The role of transforming growth factor β in T helper 17 differentiation

Summary

T helper 17 (Th17) cells play critical roles in inflammatory and autoimmune diseases. The lineage‐specific transcription factor ROR γt is the key regulator for Th17 cell fate commitment. A substantial number of studies have established the importance of transforming growth factor β (TGF‐β) ‐dependent pathways in inducing ROR γt expression and Th17 differentiation. TGF‐β superfamily members TGF‐β ₁, TGF‐β ₃ or activin A, in concert with interleukin‐6 or interleukin‐21, differentiate naive T cells into Th17 cells. Alternatively, Th17 differentiation can occur through TGF‐β‐independent pathways. However, the mechanism of how TGF‐β‐dependent and TGF‐β‐independent pathways control Th17 differentiation remains controversial. This review focuses on the perplexing role of TGF‐β in Th17 differentiation, depicts the requirement of TGF‐β for Th17 development, and underscores the multiple mechanisms underlying TGF‐β‐promoted Th17 generation, pathogenicity and plasticity. With new insights and comprehension from recent findings, this review specifically tackles the involvement of the canonical TGF‐β signalling components, SMAD2, SMAD3 and SMAD4, summarizes diverse SMAD‐independent mechanisms, and highlights the importance of TGF‐β signalling in balancing the reciprocal conversion of Th17 and regulatory T cells. Finally, this review includes discussions and perspectives and raises important mechanistic questions about the role of TGF‐β in Th17 generation and function.

Cytokine milieu for T helper 17 differentiation

T helper 17 (Th17) cells were identified in 2005 as a new subset of CD4⁺ T helper cells producing interleukin‐17 (IL‐17).1, 2 In response to environmental cues, naive T cells differentiate into Th17 effector cells (Fig. 1). The Th17 differentiation requires both inflammatory cytokine [e.g. IL‐6, or IL‐21] and transforming growth factor β (TGF‐β) to efficiently induce the lineage‐specific transcription factor retinoic‐acid‐receptor‐related orphan nuclear receptor γt (RORγt) as well as IL‐17 production.3 T helper 17 cells are critically involved in host defence, inflammation and autoimmunity.4 The Th17 responses contribute to the pathogenesis of diverse human diseases, including psoriasis, asthma, rheumatoid arthritis, inflammatory bowel disease and multiple sclerosis.5, 6 A better understanding of Th17 differentiation would therefore lay the basis for therapeutic interventions.7

Figure 1.

T helper 17 (Th17) differentiation. Interleukin‐6 (IL‐6) or IL‐21 activates signal transducer and activator of transcription 3 (STAT3) to potentiate Rorc expression, which encodes the retinoic‐acid‐receptor‐related orphan nuclear receptor γt (ROR γt) transcription factor. IL‐6 activates STAT3 through classic, trans or cluster signalling pathways. Additional transforming growth factor‐β (TGF‐β) or activin A degrades SKI to reverse the SKI/SMAD4 complex imposed repression on Rorc transcription. Unleashed ROR γt binds to the Il17a, Il17f loci and drives T cells to produces Th17 cytokines, e.g. IL‐17A, IL‐17F, IL‐21 and TGF‐β. Dendritic cells produce multiple cytokines to promote Th17 differentiation, such as IL‐6, IL1‐β, IL‐23 and activin A. Dendritic cells also express integrin α _v/β ₈ to process the latent form of TGF‐β to the active form.

Inflammatory cytokines

Interleukin‐6 is characterized as an inflammatory cytokine, together with TGF‐β featuring Th17 induction activity.8, 9 Engagement of IL‐6 with its receptor leads to activation of signal transducer and activator of transcription 3 (STAT3), which further potentiates RORγt transcription.4 Although various haematopoietic and non‐haematopoietic cells can produce IL‐6, a recent study showed that Sirpα ⁺ dendritic cells are essential for pathogenic Th17 cells by trans‐presenting IL‐6 to T cells during the process of cognate interaction.10

As an alternative cytokine, IL‐21 also can stimulate STAT3 activation. Interleukin‐21 in combination with TGF‐β leads to RORγt expression and Th17 differentiation independent of IL‐6.11, 12, 13, 14 In addition, IL‐6 induces IL‐21 during Th17 differentiation to further enhance the Th17 programme.

Other cytokines also play important roles in the development of Th17 cells. Interleukin‐1 signalling is indispensable for Th17‐mediated autoimmunity in experimental autoimmune encephalomyelitis (EAE),15 and is necessary for early Th17 differentiation in vivo.16 It is also critical for the expansion, but not the generation, of autoreactive granulocyte–macrophage colony‐stimulating factor (GM‐CSF) ‐positive Th17 cells.17 Additionally, IL‐1 enhances Th17 differentiation by inhibiting TGF‐β‐induced Foxp3 expression.18

Interleukin‐23 is capable of activating the STAT3 pathway and promoting Th17 lineage maintenance but not commitment.19 Importantly, T cells exposed to IL‐23 have promoted expression of GM‐CSF and IL‐23 receptor (IL‐23R) to become more pathogenic.20, 21

It is still unclear if there are other unrecognized cytokines that can act functionally like IL‐6 or IL‐21 to initiate Th17 differentiation with TGF‐β. It has also been elusive whether STAT3 activation is the only major function for IL‐6 and IL‐21 to induce Th17, or whether these cytokines have other important targets to synergize with TGF‐β signalling and induce Th17 differentiation.

Requirement of TGF‐β for Th17 differentiation

Transforming growth factor‐β is a regulatory cytokine, exerting pleiotropic functions in T‐cell development, homeostasis, tolerance and differentiation.22, 23 The TGF‐β is produced as an inactive form in complex with latency‐associated peptide and latent TGF‐β‐binding protein, and is further activated by disengaging from this large latent complex. Multiple mechanisms are involved during this process, including plasmin, matrix metalloproteases, thrombospondin‐1 and integrins.22, 24, 25 Integrin binds to latency‐associated peptide and induces the release of mature TGF‐β, which provides an additional layer of mechanistic regulation. For example, although dendritic cells do not produce TGF‐β, these cells are important for α _v/β ₈ integrin‐mediated TGF‐β activation to exert biological functions such as inducing Th17 differentiation.26 The α _v/β ₈ integrin transcription in antigen‐presenting cells depends on interferon regulatory factor 8 (IRF8). Mice with IRF8 deficiency in antigen‐presenting cells, but not T cells, fail to generate Th17 cells and were resistant to EAE.27

Transforming growth factor‐β established its requirement in murine models shortly after Th17 cells were identified.8, 9, 28 Mice that were TGF‐β ₁ deficient exhibited profoundly diminished Th17 cells in all tissue sites.27 Mice with TGF‐β ₁ deficiency in T cells were also defective in Th17 differentiation as well as the onset of EAE.29 Mice with TGF‐β signalling blockade by a dominant negative form of TGF‐β receptor II (CD4dnTGFβRII) or by a TGF‐β‐specific antibody failed to differentiate naive T cells to Th17 cells and were protected from EAE,30 whereas TGF‐β transgenic mice resulted in enhanced Th17 differentiation and more severe EAE.8 These data strongly suggest that TGF‐β is indispensable for Th17 differentiation.

Initially, human cells were considered not to require TGF‐β but only IL‐6 and IL‐1β, or IL‐1β and IL‐23 for Th17 differentiation.31, 32 Naive CD4⁺ T cells (defined by CD4⁺ CD45RA⁺ CD45RO⁻ 32 or CD4⁺ CD45RA⁺ CCR7⁺ CD25⁻ 31) used in these studies were sorted from peripheral blood, and so raised the concern of naiveté.33 In addition, there was possible TGF‐β contamination from the serum of culture medium. Therefore in later studies, naive cord blood CD4⁺ T cells (defined by CD3⁺ CD4⁺ CD25⁻ HLA‐DR⁻ CD45RA⁺,34 CD3⁺ CD4⁺ CD45RA⁺ CD45RO⁻,35 or CD4⁺ CD25⁻ CD62L⁺ CD45RA^hi 36) and serum‐free medium were used. With minimized TGF‐β source contamination from serum or platelets, and cord‐blood‐originated naive CD4⁺ T cells, these studies clarified that TGF‐β is indeed required for human cell Th17 differentiation.34, 35, 36 CD161⁺ CD4⁺ T‐cell precursors in umbilical cord blood and thymus were reported to constitutively express RORγt, IL‐23R and CCR6.37, 38 These cells produced IL‐17 in response to IL‐1β and IL‐23 without the need for TGF‐β.37, 39 Because most of the CD161⁺ cells express CD45RO,37, 40 these precursor cells might have a memory T‐cell phenotype. Both IL‐1β and IL‐23 could contribute to cell activation or expansion rather than to de novo Th17 differentiation. Moreover, TGF‐β is potent for skewing these CD161⁺ cells from Th1 towards Th17 after IL‐1β and IL‐23 stimulation.39 Collectively, these data suggest that TGF‐β plays an essential role in human Th17 differentiation.

TGF‐β source, TGF‐β superfamily and Th17 cell pathogenicity

There are three isoforms of TGF‐β: TGF‐β ₁, TGF‐β ₂ and TGF‐β ₃.41 Many types of cell produce TGF‐β ₁, and T cells are the important source of TGF‐β ₁ to promote Th17 differentiation and to develop EAE,29 although the role of TGF‐β ₁ in EAE induction is controversial.42, 43, 44 Despite the fact that Th1, Th2 and Th17 cells can express TGF‐β ₁ after polarization, Th17 effector cells have the most increased production of TGF‐β ₁. In vivo Th17 differentiation requires the autocrine TGF‐β ₁.45 However, this finding is not consistent with the observation that autocrine TGF‐β produced by in vitro differentiated Th17 cells under IL‐6 + IL‐1β + IL‐23 conditions is not essential, as TGF‐β antibody blockade does not significantly reduce Th17 differentiation.46 Therefore, further debate on the role of autocrine TGF‐β produced by Th17 cells continues.

Foxp3⁺ regulatory T (Treg) cells could serve as another source of TGF‐β ₁ production, which may induce naive T cells to differentiate into Th17 cells under in vitro co‐culture conditions.9 However, mice with TGF‐β ₁ deletion specifically in Foxp3⁺ Treg cells show in a similar level of Th17 differentiation in an in vivo model of EAE, arguing that Foxp3⁺ Treg‐cell‐derived TGF‐β ₁ is not essential for Th17 differentiation.45

The TGF‐β ₁ produced by Treg cells or activated CD4⁺ T cells does not substantially affect the T‐cell activation in spleen and lymph nodes, but it is required for inhibiting Th1 differentiation in the gut.45 A subset of Treg cells, CD103⁺ Treg cells, possess very high suppressive activity and exist at sites of inflammation and mucosal tissues including gut,47, 48, 49 suggesting that the different microenvironment and local cytokine milieu may determine how TGF‐β influences Th17 differentiation and propagates disease progression.50

The sources of TGF‐β include stromal cells, immune cells and cancer cells, which provide a basis for versatile regulation in local immune responses.23 For example, gliadin‐specific Th17 cells from individuals with coeliac disease simultaneously express TGF‐β. This autocrine TGF‐β plays a positive regulatory role in IL‐17 production in intestinal mucosa.51 TGF‐β prevails in the intestine, and intestinal epithelial cells and dendritic cells are important sources of bioactive TGF‐β. In addition, CD103⁺ dendritic cells express a high level of integrin α _v/β ₈, and are specialized to activate latent TGF‐β.52, 53, 54 In healthy humans, about 4 ng/ml TGF‐β ₁ circulates in plasma, therefore it may act in the manner of long‐range endocrine interactions between distant tissues. However, TGF‐β ₂ and TGF‐β ₃ are probably derived from local synthesis, as they are generally undetectable in plasma, and appear to act as local autocrine or paracrine regulators.55

Transforming growth factor‐β not only promotes Th17 differentiation but also determines the pathogenicity of Th17 cells. Researchers observed that TGF‐β ₁‐induced Th17 cells are less pathogenic in EAE,20 whereas TGF‐β ₃ leads to Th17 cells that are more pathogenic.21 TRIM28‐deficient T cells cause aberrant overexpression of TGF‐β ₃ and contribute to the development and accumulation of pathogenic Th17 cells.56

Other than TGF‐β family cytokines, a TGF‐β superfamily member, activin A, was also reported to be capable of inducing Th17 differentiation in combination with IL‐6.57, 58

Because there are more than 33 human TGF‐β superfamily members, including TGF‐β, bone morphogenetic proteins and activins, and they share similar structures,23 it would be important to identify if there are more cytokines that could be functionally similar to TGF‐β ₁, TGF‐β ₃ and activin A during Th17 differentiation.

The diverse functions of the TGF‐β superfamily rely on their specific receptor signalling, which goes through different heteromeric type I and type II receptor complexes. Receptors TGFβRI (ALK5) and TGFβRII are mostly designated to TGF‐β, whereas receptors ACTRIB (ALK4) and ACTRIIs receive signalling from activins.23, 59 The differences in receptors could explain why activin A alone fails to induce Foxp3 whereas TGF‐β can induce Foxp3.60 However, TGF‐β ₁ and TGF‐β ₃ share identical receptors, but lead to distinct pathogenic profiles in Th17 cells.21 This suggests that there might be other undefined receptors or mechanisms in existence that require further investigation.

T helper 17 cells are recognized as important pathogenic effector cells in most EAE models.61 EAE is attenuated when mice are injected with anti‐IL‐17A antibody2, 62, 63 and IL‐17RA‐deficient mice are resistant to EAE.64 Interleukin‐17A deficiency results in reduced clinical EAE, although it does not fundamentally impede the incidence of disease induction.65, 66 Cytokines critical for Th17 differentiation, such as IL‐6,67, 68 IL‐21,11, 12 IL‐115 and IL‐23,69 are also required for EAE. Increasing evidence shows that several Th17‐expressed genes are especially vital for the pathogenicity of Th17 cells. For example, IL‐23R expression is shown to be critical for the generation of encephalitogenic Th17 cells,70, 71 and GM‐CSF‐deficient mice are resistant to EAE.72, 73, 74, 75 It appears that there are certain molecular networks associated with the pathogenicity of Th17 cells. Different TGF‐β signalling pathways produce different pathogenic programmes.21, 46, 76 As the Th17 cells are highly heterogeneous, the diversity of TGF‐β superfamily ligands and receptors provides a tool for investigating the essential mechanisms of Th17 pathogenicity.

TGF‐β‐independent Th17 cell differentiation

There are also studies showing that TGF‐β is dispensable for murine Th17 differentiation. In the presence of anti‐TGF‐β antibodies, STAT6 and T‐bet double‐deficient T cells can still differentiate into Th17 cells with IL‐6 alone.77, 78 These observations raise the debate on the requirement of TGF‐β in Th17 differentiation. However, TGF‐β antibody blockade, but not TGF‐β receptor signalling deficiency, could not rule out the possibility that there is still TGF‐β or that TGF‐β superfamily receptor signalling exists in these settings. Later, Ghoreschi et al.46 reported that although TGF‐βRI‐deficient mice or CD4dnTGFβRII mice were not able to induce EAE, there were similar amounts of Th17 cells in intestinal lamina propria compared with wild‐type mice. They also found that Th17 cells could be generated in vitro without TGF‐β using a combination of cytokines (IL‐6, IL‐1β and IL‐23) and that these Th17 cells were more pathogenic. These data suggest an alternative TGF‐β‐independent Th17 differentiation pattern. It also provides insight and opportunity to investigate the mechanism of TGF‐β‐dependent and TGF‐β‐independent Th17 differentiation and Th17 cell pathogenicity. Compared with TGF‐β, this cytokine combination had a relatively low ability to induce in vitro Th17 differentiation,16, 46 but enough to raise the argument that TGF‐β may not be necessary under certain environmental contexts.

To date, the mechanisms of how these Th17 cells are induced by cytokine combinations without requiring TGF‐β signalling, and how the downstream receptor signalling of IL‐6, IL‐1β and IL‐23 synergized, are still perplexing. Notably, TGFβRI ablation, but not TGF‐β antibody blockade, dramatically reduced the proportion of Th17 cells from 30% to 13%.46 This suggests that TGFβRI signalling at least partially contributes to Th17 differentiation under this cytokine regimen. It is very likely that there is TGFβRI ligand produced, which enhances the Th17 programme in an autocrine pattern under these in vitro culture conditions. Further studies should especially focus on TGF‐β superfamily cytokines that can be secreted by these Th17 cells and also can signal through TGFβRI. For those Th17 cells deficient in TGFβRI, it is also possible that they are primed by the combination of cytokines, and then secondarily induced by autocrine cytokines through other TGF‐β superfamily receptors except TGFβRI. Further studies are warranted to uncover the requirements and central mechanisms of TGF‐β signalling in Th17 differentiation.

Mechanisms underlying TGF‐β promoted Th17 differentiation

Transforming growth factor‐β induces naive T cells to express the regulatory transcription factor Foxp3 upon T‐cell receptor activation.79 Foxp3 can directly interact with RORγt and antagonize the Th17 master regulator RORγt.80, 81, 82 Although IL‐6 can activate STAT3 and inhibit Foxp3 expression,8 it cannot sufficiently induce the RORγt transcriptional programme. TGF‐β is required for the full differentiation capacity of Th17 cells. Confounded by these observations, extensive studies have been performed to explain the paradoxical role of TGF‐β in Th17 differentiation in the last decade (Fig. 2).

Figure 2.

Transforming growth factor‐β (TGF‐β) promotes T helper 17 (Th17) differentiation through multiple mechanisms. Signal transducer and activator of transcription 3 (STAT3), induced by interleukin‐6 (IL‐6) or IL‐21, potentiates the Th17‐related transcriptional programme by binding to Rorc, Il17a, and Il21 loci. TGF‐β receptor signalling phosphorylates SMAD2 and SMAD3. The SMAD2/3 complex interacts with SMAD4 and translocates into the nucleus. Meanwhile, SMAD2 and SMAD3 act as a co‐activator and co‐repressor of STAT3, and oppositely modify STAT3‐induced transcription. TGF‐β signalling triggers SKI degradation to unleash SMAD4 and SKI complex repressed ROR γt expression. TGF‐β suppresses SOCS3 to prolong STAT3 activation. TGF‐β suppresses GFI‐1 to promote Il17a transcription. TGF‐β suppresses Eomes through the JNK‐c–JUN pathway to enhance Rorc and Il17a transcription. TGF‐β drives RhoA‐ROCK2 to phosphorylate IRF4 and to up‐regulate the expression of ROR γt, IL‐17 and IL‐21. TGF‐β induces TAZ to co‐activate ROR γt and to reduce Foxp3 through proteasomal degradation.

Essential transcription factors in Th17 differentiation

T helper 17 differentiation is controlled by several key transcription factors.83 The Th17 lineage‐specific regulator RORγt (encoded by Rorc gene) directs IL‐17 production by binding to the IL‐17 gene locus.3, 81 RORγt‐deficient T cells fail to become Th17 cells; reducing the transcriptional activity of RORγt by antagonists such as digoxin effectively inhibits Th17 differentiation.84 Post‐translational regulation of RORγt by small‐molecule RORγt antagonists,85 or by ubiquitin ligase UBR5 in response to TGF‐β signalling can limit the abundance of RORγt and so prevent rampant IL‐17 production.86

STAT3 regulates the transcription of its multiple target genes, including Rorc, Il17a and Il17f. STAT3 is required for Th17 differentiation and its deficiency results in abrogated Th17 generation, whereas overexpression of STAT3 significantly boosts Th17 differentiation.14, 87, 88, 89

The cooperative binding of basic leucine zipper transcription factor, ATF‐like (BATF) and IRF4 is critical for initial chromatin accessibility and combinatorial regulation with STAT3 during Th17 differentiation.83, 90 Either IRF4 or BATF ablation results in failure of RORγt induction and defective Th17 differentiation.91, 92, 93

SMAD‐dependent pathways

The canonical TGF‐β pathway signals through the TGF‐β receptor and phosphorylates receptor‐regulated SMADs (R‐SMAD), SMAD2 and SMAD3. The heterodimeric complex then interacts with the common mediator SMAD (co‐SMAD), SMAD4, to generate a heterotrimeric complex. This new complex translocates into the nucleus for downstream gene modulation. Complementarily, TGF‐β can activate non‐canonical pathways such as phosphoinositide 3 kinase/AKT/ mechanistic target of rapamycin (mTOR), mitogen‐activated protein kinase (MAPK) pathways (extracellular signal‐regulated kinase, Jun N‐terminal kinase and p38 MAPK), and nuclear factor‐κB pathways for T‐cell function.94, 95, 96

SMAD2‐deficient T cells have reduced ability for Th17 differentiation.57, 97 SMAD2 may modulate IL‐6R expression and STAT3 activation in the presence of TGF‐β.39 Another study found that SMAD2 binds to and synergizes with RORγt and positively regulates Th17 differentiation.57

SMAD3 appears to act in opposition to SMAD2. Deficiency of SMAD3 leads to enhanced Th17 differentiation both in vitro and in vivo.98 SMAD3 is also found to bind directly to RORγt, and inhibits RORγt transcriptional activity.

There is also an argument that neither SMAD2 nor SMAD3 is essential for Th17 differentiation. In one report, no defect in Th17 differentiation in either SMAD2‐ or SMAD3‐deficient mice was observed, whereas the non‐canonical MAPK pathway was implicated.99 In another report, Th17 differentiation in SMAD2‐deficient mice was reduced, but remained intact in SMAD3‐deficient mice, and was abolished in SMAD2 and SMAD3 double knockout mice.100 The deficiency of SMAD2 and SMAD3 did not alter Rorc gene expression but increased the level of the inhibitory cytokine, IL‐2.101 In the presence of IL‐2 blocking antibody, CD4⁺ T cells carrying SMAD2/3 double knockout restored half of the efficiency to differentiate into Th17 cells. These results suggest that SMAD2 and SMAD3 are partially and redundantly required for Th17 differentiation. A recent report supports the idea that SMAD2 plays a positive role, whereas SMAD3 plays a negative role in regulating Th17 differentiation.102 SMAD2 serves as a co‐activator, and SMAD3 serves as a co‐repressor in interactions with STAT3 to modulate Rorc and Il17a gene expression under the context of TGF‐β determined phosphorylation status.

These conflicting results regarding SMAD2‐ or SMAD3‐deficient mice among different studies appear confusing. This could be due to differences between wild‐type control cells and knockout cells that were compromised under certain in vitro culture conditions or treatments, including TGF‐β dose. Different doses of TGF‐β result in different R‐SMAD phosphorylation dynamics,103 and different efficiencies in Th17 induction.4 Nonetheless, all these studies acknowledge the involvement of the canonical TGF‐β pathway in Th17 differentiation.

SMAD4 is recognized as a canonical TGF‐β signalling central component. So it is reasonable to predict a TGF‐β signalling defect after deletion of SMAD4. Indeed, TGF‐β‐induced Foxp3⁺ Treg cell differentiation is impaired in SMAD4‐deficient mice. Astonishingly, however, Th17 differentiation is unaffected by the loss of SMAD4 under Th17 polarizing conditions.104 This observation was confirmed later by another study.105 Therefore, it appears that SMAD4 is not required for Th17 differentiation, and that TGF‐β induces Th17 differentiation through an SMAD4‐independent mechanism. However, in later studies, SMAD4 deletion counterbalances the severe autoimmunity caused by the loss of TGF‐β receptor II signalling, suggesting that SMAD4 could regulate T‐cell function on its own in the absence of TGF‐β receptor signalling.106 Inspired by the notion, Zhang et al.58 fortuitously found that SMAD4‐deficient naive T cells efficiently differentiated into Th17 cells when provided with IL‐6 alone in the absence of TGF‐β signalling. Although TGF‐β receptor II mice were unable to generate Th17 cells in the spinal cord and were protected from EAE, the additional deletion of SMAD4 fully restored Th17 differentiation as well as encephalomyelitis. These data suggest that SMAD4 is playing a decisive role in licensing Th17 differentiation. A series of experiments demonstrate that SMAD4 and repressor SKI interact and cooperatively suppress Th17 differentiation primarily through direct binding to multiple loci of Rorc. SMAD4 itself does not possess the suppressive activity but it is required for recruiting SKI to specific chromatin loci. The suppression effect can be offset in the presence of TGF‐β through TGF‐β‐directed degradation of SKI. These observations provide novel mechanistic insight into TGF‐β‐dependent Th17 differentiation. CD4⁺ T cells with CRISPR/Cas9‐disrupted SKI can differentiate into Th17 cells without the need for TGF‐β signalling. This suggests one important function for TGF‐β in Th17 differentiation is to degrade SKI. Further studies on SKI conditional knockout mice would be helpful to elaborate on the mechanistic details. Importantly, a single amino acid substitution, which released SKI from SMAD4 interactions and caused unchecked Th17 differentiation, mutually affirms the importance of losing SKI interactions in generating Th17 cells. Also, Ong et al.107 found that the absence of SKI in CD4⁺ T cells led to increased Th17 differentiation at low levels of TGF‐β. Indeed, low doses of TGF‐β can effectively degrade SKI and the degradation efficiency correlated with the induction efficiency of Th17 differentiation.58 The SKI/SMAD4 complex is highlighted as a switch for Th17 differentiation, and it might be also helpful to explain the generation of pathogenic Th17 cells by TGF‐β‐independent cytokines.46

Many questions remain for further mechanistic details, such as: why SMAD4 and SKI primarily suppress the Rorc gene but not other genes? Is it possible that SMAD4 affects the genome accessibility on the Rorc locus? Is the binding on the Rorc locus constitutive or dependent upon specific T‐cell receptor or cytokine stimulation? As SKI could serve as a potential therapeutic target for many diseases,108 it would be important to investigate how SKI is degraded by TGF‐β signalling in T cells. SMAD2 and SMAD3 were reported to mediate the ubiquitination and degradation of SKI,109 which may explain their partial requirement in T cells.58 However, TGF‐β signalling does not always trigger SKI degradation in all cell types.110 These questions must be addressed in future investigations.

SMAD‐independent pathways

In addition to SMADs, there are several studies that offer insights from non‐SMAD pathways to explain the role of TGF‐β in promoting Th17 differentiation.

It has been well known that both Th1 and Th2 cytokines potently inhibit Th17 differentiation.1, 2 TGF‐β is capable of inhibiting Th1 and Th2 differentiation through their lineage transcription factors T‐bet and GATA3.29, 111, 112 TGF‐β may exert its function by preventing Th1 and Th2 programmes to enhance Th17 differentiation. In T‐bet and STAT6 double‐deficient T cells, IL‐6 alone is sufficient to induce Th17 differentiation.77, 78 These results clearly showed that one of the mechanisms for TGF‐β in promoting Th17 differentiation is to negatively regulate Th1 and Th2 programmes. Currently, it is not clear whether the double deficiency of T‐bet and STAT6 can produce similar results in C57BL/6 mice as in these BALB/c mice. Notably, in these double knockout T cells, the addition of either TGF‐β or anti‐TGF‐β antibody had no effect on IL‐17 production in one report,78 but had around four times the promotion or reduction effect on IL‐17 production in another report.77 This discrepancy and genetic background difference obscure the interpretation of how exactly TGF‐β plays the role in Th17 differentiation. As SKI/SMAD4 also functions as a suppressor of RORγt expression, and SMAD4‐deficient T cells also can differentiate into Th17 cells with IL‐6 alone, it is not clear whether these two mechanisms are required for each other. For example, whether T‐bet and STAT6 double‐deficient T cells have disrupted SMAD4 or SKI function. STAT6 downstream target GATA3 was also reported to control TGF‐β‐induced Foxp3, suggesting possible cross talk between Th1/Th2 and the downstream of TGF‐β pathways.113, 114

Interleukin‐6‐induced suppressor of cytokine signalling 3 (SOCS3) is a negative feedback regulator of STAT3, and TGF‐β can suppress the transcription of SOCS3.115 Small interfering RNA knockdown of SOCS3 promotes Rorc expression and in vitro Th17 differentiation in the presence or absence of TGF‐β. These data indicate that TGF‐β suppresses SOCS3 to prolong STAT3 activation and to promote Th17 generation.

Under Th17 skewing conditions, protein kinase ROCK2 phosphorylated IRF4 to up‐regulate the expression of RORγt, IL‐17 and IL‐21. The TGF‐β is required to drive the activation of RhoA/ROCK2 to promote Th17 differentiation.116

To identify the SMAD‐independent mechanisms underlying TGF‐β in Th17 cells, Ichiyama et al.117 used SMAD2 and SMAD3 double‐deficient mice and identified the transcription factor Eomesodermin (Eomes), which directly binds to the Rorc and Il17a promoter region controlling Th17 differentiation. Ablation of Eomes expression by short hairpin RNA led to the induction of Th17 cells in the absence of TGF‐β, while overexpression of Eomes substantially constrained Th17 differentiation.117, 118 Eomes is subjected to the suppression of TGF‐β signalling through Jun N‐terminal kinase‐c‐Jun pathway, which provides an important mechanism for SMAD‐independent Th17 cell differentiation.

In summary, these SMAD‐dependent and SMAD‐independent pathways prove that TGF‐β induces Th17 differentiation not through a single pathway but rather that multiple mechanisms are involved to achieve efficient Th17 differentiation.

In recent years, there has been an increased understanding of environmental cues shaping the Th17 differentiation.119, 120, 121 For example, serum glucocorticoid kinase 1 (SGK1) can be up‐regulated by sensing the NaCl in a high‐salt diet through phosphorylation of p38 MAPK, which promotes IL‐23R expression, induces pathogenic Th17 cells, and exacerbates murine EAE.122, 123 As TGF‐β ₁ is able to stimulate the SGK1 expression, and SGK1 is highly specifically expressed under Th17 conditions,122 it could be another explanation for TGF‐β‐induced Th17 differentiation. Although there have been arguments regarding the physiological underpinnings of this model,124, 125, 126 these data provide a novel target for drug development and methods for the treatment of autoimmune diseases.

Although there are possibly more unrevealed TGF‐β signalling pathways for promoting Th17 differentiation, further studies will help to elucidate the precise mechanism of Th17 differentiation, as well as the pathogenicity of Th17 cells. Also, investigating the prominent pathways under different pathological conditions and environmental cues would be helpful for the intervention in Th17 differentiation to protect health.

TGF‐β signalling in Th17 cells plasticity

In addition to Th17 cells, naive T cells can differentiate into several types of helper T cells (Fig. 3). Increasing evidence has implicated the plasticity of T‐cell subsets.127 As TGF‐β is essential for the development of both Th17 and Foxp3⁺ Treg cells, the TGF‐β signalling pathway provides a dichotomous role and intimate link between Th17 and Foxp3⁺ Treg cells.128, 129 TGF‐β can induce Foxp3 expression, whereas the presence of IL‐6 or IL‐21 can prevent the conversion of Th17 to Foxp3⁺ Treg cells and promotes Th17 lineage commitment.8, 11, 12

Figure 3.

T helper differentiation. Diagram of T helper differentiation pathways for T helper 1 (Th1), Th2, Th9, Th17, Th22, regulatory T (Treg), regulatory T type 1 (Tr1) and T follicular helper (Tfh) cells. Cytokines that play an important role in inducing CD4⁺ T‐cell differentiation are listed above the cell types, and cell‐specific transcription factors are listed in a box below the cell types.

The reciprocal differentiation of Th17 cells and Foxp3⁺ Treg cells is involved in multiple immune mechanisms. TGF‐β signalling plays a critical role in the fate decision between Th17 and Foxp3⁺ Treg cells. For example, TGF‐β‐induced Foxp3⁺ Treg cells promote Th17 development through IL‐2 regulation.101, 130, 131 Mice deficient in tumour necrosis factor receptor‐associated factor 6 have less IL‐2 and therefore have promoted Th17 cell differentiation.132 In another study, TGF‐β down‐regulates the expression of a transcriptional repressor, Growth Factor Independent 1 (GFI‐1), which associates with histone lysine‐specific demethylase 1 to directly bind the intergenic region of Il17a/Il17f loci, and so represses IL‐17 expression.49 Another example in which TGF‐β regulates the Th17 and Foxp3⁺ Treg cell response is the TGF‐β‐induced Tafazzin (TAZ). TAZ is a co‐activator of RORγt, and decreases acetylation of Foxp3 to regulate the reciprocal balance between Th17 and Foxp3⁺ Treg cells.133 In addition, TGF‐β‐induced death‐associated protein kinase promotes the proteasomal degradation of hypoxia‐inducible factor 1 (HIF‐1), a key metabolic sensor,134 to regulate the balance of Th17 and Foxp3⁺ Treg cells, which ultimately prevents Th17‐mediated pathology in autoimmunity.135 In another model, TGF‐β promoted the colon‐specific trafficking molecule G Protein‐Coupled Receptor 15 (GPR15), which inhibits G Protein‐Coupled Receptor 174 (GPR174) on Foxp3⁺ Treg cells, and supports the accumulation and retention of Foxp3⁺ Treg cells in the colon to dampen Th17 responses.136 Finally, TGF‐β triggers the degradation of the newly identified T‐cell player SKI, and promotes Th17 differentiation.58 However, the deletion of SKI in Foxp3⁺ T cells also leads to increased Foxp3⁺ Treg cell conversion.107 These observations suggest that residual amounts of SKI may exist and play an important role in Foxp3⁺ Treg cells, possibly balancing the differentiation of Th17 and Foxp3⁺ Treg cells. Further study on the detailed role of SKI could be beneficial for the understanding of Th17 and Foxp3⁺ Treg cell reciprocal differentiation.

T helper 17 cells can also transdifferentiate into another Treg cell subset, type 1 regulatory T cells (Tr1) to halt inflammation.137, 138 Besides, TGF‐β converts Th1 cells into Th17 cells through the induction of RUNX1 expression.139 Conversely, Th1 lineage‐specific transcription factor T‐bet can repress RUNX1‐mediated RORγt transactivation to prevent Th17 differentiation.140 TGF‐β is required for human memory Th17 cells to produce Th9 cell cytokine, IL‐9.141 Moreover, murine Th17 cells in Peyer's patches can convert to the T follicular helper cells (Tfh) phenotype,142 possibly because TGF‐β is critical for providing additional signals for STAT3 and STAT4 to promote human Tfh differentiation.143

Currently, it is unclear if all T helper subsets have the potential to convert to each other under specific circumstances. Although Th17, Foxp3⁺ Treg, Tr1, Th9 and Tfh cells are distinct subsets, all of them can be promoted by TGF‐β.138 It is possible that these subsets share similar key factors associated with the TGF‐β signalling pathway, and the factors may be targeted as a switch in balancing autoimmunity and self‐tolerance. Hence it is important to investigate the underlying mechanisms and identify the key factors in the TGF‐β‐directed T helper programming network.

The importance of metabolic control in T‐cell function and plasticity has gained more and more attention in recent years.144, 145, 146, 147 The mTOR forms two structurally and functionally distinct complexes, mTORC1 and mTORC2; mTOR senses diverse environmental cues such as growth factors, oxygen, nutrients, energy and stress, and integrates into protein synthesis, lipid, nucleotide and glucose metabolism.148 Loss of mTORC1 results in failure of Th17 differentiation, whereas mTORC2 deficiency does not affect Th1 or Th17 differentiation.149, 150 Interestingly, if both mTORC1 and mTORC2 are inhibited, T cells skew toward Foxp3⁺ Treg cell differentiation. Although it remains unclear whether TGF‐β plays a pivotal role in regulating the mTOR pathway under pathological conditions, the cross talk between TGF‐β and mTOR signalling is certainly worthy of investigation. For example, mTORC2 could modulate SMAD2/3 linker phosphorylation to regulate TGF‐β or activin signalling.151

The mTOR is required for the induction of HIF1α, a key regulator mediating T‐cell glycolytic activity.152 HIF1α controls the T‐cell fate decision between Th17 and Foxp3⁺ Treg cells. HIF1α deficiency alters the dichotomy between these two T‐cell lineages by diminishing Th17 and promoting Foxp3⁺ Treg cell differentiation.134, 152 In vitro experiments shows that TGF‐β has the ability to induce the expression of HIF1α.134, 153 However, the proportion and importance of TGF‐β‐contributed induction of HIF1α in vivo is currently unknown.

Myc is another important target of mTOR. Myc controls glycolysis and glutaminolysis. It is required for metabolic reprogramming upon T‐cell activation.154 The fact that Myc could be repressed by TGF‐β,155, 156, 157 suggests a possible link between TGF‐β and T‐cell metabolism.

Aryl hydrocarbon receptor (AHR) responds to environmental toxins and metabolites. It boosts either Th17 or Foxp3⁺ Treg cell differentiation in a ligand‐specific fashion.158, 159 Although AHR‐deficient T cells could differentiate into Th17 cells, they fail to produce the Th17 cytokine, IL‐22.159 AHR is highly expressed in Th17 cells and Tr1 cells,160 but not Treg cells.158, 159 AHR expression is positively correlated with TGF‐β concentration in the presence of IL‐6, but is not required for IL‐22 production in the absence of TGF‐β.161 TGF‐β suppresses IL‐22 production in Th17 cells through c‐Maf but not AHR.161 There are also studies suggesting that the regulation of AHR by TGF‐β appears to be cell‐type specific.162, 163, 164 These paradoxes warrant further investigation on the role of TGF‐β in AHR modulation.

These metabolic checkpoints and pathways orchestrate the balance between Th17 cells and other subsets.165, 166, 167 The understanding of Th17 plasticity may provide tools for immunotherapies of autoimmune disease.168, 169 However, there are significant gaps between the TGF‐β signalling and T‐cell metabolic control. Therefore, it is important to investigate how TGF‐β shapes and affects metabolic reprogramming in T cells.

Concluding remarks and featured questions

The unravelling of the key factors and pathways responsible for both TGF‐β‐dependent and TGF‐β‐independent Th17 differentiation is essential for understanding the development, pathogenicity and plasticity of Th17 cells. With more understanding of the mechanisms underlying TGF‐β‐promoted Th17 differentiation in the future, the following questions should be addressed:

1. What are the dominant cytokine sources for tissue‐resident Th17 and pathogenic Th17 cells in different tissues or organs during health and disease conditions?

2. Is the dysregulation of TGF‐β or TGF‐β signalling responsible for the Th17 cells that have gone rogue?

3. How is Th17 differentiation regulated by various cytokine regimens? Does the SKI/SMAD4 complex play a role in other cytokine signalling pathways besides IL‐6 plus TGF‐β ₁?

4. Are there any other unknown key transcriptional suppressors or activators of RORγt that also respond to TGF‐β?

5. How exactly is the pathogenicity of Th17 cells regulated? What are the functions of TGF‐β signalling in pathogenic Th17 cells? Does the TGF‐β–SKI–SMAD4 axis play a role in the generation of pathogenic Th17 cells?

6. Are there specific markers and functional differences to distinguish initially committed and subsequently converted Th17 cells?

7. Within the context of autoimmune disease, is it appropriate to always assume Treg cells as friends while considering Th17 cells as foes? Do these cells work together during the initiation or progression phase of the disease?

8. Do all T helper cell subsets have the plastic potential to become any other subset? Is it possible to develop a drug to convert some, if not all, of the different subsets into one subset simultaneously?

9. Is TGF‐β or TGF‐β signalling involved in the metabolic control of T‐cell function?

Further comprehensive studies on the role of TGF‐β in controlling Th17 differentiation and function will allow us to gain insights on how Th17 differentiation impacts immune homeostasis under physiological context and facilitates the development of therapeutic intervention methods for inflammatory and autoimmune diseases.

Disclosures

The author declares no conflict of interest.