In recent years, the understanding of regulatory T cell (Treg cell) biology has expanded considerably. Key observations have challenged the traditional definition of Treg cells and have provided insight into the underlying mechanisms responsible for the development of autoimmune diseases, with new therapeutic strategies that improve disease outcome. This Review summarizes the newer concepts of Treg cell instability, Treg cell plasticity and tissue-specific Treg cells, and their relationship to autoimmunity. Those three main concepts have changed the understanding of Treg cell biology: how they interact with other immune and non-immune cells; their functions in specific tissues; and the implications of this for the pathogenesis of autoimmune diseases.
More than 20 years after their ‘re-discovery’, regulatory T cells (Treg cells) have emerged as an important component in understanding of the immune response to pathogens and the mechanisms of peripheral tolerance that control the development of allergies and autoimmune diseases. In mice and humans, Treg cells are characterized by high expression of the cytokine IL-2 receptor α-chain (CD25) and expression of the transcription factor Foxp3, which is required for their development, function and stability1,2,3,4. In humans, as CD25 is also expressed by activated CD4+ T cells, absence of the IL-7 receptor α-chain (CD127) is used as a marker complementary to CD25 to more precisely identify human Treg cells5. Moreover, numerous surface receptors have been described that are variably specific for defined Treg cell subsets, indicative of the heterogeneity of this population6. Treg cells can be broadly classified into two groups on the basis of their developmental origin. Thymic Treg cells (tTreg cells), also known as ‘natural Treg cells’, are generated in the thymus as a separate lineage at the stage of CD4+ single-positive thymocytes and are thought to show enrichment for T cell antigen receptors (TCRs) with high affinity for self peptides7. Although the details of their suppressive mechanisms are still not completely understood, and these mechanisms are most probably dependent on the microenvironment and the target population to be controlled, in general these cells perform their function both by cell-contact mechanisms that involve specific cell-surface receptors and by the secretion of inhibitory cytokines such as IL-10, TGF-β and IL-357. Induced Treg cells (iTreg cells) develop from conventional CD4+ T cells in the periphery after antigen encounter and in the presence of specific factors such as TGF-β and IL-28. So far there is (are) no definitive protein marker(s) that distinguish(es) between these two Treg cell populations in vitro or in vivo, although there are important differences in their epigenetic signatures, and particularly in the Foxp3 locus, that make iTreg cells intrinsically unstable in inflammatory and/or stress conditions8,9.
From a historical perspective, the initial observations made in the 1970s that led to the definition of Treg cells in the 1990s are tightly linked to autoimmune diseases. Seminal work by Nishizuka and Sakakura back in 1969 demonstrated that thymectomy in healthy neonatal mice led to inflammation and severe organ-specific autoimmune pathology, which indicated the involvement of a thymus-derived population in control of self-tolerance10. That observation was further confirmed in adult rats that underwent thymectomy and were subjected to sublethal irradiation11. Most notably, inoculation with CD4+ T cells from healthy syngeneic mice inhibited disease in both systems, and once autoimmunity was established, CD4+ T cells isolated from these sick mice and adoptively transferred into T cell–deficient mice were able to induce disease12. These experiments demonstrated that T cells are able not only to serve as inducers of autoimmune disease but also to inhibit such disease. That led to the hypothesis that in the periphery of normal mice there are two populations of CD4+ T cells: one population potentially capable of inducing autoimmunity, and a second population of ‘suppressor’ cells, different from helper T cells, that would inhibit autoreactive T cells. Treg cell research during those years floundered, in part due to the failure to find specific cell markers that could define this population and the ambiguity observed in their suppressive mechanisms. The identification of a population of CD4+ T cells responsible for controlling autoreactive responses13, as well as the search for a cell-surface marker that would define this population, indicated that a subset of CD25-expressing CD4+ T cells present in the periphery of healthy mice was responsible for the inhibition of autoimmunity. The transfer of cell suspensions from the spleen of BALB/c mice depleted of CD25+ cells into athymic nude mice induced autoimmunity that affected several organs, and co-transfer of CD25+CD4+ T cells inhibited disease4. In 2001, human Treg cells were identified in the thymus and peripheral blood of healthy people as those CD4+ T cells with very high expression of CD25 and phenotype and function very similar to that of their rodent counterpart14,15. Although there is still no definitive marker for the isolation of human Treg cells, these studies, together with the discovery of Foxp3 as the master transcription factor of Treg cells1,3,16, laid the groundwork for the beginning of in-depth analysis of Treg cell biology in health and disease (Box 1).
As for Treg cell function in human autoimmune disease, early studies of patients revealed that in various autoimmune disorders, there is a defect in either the number or the function of Treg cells isolated from the peripheral blood, with those data being supported by in vivo models of disease (a comprehensive list of autoimmune diseases is in ref. 17). Although some of these early data were misleading due to the confounding variable of the identification of human Treg cells based on the positive expression of CD25 as the only marker for Treg cell identification, subsequent work has clearly demonstrated that most autoimmune diseases, such as type 1 diabetes18,19,20, multiple sclerosis21,22, systemic lupus erythematosus23, myasthenia gravis24, rheumatoid arthritis25 and others17, display defects in either the number or the function of tTreg cells in peripheral blood.
Investigations during the past two decades have advanced knowledge of the mechanisms that underlie the development of tTreg cells in the thymus, as well as their gene-expression signature, their role in controlling both immune and non-immune cell responses and their effect on immune system–mediated diseases (reviewed in refs 7,26), and, finally, the identity of the tissues in which they exert functions beyond suppression27,28. Moreover, the traditional view that Treg cells are a terminally differentiated population not capable of secreting pro-inflammatory cytokines and whose only function is to suppress T cell responses has been challenged by recent data. Specifically, Treg cells possess some degree of plasticity and instability, although in-depth understanding of the molecular mechanisms that drive these two states and the relationship between them in disease settings remains incomplete.
This Review will provide an update on recent discoveries in Treg cell biology in the context of autoimmune disease. From our perspective, three main observations have changed the understanding of what Treg cells are, how they function in peripheral lymphoid organs and non-immune tissues, how they relate to other immune and non-immune cells, and how their phenotype and function can be modulated, with clear consequences for the development of new therapeutic strategies for various autoimmune diseases. These observations are as follows: 1) the instability of Treg cells and their acquisition of an effector phenotype after losing Foxp3 expression under inflammatory conditions; 2) the plasticity of the phenotype of Treg cells, with their acquisition of effector-like properties while maintaining Foxp3 expression; and 3) the discovery of tissue-specific Treg cells, which demonstrates again that Treg cells, like other immune cells, are influenced by their environment. These observations have arisen at a time during which the development of high-throughput genomic, epigenetic and proteomic technologies is allowing the analysis of rare cell populations at a single-cell level, which will undoubtedly improve knowledge of basic Treg cell biology in general and of human Treg cell biology in particular.
Foxp3 as a master regulator of Treg cell phenotype and function
The discovery of Foxp3 as the master regulator of Treg cell development and function was critical for the understanding of Treg cell biology1,2,3. Inactivating mutations in Foxp3 result in the spontaneous development of severe autoimmunity with a scurfy phenotype in mice29 and IPEX syndrome (‘immune dysregulation, polyendocrinopathy, enteropathy, X-linked’) in humans30,31. Foxp3 is necessary for the development, maintenance and function of tTreg cells1,2,3, although alone it is not sufficient for full recapitulation of the Treg cell phenotype9,32. Beyond Foxp3, the second requirement necessary for establishment of the Treg cell functional program is the generation of a specific epigenetic signature acquired during development and finalized in the periphery9,33. Both have essential roles in maintaining Treg cell function, and alteration of Foxp3 or epigenetic modifications in auto-inflammatory conditions are probably the cause of the Treg cell instability and aberrant plasticity observed in several autoimmune settings (Fig. 1). Due to the importance of Foxp3 in Treg cell maintenance and the prevention of autoimmunity, regulation of Foxp3 expression is a matter of active research. Foxp3 is subjected to two main layers of regulation, transcriptional and post-translational, both of which are responsive to positive and negative regulation by factors in the tissue environment, including cytokines, metabolic mediators and inflammatory factors.
Foxp3 expression is a tightly regulated process, and while the mechanisms by which Foxp3 is transcriptionally regulated by transcription factors during tTreg cell development and in mature Treg cells have been clearly established (reviewed in ref. 34), data have shown that in rodents, Foxp3 transcription is also controlled epigenetically. The Foxp3 locus contains several conserved non-coding sequences (CNSs) that are critical for the initiation and maintenance of Foxp3 transcription35. Among the three CNSs described so far, only CNS2 has been demonstrated to prevent autoimmunity. It is a TCR-responsive enhancer with binding sites for Runx1–CBFβ transcription-factor complexes that is important in maintaining Foxp3’s stability. CNS2-deficient mice develop spontaneous autoimmunity, which emphasizes the importance of CNS2 in the stability and function of Treg cells36,37. Furthermore, CNS2 contains a conserved CpG island (TSDR) that is hypomethylated specifically in Treg cells; thus, it is transcriptionally active in tTreg cells but is hypermethylated in naive or effector T cells38,39. This region has been widely used to distinguish true tTreg cells from conventional T cells transiently upregulating Foxp3 expression and iTreg cells38,39.
Three main post-translational modifications have been described for Foxp3 protein: acetylation, phosphorylation and ubiquitination. These modifications affect the stability and DNA-binding capacity of Foxp3 and thus modulate Treg cell function and the development of autoimmunity. The acetylation of specific lysine residues by lysine acetyltransferases globally stabilizes Foxp3 expression and promotes Treg cell function by favoring the binding of Foxp3 to its transcriptional targets40 and allowing it to avoid proteasomal degradation40. Several acetylases and deacetylases, such as TIP6041 and p30040,42, have been shown to interact with Foxp3 and control its acetylation. Other post-translational processes include phosphorylation at serine and threonine residues by several kinases, including PIM-1, PIM-2 and CDK243,44,45,46, and ubiquitination at lysine residues, which targets Foxp3 for proteasomal degradation47. Inflammatory stimuli result in proteasome-dependent degradation of Foxp3 mediated by the ubiquitinase Stub1, which binds to Foxp3 and promotes its Lys48-linked ubiquitination48. Overexpression of Stub1 abolishes the suppressive ability of Treg cells in vitro and in vivo and confers onto Treg cells a phenotype with similarities to that of the TH1 subset of helper T cells; this raises the question of whether the effector T cell–like Treg cell phenotype in this setting is an intermediate stage en route to instability and loss of Foxp3 protein expression. In contrast, the deubiquitinating enzyme USP7 has high expression in Treg cells and is associated with Foxp3 in the nucleus, where it regulates the turnover of Foxp3. Under inflammatory conditions, this enzyme is downregulated, which facilitates the degradation of Foxp340. Furthermore, conditional deletion of USP7 in Treg cells leads to lethal autoimmunity with a decreased number of Treg cells in the periphery that display an aberrant TH1-like phenotype in vitro and in vivo49. Targeting USP7 decreases recruitment of the acetyltransferase TIP60 to the CNS2 region of Foxp3. TIP60 promotes acetylation-dependent dimerization of Foxp350,51, and in the absence of TIP60, lethal autoimmunity develops52. Other factors have also been shown to control Foxp3 expression, such as HIF-1α, which is induced by stimulation with IL-6 and the TCR and inhibits Foxp3 through ubiquitination53.
Both ubiquitination and acetylation target lysine residues, so they might compete in the regulation of Foxp3 expression. In fact, hyperacetylation of Foxp3 prevents its polyubiquitination and proteasomal degradation, which increases its stability40. Such data suggest that post-translational modifications of Foxp3 have a crucial role in modulating the plasticity or instability of Treg cells, which adds another layer of complexity to regulation of the Treg cell functional program, with potential consequences for the development of autoimmunity.
Foxp3 works in concert with other transcription factors and proteins, forming multiprotein complexes that determine the transcriptional signature and effector functions of Treg cells. Hundreds of protein partners have been described, including the transcription factors Gata-3, NFAT, Runx1, Eos and others54. Global analysis of the Foxp3-interaction network suggests a model in which Foxp3 and its binding partners form multi-protein complexes that bind to pre-existing DNA enhancers55 and regulate transcription positively or negatively depending on the interacting proteins recruited56. As mentioned above, Foxp3 also associates with proteins that mediate epigenetic modifications, such as TIP60, Sirtuin 1 or HDAC741,57, which alter the acetylation state of partner loci and Foxp3 itself, with consequences for the binding of transcription factors and histone modifications. In this context, disruption of Foxp3’s interactions with specific proteins diminishes Treg cell function and leads to autoimmune responses due to increased polyubiquitination of Foxp358,59. Other evidence supporting the proposal of the importance of Foxp3-interacting partners in Treg cell function includes the observation that different FOXP3 mutations result in a wide range of IPEX disease severity, reflective of the relative importance of the residues affected in the integrity of the Foxp3 protein and the protein partners that form the DNA-binding complexes. For example, the most common mutation in IPEX, p.A384T, which disrupts the sequence specificity of Foxp3’s DNA binding and alters Foxp3’s interactions with specific target genes31, inhibits Treg cell function but preserves the ability of Treg cells to repress inflammatory cytokine production, due in part to a specific inhibition of Foxp3’s interaction with TIP6060. Furthermore, experiments with mice have shown that this mutation specifically in Treg cells perturbs the binding of Foxp3 to specific target genes, including Batf, which is partly responsible for the induction of a unique pattern of tissue-restricted inflammation in certain non-lymphoid tissues due to defective function of these Treg cells61. Additionally, a Foxp3 reporter mouse that expresses a fusion of Foxp3 with enhanced green fluorescent protein (EGFP) at its amino terminus (EGFP-Foxp3), which disrupts the interaction of Foxp3 with many cofactors, including TIP60, p300 and Eos, does not develop apparent autoimmunity, but its Treg cells display alterations in function in vivo, with autoimmune-prone mice of the non-obese diabetic (NOD) strain developing diabetes faster than their wild-type counterparts do58. Interestingly, these same Treg cells are potent suppressors of antibody-mediated arthritis due to ‘preferential’ interaction of EGFP-Foxp3 with the transcription factor IRF462. These findings not only demonstrate that certain cofactors are crucial for Treg cell–mediated function but also suggest that Treg cells might be ‘tuned’ to control particular types of inflammation by modulating the constituents of Foxp3 protein complexes under specific environmental conditions. Furthermore, we are tempted to speculate that the variety of functions that Treg cells perform in different tissue environments could be accompanied by the formation of specific multi-protein complexes with tissue-specific proteins that would act cooperatively with Foxp3 in performing Treg cell functions in specific environments.
Treg cell instability
Both the expression of Foxp3 and its stability have crucial roles in the maintenance of Treg cell function63. Thus, conditional deletion of a Foxp3 allele in mature Treg cells results in effector T cells that are capable of causing inflammatory tissue lesions64. Although the instability of Foxp3 in iTreg cells has been widely observed and is intrinsic to their developmental origin8,9, tTreg cells have been investigated to determine how the instability of Foxp3 expression under basal or inflammatory conditions in specific tissues affects the development and resolution of autoimmunity65,66,67. Loss of Foxp3 expression by tTreg cells has been observed in vitro68,69, in the adoptive transfer of cells into lymphopenic hosts70, in infectious settings71 and in graft-versus-host disease72. A fate-mapping mouse model in which a yellow fluorescent protein reporter marks all cells that at any time expressed Foxp3 in both homeostatic conditions and autoimmune inflammatory conditions has been generated. In this model, a small proportion of apparently stable Treg cells lost Foxp3 expression and acquired an effector-memory phenotype with different levels of secretion of the pro-inflammatory cytokines IFN-γ and IL-1767. These ‘ex-Foxp3 cells’ were able to induce autoimmunity in an adoptive transfer model on the NOD background and consisted of a mixed population, on the basis of their level of TSDR demethylation, which would suggest that not all ex-Foxp3 cells in this model were once de facto tTreg cells. Subsequent data have suggested that this fate-mapping mouse model was detecting a proportion of Foxp3+ tTreg cells that were either transiently upregulating Foxp3 or were not fully committed to the tTreg cell lineage73. Further data in support of the proposal of tTreg cell instability have shown that Foxp3 expression is lost in tTreg cells specific for an epitope of myelin oligodendrocyte glycoprotein (MOG; amino acids 38–49) during the development of experimental autoimmune encephalitis (EAE), with an increased frequency of ex-Foxp3 cells in the central nervous system at the preclinical and peak stages of EAE that decreases during EAE resolution. These ex-Foxp3 Treg cells express IFN-γ and are able to transfer EAE74. It remains to be determined whether the decrease in ex-Foxp3 cells during disease resolution is due to re-acquisition of Foxp3 expression by ex-Foxp3 cells.
Another genetic fate-mapping mouse model has provided evidence that the majority of mature tTreg cells in the spleen and lymph nodes are relatively stable under homeostatic conditions66. This model, based on inducible labeling of Foxp3+ cells upon treatment with tamoxifen, marks all those cells that expressed Foxp3 at the moment of tamoxifen administration, and, in contrast to models with continuous labeling67, prevents the detection of cells transiently expressing Foxp3. Although Foxp3 expression is stable under homeostatic conditions, depriving cells of growth factor or blocking the IL-2 receptor, which induces autoimmunity75, results in substantially decreased Foxp3 expression per cell in mature tTreg cells, and a small population loses Foxp3 expression completely. However, they do not produce pro-inflammatory cytokines66, suggestive of some degree of instability under specific environmental settings. This apparent discrepancy might arise from the different fate-mapping mouse models used and the type of labeling of Foxp3 cells, which could lead to the labeling of uncommitted Treg cells67 or the absence of labeling of ex-Foxp3 cells that appeared before tamoxifen administration66. The discordance in results could also depend on the inflammatory stimuli used to assess Foxp3’s stability. Regardless, in both fate-mapping models, there is a small population of tTreg cells that lose Foxp3 expression, and those Treg cells that remain ‘stable’ display diminished Foxp3 expression at the single-cell level66. As decreased levels of Foxp3 in Treg cells isolated from inflammatory sites have been observed in mouse models of autoimmunity63,76 and in patients with autoimmune diseases24,77,78,79,80,81, further work with these models is warranted in order to identify the mechanisms and consequences of long-term decreases in Treg cell Foxp3 expression. Further data have confirmed the observation that most mature tTreg cells are stable under steady-state conditions, in a new fate-mapping mouse model in which Foxp3 lacks CNS1, but these cells become unstable when stimulated in vitro and in vivo in a model of EAE, losing Foxp3 expression and acquiring TH1 cell– and TFH cell–like features73. While epigenetic changes such as re-methylation of the CNS2 region could account for the loss of Foxp3 expression in these settings67,73, the molecular mechanisms responsible for the decrease in Foxp3 protein and the potential contribution of post-translational modifications of Foxp3 protein on the generation of ex-Foxp3 cells remain to be explored.
Finally, deletion of specific Foxp3 partners can also precipitate the appearance of ex-Foxp3 cells. For example, Treg cell–specific deletion of the chaperone GP96 on the NOD background leads to lethal autoimmunity due to defective suppressive abilities of Treg cells in models of diabetes and colitis. In this system, Treg cells progressively lose Foxp3 expression and gain IFN-γ secretion, although they maintain their specific TSDR-demethylation pattern59. Mice with Treg cell–specific deletion of the transcription factor Helios develop systemic autoimmune pathology characterized by increased germinal-center formation, lymphocytic infiltration into non-lymphoid organs and glomerulonephritis82. Although Helios does not form protein complexes with Foxp3 or bind to the Foxp3 locus82, Helios-deficient Treg cells have increased expression of IFN-γ and IL-17 and are unstable, with decreased expression of Foxp3 and a tendency to completely lose Foxp3 expression82,83. Treg cells deficient in another transcription factor, Eos, exhibit increased expression of IL-2 and IFN-γ along with reduced suppressive ability, while forced overexpression of Eos in Treg cells prevents Treg cell instability, even in inflammatory environments. EosloFoxp3+ tTreg cells are detectable in vivo and have regulatory function, with the ability to acquire helper T cell–like effector characteristics while maintaining Foxp3 expression; however, such cells exhibit specific changes in their global DNA-methylation pattern84.
Treg cell plasticity
Plasticity is a property inherent to most if not all immune cells, which allows them to adapt their phenotype and function to the changing environment and extracellular ‘danger’ signals. Thus, it is not surprising that Treg cells possess some degree of plasticity. Treg cells have the ability to acquire features specific to the type of immune response they control, which is driven mostly by ‘master’ transcription factors and is regulated by environmental signals. Thus, Treg cells acquire expression of the transcription factor T-bet to restrain type 1 inflammation during infection85,86, and they utilize the transcription factors IRF4 and STAT3 to inhibit TH2 cell responses87 and TH17 cell responses88, respectively. While this modality of plasticity seems to be advantageous for the host and beneficial to the outcome of the immune response, aberrant plasticity of Treg cells is also observed in several autoimmune diseases, with Treg cells expressing pro-inflammatory cytokines, acquiring helper T cell–like phenotypes and displaying diminished function in most cases but maintaining Foxp3 expression21,83,89,90,91,92,93,94,95. Paradoxically, these helper T cell–like Treg cells utilize the same transcription factors used by Treg cells to inhibit specific types of immune responses. Therefore, the secretion of IFN-γ by TH1-like Treg cells requires T-bet expression21,91,92,93, while IL-6-driven TH17-like Treg cells require STAT3 for the secretion of IL-1772, and IL-4-driven TH2-like Treg cells upregulate IRF4 and Gata-395,96. In most cases, helper T cell–like Treg cells have a demethylated TSDR in the Foxp3 locus even though they share effector features, which suggests that their phenotype might be reversible21,90,93. They display alterations in their epigenetic signature characteristic of those of Treg cells9,32,34,97, which might be the underlying mechanism that allows the secretion of pro-inflammatory cytokines. Current efforts are focused on understanding the signaling pathways that drive this plasticity in specific autoimmune diseases, to harness this flexibility to the treatment of human disease92 and the role of Treg cell plasticity in autoimmune disease–related tissues.
TH1-like Treg cells
Perhaps the best-characterized tTreg cell plasticity event is the acquisition of TH1 cell–like features. In mouse models and patients with autoimmune diseases, such as type 1 diabetes93, multiple sclerosis21,91, autoimmune hepatitis98 and Sjogren syndrome99, there is an increased frequency of IFN-γ+Foxp3+ tTreg cells in the periphery that display lower suppressive ability than that of Treg cells from healthy age-matched subjects. TH1-like Treg cells upregulate the transcription factor T-bet and other TH1 cell markers, such as CCR5 and CXCR3. Furthermore, in the apolipoprotein E–deficient (Apoe–/–) mouse model of atherosclerosis, which shares various pathogenic similarities with autoimmune disorders, there is accumulation of IFN-γ-producing TH1-like Treg cells in the aorta that display altered suppressive ability in vitro and in vivo90. Studies of an in vitro model in which TH1-like Treg cells are generated through the use of IL-12 have demonstrated that TH1-like Treg cells possess an activated PI(3)K and Akt kinase pathway and the transcription factor FoxO, which is partly responsible for their secretion of IFN-γ and decreased suppressive capacity91. Interestingly, Treg cells isolated from patients with relapsing–remitting multiple sclerosis also display an activated PI(3)K–Akt–FoxO pathway ex vivo, and their suppressive ability is corrected by blockade of the PI(3)K pathway94. In vivo, activation of PI(3)K–Akt by Treg cell–specific deletion of the PI(3)K phosphatase PTEN elicits a type 1 autoimmune disorder, with Treg cells downregulating their expression of CD25 and Foxp3 and displaying reduced functionality100,101. Moreover, FoxO itself has been linked to the regulation of Treg cell plasticity. Mice with Treg cell–specific deletion of FoxO succumb to a lethal autoimmunity similar to that observed in scurfy mice94, with Treg cells displaying a TH1-like Treg cell phenotype and being unable to prevent disease in a colitis model. IFN-γ seems to be involved in the defective function of Treg cells, as Foxo1–/–Ifng–/– mice partially recover from the wasting syndrome associated with this model94.
Transcriptomic analysis of IFN-γ+ TH1-like Treg cells at the population level91 or single-cell level90 has demonstrated that they exhibit lower expression of genes encoding immunosuppressive molecules than that of Treg cells and have altered expression of costimulatory molecules, migratory properties and specific signaling pathways. It is unknown whether TH1-like Treg cells have a role in disease pathogenesis and/or protection in specific tissues. TH1-like Treg cells have been observed among MOG-specific Treg cells infiltrating the central nervous system of mice during EAE development. These Treg cells are not able to suppress the infiltration of the central nervous system by MOG-specific effector T cells or prevent disease onset; they secrete IFN-γ at the onset and peak of disease but at a reduced frequency during the recovery phase102. In contrast, T-bet+Foxp3+ cells in pancreatic tissue seem to be protective in a mouse model of type 1 diabetes103. Finally, while most work on TH1-like Treg cells has suggested that IL-12 and/or type 1 cytokines induce the phenotype, the fact that PI(3)K–Akt, a major pathway that integrates diverse environmental signals into cell function, is involved in their generation suggests that other environmental cues might have the ability to induce Treg cell plasticity, as it has been observed, for example, with increased concentrations of dietary salt or NaCl104.
TH17-like Treg cells
A small proportion of human peripheral Treg cells produce IL-17 in healthy people and upregulate the gene encoding the transcription factor RORγt (RORC) (TH17-like Treg cells) ex vivo105 while conserving their suppressive ability105,106. Given the well-established developmental relationship between Treg cells and TH17 cells, it remains to be determined whether TH17-like Treg cells are a transient stage in the de-differentiation of tTreg cells into TH17 cells, as has been suggested69,107. In support of that proposal, the conversion of Foxp3+ Treg cells into TH17 cells has a crucial role in the pathogenesis of autoimmune arthritis in a collagen-induced arthritis mouse model. This conversion is driven by IL-1β108 and IL-6, and Foxp3+IL-17+ Treg cells are observed in the synovium of subjects with active rheumatoid arthritis109. Moreover, the conversion of Treg cells into TH17 cells has also been reported in the CD18hypo PL/J mouse model of psoriasis108, and TH17-like Treg cells have been observed in skin tissue of psoriatic patients89.
Perhaps the tissue in which TH17-like Treg cells have been best identified is the gastrointestinal tract. The lamina propria seems to be enriched for iTreg cells with high expression of the TH17 cell–defining transcription factor RORγt. These TH17-like Treg cells seem to have a beneficial function, as their absence exacerbates pathogenesis in several models of mucosal autoimmunity110,111. Furthermore, studies have suggested that Foxp3+RORγt+ Treg cells control glomerulonephritis112, and a lack of TH17-like Treg cells results in increased mortality and organ pathology associated with systemic lupus erythematosus113. In addition to the importance of IL-1β and IL-6 in promoting the secretion of IL-17 by Treg cells114,115, other environmental factors, including indoleamine 2,3-dioxygenase116, ligation of the pattern-recognition receptor TLR2117 and certain infections118, have been shown to modulate the conversion of TH17-like Treg cells either indirectly or directly.
TH2-like Treg cells
Patients with systemic sclerosis display an increased frequency of TH2-like Treg cells in the skin but not in the peripheral blood, characterized by the secretion of IL-4 and IL-13 and upregulation of Gata-3 and IRF-4. Peripheral Treg cells from these patients have high expression of ST2, the receptor for the alarmin IL-33. The skin shows enrichment for this cytokine, which suggests that it might have a role in the reprogramming of Treg cells into a TH2-like phenotype95. TH2-like Treg cells have also been observed in mutant (Il4raF709) mice susceptible to allergy and in Treg cells from the peripheral blood of food-allergic patients. TH2-like Treg cells secrete IL-4 and/or IL-13 and upregulate the transcription factors IRF4 and Gata-396. Moreover, mice with Treg cell–restricted deletion of the ubiquitin ligase Itch show autoimmune features, and Treg cells are not able to control type 2 inflammation. This defect is associated with the acquisition of a TH2-like phenotype by the Treg cells and with increased expression of Gata-3, activation of STAT6 and secretion of IL-4119.
Many questions remain to be answered about Treg cell plasticity. Does Treg cell plasticity reflect initial heterogeneity of the Treg cell population, with only a specific subset of Treg cells being able to acquire effector-like properties65? Do helper T cell–like Treg cells represent a transient stage on the path to becoming ex-Foxp3 cells? Can Treg cell plasticity be harnessed for the design of new therapeutic strategies aimed at modulating Treg cell function in diverse disease settings? What is the role of helper T cell–like Treg cells in autoimmune tissues during disease development and progression?
Tissue-resident Treg cells
One of the main recent advances in the Treg cell field has been the discovery that Treg cells populate specific tissues in the body during physiological and stress conditions. The characterization of these populations and their mechanisms of action will have important implications for understanding the development, maintenance and resolution of autoimmunity in specific organs. Tissue-resident Treg cells actively perform non-immunological functions and work at maintaining tissue homeostasis and wound repair, roles that could be important for tissue homeostasis in autoimmunity settings. Studies have begun to define the specific phenotype of tissue-resident Treg cells in the muscles, skin, lungs, gastrointestinal tract, liver and adipose tissue, among other locations (reviewed in ref. 28). In most cases, these Treg cells seem to be of thymic origin120,121,122, with considerable oligoclonality of their TCR repertoire, which indicates that particular antigens in the tissue might be responsible for the accumulation of Treg cells in specific niches120,122,123. They are characterized by the expression of tissue-specific transcription factors and mediators that drive the function of other cells in that tissue, in support of the notion that the microenvironment exerts important effects on the phenotype of Treg cells in a given location. The molecular mechanisms by which tissue-resident Treg cells acquire their tissue-specific program have only begun to be explored. Epigenetic analysis of mouse Treg cells isolated from various tissues and compared with Treg cells from the lymph nodes has demonstrated that tissue Treg cells undergo extensive epigenetic reprograming that is globally tissue specific, but that there is a common tissue Treg cell population characterized by expression of the activation marker KLRG1 and ST2124.
The best-studied tissue-specific Treg cell population is the one that is resident in visceral adipose tissue (VAT). VAT Treg cells serve important roles in the defense against associated metabolic disorders. In mice, lean fat tissue is populated by T cells with a TH2 cell phenotype and shows enrichment for Treg cells that maintain the predominance of resident anti-inflammatory macrophages. Treg cells accumulate over time and acquire a TH2-like phenotype, with expression of Gata-3, BATF, IRF4, CCR4 and IL-10121, as well as of PPARγ, a transcription factor that controls adipocyte differentiation and mediates the accumulation, phenotype and function of VAT Treg cells121,125. Treg cell–specific deletion of PPARγ results in the specific loss of VAT Treg cells, which demonstrates the important role of this transcription factor in the development and/or maintenance of VAT Treg cells121. The frequency of VAT Treg cells diminishes with age in obese mice, which suggests that they have a role in modulating obesity-associated fat-tissue inflammation123. Given that obesity is an established risk factor in autoimmune disease, it will be important to define the features and functions of VAT Treg cells in the development of autoimmunity.
Similarly, lung and muscle Treg cells express amphiregulin, the ligand for epidermal growth factor receptor, to promote tissue repair. Muscle Treg cells accumulate in acutely injured skeletal muscle in mouse models of muscle injury and muscular dystrophy, in which they control muscle inflammation after injury and promote tissue repair by acting on immune and non-immune cells27,120. Data in a zebrafish model further support the proposal of a dual role for Treg cells in the immunoregulation and tissue regeneration observed in mice. In zebrafish, Treg cell–like cells126 rapidly migrate to damaged organs in models of spinal-cord, heart and retina regeneration, and their function is dependent on the secretion of organ-specific regenerative factors. Treg cell–specific ablation inhibits organ regeneration127.
Treg cells populate healthy skin and help maintain homeostasis. In healthy people and rodents, skin Treg cells possess an effector–memory phenotype, with expression of Treg cell–specific surface markers and greater proliferative capacity ex vivo than that of their blood counterparts. They express IL-17 and IL-10 and display a demethylated TSDR, which would suggest that they are tTreg cells122. Interestingly, Treg cells reside in close apposition to hair follicles, where they contribute to the activation of hair-follicle stem cells by promoting their proliferation and differentiation through the Notch signaling pathway. Ablation of Treg cells completely abolishes the ability of the hair follicle to regenerate, which indicates that Treg cells are a key component of the skin stem-cell niche128.
The maintenance of local tissue homeostasis is of particular importance in the intestinal mucosa, where the immune system must be able to effectively discriminate between pathogens and dietary factors or commensal flora. Accordingly, deregulated immune responses to luminal flora in genetically susceptible people are generally recognized as key factors in the pathogenesis of inflammatory bowel disease. While the role of commensal flora in inducing Treg cells has been well characterized, the dynamics and characterization of Treg cells in the mucosal tissue has only begun to be appreciated. Thus, the lamina propria seems to show enrichment for Treg cells with high expression of RORγt110,129, and although most of these are not tTreg cells, they seem to have a positive function in maintaining tissue homeostasis, and their absence exacerbates pathogenesis in several models of mucosal autoimmunity.
How do Treg cells maintain tissue integrity, and what is the involvement of tissue-specific Treg cells in autoimmune disease? One could speculate that defects in specific tissue-resident Treg cells would be involved in the development, maintenance or resolution of autoimmune disorders that affect that specific tissue; for example, is there a specific role for muscle-resident Treg cells in the pathology of myasthenia gravis, or for VAT Treg cells in autoimmune disorders in which obesity is an important risk factor, or for skin Treg cells in psoriasis? Finally, while most research performed so far has studied mice, and although there are intrinsic difficulties with translating these observations to humans, it will be crucial to determine the phenotype and function of tissue-specific Treg cells in healthy people and in patients with autoimmune diseases to understand the role of tissue-resident Treg cells in the development and maintenance of human autoimmunity.
Intense research in the Treg cell field has improved knowledge of their biology and has provided evidence of the existence of instability and plasticity within the Treg cell compartment. Accumulating data in recent years have shown the presence of ex-Foxp3 cells and helper T cell–like Treg cells in various autoimmune pathologies. Current efforts are focused on determining the molecular mechanisms responsible for the induction of each of these functional states in the periphery and autoimmune organs, as well as the environmental triggers that drive them. The ultimate goal of these studies is to harness the mechanisms that generate the instability and plasticity of Treg cells to modulate Treg cell function not only in autoimmune pathologies but also in other diseases such as cancer. As for tissue-resident Treg cells, although research has begun to elucidate the mechanisms and factors that control their function in tissues, many questions remain to be answered, mostly those related to the mechanisms that tissue Treg cells employ to maintain tissue integrity in autoimmune diseases and the characterization of tissue-resident Treg cells in humans (Box 2). Answers to these questions will undoubtedly improve the design of Treg cell–specific therapeutic options for patients with autoimmune diseases.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank members of the Hafler and Dominguez-Villar laboratories for critical reading of the manuscript.