Introduction

Spondyloarthritis (SpA) refers to a cluster of inflammatory conditions which share clinical genetic and pathophysiological characteristics Disease classification by the recently established Assessment of Spondyloarthritis International Society (ASAS) criteria is based on clinical outcome of patients, with subdivision in axial SpA (including both non-radiographic and radiographic axial SpA, also known as ankylosing spondylitis (AS)) and peripheral SpA. Typical symptoms include rheumatic/articular features (arthritis, dactylitis, sacroiliitis, and spinal inflammation) next to extra-articular manifestations, such as anterior uveitis, psoriasis, and inflammatory bowel disease (IBD). Interestingly, about 50% of SpA patients present microscopic signs of ileum (and colon) inflammation without showing overt GI symptoms . Notably, for axial SpA patients, predictive factors including (younger) age, progressive disease, male sex, and higher disease activity as measured by Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) are associated with an increased likelihood of presenting this subclinical inflammatory gut signature . In addition, it was shown that SpA patients with chronic gut inflammation display a higher degree of bone marrow edema of the sacroiliac joints (visible by magnetic resonance imaging (MRI)) as compared to those with normal gut architecture . Eventually, on average 6.5% of SpA patients develop clinically established IBD in the 5-year follow-up . These data further underscore the existence of a significant gut–joint axis in SpA pathology.

In contrast to rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE), being typical autoimmune diseases, SpA is generally considered to be autoinflammatory in nature . In this regard, it is postulated that primary triggers induced by bacterial components (microbial antigens, Toll-like receptor ligands) and/or exposure to biomechanical stress (mechanical strain) can lead to dysregulated innate inflammatory processes in targeted tissues of susceptible individuals . Inflammatory cytokines play a crucial role in these disease processes as shown by the marked efficacy of biologicals directed against tumor necrosis factor (TNF)-α and more recently interleukin (IL)-23/IL-17 pathway mediators .

These cytokines can be produced by a diverse range of immune cells, some extending beyond the classical spectrum of lineage-defined subsets. Indeed, with the recent advances in multicolor flow cytometry and imaging techniques, specific new subsets of immune cells such as innate-like T cells, innate lymphoid cells (ILCs), and myeloid subsets have been identified, of which some have been suggested to play a key role in SpA. Especially, cells showing an IL-17 signature (the so-called Type-17 cells) have acquired increased attention in recent years . In this chapter, we will provide an overview of the emerging new immune subsets and critically evaluate their potential contribution to SpA disease pathology ( Fig. 1 ).

New immune cells in spondyloarthritis: Key players or innocent bystanders? A schematic overview of immune cell subsets, highlighting their main phenotypical characteristics and cytokine profile and their hypothesized role in SpA (as discussed throughout the text). Abbreviations: ILC3 (Group 3 innate lymphoid cells), MAIT (mucosal-associated invariant T cell), iNKT (invariant natural killer T), KIR3DL2 ⁺ (killer cell immunoglobulin-like receptor 3DL2 ⁺ ) T cell, and KLRG1 ⁺ (killer cell lectin-like receptor G1 ⁺ ) T cell. PLZF (promyelocytic leukemia zinc finger), iTCR (invariant T cell receptor), IL-23R (IL-23 Receptor). Hi: high expression, int: intermediate expression, Lo: low expression, ?: unknown pro- or anti-inflammatory role in SpA as indicated.

KIR3DL2 ⁺ and KLRG1 ⁺ T cells

As established for over more than four decades now, HLA-B27, an MHC class I gene, shows the strongest genetic association with SpA, accounting for roughly 30% of heritance . Although from an immunological perspective, CD8-restricted T lymphocytes, potentially autoreactive toward joint-associated self-antigens, would be apparent candidates in the disease processes, solid evidence for this “arthritogenic peptide” hypothesis was never established . It rather seems that the unusual biochemical properties of HLA-B27 molecules provide a better rationale for their link with SpA. Specifically, HLA-B27 molecules show an increased tendency to misfold and to form aberrant disulfide-linked heavy chain homodimers . HLA-B27 misfolding can potentially initiate an unfolded protein response (UPR) in the endoplasmic reticulum (ER) and/or can activate autophagy pathways, processes responsible for the degradation and recycling of cellular organelles . Interestingly, in HLA-B27 transgenic rats presenting marked SpA disease features, it was shown that HLA-B27 misfolding and UPR activation can be responsible for an enhanced IL-23 expression by myeloid cells. Direct evidence for this condition in SpA patients is currently lacking .

Alternatively, B27 dimers can activate a particular subset of T helper cells (and natural killer (NK) cells) which express the killer cell immunoglobulin-like receptor (KIR)3DL2 . Paul Bowness and his colleagues have shown increased numbers of circulating KIR3DL2 ⁺ CD4 T cells and NK cells in SpA patients . Moreover, KIR3DL2 ⁺ CD4 T cells survive, proliferate, and produce IL-17 (next to e.g., interferon (IFN)-γ and TNF-α) upon stimulation with cells expressing B27 dimers, and particularly this IL-17 ⁺ cell subset is expanded and enriched in peripheral blood and knee joint aspirates of SpA patients .

Apart from a clear association with HLA-B27, other T cells also expressing natural killer receptors (NKRs) were associated with SpA pathology. Melis et al. showed that T cells positive for killer cell lectin-like receptor G1 (KLRG1) are enriched in the synovial fluid (SF) of SpA but not crystal-induced arthritis patients . Mechanistically, it is suggested that KLRG1 ⁺ (predominantly CD4 ⁺ ) T cells are attracted to the joint compartment, as they possess a significant migratory activity toward SpA patient-derived SF in vitro. This could probably be explained by their particular chemokine receptor expression profile, being positive for CCR6, CCR5, and CXCR3. In the SpA SF, very high levels of soluble E-cadherin (sE-cadherin) molecules could be detected, which are able to compete for binding to KLRG1 with natural ligands such as membrane-bound E-cadherin and αEβ7 integrins. It has been shown that KLRG1 signaling can lead to functional inhibition of T cell function . However, functional in vitro assays indicated that sE-cadherin enhanced TNF-α production by KLRG1 ⁺ CD4 ⁺ T cells suggesting that sE-cadherin might contribute to the local proinflammatory environment in the joint . In contrast to KIR3DL2 ⁺ CD4 T cells, these cells do not produce typical Th17 cytokines upon TCR stimulation. Taken together, the findings of increased percentages of pro-inflammatory KIR3DL2 ⁺ and KLRG1 ⁺ CD4 ⁺ T cells in the chronically inflamed joint fluid suggest a clinically relevant role for these cells in SpA pathology. However, the precise impact of NKR-expressing cells for instance in driving disease chronicity/progression is not clear and should be further investigated.

KIR3DL2 ⁺ and KLRG1 ⁺ T cells

As established for over more than four decades now, HLA-B27, an MHC class I gene, shows the strongest genetic association with SpA, accounting for roughly 30% of heritance . Although from an immunological perspective, CD8-restricted T lymphocytes, potentially autoreactive toward joint-associated self-antigens, would be apparent candidates in the disease processes, solid evidence for this “arthritogenic peptide” hypothesis was never established . It rather seems that the unusual biochemical properties of HLA-B27 molecules provide a better rationale for their link with SpA. Specifically, HLA-B27 molecules show an increased tendency to misfold and to form aberrant disulfide-linked heavy chain homodimers . HLA-B27 misfolding can potentially initiate an unfolded protein response (UPR) in the endoplasmic reticulum (ER) and/or can activate autophagy pathways, processes responsible for the degradation and recycling of cellular organelles . Interestingly, in HLA-B27 transgenic rats presenting marked SpA disease features, it was shown that HLA-B27 misfolding and UPR activation can be responsible for an enhanced IL-23 expression by myeloid cells. Direct evidence for this condition in SpA patients is currently lacking .

Alternatively, B27 dimers can activate a particular subset of T helper cells (and natural killer (NK) cells) which express the killer cell immunoglobulin-like receptor (KIR)3DL2 . Paul Bowness and his colleagues have shown increased numbers of circulating KIR3DL2 ⁺ CD4 T cells and NK cells in SpA patients . Moreover, KIR3DL2 ⁺ CD4 T cells survive, proliferate, and produce IL-17 (next to e.g., interferon (IFN)-γ and TNF-α) upon stimulation with cells expressing B27 dimers, and particularly this IL-17 ⁺ cell subset is expanded and enriched in peripheral blood and knee joint aspirates of SpA patients .

Apart from a clear association with HLA-B27, other T cells also expressing natural killer receptors (NKRs) were associated with SpA pathology. Melis et al. showed that T cells positive for killer cell lectin-like receptor G1 (KLRG1) are enriched in the synovial fluid (SF) of SpA but not crystal-induced arthritis patients . Mechanistically, it is suggested that KLRG1 ⁺ (predominantly CD4 ⁺ ) T cells are attracted to the joint compartment, as they possess a significant migratory activity toward SpA patient-derived SF in vitro. This could probably be explained by their particular chemokine receptor expression profile, being positive for CCR6, CCR5, and CXCR3. In the SpA SF, very high levels of soluble E-cadherin (sE-cadherin) molecules could be detected, which are able to compete for binding to KLRG1 with natural ligands such as membrane-bound E-cadherin and αEβ7 integrins. It has been shown that KLRG1 signaling can lead to functional inhibition of T cell function . However, functional in vitro assays indicated that sE-cadherin enhanced TNF-α production by KLRG1 ⁺ CD4 ⁺ T cells suggesting that sE-cadherin might contribute to the local proinflammatory environment in the joint . In contrast to KIR3DL2 ⁺ CD4 T cells, these cells do not produce typical Th17 cytokines upon TCR stimulation. Taken together, the findings of increased percentages of pro-inflammatory KIR3DL2 ⁺ and KLRG1 ⁺ CD4 ⁺ T cells in the chronically inflamed joint fluid suggest a clinically relevant role for these cells in SpA pathology. However, the precise impact of NKR-expressing cells for instance in driving disease chronicity/progression is not clear and should be further investigated.

Entheseal resident RORγt ⁺ T cells

In a mouse SpA model, Sherlock et al. have identified a specific subset of unconventional T cells, believed to drive inflammation upon induced systemic overexpression of IL-23 (by minicircle DNA technology) . These cells expressed the IL-23R, were double negative for CD4 and CD8, and stained positive for retinoic acid receptor-related orphan receptor-γt (RORγt), a key transcription factor originally linked to Th17 cells . Diseased mice showed a progressive polyarthritis, with both peripheral and axial components, next to psoriasis-like and aortic root inflammation. Interestingly, typical hallmarks of SpA such as enthesitis, sacroiliitis, and new bone formation (bone remodeling) were clearly present as shown by comprehensive histological and imaging techniques. Rather unexpectedly, no noticeable signs of gut inflammation were described. This latter probably reflects a technical issue as intestinal inflammation has been shown to be induced in similar IL-23-driven disease models . Mechanistically, this particular IL-23R ⁺ T-cell subset seemed to drive inflammatory pathways in affected tissues by secretion of the pro-inflammatory cytokines IL-17 and IL-22. Further analyses demonstrated that both IL-17 and IL-22 could drive joint inflammation upon IL-23 induction, but IL-22 principally caused the osteoproliferative changes observed in these animals. There is currently no evidence that these cells have a human equivalent and it might be hard to prove this in patients given their rare anatomical distribution. Because of the non-unique phenotypic characteristics, for example, they also express the innate immunity-associated transcription factor promyelocytic leukemia zinc finger (PLZF), it is reasonable to speculate that these cells are related to or overlapping with other innate-like T cells such as iNKT or MAIT cells (see next paragraph). Nevertheless, this well-designed study provides convincing evidence for the role of IL-23-responsive immune cells in the development of SpA features.

iNKT and MAIT cells

Invariant natural killer T (iNKT) and mucosal-associated invariant T (MAIT) cells are two prominent subsets of innate-like T cells showing rapid responses upon T cell receptor (TCR) activation but also TCR-independent (mainly cytokine-driven) triggering, thus acting as a “bridge” between innate and adaptive immune responses . A common feature of these cells is the expression of a semi-invariant TCR (iNKT: Vα14-Jα18 in mice, Vα24-Jα18 in humans; MAIT: Vα33Jα19 in mice; and Vα7.2Jα19 in humans) and their antigen restriction towards non-polymorphic MHC-like molecules (CD1d and MR1, respectively) . The nature of these antigens profoundly differs, with iNKT cells recognizing glycolipid molecules, predominantly of foreign origin (α-glucuronosyl ceramide structures, diacylglycerols, cholesteryl sugars, and phospholipids), although endogenous lipid antigen have also been recognized (β-anomeric forms of glycosylceramides and phosphoglycolipids) . iNKT cells can contribute to defense mechanisms against pathogenic microbial infections by recognizing CD1d-restricted bacterial lipid ligands . On the contrary, other reports suggest that this iNKT immunosurveillance is strictly dependent on cytokine-driven activation pathways . MAIT cells can be activated by vitamin B2 (riboflavin) metabolites such as ribityllumazines and pyrimidines . This is related to the ability of MR1 to capture, stabilize, and present chemically unstable pyrimidines to MAIT cells . MAIT cells can also recognize Vitamin B9 (folic acid) derivatives, but these are non-stimulatory and rather inhibitory by competition . Many vitamin biosynthetic pathways are unique to bacteria and yeast organisms, suggesting that MAIT cells recognize these ligands to detect microbial infections.

iNKT cells are present at low frequencies in lymphoid tissues (0.01–0.5% of the T lymphocytes in spleen bone marrow, thymus, lymph nodes, and blood in mice) but are more abundantly present in the liver and white adipose tissues (up to 20–40% of lymphocytes). Remarkably, they appear to be less frequent in the corresponding organs in humans, with large inter-individual differences in the blood compartment (range 0.01–4%) . Different subsets of iNKT cells have been described. Based on transcriptional expression patterns and cytokine profiles, iNKT cells can be subdivided into T-bet ⁺ NKT1, GATA3 ⁺ NKT2, and RORγt ⁺ NKT17 cells similar to T helper cells and ILCs . But there are also reports on other, mainly regulatory subsets, including NKT10 cells (dominant IL-10 producers) , Bcl6-positive follicular helper NKT cells (dominant IL-21 producers) , and E4BP4 ⁺ iNKT cells (IL-10 producers in adipose tissue) . MAIT cells, predominantly consisting of CD8 ⁺ and CD4CD8 double-negative populations, are hardly detectable in specific pathogen-free laboratory animals but they appear to be a very substantial subset of T cells in humans, accounting for 1–10% of T cells in peripheral blood and being predominantly present in liver and mucosal tissues . Hence, it appears that MAIT cells outnumber iNKT in humans.

The study of iNKT and MAIT cells in SpA pathology has been limited, but some interesting results have been reported. Jacques et al. showed that iNKT cells attenuate joint and gut inflammation in TNF ^ΔARE mice, an animal model characterized by a dysregulated TNF expression leading to the development of peripheral and axial arthritis, next to a Crohn’s like ileitis . Activation of iNKT cells in these mice was induced by a crosstalk with inflammatory CD1d ^high dendritic cells, a population which was also found to be enriched in SF samples of SpA patients . In contrast to this protective role for iNKT cells in SpA pathology, the lack of MAIT cells in MR1 knock-out mice led to reduced severity and susceptibility of, respectively, collagen-induced arthritis (CIA) and collagen antibody-induced arthritis (CAIA), two (RA-like) disease models . At present, functional alterations of iNKT and MAIT cells in human SpA pathology have not been reported, but in patients with other rheumatic diseases and IBD, quantitative (mainly a drop in frequency in circulation) and qualitative alterations of these cell types have been described . Finally, iNKT and MAIT cells display plasticity suggesting that they could be skewed from an immunoprotective toward a predominant pro-inflammatory profile in SpA. These unique characteristics and their association with joint and gut pathology warrant further investigation into their potential role as regulators or effector cells in human SpA.

Group 3 innate lymphoid cells (ILC3)

ILCs form a relatively rare population of innate immune cells, which lack typical markers associated with T cells, B cells, myeloid cells, or granulocytes, making them a recently discovered lineage-negative immune cell subset. As common for other innate(-like) cells, ILCs are enriched at barrier surfaces, such as skin, lung, and intestine, but they are also abundantly present in adipose and certain mucosal-associated lymphoid tissues. In this way, they are ideally positioned/located to promote early immune responses to different kinds of pathogenic challenges . Recent data now also provide evidence for a broader, more regulatory role for these cells in the resolution of inflammation and limiting/restraining chronicity. For instance, they seem responsible for repair of damaged tissues (described for lung and gut) and able to influence metabolic processes and to directly modulate innate and adaptive immune cells to non-harmful environmental stimuli (commensal bacteria or dietary) in the gut (reviewed in ). Although the terminology was quite confusing at the time of their discovery, ILCs are now generally classified into three subsets, termed group 1, 2, or 3 ILCs . As also described for iNKT cells, ILC subsets mirror T helper cell counterparts in terms of their cytokine-producing capabilities and transcriptional phenotype. Group 1 ILCs (ILC1) constitutively express the transcriptional Th1 cell-related regulator T-bet and promote immunity toward intracellular pathogens and tumor cells. They respond to IL-12 and predominantly secrete IFNγ and TNFα . Group 2 ILCs (ILC2s), which express high levels of the typical Th2-related transcription factor GATA3, respond toward extracellular (multicellular) parasites via IL-4, IL-5, IL-9, IL-13, and amphiregulin production. ILC2 cells are activated by the cytokines IL-25, IL-33, and thymic stromal lymphopoietin (TSLP) . Finally, Group 3 ILCs (ILC3s) produce IL-17 and IL-22 rapidly in response to either extracellular bacteria or fungi. Akin to Th17 cells, these cells express RORγt and respond to IL-1β, IL-6, and IL-23 . The most substantial heterogeneity for ILCs has been described within this group, including CCR6 ⁺ lymphoid tissue-inducer (LTi) cells (enriched in intestine and lymphoid tissues), CCR6 ⁻ T-bet ⁺ ILCs which can be positive for the natural cytotoxicity receptors NKp46 and NKp44 (cells exclusively found in the skin and intestinal lamina propria) and even RORγt ⁻ ILC3s (by some authors delineated as exILC3). ILCs are known for their protective and regulatory immune responses, but they can also promote chronic inflammation as observed in several disease mouse models, and dysregulated ILC characteristics have been noted in patient populations with chronic inflammatory diseases (psoriasis, asthma, and IBD) .

Recently, Ciccia et al. characterized ILC1, ILC2, and ILC3 cells in ileal and bone marrow biopsies and also in peripheral blood and SF mononuclear cells obtained from patients with AS . No differences in ILC1 and ILC2 cells were noted; however, a prominent expansion of Group 3 ILC in AS patients, defined by the authors as Lyn ⁻ IL ⁻ 23R ⁺ NKp44 ⁺ T-bet ⁺ RORγt ⁻ cells based on both confocal microscopy and flow cytometric analyses, was demonstrated. Although of interest, the particular phenotype of the described ILC3 precludes firm conclusions. Given their lack of RORγt expression, it has to be determined whether these cells truly represent ILC3 cells (if so, one might potentially classify them as exILC3). Interestingly, this report suggests that the defined ILCs express the homing integrin α4β7, with its counter-receptor MAdCAM1 significantly upregulated in the HEVs of the gut and the BM of patients with AS, consistent with earlier work . The ILC3 cells, which produce IL-17 and IL-22, look gradually expanded in acute to chronic inflamed gut tissues of AS patients, with normal levels being detected in AS patients without microscopic signs of gut inflammation. Moreover, ILC numbers in the gut were correlated with disease activity as assessed by BASDAI. However, theoretically, these observations could underscore a previously proposed “shuttle” hypothesis , meaning that specific subsets of pro-inflammatory cells differentiate in the gut and might be able to “transfer disease” to extra-intestinal sites (joint, skin, …) . This would be consistent with the established joint–gut axis in SpA. Recent evidence of intestinal dysbiosis in SpA patients further indicates that a dysregulated balance between regulatory and effector function of these cells upon response to aberrant microbial encounter could be a trigger for uncontrolled intestinal inflammation. Alternatively, as mentioned, these ILC3 cells can also play an important role in regulating tissue repair. In this regard, it cannot be ruled out that these observations by Ciccia et al. reflect a mechanism of the body to counter damage caused by the ongoing inflammatory processes in the gut. In patients with IBD, both reduced as well as increased numbers of ILC3 have been reported arguing for a complex role of these cells in ongoing intestinal inflammation .

Altogether, the exact role of ILC3s in AS has to be further investigated, including translational research, taking into account the substantial heterogeneity and potential lineage plasticity of this subset.