T helper 17 cell heterogeneity and pathogenicity in autoimmune disease

Abstract

Th17 cells have been proposed to represent a new CD4⁺ T cell lineage important for host defense against fungi and extracellular bacteria, and the development of autoimmune diseases. Precisely how these cells arise has been the subject of some debate with apparent species-specific differences in mice and men. Here, we describe evolving views of Th17 specification, highlighting the contribution of TGF-β and the opposing roles of STAT3 and STAT5. Increasing evidence points to heterogeneity and inherent phenotypic instability in this subset. Ideally, better understanding of expression and action of key transcription factors and the epigenetic landscape of Th17 can help explain the flexibility and diversity of IL-17-producing cells.

Recognition of complexity of Th cell subsets

Activated CD4⁺ T cells differentiate into distinct functional subsets, characterized by heritable patterns of cytokine secretion and the expression of specific transcription factors or so-called “master regulators” [1, 2]. Along with classical T helper (Th)1 and Th2 cells [3], new subsets of T cells continue to be recognized [4–6]. Of these “new” subsets, CD4⁺ T cells that preferentially produce IL-17 have attracted tremendous attention because of the connection with autoimmune disease [7–9]. The importance of autoreactive Th17 cells in mouse models of multiple sclerosis, psoriasis, rheumatoid arthritis and inflammatory bowel disease led to the establishment of treatments affecting cytokines of the Th17 network in autoimmune disease [10–13]. In humans, therapeutics targeting IL-12/IL-23p40, IL-17 or the IL-6 receptor are already in clinical practice for some of these autoimmune diseases. Given this onslaught of studies, it is useful to review the history of this field and to consider current controversies and discuss future challenges. In this article we examine the heterogeneity and pathogenicity of Th17 cells. Therefore, we focus on the IL-17-promoting cytokines IL-6, IL-1β, IL-23 and TGF-β and their signaling pathways causing Th17 cell diversity. We also discuss the opposite roles of STAT3, which can be activated by IL-6 and IL-23 and STAT5, which can be activated by IL-2, on Th17 differentiation.

The criticality of IL-23 in driving autoimmunity

The heterodimeric cytokine IL-23 (p40/p19), which shares the p40 subunit with IL-12 (p40/p35), has been identified as the crucial cytokine in Th17-associated pathology. In attempts to decipher the role of IL-12 subunits in mouse models of autoimmune disease, it was noted that mice deficient in IL-12p40 were resistant to experimental autoimmune encephalomyelitis (EAE) [14, 15]. However, mice lacking the IL-12p35 subunit developed more severe disease and in this respect were similar to mice lacking interferon (IFN)-γ. In contrast, gene targeting of another partner of p40, the p19 subunit of IL-23, resulted in resistance to EAE [15]. Therefore, an immune mechanism independent from IL-12 and Th1 cells must be responsible for T cell-mediated inflammatory autoimmunity. This result was a serious challenge to a prevailing, simple dichotomous view of Th1 versus Th2 cell differentiation, and launched a new era in the understanding of autoimmune mechanisms. Strengthening the connection between IL-23 and human autoimmunity is a considerable body of genetic evidence that links Il23R polymorphisms with susceptibility to a range of autoimmune diseases including: Crohn’s disease, spondyloarthropathy, autoimmune thyroid disease, multiple sclerosis, psoriasis, psoriatic arthritis, acquired aplastic anemia, and Behcet’s disease [16–22]. As will be discussed later, IL-23 activates Jak2 and Stat3, and polymorphisms of the genes encoding these signaling molecules are also linked to human autoimmunity [23–25]. Thus, in addition to mouse models, which may or may not accurately reflect immunopathogenetic mechanisms of human disease, independent genetic evidence has now linked IL-23 signaling to actual autoimmune disease in people.

Connecting the dots: IL-23 and IL-17 axis

First cloned in 1995, IL-17 was under appreciated for many years, even though there were clear data showing that it was associated with delayed-type hypersensitivity reactions and autoimmunity [8, 26]. The first work linking IL-23 to IL-17 production were in vitro studies using memory CD4⁺ T cells, and the authors posited that IL-23 might act during a secondary immune response to “promote an activation state with features distinct” from Th1 and Th2 cells [27]. Deficiency of IL-23 was then found to be associated with reduced production of IL-17 at sites of immune pathology [28]. Much later, the combination of IL-1β and IL-23 was noted to potently induce IL-17 in mouse γδ T cells [29, 30]. However, it proved much more challenging to drive naïve CD4⁺ T cells to differentiate to selectively produce IL-17. Initial studies using naïve T cells, argued that polarized Th1 and Th2 cells did not secrete IL-17. Stimulating cells with IL-23, while blocking IFN-γ and IL-4, resulted in small numbers of IL-17 producing T cells [31, 32]. The presence of cells that selectively produced IL-17 and not IFN-γ or IL-4 led the authors to propose that these cells represented a new helper cell lineage, Th17 cells.

Later, IL-17-producing T cells were found to express the transcription factor retinoic acid orphan receptor (ROR) gamma thymus (RORγt), which along with ROR-α, was critical for Th17 development and was required for the development of EAE [33, 34]. RORγt was thus argued to be the master regulator of Th17 differentiation, a notion that supported the lineage sovereignty of Th17 cells. However, other factors are also important for Th17 differentiation and they include: Batf [35], IRF4 [36], Runx1 [37], IκBδ [38] and IKKα [39]. Activated by IL-6 and IL-23, STAT3 was also recognized to be essential for IL-17 production in mouse and human cells, its importance being vividly illustrated in humans with hyper-IgE syndrome [40–43]. These individuals lack efficient Th17 cell development and suffer from infections with certain bacteria and fungi. In mice, deletion of STAT3 in T cells limits Th17 cell differentiation and pathology [44–49].

While an IL-23–IL-17 connection seems reasonably clear, it is also worth pointing out that IL-23 does not necessarily equal IL-17. Despite the presence of IL-17-producing T cells in EAE, absence of Il17f has only minor effects on EAE development and severity [50–52]. Mice deficient in Il17a or IL-17RA develop attenuated signs of EAE [51–53], but the severity of disease is greater than in mice lacking p40. In this regard, it is intriguing that the Il17 locus has not emerged as a susceptibility gene for human autoimmune disease in the many genome-wide association studies that have been reported. Taken together, the data suggest that in terms of immunopathology, IL-23 should not be equated with IL-17. IL-6 signaling through STAT3 is crucial for both IL-23 receptor expression and for IL-17A and IL-17F induction, and defective IL-6 signaling is also associated with a more profound block in the development of EAE suggesting that IL-23 has effects beyond simply promoting IL-17 production [14, 15, 46, 54–58]. Thus, while it is clear that IL-23 can drive IL-17 production, it is also clear that IL-23 and IL-17 are not synonymous in terms of autoimmunity [13].

A more efficient recipe for Th17 differentiation: TGF-β and IL-6

While the importance of IL-23 in driving IL-17-mediated pathology was (and remains) clear, what was less obvious was how to efficiently generate IL-17-producing cells. Especially relevant was that naïve CD4⁺ T cells express little or no receptor for IL-23. So, what are the initiating factors that induced receptor expression and first specify Th17 fate commitment?

In vitro, culture of regulatory T (Treg) cells with naïve CD4⁺ T cells was found to efficiently drive generation of IL-17 producing cells, which was attributed to Treg cell production of TGF-β [59]; it was quickly confirmed that TGF-β and IL-6 efficiently generated populations of IL-17-producing cells from naïve precursors [60, 61]. It was notable that although TGF-β (especially with addition of exogenous IL-2) upregulated Foxp3 and generated induced Treg (iTreg) cells, IL-6 inhibited Foxp3. This suggested that Tregs and Th17 cells are developmentally related [5, 62, 63].

The next challenge was to establish whether TGF-β and IL-6 were requisite for in vivo Th17 generation and to discern how these factors drove Th17 specification mechanistically. Several approaches have been used including transgenic expression of dominant negative TGF-β receptor II, deletion of TGF-β in T cells using CD4-cre, and, most recently, selective deletion of TGF-β1 in activated T cells and Treg cells [64–66]. Collectively, these models do support the idea that T cell-derived TGF-β is important for Th17 differentiation. Accordingly, genetic models in which TGF-β was absent were also resistant to EAE, leading to the view that T cell-derived TGF-β is critical for the generation of encephalitogenic Th17 cells.

Although resistance to EAE and reduced generation of Th17 cells is seen in models in which there is absence of TGF-β or disruption of TGF-β signaling in T cells, there is another consequence, namely profound autoimmunity. This pathology is characterized by overproduction of multiple cytokines especially IFN-γ and IL-4, factors known to attenuate Th17 differentiation. More importantly, IFN-γ limits the severity of EAE. This raises the question is TGF-β really important because it provides an obligate instructive signal to cells to become IL-17 producers or is TGF-β primarily limiting production of other cytokines, which in turn indirectly attenuate EAE?

It is notable that despite the association between Th17 cells and inflammatory disease in both mice and humans, TGF-β and IL-6-induced Th17 cells are weakly pathogenic in mice in the setting of EAE [67–69]. One explanation for this is that TGF-β and IL-6-induced Th17 cells make IL-10, with subsequent exposure to IL-23 being required for pathogenicity [68, 70]. Accordingly, the pathogenicity of Th17 cells generated with TGF-β can be enhanced by neutralizing IL-10 [68]. Taken together, the data suggest that the combination of IL-6 and TGF-β produces a population of cells that selectively produces IL-17 and not IFN-γ, but alone this cocktail may not sufficient to generate cells with pathogenic potential. The anti-inflammatory phenotype of this Th17 cell population is characterized by the production of cytokines like IL-9 and IL-10, which in vivo can be found in RORγt⁺FoxP3⁺ Th17 cells [68, 71–73]. Thus, a population of Th17 cells with a phenotype similar to cells that received TGF-β signals does exist in vivo; however their exact function in autoimmunity and host defense has to be determined. It should also be noted that Th17 cells generated by TGF-β in vitro can become IFN-γ producers in vivo and attain diabetogenic or colitogenic potential after adoptive transfer into immunocompromised mice [74, 75].

Those pesky human cells

Shortly after the recognition of factors that promote differentiation of mouse Th17 cells, efforts were made to generate human IL-17-producing CD4⁺ T cells, which like their murine equivalent, express the master regulator RORγt. The cocktail of IL-6, IL-1 and IL-23 was noted to readily induce IL-17-secreting cells from naïve cells, but in some cases TGF-β was found to inhibit rather than promote Th17 development [76–81]. One argument raised was that perhaps the human cells were not quite as “naïve” as their mouse counterparts [82]. Later, when cord blood naïve T cells were used rather than adult blood along, low doses of TGF-β were found to induce Th17 development [83, 84].

Another complexity revealed in these studies was that human Th17 cells readily produced IFN-γ [78, 79]. Although pure populations of IL-17⁺IFN-γ⁻ cells are easily generated in the mouse, this is harder to do in human CD4⁺ T cells. At the same time though, in lesional tissue in the setting of autoimmunity, it is quite common to see IL-17 and IFN-γ double producers; this is the case in both mouse and man [85, 86]. So it is by no means clear that in the setting of autoimmunity populations of pure IL-17⁺IFN-γ⁻ cells are necessarily the relevant players.

To beta or not to beta; that is the question: Murine Th17 revisited

Given the perplexing data from human cells and the complex effects of TGF-β, we and others revisited the requirement for TGF-β in Th17 differentiation. In T cells deficient in T-bet and STAT6 expression, IL-6 alone was found to induce IL-17 production even in the absence of TGF-β signaling [87]. This was interpreted to indicate that the TGF-β acts indirectly to regulate IL-17 by suppressing factors that drive other cell fates [88], especially Th1 and Th2 cell differentiation.

As indicated above, work from both mouse and human cells indicated that STAT3 is essential for Th17 differentiation. What is the mechanism by which STAT3 participates in the development of this subset of cells? Second generation or deep sequencing technology has permitted mapping and quantitation of transcription factor binding on a genome-wide scale [89]. This technology can also be used to measure epigenetic changes and gene expression and when used in conjunction with cells from knockout mice, one can comprehensively define the genomewide actions of a given transcription factor [90]. This is precisely what was done for STAT3 in Th17 cells. STAT3 bound to and regulated multiple genes that contribute to the Th17 phenotype including: the Il17 locus itself, Il21, and Il23r but also CCR6 [46, 49, 69]. STAT3 also bound to genes encoding transcription factors critical for Th17 differentiation including Rorc (which encodes Rorγt), Irf4, Batf and Nfibiz [49].

Given STAT3’s prominent role and its ability to bind and regulate Il23r expression along with key transcription factors and the contradictory information on the requirement for TGF-β in human IL-17 regulation, the question of whether TGF-β is really required for initiating Th17 differentiation has been recently revisited. By culturing cells in serum-free medium and using multiple strategies to block TGF-β signaling (including antibodies and genetic and pharmacologic approaches), the IL-23 receptor was found to be induced in the absence of TGF-β1 and addition of IL-23 then further induced receptor expression. Consequently, the combination of IL-1, IL-6 and IL-23 was able to induce IL-17 production in a TGF-β1-independent manner [69]. Consistent with these findings, Th17 cells are also present in the gut of mice with deficient TGF-β signaling [69, 91]. In this regard, it is worth pointing out that conditional deletion of TGF-β in T cells did not significantly alter the proportion of gut Th17 cells [66]. Moreover, deletion of SMAD2, SMAD3 or SMAD4 in T cells does not prevent Th17 development [92–94]. The expression of RORγt is also not reduced in SMAD2- and SMAD3-deficient T cells, and IL-17 can be induced in these cells when IL-2 is neutralized [94].

Th17 differentiation and the balance of STAT3 vs STAT5 activation

As already mentioned, inclusion of Treg cells in naïve CD4 T cells cultures promotes Th17 differentiation [59]. However, more recent work argues that Treg cell production of TGF-β is dispensable for Th17 differentiation [66, 95] and there is an alternative explanation for the original finding regarding Treg cells promoting Th17 cells. IL-2 was recently found to potently inhibit Th17 differentiation [45] whereby blocking IL-2 or IL-2 signaling in vitro and in vivo results in marked expansion of IL-17-producing cells. Two very recent papers conclude that the mechanism by which Treg cells promote Th17 differentiation is via their ability to act as IL-2 “sinks”, offering a re-interpretation of the original Treg cell finding [95, 96]. In this scenario, consumption of IL-2 by CD25⁺ Tregs promotes the differentiation of Th17 cells in vitro and in vivo, and in this manner Tregs can help clear C. albicans infection in mice by enhancing IL-17 production [96].

In aggregate, these results indicate that IL-2 is an important regulator of IL-17 production. It then becomes an important question to define the mechanisms through which it inhibits IL-17 expression. STAT5 is responsible downstream of IL-2. Even though STAT5 directly regulates the Foxp3 gene, this is not the mechanism underlying the inhibition of IL-17 and nor is IL-2–STAT5 acting primarily to downregulate Rorγt. Rather, STAT5 acts directly on the Il17 gene. As STAT3 directly binds the promoter and enhancer elements in Il17, STAT5 binding to this region was also assessed [46]. Surprisingly, STAT5 binds the same elements as STAT3, and furthermore, displaces STAT3. By varying the levels of cytokines that activate STAT3 versus STAT5, it was found that the balance of activation of these two transcription factors determines the outcome fate of the T cells. That is, under circumstances in which there is minimal activation of STAT5, the threshold for Th17 differentiation is greatly reduced. In this context, it should be borne in mind that multiple factors can activate both STAT3 and STAT5, so a number of factors might influence the proclivity of cells to make IL-17. Recall for instance that different preparations of media tend to promote or inhibit Th17 differentiation [97]. Specifically, unknown factors in media complemented with fetal bovine serum for instance seem to activate STAT5 and thereby inhibit Th17 differentiation. Presumably other factors in vivo can also tip the balance.

Instability and heterogeneity of Th17 cells

After the identification of Th17 cells as a putative new subset linked to autoimmunity, several groups generated Th17 cells in vitro and adoptively transferred the cells to provoke autoimmune disease. In addition to the aforementioned problem that not all Th17 cells are pathogenic, another issue emerged. Specifically, after transfer in vivo, Th17 cells quickly acquired the ability to produce IFN-γ and lost their ability to produce IL-17 [74, 98]. This was recapitulated in vitro in that, with multiple rounds of culture, Th17 cells begin to produce IFN-γ. TGF-β inhibited this tendency, but Th17 cells express IL-12 receptors and readily produce IFN-γ in response to exposure to IL-12 [75]. The tendency to produce IFN-γ was explained by the accessibility of the Ifng locus, as measured by DNase hypersensitivity mapping, which was surprisingly similar in Th17 and Th1 cells [99]. Moreover, upon stimulation with IL-12 the Ifng locus in Th17 cells undergoes rapid epigenetic remodeling with increased H3K4me3 and decreased H3K27me3. Moreover, the Tbx21 gene (which encodes T-bet, a Th1 “master regulator”) shows bivalent (H3K4me3 and K3K27me3) epigenetic marks in Th17 cells, indicating that this transcription factor gene is poised for expression [100]. Further evidence for plasticity and instability of Th17 cells was provided by IL-17A reporter mice [86]. That is, a substantial proportion of IFN-γ⁺ T cells during EAE were previous IL-17 producers and the appearance of IL-17⁺IFN-γ⁺ T cells required IL-23 signaling.

Another issue that arises when generating Th17 cells in the absence of TGF-β (i.e. with IL-1, IL-6 and IL-23) was that their transcription profile is quite different from conventional, TGF-β-induced Th17 cells. Importantly, in the absence of TGF-β, the cells express transcription factors expressed by Th1 cells, T-bet and Hlx1 [69]. IL-6, IL-1 and IL-23-generated Th17 cells also express IL-33 [69], which is associated with inflammatory immune responses. Adoptive transfer experiments showed that Th17 cells made in this manner are much more pathogenic than TGF-β-induced cells, which express higher levels of c-Maf and Ahr, CCL20, IL-9, and IL-10 [68, 69, 71, 101]. This is notable in that human Th17 cells and clones can co-express RORgt and T-bet [102], as do lesional T cells from patients with multiple sclerosis also express T-bet along with Rorγt [85].

Concluding remarks

Although IL-17 was discovered in 1995 and IL-23 was discovered in 2000, we still have much to learn about what it means for a T cell to make IL-17. An obvious question is whether these cells really represent a “lineage”. Clearly differentiated Th cells seem to be more flexible with respect to lineage commitment than we initially thought and there is accumulating evidence that subsets can express more than one “master regulator” [2]. For instance, we now know that a subset of FoxP3⁺ Tregs can co-express T-bet, which is required for appropriate trafficking [103]. Moreover, GATA3⁺ Th2 cells have been reported to convert into a phenotype that co-expresses IL-4, IFN-γ, GATA3 and T-bet, allowing them to serve a protective role in viral infection [104]. Th17 cells are no exception to this emerging complexity; on the contrary, they appear to be particular unstable and readily become Th1 cells. IL-17 producing cells may be a heterogenous collection and it is possible that this heterogeneity may be influenced by the site and circumstances of induction. For instance, perhaps in environments with high levels of TGF-β, populations that express IL-17, and possibly IL-10, but little IFN-γ, might be generated. One could imagine that such cells would be useful in terms of host defense but would have limited pathological potential (perhaps at mucosal sites). On the contrary, T-bet⁺Rorγt⁺ T cells might be more relevant in terms of autoimmune mechanisms, which in many circumstances represent a mixed picture of Th1 and Th17-mediated pathology. Certainly, in human disease this seems to be the case. Evidence that allergic asthma is promoted by IL-17-producing IL-4⁺ Th cells that co-express GATA3 and RORγt is particular intriguing in this respect [105]. It would not be surprising to learn in the future that “Th17” cells represent a spectrum of cells with subtly different functionalities.

Another unresolved question is whether TGF-β “promotes” Th17 differentiation or if it primarily acts by limiting expression of T-bet, GATA3 and other transcription factors associated with other fates. Since TGF-β is ubiquitous, there are probably no circumstances in which Th17 cells arise in the absence of this cytokine. The question is more of a mechanistic problem. Does deletion of TGF-β impair Th17 differentiation because Th1 differentiation is exacerbated? Is the protective effect of eliminating TGF-β in EAE simply a reflection of the fact that IFN-γ is protective in such models? Or does TGF-β have more direct roles in actively instructing cells to become IL-17 producers? The finding that STAT3 and STAT5 compete to regulate IL-17 production adds a new level of complexity to regulation of this gene; the finding raises the possibility that multiple factors might fine tune expression.

The understanding of cytokine signaling and its role in controlling T cell responses are the basis for the development of effective, safe and more specific therapies. IL-23 is clearly functionally and genetically linked to a number of autoimmune diseases. The insights about the pathogenic potential of IL-23 in experimental mice led to the establishment of therapies targeting IL-23. Targeting IL-12 and IL-23 with an antibody against the shared subunit, p40, is highly effective in autoimmune diseases as psoriasis and Crohn’s disease, but it remains to be determined whether selectively targeting IL-23 will be as effective. Surprisingly, in patients with relapsing-remitting multiple sclerosis the effect of anti-p40 therapy was disappointing. This may reflect the complexity of the pathogenesis of human autoimmune disease that may be different from experimental mouse models [106]. Antagonizing the action of IL-6 is effective in the treatment of some autoimmune diseases, but whether this is related to effects on Th17 cells has not been established. The efficacy and safety of neutralizing IL-17 itself is under clinical investigation in autoimmune settings, but initial reports suggest different responses in psoriasis and rheumatoid arthritis [107]. Clearly more work is required to fully understand the relevance of the IL-23/IL-17 connection in the spectrum of autoimmune disease. However, one thing’s for sure - we will continue to learn a great deal about Th17 cells in human autoimmunity. With any luck, we will soon be able to better sort out when and where to use drugs that target these cells and improve the treatment of these diseases.

Figure 1. Heterogeneity of Th17 cells.

The conventional mode of inducing Th17 differentiation is to culture cells with IL-6, IL-1β and TGF-β. IL-6 acts to upregulate IL-23R and IL-21 expression. Subsequently, IL-6, IL-21 and IL-23, acting through STAT3, directly regulate many of the phenotype-defining Th17 genes. Though low levels of TGF-β clearly promote IL-17 production (a), a major unresolved question is its mechanism of action with respect to this cytokine. One can view TGF-β as providing an important positive instructive signal. However, TGF-β is not required to induce IL-23R expression. In addition, it is possible to generate Th17 cells with little or no TGF-β (b). Thus, an alternative explanation is that the major action of TGF-β is to inhibit IL-2, IFN-γ, T-bet and other factors that antagonize IL-17 production (c). Of note is that Th17 cells generated with little or no TGF-β are phenotypically and functionally distinct from conventional Th17 cells in terms of the transcription factors they express and the cytokines they produce. Also, Th17 cells are inherently unstable and can readily express IFN-γ, which further inhibits IL-17 production. Precisely, what the in vivo correlates of these different populations of Th17 cells are has not been defined exactly. However, in sites of autoimmune-mediated damage, cells that express RORγt and T-bet are present (d). IL-2 is key factor that negatively regulates IL-17 production. The inhibition of IL-17 by IL-2 is mediated through STAT5, which can directly compete with STAT3 to limit Il17 transcription. Tregs can provide TGF-β for IL-17 production, but more importantly promote Th17 cell differentiation by consuming IL-2 (e). Therefore, Treg participate in host defense and pathogen clearance by inducing IL-17.