One of the most satisfying aspects of the genome-wide association studies conducted in patients with ankylosing spondylitis (AS) has been the large number of genes that have been shown to affect susceptibility to disease (1). The first gene of this kind to be identified was the interleukin-23 receptor (IL-23R). A potential role of IL-23 signaling in the pathogenesis of AS was further supported by additional genes associated with AS, including the following: 1) IL-12B, which encodes the 40-kd (p40) chain, a component of both IL-12 and IL-23; in the latter, it is paired with a 19-kd protein (p19); 2) caspase recruitment domain 9 (CARD-9), which encodes a major signaling intermediate in the pathway whereby fungal products induce IL-23 production from dendritic cells (DCs); and 3) STAT-3, which encodes the major signaling molecule activated by the IL-23R. Additionally, there are suggested associations (which need further confirmation) with JAK-2, which encodes another important signaling molecule downstream of IL-23R, and prostaglandin E receptor 2 subtype EP2 (PTGER-4), which encodes a prostaglandin receptor that drives IL-23 secretion by DCs (2). Taken together, these genetic associations strongly implicate IL-23 and the cells upon which it acts in the pathogenesis of AS.
Hitherto, there have been few reports on altered production of IL-23 in AS patients, but Zeng et al, whose study is published in this issue of Arthritis & Rheumatism (3), report higher levels of IL-23 production by in vitro–derived macrophages from AS patients. Similar experiments using DCs showed that those from AS patients maintained IL-23 production as compared to those from rheumatoid arthritis patients, though a statistically significant difference from the production by cells from healthy subjects was not seen (4).
If it is accepted that there is a relative increase in IL-23 production in spondylarthritis (SpA), what might account for this? Zeng et al pursued the idea that induction of the unfolded protein response (UPR) in HLA–B27+ macrophages might be responsible, but they did not find any evidence that this response was occurring, since they did not see up-regulation of the 3 genes that are generally up-regulated by the UPR. Importantly however, only HLA–B messenger RNA levels were measured, where a ∼2-fold increase was seen. Whether this increase led to an appreciable increase in protein expression was not explored. This line of investigation followed the hypothesis first developed by Colbert et al (5), who showed that the B27 heavy chain folds inefficiently in the endoplasmic reticulum (ER) and can give rise to ER stress (5). Thus, when B27 expression was up-regulated by, for example, treating B27-transgenic rat cells with interferon-γ (IFNγ), the UPR was induced, and production of certain cytokines (including IL-23) was enhanced.
It has also been shown that IL-23 secretion in response to Toll-like receptor (TLR) agonists is very sensitive to induction of the UPR, with marked enhancement, in contrast to the decreased IL-12 production, which is seen under the same experimental conditions (6). Central to this enhancement is the transcription factor CCAAT/enhancer binding protein homologous protein 10 (CHOP), which binds to the IL-23p19 promoter. Short hairpin RNA–mediated knockdown of CHOP expression inhibits the enhanced production of IL-23 seen with the combination of TLR agonists and UPR. However, the UPR is only one mechanism whereby CHOP is up-regulated, and other forms of cellular stress, including viral infection, can induce CHOP through the integrated stress response, which does not include all of the components of the UPR. Therefore, while B27 misfolding and induction of the UPR may be one way to enhance IL-23 production, it is not the only one, and other inducers of the integrated stress response, including several relevant to the pathogenesis of SpA (such as intracellular infection), should be borne in mind. It would be interesting to know to what extent, if at all, the increased production of IL-23 from AS macrophages described by Zeng et al was CHOP-dependent.
Aside from the cellular biology responsible for altered IL-23 production, the study by Zeng et al supports the emerging role of IL-23 in AS and raises the question of how IL-23/IL-23R signaling may affect the pathogenesis of the disease. The major hypothesis here is that altered IL-23 levels and/or altered IL-23R signaling lead to enhanced IL-17 production in AS. The theoretical background for this hypothesis is provided by murine studies, which have delineated the function of IL-23 in differentiation of Th17 cells. The situation was initially confusing, since differentiation of Th17 cells appeared not to require IL-23, but instead, transforming growth factor β (TGFβ) combined with IL-6 and/or IL-1.
It has now been convincingly shown that Th17 cells can be differentiated in the absence of TGFβ, using a combination of IL-23 and IL-6 (7). Furthermore, such cells have different properties from those of cells differentiated in the presence of TGFβ. In particular, they are capable of inducing diseases such as experimental autoimmune encephalomyelitis upon transfer (whereas their TGFβ-induced counterparts do not), and this pathogenicity correlates with their ability to produce other cytokines, such as IFNγ and granulocyte–macrophage colony-stimulating factor (GM-CSF). This explains why IL-23 is critical for several murine models of autoimmunity, including arthritis, since the disease is essentially abolished in p19 gene–targeted mice. Accordingly, there is ample evidence implicating Th17 cells in the pathogenesis of various forms of experimental arthritis, and the properties of IL-17 are those you would expect of an arthritogenic cytokine, with monokine-like effects on cartilage and a role in osteoclastogenesis. Finally and most importantly, IL-17A blockade has recently been shown to lead to clear clinical improvement in AS, providing direct evidence for the involvement of IL-17 in SpA (8).
What is the direct evidence that alterations in the IL-23 signaling pathways lead to an enhanced IL-17 response by T cells from AS patients or patients with other forms of SpA? First, the principal IL-23R polymorphism associated with AS (R381Q)—a protective polymorphism—has recently been shown to decrease responses to IL-23, particularly downstream STAT-3 phosphorylation, thus leading to decreased numbers of IL-17+ and IL-22+ T cells in healthy subjects carrying R381Q (9). Second, AS is also associated with genetic variants of the receptors for IL-1, which act together with IL-23 in the differentiation of Th17 cells. Third, several, but not all, investigators have reported increased numbers of CD4+IL-17+ T cells in peripheral blood from AS patients and increased production of IL-17 by AS patient T cells in vitro (10). Similar evidence of increased numbers of IL-17/IL-22+ cells has been obtained in peripheral blood and synovial fluid samples from patients with psoriatic arthritis, as well as in synovial fluid from patients with reactive arthritis. And finally, HLA–B27+ individuals and AS patients display an expanded population of KIR-3DL2+CD4+ T cells, which is enriched for IL-17 production. Killer cell immunoglobulin-like receptor 3DL2 (KIR-3DL2) recognizes HLA–B27 in a homodimeric, but not a conventional heterodimeric, form, and it is suggested that expression of these homodimers may drive the expansion of this specific Th17 subset (11).
Despite these indications, formal proof that IL-23 exerts pathogenic effects through IL-17 production by canonical α/β T cells in AS is still awaited, since such Th17 cells were not detected in the axial and peripheral lesions of AS patients (12, 13) and since T cell–directed therapies, such as abatacept, have not proved efficacious in this disease.
There are alternative possibilities: IL-23 can promote IL-17 production not only by α/β T cells, but also by a variety of other cells. Based on original observations in SCID mice and IL-23R-green fluorescence protein reporter mice, it has now been demonstrated that γ/δ T cells, innate lymphoid cells, and mast cells can all produce large amounts of IL-17 upon triggering by IL-23. More importantly, direct analysis of affected tissues in humans with SpA as well as in associated diseases, such as psoriasis and inflammatory bowel disease (both of which are also genetically associated with the same single-nucleotide polymorphisms in the IL-23R), has recently demonstrated that innate immune cells, such as mast cells, neutrophils, and innate lymphoid cells, are the major producers of IL-17 at sites of inflammation (12–14). In addition, many of these IL-23–responsive cells produce not only IL-17, but also associated cytokines, so that the relative contribution of IL-17 to pathology, as compared to other cytokines, may differ from one condition to another (Figure 1).
As noted above, the pathogenicity of Th17 cells in murine studies has been linked to the production of IFNγ and GM-CSF. IL-22 is particularly relevant to arthritis, since its receptor, IL-22R, is expressed exclusively on mesenchymal cells. Thus, the cytokine is an important link, providing signals from the immune system to the stroma. The effects of IL-22 on synovial stromal cells have yet to be investigated; IL-22 can either be protective, as seen in models of colitis (15), or pathogenic, as illustrated by its effects on keratinocytes in patients with psoriasis (16).
In summary, the data presented by Zeng et al (3) concerning increased IL-23 production by AS macrophages add support to the notion of a pathogenic IL-23/IL-17 axis in SpA. As indicated above, however, further investigation is required to define which mechanisms contribute to altered IL-23 production, which cells respond to the excessive production of IL-23 in SpA, and which IL-23–induced cytokines apart from IL-17 may contribute to important facets of the disease.