Protection Against Type 1 Diabetes Upon Coxsackievirus B4 Infection and iNKT-Cell Stimulation: Role of Suppressive Macrophages

Female proinsulin 2–deficient (Proins2^−/−) NOD mice, Vα14 transgenic NOD mice expressing the Vα14-Jα18 T-cell receptor α chain, and BDC2.5 Cα^−/− mice were previously described (15,21,25,26). NOD Vα14 were crossed with Proins2^−/− NOD mice to generate Vα14 Proins2^−/− NOD. Mice were bred and housed in specific pathogen-free conditions. This study was approved by the local ethics committee on animal experimentation (P2.AL.171.10).

CVB4 Edwards strain 2 was injected intraperitoneally at the dose of 1 × 10⁵ plaque-forming units (PFU)/mouse. When indicated, mice were treated with a single injection of αGalCer 2 µg/mouse i.p. (Alexis) diluted in PBS/Tween 0.05% at the time of CVB4 infection. For short-term blockade of IFN-γ and IL-4, mice were injected with purified anti-IFN-γ monoclonal antibody (mAb) (R46A2) 0.5 mg i.p., anti-IL-4 mAb (11B11) 0.5 mg i.p., or corresponding isotype controls on days −1 and +1 of virus infection for PCR analysis and on days −1, +1, +3 for diabetes incidence. IL-13 was blocked with soluble extracellular domain of IL-13 receptor 10 µg i.p. b.i.d. on days −1, 0, and +1 of infection. A selective inducible nitric oxide synthase (iNOS) inhibitor, 1400W 10 mg/kg/day i.p. (Calbiochem), and a selective arginase I inhibitor, Nω-hydroxy-nor-l-arginine (nor-NOHA) 20 mg/kg/day i.p. (Calbiochem), were injected starting from the day of infection and up to day 8. To inhibit IDO, mice were given 1-methyl-tryptophan (1MT) 4 mg/mL (Sigma) in drinking water 2 days before infection and for up to 8 days afterward. In some experiments, 2 × 10⁵ macrophages were isolated from the pancreas of Proins2^−/− mice treated with CVB4 or CVB4 + αGalCer 2 days earlier and transferred intravenously into recipient Proins2^−/− mice infected with CVB4 1 day earlier.

Pancreata were recovered and homogenized in liquid maintenance medium 199 (#11825-015; Gibco) complemented with distilled water (65%), sodium bicarbonate (2.7%), PBS (11.5%), penicillin/streptomycin (1.6%), and l-glutamine (1.6%) and centrifuged at 2,300 rpm for 20 min. Tenfold serial dilutions of the supernatant were overlaid on HeLa cell monolayers and incubated for 2 h at 37°C. The monolayers were washed with PBS and overlaid with mixed equal portions of maintenance medium containing FCS (PAA Laboratories) instead of PBS and 2.4% suspension of Avicel (RC581; BMC Biopolymer). Two days later, the overlay was removed, and the monolayers were fixed with formaldehyde and colored with crystal violet ammonium oxalate solution.

Overt diabetes was defined as two positive urine glucose tests of glycemia >200 mg/dL 48 h apart (Glukotest kit; Roche). For histology analysis, paraffin-embedded sections were cut at three levels (200-µm intervals) and stained with hematoxylin-eosin. Insulitis severity was scored in a blinded fashion by two examiners with the following criteria: grade 0, no infiltration; grade 1, peri-islet lymphocytic infiltration; grade 2, <50% islet lymphocytic infiltration; and grade 3, >50% islet lymphocytic infiltration. At least 40 islets from each mouse were analyzed.

Pancreata were perfused with 5 mL Collagenase P solution (0.75 mg/mL) (Roche), dissected free from surrounding tissues, digested for 10 min at 37°C, and washed twice with RPMI medium-10% FCS. Islets were then purified on a Ficoll gradient and incubated with 1 mL nonenzymatic cell dissociation buffer (Invitrogen) for 10 min at 37°C and dissociated into a single-cell suspension by pipetting.

The following mAbs were used: CD45 (30F11), CD11b (M1/70), Ly-6G (RB6–8C5), F4/80 (BM8), CD115 (AFS98), IL-4 (11B11), and IL-13 (eBio13A) (all from eBiosciences) and CD11c (HL3), 120G8, Ly-6C (AL21), IFN-γ (XMG1.2), CD62L (Mel-14), CD4 (GK1.5), CD8 (53-6.7), and anti-human Ki-67 (B56) (all from BD Biosciences). Stainings were performed in 5% FCS in PBS for 20 min at 4°C. Nonspecific Fc binding was blocked with an anti-CD16/CD32 antibody (24G2). Allophycocyanin–conjugated αGalCer-loaded CD1d tetramer was prepared in our laboratory. For cytokine stainings, cells were stimulated with phorbol myristic acid 10 ng/mL and ionomycine 1 µg/mL in the presence of Brefeldin A 1 mg/mL for 4 h at 37°C (all from Sigma). After the surface staining, cells were fixed, permeabilized for 30 min with cytofix-cytoperm kit (BD Biosciences), and incubated with intracellular mAbs for 30 min. Cells were either analyzed with a BD Fortessa flow cytometer or sorted with a BD ARIA II sorter.

RNA was extracted with an RNeasy mini kit (Qiagen) and reverse transcribed with Superscript III (Invitrogen). Quantitative RT-PCR was performed with SYBR Green (Roche) and analyzed with a LightCycler 480 (Roche). Relative expression was calculated by the 2–∆∆Ct method and normalized to the expression of the housekeeping gene GAPDH. The stability of GAPDH expression was confirmed by comparison with HPRT mRNA (Supplementary Fig. 1).

Single-cell suspensions of pancreatic islets were cultured in the presence of islet-specific glucose-6-phosphatase catalytic subunit–related protein (IGRP)_206–214 peptide (VYLKTNVFL) 10 μmol/L for 4 h 30 min at 37°C in the presence of Brefeldin A 1 mg/mL. After surface staining, cells were fixed and permeabilized, and intracellular IFN-γ staining was performed. The proliferation was assessed in a thymidine incorporation assay. Sorted naïve BDC2.5⁺ CD4 T cells (3 × 10⁴) were cultured with anti-CD3/CD28 beads (Invitrogen) and 3 × 10⁴ macrophages isolated from the pancreas of mice either infected with CVB4 alone or infected and treated with αGalCer. After 48-h culture, wells were pulsed with 1-μCi tritiated thymidine ([³H]-TdR) overnight. [³H]-TdR incorporation was measured with a TopCount counter (PerkinElmer) at the Cochin cytometry and immunobiology facility. 1MT was prepared as a 20 mmol/L stock solution in 0.1 mol/L NaOH and added to the T-cell culture at a final concentration of 200 μmol/L.

Diabetes incidence was plotted according to the Kaplan-Meier method. Incidences between groups were compared with the log-rank test. For other experiments, comparison between means was performed with the nonparametric Mann-Whitney U test. P < 0.05 was considered statistically significant. All data were analyzed with Prism version 5 software (GraphPad Software).

As previously described by Serreze et al. (11), 50% of NOD mice developed diabetes after CVB4 infection at 10 weeks of age, whereas the remaining mice stayed diabetes free for up to 30 weeks of age (Fig. 1A). Analysis of a subset of diabetic mice showed that diabetes acceleration after CVB4 infection was significant compared with noninfected mice (Fig. 1B). Similar data were obtained with Proins2^−/− mice (Fig. 1C and D), which were used because they develop accelerated spontaneous type 1 diabetes compared with wild-type NOD mice, thus shortening the duration of experiments. Proins2^−/− mice were used at 5–6 weeks of age when they already exhibited moderate insulitis. Of note, one single injection of the iNKT-cell agonist αGalCer at the time of infection strongly decreased diabetes incidence in both NOD and Proins2^−/− mice (13 and 25%, respectively). Insulitis was also reduced in CVB4 + αGalCer Proins2^−/− mice compared with CVB4 infection only (Fig. 1E). Because islet β-cell infection could promote diabetes development (12,27), we analyzed CVB4 pancreatic and islet β-cell infection. Viral titers determined by PFU were similar in the pancreas from infected mice treated or not with αGalCer (Fig. 1F and Supplementary Fig. 2A). Although in situ hybridization showed CVB4 infection of exocrine tissue, it did not reveal islet β-cell infection on day 3 and 7 postinfection (data not shown). Altogether, the data show that iNKT-cell activation decreased the incidence of diabetes following CVB4 infection in NOD and Proins2^−/− mice.

To study the mechanism of diabetes prevention by activated iNKT cells, the inflammatory response was analyzed in pancreatic islets of CVB4-infected mice. At day 2 postinfection, the expression of proinflammatory cytokines, such as IL-1β, IL-6, and TNF-α, was significantly decreased in Proins2^−/− and NOD mice from the CVB4 + αGalCer group compared with the untreated infected mice (Fig. 2 and Supplementary Fig. 2B). No differences were found in the mRNA level of the suppressive cytokines IL-10 and transforming growth factor-β (data not shown). Strikingly, αGalCer treatment at the time of infection induced a strong upregulation of the suppressive enzymes iNOS, IDO1 and IDO2 (hereafter referred as IDO), and arginase I. Although arginase I was induced by αGalCer alone, the strongest upregulation of iNOS and IDO required both αGalCer and CVB4 infection. Of note, Ym1/Ym2 mRNA was also upregulated in the islets from CVB4 + αGalCer mice, suggesting the presence of alternatively activated macrophages (28,29). The expression of these molecules was transitory, with a peak on day 2 postinfection (Supplementary Fig. 3). Thus, the activation of iNKT cells during CVB4 infection favors the establishment of a less inflammatory and more immunosuppressive environment in pancreatic islets compared with CVB4 infection only.

We next investigated which cell populations expressed the immunosuppressive enzymes iNOS, IDO, arginase I, and Ym1/Ym2. On day 2 postinfection, the highest expression of these molecules was detected in myeloid cells isolated from pancreatic islets (islet myeloid cells) (CD11b⁺/CD11c^–) of Proins2^−/− mice from the CVB4 + αGalCer group compared with CD45⁺ DCs (CD11c⁺), plasmacytoid DCs (CD11c^low/120G8⁺), remaining CD11b^–/CD11c^–, and CD45^– cells (Fig. 3A and B). Of note, these enzymes were not detected in sorted populations of pancreatic lymph node (PLN) and spleen from the same mice tested up to 10 days posttreatment (data not shown).

Further characterization of islet CD11b⁺/CD11c^– myeloid cells showed that they were predominantly F4/80⁺ and Ly-6G^– macrophages and that their frequency did not differ between CVB4- and CVB4 + αGalCer–treated mice (Fig. 4A). Of note, the kinetics of their infiltration into islets correlated with the pattern of expression of inflammatory and suppressive molecules in total islets (Supplementary Fig. 4A). Macrophages were further stained with Ly-6C and CD115 mAbs because these molecules can be upregulated on suppressive macrophages (30,31). However, after infection, Ly-6C and CD115 were upregulated to the same extent independently of αGalCer treatment. Similar data were obtained in NOD mice (data not shown). Analysis of sorted Ly-6C⁺/CD115⁺ and Ly-6C^–/CD115^– pancreatic infiltrating cells revealed that the inflammatory cytokines and suppressive enzymes were mainly expressed by Ly-6C⁺/CD115⁺ macrophages (Fig. 4B and data not shown). Of note, Ly-6C⁺ macrophages from spleen and PLN did not express these suppressive enzymes (Supplementary Fig. 4B). Altogether, these results indicate that CVB4 infection induces the infiltration of Ly-6C⁺/CD115⁺ macrophages into pancreatic islets, which resemble classically activated macrophages (CAMϕ) strongly expressing inflammatory cytokines. However, when iNKT cells are activated, macrophages express suppressive enzymes characteristic of myeloid-derived suppressor cells (MDSCs).

Because iNKT-cell manipulation leads to infiltration of suppressive macrophages and has a major impact on the development of diabetes after CVB4 infection, iNKT cells were analyzed in pancreatic islets (Fig. 5). Vα14 transgenic Proins2^−/− mice were used for this study because they exhibit a 10-fold increased frequency and number of iNKT cells, making them easier to detect. CVB4 infection induced the activation of pancreatic iNKT cells that upregulated CD69 without increasing their proliferation (no Ki-67 upregulation) and their cytokine production. In contrast, αGalCer injection led to a strong iNKT-cell proliferation and massive production of IFN-γ, IL-4, and IL-13. We next investigated the role of these cytokines in the induction of MDSC and alternatively activated macrophages in mice from the CVB4 + αGalCer group because IFN-γ can induce the expression of iNOS and IDO, whereas IL-4 and IL-13 can induce arginase I and Ym1/Ym2 expression. The blockade of IL-4 by specific mAb did not alter the expression of any of these enzymes at day 2 postinfection (Fig. 6). In contrast, blocking IFN-γ by a specific mAb significantly reduced the expression of iNOS, IDO, and Ym1/Ym2, whereas blocking IL-13 significantly decreased the expression of arginase I. Thus, cytokines produced by iNKT cells might be the key mediators in the induction of pancreatic suppressive macrophages after CVB4 infection.

To determine the role of suppressive enzymes in the protection against diabetes, Proins2^−/− mice from the CVB4 + αGalCer group were treated with specific inhibitors of iNOS, IDO, and arginase I or control vehicles. Treatment with the iNOS inhibitor 1400W or the arginase I inhibitor nor-NOHA did not abrogate the protection against diabetes (Fig. 7A). However, although the IDO inhibitor 1MT did not affect diabetes incidence of control Proins2^−/− mice (Fig. 7B), it abolished the protection of mice from the CVB4 + αGalCer group because 86% became diabetic compared with only 27% from the CVB4 + αGalCer group treated with vehicle. Of importance, combined treatment with the three inhibitors resulted in an incidence of diabetes similar to that of the IDO inhibitor only, suggesting that iNOS and arginase I did not play a significant role in the protection against diabetes in this setting. Consistent with the incidence of diabetes, 1MT treatment increased the severity of insulitis in mice from the CVB4 + αGalCer group (Fig. 7C). Because IDO expression in islets was induced by IFN-γ, we next tested the role of IFN-γ in the protection against diabetes. Short-term IFN-γ blockade during the period when IDO is detected in islets increased diabetes incidence of CVB4 + αGalCer–treated mice (Fig. 7D). Of note, both 1MT treatment and IFN-γ blockade induced a moderate increase of diabetes incidence in mice from the CVB4 group (Supplementary Fig. 5). Altogether, these results show that IDO plays a critical role in the inhibition of diabetes development during CVB4 infection and iNKT-cell activation.

The suppressive capacity of macrophages was then assessed in vitro in a T-cell proliferation assay (Fig. 7E). Macrophages were isolated from the pancreas of Proins2^−/− mice from the CVB4 and CVB4 + αGalCer groups on day 2 postinfection and cultured with naïve BDC2.5 CD4 T cells stimulated with anti-CD3/CD28 beads. Macrophages from CVB4-infected mice enhanced BDC2.5 CD4 T-cell proliferation, even though this increase did not reach statistical significance. On the contrary, macrophages from mice of the CVB4 + αGalCer group significantly suppressed T-cell proliferation, and the suppression was reversed when 1MT was added to the culture. To demonstrate the role of pancreatic islet–infiltrating macrophages in vivo, macrophages were sorted from pancreatic islets of either CVB4- or CVB4 + αGalCer–treated Proins2^−/− mice on day 2 postinfection and transferred into CVB4 recipients infected the day before. The transfer of macrophages from the CVB4 group significantly increased diabetes incidence of the CVB4-infected recipient mice, whereas the transfer of macrophages from the CVB4 + αGalCer group significantly reduced diabetes incidence (Fig. 7F). Altogether, these results highlight the dual role of pancreatic macrophages after CVB4 infection.

To further decipher the immune mechanisms involved in the development of diabetes after CVB4 infection and its inhibition by iNKT-cell activation, we evaluated specific anti-islet T-cell responses. CVB4-infected Proins2^−/− and NOD mice were analyzed 1–3 weeks postinfection when diabetes was clearly established. Untreated, αGalCer-treated, or CVB4-infected nondiabetic mice were tested in parallel. CVB4 infection dramatically increased the total numbers of both CD4 and CD8 T cells infiltrating the pancreas of Proins2^−/− and NOD mice compared with untreated and mice treated only with αGalCer (Fig. 8A and B and Supplementary Fig. 6A). IGRP-specific IFN-γ–producing CD8 T cells were detected in all Proins2^−/− and NOD mice that became diabetic after CVB4 infection, regardless of the treatment received. In contrast, the frequency of IGRP-specific IFN-γ–producing CD8 T cells remained low in untreated or nondiabetic infected mice (Fig. 8C and D and Supplementary Fig. 6B). Treatment with IDO inhibitor, which abolished diabetes protection, induced a strong anti-IGRP effector CD8 T-cell response only in diabetic mice. Altogether, the results strongly suggest that diabetes development after CVB4 infection is caused by anti-islet T-cell responses.

The current study shows that although CVB4 infection can rapidly induce diabetes in a subset of Proins2^−/− and NOD mice, the concomitant activation of iNKT cells inhibited diabetes development after CVB4 infection very efficiently. This prevention was associated with the reduction of inflammatory cytokine expression and the induction of suppressive enzymes in pancreatic infiltrating macrophages. Pancreatic iNKT cells produced IFN-γ, which was required for the induction of high levels of IDO that prevented diabetes onset. The development of diabetes was associated with an increased anti-islet T-cell response, whereas the frequency of these IFN-γ–producing auto-reactive T cells remained very low in nondiabetic mice.

Numerous studies have shown the pathogenic role of macrophages in type 1 diabetes (32–35). CD11b⁺, F4/80⁺, Ly-6C⁺, and CD115⁺ inflammatory macrophages (36) highly infiltrated pancreatic islets after CVB4 infection in both Proins2^−/− and NOD mice but differed in their regulatory functions between mice treated and mice not treated with αGalCer. Pancreas of mice that developed diabetes after CVB4 infection harbored macrophages with a classically activated phenotype expressing high levels of IL-1β, IL-6, and TNF-α, which can favor diabetes development (37–39). According to these previous articles, it is not surprising that as a consequence of macrophage depletion and low expression of inflammatory cytokines, fewer mice developed diabetes after CVB4 infection (Supplementary Fig. 7).

In contrast to the production of proinflammatory cytokines by the pancreatic macrophages from CVB4-infected mice, in the CVB4 + αGalCer group, macrophages produced suppressive enzymes, including IDO, which was required for the decrease of diabetes incidence in these mice. The high IDO expression was a result of IFN-γ, a cytokine highly produced by αGalCer-activated iNKT cells. IFN-γ alone was not sufficient for diabetes protection because αGalCer-treated mice had diabetes incidence similar to that of untreated mice despite high levels of IFN-γ. Of importance, although CVB4 infection induced some IDO, αGalCer treatment alone did not induce any, suggesting that viral infection might be the initiator of IDO induction. Of note, previous studies have suggested the role of IDO in the control of diabetes development. The poor induction of IDO in DCs from prediabetic NOD mice was shown to favor diabetes development (40), whereas transplanted pancreatic islets are protected by IDO-expressing neighboring fibroblasts (41).

CVB4-induced diabetes depends on lymphocytes because NOD Scid, NOD Igµ^−/− (11), and NOD Cα^−/− (data not shown) mice do not develop diabetes after CVB4 infection. Similarly, CVB4-infected Proins2^−/− Scid mice did not become diabetic (data not shown). Diabetes development only in a subset of CVB4-infected Proins2^−/− mice could result from differences in insulitis level before infection, as suggested in NOD mice (11). Although CD8 T-cell numbers greatly increased in islets of all infected mice, IFN-γ–producing anti-IGRP CD8 T cells were only detected in the pancreas of diabetic mice, therefore strengthening the role of T cells in diabetes induced by CVB4 as previously suggested (11,42). The inhibition of IDO in CVB4 + αGalCer–treated mice resulted in a strong anti-IGRP CD8 T-cell effector response and a high incidence of diabetes, showing the protective role of IDO in diabetes. Most often, IDO induces tolerance by suppressing T-cell proliferation through tryptophan depletion or by inducing apoptosis through by-products of tryptophan degradation, such as kynurenines (43,44). However, the numbers of T cells did not differ in mice expressing a low or high level of IDO, suggesting that IDO does not affect the proliferation of T cells but, rather, induces their local inhibition in the pancreas. The present data show that IDO inhibiting the function of CD8 T cells is reminiscent of a previous study in the allograft setting (45). Moreover, local regulation by IDO has been previously observed in other infectious contexts as well as in cancer (46).

Of note, the present study suggests that IFN-γ can play a protective or deleterious role in diabetes development, depending on the setting (Fig. 8E). We hypothesize that a strong burst of IFN-γ early after viral infection upregulates IDO expression, which is initially induced by CVB4 infection to downregulate the inflammation after viral infection. In contrast, in the absence of iNKT-cell activation, the production of inflammatory cytokines could promote the recruitment and activation of pathogenic T cells, which would accumulate and produce IFN-γ. At this later point, IDO is no longer expressed in the pancreas, and IFN-γ production would lead to islet β-cell death rather than to IDO upregulation. Of note, we did not detect any difference of viral load in the pancreas of infected mice treated or not with αGalCer. However, we cannot rule out that IFN-γ, produced by αGalCer treatment, might contribute to the prevention of diabetes by dampening CVB4 infection specifically in islet β-cells. The infection of islet β-cells could lead to type 1 diabetes development because it would favor the engulfment of infected islet β-cells by allophycocyanin, presentation of islet β-cell autoantigens to autoreactive T cells, and initiation of diabetes (27,47,48). Indeed, studies in mouse models and human samples have shown that IFNs, including IFN-γ, are important in regulating islet β-cell permissiveness of infection and replication in islet β-cells (12,49,50). Of note, some reports have described islet β-cell infection by enterovirus in recent type 1 diabetic patients (51,52). Therefore, IFN-γ production in the islets of CVB4 + αGalCer–treated mice could prevent diabetes not only by inducing IDO but also by limiting islet β-cell infection.

IDO-mediated suppression could involve several mechanisms, including regulatory FoxP3⁺ T-cell induction in pancreatic islets (53). However, in the present study, the percentages of FoxP3⁺ CD4 T cells; their expression of CTLA-4, CD103, GITR, and OX40; and their production of IL-10 and transforming growth factor-β were similar in the islets, PLN, and spleen of infected Proins2^−/− and NOD mice with or without αGalCer treatment (Supplementary Fig. 8 and data not shown). Tryptophan metabolites can also render DCs tolerogenic with intermediate expression of stimulatory major histocompatibility complex II, CD86, and upregulation of the suppressive programmed death ligands 1 and 2 (54). These molecules were highly upregulated by DCs in the pancreas, PLN, and spleen after the infection. However, the expression of these molecules was similar between CVB4-infected mice treated or not with αGalCer. Similarly, the production of IL-10 and IL-12 by DCs did not differ in any of the organs studied (data not shown).

In conclusion, this study shows how the manipulation of iNKT cells can lead to the induction of suppressive macrophages in the pancreas of CVB4-infected mice and highlights the role of macrophages in the development or prevention of diabetes after CVB4 infection. The outcome of this study can help in the design of novel therapeutic approaches to manipulating iNKT cells and suppressive macrophages.

L.G. was supported by doctoral fellowships from the Ministère de l’Education Nationale. This work was supported by funds from INSERM and CNRS and grant ANR-09-GENO-023 and Laboratoire d’Excellence INFLAMEX to A.L. A.L. is the recipient of an Assistance Publique-Hôpitaux de Paris-CNRS Contrat Hospitalier de Recherche Translationnelle.

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

L.G. researched data, performed experiments, and wrote the manuscript. J.D. researched data and reviewed the manuscript. L.B. and P.G.L. performed experiments. R.K.P. provided the IL-13 inhibitor and reviewed and edited the manuscript. N.v.R. provided clodronate liposomes and reviewed and edited the manuscript. M.F.-T. provided CVB4 and reviewed and edited the manuscript. A.L. designed and supervised the study and wrote the manuscript. L.G. and A.L. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank R. Mallone from INSERM U1016, for IGRP peptide and the staff of the mouse facility for animal care.