Introduction
IL-17 has recently been implicated in the pathogenesis of multiple autoimmune diseases, including rheumatoid arthritis (RA) and the mouse model collagen-induced arthritis (CIA). Patients with RA have higher levels of IL-17 in their serum and synovial fluid than normal controls or patients with osteoarthritis (OA) [
1‐
3]. IL-17-producing Th17 cells are present in the T cell-rich areas of RA synovium [
4] and induce the expression of receptor activator of NF-kB ligand (RANKL), which aids bone resorption [
2,
5,
6]. Furthermore, high levels of mRNA for IL-17 and TNF-α in the RA synovium are predictive of joint damage progression, while high levels of interferon (IFN)-γ mRNA are predictive of protection from damage [
7]. These findings indicate that IL-17 is a key pathogenic cytokine that is relevant to the downstream events associated with autoimmune joint inflammation. In addition, studies that have employed strategies to up-regulate, neutralize or delete IL-17 have shown, quite consistently, that Th17 cells have a pathogenic role in CIA [
8‐
10].
RA and CIA are complex diseases with requirements for systemic and target organ specific T cell and B cell activation, and these processes are positively and negatively regulated by multiple cytokine networks.
In vitro studies show that Th17 development is down-regulated by IFN-γ and IL-4, cytokines derived from Th1 and Th2 cells, respectively [
11,
12].
The role of IFN-γ in animal models of arthritis is complex, with evidence for both protective and pathogenic functions. Previous studies have found that mice deficient in either IFN-γ or IFN-γ receptor develop more severe CIA than wild type counterparts [
13‐
16]. Proteoglycan-induced arthritis, on the other hand, is dependent on IFN-γ and independent of IL-17 [
17,
18]. IFN-γ clearly has the ability to induce inflammation in some settings, but it can also inhibit Th17 differentiation and thereby reduce inflammation. The net effect of IFN-γ may depend on the phase of disease and the location - such as the joint versus the spleen or lymph node. By administering neutralizing antibodies at different time points, one study suggested that IFN-γ has pathogenic effects in the early phase of disease but protective effects in the later stages [
19]. Although this study did not measure IL-17, one plausible interpretation of these results is that IFN-γ possibly takes on a protective role after Th17 cells become overabundant and highly pathogenic.
Similar to IFN-γ, evidence for the role of IL-4 in arthritis is complex. IL-4-based interventions can prevent or alleviate joint inflammation and bone damage in multiple animal models of arthritis [
20‐
22]. We have shown previously that systemic injection of dendritic cells genetically engineered to produce IL-4 (IL-4 DCs) attenuates CIA [
21]. Further mechanistic studies revealed that IL-4 secreted from IL-4 DCs is a potent suppressor of IL-17 production by T cells from the early phase of CIA [
23]. These results suggest that endogenous IL-4 could also play a protective role in arthritis by suppressing IL-17 in the early phase of disease. However, it leaves open the possibility that it could also have pathogenic effects by suppressing production of IFN-γ, once IFN-γ has taken on a protective role. In addition, IL-4 reduces bone damage in established CIA, and is necessary for the development of arthritis, possibly due to the important role of IL-4 in B cell activation and antibody production [
20,
24]. Thus, like IFN-γ, IL-4 may have both protective and pathogenic roles in CIA, depending on the stage of disease, location of IL-4 production and relative abundance of other cytokines.
This suggests that in vivo the balance of IFN-γ, IL-4 and IL-17 is important in the pathogenesis of CIA. The experiments described in the current paper were designed to further test this hypothesis, in a CIA model in which not all immunized mice develop clinical arthritis. We measured systemic cytokine levels at several time points, as well as in vitro cytokine production from lymphoid organs and joints during the peak of disease. Our data shows that disease correlates with the systemic IL-17/IFN-γ ratio rather than the absolute level of any single cytokine. Administration of neutralizing antibodies to IFN-γ and/or IL-4 differentially altered the cytokine responses and course of disease.
Materials and methods
Mice
Male 8- to 10-week-old DBA1 mice (Jackson laboratories, Bar Harbor, Maine, USA) were housed in specific pathogen free conditions. All procedures were approved by the University Committee for the Use and Care of Animals of the University of Michigan.
Collagen-induced arthritis
Complete Freund's adjuvant (CFA) was prepared by mixing heat inactivated mycobacterial strain H37Ra in incomplete Freund's adjuvant at 4 mg/ml. Lyophilized chicken collagen (Chondrex, Redmond, WA, USA) was dissolved overnight in acetic acid at 4 mg/ml. CFA and collagen were mixed at 1:1 to form an emulsion. 100 μg of collagen was injected intradermally at the base of the tail. Mice were scored for arthritis every other day from day 15 after immunization.
Scoring was performed as follows: 0 = no swelling or redness of paws or digits; 1 = swelling and redness in one to two digits; 2 = swelling and redness over ankle or three or more digits or midfoot; 3 = swelling and redness over ankle and midfoot or digits and midfoot; 4 = swelling and redness over entire foot or ankylosis.
Neutralizing antibody protocol
For these experiments, neutralizing rat antibodies to mouse IFN-γ (clone R46A2) or IL-4 (clone 11B11) were purified from hybridomas (ATCC, Manassas, VA, USA) and used at 100 μg/mouse/per day. Neutralizing antibody to IL-17 (clone M210) was a kind gift from Amgen (Thousand Oaks, CA, USA). The antibodies were injected intraperitoneally from day 10 to 20. Rat IgG at 100 μg/mouse/day was used as a control.
Tissue collection and assays
Mice were sacrificed by CO2 inhalation. Blood was collected by cardiac puncture into serum separator tubes, and serum was frozen at -80°C for cytokine assays to be performed at a later date. For some assays 100 μl of blood was collected serially from tail bleeds on days 0, 14, 28 and 42. Spleens and inguinal lymph nodes were collected and single cell suspensions of these tissues were made and used in re-stimulation assays to assess antigen-specific responses. Re-stimulation was performed by culturing single cell suspensions of spleens or lymph nodes for five days with 100 μg/ml of chicken collagen. Supernatants were collected at day 5 of culture and analyzed for various cytokines. Paws were collected by incising at the fur line. The paws were cut up into small pieces and cultured in medium overnight at 37°C. Supernatants were collected for various cytokine assays.
ELISA
IFN-γ, IL-4 and IL-17 were measured by ELISA. Plates were coated with anti-IFN-γ antibody (clone R46A2 or XMG1.2, Biolegend, San Diego, CA, USA), anti-IL-4 antibody (clone11B11) or anti-IL-17 antibody (clone TC11-18H10.1, Biolegend, San Diego, CA, USA). Plates were blocked and then loaded with tissue culture supernatants or serum. The plates were washed and developed with biotin conjugated detection anti IL-17 antibody (clone TC11-8H4, Biolegend, San Diego, CA, USA), anti-IFN-γ detection antibody (clone XMG1.2 or R4-6A2 Biolegend, San Diego, CA, USA), or anti IL-4 detection antibody (clone BVD6-24G2, BD Pharmingen, San Jose, CA, USA) and streptavidin horseradish peroxidase followed by tetramethylbenzidine (TMB). Colorimetric intensity was then quantitated in a Biorad (Hercules, CA, USA) ELISA plate reader using KC4 software (Biotek, Winooski, VT, USA). IL-10 ELISA was performed using a kit from BD Pharmingen (San Jose, CA, USA), following the manufacturer's protocol.
Flow cytometry
Splenocytes or single cell suspensions of draining inguinal lymph nodes were cultured with collagen overnight and stained with fluorescent labeled anti-IL-17 (clone TC11-18H10.1, Biolegend (San Diego, CA, USA), anti-CD4 (clone GK1.5, Biolegend, San Diego, CA, USA), anti-IFN-γ (clone XMG 1.2, Biolegend, San Diego, CA, USA), and anti-IL-4 (clone 11B11, Biolegend, San Diego, CA, USA) antibodies following six-hour stimulation with phorbol 12-myristate 13-acetate (PMA)/ionomycin/BrefeldinA. The cells were then analyzed in FACS Calibur and data analyzed using Cell Quest software (BD, San Jose, CA, USA).
Histologic scoring
Mouse hind paws were used for histology scoring. The paraffin-embedded tissue was sectioned in an axis longitudinal to the tibia. Three sections from the center of each paw were stained with H&E and scored by two independent blinded observers. Inflammatory infiltrate, synovitis (synovial hyperplasia), cartilage destruction and bone involvement were each scored on a scale of 0 to 3. 0 = no change, 1 = mild, 2 = moderate and 3 = severe.
Statistical analysis
Serum cytokine analysis was performed for each mouse in triplicate. For some experiments the ELISA data for an entire group were pooled and expressed as mean +/- standard deviation. ELISA assays on culture supernatants were performed in triplicate. Data are presented as mean +/- standard error of the mean. Significance was analyzed by using the Student's t-test or analysis of the variance.
Discussion
In the past few years the pathogenic role of IFN-γ in immune-mediated diseases such as RA and CIA has been called into question. Several studies have shown increased IL-17, as well as inflammatory mediators induced by IL-17, in the RA synovium [
2,
4‐
6,
34‐
38]. IL-6 + transforming growth factor (TGF)-β, IL-21 and IL-23 are important in the generation, expansion and maintenance of Th17 cells [
28,
39‐
42]. All of these Th17-associated cytokines are found in RA synovial tissue. In addition, IL-17 can synergize with TNF-α and IL-1, two cytokines that are known to play an important role in the pathogenesis of RA. Thus there is considerable interest in new strategies to inhibit Th17 cells, and the safest approach is likely to be one that restores the immune homeostasis and T cell subset balance, thereby minimizing autoimmune inflammation without crippling anti-microbial and anti-tumor responses.
IFN-γ and IL-4 suppress Th17 differentiation
in vitro, as well as secretion of IL-17 from committed Th17 cells. Thus IFN-γ and IL-4 represent endogenous mechanisms for regulating Th17-mediated inflammation, which may play a role in preventing or controlling autoimmunity [
11,
12,
23]. In fact, knocking out IFN-γ worsens CIA and renders resistant strains of mice susceptible to disease, which is associated with increased IL-17 production [
25,
26]. Furthermore, we have found that treatment with exogenous IL-4, in the form of DCs genetically engineered to produce IL-4, suppresses IL-17 and reduces the incidence and severity of CIA [
21,
23].
Although there have been significant advances in understanding the development and maintenance of Th17 cells in vitro, the endogenous regulation of Th17 responses during the development of arthritis is still under investigation. Various antigenic stimuli can trigger IL-17 responses in vivo and not all of them will result in systemic or organ specific autoimmunity in animal models, implying that endogenous regulation of IL-17 responses is important in the prevention or attenuation of autoimmunity. Herein we present data on the regulation of IL-17 responses by the Th1 cytokine IFN-γ and the Th2 cytokine IL-4 in CIA.
Our studies show that the level of systemic IL-17 is not directly associated with arthritis, but the ratio of systemic IL-17/IFN-γ is an important predictor of target organ damage (Figures
1a to
1c). These results suggest that disease outcome is not determined solely by the absolute level of the pathogenic cytokine, but rather by the balance between pathogenic and protective signals. How these competing signals regulate disease pathogenesis at the molecular level is not clear. One possibility is that these signals modulate trafficking of Th17 cells to the joint, either by altering expression of chemokines by cells of the synovium or expression of chemokine receptors by T cells. Once in the joint, Th17 cells can then induce inflammation and recruitment of other inflammatory cells. Interestingly, after immunization similar high levels of IL-17 were detectable in the serum of both arthritic and non-arthritic animals, but IL-17 was only found in arthritic joints (Figures
1a and
1d). Non-arthritic paws from arthritic mice or paws from non-arthritic mice do not have detectable IL-17. In addition, arthritic joints had higher levels of IFN-γ, and IL-4 than non-arthritic joints, suggesting that once target organ inflammation is initiated there is recruitment of both inflammatory and anti-inflammatory cell types. Further studies are needed to determine the effect of the systemic Th1/Th17 balance on T cell homing and recruitment to the joint.
Our results implied that the balance between Th1 and Th17 cells played an important role in disease outcome, so neutralizing antibody to IFN-γ was administered to perturb this balance. Consistent with previous data suggesting that IFN-γ negatively regulates IL-17 responses and clinical arthritis, mice that had received anti-IFN-γ antibody had accelerated arthritis associated with elevated levels of IL-17 (Figures
2a and
2b). The arthritic paws from these mice had elevated levels of IFN-γ, IL-4, and IL-17 (Figure
2c). A similar elevation of IFN-γ, IL-4, and IL-17 was seen in arthritic joints in CIA mice not treated with cytokine-neutralizing antibodies (Figure
1d). This indicates that the systemic cytokine response is distinct from that in the target organ and the balance of Th1/Th17 response is critical in the systemic immune events leading up to target organ damage. Once the target organ damage is initiated, the IFN-γ, IL-4, and IL-17 responses within the arthritic joint, in the presence or absence of anti-IFN-γ, are similar. This is consistent with a recent study demonstrating that Th17 cells were predominant in the draining inguinal lymph nodes whereas IL-17-producing γδ T cells were predominant in the inflamed joint [
27].
Although increased systemic IL-17 production was associated with accelerated disease, the increase in systemic IL-4 (Figure
2b) was potentially surprising, in view of previous reports of the suppressive effects of IL-4-based therapies on arthritis and the known ability of IL-4 to suppress IL-17 responses
in vitro. However, the appearance of IL-4 in this situation could represent a back-up mechanism for immune regulation that was able to emerge only with neutralization of IFN-γ. To evaluate the protective or pathogenic role of IL-4 in accelerated CIA, neutralizing anti-IL-4 antibody was administered in conjunction with neutralizing anti-IFN-γ antibody. This resulted in significantly increased severity and accelerated onset of arthritis over mice that received neutralizing antibodies to IFN-γ alone (Figure
3a). Consistent with previous reports, mice that received neutralizing antibody to IL-4 alone did not have an accelerated course of arthritis [
32]. Thus endogenous IFN-γ seems to play a more prominent role than IL-4 in down-regulating arthritis. This could be due to the fact that the immune response to collagen in DBA mice is primarily Th1 and Th17, with undetectable levels of IL-4 in the serum or supernatants of collagen-stimulated lymphoid organs (data not shown). However, because IFN-γ is a potent suppressor of Th2 as well as Th17 development, neutralizing IFN-γ allows for the unmasking of the Th2 response. Therefore, the experiments involving the administration of neutralizing antibodies to both IFN-γ and IL-4 suggest a secondary protective role of endogenous IL-4 in CIA.
As both IFN-γ and IL-4 suppress IL-17
in vitro, one would expect that the increased severity of arthritis seen with anti-IFN-γ + anti-IL-4 would be associated with increased IL-17. However, there was no further increase in the serum levels of IL-17 in the anti-IFN-γ + anti-IL-4 groups, in comparison to the anti-IFN-γ group (Figure
3b). In addition, the arthritis in the anti-IFN-γ group was associated with significantly elevated IFN-γ, IL-4, and IL-17 levels in the joints, whereas the arthritis in the anti-IFN-γ + anti-IL-4 group was associated with significant increase in IFN-γ and IL-4 and a modest increase in IL-17 responses (Figure
3c).
Th17 cells that did not co-express IL-10 were found to have a higher pathogenic potential in the mouse model of multiple sclerosis than Th17 cells that expressed IL-10 [
33]. It is possible that the increased arthritis in the presence of anti-IFN-γ + anti IL-4 is associated with the generation of a more aggressive phenotype of Th17 cells, one that may be associated with reduced levels of IL-10. However, IL-10 levels were higher in the paws of mice that received anti-IFN-γ + anti-IL-4 than in paws of mice that received anti-IFN-γ alone (Figure
3d). This would suggest that the phenotype of Th17 responses in CIA in the absence of IL-4 is independent of IL-10.
As mentioned earlier, IL-4 possibly exerts a protective role in CIA through effects on cartilage and bone. There was a similar degree of inflammatory infiltrate and synovitis in the absence of IFN-γ alone or in the absence of IFN-γ + IL-4 (Figures
4 and
5). However, there was more bone and cartilage destruction in the absence of IFN-γ + IL-4, suggesting a more aggressive phenotype of IL-17 responses in the absence of both Th1 and Th2 responses (Figures
4 and
5). It is also possible that IL-4 could have direct protective effects on bone and cartilage damage, independent of regulation of IL-17. Intra-articular delivery of adenoviral vector associated Th2 cytokines, IL-4 and IL-13, has been shown to slow bone and cartilage damage in rat adjuvant arthritis [
43,
44]. In addition, IL-4 can directly down-regulate osteoclastogenesis through inhibition of RANKL activity [
45,
46]. Interestingly, in patients with RA a polymorphism in the IL-4 receptor that results in reduced responsiveness to IL-4 is associated with rapidly erosive disease, suggesting that IL-4 plays a protective role in RA [
47].
The role of endogenous IFN-γ and IL-4 in the differential regulation of IL-17, leading up to arthritis, was evaluated in experiments involving administration of anti-IL-17 antibody along with anti-IFN-γ or anti-IFN-γ + anti-IL-4 antibodies. Administration of anti-IL-17 antibody completely abrogated the arthritis associated with anti-IFN-γ alone (Figures
6a and
6d). In contrast, the augmented arthritis with anti-IFN-γ + anti-IL-4 antibodies was only partially suppressed with anti-IL-17 antibody (Figures
6a and
6d). Interestingly, although the anti-IFN-γ group had similar levels of serum and elevated levels of paw IL-17 in comparison to the anti-IFN-γ + anti-IL-4 groups (Figures
3b and
3c) and the arthritis was completely dependent on IL-17, the numbers of Th17 cells
in vivo were not increased (Figures
6a, c and
6d). This suggests that IFN-γ plays a major role in the regulation of IL-17 secretion and has only a modest effect on the number of Th17 cells
in vivo. The anti-IFN-γ + anti-IL-4 group had an elevated number of Th17 cells
in vivo (Figure
6c) and yet did not have increased levels of serum and paw IL-17 levels (Figures
3c and
3d), suggesting that the Th17 cells generated under this condition produced less IL-17 on a per cell basis.
Further, the neutralization of endogenous IFN-γ or IFN-γ + IL-4 each lead to joint inflammation by distinct pathways, one completely dependent on IL-17 and the other only partially mediated by IL-17 (Figure
6a). The cytokine responses within the arthritic joints are also different: anti-IFN-γ is associated with elevated IL-17, whereas anti-IFN-γ + anti-IL-4 is associated with a less striking elevation of IL-17 (Figure
3c). These data provide insight into the heterogeneity of systemic as well as joint-specific immune events underlying inflammatory arthritis.
Another intriguing finding was the significant level of IL-4 in the arthritic joints (Figures
1d,
2c and
3c), in the absence of detectable amounts of IL-4 in the lymphoid organs and only small amounts of IL-4 in the serum (Figures
2b and
3b). The levels of IL-4 correlated with levels of IL-17 in the joints (Figure
1e). It is possible that the source of IL-4 in the sera and the joints may be different - T cells in the peripheral circulation versus mast cells in arthritic joints. Whatever the source of IL-4 in the joints, our results suggest that its role in CIA is primarily regulatory rather than pathogenic, at least in the absence of IFN-γ.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SS and LAC performed the experiments, analyzed data and prepared the manuscript; PW, DBD and JE provided technical help with the experiments; JMJ and MJW performed the scoring of histopathology slides; and DAF reviewed experiment design, data and manuscript.