Innate pro–B-cell progenitors protect against type 1 diabetes by regulating autoimmune effector T cells

Significance

Immunoregulatory poperties have been principally ascribed to various mature immune cell types, including regulatory B cells. An immature B-cell progenitor population endowed with suppressive properties per se or after differentiation into more mature regulatory B cells has not been demonstrated as yet. We now describe a pro–B-cell progenitor population that emerged upon stimulation with the Toll-like receptor-9 ligand CpG and prevented disease upon adoptive transfer into autoimmune type 1 diabetes-prone mice. Effector T cells were the target of immunoregulatory pro-B cells and of their mature progeny. Such protective pro-B cells could be instrumental for cell therapy of autoimmune diseases.

Abstract

Diverse hematopoietic progenitors, including myeloid populations arising in inflammatory and tumoral conditions and multipotent cells, mobilized by hematopoietic growth factors or emerging during parasitic infections, display tolerogenic properties. Innate immune stimuli confer regulatory functions to various mature B-cell subsets but immature B-cell progenitors endowed with suppressive properties per se or after differentiating into more mature regulatory B cells remain to be characterized. Herein we provide evidence for innate pro-B cells (CpG-proBs) that emerged within the bone marrow both in vitro and in vivo upon Toll-like receptor-9 activation and whose adoptive transfer protected nonobese diabetic mice against type 1 diabetes (T1D). These cells responded to IFN-γ released by activated effector T cells (Teffs), by up-regulating their Fas ligand (FasL) expression, which enabled them to kill Teffs through apoptosis. In turn, IFN-γ derived from CpG-proBs enhanced IFN-γ while dramatically reducing IL-21 production by Teffs. In keeping with the crucial pathogenic role played by IL-21 in T1D, adoptively transferred IFN-γ–deficient CpG-proBs did not prevent T1D development. Additionally, CpG-proBs matured in vivo into diverse pancreatic and splenic suppressive FasL^high B-cell subsets. CpG-proBs may become instrumental in cell therapy of autoimmune diseases either on their own or as graft complement in autologous stem cell transplantation.

A growing body of evidence attests that immune cells with immunoregulatory functions do not exclusively belong to mature populations of diverse lineage, but also comprise several hematopoietic progenitor subsets. The first subset to be recognized comprised myeloid progenitors that acquired suppressive properties in tumoral and inflammatory environments (1) and played either detrimental or beneficial roles in different pathological situations. We have reported previously that mobilization with hematopoietic growth factors conferred tolerogenic properties to multipotent hematopoietic progenitors at the multipotent progenitor (MPP2) stage of differentiation that enabled them to promote the expansion of regulatory T cells (2, 3), thereby preventing spontaneous autoimmune type 1 diabetes (T1D) in the nonobese diabetic (NOD) mouse model. Moreover, parasitic infections were shown to stimulate via IL-25 the emergence of Th2 cytokine-secreting MPPs (MPP^Th2) that ultimately differentiated into mature cell types with pro-Th2 functions, thus contributing to parasitic clearance (4).

Direct interactions between pathogens and hematopoietic stem cells occur through Toll-like receptor (TLR) activation, driving their differentiation along myeloid pathways to enforce anti-infectious defenses (5, 6). TLR agonists also promote hematopoiesis by enhancing the production of the hematopoietic growth factor G-CSF, with whom they synergize to mobilize hematopoietic stem cells from the bone marrow to the periphery (7).

TLR-mediated innate-type stimulation by infectious (8) and parasitic (9) agents also plays a major role in promoting the emergence of regulatory B cells (Bregs), along with acquired-type stimulation, such as B-cell receptor (BCR) engagement concomitant or not with CD40 activation (10, 11). Such induced regulatory B-cell functions are believed to be more robust than those expressed by naive and resting B cells, which can nevertheless tolerize naive T cells and induce regulatory T cells (Tregs) (12, 13).

Bregs are a heterogeneous lymphocyte subset present among all major B-cell populations (14–17). The rare so-called B10 cells identified by their CD19⁺CD1d^hiCD5⁺ phenotype (18–20), peritoneal CD5⁺ B1a cells (21, 22), large follicular B cells, and activated transitional, marginal zone (MZ) B cells can all acquire regulatory properties. The most immature Breg subset described so far is composed of B220⁺IgM⁺CD21^lowCD93⁺CD23⁺ transitional T2 MZ precursor B (T2 MZP-B) cells, which are continuously produced in adult bone marrow and home to the MZ of the spleen, where they differentiate into IgM^highCD1d^highCD21^highCD23^low MZ B cells (23, 24).

The differentiation pathways of the various Breg subsets remain unknown. Only functional precursors, named “B10pro,” which are mature B cells requiring additional BCR-activating antigenic signals to produce immunosuppressive IL-10 but cannot be distinguished from B10 cells by phenotypic criteria, have hitherto been identified (25). It is unknown whether Bregs derive from one or several progenitors or solely from conventional B-cell subsets. Moreover, an immature B-cell progenitor population endowed with suppressive properties per se or after differentiation into more mature Bregs has not been demonstrated as yet.

Herein we describe a hematopoietic progenitor population that emerges transiently in vitro and in vivo in the bone marrow of NOD mice, after activation with the TLR-9 agonist CpG and whose adoptive transfer into NOD mice prevents T1D onset. These cells were c-kit^lowSca-1^lowCD127⁺B220⁺CD19⁺IgM⁻CD1d^intCD43⁺, a phenotype consistent with a pro–B-cell stage of differentiation, except for their CD1d expression. The cells differentiated in vivo exclusively into B lymphocytes, at various stages of maturation.

Functionally, these TLR-induced hematopoietic progenitors suppressed pathogenic effector T cells (Teffs) by reducing their IL-21 production and by inducing their apoptosis via Fas ligand (FasL). Additionally, the B-cell progeny of CpG-induced proBs continued to express high levels of FasL and to suppress Teffs, and may play a major role in the durable protection against T1D provided by the progenitors in vivo.

Results

c-kit^lowSca-1^lowB220⁺ Bone Marrow Progenitors Emerging upon Exposure to CpG Prevent T1D.

Incubation of bone marrow cells from NOD mice with the oligonucleotide CpG 1668 (CpG-B), but not with control GpC oligonucleotide, led to the transient emergence of a c-kit^lowSca-1⁺ cell population within 18 h. These cells were heterogeneous in terms of size, Sca-1 and B220 expression (Fig. 1A), and could be further sorted into a small-size Sca-1^lowB220⁺IgM⁻ and a large-size Sca-1^highB220⁻IgM⁻ fraction. Adoptive transfer to NOD mice at 6 wk of age revealed that only small-size CpG-induced c-kit^lowSca-1^lowB220⁺IgM⁻ cells protected against T1D in vivo, whereas their large-size c-kit^lowSca-1^highB220⁻IgM⁻ counterpart had no such effect (Fig. 1B). c-kit⁺Sca-1⁺ cells isolated from bone marrow cells incubated with the control oligonucleotide GpC that were negative for B220 and IgM expression, did not prevent the onset of T1D (Fig. 1B). Note that small-size c-kit^lowSca-1^lowB220⁺IgM⁻ cells remained protective in 66% of mice when injected as late as at 16 wk of age, which corresponds to disease onset (Fig.1C).

Fig. 1.

Phenotypic characterization and prevention of T1D in vivo by a CpG-induced c-kit^lowSca-1^lowB220⁺IgM⁻ bone marrow subset. (A) Bone marrow (BM) cells, incubated for 18 h with CpG-B (1 μg/mL), were magnetically selected for c-kit⁺ cells, further stained for Sca-1, B220, and IgM and electronically sorted into large-size (FSC^highSSC^high) c-kit^lowSca-1^highB220⁻IgM⁻ and small-size (FSC^lowSSC^low) c-kit^lowSca-1^lowB220⁺IgM⁻ cells. BM cells incubated with the control oligonucleotide GpC were electronically sorted as c-kit⁺Sca-1⁺ cells that were B220⁻IgM⁻. (B) Diabetes incidence in NOD mice injected with PBS or after intravenous transfer of the above sorted subsets at 6-wk of age, P = 0.0021 by Kaplan–Meier estimates when comparing cumulative incidence curves for controls and for CpG-induced c-kit^lowSca-1^lowB220⁺IgM⁻ progenitor recipients, not significant for other groups. (C) Incidence of T1D in NOD mice injected intravenously either with PBS or with 60 × 10³ CpG-induced c-kit^lowSca-1^lowB220⁺IgM⁻ progenitors at 16 wk of age, P = 0.0328. Results in B and C are pooled from two experiments. (D) Phenotypic characterization of the protective bone marrow progenitor subset by flow cytometry analysis of expression of CD19, CD127, CD43, IgM, CD1d, and CD5. Cells were stained with specific antibodies (open histograms) or control isotype antibodies (filled histograms) after FACS sorting. CD1d level on CpG-induced c-kit^lowSca-1^lowB220⁺IgM^- progenitors (red histogram) was compared with that measured on splenic follicular B cells (FoB, blue histogram) and MZ B cells (MZB, black histogram). (E) Percentage of c-kit^lowSca-1^lowB220⁺IgM⁻ cells emerging in the bone marrow of NOD mice after 18-h incubation with different TLR-agonists. CpG-B was tested in bone marrow cultures of both NOD and NOD MyD88^−/− mice. Results are expressed as mean ± SEM of three experiments. *P < 0.05, when comparing BM incubated with different stimuli to unstimulated BM.

Protective c-kit^lowSca-1^lowB220⁺ Cells Share the Characteristics of B-Cell Progenitors at the Pro-B Stage.

The protective c-kit^lowSca-1^lowB220⁺ subset expressed CD19 at low levels together with the IL-7Rα chain CD127 and CD43 but no IgM, in keeping with an immature B-cell phenotype at a pro–B-cell stage of differentiation. CpG-induced cells were also positive for CD1d, at intermediate levels between CD1d^low follicular B cells and CD1d^high MZ B cells, but CD5 was not consistently detected (Fig. 1D). Among the TLR agonists tested, those activating TLR-2, -4, -5, -6, and -7 induced a progenitor population with the same phenotype as that induced by CpG, conversely to compounds targeting TLR-1 and -3 (Fig. 1E). CpG-A (1586) and -B (1826 or 1668) had a similar effect. The protective pro-B population did not arise in the bone marrow of myeloid differentiation primary response gene 88 (MyD88)-deficient NOD mice incubated with CpG-B (Fig.1E).

At variance with the data of Welner et al. (26), who used C57BL/6 mice to establish that incubation with CpG-A for 48 h directed common lymphocyte progenitors toward differentiation into PDCA-1⁺ plasmacytoid dendritic cells (PDCs) at the expense of the B-cell lineage, the protective pro-B subset induced in the NOD mouse after a 18-h incubation with CpG-B did not express the dendritic cell markers CD11c or PDCA-1.

CpG-Induced Pro-B Cells Mature Exclusively into B Cells in Vivo.

Small-size c-kit^lowSca-1^lowB220⁺IgM⁻ cells, isolated from congenic NOD CD45.2 bone marrow cells incubated with CpG-B were electronically sorted before transfer to NOD mice (70,000 progenitors per mice). The protective progenitors were then analyzed in different tissues for migration and in vivo differentiation potential. Approximately 50% of injected CD45.2⁺ cells were found as early as 5 d after adoptive transfer in the pancreas, whereas only 7% were located in the pancreatic lymph nodes (PLNs) at the same time point (Fig. 2A). CD45.2⁺ cells gradually decreased over time in pancreas and PLNs, becoming undetectable at day 30 posttransfer. Conversely, between day 10 and day 20 posttransfer, a large proportion of CD45.2⁺ cells, representing ∼65% of injected cells, migrated to the spleen where their numbers remained stable afterward. CD45.2⁺ cells remained detectable up to 3 mo posttransfer in the spleen. A minor percentage was present in the peritoneal cavity, but not in bone marrow, blood, or liver.

Fig. 2.

Time-dependent evaluation of recovered CD45.2⁺ cells in pancreas, PLNs, and spleen. (A) Absolute cell counts of total CD45.2⁺ cells recovered from the pancreas, PLNs, and spleen of CD45.1⁺ mice at different time points after injection of 70,000 CD45.2⁺ protective progenitors per mouse. Results are expressed as the mean ± SEM of two to four mice per point. (B) FACS analysis of the maturation status of CD45.2⁺ cells in recipient mice over time. The proportion of IgM^low/+CD45.2⁺ cells in pancreas, PLNs and spleen (Left) as well as their expression of c-kit and B220 (Right) at indicated time points after progenitor injection in CD45.1⁺ NOD recipients are depicted. A representative experiment of two to four is shown.

The differentiation status of injected CD45.2⁺ cells was followed by their expression of IgM, c-kit, and B220 (Fig. 2B). CD45.2⁺ cells have matured into IgM^lowc-kit⁺B220⁺ cells in the pancreas from day 5 until day 20, and in PLNs as far as day 5. Thereafter in PLNs, between day 10 and day 20, CD45.2⁺ cells gradually lost the expression of c-kit, while acquiring higher IgM expression. Finally, in the spleen, the bulk of CD45.2⁺ cells were c-kit⁻B220⁺IgM⁺, corresponding to mature B cells as the final progeny.

The mature c-kit⁻IgM⁺B220⁺CD45.2⁺ B-cell progeny was further analyzed by FACS (Fig. 3A) in the spleen and PLNs at day 30 and at day 20 posttransfer, respectively, where it consisted exclusively of B cells sharing the CD19⁺B220⁺CD43⁻IgM⁺IgD⁺ phenotype. Neither CD11b⁺ nor CD11c⁺ cells were found within CD45.2⁺ cells (Fig. 3B). In the spleen, these CD45.2⁺ cells represented ∼0.1% of all cells and 0.3% of B cells and were CD1d⁺CD5^-/low (Fig. 3B). These cells were mostly follicular (40%) and MZ (15%) B cells, as well as T2 MZ B precursors (15%) and CD21⁺CD23^high cells (15%) (Fig. 3 B and C), as defined by their relative expression of CD93, IgM, CD21, and CD23 (27).

Fig. 3.

Phenotypic characterization of the IgM⁺CD45.2⁺ progeny of CpG-induced pro–B-cell progenitors. (A) Flow cytometry analysis of the IgM⁺CD45.2⁺ progeny in CpG-proB–injected recipients compared with untreated CD45.1⁺ controls, for cell-surface expression of the B-cell markers CD19, B220, CD43, IgM, and IgD, performed in spleen at day 30 and in PLNs at day 20 after injection of the CD45.2⁺ progenitors. (B) Flow cytometry analysis of the expression of CD11b and CD11c within splenic CD45.2⁺ cells and of the frequency of the B220⁺CD45.2⁺ splenic progeny of the CpG-induced proB cell progenitors, recovered 1 mo after progenitor injection, in diverse B-cell subfractions according to the expression of CD5 and CD1d, or CD21, IgM, CD93, and CD23. (C) Histogram representation of the percentages of the different B-cell subsets as defined in B, among injected CD45.2⁺ cells, CD45.1⁺ cells of the recipients and cells from noninjected age-matched control mice. Data from A and B are from one representative experiment of at least three. Data in C are expressed as mean ± SEM of three experiments.

Furthermore, the adoptive transfer of the progenitors had no significant effect on the proportion of the various B-cell subsets in the spleen of the recipients, relative to age-matched noninjected controls (Fig. 3C).

CpG-proBs Suppress Teff Proliferation and Trigger Their Apoptosis.

To assess whether CpG-proBs modulated Treg or Teff proliferation, we cocultured c-kit^lowSca-1^lowB220⁺IgM⁻ CpG-proBs with either CD4⁺CD25^high Tregs (all Foxp3⁺) or CD4⁺CD25⁻ Teffs, each T-cell population electronically sorted from the spleen of NOD mice. CpG-proBs did not modify the proliferation of Tregs, but suppressed the expansion of Teffs, as measured by dilution of carboxyfluorescein diacetate succinimidyl ester (CFSE) staining (Fig. 4A) after 5 d of coculture at a Teff:CpG-proB ratio of 2:1. We assessed by FACS analysis among gated B220⁺ or CD4⁺ cells that during the 5 d of coculture, the CpG-proB cells remained essentially alive, excluding Topro III (Fig. 4B), whereas only 12% of T cells cultured alone but ∼70% of cocultured T cells died, incorporating ToproIII (Fig. 4C). In contrast, neither the c-kit⁺Sca-1⁺ subset sorted from bone marrow cells incubated with the control oligonucleotide GpC nor control pro-B cells purified ex vivo from bone marrow cells as a c-kit⁺Sca-1⁻B220⁺CD24^hiCD43^hi population shared this suppressive function (Fig. 4D). This result occurred independently from IL-2 (Fig. 4E) and coincided with apoptosis of Teffs evidenced both by annexin-V staining and To-ProIII incorporation (Fig. 4 F and G). CpG-proBs expressed a variety of molecules inducing death and tolerance, such as tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and programmed death ligand (PDL)-1, PDL-2, respectively. However, none of the corresponding neutralizing antibodies restored Teff proliferation, apart from anti-FasL antibody, which allowed a recovery of cell divisions (Fig. 4 H and I). Notably, FasL was expressed not only on CpG-proB cells derived from bone marrow cultures with CpG (Fig. 5C) but also on the pancreatic CD45.2⁺IgM^lowB220⁺c-kit⁺ cells recovered between day 5 and day 20 after adoptive transfer of the progenitors and on the splenic mature IgM⁺B220⁺CD45.2⁺ B-cell progeny recovered 30 d after progenitor injection. Notably, FasL levels of the CpG-proB progeny were even higher than those exhibited by the recipient’s IgM⁺CD45.2^- cells analyzed at the same time in the corresponding tissues (Fig. 4J).

Fig. 4.

Mechanisms underlying the suppressive properties of CpG-proBs. (A) Proliferation of sorted CFSE-loaded CD4⁺CD25⁻ Teffs 5 d after activation with anti-CD3 + anti-CD28 when cocultured at a 2:1 ratio with CpG-proBs. (B and C) Viability of gated B220⁺ CpG-proBs (B) and gated CD4⁺ T-cells (C) in the coculture. Dead and live cells are distinguished by their capacity to take up or exclude Topro III, respectively. (D) Proliferation of sorted CFSE-loaded CD4⁺CD25⁻ Teffs 5 d after activation with anti-CD3 + anti-CD28 when cocultured at a 2:1 ratio with control proBs or c-kit⁺Sca-1⁺ cells from GpC-treated bone marrow. (E) Teff proliferation in coculture with CpG-proBs with or without 10 U/mL of IL-2. (F and G) Apoptosis measured by annexin V staining of phosphatidylserine and Topro III incorporation among gated Teffs cocultured with CpG-proBs (ratio 2:1) at day 5 after anti-CD3 + anti-CD28 activation. Numbers in the Upper Right quadrants represent the percentage of dead cells. Histograms in G represent the mean ± SEM of four different experiments, *P = 0.0286. (H and I) Reversal of the antiproliferative effect of CpG-proBs in coculture with Teffs by a neutralizing anti-FasL antibody (5 μg/mL). T-cell proliferation assessed at day 5 by the numbers of cells in which the CFSE content is diluted in response to anti-CD3 + anti-CD28, either alone or in coculture with CpG-proBs (Teff:CpG-proB ratio at 2:1) with or without neutralizing anti-FasL antibody. Data in I are expressed as means ± SEM of three to five experiments. **P = 0.008. (J) FasL expression by the c-kit⁺B220⁺IgM^lowCD45.2⁺ cells found in the pancreas 5 d after transfer, compared with that of the c-kit^-B220⁺IgM⁺CD45.2⁺ splenic progeny sorted after 1 mo, and to that of the recipient’s IgM⁺CD45.2^- cells found in pancreas and spleen at the same timings, respectively, relative to staining with isotype control antibodies (K and L) Ability of the splenic CD45.2⁺B220⁺IgM⁺ progeny in coculture with Teffs (ratio Teff:CpG-proB progeny of 3:1) to induce Teff apoptosis (K), assessed by Topro III incorporation and Annexin V staining (numbers in the Upper Right quadrant represent percentages of dead cells) and to suppress the proliferation of CFSE-loaded Teffs (L) at day 5 of coculture. Results shown in A–F, H, and J–L are from one representative experiment of two to five.

Fig. 5.

Contribution of Teff- and CpG-proB-derived IFN-γ to suppression, cytokine switching and protection against T1D induced by CpG-proBs. (A) CpG-proBs cocultured with CFSE-loaded Teffs (Teff:CpG-proB ratio at 2:1) derived from NOD mice do not suppress Teff proliferation when measured at day 3 but suppress Teff proliferation measured at day 5. (B) CpG-proBs suppress the proliferation of CFSE-loaded Teffs in cocultures on day 5 only when derived from WT-NOD mice, not when derived from IFN-γ^−/− NOD mice. (C) FasL expression by gated B220⁺ CpG-proBs cocultured with Teffs from either WT or IFN-γ^−/− mice on days 0 and 5. (D) IFN-γ assays in supernatants at day 5 of cocultures set up with CpG-proBs combined either with IFN-γ–competent or IFN-γ–deficient Teffs. *P < 0.05. (E) Intracytoplasmic levels of IFN-γ measured by FACS in gated B220⁺ CpG-proB cells before (d0) and after culture for 5 d (d5) alone or together with Teff at a cell ratio of 2 Teff for 1 CpG-proB. (F) Cell-sorting procedure and phenotype analysis by flow cytometry of CpG-proBs prepared from bone marrow cells of IFN-γ^−/− NOD mice incubated for 18 h with CpG-B. (G) Nuclear eomesodermin expression measured by FACS at day 5 in gated CD4⁺ Teffs sorted from WT-NOD donors and cultured alone or cocultured (Teff:CpG-proB ratio at 2:1) with CpG-proBs derived either from WT or from IFN-γ^−/− NOD mice. (H) qRT-PCR determination of IFN-γ (Left axis) and IL-21 (Right axis) mRNA levels relative to 18S in Teffs cultured for 5 d alone or together with CpG-proBs isolated from WT or IFN-γ–deficient NOD mice. Data are expressed as means ± SEM of three experiments. *P < 0.05. (I) Diabetes incidence in control NOD mice injected at 6 wk of age with PBS (●, n = 18 mice) or with CpG-proBs (60,000 cells per mouse) prepared from IFN-γ-deficient NOD mice (□, n = 16 mice). n.s., not significant. Data are pooled from two experiments.

Moreover, when electronically sorted 1 mo after adoptive transfer of the progenitors, the splenic CD45.2⁺B220⁺IgM⁺ B-cell progeny triggered Teff apoptosis during coculture (Fig. 4K) and suppressed Teff proliferation (Fig. 4L).

Teffs Control Their Own Expansion by Producing IFN-γ, Which Confers Suppressive Properties to CpG-proBs.

CpG-proBs required 5 d of coculture with Teffs to express optimal suppressive activity. On day 3, the cells had no effect on their target cells, emphasizing the requirement of prior conditioning by activated Teffs (Fig. 5A). In addition, even after 5 d the cells failed to inhibit the proliferation of Teffs sorted from IFN-γ–deficient NOD mice (Fig. 5B). This lack of effect coincided with a significantly reduced up-regulation of FasL on the surface of CpG-proBs (Fig. 5C), in support of the conclusion that IFN-γ derived from activated inflammatory Teffs renders CpG-proBs suppressive by up-regulating their FasL expression.

Progenitor-Derived IFN-γ Reduces IL-21 Production by Teffs During Coculture with CpG-proBs and Is Required in Vivo for Protection Against T1D by CpG-proBs.

After 5 d of coculture with CpG-proBs, spared Teffs were electronically sorted and their mRNA was extracted. Real-time RT-PCR analysis revealed that over 30 genes of 89 were modulated in the Lonza Th17 pathway-oriented transcripts (Table 1). In keeping with the enhanced FasL expression on the surface of CpG-proBs, FasL mRNA levels were increased 55-fold in Teffs, suggesting enhanced suicide. Transcripts for cytotoxicity-related molecules, such as granzyme B and the transcription factor eomesodermin, were likewise amplified. The latter is known for inhibiting Th17 differentiation (28), which is further hampered in our experimental setup by the down-regulation of IL-12Rβ1, the binding site of IL-23 (29). Instead, Th1-oriented gene expression was strongly up-regulated, including IFN-γ, t-Bet (Tbx21), and IL-12Rβ2, the target of IL-12 p70, whereas transcription of genes associated with a Th2 profile, such as GATA-3 and IL-13, was reduced. In addition, up-regulation of eomesodermin, which causes the repression of IL-21 production (28), occurred together with decreased transcription of inducible costimulator (ICOS), a positive regulator of IL-21 (30). Both effects culminated in a dramatic 77-fold reduction of mRNA encoding IL-21, an autocrine factor abundantly produced by Th17 cells that promotes or sustains Th17 lineage commitment (31). Given that IL-21 does also inhibit IFN-γ production in developing Th1 cells by repressing eomesodermin expression (32), the regulatory loop initiated by CpG-proBs clearly favors Th1 over Th17 differentiation of Teffs.

Table 1.

Modulation of gene expression in Teffs after coculture with CpG-proBs

Gene name	Fold-change	Statistical significance
FasL	55.676342	0.000699
Eomes	6.314982	0.003092
Gzmb	4.599565	0.006441
S1pr1	−5.795792	0.007901
Ifng	14.098450	0.008533
Ccr4	−21.421090	0.008623
Gata3	−5.284992	0.011812
Socs3	−7.477339	0.015402
Il21	−77.062769	0.016882
Il12rb2	2.388677	0.017140
Ccl5	4.384641	0.018714
Il17rd	−3.359100	0.019037
Cxcl10	1.806543	0.020413
Csf2	2.887744	0.020796
Stat3	1.579430	0.021611
S100a9	2.726780	0.022745
Fas	−3.538830	0.023138
Il12rb1	−2.424316	0.023798
Il13	−11.022945	0.023816
Rn18s	1.052367	0.025393
Cd4	−1.403812	0.026754
Itgal	1.129510	0.027668
Cd28	1.244431	0.030131
Il6	1.899298	0.030779
Tgfb1	1.081882	0.031073
Jak2	−2.067066	0.031177
Prkcq	−1.883761	0.034156
Icos	−1.975810	0.035781
Il2	2.716978	0.035822
Tbx21	1.766007	0.036636
Smad7	−1.555453	0.038206
Il10	1.849752	0.039904
Lif	−2.066163	0.042685
Stat5a	−9.092846	0.045780

After 5 d of culture with or without CpG-proBs (ratio 2:1), live Teffs were electronically sorted, their mRNA was extracted, and reverse transcription performed followed by qPCR analysis. Global Pattern Recognition online analysis provided the ratio of gene modulation and their statistical significance. n = 5 samples from different experiments per cell culture condition.FasL, fas ligand (TNF superfamily, member 6); Eomes, eomesodermin; Gzmb, granzyme B; S1pr1, sphingosine-1-phosphate receptor 1; Ifng, interferon gamma; Ccr4, C-C chemokine receptor type 4; Gata 3, gata binding protein 3; Socs3, suppressor of cytokine signaling 3; Il21, interleukin 21; Il12rb2, interleukin 12 receptor, beta 2 subunit; Ccl5, chemokine (C-C motif) ligand 5; Il17rd, interleukin 17 receptor D; Cxcl10, chemokine (C-X-C motif) ligand 10; Csf2, colony Stimulating Factor 2; Stat3, signal transducer and activator of transcription 3; S100a9: S100 calcium binding protein A9; Fas, Fas (TNF receptor superfamily member 6); Il12rb1, interleukin 12 receptor, beta 1 subunit; Il13, interleukin 13; Rn18s, 18S ribosomal RNA; Cd4, CD4 antigen; Itga1, integrin, alpha 1; Cd28, CD28 antigen; IL6, interleukin 6; Tgfb1, transforming growth factor, beta 1; Jak2, Janus Kinase 2; Prkcq, protein Kinase C, theta; ICOS, inducible T-cell co-stimulator; IL-2, interleukin 2; Tbx21, T-box transcription factor 21; IL-10, interleukin-10; LIF, leukemia inhibiting factor; Stat5a, signal transducer and activator of transcription 5a.

Cytokine assays in supernatants from cocultures set up with Teffs and CpG-proBs from wild-type or IFN-γ–deficient mice (Fig. 5D) suggested that optimal production of IFN-γ depended on or originated from both cellular sources. Intracytoplasmic flow cytometry revealed that CpG-proBs already produced IFN-γ when sorted from CpG-treated bone marrow cells, a production that was further enhanced during coculture with activated Teffs, but declined in CpG-proBs cultured alone (Fig. 5E). Taking advantage of the fact that similar progenitors emerged in bone marrow cells from IFN-γ–deficient NOD mice cultured with CpG-B (Fig. 5F), we evaluated how progenitor-derived IFN-γ affected IL-21 production by Teffs by coculturing wild-type target cells with CpG-proBs recovered from mice that were either competent or deficient for the production of IFN-γ. We found that this cytokine was essential both for increased eomesodermin expression (Fig. 5G) and enhanced IFN-γ transcription in Teffs themselves, as well as for the decrease of IL-21 mRNA levels (Fig. 5H). The importance in vivo as well of IFN-γ production by CpG-proBs was further emphasized by the failure of CpG-proBs prepared from IFN-γ–deficient NOD mice to provide long-lasting protection against T1D in NOD mice (Fig. 5I). Hence, IFN-γ, generated by both Teffs and CpG-proBs, plays a key role in the regulatory cross-talk between the two cell populations.

Finally, sphingosine 1-phosphate receptor, which is required for Teff egress from secondary lymphoid organs, was reduced fivefold upon coculture (Table 1), suggesting that exposure to CpG-proBs not only killed Teffs, and hampered their Th17 differentiation, but eventually prevented their recruitment to target tissues.

CpG-proB–Induced Protection Against T1D Is Associated with Decreased IL-21 Production and Effector Memory T-Cell Loss but Not with Modulation of IL-10 Production.

IL-21 has recently been reported for its deleterious role in both early and late phases of T1D in the NOD mouse (33–35). Pancreatic IL-21 levels were significantly reduced in NOD mice having received CpG-proBs (Fig. 6A). Plasmatic levels of IL-21 were below detection limits. These data suggested that the CpG-proBs and their progeny, which prevented T1D onset, also targeted IL-21–producing Teffs in vivo in the NOD mouse. This finding could result from their killing Teffs or their reducing the capacity of T-cells to produce IL-21. We found that activated CD4⁺CD44⁺CD62L⁺ memory effectors were significantly diminished in the spleen, PLNs, and particularly in the pancreas of the CpG-proB recipients (Fig. 6B) as soon as 2 wk after progenitor transfer, without a significant modification of CD4⁺CD25⁺Foxp3⁺ Tregs [14.01 ± 0.49 vs. 10.92 ± 1.5% of spleen CD4⁺ cells (mean ± SEM, n.s.) and 12.4 ± 1.55 vs. 8.84 ± 1.03% of pancreatic lymph node CD4⁺ cells (mean ± SEM, n.s.)], in control mice relative to CpG-proB recipients. Moreover, pancreatic CD4⁺ cells, 2 wk after progenitor transfer, became more efficient IFN-γ producers, while decreasing their secretion of IL-21 (Fig. 6C). Hence, both Teff numbers and their capacity to produce IL-21 were similarly reduced in vitro in CpG-proB:Teff cocultures and in vivo in CpG-proB recipients.

Fig. 6.

In vivo effects of CpG-proBs on pancreatic IL-21 levels in NOD recipients and on the incidence of effector memory T cells. (A) IL-21 levels (pg/mL) were quantified by ELISA in homogenates of whole pancreas, collected from control mice vs. CpG-proB recipients, 9 wk after progenitor transfer [i.e., in 15-wk-old animals (n = 4 mice per group, P = 0.0286)]. (B) Proportions of effector memory T cells in the spleen (Left), PLNs (Center), and pancreas (Right) of NOD controls and CpG-proB-recipients were assessed 2 wk after progenitor transfer and expressed as percent CD44⁺CD62L⁺ cells in CD4⁺ cells (n = 8 mice per group, P = 0.0286 in all three sites). (C) Intracytoplasmic IL-21 and IFN-γ production was measured 2 wk after progenitor transfer by FACS in pancreatic CD4⁺ cells after PMA + ionomycin stimulation for 5 h in controls and CpG-proB recipients. (D) Intracytoplasmic IL-10 production was measured by FACS in spleen and PLN CD4⁺ cells 2 wk after progenitor transfer after PMA + ionomycin stimulation for 5 h and in spleen and PLN CD19⁺ cells after 48 h stimulation with LPS (10 μg/mL) followed by PMA + ionomycin stimulation for 5 h. Results in C and D are from one representative experiment of three.

Conversely, no significant modulation of IL-10 production was observed either in spleen or PLN CD4⁺ T cells after activation ex vivo with phorbol 12-myristate 13-acetate (PMA) + ionomycin or among CD19⁺ cells gated from spleen cells activated for 48 h with LPS followed by PMA + ionomycin (Fig. 6D).

Suppressive ProB-Cell Progenitors Emerge in Vivo in CpG-Treated Mice.

Finally, we examined whether suppressive B-cell progenitors were also generated in vivo in mice having received TLR-9 activation. As shown in Fig. 7A, c-kit^lowSca-1^lowB220⁺IgM⁻ small-size cells were effectively detected in the bone marrow of NOD mice 18 h after injection of CpG-B (30 μg/mouse, i.p.), similarly to in vitro kinetics. Electronically sorted cells shared the same phenotype and suppressive effect on Teffs (Fig. 7B) as their culture-derived counterpart.

Fig. 7.

In vivo emergence of suppressive CpG-proBs in CpG-treated mice. (A) Small-size c-kit⁺Sca-1⁺B220⁺IgM⁻ cells emerge in the bone marrow of NOD mice 18 h post injection of CpG-B (30 μg/mouse, i.p.). (B) CpG-proBs sorted from the bone marrow of CpG-injected NOD mice suppress Teff proliferation when cocultured at a Teff:CpG-proB ratio of 1:1, as measured by dilution of incorporated CFSE (Left). The data are representative of three experiments.

Discussion

We have shown herein that bone marrow cells respond to the TLR-9 agonist CpG by giving rise to a transient population of c-kit^lowSca-1^low progenitors at the pro-B stage that, conversely to control Sca-1⁻ proBs, provide protection against T1D upon adoptive transfer into NOD mice. We refer to this unique protective B-cell progenitor population as CpG-proBs. Progenitor populations with similar phenotypes, of which the suppressive properties remain to be established, similarly emerge upon activation of TLR-2, -4, -5, -6, and -7, but compounds targeting TLR-1 and -3 as well as GpC had no such effect. The fact that CpG-A, a poor direct activator of B cells, triggers the emergence of a similar progenitor population, may suggest that other non–B-cells may be the primary target of CpG in the bone marrow, leading to the emergence of the CpG-proB cells. CpG-proBs are highly efficient for immunotherapy because a single adoptive transfer of less than 100,000 could prevent T1D in the NOD mouse, at both early and prediabetes stages.

The immunosuppressive functions of CpG-proBs develop in vitro through a cross-talk with activated T cells, which produce IFN-γ and thereby up-regulate FasL expression on CpG-proBs. As a result, CpG-proBs become capable of inducing apoptosis of target T cells, including diabetogenic Teffs. In addition, CpG-proBs induced a shift of Teff functions toward cytotoxicity and a cytokine profile dominated by a clear-cut enhancement of IFN-γ and its target genes IL-12Rβ2, granzyme B, and the t-Box transcription factors t-Bet and eomesodermin. Up-regulation of IFN-γ and of eomesodermin (32), a negative regulator of IL-21 production, together with down-regulation of ICOS, a positive regulator of IL-21 (30) in target Teffs, led to their dramatically reduced IL-21 production. CpG-proBs from IFN-γ–deficient NOD mice were far less potent in shifting the cytokine profile of Teffs in vitro and similarly unable to provide protection against T1D when transferred to NOD recipients. IFN-γ clearly plays a key role in vitro in the cross-talk between CpG-proBs and Teff, inasmuch as Teff-derived IFN-γ promotes FasL-dependent cytotoxicity of CpG-proBs toward Teffs; CpG-proB–derived IFN-γ accounts for the cytokine switch within the Teff population. However, whether IFN- γ is actually playing such a key role in vivo remains to be fully demonstrated at this stage. Indeed, the CpG-proB cells that emerge from cultures set up with bone marrow cells from IFN-γ–deficient mice might also be modulated by various IFN-γ–dependent factors through mechanisms that remain to be identified. Hence, a definitive proof of the key role of T-cell–derived IFN-γ in the differentiation of pro-B cells with suppressive functions would require experiments with CpG-proBs generated from mice in which only B cells would lack IFN-γ-R (chimeric mice or mice with conditional knock-out).

Nevertheless, both the reduction of Teffs and the cytokine regulation, with enhanced IFN-γ and reduced IL-21 production, could be evidenced in vivo within pancreatic CD4⁺ T-cells from CpG-proB recipients relative to age-matched noninjected controls.

The role of IFN-γ in T1D developed by NOD mice is complex. Genetic absence of IFN-γ delays but does not prevent diabetes in NOD mice (36). Although IFN-γ is cytotoxic for pancreatic β-cells, its production in the spleen leads to a paradoxical protective effect against T1D in bacillus Calmette–Guérin-treated mice (37, 38). Furthermore, IFN-γ restores normoglycemia in Ig-GAD_206–220–treated NOD mice (39) because of its capacity to control the balance between Th1 and Th17 differentiation and particularly to reduce IL-21 production.

IL-21 is a cytokine (40) whose receptor belongs to the common γ-chain receptor family. It is produced by Th17 cells (as well as follicular helper T cells and NKT cells) and thereby serves as an autocrine amplification factor for IL-23R–expressing T cells favoring Th17 differentiation. IL-21’s role remains debated (41, 42) in the resistance of Teffs to regulation by Tregs, which has been reported in the NOD mouse during progression of T1D. IL-21 plays a pathogenic role in various autoimmune diseases, such as colitis, gastritis, experimental autoimmune encephalomyelitis, and rheumatoid arthritis. It is encoded by idd3, a gene locus of predisposition to diabetes in the NOD mouse (43). Polymorphisms for IL-21 and its receptor are also linked to prevalence of T1D in humans (44).

Recent data from several laboratories have demonstrated a key role for IL-21 in the pathogenesis of T1D. This cytokine is produced at high levels in NOD mice (45) and accelerates disease onset upon transgenesis, but IL-21R deficiency protects against T1D (34, 35). Furthermore, neutralization of IL-21 with either IL21R-Fc or anti–IL-21 prevents disease development both at early and late prediabetes stages (33). In this line of evidence it has been reported that the IL-21 receptor controls both antigen transport and expansion of autoreactive CD4⁺ T cells as well as their ability to facilitate autoreactive CD8⁺ T-cell recruitment to pancreatic islets (46).

In contrast, several observations indicate that IL-10 does not play a significant role in the protective properties of CpG-proBs. Immediately after their recovery from bone marrow cultures, CpG-induced pro-B cells produced IFN-γ but no IL-10, which was generated only in response to PMA + ionomycin. IL-10 receptor blocking by a specific antibody had no effect on the suppression of Teffs promoted by CpG-proBs. Moreover, the B-cell progeny of CpG-proBs did not produce IL-10, even after PMA + ionomycin activation alone or in combination with LPS or anti-CD40 or CpG. The percentage of IL-10–producing cells was also not increased among host-derived B-cell populations in CpG-proB-recipients relative to control mice. This finding is in clear contrast with the IL-10–driven regulatory mechanisms often ascribed to Bregs (10). However, it has been reported that mature Bregs can also use regulatory pathways that do not depend on IL-10, such as cell-to-cell interactions involving MHC (47) or B7 molecules (48) and blockade of Teff expansion via FasL-induced apoptosis (49–51). Evidently, the inflammatory signals and cytokines they encounter in their microenvironment engender diverse immunosuppressive mechanisms. For example, LPS-activated mature B cells prevented T1D in NOD mice (52) by FasL-dependent apoptosis of diabetogenic Teffs, similarly to the mechanism of action reported here. In addition, these LPS-activated B cells down-regulated Th1 cytokines, possibly through TGF-β production, but Th2 cytokines were not increased. This mechanism could not account for the action of CpG-proBs that was not associated with detectable TGF-β production. However, other reports have shown that TLR activation of B cells could instead foster Th1 differentiation, through MyD88-dependent B-cell activation driving IFN-γ–producing T-cell differentiation and leading to an IgG2c primary antibody response (53). MyD88-dependent activation can therefore induce similar activities in mature B cells and pro–B-cell progenitors.

Here we provide evidence for the rapid recruitment of the early progeny of CpG-proB-cell progenitors to the pancreas as FasL^highc-kit⁺B220⁺IgM^low cells, leading to a reduction in the numbers of active IL-21–producing Teffs as early as 15–20 d after injection. In addition, the suppressive properties and high FasL expression are maintained throughout the differentiation process into several mature B-cell subsets in the spleen and the PLNs, thereby providing long-term protection against T1D.

Indeed, CpG-proBs differentiated into various B-cell subsets, but not into CD1d^hiCD5⁺ B10 cells, suggesting that they are not the common precursors of this regulatory cell lineage. Surprisingly, follicular B cells, MZ B cells, and T2-MZP-Bs largely prevailed among the suppressive CpG-proB progeny in NOD recipients. Notably, tolerogenic properties of T2-MZP-Bs have been abundantly reported in several experimental models of autoimmune diseases (23, 24, 54–57).

Finally, because CpG-proBs appeared in the bone marrow of NOD mice in vivo upon injection of CpG, the possibility that they constitute early progenitors for different regulatory B-cell subsets linking innate and adaptive immune responses, capable of exerting the appropriate feedback control to prevent autoimmune and inflammatory responses might be worth examining. However, unless injected early and repeatedly (58, 59), an administration protocol that creates disorders of the lymphoid architecture (60) and mobilizes bone marrow cells to the periphery, causing aplasia, TLR agonists and particularly CpG are not protective against T1D in the NOD mouse but rather accelerate disease onset (61–63). Indeed, the immunotherapeutic influence of CpG-proBs is limited in vivo by their transient survival and by the proinflammatory properties of most myeloid cells that emerge concomitantly in CpG-treated mice. Therefore, the regulatory potential of CpG-proBs, like that of Tregs and Bregs, is most probably overwhelmed by the pathogenic process taking place during the accelerated development of T1D in NOD mice treated by CpG. For this reason, the protective properties of this CpG-proB population in the spontaneous T1D model are only revealed after cell-sorting and adoptive transfer.

Conclusion

This unique induced B-cell progenitor subset adds to the ever expanding list of hematopoietic progenitors that can emerge in pathophysiological settings, such as infections, inflammation, and cancer to exert immunoregulatory and therapeutic functions, either directly or indirectly by differentiating into regulatory mature cell types (1–4). Such tolerogenic progenitors could become instrumental in cell therapy of autoimmune diseases, injected either on their own or in combination with stem cell grafts in the context of autologous stem cell transplantation that is already implemented in patients with autoimmune diseases (64–66).

Materials and Methods

Mice.

Wild-type CD45.1 and congenic CD45.2 NOD mice were bred in our animal facility under specific pathogen-free conditions. MyD88^−/− and IFN-γ^−/− NOD mice were gifts from N. Thieblemont (CNRS UMR8147, Paris) and from L. Chatenoud (INSERM U1013, Paris) and C. Benoist (Harvard Medical School, Boston), respectively, and bred in the same animal facility. Live animal experiments were approved by the Necker Animal Experimentation and Ethics Committee. Female CD45.1 NOD mice were used as a model of spontaneous diabetes and received intravenous progenitor cell transfers at 6- or 16-wk of age. Mice were screened for glycosuria (Glucotest; Boehringer-Mannheim) twice a week or glycemia (Haemoglukotest and Reflolux F; Boehringer-Mannheim) and were considered diabetic when nonfasting blood glucose levels were >250 mg/dL in two consecutive readings.

Isolation of TLR-Activated Bone Marrow Progenitors.

Bone marrow cells isolated from tibiae and femurs of 8- to 12-wk-old NOD mice were incubated in RPMI-1640 medium (PAA) supplemented with 10% (vol/vol) FCS and 1% antibiotics (penicillin and streptomycin) for 18 h with 1 μg/mL of the oligodeoxynucleotides CpG 1585 (CpG-A) (InvivoGen), CpG 1668 (CpG-B) (Eurogentec),and its negative GpC control (InvivoGen), or with the respective agonists of TLR1-9 supplied in a commercial kit (InvivoGen): Pam3CSK4 (0.5 μg/mL), FSLI (1 μg/mL), HKLM (2 × 10⁶ cells/mL), Poly I:C (HMW) (5 μg/mL), Poly I:C (LMW) (5 μg/mL), LPS-EK standard (1 μg/mL), FLA-ST standard (1 μg/mL), and ssRNA40/LyoVec (2 μg/mL) as well as CpG 1826 (CpG-B) (1 μg/mL). c-kit⁺ bone marrow cells were sorted by c-kit⁺ immunomagnetic separation using a RoboSep automaton (StemCell Technologies). Sorted cells were further stained with appropriate fluorochrome-conjugated mAbs against B220, IgM, Sca-1, and electronically sorted into small-size, c-kit^lowSca-1^lowB220⁺IgM⁻ and large-size c-kit^lowSca-1^highB220⁻IgM⁻ cell subsets using a FACS-Aria II (BD Biosciences).

Isolation of Immune cells from the Pancreas.

Pancreata isolated from control and CpG-recipient mice were finely minced and stirred in PBS containing 400 μg/mL Liberase and 50 μg/mL DNase (both from Roche Diagnostics) for 30 min at 37 °C. Cells were filtered through 0.22-μm filters and centrifuged at 400 × g for 8 min before use.

Staining of Cells for Flow Cytometry Analysis.

To block nonspecific Fc receptor binding, cells were preincubated for 10 min at room temperature with FcR blocker 2.4G2 mAb. Cells were then stained with appropriately labeled mAbs against CD4, B220, CD21, CD23, CD24, IgM, CD1d, CD5, CD43, CD44, CD93 (eBioscience), CD19, CD127, IgD, CD25, CD62L, Mac-1/CD11b, Gr-1, CD11c, c-kit (CD117), Sca-1 (anti-Ly6A/E), CD45.1, CD45.2 (BD Biosciences), and PDCA-1 (Miltenyi Biotec). Nuclear Foxp3 and eomesodermin expression was measured by FACS analysis as per the manufacturer’s instructions (eBioscience). Intracytoplasmic expression of cytokines was assessed after a 5-h stimulation with PMA (10 ng/mL) plus ionomycin (500 ng/mL) in the presence of Brefeldin A (2 mg/mL), followed by permeabilization with saponin and subsequent staining with specific antibodies including APC-labeled anti-IL-10 (from BD Biosciences) or anti-IL-21 (from eBioscience) and PE-labeled anti–IFN-γ (from BD Biosciences) or isotype controls. Topro III (Invitrogen) was used for assessing dead and live cells and in association with Annexin V (BD Biosciences) to assess apoptosis and necrosis. Total FasL expression was measured by FACS analysis after cell permeabilization with saponin, using APC-conjugated anti-FasL (clone MFL3; eBioscience). Membrane and intracellular antigen expression was analyzed in a FACS Canto II cytometer (BD Biosciences) using FlowJo software (Treestar).

Proliferation Assays.

CD4⁺CD25^high (all Foxp3⁺) or CD4⁺CD25⁻ spleen cells were electronically sorted from the spleen of WT- or IFN-γ^−/− NOD mice. They were loaded with CFSE (Life Technologies) and cultured (5 × 10⁴ cells per well) in RPMI medium 1640 supplemented with 5% (vol/vol) FCS (Life Technologies), 1% antibiotics, and 5 × 10⁻⁵ M β-mercaptoethanol. Cells were plated in 96-well round-bottomed culture plates, either alone or with sorted CpG-proBs at 1:1 and 2:1 T:CpG-proB cell ratios, and stimulated with 2.5 μg/mL of anti-CD3 mAb (clone 145–2C11) and 5 μg/mL of anti-CD28 for 5 d, with or without 10 U/mL of IL-2. Neutralizing antibodies against FasL (clone MFL4; eBioscience), IL-10R (clone 1B1.3a; BD Pharmingen), or isotype controls (BD Bioscience) were added at 5, 10, and 20 μg/mL to assess the role of the matching antigens.

Cytokine Measurements.

Cytokines were measured at day 5 in coculture supernatants using Flow Cytomix analyte detection reagent from eBioscience. For determination of pancreatic IL-21 levels, control and recipient NOD mice were killed at 15 wk of age, 9 wk after adoptive transfer of progenitors. Briefly, pancreases were collected from NOD mice of the different experimental groups and immediately snap-frozen in liquid nitrogen. The pancreases were kept at −80 °C until homogenization in anti-protease buffer using a Polytron device and centrifuged to remove debris. IL-21 levels were measured in these tissue homogenates by specific ELISA (R&D Systems).

Isolation of mRNA and Real-Time RT-PCR.

After coculture, live CD4⁺ cells were electronically sorted and their mRNA was extracted using RNAqueous-4PCR (Ambion). Reverse transcription was performed with high capacity cDNA reverse-transcription kits (Applied Biosystems) followed by quantitative PCR (qPCR) analysis in mouse Th17-oriented 96 StellARray qPCR plates with SYBR from Lonza in a AB 7900 HT real-time PCR system (Applied Biosystems), according to the manufacturer’s recommendations. Results were analyzed on-line with the global pattern recognition analysis tool. Alternatively, TaqMan primers and probes for murine IL-21 and IFN-γ were purchased from Applied Biosystems and samples were analyzed after preamplification with the Taqman PreAmp Master Mix kit, using the StepOne Plus analyzer (Applied Biosystems) with 18S as endogenous control.

Statistical Analysis.

Data were analyzed with nonparametric Mann–Whitney test and Kaplan–Meier estimates and logrank analysis for diabetes incidence curves, using Prism software (GraphPad). P values < 0.05 were considered statistically significant.