Research carried out over the past 10–20 years has greatly expanded our understanding of how T-cell subsets differentiate and acquire a degree of stability that allows them to be considered a distinct T-cell lineage. This is the case for CD4+ regulatory T (TReg) cells that express the transcription factor forkhead box P3 (FOXP3), which is essential for the normal development and function of these cells1, 2, 3. Together with other suppressive immune-cell populations, FOXP3-expressing TReg cells have an essential role in maintaining homeostasis of the immune system and in preventing the autoimmune reactivity of self-reactive T cells that have escaped negative selection in the thymus.
FOXP3+ TReg cells can differentiate during T-cell development in the thymus at the stage of CD4 single-positive thymocytes4. It has been reported that antigen presentation by either cortical or medullary thymic epithelial cells is sufficient for inducing FOXP3 expression in developing thymocytes5, 6. These thymically derived or natural TReg cells retain a stable phenotype following export into the periphery. There, they can become activated by specific antigen and acquire some of the phenotypical properties of effector memory T cells, such as the capacity to migrate into inflamed peripheral tissues, while maintaining FOXP3 expression and their suppressive function7. Therefore, this T-cell subset has the defining features of a stable T-cell lineage.
In addition, there is now compelling evidence indicating that FOXP3+ TReg cells can also arise from the conversion of naive conventional CD4+ T cells in the periphery (Fig. 1) following recognition of antigen under tolerogenic conditions7, 8. It can be assumed that such de novo-generated TReg cells have an important role in acquired tolerance to, for example, food antigens or commensal gut flora9, 10. However, the extent to which de novo-induced TReg cells contribute to the peripheral TReg-cell pool and whether they have a stable phenotype that is comparable to that of natural TReg cells are issues that have not yet been thoroughly investigated.
Figure 1 | The origin of regulatory T cells in the thymus and the periphery.
The expression of the transcription factor forkhead box P3 (FOXP3) is thought to best define a subset of CD4+ regulatory T (TReg) cells that have suppressive functions. Natural TReg cells differentiate in the thymus from T-cell precursors in a process that is not yet completely understood but is known to involve interactions with thymic epithelial cells in the cortex and the medulla. Cytokines that signal through the common cytokine-receptor 
-chain (c) subunit (such as interleukin-2; IL-2) are also known to participate in the generation of TReg cells, whereas the role of transforming growth factor-
(TGF) in this process is still debated. TReg cells are thought to constitute a separate T-cell lineage and to maintain FOXP3 expression throughout their life cycle in an IL-2- and TGF-dependent manner. Following antigen encounter in the periphery, naive TReg cells (similarly to FOXP3- T cells) acquire features of effector memory cells. TReg cells can also be induced de novo from naive peripheral FOXP3- T cells. In vitro, this conversion can be driven by treatment with TGF and may be enhanced by the vitamin A metabolite retinoic acid. However, these cells have an unstable phenotype and transient expression of FOXP3. Conversion can also occur in vivo following various tolerogenic protocols, such as oral antigen administration. It has been reported that TGF and retinoic acid have a role in TReg-cell conversion in vivo, although other factors are probably also involved. In addition, the stability of in vivo-converted TReg cells is still unclear.
Antigenic stimulation of conventional CD4+ T cells in vitro in the presence of transforming growth factor- (TGF) leads to the induction of FOXP3 expression and the acquisition of suppressor function11. However, this acquired TReg-cell phenotype is unstable, as most of these cells lose FOXP3 expression following restimulation with antigen in the absence of exogenous TGF (Ref. 12). This indicates that TGF is not sufficient for imprinting T cells with the permanent expression of FOXP3 that is needed for a stable TReg-cell phenotype. Moreover, although conventional human T cells transiently upregulate the expression of FOXP3 following antigenic stimulation13, 14, 15, 16, 17, 18, whether this transient FOXP3+ population has suppressive function is currently disputed. Some groups have reported that these cells acquire transient suppressive function13, 18, whereas others have suggested that FOXP3 expression in human T cells does not always directly correlate with suppressive capacity15, 16, 17.
Stable FOXP3 expression is clearly a prerequisite for the maintenance of suppressive properties in TReg cells. Given the potential for these cells to be manipulated in a therapeutic context, it is essential that the factors governing the expression of this lineage-specification factor be defined. In this Progress article, we summarize the current knowledge of how FOXP3 expression — and therefore the TReg-cell phenotype — is controlled at the molecular level, highlighting mechanisms of transcriptional regulation and the importance of epigenetic modification of the FOXP3 locus.
Requirements for FOXP3 expression
Although various signals that induce the expression of FOXP3 have been identified, the precise mechanisms by which the expression of this protein is controlled in TReg cells are not well understood. So far, it has been established that the synergistic action of signals downstream of the T-cell receptor (TCR), co-stimulatory molecules and cytokine receptors is required for the active transcription of FOXP3 (Fig. 2).
Figure 2 | Multiple signalling pathways converge for the induction of forkhead box P3 expression.
Signals that are triggered following ligand binding to the T-cell receptor (TCR), CD28, cytokine receptors that contain common cytokine-receptor -chain (c; here represented by the interleukin-2 (IL-2) receptor) and the transforming growth factor- (TGF) receptor, and by the initiation of the cyclic AMP pathway together regulate the expression of the transcription factor forkhead box P3 (FOXP3). These events result in the activation of transcription factors that are involved in FOXP3 expression, including cAMP-responsive-element-binding protein (CREB), activating transcription factor (ATF), SP1, nuclear factor of activated T cells (NFAT), activator protein 1 (AP1), TGF-inducible early gene 1 (TIEG1), mothers against decapentaplegic homologue 3 (SMAD3) and signal transducer and activator of transcription 5 (STAT5). However, the contribution of each pathway (either beneficial or inhibitory) to FOXP3 expression might differ between different types of T cell; for example, CD28 stimulation is important for the thymic development of FOXP3+ regulatory T cells, whereas the activation of the phosphoinositide 3-kinase (PI3K)–AKT–mammalian target of rapamycin (mTOR) pathway inhibits FOXP3 expression in peripheral naive T cells by an unknown mechanism. ERK, extracellular-signal-regulated kinase; JAK, Janus kinase; JNK, JUN N-terminal kinase; PKA, protein kinase A; PKC
, protein kinase C; PLC, phospholipase C.
TCR signalling. TCR signalling pathways contribute to the induction of FOXP3 expression in both natural and de novo-induced TReg cells. In human T cells, TCR activation has been shown to lead to the binding of the transcription factors nuclear factor of activated T cells (NFAT) and activator protein 1 (AP1) to the FOXP3 promoter14. In mouse T cells, TCR activation results in the binding of cyclic-AMP-responsive-element-binding-protein (CREB) and activating transcription factor (ATF) to an intronic enhancer element in the Foxp3 gene19. Accordingly, mice that are deficient for protein kinase C (PKC) and calcineurin A, which are both involved in NFAT activation, have a marked reduction in the number of TReg cells. In addition, PKC was found to stimulate the activity of the FOXP3 promoter20, which suggests that PKC promotes TReg-cell development by enhancing FOXP3 expression through the activation of the Ca2+–calcineurin–NFAT pathway. These data are supported by the finding that high levels of the calcineurin inhibitor cyclosporine block TReg-cell induction and function in both mice and humans14, 21, 22, 23. However, low-dose cyclosporine therapy has recently been reported to increase the size of the TReg-cell population in patients with atopic dermatitis24, which suggests that partial inhibition of the calcineurin pathway might be beneficial for the induction of TReg cells. Partial inhibition of calcineurin activity might mimic suboptimal TCR stimulation, which was found to result in more efficient induction of TReg cells than strong TCR activation both in vitro25 and in vivo8. Moreover, premature termination of TCR signalling and inhibition of phosphoinositide 3-kinase, AKT or mammalian target of rapamycin induced the expression of FOXP3 by mouse T cells26, 27. Together, these data indicate that the duration and strength of the TCR signal are crucial determinants of FOXP3 expression, with shorter and weaker TCR stimulation favouring TReg-cell development.
Co-stimulation. In addition to TCR signalling, specific co-stimulatory signals are essential for FOXP3 expression in both natural and de novo-induced TReg cells. CD28 stimulation of TCR-activated thymocytes induces the expression of FOXP3 and the initiation of the TReg-cell differentiation programme28. By contrast, de novo conversion of conventional T cells into TReg cells in the periphery is impaired by co-stimulation29, suggesting that the requirement for co-stimulation for the induction of FOXP3 expression differs between natural and de novo-induced TReg cells.
Cytokine-mediated signals. Specific cytokine-mediated signals are also essential for the expression of FOXP3. The best evidence in support of this is the finding that there is a complete lack of TReg cells in mice that are deficient for the common cytokine-receptor -chain (c), which transmits signals that are mediated by interleukin-2 (IL-2) and several other cytokines30. However, it is not clear which of the cytokines that use the c for signal transduction are involved in FOXP3 expression.
There is substantial evidence suggesting that the signalling cascade that is activated following binding of IL-2 to its receptor, which involves Janus kinase 1 (JAK1), JAK3 and signal transducer and activator of transcription 5 (STAT5), has an integral role in inducing FOXP3 expression (Fig. 2). Indeed, Stat5-/- mice have strongly reduced FOXP3 expression and Jak3 -/- mice lack FOXP3 expression completely31, 32. Moreover, IL-2-induced STATs bind directly to evolutionarily conserved regions in the FOXP3 locus and induce the expression of this gene32, 33, 34. Although IL-2 is crucial for the maintenance of homeostasis and competitive fitness of natural TReg cells in the periphery, it is dispensable for the expression of FOXP3 in the thymus30, 35, which suggests that other cytokines that signal through c-containing receptors, such as IL-7 and IL-15, might compensate for IL-2 during the development of natural TReg cells34.
In addition to c cytokines, TGF also has an important role in TReg-cell biology. Although the contribution of TGF to the development of TReg cells in the thymus is still controversial36, 37, TGF seems to have a central role in the maintenance of FOXP3 expression and homeostasis of natural TReg cells36. Recently, it was shown that TGF-inducible early gene 1 (TIEG1; also known as KFL10) can bind to the FOXP3 promoter and cooperate with itchy E3 ubiquitin protein ligase homologue (ITCH) to induce FOXP3 expression38. In addition, the TGF-induced transcription factor mothers against decapentaplegic homologue 3 (SMAD3) has been shown to control the activity of a newly identified FOXP3 intronic enhancer element in cooperation with NFAT39 (Fig. 2).
Several factors have been reported to promote TGF-dependent de novo induction of FOXP3 expression. Exposure of T cells to the vitamin A metabolite retinoic acid, which is produced by intestinal lamina-propria dendritic cells and other cell types9, 10, increases the expression and phosphorylation of SMAD3 (Ref. 40). This, in turn, leads to an increase in TGF-induced FOXP3 expression while preventing the differentiation of the inflammatory T helper 17 (TH17)-cell subset9, 10, 29, 40. Importantly, the de novo induction of TReg cells can occur even at high levels of co-stimulation if retinoic acid is present, suggesting that retinoic acid attenuates the inhibitory effect of co-stimulation on the induction of FOXP3 expression29. These data are supported by a recent report41 showing that retinoic acid indirectly enhances the induction of FOXP3 expression by inhibiting the production of counter-regulatory cytokines by CD44hi effector memory T cells. Notch-mediated signals also cooperate with TGF-mediated signals to regulate FOXP3 expression in conventional T cells42 by directly targeting the FOXP3 promoter through mechanisms that depend on recombination-signal-binding protein for immunoglobulin-
J-region (RBPJ) and hairy and enhancer of split 1 (HES1)43.
In addition to the factors that promote TGF-dependent de novo induction of FOXP3 expression, there are various mechanisms that negatively regulate TReg-cell differentiation; lineage-specifying factors of the TH1- and TH2-cell subsets are the most prominent of these mechanisms44, 45, 46. More specifically, direct binding of the IL-4-induced proteins GATA-binding protein 3 (GATA3) and STAT6, as well as of the interferon--induced protein interferon-regulatory factor 1 (IRF1), to conserved binding sites in the FOXP3 promoter has been reported to repress its transcriptional activity45, 47, 48, which indicates that reciprocal developmental pathways can control the generation of effector cells and TReg cells.
Although these findings support the hypothesis that a delicate balance of TCR-mediated, co-stimulatory and cytokine-mediated signals is mandatory for the transcription of the FOXP3 gene, little is known about the mechanisms by which these signals induce FOXP3 expression at the molecular level. How do the molecular signals that are required for the induction of FOXP3 expression in developing thymocytes differ from those that are required for the conversion of conventional T cells into TReg cells? And how do constitutive signals, such as TCR–MHC interactions and basal cytokines, maintain FOXP3 expression in natural TReg cells under steady-state conditions?
microRNA and TReg cells
In the past decade, the discovery of microRNAs has revealed another level of complexity in the mechanisms that regulate gene expression. These small, non-coding RNAs inhibit protein expression either by enhancing the degradation of target mRNA species or by repressing mRNA translation through a process known as RNA interference. microRNAs are involved in the regulation of a broad range of cellular processes, including decision making during lineage differentiation49. In the immune system, microRNAs have a crucial role in leukocyte development and in the modulation of immune responses50. This is also true for the TReg-cell subset, in which microRNA-mediated regulation seems to be particularly important.
First, microRNAs are involved in TReg-cell development in the thymus51. In addition, microRNAs are fundamental for TReg-cell function, as TReg-cell-specific deletion of the microRNA pathway led to the development of fatal autoimmune diseases, which was caused by a loss of suppressor function in the peripheral TReg-cell pool52, 53, 54. Although experimental evidence is still lacking, it is possible that the contribution of microRNA-mediated regulation to TReg-cell function is controlled by FOXP3, which has been shown to bind close to microRNA-encoding intergenic regions55 and to contribute to the distinct microRNA expression profile that is found in TReg cells51. Whether, in turn, the expression of FOXP3 is controlled by microRNAs, as recently suggested52, awaits further investigation. Although it is still unclear how microRNAs control TReg-cell development and function, these early studies certainly indicate that this topic will be the subject of further research.
Epigenetic regulation of FOXP3
Recently, several groups have observed that epigenetic regulation is crucial for controlling the expression of the FOXP3 locus. Epigenetic modifications, which can target histones or the DNA directly, affect gene transcription by altering the accessibility of distinct DNA regions to transcription factors and other DNA-binding molecules. Histones can be modified by site-specific acetylation and by methylation, modifications that are essential for determining the overall chromatin structure. In addition, CpG motifs in the DNA, which are rare and are often clustered in CpG-rich regions within promoters, can be methylated or demethylated. When the CpG motifs are methylated, they are often associated with chromatin-remodelling factors, such as methyl-DNA-binding proteins, which results in the condensation of chromatin. The opposite occurs following the demethylation of CpG motifs, which results in the relaxation of chromatin and an increased accessibility of target sequences, thereby allowing the binding of specific transcription factors. These epigenetic mechanisms of transcriptional control have been of great interest in recent years, as they are believed to imprint the activity state of specific gene loci, such that an environmentally induced phenotype might become heritable and be maintained over numerous cell divisions. These mechanisms are discussed in more detail in the review by Wilson et al. in this issue of Nature Reviews Immunology56. Here, we focus on the epigenetic modifications of the FOXP3 locus that allow stable FOXP3 expression. In addition, in Box 1 we highlight how post-translational modifications of the FOXP3 protein and other downstream events influence FOXP3 function and the chromatin remodelling of its target genes.
In mice and humans, distinct regions of the FOXP3 locus have a pattern of DNA methylation and specific histone modifications that differ between TReg cells and conventional T cells. Sequence analyses have revealed three highly conserved non-coding regions in the FOXP3 locus (Fig. 3), all of which have been found to be subject to epigenetic modifications and to be involved in regulating the transcription of FOXP3.
Figure 3 | The FOXP3 locus is subject to epigenetic control.
a | The forkhead box P3 (FOXP3) gene, which is located on the X chromosome, is highly conserved, as illustrated in the interspecies conservation plot (obtained from the University of California Santa Cruz Genome assembly web site; segment shown is located between bases 48,994,000 and 49,009,118). b | Three conserved non-coding regions that undergo epigenetic modifications and are involved in the regulation of FOXP3 transcription are highlighted. Epigenetic modifications that occur in these three regulatory regions, including histone acetylation and DNA methylation, are depicted for FOXP3- conventional T cells, transforming growth factor- (TGF)-induced FOXP3+ T cells and natural TReg cells (natural TReg cells show a stable FOXP3+ phenotype). Note that the TGF-sensor region does not contain CpG motifs. The hypothetical open chromatin conformation that is induced by permissive histone modifications and DNA demethylation allows the binding of transcription factors to regulatory sites and thereby enables the induction and stabilization of FOXP3 expression. Upstream signalling pathways that affect these regions when activated are depicted (if known). AP1, activator protein 1; ATF, activating transcription factor; IL-2, interleukin-2; CREB, cyclic-AMP-responsive-element-binding protein; NFAT, nuclear factor of activated T cells; SMAD3, mothers against decapentaplegic homologue 3; STAT5, signal transducer and activator of transcription 5; TCR, T-cell receptor; TIEG1, TGF-inducible early gene 1.
FOXP3 promoter. The FOXP3 promoter, which is located 6.5 kb upstream of the first coding exon of FOXP3, is a classic TATA- and CAAT-box-containing promoter that is activated in response to TCR signalling through binding of NFAT and AP1 (Ref. 14). TReg cells and resting conventional T cells show differences in the epigenetic modification of the FOXP3 promoter in both mice and humans: the CpG motifs in the FOXP3 promoter are almost completely demethylated in TReg cells, whereas they are weakly methylated in resting conventional T cells19, 57. Furthermore, the FOXP3 promoter shows a stronger association with acetylated histones in TReg cells than in conventional T cells14, 19, suggesting that the FOXP3 promoter is more accessible in TReg cells. Following in vitro activation of conventional mouse T cells, Foxp3 promoter methylation is increased19, which might further restrict the accessibility of the promoter and prevent the induction of FOXP3 expression in these cells.
TGF
sensor. The second highly conserved non-coding region in the FOXP3 locus has been identified as a TGF-sensitive element that contains binding sites for NFAT and SMADs. The chromatin in this region is also in an accessible state in cells that express FOXP3, as indicated by the increased levels of acetylated histone H4 in this region in both natural and TGF-induced TReg cells39. Moreover, TGF-induced chromatin remodelling of this region might even affect the accessibility of the upstream FOXP3 promoter, as the level of promoter demethylation was found to be slightly increased in TGF-treated mouse T cells19. A similar opening of the FOXP3 promoter (that is, a slight increase in demethylation and in the association with acetylated histones) was observed in activated conventional human T cells that transiently express FOXP3 (Refs 14, 57). This effect was observed even without the addition of exogenous TGF. However, this activation-induced opening of the FOXP3 promoter might have been due to low levels of TGF in the culture medium, as neutralization of TGF was recently reported to completely abrogate the activation-induced transient expression of FOXP3 in human T cells16. This finding suggests that a different sensitivity to low levels of TGF in mouse and human T cells might account for the differences in activation-induced FOXP3 expression that are observed in the two species.
TReg-cell-specific demethylated region. The most striking differences regarding the methylation pattern at the FOXP3 locus have been observed in a third, highly conserved CpG-rich region. This region was found to be fully demethylated in TReg cells and methylated in conventional T cells12, 19, 58, 59, and in this article is referred to as the TReg-cell-specific demethylated region (TSDR). In addition, acetylated histones H3 and H4, and trimethylated lysine 4 in histone H3, were found to accumulate at the TSDR12. The TSDR has enhancer activity that is markedly decreased following methylation60; this is supported by the finding that the transcription factor CREB binds to the TSDR when this region is demethylated19. Together, these data suggest a role for DNA methylation in the molecular regulation of FOXP3 expression. However, the methylation state of the TSDR seems to be irrelevant for determining the level of FOXP3 expression, as TGF-induced TReg cells express levels of FOXP3 that are comparable to those of natural TReg cells, despite a lack of TSDR demethylation12, 60. Evidence is now accumulating that TSDR demethylation does not act as an on/off switch, but instead determines the stability of FOXP3 expression12, 59, 60, a concept that is consistent with the known role of epigenetic regulation in T-cell lineage decisions61.
So, demethylation of the TSDR corresponds with stability of FOXP3 expression (as in natural TReg cells), whereas T cells that express FOXP3 only transiently (TGF-induced TReg cells and recently activated conventional human T cells) have a methylated TSDR12, 58, 60. These data are supported by the finding that drug-mediated DNA demethylation in conventional T cells led to the induction of stable FOXP3 expression and a TReg-cell phenotype19, 59, 60, and only the fraction of cells that expressed FOXP3 was found to have a demethylated TSDR60. Interestingly, TReg cells that were induced in vivo from conventional CD4+ T cells following targeting of antigen to steady-state dendritic cells that were expressing DEC205 also had stable FOXP3 expression and, correspondingly, demethylation of the TSDR60. It remains to be determined whether other protocols that are known to induce tolerance in vivo (such as oral administration of antigen and allergen-specific immunotherapy) also induce TReg cells with a stable phenotype. It is probable that the methylation status of the TSDR, the FOXP3 promoter and potentially additional regulatory regions will be valuable biomarkers for the detection of stably suppressive TReg cells57, 58, 62. This is especially relevant in humans, as transient expression of FOXP3 is observed in activated conventional human T cells.
Concluding remarks
In view of the potential clinical applications of TReg cells, elucidating the combination of signals that imprint FOXP3 expression and TReg-cell function in vivo remains an important challenge. Future studies should aim to confirm that the imprinted differentiation state found in natural TReg cells is as fixed as we currently believe it to be, as existing concepts of the heritability of epigenetic patterns have been challenged by recent reports of fluctuating methylation states of dynamic promoters63 and by indications that TReg cells can become IL-17-producing cells following exposure to inflammatory stimuli64, 65, 66. The conversion of TReg cells to TH17 cells can be induced by dendritic-cell-derived IL-6, which downregulates the expression of FOXP3 in a STAT3-dependent manner64, 66. However, one study65 reported that, although IL-6 or TGF did not affect the generation of IL-17-producing cells, their differentiation was increased by exogenous IL-1, IL-23 and IL-21, an effect that could be prevented by histone deacetylase inhibitors. As recent work suggests that chromatin-modifying agents, such as DNA methyltransferase and histone deacetylase inhibitors, can be used to manipulate TReg-cell biology19, 59, 60, 67, 68, there is promise that the programming and reprogramming of TReg cells at the epigenetic level may be an important approach for the development of drugs that target these cells.
