T Cell Receptor Signaling in the Control of Regulatory T Cell Differentiation and Function

Antigen affinity and Treg cell differentiation

Following the identification of CD25 and Foxp3 as markers for Treg cells, multiple lines of investigation have supported a role for high-affinity TCR signaling in tTreg cell development. Using mice that express green fluorescent protein (GFP) driven by the nuclear receptor subfamily 4 group A member 1 (Nr4a1) locus, the expression of which is driven by TCR signaling, investigators have shown that tTreg cells exhibit a greater activation phenotype than conventional CD4⁺ T cells¹⁷. Furthermore, Treg cells expressed higher levels of GFP in the periphery¹⁷, suggesting that tTreg cells continuously sample high-affinity antigens. These findings are in line with the earlier observation that conventional T cells engineered to express transgenic TCRs isolated from Treg cells are more likely to proliferate in an MHC class II-dependent manner in a lymphopenic host¹⁸. Importantly, in models of transgenic expression of TCRs with defined antigen specificity, Treg cells appear to require distinct specificities compared with conventional T cells for their differentiation or function¹⁹, and expression of cognate antigens triggers tTreg cell differentiation^{20, 21}. In one such transgenic system, expression of an agonist but not a partial agonist induced tTreg cell generation, implying a stringent antigen affinity requirement in tTreg cell differentiation²². In another study, a panel of TCRs with a wide range of reactivity to a model self antigen were evaluated for their ability to drive tTreg cell development²³. Although the spectrum of antigen reactivity that could trigger Foxp3 expression appears to be broad, the efficiency of tTreg cell generation is positively associated with TCR affinity for the model self antigen, suggesting that tTreg cell differentiation is preferentially induced under conditions of heightened TCR reactivity.

In addition to the thymic development of Treg cells, antigen stimulation of mature naive T cells induces pTreg cell differentiation²⁴. This pathway is a possible means to re-direct potentially pathogenic T cells that fail to undergo clonal deletion in the thymus into anti-inflammatory protective cells and to regulate immune responses to “non-self” antigens, including commensal microbiota-derived antigens. Administering cognate antigens under sub-immunogenic conditions, as well as chronic antigen exposure, triggers pTreg cell differentiation^{25, 26} and although a high dose of a weak agonist could promote Foxp3 expression, stable pTreg cells are induced by a low dose of a strong agonist²⁷.

The emerging consensus that a broad range of high affinity antigens drives tTreg and pTreg cell differentiation highlights the idea that the Treg cell fate is an essential, biologically critical addition to the classical T cell fates of clonal deletion or effector T cell differentiation induced by agonist antigens (Figure 1). Such consideration poses an important question whether the distinct T cell fates are determined by unique modes of TCR signaling, by the presence of cell fate-specifying accessory signals or a combination thereof.

Stage-dependent tTreg cell differentiation

The pro-apoptotic protein Bim is induced by TCR stimulation, and promotes clonal deletion of autoreactive T cells in the thymus. Quantitative studies of autoreactive thymocytes in Bim-deficient mice have revealed that approximately 50% of TCR-signalled CD4CD8 double-positive (DP) and 50% of the positively selected single-positive (SP) T cells are deleted in the cortex and medulla, respectively^{28, 29}. Thus, developing thymocytes are continuously exposed to self antigens that can trigger T cell clonal deletion. Developing tTreg cells, however, reside almost exclusively in the medulla, and the later appearance of tTreg cells compared with conventional (SP) T cells in neonatal mice is associated with the delayed medulla maturation³⁰. Indeed, tTreg cell generation is impaired in the absence of RelB³¹, which is an NF-κB family transcription factor crucial for the differentiation of medullary thymic epithelial cells (mTECs). Autoimmune regulator (Aire) is expressed in mTECs, and triggers the ectopic expression of genes encoding tissue-specific antigens³². Antigen presentation by Aire-positive mTECs promotes tTreg cell differentiation³³, and Treg cells generated in an Aire-dependent manner early after birth have a specific function in maintaining T cell self tolerance³⁴. Taken together, these findings support the idea that Treg cells are specifically induced by medulla-associated agonist antigens.

Why the medulla, but not the cortex, is permissive for Treg cell differentiation is not completely understood. A recent study showed that strong TCR stimulation can induce Bim expression in CCR7⁻ cortical and CCR7⁺ medullary thymocytes²⁸. However, CCR7⁺ cells, but not CCR7⁻ cells, can mount an NF-κB-dependent transcriptional response that opposes Bim-mediated apoptosis²⁸, allowing the medullary thymocytes to survive. Considering the important functions of NF-κB family members in Foxp3 expression (see below), it is conceivable that a finely tuned NF-κB signaling pathway accounts in part for the selective tTreg cell induction in the medulla by coupling T cell survival signals with Treg cell differentiation. Future studies will determine how TCR-induced NF-κB signaling is differentially regulated at distinct stages of T cell development. These findings imply that tTreg cell differentiation is better viewed as a cell fate diversification of CD4⁺ T cells subsequent to their positive selection rather than as a cell fate choice that is alternative to naïve CD4⁺ T cell differentiation and concomitant with positive selection. In contrast, clonal deletion may both precede and coincide with Treg cell differentiation, and may represent an alternative cell fate choice during the differentiation of CD4 SP thymocytes in the medulla (Figure 1a).

Inducing TCR signaling modules

In line with a critical role for agonist-driven TCR stimulation in tTreg cell generation, mutations in several TCR signaling molecules lead to impaired Treg cell differentiation. The tyrosine kinase Zap70 propagates antigen-initiated signals by bridging the TCR complex and downstream signaling pathways (Box 1). SKG mice, which express a phosphorylation-defective Zap70 mutant, have reduced numbers of tTreg cells³⁵. Similar observations were made in mice harboring Zap70 mutations that affect its kinase activity or its recruitment of the adaptor protein LAT^{36, 37}. However, positive and negative selection of T cells were equally affected in these mice, suggesting that the Zap70 mutants do not selectively impair the TCR signaling involved in tTreg cell differentiation.

The transmembrane protein LAT resides in a major complex that amplifies TCR-induced signals (Box 1). Mice expressing a LAT mutant that is defective in binding to phospholipase-γ1 (PLCγ1) were nearly completely devoid of tTreg cells, while the development of conventional CD4⁺ T cell was only partially impaired³⁸. Similar observations made upon analysis of mice with a T cell-specific ablation of PLCγ1³⁹ revealed a crucial role for the LAT–PLCγ1 signaling axis in tTreg cell differentiation.

PLCγ1 hydrolyzes phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5)P₂) to generate the diffusible second messenger inositol-1,4,5-triphosphate (Ins(1,4,5)P₃) and the membrane-associated second messenger diacylglycerol (DAG). Ins(1,4,5)P₃ mobilizes calcium from the endoplasmic reticulum, and calcium signaling facilitated by sensor stromal interaction molecule (STIM) proteins is required for tTreg cell differentiation⁴⁰. DAG enables the recruitment of several signaling-related proteins to the plasma membrane including protein kinase-θ (PKCθ) and RAS guanyl nucleotide-releasing protein (RasGRP). PKCθ induces activation of IKK via the CARMA1, BCL-10 and MALT1 signaling complex and the resulting activation of NF-κB family members is required for tTreg cell differentiation^41–44. RasGRP activates the small GTPase Ras, which is a crucial activator of mitogen-activated protein kinase (MAPK) signaling pathways. Thymic Treg cell differentiation is compromised in RasGRP-deficient mice or in mice expressing a transgene encoding a dominant negative form of the downstream effector molecule Raf^{45, 46}. DAG is metabolized by DAG kinase (DGK) as a means to attenuate PLCγ1-induced signaling. Mice devoid of the ζ isoform of DGK have increased numbers of tTreg cells^{47, 48}, which is associated with enhanced IKK and ERK activities. Indeed, ectopic activation of IKK or ERK in T cells promotes tTreg cell generation^{47, 49}, demonstrating that these signaling pathways comprise limiting steps of TCR-induced tTreg cell differentiation.

Inhibitory TCR signaling modules

Mice deficient in Bim or the transcription factor Nur77 (encoded by the Nr4a1 gene) are defective in T cell negative selection. However, the majority of medullary autoreactive T cells rescued from clonal deletion in these mice are not diverted to the Treg cell lineage^{29, 50}. Thus, strong TCR stimulation per se does not always initiate the tTreg cell differentiation program. Intriguingly, recent studies have revealed inhibitory functions of distinct TCR signaling modules in tTreg cell generation, suggesting a requirement for elaborately measured TCR stimulation for Treg cell fate specification.

TCR signaling is initiated by tyrosine phosphorylation at ITAM motifs in the CD3 chains of the TCR complex (Box 1). Mice expressing a phosphorylation-deficient CD3ζ mutant have attenuated TCR signaling, but exhibit enhanced Treg cell differentiation⁵¹. While diminished negative selection in these mice likely spares self-reactive precursor cells, which in turn may account for their increased tTreg cell numbers, these findings suggest that there are distinct signaling requirements for negative selection and Treg cell differentiation, and that certain aspects of CD3ζ signaling may interfere with tTreg cell differentiation. In support of this conclusion, a recent study of mice deficient in SHARPIN, a component of the linear-ubiquitin-chain-assembly complex, showed diminished tTreg cell generation in association with enhanced anti-CD3-triggered phosphorylation of CD3ζ⁵².

While the IKK and calcium signaling pathways are largely unperturbed in CD3ζ mutant T cells, activation of Akt is substantially attenuated⁵¹. Early studies showed that ectopic expression of an active form of Akt in T cells inhibited tTreg cell generation⁵³. Akt signaling is induced by PI3K, which generates the lipid messenger phosphatidylinositol-3,4,5-trisphosphate (PtdIns(3,4,5)P₃) and recruits Akt to the plasma membrane, as well as by the mTOR complex 2 (mTORC2), which activates Akt at the plasma membrane (Box 1). Mice expressing a kinase-defective p110δ isoform of PI3K and mice devoid of the mTORC2 component Sin1 have increased numbers of tTreg cells^{54, 55}. Together, these findings indicate that the Akt signaling pathway inhibits tTreg cell differentiation.

A “hit and run” model of tTreg cell differentiation

The divergent sensitivity of tTreg cell-promoting or -inhibitory signaling pathways that is observed in the context of attenuated TCR signaling caused by the CD3ζ mutation raises an intriguing possibility that tTreg cell generation is facilitated by antigens that induce distinct or partial TCR activation. Considering that high affinity antigens are most efficient in inducing tTreg cell differentiation, partial TCR signaling is likely a consequence of transient engagement of antigens. Indeed, transient, but not continuous, TCR stimulation induces robust Foxp3 expression in CD4⁺ SP T cells^{56, 57}. Likewise, TCR signaling is superfluous for CD4⁺ SP T cells that express CD25 and have experienced prior TCR stimulation to differentiate into Treg cells^{58, 59}. In addition, studies of differentiation of thymic precursor cells expressing transgenic Treg cell-derived TCRs have revealed substantial intraclonal competition^{60, 61}, suggesting that tTreg cell differentiation is facilitated by ligands present in limiting amounts in the medulla. In contrast, multiple encounters with antigen are required for negative selection of developing thymocytes⁶². Consistent with the idea of transient high affinity TCR engagement favoring Treg cell differentiation, diminished MHC class II expression on mTECs in mice co-expressing a transgenic TCR and a transgene-encoded cognate antigen under the control of the Aire regulatory elements favors Treg cell differentiation over clonal deletion⁶³. Taken together, these findings support a “hit and run” signaling model of tTreg cell differentiation⁶⁴, in which transient TCR stimulation creates a time window for the activation of the signaling pathways that promote tTreg cell differentiation, but not for the activation of the opposing signaling pathways (for example, strong Akt activation), whereas continuous antigen stimulation induces both signaling pathways and results in clonal deletion (Figure 2a). Future dissection of key signaling pathways induced downstream of TCR engagement will help test this hypothesis.

The Goldilocks Principle of Treg cell differentiation

pTreg cell differentiation is induced upon naïve T cell activation under sub-immunogenic conditions with a low dose of high affinity agonist antigen²⁷. The role of individual TCR signaling pathways in pTreg cell generation in vivo remains largely unexplored. Nevertheless, a number of studies have examined TCR signaling pathways in the control of in vitro-induced Treg (iTreg) cell differentiation and revealed several parallels with tTreg cell differentiation. Whereas IKK and calcium signaling pathways promote iTreg cell differentiation^{49, 65, 66}, strong Akt activation antagonizes Foxp3 induction in peripheral T cells^{53, 56}. In line with these findings, strong, but not weak, TCR stimulation of CD4⁺ T cells represses the expression of the PtdIns(3,4,5)P₃ phosphatase Pten⁶⁷, which is an inhibitor of the Akt signaling pathway, and this may in part explain how suboptimal TCR stimulation promotes iTreg cell generation. TCR-induced Pten repression involves the LAT-dependent activation of the Tec-family tyrosine kinase Itk (Box 1). Accordingly, Itk deficiency results in enhanced iTreg cell differentiation⁶⁷. The absence of Itk also results in an increased tTreg cell frequency in vivo⁶⁸, but it remains to be determined whether Pten-mediated repression of Akt is affected. An important downstream target of Akt is the tuberous sclerosis (TSC) complex that suppresses mTOR complex 1 (mTORC1) activation by functioning as a GTPase-activating protein for the small GTPase Rheb⁶⁹. T cell-specific deletion of mTOR, but not Rheb or Sin1 alone, leads to enhanced iTreg cell generation^{55, 70}, suggesting that both mTORC1- and mTORC2-dependent signaling pathways repress iTreg cell differentiation. The shared theme of restrained TCR stimulation in tTreg, pTreg and iTreg cell differentiation implies that TCR-triggered Treg cell lineage commitment follows the “Goldilocks Principle”, as previously proposed for T cell positive selection⁷¹, with TCR signaling being not too high, not too low, but just right.

Accessory signals and Treg cell fate control

As discussed, clonal deletion and Treg cell differentiation represent two competing cell fates of a partially overlapping pool of self-reactive precursor cells^{29, 50}. Hence, these distinct T cell fates are unlikely to be determined by TCR signaling alone. Indeed, the co-stimulatory receptor CD28 has a well-established role in tTreg cell differentiation. In the CD28 cytoplasmic tail, a sequence motif responsible for Lck binding and NF-κB activation, but not the motif involved in PI3K signaling, promotes tTreg cell generation⁷². In a transgenic model of agonist antigen-induced tTreg cell differentiation, CD28 deficiency results in increased clonal deletion⁷³. This observation suggests an important function of CD28-dependent NF-κB activation in sparing self-reactive tTreg cell progenitors from apoptosis as well as in promoting Foxp3 expression (see below). The co-stimulatory receptor CD27 also prevents apoptosis of developing tTreg cells, and likely acts subsequently to CD28 engagement⁷⁴. Furthermore, TCR and CD28 signaling induce the expression of TNF receptor superfamily proteins including GITR, OX40 and TNFR2, which collectively promote tTreg cell generation⁷⁵.

However, in contrast to tTreg cells, strong co-stimulation through CD28 suppresses pTreg or iTreg cell differentiation⁷⁶. In fact, signaling from various inhibitory receptors is essential for Treg cell generation from mature T cells. CD5 is induced by TCR stimulation and functions as a negative feedback regulator of TCR signaling. CD5 deficiency results in enhanced tTreg cell differentiation⁷⁷, but impairs pTreg cell generation through the enhancement of mTORC1 signaling⁷⁸. Likewise, the co-inhibitory receptor CTLA4 promotes iTreg cell generation⁷⁹. In addition, the inhibitory ligand PDL1 is required for optimal iTreg and pTreg cell differentiation, which is associated with enhanced Pten expression and the suppression of Akt signaling⁸⁰. How co-stimulation promotes and opposes thymic and peripheral Treg cell differentiation, respectively, is not well understood. One potential scenario to explore could be a differential sensitivity of co-stimulatory and co-inhibitory signaling pathways in thymic or peripheral Treg cell precursors.

In addition to co-stimulatory receptors, cytokines have important functions in controlling Treg cell differentiation. The immunoregulatory cytokine TGFβ inhibits Bim-mediated T cell clonal deletion⁸¹ and, therefore, likely preserves a larger pool of self-reactive thymic Treg precursor cells. In addition to specific TCR-induced signals, Foxp3 expression has been suggested to promote the death of thymocytes during Treg cell differentiation⁸², aside from its well established function of specifying the Treg cell lineage. Importantly, TCR stimulation and Foxp3 expression is associated with the induction of the high affinity IL-2 receptor CD25, and IL-2 signaling rescues developing Treg cells from Foxp3-induced death⁸², indicating a critical role for IL-2 in Treg cell survival. Combined deficiency of TGFβ and IL-2 almost completely abolishes tTreg cell generation⁸³, supporting their crucial functions in diverting the T cell fate of clonal deletion to tTreg cell generation.

The TGFβ and IL-2 duo also plays an important role in pTreg and iTreg cell differentiation. In addition to inducing Foxp3 expression^{59, 84, 85}, TGFβ and IL-2 potently inhibit alternative cell fate choices, namely, effector T helper (Th) cell differentiation into Th1 and Th2 (TGFβ), and Th17 lineages (IL-2)^{86, 87}. Taken together, these studies reveal that prevention of alternative T cell fates through the engagement of accessory signals likely represents a major strategy underlying the TCR-driven differentiation of Treg cells, including tTreg cells (Figure 2b).

Transcriptional control of Treg cell differentiation

Study of Treg cell fate specification has mostly focused on transcriptional regulation of the Treg cell lineage factor Foxp3. The apparently opposing functions of TCR-induced signaling modules in Treg cell generation are in line with the involvement of several TCR-regulated transcription factors in Foxp3 induction. Both NF-κB and the NF-κB cofactor IκB_NS, as well as Foxo proteins, promote Foxp3 expression by binding to regulatory elements in the Foxp3 locus^{49, 57, 65, 88–91}. In addition, NFAT regulates Foxp3 expression, which is mediated in part by its binding to the Foxp3 conserved noncoding sequence 1 (CNS1) element, which is necessary for pTreg cell differentiation⁹². Furthermore, the TCR-induced Nr4a1 family of transcription factors transactivates Foxp3⁹³, suggesting an additional mechanism by which TCR-induced signaling cooperates to promote Treg cell differentiation.

Members of the E-protein family of transcription factors are important regulators of lymphocyte development. TCR signaling down-regulates the activity of E-box transcription factors in part due to the induction of Id proteins that serve as the E-protein inhibitors. Deletion of E-box proteins or overexpression of Id1 results in enhanced Treg cell differentiation^{94, 95}, whereas Id protein deficiency inhibits Treg cell generation⁹⁶. Mechanistically, TCR-induced E-protein repression appears to be crucial for optimal activation of several Treg cell-promoting signaling pathways including NF-κB⁹⁴, and thus facilitates Treg cell generation.

Epigenetic control of Treg cell differentiation

In addition to the recruitment of transcription factors, deposition of active histone markers histone H3 lysine 4 trimethylation (H3K4me3) and H3K27ac occur at the Foxp3 promoter in Treg cells⁹⁷. However, enrichment of H3K4me1 takes place at the Foxp3 promoter prior to Foxp3 induction in mature SP thymocytes and naïve CD4⁺ T cells, and is dependent on the intronic Foxp3 CNS3^{88, 97}. CNS3 deficiency alters the Treg cell TCR repertoire, with an enrichment of Treg cells that exhibit heightened TCR signaling⁹⁷. These findings reveal an epigenetic mechanism via the cis CNS3 element in the Foxp3 locus that broadens the TCR repertoire of Treg cells.

In addition to Foxp3 expression, Treg cell differentiation is associated with the exploitation of a pre-existent enhancer landscape by Foxp3⁹⁸, and acquisition of a Treg cell-specific CpG DNA methylation patterns at a number of genomic loci^99–101. The epigenetic specification of Treg cells is induced by TCR signaling, but is independent of Foxp3 expression^{98, 99}. In fact, the CNS2 region in the Foxp3 locus itself is demethylated in differentiated Treg cells, which promotes heritable Foxp3 expression and maintenance of a stable Treg cell lineage⁸⁸. Site-specific DNA methylation is controlled by the balanced activity of DNA methyltransferases and DNA demethylases, which likely act in concert with DNA-binding transcription factors and accessory proteins. T cell-specific deletion of the DNA methylation maintenance enzyme Dnmt1 results in enhanced Foxp3 expression even in CD8⁺ T cells, suggesting that Dnmt1 repression participates in DNA demethylation in Treg cells¹⁰². How Dnmt1 activity is inhibited at demethylation sites in Treg cells is unknown. Recent studies of epigenetic control of the myeloid cell-specific transcription factor CEBPA have revealed that Dnmt1 can be sequestered and inhibited by locus-specific non-coding RNA transcripts¹⁰³, providing an intriguing mechanism for site-specific regulation of DNA methylation. Whether the Treg cell-specific epigenome is established through a similar mechanism, for example via TCR-induced regulatory RNA species in targeted loci that sequester Dnmt1, remains to be determined.

TCR signaling and Treg cell subsets

The involvement of strong self reactivity in tTreg cell differentiation has led to the classification of Treg cells as one of the agonist-selected T cell lineages, which also include CD8αα-expressing intestinal intraepithelial lymphocytes (CD8αα⁺ IELs)¹⁰⁴. Although these T cell populations are all considered “antigen-experienced”, Treg cells are abundant in secondary lymphoid organs, whereas CD8αα⁺ IELs are preferentially tissue-resident. Based on the expression of T cell activation and homing molecules, including CD45RA in humans or CD62L and CD44 in mice, Treg cells can be divided into resting Treg cells and activated Treg cells^{105, 106}, which exhibit distinct patterns of cell migration and tissue localization (Table 1). Treg cells among the recent thymic emigrants have been shown to display a resting Treg cell phenotype¹⁰⁷. Considering that agonist-selected T cells, such as IELs, typically display an activated phenotype and reside in target tissues, the question arises how are lymphoid organ-homing resting Treg cells generated.

Firstly, unlike IELs whose differentiation commence at the immature DP stage with predominant specificity towards MHC class I molecule-associated antigens¹⁰⁸, tTreg cells are differentiated from MHC class II-restricted CD4⁺ SP T cells in the medulla. Therefore, tTreg cells acquire properties associated with T cell maturation including the expression of lymph node homing molecules such as the chemokine receptor CCR7²⁸. Secondly, TCR signaling sensitivity decreases as cells mature from cortical DP cells to medullary SP cells¹⁰⁹, and as discussed above, intermediate level of TCR signaling is required for Treg cell differentiation. Thus, TCR stimulation is probably not strong enough to induce an overtly activated phenotype in tTreg cells. Thirdly, TCR stimulation of Treg cells results in attenuated signaling including calcium flux and activation of Akt compared with conventional T cells^{110, 111}. Indeed, among Foxp3 target genes are molecules of the TCR signaling pathway. For instance, whereas TCR stimulation induces Zap70 expression in conventional T cells, Foxp3 binds to the promoter region of the Zap70 gene and inhibits its expression⁹⁹. Antigen-induced signaling in Treg cells may be further suppressed by high expression of inhibitory receptors such as CTLA4 and CD5^{5, 6, 112}. Notably, the Ctla4 locus is demethylated in tTreg cells⁹⁹, which promotes its high expression and therefore supports a role for the Treg cell-specific epigenome in signal modulation. Such genetic and epigenetic mechanisms of TCR signal reprogramming likely dampen antigen-induced Treg cell activation in the periphery, and promote the maintenance of resting Treg cells (Figure 3).

Despite attenuated responses to TCR stimulation, Treg cells with an activated phenotype are present in non-lymphoid tissues as well as tissue-draining lymph nodes as a likely consequence of resting Treg cell activation by self antigens (Figure 4a) and, probably by cytokines¹¹³. In addition, inflammatory signals augment activated Treg cell differentiation from resting Treg cells¹⁰⁷ possibly by enhancing antigen presentation and supplying T cells with accessory signals of activation. The homeostatic properties of activated Treg cells in the steady state have been examined in mouse parabiosis experiments, which showed that these cells recirculate and are not locally maintained long term in tissues such as the liver and intestine^{107, 114}. Nonetheless, the half-life of activated Treg cells is substantially higher than that of resting Treg cells¹¹⁴. Furthermore, activated Treg cells can proliferate in response to a neo-tissue-antigen, with a fraction of them becoming long-lived memory Treg cells¹¹⁵. Although the question remains whether activated Treg cells exhibit enhanced functional capabilities akin to conventional memory T cells, TCR stimulation and probably cell division facilitates resting Treg cell differentiation into activated Treg cells.

Recent studies of mice with Treg cell-specific deletion of the TCRα chain have indeed revealed critical functions for TCR signaling pathways in the generation of activated Treg cells^{116, 117}. TCR-depleted activated Treg cells fail to acquire T cell activation markers or repopulate non-lymphoid tissues, whereas the transcriptome of resting Treg cells is minimally affected by the absence of TCRα. In fact, TCRα-deficient resting Treg cells maintain high Foxp3 expression and the Treg cell-specific epigenome^{116, 117}. Thus, continuous TCR signaling is dispensable for preserving the two defining characteristics of Treg cells at the resting state.

Notably, the ability of resting Treg cells to sense IL-2 is unperturbed¹¹⁶, suggesting that the Treg cell identity is promoted by both IL-2 and TCR signaling. The hypomethylated CNS2 element in the Foxp3 locus is required for sustaining Foxp3 expression in TCR-stimulated Treg cells that undergo cell division⁸⁸. Importantly, CNS2 is crucial to preserve the Treg cell identity under the in vivo condition of chronic inflammation, in particular when IL-2 amounts are limiting^{118, 119}. In line with these observations, STAT5 recruitment to the Foxp3 locus, including CNS2, is observed following IL-2 stimulation. Furthermore, CNS2 contains binding sites for a number of TCR-induced transcription factors including NFAT, and NFAT activation has been implicated in stabilization of Foxp3 expression in activated Treg cells¹¹⁹. Together, these findings suggest that although cytokine, but not TCR, signaling is important for resting Treg cell maintenance, both signals likely participate in preserving the identity of activated Treg cells (Figure 4b).

TCR signaling and Treg cell function

Treg cells deficient in several proximal TCR signaling molecules including Zap70 and Lat exhibit impaired suppressor activity^{37, 120}. Furthermore, Treg cell-specific deletion of the TCRα chain results in a loss of Treg cell function in mice^{116, 117}. Gene expression studies have revealed that TCR expression is required for maintaining approximately 25% of the Treg cell transcriptional signature¹¹⁶. In line with these observations, Treg cell-specific deletion of Ubc13, a Lys63-specific ubiquitin-conjugating enzyme involved in TCR-induced NF-κB activation, impairs the in vivo suppressive function of Treg cells¹²¹. Likewise, Treg cells devoid of the calcium sensor STIM proteins that promote NFAT activation are defective in repressing conventional T cells⁴⁰. These observations suggest that TCR activation-dependent transcriptional programs are crucial for Treg cell function (Figure 4b).

Compared with conventional T cells, Treg cells have elevated steady-state mTORC1 activity that is dependent on TCR signaling¹¹⁷. Treg cell-specific deletion of the essential mTORC1 component Raptor results in loss of Treg cell suppressive activity¹²², which is associated with a loss mTORC1-mediated regulation of cholesterol metabolism in Treg cells¹²². TCR stimulation of Treg cells is required for their suppressive activity in vitro, which is associated with enhanced Treg cell adhesion to antigen-presenting cells¹²³. The TCR-augmented interaction between Treg cells and antigen-presenting cells is mediated by membrane-proximal inside-out activation of integrins, which is largely independent of the enzymatic activity of Zap70¹²⁴. Together, these observations suggest that TCR signaling may promote Treg cell suppressive functions by affecting gene expression, cell metabolism and cell adhesion. Although TCR expression is not required for the maintenance of resting Treg cells^{116, 117}, resting Treg cells receive continuous TCR signaling as they express high levels of Nur77 in a TCR-dependent manner^{107, 116}. Thus, it is possible that compromised Treg cell-mediated suppression in the aforementioned TCR signaling mutants can be due to a defective function of both resting Treg and activated Treg cells.

As discussed above, TCR stimulation not only induces signaling pathways to activate transcription factors including NF-κB and NFAT, but also suppresses the activity of Foxo proteins via Akt (Box 1). Mice with Treg cell-specific deletion of Foxo1 develop a fatal inflammatory disease¹²⁵. Likewise, ablation of the Akt inhibitor Pten in Treg cells triggers Foxo1 hyperphosphorylation and the loss of Treg cell suppressive function^{126, 127}. In line with a well-established role of Foxo1 in the control of homeostasis and migration of conventional naïve T cells⁶⁴, resting Treg cells are depleted in the absence of Foxo1 (M.O.L., unpublished observations). Foxo1 suppresses the expression of the pro-inflammatory cytokine IFNγ, and promotes expression of several proteins involved in T cell trafficking including the transcription factor Klf2 and chemokine receptor CCR7¹²⁵. Indeed, CCR7 is highly expressed in resting Treg cells¹⁰⁷, and CCR7-dependent Treg cell migration is essential for its function in vivo¹²⁸. Resting Treg cells express high levels of CD25 and are preferentially dependent on IL-2 to maintain their homeostasis¹⁰⁷. A recent study showed that Treg cells expressing phosphorylated STAT5 reside in discrete clusters with IL-2⁺ conventional T cells within secondary lymphoid tissues¹²⁹, and provide a local negative feedback mechanism, possibly involving IL-2 consumption, to repress autoreactive T cell responses. Thus, it is conceivable that CCR7 may be required for resting Treg cells to migrate to functional niches, where they promote immune tolerance by, among other mechanisms, consuming IL-2.

Compared with resting Treg cells, activated Treg cells exhibit predominant cytoplasmic Foxo1 localization, and enhanced Foxo1 phosphorylation at Akt target sites¹¹⁴. Treg cell-specific expression of an Akt-insensitive Foxo1 mutant prevents the downregulation of lymphoid organ homing molecules, and impedes activated Treg cell homing to non-lymphoid organs, causing CD8⁺ T cell-mediated autoimmune diseases¹¹⁴. Together, these findings imply complementary activities of resting Treg and activated Treg cells in the maintenance of immune tolerance, which are promoted and repressed by Foxo1, respectively, and are regulated by the Akt signaling pathway downstream of TCRs (Figure 4b).

Tumors

In addition to the well-established function of Treg cells in promoting immunological self tolerance, Treg cells control immune responses to foreign antigens and antigens associated with the altered self, such as in the case of tumors. Increased frequencies of tumor Treg cells are often correlated with poor prognosis of cancer patients, as a likely consequence of Treg cell suppression of anti-tumor immunity¹³⁰. Recent studies have started to unravel the origin and antigen specificity of tumor-associated Treg cells in mouse models.

TCR profiling studies have revealed that tumor-infiltrating Treg cells and conventional CD4⁺ T cells have a largely non-overlapping TCR repertoire^{131, 132}, suggesting that it is unlikely for the Treg cells to have converted from conventional T cells. Likewise, in a transgenic model of prostate cancer, Treg cells from multiple individual tumors recurrently express several TCRs distinct from those of conventional T cells¹³³. The differentiation of Treg cells from transgenic T cells expressing one such TCR was shown to take place in the thymus in an Aire-dependent manner, whereas mature conventional T cells failed to convert to Treg cells in tumor-bearing mice¹³³. These findings suggest that most of the Treg cells that infiltrate at least some solid tumors are self antigen-specific tTreg cells in mice.

The enrichment of antigen-specific Treg cells in tumors implies that tumor progression may release a high levels of antigen that trigger Treg cell proliferation. Thus, interfering with the TCR signaling pathway might impair the homeostasis or function of tumor-associated activated Treg cells. Compared with activated Treg cells from healthy tissues, tumor-infiltrating activated Treg cells downregulate Foxo1 target genes more substantially¹¹⁴, rendering them more susceptible to Foxo1 repression of Treg cell migration to tumors. Indeed, expression of a Foxo1 mutant that is insensitive to AKT-mediated phosphorylation in Treg cells at a dose that does not cause overt autoimmune diseases is sufficient to deplete tumor-associated activated Treg cells, and to activate effector CD8⁺ T cells and inhibit tumor growth¹¹⁴. Interestingly, Treg cell specific-deletion of the leukocyte-specific PI3K isoform p110δ also potentiates anti-tumor immunity¹³⁴, but whether p110δ blockade impairs the generation of tumor-associated activated Treg cells remains to be determined. Nevertheless, these findings suggest that the PI3K–Akt–Foxo1 signaling pathway in Treg cells might be manipulated to break tumor immune tolerance.

Pregnancy

Pregnancy poses a unique challenge to the maternal immune system in placental mammals as the fetus and placenta may harbor “foreign” antigens inherited from the paternal genome. Studies of pregnant mice exposed to conceptus-specific antigens have shown that conventional T cells become activated and proliferate, but do not differentiate into pathogenic effector cells¹³⁵. Instead, Treg cells are enriched among antigen-specific CD4⁺ T cells, and their accumulation is due to both the peripheral conversion of convetional T cells and to expansion of the pre-existing Treg cell population¹³⁶. Interestingly, conceptus-specific Treg cells persist after fetus delivery, and rapidly expand during secondary pregnancy¹³⁶. Such an elevated memory-like Treg cell response may account for the enhanced fetal tolerance in recurrent human pregnancy. pTreg cell differentiation is dependent on the Foxp3 CNS1 enhancer element⁸⁸. Remarkably, the presence of CNS1 appears to be unique among placental mammals and may have been introduced by retro-transposon insertion coinciding with eutherian evolution¹³⁷. Female mice lacking the CNS1 element fail to induce conceptus-specific pTreg cells, and have higher embryo resorption rates when mated with allogenic males¹³⁷. These findings have validated an important function for Treg cells in fetomaternal tolerance, as previously suggested (see refs ^{138, 139}).

Close contact between tissues and blood supplies of fetus and mother makes the fetomaternal antigen exchange a two-way process. As the fetus harbors half of the maternal genome, many non-inherited maternal antigens (NIMA) are “foreign” to the fetus. Previous studies have revealed that maternal cells that cross the placenta induce Treg cell differentiation, and suppresses fetal anti-maternal immune responses in humans¹⁴⁰. By tracking antigen-specific T cell responses, a recent study in mice has revealed that maternal cells that form microchimerism induce the differentiation of NIMA-specific pTreg cells in female offspring during development¹⁴¹. Importantly, these females exhibit increased numbers of NIMA-specific pTreg cells during pregnancy sired by allogeneic males with overlapping NIMA specificity, which promotes cross-generation fetal tolerance. Together, these findings imply that the pTreg cell differentiation pathway may have been selected during evolution to reinforce reproductive fitness of placental mammals.

Commensals

The intestine harbor trillions of commensal bacteria that form symbiotic relationship with mammalian hosts to promote nutrient extraction from food as well as organismal homeostasis. Thus, this symbiotic relationship can be considered as a distinct antigenic state. Indeed, although intestinal Treg cells are present in germ-free animals, recent studies have revealed that commensals can induce the accumulation of antigen-specific Treg cells^{142, 143}. TCRs isolated from colonic Treg cells from specific pathogen-free (SPF) mice recognize commensal antigens in the intestines of these animals¹⁴². In addition, Treg cells isolated from mice exposed to Clostridium sp. are more suppressive in the presence of fecal material containing the same bacterial species than in its absence¹⁴⁴, in line with antigen-specific stimulation of Treg cells. The origin of commensal-reactive Treg cells has also been investigated by creating retrogenic mice expressing the Treg cell TCRs. In contrast to TCRs from lymph node Treg cells, commensal-reactive TCRs from colonic Treg cells failed to promote thymic Treg cell differentiation, but facilitated pTreg cell generation in the presence of commensals¹⁴². These findings are corroborated by the observations that Treg cell frequencies are reduced in mice deficient for the Foxp3 CNS1 element⁸⁸, or devoid of a binding site for TGFβ-induced SMAD transcription factors in CNS1¹⁴⁵. Notably, CNS1-deficient mice develop colitis¹⁴⁶, supporting an important function for pTreg cells in the control of immune tolerance to commensals.