Id2 and Id3 maintain the regulatory T cell pool to suppress inflammatory disease

Abstract

Regulatory T (T_reg) cells suppress the development of inflammatory disease, but our knowledge of transcriptional regulators that control this function remains incomplete. Here we show that expression of Id2 and Id3 in T_reg cells was required to suppress development of fatal inflammatory disease. We found that T cell antigen receptor (TCR)-driven signaling initially decreased the abundance of Id3, which led to the activation of a follicular regulatory T (T_FR) cell–specific transcription signature. However, sustained lower abundance of Id2 and Id3 interfered with proper development of T_FR cells. Depletion of Id2 and Id3 expression in T_reg cells resulted in compromised maintenance and localization of the T_reg cell population. Thus, Id2 and Id3 enforce T_FR cell checkpoints and control the maintenance and homing of T_reg cells.

Homeostasis of the immune system requires careful control mechanisms at mucosal barriers, sites exposed to abundant foreign antigens. Immune system cells must provide protection against a broad range of invading pathogens but also ensure tolerance to self antigens and innocuous non-self antigens^1–3. Failure of the immune system to enforce tolerance readily leads to the development of autoimmune disease and allergies, including asthma and atopic dermatitis. Allergy is characterized by the expression of T_H2 cell cytokines, high concentrations of serum immunoglobulin E (IgE) and eosinophilia^4,5. T_reg cells are prominent among the cell types that suppress spontaneous inflammation and are characterized by expression of the transcription factor Foxp3 (refs. 6–11). Absence of Foxp3 in mice and FOXP3 in humans rapidly results in the development of multiorgan autoimmunity, inflammatory bowel disease and allergy. T_reg cells develop in the thymus (tT_reg cells) as well as in the peripheral organs (pT_reg cells)^1–3. pT_reg cells act primarily to control the development of mucosal inflammation¹². T_reg cells are also essential in regulation of humoral immunity; loss of T_reg cells leads to elevated concentrations of autoantibodies, hyper-IgE syndrome, increased numbers of follicular helper T (T_FH) cells and spontaneous development of germinal centers (GCs)¹³. Recent studies have identified a subset of T_reg cells named T_FR cells that control GC reactions, characterized by the expression of Cxcr5, Bcl6, Pdcd1 and Prdm1 (refs. 14–16).

Members of the helix-loop-helix (HLH) family regulate many developmental trajectories in the thymus¹⁷. These include E proteins as well as Id proteins. E proteins function as transcriptional activators or repressors with the ability to bind specific DNA sequences termed E-box sites. Four E proteins have been identified and characterized: E12, E47, HEB and E2-2. E12 and E47 are encoded by the Tcf3 locus and are generated by differential splicing¹⁸. HEB and E2-2 are related to the Tcf3 gene products but diverge substantially in the N-terminal transactivation domains. DNA-binding activity of E proteins is regulated by the Id proteins^19,20. Four Id proteins named Id1, Id2, Id3 and Id4 contain an HLH dimerization domain but lack the basic DNA-binding region. Interactions between Id proteins and E proteins suppress DNA-binding activity of E proteins. Id2 and Id3 are particularly important in modulating the developmental progression of T lineage cells^21–26.

Here we found that depletion of Id2 and Id3 expression in T_reg cells resulted in the early onset of fatal T_H2 cell–mediated inflammatory disease. We found that upon TCR-mediated signaling in T_reg cells, expression of Id2 and Id3 declined, leading to higher binding activity of E proteins and induction of a T_FR cell–specific program of gene expression, including Cxcr5 and Il10. Loss of Id2 and Id3 in T_reg cells resulted in compromised T_reg cell homeostasis, increased susceptibility to cell death upon stimulation and aberrant tissue localization. Taken together, we propose that Id2 and Id3 maintain the T_reg cell pool and act as gatekeepers to enforce multiple checkpoints during T_FR cell differentiation.

RESULTS

Id2 and Id3 expression in T_reg cells

As a first approach to explore potential roles for Id2 and Id3 in T_reg cells, we analyzed their expression patterns using Id2-YFP and Id3-GFP reporter mice^26,27. We found that the thymic T_reg cell population can be segregated into Id2⁺Id3⁺ as well as Id2⁻Id3⁺ compartments (Fig. 1a). In maturing T_reg cells, which first express CD69 and then upregulate CD62L, Id2 expression declined, leading to an increase of the Id2⁻Id3⁺ compartment (Fig. 1a). In the peripheral lymphoid organs, the majority of T_reg cells consisted of Id2⁻Id3⁺ cells (Fig. 1a). To examine the dynamics of Id2 and Id3 expression upon stimulation, sorted T_reg cells carrying the Id2-YFP and Id3-GFP reporters were activated in vitro by exposure to anti-CD3e and anti-CD28 in the presence of non–T_reg cells as well as antigen-presenting cells (APCs) (Fig. 1b). The most pronounced change occurred in Id3 expression, which declined substantially upon exposure to TCR-mediated signaling (Fig. 1b). Thus, the majority of T_reg cells isolated from peripheral organs expressed abundant Id3 but lacked Id2, but upon in vitro stimulation, Id3 expression declined in a fraction of cells, leading to Id2^loId3^lo and Id2^intId3^lo T_reg cell populations.

Figure 1.

Ablation of Id2 and Id3 expression in T_reg cells leads to the early onset of fatal inflammatory disease. (a) Flow cytometric analysis of CD69 versus CD62L expression gated on the CD4⁺CD25⁺ T_reg cell population derived from the thymus (CD4⁺CD25⁺CD8⁻TCRβ^hi (left). CD4SP, CD4⁺CD8⁻. GFP versus YFP expression, gated on CD69⁺CD62L⁻ or CD69⁻CD62L⁺ T_reg cells derived from the thymus (middle). GFP versus YFP expression in CD4⁺CD25⁺ T_reg cells isolated from the spleens (Spl) of control and Id2^YFP/+Id3^GFP/+ mice (right). Numbers in quadrants indicate percent cells in each compartment. (b) Flow cytometric analysis of Foxp3 expression on sorted CD4⁺CD25⁺ and CD4⁺CD25⁻ T cells (left; number indicates the percentage of Foxp3⁺ cells gated on the CD4⁺CD25⁺ T cells); sorted CD4⁺CD25⁺ T_reg cells from Id2^YFP/+Id3^GFP/+ mouse (CD45.2) were cocultured with CD4⁺CD25⁻ T cells (CD45.1) and stimulated with anti-CD3ε plus anti-CD28 in the presence of T cell–depleted splenocytes (APCs). Flow cytometric analysis of GFP versus YFP expression gated on CD45.2⁺CD45.1⁻CD4⁺TCRβ⁺CD25⁺ cells for the indicated time points (bottom). Percentage of Id3-GFP⁺Id2-YFP⁻ cells and Id3-GFPId2-YFP⁻cells (middle). (c) Quantitative real-time PCR analysis of Id2 and Id3 transcript levels in sorted CD4⁺TCRβ⁺CD25⁺YFP⁺ T_reg cells derived from the lymph nodes of Id2^+/+Id3^+/+Foxp3^Cre (Ctrl), Id2^fl/flFoxp3^Cre or Id3^fl/flFoxp3^Cre mice. (d) Survival plot of Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/−Foxp3^Cre/CreId2^fl/flFoxp3^Cre/CreId3^fl/flFoxp3^Cre/CreId2^fl/flId3^fl/flFoxp3^Cre/+ and littermate control (Ctrl) mice, monitored during a 25-week period. Data are representative of two experiments (a,b; error bars in b, s.d.; n = 3 technical replicates), one experiment (c; error bars, s.d.; n = 3 technical replicates) and one experiment (d; n = 11 independent biological replicates per group).

Id2 and Id3 expression suppresses fatal inflammation

To evaluate the roles of Id2 and Id3 in T_reg cell function, we crossed Id2^loxP/loxP (Id2^fl/fl) and Id3^fl/fl mice with Foxp3^Cre mice^23,28,29. The Foxp3^Cre mice used throughout this study carried an internal ribosomal entry site followed by a sequence encoding a Cre-YFP functional fusion protein located downstream of the Foxp3 termination codon²⁹. The Foxp3 locus is X chromosome–linked, and consequently gender-based differences in excision of the loxP-flanked alleles impact the population of T_reg cells present in these mice. Here will refer to male mice carrying the Foxp3-Cre-YFP allele as Foxp3^Cre mice and to female mice as Foxp3^Cre/Cre or Foxp3^Cre/+ mice. Mice deficient for Id2 or Id3 alone in T_reg cells did not show evidence of inflammatory disease (Supplementary Fig. 1 and data not shown). To examine whether Id2 and Id3 compensate for each other, we sorted T_reg cells from Id2^fl/flFoxp3^Cre mice as well as Id3^fl/flFoxp3^Cre mice. We isolated RNA from the sorted populations and examined it by real-time PCR for Id2 and Id3 expression. In the absence of Id2 expression, the abundance of Id3 increased, whereas in T_reg cells deficient for Id3 expression, abundance of Id2 increased, as compared to controls (Fig. 1c).

Next, we generated Id2^fl/flId3^fl/flFoxp3^Cre as well as Id2^fl/flId3^fl/fl Foxp3^Cre/Cre mice. Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice were born with Mendelian frequencies and were healthy up to 4 weeks of age. However, 6–8-week-old mice with no sex bias developed dermatitis with swelling and fur loss across the eyelids, and the majority of Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice died within ten weeks (Fig. 1d and Supplementary Fig. 2a). Neither Id2^fl/+Id3^fl/flFoxp3^Cre nor Id2^fl/flId3^fl/+Foxp3^Cre mice displayed early morbidity (data not shown). Approximately one-third of Id2^{fl/ fl}Id3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice showed intense dermatitis with scaling and fur loss throughout the body but not in the skin of the tail (data not shown). Similar to the case with human atopic dermatitis, Id2^fl/flId3^fl/flFoxp3^Cre mice displayed dermatitis as evidenced by compulsive scratching of the face and back (Supplementary Video 1).

Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice also showed splenomegaly and lymphadenopathy of subcutaneous lymph nodes accompanied by increasing cell numbers, whereas the size of mesenteric lymph nodes (MLNs) was comparable to that of littermate control mice (Supplementary Fig. 2a,b). As the Foxp3 locus is located on the X chromosome, half of the T_reg cells in heterozygous female mice that carry one copy of Foxp3^Cre express normal amounts of Id2 and Id3 (ref. 30). We found that heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice lacked dermatitis across the eyelids, lacked detectable splenomegaly and lymphadenopathy, and had normal percentages of naive CD4⁺ and CD8⁺ T cells (Supplementary Fig. 2c,d and data not shown). Histological analysis of tissue samples from Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice revealed leukocytic infiltrates in the lung, esophagus and eyelids, with an associated reactive hyperplasia of the adjacent mucosal epithelial cells (Fig. 2a). The inflammatory infiltrates in the eyelid were localized to the subepithelial area and there was no inflammation in the lachrymal gland (Supplementary Fig. 3a). The inflammatory infiltrates in the lungs were peribronchial and perivascular, accompanied by production of mucus in the bronchial mucosal goblet cells (Fig. 2b). Many of the peribronchial and perivascular infiltrates of inflammatory cells in the lungs of Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice displayed expression of major basic protein (MBP), indicative of eosinophils, accompanied by infiltrating B cells and T cells (Fig. 2b and Supplementary Fig. 3b). In a similar fashion, inflammatory infiltrates across subepithelial regions of eyelids and the esophagus of Id2^fl/flId3^fl/flFoxp3^Cre as well as Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice were composed primarily of eosinophils (Fig. 2c). We also found considerable eosinophilic infiltration in the hilar lymph nodes as well in the pancreas but not across the pancreatic islands (Supplementary Fig. 3c,d). We did not observe esophagitis in Foxp3 mutant mice, which indicated that pathologies associated with de pletion of Id2 and Id3 expression differ from that observed in mice lacking Foxp3 (Supplementary Fig. 3e). Notably, the larger bronchioles were filled with denuded epithelial cells and cell debris in dying Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice (Supplementary Fig. 3a).

Figure 2.

Spontaneous inflammation in the lung, eyelid, skin and esophagus in mice depleted for the expression of Id2 and Id3 in T_reg cells. (a) Hematoxylin and eosin (H&E) staining of the lung, esophagus and eyelid derived from control (Ctrl; female Id2^fl/flId3^fl/flFoxp3^Cre/+) and Id2^fl/flId3^fl/flFoxp3^Cre mice. Original magnification: ×200 (lung), ×200 (esophagus) and ×50 (eyelid). (b) Periodic acid Schiff (PAS) staining and MBP immunostaining of lungs of indicated genotypes (top; arrows indicate muous-producing goblet cell (PAS staining) and MBP-positive eosinophil). Original magnification: ×400 (PAS) and ×200 (MBP). Quantification of PAS-stained epithelial cells per bronchus or MBP-positive eosinophils in peribronchial area (bottom). (c) MBP staining of eyelid and esophagus. Original magnification: ×200 (eyelid) and ×200 (esophagus). Data are representative of six experiments (a), two experiments (b; mean ± s.d. (n = 4 biological replicates) and two experiments (c). *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

In a minor fraction of mice (<20%), other organs such as liver, kidney, stomach, pancreas and colon displayed low levels of inflammation (data not shown). Notably, three of five Id2^fl/flId3^fl/flFoxp3^Cre mice that survived beyond 12 weeks of age showed evidence of focal pneumonia, characterized by areas in the lung that were filled with eosinophilic foamy macrophages, and had scarring and increased deposition of collagen surrounding the bronchioles (Supplementary Fig. 4). We also observed mild to moderate colitis with an increase in numbers of mucosal MBP-expressing eosinophils (Supplementary Fig. 4).

Consistent with the pathology in the lung, we found an increase in the total number of eosinophils and lymphocytes in the bronchial alveolar lavage (BAL) fluid and markedly elevated numbers of eosinophils, neutrophils, T cells and B cells in the lung derived from Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice (Fig. 3a,b). The increase in eosinophil cellularity in the lung was closely associated with elevated numbers of type 2 innate lymphoid cells (ILC2) (Fig. 3b)^31–33. In addition, we observed substantially increased numbers of eosinophils, basophils and neutrophils in the lymph nodes and spleen derived from Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice as compared to littermate control mice (Fig. 3c). These observations indicate that the absence of Id2 and Id3 expression in T_reg cells leads to spontaneous T_H2 cell–mediated fatal airway inflammation in the lungs as well as inflammation of the eyelids, the skin and the esophagus.

Figure 3.

T_H2 cell–mediated inflammation in the lungs of mice depleted for Id2 and Id3 expression in T_reg cells. (a) May-Giemsa staining images of cells (left) and cell numbers (right) in BAL fluid derived from 6–8-week-old control (Ctrl) or Id2^fl/flId3^fl/flFoxp3^Cre mice (left). Green, red and blue arrowheads indicate macrophages (MF), lymphocytes (Lym) and eosinophils (Eos), respectively; Neu, neutrophil. Insets show magnification of boxed areas. Original magnification, ×40 (×120 in insets). (b) Numbers of hematopoietic cells (CD45⁺), macrophages (MF), eosinophils (Eos), basophils (Bas), neutrophils (Neu), mast cells (Mas), T cells (T) and B cells (B) isolated from the left lobes of the lungs derived from Id2^fl/flId3^fl/flFoxp3^Cre mice (left). Numbers of type 2 innate lymphoid cells (ILC2; CD45⁺LineageCD25⁺IL7R⁺Sca1^hiThy1.2^hi) isolated from the left lobes of the lungs derived from Id2^fl/flId3^fl/flFoxp3^Cre mice (right). (c) Flow cytometric analysis of CD19 versus TCRβ expression of cells derived from the LNs isolated from 6–8-week-old control (Id2^fl/flId3^fl/flFoxp3^Cre/+) or Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice. Adjacent panels indicate CD11c versus SiglecF, Fcer1 versus SiglecF and Mac-1 versus Gr-1 expression, gated on the TCRβ⁻CD19⁻ compartment derived from 6–8-week-old control (Id2^fl/flId3^fl/flFoxp3^Cre/+) or Id2^fl/flId3^fl/flFoxp3^Cre/Cre lymphocytes isolated from subcutaneous LNs. Numbers in plots indicate percent CD19⁻TCRβ⁻, CD11c⁻SiglecF⁺, Fcer1⁺SiglecF⁻ and Mac-1⁺Gr-1⁺ cells. Right, percentages of eosinophils (SiglecF⁺CD11c⁻), basophils (Baso; Fcer1^hiSiglecF⁻) and neutrophils (Neut; Mac-1^hiGr-1^hi) in subcutaneous LN cells and splenocytes. Data are representative of four experiments (a,c; mean ± s.d.; in a, n = 8 independent biological replicates; in b, control n = 5, Id2^fl/flId3^fl/flFoxp3^Cren = 4 independent biological replicates; in c, control n = 6, Id2^fl/flId3^fl/flFoxp3^Cren = 9 independent biological replicates). *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

Id2 and Id3 suppress T_H2-mediated inflammation

To determine how depletion of Id2 and Id3 expression in T_reg cells leads to inflammatory disease, we examined the lymphoid organs. We noticed increased numbers of T cells and B cells in the spleen and subcutaneous LNs, in which we observed lower percentages of CD62L^hiCD44^lo naive CD4⁺ and CD8⁺ T cells as well as a significant increase in the percentages of T_FH cells, GC B cells and IgG1 class–switched B cells (Fig. 4a,b and data not shown). We observed a substantial increase in T_H2 cell–mediated production of cytokines, such as interleukin 4 (IL-4), IL-5 and IL-13 in the lymph nodes and spleen derived from Id2^fl/flId3^fl/flFoxp3^Cre mice (Fig. 4c). Consistent with the markedly increased numbers of T_FH cells and GC B cells, and the augmented expression of T_H2 cell cytokines, 6–8-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice exhibited a spontaneous increase in IgM, IgG1 and IgE serum concentrations but not in IgG2b, IgG3 and IgA (Fig. 4d). Titers of anti-nuclear antibodies in serum isolated from Id2^fl/flId3^fl/flFoxp3^Cre mice, however, were comparable to those in the littermate control (data not shown). These data indicate that expression of Id2 and Id3 in T_reg cells is required to suppress spontaneous GC development and T_H2 cell–mediated inflammatory disease.

Figure 4.

Spontaneous germinal center formation and T_H2 cell-mediated inflammation in mice depleted for the expression of Id2 and Id3 in T_reg cells. (a) αβ T cell (αβT) and B cell numbers for spleen (Spl) and subcutaneous LN (sLN) isolated from 8-week-old littermate control and Id2^fl/flId3^fl/flFoxp3^Cre mice. (b) Flow cytometric analysis of splenocytes derived from 8-week-old littermate control (Id2^fl/+Id3^fl/+Foxp3^Cre) and Id2^fl/flId3^fl/flFoxp3^Cre mice, showing expression of CD44 versus CD62L and CXCR5 versus PD-1, gated on the CD4⁺TCR[3⁺YFP⁻CD25⁻ population (left). Frequency of germinal center B cells in 8-week-old littermate control (Id2^fl/+Id3^fl/+Foxp3^Cre) and Id2^fl/flId3^fl/flFoxp3^Cre mice (middle). Numbers in quadrants and gates indicate percent cells in each compartment. Frequencies of T_FH (CXCR5^hiPD-1^hi) cells in the CD4⁺ T cell compartment and GC B (Fas^hiGL7^hi) cells (GCB) in the B cell compartment (right). (c) Cytokine expression in CD4⁺ T cell derived from littermate control (Id2^fl/flId3^fl/flFoxp3^Cre/+) and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice (left) Percentages of CD4⁺ T cells producing IL-4, L-5, IL-13, IFN-γ and IL-17 in 7–9-week- old mice for the indicated genotypes (right). (d) Enzyme-linked immunosorbent assay (ELISA) of IgM, IgG1, IgG2a, IgG2b, IgG3, IgA and IgE in serum derived from 6–9-week-old mice for the indicated genotypes. Data are representative of four experiments (a; mean ± s.d.; n = 8 biological replicates), | three independent experiments (b; mean ± s.d.; control n = 4, Id2^fl/flId3^fl/flFoxp3^Cre n = 5 biological replicates), two experiments (c; mean ± s.d.; n = 4 biological replicates), one experiment (d; mean ± s.d.; control n = 8, Id2^fl/flId3^fl/flFoxp3^Cren = 7 biological replicates) *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

Id2 and Id3 modulate CXCR5 and Foxp3 expression

The T_reg population can be segregated into distinct functional subsets, including effector T_reg (eT_reg) cells³⁴. Previously, low Id3 expression has been observed in the Blimp-1⁺ eT_reg cell population³⁵. Consistent with these observations, we found that a fraction of ICOS⁺ eT_reg cells displayed low Id3 expression and Id2^loId3^lo T_reg cells expressed more CXCR5 (Fig. 5a). In line with this, T_reg cells that lacked Id2 and Id3 expression exhibited increased expression of CXCR5, PD-1, ICOS, CXCR3, CXCR4 and CTLA-4 but decreased CD62L expression as compared to Id-sufficient T_reg cells (Fig. 5b and Supplementary Fig. 5a–c)³⁶. T_reg cells derived from healthy heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice also displayed increased expression of CXCR5 and PD-1, but expression of CXCR3, CXCR4, CD62L and CTLA-4 was comparable to that in wild-type T_reg cells (Fig. 5c,d and Supplementary Fig. 5a–c). Id2^fl/flId3^fl/flFoxp3^Cre mice displayed significantly reduced expression of Foxp3 protein (Fig. 5e). We observed this downregulation of Foxp3 even in heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice, although abundance of Foxp3 transcript in T_reg cells isolated from Id2^fl/flId3^fl/flFoxp3^Cre/+ mice did not differ from that for Id2^+/+Id3^+/+Foxp3^Cre/+ mice (Supplementary Fig. 5d,e). These data indicate that loss of Id2 and Id3 expression in T_reg cells leads to decreased abundance of Foxp3 and increased proportions of CXCR5⁺PD-1⁺ T_reg cells.

Figure 5.

Depletion of Id2 and Id3 expression in T_reg cells modulates the expression of CXCR5, Foxp3 and Nrp1. (a) Flow cytometric analysis of Id3-GFP versus ICOS, gated on the CD4⁺CD25⁺ T_reg cell compartment derived from from Id3^gfp/+ splenocytes (left). Numbers in quadrants indicate percent cells in each compartment. Expression of CXCR5 in the T_reg compartments as characterized by the expression of Id2 and Id3 (right). (b) Flow cytometric analysis of T_reg cells (Foxp3-YFP⁺CD25⁺CD4) derived from LNs isolated from 8-week-old control (Id2^fl/+Id3^fl/+Foxp3^Cre) and Id2^fl/flId3^fl/flFoxp3^Cre mice for indicated cell surface markers. Numbers in plots indicate percent Foxp3-YFP⁺CD25⁺, CXCR5⁺PD-1⁺ and CXCR5⁺PD-1⁻ cells. Spl, spleen. (c) CXCR5 expression patterns for YFP⁺T_reg, CD25⁺YFP⁻T_reg and CD25⁻non-T_reg CD4⁺CD8⁻ thymocytes derived from heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice (left). CXCR5 expression derived from Id2^fl/flId3^fl/flFoxp3^Cre/+ mice (bottom). MFI, mean fluorescence intensity. (d) CXCR5 and PD-1 expression in T_reg cells derived from the thymus and LNs isolated from control Id2^+/+Id3^+/+Foxp3^Cre/+ and Id2^fl/flId3^fl/flFoxp3^Cre/+ mice. Numbers in plots indicate percent CXCR5⁺PD-1⁺ and CXCR5⁺PD-1⁻ cells. (e) Flow cytometric analysis for Foxp3 and CTLA-4 expression in CD4 T cells derived from littermate control and Id2^fl/flId3^fl/flFoxp3^Cre mice. Numbers in plots indicate percent Foxp3⁺ cells. (f) Helios and CD25 expression (top) and Nrp1 (bottom) in Foxp3⁺ or Foxp3⁺Helios⁺ T_reg cells derived from 8-week-old control and Id2^fl/flId3^fl/flFoxp3^Cre LNs and thymi. Numbers in plots indicate percent Helios⁺ cells. Percentages of Helios⁺ cells in Foxp3⁺ T_reg cells isolated from control, 4-week-old or 8–9-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice (right). Data are representative of two experiments (a), three independent experiments (b; mean ± s.d.; control n = 3, Id2^fl/flId3^fl/flFoxp3^Cren = 5 biological replicates), two experiments (c; mean ± s.d.; n = 3 biological replicates), three experiments (d), three (Foxp3) and five (CTLA-4) experiments (e; mean ± s.d.; Foxp3 n = 3, CTLA-4 control n = 7, Id2^fl/flId3^fl/flFoxp3^Cren = 9 biological replicates), and three experiments (f; mean ± s.d.; n = 4 biological replicates). *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

Recent studies have indicated that T_reg cells generated in the thymus (tT_reg cells) abundantly express Helios or Neuropilin-1 (Nrp1)^37,38. We found that the proportions of Helios⁺ T_reg cells were comparable between 4-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice versus control mice but reduced in 8–9-week-old Id2^fl/flId3^fl/flFoxp3^Cre as compared to control mice. These data indicate that Id2 and Id3 maintain the Helios⁺ T_reg cell pool (Fig. 5f). Recent observations also have indicated a critical role for Nrp1 expression in maintaining stability of T_reg cells in inflammatory sites³⁹. We found that Nrp1 expression was substantially reduced in T_reg cells derived from Id2^fl/flId3^fl/flFoxp3^Cre mice or Id2^fl/flId3^fl/flFoxp3^Cre/+ mice (Fig. 5f and Supplementary Fig. 5f). These data indicate that the expression of a subset of genes associated with T_reg cell function, including CXCR5, PD-1, Foxp3 and Nrp1, are directly affected by loss of Id2 and Id3 expression.

Id2 and Id3 expression affects T_reg cell localization

Data described above indicate that loss of Id2 and Id3 expression leads to the abnormal proportions of T_reg cells that express CXCR5. To explore the possibility that aberrant expression of CXCR5 affects localization and/or homing of T_reg cells, we examined the localization of T_reg cells in spleens from control or Id2^fl/flId3^fl/flFoxp3^Cre and Id2^fl/flId3^fl/flFoxp3^Cre/Cre mice. Increased numbers of Foxp3⁺ cells were localized to the B cell area in the spleens derived from Id2^fl/flId3^fl/flFoxp3^Cre mice versus control mice (Supplementary Fig. 6). T_reg cells depleted for Id2 and Id3 expression also appeared to home more deeply into the B cell follicles across LNs (Supplementary Fig. 6). To examine how depletion of Id2 and Id3 expression affects T_reg localization across the lung tissues, we analyzed T_reg cells derived from 4-week-old and 7–8-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice. Whereas the Foxp3⁺ pool was not significantly affected in the lungs derived from healthy 4-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice, the percentage of Foxp3 T_reg cells localized across the lungs in 7–8-week-old Id2^fl/flId3^fl/flFoxp3^Cre mice was significantly decreased (Supplementary Fig. 7a). To compare the abilities of T_reg cells deficient for Id2 and Id3 expression to home to the lymphoid organs versus lung tissues, we transferred T_reg cells isolated from heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice into wild-type CD45.1 recipient mice. We found that the homing potential of T_reg cells depleted for Id2 and Id3 expression was reduced in the lung, as compared to the lymph nodes or spleen (Supplementary Fig. 7b).

Roles of Id2 and Id3 in T_reg-mediated immunosuppression

To determine whether the development of fatal T_H2 cell–mediated inflammation upon depletion of Id2 and Id3 expression in T_reg cells was caused by loss of suppressive functions of T_reg cells, we examined the in vitro suppressive function of T_reg cells sorted from 4-week-old Id2^fl/flId3^fl/flFoxp3^Cre or Foxp3^Cre mice. We found that T_reg cells depleted for Id2 and Id3 expression displayed normal in vitro immunosuppressive function (Supplementary Fig. 8).

To examine whether T_reg cells depleted for Id2 and Id3 expression have the ability to suppress T cell–mediated inflammation in vivo, we injected sorted CD4⁺CD25⁻CD45RB^hi T cells (T_conv cells) derived from CD45.1 mice into Rag1^−/− recipient mice with or without purified T_reg cells isolated from 4-week-old Id2^fl/flId3^fl/flFoxp3^Cre or Foxp3^Cre mice (Fig. 6a). As expected, Rag1^−/− recipient mice transferred with T_conv cells alone showed severe colitis accompanied by body-weight loss (Fig. 6a,b). Cotransfer of T_reg cells isolated from Id2^fl/flId3^fl/flFoxp3^Cre mice resulted in partially impaired suppressive activity (Fig. 6a,b). As expected, we found a large fraction of CD45.1⁻Foxp3⁺ T_reg cells in the MLNs of recipient mice but fewer Id-deficient T_reg cells in the MLNs of recipient mice (Fig. 6c). These findings are consistent with the notion that expression of Id2 and Id3 is required to maintain the T_reg pool. Overall these data indicate that T_reg cells depleted for Id2 and Id3 expression have the ability to suppress, albeit partially, T cell– mediated inflammation in a transfer colitis model.

Figure 6.

Impaired homeostasis of T_reg cells deleted for Id2 and Id3 expression after transfer into Rag1^−/− mice. (a) Body weight in indicated populations of mice after CD4⁺CD25⁻CD45RB^hi T cell (T_conv) from CD45.1 mice were intravenously injected into Rag1^−/− recipient mice with or without CD4⁺CD25⁺Foxp3-YFP⁺ T cell (T_reg) from 4-week-old Id2^fl/flId3^fl/flFoxp3^Cre or wild-type Foxp3^Cre mice (CD45.2) (n = 4 each group). WT, wild type. (b) H&E staining of large intestine (left). Blue and green arrowheads indicate crypt abscess and infiltration of inflammatory cells in the submucosal muscle layer, respectively. IBD score⁵⁰ in large intestine on day 44 after transfer (right). Original magnification, ×100 (c) Flow cytometric analysis of Foxp3 and CD45.1 expression gated on CD4⁺TCRβ⁺ cells in mesenteric LNs (left) and analysis of CD45.1Foxp3⁺ cells (right). Data are representative of one experiment (a–c; error bars, s.d.; n = 4 independent biological replicates). *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

Id2 and Id3 maintain T_reg cell homeostatic potential

To assess directly the roles of Id2 and Id3 under noninflammatory conditions, we analyzed heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice carrying the IRES-YFP reporter as described above. As inactivation of the X chromosome can occur on either allele, 50% of the T_reg cells in heterozygous female Foxp3^Cre-YFP/+ mice expressed the Foxp3^Cre-YFP marker, whereas the rest of the T_reg cell population lacked expression of Foxp3^Cre-YFP. This setting allows for a direct functional comparison of T_reg cells that are deficient for Id2 and Id3 expression versus wild-type cells under noninflammatory conditions. In the thymi derived from female Id2^fl/flId3^fl/flFoxp3^Cre/+ mice, the YFP⁺/CD25⁺ ratio of the CD4⁺ T cell population was comparable to that of Id2^+/+Id3^+/+ Foxp3^Cre/+ mice, whereas the YFP⁺/CD25⁺ ratio in the spleen and LNs was substantially reduced (Fig. 7a). Similarly, the fraction of YFP⁺ cells as compared to the proportion of Foxp3⁺ cells was substantially lower in spleen and LNs, indicating that the Id-sufficient Foxp3⁺ population outcompeted the Id-deficient population (Fig. 7a). We did not detect obvious differences in CD25, Annexin-V or Ki67 expression in Id2- and Id3-deficient YFP⁺ T_reg cells derived from Id2^fl/flId3^fl/flFoxp3^Cre/+ versus Foxp3^Cre/+ mice (data not shown). Notably, we found that when cultured in the presence of anti-CD3 and anti-CD28, the fraction of T_reg cells depleted for Id2 and Id3 expression (YFP⁺ cells) among Foxp3⁺ populations declined progressively over time and was accompanied by elevated numbers of apoptotic cells and decreased numbers of Ki67⁺ cells at 24 h (Fig. 7b). These observations indicate that in the absence of Id2 and Id3, T_reg cells are more susceptible to cell death upon TCR stimulation. Collectively, these data indicate that Id2 and Id3 expression is required to maintain the peripheral T_reg cell pool.

Figure 7.

Id2 and Id3 modulate the homeostasis and survival of T_reg cells. (a) Flow cytometric analysis for YFP and CD25 expression on cells isolated from thymocytes (Thy) and LNs (LN) derived from Id2^+/+Id3^+/+Foxp3^Cre/+ or Id2^fl/flId3^fl/flFoxp3^Cre/+ mice (left; cells were gated on the CD4⁺CD8⁻TCRβ^hi, CD4⁺CD8⁻ or CD4⁺ populations). Numbers in plots indicate percent YFP⁺CD25⁺ and YFP⁻CD25⁺ cells. Ratios of YFP⁺ T_reg cells versus CD25⁺YFP⁻T_reg cells and YFP⁺Foxp3⁺ T_reg versus Foxp3⁺ T_reg cells in indicated tissues (right). (b) Flow cytometric analysis of the Foxp3⁺CD4⁺ population in splenocytes derived from heterozygous female Id2^+/+Id3^+/+Foxp3^Cre/+ or Id2^fl/flId3^fl/flFoxp3^Cre/+ mice cultured and stimulated with anti-CD3ε plus anti-CD28 antibodies in the presence of human IL-2 (top left). Numbers indicate the fraction of YFP⁺Foxp3⁺ and YFP⁻Foxp3⁺ cells. Percentages of Ki67⁺ cells in the YFP⁺ T_reg cell populations analyzed 24 h after stimulation for the indicated genotypes (top right). Ratios of YFP⁺ cells for the indicated time points as compared to the zero time point (bottom left). Percentages of annexin-V⁺ cells gated on the YFP⁺ T_reg cell population (right). Data are representative of three independent experiments (a; mean ± s.d.; Id2^+/+Id3^+/+Foxp3^cre/+n = 5, Id2^fl/flId3^fl/flFoxp3^cre/+n = 7 biological replicates) and one experiment (b; mean ± s.d.; n = 3 technical replicates). *P < 0.05, **P < 0.01 (two-tailed unpaired Student’s t test).

Id2 and Id3 regulate T_FR cell–specific transcription signatures

As a first approach to identify critical target genes affected by the depletion of Id2 and Id3 in T_reg cells, we isolated by cell sorting YFP⁺ T_reg cells from subcutaneous LNs derived from 3-week-old Id2^fl/flId3^fl/fl Foxp3^Cre or Id2^fl/+Id3^fl/+Foxp3^Cre mice. 792 genes were differentially expressed (greater than twofold; fragments per kilobase of exons per million fragments mapped; FPKM) in YFP⁺ T_reg cells derived from Id2^fl/flId3^fl/flFoxp3^Cre mice when compared to those in Id2^fl/+Id3^fl/+Foxp3^Cre mice (Fig. 8a). Hierarchical clustering showed that a large fraction of transcripts that were differentially expressed were associated with genes encoding for metabolic process, transcription, T cell lymphocyte activation, cytokine/chemokine inflammation, cell cycle and cell death by Gene Ontology (GO) analysis (Supplementary Fig. 9). Prominent among the putative targets were Cxcr5, Ikzf3, Cebpa, Hif1a, Pparg, Il10, Il10ra, Ccr2, Nrp1 and Ccr6 (Fig. 8a–c and Supplementary Fig. 10a). However, expression of genes encoding transcriptional regulators, such as Tbx21, Irf4, Gata3, Stat3, Prdm1 and Sh2d1a, was not affected (data not shown). Similarly, cytokine genes including Il2, Il4, Il5, Il13, Ifng, Il17 and Il21 were not modulated upon depletion of Id2 and Id3 expression in the T_reg cell pool (data not shown). Similar to the case in T_FR cells, we found elevated expression of Gzma, Il10 and Ctla4. On the other hand, expression of Bcl6 and Prdm1 was not altered¹⁵ (Fig. 8a,b and Supplementary Fig. 10a).

Figure 8.

Unique patterns of gene expression in T_reg cells depleted for Id2 and Id3. (a) RNA-seq analysis of mRNA expression in sorted T_reg cells isolated from Id2^fl/+Id3^fl/+Foxp3^Cre versus Id2^fl/flId3^fl/flFoxp3^Cre mice (log₂; FPKM values for RefSeq genes). Points represent genes upregulated in Id2^fl/flId3^fl/flFoxp3^Cre T_reg cells > twofold (red; P < 0.05 (calculated by Cuffdiff)), genes downregulated in Id2^fl/flId3^fl/flFoxp3^Cre T_reg cells > twofold (blue; P < 0.05) and not significantly changed (gray) as compared to control cells. Selected genes are labeled. (b) RNA-seq analysis at Cxcr5, Ccr2, Il10, Apoe, Bcl6 and Prdm1 loci, presented in reads per million reads aligned (RPM). Arrows indicate transcription start site and direction of transcription. (c) Quantitative real-time PCR analysis of genes in sorted Foxp3-YFP⁺ T_reg cells derived from LNs isolated from heterozygous female Id2^+/+Id3^+/+Foxp3^Cre/+ or Id2^fl/flId3^fl/flFoxp3^Cre/+ mice, presented normalized by the abundance of Hprt1 transcript. (d) E2A occupancy (E2A binding) and deposition of H3K4me1 across the Cxcr5 and Il10 loci in DN3, A12 and proB cells. Numbers in plots indicate total tags observed. DN3 cells; Rag2^−/− thymocytes, A12 cells; E2a^/ E47-reconstituted T cell line, pro-B cells; Rag2^−/− bone marrow B cells. Arrows indicate transcriptional start site and direction of transcription. Data are representative of three experiments (a,b), one experiment (c; mean ± s.d.; n = 3 technical replicates) and one experiment (d).

To identify target genes directly affected by depletion of Id2 and Id3 expression rather than genes affected by inflammatory disease, we isolated RNA from YFP⁺ T_reg cells from heterozygous female Id2^fl/flId3^fl/flFoxp3^Cre/+ and Id2^+/+Id3^+/+Foxp3^Cre/+ mice and examined it by real-time PCR. We found that ablation of Id2 and Id3 expression in the T_reg cell pool in healthy mice affected Cxcr5, Ikzf3, Hif1a, Myb, Apoe, Il10, Il10ra, Ccr2, E2f2, Nrp1 and Il18r1 transcript abundance (Fig. 8c). In contrast, Ikzf4, Ccr6, Cebpa, Pparg1 and S1pr1 transcript abundance was not altered in T_reg cells depleted of Id2 and Id3 (data not shown). To determine whether the differences in transcript abundance directly relate to transcription factor occupancy across putative enhancer elements, we inspected previous genome-wide studies for E2A occupancy as well as deposition of monomethylation of histone H3K4 (H3K4me1) in CD4⁻CD8⁻double-negative (DN3) thymocytes, pro-B cells and T cell lymphomas^26,40. Notably, E2A-bound sites were closely associated with putative enhancer regions across the Cxcr5, IL10, Hif1a, Ikzf3, Myb, Il10ra, Nrp1, E2f2 and Bmp7 loci, which indicated that this ensemble of loci is directly regulated by E proteins (Fig. 8d and Supplementary Fig. 10b). These data indicate that a subset of genes associated with T_FR cells, including Il10 and Cxcr5, but not Bcl6 and Prdm1, are directly regulated by the expression of Id2 and Id3 in T_reg cells to orchestrate the developmental progression of the T_FR cell population (Supplementary Fig. 10c).

DISCUSSION

Here we demonstrated that expression of the antagonist HLH proteins, Id2 and Id3, in T_reg cells is required for suppression of early onset inflammation at mucosal barriers. How does depletion of Id2 and Id3 expression in T_reg cells lead to the development of severe inflammation? Our data indicate that expression of Id2 and Id3 is essential to maintain the T_reg cell population. In 3– 4-week-old mice that lacked expression of Id2 and Id3, we observed T_reg cells in near wild-type numbers. However, in 7–8-week-old mice, the T_reg cell population depleted for expression of Id2 and Id3 was severely reduced across the lung tissues. Homing of T_reg cells to the lung in transfer experiments was also compromised in the absence of Id2 and Id3 expression. Finally, we found that activated T_reg cells depleted for expression of Id2 and Id3 displayed decreased viability and impaired proliferation upon stimulation. Collectively, these data indicate that Id2 and Id3 coordinate the survival, expansion and homing of T_reg cells under inflammatory conditions.

In addition to the roles of Id2 and Id3 in suppressing T_H2 cell– mediated inflammation, our observations also point to a role for Id2 and Id3 in T_FR cell development. How may Id2 and Id3 regulate T_FR differentiation? We suggest that Id2 and Id3 expression enforces distinct T_FR checkpoints. Upon TCR-mediated signaling, Id2 and Id3 levels readily decline permitting increased E-protein binding activity and the induction of a T_FR-specific transcription signature, including the activation of Cxcr5 and IL-10 expression. We propose that abundance of Id2 and/or Id3 is elevated and required again in differentiating T_FR cells to indirectly modulate Bcl-6 and Blimp-1 protein abundance, which ultimately leads to the development of a mature T_FR cell population.

There are interesting parallels between the roles of E and Id proteins in development of T_FR cells and T_FH cells. Differentiation of T_FH cells is characterized by distinct developmental stages. First, upon activation, effector T_FH cells induce the expression of CXCR5 to migrate to the border that separates the T cell areas from the B cell follicles. Followed by ICOS-ICOSL interaction with B cells, T_FH cells continue to differentiate, acquire Bcl-6 expression and migrate to GCs^41,42. Two critical factors control the early developmental progression of naive CD4⁺ T cells toward the effector T_FH cell type: Id3 and a heterodimeric complex involving the basic HLH protein Ascl2 and E proteins to activate the expression of CXCR5 and other critical target genes^26,43,44. Based on these observations, we suggest that a common genetic circuitry that involves the E-Id protein module controls the homing of both T_FR cells and T_FH cells.

In sum, higher Id2 and Id3 expression enforces the early T_FR cell checkpoint, whereas at intermediate developmental stages the abundance of E protein versus Id protein controls developmental progression toward a fully mature T_FR cell fate. We note this property of Id proteins and E proteins is not unique to T_reg cells. Previous studies have demonstrated critical roles for E and Id proteins acting as gatekeepers at the pre-TCR, TCR and B cell antigen receptor checkpoints^{21,23,45–47}. During the developmental progression of the CD8⁺ and T_FH cell lineages, the Id proteins also control key developmental transitions^27,48,49. Each of these checkpoints involves the regulation of Id protein expression by antigen receptor–mediated signaling. Thus, a picture is now emerging in which antigen receptors directly connect to the E protein and Id protein machinery to orchestrate the developmental transitions of cells that comprise the adaptive immune system under healthy as well as inflammatory conditions.

ONLINE METHODS

Mice

C57BL/6, Id3^fl/fl, Id2^fl/fl Id3^−/−, Foxp3 ^Cre-YFP and Id3^GFP/+ Id2^YFP/+ mice were bred and housed in specific pathogen-free conditions in accordance with the Institutional Animal Care and Use Guidelines of the University of California, San Diego.

Flow cytometry

Single-cell suspensions from thymus, lymph nodes and spleen were stained with the following: FITC-, PE-, APC-, APC-Cy7-, Pacific Blue-, Alexa Fluor 700-, Alexa Fluor 780-, PerCP-Cy5.5-, PE-Cy7- or biotin-labeled monoclonal antibodies were purchased from BD PharMingen including CD11c (HL3), CD44 (IM7), CXCR5 (2G8), CD95 (Jo-2), GL-7 (GL7), SiglecF (E50–2440), IL-5 (TRFK5). CD8 (53-6.7), CD19 (ID3), CD38 (90), CD62L (MEL-14), CD44 (IM7), B220 (RA3–6B2), Mac1 (M1/70), Gr1 (RB6–8C5), Nk1.1 (PK136), Ter119 (TER119), TCRβ (H57), TCRγδ (GL3), PD-1 (J43), ICOS (7E.17G9), CXCR4 (2B11), IL-4 (11B11), IL-13 (eBio13A), IL-17 (eBio17B7), IFN-γ(XMG1.2), FOXP3 (FJK-16s) and TBR2 (Dan11mag) were purchased from eBioscience. FceR1 (MAR-1), CD3e (2C11), CD4 (GK1.5), CD8 (53-6.7), CD11b (M1/70), CD25 (PC61), CD122 (TMb1), CTLA4 (UC10-4B9), CCR6 (29-2L17), Helios (22F6) and PD-1 (RMP1–30) were obtained from BioLegend. Anti-Nrp1 (AF566) was purchased from R&D Systems. Biotinylated antibodies were labeled with streptavidin-conjugated Qdot-665 or Qdot-605 (Invitrogen). Clone 2.4 G2 anti-CD32:CD16 (eBioscience) was used to block Fc receptors. Dead cells were removed from analysis and sorting by staining with propidium iodide (PI) (Sigma-Aldrich). Isolation of lung single cells was performed as described⁵¹. Mouse T_reg cell staining kit (eBioscience) was used for intracellular Foxp3, CTLA4 and Helios staining. Samples were collected on a LSRII (BD Biosciences) and were analyzed with FlowJo software (Tree Star). Cells were sorted on a FACSAria (BD). For intracellular staining of IFN-γ, IL-4, IL-5, IL-13 and IL-17, isolated cells from spleen and lymph nodes were treated with phorbol 12-myristate 13-acetate (PMA) plus ionomycin (4 h) and Golgi stop (2 h). After culture, cells were stained with anti-CD4, CD8, TCRβ and CD3e. Intracellular staining was performed with the BD Biosciences Cytofix/Cytopermkit.

RNA sequencing

Total RNA was isolated using the RNeasy-Mini Kit (Qiagen) and was processed for deep sequencing as described previously with slight modifications. Briefly, mRNA was purified from total RNA using Dynabeads mRNA purification kit (Invitrogen). The mRNA was then subjected to first-strand cDNA synthesis using a combination of random hexamers and oligo(dT). Second-strand synthesis was performed using dUTP instead of dTTP. The double-stranded cDNA was sonicated using S220 Focused-ultrasonicator (Covaris). Sonicated cDNA was next ligated to adaptors and size-selected. The size-selected cDNA was treated with uracil-N-glycosylase and then multiplex-amplified by PCR. cDNA libraries were sequenced with a HiSeq 2000 sequencer (Illumina). Sequence reads from each cDNA library were analyzed with Galaxy (http://galaxyproject.org/). They were mapped to mouse genome build NCBI37 mm9 using TopHat. The expression values were obtained using Cufflink and Cuffdiff. GO analyses were performed using Homer (http://homer.salk.edu/homer/index.html). Visualization files were generated using Homer and read pile-ups visualized using the University of California Santa Cruz Genome Browser.

Histology and immunostaining

Tissues were fixed in 4% paraformaldehyde (Electron Microscopy Sciences). Fixed tissues were embedded in paraffin and sliced, followed by hematoxylin and eosin (H&E) staining, periodic acid Schiff (PAS) staining and MBP staining. For MBP immunostaining, we used a monoclonal antibody directed against MBP⁵¹. For quantification of MBP⁺ or PAS⁺ cells, we chose five bronchioles and count positive cells in one mouse. For Foxp3 staining, spleen and lymph nodes were embedded in optimal cutting temperature (OCT) compound and snap-frozen. 5-µm serial sections were stained with anti-CD45R (B220; supernatant isolated from the medium of cultured hybridomas) or anti-Foxp3 (FJK-16s). Secondary antibody used was donkey anti-rat IgG Alexa Fluor 488 (Molecular Probes). The images were captured by Olympus Provis AX80 microscopy. The digital images were acquired using RETIGA EXi (Nippon Rooper). Image process was done by Photoshop (Adobe). IBD scores were calculated as described⁵⁰.

Enzyme-linked immunosorbent assay

Serum IgM, IgG1, IgG2a, IgG2b, IgG3 and IgA were done as described⁵². For IgE, ELISA was performed using Mouse IgE Standard Set Module (BioLegend).

RNA isolation and reverse transcriptase PCR

First, cDNA was obtained from purified YFP⁺ or CD25⁺ T_reg cells from Id2^+/+Id3^+/+Foxp3^Cre/+ or Id2^fl/flId3^fl/flFoxp3^Cre/+ mice. Total RNA was isolated from sorted cells with RNeasy kit (Quagen), followed by reverse transcription with the SuperScript III RT-PCR system (Invitrogen) and quantitative real-time reverse transcription (RT)-PCR.

In vitro and in vivo suppression and in vitro stimulation

For in vitro suppression assay, YFP⁺ T_reg cells and responder CD4⁺CD25⁻ T cells were purified by sorting. 2 × 10⁴ Responder CD4⁺ T cells were cocultured with T_reg cells at indicated ratios and were stimulated with soluble anti-CD3 antibody (1 µg/ml) in the presence of T cell–depleted splenocytes treated with mitomycin C (1 × 10⁵) in 96-well plates for 72 h. Cells were pulsed with [³H]thymidine deoxyribose (1 µCi/well) for the last 12 h. In vivo suppression assay was performed as described⁵³. For in vitro stimulation, total lymphocytes were cultured in vitro and stimulated with anti-CD3ε (0.5 µg/ml) plus anti-CD28 (1 µg/ml) antibodies in the presence of human IL-2 (100 U/ml) as described³⁰.

Statistical analysis

P values were calculated with the two-tailed Student’s t-test for two-group comparison as applicable with Microsoft Excel software. The statistical significance level was 0.05.