, the ability of Nod2 to positively regulate IL-10 production by CD4⁺ T cells and in the context of arthritis has not been described until now. Interestingly, autocrine production of IL-10 by self-reactive T cells has been proposed as a mechanism of self-regulation meant to suppress the development of pathogenic autoreactive Th17 responses (49). These studies and our data support a potential function for Nod2 within CD4⁺ T cells in dampening development of pathogenic Th17 cells. Because T cells express Nod2 during peripheral activation (9) as well as during thymic development (50), future studies that elucidate how Nod2 modulates arthritogenic T cell responses that could arise from compromised peripheral and even central tolerance mechanisms will be of importance to our understanding of Nod2 in arthritis and autoimmunity.

Another interesting observation from this work is that different immunological mechanisms appear to underlie ileitis and arthritis in SKG mice, as exemplified by the discordant effects of Nod2 deficiency in the gastrointestinal tract ileum versus the joint. Although Nod2^−/−SKG mice developed worsened arthritis, they had dramatically reduced ileitis compared with SKG mice, a phenotype that could be related to different roles for T cells and/or microbiota. The underlying mechanisms of ileitis development in SKG mice are continuing to be defined but appear to be T cell independent (34) and related to their inherent intestinal dysbiosis (32). In contrast, arthritis development in SKG mice is T cell mediated (24) and independent of dysbiosis, as evidenced by cohousing studies of SKG mice with WT-BALB/c mice, which alleviated ileitis but not arthritis (29). Moreover, induction of arthritis was unaffected in germ-free mice (29), whereas ileitis was mitigated. We would interpret our observations similarly to those of a recent report of compounding genetic interactions between Nod2 and development of colitis in predisposed SAMP1Yit/Fc mice (51). That given, that distinct mechanisms pertaining to microbiota control of ileitis versus arthritis occur in SKG mice would be further supported by our data demonstrating that worsened arthritis in Nod2^−/−SKG mice was maintained in mice cohoused with SKG mice. These data coupled with diminished ileitis in Nod2^−/−SKG mice would support an endogenous protective function of Nod2 in shaping T cell responses in arthritis, which is likely independent of microbiota composition.

Collectively, these findings illustrate how Nod2 may serve multiple mutually exclusive functions in pathways and tissues such as the gut or joint that can result in contrasting effects on disease. This paradigm may offer new insight into interpretation of prior work by our laboratory and others that demonstrated promotion of arthritis by Nod2. Prior studies were based on different types of animal models, wherein arthritis was triggered by localized injury to the joint with PGN (20–23) or immune complexes (52). In these scenarios, Nod2 deficiency was protective and the inflammatory function of Nod2 was attributed to its muramyl dipeptide–sensing (53) or PGN-sensing function (20–23) and activation of the canonical Rip2 pathway (52). Moreover, synovial inflammation is exacerbated by mechanistic actions of Nod2 within innate cells (i.e., neutrophils and monocytes) or tissue-resident cells such as synovial fibroblasts and chondrocytes (13, 54) versus T cell–mediated disease, as is exemplified in SKG mice reported in this study. Indeed, the protective effect of Nod2 in SKG mice required the skg mutation and could be transferred with Nod2-deficient SKG T cells, thereby underscoring the T cell–intrinsic mechanism of arthritis modeled in this study. Importantly, Nod2-sufficient SKG CD4⁺ T cells were able to trigger arthritis in lymphopenic hosts (i.e., Rag1^−/− or Nod2^−/−Rag1^−/−), irrespective of the recipient’s genotype. These data indicate that the mechanism of protection of Nod2 in SKG mice originates from CD4⁺ T cells rather than other cellular sources of Nod2 in the joint or periphery. Our studies reveal a novel role for Nod2 in T cell homeostasis that is independent of its role in microbial sensing or in acute inflammatory responses.

In conclusion, we present evidence for a unique function of Nod2 in attenuation of autoreactive T cell responses that cause arthritis. These data broaden our understanding of the biological functions of Nod2 to include roles in Th17 differentiation and T cell homeostasis that could help clarify the pathophysiology of arthritis. Understanding how Nod2 may participate in immune tolerance mechanisms could elucidate hitherto unappreciated mechanisms of protection that may be exploited for development of novel treatments for patients with Blau syndrome as well as other forms of arthritis such as rheumatoid arthritis.

We thank Brieanna Brown and Sharon Osterbind at the VA Portland Health Care Systems as well as Fanny Polesso (Oregon Health & Science University) for technical contributions. We are grateful to Dr. Cong-Qiu Chu (VA Portland Health Care System, Oregon Health & Science University) for helpful discussions. We are grateful for support from the Portland VA Research Foundation.

The authors have no financial conflicts of interest.