Insufficient IL-10 Production as a Mechanism Underlying the Pathogenesis of Systemic Juvenile Idiopathic Arthritis

Systemic juvenile idiopathic arthritis (sJIA) is a childhood-onset immune disorder characterized by arthritis and systemic inflammatory features, including fever, rash, and lymphadenopathy. sJIA is a subtype of juvenile idiopathic arthritis (JIA), the most common chronic rheumatic disease in childhood. sJIA represents roughly 10% of all forms of JIA and is unique in presenting severe systemic symptoms that often overshadow joint inflammation (1–5). Several findings have led to the consideration of sJIA as a polygenic autoinflammatory disease driven by proinflammatory cytokines related to the innate immune system, such as IL-1β, IL-6, and IL-18 (1, 4, 6–9). A striking feature of sJIA is its association with macrophage activation syndrome, a potentially life-threatening complication of several systemic inflammatory disorders, which is characterized by excessive activation of T cells and macrophages and massive inflammatory responses (3, 10, 11).

The cause of sJIA is largely unknown. One concept is that sJIA results from a relatively harmless trigger in children who cannot adequately suppress immune responses because of genetic or acquired defects (1). Increased susceptibility for sJIA has been linked to polymorphisms in the IL-10 and IL-10 promotor genes associated with lower expression rates of the cytokine (12, 13). Because IL-10 is an anti-inflammatory cytokine with a crucial role in preventing excessive immune responses and autoimmune pathologies (reviewed in Ref. 14), insufficient IL-10 production may cooperate with other cytokine-related abnormalities in the onset of sJIA. In literature, contradictory results are reported regarding IL-10 in sJIA. A decreased IL-10 production in sJIA patients, as compared with healthy controls (HC), was shown when whole blood cells were stimulated with LPS or PHA (15). In contrast, when PBMCs of sJIA patients were left unstimulated or were stimulated with anti-CD3 Abs, anti-CD28 Abs, or PMA, they produced more IL-10 than cells from healthy donors (16). Quantification of IL-10 mRNA in PBMCs and purified monocytes revealed higher IL-10 mRNA levels in sJIA patients than in HC (17, 18). IL-10 plasma levels in sJIA patients were increased compared with HC but were lower compared with levels found in patients with polyarticular JIA (19).

In the current study, we explored the hypothesis that defects in IL-10 may underlie the pathogenesis of sJIA. We took advantage of our recently developed mouse model for sJIA (20) and analyzed IL-10 production in a cohort of patients. The mouse model relies on injection of wild type (WT) and IFN-γ–deficient (IFN-γ knockout [KO]) mice with CFA containing heat-killed mycobacteria. Whereas IFN-γ KO mice develop sJIA-like symptoms, WT mice only experience a subtle and transient inflammatory reaction (20). We found lower IL-10 levels in tissues and cells of CFA-injected IFN-γ KO mice and therefore investigated whether an sJIA-like syndrome would develop in WT mice injected with anti–IL-10R neutralizing Abs. In sJIA patients, we explored possible defects in IL-10 production by analyzing plasma levels in parallel with proinflammatory parameters and by examining cell-specific IL-10 production both ex vivo and after in vitro stimulation.

IFN-γ KO and WT BALB/c mice were bred under specific pathogen–free conditions in the Experimental Animal Centre of KU Leuven. The sJIA mouse model was induced in mice 6–8 wk old by s.c. injection of CFA (Difco) with added heat-killed mycobacteria (1.5 mg/ml) as previously described (20). Mice were euthanized the day on which overt signs of inflammation occurred, between 16 and 21 d after CFA injection. Age- and sex-matched noninjected littermates were included as controls. mAbs against the IL-10R (anti–IL-10R, clone 1B1.2) were purified from hybridoma culture (kindly provided by Dr. O. Leo, University of Brussels, Gosselies, Belgium). Mice were treated i.p. with 200 μg of anti–IL-10R Ab or PBS twice a week, starting on day −1. All experiments were approved by the Ethics Committee of KU Leuven (P182/2014).

A total of 29 patients (18 active sJIA and 11 inactive sJIA patients) and 21 age-matched HC were recruited from the University Hospital of Leuven (patient characteristics in Supplemental Table I). Informed consent was given according to the Declaration of Helsinki, and the study was approved by the Ethics Committee of KU Leuven (S58814). sJIA patients met the classification criteria of the International League of Associations for Rheumatology (2) and were divided according to their disease state into active and inactive patients. Samples from active patients were collected at the occurrence of disease before therapeutic intervention or at disease flares. Inactive patients were defined by the absence of fever, rash, arthritis, and inflammatory parameters.

Murine blood samples were obtained by cardiac puncture with heparin (LEO Pharma); complete blood cell analysis was performed with a Cell-Dyn 3700 apparatus (Abbott Diagnostics). Spleen and lymph nodes were fragmented and passed through a cell strainer to obtain single-cell suspensions. Plasma from patient samples was separated within 2 h after withdrawal of EDTA-anticoagulated blood. Human PBMCs were obtained by Lymphoprep density centrifugation and frozen in liquid nitrogen in cryopreservation medium (90% FBS, 10% DMSO).

Cells were freshly isolated (mouse) or thawed (human) and suspended in RPMI 1640 medium containing 10% FBS, penicillin (100 U/ml; Life Technologies), streptomycin (100 μg/ml; Life Technologies), and 50 μM 2-ME. Cells were stimulated with anti–CD3/CD28 Ab-coated beads (aCD3/CD28) (10 μl/ml; Life Technologies), IFN-γ (150 or 300 ng/ml; PeproTech), CpG oligodeoxynucleotides (CpG ODN2006, 1 μg/ml; Integrated DNA Technologies), LPS (200 ng/ml; Sigma-Aldrich), or recombinant human IL-2 (100 ng/ml; PeproTech) and IL-12 (10 ng/ml; PeproTech) at 37°C and 5% CO₂. For stimulation of osteoclast generation, splenocytes were resuspended in alpha-MEM containing 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml). A total of 5 × 10⁵ cells in a total volume of 400 μl were seeded in chamber slides (Thermo Fisher Scientific) and incubated for 3 d with the osteoclast-differentiating factor receptor activator of NF-κB ligand (RANKL; 100 ng/ml; R&D Systems) and M-CSF (20 ng/ml; R&D Systems). After 3 d, the medium and stimuli were replaced and cells were stimulated for three more days, followed by staining of the cells for tartrate-resistant acid phosphatase (TRAP) as previously described (21). Cells staining red were considered to contain TRAP, and TRAP⁺ cells with three or more nuclei were defined as osteoclasts.

RNA was extracted from tissues using the PureLink RNA Mini Kit (Invitrogen). cDNA was obtained by reverse transcription using SuperScript II Reverse Transcriptase and random primers (Invitrogen) according to the manufacturer’s protocol. mRNA levels were analyzed by quantitative PCR using a TaqMan Gene Expression Assay (Applied Biosystems) and a 7500 Real-Time PCR System apparatus. Expression levels of IL-10 (Mm.PT.58.13531087) were normalized for GAPDH (Mm99999915_g1) RNA expression by the 2^−ΔΔCT method. Murine IL-6 and IL-17 protein levels were measured by ELISA (DuoSet; R&D Systems). IL-10 protein levels were measured by ELISA (DuoSet; R&D Systems) or Meso Scale Discovery (MSD) assay kit (Meso Scale Diagnostics). Human C-reactive protein (CRP), IL-6, and IL-18 levels were measured using MSD according to the manufacturer’s protocol.

Cells were incubated with FcR block (Miltenyi Biotec) and stained with mAbs against mouse CD3 (clone 145-2C11 or 17A2; BioLegend), CD4 (GK1.5; BD Biosciences), CD8 (53-6.7; BD Biosciences), CD11b (M1/70; BD Biosciences), CD115 (AFS98; BioLegend), Ly-6G (1A8; Thermo Fisher Scientific), CD19 (6D5 or 1D3; BioLegend or Thermo Fisher Scientific), CD122 (TM-b1; Thermo Fisher Scientific), CD49b (DX5; BD Biosciences), TCR γδ (eBioGL3; Thermo Fisher Scientific), Foxp3 (REA788; Miltenyi Biotec), TER119 (Ter-119; Thermo Fisher Scientific), CD71 (C2; BD Biosciences), CD34 (RAM34; BD Biosciences), B220 (RA3-6B2; BD Biosciences), CD24 (M1/60; BioLegend), IgM (II41; Thermo Fisher Scientific), IgD (11-26c.2a; BD Bisociences), CD21/35 (7G6; BD Biosciences), CD23 (B3B4; Thermo Fisher Scientific), CD1d (1B1; Thermo Fisher Scientific), and CD5 (53-7.3, BioLegend) or mAbs against human CD3 (SK7, HIT3a; Thermo Fisher Scientific), CD4 (RPA-T4, OKT4; Thermo Fisher Scientific), CD14 (61D3; Thermo Fisher Scientific), CD19 (HIB19; BD Biosciences), CD56 (NCAM 16.2, HCD56; BD Biosciences, BioLegend), CD24 (HL5; BD Biosciences), and CD38 (HB7; Thermo Fisher Scientific). Dead cells were excluded using Zombie Aqua 516 (BioLegend), DAPI (Sigma-Aldrich), or Fixable Viability Dye eFluor 780 (Thermo Fisher Scientific). For intracellular staining, cells were stimulated for 4 h in RPMI 1640 + 10% FBS containing ionomycin (750 ng/ml, Bio-Techne), phorbol 12,13-dibutyrate (500 ng/ml; Bio-Techne) or PMA (100 ng/ml; Sigma-Aldrich), and brefeldin A (2 μg/ml; Bio-Techne). After stimulation, cells were stained for surface markers and viability in the presence of brefeldin A for murine samples. Cells were fixed and permeabilized according to the manufacturer’s protocol (BD Biosciences) and stained intracellularly with Abs against murine IL-10 (JES5-16E3; BD Biosciences) or human IL-10 (JES3-19F1; BD Biosciences). Unstimulated cells were used as negative controls for cytokine staining. Flow cytometry was performed on a BD LSRFortessa ×20 or BD FACSymphony, and sorting was performed on a BD FACSAria III; all were equipped with Data-Interpolating Variational Analysis software. Results were analyzed with FlowJo (V10).

Data were analyzed using GraphPad Prism 7 software. For comparison of two different groups, the two-tailed nonparametric Mann–Whitney U test was used for unpaired data. The Wilcoxon matched-pairs signed rank test was used for paired data. When three or more groups were included, data were first checked for statistical difference with the nonparametric Kruskal–Wallis test (unpaired data) or Friedman test (paired data). For analysis of grouped data, two-way ANOVA followed by Tukey multiple comparisons test was used.

As previously described, CFA induces an sJIA-like disease in IFN-γ KO mice, whereas WT mice develop a more subtle and transient inflammation. Proinflammatory cytokines are induced by CFA and are often higher (i.e., IL-6 and IL-17) in the diseased IFN-γ KO mice (20). In the current work, the production of IL-10, an anti-inflammatory cytokine, was comprehensively studied by different approaches. When mRNA levels of IL-10 were determined in spleen, lymph nodes, liver, and lung tissue, lower levels were found in the diseased IFN-γ KO mice as compared with WT mice, and the differences reached statistical significance in spleen and liver tissue (data not shown). To verify IL-10 production at the protein level, we first performed ELISA on plasma of the mice but failed to detect IL-10. However, when splenocytes and lymph node cells were stimulated in vitro with aCD3/CD28, an approach by which we previously demonstrated increased production of proinflammatory cytokines in the diseased IFN-γ KO mice (20), we found a clear-cut lower production of IL-10 in CFA-challenged IFN-γ KO mice versus WT counterparts (Fig. 1A). We subsequently focused on spleen tissue and analyzed IL-10 production by intracellular flow cytometry in CD11b⁺ cells (which constitute monocytes, macrophages, neutrophils, and dendritic cells), CD4⁺ effector and regulatory T (T_reg) cells, CD8⁺ T cells, γδ T cells, B cells, and NK cells. We observed a significant decrease in IL-10–producing CD4⁺ T_reg cells, B cells, and NK cells in diseased IFN-γ KO CFA mice compared with WT counterparts (Fig. 1B). In the other cell types analyzed, no differences between diseased and control mice were observed (Supplemental Fig. 1A). When IL-10⁺ cell types were plotted with respect to all live cells, B cells were found to be the most abundant IL-10 producers in this mouse model, followed by CD11b⁺ cells and CD4⁺ T cells (Fig. 1C). Analysis of the phenotype of these IL-10–producing B cells in the spleen revealed that they cannot be assigned to one specific B cell subset but are mainly divided across the follicular B cells, marginal zone B cells, and type 1 transitional B cells (data not shown).

Next, we evaluated whether IFN-γ affects the amount of IL-10 produced per cell by taking the median fluorescence intensity of IL-10 within IL-10⁺ cells (Supplemental Fig. 1B). In IL-10⁺ B cells, the median fluorescence intensity of IL-10 was significantly lower in IFN-γ KO mice than in WT mice, both in naive conditions and after CFA injection. Also, in IL-10⁺ CD4⁺ T_reg cells, the amount of IL-10 produced by CFA-injected IFN-γ KO mice was significantly lower compared with their WT counterparts. In the other cell types analyzed, no differences between diseased and control mice could be observed. Thus, in the absence of IFN-γ, B cells and CD4⁺ T_reg cells of CFA-injected mice produce less IL-10 per IL-10⁺ cell.

To further clarify the link between the absence of IFN-γ and a decreased IL-10 production in the CFA-injected mice, splenocytes were stimulated for 72 h with aCD3/CD28 in the presence of two concentrations of IFN-γ (150 ng/ml and 300 ng/ml) (Supplemental Fig. 1C). In agreement with our previous experiment (Fig. 1A), lower levels of IL-10 were found in supernatant of splenocytes derived from IFN-γ KO mice as compared with WT mice. IL-10 levels in IFN-γ KO mice were not increased by adding IFN-γ, suggesting that the defective IL-10 production is not a direct result of the absence of IFN-γ. Interestingly, when enriched splenic CD4⁺ T cells were stimulated with aCD3/CD28, differences in IL-10 production between IFN-γ KO and WT mice were no longer seen (Supplemental Fig. 1D), from which we conclude that despite the stimulation with aCD3/CD28, the defective IL-10 production by total splenocytes cannot be attributed to T cells.

Our results provide evidence for a defective IL-10 production in diseased, CFA-injected IFN-γ KO mice as compared with their WT counterparts, in different organs and immune cells. In the next set of experiments, we investigated whether the lower IL-10 production may underlie sJIA disease development in this mouse model by injecting CFA-challenged WT mice with IL-10R-neutralizing Abs. sJIA-like symptoms and laboratory parameters were recorded and are shown in Fig. 2.

First of all, anti–IL-10R injection resulted in weight loss with an earlier onset than seen in IFN-γ KO mice (Fig. 2A). Furthermore, a significant splenomegaly and lymphadenopathy was seen in CFA-challenged, anti–IL-10R-treated WT mice and was comparable with those of CFA-challenged IFN-γ KO mice (Fig. 2B). Two mice of the anti–IL-10R group with a highly increased spleen weight (Fig. 2B, gray triangles) also developed anemia (Fig. 2C, gray triangles), which was clinically evident by their pale ears, an sJIA feature usually not observed in WT mice (20). Other hematological abnormalities typical of sJIA (i.e., thrombocytosis and neutrophilia) were demonstrated in anti–IL-10R-injected mice (Fig. 2C). In part of the mice, the immature character of RBCs and WBCs was analyzed. CFA-injected IFN-γ KO mice showed an increase in immature TER119⁺CD71⁺ RBC [Fig. 2D and Avau et al. (20)], which was also visible in the anemic, anti–IL-10R-treated WT mouse (Fig. 2D). The percentage of immature CD34⁺ WBCs, determined in the spleen, was equally increased in IFN-γ KO CFA and WT CFA mice treated with anti–IL-10R.

To further determine the parallels between anti–IL-10R-treated WT mice and IFN-γ KO mice subjected to CFA, a flow cytometric analysis was performed on splenocytes. Most CFA-induced changes (i.e., decreased B cells, T cells, and NK cells and increased neutrophils) were observed in WT mice after CFA challenge and were more pronounced in anti–IL-10R-treated WT mice (Fig. 2E). An in-depth analysis of disease mechanisms was performed in part of the mice by measuring IL-6 levels in the plasma and IL-17–producing capacity of lymph node cell cultures stimulated with aCD3/CD28. Although the inflammatory cytokines were elevated in anti–IL-10R-treated mice, levels did not reach those of IFN-γ KO mice after CFA (Fig. 2F).

We previously reported development of arthritis from day 25 onwards in CFA-challenged IFN-γ KO mice but not in WT mice (20). In the current study, WT CFA mice treated with anti–IL-10R did not show any joint inflammation, which may be explained by the transient effect of anti–IL-10R treatment. Indeed, most of the effects of anti–IL-10R treatment were abolished after 2–3 wk, which is probably due to the development of Abs against the injected rat anti–IL-10R IgGs in mice challenged with CFA (an adjuvant that was initially used for the production of Abs in laboratory animals). To detect arthritis at an early time point, we determined the percentage of osteoclast precursor cells in the spleen, which can be considered as a parameter to predict possible development of arthritis (22, 23). To this end, mice were euthanized at day 16 after CFA injection, and osteoclast precursor cells [defined as CD3⁻B220⁻CD11b⁺CD115⁺ cells (22, 23)] were studied by flow cytometry. In CFA-injected IFN-γ KO mice and in part of the anti–IL-10R-treated mice, the percentage of osteoclast precursor cells was increased compared with CFA-injected WT mice (Fig. 2G; NS for anti–IL-10R treatment). In addition, induction of osteoclast generation via stimulation with RANKL and M-CSF was analyzed to confirm this observation (Fig. 2H). In line with the increase in osteoclast precursor cells, a trend toward increased numbers of multinucleated osteoclasts was observed in CFA-injected, anti–IL-10R-treated mice and IFN-γ KO mice, indicating that these mice are more prone to develop arthritis.

To summarize, neutralization of IL-10 signaling in WT mice resulted in more pronounced sJIA characteristics, with striking parallels between CFA-challenged IFN-γ KO mice and anti–IL-10R-treated WT mice (Table I), indicating that endogenous IL-10 plays a protective role in sJIA disease development in this mouse model.

In literature, higher IL-10 plasma levels in sJIA patients as compared with HC have been reported (19, 24). However, increased IL-10 levels in sJIA patients versus HC may be part of a homeostatic response and do not rule out a potential imbalance of IL-10 relative to proinflammatory cytokines. With this in mind, we used a different strategy to study the potential imbalance in IL-10 by analyzing plasma IL-10 levels in relation to known markers of inflammation in sJIA. Levels of IL-10 and the inflammatory molecules CRP, IL-6, and IL-18 were analyzed in plasma of HC, active sJIA patients, and corresponding inactive sJIA patients after treatment (Fig. 3A, 3B). In the active sJIA group, only 2 out of 18 sJIA patients showed an increased IL-10 production, whereas all other patients (active and inactive) had levels comparable with those in HC (Fig. 3A). In contrast, we observed very high levels of CRP and IL-6, general markers for systemic inflammation (25–27), in patients with active sJIA. We also confirmed that IL-18, a cytokine very prominent in sJIA (10), was strongly increased in patients with active disease (Fig. 3B). Of note, although inactive sJIA patients had very low CRP levels comparable with HC, IL-6 and IL-18 levels were still moderately increased (Fig. 3B).

Together, these data show an increased production of proinflammatory cytokines in active sJIA patients, whereas IL-10 production is not increased and comparable with levels in HC. The data are indicative of an imbalance between pro- and anti-inflammatory mediators in sJIA patients, which is visualized in Fig. 3C, showing the ratios of IL-10 to the respective inflammatory mediators.

We searched for the cell-specific defects in IL-10 production by quantifying IL-10 ex vivo in PBMCs from HC, active sJIA patients, and corresponding inactive treated sJIA patients using intracellular flow cytometry (Fig. 4). We investigated IL-10 production by T cells (CD3⁺), NK cells (CD3⁻CD56⁺), monocytes (CD14⁺), and B cells (CD19⁺). Within the T cells, NK cells, and monocytes, we could not detect any differences in IL-10 production between the different groups (Fig. 4). Also, in the rest fraction (CD3⁻CD19⁻CD14⁻CD56⁻), no profound differences could be observed (data not shown). In contrast, in B cells, we observed a nonsignificant decrease in IL-10 production by sJIA patients. Blair et al. (28) showed that within the B cell fraction of healthy donors, CD19⁺CD24^hiCD38^hi cells possess regulatory capacities and are the most important IL-10 producers. Also, other studies point to the regulatory and IL-10–producing capacities of this B cell subset (29–33). Interestingly, when we looked in more detail at IL-10 production by this specific B cell subtype, a significantly impaired IL-10 production was seen in sJIA patients as compared with HC (Fig. 4).

We also studied the potential of PBMCs from HC and active and inactive sJIA patients to produce IL-10 after stimulation with CpG oligodeoxynucleotides (CpG), LPS, aCD3/CD28, and IL-2 + IL-12, triggers known to induce IL-10 production by B cells, monocytes, T cells, and NK cells, respectively (34–37). After 48 h, we measured the amount of IL-10 produced in the supernatant of the stimulated PBMCs (Fig. 5A). Stimulation with CpG, a known inducer of IL-10 in B cells (34), elicited less IL-10 production in sJIA patients when compared with HC, yet it was only significant in the active sJIA patients. IL-10 production induced by the other triggers was unaltered between the different groups. To verify the cell type responsible for this lower IL-10 production after CpG stimulation, we performed intracellular flow cytometry on CpG-stimulated PBMCs. In this study, the data revealed that, despite the high variation, B cells from both active and inactive sJIA patients produced significantly less IL-10 than HC cells after stimulation with CpG (Fig. 5B). To confirm the defective IL-10 production by B cells, PBMCs were sorted in B cells, CD4⁺ T cells, monocytes, and NK cells and stimulated with CpG, aCD3/CD28, LPS, and IL-2 + IL-12, respectively (Fig. 5C). We observed a trend toward a lower IL-10 production by CpG-stimulated B cells of sJIA patients compared with HC, which was only minimally seen in other cell types analyzed.

sJIA is thought to be caused by an excessive inflammatory immune reaction to a generally harmless trigger in predisposed children. We previously demonstrated that injection of IFN-γ KO mice with adjuvant containing heat-killed mycobacteria leads to a chronic immune syndrome with clinical and pathological features reminiscent of sJIA, including a storm of proinflammatory cytokines (20). Although the model is well recognized as a clinically relevant animal model of sJIA (38–40), its applicability may be disputed in view of the absence of documented IFN-γ mutations in sJIA patients. However, in the current study, we demonstrated that upon adjuvant challenge, IFN-γ KO mice acquire cell-specific defects in IL-10, a key cytokine in anti-inflammatory immune responses. The adjuvant-induced IL-10 defects in IFN-γ KO mice are not merely a consequence of the disease but rather underlie disease pathogenesis as neutralization of IL-10 signaling in WT mice leads to an sJIA-like disease as well. In fact, as shown in Table I, there is a striking similarity between disease development in CFA-injected IFN-γ KO mice and CFA-challenged WT mice in which IL-10 is neutralized. The data indicate that the susceptibility of IFN-γ KO mice to develop sJIA may, at least partially, be explained by acquired defects in IL-10 production. The development of sJIA in IFN-γ–competent mice with neutralized IL-10 is in accordance with previous reports of low-expression IL-10 polymorphisms in sJIA patients (12, 13) and can therefore be considered as an alternative, clinically relevant model.

In the sJIA mouse model, less IL-10–producing B cells, CD4⁺ T_reg cells, and NK cells were observed, with B cells being the main IL-10–producing cell type. Although it has been proposed that CD1d^hiCD5⁺CD19^hi cells are the major IL-10–producing B cells in murine splenocytes (41), the IL-10–producing B cells in our model could not be assigned to a particular phenotype. We also observed an almost complete loss of IL-10–producing NK cells in IFN-γ KO mice, which may be in line with the general NK cell defects in the absence of IFN-γ [as originally described by Dalton et al. (42) and confirmed by Avau et al. (20)]. Furthermore, not only were less IL-10–producing cells observed in CFA-injected IFN-γ KO mice, but the amount of IL-10 produced per IL-10⁺ B cell and CD4⁺ T_reg cell was also significantly decreased, suggesting that their regulatory effector functions are diminished in the diseased mice.

The defective IL-10 production in CFA-injected IFN-γ KO mice was a rather unexpected observation because IFN-γ is generally considered as an inhibitor of IL-10 (43–45). However, IFN-γ has been found to stimulate IL-10 production in Th1 cells and neutrophils under certain in vivo conditions (46, 47). Furthermore, we recently reported that IFN-γ significantly increases IL-10 production when mouse and human B cells (from naive mice and healthy donors, respectively) are triggered with the TLR9 ligand CpG. This stimulatory effect of IFN-γ on CpG-induced IL-10 was restricted to B cells and found to be dependent on the MAPKs p38 and JNK (48). Because CFA is an oil adjuvant with heat-killed mycobacteria containing the TLR9 ligand CpG (49), it is intriguing to speculate that the defective IL-10 production by B cells in CFA-injected IFN-γ KO mice, as opposed to WT mice, may be conditioned by CpG.

In previous studies, IL-10 levels in plasma of sJIA patients were analyzed and compared with HC, showing an increased IL-10 production in sJIA patients (19, 24). However, because IL-10 is an important anti-inflammatory cytokine that serves to downregulate inflammatory processes, it seems reasonable that IL-10 production increases together with proinflammatory mediators in sJIA patients. Therefore, it might be more informative to analyze IL-10 levels in comparison with proinflammatory markers in these patients. In contrast to previous publications, we did not observe an increased IL-10 production in plasma of sJIA patients when compared with HC (19, 24). Levels of CRP, IL-6, and IL-18, in contrast, were increased in patients with active sJIA, as reported previously (8, 10, 19, 24). The fact that the increase in proinflammatory cytokine production is not accompanied by an increased IL-10 production points to an imbalance between pro- and anti-inflammatory mediators in sJIA patients and suggests that sJIA patients are not capable of producing sufficient IL-10 to adequately regulate the amount of inflammation they experience.

Human peripheral B cells can be divided into different subsets, mainly depending on their surface expression of CD19, CD24, CD38, and CD27. According to these markers, three main B cell subsets can be identified: immature transitional B cells (CD19⁺CD24^hiCD38^hi), naive B cells (CD19⁺CD24^intCD38^int), and memory B cells (CD19⁺CD27⁺). To date, human B cells with regulatory functions have been essentially described within the CD24^hiCD38^hi B cell subset, which is therefore seen as the regulatory B cell subset (32, 50, 51). CD24^hiCD38^hi B cells are able to produce IL-10 and thereby suppress CD4⁺CD25⁻ T cell proliferation as well as the release of IFN-γ and TNF-α by these cells. In addition, they inhibit naive T cell differentiation into Th1 and Th17 cells and stimulate the conversion of CD4⁺CD25⁻ T cells into T_reg cells, partially through the production of IL-10 (33). In active sJIA patients, we observed a significantly lower amount of IL-10–producing regulatory B cells compared with HC, which is in line with our observation of a defective IL-10 production by B cells in the mouse model. A defective IL-10 production by CD24^hiCD38^hi B cells in sJIA patients may account for the enrichment of both Th1 and Th17 cell populations in the blood of sJIA patients, which was observed both in active and inactive sJIA patients (52). Correspondingly, we also observed a lower number of IL-10–producing CD24^hiCD38^hi B cells in PBMCs of inactive sJIA patients, although not significantly. A defective IL-10 production by CD24^hiCD38^hi B cells has previously been reported in patients with systemic lupus erythematosus and psoriasis after stimulation with CD40L or CpG and CD40L, respectively. The total number of CD24^hiCD38^hi B cells was unaltered in systemic lupus erythematosus patients and even increased in patients with psoriasis (28, 50). In sJIA patients, we observed similar numbers of CD24^hiCD38^hi B cells compared with HC (data not shown). Also, the total percentage of B cells was similar in all groups (data not shown), consistent with the study of Fall et al. (53) but in contrast to Macaubas et al. (54), who reported decreased B cells in sJIA.

In line with the observations ex vivo, B cells from active and inactive sJIA patients produce less IL-10 compared with controls after in vitro stimulation with CpG. Unmethylated CpG motifs are prevalent in bacterial and viral genomic DNA and trigger an immune response by activating TLR9 (34). The reduced IL-10–producing response of sJIA B cells after triggering of TLR9 with CpG might be partially responsible for the inability of sJIA patients to dampen the ongoing inflammation. Indeed, it is suggested that sJIA is caused by an excessive inflammatory reaction to a generally harmless trigger, which often contains CpG motifs, and that inadequate downregulation of immune activation is central to sJIA disease development (1).

In this study, we provided evidence for a defective anti-inflammatory IL-10 response in sJIA, both in the mouse model and in patients. The IL-10 neutralization experiments in the mouse model endorse the hypothesis that these IL-10 defects may underlie disease development rather than being a consequence of the disease. sJIA is considered an autoinflammatory disease, and therefore, most attention has been given to the role of innate immune cells in sJIA pathogenesis. To our knowledge, we are the first to elaborate on the importance of regulatory B cells in sJIA patients and showed an impaired IL-10 production by these cells. Our data allow us to speculate that this defect in IL-10 production plays a role in disease development.

We thank Dr. Ellen Brisse for revising the manuscript.

The authors have no financial conflicts of interest.