Th17 Cells Expressing KIR3DL2⁺ and Responsive to HLA-B27 Homodimers Are Increased in Ankylosing Spondylitis

The spondyloarthritides are a group of common inflammatory rheumatic diseases characterized by axial and peripheral arthritis, extraskeletal manifestations including uveitis, and an extremely strong association with HLA-B27 (B27) (1, 2). Ankylosing spondylitis (AS) is the most common of the spondyloarthritides, and 94% of patients are B27⁺ compared with 9.4% of control subjects, with an odds ratio of 171 (1). Reactive arthritis (ReA) is a related form of spondyloarthritis (SpA) that follows gastrointestinal or genitourinary infection with certain Gram-negative bacteria including Salmonella and Chlamydia species. The pathogenic role of B27 remains unknown (reviewed in Ref. 2). Like all classical class 1 proteins, HLA-B27 presents peptide Ags to CD8 T cells, but it also shows the unusual propensity to form H chain homodimers (“B27₂”). This is mediated by an unpaired cysteine at position 67, and we have shown that B27 dimers, unlike classical HLA-B27, interact with the killer-cell Ig-like receptor (KIR) KIR3DL2.

Recent genomewide association studies have implicated additional genes whose products are involved in innate and/or adaptive immune function, including the IL-23R (3). IL-23R is expressed by a subset of proinflammatory CD4 T cells known as Th17 cells (4), whose development is stimulated by cytokines including IL-23. The signature cytokine produced by these cells is IL-17A (henceforth termed IL-17). Increased levels of IL-17 have been described in AS serum (5, 6) and synovial fluid (7, 8). Two recent studies have found increased frequencies of Th17 cells in AS (9, 10). Th17 cells have been implicated in the pathogenesis of other autoimmune diseases including psoriasis, rheumatoid arthritis (RA), multiple sclerosis, and inflammatory bowel disease (11, 12). Th17 cells have been identified both in blood and at sites of inflammation including psoriatic skin (13) and rheumatoid joints (14), and have potent proinflammatory actions, including stimulation of human macrophages to produce TNF and IL-1β (15), and neutrophil recruitment (16, 17). Th17 cells also play an important role in protective immune responses to certain microbial pathogens such as Klebsiella pneumoniae (16) and Bordetella pertussis (18).

KIR are a family of MHC class I-binding receptors expressed by NK and minor subsets of T cells. KIR3DL2 is a three-Ig domain KIR whose cognate ligands include HLA-A3 and A11 (19). Our studies have shown that KIR3DL2 does not recognize HLA-B27/β₂-microglobulin/peptide heterotrimeric complexes but binds to B27₂, which are expressed both on leukocytes of AS patients and on transfected cell lines (20, 21).

We have previously shown an increase in KIR3DL2-expressing CD4 T cells in the peripheral blood of B27⁺ SpA patients (22), but the role of these cells in promoting inflammation in SpA and their interaction with B27₂ has not been defined. KIR ligation by HLA class I can inhibit activation-induced cell death of NK and T cells (22, 23). Thus, we hypothesized that interaction of B27₂ with KIR3DL2 expressed on CD4 T cells might have a net proinflammatory effect by promoting T cell survival and subsequent cytokine production. We asked whether KIR3DL2 interaction with B27₂ affected the proliferation, survival, and cytokine production of Ag-stimulated CD4 T cells. We also investigated to what degree KIR3DL2⁺-expressing CD4 T cells contributed to production of IL-17 in AS.

In this article, we show that B27₂-expressing APCs stimulate the survival and proliferation of superantigen-activated KIR3DL2⁺-expressing CD4 T cells, which can produce IL-17. KIR3DL2⁺ CD4 T cells are expanded, enriched for IL-17 production, and account for the majority of CD4 T cell IL-23R expression in the blood of patients with AS/SpA. These cells are further enriched in SpA synovial fluid and constitute a therapeutic target in SpA.

Fifteen milliliters heparinized venous blood was obtained from 25 patients with AS fulfilling the modified New York criteria. Supplemental Table I shows patient demographics and medication. None had received anti-TNF therapy. Paired knee synovial fluid samples were obtained from three AS and three acute ReA patients. Of 28 SpA patients, 22 were male and 27 were HLA-B*2705+ve by molecular typing. Seventeen healthy HLA-B27⁻ and six HLA-B*2705⁺ healthy control subjects were studied. As inflammatory disease control patients, we studied 10 patients with RA and 10 patients with ulcerative colitis (UC) without SpA. Appropriate ethical permission (Centre of Research Ethical Campaign 06/Q1606/139 and Oxfordshire Research Ethics Committee B 07/Q1605/35) and individual informed consent were obtained.

PBMC and synovial fluid mononuclear cells (SFMC) were separated by density gradient centrifugation. Cells were stimulated with PMA and ionomycin for 6 h or overnight with anti-CD3 and anti-CD28 beads (Miltenyi Biotec), staphylococcal enterotoxin B (SEB), or LPS (100 ng/ml) with brefeldin A (1 μg/ml; all Sigma). Cells were infected with Salmonella typhimurium araA^−/− (multiplicity of infection, 10 bacteria/cell) for 1 h in antibiotic-free medium, before culture overnight at 37°C, 5% CO₂ in RPMI 1640 containing 10% FCS, antibiotics (penicillin, streptomycin [200 mM] and gentamicin sulfate [50 μg/ml]) and l-glutamine (2.5 mM). Cells were stained for surface markers (anti-IgG2a, anti-DX31, anti-CD3 PerCP; BD Biosciences), anti-CD4 Pacific Blue (PB)/FITC (BioLegend), anti-CD14 PB, anti-CD56 or anti-CD161 PE (Beckman Coulter), anti–IL-23R (FITC; FAB14001F; R&D Systems), and anti-CCR6 PE (BD Biosciences). KIR3DL2 mAb DX31 was a kind gift of Jo Phillips (DNAX). CD4 T cells were stained for expression of “other KIR” using a mix of PE-conjugated CD158a, CD158b, and Z27 mAbs (BD Biosciences and Beckman Coulter). Dead cells, monocytes, and B cells were excluded by staining with PB live-dead stain (Invitrogen), anti-CD14, and anti-CD19 (BioLegend). Intracellular cytokine staining was performed using standard protocols (BD Biosciences) and anti–IL-17A (allophycocyanin/PE; eBioscience), anti–IFN-γ, and anti–TNF-α (FITC; BD Biosciences). Analysis was done with FlowJo software (version 8.8.6). A total of 0.5 × 10⁶ events were acquired, except cell viability and proliferation experiments, when all cells were acquired. IL-17⁺ KIR3DL2⁺ and IL-17⁺ KIR3DL2⁻ CD4 T cell numbers/million cells were calculated after exclusion of doublets. The KIR3DL2⁺ CD4 T cell gate was set at 10× IgG2a background staining. ELISAs for IL-17 and IFN-γ were performed following the manufacturer's instructions (eBioscience).

We used GraphPad Prism version 4 software for statistical analysis. Based on the work of Shen et al. (10), a sample size of 26 patients and 18 B27⁻ control subjects was chosen to achieve an α value of 5% in comparisons of IL-17–producing KIR3DL2⁺ CD4 T cells. One-tailed unpaired t tests with Welch’s correction and ANOVAs with a Bonferroni posttest were performed where shown.

FACS-sorted PBMC and SFMC KIR3DL2⁺ CD4 T lines were maintained as described previously (21). Short-term 5- to 10-d FACS-sorted CD4 T cells or negatively selected CD4 T cell lines (Miltenyi Biotec) were maintained by stimulation with anti-CD3 and anti-CD28 beads in DMEM with 10% FCS, supplements, and 40 IU/ml rIL-2 (D10/rIL-2) with or without rIL-23 and rIL-1 (10 ng/ml of each cytokine; R&D Systems).

LBL.721.220 parental B lymphocyte-derived cell lines (220) transfected with B7, B27, B27 C67S, and B27 with human tapasin have been described previously (21, 24). CD4 T cells were labeled with CFSE following the manufacturer’s instructions (Invitrogen). A total of 100,000/500,0000 gamma-irradiated 220 APCs were incubated with T cells (at a 1:1 ratio) and 100 ng/ml SEB (Sigma, Poole, U.K.) in 100 μl D10/rIL-2. Day 3 supernatants were taken for IL-17 and IFN-γ ELISA. On day 6, cells were stimulated with PMA and ionomycin for 5 h before FACS staining.

We first asked whether superantigen-stimulated KIR3DL2⁺ Th17 T cells could be preferentially expanded in the presence of cells expressing HLA-B27 H chain homodimers (B27₂). We have shown previously that the LBL721.220 cells (hereafter abbreviated to 220), transfected with B27, protect KIR3DL2⁺ NK cells from apoptosis (22). This cell line lacks functional tapasin and expresses high levels of B27₂, the levels being abrogated by mutation of the unpaired cysteine at position 67 to serine and reduced 2-fold if tapasin is overexpressed by cotransfection (21). CFSE-labeled CD4 T cells from AS patients and control subjects were cultured with SEB and equal numbers of 220 cells transfected with B27 or control HLA for 5 d. We compared the proliferation of viable CD4 T cells expressing KIR3DL2 with KIR3DL2−ve CD4 T cells and with CD4 T cells expressing “other” KIR (stained for expression with a mix of mAbs recognizing KIR3DL1/DS1, KIR2DL1/DS1, and KIR2DL2/3/2DS2; Fig. 1A). The greatest proliferation was observed in the KIR3DL2⁺ population stimulated with SEB and B27₂⁺ expressing 220B27. Less proliferation of KIR3DL2⁺ cells was seen when CD4 T cells were stimulated with 220 or control 220 transfectants lacking B27₂ (220B27C67S), expressing reduced levels of B27₂ (220B27 transfected with human tapasin) or expressing other HLA class I (220B7). The percentage of viable CD4 T cells expressing KIR3DL2 was consistently greater when CD4 T cells from patients with AS were stimulated with SEB and 220B27 (Fig. 1B). KIR3DL2⁻ CD4 T cells and CD4 T cells expressing “other KIR” did not show the same stimulation of proliferation with 220B27 cells (Fig. 1A and data not shown). The same effect was observed with FACS-purified KIR3DL2⁺ CD4 T cell lines and was inhibited with the KIR3DL2-specific mAb DX31, showing direct involvement of KIR3DL2 (Fig. 1C). HC10, an Ab to HLA class I H chains that also recognizes B27₂, also inhibited cell survival (data not shown). LBL.721.221 cells expressing tapasin, transfected with HLA-B27, also stimulated proliferation of KIR3DL2⁺ CD4 T cells more than LBL721.221 transfected with control HLA (data not shown). LBL721.221 HLA-B27 transfectants also express cell surface B27₂ (25).

Fig. 2A shows that bulk CD4 T cells from AS patients, stimulated with SEB presented by 220B27, produced more IL-17 than cells stimulated with control APCs with reduced or no B27₂ expression. Fig. 2B shows that KIR3DL2⁺ CD4 T produce more IL-17 under these stimulation conditions. Fig. 2C shows that, for a KIR3DL2⁺ CD4 T cell line, both cell survival (total number) and numbers of cells producing IL-17 on stimulation can be inhibited by DX31 mAb.

We have previously described increased numbers of KIR3DL2⁺-expressing CD4 T cells in the peripheral blood of B27⁺ SpA patients, compared with healthy control subjects and RA patients (22). We asked whether these cells were capable of producing IL-17 ex vivo. We first studied 26 SpA patients (24 with AS, 2 with ReA, 25 B27⁺), together with 6 B27⁺ and 18 B27⁻ healthy control subjects, and 10 RA patients and 10 UC patients without SpA as inflammatory disease control patients. The patient demographics are shown in Supplemental Table I. Fig. 3A shows enumeration by intracellular cytokine staining and FACS of IL-17–producing CD4 T (Th17) cells from the peripheral blood of a representative AS patient and a B27⁻ healthy control subject. As compared with only 1.54% of KIR3DL2⁻ CD4 T cells, 3.31% of AS KIR3DL2⁺ CD4 T cells produced IL-17 after stimulation with PMA and ionomycin ex vivo. For the healthy control subject, 1.09% of KIR3DL2⁺ and 0.45% of KIR3DL2⁻ CD4 T cells produced IL-17. Because 16.5% of patient CD4 T cells expressed KIR3DL2, this corresponds to 30% of all the Th17 cells in this particular AS patient. By contrast, for this particular healthy control subject, only 6.4% of IL-17–producing CD4 T cells expressed KIR3DL2. Fig. 3B shows that KIR3DL2⁺ CD4 T cells contain a significantly greater proportion of IL-17–producing cells for both SpA patients and, to a lesser extent, healthy control subjects. We observed no difference in IFN-γ production by KIR3DL2⁺ CD4 T cells between SpA patients and healthy control subjects (Fig. 3B). By contrast, with IL-17, KIR3DL2⁺ CD4 T cells from SpA patients were not enriched for TNF-α production compared with KIR3DL2⁻ CD4 T cells (data not shown).

Fig. 3C shows that the absolute number of peripheral blood KIR3DL2⁺ Th17 cells was also increased in SpA patients compared with healthy B27⁻ control subjects, and RA and UC inflammatory disease control patients. We also studied six B27⁺ healthy control subjects. Both the absolute numbers (Fig. 3C) and percentages of KIR3DL2⁺ Th17 were increased compared with healthy B27⁻ individuals, although they were lower than the SpA patients. Interestingly, healthy B27+ve control CD4 T cells also expressed more KIR3DL2 than B27−ve control subjects (mean values ± SD; B27⁺ control subjects: 11.2 ± 2.3% compared with 7.9 ± 3.2% on B27⁻ control subjects; p = 0.01; SpA patients: 15.12 ± 10%; p = 0.0018 compared with B27⁻ control subjects). Fig. 3D shows that the proportion of Th17 cells expressing KIR3DL2 was increased both in SpA patients and healthy B27⁺ control subjects compared with healthy B27⁻, RA, and UC disease controls. Thus, a mean of 36% of all Th17 cells in SpA patients expressed KIR3DL2 compared with 12.7, 16, and 17.4% in healthy B27⁻, RA, and UC controls, respectively. Strikingly, although KIR3DL2⁺ cells comprised the minority of CD4 T cells, the observed expansion within the KIR3DL2 population accounted for 70% of the total difference in Th17 numbers between SpA patients and control subjects. Consistent with two previous studies (9, 10), the absolute number of Th17 cells in the bulk CD4 T cell population was also increased in SpA patients compared with healthy B27⁻ control subjects, but to a lesser extent than the absolute number of Th17 cells in the KIR3DL2⁺ CD4 T cell population (Supplemental Fig. 1).

We next determined whether IL-17–producing KIR3DL2⁺ CD4 T cells expressed phenotypic markers consistent with the Th17 phenotype. Fig. 4A shows a representative FACS stain from an AS patient of IL-23R and CCR6 expression by KIR3DL2⁺ and KIR3DL2⁻ CD4 T cells. In this patient, 72% of KIR3DL2⁺ CD4 T cells coexpressed IL-23R and CCR6 compared with 5% of KIR3DL2⁻ CD4 T. Overall, a mean of 32% of KIR3DL2⁺ CD4 T cells expressed IL-23R, compared with only 3.1% of KIR3DL2−ve CD4 T cells in the SpA patients studied (Fig. 4B). Although KIR3DL2⁺ CD4 T cells comprised a minority of peripheral blood CD4 T cells, they contained a mean of 57% of the total number of IL-23R⁺ CD4 T cells in SpA patients. By contrast, a lower proportion of KIR3DL2⁺ CD4 T cells from B27⁻ control subjects expressed IL-23R (15%; Fig. 4B). There was no difference in the proportion of KIR3DL2⁻ CD4 T cells expressing IL-23R between SpA patients and healthy control subjects (Fig. 4B). In addition to IL-23R, KIR3DL2⁺ CD4 T cells were enriched for expression of CCR6 and CD161 Th17 phenotypic markers (Fig. 4C and data not shown). By contrast with IL-23R and CCR6, we observed no difference in the proportion of KIR3DL2⁺ CD4 T cells expressing CD161 between SpA patients and healthy controls (data not shown).

We sought to investigate whether KIR3DL2⁺ cells produced more IL-17 than KIR3DL2⁻ CD4 T cells in response to TCR ligation and whether KIR3DL2⁺ production of IL-17 was promoted by the Th17 cytokines IL-23 and IL-1. We first studied IL-17 secretion by equivalent numbers of FACS-sorted KIR3DL2⁺ and KIR3DL2⁻ CD4 T cells stimulated with Abs against CD3 and CD28 for 5 d by ELISA. Fig. 4D shows that, with this stimulus, KIR3DL2⁺ CD4 T cells from SpA patients produce 4-fold more IL-17 than KIR3DL2⁻ CD4 T cells or KIR3DL2⁺ CD4 T cells from healthy B27⁻ control subjects (mean ± SD; SpA KIR3DL2⁺ CD4 T: 344 ± 221 versus 80.63 ± 81.7 pg/ml/50,000 cells for KIR3DL2⁻; KIR3DL2⁺CD4 T cells from healthy control subjects: 86.4 ± 32.5 pg/ml/50,000 cells, n = 7). By contrast with the KIR3DL2⁺ population, there was no difference in production of IL-17 between KIR3DL2⁻ CD4 T cells from AS patients and healthy control subjects. We also observed no difference in production of TNF-α and IFN-γ by this subset between patients and control subjects (data not shown). Thus, although KIR3DL2⁺ cells constituted a minority of CD4 T cells, IL-17 production by this subset accounted for the majority of the increase in production of this cytokine in SpA patients (Fig. 4D, Supplemental Fig. 2A, 2B). FACS-sorted KIR3DL2⁺ CD4 T cells from AS patients secreted up to 58% of all CD4 T cell IL-17 (Supplemental Fig. 2A). IL-17 production was further enhanced by addition of the Th17-promoting cytokines IL-23 and IL-1, and this effect was more pronounced (>5-fold increase) for KIR3DL2⁺ cells from SpA patients (Fig. 4D).

We next asked whether KIR3DL2⁺ IL-17–producing CD4 T cells were capable of producing additional cytokines. Fig. 5A shows that ex vivo-stimulated KIR3DL2⁺ Th17 from the peripheral blood of an AS patient also produce TNF-α and IFN-γ, in addition to IL-17. The majority of IL-17−producing KIR3DL2⁺ CD4 T cells also produced TNF-α. KIR3DL2⁺ Th17 were enriched for production of IFN-γ compared with KIR3DL2⁻ Th17 (Fig. 5A, 5B). Increased proportions of KIR3DL2⁺ CD4 T cells staining for both IL-17 and IFN-γ production were found in SpA patients compared with healthy B27⁻ and disease controls (Fig. 5B and data not shown). Fig. 5C shows a representative FACS stain for IL-17, TNF-α, and IFN-γ production by a KIR3DL2⁺CD4 T cell line from a B27⁺ ReA patient. Most IL-17–producing cells from FACS-sorted KIR3DL2⁺CD4 T cells from this and other SpA patients also produced TNF-α, with some also producing IFN-γ. KIR3DL2⁺ CD4⁺ T cells also produced IL-17 after stimuli known to induce SpA, such as infection with Salmonella typhimurium or stimulation with anti-CD3 or superantigen (Supplemental Fig. 3).

Because the hallmark of SpA is inflammation of entheses and synovial joints, we next asked whether the joints of SpA patients were enriched for KIR3DL2⁺ Th17 cells. Fig. 6A shows that synovial fluid KIR3DL2⁺ CD4 T cells from the inflamed knee of a patient with AS produce IL-17. Ninety-five percent of IL-17–producing KIR3DL2⁺ CD4 T cells also produced TNF-α, and 60% produced IFN-γ.

Fig. 6B shows data from paired peripheral blood and knee synovial fluid samples from the above and two further AS patients, together with three B27⁺ patients with acute ReA after Salmonella sp. or Chlamydia sp. infection. A total of 15.5 ± 4.4% of synovial fluid CD4 T cells expressed KIR3DL2, compared with 8.4 ± 1.4% of peripheral blood CD4 T from matched samples from these patients. In all but one case, SFMC were further enriched for IL-17 production and CD161 expression compared with peripheral blood, and all patients’ SFMC KIR3DL2⁺ CD4 T cells showed increased CCR6 and IL-23R expression. Synovial fluid KIR3DL2⁺ CD4 T cells also frequently produced more TNF-α and IFN-γ than matched peripheral blood KIR3DL2⁺ CD4 T cells (Fig. 6B). Increased proportions of KIR3DL2⁺ CD4 T cells producing both IL-17 and IFN-γ were found in the synovial fluid of patients compared with peripheral blood. Fig. 6C shows that FACS-sorted KIR3DL2⁺ CD4 T cells from SFMC produced more IL-17 than KIR3DL2⁻ CD4 T cells, when stimulated with anti-CD3 and anti-CD28 for 5 d.

The spondyloarthritides, of which AS is the most common, are a group of inflammatory arthritides strongly associated with possession of HLA-B27. In this article, we show that CD4 T cells expressing the killer cell Ig-like receptor KIR3DL2⁺ specifically survive, proliferate, and produce IL-17 on stimulation with cells expressing B27 dimers (termed B27₂). We also show that the KIR3DL2⁺ CD4 T cell subset is expanded and enriched for IL-17 production in the peripheral blood and synovial fluid of patients with SpA. Thus, IL-17–producing KIR3DL2⁺ CD4 T cells accounted for 70% of the observed increase in Th17 numbers in SpA subjects (compared with control subjects), despite this subset comprising a mean of only 15% of CD4 T cells in the peripheral blood of SpA patients. Furthermore, KIR3DL2⁺ from AS patients secreted more IL-17 than KIR3DL2⁻ lines and KIR3DL2⁺ CD4 T cells from healthy control subjects ex vivo after TCR stimulation.

An important role for Th17 cells in the pathogenesis of AS has recently been suggested by the strong genetic linkage with IL-23R polymorphisms (3), and by the findings of increased IL-17 levels and Th17 cell numbers in AS patients (7, 9, 10). IL-23R is typically, although not exclusively, expressed by Th17 cells. Strikingly, we show in this article that the KIR3DL2⁺ CD4 T population accounts for the majority of peripheral blood CD4 T cell IL-23R expression in SpA patients. We also show that IL-23 (together with IL-1) stimulates KIR3DL2⁺ CD4 T cell IL-17 production. Increased IL-23 expression has been demonstrated in the terminal ileum of AS patients (26), and IL-23 has been shown to be critical for maintenance of pathogenic Th17 cells producing IFN-γ in a murine colitis model (27). IL-17 acts on a variety of cell types to induce secretion of multiple proinflammatory cytokines, chemokines, and prostanoids (15–17). Our data support the concept that IL-17–producing T cells contribute to the pathogenesis of AS.

The B27-responsive IL-17–producing CD4 T cells described in this article carry Th17 phenotypic markers including CCR6, IL-23R, and CD161, but also commonly produce TNF-α and/or IFN-γ (furthermore, production of all three cytokines is increased in response to IL-23). There is now mounting evidence not only that such polyfunctional cells are common and increased in SpA, but that cells with an overlapping Th17 and Th1 phenotype may have increased pathogenicity (27, 28). Interestingly, the chemokine CCL20, a potent chemotactic factor for CCR6⁺ T cells, is increased in SpA (and RA) synovial fluid (29). Forty-five percent of KIR3DL2⁺T cells in SpA patients in our study are CCR6⁺. CCL20 has been shown to be secreted by human articular chondrocytes on dynamic compression (30). This may be of relevance given the predilection of inflammation in SpA for the spine and entheses, areas of high mechanical stress.

We have previously shown that B27 can form H chain dimers (B27₂) (31). Unlike classical HLA-B27, B27₂ bind the KIR KIR3DL2 (20, 22, 32). In this article, we show that B27₂-expressing APCs stimulate the survival, proliferation, and IL-17 production of KIR3DL2⁺ CD4 T cells. We provide evidence that this effect is mediated directly through the KIR3DL2/ B27₂ interaction, because the effect is reduced by reducing B27₂ expression and inhibited by a KIR3DL2-specific mAb. The molecular basis requires further elucidation, although we show that cell death by apoptosis is reduced and cell proliferation is increased. We have observed similar antiapoptotic effects of KIR3DL2 ligation by B27₂ on the survival of NK cells and of CD4 T cells that do not make IL-17 (22 and unpublished observations), so it is likely that B27₂-KIR interactions primarily exert their effect by promoting leukocyte survival rather than other effects such as effects on T cell polarization.

In this study, we also show that some of these effects are also found in healthy B27 individuals. Thus, we found a modest expansion of total KIR3DL2⁺ and of KIR3DL2⁺ Th17 numbers in healthy HLA-B27⁺ individuals, without clear evidence of increased cytokine production. To our knowledge, these data show for the first time evidence of an immunological phenotype in healthy HLA-B27⁺ individuals, and further support a direct pathogenic role for HLA-B27. We propose that this increase in KIR3DL2⁺CD4 T cells, potentially skewed toward IL-17 production, may predispose to SpA.

We show in this study that AS patients have increased KIR3DL2⁺ Th17 cells, confirming previous findings of increased numbers of circulating KIR3DL2⁺ CD4 T cells (22), and increased Th17 cells (9, 10), and for the first time, to our knowledge, link both these findings to each other. Expression of HLA-B27 likely explains both findings, because we have shown that KIR3DL2⁺ IL-17–secreting CD4 T cells selectively proliferate on coculture with B27₂-expressing cells. Our data are consistent with the hypothesis that Th17 cells play a significant role in AS pathogenesis. The mAb ustekinumab against the p40 subunit of IL-12 and IL-23 has recently been shown to be effective in the treatment of a number of inflammatory diseases including psoriatic arthritis (33). Our study provides scientific rationale for treatment trials of anti-Th17 therapy in AS.

We thank the Thames Valley Comprehensive Local Research Network and Claire Farrar for patient recruitment, Jo Phillips for KIR3DL2 (DX31) mAb, and Prof. Brigitte Askonas and Sarah Rowland Jones for comments.

The authors have no financial conflicts of interest.