Translational Mini-Review Series on Th17 Cells: Induction of interleukin-17 production by regulatory T cells

Abstract

Uncommitted (naive) CD4⁺ T helper cells (Thp) can be induced to differentiate to specific lineages according to the local cytokine milieu, towards T helper type 1 (Th1), Th2, Th17 and regulatory T cell (T_reg) phenotypes in a mutually exclusive manner. Each phenotype is characterized by unique signalling pathways and expression of specific transcription factors, notably T-bet for Th1, GATA-3 for Th2, forkhead box P3 (FoxP3) for T_regs and receptor-related orphan receptor (ROR)α and RORγt for Th17 cells. T_regs and Th17 cells have been demonstrated to arise from common precursors in a reciprocal manner based on exposure to transforming growth factor (TGF)-β or TGF-β plus interleukin (IL)-6 and carry out diametrically opposing functions, namely suppression or propagation of inflammation, respectively. However, while epigenetic modifications in Th1 and Th2 differentiated cells prevents their conversion to other phenotypes, Th17 cells generated in vitro using TGF-β and IL-6 are unstable and can convert to other phenotypes, especially Th1, both in vitro and in vivo. T_regs are generated from naive precursors both in the thymus (natural, nT_regs) and in the periphery (induced, iT_regs). The highly suppressive function of T_regs enables them to control many inflammatory diseases in animals and makes them particularly attractive candidates for immunotherapy in humans. The stability of the T_reg phenotype is therefore of paramount importance in this context. Recent descriptions of T_reg biology have suggested that components of pathogens or inflammatory mediators may subvert the suppressive function of T_regs in order to allow propagation of adequate immune responses. Unexpectedly, however, a number of groups have now described conversion of T_regs to the Th17 phenotype induced by appropriate inflammatory stimuli. These observations are particularly relevant in the context of cell therapy but may also explain some of the dysregulation seen in autoimmune diseases. In this paper, we review T_reg to Th17 conversion and propose some potential mechanisms for this phenomenon.

Introduction

Random rearrangement of T cell receptor (TCR) genes in the thymus during ontogeny unsurprisingly generates some T cells with cognate specificity for self-antigens, imparting an inherent potential in the immune system for self-reactivity and autoimmune disease. While this capacity is reduced by the negative selection of autoreactive thymocytes by the AIRE (autoimmune regulator protein)-directed [1] ectopic expression of tissue specific antigens (TSAs) on medullary thymic epithelial cells (mTECs) and dendritic cells (DCs) [2,3] (‘central tolerance’), this is an incomplete process, with thymic émigrés containing a proportion of autoreactive cells. As a result, the mature T cell repertoire retains the capacity for autoimmunity. In order to prevent the development of autoimmune diseases, peripheral mechanisms of retaining tolerance to self-antigens (‘peripheral tolerance’) exclude recirculating T cells from the sites of their cognate antigens, fail to provide adequate co-stimulation to T cells and actively ‘suppress’ the activation of autoreactive cells.

Although a number of immunoregulatory cells have been described in the literature, [4–15], it is thought that CD4⁺ T cells expressing high levels of the interleukin (IL)-2 receptor α chain, CD25 are the most important in the maintenance of peripheral tolerance. These CD4⁺CD25^hi regulatory T cells (T_regs) are derived developmentally from the neonatal thymus [16], but can also be generated directly from naive precursors in the periphery through appropriate activation and cytokine receptor engagement (see below). The former, referred to as natural (n)T_regs, develop in response to self-antigens expressed in the thymus and maintain peripheral self-tolerance while the latter, referred to as induced (i)T_regs, are thought to develop in response to environmental antigens and maintain tolerance to non-self components such as gut flora and ingested material. These two populations have few characteristics that can distinguish them in the peripheral blood (differences between nT_regs and iT_regs are summarized in the review by Horwitz et al.[17]), therefore for the purposes of the present paper they will be considered together. The critical, non-redundant, importance of T_regs in mammalian biology is highlighted by the development of life-threatening autoimmune diseases in both humans and mice who are deficient in this population (as a result of mutations in the FOXP3 and foxp3 genes, respectively; see below) [15,18–20].

T_reg function

While the precise means of T_reg function are not entirely understood it is likely that they possess a functional repertoire of suppressive mechanisms, which would be consistent with diverse descriptions of suppression through direct cell-to-cell contact, production of soluble mediators [21–23] and activity through intermediary cells [24,25]. As a result, T_regs have the in vitro ability to inhibit proliferation and production of cytokines [notably IL-2 and interferon (IFN)-γ] by non-regulatory, traditional T cells (CD4⁺CD25^-) [26–29] as well as responses of CD8⁺ T cells, monocytes and natural killer (NK) cells [26,30,31]. These predicates translate in vivo to a greater number of functions other than the maintenance of tolerance to self-components (i.e. prevention of autoimmune disease) [32] and include control of allergic diseases [33], maintenance of gastrointestinal (GI) tolerance [34] and maternal acceptance of semi-allogeneic fetal antigens [35]. A detailed review on T_reg functions is provided by O'Connor et al. in this series [36].

T_regs in infection and T_reg subversion

T_regs can clearly regulate responses to infectious agents [37,38]; however, appropriate responses to pathogens should balance the need for antigen clearance against excessive tissue injury engendered by an over-exuberant immune system (Fig. 1). Excessive T_reg activity is observed in persistent infections such as murine models of Leishmaniasis, malaria and tuberculosis [39–41] and in human diseases such as upper GI persistence of Helicobacter pylori, human immunodeficiency virus (HIV) and hepatitis C virus (HCV) infections [42–45], suggesting the possibility of a link between pathogen persistence and T_reg-mediated suppression. Subversion of T_reg function for the generation of appropriate immune responses to effect efficient pathogen clearance may therefore be an advantage or, indeed, a necessity.

Fig. 1.

Regulatory T cells during infection. In the context of infection, excessive regulatory T cell (T_reg) activity may leave the host susceptible to overwhelming infection (left side of the swing), while too little T_reg activity may result in excessive tissue injury from unbridled inflammation (right side of the swing).

Indeed, accumulating evidence supports the assertion that interactions between T_regs and an infective/inflammatory environment leads to the subversion of their suppressive function. The salient experiments demonstrate a direct effect of Toll-like receptor (TLR) ligation on T_regs to block their suppression [46,47] and modulation of dendritic cell (DC) activity by lipopolysaccharide (LPS) to induce restricted T_reg activity [48] in a manner that is independent of direct ligation of the TLR on T_regs[49,50]. Indeed, appropriately activated DC can break the ‘anergic’ state of T_regs and promote proliferation in this usually hypoproliferative population [51]. Our own (unpublished) observations and those of others suggest that proinflammatory cytokines, in particular IL-1β, IL-6 and tumour necrosis factor (TNF)-α, released by DC following interaction with pathogens, can subvert the suppressive effects of T_regs. Both IL-1β and IL-6 can block T_reg-mediated suppression of effector cell proliferation [48,52], although IL-6 may require the presence of IL-1 to overcome regulation [49]. There are some data from humans to suggest that TNF-α can inhibit T_reg function [53] with some supporting, but circumstantial, evidence showing a numerical increase in forkhead box P3 (FoxP3)⁺ T cells and restoration of defective regulatory function in patients with rheumatoid arthritis treated with anti-TNF-α therapy [54]. The inevitable question is whether subverted T_regs remain ‘dormant’ T_regs or undergo a stable change of phenotype to an alternative lineage.

T helper type 17 (Th17) cells, development, function and stability

IL-17 is a proinflammatory cytokine with non-redundant functions in the clearance of extracellular pathogens (see also [55] for further detail). This is seen readily in both IL-17R-deficient mice, which demonstrate great susceptibility to lethal bacterial infections [56,57], and in IL-17-deficient humans as part of the hyper-immunoglobulin E (IgE) syndrome (HIES), where recurrent infections are a feature [58,59]. The significant proinflammatory features of IL-17 have been reviewed previously, as has the compelling evidence for the role of IL-17 in inflammatory/autoimmune conditions of mice and the considerable body of evidence suggesting an important role for IL-17 in the aetiopathogenesis of inflammatory and autoimmune diseases in humans [60,61]. The majority of IL-17 is derived from a population of CD4⁺ Th cells discrete from Th1 and Th2, known as Th17.

There has been much interest in the differentiation of Th17 cells from naive precursors and it is now understood that Th17 commitment is linked reciprocally to that of T_regs. While transforming growth factor (TGF)-β differentiates murine naive CD4⁺ T cells to T_regs, the presence of IL-6, in addition to TGF-β, skews the commitment towards Th17 [62–64]. There is greater debate regarding human Th17 differentiation. These pathways of differentiation are discussed in more depth in the review by de Jong and Lord in this series [65]. However, it is important to note that the evidence indicates that Th17 cells are unstable or that the phenotype may be an intermediately differentiated state. In particular, bulk CD4⁺ T cells primed to produce IL-17 by polyclonal activation (anti-CD3 and anti-CD28) in the presence of IL-23 can be redirected away from IL-17 production towards a Th1 phenotype by subsequent TCR activation in the absence of IL-23 or by induction of the Th1 specifying transcription factor, T-bet, suggesting that the Th17 state may be either unstable or a non-terminally differentiated one [66]. This is corroborated by in vivo murine data demonstrating that the adoptive transfer of highly purified islet-specific Th17 cells, devoid of IFN-γ producing populations, causes type 1 diabetes mellitus in recipient mice through the conversion of the Th17 population to a Th1 phenotype (as characterized by cytokine and transcription factor profiles) [67]. This is also observed in experimental autoimmune encephalomyelitis (EAE) models, where fate-mapping of adoptively transferred Th17-skewed cells reveals a significant conversion in vivo to the Th1 lineage [68]. All these findings suggest that there is considerably more plasticity among ‘skewed, lineage-committed’ Th17 cells than thought previously, and contrasts with Th1 and Th2 lineages which are resistant to further differentiation as a result of epigenetic modifications of gene loci associated with the reciprocal lineage [69], ensuring that Th1 and Th2 phenotypes remain stably expressed.

T_reg to Th17 conversion

A number of groups, including our own, have investigated the subversion of T_regs by inflammatory cytokines in both mouse and man and found that, in addition to reduced suppressive activity on target cells, inflammatory cytokines direct T_regs to differentiate into the Th17 lineage and produce IL-17. That this conversion is not the result of outgrowth of a contaminating Th17 precommitted population is indicated by the demonstration of double-positive cells for the T_reg transcription factor FoxP3 and IL-17 (our unpublished observations), which is suggestive of an intermediate, transitional, stage. The conversion of T_regs to Th17 cells has now been reported by a number of groups, in both mouse and human, as shown in Table 1[70–79], albeit with a very interesting difference. While murine T_regs are converted predominantly to Th17 under the influence of IL-6 [70,74,75,78], the same cytokine which skews naive Th differentiation away from T_reg and towards Th17 [62–64], human T_regs are largely resistant to IL-6 and differentiate to the Th17 lineage in an IL-1-dependent manner [73,76,77]. Of note here, one recent murine study has shown that IL-1 signalling is also essential for Th17 lineage differentiation in mice, and that IL-6 induces IL-1R expression on T cells. In this report, IL-1r1^−/− animals had higher percentages of FoxP3⁺ T cells compared to wild-type counterparts, and in an EAE model wild-type, but not IL-1r1^−/−, FoxP3⁺ T cells produced IL-17 in the central nervous system (CNS), suggesting a greater similarity in Th17 differentiation and T_reg to Th17 conversion between humans and mice than thought previously [79]. Murine T_regs can be directed towards the Th17 lineage through receptor–ligand interactions on DC that activate them to produce the appropriate cytokine environment, including (Curdlan-induced) Dectin-1 activation [72] and B7 cross-linking on DC [78]. Conversely, murine T_regs can be protected from IL-6-driven Th17 conversion following exposure to TGF-β and IL-2, as these cytokines in concert reduce surface expression of the IL-6 receptor [75]. As a result, it has been proposed that TGF-β iT_regs are more resistant to Th17 conversion in mice than nT_regs[75]. This is the only publication that demonstrates a potential difference between nT_regs and iT_regs in the propensity to convert to the Th17 lineage and should be accepted only with the caveats that the observed effect cannot be said categorically to be due to inherent differences between nT_regs and iT_regs and not the result of TGF-β and IL-2 signalling per se, and that the concentrations of TGF-β and IL-2 used in iT_reg generation in vitro are orders of magnitude higher than those seen in vivo.

Table 1.

Factors inducing regulatory T cells (T_reg) to T helper type 17 (Th17) conversion in mouse and human.

Species	Reference	Condition mediating Treg to Th17 differentiation†	Additional information
Mouse	[71]	IL-6	Naturally occurring Tregs were more susceptible to Th17 conversion than in vitro IL-2/TGF-β generated Tregs
	[70]	IL-6	Enhanced by IL-1 and IL-23
	[66]	IL-6	LPS activated DC produced sufficient IL-6 to also have this effect
	[75]	IL-1	Il-1r1−/− animals had higher percentages of FoxP3+ T cells compared to wild-type counterparts. In an EAE model, wild-type, but not Il-1r1−/−, FoxP3+ T cells produced IL-17 in the CNS
	[74]	B7-DC cross-linking antibody in an IL-6-dependent manner	Adoptively transferred Th17-converted Tregs induced antigen-specific diabetes mellitus in recipient animals
	[68]	Dectin-1 activated DC	IL-23 produced by the DC was critical in this model
Human	[72]	IL-1β plus IL-2‡	Both naive and memory Tregs converted to Th17, although memory Tregs were more efficiently skewed. IL-6 did not convert Tregs to Th17
	[69]	Mainly IL-1β plus either IL-2 or IL-15‡	Histone deacetylase enzyme inhibition protected Tregs from Th17 conversion
	[73]	Mainly IL-1β plus IL-2‡	IL-17 concentrations were enhanced by IL-6, IL-21 and IL-23 in addition
	[67]	IL1β plus IL-6 and IL-2‡	HLA-DR- Tregs were more potently converted to Th17 than HLA-DR+ Tregs. IL-6 was more potent than IL-1β in DR- Treg skewing to Th17 but the two synergized together

^†In addition to cell activation through the T cell receptor (TCR).

^‡Interleukin (IL)-2/IL-15 do not convert T_regs to Th17 but enhance the efficiency of conversion by other cytokines. CNS: central nervous system; DC: dendritic cell; EAE: experimental autoimmune encephalomyelitis; HLA-DR: human leucocyte antigen D-related; FoxP3: forkhead box P3; LPS: lipopolysaccharide; TGF: transforming growth factor.

Some of these reports have demonstrated that Th17 cells derived from T_regs share common features with Th17 cells generated from naive precursors, including expression of the chemokine receptor CCR6 [73,76,80]. CCR6 is a chemokine receptor expressed on the surface of Th17 cells, under the control of the Th17 transcription factor receptor-related orphan receptors (ROR)α and RORγt, which directs their migration into sites of inflammation [81]. Interestingly, although ‘converted’ T_regs also express CCR6 (as well as other chemokine receptors in common with Th17 cells [82]), in contrast to Th17 cells they do not express CCL20 [macrophage inflammatory protein (MIP)-3α][81], which is the only known ligand for CCR6 [83]. Th17 cells therefore recruit other Th17 cells and T_regs into sites of inflammation through secretion of CCL20 [81]. Indeed, chronically inflamed tissues in human diseases are characterized by the presence of infiltrating Th17 cells expressing CCR6 [84], and mice are protected from developing EAE if the CCR6–CCL20 interaction is neutralized [81]. Expression of CCR6 by Th17 cells derived from conversion of T_regs[73,76] is therefore a significant finding, and suggests that these Th17 cells have the capacity to migrate to sites of inflammation containing CCL20. However, these observations should be tempered by murine data showing that IL-17 is produced from both CCR6^- and CCR6⁺ T_regs at sites of disease (in this case, the CNS) [81].

In humans, the biological relevance of T_reg to Th17 conversion seen in vitro is unknown; however, human memory phenotype (CD45RO⁺) FoxP3⁺ T_regs isolated ex vivo have been shown very recently to secrete IL-17 and to express the Th17 transcription factor RORγt constitutively [85], suggesting that IL-17 production from T_regs also occurs in vivo. The reversal of the regulatory function of T_regs, and skewing of phenotype towards production of IL-17, a cytokine known to be important in human autoimmune diseases [60], may provide a link between the loss of regulation and high levels of IL-17 seen in some of these disorders. In addition, mice in which the IL-1 receptor antagonist gene has been silenced develop spontaneous autoimmune T cell-mediated arthritis, an IL-17-mediated condition [86,87], due to excessive IL-1 signalling [88]. These mice do not exhibit arthritis when kept germ-free, but rapidly develop pathological features when exposed to a single species of indigenous gut flora (Lactobacillus bifidus) or to signalling through TLRs [89]. The epidemiological association between infections and the development of human autoimmune diseases could indicate a similar mechanism through altered T_reg function and the promotion of IL-17, potentially also mediated through IL-1 or associated TLR signalling pathways.

Demonstrations of the capacity of T_regs to convert to the Th17 lineage also suggests that infiltrating CD4⁺ cells bearing the phenotype of T_regs (CD4⁺CD25⁺FoxP3⁺) at sites of infection [42] where IL-1β or IL-6 are highly expressed may not necessarily effect a suppressive function, but might instead participate in clearance of the inciting pathogen through conversion to the Th17 lineage. The stability of the Th17 phenotype in this model is an important consideration: given that Th17 cells generated from naive precursors are not stable either in vitro or in vivo[66–68], prolonged T_reg-derived Th17 persistence at sites of inflammation may engender excessive tissue injury. Although this has not been addressed sufficiently in the literature, some available data suggest that restoration of suppressive function may be possible upon exposure to IL-2 [71].

The T_reg–Th17 axis and mechanisms of conversion

In the context of concerted efforts to use expanded populations of T_regs for adoptive therapy in human inflammatory diseases, descriptions of T_reg to Th17 conversion are important observations, as transition of adoptively transferred cells from an anti- to a proinflammatory lineage may exacerbate, rather than ameliorate, disease. Therefore, an understanding of the mechanisms underlying this conversion and methods to stabilize the T_reg phenotype have become important aspects of T_reg biology. Although the mechanisms have not been elucidated fully, there are some suggestions in the available literature, which will be reviewed here. These include upstream signalling and transcription factor interactions.

Transcription factor interactions – FoxP3 and RORγt

Several members of the retinoic acid receptor (RAR) orphan receptor (ROR) family have been described as transcription factors expressed specifically in Th17 cells. These include RORα and RORγt [90–92], which are encoded by the genes RORA and RORC. RORγt is induced by TGF-β and IL-6 in naive Thp and leads to transcription of IL-17 [90]. As expected, overexpression of RORγt promotes Th17 differentiation. However, while RORγt-deficient mice have reduced numbers of Th17 cells, the population is not depleted [90]. This is because RORα is also expressed highly in TGF-β/IL-6-induced Th17 cells [91]. This related transcription factor synergizes with RORγt to induce Th17 differentiation, and elimination of both RORα and RORγt (double-deficient animals) at the same time is required to deplete Th17 differentiation effectively and protect against Th17-driven autoimmune diseases [91].

The Scurfy mouse (sf), an X-linked mutant strain, described in 1949 (loc. cit. [93], exhibits a series of autoimmune features including skin scaliness, diarrhoea and death (between 2 and 4 weeks after birth) in association with CD4⁺ T cell hyperproliferation, multi-organ CD4⁺ cell infiltration [94] and over-production of several inflammatory cytokines [95]. This fatal autoimmune lymphoproliferative syndrome maps to a gene locus on the X chromosome called foxp3, which has been described as a member of the forkhead/winged-helix family of transcription factors [96]. The foxp3 gene is highly conserved between species and a mutation in the human gene, FOXP3, has been identified as the causative factor responsible for the human equivalent of Scurfy, the immunodysregulation, polyendocrinopathy and enteropathy, X-linked syndrome (IPEX), also known as X-linked autoimmunity and allergic dysregulation syndrome (XLAAD) [19,97,98]. Both the mouse and human disease lack discrete circulating T_regs, which suggests that foxp3 and FOXP3 are essential for normal T_reg development in the two species, respectively. This position is strengthened by the failure of foxp3 knock-out mice to develop circulating T_regs; these animals develop a Scurfy-like syndrome from which they can be rescued by the adoptive transfer of T_regs from a foxp3 replete animal [99]. Furthermore, ectopic or over-expression of foxp3 in CD4⁺CD25^- mouse cells results in development of a T_reg phenotype [97,99,100]. In mice, FoxP3 expression is a good phenotypic marker of T_regs[101,102]; in humans, however, FoxP3 does not allow the unambiguous identification of T_regs[103], as FoxP3 is induced during TCR stimulation in conventional CD4⁺ T cells [104–106] (in much the same manner as CD25) and there is some debate as to whether the induced CD4⁺CD25⁺FoxP3⁺ population is suppressive or anergic [104,105]. The biology of FoxP3 is still understood incompletely, but there are suggestions that FoxP3 may function as a transcriptional inhibitor through associations with nuclear factor of activated T cells (NFAT) and nuclear factor-kappa B (NF-κB) [107,108].

Recent evidence suggests that similar mechanisms may regulate the commitment of Thp between T_reg and Th17. In human cells, FoxP3 exists in two separate but equally expressed isoforms: one (FoxP3), which is encoded by a full length mRNA and the other a truncated form lacking exon 2 (FoxP3Δ2), which is coded by a splice variant mRNA [104,109]. T_regs, perhaps unexpectedly, also express Th17-specifying transcription factors, notably RORα[110] and RORγt [111]. However, co-immunoprecipitation experiments have shown that FoxP3 binds to RORα and RORγt and inhibits their biological activity in a dose-dependent fashion [110,111]. This interaction is mediated through a (LxxLL) motif in the FoxP3 second exon; as expected, the FoxP3Δ2 isoform is unable to bind RORα or RORγt [110,111].

A similar interaction has subsequently been described, by the same group and others, in murine cells. Specifically, both FoxP3 and RORγt are co-expressed in naive CD4⁺ T cells exposed to TGF-β, where FoxP3 inhibits RORγt directly through a physical interaction, repressing the Th17 programme [111]. In these experiments exposure of Thp to TGF-β leads to rapid induction of RORγt [92], but the binding of RORγt to the IL-17 promoter is suppressed by interaction with FoxP3 [112]. Upon addition of exogenous IL-6 or IL-21, the inhibitory effect of FoxP3 on IL-17 induction is circumvented [111] and FoxP3 levels are reduced [112]. The interaction between FoxP3 and RORγt in murine cells is also dependent upon the second exon of FoxP3 [111,112]. These observations have also been confirmed by another, independent group [74].

These interactions can, in part, explain the conversion of T_regs to Th17, at least in mice. While TGF-β induces both FoxP3 and RORγt expression, IL-6 does not alter expression of RORγt but inhibits FoxP3. As a result, exposure of T_regs to IL-6 down-modulates FoxP3 preferentially and reduces the physical inhibition of RORγt, permitting binding to the IL-17 gene promoter. In addition, very recent murine data suggest that IL-1 regulates expression of RORγt [79].

Signal transducer and activator of transcription (STAT) 3 and 5

The Janus kinase/signal transducer and activator of transcription (JAK/STAT) pathway is a receptor-coupled signal transduction mechanism linking cytokine–receptor interactions to gene expression. There are seven STAT (STAT1-4, 5A, 5B and 6) and four JAK [JAK1-3 and TYK2 (tyrosine kinase 2)] proteins in humans (reviewed in [113]). Specific JAKs are associated with the cytoplasmic tails of multimeric cytokine receptors, and are activated upon ligand-induced receptor oligomerization [113,114]. Activated JAKs phosphorylate specific tyrosine residues on cytoplasmic tails of their associated cytokine receptors, creating docking sites for the SH2 (Src-homology-2) domain of STAT proteins, and then activate the docked STATs through tyrosine phosphorylation. Activated STATs dimerize and translocate to the nucleus to regulate gene transcription by binding to specific promoter regions. Suppressor of cytokine signalling (SOCS) proteins induced by STAT signal transduction feed back to regulate STAT signalling negatively. The JAK/STAT/SOCS pathway is depicted in Fig. 2.

Fig. 2.

The Janus kinase/signal transducer and activator of transcription/suppressor of cytokine signalling (JAK/STAT/SOCS) axis. Cytokine engagement of their cognate receptors (a) leads to receptor dimerization, which activates JAK kinases. Activated JAKs phosphorylate the cytoplasmic tails of the receptor (b), creating docking sites for Src-homology-2 (SH2)-domain-containing proteins such as STATs (c). Docked STATs are activated by JAKs, dissociate from the receptor and dimerize (d). Dimerized STATs migrate to the nucleus and associate with specific STAT binding sites (e) and (positively or negatively) regulate expression of STAT-responsive genes, including suppressor of cytokine signalling (SOCS) proteins. SOCS proteins also contain an SH2-domain and feed back to inhibit this pathway through competition with STAT proteins for SH2-domain-docking sites (f), SH2-recruitment to the receptor cytoplasmic domain followed by JAK inhibition (g) or direct inhibition of JAK activity (h). In addition, SOCS proteins contain a SOCS box motif which contains an E3 ubiquitin–ligase complex, through which the SOCS protein can ubiquitinate JAK proteins and target them for proteosomal degradation (i).

The IL-2-induced JAK3/STAT5 pathway is an indispensible signal transduction mechanism for the direct induction of FOXP3 in T_regs[115], as animals deficient in IL-2, IL-2Rα (CD25), IL-2Rβ (CD122) or STAT5 have depleted numbers of T_regs, fail to express FoxP3 and develop autoimmune diseases [115–118]. In these models, animals can be rescued from autoimmunity with restored T_reg development if IL-2 signalling is re-established, for example, by reconstitution with bone marrow from an IL-2Rβ mutant that activates STAT5 exclusively or by ectopic activation of foxp3 in CD4⁺ T cells [116]. Similarly, STAT5 deficiency in humans results in loss of T_regs and immune dysregulation, while overexpression of STAT5 in CD4⁺CD25^- cells leads to elevated levels of FoxP3 [119,120]. These observations can be explained by the direct binding of STAT5 to the foxp3 promoter [115,116], instigating an IL-2-directed STAT5-dependent positive regulation of foxp3. Indeed, T_regs show an obligate requirement, both in vivo and in vitro, for IL-2 and the structurally related IL-2 family members, IL-15 and IL-7, for maintenance of FoxP3 expression and suppressive function [119,121–123].

Th17 development from naive precursors is dependent upon signal transduction through STAT3. In mice, RORC is a STAT3 target gene and Th17 differentiation is induced by STAT3 signalling cytokines, notably IL-6, IL-21 and IL-23, and can be abrogated effectively by a deficiency in STAT3 [124]. In humans, STAT3 deficiency from dominant negative mutations in the STAT3 gene occurs in the hyperIgE (HIES or Job) syndrome (OMIM 147060), which is characterized by morphological abnormalities, recurrent infections (particularly with Staphylococcus aureus and Candida sp.) and a deficiency of Th17 cells [59,125–127]. Patients with HIES not only have reduced Th17 numbers, but their naive Th cells are resistant to Th17 differentiation under appropriate stimulatory conditions, with concomitant impairment of RORγt expression relative to healthy controls [59,126].

There are reasons to suspect that the STAT3/STAT5 signalling pathways are important in the conversion of T_regs to Th17. First, there is evidence to suggest that STAT5 and STAT3 cross-regulate the conversion of naive T cells to the T_reg and Th17 lineages. Specifically, in mice, IL-2-induced STAT5 inhibits Th17 differentiation from IL-6 and TGF-β stimulated naive Th cells [128]. It should be noted that, although this inhibition is STAT5-dependent, it is unclear whether the mechanism is a direct STAT5 inhibition of IL-17 associated genes or an indirect effect of STAT5-induced FoxP3-directed inhibition of Th17 transcription factors (as described above). Similarly, STAT3-activating cytokines, such as IL-6 and IL-27, regulate FoxP3 negatively in a STAT3-dependent manner [62,115,129]. This enables IL-6-activated STAT3 to inhibit both FoxP3 expression and enable IL-17 production in naive T cells stimulated with TGF-β[74]. Not surprisingly, therefore, humans with HIES (who have mutations in STAT3) have a higher than normal percentage of cells bearing the phenotype of T_regs[59], while mice deficient in the IL-2 signalling cascade (notably IL-2 or STAT5) have a reduction in T_regs and an excess of Th17 cells in association with autoimmune disease.

Given that there appears to be functional antagonism between the STAT3 and STAT5 pathways during the polarization of naive T cells towards T_reg or Th17, it can be hypothesized that the plasticity of differentiated T_regs may be regulated by the dominant STAT signal induced by local cytokines.

Interferon regulatory factor-4 (Irf-4)

There are reasons to suspect the involvement of other signalling pathways in the conversion of T_regs to Th17. These include the Irf-4 transcription factor. Irf-4 is a lymphocyte-restricted member of the Irf family of transcription factors [130] that is critical for the function of mature B and T cells [131]. In T cells, Irf-4 binds to the regulatory regions of cytokine genes, notably IL-2, IL-4, IL-10 and IL-13, and enhances their expression [132]. Involvement of Irf-4 in Th17 polarization in mice is suggested by a failure of Th17 skewing in Thp from mice that are Irf-4-deficient [133]. T cells from these mice do not respond to Th17 polarizing conditions (TGF-β plus IL-6) in the same manner as their wild-type counterparts, maintaining low levels of RORγt, and fail to induce experimental allergic encephalomyelitis (EAE) in vivo[133]. Of particular note, while exposure of Thp from Irf-4^−/− animals to TGF-β up-regulates FoxP3 in a normal manner, these cells are subsequently resistant to down-regulation of FoxP3 by IL-6, resulting in failure of Th17 differentiation [133]. Irf-4 is therefore a critical factor in the reciprocal differentiation of T_regs and Th17 cells from common precursors. This assertion is reinforced by the promotion, by Irf-4, of IL-21 [134,135], a stabilizing factor for the Th17 phenotype, and the development of IL-17 driven diseases (such as inflammatory arthropathies) in Irf-4-overexpressing animals [134]. As a result, there is the possibility that Irf-4 may also be an important transcription factor for the conversion of T_reg-committed cells to a Th17 phenotype under the influence of inflammatory cytokines. This notion is enhanced by the recent finding that IL-1 induces the expression of Irf-4 during early stages of murine Th17 polarization [79].

Conclusions

The potent suppressive nature of T_regs and their ability to ameliorate a wide array of inflammatory conditions in animals has led to considerable efforts directed towards their utilization as therapeutic tools in humans. However, the propensity of T_regs to convert to a Th17 phenotype in the presence of inflammatory mediators is a real concern, as they would have the potential to exacerbate conditions they are intended to treat. To date, only the rudimentary mechanisms of this phenomenon have been identified, but a greater understanding of the mechanisms underlying T_reg to Th17 conversion may identify targets for modification and pharmacological intervention that might stabilize T_regs intended for clinical use and inhibit their proinflammatory potential in vivo.

Disclosure

There are no conflicts of interest: the authors have been supported by grants from the Medical Research Council and the British Heart Foundation.