Foxp3+ CD4+ T helper cells called regulatory T (T reg) cells play a key role in controlling reactivity to self-antigens and onset of autoimmunity. T reg cells either arise in thymus and are called natural T reg (nT reg) cells or are generated in the periphery through induction of Foxp3 and are called inducible T reg (iT reg) cells. The relative contributions of iT reg cells and nT reg cells in peripheral tolerance remain unclear as a result of an inability to separate these two subsets of T reg cells. Using a combination of novel TCR transgenic mice with a defined self-antigen specificity and conventional mouse models, we demonstrate that a cell surface molecule, neuropilin-1 (Nrp-1), is expressed at high levels on nT reg cells and can be used to separate nT reg versus iT reg cells in certain physiological settings. In addition, iT reg cells generated through antigen delivery or converted under homeostatic conditions lack Nrp-1 expression. Nrp-1lo iT reg cells show similar suppressive activity to nT reg cells in controlling ongoing autoimmune responses under homeostatic conditions. In contrast, their activity might be compromised in certain lymphopenic settings. Collectively, our data show that Nrp-1 provides an excellent marker to distinguish distinct T reg subsets and will be useful in studying the role of nT reg versus iT reg cells in different disease settings.
Foxp3+ regulatory T (T reg) cells constitute a population of CD4+ T cells that plays a key role in maintaining immune homeostasis and tolerance (Bennett et al., 2001). Foxp3+ T reg cells arise both in the thymus (natural T reg [nT reg] cells) and in the periphery (inducible T reg [iT reg] cells) as a consequence of exposure to antigens (Bluestone and Abbas, 2003). TCR repertoire analysis and intracellular expression of Helios, a member of the ikaros family of transcription factors, suggest that iT reg cells could constitute up to 30% of total T reg cells in the periphery (Hsieh et al., 2004; Pacholczyk et al., 2006; Wong et al., 2007) and may be important in regulating autoimmunity (Feuerer et al., 2010). However, the use of Helios as a marker to distinguish iT reg cells has been controversial as a result of its inconsistent expression on iT reg cells in distinct immune settings (Thornton et al., 2010; Akimova et al., 2011; Gottschalk et al., 2012).
iT reg cells are induced in vivo in several experimental settings including the treatment of mice with minute doses of antigen delivered using osmotic pump or by directing antigen to DEC-205+ dendritic cells (Apostolou and von Boehmer, 2004). Similarly, a study by Mucida et al. (2005) showed that OVA-specific Foxp3+ T reg cells appeared in the gut of mice administered with antigen orally. Importantly, unlike in the in vitro induced cells, in vivo generated iT reg cells exhibited a demethylated Foxp3 T reg cell–specific demethylation region (TSDR; Polansky et al., 2008). Functionally, iT reg cells have been shown to be protective in some studies (Mottet et al., 2003; Huter et al., 2008; Haribhai et al., 2011) although others have reported less efficacy (Hill et al., 2007). This might be a result of different disease models or the use of in vivo versus in vitro generation of iT reg cells. Thus, the exact contribution of iT reg cells in maintaining peripheral tolerance remains unclear.
In this study, we addressed the phenotype and function of nT reg versus iT reg cells using a newly defined marker, neuropilin-1 (Nrp-1), which was found to be selectively expressed on nT reg cells and not on in vivo or in vitro generated iT reg cells. By separating Foxp3+ T reg cells into Npr-1lo iT reg and Nrp-1hi nT reg cells, we were able to analyze their regulatory functions in suppressing autoimmune responses in vivo.
RESULTS AND DISCUSSION
Spontaneous generation of iT reg cells in myelin basic protein (MBP)–TCR-transgenic (Tg) mouse
In the current study, we took advantage of an MBP-specific TCR-Tg mouse developed in our laboratory to study T reg subsets. The MBP-specific TCR-Tg mouse (1B3) was created by cloning the rearranged TCR-α (Vα5) and -β (Vβ14) chains of a CD4+ T cells from infiltrated peripheral nerves of a nonobese diabetic (NOD) mouse deficient for the co-stimulatory molecule B7-2 (NOD-B7-2KO; Louvet et al., 2009; Fig. S1). 1B3 mice bred onto recombination activation gene (RAG)–deficient background developed a rapid and fatal experimental autoimmune encephalomyelitis (EAE). Examination of Foxp3+ T reg cells showed that NOD.1B3.RAG+ mice expressed T reg cells in the thymus but not when mice were bred onto the RAG-deficient background (Fig. 1 a). Interestingly, a significant number of Foxp3+ T reg cells were observed in the periphery of the RAG-deficient 1B3 mice based on both intracellular staining for Foxp3 and on crossing the 1B3.RAG onto Foxp3-GFP background (Fig. 1 a and not depicted). The Foxp3+CD4+ T cells were completely absent at birth in 1B3.RAG−/− mice and began to emerge only in the periphery (spleen and LN) at ∼3 wk, consistent with the development of these cells outside the thymus. The frequency of T reg cells in 1B3.RAG−/− mice in the periphery reached comparable levels to RAG+/− littermate controls (1B3 and WT) in mice that survived to 6 wk (Fig. 1 b and not depicted). Foxp3+ T reg cells from 1B3.RAG−/− expressed similar levels of canonical T reg markers such as Foxp3, GITR, CTLA-4, and CD25 (Fig. 1 a and not depicted). Together, the data suggest that the T reg cells in IB3.RAG−/− develop in the periphery but not in the thymus. The spontaneous generation of iT reg cells was likely a result of the ubiquitous expression of MBP in the periphery that led to the initial induction of iT reg cells (Voskuhl, 1998). Interestingly, MBP is also expressed in the thymus but T reg cells were not generated. This contrasts with earlier studies using HA-specific TCR Tg mouse, where self-peptide expression in thymus induced thymocytes to differentiate into T reg cells (Larkin et al., 2008). However, recent studies of an insulin-specific TCR Tg mouse demonstrated similar findings to ours, reflecting differences in recognition of distinct natural auto-antigens in the thymus most likely based on TCR affinities (Fousteri et al., 2012).
1B3.RAG−/− T reg cells are generated in periphery
To further confirm that T reg cells in the 1B3.RAG−/− mice were indeed iT reg cells, we compared the development of Foxp3+CD4+ T reg cells in the peripheral lymphoid organs in euthymic versus athymic 1B3.RAG−/− mice. As shown in one representative cohort of five mice, the number of peripheral T reg cells appeared similar in the athymic animals as compared with control (sham-operated) littermates (Fig. 1 c). Next, we injected thymocytes from a 2-wk-old 1B3.RAG−/− into thymuses of congenic WT mice and monitored their T reg development to rule out the possibility that interclonal competition restricted the development of T reg cells in the thymus (Bautista et al., 2009; Leung et al., 2009). As shown in Fig. 1 d, <0.5% of CD4+ single positive 1B3.RAG−/− cells became Foxp3+ after 4 d of transfer in this setting. This is significantly less than the percentage previously reported by Bautista et al. (2009) and Leung et al. (2009) where a 100-fold greater frequency of Foxp3+ cells was detected using the TCR-Tg cells cloned from a Foxp3+ cell TCR. These data strongly suggest that the absence of T reg cells in the thymuses of 1B3.RAG−/− mice was not a result of the interclonal competition.
We also ruled out that the development of iT reg cells in the 1B3 model was a consequence of the inflammatory environment in the RAG-deficient 1B3 background. Radiation bone marrow chimeras were generated by mixing 1B3.RAG−/− bone marrow with congenically marked WT bone marrow (1:1). Foxp3+ T reg cells in the thymus were derived solely from the WT bone marrow, whereas 1B3.RAG−/− bone marrow failed to generate any T reg cells in the thymus. However, there was a significant number of 1B3.RAG−/− bone marrow–derived T reg cells in spleen and LN (Fig. 1 e). Most importantly, the mixed chimeras showed no signs of disease or evidence of autoimmunity (unpublished data), consistent with the conclusion that the iT reg cells develop as a consequence of immune homeostasis and not fulminant autoimmunity.
Nrp-1 is differentially expressed on natural and iT reg cells
To define the transcriptional characteristics and distinctly regulated genes between nT reg and iT reg cells, we compared 1B3.RAG−/− iT reg cells with T reg cells isolated from periphery of WT or 1B3 mice. Approximately 700 transcripts were differentially expressed (at least threefold) between all T reg cells and iT reg cells (fold discovery rate cutoff at 0.01). A more systematic analysis of transcripts showed that a subset of transcripts was selectively expressed at reduced levels in iT reg cells (highlighted in box in Fig. 2 a), which included PD-1 (programmed cell death -1, pdcd1), Nrp-1, Helios (Ikzf2), and CD73 (Nt5e; Fig. 2 b and Fig. S2). Interestingly, multiple canonical T reg markers, including Foxp3, CD25, GITR, and CTLA-4, did not differ among the different T cell subsets (Fig. 2 a and not depicted).
Further analyses were performed on Nrp-1 as prior studies have shown reduced Nrp-1 on different iT reg cell populations generated in vivo (Feuerer et al., 2010). Nearly 70% of Foxp3+ T reg cells expressed Nrp-1 in WT NOD or B6 mice correlating with Helios expression (Fig. 2 c; Thornton et al., 2010), suggesting that Nrp-1hi T reg cells were thymically derived, whereas Nrp-1lo T reg cells represented a peripherally derived T reg cell subset. To further confirm that Nrp-1loFoxp3+ T reg cells in conventional mice resembled the iT reg cells seen in 1B3.RAG−/− mice, we examined selected mRNAs differentially expressed between 1B3 T reg cells and 1B3.RAG−/− iT reg cells in the microarray analyses. Several genes, including Eos, Tnfrsf9, and Lypd1, showed detectable mRNA levels in the T reg subsets by real time PCR. All of these transcripts showed elevated expression in Nrp-1hi compared with Nrp-1lo T reg cells (Fig. 2 d). Importantly, the Nrp-1hi and Nrp-1lo T reg cells expressed the same high levels of Foxp3. Although there were both Nrp-1hi– and Nrp-1lo– expressing CD4+CD8−Foxp3+ T reg cells in the thymus, the overwhelming majority of the Nrp-1lo cells were present within the immature CD24hiQa-2lo subset, whereas the vast majority of the mature CD24loQa-2hihi were Nrp-1+ (Fig. 2 e and Fig. S3 a). Although these results cannot rule out that a rare Nrp-1lo population might emigrate from the thymus, the bulk of the data in Figs. 1 and 2 are most consistent with Nrp-1 distinguishing thymically derived nT reg from the peripheral iT reg population.
iT reg cells generated in vitro or in vivo lack Nrp-1 expression
Stimulation of CD4+CD25− conventional T (T conv) cells with either antigen-loaded APCs or anti-CD3 plus anti-CD28 beads in the presence of TGF-β results in the induction of Foxp3+ T reg cells (Chen et al., 2003). Despite expressing high levels of Foxp3, in vitro generated T reg cells expressed little or no Nrp-1 (Fig. 3 a). Several models of antigen-induced conversion of T conv cells into iT reg cells in vivo have been developed by altering antigen dose or route of antigen administration (Apostolou and von Boehmer, 2004). First, we examined Nrp-1 expression on iT reg cells generated in vivo through the homoeostatic conversion of naive T cells. Naive CD4+ T cells were transferred into RAG-deficient mice and monitored for Nrp-1expression as they converted into T reg cells. Foxp3+ T reg cells induced 2 wk after transfer exhibited an Nrp-1lo phenotype (Fig. 3 b).
Next, we examined the expression of Nrp-1 in the setting of prolonged low-dose antigen administration (Apostolou and von Boehmer, 2004) by exposing T reg–deficient, OVA-specific TCR Tg (DO11) RAG−/− mice to a subimmunogenic regimen of OVA peptide (0.1 or 1 µg per day) over a 2-wk time period. As shown in Fig. 3 c, T reg cells from DO11.RAG+/+ exhibited both Nrp-1hi– and Nrp-1lo– expressing T reg cells, whereas iT reg cells generated in OVA-treated DO11.RAG−/− mice were predominantly Nrp-1lo. These results confirmed our findings from 1B3.RAG−/− mouse that iT reg cells are largely deficient in Nrp-1 expression. Finally, we examined Nrp-1 expression on resident T reg cells in the gut-associated lymphoid tissue, which is a preferential site for the peripheral induction of Foxp3+ T reg cells. As shown in Fig. 3 d, analysis of CD4+ T cells from the colonic lamina propria revealed that a majority of Foxp3+ T reg cells in this tissue expressed low levels of Nrp-1, a phenotype distinct from LN T reg cells (Fig. 3 d and Fig. S3 b). Together, the in vitro and in vivo data, combined with the resident gut T reg cell data, strongly support the conclusion that Nrp-1 distinguishes thymically derived nT reg cells from iT reg cells. We suggest that Nrp-1 may be a more reliable marker than Helios under certain conditions, as we and others (Akimova et al., 2011; Gottschalk et al., 2012) have seen that iT reg cells up-regulate Helios expression in some settings (not depicted). Moreover, unlike Helios, Nrp-1 is a cell surface molecule that can be efficiently used to separate inducible versus nT reg cell subsets under homeostatic conditions.
Nrp-1 expression on T reg cells is not significantly altered by activation
Next, we tested whether differential Nrp-1 expression on T reg cells was a consequence of activation status of the cells. T reg cells from 1B3 mice were sorted into Nrp-1hi and Nrp-1lo T reg cells and activated with APCs loaded with 1B3 cognate peptide for 72 h. As shown in Fig. 4 a, both Nrp-1hi and Nrp-1lo cells maintained Nrp-1 expression (either high or low, respectively) on both CD69+ and CD69− subsets. Interestingly, Nrp-1 was up-regulated on some activated T conv cells suggesting a transient expression, unlike nT reg cells where it is stably expressed. Similar results were observed with NOD T reg cells using anti-CD3 and anti-CD28 mAb activation (unpublished data).
To test stability of Nrp-1 expression in vivo, Nrp-1hi or Nrp-1lo T reg cells from NOD.Thy1.1 mice were sorted and transferred into Thy1.2 recipients. Under these homeostatic conditions, both the T reg subsets exhibited a stable phenotype and maintained equivalent levels of Foxp3. Most importantly, levels of Nrp-1 expression, high versus low, remained stable and similar to the levels of the injected cell subsets (Fig. 4 b). A similar trend was seen when Nrp-1hi or Nrp-1lo T reg cells were transferred into lymphopenic hosts although Nrp-1 expression was up-regulated on some Nrp-1lo T reg cells under these conditions (Fig. 4 c). These results suggest that low Nrp-1 expression may not be an immutable feature of iT reg cells and could be up-regulated on a subset of cells in certain proinflammatory environments. Further studies will be required to determine if specific conditions can up-regulate Nrp-1 expression on iT reg cells.
Epigenetic modulation in the Foxp3 locus leading to demethylation of CpG islands in the region of Foxp3 locus (TSDR) is a hallmark of nT reg cells and is thought to reflect stable, constitutive Foxp3 expression in this population. Bisulfite sequence analysis confirmed that 100% of CD4+Foxp3.GFP− T conv cells had methylated CpG sites, whereas >98% of CpG sites were demethylated in Nrp-1hi nT reg cells (Fig. 4 d). Notably, Nrp-1lo iT reg cells also displayed a highly demethylated TSDR but had slightly but significantly lower demethylation than Nrp-1hi T reg cells (87.5 compared with 98% , P < 0.001). This might reflect the newly generated Foxp3+ cells among Nrp-1lo T reg cells, which have not yet fully stabilized the Foxp3 expression. In vivo generated iT reg cells previously have been shown to exhibit demethylated TSDR (Polansky et al., 2008) and methylated TSDR (Haribhai et al., 2011), which highlights differences in the models used and the heterogeneity of this population. By comparison, a study by Miyao et al. (2012) showed that iT reg cells once stabilized in vivo display demethylated TSDR.
iT reg cells can suppress ongoing autoimmune response
Despite the presence of Foxp3+ iT reg cells in the periphery of 1B3.RAG−/− mice, the animals developed EAE by 3–4 wk of age. These results suggested that the peripherally generated T reg cells could not control autoimmunity in this setting. However, iT reg cells from 1B3.RAG−/− exhibited excellent suppressive ability and inhibited proliferation of 1B3 T conv cells (responder T [T resp] cells) equally or better than nT reg cells (from 1B3 mice) at various ratios of T resp/T reg cells tested in vitro (Fig. 5 a). We speculate that iT reg cells in 1B3.RAG−/− mice were being generated in inflammatory and lymphopenic environment, which could compromise their activity. Consistent with this hypothesis, we observed that Nrp-1lo T reg cells from the 1B3 mouse were less effective in controlling EAE when transferred together with naive T cells into RAG−/− mice when compared with Nrp-1hi cells. Not only did Nrp-1lo T reg cells fail to control EAE, the cells down-regulated Foxp3 expression in this lymphopenic setting (Fig. 5, b and c).
However, the Nrp-1lo T reg cells were able to suppress autoimmunity in both a lymphodeficient and a lymphosufficient autoimmune setting when using another TCR Tg mouse, the chromogranin A–specific BDC2.5 model. Previous studies have shown spontaneous diabetes could be prevented by injection of islet antigen-specific BDC2.5 TCR Tg T reg cells into NOD.CD28−/− or BDC2.5 T effector cell reconstituted NOD.RAG−/− mice before the disease onset (Tang et al., 2004). Therefore, we transferred Nrp-1hi or Nrp-1lo BDC2.5 T reg cells into prediabetic NOD.CD28−/− mice and monitored diabetes. Both Nrp-1hi and Nrp-1lo BDC2.5 T reg cells were equally effective in preventing the development of diabetes in CD28−/− mice (Fig. 5 d). Similar preliminary results have been observed in the RAG-deficient adoptive transfer setting (not depicted). Different results in the alternative model systems (1B3 vs. BDC2.5) may reflect differences in TCR affinity or specificity, which may affect T reg cell function or stability. This concept is supported by the reduced stability of the iT reg population seen in this environment (Fig. 5 c; Duarte et al., 2009). Hence, careful analyses will determine whether iT reg cells with distinct TCR specificity function differently in various immune settings. This will be crucial in understanding whether iT reg cells merely complement nT reg cells or play a larger role in maintaining tolerance, especially given data suggesting that blocking of Nrp-1 abrogates suppression of proliferation of responder T cells by T reg cells (Sarris et al., 2008).
In summary, the extensive studies described in this paper demonstrate that the cell surface marker Nrp-1 distinguishes nT reg from iT reg cells generated in several different immune settings including low-dose antigen-induced iT reg cells, T reg cells in intestinal lamina propria, iT reg cells generated spontaneously in 1B3.RAG−/− mice or through conversion in vivo, and T reg cells generated in vitro using TGF-β. Moreover, our thymus data showing that mature Foxp3+ T reg cells express the highest levels of Nrp-1 strongly suggest that the overwhelming majority of thymically derived T reg cells express Nrp-1. Although it remains possible that a small subset of Nrp-1lo T reg cells leave the thymus, T reg cells generated in the periphery are uniformly Nrp-1lo, whereas Nrp-1hi T reg cells are always thymically derived. This ability to separate different subtypes of T reg cells will be useful in studying their role and contribution in autoimmune disease. Finally, although human T reg cells in peripheral blood do not express Nrp-1, T reg cells in secondary lymphoid organs have been shown to express Nrp-1 (Battaglia et al., 2008; Milpied et al., 2009). A similar separation of T reg cells into natural and iT reg subsets in human would be useful in the study of human T reg cells and their use in clinical cell-based therapies.
MATERIALS AND METHODS
NOD.POT (P0-specific TCR-Tg mouse), NOD.Foxp3-GFP-Cre reporter mice have been previously described (Louvet et al., 2009; Zhou et al., 2009). C57BL/6, NOD, NOD-B7-2−/−, NOD-TCRα−/−, and NOD-RAG2−/− mice were bred in our facility. All mice were housed in a specific pathogen-free University of California, San Francisco (UCSF) facility. The Institutional Animal Care and Use Committee of UCSF approved all animal experiments.
For analysis of surface markers, cells were stained in FACS buffer (PBS containing 2% wt/vol BSA) and the appropriate antibodies. Unless stated, all antibodies for cell surface markers and the anti-Helios mAb were purchased from BioLegend. Anti–Nrp-1 Ab was purchased from R&D Systems. Anti–PD-1 mAb and anti-Foxp3 mAb was purchased from eBioscience. Foxp3 staining was done according to the manufacturer’s instructions (eBioscience). For cytokine analysis, cells were incubated for 3–4 h at 37°C with 0.5 M ionomycin, 10 ng/ml of PMA, and 3 M monensin before being fixed and made permeable (Foxp3 Staining kit; eBioscience) and were stained for intracellular proteins. Stained cells were analyzed with FACSCalibur or an LSR II (BD) and data were analyzed using FlowJo software (Tree Star). T cells were sorted with a MoFlo high-speed cell sorter (Dako) or FACSAria (BD).
Generation of 1B3 TCR-Tg mice.
Cells from sciatic nerves of neuropathic NOD-B7-2−/− mice were prepared as previously described (Louvet et al., 2009). In brief, CD4+CD8− cells were sorted and expanded with anti-CD3– and anti-CD28–coated beads supplemented with 2,000 IU/ml rhIL-2, as previously described (Tang et al., 2004). The expanded CD4+ T cells were fused to the TCR-negative thymoma partner BW1100 (provided by E. Palmer, University Hospital Basel, Basel, Switzerland) using standard protocols. TCR-expressing hybridoma clones were tested for antigen specificity by incubating 50,000 cells with 0.2 × 106 APCs (NOD-TCR-αKO spleen cells, irradiated NOD spleen cells, or NOD BM-derived DCs) with different nerve lysates for 16–24 h, followed by staining with CD69. From the 1B3 hybridoma cDNA, the full-length coding sequences for the TCR-α and -β chains were cloned by PCR and were subcloned into a CD2 and CD4 expression vector, respectively, allowing expression of the transgenes in both CD4 and CD8 T cells (Wang et al., 2001). Tg mice were generated by microinjection of CD2–TCR-α and CD4–TCR-β constructs into NOD embryos as described earlier (Louvet et al., 2009). Vectors were provided by N. Killeen (UCSF, San Francisco, CA). Overlapping MBP peptides spanning the sequence of mouse MBP were synthesized by Genemed or were synthesized by J. Wouter and C.J.M. Melief (Leiden University, Leiden, Netherlands).
Sorted T reg cells (CD4+CD25+GITRhi with >95% purity) from WT, 1B3, or 1B3.RAG−/− mice were used for isolation of Total RNA. Total RNA from each paw was amplified for one round using the Message Amp II amplified RNA (aRNA) Amplification kit (Ambion), labeled using a modified protocol with the ULS aRNA Fluorescent Cy5 labeling kit (Kreatech Diagnostics), and purified using RNA Clean and Concentrate spin columns according to manufacturer’s protocol (Zymo Research Corporation). aRNA was hybridized to 4 × 44K Mouse Whole Genome Microarrays (Agilent Technologies) for 18 h according to the manufacturer’s protocol. Microarrays were washed in 0.01× SSC/0.005% Triton X-102 at 40°C, dried with HEPA filtered compressed nitrogen and scanned on a DNA Microarray Scanner (Agilent Technologies) at a 5-µm resolution. Data were extracted from the scanned image using Feature Extraction Software (Agilent Technologies) and were further processed for background subtracted and signal normalization.
Thymectomy and intrathymic injections.
The thymus was removed from 2–3- or 10–14-d-old mice by suction under anesthesia. The wound was glued and the mice were kept at 30°C overnight and then returned to their mothers. Thymectomy was confirmed by visual examination at autopsy and only mice with complete thymectomy were included in the experiments. For intrathymic injections mice were anesthetized with acepromazine, ketamine, and xylazine, the sternum was cut open to access the thymus. 1 × 107 total thymocytes were injected in 10-µl suspension into the thymus of 4–6-wk-old WT NOD.Thy1.1 mice. Animals were placed on a heat pad until they regained consciousness. After 4 d, mice were sacrificed and single cell suspensions were prepared from whole thymuses and stained for surface markers and intracellularly for Foxp3 and analyzed by flow cytometry.
Mixed bone marrow chimera.
1B3.RAG−/− or NOD bone marrow was mixed with NOD.Thy1.1 bone marrow at a ratio of 1:1 (1 × 106) and co-injected into sublethally irradiated RAG−/− recipients (400 rads) to generated mixed chimera. After 6 wk, the thymus, peripheral LN and spleen were analyzed by flow cytometry for expression of Nrp-1, PD-1, Helios, and Foxp3 on CD4+ T cells in Thy1.1 or Thy1.2+ fractions.
T reg cells in prevention of diabetes in vivo in NOD.CD28−/− mice.
Nrp-1hi or Nrp-1lo T reg cells were sorted from NOD.BDC2.5.Foxp3-GFP mouse. 5 × 104 sorted T reg cells were adoptively transferred in 5–6-wk-old prediabetic male NOD.CD28−/− mice. Nonfasting blood glucose levels in recipient mice were monitored every week using a One-touch Ultra glucometer (Lifescan Inc.). Mice with two consecutive blood glucose readings over 250 mg/dl were considered diabetic.
Foxp3+ iT reg cell in vitro activation.
For in vitro activation, T cells from 1B3.Foxp3-GFP mice were sorted into T conv (CD4+Foxp3−) or T reg cells (separated into Nrp-1lo, Nrp-1hi, and total) and activated in vitro with irradiated splenocytes loaded with 1B3 peptide (SRPGLCHMYKDS) or with anti-CD3 and anti-CD28 beads for 72 h. At the end of incubation, cells were stained with CD4 and Foxp3 or with CD69 and Nrp-1.
Foxp3+ iT reg cell conversion.
Freshly sorted T conv cells (CD4+CD25−Foxp3−) from spleen and LN of 1B3.Foxp3-GFP mouse were cultured for 5 d with anti-CD3 and anti-CD28 microbeads or irradiated splenic cells loaded with 1B3 cognate (SRPGLCHMYKDS) peptide and IL-2 in the presence or absence of TFG-β. After incubation T cells were then collected and live cells were stained for intracellular Foxp3 and surface markers. For in vivo conversion, 106 naive T cells (CD4+CD25−CD62L+GITR−) were adoptively transferred into RAG−/− mice. 2 wk later, CD4+ T cells from spleen and LN were stained for Foxp3 and Nrp-1.
Induction of iT reg cells in DO11.RAG−/− mice.
Osmotic pump (ALZET1002; Durect Corporation) infusing either 0.1 µg or 1 µg OVA peptide (323–339) or PBS per day was implanted s.c. in DO11.RAG−/− mice as described previously (Apostolou and von Boehmer, 2004). After 14 d, inguinal LN and spleen were harvested and CD4+ cells were surface stained for Nrp-1 and intracellularly for Foxp3. 3 DO11.RAG−/− mice per group were used and unimmunized DO11.RAG+/+ was used as control.
Isolation of lamina propria cells.
Colons were harvested from 8–10-wk-old NOD mice sliced and digested by 0.07 U/ml Liberase Blendzyme III and 0.5 mg/ml DNase 1. LN cells were also subjected to similar treatment. The cells were resuspended in 40% Percoll and carefully overlaid onto 70% Percoll. The interface containing the lamina propria leukocytes was collected. Harvested cells were surface stained for CD4, Nrp-1, and intracellularly for Foxp3.
TSDR methylation assay.
A quantitative real-time PCR method was developed to analyze the methylation status of key CpG dinucleotides in the TSDR of the FOXP3 gene where demethylation has been shown to correlate with stable T reg cells (Floess et al., 2007). T conv, Nrp-1lo, and Nrp-1hi (2 × 105 each) were sorted from Foxp3.GFP mouse and processed using the EZ DNA Methylation-Direct kit (Zymo Research) according to the manufacturer’s protocol. Purified bisulfite-treated DNA was used in a quantitative PCR reaction. Primers and probes were designed using Primer Express software (Applied Biosystems) with TSDR DNA sequence that had been bisulfite converted in silico using Methyl primer Express software (Applied Biosystems). Sequences used were the following: forward primer, 5′-GGTTTATATTTGGGTTTTGTTGTTATAATTT-3′; and reverse primer, 5′-CCCCTTCTCTTCCTCCTTATTACC-3′. Probe sequences were: methylated (CG) probe, 5′-TGACGTTATGGCGGTCG-3′; and unmethylated (TG) probe, 5′-ATTGATGTTATGGTGGTTGGA-3′. PCR was performed with 10 µl Universal Master Mix II (Applied Biosystems), 1 µl eluted DNA, primer/probe mix, and enough water to bring total volume to 20 µl. Final concentration of primers was 900 nM and concentration of probes was 150 nM. Reactions were run for 10 min at 95°C for 10 min and 50 cycles of 95°C for 15 s and 61°C for 1.5 min (7500 Real-Time PCR System; Applied Biosystems). Percent demethylation was calculated using the formula percent demethylation = 100/[1 + 2(CtTG−CtCG)], where CtTG represents the threshold cycle of the TG (unmethylated) probe and CtCG represents the threshold cycle of the CG (methylated) probe (Cottrell et al., 2007).
Statistical analysis was done using Prism software (GraphPad Software). Statistical analysis was performed using a two-tailed Student’s t test.
Online supplemental material.
Fig. S1 depicts the generation and characterization of 1B3 TCR Tg mouse. Fig. S2 shows the heat map of selected genes that were differentially regulated between 1B3.RAG−/− and NOD and 1B3 T reg cells. Fig. S3 shows the gating strategy for Fig. 2 e and Fig. 3 d. Table S1 shows the full list of genes differentially regulated between 1B3.RAG−/− and NOD and 1B3 T reg cells with a fold discovery rate of ≤0.01.
We thank M. Lee, N. Grewal, and S. Jiang, for technical assistance; J. J. Lozano for helping with microarray analysis; D. Fuentes for animal husbandry; and A. Abbas, Q. Tang, H. Bour-Jordan, M. Anderson, and members of the Bluestone laboratory for discussions.
This work was supported by the US National Institutes of Health (R01 AI50834 and P30DK63720) and a grant from Pfizer.
The authors declare no competing financial interests.
M. Yadav designed and did experiments, analyzed data, and wrote the manuscript; C. Louvet designed and did experiments; J.M. Gardner and D. Davini did experiments; R. Head and D. Weiss did Microarray design, data generation, analysis, and interpretation; M. Martinez-Llordella, D.J. Kuster, and D. Von Schack did microarray analysis and interpretation; S. Bailey-Bucktrout helped with data analysis; F.M. Sverdrup and B.A. Anthony performed the TSDR analyses; and J.A. Bluestone designed experiments, supervised the work, analyzed data, and wrote the manuscript.
experimental autoimmune encephalomyelitis
- iT reg cell
inducible T reg cell
myelin basic protein
- nT reg cell
natural T reg cell
recombination activation gene
- T conv cell
conventional T cell
- T reg cell
regulatory T cell
T reg cell–specific demethylation region
C. Louvet’s present address is INSERM U1064, Université de Nantes, Nantes 44093, France.