Continuous requirement for the T cell receptor for regulatory T cell function

Abstract

Foxp3⁺ regulatory T cells (T_reg cells) maintain immunological tolerance and their deficiency results in fatal multi-organ autoimmunity. Although heightened T cell receptor (TCR) signaling is critical for the differentiation of T_reg cells, the role of TCR signaling in T_reg cell function remains largely unknown. Here we demonstrate inducible ablation of the TCR results in T_reg cell dysfunction which cannot be attributed to impaired Foxp3 expression, decreased expression of T_reg cell signature genes or altered ability to sense and consume interleukin 2. Rather, TCR signaling was required for maintaining the expression of a limited subset of genes comprising 25% of the activated T_reg cell transcriptional signature. Our results reveal a critical role for the TCR in T_reg cell suppressor capacity.

Regulatory CD4⁺ T cells expressing the transcription factor Foxp3 play an essential role in maintaining immune tolerance¹. In the thymus, increased affinity T cell receptor (TCR) engagement in immature CD4 single positive thymocytes is required for initiation of the T_reg cell differentiation program and induction of Foxp3 expression². As a consequence, T_reg cells exported to the periphery exhibit a TCR repertoire skewed towards self-recognition^{3, 4}. However, the requirement for TCR signaling in mediating T_reg cell suppressive function in the periphery remains largely unclear.

In comparison to conventional CD4⁺ T cells, T_reg cells exhibit impaired calcium flux, Akt activation and Erk phosphorylation upon TCR stimulation, and Foxp3 is known to potently repress at least some TCR-induced genes, as well as some genes involved in the TCR signaling pathway^5–8. At the same time, Foxp3⁺ T_reg cells have high basal expression of several cell surface molecules which are known to contribute to T_reg cell function (such as CD25, CD39 and CTLA-4) and whose expression in conventional CD4⁺ T cells is dependent upon TCR stimulation^9–14. Whether or not high affinity TCR interactions with self contribute to constitutive expression of these genes and consequently to T_reg cell function is not known.

T_reg cells—despite their intrinsically dampened response to TCR stimulation—acquire an activated phenotype and expand in response to their cognate antigens in settings of immune activation, such as infection and autoimmunity^{15, 16}. These observations imply that self-antigen recognition helps maintain T_reg cells of particular specificities and may potentiate their suppressive capacity during immune challenge¹⁷. Nevertheless, strict reliance on TCR expression for T_reg cell activation—as opposed to preferential activation of antigen-specific T_reg cells—has not been demonstrated, nor has TCR engagement in vivo been shown to be required for T_reg cell function in any context.

We used inducible genetic ablation of cell surface TCR complexes to directly address the requirement of TCR expression for T_reg cell immunosuppressive capacity. Notably, the TCR was largely dispensable for Foxp3 expression, lineage stability and for high expression of many T_reg cell signature genes. Nevertheless, these features were not sufficient to preserve T_reg cell function and to prevent immune activation. Loss of suppressor capacity in the absence of TCR was not due to impaired T_reg cell ability to gain access to interleukin 2 (IL-2), and accordingly administration of exogenous IL-2 failed to rescue systemic autoimmunity. Instead, TCR expression was essential for T_reg cell activation and maintenance of a limited set of genes which were found to be expressed almost exclusively in activated T_reg cells. Among these genes, expression of the transcription factor IRF4 contributed to optimal T_reg cell function and homeostasis. These results demonstrate an essential role for the TCR in eliciting the suppressor function of differentiated T_reg cells.

Results

Maintenance of T_reg cell identity in the absence of the TCR

In order to investigate the consequences of TCR signaling in T_reg cell function, we crossed Foxp3^eGFP-CreERT2 mice to Cα^FL mice to allow inducible deletion of Cα encoding the TCRα constant chain specifically in T_reg cells^{18, 19}. In this model, Cre-induced loss of the conditional Cα allele upon tamoxifen administration eliminates TCRα expression, preventing cell surface TCRαβ heterodimer formation. Tamoxifen was administered via oral gavage on days 0 and 1 and mice were analyzed on day 9. Allelic exclusion at the Cα locus in heterozygous Foxp3^CreERT2Cα^FL/WT mice yielded a minor population of TCR-deficient (TCR⁻) T_reg cells (~25%), while in homozygous Foxp3^CreERT2Cα^FL/FL mice the majority of T_reg cells (~60–70%) lacked cell surface TCRβ (Fig. 1a). Although we cannot definitively exclude the possibility that few residual TCR complexes were present in minute amounts (below the detection limit of flow cytometric analyses) on the TCR⁻ T_reg cells, functional in vitro analyses confirmed loss of TCR crosslinking-dependent activation of TCR⁻ T_reg cells (Supplementary Fig. 1a–d).

Figure 1. Maintenance of T_reg cell identity in the absence of the TCR.

(a, b) TCRβ expression (a) and Foxp3 median fluorescence intensity (MFI) (b) in CD4⁺Foxp3⁺ lymph node cells from 8–10 week old Foxp3^CreERT2Cα^WT/WT (black line, WT/WT), Foxp3^CreERT2Cα^FL/WT (blue line, FL/WT) and Foxp3^CreERT2Cα^FL/FL (red line, FL/FL) mice gavaged on days 0 and 1 with tamoxifen and analyzed on day 9. The gray histogram in (a) shows TCRβ staining on CD4⁻TCRβ⁻ cells. (c) Percentages of TCRβ⁺Foxp3⁺ and TCRβ⁻Foxp3⁺ T cells among CD4⁺YFP⁺ cells sorted to >99% purity from the spleens and lymph nodes of Foxp3^CreERT2Cα^FL/WTRosa^YFP mice on day 13 following treatment with tamoxifen on days 0 and 1 and i.p. injections of IL-2 neutralizing or control antibody on days 4 and 8. Data are representative of two independent experiments with four or more (a,b) or two or more (c,d) mice per group in each. ***, P < 0.001; **, P < 0.01; *, P < 0.05. P-values were calculated using an unpaired t-test.

Because binding sites for the transcription factors NFAT and c-Rel have been identified within the Foxp3 locus, and TCR engagement-driven NF-κB signaling is critical for induction of Foxp3 expression, we speculated that the TCR might be essential to maintain Foxp3 expression^20–22. However, Foxp3 expression was only marginally reduced in TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/WT mice, and was not reduced at all in TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/FL mice, compared to TCR⁺ T_reg cells present in the same mice and in Foxp3^CreERT2Cα^WT/WT animals (Fig. 1b). Similarly, the expression of several T_reg cell signature molecules, including CD25, GITR, CD39 and CD73, was largely unaffected in TCR⁻ T_reg cells from both Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice (Supplementary Fig. 1e). These results indicates that, at steady state, continuous TCR-mediated recognition of self does not contribute significantly to Foxp3-dependent maintenance of expression for these genes^{11, 23}. In contrast, CTLA-4 expression was notably diminished in TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/WT mice, although not in TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/FL mice, compared to TCR⁺ T_reg cells in these mice and in Foxp3^CreERT2Cα^WT/WT animals (Supplementary Fig. 1e).

Compared to Foxp3^CreERT2Cα^WT/WT mice, percentages and absolute numbers of Foxp3⁺ cells in the spleens and lymph nodes of Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice were unaltered (Supplementary Figure 1f). However, to address the possibility that a portion of T_reg cells completely lost Foxp3 expression upon ablation of the TCR and that these “former” T_reg cells were not accounted for in this experimental setup, we crossed Foxp3^CreERT2Cα^FL/WT mice with mice expressing the recombination reporter Rosa26^YFP allele. TCRβ and Foxp3 expression was assessed in CD4⁺YFP⁺ cells sorted from the spleens and lymph nodes of Foxp3^CreERT2Cα^FL/WTRosa^YFP mice on day 9 or 50 following tamoxifen administration on two consecutive days. YFP-expressing CD4⁺TCRβ⁻ and CD4⁺ TCRβ⁺ cell subsets contained similarly minor frequencies of Foxp3⁻ cells at both time points (data not shown). Furthermore, following in vivo IL-2 neutralization – a condition known to promote loss of Foxp3 expression – in Foxp3^CreERT2Cα^FL/WTRosa^YFP mice, CD4⁺YFP⁺TCRβ⁻ cells retained a higher percentage of Foxp3⁺ cells than did CD4⁺YFP⁺TCRβ⁺ cells. Together, these data indicate that TCR signaling is dispensable for the maintenance of the T_reg cell phenotype and lineage stability, and moreover, that TCR signaling can drive the loss of Foxp3 when IL-2 amounts are limiting (Fig. 1c)¹⁹.

Requirement for the TCR in T_reg cell effector differentiation

Although the T_reg cell phenotype was largely preserved upon ablation of the TCR, we observed a relative enrichment of naïve-like CD44^loCD62L^hi cells among TCR⁻ T_reg cells in lymph nodes and spleens of Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice (Fig. 2a and Supplementary Fig. 1g). T_reg cell proliferation is almost exclusively restricted to the CD44^hi subset, and in part, the enrichment we observed appeared to be a consequence of the severely impaired proliferative capacity of CD44^hi T_reg cell in the absence of the TCR (Fig. 2b)²⁴. The minor population of TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/WT mice was predominantly non-dividing: these cells showed minimal expression of Ki67, failed to incorporate BrdU over a 24 hour labeling period and contained the largest percentage of CD44^loCD62L^hi cells among all TCR⁺ and TCR⁻ T_reg cell populations in Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT, and Foxp3^CreERT2Cα^FL/FL mice (Fig 2a,b and Supplementary Fig. 1h). In Foxp3^CreERT2Cα^FL/FL mice, however, TCR⁻ T_reg cells exhibited considerable proliferative activity, albeit reduced in comparison to TCR⁺ T_reg cells present in the same mouse (Fig 2a). Increased Ki67 staining in CD44^hi Treg cells inversely correlated with a decreased percentage of CD44^loCD62L^hi cells within all TCR⁺ and TCR⁻ T_reg cell populations present in Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT, and Foxp3^CreERT2Cα^FL/FL animals (Fig 2a, b).

Figure 2. Requirement for the TCR in T_reg cell differentiation and expansion.

(a, b) CD44 and CD62L expression on TCR⁺ and TCR⁻ CD4⁺Foxp3⁺ (a) and Ki67 expression on CD44^hi CD4⁺Foxp3⁺ (b) lymph node cells isolated on day 9 from Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT, and Foxp3^CreERT2Cα^FL/FL mice treated with tamoxifen on days 0 and 1. Data in (a) and (b) are representative of two independent experiments with four or more mice per group in each. (c) CD44 and CD62L expression on TCRβ⁺ and TCRβ⁻ CD4⁺Foxp3⁺ lymph node cells in 2.5 week old Foxp3^CreCα^WT/WT (WT/WT), Foxp3^CreCα^FL/WT (FL/WT) and Foxp3^CreCα^FL/FL (FL/FL) mice (left) and differentiation marker expression on TCRβ⁺ and TCRβ⁻ CD4⁺Foxp3⁺ lymph node cells in 2.5 week old Foxp3^CreCα^FL/WT mice (right). Data is representative of three independent experiments involving a total of six or more mice in each group. (d,e) Ki67 expression in lymph node TCRβ⁺ and TCRβ⁻ CD4⁺Foxp3⁺ cells (d) and percentages of TCRβ⁺ and TCRβ⁻ CD4⁺Foxp3⁺ cells among total CD4⁺ cells (e) in 2.5 week old Foxp3^CreCα^WT/WT (WT), Foxp3^CreCα^FL/WT (Het), and Foxp3^CreCα^FL/FL (KO) mice. The data in (d) and (e) represent an aggregate of three independent experiments with a total of six or more mice per group. **, P < 0.001; *, P < 0.01. P-values were calculated using an unpaired t-test.

To address the possibility that continuous peripheral differentiation of naïve-like T_reg cells into CD44^hi cells was impeded in the absence of TCR expression and that such a differentiation block contributed to the predominantly CD44^loCD62L^hi phenotype of TCR⁻ T_reg cells, we bred Cα^FL mice to Foxp3^YFP-Cre mice to induce ablation of the TCR in newly generated, “naïve” T_reg cells^{25, 26}. We reasoned that if the TCR was critical for T_reg cell effector differentiation in the periphery, TCR⁻ T_reg cells in these mice would retain a CD44^loCD62L^hi naïve-like phenotype.

Immature HSA^hiCD4⁺Foxp3⁺ cells in the thymi of Foxp3^CreCα^FL/FL mice showed similar cell surface expression of TCR complexes as their wild type counterparts, while more mature HSA^loCD4⁺Foxp3⁺ thymocytes showed only slightly reduced TCR expression (Supplementary Fig. 2a.). Among Foxp3⁺ cells present in the spleens and lymph nodes of Foxp3^CreCα^FL/WT mice, ~5–15% were TCRβ⁻ whereas ~80% of Foxp3⁺ cells in Foxp3^CreCα^FL/FL mice lacked surface TCRβ expression (Fig 2e and Supplementary Fig. 2b). We again observed a slight decrease in the amount of Foxp3 protein in the TCR⁻ T_reg cells in Foxp3^CreCα^FL/WT mice, but not in Foxp3^CreCα^FL/FL mice, while expression of T_reg cell signature genes was variably affected in the TCR⁻ T_reg cell populations in both mice as compared to TCR⁺ Treg cells in Foxp3^CreCα^WT/WT, Foxp3^CreCα^FL/WT, and Foxp3^CreCα^FL/FL mice (Supplementary Fig. 2c,d).

Despite their generally intact T_reg cell surface phenotype, nearly all TCR⁻ T_reg cells in healthy Foxp3^CreCα^FL/WT mice had a naïve-like CD62L^hiCD44^lo phenotype and lacked expression of all T_reg cell differentiation markers tested, including KLRG1, CD103 and CXCR3 (Fig. 2c). Notably, this pattern was also observed under severe inflammatory conditions in Foxp3^CreCα^FL/FL mice, which were moribund by three weeks of age (Fig. 2c and Supplementary Fig. 3e,f). Lack of CD44^hi cells among TCR⁻ populations in Foxp3^CreCα^FL/WT and Foxp3^CreCα^FL/FL mice correlated with decreased proliferation and markedly diminished percentages and numbers of TCR⁻ T_reg cells in lymph nodes and, particularly, in the spleens and tissues such as liver and lung as compared to TCR⁺ T_reg cells in the same mice and in Foxp3^CreCα^WT/WT animals (Fig. 2d,e and Supplementary Fig. 2g,h). Together, these data are consistent with an absolute requirement for TCR expression—loss of which cannot be compensated for even in conditions of extreme immune activation—for peripheral effector differentiation of naïve-like T_reg cells and acquisition of an activated CD44^hi phenotype.

TCR-dependent effector function of mature Treg cells

Foxp3 expression and T_reg cell expansion are facilitated by IL-2R signaling^27–29. The increase in Foxp3 protein expression and proliferative activity of TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/FL mice vs. Foxp3^CreERT2Cα^FL/WT mice led us to suspect that ablation of TCR expression, even on mature T_reg cells, might precipitate immune activation and elevate production of IL-2 and other cytokines by activated CD4⁺ T cells. Indeed, analysis of Foxp3^CreERT2Cα^FL/FL mice treated twice with tamoxifen and analyzed on day 9 post-treatment revealed increased percentages of CD44^hi T cells and increased numbers of IL-2-producing CD4⁺ T cells compared to tamoxifen-treated Foxp3^CreERT2Cα^WT/WT and Foxp3^CreERT2Cα^FL/WT mice (data not shown).

To more rigorously assess the role for TCR expression in mature T_reg cell function, we administered four doses of tamoxifen to mice (on days 0, 3, 7 and 10) in order to maximize Cre-ERT2-mediated recombination. On day 13, we noted deletion of TCR expression in ~75–80% of T_reg cells in Foxp3^CreERT2Cα^FL/FL and ~25–30% in T_reg cells in Foxp3^CreERT2Cα^FL/WT mice (Fig. 3a,b). Despite normal or even increased percentages of total Foxp3⁺ cells in the lymph nodes and spleens of Foxp3^CreERT2Cα^FL/FL mice, elevated numbers of CD4⁺Foxp3⁻ and CD8⁺ T cell were found in the lymph nodes, and increased percentages of CD44^hi CD4⁺ and CD8⁺ T cells were found in the lymph nodes and spleens of these mice (Fig. 3ca–d). CD8⁺ T cells and CD4⁺ T cells from Foxp3^CreERT2Cα^FL/FL mice produced increased amounts of interferon-γ (IFN-γ) and IFN-γ, IL-2, IL-4, IL-13, IL-5 and IL-17, respectively, when compared to T cells from Foxp3^CreERT2Cα^WT/WT and Foxp3^CreERT2Cα^FL/WT mice (Fig. 3e and data not shown).

Figure 3. TCR-dependent effector function of mature T_reg cells in adult mice.

(a,b) TCRβ and Foxp3 expression among lymph node CD4⁺ cells (a) and percentages of Foxp3⁺ cells among lymph node and splenic CD4⁺ cells (b) in Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice on day 13 following tamoxifen treatment on days 0, 3, 7 and 10. In (b), percentages of TCRβ⁺Foxp3⁺ cells among CD4⁺ cells in Foxp3^CreERT2Cα^WT/WT (light gray bars), Foxp3^CreERT2Cα^FL/WT (dark gray bars) and Foxp3^CreERT2Cα^FL/FL (black bars) are shown together with percentages of total Foxp3⁺ cells among CD4⁺ cells (white bars) within each genotype. Error bars in (b) indicate s.e.m. Data in (a) is representative of three experiments with three mice or more per group in each. Data in (b) represent an aggregate of three experiments with a total of nine mice or more mice per group. (c–d) Numbers of (c) and CD44 expression (d) and cytokine production (e) by CD4⁺Foxp3⁻ and CD8⁺ T cells in Foxp3^CreERT2Cα^WT/WT (white circles), Foxp3^CreERT2Cα^FL/WT (gray circles) and Foxp3^CreERT2Cα^FL/FL (black circles) mice on day 13 following tamoxifen treatment on days 0, 3, 7 and 10 in spleens and lymph nodes for (c,d) and spleens for (e). Data represent an aggregate of three experiments with nine or more mice per group. ***, P < 0.001; **, P < 0.01; *, P < 0.05. P-values were calculated using an unpaired t-test.

Immune activation in Foxp3^CreERT2Cα^FL/FL mice was milder than that resulting from complete T_reg cell depletion in Foxp3^DTR mice, which express the human diphtheria toxin receptor (DTR) concomitantly with Foxp3³⁰. Thus, it was possible that the large numbers of TCR⁻ T_reg cells in Foxp3^CreERT2Cα^FL/FL mice retained measurable TCR-independent suppressor capacity and were still capable of immune regulation. Alternatively, the minor population of remaining TCR-sufficient T_reg cells in these mice might have limited the activation of effector T cells and the associated autoimmunity to some degree. To address these possibilities, we attempted to reduce the T_reg cell percentages among CD4⁺ cells in Foxp3^DTR mice to approximate the percentages of the residual TCR-sufficient T_reg cells in Foxp3^CreERT2Cα^FL/FL mice following four doses of tamoxifen³¹. We reasoned that if TCR⁻ T_reg cells were capable of significant suppression, autoimmunity in Foxp3^DTR mice subjected to only partial depletion of T_reg cells would be more severe than in Foxp3^CreERT2Cα^FL/FL mice. As diphtheria toxin (DT) injection depletes T_reg cells within 24 hours, while tamoxifen-induced Cre-ERT2 mediated recombination progressively increases over a four day period, we treated Foxp3^DTR, Foxp3^CreERT2Cα^WT/WT and Foxp3^CreERT2Cα^FL/FL mice each with DT four days after their first dose of tamoxifen (which like DT was administered to mice of all genotypes) in order to account for the time needed for complete Cre-ERT2-mediated deletion of Cα (Supplementary Fig. 3a,b and data not shown). Partial depletion of the T_reg cell compartment in Foxp3^DTR mice resulted in CD4⁺Foxp3⁻ T cell activation and cytokine production grossly comparable to that observed in Foxp3^CreERT2Cα^FL/FL mice harboring populations of TCR-sufficient T_reg cells of a similar or even larger size (Supplementary Fig. 3b–d). Together, these results demonstrate that T_reg cells require continuous TCR expression for the effective elaboration of their suppressor function, and suggest that TCR⁻ T_reg cells, which are abundant in Foxp3^CreERT2Cα^FL/FL mice, are grossly devoid of detectable suppressor capacity.

TCR⁻ T_reg cell dysfunction is not secondary to impaired IL-2R signaling

We considered that the apparent loss of T_reg cell suppressive capacity in the absence of the TCR might be an indirect consequence of an impaired ability to localize in a TCR- and antigen-dependent manner to sites of CD4⁺ T cell activation and to, thereby, acquire IL-2, a cytokine known to be critical for T_reg cell function and homeostasis. This scenario would explain the decreased expression of Foxp3 and minimal proliferation of TCR⁻ T_reg cells in healthy Foxp3^CreERT2Cα^FL/WT mice, in which IL-2 amounts were not elevated and would not be able to partially remedy these defects³².

However, direct ex vivo analysis of phosphorylation of Stat5—which occurs downstream of IL-2 signaling in T_reg cells—in spleen and lymph nodes showed at least equivalent proportions of p-Stat5 in TCR⁻ and TCR⁺ T_reg cells in both Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice (Fig. 4a). In Foxp3^CreERT2Cα^FL/FL mice, TCR⁻ T_reg cells had increased p-Stat5 compared to TCR⁺ T_reg cells, mirroring expression of CD25 and CD62L, which remained high on TCR⁻ cells, but was decreased on the residual, activated TCR⁺ T_reg cells present in these mice (Fig. 4a and Supplementary Figure 1e,f). These results are consistent with the observation that p-Stat5⁺ T_reg cells are largely found within the CD62L^hiCD44^lo (and CD25^hi) T_reg cell subset which, in contrast to the activated CD44^hiCD62L^lo (and CD25^int) T_reg cell subset, rely on IL-2R signaling rather than co-stimulatory receptor engagement for their maintenance²⁴.

Figure 4. T_reg cell TCR expression is dispensable for IL-2R signaling in vivo.

(a) p-Stat5 expression in lymph node TCRβ⁺ and TCRβ⁻ CD4⁺Foxp3⁺ cells in Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice on day 9 following tamoxifen administration on days 0 and 1. Data is representative of three experiments with three or more mice per group in total. (b) Foxp3 median fluorescence intensity (MFI) in TCRβ⁺ and TCRβ⁻ Foxp3⁺ cells (upper panel) and percentages of TCRβ⁺ and TCRβ⁻ Foxp3⁺ cells among CD4⁺ cells (lower panel) in lymph nodes of Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice on day 13 following tamoxifen treatment on days 0, 3, 7 and 10 and i.p. injections of IL-2 neutralizing or isotype antibody on days 4 and 8. The data in (b) is representative of two experiments with two or more mice per group each. ***, P < 0.001; **, P < 0.01; *, P < 0.05. P-values were calculated using an unpaired t-test.

In vitro analysis confirmed that lack of TCR expression did not significantly influence Stat5 phosphorylation in response to IL-2, nor impair T_reg cell ability to capture and deplete IL-2 from culture media, suggesting that in the absence of TCR expression T_reg cell-mediated IL-2 deprivation may not be a significant mechanism of immune suppression (Supplementary Fig. 4a,b). Furthermore, treatment of Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice with neutralizing anti-IL-2 (or isotype control) antibody reduced Foxp3 expression and decreased cell percentages among total CD4⁺ cells comparably between TCR⁺ and TCR⁻ T_reg cells (Fig. 4b). Together, these data suggest that TCR⁻ T_reg cells efficiently capture IL-2 during immune activation and at steady state. Although it remains to be determined what signal(s) drive the proliferation of TCR⁻ T_reg cells selectively in diseased Foxp3^CreERT2Cα^FL/FL mice, we observed increased CD80 and CD86 expression on lymph node dendritic cells (DCs) in Foxp3^CreERT2Cα^FL/FL mice, and activated DCs were able to induce limited proliferation of TCR⁻ T_reg cells in vitro (Supplementary Fig. 4c,d).

Lastly, administration of IL-2-anti-IL-2 complexes to Foxp3^CreERT2Cα^FL/FL mice did not diminish to any measurable extent the activation and lymphoproliferation of effector T cells caused by loss of TCR expression in T_reg cells (Supplementary Fig. 5). Conversely, IL-2 depletion did not further exacerbate autoimmunity (data not shown). Notably, this was the case in spite of a 1.5-fold expansion of TCR⁻, but not TCR⁺, T_reg cells following IL-2 administration (a likely consequence of higher CD25 expression and heightened IL-2 responsiveness by TCR⁻ T_reg cells) and a greater than two-fold reduction in TCR⁻ T_reg cells following IL-2 depletion (Fig. 4b and Supplementary Fig. 5b). These observations further confirm that TCR⁻ T_reg cells—even when present in elevated numbers—possess minimal suppressive capacity. Together, these results indicate that neither TCR-dependent interactions with antigen presenting cells nor continuous TCR-mediated localization within lymphoid organs are required for T_reg cell acquisition of IL-2, and that T_reg cell dysfunction in the absence of TCR cannot be attributed to altered IL-2R signaling.

TCR expression promotes T_reg cell adhesive properties in vitro

The in vitro suppressive capacity of T_reg cells requires TCR engagement, possibly involving pathways independent of Zap70 catalytic activity—which is essential for conventional T cell effector function—but dependent on membrane proximal inside-out activation of integrins and subsequent enhancement of T_reg cell interactions with antigen presenting cells^{8, 33, 34}. To test this possibility, we cultured TCR⁺ or TCR⁻ T_reg cells isolated from Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT or Foxp3^CreERT2Cα^FL/FL mice with DCs and assessed DC-T_reg cell conjugate formation. We did not detect any differences in conjugate formation between DCs and TCR⁺ or TCR⁻ T_reg cells following 30 min of incubation (data not shown). However, TCR⁻ T_reg cells isolated from Foxp3^CreERT2Cα^FL/WT or Foxp3^CreERT2Cα^FL/FL mice were less efficient than TCR⁺ T_reg cells at forming conjugates with DCs (3.7% vs. 7.5% of Treg cells) following overnight culture (Supplementary Fig. 6a). Conjugate formation was unaffected by the presence or absence of MHC class II molecules on DCs, which may imply that the increased adhesion of TCR⁺ vs. TCR⁻ T_reg cells in this assay was not a result of MHC class II-TCR interactions and may instead have been a consequence of the overall heightened activation status of TCR⁺ vs. TCR⁻ T_reg cells (Supplementary Fig. 6a). As LFA-1 expression is higher on CD44^hi T_reg cells in comparison to CD44^loCD62L^hi T_reg cells, it is possible that superior conjugate formation by TCR⁺ T_reg cells, which are enriched for CD44^hi cells compared to TCR⁻ T_reg cells, may be due, at least in part, to increased expression of this integrin (Supplementary Figure 6b). Although further work is necessary to determine precisely how TCR engagement in vivo affects signaling pathways to modulate T_reg cell adhesive properties, our results indicate that TCR expression contributes to optimal T_reg cell contact-dependent interactions with APCs, which may contribute to TCR-dependent immunosuppressive function.

TCR signaling modulates the effector T_reg cell transcriptional signature

In order to explore whether TCR signals, apart from influencing T_reg cell adhesion, might drive transcriptional events to license suppressor function in vivo, we conducted gene expression analysis of TCR⁺ and TCR⁻ T_reg cells. Flow cytometric analysis showed that loss of TCR expression had a stronger effect on effector-like CD44^hiCD62L^lo T_reg cells compared to naïve-like CD44^loCD62L^hi T_reg cells (data not shown) prompting us to investigate the gene expression profiles of these two populations separately within the TCR⁺ and TCR⁻ T_reg cell populations isolated from healthy Foxp3^CreERT2Cα^FL/WT mice (to avoid confounding effects of immune activation). We found 155 genes were significantly down-regulated, and only five genes were significantly up-regulated, by at least 2-fold in effector-like CD44^hiCD62L^lo TCR⁻ T_reg cells compared to CD44^hiCD62L^lo TCR⁺ T_reg cells (Fig. 5a). Sixteen genes were significantly down-regulated in the naïve-like CD44^loCD62L^hi TCR⁻ T_reg cells compared to CD44^loCD62L^hi TCR⁺ T_reg cells (all of them also down-regulated in CD44^hiCD62L^lo TCR⁻ vs. TCR⁺ T_reg cells), whereas one gene was up-regulated (Fig 5a). 535 genes showed higher expression (2-fold or greater) in effector-like CD44^hiCD62L^lo TCR⁺ T_reg cells compared to naïve-like CD44^loCD62L^hi TCR⁺ T_reg cells, and expression of 136 of them (25%) was TCR-dependent (Fig. 5b). Importantly, 127 of the 155 genes (82%) down-regulated in the absence of TCR in CD44^hiCD62L^lo T_reg cells showed at least 2-fold higher expression in effector-like vs. naïve-like T_reg cells.

Figure 5. TCR signaling maintains the effector T_reg cell transcriptional signature.

(a) Genes differentially expressed in CD44^hiCD62L^lo vs. CD44^loCD62L^hi TCRβ⁺ T_reg cells compared to genes differentially expressed in CD44^hiCD62L^lo TCRβ⁻ vs. CD44^hiCD62L^lo TCRβ⁺ Treg cells (two replicates, upper panel) or CD44^loCD62L^hi TCRβ⁻ vs. CD44^loCD62L^hi TCRβ⁺ Treg cells (three replicates, lower panel). The indicated cell subpopulations were sorted on day 14 from Foxp3^CreERT2Cα^FL/WT mice (five or more mice per replicate) treated with tamoxifen on days 0, 1 and 3 using an Aria II flow cytometer. Gene expression was analyzed using Affymetrix 430 2.0 mouse gene expression arrays. Genes down- or up-regulated in the absence of TCR are shown in red or blue, respectively. The number of genes significantly differentially expressed by 2-fold or more are indicated, q-value < 0.01. (b) Genes down-regulated 2-fold or more in CD44^hiCD62L^lo TCRβ⁻ vs. TCRβ⁺ T_reg cells (‘TCR’) among genes up-regulated 2-fold or more in CD44^hiCD62L^lo vs. CD44^loCD62L^hi T_reg cells (‘Up’) are shown as white portion of the black bar (left) and ‘Up’ genes among ‘TCR’ genes are shown as the gray portion of the white bar (right). (c) Cumulative distribution function plot of TCR-dependent genes vs. all genes differentially expressed in Foxp3^GFPKO vs. Foxp3⁺ CD4⁺ T cells. P < 10⁻²⁰, two-sample Kolmogorov-Smirnov test. (d) Treg cell TCR-dependent genes encoding transcription factors, cell surface, and secreted molecules.

Foxp3 has been proposed to solidify and amplify a transcriptional program initiated in T_reg cell precursors by TCR engagement^{9, 10, 35, 36}. We compared expression of the 155 genes identified above as being maintained by TCR in T_reg cells (which we call TCR-dependent genes) to Foxp3-dependent genes, identified as upregulated in wild-type T_reg cells compared to Foxp3^null-expressing T cells from Foxp3^GFPKO mice (which express a Foxp3 reporter null allele).³⁶ We found that a significantly enriched percentage of the TCR-dependent genes, compared to all genes, were also Foxp3-dependent (Fig. 5c). These observation suggest that the TCR-driven transcriptional program in T_reg cells is enhanced by Foxp3 expression, but that Foxp3 alone is not sufficient to maintain the full effector T_reg cell transcriptional signature.

Examination of the TCR-dependent genes identified several transcription factors that were upregulated in TCR⁺ compared to TCR⁻ CD44^hiCD62L^lo T_reg cells as well as compared to TCR⁺ and TCR⁻ CD44^loCD62L^hi T_reg cells, including NFATc1, c-Rel, Bcl6 and IRF4 the latter two of which were previously shown to be important for T_reg cell effector differentiation and function (Fig 5d).^37–39 Of the 155 TCR-dependent genes, we identified only one gene encoding an adhesion molecule, Vcam1, which was similarly upregulated in CD44^hiCD62L^lo T_reg cells compare to CD44^loCD62L^hi T_reg cells in a manner than depended upon TCR expression (Fig. 5d). Several genes encoding potential effector molecules were also found to be upregulated in CD44^hiCD62L^lo vs. CD44^loCD62L^hi T_reg cells in a TCR-dependent manner, including IL-1R2, a decoy receptor for IL-1, and the immune inhibitory molecules CD83, CD200, and LAG-3, as well as IL-10 and EBI3, a subunit of the cytokines IL-27 and IL-35, all of which have been implicated in T_reg cell function^{26, 40–45}. In addition, the chemokine-encoding genes Cxcl10 and Ccl1, as well as Ccr8 encoding the receptor for CCL1, were significantly down-regulated in TCR⁻ compared to TCR⁺ T_reg cells, suggesting that T_reg cells may signal each other through the expression of chemokines and their corresponding receptors, or may recruit into close proximity the targets of their suppressive acitivity⁴⁶.

All together, these data indicate that, under physiologic conditions, a substantial portion of the effector—but not the naïve-like—T_reg cell transcriptional program characterized by elevated expression of several potential T_reg cell effector molecules, is maintained by continuous TCR signaling.

IRF4 expression promotes T_reg cell function and homeostasis

To begin to assess the importance of the TCR-dependent transcriptional program for continuous T_reg cell function in vivo, we focused on IRF4 as a downstream target of the TCR signaling pathway in T_reg cells. We confirmed that elevated IRF4 expression in T_reg cells was restricted to CD44^hi cells and was reduced to basal levels upon TCR ablation in both Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice (Supplementary Fig. 7a). Foxp3^YFP-CreIrf4^FL/- mice, in which IRF4 is constitutively deleted in T_reg cells, have been shown to develop a severe T_H2 cytokine-dominated autoimmunity by 8 weeks of age³⁷. T_reg cells in Irf4^−/− mice were previously demonstrated to have an almost exclusively naïve-like phenotype, and we similarly found that T_reg cells in Foxp3^YFP-CreIrf4^FL/FL mice were largely CD44^loCD62L^hi even in the context of severe inflammation (data not shown), suggesting impaired differentiation, survival and/or expansion of effector T_reg cells³⁸. In order to determine whether fully differentiated T_reg cells require IRF4 expression downstream of TCR engagement for their in vivo suppressive function, we administered tamoxifen on days 0, 3, 7 and 10 to Foxp3^CreERT2Irf4^FL/FL and Foxp3^CreERT2Irf4^WT/WT littermates⁴⁷. On day 13, we observed a slight but reproducible decrease in IRF4 protein expression in lymph node CD44^hi T_reg cells in Foxp3^CreERT2Irf4^FL/FL mice compared to Foxp3^CreERT2Irf4^WT/WT mice, and an increase in GFP expression in the T_reg cells as a sum of fluorescence of the Irf4-deletion GFP reporter (whose expression is switched on in the Irf4 locus upon Cre-mediated deletion of the Irf4^FL allele) and the GFP-CreERT2 fusion protein expressed concomitantly with Foxp3 (Fig. 6a and Supplementary Fig. 7b. qPCR analysis indicated a ~50% reduction in Irf4 mRNA transcript amount in T_reg cells sorted from pooled spleens and lymph nodes of tamoxifen-treated Foxp3^CreERT2Irf4^FL/FL compared to Foxp3^CreERT2Irf4^WT/WT mice (Fig. 6a).

Figure 6. IRF4 expression contributes to optimal T_reg cell suppressive capacity and homeostasis.

(a) IRF4 expression in lymph node CD44^hiCD62L^lo (‘CD44^hi’) and CD44^loCD62L^hi (‘CD62L^hi’) CD4⁺Foxp3⁺ cells in Foxp3^CreERT2Irf4^WT/WT and Foxp3^CreERT2Irf4^FL/FL mice (left) and qPCR analysis of IRF4 mRNA transcript amounts in CD4⁺eGFP⁻ and CD4⁺eGFP⁺ cells sorted from pooled spleens and lymph nodes (right). qPCR is representative of two experiments with four or more mice per group each (b) IRF4 expression in colonic lamina propria CD4⁺Foxp3⁺ cells in Foxp3^CreERT2Irf4^WT/WT and Foxp3^CreERT2Irf4^FL/FL mice (left); gray histogram represents CD4⁺Foxp3⁻ cells. Percent Foxp3⁺ among CD4⁺ cells in the large intestine lamina propria in Foxp3^CreERT2Cα^WT/WT and Foxp3^CreERT2Irf4^WT/WT (white circles) and Foxp3^CreERT2Cα^FL/FL and Foxp3^CreERT2Irf4^FL/FL (black circles) mice, as indicated (right). (c,d) Numbers, percent CD44^hi, percent Ki67⁺ (c) and percent cytokine-producing cells (d) of CD4⁺Foxp3⁻ T cells in lymph nodes (c) and spleens (d) of Foxp3^CreERT2Cα^WT/WT (white circles), Foxp3^CreERT2Irf4^WT/WT (white squares), Foxp3^CreERT2Cα^FL/FL (black circles), and Foxp3^CreERT2Irf4^FL/FL (black squares) mice. All mice were analyzed on day 13 following tamoxifen treatment on days 0, 3, 7 and 10. Histograms (a,b) are representative of two experiments with four or more mice per group each. All other data in (b–d) is combined from two experiments with four or more mice per group each. ***, P < 0.001; **, P < 0.01; *, P < 0.05. P-values were calculated using an unpaired t-test.

We hypothesized that the suboptimal ~50% reduction in mRNA amount and the only slight reduction in IRF4 protein expression in the spleens and lymph nodes of Foxp3^CreERT2Irf4^FL/FL compared to Foxp3^CreERT2Irf4^WT/WT mice might have resulted from a competitive disadvantage of T_reg cells lacking IRF4 protein, which might lead to preferential expansion of IRF4-sufficient T_reg cells remaining in Foxp3^CreERT2Irf4^FL/FL mice. This would be consistent with the reported observation that the survival and expansion of strongly antigen-stimulated CD8⁺ T cells was greatly impaired in the absence of IRF4⁴⁸. In order to assess whether IRF4 might similarly contribute to the maintenance of T_reg cells that have been strongly activated, we looked in the colonic lamina propria, in which nearly all T_reg cells were CD44^hi and likely had recently experienced TCR engagement based on robust IL-10 production (data not shown). Indeed, we observed decreased percentages of colonic lamina propria T_reg cells in Foxp3^CreERT2Irf4^FL/FL compared to Foxp3^CreERT2Irf4^WT/WT mice, similar to the decrease in T_reg cell percentages seen in the colonic lamina propria of Foxp3^CreERT2Cα^FL/FL compared to Foxp3^CreERT2Cα^WT/WT mice (Fig. 6b). As opposed to spleen and lymph node T_reg cells, and consistent with their decreased percentages, colonic lamina propria T_reg cells in Foxp3^CreERT2Irf4^FL/FL mice exhibited a substantial reduction in IRF4 protein expression compared to T_reg cells in Foxp3^CreERT2Irf4^WT/WT mice (Fig. 6b). We hypothesized that the reportedly low influx of circulating T_reg cells into the colonic lamina propria at steady state may have precluded IRF4-sufficient cells from becoming activated and repopulating to wild type percentages the T_reg cell niche in this tissue²⁴.

Despite the only modest decrease in IRF4 expression in T_reg cells isolated from the spleens and lymph nodes of Foxp3^CreERT2Irf4^FL/FL mice, we were able to detect a very mild, but statistically significant increase in the percentage of CD44^hi and Ki67⁺ lymph node Foxp3⁻CD4⁺ T cells as well as increased IFN-γ, IL-4 and IL-13 production by splenic Foxp3⁻CD4⁺ T cells, suggesting that IRF4 expression downstream of TCR signaling in T_reg cells contributes to T_reg cells suppressive function (Fig. 6c,d). An increase in the production of T_H2 cytokines was consistent with the phenotype of mice with constitutive ablation of IRF4 in T_reg cells, whereas the increased IFN-γ was likely a consequence of the substantial T_H1 bias in the C56B/L6 adult mice prior to induced Irf4 deletion³⁷. As a control, we confirmed that in the absence of tamoxifen treatment, CD4⁺Foxp3⁻ T cell expression of CD44, Ki67, IFN-γ, IL-4 and IL-13 in Foxp3^CreERT2Irf4^FL/FL and Foxp3^CreERT2Irf4^WT/WT mice were indistinguishable, as were T_reg cell percentages in the colonic lamina propria (data not shown), suggesting that the modest differences we observed between Foxp3^CreERT2Irf4^FL/FL and Foxp3^CreERT2Irf4^WT/WT mice upon tamoxifen treatment were not a consequence of the Irf4^FL allele itself. Together, these data indicate that even partial loss of IRF4 expression downstream of TCR engagement in T_reg cells interferes with optimal suppressive function of these cells.

Discussion

Despite major progress in understanding the molecular mechanisms of TCR engagement-driven differentiation of T_reg cells, the role of the TCR in T_reg cell function in vivo has remained unclear. Here, we demonstrate that TCR signaling in differentiated T_reg cells is dispensable for the maintenance of Foxp3 expression and for expression of many T_reg cell characteristic markers. Although the bulk of the T_reg cell-specific gene signature was also preserved in the absence of TCR, suppressor function was critically dependent on the TCR.

Given that antigen-activated CD4⁺Foxp3⁻ T cells in lymphoid organs are thought to produce IL-2 in a spatially restricted manner, we considered the possibility that T_reg cells might analogously require their TCRs to correctly position themselves to gain preferential access to IL-2, which might elicit their suppressive function by stimulating IL-2R. However, our in vivo and in vitro data suggests that T_reg cells acquire and likely effectively deplete IL-2 in a TCR-independent manner, and thus may instead rely predominantly on CCR7 expression to ensure sufficient IL-2 exposure, as recently proposed²⁴.

Our observation that newly generated T_reg cells in Foxp3^CreCα^FL/WT and Foxp3^CreCα^FL/FL mice remained naïve-like upon loss of TCR and did not populate non-lymphoid tissues suggests that effector differentiation and expansion are TCR- and likely antigen-dependent processes. This finding may help explain the observation that distinct T_reg cell TCR repertoires are displayed by T_reg cells populations found in distinct lymphoid organs and tissues in adult mice^{49, 50}. Furthermore, the fact that inducible ablation of the TCR resulted in a far more pronounced change in gene expression in the effector-like vs. the naïve-like T_reg cell subset suggests that continuous TCR signaling may be selectively driving the homeostasis and suppressor function of effector-like T_reg cells. As inducible deletion of the TCR in differentiated T_reg cells precipitated autoimmunity, our data may suggest that all—or the bulk—of T_reg cell suppressor function in vivo is mediated by the CD44^hi effector-like T_reg cell subset.

We found that partial inducible ablation of Irf4, expressed downstream of TCR engagement in CD44^hi T_reg cells, resulted in a very mild, but highly reproducible immune activation. Although this result suggests that IRF4 expression is important for TCR-dependent Treg cell function, given the suboptimal Irf4 deletion and modest immune activation, to what extent and how TCR-dependent IRF4 induction in T_reg cells contributes to their ability to suppress spontaneous autoimmunity remains to be determined. IRF4 may predominantly function to control maintenance of highly activated T_reg cells, which was particularly evident in the colonic lamina propria and which, when altered, may affect the ability of the T_reg cell pool to suppress. Alternatively, IRF4 may play a more direct role in promoting T_reg cell suppressive activity, perhaps by driving expression of certain T_reg cell effector molecules. Although we observed decreased ICOS expression on colonic T_reg cells lacking IRF4, the remaining IRF4-sufficient T_reg cells present in lymphoid organs of Foxp3^CreERT2Irf4^FL/FL mice complicate a more rigorous identification of IRF4 targets (data not shown).

It is important to note, however, that the fact that naïve-like T_reg cells do not express IRF4 and are overwhelmingly unaffected by TCR deletion on a transcriptional level is not necessarily an indication that these cells are non-functional or that they are not experiencing TCR engagement. Indeed, several genes including Egr1, Egr2 and Nr4a1 were found to be down-regulated in this T_reg cell subset in the absence of TCR.

Future experiments will be necessary to elucidate the contributions of other individual T_reg cell TCR-dependent genes to maintenance of immune tolerance in the steady state, and to the restraint of immune responses directed against commensal microorganisms, food and environmental antigens and pathogens. In this regard, it is noteworthy that while IL-10 production by T_reg cells has been implicated in the control of inflammatory responses at environmental interfaces such as the gut, lungs and skin, it has also been shown to be dispensable for T_reg cell control over systemic autoimmunity²⁶. Likewise, we found that while constitutive deletion of calcineurin B1 in Foxp3^CreCnb1^FL/FL mice (which eliminated calcineurin-dependent NFAT activation in T_reg cells) resulted in lethal early onset autoimmunity, highly efficient inducible deletion in adult lymphoreplete Foxp3^CreERT2Cnb1^FL/FL mice had no detectable adverse consequences on T_reg cell function (data not shown). Together, these findings suggest that TCR engagement on T_reg cells may drive a focused transcriptional program, select aspects of which are required in a context dependent manner for mediation of a broad range of T_reg cell immunosuppressive functions.

Online Methods

Mice

Mice were bred and housed in the pathogen-free animal facility at Memorial Sloan-Kettering Cancer Center and were used in accordance with institutional guidelines. Unless otherwise noted, 8–10 week old mice were used for all experiments. Foxp3^YFP-Cre, Foxp3^eGFP-CreERT2, Rosa26^YFP, Foxp3^DTR, and Irf4^FL mice have been previously described^{19, 26, 30, 47}. We thank Marc Schmidt-Supprian and Klaus Rajewsky for kindly providing Cα^FL mice. For tamoxifen administration, 40mg tamoxifen were dissolved in 100ul ethanol and subsequently in 900uL olive oil (Sigma-Aldrich) and sonicated 4 × 30 seconds in a Bioruptor Twin (Diagenode). Mice were orally gavaged with 200 µl tamoxifen emulsion per treatment. For diphtheria toxin (DT) injections, DT was dissolved in PBS and 200 µl of indicated doses were injected i.p per mouse. For in vivo IL-2 depletion, mice were gavaged with tamoxifen on days 0 and 1 or on days 0, 3, 7 and 10 and injected i.p. on days 4 and 8 with a 0.5mg 1:1 mix of IL-2 neutralizing JES6- 1A12 and S4B6-1 antibodies or IgG2a isotype (BioX-Cell). For administration of IL-2-anti-IL2 complexes, 1ug recombinant mouse IL-2 (R&D Systems) was incubated for 10 minutes at room temperature with 5 µg JES6-1 (BioXCell) and diluted to 200 µl in PBS immediately before i.p injection. Mice were gavaged with taxomofen on days 0, 3, 7 and 10 and received IL-2-anti-IL-2 complexes or PBS on days 5, 7, 9 and 11.

FACS staining and cell isolation

Cells were stained with LIVE/DEAD Fixable Yellow Dead Cell Stain (Molecular Probes) and antibodies listed in Supplementary Table 1. For BrdU experiments, mice were injected i.p. with 1mg BrdU in 1mL PBS and staining was performed using the BD Pharmingen BrdU Flow Kit. Flow cytometric analysis was performed using an LSRII flow cytometer (BD Bioscience) and FlowJo software (Tree Star). For cell isolation, CD4⁺ T cells were purified from pooled spleen and lymph node cell suspensions using magnetic Dynabeads (Invitrogen) and further sorted using an Aria II cell sorter (BD Bioscience). Intracellular staining was performed using eBioscience Fixation Permeabilization buffers. For cytokine staining lymph node and spleen cells were stimulated with soluble anti-CD3 clone 2C11 (5ug/ml) and anti-CD28 clone 37.51 (5ug/ml) in the presence of 1ug/mL brefeldin A for 4–6 hours at 37°C, 5% CO².

In vitro proliferation assay

Dendritic cells (DCs) were expanded in vivo by subcutaneous injection of B16 melanoma cells secreting Flt3 ligand into the left hind flank of B6 mice. Once tumors were visible, spleens from injected animals were dissociated in RPMI 1640 medium containing 1.67 units/mL liberase TL (Roche) and 50 µg/mL DNAse I (Roche) for 20 min at 37 °C with shaking. EDTA was added at a final concentration of 5 mM to stop digestion and the resulting homogenate was processed for CD11c⁺ cell isolation using the MACS mouse CD11c (N418) purification kit (Miltenyi Biotec). FACS purified CD4⁺eGFP⁺ cells from tamoxifen treated Foxp3^CreERT2Cα^FL/WT mice were labeled with CellTrace Violet (Molecular Probes) according to the manufacturer’s instructions and plated in triplicate in 96-well flat bottom plates (5 × 10⁴ cells/well) in media containing 25 U/mL human recombinant IL-2 (PeproTech) with or without equal numbers of DCs, with or without 100ng/mL LPS (E. coli 0111:B4, Sigma-Aldrich.)

In vitro IL-2 stimulation and pSTAT5 detection

Foxp3^CreERT2Cα^FL/WT mice were treated with tamoxifen on days 0 and 1, and on day 9 CD4⁺ T cells were purified from pooled spleen and lymph node cell suspensions using magnetic Dynabeads (Invitrogen). Cells were stained with anti-CD4 and anti-TCRβ, washed, and plated in 96-well V-bottom plates (1 × 10⁶ cells/well) in RPMI containing 10% FBS with or without increasing concentrations of IL-2 for 20 minutes at 37oC. Cells were subsequently processed using BD Phosflow Lyse/Fix Buffer and Perm Buffer III (BD Biosciences) and stained with anti-p-STAT5 Y694 antibody according to the manufacturer’s instructions. T^reg cells were identified by eGFP expression. For ex vivo p-STAT5 staining, spleen and lymph nodes were dissociated at 4°C in PBS (0.5% BSA) containing anti-CD4 and anti-TCRβ antibodies, stained for 10 minutes on ice, and washed twice before fixation.

In vitro IL-2 depletion

Foxp3^CreERT2Cα^FL/WT mice were treated with tamoxifen on days 0 and 1, and on day 9 pooled spleens and lymph nodes were enriched for CD4⁺ cells and subsequently sorted to >99% purity into eGFP⁺TCRβ⁺, eGFP⁺TCRβ⁻, and eGFP⁻TCRβ⁺ populations. Each population was divided among 8 wells of a V-bottom 96-well plate (250,000 cells/well) in 25uL RPMI(10%) with or without increasing doses of recombinant human IL-2 for 2 hours at 37°C. IL-2 depletion from the media was assessed using the BD Cytometric Bead Array and Human IL-2 Enhanced Sensitivity Flex Set (BD Biosciences) according to the manufacturer’s instructions.

In vitro conjugation assay

Foxp3^CreERT2Cα^WT/WT, Foxp3^CreERT2Cα^FL/WT, and Foxp3^CreERT2Cα^FL/FL mice were treated with tamoxifen on days 0 and 1 and on day 9, CD4⁺ T cells were isolated from spleen and lymph nodes using the Dynabeads Untouched Mouse CD4 Cells negative selection kit (Invitrogen.) TCR⁺ T^reg cells were sorted from Foxp3^CreERT2Cα^WT/WT mice based on eGFP expression alone. TCR⁻ T^reg cells were sorted from Foxp3^CreERT2Cα^FL/WT and Foxp3^CreERT2Cα^FL/FL mice as eGFP⁺ TCRβ⁻. T^reg cells were subsequently labeled with CFSE and Flt3L-expanded DCs from Ab (MHC class II)^+/+ and Ab (MHC class II)^−/− mice were labeled with CellTrace Violet. 10⁴ T^reg cells and 6 × 10⁴ DCs were cultured together in a 96-well round bottom plate in RPMI 10% supplemented with 500 U/mL IL-2. Concanavalin A was used at a final concentration of 2.5 ug/mL. Following a 10 hr culture at 37°C, cells were gently resuspended before flow cytometric analysis.

Gene expression analysis

eGFP⁺ TCRβ⁺ and TCRβ⁻ CD44^hiCD62L^lo (two replicates) and CD44^loCD62L^hi (three replicates) cell populations were sorted from tamoxifen-treated Foxp3^CreERT2Cα^FL/WT mice (five or more mice per replicate) using an Aria II flow cytometer. Complementary DNA (cDNA) libraries were amplified and hybridized to Affymetrix 430 2.0 chips. Arrays were normalized using RMA, and genes were considered differentially expressed if they had a q value <0.01 after Benjamini-Hochberg FDR correction. Differential gene expression in T^reg cells from Foxp3^GFPKO vs. Foxp3⁺ mice has been previously described.