Translational Mini-Review Series on Th17 Cells: Are T helper 17 cells really pathogenic in autoimmunity?

Abstract

In this review the authors discuss the evidence for T helper type 17 (Th17) cells as pathogenic T cells in autoimmunity. Studies with cytokine-deficient mice or blocking of interleukin (IL)-17, IL-21 and IL-22 have resulted in a conflicting data set. Although in the experimental autoimmune encephalomyelitis model the role of Th17 cells remains a point of debate, this IL-17-producing T cell in experimental arthritis is clearly contributing to inflammation and destruction.

Introduction: T helper type 17 (Th17) cells in infection and autoimmunity

Since the discovery of the T helper type 17 (Th17) cell as a new subset of effector T cells, the role of this CD4⁺ interleukin (IL)-17-producing T cell has been studied extensively in both infection and autoimmunity (reviewed by Crome et al. in this series [1]). During infection, Th17 cells provide protection against pathogens by attracting other inflammatory cells to the side of inflammation. Of great importance in this process is the secretion of the Th17 cytokine IL-17, which is a potent inducer of chemokines such as IL-8/CXCL8 that attracts neutrophils. In autoimmunity, the exact role of the Th17 cell in the process of inflammation and subsequent tissue damage is less clear, and a debate is ongoing as to whether the Th17 cells are really pathogenic in autoimmunity. The regulation of IL-17/Th17 cells and other T helper cell subsets (Th1, Th2) is discussed elsewhere in this series [2,3]. In this review, we will focus upon the relative contribution of Th1 and Th17 cells and their respective cytokines interferon (IFN)-γ and IL-17/IL-21/IL-22 to autoimmune pathogenesis, as demonstrated in various animal models for autoimmunity such as experimental arthritis and experimental autoimmune encephalomyelitis (EAE), a model for multiple sclerosis (MS).

Discovery of IL-23: shift in concept from Th1 to Th17

For a naive CD4⁺ T cell to become activated, it needs to encounter an antigen presented by an antigen-presenting cell (APC), co-stimulation to prevent the T cell from becoming anergic and, depending upon the cytokine signals it receives, this activated T cell will change into a differentiated T helper subset. In the past, the concept of T helper cells was restricted to a dichotomy of Th1 and Th2 cells: Th1 cells differentiate in the presence of IL-12 and produce IFN-γ, whereas Th2 cells are stimulated by IL-4 and secrete IL-4, IL-10 and IL-13 (Fig. 1). At that time, autoimmune diseases such as rheumatoid arthritis (RA) and multiple sclerosis (MS) were regarded as typical Th1 diseases, but that concept changed after the discovery of a third T helper subset, the Th17 cell. The differentiation of Th17 cells is discussed in more detail by de Jong et al. in this series [4].

Fig. 1.

T helper type 17 (Th17) formation from precursor Th0 cells and major products of Th17 cells.

Studies on the role of IL-12 by blocking or knocking out the two subunits of the IL-12 protein, IL-12p35 and IL-12p40 resulted in confusing results and led finally to the discovery of a new T helper cell: the Th17 cell. Our group demonstrated previously a dual role for IL-12 during collagen-induced arthritis (CIA): amelioration of arthritis when anti-IL-12 antibodies were injected around the onset of disease, but acceleration of arthritis after prolonged treatment [5]. This striking contrast was also shown with injections of recombinant IL-12, which resulted in aggravation of CIA when given early, and suppression when injected during established arthritis. A few years later, again in conflict with the ruling concept, it was demonstrated that mice deficient for IL-12p35 were not protected against autoimmune models of arthritis or EAE. Further studies demonstrated a new subunit called IL-23p19, which also binds to IL-12p40, thereby forming the IL-23 protein. Subsequently, it was demonstrated that IL-23p19^−/− but not IL-12p35^−/− mice are resistant to EAE and CIA [6,7]. These findings suggested that, in contrast to the previous concept, not IL-12/Th1 but the IL-23/Th17 axis is crucial in a variety of animal models of autoimmunity.

The Th17 cell and its cytokines in pathological processes of autoimmunity

IL-17 and its actions

Th17 cells are regulated by a combination of cytokine stimuli such as IL-6, transforming growth factor (TGF)-β, IL-1β and IL-23, and these cells produce a wide range of cytokines: IL-17, IL-17F, IL-21 and IL-22 (Fig. 1). IL-17 contributes to inflammation by inducing chemokines such as IL-8/CXCL8 (KC/CXCL1 in mouse), CCL3/major inflammatory protein (MIP)-1α and CCL2/monocyte chemoattractant protein (MCP)-1 to recruit polymorphonuclear leucocytes (PMNs) and macrophages to the site of inflammation. Recently it was described that IL-17 itself can also function as a chemokine, acting directly on the IL-17RA and IL-17RC of monocytes to induce their migration [8]. IL-17F has similar effects to IL-17A (generally called IL-17), but is clearly less potent than its more famous family member [9]. Deletion or blocking of IL-17 during experimental arthritis results in protection against joint inflammation and bone erosion [10–14], and EAE develops with suppressed incidence and severity in mice deficient for IL-17 [15].

Although a range of publications report the benefits of IL-17 inhibition, some recent reports dispute the pathogenicity of the Th17 cells. In an elegant set of transgenic mice, Haak et al. [16] provided evidence that neither the T cell-specific overexpression of IL-17 nor its complete loss had a major impact on the disease course of autoimmune neuroinflammation. It might be argued that IL-17F could compensate for this. However, blocking of IL-17 in IL-17F-deficient mice during EAE also had, at best, a marginal effect on disease development. In line with this, O'Connor et al. [17] demonstrated that adoptive transfer of myelin-specific T cells depleted of Th17 cells remained highly pathogenic. This is in striking contrast to an earlier paper by Komiyama et al. [15], who reported that transfer of EAE with T helper cells from IL-17^−/− mice was not efficient. This difference might be caused by the way in which the differentiated T cells were prepared. Where Komiyama simply uses the entire T cell population of IL-17-deficient mice and only restimulates the cells in vitro with antigen for 72 h before transfer, O'Connor tried to manipulate the mixed Th1/Th17 response to generate clean T cell populations using additional Th1- (IL-12 and IL-18) or Th17-polarizing (IL-23, IL-6, TGF-β) conditions during antigen restimulation in vitro. Th1-polarized cells clearly initiated EAE, but Th17 cells could only induce mild disease with delayed onset. Because the transferred Th17-enriched donor cells showed a striking shift towards IFN-γ-producing cells when isolated from the central nervous system (CNS) of the recipient mice, we cannot exclude that a more stable Th17 population can be pathogenic and be used to transfer EAE adoptively.

Another key observation of the study by O'Connor et al. [17] was the fact that only Th1 cells can access the non-inflamed CNS, paving the way subsequently for Th17 cells to enter affected tissue. It is conceivable that Th17 cells are contributing to disease mainly in the context of Th1-initiated disease. Other events, apart from Th1 entry, may also enhance accessibility for Th17 cells, such as immune complex-mediated tissue damage. The latter certainly plays a role in some arthritis models and may partly explain differences in Th17 dependencies in arthritis and encephalomyelitis.

IL-21 and IL-22

The discussion on the pathogenicity of Th17 cells is not restricted to debates on IL-17 studies. For the other Th17 cytokines, conflicting results have also been published. IL-21 is an autocrine cytokine that functions as a positive feedback signal for the Th17 cell. However, the IL-21R is expressed not only on Th17 cells, but on a variety of cells such as B and natural killer (NK) cells and dendritic cells (DCs), resulting in pleiotropic effects of IL-21. IL-21 is important for normal B cell homeostasis, and IL-21 also has shown unexpected immunosuppressive effects, inducing the expression of IL-10 [18]. Blocking of IL-21 during CIA using an IL-21R.Fc ameliorated arthritis severity [19], and anti-IL-21 treatment in RA synovial cell cultures inhibited the release of tumour necrosis factor (TNF)-α, IL-1β and IL-6 [20]. IL-21 administration before onset of EAE resulted in increased severity by boosting NK cell function, but did not show any effects after onset administration of IL-21 [21]. However, blockade of IL-21 during EAE reduced regulatory T cell (T_reg) numbers and function, resulting in proliferation of autoreactive T cells and enhanced inflammation [22]. These results were confirmed in IL-21R-deficient mice that showed earlier and more severe development of EAE due to a defect of T_reg cells [23,24].

For IL-22 the same story applies: publications vary from IL-22 being pathogenic to IL-22 as a protective factor. IL-22 is a member of the IL-10 subfamily, and is produced by activated Th17 cells. Experimental arthritis in mice deficient for IL-22 is suppressed despite increased levels of antigen-specific antibodies [25], whereas another group demonstrated that IL-22-deficient mice were fully susceptible to EAE [26]. In contrast, a protective role of IL-22 has been described in inflammatory bowel disease (IBD) and experimental hepatitis, suggesting more IL-10-like characteristics. However, not only Th17 cells but also NK T cells were an important source of this protective cytokine in these models [27,28].

In vivo studies resulted in conflicting views on the pathogenic or protective role of the Th17 cell and its cytokines, but data from in vitro studies have mainly underlined the pathogenic character of the Th17 cell. In vitro, it was shown that Th17 cytokines induce the expression of proinflammatory cytokines and chemokines to promote joint inflammation and cartilage destruction, especially in synergy with other cytokines such as TNF and IL-1. Similar to IL-17, both IL-21 and IL-22 can induce the expression of receptor activator of nuclear factor-kappaB (NF-κB) ligand (RANKL), a regulator of osteoclastogenesis [29], thereby contributing to bone destruction in the arthritic process. In contrast to Th1 cells, Th17 cells themselves also highly express RANKL [30], but this expression alone is not sufficient to promote osteoclastogenesis.

Th1 cells and IFN-γ: contribution to inflammation and tissue destruction

Although Th17 cells appear to be quite capable of inducing inflammation and tissue destruction, the Th17 cell is not the only CD4⁺ T cell that can do this. Th1 cells producing IFN-γ can be destructive in the absence of IL-17 or Th17 cells. IFN-γ has been shown to up-regulate Fc gamma receptor expression in the joint, leading to enhanced cartilage erosion and chondrocyte death in passive immune complex-mediated experimental arthritis [31]. Autoimmune proteoglycan-induced arthritis (PGIA) appears to be mainly a Th1-mediated model, in contrast to most Th17-driven arthritis models. The reason for this peculiar phenotype is largely unexplained. The PGIA model is suppressed but not completely prevented in IFN-γ-deficient mice, whereas PGIA could be elicited easily in IL-17-deficient mice, to demonstrate that severe and destructive arthritis can occur independently of IL-17 [32]. Interestingly, IFN-γ-deficient mice showed not a full absence, but merely a delayed onset of disease, suggesting an early role for Th1 cells in this model. A role of IL-17 in late disease was unmasked in IFN/IL-17 double knock-outs, showing complete suppression of both early and late arthritis (A. Finnegan, personal communication). In line with this, IL-17 levels were low in early murine methylated bovine serum albumin (mBSA)-induced arthritis, but much higher in subsequent T cell-driven flares, and these exacerbations could be blocked fully with anti-IL-17 antibodies [12]. Remarkably, T-bet deficiency had no effect on PGIA, which might indicate that early IFN-γ dependency reflects processes other than Th1 cell involvement.

Previous literature had already identified that blocking of IFN-γ in autoimmune models can lead to aggravation of disease rather than protection. This was found in encephalomyelitis [33] as well as in collagen arthritis [34]. In addition, numerous groups have shown that autoimmune collagen arthritis can be induced in resistant mouse strains, when IFN-γ is knocked out. In the absence of IFN-γ, the controlling role of IFN-γ on Th17 development is lacking and florid Th17-dependent arthritis occurs. It needs to be stressed that this is found under conditions of excessive boosting with mycobacteria (also discussed in next paragraph). Further, it illustrates a delicate balance between Th1 and Th17 activity, and that targeting only one of these CD4⁺ T cell lineages can still result in florid inflammation and tissue destruction.

Impact of environment and co-stimuli

Using another model of autoimmune inflammation, Luger et al. [35] showed that conditions of disease induction affect dominant Th17 or Th1 effector category. After immunization with retinal antigen interphotoreceptor retinoid binding protein (IRBP) in complete Freund's adjuvant an IL-17-dependent uveitis develops, whereas induction of experimental autoimmune uveitis with IRBP-pulsed mature dendritic cells displayed the generation of an IFN-γ-producing effector response. The data indicate that conditions of initial antigen exposure, including the quality/quantity of Toll-like receptor (TLR) stimulation and/or type of APCs determine dominant effector phenotype. Similarly, IL-1Ra^−/− mice develop spontaneous Th17-dependent arthritis under normal housing conditions, but arthritis is lost completely under germ-free conditions. It suggests that bacterial flora drives the generation of autoimmune Th17 cells under conditions of excess IL-1. Crossing studies with various TLR-deficient mice identified that TLR-4, in particular, was responsible for skewing to Th17 responses and development of an erosive arthritis phenotype [36].

Elegant studies by Kroenke et al. [37], using either IL-12p70 or IL-23 polarized T cells, demonstrated that both subsets could induce an adoptive transfer of encephalomyelitis which was macroscopically indistinguishable. However, histology identified the dominance of neutrophils in IL-23-driven IL-17-dependent lesions, whereas a macrophage rich infiltrate was found in IL-12p70-driven lesions. It remains to be seen which type of encephalomyelitis is the most tissue-destructive. Of note, IL-17 alone is arthritogenic, but not overly erosive. However, when IL-17 is overexpressed under conditions of a concomitant mild immune complex arthritis, IL-17 greatly amplifies joint inflammation and joint destruction, in particular [38].

Pathogenic character of Th17 cells is plausible, but its dominance is not yet proven

In summary, in arthritis research we can find many indications for a pathogenic role of Th17 cells. Targeting IL-17 or the other Th17 cytokines IL-21 or IL-22 resulted in most cases in suppression of inflammation and joint destruction (Table 1). No eminent role was found for the IL-17-producing T cells in models that are only strongly dependent upon Th1 cells [32].

Table 1.

Demonstrated role of interleukin (IL)-17, IL-21 and IL-22 in experimental arthritis and experimental autoimmune encephalomyelitis (EAE).

	Experimental arthritis	EAE
IL-17−/− or IL-17R−/− mice	↓ Incidence and severity of CIA [11]	↓ Incidence and severity of EAE [15]
	↓ Severity of chronic SCW arthritis [13]
Anti-IL-17 treatment	↓ Severity of CIA4 [10]	Minimal effects on EAE [16]
	↓ AIA [12]
	↓ Severity in IL-1Ra−/− mice [14]
	– No effect in PGIA [32]
IL-21−/− or IL-21R−/− mice	↓↓ Full protection in K/BxN mice [29]	↑ Incidence and severity of EAE [23]; [24]
Anti-IL-21 treatment	↓ Severity of CIA [19]	↑ Severity of EAE [22]
	↓ Cytokines by synovial tissue cultures [20]
Addition of IL-21	–	↑ EAE before onset, no effect after onset [21]
IL-22−/− mice	↓ Incidence and severity of CIA [25]	Normal susceptibility to EAE [26]
Anti-IL-22 treatment	–	–
Adoptive transfer of T cells	–	Depleted of IL-17+ T cells still pathogenic
	O'Connor [17]
	IL-17−/− T cells not efficient for transfer of EAE
	Komiyama [15]

CIA: collagen-induced arthritis; PGIA: proteoglycan-induced arthritis; SCW: streptococcal cell wall.

In contrast, animal studies on EAE have resulted in a conflicting set of data (Table 1). Blocking of IL-17 or IL-22 results in mild or no suppression of the disease and suggests a relatively small contribution of Th17 cells to EAE pathology. Addition of IL-21 before the onset of disease enhanced EAE severity but, unexpectedly, blocking the IL-21/IL-21R pathway also resulted in aggravation of EAE. As discussed above, entry of Th17 cells is limited in non-inflamed CNS and becomes significant only after an initiating influx of Th1 cells, herein creating a mixed Th1/17 pathology and reduced dominance of Th17 in this disease. An amplifying role of immune complexes to facilitate influx in an affected tissue is obvious in arthritis, but less clear in encephalomyelitis.

Th17 cells have been demonstrated at the site of inflammation and destruction, but this presence does not necessarily mean that Th17 cells contribute to the pathological process. To prove that Th17 cells indeed play a pathogenic role in autoimmune models, we need to target specifically the Th17 cell, without affecting other T cell populations. A complicating factor in the Th17 research is that all Th17 cytokines and cell surface markers are not unique for Th17 cells, as they are also expressed on, for example, invariant NK (iNK) T cells and γδ T cells. The latter can also be a significant source of IL-17 production and dependency. What we can conclude at present is that Th17 cells are very likely to contribute to experimental arthritis, but that more in-depth research is needed, particularly in the field of EAE, to explain the exact pathogenic role of Th17 cells. Future tools will hopefully help us to demystify the exact function of Th17 cells in autoimmune processes.

Disclosure

Marije I Koenders and Wim B van den Berg have nothing to disclose.