Inhibition of Th17 Cells Regulates Autoimmune Diabetes in NOD Mice

NOD/LtJ and NOD-RAG^−/−mice (NOD.129S7-Rag1^tm1Mom/J) were obtained from Jackson Labs (Bar Harbor, ME). For remission and transplant studies, spontaneously diabetic NOD females were identified by monitoring our colony two to three times per week, with animals considered diabetic after two consecutive blood glucose readings >18 mmol/l or one reading >25 mmol/l with a One Touch Ultra Glucometer (LifeScan, Mississauga, ON, Canada). All animals were cared for according to the guidelines of the Canadian Council on Animal Care, and ethical approval was obtained from the animal welfare committee at the University of Alberta.

A mouse anti–IL-17 monoclonal antibody (clone 50104), mouse IgG2A isotype control antibody, and IL-25 (IL-17E) were all obtained from R&D Systems (Minneapolis, MN).

For studies examining IL-17 neutralization, anti–IL-17 or isotype control was administered at 100 μg i.p. on alternating days over a 12-day period (six total injections). For studies using IL-25, the recombinant cytokine was administered at 1 μg s.c. daily for 25 days (an equivalent volume of saline was given to vehicle-administered animals).

Spontaneous diabetic mice were assigned randomly to one of the following treatment groups: anti–IL-17 (100 μg i.p. every other day), IL-25 (1 mg s.c. daily), or control (IgG at 100 μg i.p. for anti–IL-17 and vehicle for IL-25). Treatment was continued until eight consecutive daily readings >25 mmol/l were obtained, after which point the experiment was terminated.

For anti–IL-17 studies, transplanted animals received either anti–IL-17 or isotype control treatment (100 μg i.p.) on days 0, 4, 8, 12, and 16 posttransplant. For IL-25 studies, transplanted animals received either IL-25 (1 μg s.c.) or vehicle daily through day 15 posttransplant.

At 1 month after the completion of treatment, pancreatic tissue was harvested, fixed in formaldehyde, processed, and embedded in paraffin (20). Sections (10 μm) were stained using hematoxylin and eosin. All samples were blinded before being scored by a single pathologist using the scheme outlined by Yoon et al. (21). Representative islets were photographed using a Zeiss microscope at ×200 magnification.

Before immunostaining, cryostat sections (10 μm) of pancreata were fixed in acetone and blocked using 20% goat serum. Rat anti-mouse CD4 (clone GK1.5; Ebioscience, Mississauga, ON, Canada), rat anti-mouse CD8 (clone 53–6.7; Ebioscience), rat anti-mouse Foxp3 (clone FJK-16s; Ebioscience), and polyclonal guinea pig anti-insulin (Dako, Mississauga, ON, Canada) were used as primary antibodies. To detect bound antibodies, biotinylated goat anti-rat antibodies (for CD4, CD8, and Foxp3; Cedarlane) and tetramethylrhodamine isothiocyanate–labeled goat anti–guinea pig antibodies (1:200, for insulin; Cedarlane) were used. Polyclonal rabbit anti–IL-17 (Cedarlane Labs, Mississauga, ON, Canada) was used to detect IL-17 on formalin-fixed sections. All slides were mounted using ImmunoGold mounting medium with 4′,6-diamidino-2-phenylindole for nuclear counterstaining (Invitrogen, Mississauga, ON, Canada).

Recombinant GAD65 was prepared as previously described and kindly provided by Dr. John Elliott (22). Serum was harvested via tail vein bleeds from treated animals, and GAD65 autoantibody levels were determined using the method described by Ma et al. (23). Serum samples were diluted 1:50 in blocking buffer. Reactions were stopped after 20 min using 1 mol/l H₂SO₄and immediately analyzed at 450-nm wavelength.

Mouse islets were isolated using established methods (24). Spontaneously diabetic NOD females were maintained on daily insulin injections for 2–3 weeks to collect a sufficient cohort to transplant, with insulin withdrawal the day before islet transplantation. A total of 500 NOD-RAG^−/−islets (syngeneic but without insulitis) were transplanted under the left kidney capsule.

An enzyme-linked immunosorbent assay (ELISA) for TGF-β1 and enzyme-linked immunospot (ELISpot) kits for IFN-γ, IL-4, and IL-17 were purchased from Ebioscience. At 1 month after the completion of treatment, animals were euthanized, and peripheral blood mononuclear cells were purified from spleen or pancreatic lymph nodes using Lympholyte-M (Cedarlane). An aliquot of cells was stained for Treg cell activity using the murine CD4/CD25/FoxP3 flow cytometry kit available from Ebioscience. A total of 1,000,000 splenocytes (or 2 × 10⁵pancreatic lymph node cells) were incubated with 20 μg/ml GAD65 in RPMI medium supplemented with 10% FCS (Invitrogen) for 48 h. After this incubation period, ELISpot plates were washed and processed according to the manufacturer's protocols. Supernatants were harvested for TGF-β1 ELISA and processed according to the manufacturer's instructions. Developed plates were read using an ImmunoSpot ELISpot reader, and the total number of spots per well was quantified using ImmunoSpot 4.0.17 software (Cellular Technology, Shaker Heights, OH). Lymphocytes from individual animals in each experimental group were analyzed in triplicate. Four-color flow cytometric analysis was carried out using Cell Quest Pro (BD Biosciences, Mississauga, ON, Canada).

At 1 month after the completion of treatment, splenocytes were harvested and purified with Lympholyte-M. CD4-positive T-cells were extracted from splenocyte suspensions using magnetic beads (negative selection; Miltenyi Biotec, Mississauga, ON, Canada) and confirmed to be >90% pure by flow cytometry (data not shown). Naïve NOD-RAG^−/−males received either 1 × 10⁷splenocytes from a spontaneously diabetic NOD mouse (“diabetic splenocytes”), 1 × 10⁷diabetic splenocytes combined with 2 × 10⁶CD4-positive splenocytes from normoglycemic NOD mice previously treated with anti–IL-17, or 1 × 10⁷diabetic splenocytes combined with 2 × 10⁶CD4-positive splenocytes from normoglycemic NOD mice previously treated with IL-25. Blood glucose levels were monitored three times per week thereafter.

All statistical analyses in this study were carried out using SigmaPlot 10 and SigmaStat 3.5 (Systat), and results are expressed as the means ± SE. Mann-Whitney rank-sum tests were used, and ANOVA performed on ranks with Bonferroni post hoc analysis was used to analyze multiple groups. Kaplan-Meier survival analyses were compared using the log-rank test.

To explore the potential contribution of Th17 cells to the natural development of type 1 diabetes, either neutralizing anti–IL-17 antibodies or IL-25 were administered to 5-week-old NOD females to investigate the role of this population in the initiation phase of autoimmunity and to 10-week-old NOD females to investigate the role of Th17 cells during the effector phase of autoimmunity. Dosing regimens for anti–IL-17 (100 μg i.p. days on alternating days for 2 weeks) and IL-25 (1 μg s.c. each day for 25 days) were based on results obtained in the EAE model, which effectively controlled antigen-specific Th17 cells (15,18). As shown inFig. 1,A, IL-17 neutralization did not alter diabetes progression when treatment was initiated at 5 weeks of age, and a similar result was obtained using IL-25 beginning at 5 weeks of age (Fig. 1,C). In contrast, both anti–IL-17 (Fig. 1,B) and IL-25 (Fig. 1 D) prevented diabetes development in the majority of treated animals by 6 months of age when treatment was initiated at 10 weeks of age (P = 0.001 by log-rank test for each group vs. controls; n = 10 per group). These data suggest that Th17 cells are involved in the natural progression of type 1 diabetes, particularly during the effector phase of disease development.

To further understand the mechanism of diabetes prevention using anti–IL-17 or IL-25 treatment during the effector phase of type 1 diabetes development, detailed histological analysis was carried out in normoglycemic animals 1 month after the completion of treatment (serial sections from six to eight animals per group). Insulitis scoring by a pathologist using blinded samples indicated that both treatment strategies significantly reduced the degree of islet inflammation (Fig. 2). The majority of the pancreata in normoglycemic control animals had destructive insulitis (mean score of 2.6 ± 0.3), whereas insulitis was significantly reduced in both anti–IL-17–treated (mean score 2.0 ± 0.3) and IL-25–treated animals (1.5 ± 0.3; P < 0.05 for anti–IL-17, and P < 0.02 for IL-25 vs. controls; representative hematoxylin and eosin–stained sections from each cohort are shown inFig. 2,B–D). To determine the contribution of different T-cell populations within the peri-islet infiltrates in both anti–IL-17– and IL-25–treated animals, immunofluorescence staining for CD4, CD8, Foxp3, and IL-17 was carried out in combination with insulin staining and compared with control animals. As shown inFig. 3, insulitic lesions in normoglycemic control NOD mice were composed primarily of CD4-positive cells, with a smaller proportion of CD8-positive lymphocytes and low Foxp3 staining. Examination of multiple sections of control NOD pancreata for IL-17 staining consistently showed a small number of IL-17–positive cells within the peri-islet infiltrates, regardless of the degree of insulitis. Pancreatic sections from normoglycemic animals treated with either anti–IL-17 or IL-25 demonstrated a reduced level of insulitis that was composed primarily of CD4-positive cells, a reducednumber of CD8-positive cells, and enrichment in Foxp3-positive cells compared with control animals. Very few IL-17–positive cells were observed in anti–IL-17–treated animals, with only a small proportion of inflamed islets staining positive for IL-17 (representative sections inFig. 3). Analysis of multiple sections from IL-25–treated animals for IL-17 staining did not reveal any IL-17–positive cells (representative section inFig. 3). Overall, these data indicate that inhibition of Th17 cells with both anti–IL-17 and IL-25 treatment regimens significantly reduces islet-specific inflammatory T-cell infiltration and increases the proportion of Foxp3-positive cells around the islets.

Inhibition of Th17 populations in other models of autoimmunity has resulted in a measurable reduction in autoantibody formation (25,26). To determine the impact of anti–IL-17 and IL-25 therapies on autoreactive B-cells in NOD mice, serum samples were analyzed for anti-GAD65 autoantibodies, a later-stage marker of autoimmunity in this model (27). Initially, serum samples obtained from anti–IL-17– and IL-25–treated animals 1 month after the completion of treatment (effector phase) were analyzed, and a significant reduction in GAD65 autoantibodies was observed in both anti–IL-17–treated (mean optical density 1.11 ± 0.02) and IL-25–treated (1.03 ± 0.03) animals compared with controls (1.34 ± 0.08; P < 0.02 vs. anti–IL-17, and P < 0.005 vs. IL-25 by ANOVA) (Fig. 4 A). Based on these data, we chose to prospectively monitor IL-25–treated animals for GAD65 autoantibody levels to determine whether this observation persisted over time (serum samples were not collected prospectively in anti–IL-17 studies and thus could not be analyzed). Animals treated with IL-25 exhibited a persistent reduction in GAD65 autoantibody formation up to 90 days after the completion of treatment (P < 0.001 vs. control by ANOVA). These data indicate that inhibition of Th17 cells in type 1 diabetes can prevent autoantibody formation.

In the NOD mouse model, the initiation and effector phases of disease, before the onset of hyperglycemia, carry the lowest threshold for disease prevention (rev. in (28)). However, once the autoimmune response has matured and resulted in hyperglycemia, reversal of diabetes and prevention of recurrent autoimmunity after β-cell replacement represent significant barriers, with only a few therapeutic strategies regulating type 1 diabetes in the NOD mouse at these late-stage disease time points (28). Thus, to understand the role of Th17 cells after the development of overt diabetes, a series of experiments were conducted in two different models: at the time of diabetes onset (attempt to reverse new-onset diabetes) and after a period of rest after diabetes onset with subsequent β-cell replacement via syngeneic islet transplantation (recurrent autoimmunity). Data inFig. 5,A illustrates that anti–IL-17 had no effect once diabetes was established, with all animals remaining persistently diabetic throughout the treatment period. However, daily treatment with IL-25 resulted in remission in 90% of treated animals, versus none of the controls (P < 0.0001 by ANOVA, and P = 0.002 by Fisher's exact test) (Fig. 5 B). Ultimately, most animals returned to hyperglycemia by 10 days after initiation of treatment, despite ongoing therapy, although one animal did exhibit persistent normoglycemia for >100 days even after IL-25 treatment withdrawal at day 30 (data not shown). This enhanced efficacy of IL-25 compared with anti–IL-17 was also observed in recurrent autoimmunity after syngeneic islet transplantation, where IL-25 nearly doubled islet graft survival time from 4.2 ± 0.8 days in control animals to 7.2 ± 0.2 days in treated animals (P = 0.0013 by log-rank test). These studies indicate that IL-25, which is known to directly inhibit Th17 populations, is superior to IL-17 neutralization in regulating a mature autoimmune response after the onset of hyperglycemia.

Although both anti–IL-17 and IL-25 therapies were able to reduce the incidence of type 1 diabetes during the effector phase leading into type 1 diabetes, only IL-25 therapy was able to control diabetes once the disease was established. To further investigate the different mechanisms by which these two therapies function, splenocytes from normoglycemic treated animals in the prevention studies (Fig. 1) were examined ex vivo for autoreactive T-cell populations using GAD65-stimulated ELISpot assays at 1 month after the completion of treatment. Although no difference in IFN-γ–secreting GAD65-responsive splenocytes was observed compared with controls (Fig. 6,A), a significant reduction in IL-4–secreting GAD65-responsive splenocytes was observed in IL-25–treated animals compared with both anti–IL-17–treated and control animals (P < 0.02) (Fig. 6,B). Paradoxically, anti–IL-17 treatment resulted in an increased frequency of IL-17–secreting GAD65-reponsive splenocytes, whereas the opposite occurred after IL-25 treatment, where a significant reduction in this autoreactive Th17 population was observed (P < 0.001 for anti–IL-17 vs. control and IL-25, and P < 0.05 by ANOVA for IL-25 vs. control) (Fig. 6 C).

Next, a series of adoptive transfer experiments was carried out using immunodeficient NOD-RAG^−/− recipients. In this model, transfer of 1 × 10⁷splenocytes from a recent-onset diabetic NOD mouse results in hyperglycemia in all recipients (control mean diabetes onset at 42.8 ± 2.3 days post-transfer) (Fig. 6,D). To evaluate whether the protective effects of anti–IL-17 or IL-25 treatment could dominantly control autoreactive T-cell populations, 2 × 10⁶purified CD4-positive splenic T-cells, harvested from either anti–IL-17 or IL-25 treated animals 1 month after the completion of treatment, were coinjected with 1 × 10⁷diabetic splenocytes into naïve normoglycemic NOD-RAG^−/−recipients. In these experiments, cotransfer of CD4-positive splenocytes from animals previously treated with anti–IL-17 resulted in no delay in diabetes development (anti–IL-17 mean diabetes onset at 41.8 ± 4.8 days), demonstrating that anti–IL-17 treatment does not alter the CD4-positive T-cell compartment sufficiently to regulate effector diabetogenic splenocytes. In contrast, CD4-positive splenocytes harvested from animals previously treated with IL-25 exerted a dominant protective effect against diabetogenic splenocytes, resulting in diabetes prevention in 70% of reconstituted recipients by 80 days post-transfer (Fig. 6 D).

Based on the results obtained in these adoptive transfer studies, further experiments were conducted to characterize the splenic- and pancreatic lymph node–derived lymphocyte populations in animals previously treated with either anti–IL-17 or Il-25. Flow cytometric analysis of these different lymphocyte sources revealed a significant increase in CD4/CD25/Foxp3-positive Treg cells compared with vehicle-treated controls (Fig. 7), whereas no difference in the relative frequency of CD8-positive T-cells was observed (data not shown). To further understand the function of these lymphocytes, splenic- and pancreatic lymph node–derived lymphocytes were incubated with GAD65 for 48 h, and afterward the supernatant was assayed for TGF-β1 activity, a key cytokine that drives expansion of the Treg cell population. Data inFig. 7 C demonstrate that prior treatment with either anti–IL-17 or IL-25 resulted in a significant increase in TGF-β1 activity from pancreatic lymph node lymphocytes, with IL-25 treatment resulting in greater TGF-β1 activity even when compared with animals previously treated with anti–IL-17. No difference in TGF-β1 activity was observed among splenocyte populations within the three treatment groups. Taken together, these experiments indicate that IL-25 treatment alters the T-cell repertoire in NOD mice, reducing the frequency of autoreactive GAD65-responsive Th2 and Th17 cells and promoting the development of a dominantly protective CD4-positive T-cell population with a relative enrichment in Treg cell populations.

The recently characterized pathogenic Th17 population has been linked to a number of organ-specific autoimmune diseases (29) and is currently being investigated as a clinical therapeutic target in autoimmunity. The current study demonstrates for the first time that inhibition of Th17 cells, either with neutralizing anti–IL-17 or with recombinant IL-25, can impact the course of diabetes in NOD mice, indicating that the Th17 population is a major contributing factor in this model. Initially, the impact of these treatments was investigated in two age-groups that correlate with different stages in the development of autoimmunity. The impact on diabetes was most evident when treatment with anti–IL-17 or IL-25 was introduced during the active phase of autoimmunity preceding the development of overt disease. The observation that these treatments could reduce autoimmune β-cell destruction was confirmed by scoring of insulitis and measurement of GAD65 autoantibodies, which was significantly reduced in both anti–IL-17– and IL-25–treated animals. Taken together, these data indicate that the Th17 population is less involved in the initiation of autoimmune diabetes and primarily contributes to the active phase of disease development in this model.

Further characterization of the peri-islet T-cell infiltrates revealed that both anti–IL-17 and IL-25 treatments reduced the degree of CD4- and CD8-positive infiltrates while increasing the proportion of Foxp3-positive cells, a marker of the Treg cell population (Fig. 3). These data were confirmed by flow cytometric analysis of splenocytes and pancreatic lymph node lymphocytes, which revealed that either treatment resulted in a relative increase in the Treg cell population (Fig. 7). Interestingly, the frequency of Th17 cells within the pancreas was very low at all stages of disease, with only a small fraction of the peri-islet mononuclear cell infiltrate staining positive for IL-17, the characteristic marker of this population (Fig. 3). We observed that treatment with anti–IL-17 reduced the frequency of peri-islet IL-17–positive cells even further, whereas treatment with IL-25 resulted in no identifiable IL-17–positive infiltrating cells (Fig. 3). The relatively low frequency of Th17 cells within the insulitic lesions suggests that this population is more involved in directing an immune response in the secondary lymphoid tissues rather than participating directly in β-cell destruction. This concept is supported by data using the EAE model, which demonstrates that pathogenic Th17 cells facilitate the development of autoreactive effector T-cells during antigenic priming within cervical draining lymph nodes and that the Th17 population remains dominant within this compartment throughout the course of disease (30,31). Our finding that inhibition of Th17 cells with anti–IL-17 or IL-25 leads to enrichment in peri-islet Foxp3-positive cells and pancreatic lymph node lymphocytes is also consistent with the Th17/Treg paradigm because it is known that these populations exert mutually opposing effects (7). Thus, our data suggest that inhibition of Th17 cells allows the Treg cell population to expand and counteract the development of autoimmunity in this model.

To explore the impact of Th17 inhibition on the autoimmune response in more detail, B- and T-cell reactivity to the later-stage autoantigen GAD65 was analyzed (27). Previous treatment with either anti–IL-17 or IL-25 prevented GAD65 autoantibody formation, and this effect was durable over time (Fig. 4). This result is in keeping with data generated in a mouse model of inducible autoimmune myasthenia gravis, which demonstrated that IL-17 knockout animals are resistant to autoantibody formation primarily as a result of the loss of help signals from Th17 cells during the generation of autoreactive B-cells (32). Data in the current study also indicate that inhibition of the Th17 population resulted in a similar effect in autoreactive T-cells. Ex vivo analysis of splenocytes from animals previously treated with anti–IL-17 or IL-25 using ELISpot assays established that inhibition of Th17 cells alters the frequency of GAD65-responsive T-cell populations (Fig. 6). Surprisingly, no difference was observed in the frequency of GAD65-responsive IFN-γ–secreting splenocytes between both treatment groups and controls, despite the marked difference in disease outcomes as observed inFig. 1. Our data indicate that treatment with these Th17-inhibitory therapies resulted in enrichment in Treg cell populations, which likely control the pathogenic effects of autoreactive Th1 cells in the vicinity of the islet.

As expected, treatment with IL-25 significantly reduced the frequency of GAD65-responsive IL-17–secreting cells (Fig. 6,C), which is consistent with previous studies illustrating that this cytokine can inhibit formation of the Th17 population (18). The opposite effect was observed in splenocytes harvested from animals previously treated with anti–IL-17, which possessed a significant increase in the number of GAD65-responsive IL-17–secreting cells (Fig. 6,C). The relative importance of this expanded Th17 population in these anti–IL-17–treated animals is unclear because this treatment prevented diabetes development (Fig. 1 B). A possible explanation for these data are the findings that IL-17 secretion exerts autocrine and paracrine effects on Th17 cells, sending a negative feedback signal to inhibit proliferation (33). Thus, IL-17 neutralization could in theory remove that negative signaling event and enhance the development of Th17 cells. If this were the only effect of anti–IL-17 treatment, one would anticipate that autoimmunity would be accelerated in treated animals, rather than prevented, as shown in this study and others (9,14,15). Another possibility involves the “Sprent effect,” where anticytokine monoclonal antibody treatment paradoxically results in enhanced receptor signaling (34). Further investigation of the mechanism of anti–IL-17 therapy on autoimmunity in animal models and ongoing clinical trials will likely reveal the dominant pathway involved in modulation of autoimmunity using this approach.

The most intriguing finding in the current study relates to the potent effect of IL-25 treatment on the mature autoimmune response in NOD mice after the onset of hyperglycemia. It is well known that therapeutic interventions that can prevent diabetes in this model rarely mediate late-stage autoimmunity, resulting in reversal of disease (28). We explored the impact of both anti–IL-17 and IL-25 treatment using two different models of mature autoimmune responses, either immediately after disease onset or after a period of rest and subsequent syngeneic islet transplantation. As shown inFig. 5, IL-25 therapy restored normoglycemia in newly diabetic animals and significantly delayed recurrent autoimmunity after islet transplantation, whereas anti–IL-17 had no effect in either setting. These findings were particularly surprising given the short serum half-life of IL-25 after subcutaneous injection, with a peak concentration at 1 h and no detectable levels by 6 h postinjection (see supplemental Fig. A1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08–1113/DC1). It is likely that the therapeutic effect of IL-25 treatment persists longer because it inhibits Th17 populations and thus may result in a protracted immunological impact. Measuring this effect ex vivo after IL-25 treatment directly would be challenging; however, adoptive transfer studies using CD4-positive splenocytes harvested from animals previously treated with IL-25 indicated that IL-25 induces the formation of a dominant protective T-cell population that can control the mature autoimmune response, even after treatment withdrawal (Fig. 6 D). Our data indicate that this result was likely due to a reduction in Th17 cells, expansion of Treg cells, and/or a combination of both effects. That anti–IL-17 had no effect on the mature autoimmune response in the current study is not all that surprising, given that it only neutralizes the effector molecule of the Th17 population without sending any direct negative signals. Thus, if the Th17 population has already developed and expanded, neutralizing the effector molecule IL-17 will have minimal impact on the course of disease. In fact, as mentioned previously, IL-17 has been shown to exhibit negative feedback to the Th17 population (33), so neutralizing this cytokine at later disease points may be disadvantageous.

In summary, the current study indicates that the Th17 population is involved in the pathogenesis of autoimmune diabetes in the NOD mouse and that intervention with treatments that inhibit this subset can alter the course of disease, even after the autoimmune response has evolved into overt diabetes. Based on these data, further exploration of this subset in the NOD model and in patients with type 1 diabetes is warranted. The current study also indicates that further ongoing development of agents like IL-25 that can directly inhibit the Th17 population will be superior to approaches using IL-17 neutralization in autoimmune diabetes. Whereas recombinant IL-25 may not represent the best therapeutic molecule because of its short half-life and potential role in allergic airway inflammation (35), this study and others provide proof of concept that this therapeutic target should be developed further for use in autoimmune disease (18).

This work was supported by Juvenile Diabetes Research Foundation (JDRF) Clinical Center Grant 4-2001-920. J.A.E. was supported by fellowships from the American Society for Transplantation, JDRF, Alberta Heritage Foundation for Medical Research (AHFMR), and by the Rhind Autoimmunology Award. S.M. was supported by studentship awards from the AHFMR and the Canadian Institutes for Health Research. C.T. was the recipient of a grant from the Swiss National Science Foundation, an FS Chia Scholarship, and an AHFMR fellowship. A.M.J.S. is an AHFMR scientist and was supported through grants from the JDRF and National Institutes of Health.

No potential conflicts of interest relevant to this article were reported.