In autoimmune diseases such as rheumatoid arthritis (RA), regulatory T cells (Tregs) fail to constrain autoimmune inflammation; however, the reasons for this are unclear. We investigated T cell regulation in the RA joint. Tregs from RA synovial fluid suppressed autologous responder T cells; however, when compared with Tregs from healthy control peripheral blood, they were significantly less suppressive. Despite their reduced suppressive activity, Tregs in the RA joint were highly proliferative and expressed FOXP3, CD39, and CTLA-4, which are markers of functional Tregs. This suggested that the reduced suppression is due to resistance of RA synovial fluid responder T cells to Treg inhibition. CD161+ Th17 lineage cells were significantly enriched in the RA joint; we therefore investigated their relative susceptibility to Treg-mediated suppression. Peripheral blood CD161+ Th cells from healthy controls were significantly more resistant to Treg-mediated suppression, when compared with CD161- Th cells, and this was mediated through a STAT3-dependant mechanism. Furthermore, depletion of CD161+ Th cells from the responder T cell population in RA synovial fluid restored Treg-mediated suppression. In addition, CD161+ Th cells exhibited pathogenic features, including polyfunctional proinflammatory cytokine production, an ability to activate synovial fibroblasts, and to survive and persist in the inflamed and hypoxic joint. Because CD161+ Th cells are known to be enriched at sites of autoinflammation, our finding that they are highly proinflammatory and resistant to Treg-mediated suppression suggests an important pathogenic role in RA and other autoimmune diseases.
Regulatory T cells (Tregs) represent an important peripheral tolerance mechanism that serves to prevent autoimmunity. However, in autoimmune diseases such as rheumatoid arthritis (RA), there is clearly a failure of Tregs to constrain autoimmune inflammation. A deficit in overall Treg numbers or a defect in Treg function are both plausible contributors to the chronic inflammation observed in the RA joint. Several studies examining Treg phenotype and function in RA have drawn incongruent conclusions. Conflicting data may in part be the result of some studies examining cells taken from the peripheral blood of RA patients, whereas others examined cells from the site of inflammation in the RA synovial fluid (SF). In addition, differences in isolation strategies for purifying Tregs may account for discrepancies between studies. Studies examining the peripheral blood of RA patients have reported a decrease in the frequency of Tregs compared with healthy control (HCs) (1–3); however, studies that extracted lymphocytes from the SF of RA patients have reported enhanced frequencies of Tregs (3–5).
Inflammation persists in RA despite high frequencies of Tregs in the joint, and an explanation as to why Tregs fail to constrain or resolve the chronic inflammatory response remains elusive. Suppression assays performed by two independent groups illustrated that Tregs from the SF of RA patients suppressed effector cell proliferation and reduced the concentrations of TNF and IFN-γ produced in the supernatants (4, 5), suggesting that Tregs in the synovial joint are in fact suppressive yet fail to constrain inflammation. On the other hand, Tregs from the peripheral blood of RA patients have been shown to be functionally impaired (6–9).
The RA joint is a complex environment where immune cells, fibroblasts and the hypoxic environment collaborate to result in inflammation and joint destruction. T cells are thought to play a key role in RA pathogenesis (10–12) with both experimental models and clinical research focusing on the role of Th17 cells, which produce IL-17 as their signature cytokine (13–17). Indeed, Th17 cells have recently been targeted by the use of anti–IL-17 in RA, although with somewhat disappointing results (18). It has emerged that T cell subsets can exhibit functional plasticity, particularly at inflamed sites, making it difficult to identify discrete T cell subsets on the basis of their cytokine production (19–21). Th17 cells appear to be particularly unstable under inflammatory conditions and can become so-called ex-Th17 cells or nonclassical Th1 cells, which start to produce IFN-γ with the eventual loss of IL-17 expression (22–27). CD161, a lectin-like receptor, is a marker of human Th17 type cells (28), which is retained even after they switch to an ex-Th17 cell that coproduces IL-17 and IFN-γ, or IFN-γ alone (29). This marker can therefore be used to track Th17 plasticity and to identify Th17 lineage cells. CD161+ Th cells have been shown to be enriched in a variety of human autoimmune diseases, including RA (30), juvenile idiopathic arthritis (JIA) (31), inflammatory bowel disease (28, 32), and multiple sclerosis (33, 34), suggesting that they have a pathogenic role, although the reasons for their accumulation and pathogenicity are still not clear. Synergy between different cytokines derived from Th17-type cells is likely to be mechanistically important for Th17-mediated pathogenic effector functions in RA, or autoimmunity in general (35–37). Given the key pathogenic role of CD161+ Th17 lineage cells in autoimmune disease, it is important to understand how these cells are regulated.
In this study, we investigated the regulation of effector T cells taken from the RA SF by Tregs. We found that Tregs from RA SF exerted suppressive function, but to a reduced extent compared with peripheral blood Tregs from HCs. We were able to attribute this reduced suppression to the fact the CD161+ Th cells, which were enriched in the RA joint and exhibited a variety of pathogenic features, were resistant to Treg suppression.
Materials and Methods
All patients fulfilled the American College of Rheumatology 2010 (38) criteria for a diagnosis of RA and had active disease at time of sampling (as classified by disease activity score in 28 joints [DAS28] scores or erythrocyte sedimentation rate and C-reactive protein). Fully informed, written consent was obtained from each patient. This study was approved by the St. Vincent’s University Hospital Ethics and Medical Research Committee. Peripheral blood and SF mononuclear cells (SFMC) were isolated by density gradient centrifugation (Lymphoprep; Axis-Shield poC). Twenty-nine patients were female and 19 were male, with a median age of 59 y (range, 19–83 y). These patients had a disease activity score (DAS28) of 4.2 ± 0.26 with 22.4% naive to treatment, 65.3% on synthetic disease-modifying antirheumatic drugs (DMARDs; e.g., methotrexate, sulfasalazine) and 40.08% on biologic DMARDs. No significant difference in DAS28 was observed between patients on no treatment compared with DMARD or biologic. Synovial tissue biopsies with paired PBMCs were obtained at arthroscopy using local anesthetic and a Wolf 2.7-mm telescope (RWolf) from four patients (three female and one male; median age, 50 y; range, 45–60 y). These patients had a DAS28 of 5.75 ± 0.89; one patient was naive to therapy and three patients were taking methotrexate. Lymphocytes were isolated from biopsy specimens by gentle mechanical and enzymatic digestion on a GentleMACS dissociator (Miltenyi Biotech) using a tumor dissociation kit (Miltenyi Biotech), according to the manufacturer’s protocol.
Lymphocytes were identified by forward and side scatter, and dead cells and doublets were excluded. Cells were stained extracellularly with amine-binding markers for dead cells (viability dye; eBioscience) and fluorochrome conjugated Abs specific for CD4 (eBioscience), CD3 (BD Biosciences), CD8 (eBioscience), and CD45RO (Miltenyi Biotec) or CD45RA (eBioscience), CD127 (eBioscience), CD25 (eBioscience), CD39 (eBioscience), and CD161 (eBioscience and BioLegend). For FOXP3, CTLA-4, Ki67, Bcl2 (eBioscience), and HIF-1α (R&D Systems) analysis, cells were surface stained ex vivo and then fixed and permeabilized for intranuclear staining (FOXP3 staining buffer kit; eBioscience). For intracellular cytokine analysis, cells were stimulated for 5 h with PMA (50 ng/ml) and ionomycin (500 ng/ml) in the presence of brefeldin A (5 μg/ml; Sigma-Aldrich). Cells were then surface stained, fixed, and permeabilized using Caltag Fix and Perm Kit (Biosciences) before being stained intracellularly for cytokines: GM-CSF (BioLegend), IFN-γ and IL-2 (BD Biosciences), IL-17 (eBioscience and BD Biosciences), IL-4, and TNF (eBioscience).
The CD4+ T cell population in PMA/ionomycin–stimulated samples was gated on the basis of CD3+CD8− cells because CD4 was downregulated upon stimulation. Cells were acquired on a BD FACSCanto II or LSRFortessa (BD Biosciences) and analyzed with FlowJo software with gating set on unstimulated control samples or fluorescence minus one controls, as appropriate. Presentation of polyfunctional cytokine production was performed using SPICE version 5.3 software (39). For phospho-STAT staining, CD4+CD45RO+CD25−CD161+ and CD4+CD45RO+CD25−CD161− Th cells were sorted from HC PBMCs as described below. Cells were fixed, permeabilized, and stained for pSTAT3 according to the manufacturer’s protocol (BD Biosciences).
Assessment of telomere length by flow fluorescence in situ hybridization
HC PBMCs and RA SFMCs were stained with fluorochrome-conjugated Abs specific for CD45RA, CD161, CD4, and CD25 (eBioscience). CD4+CD45RA−, CD4+CD45RA−CD25−CD161+, and CD4+CD45RA−CD25−CD161− subpopulations were sorted on a MoFlo Legacy cell sorter (Beckman Coulter). Equalized numbers of cells were divided into triplicate tubes and combined with 2 × 105 CFSE-labeled murine thymocytes to act as an internal control. Cells were fixed in 4% PFA (Santa Cruz Biotech), washed in hybridization buffer containing 70% formamide (Sigma-Aldrich), and stained with the PNA Cy5 telomere probe (Cambridge Research Biochemicals) at a final concentration of 750 ng/ml, according to the manufacturer’s protocol. Cells were immediately acquired on a FACSCanto II flow cytometer. The relative telomere lengths of CD161+ and CD161− subpopulations were expressed as a percentage of the total CD4+CD45RA− memory population (control).
Treg activation experiments
The frequency and phenotype of Tregs from HC PBMCs were assessed on day 0 using fluorochrome-conjugated Abs specific for CD4, CD25, CD127, CD39, CTLA-4, and FOXP3 and analyzed by flow cytometry. To determine the effect of Treg activation on phenotype, PBMCs were treated with SF from RA patients with active disease, anti-CD3, SF, and anti-CD3 or left untreated for 6 d. Treg frequency and phenotype were reassessed by flow cytometry, as on day 0.
CD4+CD25+CD127loCD39+ Tregs (which were routinely >90% FOXP3+) and Treg-depleted CD4+CD4+CD25− responder cells were sorted from HC PBMCs and RA SFMCs on a MoFlo Legacy (Beckman Coulter) or FACSAria Fusion (BD Biosciences) cell sorter, with purities routinely >98%. For analysis of suppression in CD161+ and CD161− responder cell subpopulations, leukocyte-enriched buffy coats from anonymous healthy donors were obtained with permission from the Irish Blood Transfusion Board, St. James’s Hospital, Dublin. CD4+ T cells or CD4+CD45RO+ memory T cells were purified from PBMCs using magnetic cell sorting (Miltenyi Biotec). CD4+ T cells were stained with fluorochrome-conjugated Abs specific for CD39, CD25, CD4, and CD127 (eBioscience) to sort Treg and total responder cells. CD4+CD45RO+ cells were stained with fluorochrome-conjugated Abs specific for CD45RA, CD161, CD4, and CD25 (eBioscience) to sort CD4+CD45RA−CD25−CD161+ and CD4+CD45RA−CD25−CD161− subpopulations. To purify CD161-depleted responder cells (CD4+CD25−CD161−), SFMCs from RA patients were stained with fluorochrome-conjugated Abs specific for CD161, CD4, and CD25 (eBioscience). Cells were purified using a MoFlo Legacy (Beckman Coulter) or FACSAria Fusion (BD Biosciences) cell sorter, with purities routinely >98%. Sorted responder cells were labeled with cell trace violet (CTV; Life Technologies), according to the manufacturer’s protocol, and cultured at 5 × 104 cells/well alone or in coculture with equal numbers of unlabeled Tregs in the presence of anti-CD3 (1 μg/ml; eBioscience) and 3 × 105 CFSE-labeled irradiated (60 Gy) allogeneic PBMCs to serve as APCs.
To investigate the resistance of suppression of CD161+ responder cells, various factors were added to suppression assays. To determine the role of soluble factors on suppression, supernatants were harvested and pooled on day 5 from triplicate wells of CD161+ or CD161− responder cells stimulated with irradiated APCs and anti-CD3. Supernatants were diluted 1 in 4 and added into suppression assays. In addition, sorted CD161+ and CD161– Th cell subpopulations were treated with recombinant human IL-10 or TGF-β at the indicated concentrations and cultured with anti-CD3. To determine the effect of CD161, sorted CD161+ and CD161− Th subpopulations were preincubated with a CD161 blocking Ab or matched isotype control (BD Biosciences; clone DX12) before being cultured alone or cocultured with Tregs in the presence of irradiated APCs and anti-CD3. To inhibit STAT3 selectively in the responder cell population of the suppression assay, responder cells were pretreated with Stattic (0.47 μM; Sigma-Aldrich) for 18 h and washed before coculture with Tregs, irradiated APCs, and anti-CD3. Cells were maintained in humidified incubators at 37°C with 5% CO2 for 4–5 d. Proliferation of responder cells was measured by CTV dilution. Cells were restimulated with PMA and ionomycin in the presence of brefeldin A to assess cytokine production. CFSE-labeled irradiated APCs and unlabeled Tregs were excluded from the analysis. Percent suppression of proliferation (as measured by CTV dilution) was calculated as the difference in proliferation of responder cell cultured alone and cocultured with Tregs, over the frequency of proliferating responder cells cultured alone: Responder − (Responder + Treg)/(Responder) × 100. Percent suppression of cytokine production was assessed in a similar manner from proliferating cells only. In addition, the concentrations of cytokines present in the supernatants were quantified by ELISA (eBioscience Ready-Set-Go IL-17A, IFN-γ, GM-CSF, and TNF).
Fibroblast cell culture and assays
Primary RA fibroblasts were isolated from synovial tissue as described previously (40). A K4-immortalized synoviocyte fibroblast cell line was cultured and used between passages 30 and 40. CD4+CD25−CD45RA−CD161+ and CD4+CD25−CD45RA−CD161+ T cells were cell sorted from HC PBMCs. Cells were cultured for 5 d in the presence of anti-CD3 and irradiated APCs, after which cell culture supernatants were harvested. Fibroblast cells were rendered quiescent by omitting serum for a minimum of 8 h before stimulating with T cell supernatants (diluted 1:4) for 24 h. The concentrations of IL-8 and IL-6 were quantified by ELISA (eBioscience). No significant differences were found between the concentrations of IL-6 and IL-8 in the CD161+ and CD161− Th cell supernatants and the average concentrations added to fibroblast cultures were <500 pg/ml (data not shown). Cells were trypsinized, stained with fluorochrome-conjugated Abs specific for ICAM-1, and RANKL (eBioscience) and analyzed by flow cytometry.
Statistical analyses were performed using Prism 5 software; two groups within a sample were determined by Student paired t test with two-tailed p values; ≥3 groups were analyzed by one-way ANOVA with Tukey multiple comparison test. Statistical differences between two groups, each containing more than one variable, were determined by two-way ANOVA with Bonferroni posttests; p values < 0.05 were considered significant and denoted with asterisks in the figures.
Enrichment of Tregs in the RA joint is associated with their enhanced proliferation, survival, and a suppressive phenotype
The unresolved inflammation that persists in the RA joint, together with the demonstration that Tregs are not depleted in RA patients (4, 5), indicates that Tregs are unable to constrain the inflammation; however, the reasons for this are still unclear. We therefore analyzed the phenotype and function of Tregs in RA. We first examined the frequency of Tregs from RA SF and peripheral blood from either RA patients or HCs using multiparameter flow cytometry, and Tregs were identified as being CD4+CD127loCD25+FOXP3+ (Supplemental Fig. 1A). We observed a significant increase in the frequency of Tregs within RA SF compared with PBMCs from either RA patients (p < 0.001) or HCs (p < 0.001; Fig. 1A, dot plots shown in Supplemental Fig. 1B). In addition, we found higher FOXP3 expression in Tregs from RA SF compared with matched peripheral Tregs (Supplemental Fig. 1B). There was no significant difference in the frequency of Tregs in the peripheral blood of HC and RA patients (Fig. 1A). Enrichment of Tregs in the RA joint could be explained by increased recruitment, survival, or local proliferation. To investigate these possibilities, we examined the proliferation of Tregs by analyzing expression of the cell cycle molecule Ki67. The frequency of Ki67+ (proliferating) Tregs was significantly higher in RA SF compared with HC PBMCs (p < 0.001) or RA patients (p < 0.01; Fig. 1B, dot plots illustrating Ki67 staining shown in Supplemental Fig. 1C). Interestingly, within the RA SF, Tregs were also more highly proliferative than non-Treg responder cells were (p < 0.001; Fig. 1B). Because Tregs from peripheral blood have previously been shown to be prone to apoptosis (41), we next investigated the propensity of Tregs to survive in the environment of an inflamed and hypoxic joint. We found that in HC PBMCs, the levels of the antiapoptotic marker Bcl-2 were significantly lower in Tregs compared with responder T cells (p < 0.05; Fig. 1C). In contrast, there was no significant difference in Bcl-2 expression between Treg and responder T cells within RA SF, indicating that Tregs from the RA joint were not at a survival disadvantage as in peripheral blood (Fig. 1C with representative histograms in Supplemental Fig. 1D). Because the RA joint is known to be profoundly hypoxic, we examined expression of HIF-1α, a transcription factor that plays a key role in mediating the cellular response to hypoxia. Interestingly, we found that expression of HIF-1α was greater in Tregs from RA SF compared with those from HC peripheral blood (p < 0.05; Fig. 1D). We next examined the expression of CD39 and CTLA-4 on Tregs, because both of these can mediate suppressive function. There was a significant increase in the expression of the ectonuclease CD39 on Tregs from RA SF compared with PBMCs from RA patients (p < 0.001) or HCs (p < 0.05; Fig. 1E, representative dot plots shown in Supplemental Fig. 1E). CTLA-4 expression was also significantly higher in Tregs from SF compared with PBMCs from RA patients (p < 0.001) or HCs (p < 0.001; Fig. 1F with representative dot plots in Supplemental Fig. 1F). To mimic the effect of the joint environment, we activated T cells from HCs for 6 d with anti-CD3 in the presence of RA SF. The frequency of Tregs increased significantly (Supplemental Fig. 2A), as did their expression of CTLA-4 (Supplemental Fig. 2B), although there was no significant change in the frequency of CD39 expression (Supplemental Fig. 2C). This finding suggested that the phenotype observed in RA SF Tregs could be due in part to activation within an inflammatory milieu. These findings demonstrate that Tregs within RA SF were enriched, highly proliferative ex vivo, and exhibited prosurvival features. In addition, the high levels of expression of CD39 and CTLA-4 within Tregs from the RA joint may be indicative of good suppressive function.
Tregs from RA SF exhibited a reduced ability to suppress responder cell proliferation and cytokine production
Having demonstrated that Tregs from the RA joint exhibited a suppressive phenotype, we next investigated their ability to suppress autologous responder T cells. We used a flow cytometric assay to distinguish between Treg and responder T cell proliferation and cytokine production. CD4+CD25− responder cells and CD4+CD127loCD25+CD39+ Tregs were sorted from RA SF and HC peripheral blood. CD39 was included as an additional marker so that Tregs from SF and peripheral blood, which contained different frequencies of CD39+ Tregs, could be compared directly. Responder T cells were labeled with the cell tracer CTV and stimulated with anti-CD3 and irradiated APCs in the presence or absence of autologous Tregs at a ratio of 1:1. After 5 d, cells were restimulated with PMA and ionomycin, and stained for intracellular cytokines. Responder cell proliferation was measured by dilution of CTV. Tregs from HC peripheral blood suppressed autologous responder cell proliferation; however, Tregs from RA SF appeared to have a relatively reduced suppressive capacity (Fig. 2A, collated data in Fig. 2B). Furthermore, comparison of the percentage suppression by Tregs from RA SF versus HC peripheral blood revealed that Tregs from RA SF were significantly less capable of suppressing proliferation (p < 0.05; Fig. 2B).
Suppression of cytokine production by Tregs was measured by gating on cytokine+ proliferating cells. The vast majority of IFN-γ production was suppressed by HC Tregs, whereas as RA synovial Tregs suppressed IFN-γ to a lesser extent (Fig. 2C, collated data in Fig. 2D). Overall Tregs from RA SF were significantly less able to suppress IFN-γ compared with those from HC blood (Fig. 2D; p < 0.05). To validate the assay, suppression of IFN-γ was also measured in cell culture supernatants by ELISA. Consistent with the data from the flow cytometric assay, Tregs from HC peripheral blood and RA SF suppressed IFN-γ production; however, SF Tregs exhibited significantly less suppressive capacity compared with those from HC peripheral blood (p < 0.05; Fig. 2E). Tregs from HC blood also efficiently suppressed TNF (p < 0.001), GM-CSF (p < 0.05), and IL-17 (p < 0.01) production; however, Tregs from RA SF significantly suppressed autologous responder cell IL-17 (p < 0.05), but not TNF or GM-CSF (Fig. 2F–H). Furthermore, comparison of the percentage suppression by Tregs from RA SF versus HC peripheral blood revealed that Tregs from RA SF were significantly less capable of suppressing TNF (p < 0.01) and GM-CSF (p < 0.05), but not IL-17 (Fig. 2F–H). Collectively, these data demonstrate that Tregs from RA SF have significantly reduced suppressive activity compared with those from HC peripheral blood.
CD161+ Th cells are enriched in RA synovial tissue and fluid, exhibit increased expression of inflammatory cytokines, Bcl-2, HIF-1α, and longer telomeres than CD161- Th cells
Because Tregs from the RA joint displayed an activated and suppressive phenotype, we hypothesized that the reduced suppression observed may be due to resistance of the responder T cells to suppression by Tregs rather than impaired Treg function. We therefore investigated the characteristics of effector/responder CD4+ T cells within the RA joint. CD161 is a marker of Th17 lineage cells (28), which have been suggested to be important in the pathogenesis of RA. We examined the frequency of CD161+ Th cells in RA SF and peripheral blood from RA patients and HCs. We observed a strikingly higher frequency of CD161+ Th cells in RA SF peripheral blood from RA patients and HCs (Fig. 3A). To confirm the enrichment of CD161+ Th cells in SF, we isolated cells from synovial tissue biopsies and matched peripheral blood and examined the frequency of CD4+CD161+ T cells. To our knowledge, we demonstrate for the first time a significant increase in the frequency of CD4+CD161+ cells from synovial tissue compared with matched peripheral blood (Fig. 3A). We next compared the relative capacity of CD161+ versus CD161− Th cells from the RA joint to secrete cytokines. There was no significant difference in the frequency of IFN-γ+ cells between CD161+ and CD161− Th cells. The fact that a significant proportion of CD161+ Th17 lineage cells within the RA joint secreted IFN-γ indicated that these cells had undergone functional plasticity and switched to become ex-Th17 or nonclassical Th17 cells. However, CD161+ Th cells secreted significantly more TNF (p < 0.01), GM-CSF (p < 0.01), and IL-17 (p < 0.05) when compared with CD161− Th cells (Fig. 3B). These findings demonstrate that CD161+ Th17 lineage cells constitute a significant proportion of the Th cell pool within the RA joint. Although the majority of these CD161+ Th17 lineage cells no longer produced IL-17, they had increased production of multiple other proinflammatory cytokines, including IFN-γ, TNF, and GM-CSF, and are thus likely to have a pathogenic role in the RA joint.
We investigated the potential of CD161+ Th cells to survive and persist in the hypoxic RA joint. The expression of the prosurvival factor Bcl-2 was analyzed and compared between CD161+ and CD161− memory Th cells in HC peripheral blood and RA SF. We observed significantly higher expression of Bcl-2 in CD161+ compared with CD161− Th cells in both HC blood and RA SF (Fig. 3C). HIF-1α expression was significantly increased in CD161+ compared with CD161− Th cells derived from the hypoxic joint (Fig. 3D). By contrast, no difference in expression of HIF-1α was observed between CD161+ and CD161− Th cells taken from normoxic HC PBMCs (Fig. 3D). Next, telomere length was measured using flow fluorescence in situ hybridization to provide a relative measure of the replicative capacity of cells. CD161+ Th cells had significantly longer telomeres than CD161− Th cells did in HC peripheral blood and RA SF (Fig. 3E). In summary, the data indicate that CD161+ Th cells, which are enriched in the RA joint, have enhanced capacity to produce proinflammatory cytokines and increased expression of Bcl2 and HIF-1α in addition to enhanced replicative capacity, shown by relatively longer telomeres. To our knowledge, these data are the first to suggest that CD161+ Th cells may be markedly more pathogenic because of their increased production of inflammatory cytokines such as TNF, GM-CSF, and IL-17, which all have a prominent role in RA pathology. Moreover, CD161+ Th cells may persist for protracted periods during inflammation because of their prosurvival features, which include increased Bcl-2 and HIF-1α expression and relatively longer telomeres.
Polyfunctional cytokine production is a feature of CD161+ Th cells
Having shown that CD161+ T cells from RA SF exhibited increased cytokine production compared with CD161− Th cells, we next investigated the cytokine polyfunctionality of CD161+ and CD161− Th cells. HC (n = 20) peripheral blood cells were cultured for 3 d with anti-CD3 before restimulation for cytokine production. Cytokine polyfunctionality was assessed using multiparameter flow cytometry and SPICE software. The orange and red inner pie segments, which indicate cells producing three or more cytokines, are noticeably larger in the CD161+ Th helper cells compared with the CD161− Th cells, whereas the gray (one cytokine) and yellow segments (two cytokines) were relatively increased in the CD161− Th cells (Fig. 4A). In particular, the proportion of cells producing GM-CSF, IFN-γ, IL-2, and TNF (dominant red pie segment overarched by pink, green, blue, and purple) was relatively increased in the CD161+ Th cells. Fig. 4B shows the percentage of cells that produced one, two, three, four, five, or six cytokines. There were significantly more CD161− Th cells that produced only one or two cytokines (p < 0.001), whereas the proportion of polyfunctional cells that produced three, four, five, or six cytokines was significantly increased in the CD161+ Th cell population (Fig. 4B; p < 0.001). Significantly increased frequencies of CD161+ Th cells were GM-CSF+IFN-γ+IL-2+TNF+ and IFN-γ+IL-2+TNF+ compared with CD161− Th cells (Fig. 4C). Whereas the CD161− Th cell population contained significantly more bifunctional IL-2+TNF+ and monofunctional IL-2+ and IL-4+ cells compared with CD161+ Th cells. These data indicate that CD161+ Th cells, which are enriched in the RA joint, are inherently more polyfunctional.
Supernatants from CD161+ Th cells have an increased capacity to activate RA synovial fibroblasts
Synovial fibroblasts play a key role in perpetuating inflammation and mediating damage in the RA joint (42–47); therefore, we investigated the differential ability of CD161+ and CD161− Th subpopulations to induce a proinflammatory phenotype in synovial fibroblasts. Culture supernatants harvested from sorted CD161+ or CD161− Th cells from HC peripheral blood and stimulated in culture with anti-CD3 were added to the K4 synovial fibroblast cell line or primary RA synovial fibroblasts. Induction of ICAM-1 (an adhesion molecule that promotes trafficking of lymphocytes into the joint) and RANKL (which is responsible for bone degradation) on fibroblasts was measured by flow cytometry, and secretion of IL-6 and IL-8 was measured by ELISA after 24 h. Cell culture supernatants from CD161+ Th cells exhibited a significantly increased capacity to induce IL-8 (p < 0.05) and IL-6 (p < 0.05) secretion from the K4 fibroblast cell line (Fig. 5A), and similar results were observed using primary RA synovial fibroblasts (Fig. 5B). Significantly increased induction of ICAM-1 and RANKL expression was observed in primary RA synovial fibroblasts in response to addition of supernatants from cultured CD161+ compared with CD161− Th cells (Fig. 5C, 5D). Thus, soluble factors produced by CD161+ Th cells had increased capacity to induce proinflammatory cytokine production and adhesion molecule expression in RA synovial fibroblasts, and to promote bone remodeling. Taken together with earlier findings, these data suggest that CD161+ Th cells are likely to play a highly pathogenic role in the RA joint.
CD161+ Th cells from HCs are resistant to Treg-mediated suppression of proliferation and cytokine production
Having demonstrated that CD161+ Th17 lineage cells were enriched in the RA joint and that they exhibited a variety of potentially pathogenic characteristics, we next investigated whether they were susceptible to regulation by Tregs. CD4+CD45RO+ T cells from HCs were sorted into CD161+ and CD161− Treg-depleted responder subsets that were labeled with the tracer dye CTV, cocultured in the presence or absence of sorted CD4+CD127loCD25+CD39+ Tregs, and stimulated with anti-CD3 and irradiated APCs. Representative histograms show that CTV dilution (proliferation) within CD161+ and CD161− populations cultured alone was similar on day 5 (Fig. 6A). Although coculture of Tregs inhibited proliferation of CD161− Th responder cells, proliferation of CD161+ Th responder cells was relatively unaffected (Fig. 6A). Overall, highly significant resistance to suppression of proliferation by Tregs was observed in CD161+ Th cells (n = 15; Fig. 6B). Both responder cell populations had similar frequencies of TNF-producing cells, which were efficiently suppressed by Tregs in the case of CD161− responders but not CD161+ Th cells (Fig. 6C). Collated data demonstrated resistance to Treg-mediated suppression of IFN-γ (p < 0.01), TNF (p < 0.01), and GM-CSF (p < 0.05) production by CD161+ responder cells (Fig. 6D). Similar results were found when cytokines in the supernatants were measured by ELISA (Fig. 6E). Our findings demonstrate that CD161+ Th cells from HC peripheral blood were significantly less suppressed by Tregs than CD161− Th cells were. These findings suggest that CD161+ Th cells, which are known to be enriched in the RA joint and have been shown to have pathogenic and prosurvival features, might not be adequately constrained by Tregs in the RA joint.
Depletion of CD161+ T cells from RA SF responder cells restores Treg-mediated suppression
Having shown that CD161+ Th cells were enriched in RA SF and that CD161+ Th cells from HCs were relatively resistant to Treg-mediated suppression; we hypothesized that this might account for the impaired suppressive function of Tregs in RA SF. Constraints in cell numbers that can be recovered from SF precluded us from directly assessing the ability of Tregs to suppress CD161+ versus CD161− Th cells. Therefore, we indirectly determined the contribution of CD161+ Th cells by comparing the Treg-mediated suppression of total versus CD161-depleted responder T cells. Total CD4+CD25− responder cells and CD161-depleted responder T cells (CD4+CD25−CD161−) were sorted from RA SF and labeled with CTV. Responder T cells were stimulated with anti-CD3 in the presence of irradiated APCs and cultured in the presence or absence of CD4+CD25+CD127loCD39+ Tregs at a 1:1 ratio. CD161-depleted responder cells were more susceptible to Treg suppression of proliferation than total responder cells were (Fig 7A, collated data in Fig. 7C; p < 0.05). Furthermore, IFN-γ production by CD161-depleted responder T cells was markedly more suppressed than total responder T cell IFN-γ (Fig. 7B, collated data in Fig 7C). These data suggest that the reduced suppression of autologous responder T cells by Tregs from RA SF was due to resistance to suppression by responder T cells rather than an impairment of Treg function. Furthermore, the data suggest that the CD161+ CD4 T cells were responsible for the resistance of SF responder T cells to Treg-mediated suppression.
Blocking STAT3 restores Treg-mediated suppression of CD161+ Th cells
Next the mechanism by which CD161+ Th cells evaded Treg-mediated suppression was investigated. The expression of TRAIL was previously associated with impaired Treg suppression in RA (8); however, we observed no differential expression of TRAIL between CD161+ and CD161− Th cells either ex vivo or after activation (Supplemental Fig. 3A). Because the expression of CD161 was associated with resistance to Treg suppression, we next determined the effect of blocking CD161 on susceptibility to Treg suppression. A neutralizing Ab to CD161 or matched isotype control Ab was added to cocultures of Tregs and CD161+ or CD161− responder cells; however, no effect on the suppression of either subpopulation was observed (Supplemental Fig. 3B). The question of whether soluble factors produced by CD161+ Th cells had a role in mediating their resistance to suppression was also investigated. Cell culture supernatants from previously stimulated CD161+ or CD161− Th cells were added to cocultures of Tregs and either CD161+ or CD161− responder cells; however, no effect on the suppression of either subset was observed (Supplemental Fig. 3C). In addition, there was no difference in the susceptibility of CD161+ versus CD161− Th cells to inhibition by Treg-associated soluble factors IL-10 and TGB-β (Supplemental Fig. 3D). Having excluded a role for soluble factors in mediating resistance to suppression, we next addressed whether contact with CD161+ Th cells reduced the viability of Tregs or their expression of FOXP3, CD39 or CTLA-4. However we found no evidence of alterations in Treg viability or expression of FOXP3, CD39, and CTLA-4 after coculture with either CD161+ or CD161− Th cells (Supplemental Fig. 3E).
Resistance to Treg-mediated suppression has previously been associated with activation of STAT3 in responder T cells (48). We therefore measured STAT3 phosphorylation in CD161+ and CD161− Th cells ex vivo. We found a significantly increased expression of pSTAT3 in CD161+ Th cells relative to CD161− Th cells (Fig. 8A). We next determined the effect of blocking STAT3 on the resistance of CD161+ Th cell cells to Treg-mediated suppression, using the STAT3 inhibitor Stattic. Stattic was previously shown to be specific for STAT3 at the concentration used in this study (0.47 μM) (49). CD161+ or CD161- responder cell populations were sorted, CTV-labeled, and pretreated with Stattic or DMSO vehicle, followed by washing and activation in the presence or absence of Tregs. Interestingly, pretreatment of CD161+ responder cells with Stattic rendered them more susceptible to Treg-mediated suppression of proliferation and cytokine production (TNF p < 0.01; GM-CSF p < 0.05) while having no significant effect on the suppression of CD161− responder cells (Fig. 8B, representative plots in Fig. 8C, 8D). These data indicate that the resistance to Treg-mediated suppression of CD161+ Th cells was dependent on activation of STAT3.
This study has provided new insight into the failure of Tregs to constrain autoimmune inflammation in the RA joint. We performed a detailed examination of T cell regulation in RA using SF cells taken from inflamed joints. Our work has contributed to understanding Treg-mediated suppression in the RA joint by using stringent sorting techniques and a flow cytometry–based suppression assay for robust assessment of the suppression of proliferation and several proinflammatory cytokines. We showed that CD161+ Th17 lineage cells, which were enriched in the RA synovial tissue and fluid, exhibited pathogenic characteristics. In addition, CD161+ Th17 lineage cells were inherently resistant to suppression, even in HCs, and their depletion could restore the impaired Treg suppression in cocultures of Treg and responder T cells from RA SF. Finally, the resistance to Treg-mediated suppression of CD161+ Th cells was found to be STAT3 dependent.
We found that the frequency of Tregs in RA SF was significantly increased compared with peripheral blood, which is consistent with previous findings (4, 5). Factors that may contribute to the accumulation of Tregs in the RA joint include increased trafficking, local proliferation, or enhanced survival. Blood-derived Tregs have been shown to be anergic and relatively prone to apoptosis in vitro (41). However, our data indicated that Tregs in the RA joint were highly proliferative; indeed, they were proliferating to a greater extent than non-Treg effector T cells. In addition, similar Bcl2 levels in SF Tregs and non-Treg effector T cells indicated that they were not more prone to apoptosis than non-Tregs were, unlike in peripheral blood.
We demonstrated for the first time, to our knowledge, that the expression of HIF-1α was increased in Tregs taken from the RA joint relative to peripheral blood. This is likely to be important for the survival of Tregs, because the inflamed joint has been shown to be hypoxic (50) and HIF-1α is a key factor that senses oxygen levels and coordinates the cellular response to hypoxia. Studies in mice have indicated that HIF-1α promotes Th17 cells but antagonizes Treg differentiation (51, 52). However, consistent with our findings, others have demonstrated that hypoxia can induce the expression of FOXP3 via HIF-1α (53, 54). Expression of HIF-1α is required for Treg function in colitis, because Hif1a-deficient Tregs failed to control colitis (54). Thus, it seems that, although hypoxia and HIF-1α may inhibit the differentiation of Tregs, it may nonetheless be necessary for the survival and function of either thymic or peripheral Tregs at sites of inflammation. The increased rate of proliferation of Tregs in the RA joint, together with their ability to survive in the joint environment may explain their accumulation.
We showed that Tregs, which are found at high frequency in RA SF, expressed the suppressive molecules CD39 and CTLA-4. In vitro experiments suggested that this activated Treg phenotype could be induced by activation in the presence of RA SF. These findings suggested that RA synovial Tregs should have good suppressive function, yet they fail to constrain joint inflammation. We have previously shown that CD39+ Tregs expressed higher levels of FOXP3 and could suppress IL-17, whereas CD39− Tregs expressed less FOXP3 and did not suppress IL-17 by virtue of the fact that the Tregs themselves secreted IL-17 (55). Importantly, in the current study we were able to distinguish the contribution of Tregs and responder T cells to proliferation and cytokine production using cell tracer dyes. Tregs from RA SF exhibited suppressive activity on autologous responder cell proliferation and cytokine production, consistent with other studies in RA SF (4, 5). However, the current study has robustly shown, for the first time, to our knowledge, that suppression of proliferation and cytokine production from responder T cells by autologous CD39+ Tregs was significantly reduced in RA SF compared with HC peripheral blood. Given the suppressive phenotype of RA SF Tregs, we hypothesized that the reduced suppression might be due to a resistance of RA SF responder T cells to suppression by Tregs, rather than impaired capacity of the Tregs. This view is consistent with studies performed in JIA SF, where crossover studies demonstrated that reduced Treg suppression was due to resistance of responder cells to suppression rather than defective Tregs (56, 57). However, this contrasts with previous studies showing that Tregs from the blood of RA patients were inherently defective in their suppressive capacity (6, 7, 9). It is possible Tregs are activated in the inflammatory milieu of the RA joint, allowing them to overcome defects following their recruitment from the blood. Alternatively, different Treg sorting strategies could account for discrepancies in results between studies. We sorted CD39+ Tregs from both RA SF and HC blood to ensure that the Treg populations were comparable; however, it is possible that this stringent sorting strategy excluded less suppressive CD39− Tregs, which could account for the diminished suppression by RA Tregs observed by others.
Given the important role that has been ascribed to Th17 cells in RA and autoimmunity in general, it is important to understand their regulation. We used the marker CD161 to identify Th17 lineage cells, which include bona fide Th17 cells as well as ex-Th17 or non-classical Th1 cells (29). Interestingly, we found that over a large number of healthy donors, CD161+ Th cells were significantly less susceptible to Treg-mediated suppression. To our knowledge, we showed for the first time that CD161+ Th cells were significantly enriched in tissue biopsies from inflamed RA joints. In agreement with another study (30), we observed significantly increased frequencies of CD161+ Th17 lineage cells in RA SF. This prompted the question of whether the reduced suppression that we had observed using Tregs and responder cells in RA SFs could be explained by resistance to Treg-mediated suppression of the enriched CD161+ Th cell population within SF. Interestingly, we found that depletion of CD161+ Th cells from the RA SF responder cells significantly increased the suppression of proliferation by autologous Tregs. To our knowledge, our data indicate for the first time that CD161+ Th17 lineage population as a whole is resistant to Treg-mediated suppression in the RA joint and in HCs. Of course, within HCs the CD161+ Th cell population represents a minor component of the total responder Th cell population and would therefore not have a significant effect on Treg suppression of the total population. However, at sites of autoimmune inflammation such as in RA, JIA, multiple sclerosis, and inflammatory bowel disease where the proportion of CD161+ Th cells has been shown to be significantly increased (28, 30–34), this resistance of CD161+ Th cells to suppression is likely to be of key importance and might explain the inability of Tregs to constrain inflammation at such sites. Because CD161+ Th17 lineage cells are plastic and can switch to become so called ex-Th17 or nonclassical Th1 cells, it is possible that the resistance to Treg-mediated suppression observed within the CD161+ Th cell population could be dissected further, and this is the subject of ongoing investigation. There is currently disparity in the literature regarding the relative susceptibility of Th17 cells to Treg-mediated suppression. To our knowledge, this study is the first to demonstrate robustly that the CD161+ Th subpopulation is resistant to Treg-mediated suppression of proliferation and cytokine production. Other studies have suggested likewise. Using expanded clones of discrete Treg, Th1, Th17, and ex-Th17 cells, one group showed that proliferation of Th17 lineage clones were less susceptible to suppression (58). In contrast, another study documented that ex vivo Th17 cells were suppressed by Tregs, showing that IL-17 was reduced in CD161+ Th17 cells cocultured with Tregs, and that IFN-γ was reduced in CXCR3+ Th1 cells cocultured with Tregs in three donors (59). The present study directly compared differential Treg-mediated suppression of CD161+ and CD161− Th cells by examining proinflammatory cytokines produced by both subpopulations, therefore enabling a direct comparison to be drawn between the susceptibility to suppression in each subpopulation. It is important to understand the mechanism by which Th17 lineage cell evaded Treg-mediated suppression. We excluded a number of possible mechanisms, including a role for soluble factors and the expression of CD161. Interestingly, however, blockade of STAT3 in the CD161+ responder cells reversed their resistance to suppression, indicating that the resistance was STAT3 dependent. Further investigation will be required to determine the specific downstream effects of STAT3 on Treg suppression mechanisms. These findings are consistent with a study showing that activation of STAT3 by IL-6 in T cells induced resistance to Treg-mediated suppression (48). Furthermore, STAT3 activation has been shown to promote inflammation in RA (60).
Given that CD161+ Th cells within the RA joint do not appear to be effectively regulated by Tregs, it is particularly important to consider their potential pathogenic effects. Much of the focus of the pathogenicity of Th17 cells has focused around the production of their signature cytokines IL-17A and IL-17F. However, it is becoming increasingly evident that T cell subsets are heterogeneous and plastic in terms of cytokine production; therefore, complete assessment of T cell cytokine production requires analysis of multiple cytokines that can be produced by subsets of T cells. Polyfunctional cytokine production appeared to be an inherent feature of CD161+ Th cells, because those from HCs exhibited significantly increased polyfunctionality compared with CD161− Th cells. Furthermore, CD161+ Th cells from the RA joint were found to be the predominant Th cell source of GM-CSF and TNF. Given the established role of TNF and emerging prominence of GM-CSF in autoimmune inflammation (61–64), this finding is significant in the context of RA. Synergistic effects of cytokines are likely to be of key importance in mediating inflammation and fibroblast hyperplasia in the RA joint, as has been shown for combinations of IL-17 and TNF (36, 65). Indeed, we found that the supernatants taken from the polyfunctional CD161+ Th cells induced higher expression of IL-6, IL-8, RANKL, and ICAM-1 than those from CD161− Th cells. These molecules are likely to have important pathogenic effects in driving inflammation and leukocyte recruitment and promoting bone remodeling. In light of the relatively poor efficacy of anti–IL-17 therapy for RA, targeting single cytokines might not be an effective therapeutic strategy to abrogate the proinflammatory effects of T cells. These data herein demonstrate that targeting CD4+CD161+ Th cells that produce multiple proinflammatory cytokines may be a more effective strategy.
We show in this study that in addition to their increased cytokine production and resistance to Treg-mediated suppression, CD161+ Th cells possess important characteristics that may promote their survival and accumulation within the inflamed joint once they have been recruited to the site. The RA joint is profoundly hypoxic, and hypoxia drives inflammation (50, 60). To survive in hypoxia, cells must stabilize HIF-1α, which then co-ordinates the switch to anaerobic glycolysis. To our knowledge, we showed for the first time that CD161+ Th cells exhibit increased expression of HIF-1α in the RA joint. This adds to the growing awareness that hypoxia and metabolism play a key role in regulating T cells and is consistent with recent studies showing that HIF-1α expression promotes Th17 cells (51, 52). Survival of CD161+ Th cells in the RA joint may be promoted relative to other effector cells due to their increased expression of the antiapoptotic molecule Bcl-2. Replicative senescence resulting from telomere shortening after chronic stimulation represents a constraint for T cell persistence (66). We showed that CD161+ Th cells in the RA joint had relatively longer telomeres than their CD161– counterparts did, suggesting that they have an increased ability to undergo multiple rounds of proliferation. Thus, CD161+ Th cells have a number of features likely to promote their persistence, and these features might explain their enrichment in the inflamed hypoxic RA joint.
This study has demonstrated that CD161+ Th cells are not only enriched but also highly pathogenic in the RA joint. In addition, CD161+ Th cells appear to be resistant to suppression by Tregs, and their depletion was able to restore the impaired suppression observed in cocultures of Tregs and responder cells taken from the RA joint. These findings have important implications for autoimmune inflammation in general because CD161+ Th cells have been shown to be enriched at sites of autoinflammation where they might not be adequately constrained by Tregs. Furthermore, CD161+ Th cells may represent a valid therapeutic target in RA.
We thank Dr. Neil Marshall for guidance with SPICE software.
This work was supported by Science Foundation Ireland Starting Investigator Grant B1593 (to J.M.F.).
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.