Functional Waning of Naturally Occurring CD4⁺ Regulatory T-Cells Contributes to the Onset of Autoimmune Diabetes

NOD, C57BL/6, NOD.TCRα^−/−, and NOD.BDC2.5 mice were bred and maintained in pathogen-free conditions at the McGill University animal facility. NOD.TCRα^−/− and NOD.BDC2.5 TCR mice were a gift from Dr. Christophe Benoist (Harvard University, Boston, MA).

T-cells were stained with a variety of fluorochrome-conjugated or biotinylated monoclonal antibodies (mAbs), as previously described (22,23): anti-CD4 (clone RM5), anti-CD25 (clone PC61.5) (eBioscience, San Diego, CA), anti-Vβ4 (clone CTVB4), anti-CD69 (clone H1.2F3), anti-CD44 (clone IM7), anti-CD62L (clone MEL14) (BD Bioscience, Mississauga, ON, Canada), and anti-Foxp3 (eBioscience). Stained cells were acquired on a FACSCalibur flow cytometer (BD, San Jose, CA). In adoptive transfer experiments, pancreata were digested with collagenase-V (Invitrogen, Burlington, ON, Canada), and T-cells were separated from the digested tissue by centrifugation on a Lympholyte-M gradient (Cederlane) and then stained accordingly.

CD4⁺ T-cell subsets were isolated from spleen, thymus, or lymph nodes on a FACSAria (BD) or AutoMACS (Miltenyi, Auburn, CA) cell sorter as previously described (22,23). The final purity was assessed on a FACSCalibur (BD) and was routinely >98%.

Cells from the spleen and pancLN were stimulated 4–5 h with PMA and ionomycin and were treated with monensin (2 μmol/l) for the last 2–3 h of culture. After surface staining, cells were fixed and permeabilized, and then intracytoplasmic staining was performed using anti-mouse IL-2 mAb (JES6–5H4), anti-mouse tumor necrosis factor-α (TNF-α) mAb (MP6), anti-mouse IL-17 (17B7), or phycoerythrin-labeled rat IgG1 isotype control (eBioscience). Stained cells were analyzed on a FACSCalibur flow cytometer (Becton Dickinson) using CellQuest software.

Purified CD4⁺ T-cell subsets were transferred intravenously, either alone or in combination, into recipient mice (1–3 × 10⁵/mouse). In some experiments, donor T-cells were labeled with carboxyfluorescein succinimidyl ester (CFSE) to assess cell proliferation in vivo, as previously shown (24).

Recipient mice were tested every 2–3 days for diabetes, as previously described (25). Overt diabetes was defined as glycemia >300 mg/dl. Hemoglukotest kits were provided by Roche Diagnostics (Laval, Canada).

Hematoxylin-eosin (H-E)-stained histological slides of pancreatic tissues were prepared and graded for insulitis as previously described (14,25,26). Three randomly obtained levels of pancreas were analyzed in double-blind fashion, and 12–15 islets per section were scored per time point. Statistical analysis was performed with the χ² test.

Pancreata were harvested, washed with PBS, and treated with a digestion buffer (Hanks’ balanced salt solution [HBSS], 1 mg/ml collagenase V, and 1 μg/ml DNase I) for 25 min at 37°C. The digestion was quenched with fetal bovine serum (Invitrogen). Cell suspensions were then extensively washed in HBSS, resuspended in nonenzymatic dissociation solution (Invitrogen), and incubated for 10 min at 37°C. The cells were resuspended in PBS, stained with Pe-Cy7-conjugated CD4 and antigen-presenting cell (APC)-conjugated CD25 mAbs, and then stained intracellularly with phycoeryhrin-conjugated Foxp3 and fluorescein isothiocyanate-conjugated Ki-67 (clone B56) (BD Bioscience). Data were analyzed on a FACSCalibur flow cytometer (BD) using the Flowjo software.

Some reports suggested that the ability of nT_reg cells to suppress type 1 diabetes is either absent or poorly developed, thus promoting disease onset (14,20,21). To determine whether NOD mice developed CD4⁺Foxp3⁺ nT_reg cells in their thymus, the frequency of thymic CD4⁺Foxp3⁺ nT_reg cells was assessed in CD4^SP T-cells in pre-diabetic NOD and BDC2.5 mice relative to type 1 diabetes–resistant C57BL/6 (B6) mice. We show that the proportion of CD4⁺Foxp3⁺ nT_reg cells represents ∼1 and 5% of CD4^SP T-cells in BDC2.5 and NOD mice, respectively, and does not decrease in adult mice (Fig. 1A). Total numbers of CD4⁺Foxp3⁻ or CD4⁺Foxp3⁺ T-cells were stable between 10 and 25 days of age and seemingly increased between 25 to 50 days of age in NOD mice only to reach levels not significantly different from C57BL/6 mice (data not shown). No differences were observed in CD25⁺ or CD25⁻ Foxp3⁺ T_reg cells from thymi of pre-diabetic mice compared with C57BL/6 mice (data not shown).

To assess NOD thymic T_reg cell function, we transferred 5 × 10⁵ thymic CD4⁺CD25⁻ T_eff cells in the presence or absence of 5 × 10⁴ thymic CD4⁺CD25⁺ nT_reg cells from neonatal NOD mice into NOD.scid recipient mice. CD4⁺CD25⁻ T_eff cells from pre-diabetic mice transferred diabetes at a 50% incidence 5–6 weeks after transfer, reaching a total incidence of 80–90% at 10 weeks after transfer (Fig. 1B). However, when thymic CD4⁺CD25⁺ nT_reg cells were co-transferred, type 1 diabetes onset and incidence were dramatically reduced (Fig. 1B) and β-islet cells were preserved (data not shown), suggesting that thymic nT_reg cells, which are devoid of potential peripherally activated CD25⁺ T_eff cells, are protective in NOD mice. To further assess thymic nT_reg function in a synchronous, antigen-specific system, similar experiments were performed in the BDC2.5 TCR transgenic transfer model of type 1 diabetes (26). To this end, we transferred thymic CD4⁺CD25⁻ T_eff cells from pre-diabetic, 3- to 4-week-old BDC2.5 mice, either alone or in the presence of thymic CD4⁺CD25⁺ T_reg cells, into NOD.TCRα^−/− recipients. Thymic CD4⁺CD25⁻ T_eff cells transferred diabetes as early as 11 days after transfer, with an incidence as high as 60–75% at 30 days after transfer, and correlated with a prominent infiltration of mononuclear cells into the islets and general loss of islet morphology (Fig. 1B–D). In contrast, the presence of thymic nT_reg cells in transferred CD4⁺ T-cells completely abrogated the onset of disease and protected from type 1 diabetes for at least 3–4 weeks after T-cell transfer (Fig. 1B). Although not completely preventing the immune cell infiltration of the islets, nT_reg cells nonetheless reduced the severity of insulitis with a greater proportion of islets with preserved morphology (Fig. 1C–D). Thus, nT_reg cells develop in the NOD thymus and can control the onset of type 1 diabetes.

We hypothesized that alterations in the numbers or function of peripheral T_reg cells may precipitate type 1 diabetes. We first enumerated Foxp3⁺ T-cells in lymph nodes and spleen of pre-diabetic NOD and BDC2.5 mice of various ages. In NOD mice, Foxp3⁺ T_reg cell represented an average of 16–20% of CD4⁺ T-cells in the spleen and 7–10% in the pancLN with variations between ages not significantly different from type 1 diabetes–resistant B6 mice (Fig. 2A). Similarly, nT_reg cells accounted for 9–11% and 4–7% of CD4⁺ T-cells from spleen and pancLN BDC2.5 mice, respectively (Fig. 2C). The total pool of nT_reg cells in pancreatic sites gradually increased with age in both NOD and BDC2.5 mice but was comparable with age-matched B6 mice (Fig. 2B; data not shown), suggesting that a quantitative defect in peripheral T_reg cells does not precede type 1 diabetes onset.

We then assessed the function of nT_reg cells from neonatal NOD mice (7–10 days old). To this end, we transferred 5 × 10⁵ CD4⁺CD25⁻ T_eff cells from pre-diabetic or diabetic mice into NOD.scid mice either alone or in the presence of titrated doses of CD4⁺CD25⁺ nT_reg cells from neonatal, pre-diabetic NOD mice. CD4⁺CD25⁻ T-cells transferred diabetes at a 50% incidence 5–6 weeks after transfer, reaching an incidence of 80–90% at 10 weeks after transfer (Fig. 2D), and correlated with a prominent infiltration of mononuclear cells into the islets and general loss of islet morphology (Fig. 2E and F). However, when CD4⁺CD25⁺ nT_reg cells were cotransferred with either pre-diabetic (Fig. 2D, left panel) or diabetic (Fig. 2D, right panel) CD4⁺CD25⁻ T_eff cells, type 1 diabetes onset and incidence were dramatically reduced at every nT_reg-to-T_eff ratio transferred. Moreover, the severity of insulitis was greatly reduced, with a greater proportion of β-islets demonstrating preserved morphology (Fig. 2E and F), albeit to a lesser extent than thymic nT_reg cells. Transfer of CD4⁺CD25⁺ T-cells from diabetic mice with pre-diabetic or diabetic CD4⁺CD25⁻ T_eff cells did not prevent type 1 diabetes onset and incidence (data not shown). Therefore, neonatal nT_reg cells are functional and have the potential of protecting NOD mice from primary and established type 1 diabetes.

We reasoned that functional changes in the periphery might nevertheless disrupt nT_reg function and precipitate type 1 diabetes over time. We evaluated the potential for an age-dependent variation in nT_reg cell suppressor function in our BDC2.5 transfer model. CD4⁺CD25⁺ T_reg cells were isolated from young or adult BDC2.5 mice and cotransferred with CD4⁺CD25⁻ T_eff cells from BDC2.5 mice into NOD.TCRα^−/− mice. We show that whereas nT_reg cells from young BDC2.5 mice completely suppressed type 1 diabetes induced by young or old T_eff cells (Fig. 3A), nT_reg cells isolated from older or overtly diabetic BDC2.5 animals were completely ineffective at suppressing type 1 diabetes induced by young or old T_eff cells (Fig. 3A). It was possible that the lack of regulation of type 1 diabetes was due to the increased frequency of activated T-cells among CD4⁺ T-cells. The expression of CD25, CD69, CD44^hi, or CD62L^lo activation markers on T-cells was not significantly different between 3- and 4-week-old and 6- and 8-week-old donors, and the reduced ability of old Treg cells to control type 1 diabetes was not due to a reduced proportion of Foxp3⁺ T-cells (Fig. 2C) or to an increased cellular frequency of activated or TNF-α–secreting Foxp3⁻ T-cells, suggesting that the waning nT_reg function in older donors was not attributable to an increased contamination of activated T_eff cells within T_reg preparations (Figs. 2C and 3B and C). CD4⁺CD25⁺ T-cells from overtly diabetic donor mice were completely inefficient in controlling the onset of type 1 diabetes in our transfer system (Fig. 3A). Furthermore, recipients of 6- to 8-week-old donor CD4⁺CD25⁻ T-cells did not demonstrate a higher diabetes incidence compared with 3- to 4-week-old donor cells, excluding the possibility of reduced pathogenicity of younger T_eff cells (Fig. 3A). Interestingly, regulation of type 1 diabetes could be restored if an additional bolus of purified CD4⁺CD25⁺ T_reg cells from 6- to 8-week-old donor mice were infused in recipient mice (data not shown), suggesting that an increase in the circulating pool of nT_reg cells could ultimately control type 1 diabetes in older NOD mice. Our results do not exclude the possibility that alterations in the pathogenic potential of T_eff cells from older mice may also contribute to type 1 diabetes onset by heightening their resistance to regulation. Overall, these results suggest that while nT_reg cells function normally in the periphery of neonatal NOD mice, the progression beyond insulitis is seemingly due to a time-dependent waning in the functional potency of T_reg cells, allowing self-reactive T_eff cells to escape regulation and initiate a destructive infiltration of the islets.

The functional impact of nT_reg cells on diabetogenic T-cells during the pre-diabetic period is not thoroughly understood. CD4⁺ T_reg cells may potentially control diabetes progression by inhibiting the activation or clonal expansion of islet-specific CD4⁺ T-cells in the pancLN, altering T_eff cell trafficking to the pancreas, or suppressing effector functions in situ (18,19,27). To determine whether nT_reg cells influence the priming of diabetogenic T-cells, NOD.TCRα^−/− recipient mice were transferred intravenously with 10⁶ CFSE-labeled CD4⁺CD25⁻ T_eff cells in the presence or absence of 10⁵ CD4⁺CD25⁺ isolated by fluorescence-activated cell sorter (FACS) from peripheral lymph nodes of 2- to 4-week-old BDC2.5 mice. At 3 days after transfer, we examined the frequency of activated and proliferating islet-specific CD4⁺ T-cells in pancLN and nondraining lymph nodes by FACS and in the presence or absence of nT_reg cells. In this system, transferred CD4⁺ T_eff cells migrate to the pancLN (Fig. 4A), and the activation of CD4⁺Vβ4⁺ T_eff cells (CD69 early marker expression) (Fig. 4B; 22–26 vs. 3% in nondraining lymph nodes), as well as the frequency of proliferating T-cells (CFSE dilution) (Fig. 4C; 32–35 vs. 7% in nondraining lymph nodes), could only be detected in pancLN, confirming that T-cell activation is antigen specific. The frequency of CD4⁺Vβ4⁺CD69⁺ T-cells (21.6 vs. 26%) and the frequency of dividing CD4⁺Vβ4⁺ T-cells (32.4 vs. 35.2%) in the pancLN remained unaffected in the presence of nT_reg cells compared with recipient mice transferred with T_eff cells alone, suggesting that antigen presentation and proximal TCR signals are not inhibited in T_reg cell-protected mice (Fig. 4A).

Previous studies have suggested that lymphopenic environments might impair the functional dissection of the mechanism of nT_reg cells. To circumvent the possible confounding effect of homeostatic proliferation in our system, similar transfer experiments were also conducted in nonlymphopenic, wild-type NOD recipient mice. Our results show that the frequency of CD4⁺Vβ4⁺ T-cells (0.41 vs. 0.71%) and Ag-driven proliferation of T_eff cells in the pancLN were unaffected by the presence of BDC2.5 T_reg cells (53.1 vs. 53.1%) (Fig. 4D and E), suggesting that T_reg cells do not control the diabetogenicity of T_eff by altering their homing to draining pancLN, early activation, or antigen-driven proliferation. However, the frequency of TNF-α and IL-17 secreting CD4⁺ T-cells in pancLN was significantly suppressed in the presence of BDC2.5 T_reg cells, suggesting that while priming and expansion of CD4⁺ T_eff cell remain unaffected, the differentiation of proinflammatory cytokine producing diabetogenic T-cells is suppressed by T_reg cells (Fig. 4F).

Recent studies show that nT_reg cells localize in sites of inflammation to mediate their protective effect (16,18,19,27). To determine whether nT_reg cells home to and expand within pancLN, NOD.TCRα^−/− or wild-type NOD recipient mice were transferred intravenously with 10⁵ CFSE-labeled CD4⁺CD25⁺ nT_reg cells in the presence of 10⁶ CD4⁺CD25⁻ T_eff cells isolated from 2- to 4-week-old BDC2.5 mice, and the frequency of proliferating T_reg cells in pancLN and nondraining lymph nodes was examined 3 days after T-cell transfer. The frequency of BDC2.5 Foxp3⁺ nT_reg cells in recipients of CD25⁺/CD25⁻ cells was found to be similar in the pancLN and the nondraining lymph nodes of both recipients (Fig. 5A and C). The proportion of proliferating Foxp3⁺ nT_reg cells was significantly greater in pancLN than in nondraining lymph nodes in immunodeficient (43 vs. 13%, respectively; Fig. 5B) and immunocompetent (52.7 vs. 20.5%, respectively; Fig. 5D) hosts, suggesting that nT_reg cells actively expand in pancreatic sites.

Because altered early T-cell priming events cannot directly explain the protective effect of nT_reg cells, we then hypothesized that nT_reg cells accumulating in pancreatic sites may have a functional impact on T_eff cells at later stages of the diabetogenic process. To this end, CD4⁺ T-cell subsets from thymus of BDC2.5 mice were adoptively transferred into NOD.TCRα^−/− mice as described above, and 30 days after transfer, the frequency of CD4⁺Foxp3⁻ T_eff and CD4⁺Foxp3⁺ nT_reg cells was examined in various lymph nodes and pancreas. Although our results do not show an influence of nT_reg cells on the initial rounds of proliferation of diabetogenic T-cells within pancLN (Figs. 4 and 5), our results indicate that the absence of CD4⁺ nT_reg cells favors the infiltration/expansion of T_eff cells in spleen (data not shown), pancLN, and pancreas, with a two- to threefold increase in the number of CD4⁺Vβ4⁺ T-cells in these sites compared with recipient mice cotransferred with T_eff and nT_reg cells (Fig. 6A). This increased accumulation of T_eff cells correlated with the onset of diabetes and may suggest a role for nT_reg cells in the control of antigen-driven recruitment or proliferation of T_eff cells in pancreatic sites. Our result show that protection from type 1 diabetes correlates with a significantly increased proportion of Foxp3⁺ nT_reg cells in these sites and mirrors the decline in T_eff cell numbers, suggesting that a crucial part of the protective role of nT_reg cells resides in their ability to migrate to or expand in sites of autoimmune attack (Fig. 6A). Although not entirely preventing the infiltration of Foxp3⁻ T_eff cells into the islets, nT_reg cells nonetheless reduced the severity of insulitis, with a greater proportion of islets with preserved morphology (Fig. 2E and F). Interestingly, examination of the few recipient mice receiving only CD4⁺CD25⁻ T_eff cells and that never developed type 1 diabetes revealed that the proportion of Foxp3⁺ cells was similar to that observed in recipient mice co-injected with thymic nT_reg cells, suggesting that this protective Foxp3⁺ T-cell subset originated from the thymic CD25⁻ T-cell donor fraction, possibly expanded/differentiated within these sites, and ultimately ensured disease protection (Fig. 6A).

Because young and adult nT_reg cells differed in their protective effect, we then compared the differential ability of Foxp3⁺ nT_reg cells from 3- to 4-week-old or 6- to 8-week-old donors to accumulate in inflamed pancreatic sites. In recipients of 3- to 4-week-old cells, the frequency of nT_reg cells in pancLN and pancreas was significantly greater than recipients receiving T_eff alone and correlated with type 1 diabetes protection (Fig. 6B and C). In stark contrast, nT_reg cells from 6- to 8-week-old donor mice accumulated less efficiently in the pancLN and pancreas and correlated with the onset of type 1 diabetes in the majority of recipient mice (Fig. 6B and C). In the few type 1 diabetes–free mice receiving nT_reg cells from 6- to 8-week-old donor mice, we observed that the frequency of nT_reg cells in the pancreas was comparable with recipient mice receiving 3- to 4-week-old nT_reg cells (Fig. 6C).

We were unable to find any difference in the homing of nT_reg and Teff cells in pancLN over time (data not shown). To directly determine whether this reduced frequency of BDC2.5 T_reg cells in pancreatic sites was due to a loss of nT_reg cell expansion in these sites over time, we examined by FACS the cellular frequency of cycling CD4⁺Foxp3⁺ T-cells in the periphery of young and old BDC2.5 mice, as determined by the Ki-67 proliferation marker (Fig. 6D and E). We show that the total proportion of cycling Foxp3⁺ T_reg cells and the fraction of cycling cells among the total pool of CD4⁺Foxp3⁺ nT_reg cells are significantly reduced in pancLNs and pancreata of old BDC2.5 mice compared with young mice (Fig. 6D and E). Collectively, our results show that nT_reg cells expand within inflamed pancreatic sites where they constrict the size of the T_eff cell pool and reduce the histopathological consequences of a destructive infiltration. Thus, a possible waning with age of nT_reg expansion within pancreatic sites could explain the onset of an uncontrolled T_eff cell infiltration of the pancreas and type 1 diabetes induction.

Foxp3⁺ nT_reg cells have been implicated as a central control point in type 1 diabetes progression, and defects in their development or function may represent a major predisposing factor for spontaneous autoimmunity in NOD mice (2,28,29). Here, we show that thymic and peripheral CD4⁺CD25⁺ nT_reg cells can suppress disease in both normal NOD and BDC2.5 antigen-specific model of type 1 diabetes. We also show that CD4⁺CD25⁺Foxp3⁺ nT_reg cells do not affect the priming of antigen-specific T_eff cells in pancLN but localize within insulitic lesions, where they suppress the infiltration of T_eff cells. The cellular potency of nT_reg cells, although fully operative in neonatal mice, declines with age despite a stable cellular frequency of Foxp3⁺ nT_reg cells in primary and secondary lymphoid tissues.

Recent studies stipulate that defective or reduced CD4⁺CD25⁺ T-cell frequencies in autoimmune-prone hosts, including NOD mice, represent the primary predisposing factor to spontaneous autoimmunity(20,21,30,31). In most studies, the CD25 surface marker is frequently used for the monitoring of nT_reg cell frequencies, albeit at a time point when pancreatic inflammation is well engaged. CD25 is an unreliable marker because activated CD4⁺ T-cells upregulate CD25, thus precluding its use as a tracking biomarker of nT_reg cells in NOD mice. We observed that Foxp3⁺CD4⁺ nT_reg cells, irrespective of CD25 expression, represent a stable pool in thymocytes, lymph nodes, or spleen of neonatal and adult NOD mice and are comparable with type 1 diabetes–resistant C57BL/6 mice, thus refuting the view that NOD mice have a developmental defect in nT_reg cells. However, it is possible that a functional deficiency in nT_reg cells may not be visible as a sudden decline in the frequency of these cells in peripheral tissues and may conceivably be resultant to gaps in the TCR repertoire or gene polymorphisms modulating various effector functions (32). Consistently, we show that nT_reg cell function wanes with time as evidenced by their inability to expand efficiently and to prevent T_eff cell infiltration in pancreatic sites, suggesting that the loss of Ag-driven homing or expansion of nT_reg cells in pancreatic environments may represent an essential checkpoint in the progression to type 1 diabetes. We cannot exclude the possibility that time-dependent changes in T_eff cells may contribute to type 1 diabetes onset (33).

An unresolved question relates to the location of nT_reg cell-mediated tolerance induction in vivo. The ability of nT_reg cells to localize within tissues to dampen the magnitude of T_eff cell responses and prevent the histopathological consequences has been observed in models of infectious disease, inflammatory bowel disease, and tumors (34,35). Our results show that nT_reg preferentially home and expand within inflamed pancLN and islets of type 1 diabetes–protected mice and that nT_reg cells control the effector functions of infiltrating diabetogenic CD4⁺ T-cells in these sites, albeit not completely preventing insulitis. Interestingly, a more significant reduction in the degree of insulitis was observed with thymic nT_reg cells compared with peripheral nT_reg cells (Fig. 1D vs. 2F), suggesting that an increased functional potency may exist in the thymic nT_reg cell compartment, as well as a potential waning of this functional potency in peripheral nT_reg cells (21,29). It is conceivable that the thymic microenvironment provides the necessary developmental and homeostatic signals that may be lacking in the periphery of NOD mice. One study showed that pancreatic BDC2.5 CD4⁺CD25⁺ T_reg cells abrogated disease induced by pancreatic CD4⁺CD25⁻ T_eff cells in NOD.scid recipients, and the majority of T_reg cells were actively suppressing in the pancreas, rather than affecting the initial priming of the autoreactive T-cells in the pancLN (19). Similarly, in vitro expanded BDC2.5 CD4⁺CD25⁺ T_reg cells suppressed type 1 diabetes induced after transfer of diabetic NOD splenocytes into NOD.scid recipients, despite the fact that insulitis was nonetheless apparent in protected mice (36,37). Interestingly, the gene expression profile of islet-infiltrating nT_reg cells differ from nT_reg cells residing in the pancLN, suggesting that the target tissue engages unique transcriptional programs in nT_reg cells, which might relate to their regulation in these sites (19). It is unclear whether these distinct nT_reg gene signatures occur as a result of their tissue localization or as a consequence of their own suppression. Target organs may confer unique regulatory pressures on infiltrating T_eff cells and may shape the type of regulation needed for disease resolution. Alternatively, different Foxp3⁺ nT_reg cell subsets may exist to operate in tissue-specific fashions, or chemokine receptors like CCR5 or CCR6 may endow nT_reg cells with a competitive advantage to enter more efficiently in pancLN (38–40).

CD4⁺ nT_reg cells may potentially suppress anti-islet T-cell responses by affecting their activation and clonal expansion in pancLN. In our system, the initial activation of islet-specific CD4⁺ T-cells is unaffected in the presence of nT_reg cells, suggesting that antigen presentation and TCR signals are not inhibited by nT_reg cells in draining pancLN, a finding consistent with those from Chen et al. (19). This is also in agreement with our observation that nT_reg cells suppress type 1 diabetes mediated by diabetic T-cells, which likely traffic directly to islets, circumventing priming in the pancLN. In addition, we were unable to detect any changes in the frequency of proliferating diabetogenic CD4⁺ T-cells in the pancLN, either in the presence or absence of nT_reg cells, suggesting that antigen-induced priming of autoreactive T-cells is not directly affected by nT_reg cells. Paradoxically, the absolute number of T_eff cells in pancreatic sites is dramatically increased in the absence of nT_reg cells, suggesting that nT_reg cells may restrain T-cell clonal expansion, survival, or homing at later events of diabetogenesis. Consistently, transfer of BDC2.5 T-cells into thymectomized NOD.B7–2^−/− recipients in conjunction with in vivo depletion of CD25⁺ T-cells resulted in an increased accumulation of T_eff cells in the pancLN compared with control mice (41). Moreover, the cellular frequency of TNF-α–and IL-17–secreting antigen-specific T-cells is significantly suppressed in the functional presence of CD4⁺Foxp3⁺ Treg cells, suggesting that although Treg cells remain unable to suppress the activation and priming of diabetogenic T-cells, they remain fully capable to inhibit the differentiation of T_eff cells in sites. Interestingly, T_reg cells were shown to control the pathogenicity of islet-specific, CD8⁺ T_eff cells by inhibiting dendritic cell maturation in the pancLN (42), and in vitro expanded BDC2.5 T_reg cells have been shown to stably interact with dendritic cells and potentially disrupt BDC2.5 T/dendritic cell cellular interactions in pancLN, suggesting that nT_reg/APC interactions may be, in part, responsible for nT_reg cell-mediated protection (27). Lastly, we cannot formally exclude a more subtle effect on T_eff cells in these sites such that T_eff cells are now imprinted with a reduced pathogenic potential, which would reveal itself once they traffic to islets.

A recent study by Chen et al. (19) used NOD mice harboring the scurfy mutation of the foxp3 gene (FoxP3^sf) to examine the functional role of nT_reg cells in type 1 diabetes. Because the mutation of FoxP3 impairs the development of nT_reg cells, NOD.FoxP3^sf displayed a significantly advanced onset of type 1 diabetes compared with normal NOD mice, implying a role for Foxp3⁺ nT_reg cells in the control of type 1 diabetes pathogenesis. However, this study did not address the possibility that the injection of wild-type, antigen-specific T_reg cells, while rescuing Foxp3-deficient NOD mice from the early onset of type 1 diabetes, were compensating for the primary deficit in nT_reg cells believed to exist in these mice or whether such injection was actually suppressing the global inflammation that likely arose as a secondary consequence of Foxp3 deficiency. Although the NOD.FoxP3^sf model provides a system devoid of nT_reg cells, the Scurfy mice also possess multiple immune defects, which likely have consequences on the physiopathology of type 1 diabetes. Chang et al. (43) have shown that a T-cell extrinsic defect may contribute to the Scurfy and IPEX syndrome because the Foxp3^sf mutation in nonhematopoietic, thymic stromal cell leads to an ErbB2-dependent defective thymopoiesis. Furthermore, it is unclear whether NOD.FoxP3^sf mice possess aberrant antigen presentation or costimulation, which could combine to reduce the activation thresholds for Foxp3^sf T_eff cells and render them resistant to suppression (44,45).

In conclusion, nT_reg cells represent a master switch regulating disease onset and progression in NOD mice because abrogation of nT_reg function can break T-cell tolerance to β-islet antigens. Despite nT_reg cells actively suppressing anti-islet T-cell responses in the neonatal immune system, this suppression is ultimately insufficient to maintain tolerance to pancreatic antigens because autoimmunity ultimately ensues in these mice, as shown in recent studies (46,47). These studies may provide insights into the cellular basis of type 1 diabetes susceptibility and may lead to the development of novel approaches to potentiate nT_reg activity in autoimmune-prone hosts.

M.T., E.S., and E.d'H. have received fellowships from the Canadian Institutes for Health Research Training Grant in Neuroinflammation. C.A.P. has received a Canada Research Chair. This study has been supported by Canadian Institutes for Health Research Grant CIHR MOP 67211 and Canadian Diabetes Association Grant GA-3-05-1898-CP.

We thank Michal Pyzik, Ekaterina Yurchenko, and Valerie Hay for advice and technical assistance.