Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Immunity. Author manuscript; available in PMC 2010 December 8.
Published in final edited form as:
PMCID: PMC2998796

How punctual ablation of Foxp3+ T cells unleashes an autoimmune lesion within the pancreatic islets


CD4+Foxp3+ regulatory T cells (Tregs) are known to control the progression of autoimmune diabetes, but when, where and how they exert their influence in this context are questions still under vigorous debate. Exploiting a transgene encoding the human diphtheria toxin receptor, we punctually and specifically ablated Foxp3+ cells in the BCD2.5/NOD mouse model of autoimmune diabetes. Strikingly, overt diabetes developed within three days. The earliest detectable event was the activation of natural killer cells (NK) directly within the insulitic lesion, notably induction of gene expression already by seven hours. Increased expression of IL-12p40 by neighboring dendritic cells followed shortly thereafter. Interferon (IFN)-γ impacted extensively on the gene-expression program of the local CD4+ effector cell population, unleashing it to aggressively attack the islets, and very crucial for the development of diabetes. Thus, Tregs rein in pancreatic autoimmunity in situ through control of a central innate immune system player, NK cells.


Foxp3+CD4+ Tregs regulate a variety of immune responses, including autoimmunity, allergy, inflammation, infection and tumorigenesis (Zheng and Rudensky, 2007; Sakaguchi et al., 2008). This cell population is required life-long to guard against autoimmunity, perhaps best illustrated by the multi-organ infiltrates that arise a few weeks after its acute ablation in adult mice (Kim et al., 2007). In particular, Tregs play a crucial role in protection from type-1 diabetes (T1D), an autoimmune disease characterized by specific attack of the insulin-producing cells of the pancreatic islets (Tang and Bluestone, 2008). For example, autoimmune diabetes is one of the major elements of the IPEX (immune dysfunction – polyendocrinopathy – entreropathy – X-linked inheritance) syndrome that afflicts humans with a defective Treg compartment due to a mutation in the FOXP3 gene (Bennett et al., 2001; Wildin et al., 2001). Moreover, transfer of Tregs can protect mice from autoimmune diabetes, whether in the NOD model or in T cell receptor (TCR) transgenic systems derived there from (Salomon et al., 2000; Tarbell et al., 2004; Tang et al., 2004; Herman et al., 2004; Tarbell et al., 2007). Conversely, genetic deficiencies or experimental manipulations that reduce numbers or activity of this regulatory population can exacerbate diabetes (Salomon et al., 2000; Chen et al., 2005).

The precise point at which Tregs impact on the behavior of effector T (Teff) cells to rein in autoimmunity, and the pathways involved, remain controversial issues (Zheng and Rudensky, 2007; Sakaguchi et al., 2008). Several junctures are possible, and have been highlighted in different experimental settings: the migration of naïve T cells to the lymph nodes (LNs) draining the target tissue(s); their activation, expansion or survival therein; differentiation to a particular T helper (Th) cell phenotype; homing of activated Teff cells to target tissues; their expansion or survival after arrival; and their ultimate destructiveness towards the tissues.

As concerns diabetes, several groups have focused on Treg influences at an early stage – initial priming of potentially diabetogenic T cells within the pancreatic LNs (PLNs). Proliferation of islet-reactive BDC2.5 Teff cells in the PLNs was inhibited by pre-administration of a large number of Tregs and, conversely, was enhanced when BDC2.5 effectors were transferred into Treg-deficient CD28−/− mice (Tang et al., 2006), consistent with previously published results issuing from related experimental manipulations (Bour-Jordan et al., 2004). In other cases, while Tregs did not inhibit the expansion of islet-reactive Teff cells within the PLNs, they did impede their early differentiation at that site, reducing the production of IFN-γ and expression of chemokine receptors needed for migration to the islets (such as CXCR3) (Sarween et al., 2004), or diminishing the fraction of T cells producing tumor necrosis factor (TNF)-α or interleukin (IL)-17 (Tritt et al., 2008). Effects on the survival of differentiated Teff cells within the PLNs have also been postulated (Tritt et al., 2008)

On the other hand, such influences of Tregs on the priming phase of islet-reactive T cells was not evident in several other studies on diabetes models, as LN Teff cells were found to proliferate equivalently in their presence or absence (e.g. (Chen et al., 2005)). Despite some initial technical difficulties in finding them, it is now clear that Tregs are prominent residents of many types of autoimmune tissular infiltrates, both murine and human (Zheng and Rudensky, 2007; Sakaguchi et al., 2008). Some investigators have argued that they are inoperative in such an inflammatory context, given that tissue damage eventually occurs (e.g. (Korn et al., 2007)), but this conclusion ignores the fact that destruction might have been worse in their absence. Tregs are readily found in pre-diabetic islet infiltrates, where they have a distinct phenotype, prompting us to propose that such tissue-resident regulatory cells are a key element in local immunoregulation, capable of staving off the terminal destruction of islet cells for a protracted period (Chen et al., 2005; Herman et al., 2004).

Some of the discrepancies highlighted above no doubt issue from the experimental strategies chosen in the different studies. Transfer systems, employed in most of the experiments, may not mimic the natural disease course, especially when they entail lymphopenic mice as T cell recipients. Chen et. al. tried to sidestep this issue by introducing the scurfy (Foxp3-deficiency) mutation into the BDC2.5 TCR transgenic diabetes model (Chen et al., 2005), but immune system adaptation can take place in the constitutive absence of Tregs. We have now addressed this issue using a more definitive approach: BAC transgenic mice expressing the diphtheria toxin (DT) receptor (R) under the dictates of Foxp3 transcriptional regulatory elements were generated directly on the NOD genetic background, allowing temporally controlled ablation specifically of Tregs. Exploiting this new resource, we have assessed where and how Tregs impact on the anti-islet autoimmune response. The strength and synchrony of the rapidly ensuing reaction were instrumental in elucidating the pathway by which Tregs keep the immune system at bay in the pre-diabetic state, in particular how they dampen effector cells within the insulitic lesion to stave off the conversion of insulitis to diabetes, a therapeutically critical disease juncture. Unexpectedly, an effect of Tregs on innate immune system players, notably NK cells, appears primary.


An ongoing need for Tregs to guard against autoimmune destruction of the pancreatic islets

Controlled lineage ablation is a powerful approach for determining the role of particular cell-types in specialized biological processes. More and more, transgene-directed expression of the human DTR in particular cells in mice is being exploited to achieve this goal: murine cells are resistant to DT for lack of a receptor, so those expressing the DTR transgene are uniquely sensitive to DT treatment in vivo (Saito et al., 2001). Importantly, since the death induced by this drug is apoptotic, cell ablation has not provoked an inflammatory response in the many contexts so far examined (Bennett and Clausen, 2007; Feuerer M et al., 2009). Recently, this strategy was successfully employed to selectively eliminate Tregs in mice (Kim et al., 2007; Lahl et al., 2007; Lund et al., 2008).

We generated a Foxp3-DTR line directly on the NOD genetic background by injecting NOD embryos with a large BAC construct harboring a chimeric DTR-eGFP-stop coding segment inserted between the first and second codons of the Foxp3 open reading-frame (Fig. 1A). Multiple founders expressing GFP in CD4+CD25+ T cells were obtained, one of which was used to propagate a stable line (NOD.Foxp3DTR+). Two consecutive daily administrations of DT into young adults of this line resulted in an 80–90% depletion of the Foxp3+ T cells in lymphoid organs such as the spleen, axillary LN and PLN (data not shown, but see below). By three days after the last injection, Treg numbers had recovered to a substantial degree (to only ~60% depletion) (data not shown). If DT was administered to young adults every other day until day 9 and histological analysis performed on day 12, only very mild infiltrates were detected in the lung and liver of Foxp3-DTR transgenic mice (but not of transgene-negative littermate-control animals) (Supplementary Fig. S1). On the other hand, the pancreatic islets of the DT-injected transgenic mice showed a strong immune infiltrate at that time-point after treatment, in contrast to the identically treated littermate controls.

Fig. 1
Generation of Foxp3-DTR BAC transgenic NOD mice and kinetics of the autoimmune attack on the pancreas

The goal of this study was to elucidate the role of Tregs in controlling the progression of autoimmune diabetes. To that end, we crossed the Foxp3-DTR transgene into the BDC2.5 TCR transgenic mouse model. The BDC2.5 line derives from a CD4+ T cell clone that is restricted by the NOD MHC-class-II Ag7 molecule and is specific for an unknown pancreatic islet β-cell protein (Katz et al., 1993; Haskins et al., 1988). This line has been instrumental in elucidating the constellation of immunoregulatory genes, molecules and cells that control the frequency and aggressivity of diabetogenic T cells (Gonzalez et al., 1997; Luhder et al., 1998; Kanagawa et al., 2002; Poirot et al., 2004a). In BDC2.5 mice, T cells invade the islets at 15–18 days of age and seed a massive infiltration therein; however, on the NOD genetic background, progression to diabetes occurs only months later in only 10–20% of animals, reflecting strong immunoregulation at play. Thus, this model has the attractions of being well studied, having a T cell compartment that can be easily tracked, being conveniently synchronous and exhibiting readily discernable immunoregulation.

BDC2.5/NOD.Foxp3DTR+ double-transgenic mice showed an efficiency of Treg depletion, including ablation of the population residing in the pancreas, similar to that mentioned above for the polyclonal NOD.Foxp3DTR+ animals (Fig. 1B). In a first set of experiments, the double-transgenic mice were assessed 24, 72 and 120 hours after the depletion of Tregs. As documented in Fig. 1C, histological changes could be observed as early as the 24-hour time-point. The innocuous, static-looking infiltrate characteristic of negative-control mice had begun to evolve such that the leukocyte mass seemed to breach its boundaries and infiltrate deeply into the islet space; in addition, the border between the endocrine and exocrine tissue became much more fuzzy (second row of panels, arrowhead). After 72 hours, the leukocytes appeared to aggressively swarm throughout the islets, even beginning to invade the exocrine pancreas (third row, arrowhead). During the next two days, the islet completely disaggregated, making it difficult to recognize cells, and the infiltrate escaped massively into the exocrine tissue (bottom row, arrowhead). This picture is very reminiscent of what we previously documented for Foxp3-defective and Rag-deficient BDC2.5 TCR transgenic mice (Chen et al., 2005).

In agreement with these findings, diabetes occurred as soon as 3–5 days after the punctual ablation of Tregs from adult BDC2.5/NOD.Foxp3DTR+ mice (Fig. 1E), clearly demonstrating an ongoing need for Tregs to guard against autoimmune attack of the pancreas.

The removal of Tregs activates Teff cells, predominately within the pancreas

Given the strength and synchrony of the disease that rapidly developed subsequent to Treg depletion, this system seemed a highly advantageous one for elucidating how Tregs rein in Teff cells in the T1D context. First, we asked where Tregs were having the greatest effect: in the PLNs, the site of ongoing T cell priming, or within the insulitic lesion, itself? We initially employed micro-array gene-expression profiling as a broad, unbiased approach to addressing this question, focusing on CD4+ T cells, as they are the primary effector cells in the BDC2.5 model. Non-Treg T cells were sorted as GFP-negative at 0, 15 and 24 hours after a single injection of BDC2.5/NOD.Foxp3DTR+ mice and BDC2.5/NOD.Foxp3DTR− littermate controls with DT. RNA samples were prepared in triplicate, were hybridized to Affymetrix M430v2 microarrays, and RMA-normalized expression values were analyzed. (All datasets have been deposited at NCBI/GEO under accession # XXX.)

A global comparison of gene-expression values revealed that far more extensive changes occurred in the pancreas-derived CD4+ Teff cells than in their PLN counterparts: at an arbitrary counting threshold of 2-fold changes, 352 genes were over-expressed in the pancreas 24 hours after Treg ablation, while 168 genes were under-represented; versus 46 and 26, respectively, in the PLN (Fig. 2A, left panels; and listed in Supplementary Table S1). A direct comparison of transcript alterations in a FoldChange/FoldChange (FC/FC) plot demonstrated that many of the same changes occurred in the pancreas and the PLNs, but the off-diagonal alignment (slope = 0.36) confirmed the dominant pancreas response (Fig 2B, left panel). At 15 hours, CD4+ Teff cells from both sites exhibited only minor alterations in gene expression (Fig. 2A and B, right panels); however, highlighting on the 15-hour plots those genes whose expression had changed by 24 hrs revealed that the response was already beginning at the earlier time-point for certain of the loci, with a bias for both over-expressed (in red) and under-represented (in blue) genes (Fig. 2C). Thus, elimination of Tregs had a dominant impact within the insulitic lesion, rather than in the draining lymph nodes, and the response was surprisingly rapid, discernible by as few as 10–12 hours or so (considering that the death of Tregs in response to DT must take at least a few hours).

Fig. 2
T cell activation and proliferation in the pancreatic lesion

To begin characterizing the effects on Teff cells, we highlighted on the FC/FC plot of Fig. 2D a T cell activation/proliferation signature previously defined on the basis of compiled analyses of T cells stimulated by antigen in vitro or in vivo (Hill et al., 2007). At 24 hours after Treg depletion, transcripts of the activation/proliferation signature were clearly altered in the pancreas, but far less so in the PLN. There was little to no change in this signature at 15 hours. For an independent verification of these conclusions, we isolated CD4+ T cells from the pancreas and PLN at 15, 20 and 24 hours after Treg removal, and analyzed them by flow cytometry for expression of the early-activation marker CD69 (Fig. 2E). Pancreas-resident CD4+ T cells showed an increase in cell-surface display of CD69 between 15 and 20 hours, consistent with the gene-expression profile. As expected, PLN-derived CD4+ T cells did not exhibit a significant change in CD69 expression at these early time-points.

Since the highlighted signature included proliferation-associated transcripts, we confirmed the above conclusions in bromodeoxyuridine (BrdU)-labeling experiments. Mice were intravenously (iv)-injected with BrdU 48 hours after DT administration, the pancreas and PLN were excised one hour later, and cells were stained for incorporation of BrdU into their DNA as an indicator of proliferation. We chose this very early time-point to exclude the possibility of division elsewhere followed by labeled-cell migration into the pancreas. Again, pancreas-resident CD4+ T cells showed a substantial early reaction: about 13% were in S phase during the labeling period, compared with only about 2% in the DT-treated DTR-negative littermate-control group (Fig. 2F). In contrast, the division of PLN-derived CD4+ T cells was not significantly different from that of their negative-control counterparts. The dichotomy in proliferation was also seen with CD8+ T cells from the pancreas versus PLN, with high rates in the former and less increase in the latter (Supplementary Fig. S2).

By these diverse criteria, then, Treg ablation can unleash a rapid activation of CD4+ Teff cells residing within the autoimmune lesion, while not in the draining LNs.

The CD4+ Teff cell response unleashed by Treg cell depletion

Thus, Tregs restrain Teff cells within the insulitic lesion, inhibiting terminal destruction of the mass of β-cells. But what are the critical effector functions thereby kept in check? And what molecules drive these functions? Many of the genes highly induced in response to Treg ablation are implicated in the function of known Th subsets, notably transcripts encoding Tbet (3.8-fold over-expressed), IL-12rb1 (2.3-fold), IFN-γ (9.9-fold), granzyme B (6.1-fold) and granzyme A (2.6-fold) for Th1 cells; and IL-17 (2.3-fold), IL-22 (9.3-fold) and IL-21 (3.7-fold) for Th17 cells (Fig. 3A and Supplementary Table 1). A number of these alterations were confirmed by flow-cytometric or quantitative-PCR analysis (Fig. 3B and C, and see below). For example, monoclonal antibody (mAb) staining for intracellular IFN-γ and IL-17 showed an increase in pancreas Teff cells at 48 hours after Treg depletion, and also an augmentation in IFN-γ production in PLN cells at this later time-point (Fig 3B and C). In contrast, there were no evident differences in the intracellular expression of TNF-α, IL-4, IL-10 and IL-2 in pancreas Teff cells (Supplementary Fig. S3).

Fig. 3
Molecular changes after Treg cell ablation

We wondered whether IFN-γ might be an important “driver” cytokine because: 1) transcripts encoding Tbet, which induces this cytokine’s expression, were up-regulated in pancreas CD4+ Teff cells already at 15 hours after Treg ablation (Supplementary Table 1); 2) transcripts encoding IFN-γ itself, though not yet induced in this population at 15 hours, were strongly up-regulated by 24 hours (Supplementary Table 1), and more of the cells were actually producing this cytokine (Fig. 3B); and 3) isolated BDC.2.5 Th1 cells rapidly provoke autoimmune diabetes upon transfer (Katz et al., 1995). Thus, we overlaid an IFN-γ-response signature [independently derived (J. Wu, D. Mathis, C. Benoist, unpublished)] onto the FC/p-value “volcano” plots of the CD4+ T cell expression values. A clear displacement of the IFN-γ-induced (red) and -repressed (blue) transcripts was already evident at 15 hours (Fig. 4A, i and ii; p=4.85−16). A control TGFβ signature showed no such bias -- if anything, the TGFβ-induced transcripts were slightly under-represented (Fig. 4A, iii and iv).

Fig. 4
Induction of IFN-γ expression is an early, predominant response after punctual Treg cell depletion

Subsequent to Treg ablation, IFN-γ-responsive genes were up-regulated in CD4+ Teff cells faster than the genes of the activation signature were, and faster than the IFN-γ gene itself was (Fig. 4B, upper and lower panel), prompting us to search for other cells producing this cytokine. We performed a careful kinetic analysis around this early time-point, quantifying IFN-γ transcript levels in CD4+, CD8+ and NK cells from various organs (Fig. 4C). All cell-types from all organs of the DT-treated DTR-negative littermate controls showed background levels of IFN-γ transcripts at all of the time-points. Unexpectedly, NK cells exhibited the earliest, greatest response. There was a burst of IFN-γ expression by NK cells from the insulitic lesion as soon as 15 hours after DT injection, which was barely detectable in the PLN NK cells and was undetectable in the spleen NK population. In clear contrast, IFN-γ transcripts were not up-regulated in either the CD4+ or CD8+ cells at this time-point, but only after a lag of about 9 hours, though only in the lesion, and never reaching the levels in NK cells.

NK cells sense the loss of Treg control

The striking induction of IFN-γ expression by NK cells in response to declining Treg numbers suggested that this innate immune system cell-type might play a primordial role in diabetogenesis in this setting. We first addressed this possibility by analyzing the percentage, proliferation and effector-molecule expression of NK cells in the PLN and pancreas during the course of Treg ablation. The percentage of NK cells in the pancreas steadily increased along with the decrease in Tregs, rising from about 2% to about 10% by 48 hours, with the first significant augmentation occurring between 15 and 20 hours after the first administration of DT (Fig. 5A, upper and lower panels). The rise in NK cells in the PLN was more moderate and lagged somewhat behind, a significant increase not being apparent until 48 hours (Fig, 5A, lower panel.) BrdU-labeling experiments at 48 hours revealed NK cells in both the pancreas and PLN to be actively proliferating (Fig. 5B). Again, the short (1 hour) BrdU pulse argued that the bulk of the BrdU labeling represented in situ proliferation rather than an influx of cells that had divided elsewhere.

Fig. 5
The loss of Treg cell control triggers NK-cell activation within the autoimmune lesion

We wondered whether the induction of IFN-γ synthesis by NK cells was accompanied by an increase in the NK-cell killing potential as well. Indeed, PCR analysis at 24 hours after DT-injection did show an augmentation in transcripts encodin granzymes A and B, especially in the insulitic lesion, and less so in the PLN (Fig. 5C). While Gzma mRNA was expressed essentially only in NK cells, Gzmb transcripts were up-regulated in pancreatic CD4+ and CD8+ T cells in addition (data not shown). An in vivo cytoxicity assay confirmed the unleashing of NK cell activity upon Treg ablation (Fig 5D). DTR-positive or DTR-negative BDC2.5/NOD mice were depleted of Tregs two days before injection of a mix of splenocyte populations (CFSElo MHCI-negative targets and CFSEhi MHCI-positive controls); sixteen hours later, NK cell activity in different organs was measured by comparing the ratio of the two populations. NK activity was significantly enhanced after Treg depletion, in the pancreatic tissue and training LN, and also in the spleen at this rather late time-point. These results provide important evidence that Treg ablation impacts on NK-cell activity, itself, and in an in vivo context.

The IFN-γ induction triggered by a loss of Treg cells is pro-diabetogenic

We have shown, then, that specific, punctual depletion of Treg cells provoked a rapid rise in NK-cell production of IFN-γ in pancreas tissue targeted by autoimmunity, which in turn substantially impacted on the gene-expression program of the pancreas-resident CD4+ Teff cells. The implication is that IFN-γ is an important trigger of diabetogenesis in this context, a notion we tested by performing mAb blocking studies. Anti-IFN-γ mAb was administered coincident with the DT treatment, and its effects were monitored 24 hours later. One consequence was a substantial reduction in the augmentation of pancreatic NK cells usually provoked by Treg ablation (Fig. 6A). In addition, there were effects on the expression of effector genes in NK and other cell populations. Anti-IFN-γ mAb treatment inhibited the induction of IFN-γ gene transcription in NK cells, CD4+ T cells and CD8+ T cells in the pancreas that usually occurred in the absence of Tregs (Fig. 6B). The induction of Th17 effector cytokines was attenuated even more, suggesting that they might be a downstream event (Fig. 6C). Most importantly, the diabetes normally triggered by Treg ablation was strongly inhibited by IFN-γ blockade – being both delayed and less frequent (Fig 6D).

Fig. 6
IFN-γ drives the autoimmune response provoked by Treg cell removal

Lastly, we sought to establish a role for NK cells, themselves, in unleashing Teff cell activities. Treatment with reagents specific for asialo-Gm1 or NKp46 did not lead to effective NK-cell depletion, therefore, we introduced the NK1.1 marker into the BDC2.5/NOD.Fox3DTR line by crossing it with a NOD.NK1.1 congenic line (Carnaud et al., 2001). As expected, ablation of Tregs from the F1 mice by DT treatment for 24 hours resulted in an augmented population of NK cells, which was substantially depleted by co-administration of anti-NK1.1 (Fig 7A). There were accompanying reductions in the expression of the genes encoding both IFN-γ and granzyme-B by pancreatic CD4+ T cells (Fig 7B), reductions that correlated nicely with the loss of NK cells (Fig 7C). In contrast, the expression of granzyme B by the remaining NK cells was unaltered (Fig 7B), arguing for its independence of NK-cell-produced mediators (notably IFN-γ).

Fig. 7
NK cells directly influence CD4+ T cell activation after Treg cell removal


This study addressed where and how Tregs impact on autoimmunity, in particular the autoimmune attack on the pancreatic islets that preludes T1D. We chose a punctual loss-of-function approach that permitted specific, controlled elimination of Tregs in pre-diabetic BDC2.5/NOD mice with an ongoing, but contained, insulitis. Acute Treg ablation had a devastating effect in this context: within hours, the “well-behaved” leukocytes within the insulitic lesion were unleashed, culminating in armed effector cells that rapidly executed terminal destruction of the mass of islet β cells. The strength and synchrony of this response facilitated dissection of the intervening molecular and cellular events. The earliest and most striking changes subsequent to Treg depletion were observed within the infiltrated islets rather than in the allied lymphoid tissue. An activated, expanding population of NK cells drove an early increase in the production of IFN-γ, which had a dominant impact on the gene-expression program of CD4+ T effector cells and ultimately provoked the development of clinical diabetes.

It may be instructive, first of all, to consider these findings in the context of previously reported experiments entailing punctual ablation of Tregs. Using a DTR-knockin approach, which resulted in ~98% depletion of this regulatory population, the Rudensky group found Tregs to be required throughout life to avoid autoimmunity in C57Bl/6 (B6) (non-autoimmunity-prone) mice: in their absence, a devastating multi-organ autoimmune disease developed in 1–2 weeks (Kim et al., 2007; Lund et al., 2008). Quite different results were published by Lahl et. al., who employed a DTR-BAC strategy that resulted in only about 90% Treg depletion: while B6 neonates lacking this population developed raging autoimmunity, adults did not (Lahl et al., 2007). Apparently, even a small Treg compartment is capable of controlling this form of autoimmunity in adult mice. Our experimental system proved to be rather like that of Lahl et. al. DT treatment of the NOD.Foxp3DTR+ line resulted in 80–90% elimination of Tregs; in 4-week-old NOD individuals, not yet showing signs of the insulitis that ultimately afflicts 100% of NOD mice, Treg depletion to this level led to mild leukocytic infiltrates in the lung and liver, accompanied by strong invasion of the pancreas, but only after 12 days. Thus, the interpretation of results from our system should not be confounded by the side effects of devastating systemic autoimmunity. Another point to keep in mind is that autoimmunity is already installed in our system, using pre-diabetic BDC2.5/NOD mice, so that a lymph-node priming event triggered by the loss of Tregs is not a requirement for disease to initiate. This is most likely the reason why we saw autoimmune manifestations already within hours, rather than in the several days reported for the B6 systems (Kim et al., 2007; Lund et al., 2008).

A key, somewhat unexpected, finding was the early activation of NK cells after Treg depletion, resulting in their expansion, a burst of IFN-γ production as rapidly as 15 hours later, and subsequent expression of effector molecules like granzymes A and B. Teff cells responded very rapidly to the increase in IFN-γ and their ultimate destructiveness towards islet β-cells depended on it. It is not yet known to what extent β-cell death might also depend on the direct killing function of NK cells, perhaps via granzyme activity. While NK cells have recently been reported to control the destructiveness of insulitis in a few particularly aggressive models of autoimmune diabetes (Poirot et al., 2004b; Alba et al., 2008), there has been no hint that their behavior is kept in check so immediately and powerfully by Tregs. That NK cells can be subjugated by Treg control has been reported in other contexts. They become activated and/or expand in mice constitutively or punctually depleted of Tregs (Kim et al., 2007; Terme et al., 2008; Ghiringhelli et al., 2005; Lund et al., 2008), and Tregs can augment various of their in vivo activities, for example their proliferation, cytotoxicity, NKG2D receptor expression, and ability to control tumor growth. However, in all of these previous studies, the secondary lymphoid organs were the focus of attention, and NK cells were thought to join the action rather late, subsequent to and dependent on critical interactions between DCs and CD4+ Teff cells (Kim et al., 2007; Terme et al., 2008). The early involvement of NK cells in the diabetogenic process provoked by Treg depletion in the BDC2.5/NOD system most likely reflects their pre-recruitment to the insulitic lesion and at least partial activation therein. It is not yet clear whether the Tregs act directly on NK cells to dampen IFN-γ production, as they have been seen to do in vitro in a different disease setting (Ghiringhelli et al., 2005). Alternatively, they might operate through the intermediary of DCs, perhaps keeping in check an IL-12/IFN-γ feed-forward loop, reminiscent of what is being observed in more and more experimental contexts (Martin-Fontecha et al., 2004; Wu et al., 2007). Indeed, our most recent studies on even earlier time-points after Treg ablation (illustrated in Supplementary Fig S4) revealed induction of the IFN-γ gene in pancreatic NK cells already at 7 hours (after DT treatment), followed shortly thereafter (12 hours) by an increase in DC expression of the IL-12p40 gene. Anti-IL-12p40 mAb treatment reduced, though did not extinguish, IFN-γ gene expression by NK cells, but did not influence granzyme B expression over the same time-frame.

Thus, Treg control of IFN-γ production, and thereby effector cell activities, directly within infiltrated islets of BDC2.5/NOD mice seems to be an important element of their prolonged restraint of the progression of insulitis to overt diabetes. This conclusion does not mean, however, that in the setting of unmanipulated mice, BDC2.5 or other, the long-delayed spontaneous emergence of effector cells from Treg control and development of diabetes necessarily requires IFN-γ. Indeed, NOD mice show only mild disease attenuation in the absence of IFN-γ or IFNγ-R (Hultgren et al., 1996; Kanagawa et al., 2000; Serreze et al., 2000); although, on the other hand, injection of an anti-IFN-γ mAb was found to have a strong effect (Debray-Sachs et al., 1991), T-bet is a requirement (Esensten et al., 2009), and IFN-γ-producing Th1 cells potently induce disease on transfer (Katz et al., 1995). It may be that Treg control of IFN-γ production keeps the autoimmune lesion in check for a protracted period (and in its absence diabetes is immediate and intense), but that with time effector cells bypass this restraint and proceed to decimate the β cells by an IFN-γ-independent mechanism. An obvious possibility would be through the slow emergence of a Th17 or similarly auto-aggressive population. The emergent population might be less effectively controlled by Tregs, so such a switch could explain observations of a reduced susceptibility of CD4+ Teff cells to Treg suppression in aging NOD mice as they develop clinical diabetes (You et al., 2005; Gregori et al., 2003). Moreover, it is possible that any such alternative CD4+ Teff cell population might expand in the long-term, but not acute, absence of IFN-γ, thus explaining the data discrepancies mentioned above.

Treg cell-based therapy is currently seen as an attractive strategy for preventing or halting the progression of T1D, as well as multiple other autoimmune disorders. For such an approach to be optimally successful, it will be highly advantageous to know precisely where and how Tregs are exerting their impact, which may well vary in different disease settings.



Mice expressing the human DTR under the control of foxp3 transcriptional control elements (Foxp3DTR) were generated by BAC transgenesis. The BAC construct spanned from 150kb upstream to 70 kb downstream of the Foxp3 transcription start-site. A DTR-eGFP cDNA with a stop codon was inserted between the first and second codons of the Foxp3 open reading-frame. The recombinant Foxp3-DTR-eGFP BAC was injected into NOD fertilized oocytes, and offspring were genotyped by PCR.

NOD/LtDOI (NOD), NOD.Foxp3DTR, and BDC2.5/NOD TCR transgenic mice (Katz et al., 1993) were bred in the specific-pathogen-free facility of the Joslin Diabetes Center. Four week old NOD and 4–6-week-old female BDC2.5/NOD.Foxp3DTR+ mice and BAC-negative control animals were ip-injected with DT (Sigma, St. Louis, MO) (40ng/g body weight), and were analyzed at the indicated time-points.

For NK cell depletion: BDC2.5/NOD.Foxp3DTR line was crossed with the NK1.1-congenic NOD line (Carnaud et al., 2001). NK cell depletion was performed by i.p.-injection of the depleting mAb against NK1.1 (clone: PK136; 0.5mg) one day before the injection of DT. B2m deficient NOD mice were used as splenocyte donors in the in vivo NK cytotoxicity assay. IFN-γ was blocked by i.p.-injecting of anti-IFN-γ mAb (clone: R4-6A2), control rat Ig together with the first injection of DT. Mice received a second injection of anti-IFN-γ mAb on day 2. IL-12p40 was blocked by i.p.-injection of mAb against IL-12p40 (clone: C17.8; 0.5mg) together with the injection of DT.

Disease assays

For histology, the pancreas was excised at different time-points after DT injection, was fixed in formalin, and stepped five-micron sections were stained with H+E. For diabetes incidence studies, mice were DT-injected on days 0, 1, 3 and 5 (or until diabetes developed, but not longer than 5 days), and diabetes was assessed by measuring blood-glucose levels (350 mg/dl on two consecutive draws).

Cell sorting and flow cytometry

Different leukocyte subsets were isolated from the spleen, PLN and pancreas at the indicated time-points after Treg depletion, and the cells stained with the stated mAbs. In some experiments, leukocytes were sorted directly into Trizol using the Moflo instrument. Isolated RNA was used for quantitative PCR or microarrays

Cell-surface and intracellular stainings were performed with m Abs against: CD3 (clone: 145-2C11, BD), CD4 (clone: RM4-5, BioLegend), CD8 (clone: 5H10, BD), CD19 (clone: 6D5, Invitrogen), CD49b (clone: HMa2, BD), CD25 (clone: PC61, eBioscience), CD69 (clone: H1.2F3, BD), Foxp3 (clone: FJK-16s, eBioscience), IFNγ (clone: XMG1.2, BD), TNF-α (clone: MP6-XP22, BD), IL-2 (clone: JES6-5H4, BD), IL-4 (clone: 11B11, BD), IL-17 (clone: TC11-18H10, BD) or IL-10 (clone JES5-16E3, BD). Foxp3 staining was performed according to the manufacturer’s instructions (eBiosciences). For intracellular cytokine staining, cells were stimulated with PMA (50ng/ml) (Sigma) and ionomycin (1nM) (Calbiochem) for 4 hours. Golgistop (BD) was added to the culture in the recommended amount during the last three hours, followed by fixing and permeabilisation according to the manufacturer’s instructions (BD). In vivo NK cell cytotoxicity assay: Treg cells were depleted in BDC2.5/NOD.Foxp3DTR+ mice by DT 2 days before 1×107 CFSE labeled target cell were i.v. injected. Target cells were a mix of splenocyte populations (CFSElo MHCI-negative targets [B2m deficient NOD cells] and CFSEhi MHCI-positive controls [WT NOD cells]); sixteen hours later, NK cell kill activity was measured in different organs by comparing the ratio of the two populations in reference to the input mix.

For BrdU staining, BrdU (1 mg, Sigma) was iv-injected 48 hours after DT administration, and the mice were sacrificed one hour later. The pancreas and PLN were removed and digested for about 15 minutes with collagenase type IV (1 mg/ml; Sigma) and DNAse (0.5 mg/ml; Sigma). Cells were fixed and permeabilized according to the manufacturer’s instructions (BD) for BrdU staining, and were then analyzed using the LSRII instrument and FlowJo software.


RNA was prepared from sorted CD4+ T cell populations from BDC2.5/NOD.Foxp3DTR+ mice using Trizol as described (Yamagata et al., 2004). RNA was amplified for two rounds (MessageAmp aRNA, Ambion), biotin-labeled (BioArray High Yield RNA Transcription Labeling, Enzo), and purified using the RNeasy Mini Kit (Qiagen). The resulting cRNAs were hybridized to M430 2.0 chips (Affymetrix). All of the cell populations analyzed were generated in duplicate or triplicate. Raw data were normalized using the RMA algorithm implemented in the “Expression File Creator” module from the GenePattern software package (Reich et al., 2006). Data were visualized using the “Multiplot” modules from GenePattern.

Supplementary Material



We thank John Stockton for producing the DTR transgenics, Joyce LaVecchio and Giridesh Buruzala for flow cytometry, and Jonathan Hill, Jasmine Perez and Kristen Leatherbee for help with the microarray analyses, Hsin-Jung Wu for providing the IFN-γ signature. This work was supported by grants from the Juvenile Diabetes Research Foundation (JDRF) (4-2007-1057) and the NIH (R01 DK59658), and Young Chair funds to DM and CB; by grants from the Sandler Program in Asthma Research and the National Multiple Sclerosis Society to DRL; and by the core facilities of Joslin Diabetes Center’s National Institutes of Diabetes and Digestive and Kidney Diseases funded Diabetes and Endocrinology Research Center and of the JDRF Center on Immunological Tolerance in Type-1 Diabetes at Harvard Medical School. MF was supported by postdoctoral fellowships from the German Research Foundation (Emmy-Noether Fellowship, FE 801/1-1) and the Charles A. King Trust Postdoctoral Fellowship.

Abbreviations used in this paper

regulatory T cells
effector T cells
natural killer
T cell receptor
lymph node
T helper
pancreatic LN
tumor necrosis factor
diphtheria toxin
monoclonal antibody


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Alba A, Planas R, Clemente X, Carrillo J, Ampudia R, Puertas MC, Pastor X, Tolosa E, Pujol-Borrell R, Verdaguer J, et al. Natural killer cells are required for accelerated type 1 diabetes driven by interferon-beta. Clin Exp Immunol. 2008 [PubMed]
  • Bennett CL, Christie J, Ramsdell F, Brunkow ME, Ferguson PJ, Whitesell L, Kelly TE, Saulsbury FT, Chance PF, Ochs HD. The immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX) is caused by mutations of FOXP3. Nat. Genet. 2001;27:2021. [PubMed]
  • Bennett CL, Clausen BE. DC ablation in mice: promises, pitfalls, and challenges. Trends Immunol. 2007;28:525–531. [PubMed]
  • Bour-Jordan H, Salomon BL, Thompson HL, Szot GL, Bernhard MR, Bluestone JA. Costimulation controls diabetes by altering the balance of pathogenic and regulatory T cells. J Clin Invest. 2004;114:979–987. [PMC free article] [PubMed]
  • Carnaud C, Gombert J, Donnars O, Garchon H, Herbelin A. Protection against diabetes and improved NK/NKT cell performance in NOD.NK1.1 mice congenic at the NK complex. J. Immunol. 2001;166:2404–2411. [PubMed]
  • Chen Z, Herman AE, Matos M, Mathis D, Benoist C. Where CD4+CD25+ T reg cells impinge on autoimmune diabetes. J Exp Med. 2005;202:1387–1397. [PMC free article] [PubMed]
  • Debray-Sachs M, Carnaud C, Boitard C, Cohen H, Gresser I, Bedossa P, Bach J-F. Prevention of diabetes in NOD mice treated with antibody to murine IFNγ J. Autoimmun. 1991;4:237–248. [PubMed]
  • Esensten JH, Lee MR, Glimcher LH, Bluestone JA. T-bet-deficient NOD mice are protected from diabetes due to defects in both T Cell and innate immune system function. J Immunol. 2009;183:75–82. [PMC free article] [PubMed]
  • Feuerer M, Herrero L, Cipolletta D, Naaz A, Wong J, Nayer A, Lee J, Goldfine A, Benoist C, Shoelson SE, et al. Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters. Nat Med. 2009 in press. [PMC free article] [PubMed]
  • Ghiringhelli F, Menard C, Terme M, Flament C, Taieb J, Chaput N, Puig PE, Novault S, Escudier B, Vivier E, et al. CD4+CD25+ regulatory T cells inhibit natural killer cell functions in a transforming growth factor-beta-dependent manner. J Exp Med. 2005;202:1075–1085. [PMC free article] [PubMed]
  • Gonzalez A, Katz JD, Mattei MG, Kikutani H, Benoist C, Mathis D. Genetic control of diabetes progression. Immunity. 1997;7:873–883. [PubMed]
  • Gregori S, Giarratana N, Smiroldo S, Adorini L. Dynamics of pathogenic and suppressor T cells in autoimmune diabetes development. J Immunol. 2003;171:4040–4047. [PubMed]
  • Haskins K, Portas M, Bradley B, Wegmann D, Lafferty KJ. T-lymphocyte clone specific for pancreatic islet antigen. Diabetes. 1988;37:1444–1448. [PubMed]
  • Herman AE, Freeman GJ, Mathis D, Benoist C. CD4+CD25+ T regulatory cells dependent on ICOS promote regulation of effector cells in the prediabetic lesion. J Exp Med. 2004;199:1479–1489. [PMC free article] [PubMed]
  • Hill J, Feuerer M, Tash K, Haxhinasto S, Perez J, Melamed R, Mathis D, Benoist C. Foxp3-dependent and independent regulation of the Treg transcriptional signature. Immunity. 2007;27:786–800. [PubMed]
  • Hultgren B, Huang X, Dybdal N, Stewart TA. Genetic absence of γ-interferon delays but does not prevent diabetes in NOD mice. Diabetes. 1996;45:812–817. [PubMed]
  • Kanagawa O, Militech A, Vaupel BA. Regulation of diabetes development by regulatory T cells in pancreatic islet antigen-specific TCR transgenic nonobese diabetic mice. J Immunol. 2002;168:6159–6164. [PubMed]
  • Kanagawa O, Xu G, Tevaarwerk A, Vaupel BA. Protection of nonobese diabetic mice from diabetes by gene(s) closely linked to IFN-gamma receptor loci. J Immunol. 2000;164:3919–3923. [PubMed]
  • Katz JD, Benoist C, Mathis D. T helper cell subsets in insulin-dependent diabetes. Science. 1995;268:1185–1188. [PubMed]
  • Katz JD, Wang B, Haskins K, Benoist C, Mathis D. Following a diabetogenic T cell from genesis through pathogenesis. Cell. 1993;74:1089–1100. [PubMed]
  • Kim JM, Rasmussen JP, Rudensky AY. Regulatory T cells prevent catastrophic autoimmunity throughout the lifespan of mice. Nat. Immunol. 2007;8:191–197. [PubMed]
  • Korn T, Reddy J, Gao W, Bettelli E, Awasthi A, Petersen TR, Backstrom BT, Sobel RA, Wucherpfennig KW, Strom TB, et al. Myelin-specific regulatory T cells accumulate in the CNS but fail to control autoimmune inflammation. Nat. Med. 2007;13:423–431. [PMC free article] [PubMed]
  • Lahl K, Loddenkemper C, Drouin C, Freyer J, Arnason J, Eberl G, Hamann A, Wagner H, Huehn J, Sparwasser T. Selective depletion of Foxp3+ regulatory T cells induces a scurfy-like disease. J Exp Med. 2007;204:57–63. [PMC free article] [PubMed]
  • Luhder F, Höglund P, Allison JP, Benoist C, Mathis D. Cytotoxic T lymphocyte-associated antigen 4 regulates the unfolding of autoimmune diabetes. J. Exp. Med. 1998;187:427–432. [PMC free article] [PubMed]
  • Lund JM, Hsing L, Pham TT, Rudensky AY. Coordination of early protective immunity to viral infection by regulatory T cells. Science. 2008;320:1220–1224. [PMC free article] [PubMed]
  • Martin-Fontecha A, Thomsen LL, Brett S, Gerard C, Lipp M, Lanzavecchia A, Sallusto F. Induced recruitment of NK cells to lymph nodes provides IFN-gamma for T(H)1 priming. Nat. Immunol. 2004;5:1260–1265. [PubMed]
  • Poirot L, Benoist C, Mathis D. Natural killer cells distinguish innocuous and destructive forms of pancreatic islet autoimmunity. Proc Natl. Acad Sci U S. A. 2004a;101:8102–8107. [PubMed]
  • Poirot L, Benoist C, Mathis D. Natural killer cells distinguish innocuous and destructive forms of pancreatic islet autoimmunity. Proc Natl. Acad Sci U S. A. 2004b;101:8102–8107. [PubMed]
  • Reich M, Liefeld T, Gould J, Lerner J, Tamayo P, Mesirov JP. GenePattern 2.0. Nat. Genet. 2006;38:500–501. [PubMed]
  • Saito M, Iwawaki T, Taya C, Yonekawa H, Noda M, Inui Y, Mekada E, Kimata Y, Tsuru A, Kohno K. Diphtheria toxin receptor-mediated conditional and targeted cell ablation in transgenic mice. Nat. Biotechnol. 2001;19:746–750. [PubMed]
  • Sakaguchi S, Yamaguchi T, Nomura T, Ono M. Regulatory T cells and immune tolerance. Cell. 2008;133:775–787. [PubMed]
  • Salomon B, Lenschow DJ, Rhee L, Ashourian N, Singh B, Sharpe A, Bluestone JA. B7/CD28 costimulation is essential for the homeostasis of the CD4+CD25+ immunoregulatory T cells that control autoimmune diabetes. Immunity. 2000;12:431–440. [PubMed]
  • Sarween N, Chodos A, Raykundalia C, Khan M, Abbas AK, Walker LS. CD4+CD25+ cells controlling a pathogenic CD4 response inhibit cytokine differentiation, CXCR-3 expression, and tissue invasion. J Immunol. 2004;173:2942–2951. [PubMed]
  • Serreze DV, Post CM, Chapman HD, Johnson EA, Lu B, Rothman PB. Interferon-gamma receptor signaling is dispensable in the development of autoimmune type 1 diabetes in NOD mice. Diabetes. 2000;49:2007–2011. [PubMed]
  • Tang Q, Adams JY, Tooley AJ, Bi M, Fife BT, Serra P, Santamaria P, Locksley RM, Krummel MF, Bluestone JA. Visualizing regulatory T cell control of autoimmune responses in nonobese diabetic mice. Nat. Immunol. 2006;7:83–92. [PMC free article] [PubMed]
  • Tang Q, Bluestone JA. The Foxp3+ regulatory T cell: a jack of all trades, master of regulation. Nat. Immunol. 2008;9:239–244. [PMC free article] [PubMed]
  • Tang Q, Henriksen KJ, Bi M, Finger EB, Szot G, Ye J, Masteller EL, McDevitt H, Bonyhadi M, Bluestone JA. In Vitro-expanded Antigen-specific Regulatory T Cells Suppress Autoimmune Diabetes. J Exp Med. 2004;199:1455–1465. [PMC free article] [PubMed]
  • Tarbell KV, Petit L, Zuo X, Toy P, Luo X, Mqadmi A, Yang H, Suthanthiran M, Mojsov S, Steinman RM. Dendritic cell-expanded, islet-specific CD4+ CD25+ CD62L+ regulatory T cells restore normoglycemia in diabetic NOD mice. J Exp Med. 2007;204:191–201. [PMC free article] [PubMed]
  • Tarbell KV, Yamazaki S, Olson K, Toy P, Steinman RM. CD25+ CD4+ T Cells, Expanded with Dendritic Cells Presenting a Single Autoantigenic Peptide, Suppress Autoimmune Diabetes. J Exp Med. 2004;199:1467–1477. [PMC free article] [PubMed]
  • Terme M, Chaput N, Combadiere B, Ma A, Ohteki T, Zitvogel L. Regulatory T cells control dendritic cell/NK cell cross-talk in lymph nodes at the steady state by inhibiting CD4+ self-reactive T cells. J Immunol. 2008;180:4679–4686. [PubMed]
  • Tritt M, Sgouroudis E, d'Hennezel E, Albanese A, Piccirillo CA. Functional waning of naturally occurring CD4+ regulatory T-cells contributes to the onset of autoimmune diabetes. Diabetes. 2008;57:113–123. [PubMed]
  • Wildin RS, Ramsdell F, Peake J, Faravelli F, Casanova JL, Buist N, Levy-Lahad E, Mazzella M, Goulet O, Perroni L, et al. X-linked neonatal diabetes mellitus, enteropathy and endocrinopathy syndrome is the human equivalent of mouse scurfy. Nat. Genet. 2001;27:18–20. [PubMed]
  • Wu HJ, Sawaya H, Binstadt B, Brickelmaier M, Blasius A, Gorelik L, Mahmood U, Weissleder R, Carulli J, Benoist C, et al. Inflammatory arthritis can be reined in by CpG-induced DC-NK cell cross talk. J Exp Med. 2007;204:1911–1922. [PMC free article] [PubMed]
  • Yamagata T, Mathis D, Benoist C. Self-reactivity in thymic double-positive cells commits cells to a CD8 alpha alpha lineage with characteristics of innate immune cells. Nat. Immunol. 2004;5:597–605. [PubMed]
  • You S, Belghith M, Cobbold S, Alyanakian MA, Gouarin C, Barriot S, Garcia C, Waldmann H, Bach JF, Chatenoud L. Autoimmune diabetes onset results from qualitative rather than quantitative age-dependent changes in pathogenic T-cells. Diabetes. 2005;54:1415–1422. [PubMed]
  • Zheng Y, Rudensky AY. Foxp3 in control of the regulatory T cell lineage. Nat. Immunol. 2007;8:457–462. [PubMed]