Synergistic Reversal of Type 1 Diabetes in NOD Mice With Anti-CD3 and Interleukin-1 Blockade: Evidence of Improved Immune Regulation

Animal experiments were approved by the Yale Institutional Animal Care and Use Committee. Female NOD/ShiLtJ or NOD/scid mice were purchased from The Jackson Laboratory and kept in a specific pathogen-free environment. Nonfasting blood glucose was measured every other day with an Easy Check monitor (Home Aide Diagnostics, Deerfield Beach, FL) beginning at 12 weeks of age. Mice were considered diabetic when blood glucose exceeded 250 mg/dL on at least one of two consecutive determinations. Treatment was started on the day of the second high glucose reading, designated as day 0.

F(ab′)₂ fragments of hamster anti-mouse CD3 mAb (clone145–2C11; ATCC, Manassas, VA) and hamster IgG, as a control for the anti-CD3 mAb (BioXCell, West Lebanon, NH), were prepared by Rockland Immunochemicals (Gilbertsville, PA). The fragments were administered to diabetic mice at a dose of 50 μg/mouse i.p. daily for 5 days (a dose that achieved remission in 50–70% of mice) (35–37). IL-1RA (Anakinra; Amgen, Thousand Oaks, CA) was administered, 10 mg/mouse i.p. daily, for 5 days mixed with the respective F(ab′)₂ before injection. The mice were randomly assigned to one of four treatment groups: 1) anti-CD3 F(ab′)₂ plus placebo (control protein; a gift from Dr. C. Dinarello; 2) IL-1RA + placebo control F(ab′)₂; 3) anti-CD3 mAb + IL-1RA; and 4) placebo + control F(ab′)₂. Glucose levels were measured every other day, and the mice were considered to be in remission if blood glucose levels were <200 mg/dL. In other experiments, an anti-mouse IL-1β mAb (Novartis, Switzerland) was used at a dose of 75 μg/mouse on days 0, 2, and 4.

Circulating cytokines were measured on day 2 after commencement of treatment in plasma by a Milliplex Map kit (Millipore, Billerica, MA) and Bioplex analyzer (Bio-Rad, Hercules, CA). Some mice were killed on days 3 or 7 after diagnosis and others were studied at day 30. Cells from spleens or pancreatic lymph nodes were counted and analyzed for expression of cell surface markers (CD4, CD8, CD19, CD11b, CD11c, CD25, and ICOS) and intracellular molecules (FoxP3) using mAbs from BD Pharmingen (Franklin Lakes, NJ) by flow cytometry (FACSCalibur, BD Immunocytometry Systems, San Jose, CA). Flow data were analyzed with FlowJo software (TreeStar, Ashland, OR). Splenocytes from some mice were also harvested aseptically and activated with phorbol 12-myristate 13-acetate (PMA) (50 ng/mL) and ionomycin (500 ng/mL) (Sigma, St. Louis, MO) for 24 h. Supernatants were collected and cytokines were measured as above. Pancreatic lymph nodes were isolated during collagenase digest, and NRP-specific CD8⁺ lymphocytes were identified by staining with phycoerythrin-conjugated H-2K^d tetramer with NRP-V7 peptide KYNKANVFL, a mimotope of islet-specific glucose-6-phosphatase catalytic subunit–related protein (IGRP) (38). TUM (KYQAVTTTL)/H-2K^d tetramer was used as a negative control (39,40). In separate experiments, day 7 splenocytes were positively sorted using anti-CD11b, or anti-CD11c magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany), and lysed for measurement of arginase 1 (Arg1) and inducible nitric oxide synthase (iNOS) gene expression by quantitative PCR (qPCR).

A total of 10 million spleen cells from nontreated diabetic NOD mice (effector cells) were cotransferred intravenously with 5 × 10⁶ spleen cells from mice that received either anti-CD3 mAb + IL-1RA or anti-CD3 mAb alone or placebo into immunodeficient NOD/scid recipients. The mice were monitored twice a week for development of hyperglycemia and declared diabetic when the glucose level was >250 mg/dL.

Pancreata were removed on days 7 or 30. They were fixed in formalin and stained with hematoxylin and eosin. An insulitis score for each islet was determined (0 = no insulitis; 1 = periinsulitis; 2 = invasive insulitis with <50% islet area affected; 3 = invasive insulitis with >50% islet area affected). Between 36 and 59 islets were scored from three to six mice in each treatment group. Other pancreata were snap-frozen and minced in precooled acid-ethanol for 48 h. Mouse insulin was measured by ELISA h (Crystal Chem, Downers Grove, IL) (41). Pancreatic proinsulin content was measured in acid-ethanol extracts using a rat/mouse proinsulin ELISA kit (Mercordia AB, Uppsala, Sweden). In other mice, pancreata and spleens were homogenized in lysis buffer, and RNA was isolated using the RNeasy Mini Kit (Qiagen Sciences, Germantown, MD). cDNA was prepared, and gene expression was measured by RT-PCR using the primers shown in Supplementary Table 1. Expression of genes of interest was normalized to actin.

Statistical analysis was performed using GraphPad Prizm, Version 5 (GraphPad Software, San Diego, CA). One-way ANOVA or Kruskal-Wallis test with Dunnett or Dunn multiple comparison posttests were used to evaluate differences between the groups, with P < 0.05 considered significant. For survival curves, a log-rank test was used. The displayed data were pooled from a minimum of three replicate experiments. Each data point presented in scatter plots was from a single mouse.

To identify the effects of IL-1 blockade on the reversal of diabetes in NOD mice, we administered IL-1RA with and without F(ab′)₂ fragments of anti-CD3 mAb 145–2C11. Figure 1A shows that diabetic NOD mice treated with this combination had significantly faster restoration of euglycemia (i.e., glucose <250 mg/dL) with median reversal time of 1 day, compared with mice treated with anti-CD3 alone (median reversal time of 14 days; P = 0.003). In addition, a larger proportion of animals (96 vs. 70%, P < 0.01) reached euglycemia. We followed eight of the anti-CD3 mAb–treated and eight of the combination-treated mice for up to 60 days and found no recurrence of T1D at a later time in the mice that had reversal at day 30 (not shown). To confirm that the synergy observed was due to a specific blockade of IL-1 action rather than a nonspecific effect of the protein, we used another anti–IL-1 reagent, a neutralizing anti–IL-1β mAb. We again saw rapid reversal of diabetes when anti-CD3 mAb was combined with anti–IL-1β mAb (P < 0.02, Fig. 1B). In these experiments, all of the mice treated with anti-CD3 mAb and placebo eventually did show reversal of diabetes.

The rapid reversal of clinical diabetes suggested an acute effect of the combination therapy on either β-cell secretory function or improved sensitivity to insulin. We extracted total insulin from the pancreata of the mice on day 7 and found that the pancreatic insulin content of the combination-treated mice was higher than that of placebo-treated mice (P < 0.05), whereas that of mice treated with either anti-CD3 mAb, or IL-1RA alone, was not (Fig. 1C). In addition, we performed insulin tolerance tests on the treated mice on day 2, but found that the decline in glucose levels were similar in mice that received the combination treatment (a decrease of 31.4 ± 3.5%) compared with mice treated with anti-CD3 mAb alone (a decrease of 38 ± 2.4%; NS) or IL-1RA alone (a decrease of 34.6 ± 2.28%; NS), indicating that improved insulin sensitivity was not the basis for the improved response (Fig. 1D).

We measured the levels of cytokines in the serum after the second dose of drugs and also characterized cytokine production by splenocytes that were activated with PMA/ionomycin at the conclusion of the treatments (Fig. 2). There were increased levels of IL-5 (P < 0.02) and IFN-γ (P < 0.02) in serum in mice that received anti-CD3 mAb alone or in combination with IL-1RA. The increased levels of these cytokines were most likely due to the activation of T cells in vivo, since they were seen in both groups that received anti-CD3 mAb. However, the level of IL-4 was also significantly greater in the combination-treated versus anti-CD3 mAb–treated mice (P < 0.05, Fig. 2) but was not significantly increased in the group that received anti-CD3 mAb alone, since 6 of 10 mice had undetectable serum levels. In the culture supernatants from splenocytes that were activated at the conclusion of treatment (day 7), we found significantly increased levels of IL-10 in mice treated with the combination compared with anti-CD3 mAb alone. The levels of IL-4, IL-5, and IFN-γ were not significantly different in the supernatants from splenocytes in the treatment groups (not shown).

The combination affected pathogenic T cells. We enumerated autoantigen-specific NRP-V7⁺ T cells in the spleen and pancreatic lymph nodes on day 7 with class I major histocompatibility complex tetramers (39,40). This analysis failed to show a difference in the number of the tetramer⁺ CD8⁺ T cells between the groups in the spleen (not shown), but the combination- and IL-1RA–treated groups had lower frequencies of autoantigen-specific CD8⁺ cells in pancreatic lymph nodes (Fig. 3A).

Previous studies by our group and others have suggested that anti-CD3 mAb induces adaptive Tregs that mediate the reversal and prevent progression of diabetes in NOD mice (27,42). To determine the effects of the drug combination on Tregs, we first enumerated the T-cell subsets in the spleens of the treated mice on day 7. The total number of splenocytes was increased in mice that received anti-CD3 mAb with or without IL-1RA (P < 0.001), and the ratio of CD4:CD8⁺ T cells was decreased by anti-CD3 mAb in combination with IL-1RA (P < 0.001, Fig. 3B). In addition, the percent and total number of Foxp3⁺CD4⁺ T cells was increased in the spleens of both the anti-CD3 mAb and combination-treated mice (P < 0.001, Fig. 3B). We also found an increase in the absolute numbers of ICOS⁺CD4⁺ T cells compared with mice that received placebo or IL-1RA alone (Fig. 3B). Upregulated FoxP3 and ICOS was due to the anti-CD3 mAb, since there were no detectable differences in cell numbers in mice that received the combination versus anti-CD3 mAb alone and we did not detect significant differences with IL-1RA alone.

Because IL-1RA and IL-10 are also products of monocytes/macrophages with anti-inflammatory properties (M2), we hypothesized that combination treatment may affect macrophage differentiation. We sorted macrophages and dendritic cells with CD11b⁺ and CD11c⁺ mAbs from day 7 spleens and analyzed expression of Arg1 and NOS2 genes, which encode arginase 1 (M2 marker) and iNOS (M1 marker), respectively, on these subpopulations. In both CD11b⁺ and CD11c⁺ cells, the relative expression of Arg1 was increased in IL-1RA–treated mice (Fig. 4A and C). The levels of NOS2 expression were not different between the groups (data not shown), but the Arg1 to NOS2 ratio was higher in IL-1RA–treated animals (Fig. 4B and D). Expression of other genes tested, which discriminate between classically and alternatively activated monocytes/macrophages or dendritic cells (IL-1RN, CCR3, IL12, IL10, IL18, and Ym1), were not significantly different (not shown).

This result suggested that IL-1RA with anti-CD3 mAb induced adaptive and innate immune cells that may have enhanced immune regulatory function. To test this directly, we performed adoptive transfer experiments mixing spleen cells from diabetic NOD mice with day 7 spleen cells from treated mice. There was a modest but significant delay (31 vs. 25 days, P < 0.05) in the time of onset of diabetes in NOD/scid recipient mice that received splenocytes from combination-treated mice but not anti-CD3 mAb–treated (Fig. 4E and F) or IL-1RA–treated (not shown) mice (Fig. 4E and F).

In addition to a more rapid reversal of diabetes, the combination treatment also had long-term effects on islet inflammation. We compared the immunologic effects and the pancreata from treated mice 30 days after treatment initiation. We measured the isotypes of immunoglobulins in the serum 25 days after completion of treatment and found that the relative concentrations of IgG1 were increased in mice that received the combination treatment compared with mice treated with anti-CD3mAb alone (P < 0.05, Fig. 5). This result is consistent with the effect of the combination on cytokine production.

We extracted and measured insulin and proinsulin in pancreata from mice at day 30 (Fig. 6A and B). Similar to observations on day 7, the increase in total insulin and proinsulin content persisted at day 30 in mice that received the combination of drugs compared with placebo, or IL-1RA (Fig. 6A and B; P < 0.001 and P < 0.01, respectively).

We compared islet inflammation and inflammatory gene expression in the pancreata from day 30 mice. The relative expression of CD45 in the pancreas was not distinguishable between the groups by qPCR, and the histological insulitis score was similar between the treatment groups (Fig. 6C and D). However, there were marked differences in the expression of immune response genes in the pancreas. The expression of IL-17 (P < 0.02), IL-6 (P < 0.05), and IFN-γ (P < 0.05) were all reduced in the pancreata from combination-treated mice compared with mice treated with placebo (Fig. 7A). The reduced expression of these cytokine genes was not significantly different in the mice treated with anti-CD3 mAb alone or IL-1RA alone compared with placebo. The absence of immune responses observed in the pancreas was not a global feature. When splenocytes from day 30 mice were activated with anti-CD3 mAb, we did not detect significant differences in levels of the six measured cytokines in culture supernatants (Fig. 7B).

We found that IL-1β modulates T-cell responses after anti-CD3 mAb in T1D. A 5-day treatment regimen with IL-1RA significantly improved the rate of reversal of T1D in hyperglycemic NOD mice treated with non-FcR binding anti-CD3 mAb compared with the anti-CD3 treatment alone, but IL-1RA itself was ineffective. Our studies with neutralizing anti–IL-1β mAb suggest that the effect of the IL-1RA is due to a specific blockade of IL-1 action. Our data suggest that the combination treatment induced long-term tolerance, since reduced inflammation was found 1 month after treatment, and the reversal did not require continuous IL-1 blockade, which may be detrimental for the immune system and even for β-cell function (43). Similar lasting effects of IL-1RA treatment on low-grade systemic inflammation and β-cell function has been reported in patients with type 2 diabetes (44).

It has been known for >20 years that IL-1β has an important role in the pathogenesis of autoimmune diabetes (45). IL-1, in synergy with other cytokines, is cytotoxic to pancreatic β-cells (5,6,10,45,46). The direct toxicity occurs by signaling through IL-1 receptors that are expressed on β-cells and activation of the nuclear factor κB and mitogen-activated protein kinase pathways, NO production, and depletion of endoplasmic reticulum calcium (14,43). IL-1β favors immune effector cell survival and expansion and reduces the regulatory compartment (29). Thus, because of the role of IL-1β as both an immune response initiator and as a direct toxin to β-cells as well as its potential role in mediating insulin resistance through the elaboration of other inflammatory mediators, it is not surprising that its blockade would be of benefit in diabetes. Indeed, Larsen et al. (47) found that hemoglobin A_1c levels were improved in patients with type 2 diabetes who were treated with IL-1RA, largely because of the direct β-cell protective effects. However, previous studies from NOD mice have shown modest effects of IL-1β antagonists or knockout of the IL-1 receptor in this model, similar to our findings in which IL-1RA or anti–IL-1β mAb did not reverse diabetes (24).

Our analysis suggests that a number of mechanisms may be involved in the effects of the combination when compared with anti-CD3 mAb alone. When IL-1RA was combined with anti-CD3 mAb, there was significant elimination of certain pathogenic T cells, increased insulin content in the pancreata of mice at the time that the drug treatment was completed, and increased regulatory function of splenocytes in preventing adoptive transfer of diabetes. This result is consistent with a local effect of the IL-1 antagonist on islet inflammation and IL-1–driven expansion of effector T cells. The selective effects of the IL-1RA on effector cells in the pancreas may reflect the increased number of these cells at that location at the time of disease onset (39,40). A reduction of tetramer⁺ CD8⁺ T cells could be partially due to downregulation of TCR caused by anti-CD3 binding; however, it was not seen in the spleen and was not observed in the mice treated with anti-CD3 alone. The reduced effector cells alone cannot account for the effect of the combination therapy, since they were reduced in the mice that received single therapies, albeit not of statistical significance. In addition to effects on effectors, there were increased regulatory mechanisms including increased IL-10 production by activated splenocytes, increased M2 splenocytes, and induction of regulatory activity. The combination therapy led to increased serum levels of IL-4 and IL-5 on day 2 and increased IL-10 production by activated cells, which may have had an early effect on antigen presentation. Our finding of increased levels of IgG1 is consistent with diversion to a Th2 phenotype when the combination is administered. As a result of the anti-CD3 mAb treatment, there was an increase in the total number of splenocytes, a decrease in the CD4:CD8 T-cell ratio, and an increase in the number of Foxp3⁺ splenic Tregs and ICOS⁺CD4⁺ T cells. ICOS⁺Foxp3⁺CD4⁺ T cells have been found to be IL-10–producing Tregs (48). Thirty days after treatment, pancreatic infiltrates were not reduced, but the expression of proinflammatory genes that distinguish the cells as pathogenic effectors was decreased. This appeared to be a site-specific effect, since stimulatory responses of splenocytes were normal in these same mice. Thus, our studies have identified selective effects of anti-CD3 mAb and IL-1RA on effector and regulatory mechanisms that are involved in T1D. Individually, these effects have either no or modest effects in reducing and inducing tolerance to autoimmune diabetes. In combination, however, they are very potent and have lasting effects.

The basis for the immune regulatory changes may be the effects of IL-1RA on macrophages and dendritic cells. Whereas the proportion of CD11b⁺ cells in the spleen was similar between the groups, these cells strongly expressed the anti-inflammatory marker Arg1 in IL-1RA–treated mice, suggesting an increased proportion of M2 cells. Similar increase was found in the CD11c⁺ population. Importantly, the Arg1 gene encodes for arginase 1, an enzyme that uses l-arginine to make urea, and thus competes with iNOS for the substrate. IL-1 is a powerful inducer of iNOS, and NO is highly toxic to β-cells (49). An early increase of splenic Arg1, and particularly of the Arg1 to NOS2 ratio together with elimination of effector T cells (also a consequence of IL-1 blockade), may be one of the important mechanisms behind a rapid restoration of euglycemia and preservation of pancreatic insulin content in the combination-treated mice. Arginase 1 has also been reported as an enzyme involved in tissue repair and recruitment of other cell types. It is also possible that in addition to macrophages and dendritic cells, other cells that express these markers (e.g., NK cells) may have been affected and contribute to the effects of the combination treatment.

In agreement with this, we found a delay in transfer of diabetes when splenocytes from mice treated with the combination (but not anti-CD3 mAb alone) were mixed with pathogenic spleen cells compared with mixtures of splenocytes from placebo-treated mice. These studies would suggest that the combination therapy induces either a greater proportion or potency of cells with regulatory function. We are not clear on which cells are responsible for the enhanced activity—both IL-10–producing T cells and modified dendritic cells or macrophages may be involved. A remarkable feature of the transfer studies was that the regulatory activity was identified in the spleen soon after treatment, since, in previous studies by Belghith et al., as well as in our previous studies (27,42), the adaptive Tregs induced by anti-CD3 mAb were generally found at a later time after anti-CD3 mAb therapy.

A challenge of confronting development of immune modulatory therapies of T1D is the loss of efficacy several months after they are given, even with continued administration (30–34,50,51). In addition to the more rapid correction of diabetes, this combination therapy had lasting effects, seen 1 month after treatment. At that time, the infiltrates were not reduced in the islets, but we showed reduced expression of genes associated with pathologic phenotype(s). These findings are consistent with reports that show that IL-1β plays an important role in the differentiation of Th17⁺ cells as well as in the expansion of CD25⁺ effectors (8,29,52). Interestingly, blockade of cytokine gene expression was not mirrored by reduced cytokine production by activated splenocytes, suggesting that the nonresponsiveness in the pancreas may be maintained by exposure to pancreatic antigen. Further studies that can directly assess the function of the infiltrating cells and the biologic significance of the lack of cytokine gene expression that we identified in the pancreata will be needed to clarify this point.

In summary, we have shown that combining IL-1 antagonists with FcR nonbinding anti-CD3 mAb is synergistic in reversal of T1D in NOD mice. The combination affects both innate and adaptive immune pathways, resulting in lasting effects. This combinatorial approach may be useful in patients in enhancing remission of disease that has been difficult to achieve with single agents.

This work was supported by National Institutes of Health grant DK057846 and grant 2009-694 from the Juvenile Diabetes Research Foundation (JDRF) and the Brehm Consortium. J.B.H. received support from the SHARE LIFE initiative of the University of Copenhagen. T.M.-P. received support from Novo Nordisk. P.S. was funded by the Canadian Institutes of Health Research. P.S. is a scientist of Alberta Innovates–Health Solutions (AIHS) and a JDRF scholar. The Julia McFarlane Diabetes Research Centre was supported by the Diabetes Association (Foothills). No other potential conflicts of interest relevant to this article were reported.

V.A. planned the experiments, performed the studies in experimental animals and the analysis of specimens and materials from the mice, reviewed data, and wrote the manuscript. O.H., J.B.H., L.O.-A., and P.P.-H. performed the studies in experimental animals and the analysis of specimens and materials from the mice. P.S. and T.M.-P. reviewed data and wrote the manuscript. K.C.H. planned the experiments, reviewed data, and wrote the manuscript. K.C.H. is the guarantor and takes full responsibility for the article and its originality.

The authors thank Charles Dinarello, University of Colorado, for providing the placebo for IL-1RA and Louise Koch, Hagedorn Research Laboratory, for expert technical assistance.