Schistosoma mansoni egg antigens induce Treg that participate in diabetes prevention in NOD mice

Authors


Abstract

Schistosoma mansoni soluble egg antigens (SEA) profoundly regulate the infected host's immune system. We previously showed that SEA prevents type 1 diabetes in NOD mice and that splenocytes from SEA-treated mice have reduced ability to transfer diabetes to NOD.scid recipients. To further characterize the mechanism of diabetes prevention we examined the cell types involved and showed that CD25+ T-cell depletion of splenocytes from SEA-treated donors restored their ability to transfer diabetes. Furthermore, SEA treatment increased the number and proportional representation of Foxp3+ T cells in the pancreas of NOD mice. We have used in vitro systems to analyze the effect of SEA on the development of NOD Foxp3+ T cells. We find that SEA can induce Foxp3 expression in naïve T cells in a TGF-β-dependent manner. Foxp3 induction requires the presence of DC, which we also show are modified by SEA to upregulate C-type lectins, IL-10 and IL-2. Our studies show that SEA can have a direct effect on CD4+ T cells increasing expression of TGF-β, integrin β8 and galectins. These effects of SEA on DC and T cells may act in synergy to induce Foxp3+ Treg in the NOD mouse.

Introduction

Parasitic helminths like Schistosoma mansoni have co-evolved with the immune systems of their mammalian hosts and have developed many strategies to suppress host responses. Intravenous egg deposition by female worms is accompanied by Th2 skewing of the host's immune system, which has been linked to the induction of DC with a suppressive phenotype and alternatively activated macrophages (MΦ) 1. Although in vitro parasite-conditioned DC have been shown to retain an immature phenotype and have been shown to mediate an in vivo Th2 response indicative of parasite-mediated effects 2, most studies failed to identify clear phenotypic changes in DC. We and others have shown that after exposure to S. mansoni soluble egg antigens (SEA), DC have a reduced ability to secrete inflammatory mediators in response to TLR stimulation providing further evidence of SEA-mediated effects in DC 3–6.

Foxp3+ Treg are important in regulating immune responses to helminth infections 7–10 and in preventing autoimmunity 11. In particular, the importance of Treg during the egg deposition phase of S. mansoni infection has been shown, with a demonstrated role in suppressing Th1 pathology while permitting a contained Th2 response 8–10. However, links uniting the effect of SEA on DC with Treg functional effects have not so far been described.

In the NOD mouse model of autoimmune type 1 diabetes we have shown that exposure of mice to either live S. mansoni infection or parasite-derived soluble antigen confers permanent protection against diabetes 3. Increase in the frequency and activation of Foxp3+ Treg has been shown to prevent diabetes in NOD mice 12, 13. Foxp3+ Treg generation is largely dependent on stimulation by immature DC in an IL-10, IL-2, and TGF-β-rich cytokine environment, and infections that create these conditions can expand and activate Foxp3+ Treg, thus regulating the host immune response 14. Foxp3+ Treg modulate T effector cell function in vitro via cell to cell contact; and CTLA-4, GITR, galectin 1, and membrane-bound TGF-β might be involved in Treg immunosuppression 11, 13. There is evidence that galectins have a role in modulating inflammatory and immune responses to schistosome infection 15, 16. Galectins have the potential to alter sensitivity to apoptosis and cytokines (in particular for TGF-β) and also to be involved in the binding of glycans in S. mansoni egg antigens 17, 18. Expression of latency-associated peptide (LAP) and integrin αvβ8, which have been shown to be involved in the activation of TGF-β19, 20, are also a measure of Foxp3+ Treg activation and/or suppressive activity.

In this study we provide evidence of phenotypic changes in DC following exposure to SEA. We have furthermore investigated the role of Foxp3+ Treg in SEA-mediated diabetes protection in the NOD mouse model. Finally, we extended our interest to the expression of galectins as potential “receptors” for SEA on T cells, suggesting that SEA-induced expansion of Foxp3+ Treg might be the result of both indirect APC-mediated and direct effect on T cells.

Results

SEA treatment increases Foxp3 Treg in the pancreas of NOD mice

We have shown that although SEA treatment prevents diabetes in NOD mice 3, a peri-insular infiltrate is still present in the pancreas 3. Furthermore, spleen cells from SEA-treated donor mice had a reduced capacity to transfer diabetes suggesting either a lack of effector T cells or dominant immunoregulation. High-dose cyclophosphamide is known to precipitate diabetes in NOD mice, an observation associated with reduction in Treg 12. We find that cyclophosphamide treatment overrides SEA-mediated diabetes prevention (data not shown), suggesting that potentially autoreactive T cells might be present in SEA-treated mice but that their autoreactivity is regulated. Therefore, we investigated if SEA administration influences Treg in the pancreas.

Following two i.p. SEA injections, there was a significant increase in CD4+Foxp3+ T cells in the pancreas of treated NOD mice. Figure 1A shows that not only was there an increased proportional representation of Foxp3+ cells among the infiltrating CD4+ T cells but also that the absolute numbers of Treg significantly increased in the pancreas of SEA-treated mice. Figure 1B shows that Treg in the pancreas of SEA-treated mice display a more activated phenotype since a higher percentage express CD25 and GITR. Foxp3+ T cells are also proportionally increased in the spleen of SEA-treated NOD mice but not in the pancreatic lymph nodes (Fig. 1C and data not shown). To further examine whether SEA induced a regulatory environment in the pancreas we used real-time PCR to analyze mRNA expression of cytokines and C-type lectin receptors (CLR) in the pancreas of untreated and SEA-treated mice. Ten days after SEA injection there were statistically significant increases in expression of TGF-β, IL-10, IL-4, IL-35, and galectin 3 (Fig. 1D). There was no change in the expression of IL-27p28 (data not shown). As not all TGF-β transcripts lead to biologically active protein, we also assessed the expression of PAI-1 21, which can be induced by TGF-β signaling through Smads, and found that it was also increased in response to SEA (Fig. 1D). The presence of IL-35 is of interest as it is expressed by Treg and is important for their function 22.

Figure 1.

SEA treatment expands Treg. Six-week-old female NOD mice were injected twice (days 0 and 7) with 50 μg SEA i.p. Infiltrating pancreatic mononuclear cells and splenocytes were analyzed on day 10 by flow cytometry and real-time PCR. (A) SEA-injected mice show increased numbers and proportional representation of Foxp3+ Treg in the pancreas. (B) Pancreatic Treg upregulate GITR and CD25 in response to SEA. (C) SEA treatment also expands Treg in the spleen. (D) Increased expression of anti-inflammatory cytokines in the pancreatic infiltrate of SEA-treated mice. Y-axis shows relative gene expression (Hprt-RE). Data points represent single (A–C) or mean (D) expression values from five to six individual mice. Similar data were obtained in two additional independent experiments. Statistical analysis by Mann–Whitney U test.

CD25+ cells are key to SEA-mediated protection in diabetes adoptive transfer

We previously showed that splenocytes from NOD mice have a reduced capacity to transfer diabetes to NOD.scid recipients following in vivo exposure to SEA and soluble worm antigen (SWA) (Fig. 2) 3. To determine whether Treg played a role in S. mansoni antigen-mediated diabetes prevention, we carried out adoptive cell transfer experiments where CD25+ T cells were depleted using flow sorting. Following CD25+ cell depletion, splenocytes from SEA-treated NOD mice regained the ability to transfer diabetes in NOD.scid recipients (Fig. 2). Interestingly, depletion of CD25+ T cells from splenocytes of SWA-treated NOD mice does not restore their ability to transfer diabetes in NOD.scid recipients (data not shown). In a second experiment we transferred purified CD4+CD25 cells isolated using Miltenyi's MACS system, and obtained similar results to the experiments where we used flow sorting depletion method (untreated CD4+CD25 transferred at 100% frequency, CD4+CD25 from SEA-treated mice at 75% and data not shown). All together these data suggest different mechanisms of protection between SEA and SWA and a possible role for Treg in diabetes prevention in SEA-treated NOD mice but not SWA-treated NOD mice. As depletion of Treg in vivo also caused the development of other forms of autoimmunity (e.g. colitis and thyroiditis) we turned to in vitro systems to model the early events favoring the protective response in NOD mice.

Figure 2.

Adoptive transfer of NOD splenocytes into NOD.scid recipients. While splenocytes from untreated female NOD mice transmit diabetes into the NOD.scid, those from SEA-treated mice have a reduced capacity to do so. Splenocytes from SEA-treated mice transfer diabetes when depleted of CD25+ cells by flow sorting. Data for whole splenocytes transfer representative of three independent experiments; data for CD25 cells are from a single experiment. Six NOD.scid mice were used in each experimental group. Statistical analysis by log-rank test of Kaplan–Meier survival curves. Only the survival curve for SEA-treated whole splenocytes significantly differs from the other groups (p<0.01) by this test.

SEA generates Treg from naïve CD4 T cells

Using BM-derived DC (BMDC) as APC to present to naïve NOD CD4+CD44lowCD25 T cells, we found that SEA induced the generation of Foxp3+ Treg (Fig. 3A). The generation of Foxp3+ cells in response to SEA is TGF-β-dependent, since its neutralization results in reduced Foxp3 expression (Fig. 3A). In this system SWA does not induce Foxp3 expression (Fig. 3A). Figure 3B presents a summary of five separate experiments demonstrating SEA-mediated induction of Foxp3 in NOD T cells. Surprisingly, cells from C57BL/6 mice, cultured in the same condition, do not generate Foxp3+CD4+ T cells (Fig. 3C). As this Foxp3 generation is TGF-β-dependent we tested whether SEA induced TGF-β expression or secretion in NOD BMDC. Both bioassay for TGF-β activity and real-time PCR for TGF-β expression revealed no changes in NOD BMDC production of this cytokine (data not shown). Furthermore, the BMDC expression of integrin β8, involved in the activation of TGF-β, is not affected by SEA (data not shown). Examination of cytokine mRNA levels did, however, show upregulated IL-10 and IL-2 expression in BMDC after in vitro exposure to SEA with no significant changes in IL-4, IL-12p35, and IL-12p40 mRNA (Fig. 3D and data not shown). Blockade of IL-10 (using a combination of anti-IL-10 and anti-IL-10 receptors) did not reverse the SEA-mediated induction of Foxp3-expressing cells (data not shown). Our findings are consistent with previous studies showing altered gene expression signatures in DC exposed to Schistosome egg antigens 4, 23.

Figure 3.

Polarization assay of naïve T cells stimulated with S. mansoni antigens. Naïve splenic T cells (5×105) were cultured for 5 days with BMDC (5×105) and 0.5 μg/mL αCD3, in the presence of SEA or SWA (10 μg/mL). In the NOD mouse (A), SEA increases Foxp3 expression in a TGF-β-dependent manner, whereas SWA does not. (B) Summary of five independent experiments using NOD mouse cells, with each data point representing the percent of CD4+ T cells expressing Foxp3 in an individual polarization. Statistical analysis by Mann–Whitney U test between indicated groups. (C) SEA does not induce Foxp3 expression in the C57BL/6 mouse. (D and E) In vitro stimulation of NOD BMDC. (D) SEA upregulates IL-2 and IL-10 expression in BMDC as determined by real-time PCR. Y-axis shows relative gene expression (Hprt-RE). Immature BMDC (1×106) were rested overnight and stimulated for 2–8 h with 10 μg/mL SEA. Expression levels for untreated BMDC are indicated at time 0 h. Data shown are means of duplicate reactions, and are representative of three independent experiments. Statistical analysis was performed by one-way ANOVA with Bonferroni Multiple Comparison test between indicated samples. (E) SEA upregulates expression of CLR on NOD BMDC. Isotype control (gray shaded), unstimulated cells (dashed line), and SEA (bold line) at 50 μg/mL shown 24 h after stimulation with MFI. Data are representative of five independent experiments.

S. mansoni encodes a TGF-β homologue, and the observed effects could have been due to its presence in SEA. However, TGF-β was not detected in SEA by ELISA or two different bioassays (data not shown). Furthermore, TGF-β-mediated collagen induction occurs in Schistosomiasis 24, but the SEA here did not upregulate expression of collagens I and III in human fibroblasts suggesting the absence of bioactive schistosome TGF-β (data not shown).

Our previous in vitro studies and those of others showed that SEA did not affect expression of CD80, CD86, and MHC class II on BMDC 3, 6. However, functional differences in cytokine secretion were observed when BMDC are cultured with SEA in the presence of LPS 3. We investigated a wider range of surface markers on BMDC and found that in vitro stimulation with SEA induces consistent, small increases in expression of CLR. Increased surface expression of monaose receptor (MR), DEC-205, and DC-SIGN was seen in SEA-exposed BMDC (Fig. 3E). Recognition and endocytosis of SEA by human DC have been shown to be mediated in part by MR and DC-SIGN 5. As previously shown by us and others, we saw no changes in the surface expression of CD80, CD86, MHC class II, or CD40.

The direct effect of SEA on T cells might favor Foxp3 induction

We found that SEA appeared to favor the survival of, and/or expand, Foxp3+ Treg from αCD3-stimulated splenocytes in vitro (Fig. 4A). This effect is coupled with a dose-dependent anti-proliferative effect seen by both 3H-thymidine incorporation and CFSE dilution (Fig. 4B and C). Schistosome antigens can induce a suppressive phenotype in DC and MΦ 5, 25. Therefore, we investigated whether the anti-proliferative effect was mediated by APC. In the absence of APC, SEA could still strikingly reduce purified CD4+ T-cell proliferation (Fig. 4D) indicating that SEA also has a direct effect on T cells. SEA has been previously demonstrated to induce antigen-induced cell death in CD4+ T cells from previously infected animals 26. It is therefore possible that at high doses of SEA, cytotoxicity from certain components may contribute to the observed inhibition of proliferation, in particular, for activated T cells. We do not believe that the inhibition of proliferation that we see is due to cytotoxicity as we do not observe increases in Annexin V staining on CD4+ T cells stimulated with SEA (data not shown).

Figure 4.

SEA directly interacts with T cells. (A) Increased proportional representation of Foxp3+ Treg following SEA. NOD splenocytes (5×105) were stimulated in vitro with 2 μg/mL αCD3 and 50 μg/mL SEA for 48 h. (B and C) SEA inhibits proliferation. Splenocytes were treated as in (A) with increasing doses of SEA and proliferation was assessed by 3H-thymidine incorporation (B) or CFSE dilution gating on CD4+ cells (C). (D) SEA directly inhibits T-cell proliferation. Purified NOD CD4+ T cells (5×105) stimulated with 2 μg/mL αCD3 and 5 μg/mL αCD28 with proliferation measured as in (B). Data in (B–D) presented as mean±SD of triplicate samples. Data are representative of either two (A and C) or three (B and D) independent experiments. Statistical analysis by one-way ANOVA with Bonferroni Multiple Comparison test between indicated samples.

We therefore investigated the direct effects of SEA on mRNA levels of immunomodulatory genes. When we analyzed the expression of galectins 1 and 3 on CD4+ T cells, a statistically significant increase of mRNA for both of these genes was also seen (Fig. 5A). This upregulation of galectin 1 was specific to NOD T cells, whereas increased galectin 3 was observed in both NOD and C57BL/6 T cells (data not shown). We also found that SEA directly induces surface expression of galectin 3 on CD4+ T cells (Fig. 5B).

Figure 5.

SEA directly induces expression of galectins and TGF-β in NOD CD4+ T cells. (A) SEA increases expression of galectins 1 and 3 in purified lymphatic CD4+ T cells. Cells (5×105) treated with 2 μg/mL αCD3 and 5 μg/mL αCD28 for 24 h using 10 μg/mL SEA, with gene expression assessed by real-time PCR. Y-axis shows relative gene expression (Hprt-RE). (B) CD4+ T cells upregulate surface galectin 3 in response to SEA (25 μg/mL). CD4+ T cells (5×105) stimulated with 2 μg/mL αCD3 and 5 μg/mL αCD28 for 48 h (numbers in brackets show MFI of viable (7-aminoactinomycin D) CD4+ cells). (C) SEA induces expression TGF-β and integrin β8 in CD4+ T cells. Cells were treated as in (A). (D) Appearance of LAP on the surface of CD4+ T cells following stimulation with SEA (25 μg/mL). Cells were treated as in (B), gating on viable CD4+ cells. Data points in (A) and (C) represent mean expression values from six independent experiments. Data in (B) and (D) are representative of at least five independent experiments. Statistical analysis by Mann–Whitney U test.

To determine if there is a link between the direct SEA effect on T cells and Foxp3 induction, we analyzed mRNA levels of factors important for Treg development. SEA directly induced TGF-β and integrin β8 mRNA in CD4+ T cells (Fig. 5C). SEA also increases expression of LAP on the surface of CD4+ T cells, confirming the presence of surface-bound TGF-β (Fig. 5D). As TGF-β is not increased in the BMDC populations treated with SEA, this observation of surface-bound TGF-β on CD4+ T cells identifies the source of this cytokine, which is instrumental in the SEA-mediated generation of Foxp3+ Treg. In this system, we demonstrate that the TGF-β required for Treg generation is actually expressed by CD4+ T cells due to a direct interaction with a parasite product.

Discussion

In vivo both S. mansoni soluble antigens, SEA and SWA, can induce Th2 responses and this has been proposed as a major mechanism for prevention of T1D in NOD mice 3. The Th2 response to SEA and SWA is well characterized and a role for Treg in reducing immunopathology during S. mansoni infection has also been shown 7–10, 27. Previous work by McKee and Pearce has shown that CD4+CD25+ cells present during S. mansoni infection express Foxp3 and do not express Th2 cytokines 9. Our observation in an adoptive transfer system that depletion of CD25+ T cells restored pathogenicity to cells derived from SEA-treated, but not SWA-treated, mice suggested that these schistosome-derived products act to prevent diabetes in different ways. This finding furthermore suggested that SEA-mediated diabetes prevention involves Treg activity.

In accord with this, we find increases in Treg (and Th2 cytokines, Fig. 1D) in the pancreas following SEA treatment, along with upregulation of TGF-β and integrin β8. Among the regulatory cytokines, TGF-β has the most complex activation mechanism. Initially held as part of an inactive complex with the LAP, TGF-β requires proteolytic cleavage or mechanical dissociation to become bioactive. TGF-β's mechanism of activation differs by cell type, and recently integrin β8-mediated mechanical dissociation of the LAP has been shown to be important for APC and lymphocytes 20. CD4+ T cells expressing surface LAP have recently been shown to exhibit potent regulatory activity mediated by TGF-β and often co-express Foxp3 28. It is interesting that SEA upregulates integrin β8 because it indicates that the parasite antigens not only induce TGF-β but also its activation. We show increased numbers and frequencies of Foxp3 Treg in the pancreas of SEA-treated NOD mice, and these cells express high CD25 and GITR amidst an active TGF-β signature. Among Foxp3+ Treg, those expressing GITR have been shown to be critically important for maintaining peripheral tolerance and restraining autoimmune attack on pancreatic islets 29. In the NOD mouse, high surface expression of CD25 has been associated with functional Treg cell activity in Foxp3+ Treg in the pancreas 30. In the context of Foxp3+ Treg function in the pancreas it is interesting that we also found evidence for increased IL-35 expression in the pancreatic infiltrate, a new cytokine demonstrated to be expressed by Treg and which enhances their regulatory function 22.

In addition to demonstrating an increased presence of Treg in the pancreas, we also showed the ability of SEA to generate Treg in vitro. We found that SEA induces a TGF-β-dependent generation of Treg from naïve T cells in NOD mice. Our data demonstrate that the induction of Foxp3 expression by SEA in naïve NOD T cells requires the presence of DC. The inability of SEA to induce Foxp3 in naïve C57BL/6 T cells could be attributable to a difference in T cells or DC or in both cell types. As we also find that C57BL/6 cells respond differently to SEA stimulation and do not upregulate LAP or TGF-β this suggests that at least part of the difference in response lies with the T cell. This does not however exclude additional differences at the DC level. The underlying basis for these differences in responsiveness between the two strains remains unclear but could include differential expression of receptors such as CLR and TLR on either cell type (Burton et al., unpublished data).

Our finding of SEA-mediated upregulation of IL-2 and IL-10 by BMDC suggests one way in which Treg development might be supported in vitro. Foxp3+ Treg display an exceptional requirement for IL-2 31. The absence of TGF-β induction by BMDC suggests a central role for the direct effects of SEA on T cells and the importance of the interaction between SEA-conditioned APC and T cells. We believe that this synergy between the NOD mouse APC and T cell response to SEA is necessary for Foxp3 induction. For this reason, and after observing the direct anti-proliferative effect of SEA on isolated T cells, we further investigated the direct effect of SEA on purified T cells. We have shown enhanced expression of TGF-β and TGF-β-activating integrin β8 in CD4+ T cells, indicating an increased capacity to activate TGF-β at the point of the immunological synapse 32. Our finding of a direct engagement of SEA and T cells raises the possibility of a bidirectional interaction between APC and T cells upon SEA stimulation, with roles for both cells in the induction and maintenance of the new phenotypes. SEA also directly upregulates the expression of galectins 1 and 3 on purified CD4+ T cells. Galectins have been shown to bind to N-acetyl-lactosamine residues present in SEA, providing a “receptor” for SEA to induce direct effects on T cells 17. Galectin 3 has also been shown to crosslink the TGF-β receptors, sensitizing the cell to TGF-β and galectin 1, which is highly expressed on Treg, has been shown to be an important effector molecule for their suppressive capacity 11, 18. Moreover, galectin 1 has a high affinity for activated Th1 and Th17 cells, but does not bind to Th2 cells due to differential glycosylation 33. Our observation that SEA influences galectin expression in NOD T cells is interesting given the role of these molecules in immunosuppression by Treg 11, 34. Other CLR and TLR, which are also expressed on T cells 35, could transmit signals following interactions with SEA and more work is needed to address the relevant receptors that mediate the effects of SEA on T cells.

In conclusion, SEA induces CD4+CD25+Foxp3+ T cells and since the removal of CD25+ T cells overrides diabetes prevention induced by this helminth product, this suggests a role for this pathway in SEA-mediated diabetes prevention in NOD mice. We furthermore highlight for the first time the potential importance of a direct interaction between parasite antigens and T cells to achieve immunomodulation.

Materials and methods

Mice

Female C57BL/6 mice were purchased from Charles River Laboratories. Female NOD/Tac and NOD.scid mice were housed and barrier bred in the Pathology Department, University of Cambridge animal facilities (Cambridge, UK) and used between 4 and 8 wk of age. All work was conducted under UK Home Office project licence regulations after approval by the Ethical Review Committee of the University of Cambridge.

Mice were routinely monitored for glucosuria using Diastix (Bayer), and were deemed diabetic if urinary glucose exceeded 5 g/L.

Preparation of SEA and SWA

Preparation of SEA and SWA was described previously 36. Briefly, for SEA, eggs were harvested from the livers of outbred infected mice, which were treated to prevent granuloma formation. Livers were homogenized through sieves and the eggs were collected, washed, and sonicated in PBS on ice, prior to centrifugation to separate the saline soluble fraction. SWA was prepared from adult worms recovered from the portal venous vasculature. Both antigens were sterile filtered and endotoxin removed to <1 EU/mg using Polymixin B beads (Sigma).

Cell culture and reagents

Cells were cultured in IMDM supplemented with 2 mM L-glutamine and 100 U/mL penicillin-streptomycin. T cells were cultured in 5–10% fetal calf serum. Anti-CD3 (αCD3) (2C11) and αCD28 (37.51) were obtained from BD Pharmingen and were used in solution at 0.5–2 and 5 μg/mL, respectively. Anti-TGF-β (1D11.16) was grown in house and was used at 20 μg/mL. Cell proliferation was measured either by incorporation of 3H-thymidine (added to cultures at 48 h, CCPM measured 12 h later) or by CFSE dye dilution at 48 h.

Flow cytometry

Cells were stained with appropriate combinations of the following antibodies: CD4 (RM4-5)-PCP-Cy5.5, CD25 (PC61)-PE, CD44 (IM7)-FITC, galectin 3 (eBioM3/38)-Alexa647, DC-SIGN (5H10)-PE, Foxp3 (FJK-16 s)-PE,-APC, CD205 (NLDC-145)-FITC, CD206 (MR5D3)-FITC, MHC Class II (OX6)-FITC, GITR (108619)-FITC. Appropriate isotype control antibodies were from the same manufacturer as for the specific antibody. Intracellular Foxp3 staining was according to the manufacturer's instructions (anti-mouse/rat Foxp3 staining set; eBioscience). Biotinylated polyclonal goat anti-human LAP and normal goat serum (R&D Systems) were used in conjunction with streptavidin-PE (BD Biosciences). For live cell discrimination, 7-aminoactinomycin D (BD Biosciences) was used. Cells were acquired using a BD FACScalibur or a BD FACScan (BD Biosciences) and analyzed using FlowJo (Tree Star) software.

Cell purification

Naïve T cells (>98% pure) were sorted from splenic cell suspensions on the basis of CD4+CD44lowCD25 using a MoFlo (Beckman Coulter). Whole CD4+ T cells were purified from pooled mesenteric and pancreatic lymph nodes of untreated mice using a CD4 T-cell isolation kit and an Automacs Pro (Miltenyi Biotech), and contained >98% CD4+CD3+ cells.

DC culture

Immature DC were differentiated from BM precursors as follows: BM was flushed from the femurs and tibias of mice with a 25-gauge needle and passed through a 70 um strainer. MΦ-type precursors were removed by adherence at 37°C for 30 min. Cells were cultured for 3 days in 10 ng/mL recombinant mouse GM-CSF (Peprotech) in 10% FBS IMDM. On day 3 all non-adherent cells were removed from culture and the adherent population was cultured for a further 2–3 days with 10 ng/mL GM-CSF 10% FBS IMDM. Cells generated on day 5–6 were 100% CD11b+, ∼90% CD11c+, and low for MHC Class II and co-stimulatory molecules.

Isolation of pancreatic-infiltrating leukocytes

Whole pancreas samples were harvested and processed from individual mice. Briefly, pancreases were torn into pieces in cold PBS containing 5% FBS, 56 mM glucose, and Complete Mini Protease inhibitors (Roche). The tissues were washed twice in cold PBS before incubation in 1 mL of pre-warmed PBS containing 15% FBS, DNAse I (10 μg/mL; Sigma) and Liberase CI (330 μg/mL; Roche). After digestion, tissues were washed and cell suspensions were prepared by forcing through a 70 μm cell strainer. Cells were then centrifuged through a 33% Percoll (GE Healthcare) gradient to obtain leukocyte populations.

RNA isolation and real-time PCR

Total RNA was isolated using an RNeasy kit, converted to cDNA using a Reverse Transcription kit, and quantified in real-time using SYBR green fast PCR (all Qiagen). cDNA was analyzed in duplicate reactions with amplification of target gene and housekeeping gene (hypoxanthine phophoribosyl transferase 1, hprt1) transcripts performed on the same reaction plate on a 7500 fast real-time PCR system (Applied Biosystems). Proprietary QuantiTect Primer Assays for all genes were purchased from Qiagen. Relative gene expression (expression relative to hypoxanthine phosphoribosyl transferase*1000 (Hprt-RE)) was determined according to the algorithm 1000*2ˆ(CTgene1−CThprt1).

Adoptive cell transfers

Cells were isolated from the spleens of 14 wk-old SEA- or SWA-protected mice, or age-matched untreated mice. To establish protection, 4-wk-old female mice were injected i.p. with 50 μg SEA or SWA once a week for 4 wk. There was a 6 wk interval between the last inoculation and the cell transfer. Purified populations of CD25 cells were sorted on a MoFlo excluding cells expressing CD25 (PC61)-PE, CD11b (M1/70)-FITC, CD11c (HL3)-FITC (all BD Pharmingen) or MHC Class II (OX-6)-FITC (AbD Serotec). By this sorting method CD4+Foxp3+ Treg are reduced from approximately 12% of CD4+ T cells to 3%. Cells were transferred i.v. into NOD.scid recipients, either 2×107 splenocytes or 5×106 CD25 cells.

Statistics

Statistical analyses were performed using GraphPad Prism 4 software. Diabetes progression was analyzed by plotting Kaplan–Meier survival curves and performing log-rank test between the groups. One-way ANOVA analyses were carried out to determine the significance of intra-assay results, with Bonferroni multiple comparison test applied to calculate significance values between samples. For inter-assay analysis or data involving multiple mouse samples, the non-parametric unpaired Mann–Whitney t-test was used to calculate p values. This test requires a minimum of five biological replicates per group to reach statistical significance. Tests performed and calculated p values are indicated in the individual figure legends.

Acknowledgements

We gratefully acknowledge Dr. Allan Richards for assistance with collagen induction assays, Sarah Gibbs for technical support, Dr. Zoltan Fehervari and Dr. Jenny Phillips for helpful discussion and Dr. Mark Travis for insight and discussion of TGF-β and integrin β8. This work was funded by the BBSRC. Oliver Burton is funded by a Herchel Smith Fellowship.

Conflict of interest: The authors declare no financial or commercial conflict of interest.

Ancillary