BALB/c, C57BL/6, Bim^-/- mice and Scurfy mice were purchased from Jackson Laboratories. 129 Foxp3-eGFP transgenic mice were purchased from Taconic farms. CB-17 scid mice were also purchased from Charles River Laboratories. All mice were maintained in NIAID animal facility and cared for in accordance with institutional guidelines.

Purified anti-CD3 (145-2C11), purified anti-CD28 (37.51), anti-CD25 (3C7), biotin-conjugated anti-CD25 (7D4), fluorescein isothiocyanate-conjugated anti-CD4 (GK1.5), phycoerythrin-conjugated anti-CD25 (PC61), unconjugated, allophycocyanin- or phycoerythrin-conjugated anti-IL-2 (JES6-5H4), anti-IL-4 (11B11) and anti-IL-4Rα (mIL4R-M1) are from BD Biosciences. Anti-Foxp3 (FJK-16S) and anti-IL-7Rα (A7R34) are from eBiosciences. The anti-FITC Multisort kit, IL-2 secretion assay kit and anti-biotin microbeads were from Miltenyi Biotec. The IL-2 Quantikine enzyme-linked immunosorbent assay (ELISA) kit and recombinant mouse IL-2, IL-7, IL-4, IL-15 and IL-21 were purchased from R&D Systems. Cell cultures were performed in complete RPMI 1640 medium (Bio-Whittaker) supplemented with 10% (vol/vol) FCS, 100 U/ml of penicillin, 100 μg/ml of streptomycin, 2 mM glutamine, 10 mM HEPES, 1 mM sodium pyruvate and 50 μM β-mercaptoethanol.

Splenocytes were harvested from 5 to 12 week old mice. Erythrocytes were osmotically-lysed using Ack lysing buffer (Bio Whitaker) and single cell suspensions were incubated with FITC-conjugated anti-CD4 and biotin-conjugated anti-CD25 followed by incubation with anti-FITC microbeads. CD4⁺ T cells were then purified by magnetic isolation using the Auto MACS sorter (Miltenyi Biotec). For isolation of CD4⁺CD25⁺ T_reg cells, after releasing the beads, the purified CD4⁺ T cell suspension was incubated with α-biotin microbeads followed by separation using the Auto MACS. In all the experiments 90 to 95% of these cells were positive for CD4 and CD25. The negative fractions were depleted of CD25⁺ cells to obtain CD4⁺CD25^- cells.

T_con cells (6 × 10⁴) or T_reg were cultured in U-bottom 96-well plates in the presence of soluble 0.75 μg/ml α-CD3 and 3 μg/ml α-CD28 for 3–4 days. Death was measured by flow cytometry after 3 days. For co-culture assays, CD4⁺CD25^- responder T cells (T_resp) (3 × 10⁴) were cultured in U-bottom 96-well plates with T_con (CD4⁺CD25^-) (3 × 10⁴) or T_reg (CD4⁺CD25⁺) in the presence of soluble 0.5–0.75 μg/ml α-CD3 and 3–4 μg/ml α-CD28 for 2–4 days. T_reg or T_con cells were used in the co-culture with responders directly in U-bottom 96-well plates. T_resp cells were CFSE-labeled to distinguish them from T_con or T_reg cells in co-culture. Proliferation was also assayed by CFSE dilution. Cell death analyses of CFSE⁺ responders were performed based on forward scatter and propidium iodide staining. All flow cytometry analyses assessing cell death were performed with events acquired at constant time, in order to count the events. The percentage of survival (Survival (%)) in all analyses is the percentage of cells that FSC^high and PI^-. When indicated, IL-2 (1000 U/ml), IL-7 (20 ng/ml), IL-4 (20 ng/ml), IL-15 (20 ng/ml), IL-21 (20 ng/ml) was added. For IL-2 blocking experiments, CD4⁺CD25^- cells were isolated and 6 × 10⁴ cells were stimulated with anti-CD3 and anti-CD28 in the presence of isotype control or cytokine blocking antibodies, 10 μg/ml each, and cultured in 96-well U-bottomed plate.

T_con or T_reg cells isolated from cultures were washed with PBS twice, fixed with fixation buffer containing Glutaraldehyde and Sodium Cacodylate. Fixed cells were pelleted and sent to electron microscopy facility at SAIC-Frederick, Inc. for imaging and analyses.

To examine the survival of T_reg cells in vitro, we used magnetically sorted CD4⁺CD25^-Foxp3^- T cells (T_con) or CD4⁺CD25⁺ (T_reg) cells and cultured with soluble anti-CD3 and anti-CD28 for 72–96 hours. We measured the frequency of surviving cells based on forward scatter and propidium iodide (PI) staining and flow cytometry analyses. There was a dramatic T_reg cell death (75–90%) in the absence of IL-2 and increasing doses of IL-2 rescued them in a concentration dependent manner (Fig. 1a, left panel). T_con cells survived well without exogenous IL-2 in the cultures, presumably because they produce IL-2 themselves (Fig. 1a, left panel). To test the effect of other γ_c cytokines, we added IL-2, IL-4, IL-7, IL-15 or IL-21 at 20 ng/ml concentration during the beginning of stimulation in T_reg cultures. We observed that the presence of the cytokines rescued T_reg death, with IL-2 having the strongest pro-survival function (Fig. 1a, right panel). On the other hand, IL-23 a non-γ_c cytokine did not have an effect on the survival of T_reg cells (Fig. 1a, right panel). Carboxyfluorescein succinimidyl ester (CFSE) labeling of the cells showed that in the absence of cytokines, few T_reg cells that remained in the culture did not undergo proliferation whereas T_con cells proliferated vigorously (Fig. 1b). IL-2 induced proliferation in T_reg cells whereas IL-4, IL-7, IL-15, IL-21 had minimal effect on proliferation even at the excessive concentration of 20 ng/ml (Fig. 1b, data not shown). However, the possibility that the combination of some or all γ_c cytokines such as IL-4, IL-7, IL-15 and IL-21 could initiate proliferation in Treg cells is not excluded. As a control for CFSE staining, un-stimulated T_reg cells isolated ex vivo is shown (Fig. 1b). Next, we hypothesized that T_con cells producing IL-2 might also serve as IL-2 source and support the survival and proliferation of T_reg cells in cultures. Therefore we stimulated the CFSE labeled T_reg cells with increasing numbers of CD4⁺ T cells and analyzed their death after 72 hours. As expected, T_reg cells died less and also proliferated in the presence of conventional CD4⁺ T cells (Fig. 1c). Even though conventional CD4⁺ T cells themselves died in the presence of T_reg cells at CD4:T_reg in 1:1 ratio, their viability was only mildly affected at 3:1 and 8:1 ratios. To test if T_con induced survival was mediated by IL-2, we blocked IL-2 using a blocking antibody and found that the survival frequencies of T_reg cells fell back to basal levels even in the presence of CD4⁺ T cells (Fig. 1d). However, the direct effect of blocking IL-2 in T_con cells cannot be ruled out. The number of proliferating T_reg cells was directly proportional to the numbers of viable CD4⁺ T cells (Fig. 1d). This is in accordance with our previous findings that T_reg cells consume cytokines from conventional CD4 T cells, and in the process, suppress them 8. Transmission electron microscopy and confocal microscopy analyses of T_regcells in the absence of cytokines showed condensed nuclei and membrane blebbing, the characteristic features of apoptosis (Fig. 1e and 1f). Thus, it is evident that T_reg cells undergo apoptosis in the absence of cytokine signaling in vitro. Our findings corroborate the observation that T_reg cells are absent in γ_c-knockout mice, implying that gamma chain signaling is important not only for the development but also for the survival of T_reg cells in the periphery. Our observation that gamma cytokines besides IL-2 can support the survival of T_reg cells potentially explains previous observation that shows only a reduced frequency and not a complete absence of Foxp3⁺ cells in CD25 deficient mice 38. In the absence of IL-2-R signaling, the combination of other γ_c cytokines could also induce both survival and homeostatic proliferation of T_reg cells in the periphery.

To determine if T_reg cells express the receptors for cytokines in addition to IL-2R-α CD25, splenocytes were isolated from Foxp3-eGFP mice and stained for cytokine receptors such as IL-4R-α (CD124), IL-7R-α (CD127) and IL-15R-α ex vivo. We found that T_reg express receptors for these cytokines (Figs. 2a,b,c,d). We could not detect IL-21R-α expression on T_con and T_reg cells, possibly due to the absence of good detecting antibody for mouse IL-21R-α (data not shown). We also found that T_reg cells showed an upregulation of chemokine receptors such as CXCR4, CCR5 and CCR7 up on α-CD3 and α-CD28 stimulation (Fig. 2e,f and 2g, upper and lower panels). These chemokine receptors on T_reg cells are likely important for the T_reg cells to be recruited to the chemokine and possibly also cytokine producing cells at the sites of inflammation.

To further assess the role of apoptosis in regulating T_reg cells in vivo, we determined the role of B cell lymphoma-2 (Bcl-2) protein in T_reg death. To this end, we measured the frequency of CD25⁺ Foxp3⁺ cells in Bcl-2 transgenic mice in CD45.1 background. We found that there was an increased percentage of T_reg cells in these mice (Fig. 3a). The specific role of Bcl-2 interacting member (Bim) protein in mediating cytokine deprivation apoptosis is well documented 39. Therefore, we tested the frequency of T_reg cells in Bim deficient mice. Surprisingly, we also found that there was an increased accumulation of T_reg cells in the spleens of Bim^-/- mice (Fig. 3b and 3c). Furthermore, when we stimulated Bim^-/- CD4⁺ T_reg cells in vitro, we found that they had increased resistance to death in the absence of cytokines (Fig. 3d). However, their suppressive phenotype remained intact in the absence of Bim. They induced a partial suppression in proliferation and cell death of CFSE labeled T_resp cells that were co-cultured with them as compared to those with T_con cells (Fig. 3e and 3f). In addition to causing cell death in CD4⁺ T_resp cells, both WT T_reg cells and BIM^-/- T_reg cells were able to induce Foxp3 in conventional CD4⁺ T cells in co-cultures (Fig. 3g). The relevance of Foxp3 induction in T_resp cells due to the presence of T_reg cells is unknown presently. However, in the presence of T_reg cells and IL-7, the frequency of induced Foxp3+ cells seems to be diminished, probably due to an increased proliferation of non Foxp3⁺ CD4⁺ T_resp cells as compared to induced Foxp3⁺ CD4⁺ T_resp cells in the presence of IL-7.

Because CD25 and Foxp3 are important for the function for T_reg cells, we tested the influence of gamma chain cytokines on the expression for these molecules. T_reg cells stimulated without any cytokine showed a substantial down-regulation of the CD25 expression whereas CD25 up-regulation was normal on T_con cells (Fig. 4a, two upper panels). In the presence of cytokines however, CD25 expression was maintained at high levels both on T_con cells and T_reg cells (Fig. 4a, two upper panels). CD25 down regulation was only partial on BIM^-/- T_reg cells in the absence of cytokines, but was further up-regulated in the presence of cytokines (Fig. 4a, two lower panels). We also found that in WT T_reg cells, Foxp3 was down regulated in the absence of cytokines whereas Foxp3 levels remained high in Bim^-/-T_reg cells upon TCR stimulation (Fig. 4b). Cytokines maintained high levels of Foxp3 expression both in WT and Bim^-/- T_reg cells (Fig. 4b). On the other hand, WT and Bim^-/- T_con population had only few Foxp3⁺ cells upon TCR stimulation, and was not up-regulated in the presence of cytokines. These findings demonstrate that cytokines are crucial not only for the survival but also for maintaining the cardinal features of T_reg cells i.e the expression of CD25 and Foxp3. We believe that T_reg cells lose the expression of Foxp3 due to the initiation of death signals in the absence of cytokines because it did not occur in the absence of death in Bim^-/- T_regcells.

Scurfy mice carry a mutation in Foxp3 and succumb to a fatal autoimmune syndrome. We tested whether a strong dependence of cytokines was a characteristic feature of all ex vivo isolated CD25⁺ cells in general. We found that these Scurfy mice harbored CD4⁺ CD25⁺ cells that are presumably activated CD4⁺ T cells owing to the autoimmune condition of the mice (Fig. 5a). However, the CD25^high cells that represent the T_reg population was reduced from 16% to 8% in CD4⁺ population as compared to WT mice (Fig. 5a). However, there was a complete absence of Foxp3⁺ cells in these mice (Fig. 5b). To investigate whether Foxp3 is important in T_reg cells for the extreme cytokine dependence for their survival, we isolated CD25⁺ cells from the Scurfy mice and tested their survival in the presence or in the absence of IL-2. We found that isolated Scurfy T_con cells, when stimulated in cultures had an impaired survival as compared to WT T_con cells. Surprisingly, however we found that Scurfy CD25⁺ cells survived as well as Scurfy T_con cells even in the absence of IL-2 whereas WT T_reg cells died substantially (Fig. 5c). Because of the lack of dependence on IL-2, we hypothesized that Scurfy CD25⁺ cells might not suppress other conventional T cells. To test this tenet, we co-cultured the CD25⁺ cells from WT mice or Scurfy mice with CFSE labeled CD4 T cells and measured the suppressive death after 3 days. Interestingly, we observed that responding T_resp cells underwent death in the presence of CD25⁺ cells from WT mice and not with CD25⁺ from Scurfy mice (Fig. 5d). It is likely that CD25⁺ cells from Scurfy mice produce cytokines due to the absence of Foxp3, which is why they do not depend on IL-2 added exogenously. To validate this theory, we stimulated WT T_reg cells and scurfy CD25⁺ cells with anti-CD3 and anti CD28 and measured IL-2 in the supernatants after 3 days. We found that while WT T_reg cells did not produce IL-2, Scurfy CD25⁺ cells produced as much cytokine as Scurfy CD4⁺ CD25^- cells approaching the level of cytokine produced by WT CD4⁺, CD25^- cells (Fig. 4e). This data is consistent with the previous observations showing suppressive effects of Foxp3 on IL-2 production 4041 and the effect of IL-2 on T_reg homeostasis 1142. Together, we show here that Foxp3 represses cytokine production in T_reg cells, which is why they are dependent on gamma cytokines from an external source for survival. Thus T_reg cells have a self-regulatory mechanism through which their inability to produce cytokines instruct them to depend on cytokines and without the cytokines, the T_reg cells are deleted.

Dr Avinash Bhandoola, University of Pennsylvania School of Medicine, Philadelphia

PA United States

I thought there was plenty of interesting new data in this work. I have a few very minor comments that should be simple to deal with, and do not need to be published. 1) I thought the abstract somewhat repetitious in places. It could be shortened. 2) The figure legends do not clearly explain 3b, particularly the bottom 2 panels. 3) I did not understand the relevance of the right-most panels in Fig. 3g (WT or Bim-/- Treg + IL-7), particularly when compared to the two preceding panels (WT or Bim-/- Treg). Is it referred to at all in the text, or otherwise explained?

Fred Ramsdell, Associate Director, Zymogenetics, Seattle, WA 98102

Overall, the manuscript is well-written and concise. The connection between g-c receptor signaling and apoptosis – and the distinction between these cytokines and proliferation – is a significant finding and generally well supported by the data. To date, the bulk of studies on Treg survival/activity and cytokines has focused on IL-2, and the extension to other gamma-c receptor using cytokines provides a more comprehensive analysis of this biology.

Whilst the experiments in Fig 5 are an interesting attempt to address the function of Foxp3 with respect to cytokine dependence, the conclusions are not fully supported by the data. The abstract states that in scurfy mice, "Foxp3 increases their (CD25+ cells) dependence on cytokines by suppressing cytokine production in Treg cells." Whilst Foxp3 does appear to directly suppress cytokine production, it does so in any T cell and the major effect of lack of Foxp3 in scurfy mice would appear to be the absence of the T_R lineage more broadly. Thus, this is not an appropriate way to test "whether Foxp3 is important in Treg cells for the extreme cytokine dependence for their survival" as these mice don't have Treg cells. Previous data has demonstrated that scurfy T cells do not express Foxp3, that they are CD25+ and that these cells do not have any Treg activity (Khattri, et. al.). In fact, these cells produce large amounts of IL-2 and many other cytokines. Importantly however, the cells remaining in scurfy mice do not appear to be in any way related to Treg cells. The data in the manuscript is consistent with data from Foxp3 transgenic mice in which Foxp3 levels are increased, but the actual number of Treg cells is decreased, as are their CD25 levels – perhaps due to Foxp3 inhibition of g-c derived (IL2 or other) secretion. This figure however does not seem necessary to me for the manuscript to be of interest.

Anne Cooke, University of Cambridge, Department of Pathology, Tennis Court Rd

Cambridge, CB21QP, United Kingdom

In this manuscript the authors have examined the role of common gamma chain (γc) cytokines in CD4⁺ CD25⁺ Foxp3⁺ T cell survival. They clearly show that cytokines other than IL-2 that signal through the γc prevent apoptosis of T reg cells following stimulation with αCD3 and αCD28. This provides a nice explanation for the presence of T reg in CD25 deficient mice. While addition of exogenous γc cytokines enabled T reg survival, they are proposed individually not to be as effective as IL-2. This reviewer was unclear whether all the cytokines had been titrated to determine efficacy. Were the cytokines titrated fully and were doses greater than 20 ng/ml used?