Critical Role of TGF-β and IL-2 Receptor Signaling in Foxp3 Induction by an Inhibitor of DNA Methylation

Under physiological conditions, CD4+ regulatory T (Treg) cells expressing the transcription factor Foxp3 are generated in the thymus [thymus-derived Foxp3+ Treg (tTregs) cells] and extrathymically at peripheral sites [peripherally induced Foxp3+ Treg (pTreg) cell], and both developmental subsets play non-redundant roles in maintaining self-tolerance throughout life. In addition, a variety of experimental in vitro and in vivo modalities can extrathymically elicit a Foxp3+ Treg cell phenotype in peripheral CD4+Foxp3− T cells, which has attracted much interest as an approach toward cell-based therapy in clinical settings of undesired immune responses. A particularly notable example is the in vitro induction of Foxp3 expression and Treg cell activity (iTreg cells) in initially naive CD4+Foxp3− T cells through T cell receptor (TCR) and IL-2R ligation, in the presence of exogenous TGF-β. Clinical application of Foxp3+ iTreg cells has been hampered by the fact that TGF-β-driven Foxp3 induction is not sufficient to fully recapitulate the epigenetic and transcriptional signature of in vivo induced Foxp3+ tTreg and pTreg cells, which includes the failure to imprint iTreg cells with stable Foxp3 expression. This hurdle can be potentially overcome by pharmacological interference with DNA methyltransferase activity and CpG methylation [e.g., by the cytosine nucleoside analog 5-aza-2′-deoxycytidine (5-aza-dC)] to stabilize TGF-β-induced Foxp3 expression and to promote a Foxp3+ iTreg cell phenotype even in the absence of added TGF-β. However, the molecular mechanisms of 5-aza-dC-mediated Foxp3+ iTreg cell generation have remained incompletely understood. Here, we show that in the absence of exogenously added TGF-β and IL-2, efficient 5-aza-dC-mediated Foxp3+ iTreg cell generation from TCR-stimulated CD4+Foxp3− T cells is critically dependent on TGF-βR and IL-2R signaling and that this process is driven by TGF-β and IL-2, which could either be FCS derived or produced by T cells on TCR stimulation. Overall, these findings contribute to our understanding of the molecular mechanisms underlying the process of Foxp3 induction and may provide a rational basis for generating phenotypically and functionally stable iTreg cells.

Under physiological conditions, CD4 + regulatory T (Treg) cells expressing the transcription factor Foxp3 are generated in the thymus [thymus-derived Foxp3 + Treg (tTregs) cells] and extrathymically at peripheral sites [peripherally induced Foxp3 + Treg (pTreg) cell], and both developmental subsets play non-redundant roles in maintaining self-tolerance throughout life. In addition, a variety of experimental in vitro and in vivo modalities can extrathymically elicit a Foxp3 + Treg cell phenotype in peripheral CD4 + Foxp3 − T cells, which has attracted much interest as an approach toward cell-based therapy in clinical settings of undesired immune responses. A particularly notable example is the in vitro induction of Foxp3 expression and Treg cell activity (iTreg cells) in initially naive CD4 + Foxp3 − T cells through T cell receptor (TCR) and IL-2R ligation, in the presence of exogenous TGF-β. Clinical application of Foxp3 + iTreg cells has been hampered by the fact that TGF-β-driven Foxp3 induction is not sufficient to fully recapitulate the epigenetic and transcriptional signature of in vivo induced Foxp3 + tTreg and pTreg cells, which includes the failure to imprint iTreg cells with stable Foxp3 expression. This hurdle can be potentially overcome by pharmacological interference with DNA methyltransferase activity and CpG methylation [e.g., by the cytosine nucleoside analog 5-aza-2′-deoxycytidine (5-aza-dC)] to stabilize TGF-β-induced Foxp3 expression and to promote a Foxp3 + iTreg cell phenotype even in the absence of added TGF-β. However, the molecular mechanisms of 5-aza-dC-mediated Foxp3 + iTreg cell generation have remained incompletely understood. Here, we show that in the absence of exogenously added TGF-β and IL-2, efficient 5-aza-dC-mediated Foxp3 + iTreg cell generation from TCR-stimulated CD4 + Foxp3 − T cells is critically dependent on TGF-βR and IL-2R signaling and that this process is driven by TGF-β and IL-2, which could either be FCS derived or produced by T cells on TCR stimulation. Overall, these findings contribute to our understanding of the molecular mechanisms underlying the process of Foxp3 induction and may provide a rational basis for generating phenotypically and functionally stable iTreg cells. Mechanisms of 5-Aza-dC-mediated Foxp3 Induction Frontiers in Immunology | www.frontiersin.org February 2018 | Volume 9 | Article 125 inTrODUcTiOn Peripherally induced Foxp3 + Treg (pTreg) cells, which are generated from precommitted CD4 + Foxp3 − CD25 + pTreg cell precursors (1,2), comprise 20-40% of the mature Treg cell pool in steady-state mice (3) and act in concert with thymus-derived Foxp3 + Treg cells (tTregs) to enforce immune tolerance (3)(4)(5)(6).
In addition to such naturally occurring pTreg cells that continuously develop in peripheral lymphoid tissues of nonmanipulated mice, Foxp3 + pTreg cells can be artificially generated in vivo from post-thymic, initially naive CD4 + Foxp3 − T cells in experimental settings of lymphopenia-driven proliferation (7,8) and subimmunogenic antigen administration (9,10). Early studies using CD25 as a surrogate Treg cell marker provided first evidence that CD4 + CD25 − T cells (11,12) can acquire a Treg cell phenotype in vitro [termed iTreg cells (13)] upon T cell receptor (TCR) stimulation in the presence of added TGF-β. After anti-Foxp3 mAbs and Foxp3-fluorochrome reporter mice became commonly available, numerous reports have extended the concept of TGF-β-/TCR-mediated Foxp3 + induction to truly naive CD4 + Foxp3 − T cells by rigorously excluding preformed Foxp3 + Treg cells. These studies established that the process of TGF-β-/TCR-mediated Foxp3 + iTreg cell generation is strictly dependent on IL-2R signaling and IL-2, which could either be exogenously added or produced by TCR-stimulated CD4 + T cells (14). Enhanced co-stimulation and inflammatory cytokines antagonize TGF-β-mediated Foxp3 induction, whereas retinoic acid augments this process through direct and indirect mechanisms (15), including interference with inhibitory cytokine signaling (16) and enhanced TGF-β-driven SMAD3 phosphorylation in developing iTreg cells (17), as well as inhibition of inflammatory cytokine secretion by bystander cells (18). Since then, TGF-β-mediated iTreg cell generation has been widely used as an experimental in vitro approach that recapitulates some aspects of Foxp3 + Treg cell development in vivo. However, TGF-β-mediated Foxp3 + iTreg cell generation in vitro fails to recapitulate the epigenetic (19)(20)(21)(22)(23) and transcriptional (24,25) signature of in vivo generated Foxp3 + Treg cells, which is reflected by variable suppressor activity (26) and unstable Foxp3 expression (19,20,27), precluding the clinical use of TGF-β-induced Foxp3 + iTreg cells.
A causal relationship between 5-aza-dC-mediated Foxp3-CNS2 demethylation and stabilization of induced Foxp3 expression has been firmly established (19-21, 32, 33), but it has remained unclear whether early events during 5-aza-dC-mediated Foxp3 + iTreg cell generation are also mediated through direct mechanisms (i.e., CpG demethylation of the Foxp3 gene) or by indirectly regulating signaling pathways that then promote Foxp3 induction. On the basis of the observation Mechanisms of 5-Aza-dC-mediated Foxp3 Induction Frontiers in Immunology | www.frontiersin.org February 2018 | Volume 9 | Article 125 that 5-aza-dC enhances surface expression of the IL-2Rα subunit CD25 and synergizes with TGF-β in promoting Foxp3 + iTreg cell generation, we hypothesized that 5-aza-dC may facilitate Foxp3 induction through TGF-βR and IL-2R signaling. Here, we show that TGF-βR and IL-2R signaling is indispensable for efficient 5-aza-dC-mediated Foxp3 + iTreg cell generation. Our results suggest that, in the absence of added cytokines, 5-aza-dC sensitizes converting CD4 + Foxp3 − T cells to undergo TGF-βR and IL-2R signaling-dependent Foxp3 upregulation driven by low amounts of FCS-/T cell-derived TGF-β and IL-2.

Mice
Mice with transgenic expression of GFP as a fusion protein with Foxp3 (Foxp3 GFP ) (43) and Foxp3 GFP mice that additionally expressed a dominant-negative TGF-βRII (dnTGF-βRII) (44) in CD4 + and CD8 + T cells (Foxp3 GFP × dnTGF-β-RII) were on the C57BL/6 CD45.1 background. All mice were housed and bred at the Animal Facility of the CRTD under specific pathogen-free conditions. All animal studies were performed in strict accordance with German Animal Welfare legislation. All protocols were approved by the Institutional Animal Welfare Officer (Tierschutzbeauftragter), and necessary licenses were obtained from the regional license granting body (Landesdirektion Dresden, Germany).

statistical analysis
Statistical significance was assessed using Prism 6 software (GraphPad) and t-test (unpaired) with Sidak-Bonferroni corrections for multiple comparisons for determination of statistical significance or two-way ANOVA (ordinary or repeated measures) and Tukey's multiple comparison test, as indicated. The formula used to calculate percent inhibition of Foxp3 + iTreg cell generation was [(Y − X)/X] × 100, where X is the proportion CD4 + T cells that exhibited Foxp3 expression under culture conditions for efficient Foxp3 + iTreg cell generation (i.e., TCR stimulation with either TGF-β or 5-aza-dC and expression of a wild-type TGF-βRII), and Y is the proportion CD4 + T cells that exhibited Foxp3 expression under inhibitory or enhancing conditions [i.e., TCR stimulation with TGF-β or 5-aza-dC, in the presence of anti-TGF-β mAb ( Figure 2D resUlTs interplay of 5-aza-dc with il-2r/TgF-βr signaling during Foxp3 + iTreg cell generation The generation of Foxp3 + iTreg cells from initially naive, TCRstimulated CD4 + Foxp3 − T cells in vitro has been shown to be strictly dependent on IL-2R and TGF-βR signaling (14). In the absence of exogenously added IL-2, production of IL-2 by TCRstimulated CD4 + Foxp3 GFP− T cells can drive TGF-β-mediated Foxp3 + iTreg cell generation in an autocrine manner, but results in a predominantly CD25 low phenotype ( Figure 1A), indicating limited IL-2 availability at day 3 of such cultures. In contrast, in the absence of added IL-2 and TGF-β, TCR stimulation in the presence of 5-aza-dC promotes the generation of Foxp3 GFP+ iTreg cells with a CD25 high phenotype ( Figure 1A). Titration experiments (5-aza-dC: 0.1, 0.25, 0.5, and 1.0 µM) revealed a positive correlation between 5-aza-dC concentrations and proportions of Foxp3 GFP+ iTreg cells (Figure 1B), whereas cell viability inversely correlated with the amount of added 5-aza-dC (Figures 1C,D). Importantly, the population size of Foxp3 GFP+ cells was found to be essentially identical, irrespective of whether live/dead cell discrimination was based on FSC/SSC gating alone (Figures 1A-D) or a viability dye (DAPI) was included (data not shown).
Mechanistically, the data presented in Figure 1 suggest that the mode of action of 5-aza-dC may involve the modulation of IL-2R and TGF-βR signaling, two signaling pathways with well-established and non-redundant roles in Foxp3 induction and Foxp3 + iTreg cell generation. Our hypothesis that 5-aza-dC may act through indirect mechanisms could be further corroborated by kinetics studies on the expression of Foxp3 mRNA (quantitative RT-PCR) and Foxp3 protein (Foxp3 GFP fluorescence) at early time points during 5-aza-dC-and TGFβ-driven Foxp3 + iTreg cell generation (Figures 1E-G). Overall, the results revealed a considerable delay in the kinetics of 5-aza-dC-induced expression of Foxp3 mRNA ( Figure 1E) and Foxp3 protein (Figures 1F,G) although the proportions of Foxp3 GFP+ iTreg cells were found to be largely comparable at 72 h (5-aza-dC: 35.2 ± 2.1%; TGF-β: 34.7 ± 1.0%). Specifically, while Foxp3 mRNA expression levels in ex vivo CD4 + Foxp3 GFP− T cells (CD25 − CD62L high ) were below the level of detection (data not shown), TCR stimulation in the presence of exogenous TGF-β resulted in low but clearly detectable amounts of Foxp3 mRNA as early as 18 h after initiation of cultures, which further increased over time (Figure 1E; blue triangles) and correlated well with Foxp3 GFP protein expression (Figures 1F,G). In contrast, during 5-aza-dC-mediated iTreg cell generation, Foxp3 mRNA became first detectable at 40 h ( Figure 1E; red circles), with expression levels at 48 h remaining 5.6-fold below those observed during TGF-β-mediated iTreg cell generation. Consistently, the kinetics of 5-aza-dC-induced Foxp3 GFP protein expression was substantially delayed, with undetectable protein expression levels at 24 and 30 h (Figures 1F,G).
Overall, the delayed kinetics of 5-aza-dC-mediated Foxp3 + iTreg cell generation is consistent with a scenario, in which 5-aza-dC promotes Foxp3 expression primarily through indirect mechanisms, such as the enhancement of IL-2R/TGF-βR signaling, rather than acting directly on the Foxp3 gene locus. Interestingly, quantitative RT-PCR interrogating the expression of genes encoding essential components of the TGF-βR signaling pathway indicated that, in striking contrast to TGFβ-supplemented cultures, 5-aza-dC fails to upregulate of Smad7 mRNA levels in TCR-stimulated CD4 + T cells ( Figure 1H). Thus, while TGF-β-mediated iTreg cell induction is associated with Smad7-mediated negative regulation of TGF-βR signaling, the lack of Smad7 mRNA up-regulation during 5-aza-dC-mediated iTreg cell induction may provide a molecular basis for sustained sensitivity of CD4 + T cells to TGF-β during the conversion process in the absence of added TGF-β.
impact of TgF-βr signaling on 5-aza-dc-Driven Foxp3 + iTreg cell generation To further dissect the role of TGF-βR signaling in 5-aza-dCmediated Foxp3 induction in vitro, we performed a series of loss-of-function studies that involved mAb-mediated TGF-β blockage as well as genetic and pharmacological abrogation of TGF-βR signaling.
Overall, the data presented in Figures 1-3 provide direct evidence of an essential role of TGF-β and TGF-βR signaling in 5-aza-dC-mediated Foxp3 + iTreg cell generation.

5-Aza-dC-Mediated Foxp3 Induction in TCR-
and IL-2R-Stimulated CD8 + T Cells 5-Aza-2′-deoxycytidine has been shown to promote Foxp3 expression and CNS2 hypomethylation in a murine CD8 + cytotoxic T cell line that normally lacks Foxp3 expression and exhibits a highly methylated CNS2 (41). The essential role of Dnmt activity and CpG methylation in repressing Foxp3 expression was further corroborated by the observation that genetic ablation of Dnmt1 facilitates the upregulation of Foxp3 expression in initially naive Foxp3 − T cells of both the CD4 and CD8 lineage in vitro, on TCR and IL-2R stimulation in the absence of added TGF-β (40). Here, we show that these observations can be recapitulated by pharmacological inhibition of CpG methylation in freshly isolated CD8 + T cells (Figure 5A; top: CD4 + T cells, bottom: CD8 + T cells). To enable a direct comparison of 5-aza-dC-driven Foxp3 induction in initially naive, TCR-stimulated CD4 + and CD8 + T cells, all cultures were supplemented with exogenous IL-2 (100 U/ml), as IL-2 represents an essential survival/growth factor for CD8 + T cells and highly pure populations of naive CD8 + T cells fail to produce adequate amounts of IL-2 under given TCR stimulatory conditions. In IL-2-supplemented CD8 + T cell stimulation cultures (Figure 5A, bottom), the frequencies of TGF-β-induced Foxp3 + T cells (24.9 ± 0.2%) were almost doubled by 5-aza-dC (42.5 ± 17.1%) and increased to 75.3 ± 2.9% upon co-administration of TGF-β and 5-aza-dC (Figure 5A), demonstrating that 5-aza-dC-mediated inhibition of CpG methylation is suitable to promote Foxp3 expression in both CD4 + and CD8 + T cells. However, we consistently observed significantly increased expression levels of CD25 upon 5-aza-dC-mediated Foxp3 induction in CD4 + T cells ( Figure 5B) but not in CD8 + T cells (Figure 5C), suggesting that 5-aza-dC is acting on IL-2 expression and/or IL-2R signaling, rather than directly acting on Cd25 gene expression. Furthermore, we did not observe any evidence for Foxp3 induction in LPS-stimulated CD19 + B cells in the presence of TGF-β and/or 5-aza-dC (data not shown), indicating that 5-aza-dC-mediated Foxp3 induction is restricted to the T cell lineage.

5-Aza-dC Enhances the Expression of CD25 and pSTAT5
The expression of CD25 on T cells is regulated by multiple mechanisms, including TCR signal strength, Foxp3 occupancy of regulatory regions within the Cd25 gene, and IL-2R signaling, with IL-2 providing a positive feedback loop that involves STAT5 binding to the Cd25 gene locus (50). Therefore, we next sought to correlate the expression of Foxp3, CD25, and pSTAT5 at early (24 h) and late (72 h) time points during TGF-β-and 5-aza-dCmediated Foxp3 + iTreg cell generation in the absence of added IL-2, i.e., experimental conditions under which IL-2 availability is limited (Figure 6). 24 h after initiation of naive CD4 + Foxp3 GFP− T cell stimulation cultures, expression of Foxp3 GFP protein remained largely undetectable (<3.0%, data not shown), but essentially all CD4 + T cells already upregulated CD25 expression (Figures 6A-C), irrespective of whether TCR stimulation cultures were left untreated (gray histograms) or supplemented with either TGF-β (blue histograms) or 5-aza-dC (red histograms); at this early time point, the majority of CD4 + T cells that had been TCR stimulated in the presence of TGF-β exhibited high expression levels of pSTAT5 (Figures 6D,F), while the pSTAT5 expression pattern of TCR-/5-aza-dC-stimulated T cells closely resembled that of untreated T cell stimulation cultures. After 72 h, and consistent with Foxp3 promoter occupancy stabilizing and amplifying Cd25 gene expression (51), TGF-β-induced Foxp3 GFP expression correlated with high expression levels of CD25 ( Figure 6B, blue histograms) and pSTAT5 ( Figure 6E, blue histograms), while CD4 + Foxp3 GFP− T cells exhibited somewhat lower CD25 expression levels and only limited STAT5 phosphorylation; in striking contrast, TCR stimulation in the presence of 5-aza-dC promoted high expression levels of CD25 ( Figure 6B, red histograms) and

DiscUssiOn
Here, we provide evidence that in the absence of added TGF-β and IL-2, initiation of Foxp3 expression during 5-aza-dC-mediated Foxp3 + iTreg cell generation from TCR-stimulated, initially naive CD4 + Foxp3 − T cells critically depends on TGF-βR and IL-2R signaling, two signaling pathways with well-established roles in Foxp3 + iTreg cell generation in vitro and pTreg cell generation in vivo. Our data are consistent with the interpretation that 5-aza-dC initially facilitates stochastic Foxp3 induction by the enhancement of TGF-βR and IL-2R signaling under experimental conditions of limited TGF-β and IL-2 availability, elucidating an additional mode of action of 5-aza-dC beyond the direct modulation of CNS2 CpG methylation within the Foxp3 gene locus. Concomitant 5-aza-dC-mediated CNS2 hypomethylation may further stabilize induced Foxp3 expression by transcription factor recruitment, resulting in the accumulation of iTreg cells with a stable Foxp3 + phenotype. Considering the dose-dependent cytotoxicity of 5-aza-dC (Figures 1C,D), the relatively high proportions of Foxp3 + iTreg cells at later time points of TCR stimulation cultures are likely to reflect a selective survival advantage of cells that succeeded in acquiring a CD25 high Foxp3 high phenotype. Consistent with this interpretation, several previous observations argue against a scenario, in which pharmacological Dnmt inhibition bypasses essential signaling pathways implicated in the process of extrathymic Foxp3 induction and Foxp3 + iTreg cell generation by directly modulating the methylation status of Foxp3 CpG elements. First, studies using CD4 + T cells from gene-targeted mice established a non-redundant role of CNS1 in extrathymic Foxp3 induction (5,6,29,30), whereas the CpG-rich CNS2, which represents the target of 5-aza-dC-mediated hypomethylation, was found to be dispensable in this process (20,32,33). Second, TCR-stimulated CD4 + effector/memory cells are largely refractory to TGF-β-/IL-2-mediated induction of Foxp3 expression, and 5-aza-dC fails to abrogate this resistance (21), indicating that other signaling pathways apart from CpG demethylation play Mechanisms of 5-Aza-dC-mediated Foxp3 Induction Frontiers in Immunology | www.frontiersin.org February 2018 | Volume 9 | Article 125 an essential role in 5-aza-dC-mediated Foxp3 induction. Third, in addition to CD4 + T cells, 5-aza-dC can drive aberrant Foxp3 expression in non-CD4 + T cells, such as NK cells (52) and CD8 + T cells (Figure 5), but rather than indiscriminately activating many loci, 5-aza-dC appears selective in that it affects Foxp3 gene expression and a limited set of Treg cell signature proteins (such as CD25 and GITR) (21,53). Notably, on TCR stimulation in the presence of 5-aza-dC, Foxp3 mRNA expression and CNS2 demethylation were observed only in the fraction of CD4 + T cells that expressed Foxp3 protein (20), arguing against overall dysregulated gene expression by 5-aza-dC. Thus, it appears that the impact of 5-aza-dC on DNA methylation is unexpectedly selective, which could at least in part be attributed to the inverse relationship between 5-aza-dC-mediated Foxp3 induction and cell death (20) (Figure 1), which may result in the loss of cells with severely dysregulated gene expression and preferential survival of Foxp3 + iTreg cells that maintained largely unperturbed gene expression. Consistently, murine Foxp3 + iTreg cells that had been generated by TCR stimulation in the presence of TGF-β in combination with 5-aza-dC have been shown to exhibit potent suppressor function both in vitro and upon adoptive transfer in vivo (21). Previous studies employing mice with constitutive Foxp3-CNS2 deficiency (Foxp3Δ CNS2 ) highlighted the important role of CNS2 in maintaining Foxp3 expression and thus Treg cell identity in vivo and provided direct evidence that efficient 5-aza-dC-mediated Foxp3 + iTreg cell generation in vitro requires CNS2 (33): under suboptimal iTreg cell differentiation conditions (i.e., addition of 5-aza-dC after 1 day of TCR stimulation for the duration of 24 h, rather than addition of 5-aza-dC at the start until the end of TCR stimulation cultures), up to 15% of CNS2-proficient CD4 + T cells with an initially naive Foxp3 GFP− CD62L high CD44 low phenotype acquired Foxp3 GFP expression, whereas 5-aza-dC-mediated Foxp3 GFP induction was completely abrogated in Foxp3Δ CNS2 CD4 + T cells. Thus, it appears that 5-aza-dC-mediated Foxp3 induction and Foxp3 + iTreg cell generation are initiated by low amounts of FCS-/T cell-derived TGF-β/IL-2 and enhanced TGF-βR/IL-2R signaling, while developmental progression is critically dependent on the recruitment of key transcription factors to CNS2. In fact, enforced 5-aza-dCmediated CNS2 demethylation in viable CD4 + T cells was found to be sufficient for CNS2 binding of CREB/ATF (41) and Ets-1 (54). Clearly, future studies are warranted to determine the exact role of CNS2 in 5-aza-dC-mediated Foxp3 + iTreg cell generation.
While 5-aza-dC may facilitate TGF-βR signaling in response to limited amounts of serum-derived TGF-β by enhancing TGF-βR expression (both mRNA and protein (48)), the observed increased expression levels of CD25 (Figures 4C,D) and pSTAT5 (Figures 6D,E) by converting CD4 + T cells provide a molecular basis for enhanced IL-2R signaling during 5-aza-dC-mediated Foxp3 + iTreg cell generation. Considering that TCR-stimulated CD8 + T cells fail to secrete significant amounts of IL-2 and lack increased CD25 expression levels in the presence of 5-aza-dC (Figure 5), it appears likely that the enhanced CD25 expression on converting CD4 + T cells (Figure 4) is driven by bystander IL-2 production during 5-aza-dC-mediated Foxp3 + iTreg cell generation. Consistently, the fraction of CD25 dim but not CD25 high cells has recently been reported to express significant amounts of IL-2 during 5-aza-dC-mediated iTreg cell generation from human TCR-stimulated CD4 + CD25 − T cells (53). However, currently, we cannot formally exclude that 5-aza-dC may additionally promote CD25 expression through the establishment of a Treg cell-specific CpG hypomethylation pattern within the Cd25 gene locus (22). Thus, rather than promoting STAT5 phosphorylation through direct mechanisms, 5-aza-dC may indirectly enhance IL-2R signaling by increasing IL-2 production in TCR-stimulated bystander cells and by facilitating sustained CD25 high expression on converting cells by demethylation of CpG regions within the Cd25 gene.
TGF-β-supplemented cultures represent a remarkably robust approach to promote Foxp3 expression in TCR-stimulated, initially naive CD4 + Foxp3 − T cells, but the unstable Foxp3 + phenotype of such iTreg cells has represented a major obstacle to clinical translation. In this context, it appears noteworthy that 5-aza-C has been approved by the FDA an epigenetic therapeutic agent to treat some human cancers (39,(55)(56)(57). Systemic administration of 5-aza-C has been reported to ameliorate clinical symptoms in murine models of undesired immunity [ovalbumin-induced airway hyperresponsiveness (58) and virus-induced ocular infection (45)], perhaps by promoting the accumulation of Foxp3 + Treg cells with increased expression of Treg cell signature proteins (CD25, FR4, and GITR) and suppressor function (45). However, the substantial cytotoxic effects inherent to nucleoside Dnmt inhibitors, which may contribute to their antitumor activity, poses a considerable limitation for their further use as therapeutic agent for in vivo administration in clinical settings of unwanted immunity (39).
However, initial reports on 5-aza-C-mediated Foxp3 induction in human CD4 + T cells (21,59) have recently been substantiated by detailed studies (53), indicating that TCR stimulation (in the absence of added TGF-β or IL-2) of initially CD4 + CD25 − T cells in 5-aza-C-supplemented cultures promotes the acquisition of a FOXP3 high phenotype with enhanced expression of Foxp3 + Treg cell signature proteins (such as CD25 and GITR). Such 5-aza-C-induced human FOXP3 + iTreg cells were hyporesponsive to TCR engagement and lacked IL-2 production, but retained Foxp3 expression and suppressor function after IL-2-driven proliferative expansion (53). Thus, the beneficial effect of 5-aza-C on extrathymic Foxp3 induction and its Foxp3-stabilizing ability to promote efficient CNS2 demethylation clearly deserves further consideration as an approach toward the in vitro generation of functionally stable Foxp3 + iTreg cells for adoptive cell therapy.
aUThOr cOnTriBUTiOns KF, NL, SD, AIG, and SS designed, performed, and analyzed the experiments. KF and NL contributed to the data interpretation and assisted in manuscript preparation. SS and KK conceived the research, guided its design, analysis and interpretation, and wrote the manuscript.

acKnOWleDgMenTs
We are indebted to M. Boernert, D. Becker, and S. Hoffmann for excellent technical assistance and to S. Boehme for excellent support in mouse husbandry.  FigUre s2 | Titration of SB431542 to inhibit TGF-βR signaling during Foxp3 + iTreg cell generation. As indicated, naive CD4 + Foxp3 GFP− T cells were T cell receptor stimulated in the presence (+TGF-β; 0.5 ng/ml) or absence (without TGF-β) of exogenously added TGF-β, with or without titrating amounts of SB431542 (2.5, 10, 40, or 80 µM), a selective inhibitor of TGF-βR activation and Smad2/3 phosphorylation. Cultures were analyzed at day 3 for Foxp3 GFP and CD25 expression among gated CD4 + T cells.