Granulocyte-Macrophage Colony-Stimulating Factor Prevents ...

The Journal of Immunology

Granulocyte-Macrophage Colony-Stimulating Factor Prevents Diabetes Development in NOD Mice by Inducing Tolerogenic Dendritic Cells that Sustain the Suppressive Function of CD4ⴙCD25ⴙ Regulatory T Cells1 Simon Gaudreau, Chantal Guindi,2 Michae¨l Me´nard,2 Gilles Besin, Gilles Dupuis, and Abdelaziz Amrani3 Autoimmune diabetes results from a breakdown of self-tolerance that leads to T cell-mediated ␤-cell destruction. Abnormal maturation and other defects of dendritic cells (DCs) have been associated with the development of diabetes. Evidence is accumulating that self-tolerance can be restored and maintained by semimature DCs induced by GM-CSF. We have investigated whether GM-CSF is a valuable strategy to induce semimature DCs, thereby restoring and sustaining tolerance in NOD mice. We found that treatment of prediabetic NOD mice with GM-CSF provided protection against diabetes. The protection was associated with a marked increase in the number of tolerogenic immature splenic DCs and in the number of Foxp3ⴙCD4ⴙCD25ⴙ regulatory T cells (Tregs). Activated DCs from GM-CSF-protected mice expressed lower levels of MHC class II and CD80/CD86 molecules, produced more IL-10 and were less effective in stimulating diabetogenic CD8ⴙ T cells than DCs of PBS-treated NOD mice. Adoptive transfer experiments showed that splenocytes of GM-CSF-protected mice did not transfer diabetes into NOD.SCID recipients. Depletion of CD11cⴙ DCs before transfer released diabetogenic T cells from the suppressive effect of CD4ⴙCD25ⴙ Tregs, thereby promoting the development of diabetes. These results indicated that semimature DCs were required for the sustained suppressive function of CD4ⴙCD25ⴙ Tregs that were responsible for maintaining tolerance of diabetogenic T cells in NOD mice. The Journal of Immunology, 2007, 179: 3638 –3647.

D

endritic cells (DCs)4 are bone marrow-derived leukocytes that are key players in the orchestration of immune responses. Several lines of evidence have indicated that DCs are involved in Ag-specific T cell responses and T cell tolerance (1, 2). In addition, a large number of data has revealed that immature DCs (iDCs) or semimature DCs can induce and maintain peripheral T cell tolerance, whereas terminally differentiated mature DCs can efficiently induce the development of effector T cells (3). The maturation stage of DCs depends on the nature of the cytokines they encounter. In this respect, GM-CSF (4), IL-10 and/or TGF-␤ (5–7), as well as 1,25 dihydroxyvitamin D3 (8) have been shown to generate tolerogenic DCs. Two types of DCs gen-

Department of Pediatric, Immunology Division, Faculty of Medicine and Health Sciences, University of Sherbrooke, Sherbrooke, Quebec, Canada Received for publication February 22, 2007. Accepted for publication July 13, 2007. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by a grant from the Juvenile Diabetes Foundation International. S.G. is recipient of a PhD scholarship from Fonds de la Recherche en Sante´ du Que´bec. C.G. and M.M. are recipients of summer studentships from Diabe`te Que´bec. G.B. is the holder of a fellowship from Association de Langue Franc¸aise pour l’e´tude du Diabe`te et des maladies Me´taboliques (ALFEDIAM). A.A. is Canadian Diabetes Association New Investigator and a recipient of a Chercheur Boursier Junior 1 and Junior 2 fellowship from the Fonds de la Recherche en Sante´ du Que´bec. 2

C.G. and M.M. contributed equally to this work.

3

Address correspondence and reprint requests to Dr. Abdelaziz Amrani, Department of Pediatric, Immunology Division, Faculty of Medicine and Health Sciences, University of Sherbrooke, 3001 12th Avenue North, Sherbrooke, Quebec, Canada J1H 5N4. E-mail address: [email protected] 4 Abbreviations used in this paper: DC, dendritic cell; iDC, immature DC; Treg, regulatory T cell; PLN, pancreatic lymph node.

Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 www.jimmunol.org

erated using a combination of GM-CSF and IL-10 have been described: iDCs that express the CD11clow/CD45Rbhigh markers and that have a plasmacytoid morphology (6) and CD11chigh iDCs (7). Immunoregulatory CD11chigh DCs generated using a combination of GM-CSF, IL-10, and TGF-␤ have also been shown to control the ability of the acute graft-vs-host disease and leukemia relapse in allogenic bone marrow transplantation in leukemic mice (9). These tolerogenic DCs share common characteristics such as low level expression of CD80/86 and MHC class II molecules and induction of T cell unresponsiveness. DCs have also been shown to behave as tolerogenic APCs for self-Ag in vivo. For example, injection of DCs exposed to Ag (but not to full maturation stimuli) induced Ag-specific protection from experimental autoimmune encephalomyelitis in mice (10). Recent studies have revealed the importance of GM-CSF in the regulation of DC functions. It has been reported that bone marrowderived DCs generated with low doses of GM-CSF were immature, resistant to maturation stimuli, and restored T cell tolerance in vitro and in vivo (4). Moreover, constitutive expression of GMCSF in the pancreas of BALB/c mice delayed and decreased diabetes induced by streptozotocin, suggesting that GM-CSF protects from diabetes by expanding a population of tolerogenic DCs (11). Other studies have shown that repetitive GM-CSF injection protected mice from development of autoimmune thyroiditis by increasing the number of semimature myeloid DCs and by inducting IL-10-secreting CD4⫹CD25⫹ regulatory T cells (Tregs) (12, 13). Therefore, in view of the involvement of DCs in peripheral tolerance, in vivo modulation of DC function with GM-CSF may represent a valuable strategy to increase and maintain a high number of maturation-resistant iDCs, thereby promoting peripheral tolerance in autoimmune diabetes.

The Journal of Immunology Autoimmune diabetes results from a breakdown of self-tolerance culminating in T cell-mediated ␤-cell destruction (14, 15). Numerous studies have attributed the breakdown of tolerance in NOD mice to defects in the number and/or function of DCs (16 – 19) and CD4⫹CD25⫹ Tregs (20, 21). The defects include an impaired development of DCs from myeloid progenitors (bone marrow iDCs) and abnormal levels of expression of costimulatory molecules under proinflammatory conditions (16, 17). These combined alterations lead to an increased capacity of DCs to stimulate CD4⫹ and CD8⫹ T cells, thereby promoting Th1 or Tc1 differentiation in vitro (18, 19, 22, 23). Consequently, the level of iDCs that induce and maintain peripheral T cell tolerance is not sufficient to trigger a sustained immunoregulatory T cell response in NOD mice. With respect to Tregs, it has been shown that the pool of CD4⫹CD25⫹ Tregs decreases with progression of diabetes (21, 24) and that these cells can afford protection against the development of diabetes (25, 26). In the present study, we have investigated whether reestablishing a sufficient number of DCs possessing tolerogenic functions would maintain CD4⫹CD25⫹ Treg number and function, thereby providing protection against the development of diabetes in NOD mice. Our data showed that treatment of NOD mice with GM-CSF increased the number of iDCs acquiring a semimature phenotype and that these iDCs had less capability to activate diabetogenic CD8⫹ T cells. We further showed that these tolerogenic DCs did not undergo complete maturation and were required to maintain permanent diabetes protection through CD4⫹CD25⫹ Tregs.

Materials and Methods Mice and treatment Female NOD and NOD.SCID mice were purchased from the The Jackson Laboratory. The 8.3-NOD mice, obtained from Dr. P. Santamaria (University of Calgary, Calgary, Alberta, Canada), have been described (27). In vivo treatments of 3-wk-old female NOD mice were done by i.p. injection of 100 ng/mouse of recombinant murine GM-CSF (R&D Systems) or PBS. Mice were injected three times per week for the first 3 wk, followed by two injections per week up to 52 wk of age. All the mice were housed and bred in our animal facilities under specific pathogen-free conditions and were used according to guidelines of the Institutional Animal Care Committee of the University of Sherbrooke (Sherbrooke, Quebec, Canada).

Monitoring of diabetes Diabetes was monitored by a urine glucose test using Uristix (Bayer) and confirmed by measuring hyperglycemia with an Advantage Accu-Check glucometer (Roche Diagnostics). The animals were considered diabetic after two positive Uristix readings or when blood glucose concentration was greater than 14 mmol/L.

Cell lines and Abs Anti-CD8␣-PE (clone 53-6.7), anti-CD4-FITC/biotin (clone GK1.5), antiCD25-FITC (clone 7D4), anti-CD11b-FITC (clone M1/70), anti CD45R/ B220-biotin (clone RA3-6B2), anti-CD11c-FITC/biotin (clone HL3), anti-CD80-biotin (clone 16-10A1), anti-CD86-biotin (clone GL1), anti-I-Ag7biotin (clone 10-3.6) Abs and streptavidin-PerCP (used for all biotin-conjugated Abs) were from BD Biosciences. The anti-Foxp3-FITC (FJK-16s) and anti-rat IgG2a (eBR2a) Ab were from eBioscience. The 145-2C11 hybridoma (anti-CD3) was obtained from Dr. P. Santamaria (University of Calgary, Calgary, Alberta, Canada).

3639 sized using fMOC chemistry, purified by reverse phase HPLC to 90% purity as confirmed by MALDI-TOF mass spectrometry.

T cell and splenic DC isolation Cell (CD4⫹ and CD8⫹ T cells and DCs) purification was done using Abcoated magnetic beads from Miltenyi Biotec. Briefly, DCs were purified from collagenase D-digested spleens (two or three organs) using antiCD11c-coated beads (Miltenyi Biotec). In the case of CD8␣⫹ and CD8␣⫺ DC purification, spleen cells were enriched for CD11c⫹ cells by depletion of T, B, and NK cells using a mixture of Abs (Miltenyi Biotec), incubated with anti-CD8␣ microbeads, and sorted by MACS into CD8␣⫹ and CD8␣⫺ CD11c⫹ fractions (Miltenyi Biotec). Purity was ⬎85% for CD11c⫹, CD8␣⫹, and CD8␣⫺ cells as determined by FACS analysis. In the case of CD4⫹, CD4⫹CD25⫺, CD4⫹CD25⫹, and CD8⫹ T cell enrichment, splenocytes were freed of RBC by hemolysis and the respective T cells were isolated by magnetic beads separation using an antiCD4⫹CD25⫹/⫺ isolation kit for CD4⫹ T cell purification or with antiCD8-coated magnetic beads for CD8⫹ T cell purification, according to the manufacturer’s protocol (Miltenyi Biotec).

Bone marrow-derived DCs Bone marrow-derived DCs from NOD mice were generated by culturing bone marrow cells in the presence of recombinant murine GM-CSF and IL-4 (5 ng/ml) (28).

Flow cytometric analysis All analyses were performed using a FACSCalibur flow cytometer using the CellQuest Pro software (BD Biosciences). In brief, cells were washed with PBS containing 0.05% sodium azide and 2.5% heat-inactivated FBS (FACS buffer). The cells were then labeled in the same buffer for 30 min at 4°C using the various Abs described in the appropriate experiments, washed, and analyzed by flow cytometry. Foxp3 expression was determined as follows. The cells were stained with a combination of anti-CD4 and anti-CD25 mAb before fixation with 4% paraformaldehyde for 1 h at 4°C. After treatment with FACS buffer containing 0.1% saponin (for intracellular staining), the cells were stained with FITC-labeled anti-Foxp3 mAb or anti-rat IgG2a for 30 min at 4°C, washed, and analyzed by FACS.

Proliferation assays Assays of 8.3-CD8⫹ T cell activation by DCs were performed as follows. Naive splenic 8.3-CD8⫹ T cells (2 ⫻ 104 lymphocytes/well) were incubated with NRP-A7- or TUM-pulsed (1 ␮g/ml), irradiated splenic DCs (2 ⫻ 104 cells/well), from GM-CSF- or PBS-treated mice, in 96-well plates for 3 days at 37°C. The supernatants were collected after 48 h and assayed for cytokine content by ELISA using commercially available kits (R&D Systems). During the last 18 h, cultures were pulsed with 1 ␮Ci of [3H]thymidine. The regulatory activity of CD4⫹CD25⫹ Tregs purified from GMCSF- or PBS-treated NOD mice was measured by adding preactivated (2 days with anti-CD3 (5 ␮g/ml) and IL-2 (10 U/ml)) CD4⫹CD25⫺ or CD4⫹CD25⫹ T cells (2 ⫻ 104 to 4 ⫻ 104 lymphocytes/well) to 8.3-CD8⫹ T cells (2 ⫻ 104 lymphocytes/well) and NRP-A7-pulsed (1 ␮g/ml) irradiated APC (1 ⫻ 105 cells/well) cocultures.

Cytokine assays Measurement of cytokines released in the culture medium by the different T cell populations was done by ELISA (R&D systems) following activation of the cells using anti-CD3-coated 96-well plates (5 ␮g/ml) with or without IL-2 (10 U/ml, Roche) for 2 days at 37°C in a humidified 5% CO2 incubator.

RT-PCR

Pancreata were fixed in formalin, embedded in paraffin, sectioned, and stained with H&E. Islet insulitis was scored as follow: 0, healthy islet; 1, peri-insulitis; 2, leukocytic infiltration of up to 25% of islet mass; 3, leukocytic infiltration of up to 75% of the islet mass; and 4, ⬍20% of islet mass remaining.

CD4⫹CD25⫺ and CD4⫹CD25⫹ T cells of GM-CSF- and PBS-treated NOD mice were purified using an anti-CD4⫹CD25⫹/⫺ magnetic beads isolation kit (Miltenyi Biotec). RNA was isolated from the sorted cells using the TRIzol reagent (Invitrogen Life Technologies). cDNA were generated using a first-strand DNA synthesis kit (Invitrogen Life Technologies) and amplified by PCR. The sequences of the primers used were as follows: Foxp3 5⬘-CAG CTG CCT ACA GTG CCC CTA G-3⬘ and 5⬘CAT TTG CCA GCA GTG GGT AG-3⬘; and HPRT, 5⬘-GT TGG ATA CAG GCC AGA CTT TGT TG-3⬘ and 5⬘-GAA GGG TAG GCT GGC CTA TAG GCT-3⬘ (29). All samples were subjected to electrophoresis using a 2% agarose gel.

Peptides

Adoptive cell transfers

The NRP-A7 and TUM peptides were from C. Servis (Biochemistry Institute, Lausanne University, Lausanne, Switzerland). They were synthe-

Whole spleen cells or DC-depleted (CD11c⫹-depleted) spleen cells from GM-CSF- or PBS-treated diabetic mice were injected into the tail vein of

Histopathology

DC AND CD4⫹CD25⫹ Treg FUNCTIONS IN NOD MICE

3640

FIGURE 1. GM-CSF treatment significantly reduced the incidence of diabetes in NOD mice. NOD mice were treated with GM-CSF (n ⫽ 27) or PBS (n ⫽ 24) and monitored for diabetes development as described in Materials and Methods.

4- to 6-wk-old NOD.SCID hosts (8 ⫻ 10 cells). In some experiments, CD4⫹CD25⫺ or CD4⫹CD25⫹ T cells (0.5 ⫻ 106 cells) from GM-CSF- or PBS-treated NOD mice were coinjected with splenocytes (8 ⫻ 106 cells) obtained from a diabetic NOD mice to NOD.SCID recipients. Recipient mice were monitored for diabetes by measuring urine glucose, starting 20 days after cell transfer. 6

FIGURE 2. Islet inflammation is dramatically reduced in GM-CSFtreated NOD mice. A, Pancreata from GM-CSF- or PBS-treated animals were scored for insulitis as described in Materials and Methods. Data represent the average ⫾ SEM. B, Evolution of islets inflammation in GMCSF- or PBS-treated NOD mice with time. A minimum of 30 islets was scored for each group of six animals (p ⬍ 0.001). C, Representative microphotographs of H&E-stained pancreatic sections from GM-CSF- or PBS-treated NOD mice.

Statistical analyses Student’s t test and ␹2 test were used to determine the statistical significance of differences between groups, which was set at the 95% confidence level. Data reported are representative of a minimum of three independent experiments. Results are shown as the mean ⫾ SEM represented by the error bars.

Results In vivo treatment of NOD mice with GM-CSF reduces insulitis and protects against diabetes The effect of GM-CSF on diabetes development was investigated in 3-wk-old prediabetic NOD mice. The mice were treated for 49 wk with GM-CSF (100 ng/mouse) or PBS (control), and the development of diabetes and insulitis was monitored over time. Results showed that the treatment with GM-CSF significantly decreased the progression of the disease (Fig. 1). At 52 wk of age, 21% of GM-CSF-treated mice developed diabetes (n ⫽ 27), whereas 86% of PBS-treated mice became diabetic over the same period of time (n ⫽ 24) ( p ⬍ 0.0001, ␹2 test). All GM-CSF-protected mice remained normoglycemic with blood glucose levels of 5–7 mmol/L. To further assess whether diabetes resistance induced by GM-CSF was associated with reduced islet inflammation, the progression of insulitis was evaluated at 8, 12, 16, and 32 wk of age. Results of histopathology of pancreas showed that, at up to 16 wk of age, GM-CSF-treated NOD mice exhibited little or no insulitis, whereas control (PBS) mice displayed typical inflammation, with both peri-insulitis and invasive insulitis (Fig. 2, A and C). GM-CSF-protected mice showed only mild inflammation of islets at 32 wk of age (Fig. 2C). In these mice, the pancreas contained a significantly high percentage of islets without lymphocytic infiltration (58%) as well as a dramatically reduced number of islets with more aggressive inflammation (8.7%) (Fig. 2B). In contrast, pancreas from 32-wk-old control mice (PBS-treated), were devoid of islets (data not shown). These results were consistent with the interpretation that GM-CSF-treatment protected NOD mice from diabetes by preventing the development of insulitis.

GM-CSF-treatment expands DCs with low Ag-presenting capacity To examine whether GM-CSF treatment stimulated the expansion of DCs in NOD mice, we analyzed splenocytes for the presence of cells expressing high levels of the CD11c⫹ marker. Results showed a marked increase in the number of cells expressing CD11c⫹ (2.27 ⫾ 0.26 ⫻ 106 cells, n ⫽ 12) in the case of DCs from GM-CSF-treated mice in comparison with CD11c⫹ DCs (1.36 ⫾ 0.17 ⫻ 106, n ⫽ 11) from PBS-treated mice (Fig. 3A). Furthermore, there was a consistent and significant ( p ⬍ 0.02) higher percentage of purified CD11c⫹ DCs per spleen in the case of GM-CSF-treated mice (2.46 ⫾ 0.27%, n ⫽ 14) as compared with PBS control mice (1.63 ⫾ 0.12%, n ⫽ 10) (Fig. 3B). The phenotypes and functions of splenic DCs isolated from GM-CSF-treated NOD and control (PBS-treated) mice were determined in the next sets of experiments. Freshly isolated splenic DCs from groups of mice exhibited an immature phenotype characterized by low levels of expression of CD80 and CD86 and MHC class II molecules (data not shown). Although the mean fluorescence intensity for CD80 and CD86 on DCs from GM-CSFtreated mice appeared to be higher than on DCs from control PBStreated mice, the differences did not reach statistical significance (data not shown). The DCs from both groups of mice also expressed comparable levels of MHC class II molecules (data not shown). Interestingly, NRP-A7-pulsed DCs from GM-CSF-treated mice were not as efficient as DCs from PBS-treated NOD mice to induce the proliferation of diabetogenic 8.3-CD8⫹ T ( p ⬍ 0.006) (Fig. 3C). The lower response of 8.3-CD8⫹ T cells to NRP-A7pulsed DCs from GM-CSF-treated mice was not due to the anergy of CD8⫹ T cells because activated 8.3-CD8⫹ T cells produced equal amounts of IL-2 and IFN-␥ as compared with DCs from PBS-treated mice (Fig. 3D). The lower proliferative response of 8.3-CD8⫹ T cells was not accompanied by the production of IL-4 (Fig. 3D), IL-10, or TGF-␤ (data not shown).


3641

FIGURE 3. GM-CSF increases the number of DCs without changing their phenotype in NOD mice. A, Number of DCs in spleen of GM-CSFor PBS-treated mice (8 –15 mice per group). Data represent the average ⫾ SEM. B, Relative percentage of DCs isolated from individual spleen of GM-CSF- or PBS-treated mice with respect to total splenocytes (10 –14 mice per group). C, Proliferation of naive 8.3-CD8⫹ T cells (2 ⫻ 104 lymphocytes/well) in response to peptide-pulsed (NRP-A7 or TUM, 1 ␮g/ ml), irradiated splenic DCs (2 ⫻ 104 cells/well) from GM-CSF- or PBStreated mice. D, Production of IL-2, IFN-␥, and IL-4 by naive 8.3-CD8⫹ T cells cultured under the same conditions.

Inhibition of full maturation and Ag-presenting function of DCs from GM-CSF-protected mice We next examine whether the GM-CSF treatment altered splenic DCs maturation. Splenic DCs of GM-CSF- and PBS-treated mice were exposed to LPS for 48 h and their phenotype and function were analyzed. As expected, LPS-activated DCs from PBS-treated NOD mice expressed high levels of CD80, CD86, and MHC class II, indicating that these DCs had undergone a full process of maturation (Fig. 4A). In contrast, the levels of expression of MHC class II and CD80/CD86 molecules were lower in LPS-activated DCs from GM-CSF-treated mice (Fig. 4A). These observations indicated that DCs from GM-CSF-treated mice did not undergo a full maturation process in response to LPS treatment in contrast to DCs from PBS-treated mice. These results suggested that DCs from GM-CSF-treated mice had acquired a semimature phenotype, a property characteristic of tolerogenic DCs. In addition, the levels of IL-10, a cytokine known to be involved in the tolerogenic functions of DCs, were significantly increased in DCs from GM-CSFtreated mice as compared with PBS-treated mice (Fig. 4B). Splenic DCs of both groups of mice produced similarly low levels of IL12p70 (Fig. 4C). Under similar conditions of assay, LPS-activated bone marrow-derived DCs of NOD mice produced large amounts of IL-12p70, whereas bone marrow-derived DCs generated from GM-CSF-treated mice produced much lower amounts of IL-12p70 (Fig. 4D). Interestingly, 8.3-CD8⫹ T cells showed a significantly lower ( p ⬍ 0.01) proliferative response to stimulation with NRPpulsed LPS-activated DCs from GM-CSF-treated mice when compared with the response from PBS-treated mice (Fig. 4E). The 8.3-CD8⫹ T cells stimulated with LPS-activated DCs of GM-CSFtreated mice produced significantly lower amounts of IL-2 and IFN-␥ ( p ⬍ 0.03 and p ⬍ 0.004, respectively) (Fig. 4F). Together, these results suggested that DCs from GM-CSF-treated, but not from PBS-treated NOD mice do not undergo full maturation process and are poor Ag-specific CD8⫹ T cell stimulator.

FIGURE 4. GM-CSF induced DCs with tolerogenic properties in NOD mice. A, Splenic DCs from GM-CSF- or PBS-treated mice (CD11c⫹ gated cells) exposed to LPS (2 days, 1 ␮g/ml) were labeled for maturation markers CD80, CD86, and I-Ag7 and analyzed by FACS. B and C, IL-10 and IL-12p70 secretion by LPS-activated (2 days, 1 ␮g/ml) splenic DCs (1 ⫻ 105 cells/well) from GM-CSF- or PBS-treated mice. D, IL-12p70 secretion by LPS-activated bone marrow-derived DCs from PBS- or GM-CSFtreated NOD mice. E, Proliferation of naive 8.3-CD8⫹ T cells (2 ⫻ 104 lymphocytes/well) in response to peptide-pulsed (NRP-A7 or TUM, 1 ␮g/ ml) irradiated splenic LPS-stimulated DCs (2 ⫻ 104 cells/well). F, Production of IL-2, IFN-␥, and IL-4 by naive 8.3-CD8⫹ T cells cultured under the same conditions. Data are representative of three to four independent experiments. Mean fluorescence intensity (MFI) is shown. Results are shown as the mean ⫾ SEM.

Splenic CD8␣⫺ DCs from GM-CSF-treated mice display markers and function of tolerogenic cells We have observed that the number of splenic DCs was significantly increased in GM-CSF-treated NOD mice (Fig. 3, A and B) and asked which DC subset was affected by GM-CSF. We have found that myeloid (CD11c⫹CD8␣⫺) and lymphoid (CD11c⫹ CD8␣⫹) splenic DC numbers were increased in GM-CSF-treated NOD mice (Fig. 5A). To determine which of these DC subsets display tolerogenic function, splenic CD8␣⫺ and CD8␣⫹ DCs from GM-CSF- or PBS-treated mice were purified and tested for their abilities to induce proliferation of 8.3-CD8⫹ T cells and their cytokines production. We found no significant difference ( p ⬎ 0.05) in the proliferation of 8.3-CD8⫹ T cells in response to CD8␣⫹ DCs from GM-CSF- or PBS-treated NOD mice (Fig. 5B). However, a marked decrease ( p ⬍ 0.001) in the proliferation of 8.3-CD8⫹ T cells was observed in the case of CD8␣⫺ DCs of GM-CSF-treated NOD mice (Fig. 5B). The analysis of cytokines production indicated that there were no significant differences in the amounts of IL-2, IFN-␥, and IL-4 produced by 8.3-CD8⫹ T cells activated either with CD8␣⫹ or CD8␣⫺ DCs from GM-CSF- or PBS-treated mice

3642


FIGURE 5. Low capacity of myeloid DCs from GM-CSF-treated NOD mice to activate diabetogenic 8.3-TCR transgenic CD8⫹ T cells in vitro. A, Number of CD11c⫹CD8␣⫹ and CD11c⫹CD8␣⫺ DCs among purified splenic CD11c⫹ DCs from PBS- (n ⫽ 10) and GM-CSF-treated (n ⫽ 12) NOD mice as determined by FACS. B, Proliferation of naive 8.3-CD8⫹ T cells (2 ⫻ 104 lymphocytes/well) in response to peptide-pulsed (NRP-A7 or TUM, 1 ␮g/ml) irradiated CD11c⫹CD8␣⫹ and CD11c⫹CD8␣⫺ splenic DCs (2 ⫻ 104 cells/well). C, Production of IL-2, IFN-␥, and IL-4 by naive 8.3-CD8⫹ T cells cultured under the same conditions. Data are representative of four independent experiments. Results are shown as the mean ⫾ SEM.

(Fig. 5C). These results suggested that in GM-CSF-treated mice, myeloid CD8␣⫺ DCs were endowed with a tolerogenic function by reducing proliferation of diabetogenic CD8⫹ T cells without affecting their cytokine production. Phenotype and function of T cells in GM-CSF-treated NOD mice T cells from NOD mice were analyzed to determine whether diabetes resistance induced by GM-CSF was associated with alterations in their phenotype and function. The number of spleen cells was not increased in GM-CSF-treated NOD mice as compared with PBS-

FIGURE 6. CD4⫹ and CD8⫹ T cells function in GM-CSF-treated NOD mice. A and B, Proliferation and cytokine production by CD8⫹ T cells from GM-CSF- or PBS-treated mice. C and D, Proliferation and production of IL-2, IFN-␥, IL-4, and IL-10 by activated CD4⫹ T cells (1 ⫻ 105 lymphocytes/ well) from GM-CSF- or PBS-treated mice. CD4⫹ and CD8⫹ T cells were exposed to anti-CD3 (2 days, 5 ␮g/ml) surface-bound mAb and cytokine production was determined in supernatants by ELISA. Data are representative of three independent experiments.

treated mice (data not shown). In addition, there were no differences in the number, percentage, or ratio of CD4⫹ and CD8⫹ T cell populations in spleen, pancreatic lymph node (PLN), and mesenteric lymph node of GM-CSF- and PBS-treated mice (data not shown). The possibility that the protective effect of GM-CSF was the result of immune deviation toward Th2 or Tc2 responses was investigated by comparing the proliferative response and cytokine production of splenic CD4⫹ and CD8⫹ T cells in response to antiCD3 stimulation. Results showed that CD8⫹ T cells of GM-CSFand PBS-treated mice displayed similar proliferative responses and produced the same amounts of IL-2 and IFN-␥ (Fig. 6, A and B).


3643

FIGURE 7. GM-CSF increases the number of CD4⫹CD25⫹ T cells in secondary lymphoid organs of NOD mice. A, Percentage of CD4⫹ T cells in GM-CSF- or PBS-treated mice expressing CD25⫹ in spleen and PLN (n ⫽ 10 –14 mice per group). Data are expressed as the average ⫾ SEM. Flow cytometry (B) and Foxp3 expression (C) in splenic CD4⫹CD25⫺ or CD4⫹CD25⫹ T cells purified from GM-CSF- or PBS-treated NOD mice. D, IL-4 and IL-10 secretion by splenic CD4⫹CD25⫺ and CD4⫹CD25⫹ T cells (1 ⫻ 105 lymphocytes/well) exposed to anti-CD3 (2 days, 5 ␮g/ml) mAb and IL-2 (10 U/ml) (n ⫽ 5– 6 mice per group). E, Ability of activated splenic CD4⫹CD25⫹ but not CD4⫹CD25⫺ T cells (0, 2, 4 ⫻ 104 lymphocytes/well) from GM-CSF- or PBS-treated mice to suppress 8.3-CD8⫹ (2 ⫻ 104 lymphocytes/well) T cell responses to NRP-A7-pulsed irradiated-NOD spleen cells (1 ⫻ 105 cells/well). Background responses to the control peptide TUM were subtracted. Data are representative of at least three independent experiments.

However, there was an absence of production of IL-4 or IL-10 in response to anti-CD3 stimulation (data not shown). These results indicated that the treatment with GM-CSF did not alter CD8⫹ T cell function and suggested that diabetes resistance induced by GM-CSF was not a consequence of immune suppression or immune deviation toward a Tc2 response. Splenic CD4⫹ T cells of GM-CSF-treated mice responded vigorously and produced higher amount of IL-4 and IL-10 than CD4⫹ T cells from PBS-treated NOD mice (Fig. 6, C and D). However, no differences were noted in the production of IL-2 and IFN-␥ by CD4⫹ T cells of both groups of mice (Fig. 6D). These results indicated that CD4⫹ T cells from GM-CSF-treated NOD mice proliferate vigorously in vitro and secrete increased amounts of the anti-inflammatory cytokines IL-10 and IL-4 as compared with PBS-treated mice, and suggest that CD4⫹ T cells are acquiring a Th2-like phenotype. GM-CSF treatment increases the Foxp3⫹CD4⫹CD25⫹ Treg population without altering their in vitro suppressive function We next investigated whether GM-CSF treatment affected the number and function of CD4⫹CD25⫹ Tregs in NOD mice. In a first series of experiments, the effect of GM-CSF on the pool of CD4⫹CD25⫹ Treg was investigated in the spleen and PLNs of GM-CSF-treated and control (PBS-treated) NOD mice. Flow cytometry analysis revealed a significant increase in the size of the CD4⫹CD25⫹ Treg population in the spleen and PLNs of GMCSF-treated mice as compared with PBS-treated mice. Results were 10.5 ⫾ 0.8% (n ⫽ 9) and 5.9 ⫾ 0.6% (n ⫽ 13), respectively ( p ⬍ 0.001), in the case of splenocytes, and 14.0 ⫾ 1.7% (n ⫽ 7) and 8.3 ⫾ 0.6% (n ⫽ 15), respectively ( p ⬍ 0.005), in the case of PLNs (Fig. 7A). In agreement with these results, the CD4⫹CD25⫹ Treg marker Foxp3 was expressed in a significantly ( p ⬍ 0.02) higher percentage of CD4⫹ T cells in GM-CSF-treated mice

(19.1 ⫾ 0.8%) as compared with PBS-treated (13.7 ⫾ 0.7%) mice (data not shown). The analysis of the expression of Foxp3 by FACS and by RT-PCR (Fig. 7, B and C) revealed that freshly purified CD4⫹CD25⫹ T cells from PBS-treated and GM-CSFtreated mice were Foxp3 positive, supporting the idea that they are Tregs rather than activated CD4⫹ T cells. In addition, splenic and PLNs Foxp3⫹CD4⫹CD25⫹ Treg populations of GM-CSF- and PBS-treated mice contained normal levels of CD62 ligand and expressed high levels of CTLA-4 and glucocorticoid-induced TNFR (data not shown). There was a significant ( p ⬍ 0.02) increase of Foxp3-positive CD4⫹CD25⫺ T cells in the spleen of GM-CSF-treated (17.9 ⫾ 0.8%) mice as compared with PBS-treated (13.8 ⫾ 0.6%) mice (Fig. 7B, top panels). Interestingly, RT-PCR showed the induction of expression of transcription factor Foxp3 gene in CD4⫹CD25⫺ T cells from GM-CSF-treated mice (Fig. 7C). When activated with combination of an anti-CD3 mAb and IL-2, CD4⫹CD25⫹ T cells of both groups of mice produced the same amount of IL-4 and IL-10, whereas CD4⫹CD25⫺ T cells from GM-CSF-treated mice produced significantly increased amounts of IL-4 ( p ⬍ 0.001) and IL-10 ( p ⬍ 0.001) (Fig. 7D). A hallmark of CD4⫹CD25⫹ Tregs is their ability to suppress the proliferation of T cells stimulated through their TCR, at least in vitro. We therefore investigated whether Foxp3⫹CD4⫹CD25⫹ Tregs from GM-CSF-treated mice displayed a better suppressive activity on the proliferation of diabetogenic CD8⫹ T cells than Foxp3⫹CD4⫹CD25⫹ Tregs from PBS-treated mice. To this end, in vitro-activated splenic CD4⫹CD25⫺ and CD4⫹CD25⫹ T cells from GM-CSF- or PBS-treated mice were assessed for their ability to inhibit proliferation and cytokine production of a highly diabetogenic 8.3-CD8⫹ T cell clone in a coculture system. Data showed that CD4⫹CD25⫹ Tregs from GM-CSF- and PBS-treated mice

3644


suppressed Ag-specific proliferation of 8.3-CD8⫹ T cells to a similar extent (Fig. 7E). The suppressive effect was also observed on IL-2 and IFN-␥ release by NRP-A7-activated 8.3-CD8⫹ T cells in vitro (Fig. 7E). However, CD4⫹CD25⫺ T cells did not inhibit NRP-A7-specific proliferation of 8.3-CD8⫹ T cells or their production of cytokines (Fig. 7E). Together, these results suggested that GM-CSF induced diabetes resistance by increasing the frequency of Foxp3⫹CD4⫹CD25⫹ Tregs without affecting their suppressive function, at least in vitro. In addition, the data confirmed that CD4⫹CD25⫹ Tregs in NOD mice display normal regulatory functions in vitro when they are properly activated. Absence of effect of GM-CSF treatment on the suppressive function of Foxp3⫹CD4⫹CD25⫹ Tregs of NOD mice in adoptive transfer experiments The in vivo suppressor function of CD4⫹CD25⫹ Tregs accumulating in GM-CSF-treated mice was tested by evaluating the ability of these cells to control autoimmune diabetes mediated by effector diabetogenic splenocytes from NOD mice after transfer into NOD. SCID recipient mice. CD4⫹CD25⫹ Tregs from GM-CSF-treated mice as well as from PBS-treated mice were able to significantly ( p ⬍ 0.002) delayed the onset of diabetes by transfer of diabetogenic T cells into NOD.SCID mice (51 ⫾ 4 days and 52 ⫾ 6 days, respectively) (Fig. 8A). In contrast, recipients of CD4⫹CD25⫺ T cells from both groups of mice with diabetogenic cells developed diabetes within an average of 3– 4 wk posttransfer (Fig. 8A). Adoptive transfer experiments repeated with cotransfer of DCs and diabetogenic T cells showed no direct impact of DCs, purified from GM-CSF-treated or control (PBS-treated) NOD mice, on effector function of diabetogenic splenocytes (Fig. 8B). Similar results were obtained in cotransfer experiments using spleen cells from 6to 8-wk-old mice (data not shown). These results suggested that CD4⫹CD25⫹ Tregs from diabetes-free GM-CSF-treated mice or PBS-treated mice delayed transfer of diabetes but were not efficient to prevent, on a long-term basis, the transfer of the disease. In vivo requirement of DCs to maintain long-term suppressive functions of CD4⫹CD25⫹ Tregs in GM-CSF-protected mice The findings that Foxp3⫹CD4⫹CD25⫹ Tregs or DCs from GMCSF-treated mice were not able to provide a long-term protection from diabetogenic spleen cells in adoptive transfer experiments led to investigate whether spleen cells of diabetes-free GM-CSFtreated NOD mice could transfer diabetes into NOD.SCID mice. The experiments were done by transfusing NOD.SCID recipients with 8 ⫻ 106 spleen cells from GM-CSF- or PBS-treated mice. Results showed that spleen cells from 32-wk-old GM-CSF-protected mice did not transfer diabetes into NOD.SCID mice (0/8 for 3 mo posttransfer) (Fig. 8C). However, transfer of diabetes was observed when spleen cells of PBS-treated diabetic NOD mice were used (10/10 within 3– 4 wk post transfer) (Fig. 8C). The transfer of diabetes was also observed when spleen cells of prediabetic PBS-treated NOD donors (6- to 8-wk-old) (6/6 within 3– 4 wk posttransfer) were used (data not shown). These results suggested that cells that reside in the spleen of GM-CSF-treated mice contained a sufficient number of cells (Tregs and/or DCs) with tolerogenic capacity to prevent the development of diabetes. Depletion of DCs from the splenocytes of GM-CSF- or PBS-treated mice were done before transfer into NOD.SCID mice to further dissect the role of splenic DCs and CD4⫹CD25⫹ Tregs in the protection from diabetes in GM-CSF-treated mice. NOD.SCID mice that received CD11c⫹-depleted splenocytes of PBS-treated mice developed diabetes slightly but not significantly later than recipients transferred with diabetogenic splenocytes (Fig. 8C). Interestingly, spleen cells from 32-wk-old GM-CSF-treated mice

FIGURE 8. DCs from GM-CSF-treated NOD mice are required to maintain long-term tolerance in NOD.SCID mice. A, Diabetes onset in NOD.SCID recipient injected with 8 ⫻ 106 cells diabetogenic spleen cells (Diab) alone or with mixture of diabetogenic cells and 0.5 ⫻ 106 CD4⫹CD25⫹ and CD4⫹CD25⫺ purified T cells from GM-CSF- or PBStreated mice (n ⫽ 6 mice per group). B, Diabetes onset in NOD.SCID recipient injected with diabetogenic cells with mixture of diabetogenic cells and 1.2 ⫻ 106 purified splenic DCs of GM-CSF- or PBS-treated mice (n ⫽ 6 mice per group). C, Adoptive transfer of 8 ⫻ 106 total spleen cells (n ⫽ 10 mice per group) or 8 ⫻ 106 CD11c-depleted spleen cells (n ⫽ 8 –10 mice per group,) from GM-CSF- or PBS-treated NOD mice.

that had been depleted of DCs were not able to provide long-term protection against diabetes in recipient mice (Fig. 8C). In these mice, diabetes incidence was not significantly decreased and the onset of diabetes occurred significantly later as compared with the mice transferred with DC-depleted splenocytes from PBS-treated mice (74 ⫾ 0.6 days (n ⫽ 10) as compared with 34 ⫾ 0.9 days (n ⫽ 9), p ⬍ 0.0001), further emphasizing the DC-dependent role of Tregs in a long-term protection against diabetes in NOD mice. Kared et al. (30) have reported that G-CSF prevent diabetes development in NOD mice over a period of 30 wk. The protective effect of GM-CSF was also tested using similar protocol. Female NOD mice received seven consecutive daily injections of GMCSF (100 or 500 ng/mouse) or PBS at ages 4, 8, 12, and 16 wk, and the development of diabetes was monitored over time. As reported in the case of G-CSF treatment (30), NOD recipients of GM-CSF (500 ng/ml) were significantly ( p ⬍ 0.02) protected from diabetes at 30 wk of age (Fig. 9). The lower dose of GM-CSF (100 ng/ml) was less efficient and 60% of the animals had become diabetic compared with 80% in PBS-treated mice at this time (Fig. 9).


FIGURE 9. Effect of two different GM-CSF doses on diabetes development in NOD mice. NOD mice (n ⫽ 10 per group) were injected i.p. with GM-CSF (100 or 500 ng/mouse) or PBS for seven consecutive days at ages 4, 8, 12, and 16 wk. The development of diabetes was monitored over time.

Discussion We showed in this study that GM-CSF, a cytokine that has been previously shown to increase the number of DCs and CD4⫹CD25⫹ Tregs in vivo (12, 13, 31), provided significant protection against the development of diabetes in NOD mice. Our data provided the first evidence that such a protective effect was dependent on expanded DCs acquiring semimature phenotype under maturation stimuli. These DCs produced high amounts of IL-10 and were required to maintain a critical number of immunoregulatory CD4⫹CD25⫹ Tregs that were important for long-term protection against diabetes. GM-CSF is a key mediator of the development of DCs (32). It expands myeloid CD11c⫹CD8␣⫺ DCs and increases the number of CD11c⫹CD11b⫹ DCs (33), two subsets of DCs that play a critical role in induction of tolerance (34). We found that the treatment of NOD mice (before islet insult) with GM-CSF protected NOD mice from diabetes and increased 1.7-fold the number of splenic CD11c⫹ DCs, most of these having a myeloid phenotype (CD11c⫹CD11b⫹CD8␣⫺). This effect of GM-CSF on DC phenotype in NOD mice was consistent with the fact that the protection from other autoimmune diseases such as thyroglobulin-induced autoimmune thyroiditis (12, 13) and experimental autoimmune myasthenia gravis (31), by GM-CSF treatment was associated with expansion of CD11c⫹CD8␣⫺ DCs. Steptoe et al. (19) have shown that autoimmune diabetes could be suppressed by transfer of Gr-1⫹ myeloid progenitor cells. The Gr-1⫹ myeloid cells were preferentially retained in the spleen, continued to express CD11b but had lost Gr-1 expression. Furthermore, many of these myeloid cells had acquired a substantial expression of MHC class II, CD80, CD86, and CD11c, a characteristic of resting DCs. In this study, we do not know whether expanded splenic DCs originated from myeloid Gr-1⫹ precursor. Although there is controversy with respect to the subsets of DCs responsible for peripheral tolerance, it remains that DCs must be in a steady or semimature state to initiate and maintain tolerance (35, 36). In this study, we found that freshly isolated splenic DCs from both groups of animals expressed low levels of CD80, CD86, and MHC class II molecules, and that total and myeloid but not lymphoid splenic DCs from GM-CSF-treated mice were less effective in inducing proliferation of diabetogenic CD8⫹ T cells without affecting their cytokines production. These differences in the capacity to activate proliferation of CD8⫹ T cells could not be explained by the costimulation pathway because the low levels of CD80, CD86, and MHC class II molecules were not significantly different from those of control NOD mice. One possible explana-

3645 tion is that iDCs from GM-CSF-treated mice use an additional inhibitory pathway such as PD-1/PD-L1, which competes with CD80/CD86-dependent costimulation. In support of this interpretation, it has been reported that GM-CSF increased the expression of PD-L1 on bone marrow-derived DCs (37). In addition, it has been observed that PD-1/PD-L1 inhibits CD4⫹ and CD8⫹ T cell priming (38) and its blockade accelerates the development of diabetes in prediabetic NOD mice (39). Finally, a recent study has shown that CTLA-4 and PD-1/PD-L1 played a synergic role in iDCs-dependent induction of peripheral CD8⫹ T cell tolerance (36). The contribution of this pathway in the induction of tolerance in NOD mice treated with GM-CSF is currently under investigation in our laboratories. Because DCs undergo abnormal maturation in NOD mice, we investigated whether the treatment with GM-CSF interfered with this process and restored DCs tolerogenic properties. We found that LPS-activated splenic DCs from GM-CSF-treated mice did not undergo a full process of maturation as compared with control mice that showed increased levels of expression of MHC class II and costimulatory molecules (CD80 and CD86). Of interest, LPSactivated DCs from GM-CSF-treated mice were less efficient in Ag presentation and activation of diabetogenic CD8⫹ T cells as compared with DCs from PBS-treated NOD mice. In agreement with our findings, others groups have reported that LPS-activated bone marrow-derived DCs generated with GM-CSF expressed low levels of CD80/CD86 molecules and suppressed T cell activation (4). In addition, semimature DCs maintained self-tolerance by sustaining a population of CD4⫹CD25⫹ Tregs (40). Similarly, exposure of bone marrow-derived DCs to low doses of GM-CSF led to the development of iDCs that were resistant to maturation stimuli in vitro. These iDCs induced T cell unresponsiveness and prolonged cardiac and islet allograft survival (41, 42). Our results suggested that the GM-CSF-associated prevention from diabetes in NOD mice was largely dependent on the state of maturation of DCs. Because DCs from GM-CSF-treated mice produced amounts of IL-10 greater than DCs from control mice, it is likely that IL-10 produced in the vicinity of DCs influenced their function and their process of maturation. This argument was supported by the findings that autocrine IL-10 prevented the spontaneous maturation of DCs, limited LPS- and CD40-mediated maturation and increased IL-10 production by DCs (43). Previous studies have shown that GM-CSF is a potent growth factor of DCs and a promoter of Th2 response in other model of autoimmunity (12, 13, 31). In agreement with previous findings in T cell-mediated autoimmune thyroiditis, we have shown that treatment with GM-CSF results in mobilizing CD11c⫹ and a skewing of the immune response to Th2-like cells. Unlike Th2 cells, Th2like cells had increased production of IL-4 and IL-10, but continued to produce similar amounts of IFN-␥ and IL-2 as compared with PBS-treated mice. Interestingly these CD4⫹CD25⫺ T cells exhibited higher levels of Foxp3 as compared with CD4⫹CD25⫺ of PBS-treated mice. The role of CD4⫹CD25⫺ Foxp3⫹ T cells in protection from diabetes in GM-CSF-protected NOD mice is not fully known. However, there is evidence that Tregs can be thymically derived or peripherally induced from CD4⫹CD25⫺Foxp3⫺ T cells (29, 44 – 46). Furthermore, retroviral transduction of Foxp3 in naive CD4⫹ T cells convert these cells toward a Treg phenotype (47). These observations support the interpretation that peripheral CD4⫹CD25⫺Foxp3⫹ T cells of GM-CSF-treated mice may originate from naive CD4⫹CD25⫺ T cells that are going to be converted to CD4⫹CD25⫹Foxp3⫹ suppressor T cells. Therefore, it is likely that CD4⫹CD25⫺Foxp3⫹ cells constitute a reservoir of committed regulatory T cells that gain CD25 expression in GMCSF-protected mice, as previously reported (48). Interestingly, a

3646 recent study has shown that IL-2 is essential to convert naive CD4⫹CD25⫺ T cells to CD4⫹CD25⫹Foxp3⫹ Tregs and that IL-4, IL-7, and IL-15 could sustain Foxp3 expression (49). These observations are consistent with the fact that more CD4⫹CD25⫺ T cells from GM-CSF-protected mice express Foxp3 and continue to produce IL-2 and IL-4, two events that contribute to their conversion to CD4⫹CD25⫹Foxp3⫹ Treg and the maintain of Foxp3 expression. Taking in account the fact that CD4⫹CD25⫹ Tregs delayed the onset of diabetes in transfer experiments of diabetogenic spleen cells, we investigated whether long-term protection from diabetes required a continual support of tolerogenic DCs. Therefore it is attracting to suggest that protection from diabetes in GM-CSFtreated NOD mice by CD4⫹CD25⫹ Tregs requires the presence of tolerogenic-resting DCs to sustain their long-term suppressive function. Our results showed that diabetes resistance induced by GM-CSF was associated with an increase in the pool of splenic CD4⫹CD25⫹Foxp3⫹ Tregs. Furthermore, CD4⫹CD25⫹ Tregs from GM-CSF- and PBS-treated mice displayed similar suppressive ability, suggesting that the inhibitory functions of Tregs were not affected by GM-CSF treatment. In this respect, it has been reported that a low number of CD4⫹CD25⫹ Tregs in B7⫺/⫺ and CD28⫺/⫺ NOD mice correlated with exacerbation and acceleration of diabetes (50). It is therefore possible that tolerogenic DCs induced by GM-CSF act indirectly through the induction/ activation of Tregs, a mechanism that would promote long-term tolerance, thereby ensuring persistent islet-specific T cell hyporesponsiveness. CD4⫹CD25⫹ Tregs are anergic and require cognate interactions with DCs for their activation in vivo (51, 52). We determined the contribution of DCs to the tolerance induced by GM-CSF treatment using adoptive transfer experiments and found that splenocytes from GM-CSF-treated mice did not transfer diabetes into NOD.SCID recipients. However, NOD.SCID hosts injected with DC-depleted splenocytes from GM-CSF-treated mice were no longer protected and developed delayed diabetes when compared with NOD.SCID injected with DC-depleted splenocytes from PBS-treated mice. This delay was likely due to the loss of Tregs initially increased in the spleen of GM-CSF-treated mice. Keeping in line with our data, a recent study has reported that a defect in splenic APCs of NOD mice could be responsible for a suppressive defect of CD4⫹CD25⫹ Tregs (53). However, the nature, phenotype and function of APCs were not investigated in this study. Our results confirmed the previous described defect of DCs in NOD mice and presented evidence that GM-CSF reestablished tolerogenic DCs required for maintaining Treg efficiency. Yamazaki et al. (52) have shown that GM-CSF-activated DCs induced a robust proliferation of Tregs in vitro and in vivo. In addition, two recent studies have reported that CD4⫹CD25⫹ Tregs interact directly with DCs and that persistent cell-cell interactions led to sustained immunosuppressive function of CD4⫹CD25⫹ Tregs (54) thereby decreasing the time of contact between DCs and effector cells (55). Our data showed that long-term tolerance induced by GM-CSF treatment in NOD mice can be attributed to re-establishment of sufficient amount of Tregs that are supported by tolerogenic GM-CSF-induced DCs. The protective effect of GM-CSF from diabetes we reported is corroborated with recent data showing that GM-CSF/IL-3-deficient mice developed insulitis and diabetes with age, suggesting that functional deficiencies of GM-CSF and IL-3 contribute to diabetes (56). In addition, other groups have shown that G-CSF affords diabetes protection from several autoimmune diseases including diabetes (30, 57, 58). The beneficial effect of G-CSF therapy in diabetes was shown to be mediated through promotion

DC AND CD4⫹CD25⫹ Treg FUNCTIONS IN NOD MICE of tolerogenic DCs and the expansion of Tregs (30). In agreement with these observations, our data showed that GM-CSF-treated (500 ng/mouse) NOD mice were protected from diabetes over 30 wk using a protocol similar to the G-CSF study (30). Of note, GM-CSF appeared to be more effective than G-CSF because a dose of GM-CSF of 500 ng/mouse that is equivalent to 25–30 ␮g/kg/day brought a protective effect similar to G-CSF used at a dose of 200 ␮g/kg/day (30). In summary, our data showed that GM-CSF modulates the immunoregulatory function of DCs, and strongly suggest that DCs act as guardians of the induction and maintenance of long-term T cell tolerance to islet Ags in the periphery through CD4⫹CD25⫹ Tregs.

Acknowledgments We thank Dr. P. Santamaria for reagents, mice, critical reading of the manuscript, and helpful discussions. We also thank Dr. A. Elandaloussi for helpful comments on the manuscript, and J. N. Silvestro and J. Cayer for animal care and technical assistance.

Disclosures The authors have no financial conflict of interest.

References 1. Banchereau, J., and R. M. Steinman. 1998. Dendritic cells and the control of immunity. Nature 392: 245–252. 2. Mahnke, K., E. Schmitt, L. Bonifaz, A. H. Enk, and H. Jonuleit. 2002. Immature, but not inactive: the tolerogenic function of immature dendritic cells. Immunol. Cell Biol. 80: 477– 483. 3. Lanzavecchia, A., and F. Sallusto. 2001. The instructive role of dendritic cells on T cell responses: lineages, plasticity and kinetics. Curr. Opin. Immunol. 13: 291–298. 4. Lutz, M. B., R. M. Suri, M. Niimi, A. L. Ogilvie, N. A. Kukutsch, S. Rossner, G. Schuler, and J. M. Austyn. 2000. Immature dendritic cells generated with low doses of GM-CSF in the absence of IL-4 are maturation resistant and prolong allograft survival in vivo. Eur. J. Immunol. 30: 1813–1822. 5. Riedl, E., H. Strobl, O. Majdic, and W. Knapp. 1997. TGF-␤1 promotes in vitro generation of dendritic cells by protecting progenitor cells from apoptosis. J. Immunol. 158: 1591–1597. 6. Wakkach, A., N. Fournier, V. Brun, J. P. Breittmayer, F. Cottrez, and H. Groux. 2003. Characterization of dendritic cells that induce tolerance and T regulatory 1 cell differentiation in vivo. Immunity 18: 605– 617. 7. Steinbrink, K., M. Wolfl, H. Jonuleit, J. Knop, and A. H. Enk. 1997. Induction of tolerance by IL-10-treated dendritic cells. J. Immunol. 159: 4772– 4780. 8. Penna, G., and L. Adorini. 2000. 1␣,25-dihydroxyvitamin D3 inhibits differentiation, maturation, activation, and survival of dendritic cells leading to impaired alloreactive T cell activation. J. Immunol. 164: 2405–2411. 9. Sato, K., N. Yamashita, M. Baba, and T. Matsuyama. 2003. Regulatory dendritic cells protect mice from murine acute graft-versus-host disease and leukemia relapse. Immunity 18: 367–379. 10. Kleindienst, P., C. Wiethe, M. B. Lutz, and T. Brocker. 2005. Simultaneous induction of CD4 T cell tolerance and CD8 T cell immunity by semimature dendritic cells. J. Immunol. 174: 3941–3947. 11. Krakowski, M., R. Abdelmalik, L. Mocnik, T. Krahl, and N. Sarvetnick. 2002. Granulocyte macrophage-colony stimulating factor (GM-CSF) recruits immune cells to the pancreas and delays STZ-induced diabetes. J. Pathol. 196: 103–112. 12. Vasu, C., R. N. Dogan, M. J. Holterman, and B. S. Prabhakar. 2003. Selective induction of dendritic cells using granulocyte macrophage-colony stimulating factor, but not fms-like tyrosine kinase receptor 3-ligand, activates thyroglobulinspecific CD4⫹/CD25⫹ T cells and suppresses experimental autoimmune thyroiditis. J. Immunol. 170: 5511–5522. 13. Gangi, E., C. Vasu, D. Cheatem, and B. S. Prabhakar. 2005. IL-10-producing CD4⫹CD25⫹ regulatory T cells play a critical role in granulocyte-macrophage colony-stimulating factor-induced suppression of experimental autoimmune thyroiditis. J. Immunol. 174: 7006 –7013. 14. Tisch, R., and H. McDevitt. 1996. Insulin-dependent diabetes mellitus. Cell 85: 291–297. 15. Santamaria, P. 2001. Effector lymphocytes in autoimmunity. Curr. Opin. Immunol. 13: 663– 669. 16. Boudaly, S., J. Morin, R. Berthier, P. Marche, and C. Boitard. 2002. Altered dendritic cells (DC) might be responsible for regulatory T cell imbalance and autoimmunity in nonobese diabetic (NOD) mice. Eur. Cytokine Network 13: 29 –37. 17. Feili-Hariri, M., and P. A. Morel. 2001. Phenotypic and functional characteristics of BM-derived DC from NOD and non-diabetes-prone strains. Clin. Immunol. 98: 133–142. 18. Marleau, A. M., and B. Singh. 2002. Myeloid dendritic cells in non-obese diabetic mice have elevated costimulatory and T helper-1-inducing abilities. J. Autoimmun. 19: 23–35.

The Journal of Immunology 19. Steptoe, R. J., J. M. Ritchie, and L. C. Harrison. 2002. Increased generation of dendritic cells from myeloid progenitors in autoimmune-prone nonobese diabetic mice. J. Immunol. 168: 5032–5041. 20. Ott, P. A., M. R. Anderson, M. Tary-Lehmann, and P. V. Lehmann. 2005. CD4⫹CD25⫹ regulatory T cells control the progression from periinsulitis to destructive insulitis in murine autoimmune diabetes. Cell Immunol. 235: 1–11. 21. Brusko, T. M., C. H. Wasserfall, M. J. Clare-Salzler, D. A. Schatz, and M. A. Atkinson. 2005. Functional defects and the influence of age on the frequency of CD4⫹CD25⫹ T-cells in type 1 diabetes. Diabetes 54: 1407–1414. 22. Lee, M., A. Y. Kim, and Y. Kang. 2000. Defects in the differentiation and function of bone marrow-derived dendritic cells in non-obese diabetic mice. J. Korean Med. Sci. 15: 217–223. 23. Strid, J., L. Lopes, J. Marcinkiewicz, L. Petrovska, B. Nowak, B. M. Chain, and T. Lund. 2001. A defect in bone marrow derived dendritic cell maturation in the nonobesediabetic mouse. Clin. Exp. Immunol. 123: 375–381. 24. Pop, S. M., C. P. Wong, D. A. Culton, S. H. Clarke, and R. Tisch. 2005. Single cell analysis shows decreasing FoxP3 and TGF␤1 coexpressing CD4⫹CD25⫹ regulatory T cells during autoimmune diabetes. J. Exp. Med. 201: 1333–1346. 25. Chen, Z., A. E. Herman, M. Matos, D. Mathis, and C. Benoist. 2005. Where CD4⫹CD25⫹ T reg cells impinge on autoimmune diabetes. J. Exp. Med. 202: 1387–1397. 26. Salomon, B., D. J. Lenschow, L. Rhee, N. Ashourian, B. Singh, A. Sharpe, and J. A. Bluestone. 2000. B7/CD28 costimulation is essential for the homeostasis of the CD4⫹CD25⫹ immunoregulatory T cells that control autoimmune diabetes. Immunity 12: 431– 440. 27. Verdaguer, J., D. Schmidt, A. Amrani, B. Anderson, N. Averill, and P. Santamaria. 1997. Spontaneous autoimmune diabetes in monoclonal T cell nonobese diabetic mice. J. Exp. Med. 186: 1663–1676. 28. Lutz, M. B., N. Kukutsch, A. L. Ogilvie, S. Rossner, F. Koch, N. Romani, and G. Schuler. 1999. An advanced culture method for generating large quantities of highly pure dendritic cells from mouse bone marrow. J. Immunol. Methods. 223: 77–92. 29. Hori, S., T. Nomura, and S. Sakaguchi. 2003. Control of regulatory T cell development by the transcription factor Foxp3. Science 299: 1057–1061. 30. Kared, H., A. Masson, H. Adle-Biassette, J. F. Bach, L. Chatenoud, and F. Zavala. 2005. Treatment with granulocyte colony-stimulating factor prevents diabetes in NOD mice by recruiting plasmacytoid dendritic cells and functional CD4⫹CD25⫹ regulatory T-cells. Diabetes 54: 78 – 84. 31. Sheng, J. R., L. Li, B. B. Ganesh, C. Vasu, B. S. Prabhakar, and M. N. Meriggioli. 2006. Suppression of experimental autoimmune myasthenia gravis by granulocyte-macrophage colony-stimulating factor is associated with an expansion of FoxP3⫹ regulatory T cells. J. Immunol. 177: 5296 –5306. 32. Hamilton, J. A., and G. P. Anderson. 2004. GM-CSF biology. Growth Factors 22: 225–231. 33. Basak, S. K., A. Harui, M. Stolina, S. Sharma, K. Mitani, S. M. Dubinett, and M. D. Roth. 2002. Increased dendritic cell number and function following continuous in vivo infusion of granulocyte macrophage-colony-stimulating factor and interleukin-4. Blood 99: 2869 –2879. 34. Ardavin, C. 2003. Origin, precursors and differentiation of mouse dendritic cells. Nat. Rev. Immunol. 3: 582–590. 35. Hawiger, D., K. Inaba, Y. Dorsett, M. Guo, K. Mahnke, M. Rivera, J. V. Ravetch, R. M. Steinman, and M. C. Nussenzweig. 2001. Dendritic cells induce peripheral T cell unresponsiveness under steady state conditions in vivo. J. Exp. Med. 194: 769 –779. 36. Probst, H. C., J. Lagnel, G. Kollias, and M. van den Broek. 2003. Inducible transgenic mice reveal resting dendritic cells as potent inducers of CD8⫹ T cell tolerance. Immunity 18: 713–720. 37. Yamazaki, T., H. Akiba, H. Iwai, H. Matsuda, M. Aoki, Y. Tanno, T. Shin, H. Tsuchiya, D. M. Pardoll, K. Okumura, et al. 2002. Expression of programmed death 1 ligands by murine T cells and APC. J. Immunol. 169: 5538 –5545. 38. Carter, L., L. A. Fouser, J. Jussif, L. Fitz, B. Deng, C. R. Wood, M. Collins, T. Honjo, G. J. Freeman, and B. M. Carreno. 2002. PD-1:PD-L inhibitory pathway affects both CD4⫹ and CD8⫹ T cells and is overcome by IL-2. Eur. J. Immunol. 32: 634 – 643. 39. Ansari, M. J., A. D. Salama, T. Chitnis, R. N. Smith, H. Yagita, H. Akiba, T. Yamazaki, M. Azuma, H. Iwai, S. J. Khoury, et al. 2003. The programmed

3647

40.

41.

42.

43.

44. 45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

death-1 (PD-1) pathway regulates autoimmune diabetes in nonobese diabetic (NOD) mice. J. Exp. Med. 198: 63– 69. Verginis, P., H. S. Li, and G. Carayanniotis. 2005. Tolerogenic semimature dendritic cells suppress experimental autoimmune thyroiditis by activation of thyroglobulin-specific CD4⫹CD25⫹ T cells. J. Immunol. 174: 7433–7439. DePaz, H. A., O. O. Oluwole, A. O. Adeyeri, P. Witkowski, M. X. Jin, M. A. Hardy, and S. F. Oluwole. 2003. Immature rat myeloid dendritic cells generated in low-dose granulocyte macrophage-colony stimulating factor prolong donor-specific rat cardiac allograft survival. Transplantation 75: 521–528. Rastellini, C., L. Lu, C. Ricordi, T. E. Starzl, A. S. Rao, and A. W. Thomson. 1995. Granulocyte/macrophage colony-stimulating factor-stimulated hepatic dendritic cell progenitors prolong pancreatic islet allograft survival. Transplantation 60: 1366 –1370. Corinti, S., C. Albanesi, A. la Sala, S. Pastore, and G. Girolomoni. 2001. Regulatory activity of autocrine IL-10 on dendritic cell functions. J. Immunol. 166: 4312– 4318. Khattri, R., T. Cox, S. A. Yasayko, and F. Ramsdell. 2003. An essential role for Scurfin in CD4⫹CD25⫹ T regulatory cells. Nat. Immunol. 4: 337–342. Chen, W., W. Jin, N. Hardegen, K. J. Lei, L. Li, N. Marinos, G. McGrady, and S. M. Wahl. 2003. Conversion of peripheral CD4⫹CD25⫺ naive T cells to CD4⫹CD25⫹ regulatory T cells by TGF-␤ induction of transcription factor Foxp3. J. Exp. Med. 198: 1875–1886. Fontenot, J. D., M. A. Gavin, and A. Y. Rudensky. 2003. Foxp3 programs the development and function of CD4⫹CD25⫹ regulatory T cells. Nat. Immunol. 4: 330 –336. Walker, M. R., D. J. Kasprowicz, V. H. Gersuk, A. Benard, M. Van Landeghen, J. H. Buckner, and S. F. Ziegler. 2003. Induction of FoxP3 and acquisition of T regulatory activity by stimulated human CD4⫹CD25⫺ T cells. J. Clin. Invest. 112: 1437–1443. Zelenay, S., T. Lopes-Carvalho, I. Caramalho, M. F. Moraes-Fontes, M. Rebelo, and J. Demengeot. 2005. Foxp3⫹ CD25⫺ CD4 T cells constitute a reservoir of committed regulatory cells that regain CD25 expression upon homeostatic expansion. Proc. Natl. Acad. Sci. USA 102: 4091– 4096. Zheng, S. G., J. Wang, P. Wang, J. D. Gray, and D. A. Horwitz. 2007. IL-2 Is Essential for TGF-␤ to Convert Naive CD4⫹CD25⫺ cells to CD25⫹Foxp3⫹ regulatory T cells and for expansion of these cells. J. Immunol. 178: 2018 –2027. Bour-Jordan, H., B. L. Salomon, H. L. Thompson, G. L. Szot, M. R. Bernhard, and J. A. Bluestone. 2004. Costimulation controls diabetes by altering the balance of pathogenic and regulatory T cells. J. Clin. Invest. 114: 979 –987. Cederbom, L., H. Hall, and F. Ivars. 2000. CD4⫹CD25⫹ regulatory T cells down-regulate co-stimulatory molecules on antigen-presenting cells. Eur. J. Immunol. 30: 1538 –1543. Yamazaki, S., T. Iyoda, K. Tarbell, K. Olson, K. Velinzon, K. Inaba, and R. M. Steinman. 2003. Direct expansion of functional CD25⫹CD4⫹ regulatory T cells by antigen-processing dendritic cells. J. Exp. Med. 198: 235–247. Alard, P., J. N. Manirarora, S. A. Parnell, J. L. Hudkins, S. L. Clark, and M. M. Kosiewicz. 2006. Deficiency in NOD antigen-presenting cell function may be responsible for suboptimal CD4⫹CD25⫹ T-cell-mediated regulation and type 1 diabetes development in NOD mice. Diabetes 55: 2098 –2105. Tang, Q., J. Y. Adams, A. J. Tooley, M. Bi, B. T. Fife, P. Serra, P. Santamaria, R. M. Locksley, M. F. Krummel, and J. A. Bluestone. 2006. Visualizing regulatory T cell control of autoimmune responses in nonobese diabetic mice. Nat. Immunol. 7: 83–92. Tadokoro, C. E., G. Shakhar, S. Shen, Y. Ding, A. C. Lino, A. Maraver, J. J. Lafaille, and M. L. Dustin. 2006. Regulatory T cells inhibit stable contacts between CD4⫹ T cells and dendritic cells in vivo. J. Exp. Med. 203: 505–511. Enzler, T., S. Gillessen, M. Dougan, J. P. Allison, D. Neuberg, D. A. Oble, M. Mihm, and G. Dranoff. 2007. Functional deficiencies of granulocyte-macrophage colony stimulating factor and interleukin-3 contribute to insulitis and destruction of ␤ cells. Blood 110: 954 –961. Zavala, F., S. Abad, S. Ezine, V. Taupin, A. Masson, and J. F. Bach. 2002. G-CSF therapy of ongoing experimental allergic encephalomyelitis via chemokine- and cytokine-based immune deviation. J. Immunol. 168: 2011–2019. Zavala, F., A. Masson, K. Hadaya, S. Ezine, E. Schneider, O. Babin, and J. F. Bach. 1999. Granulocyte-colony stimulating factor treatment of lupus autoimmune disease in MRL-lpr/lpr mice. J. Immunol. 163: 5125–5132.