Treg Cells, Life History, and Diversity

Abstract

Regulatory T cells expressing the FoxP3 transcription factor have a profound and nonredundant role in several aspects of immunological tolerance. We will review here the specification of this lineage, its population dynamics, and the diversity of subphenotypes that correlate with their diverse roles in controlling inflammation in a variety of settings.

The overall physiological function of regulatory T cells may not be to maintain immune tolerance, but more generally to ensure tissue homeostasis and control inflammation.

Sculpting a functional but self-tolerant repertoire of T cells first involves a broad roughing in the thymus, eliminating immature thymocytes whose T-cell receptor (TCR) responds to self-antigens presented by thymic antigen-presenting cells (APCs). But potentially autoreactive T cells do escape thymic negative selection, and need to be controlled by peripheral mechanisms of tolerance induction. Self/nonself discrimination poses an inherent challenge in peripheral lymphoid organs that are open to environmental antigens, and relies in part on cues from differential presentation (timing, dose, or costimulatory environment) as discussed in Xing and Hogquist (2012). Peripheral tolerance also includes dominant mechanisms, wherein particular cells exert an inhibitory influence on the activation or effector function of other cell types. Such suppressive, or apparently suppressive, effects can stem from the normal balance between effector functions that tend to inhibit each other, such as the mutual antagonism between Th1 and Th2 effectors; at some level, every cell type probably inhibits some other, without implying any particularly “suppressive” function. But the immunological constellation also includes cells whose primordial role seems to be the dominant control of other cells, at least as currently perceived. The only truly well characterized of these “regulatory” cells are FoxP3⁺ CD4⁺ T cells (Tregs), the topic of much scrutiny over the last decade. As a result, we have a broad perspective on these cells, which affect many facets of immune tolerance: to unperturbed self, to infected or tumoral self, to artificial or natural (e.g., pregnancy) grafts. Indeed, their scope also extends beyond immune tolerance, more generally to controlling a variety of inflammatory processes. As their cellular determinism and function have been extensively reviewed over the past few years (Brusko et al. 2008; Sakaguchi et al. 2008; Vignali et al. 2008; Belkaid and Tarbell 2009; Feuerer et al. 2009b; Littman and Rudensky 2010, 2011), we will present a more focused discussion of the origin, fate, and diversity of Tregs, in relation to their basic physiology and to pathological implications.

FoxP3⁺ Treg IDENTITY: A SHORT OVERVIEW

Treg cells were identified as CD4⁺ T cells expressing high levels of CD25 or low levels of CD45RB, and able to protect from autoimmune pathology induced by neonatal thymectomy or by reconstitution of lymphopenic rodents (Sakaguchi et al. 1982, 1995; Powrie and Mason 1990; Smith et al. 1991; Morrissey et al. 1993; Suri-Payer et al. 1998). The lack of specificity of these markers, shared with activated T cells, and the contrived nature of the experimental settings, initially led to skepticism as to their true relevance to immune tolerance. This skepticism was enhanced by the bitter aftertaste left by the embarrassing collapse of the “suppressor cells” house of cards in the early 1980s (Germain 2008). The breakthrough that led to general acceptance of Treg cells as a distinct phenotype/lineage was the identification of the transcription factor FoxP3 (Chatila et al. 2000; Brunkow et al. 2001; Wildin et al. 2001), and of its unique expression in Treg cells (Fontenot et al. 2003, 2005b; Hori et al. 2003; Khattri et al. 2003; Wan and Flavell 2005). Deficiencies in Foxp3 were shown to cause lymphoproliferation and multiorgan autoimmunity in scurfy mutant mice and human IPEX patients (Ziegler 2006). This pathology could be rescued by transfer of Treg cells (Fontenot et al. 2003), and was reproduced by inactivation of FoxP3 uniquely in T cells (Fontenot et al. 2005b), proving the T-cell-autonomous role of FoxP3. Aside from providing a unique molecular identifier, these results established that Tregs were not merely constructions from contrived and artifact-prone experimental systems in mice, but actually played a nonredundant role in immune tolerance and homeostasis.

FoxP3⁺ Treg cells use the αβTCR for antigen recognition, and have a broad repertoire, comparable in size to, but largely distinct in composition from, that of conventional CD4⁺ T cells (Tconv) (Hsieh et al. 2006; Pacholczyk et al. 2006; Wong et al. 2007b; Lathrop et al. 2008). Consistent with their expression of CD4, they are restricted by MHC-II molecules, although some MHC-I-restricted CD8⁺ or CD4⁺CD8⁺ FoxP3⁺ cells are observed in MHC-II-deficient mice (Krajina et al. 2004; Bienvenu et al. 2005; Fontenot et al. 2005b; Stephens et al. 2007) and in anti-CD3-treated human patients (Bisikirska et al. 2005).

Treg cells were long thought to represent an anergic phenotype, because they are largely unresponsive to the usual TCR triggers in vitro. Indeed, this nonresponsiveness extends dominantly to Tconv cells cultured together with Tregs, forming the basis for the classic in vitro suppression assay (Takahashi et al. 1998; Thornton and Shevach 1998). On the other hand, it is now recognized that this apparent anergy is really an in vitro artifact and that Treg cells are far from anergic in vivo. Indeed, Tregs actually cycle more actively than Tconv, spontaneously or in response to specific antigen (Fisson et al. 2003; Walker et al. 2003; Kretschmer et al. 2005). The anergic in vitro phenotype likely reflects the very strong dependence of Treg cells on IL2, which they are unable to synthesize but which is key to their differentiation, homeostasis, and function (Furtado et al. 2002; Malek et al. 2002; Bensinger et al. 2004; Thornton et al. 2004; Fontenot et al. 2005a; Setoguchi et al. 2005; Tai et al. 2005; Lio and Hsieh 2008; Yu et al. 2009). This dependence of Treg cells on the IL2 produced by Tconv cells provides a negative-feedback loop in which the amplification of Tregs is directly conditioned by the degree of Tconv activation. Treg cells are under strong homeostatic control. Their numbers recover within a day or two in lineage ablation experiments (Kim et al. 2007; Feuerer et al. 2009c), and they rapidly fill empty niches after transfer into lymphopenic conditions or after systemic perturbation by anti-CD3 (Nishio et al. 2010).

FoxP3⁺ Treg CELLS: DIFFERENTIATION

Two origins have been described for FoxP3⁺ cells, whose numeric and functional importance remain in question. Most FoxP3⁺ cells differentiate in the thymus from immature CD4⁺CD8⁺ precursors, as an alternative to conventional CD4⁺ T cells. The second occurs in the periphery, where a number of triggers induce the expression of FoxP3 in Tconv cells. The former are commonly referred to as “natural” or “thymus-derived” (here tTregs), the latter as “induced, adaptive, peripheral” (in view of the semantic inadequacy of the former terms, we will refer to these as pTreg).

In experimental conditions, the Foxp3⁺ pTreg phenotype results from in vivo “conversion” of mature Foxp3⁻ CD4⁺ cells in a variety of conditions: chronic and/or suboptimal stimulation by agonist peptide, exposure to agonist administered orally, response to homeostatic cues after transfer into lymphopenic hosts, or in response to helminth infection (Apostolou and von Boehmer 2004; Kretschmer et al. 2005; Finney et al. 2007; Curotto de Lafaille et al. 2008; McSorley et al. 2008; Haribhai et al. 2009, 2011; Daniel et al. 2010; Feuerer et al. 2010; Grainger et al. 2010). In addition, natural antigens from commensal bacteria can elicit particular populations of Foxp3⁺ Tregs specifically in the colon (Atarashi et al. 2010; Lathrop et al. 2011). The construction of transgenic mice expressing TCRs isolated from these microbe-responsive pTregs showed that these TCRs cannot entice Treg selection in the thymus, implying that many colonic Tregs arise from antigen-driven pTreg development (Lathrop et al. 2011). Peripherally generated pTregs proved as effective as bulk Tregs in several functional assays. Indeed, the conditions that promote their conversion can lead to superior suppressive ability, in part by ensuring an antigen-specific repertoire (Daniel et al. 2010; Haribhai et al. 2011). pTregs are also largely similar to bulk Tregs from lymphoid organs in regard to their transcriptional signatures, although there are some fine differences that reflect their mode of conversion (e.g., antigen- vs. homeostatic-driven conversion [Haribhai et al. 2009; Feuerer et al. 2010]).

Perhaps with the exception of the colonic Treg mentioned above, the true contribution of pTregs to the Treg pools and function remains incompletely defined. There has been much enthusiasm to extrapolate from in vitro results, to consider that any local expansion of Treg cells represents a conversion event, and generally that pTregs are a branch of T-cell differentiation akin to Th1/2/17 fates (Curotto de Lafaille et al. 2008). Yet, only in a few instances has the in vivo relevance of pTregs been tested with transfers of rigorously purified Tconv cells, a test essential to distinguish true pTreg generation from the expansion of preexisting tTregs. The lack of specific molecular identifiers of pTregs has also been a confounder. It was suggested that expression of the transcription factor Helios (Ikzf2) might serve as a marker of thymus-derived Tregs (Thornton et al. 2010). Unfortunately, it is now realized that Helios is expressed at substantial levels in pTregs induced peripherally by lymphopenia or agonist peptide (Verhagen and Wraith 2010; Darce et al. 2012; Gottschalk et al. 2012) and is generically induced upon activation of CD4⁺ T cells (Akimova et al. 2011; Serre et al. 2011). Analyses of TCR repertoires in unchallenged mice have shown a high degree of similarity between Tregs in the thymus and peripheral lymphoid organs (Hsieh et al. 2006; Pacholczyk et al. 2006; Wong et al. 2007b), suggesting that tTregs constitute the majority of those pools (there is admittedly a caveat to this argument, which assumes that there is little or no colonization of thymic pools by recirculation from the periphery). Similar observations were made in the context of autoimmune lesions, in which Tregs seem to mainly result from recruitment and amplification of preexisting tTregs, rather than from local conversion (Korn et al. 2007; Wong et al. 2007a; Liu et al. 2009). It will be important, in responses to infection or tumors, to obtain a better understanding of the relative contribution of the expansion of preexisting tTregs relative to de novo pTreg generation. It has been suggested that tTregs and pTregs may come into play at different times of Toxoplasma gondii lesions, pTregs perhaps becoming important in later stages, after initial control by preexisting Tregs (R Maizels, pers. comm.); on the other hand, the high prevalence of Tregs in chronic Leishmania cutaneous lesions seems solely attributable to tTregs (Suffia et al. 2006). A mutant mouse bearing a deletion in the CNS1 enhancer region of the Foxp3 locus necessary for peripheral induction showed localized Th2-driven pathologies at mucosal sites, rather than the generalized autoimmunity of fully Treg-deficient animals, indicating that pTregs may preferentially come into play at mucosal interfaces. In a further application of pTregs at organismal boundaries, it has recently been suggested that pTregs have an elective function at the foeto–maternal interface (Samstein et al. 2012).

Besides in vivo-generated pTregs, Foxp3⁺ cells can be very easily induced in vitro by TCR-mediated activation of naïve T cells in the presence of TGFβ and IL2 (“TGF-Treg”) (Chen et al. 2003; Fantini et al. 2004; Peng et al. 2004). There has been a tendency to generalize from these results to Treg physiology in vivo, and to assume that TGFβ is an important driver of Treg differentiation, which would preferentially occur in locales with higher TGFβ levels. Yet, the functionality and relevance of these in vitro-generated Foxp3⁺ cells have been questioned. Although Foxp3 expression is high in TGF-Tregs it is very unstable because of, or reflected by, incomplete CpG demethylation at the Foxp3 locus (Floess et al. 2007; Huehn et al. 2009). Perhaps not surprisingly given this instability, their functional ability has proven variable (Chen et al. 2003; Davidson et al. 2007; Floess et al. 2007; Hill et al. 2007; Aricha et al. 2008; Huter et al. 2008), although some of this variability may have reflected genetic differences between inbred strains (D’Alise et al. 2011). Deletion of a conserved enhancer motif in Foxp3 that is needed for response to TGFβ results in only modest effects on Treg numbers in vivo (mainly as a partial reduction of Treg numbers in the gut) (Zheng et al. 2010), and TGFβ proved unnecessary for thymic Treg differentiation, except in the neonatal period (Liu et al. 2008). Retinoic acid also enhances the activation of the Foxp3 locus in vitro (Benson et al. 2007; Coombes et al. 2007; Mucida et al. 2007; Sun et al. 2007; Elias et al. 2008), perhaps through direct enhancement or indirectly by curtailing inhibitory cytokines (Hill et al. 2008; Nolting et al. 2009), although in vivo effects are far more complex owing to retinoic acid's pleiotropic effects on many cell types, including effector cells (Hall et al. 2011).

In the thymus, maturing T cells at the double-positive (DP) and single-positive (SP) stages are selected into the Treg lineage, as an alternative to Tconv fate, as one of the “clonal deviation” escape pathways together with NKT cells or CD8αα IEL precursors. Treg selection requires TCR:MHC (major histocompatibility complex) molecular interactions, as evidenced by their reduction in MHC-deficient mice (Krajina et al. 2004; Bienvenu et al. 2005; Fontenot et al. 2005b; Stephens et al. 2007), but also requires costimulatory signals from CD28, which seems to amplify the probability that cells expressing a TCR committing them to Treg fate are actually selected (Salomon et al. 2000; Tai et al. 2005; Lio et al. 2010). Treg differentiation follows a two-step process, through a FoxP3-negative CD25^hi intermediate that secondarily converts to FoxP3^hi under the influence of trophic cytokines, in particular, IL2 (Burchill et al. 2008; Lio and Hsieh 2008). Engagement of the TCR by agonist ligands strongly favors the selection of Treg cells, either by inducing differentiation along the lineage (Jordan et al. 2001; Apostolou et al. 2002; Kawahata et al. 2002; Walker et al. 2003), or because FoxP3⁺ cells are inherently more resistant to clonal deletion (Liston et al. 2003; Van Santen et al. 2004; Bonasio et al. 2006), or both. It would be an oversimplification, however, to conclude that all Treg cells are necessarily self-reactive (Pacholczyk et al. 2007; Dipaolo and Shevach 2009), if only because “self-reactivity” is a relative concept, highly influenced by the mode of self-antigen presentation and the state of the responding cell. Indeed, repertoire analyses show that a significant proportion of TCR sequences are used by both Treg and Tconv cells (Hsieh et al. 2006; Pacholczyk et al. 2006; Wong et al. 2007b), indicating that a number of Treg cells are no more self-reactive than are Tconv cells.

In addition, recent work has shown a very strong degree of intraclonal competition between Treg cells expressing the same or related TCRs, which greatly limits the probability of similar Tregs to differentiate in concert (Bautista et al. 2009; Leung et al. 2009). This observation explains the long-standing riddle of the quasi-absence of Treg cells in TCR transgenic mice on a RAG-deficient background. These niche limitations may stem from strict competition for limiting ligands, or perhaps from the propensity of Tregs to suppress other cells in their vicinity, which might include other Tregs as well. Interestingly, strong control on niche size for thymic selection also seems to apply for peripheral homeostasis (Nishio et al. 2010).

In an oft-repeated but largely unsupported assertion, Treg differentiation is presented as occurring in a window of affinity of TCR engagement by MHC molecules, higher than that required for positive selection of Tconv cells, but lower than that inducing clonal deletion. Results in support of this observation have only recently come from the analysis of a Nurr77-GFP reporter whose expression correlates with the strength of the TCR signal: the reporter was consistently expressed at a higher level in Tregs than in conventional CD4⁺ T cells (Moran et al. 2011). Contradicting this simple model, however, are studies in which a low-affinity agonist variant that remained active in negative selection was unable to enhance Treg differentiation (Cozzo et al. 2011), an observation incompatible with a straightforward “window of affinity” interpretation. Rather, commitment to the Treg lineage may reflect a particular balance and intensity of TCR signals different from that eliciting apoptosis or Tconv differentiation. Activation of the NF-κB pathway leading to c-Rel seems particularly important for Treg cell differentiation, more so than for Tconv cells, because deficits in several elements that connect the TCR to NF-κB perferentially curtail Treg differentiation. Defective Treg cell selection is observed in knockouts of PKCθ, BCL10, CARMA1 TAK1, IKK2, and cREL (Schmidt-Supprian et al. 2003, 2004; Sato et al. 2006; Wan et al. 2006; Barnes et al. 2009; Long et al. 2009; Medoff et al. 2009; Ruan et al. 2009). The level of NF-κB activation in the thymus correlates with FoxP3 induction, and Treg differentiation in TAK1- or CARMA1-deficient mice can be rescued by transgenic expression of a constitutively active IKK2 (Long et al. 2009). The presence of a cREL binding site in the CNS3 segment of the first intron of Foxp3 may provide a landing point for this axis (Zheng et al. 2010), but it is also conceivable that NF-κB promotes Treg differentiation by modulating other TFs, such as Hif1α.

Opposing this positive influence of the NF-κB pathway, signaling through the AKT pathway dampens thymic and peripheral Treg differentiation. Constitutively active Akt impairs thymic differentiation of Treg cells as well as their induction by TGFβ (Haxhinasto et al. 2008; Sauer et al. 2008), consistent with a positive effect of rapamycin on Treg cell selection and expansion (Zheng et al. 2003; Battaglia et al. 2005; Qu et al. 2007; Strauss et al. 2007; Haxhinasto et al. 2008; Sauer et al. 2008). This activity was attributed to the mTORC2 complex (Delgoffe et al. 2011), and may proceed through inactivation of Foxo1/3, which are needed for optimal FoxP3 induction (Kerdiles et al. 2010; Ouyang et al. 2010). This negative impact of the PI3K/Akt pathway may also stem from activation of cell metabolism. Concordant with this view, Hif1α has recently been suggested to promote Th17 differentiation and to counterregulate Tregs by targeting FoxP3 for degradation, and perhaps also by favoring active metabolic conditions in the cell that are more favorable to Tconv expansion (Dang et al. 2011; Shi et al. 2011). Similarly, NR4a1 (a.k.a. Nur77), part of the immediate-early response to cell activation, also antagonizes Treg selection, again in the context of transcriptional activation of glycolytic pathways (Fassett et al. 2012). Thus, one might speculate that selection into the Treg cell lineage depends not so much on absolute strength of TCR signals as on the relative balance of NF-κB and of metabolic activation. The Treg phenotype might be favored by activation in a context of relative starvation of the differentiating cell. Such a view would also be consistent with the notion of peripheral induction of pTregs by suboptimal presentation in the absence of costimulatory signals, which also favor full metabolic activity. (In contrast, the requirement for CD28 during Treg selection in the thymus, which involves both a trans dependence on costimulation for the production of IL2 and a cell-autonomous cis requirement in Treg precursors [Tai et al. 2005], may reflect the need to amplify weaker TCR signals.)

FOREVER Tregs?

Phenotypic stability is an important consideration for Treg cells. The self-reactivity of the TCR expressed by many of them makes it important for their suppressive phenotype to be stable and maintain the expression of FoxP3 and the suppressor pathways it controls, lest they convert into aggressive effectors. Indeed, one hypothesis proposed that much of the pool of autoreactive T cells in autoimmune disease may consist of “ex-Treg” cells that turned rogue after losing FoxP3 expression (Zhou et al. 2009; Bailey-Bucktrout and Bluestone 2011). Initial support for Treg instability stemmed from transfer of congenically marked Treg cells into alymphoid hosts, which resulted in sizable FoxP3-negative populations (Duarte et al. 2009; Tsuji et al. 2009). These experiments carried the caveat of the artificial drive of the lymphopenic environment, and/or of homeostatically driven expansion of small numbers of contaminating Tconv cells in the inoculum. This was a particular concern because such results only concerned a fraction of FoxP3⁺ cells (Komatsu et al. 2009), and were not observed after transfers into normal hosts, even in the context of a strong autoimmune disease (Rubtsov et al. 2010). Lineage tracing experiments based on Foxp3-cre transgenes then showed that many cells in the CD4⁺ effector/memory pools had, at some point in their life history, activated the FoxP3 locus (Zhou et al. 2009). There was a pitfall to these experiments, however, in that the tracer system was continuously active, and would tag cells with a transiently active Foxp3 locus. Indeed, when similar lineage tracing experiments were performed in pulse-chase mode with a Tamoxifen-controlled Foxp3-creERT transgene that tagged FoxP3-expressing cells only during a defined time frame, the Treg phenotype of their progeny was found to be extremely stable over time, even in conditions of ongoing inflammation (perhaps with the exception of IL2 quenching) (Rubtsov et al. 2010). Thus, the large number of labeled effector/memory cells in the initial experiments likely reflected the accumulation over time of Teff cells that transiently expressed FoxP3 at the onset of activation, as has been well documented in human CD4⁺ cells. This interpretation was recently confirmed independently (Miyao et al. 2012). This transient activation of the Foxp3 locus also implies that results from conditional knockout experiments that use Foxp3-cre should be interpreted with some caution, particularly when a strong drive may affect the population dynamics of cells bearing the genetic excision. Although the vast majority of Tregs likely remain Tregs (at least over the lifetime of a mouse), it is probably worth keeping in mind that Tregs can lose FoxP3 and suppressive activity when mistreated (e.g., local IL2 deprivation). This possibility needs to be taken into account when devising therapies based on Treg transfer (Brusko et al. 2008).

DETERMINANTS OF Treg FUNCTION IN VIVO

Several elements condition the functional ability of Treg cells to perform their inhibitory functions:

1. Specificity. The overall TCR repertoire of Tregs is essentially as diverse as that of Tconv cells, and both have the same capacity for broad recognition of the antigenic universe. This TCR specificity is essential in directing Treg inhibitory activity, via direct killing of B or dendritic cells (DCs) that present the cognate antigen, or via bystander suppression of neighboring T cells responding to the same antigen source. As for other T or B lymphocytes, antigen-specific receptors expressed by Tregs can drive clonal expansion in response to exogenous antigens (Nishimura et al. 2004; Suffia et al. 2006). Treg suppression does not require matching epitopes on the T cells it suppresses, however, as shown in vitro (Takahashi et al. 2000) or by the dominant protection of a polyclonal autoimmune infiltrate afforded by monospecific Tregs (Tang et al. 2004; Tarbell et al. 2004; Chen et al. 2005). This focusing of Treg action is essential, however, and Tregs displaying a TCR specific for a particular self-antigen are far more effective at inhibiting the relevant autoimmune destruction than are polyclonal populations (Tang et al. 2004; Tarbell et al. 2004).

2. Location is essential for Tregs to control antimicrobial or autoimmune responses. In theory, Tregs could affect the generation of effector lymphocytes at several different points of their activation process: during the initial priming and differentiation in draining lymph nodes, during their migration to target organs, or at sites of tissular inflammation. Arguments have been presented for an effect on initial activation (Tang et al. 2006), cellular migration (Sarween et al. 2004; Davidson and Shevach 2011), or the control of terminal inflammation in the target organ (Chen et al. 2005; Suffia et al. 2006; Feuerer et al. 2009c). Indeed, Tregs establish a local balance with autoreactive cells, or a three-way balance with the infecting agent and the inflammatory response, which results in limitation of both immunopathology and damage by the infectious agent. TCR specificity can contribute to this homing to inflammatory locales, helping to attract and/or retain specific Tregs. Receptor specificity is complemented by an array of chemokine receptors on Tregs (Siegmund et al. 2005; Wei et al. 2006), and different Treg subpopulations express a range of chemokine receptors (Feuerer et al. 2010) that promote their differential tissue localization. For instance, CCL28 secreted by hypoxic tumors recruits CCR10⁺ Tregs (Facciabene et al. 2011), CXCR5 is expressed by Tregs that control germinal center (GC) B cells (Linterman et al. 2011), and CCR9⁺ and/or CX3CR1 are required for Treg cells to home and expand in the gut lamina propria (Menning et al. 2010; Hadis et al. 2011).

3. Effector strategies. A broad array of molecular mechanisms is used by Tregs, which involve molecules that are typical members of the Treg signature: release of inhibitory cytokines (TGFβ, IL-10, IL-35) and their carriers (Lrrc32, a.k.a. GARP), deprivation of trophic cytokines (in particular, IL2 via expression of the high-affinity IL2Rα), direct killing of dendritic cells (GrnzA/B), suppression of DC maturation (LAG-3), stripping of costimulatory ligands from APCs (CTLA-4), cAMP- or adenosine-mediated inhibition (CD39, CD73), and amino acid deprivation through induction of indoleamine 2,3-dioxygenase in DCs (Vignali et al. 2008; Shevach 2009). This diversity is probably linked to the range of cellular targets, which include cells of the innate immune system (dendritic cells, natural killer [NK] cells) as well as adaptive T or B lymphocytes. It also raises the possibility that Treg defects that partake in the determinism of autoimmune disease might be focalized, affecting only a facet of Treg function, rather than a generalized reduction in Treg numbers or activity. A specific defect in Treg activity via CD39 has been suggested to contribute to multiple sclerosis (Fletcher et al. 2009). Such focalized defects might not be apparent when assessing Treg function through tests that may primarily reflect another of these suppressive pathways.

FoxP3⁺ Treg CELLS: DIVERSITY

Correspondingly, a number of Treg subphenotypes have been delineated, with differential effector functions and tissue localization (reviewed in Feuerer et al. 2009b). There is really no such thing as a generic Treg cell. Indeed, Treg cells can even secrete proinflammatory cytokines such as IFNγ or IL-17 in contexts of microbial or parasitic challenge (Oldenhove et al. 2009), and it will probably be recognized that, in some instances, FoxP3⁺ Tregs can actually have effector rather than suppressive function.

In addition, different facets of Treg cells are brought to bear to control specialized Th effector functions. Interestingly, these programs appear to be determined in Treg cells by the same transcription factors that are central to the differentiated functions of the Tconv cells they regulate. For instance, Irf4 is required for the differentiation of B cells and of the Th2/17-type cells that help them, and the absence of Irf4 in Tregs impairs their ability to limit Th2 responses and antibody production (Zheng et al. 2009). Similarly, Treg cells expressing Tbet or STAT3 optimally suppress inflammatory Th1 and Th17 responses (Chaudhry et al. 2009; Koch et al. 2009). Treg cells can also use the Bcl6-directed transcriptional module (including Cxcr5 or PD1) that controls entry into GCs, where they limit the number of B cells, favoring the selective expansion of antigen-specific B cells (Linterman et al. 2011).

These alternative effector functions have direct relevance to pathology, as illustrated by a FoxP3 mutant mouse that is highly resistant to Th2/17-dependent arthritis, but at the price of increased susceptibility to Th1-dependent diabetes (Bettini et al. 2012; Darce et al. 2012). Some of these target-specific functions, such as those of GC or adipose tissue Tregs discussed below, correspond to distinct cell phenotypes, which include chemokine receptors needed to migrate to and reside in those locations, together with effector molecules needed there. But others are ubiquitously expressed pathways shared by all Tregs: the expression of Irf4 and the preferential suppression of Th2/17 that it controls appear broadly distributed among Tregs.

In addition, it is important not to merely consider Tregs as obtuse suppressors of any form of activity in other immune cells. Indeed, Tregs can contribute positively to the specificity and efficacy of immune responses. Acute Treg lineage ablation leads to more aggressive genital herpes infection, with insufficient interferon production and NK recruitment in the infected tissue, owing to a florid but ultimately unhelpful response that takes place in the draining lymph node (Lund et al. 2008). They can enhance the affinity maturation of the antibody response (Linterman et al. 2011). Paradoxically, given the common perception that Treg and Th17 cells represent balanced forks of differentiation, in some settings Tregs actually promote acute Th17 responses to antigen, likely by preventing the inhibitory effects of IL2 (Chen et al. 2011).

Even further from Treg’s home base of immune tolerance are the FoxP3⁺ cells found in nonlymphoid settings of inflammation, such as the atherosclerotic plaque (Ait-Oufella et al. 2006) or “fat Tregs” that colonize the visceral adipose tissue (Feuerer et al. 2009a). Although bona fide Tregs judging by their phenotype and transcription profiles, they also express distinct transcripts that reflect homing and adaptation to the adipose tissue. In a new role, these Tregs influence metabolic parameters such as insulin resistance by dampening inflammation in the adipose tissue. Other extralymphoid Treg populations may yet to be discovered, a common theme being perhaps an attraction of Treg cells for settings with NF-κB activation, found in infected locales, as well as sterile inflammation such as encountered in the adipose tissue.

A unifying perspective may be that the overall physiological function of Tregs is not really to maintain immune tolerance, but more generally to ensure tissue homeostasis (where “tissue” includes the commensal microbial self) and to bring inflammation to proper resolution. Such a definition would account well for the control exerted by Tregs on numbers and activity of T, B, NK, or adipose macrophages. In the same vein, the syndromes that develop in scurfy mice and IPEX patients are perhaps due to unconstrained homeostasis and inflammation, rather than to autoimmunity through lack of tolerance.