IL-12p35 Subunit Contributes to Autoimmunity by Limiting IL-27–Driven Regulatory Responses

The IL-12 family members are heterodimeric cytokines associated with the regulation of innate and adaptive immune responses, operating in bacterial and parasitic infections as well as in autoimmune diseases (1, 2). The release of these cytokines is carefully orchestrated with different family members eliciting opposing effects. Hence, IL-23 (comprising p19/p40 subunits) induces IL-17 differentiation, promoting an inflammatory environment and suppressing anti-inflammatory responses driven by IL-10 (3). In contrast, IL-27 (comprising p28/Ebi-3) has a defined suppressive effect and is pivotal in the induction of regulatory T cell (Treg)1 but also in the maintenance of Foxp3⁺ Tregs, either directly or via the release of IL-10 from APCs (4, 5). The results obtained using p19-deficient mice (lacking IL-23), showing that these mice do not develop experimental autoimmune encephalitis or collagen-induced arthritis (CIA) and that they have impaired delayed-type hypersensitivity (DTH), demonstrate that IL-23 plays a dominant role, over IL-12, in the induction of inflammatory disease (6–8).

Experiments using IL-12p35^−/− mice, which have an untouched IL-23/IL-17 pathway but lack IL-12 (comprising p35/p40) and IL-35 (p35/Ebi-3), have produced perplexing data. Notably, IL-12p35^−/− mice have been shown to develop exacerbated experimental autoimmune encephalitis and CIA compared with wild-type (WT) mice (9, 10). However, they are resistant to developing hapten-induced colitis and Ab-induced arthritis and they cannot clear several infectious diseases, such as Leishmania major and Leishmania donovani, being unable to mount normal DTH response toward the Ags obtained from these intracellular parasites (11–15). In the present study we took advantage of Ag-induced arthritis (AIA), an acute inflammatory arthritis model driven by DTH response, to understand the mechanisms that render IL-12p35^−/− mice incapable of mounting normal inflammatory responses. Our data demonstrate that IL-12p35^−/− mice develop a significantly milder arthritis compared with WT mice. Protection from disease, in the IL-12p35^−/− mice, was characterized by significant increases in IL-27p28 and Ebi-3 mRNA levels, as well as increased IL-27 and IL-10 secretion, compared with WT mice. Higher numbers of regulatory cells, including Tregs and regulatory B cells (Bregs), were also present in IL-12p35^−/− mice compared with WT mice. Increases in anti-inflammatory mediators were mirrored by a decrease in the numbers of proinflammatory cells, including Th17 and Th1. Inflammation was restored following in vivo neutralization of IL-27, but not after administration of IL-12. Importantly, IL-12-p35^−/− mice immunized with collagen in CFA display a significantly decreased production of IL-27, as compared with IL-12-p35^−/− mice with AIA. Our results highlight that in the absence of the IL-12/IL-35 axis, IL-27 exerts a suppressive effect over IL-23–driven inflammation in the AIA model. Moreover, we show that the outcome of disease in the absence of p35 varies according to the experimental model used. Therefore, generalization about the mechanisms of action extrapolated from the use of a single genetically modified animal or single experimental disease model should exercise caution.

Anti-CD3 for culture, anti–CD3-FITC (clone 17A2), anti–CD4-FITC (clone GK1.5), anti–CD19-PE-Cy7 (clone 1D3), anti-CD21 (clone 7G6), anti-CD23 (clone B3B4), anti-CD24 (clone 30-F1), anti–Foxp3-allophycocyanin (clone MF23), anti–IL-10-allophycocyanin (clone JES5-16E3), anti–IFN-γ-allophycocyanin (clone XMG1.2), and anti–IL-17-PE (clone TC11-18H10) were purchased from BD Pharmingen. Anti–IL-27-carboxyfluorescein (clone 234205) was purchased from R&D Systems.

IL-12p35^−/− and C57BL/6J WT mice were purchased from The Jackson Laboratory. These animals were bred and maintained under specific pathogen-free conditions at the animal facility at University College London (London, U.K.). These studies have been reviewed and approved by the UK Home Office.

Female C57BL/6 mice (8–12 wk old) were used in all experiments. Animals were immunized with 100 μg methylated BSA (mBSA; Sigma-Aldrich) dissolved in 50 μl PBS and emulsified in an equal volume of CFA (Sigma-Aldrich), containing 3 mg/ml Mycobacterium tuberculosis (Difco). This was followed, 7 d later, by intra-articular injection of 10 μl 20 mg/ml methylated BSA in the right knee. Additionally, the second knee was injected with PBS alone to observe the swelling due to trauma. Knee swelling was measured using calipers and “Ag-specific” swelling was calculated by deducting the size of the PBS-injected knee from the mBSA-injected knee. Knee joints were removed postmortem and fixed with 10% (w/v) buffered formalin and decalcified in 5% EDTA. After decalcification, the paraffin sections were stained with H&E. Two independent observers evaluated the slides histologically. The gradation of arthritis was scored from 0 to 4 according to the intensity of lining layer hyperplasia, mononuclear cell infiltration, and pannus formation as reported (16): 0, normal; 1, normal synovium with occasional mononuclear infiltrates; 2, a few layers of flat to round synovial lining cells and scattered mononuclear infiltrates; 3, clear hyperplasia of the synovium with three or more layers of loosely arranged lining cells and dense infiltration with mononuclear cells; 4, severe synovitis with pannus and erosion of articular cartilage and subchondral bone.

Female C57BL/6 mice (8–12 wk old) were used in all experiments. CIA was induced as described previously (17). Briefly, animals were immunized with 200 μg chicken collagen emulsified in an equal volume of CFA (Sigma-Aldrich), containing 3 mg/ml M. tuberculosis (Difco). This was followed, 14 d later, by 200 μg chicken collagen emulsified in an equal volume of IFA (Sigma-Aldrich).

C57BL/6 WT and IL-12p35^−/− mice received 100 ng recombinant mouse IL-12 (R&D Systems) i.p. at days −8, −7, −6, −1, 0 and 1 after AIA onset. Injection of PBS was used as a control. For neutralization of IL-27, 20 μg polyclonal goat anti–IL-27 Ab (R&D Systems) was administered i.p 4 h before AIA induction. Polyclonal goat IgG was used as an isotype control.

Cells infiltrating the AIA-affected joint were isolated from the synovial membrane 3 d after disease onset as described (18). Briefly, knee joints were removed and synovial membrane was excised under a dissecting microscope and digested with collagenase A (1 mg/ml; Sigma-Aldrich) and DNAse type IV (150 μg/ml; Roche) for 45 min at 37°C. The cells were then washed and passed through a cell strainer before using for flow cytometry and cell culture.

Intracellular cytokine analysis was performed as previously described (19). Briefly, splenocytes, inguinal lymph node (LN), or synovial cells (3 d after disease onset) were suspended at 5 × 10⁵–10⁶ cells/well and cultured in complete medium with PMA (50 ng/ml; Sigma-Aldrich), ionomycin (500 ng/ml; Sigma-Aldrich), and GolgiPlug (BD Biosciences) for 6 h. Cells were then stained for extracellular markers followed by fixation (PBS/2% PFA), permeabilization (permeabilization buffer from eBioscience), and incubation with anti-mouse IL-10-allophycocyanin, IFN-γ-allophycocyanin, or IL-17-PE mAbs. The cells were acquired with a FACSCalibur flow cytometer (BD Biosciences) and analyzed using FlowJo software.

Blood was obtained postmortem. After blood coagulation, serum was separated and analyzed for levels of anti-mBSA IgG by a sandwich ELISA assay. Microplates were coated with mBSA (Sigma-Aldrich), diluted in PBS, and incubated overnight. Nonspecific binding sites were saturated with 1% gelatin (Sigma-Aldrich) for 1 h. Serial dilutions of tested sera were incubated at 37°C for 2 h. Following washes, peroxidase-conjugated streptavidin detected bound complexes by using 3,3′,5,5′-tetramethylbenzidine (Sigma-Aldrich) and the reaction was stopped by adding H₂SO₄. Absorbance was read at 450 nm. Washing steps were conducted in 0.05% Tween 20 in PBS.

Draining LN and synovial cells were cultured at 5 × 10⁵/well for 48 h without any stimulation or with mBSA (50 μg/ml). Supernatants from cell suspensions were harvested analyzed for levels of secreted cytokines using standard sandwich ELISA kit (R&D Systems) or cytokine bead assay (CBA; Bender MedSystems) performed according to the manufacturer’s instructions.

Cells were cultured at 5 × 10⁵/well for 48 h without any stimulation or with mBSA (50 μg/ml). RNA was extracted using RNeasy columns (Qiagen) and treated with DNase I (Promega). cDNA was obtained using a reverse transcriptase kit (Qiagen). RNA (100 ng) was analyzed for the expression of IL-10, IL-12p35, IL-27p28, and Ebi-3 using TaqMan expression assays and performed according to the manufacturer’s instructions.

LNs were excised postmortem from IL-12p35^−/− and WT mice 3 d after disease onset. Tregs and effector CD4⁺ cells were isolated using a mouse CD4⁺CD25⁺ Treg isolation kit (Miltenyi Biotec). Purity and expression of Foxp3 in CD4⁺CD25⁺ regulatory and CD4⁺CD25⁻ effector cells were determined by FACS and are shown in Fig. 5C. Cells were plated with a total cell number of 2 × 10⁵/well. Effector T cell (Teff)-only wells contained 2 × 10⁵ Teffs/well; Treg-only wells contained 2 × 10⁵ Tregs/well; and mixed wells contained 1 × 10⁵ Teffs/well and 1 × 10⁵ Tregs/well (ratio 1:1). Purified cells were cultured for 60 h without any stimulation or with mBSA (50 μg/ml) or anti-CD3 (1 μg/ml). Supernatants were harvested and analyzed for TGF-β secretion by ELISA (R&D Systems) Additional cultures were left for 96 h and then pulsed overnight with 1 μCi [³H]thymidine, harvested, and counted in a scintillation counter (LKB Instruments).

B cells isolated from the spleens of WT or IL-12p35^−/− mice in the remission phase of AIA (day 12 after disease onset) were stained with CD19, CD23, CD21, and CD24. B cell subsets were FACS sorted using gates drawn according to previous reports (20). Seven days prior to cell transfer/intra-articular injection, all recipients were immunized with mBSA in CFA. Transitional 2-marginal zone precursor (T2-MZP) Bregs and MZB cells (2 × 10⁶) were transferred i.v. to syngeneic WT C57BL/6 mice on the day of intra-articular injection. The control group (no transfer) received a PBS injection.

For the statistical analysis of the data, unpaired t tests and the Fisher test were applied. Results in which p < 0.05 were considered significant.

To study the direct effect of IL-12p35 on acute inflammatory arthritis, IL-12p35^−/− mice and C57BL/6 WT mice were immunized with mBSA with CFA, followed by an intra-articular inoculation of mBSA 1 wk later (16). IL-12p35^−/− mice displayed a significantly reduced joint swelling and total arthritis index score as compared with age and gender matched WT mice (Fig. 1A–C). Histopathological evaluation of the WT mice revealed a substantial accumulation of inflammatory cells in the affected joints, causing severe cartilage damage and bone erosion. In contrast, IL-12p35^−/− mice showed only a minor cell infiltrate and an overall preservation of joint structure (Fig. 1B, 1C). Significantly lower numbers of inflammatory cells were recovered from the affected synovia of IL-12p35^−/− compared with WT mice (Fig. 1D). IL-12p35^−/− mice had reduced levels of mBSA-specific Abs as compared with WT mice (Fig. 1E), despite mBSA-induced proliferation of spleen and draining LN lymphocytes having been comparable in both mouse strains (Fig. 1F).

To gain more insight into the potential mechanisms modulating the development of AIA in IL-12p35^−/− mice, we assessed the release of cytokines associated with inflammation in the draining LNs. In response to in vitro mBSA restimulation, lymphocytes from the draining LN of IL-12p35^−/− mice displayed significantly reduced IFN-γ production compared with lymphocytes from WT mice (Fig. 2A). However, the expression of IL-23p19 mRNA (Fig. 2B) was not significantly affected. Additionally, the production of IL-23 measured by ELISA (data not shown) or assessed by intracellular staining (Supplemental Fig. 1A) was also not changed. Further cytokine analysis revealed that significantly lower amounts of IL-17, IL-6, and IL-1β were released in IL-12p35^−/− compared with the levels measured in WT mice (Fig. 2C). In contrast, increased levels of the anti-inflammatory cytokines IL-10 and IL-5 were observed in IL-12p35^−/− mice as compared with WT mice (Fig. 2C).

To assess the cellular source of cytokine production, we next analyzed T cell responses by intracellular FACS staining and demonstrated that CD4⁺ T cells from IL-12p35^−/−-immunized mice display reduced differentiation into Th1 and Th17 cells (Fig. 2D, 2E).

Next, we studied how the lack of IL-12p35 influenced the cytokine profile at the site of inflammation, namely the synovial membrane. Th1 or Th17 cells were virtually absent in the synovial infiltrate of IL-12p35^−/− mice, whereas they were abundantly present in the inflamed joints of WT mice (Fig. 2F–I). No detectable levels of IL-23 could be measured in the synovia of WT or IL-12p35^-/ mice (data not shown).

CD4⁺CD25⁺Foxp3⁺ Tregs were found with increased frequency in both the LN and at the site of inflammation in IL-12p35^−/− mice compared with WT mice (Fig. 3A–C). In addition to the increased frequency, CD4⁺CD25⁺Foxp3⁺ Tregs infiltrating the synovia of IL-12p35^−/− mice expressed increased levels of Foxp3 (Fig. 3D), which has been correlated to increased suppressive activity of regulatory cells (21).

It has been shown that IL-35 is released by CD25⁺Foxp3⁺CD4⁺ Tregs and contributes to their development and suppressive function (22). Despite their inability to release IL-35, CD4⁺CD25⁺Foxp3⁺ Tregs isolated from IL-12p35^−/− mice efficiently suppressed in vitro mBSA-mediated proliferation of CD4⁺CD25⁻ Teffs isolated from the draining LN of WT mBSA-immunized arthritic mice (Fig. 3E). These results are in agreement with those previously published showing that Tregs in Ebi-3^−/− mice, also lacking IL-35, are normal during Listeria monocytogenes infection (23). However, in agreement with the data of Vignali and colleagues (24), the suppressive capacity of IL-12p35^−/− Tregs was significantly reduced when effector cells were polyclonally stimulated with anti-CD3 (Supplemental Fig. 1B, 1C). Therefore, these data suggest that IL-35^−/− Tregs retain their capacity to suppress if activated in the presence of triggering autoantigen. Of interest, the lack of IL-35 secretion appeared to be balanced by an increased production of TGF-β during Ag-specific responses (Fig. 3E).

IL-12–producing Bregs have been shown to protect mice from colitis (25). Next, we assessed whether, in addition to Tregs, the lack of IL-12p35 also affected the number and function of T2-MZP Bregs previously identified as the predominant Breg subset in arthritic disease (20, 26–28). At day 3 after disease onset, we observed an increase in the frequencies and absolute numbers of T2-MZP Bregs and decreased numbers of marginal zone MZB cells in IL-12p35^−/− mice as compared with WT mice (Fig. 4A, 4B). The latter B cell subset has been previously attributed with pathogenic properties (29). The number of follicular B cells remained unchanged in both groups (Fig. 4C). Transfer of T2-MZP Bregs isolated from WT or IL-12p35^−/− mice, obtained from the spleen at the remission stage of AIA, to WT syngeneic mice with arthritis show equivalent suppressive capacity, thus excluding a role of IL-12p35, IL-12, or IL-35 in the suppressive function of T2-MZP Bregs (Fig. 4D).

Other groups have demonstrated that IL-23 is pivotal in the differentiation of Th17 cells and in the pathogenesis of autoimmunity (30). However, our data so far support a previously unappreciated role for IL-12p35 in governing DTH in this model of acute inflammation and joint damage. The IL-12 family of heterodimeric cytokines includes IL-12 (comprising p40/p35), IL-23 (p40/p19), IL-27 (p28/Ebi-3), and IL-35 (Ebi-3/p35). The lack of IL-12p35 ablates IL-12 and IL-35 production, but it may indirectly impinge in the release of IL-23 and IL-27. Two conceivable explanations could be put forward to explain the reduced inflammatory response, in favor of the regulatory environment, observed in IL-12p35^−/− mice. First, the IL-12p35 subunit, via the release of bioactive IL-12, contributes to the differentiation of an inflammatory environment. Second, owing to the promiscuous sharing of the same chain, we suggest that in the absence of the IL-12p35 subunit, Ebi-3 may be more “available” for coupling to p28, resulting in increased production of IL-27. This may overshadow bioactive IL-23 in the IL-12p35^−/− mice and ultimately block Th17 differentiation in this acute model of inflammatory arthritis.

To determine whether exogenous rIL-12 could restore the proinflammatory cascade in IL-12p35^−/− mice, murine rIL-12 was given i.p. at days −1, 0, and 1 of mBSA/CFA and intra-articular immunizations as previously described (31). Administration of IL-12 increased IFN-γ secretion in IL-12p35^−/− mice (Fig. 5A). However, reconstitution with rIL-12 failed to restore the production of IL-23 or IL-17 in IL-12p35^−/−mice (data not shown) and arthritis, as shown by a very mild inflammation and infiltration in the affected joint (Fig. 5B–D). These results suggest that the proinflammatory effect of the IL-12p35 subunit is not due to coupling with the p40 subunit resulting in the release of bioactive IL-12.

IL-27 is a member of the IL-12 family of heterodimeric cytokines that comprises the Ebi-3 and p28 chains. IL-27 plays an ambiguous role, as it can promote as well as inhibit autoimmune disease and DTH responses (32, 33). We next investigated the effect that the lack of IL-12p35 subunit has on the expression levels of Ebi-3 and p28, unique to IL-27. mRNA was isolated from the draining LN and analyzed for the expression of Ebi-3 and IL-27p28. Expression of mRNA encoding for both Ebi-3 and p28 was upregulated in the IL-12p35^−/− mice compared with WT mice (Fig. 6A). We also noted a significant increase in the frequencies of IL-27–producing cells in IL-12p35^−/− compared with WT mice as well as an increase in secreted IL-27 as measured by ELISA (Fig. 6B, 6C). Previous reports have identified the major producers of IL-27 as dendritic cells and B cells (34, 35). In our experiments we also identified the CD19⁺ B cell subset as producers of IL-27 (Fig. 6D). However, the major producers of IL-27 during AIA have yet to be identified, but we can confirm these cells to be CD19⁻, CD3⁻, CD11b⁻, and CD11c⁻ (data not shown).

Next, we tested whether we could restore AIA in IL-12p35^−/− mice by blocking the release of IL-27 in vivo. IL-12p35^−/− and WT mice were treated with anti–IL-27 or with an irrelevant isotype control at the time of mBSA/CFA immunization. In vivo neutralization of IL-27 resulted in a restoration of arthritis and cellular infiltration, causing joint damage to levels observed in WT mice (Fig. 7A–C).

IL-27 drives the differentiation of CD4⁺ T cells producing IL-10, leading to a decrease of Th17 differentiation (36). In vivo neutralization of IL-27 restored Th17 responses in the draining LN of IL-12p35^−/− mice to WT levels (Fig. 7D). Release of IFN-γ and IL-23 was not affected by the anti–IL-27 treatment (data not shown). Interestingly, neutralization of IL-27 production led to a significant decrease in IL-10 production in IL-12p35^−/− but not in WT mice (Fig. 7E, 7F). Moreover, anti–IL-27 administration impaired the accumulation of CD4⁺Foxp3⁺ cells in the draining LN of IL-12p35^−/− mice (Fig. 7G). The findings showing that neutralization of IL-27 in WT mice did not affect the frequencies of Th17, Tregs, or the levels of IL-10 produced in the draining LNs, suggesting that in the presence of an intact IL-12p35 pathway, the protective function of IL-27 is overshadowed by the proinflammatory effect of other members of this family, including IL-23.

One of the many outstanding questions is why during CIA, IL-12p35^−/− mice develop an exacerbated disease, whereas the same mice undergoing a different immunization regimen, for example, AIA, produce to opposite results. It has been previously shown that the CIA can be ameliorated upon IL-27 administration (37, 38). Because we have demonstrated that IL-27 is crucial in restraining inflammation in this model, we hypothesized that IL-27 could be a pivotal factor determining disease versus resistance in the two different models of arthritis. Comparison of IL-27 production in CIA, as well as in AIA, highlighted that during AIA, IL-12p35^−/− mice produce higher levels of IL-27 than do WT animals (Fig. 6B–D), whereas the scenario is reversed in CIA, with IL-12p35^−/− animals showing a reduced level of IL-27 compared with WT animals (Fig. 7H, 7I). The differential IL-27 response may help to explain the complexity of why IL-12p35^−/− animals are resistant to AIA, Ab-induced arthritis, and hapten-induced colitis (11, 14) and yet exhibit exacerbated autoimmune disease, including CIA (7).

The discrepancy between studies showing that animals lacking the IL-12p35 subunit can either be protected from or develop exacerbated arthritis and DTH responses led us to investigate how the IL-12p35 subunit contributes, if at all, to inflammation and disease. Our results have shown that in the AIA model of arthritis, mice lacking the IL-12p35 subunit, and hence IL-12 and IL-35, were unable to mount an inflammatory response and were protected from arthritis. We have demonstrated that the lack of the IL-12p35 subunit leads to an increase in the release of IL-27. The upregulation of IL-27 production resulted in further differentiation of immunoregulatory cells and secretion of IL-10, contributing to the generation of an anti-inflammatory environment.

IL-27 was originally reported as cytokine promoting Th1 development and secretion of IFN-γ (39, 40). However, more recently IL-27 has been defined as an important anti-inflammatory mediator that can induce both Treg1 and Foxp3⁺ Tregs and additionally inhibit the induction of Th17 cells (34, 41, 42). Our results also show that the increase in IL-27 is followed by an upregulation of IL-10 and regulatory cell differentiation, which was inhibited after in vivo neutralization of IL-27. Taken together, our results demonstrate that in the absence of the opposing signals driven by IL-12 and IL-35, the suppressive effect of IL-27 overpowers the pathogenic capacity of IL-23 (3, 43, 44).

We have also demonstrated that during AIA, IL-27 was increased in the absence of IL-12p35, as compared with WT animals, whereas in CIA IL-27 was reduced. This goes toward explaining why different groups have observed differential responses to inflammatory and infectious disease models (7, 9, 11, 12, 14, 45). We do not claim that the data presented in this study account for all of the differences observed in different mouse model of arthritis. This could be due to the different genetic backgrounds used (C57BL/6 versus DBA/1), animal husbandry practices, different modality of immunizations, differential time of onset, and the capacity to resolve inflammation. The differential production of IL-27 provides a mechanistic explanation of why IL-12p35^−/− mice are protected from AIA, Ab-induced arthritis, hapten-induced colitis, and L. major- and L. donovai-induced DTH response (11, 13, 14). More tailored experiments are needed for future studies to unravel the mechanisms that coordinate the differential responses in the various models of arthritis.

Several studies have identified IL-23 among the potent inducers of IL-17 and critical in the development of several autoimmune diseases (46–49). We showed that the levels of IL-23 were not significantly affected by the lack of IL-12p35. Whereas our data do not dispute the involvement of IL-23 in the pathogenesis of arthritis, or in other autoimmune disorders, they show that in the absence of IL-12/IL-35 cytokines, IL-27 promotes a tolerogenic environment despite the presence of IL-23. Our results showing higher expression of both p28 and Ebi-3 early after the onset of the inflammatory reaction, in the absence of the IL-12p35 subunit, may suggest two scenarios. First, IL-12 may directly suppress IL-27 production. Second, the Ebi-3 chain (which couples with IL-12p35 to give rise to IL-35) is more available to react with p28, thus explaining an increase in IL-27. Our results demonstrating that the administration of IL-12 fails to inhibit IL-27 (data not shown) and to restore disease support the latter explanation. However, our findings also show that IL-27 is only upregulated following immunization, demonstrating that this cascade of events takes place following activation. It has been shown that the TLR4 agonist LPS upregulates the transcription of p28 and Ebi-3 in a MyD88-dependent fashion (50, 51). Therefore, we cannot exclude that the increase in the expression of p28 and Ebi-3 genes is due to the effect of mycobacteria, a component of CFA.

Previous results have demonstrated that IL-35 mediates the suppressive function of regulatory CD4⁺CD25⁺Foxp3⁺ Tregs (22, 24). However, whereas Foxp3⁺ Tregs isolated from IL-12p35^−/− mice are unable to suppress Teff responses in vitro following anti-CD3 stimulation (Supplemental Fig. 1C and Ref. 24), they have retained the capacity to suppress in response to autoantigen stimulation. These results suggest that IL-35 produced by Tregs may downregulate nonantigen-specific responses, but that during autoreactive response this is left to other components of the “immunoregulatory machinery.” In support of this hypothesis, our studies have shown that despite being unable to produce IL-35, Tregs accumulate in the synovia of IL-12p35^−/− mice where they are likely to contribute to the control of inflammatory arthritis. Additionally, these Tregs express higher levels of Foxp3, previously shown to be “proportional” to the strength of suppression (21). Furthermore, IL-12p35^−/− CD25⁺CD4⁺ Tregs obtained from the draining LN secreted higher levels of TGF-β than did WT Tregs. Previous work has shown that IL-12p35^−/− mice are resistant to developing Ab-induced arthritis and that the mechanism for this was attributed to an increase of TGF-β levels in the joint, as compared with WT mice (14). Collectively, we suggest that IL-10 and TGF-β not only compensate for the lack of IL-35, but they are more (or equally) effective in controlling acute inflammatory arthritis. It would be very interesting to investigate the in vivo effects of adding rIL-35 to the IL-12p35^−/− mice once this reagent is easily and cheaply available.

In addition to affecting Treg homeostasis and function, the lack of IL-12p35 also altered the numerical balance of different B cell subsets involved in the maintenance of tolerance versus pathogenicity. The lack of the IL-12p35 subunit increased the number of T2-MZP Bregs, previously shown to be the major subset of B cells producing IL-10 in the context of CIA and the subset that efficiently suppresses arthritis and other arthritic disorders (20, 26, 27). Interestingly, the number of pathogenic MZB cells was reduced. MZB cells are implicated in autoreactive B cells responses (52, 53), and autoreactive MZB cells are considered as a pathogenic hallmark in various models for arthritis (29, 54). Therefore, the reduction of autoreactive MZB cells in favor of T2-MZP Bregs is likely to contribute to the generation of a tolerogenic environment. Our findings in this study also demonstrate that IL-12 is not essential for the suppressive function of T2-MZP Bregs in arthritis, unlike colitis (25). Furthermore, Vignali and colleagues (24) demonstrated that both IL-12p35 and Ebi-3 are expressed at high levels in B cells; however, our results suggest that IL-35 is not necessary for Breg suppressive function.

We have shown that IL-12p35^−/− animals have a complex phenotype where neither the lack of IL-12 nor the lack of IL-35 seems to be as important as the increase in both the levels and dominance of IL-27 following induction of AIA. Our data, although at odds with some of the current literature describing a redundant function for the IL-12 p35 chain in inflammation, have highlighted an important function of IL-27 and show that even in the presence of an intact IL-23 pathway IL-27 has an overpowering anti-inflammatory effect. The long-term ability of patients treated with these drugs, for example with ustekinumab, an anti–IL-12/IL-23p40 mAb currently used as a treatment for psoriasis with some success (55), to combat infection, cancer, and allergy remain to be seen. This poses a note of warning for therapy aimed at the neutralization of shared cytokine chains, as it may result in unpredicted and wide-reaching immunological effects, including skewing the immune response and modulation of Treg and Breg numbers.

We thank Dr. Mark Wesley for instruction on the use of the Bender MedSystems CBA technique and Jamie Evans for help with FACS sorting. We also thank Dr. P. Blair for critically reviewing this manuscript.

The authors have no financial conflicts of interest.