Regulatory T cells in psoriatic arthritis: an IL-17A-producing, Foxp3intCD161 + RORγt + ICOS + phenotype, that associates with the presence of ADAMTSL5 autoantibodies

In psoriatic arthritis (PsA), predisposing class I HLA alleles, the presence of synovial clonally proliferated CD8 + T cells and autoantibodies all point towards the loss of immune tolerance. However, the key mechanisms that lead to immune dysregulation are not fully understood. In other types of inflammatory arthritis, T regulatory cell (Treg) dysfunction and plasticity at sites of inflammation were suggested to negatively affect peripheral tolerance. We here addressed if Treg variances associate with psoriatic disease. We collected clinical data, sera and peripheral blood mononuclear cells from 13 healthy controls, 21 psoriasis and 21 PsA patients. In addition, we obtained synovial fluid mononuclear cells from 6 PsA patients. We studied characteristics of CD4 + CD25 + CD127loFoxp3 + Tregs by flow cytometry and used ELISA to quantify antibodies against ADAMTSL5, a recently discovered autoantigen in psoriatic disease. In comparison with their circulating counterparts, Tregs from inflamed joints express increased levels of ICOS, CTLA-4 and TIGIT. Furthermore, synovial fluid-derived Tregs have a distinct phenotype, characterized by IL-17A production and upregulation of CD161 and RORγt. We identified a subset of Tregs with intermediate Foxp3 expression as the major cytokine producer. Furthermore, ICOS + Tregs associate with PsA disease activity as measured by PASDAS. Lastly, we observed that presence of the Foxp3int Tregs associates with an increased abundance of anti-ADAMTSL5 autoantibodies. Tregs derived from the inflammatory environment of inflamed PsA joints exhibit a distinct phenotype, which associates with loss of peripheral immune tolerance in psoriatic disease.

www.nature.com/scientificreports/ Flow cytometry. We stained samples by incubation with 25 µl antibody mix diluted in buffer (500 ml phosphate-buffered saline + 5 ml 10% sodium azide + 5 g bovine serum albumin) for 25 min at 4 °C. Before intracellular stains, we fixed and permeabilized cells with 100 µl Fixation/Permeabilization Concentrate and Diluent (00-5123-43, 00-5223-56, eBioscience). Details of flow cytometry antibodies of both panels used are listed in Supplemental Table S1. Using fluorescence minus one (FMO) controls, we identified viable (assessed by a Fixable Viability Dye) CD3 + CD4 + CD25 + CD127 lo Foxp3 + Tregs. To further study the phenotypical and functional properaties of Tregs with reduced expression of Foxp3, we studied Tregs with intermediate (Foxp3 int ) and high (Foxp3 hi ) Foxp3 expression. Importantly, we do not propose to have identified two different subsets of Tregs, but use this exploratory dichotomization of the Treg population to enable studying the differences between Tregs with high and intermediate Foxp3 expression. Of these Foxp3 int and Foxp3 hi Treg subsets, we used FMO controls to assess median fluorescent intensity (MFI) and proportions of cell populations that express cytotoxic T-lymphocyte-associated protein 4 (CTLA-4 = CD152), CD161, inducible T-cell costimulator (ICOS = CD278), T cell immunoreceptor with Ig and ITIM domains (TIGIT), Ki67 and RORγt. In addition to FMO controls, we found that the modest expression of RORγt and CD161 required standardization by using a uniform gate based on a representative healthy control sample of CD3 + CD4 + CD25 + CD127 lo Tregs. We standardized quantification of intracellular IL-10 and IL-17A by applying a cutoff value of < 0.5% in the medium control samples. Negative controls of flow cytometry analyses are shown in Supplemental Figure S1 and S2. We performed acquisition on the BD LSRFortessa (405, 488, 561, 635 nm lasers) with BD FACSDIVA (version 8.0.1). We used FlowJo (version 10.7.1) for further analyses.

Results
Cohort. To investigate Treg phenotypical variances in psoriatic disease, we studied characteristics of Tregs in peripheral blood from 13 healthy controls (HC), 21 psoriasis patients and 21 PsA patients, and in synovial fluid of 6 PsA patients. Detailed patient characteristics are shown in Supplemental Table S2. We identified Tregs using the best available discriminative markers: CD3 + CD4 + CD25 + CD127 lo Foxp3 + (Fig. 1A) 1B). As Foxp3 is the key transcription factor of Treg development, maintenance and function, this finding raised our interest in phenotypical and functional properties of Tregs with reduced expression of Foxp3 29 . Therefore, we studied two subset of Tregs: with intermediate (Foxp3 int ) and high (Foxp3 hi ) Foxp3 expression (Fig. 1C). CD4 + CD25 + CD127 lo T cells without Foxp3 were excluded from further analyses. As compared to circulating T cells from PsA patients, in synovial fluid we observed an increase of Foxp3 int Tregs of CD3 + CD4 + CD25 + CD127 lo lymphocytes from 11 to 33% in SF (P < 0.001) and a decrease of Foxp3 hi Tregs from 80 to 50% (P < 0.001). The percentages of Foxp3 int and Foxp3 hi Tregs were strongly negatively correlated in both PBMC and SFMC of HC, PsO and PsA patients (ρ > 0.9, P < 0.05)(Supplemental Figure S3). We observed no significant differences in Foxp3 expression between healthy controls, psoriasis and PsA patients.   www.nature.com/scientificreports/ Ki67-expressing tregs are increased in inflamed PsA joints. Subsequently, we focused on differences between Foxp3 int and Foxp3 hi Tregs in peripheral blood and synovial fluid by assessing their relative frequencies and proliferative capacity. Compared to peripheral blood, synovial fluid was significantly enriched for Foxp3 int Tregs (P < 0.001)(Supplemental Figure S4A and S4B). When examining the proliferative capacity, we noted that in general Foxp3 hi Tregs had higher proliferative capacity than Foxp int Tregs. Nonetheless, both subsets (Foxp int and Foxp3 hi ) derived from synovial fluid had higher proliferative capacity compared to their peripheral blood counterparts (Supplemental Figure S4C-E).

A subset of tregs in inflamed joints in PsA upregulate CD161 and RORγt.
To further investigate the phenotype of Foxp3 int and Foxp3 hi synovial fluid-derived Tregs in PsA we studied CD161 and RORγt, as they are associated with arthritis and a pro-inflammatory potential of Tregs 7,16,30 . In PsA patients, we found that the percentage of CD161-expressing Tregs was higher in synovial fluid than in circulation and that more Foxp3 int Tregs express CD161 (4.1%), as compared to Foxp3 hi Tregs (1.3%)( Fig. 2A-C). Additionally, in PsA, synovial fluid Tregs express more RORγt than Tregs in circulation (Foxp3 int 3.0 vs. 1.0%, P = 0.048; Foxp3 hi 1.8% vs. 0.5% Treg, P = 0.026) (Fig. 2D-F). The increased expression of CD161 and RORγt by synovial Tregs was most pronounced in the intermediate Foxp3 subset of Tregs ( Fig. 2A and D). Again, no differences were found between peripheral blood Tregs of healthy controls, psoriasis and PsA patients.
High IL-10 and IL-17A production by synovial fluid-derived tregs. To study functional differences between Tregs in an inflammatory environment and in circulation of PsA patients, we measured inhibitory and pro-inflammatory cytokine production. As expected, Tregs from synovial fluid showed a modest but increased capacity to produce cytokines, both the anti-inflammatory cytokine IL-10 (3.7% vs. 1.8%, P < 0.001) (Fig. 3A) and the key pro-inflammatory cytokine IL-17A (3.2% vs. 1.7%; P = 0.002) (Fig. 3E). Moreover, Tregs from PsA patients produce less IL-10, as compared to psoriasis patients ( Fig. 3A and B). When examining the different subsets of Tregs, we found that the Foxp3 int subset was the major cytokine producer, the most notable being the elevated IL-17A producing capacity by Foxp3 int synovial fluid Tregs (5.9% vs. 1.2%, P = 0.028) (Fig. 3B-D and 3F-H). We observed that-both in peripheral blood and in synovial fluid-PsA patients have a higher proportion of single-IL-17A cytokine-producing Tregs as compared to healthy controls and psoriasis patients. A minority of Tregs produce both IL-10 and IL-17A (Supplemental Figure S5).

Synovial fluid-derived tregs express high CTLA-4, TIGIT and ICOS. Reduced expression of
immune receptors by Tregs could contribute to abnormal Treg function in inflammatory arthritis 8 . Therefore, we measured expression of two key inhibitory receptors essential for Treg suppressive function: CTLA-4 and TIGIT ( Fig. 4A-F). Both receptors were expressed more by Foxp3 hi Tregs, as compared to the intermediate Treg subset (Fig. 4A and D). In PsA patients, the proportions Tregs that express CTLA-4 were increased in synovial fluid, as compared to peripheral blood (Foxp3 int 23.5% vs. 3.3%, P = 0.000; Foxp3 hi 28.6% vs. 7.4%, P = 0.003) ( Fig. 4A . 4A  and B). Moreover, we included ICOS in our phenotypical Treg characterization, because ICOS + Tregs can play a proinflammatory, pathogenic role in inflammatory arthritis and immune diseases 31,32 . We observed a comparable expression pattern as for the inhibitory receptors: synovial fluid derived Tregs express more ICOS, as compared to Tregs in circulation (Fig. 5A-D). Furthermore, we found a difference between the Foxp3 int and Foxp3 hi Treg subsets: in peripheral blood Foxp3 int Tregs express less ICOS as compared to Foxp3 hi (3.5% vs. 8.6%, P = 0.000), but in synovial fluid both subset express similar levels (10.3% vs. 13.0%, P = 0.753) (Fig. 5A). This is a relevant finding, because we observed an association of ICOS expression on Treg with PsA disease activity as measured by PASDAS (range 0-10), which takes arthritis, enthesitis, dactylitis, C-reactive protein, physician disease activity score and two PROs into account (Fig. 5E). Both the proportion of ICOS + Tregs and the MFI of ICOS  www.nature.com/scientificreports/ significantly correlated with PASDAS in the Foxp3 int Treg subset and the Foxp3 hi Treg subset. No significant differences were found between healthy controls, psoriasis and PsA patients.

Discussion
In the inflammatory microenvironment of autoimmune disease, Treg defects and differentiation are suggested to play a role in loss of peripheral immune tolerance. However, the implications of Treg dysfunction and plasticity have not been clarified in psoriatic disease. To our knowledge, this is the first study to perform in-depth phenotypical characterization of Tregs derived from the inflammatory microenvironment of inflamed joints in patients with psoriatic disease. Here, we provide evidence for Treg variance in PsA by showing distinct phenotypical and functional properties of intra-articular Tregs as compared to Tregs in circulation: downregulation of key transcription factor Foxp3, increased cytokine production, upregulation of inhibitory immune receptors, and upregulation of markers that have been reported to associate with a pro-inflammatory potential of Tregs: CD161, RORγt and ICOS 7,14,15,30,31,[37][38][39] .
Foxp3 is the key transcription factor of Tregs and its expression is essential for Treg development, maintenance and function 29 . Our results demonstrate a significant increase of intra-articular Tregs with intermediate Foxp3 expression in PsA patients. Association of Foxp3 with PsA has been previously described by one study, that identified a hemizygous Foxp3 mutation (c.1222G > A) in familial juvenile PsA 35 . Moreover, in psoriasis patients, it was shown that enhanced loss of Foxp3 is linked to Treg differentiation into IL-17A producing cells 15 . In the broader context of autoimmune disease, multiple studies suggested that stability of Foxp3 expression is negatively affected by pro-inflammatory conditions 13,19 . These findings have clinical relevance, because Treg defects-including Foxp3 instability-could contribute to disease pathophysiology. This contribution is either through increased escape of autoreactive T cells from Treg regulation or, what has been suggested more recently, by conversion of Tregs into autoreactive, memory T cells 13,15,29 .
Furthermore, we show that the intra-articular subset of Tregs with lower Foxp3, as compared to their Foxp3 hi counterparts in the same tissue location, have lower expression of the inhibitory receptors CTLA-4 and TIGIT. These Foxp3 int Tregs produce even more of the psoriatic disease hallmark cytokine IL-17, and display the highest expression of CD161. These substantial differences between Foxp int and Foxp3 hi are relevant, since our results demonstrate that Tregs with decreased Foxp3 expression are present in large numbers in the synovial compartment. In line with the homeostatic importance of Tregs in psoriatic disease, we observed a relation of Foxp3 instability with loss of immune tolerance in psoriatic disease: presence of ADAMTSL5 antibodies in psoriasis and PsA reflected the balance between Foxp int and Foxp3 hi Tregs. However, as we did not perform functional experiments and only studied relative Foxp3 expression, we are careful to draw definite conclusions.
Based on their phenotypical characteristics, synovial fluid-derived Tregs-and in particular the Foxp3 int Treg subset-might play a role in pathogenesis or ongoing inflammation in PsA. First, because failure to upregulate inhibitory receptors has been shown to contribute to the comprised suppressive function of intra-articular Tregs in inflammatory arthritis 8 . In addition, CD161 + Tregs were previously identified as a subset capable of IL-17A and IFNγ production, and to exhibit a pro-inflammatory potential 7,30 . In fact, CD161 + Tregs are the predominant IL-17 producing Treg population in inflamed joints of inflammatory arthritis patients 7 . Furthermore, concerning the implications of RORγt + Tregs, in inflammatory bowel disease it was shown that the capacity of IL-17 + RORγt + Tregs to suppress autologous T cell proliferation is reduced by approximately 60% 14 . Also, RORγt expression associates with IL-17A producing Tregs in psoriasis 15 . These findings are not surprising, considering that Foxp3 and RORγt transcription factors drive differentiation of T cells towards Tregs or Th17 cells 36 . With regards to functional differences, intra-articular Tregs may contribute to ongoing localized inflammation by increased production of IL-17A, as compared to peripheral Tregs. IL-17 has previously been associated with unresponsiveness of Teff in the microenvironment of inflammatory arthritis 18 . Taken together, we identified a distinct phenotype of synovial fluid-derived Tregs, most pronounced in the Treg subset with downregulated * * * * * * *  Further, the association of ICOS + Tregs with PsA disease activity drew our attention, because ICOS is most commonly associated with a strongly inhibitory Treg subset 32 . However, studies in the last decade have suggested a possible role for ICOS + Tregs in pathogenesis, contributing to autoimmune rheumatic disease. In lupus, RA and spondyloarthritis, associations were found of ICOS expression by Tregs with high disease activity, with non-response to therapy, increased autoantibodies, and pro-inflammatory cytokine production 31,37-39 . Since we show that all intra-articular Tregs upregulate ICOS expression, even independent of Foxp3 expression, further investigation is warranted for ICOS + Tregs as possible therapeutic target for treatment of psoriatic disease.
Our study has several limitations. First, we were limited by the number of subjects tested, especially concerning synovial fluid mononuclear cell samples, and future studies to confirm our results are warranted. However, we deem that the evident observed dissimilarities between Tregs derived from synovial fluid and peripheral blood have enabled us to draw conclusions about Treg variances. Second, the use of DMARDs (methotrexate and golimumab) by two patients in our SFMC cohort could have influenced our results, although contradicting results have been published as to whether DMARDs affect Treg phenotype and function [40][41][42] . Third, we have not assessed the pro-inflammatory potential of Tregs in functional experiments or performed assays to evaluate the suppressive capacity of intra-articular Tregs. Fourth, with flow cytometry analyses we could only assess relative differences. Hence, we can only speculate about the implications of absolute numbers of this distinct subset of Tregs in PsA pathogenesis. Fifth, we observed markedly lower Treg CD161 expression as compared to literature, which may be the result of our gating strategy that selected only Foxp3 + CD25 + CD127 lo T cells in an attempt to exclude activated non-Treg T cells with low Foxp3 expression. Moreover, this gating strategy might have resulted in the inclusion of a minor subset of effector T cells, that upon activation might transiently express CD25 and Foxp3 43 . Nevertheless, based on what is known from literature in other types of inflammatory arthritis, we deem to have identified an interesting Treg subset that warrants further investigation.
Since we have not confirmed our hypothesis by assessing the suppressive capacity of intra-articular Tregs, it must be taken into consideration that even differentiated Tregs may be able to effectively suppress Teff 6,7,16 . If and how differentiated Tregs in inflammatory arthritis can effectively suppress Teff T cells is an increasing topic of interest and contradicting results have been published 44 . Some concluded that impaired expression of immune regulatory molecules or lack of cytokine production are key to defective Teff suppression by Tregs 8,9,45 . Others attributed failure of effective Teff suppression to the factor that Tregs are prone to apoptosis under inflammatory conditions 44 . Moreover, evidence suggested that sustained resistance of local CD4 + and CD8 + Teff in an inflammatory microenvironment could be key to ineffective Treg suppression 18 . Whether these mechanisms play a role in psoriatic disease has yet to be elucidated. In the case that an important role for Tregs derived from sites of inflammation in PsA pathogenesis is confirmed, this may facilitate identification of new treatment targets and therapies, including Treg growth factors, Treg stabilizing factors and therapies that enhance Treg function 46 . In preclinical models and clinical trials low dose IL-2 and cellular therapy with polyclonal, therapeutic Tregs have already shown promising results [46][47][48] . Moreover, research to artificially stabilize Treg Foxp3 expression in vitro for clinical applications are ongoing 49 . As treatment options for autoimmune disease are evolving, we deem it essential to further advance our understanding of the role of Tregs in psoriatic disease pathogenesis.

Conclusions
In conclusion, we show that Tregs derived from the inflammatory environment of inflamed joints in PsA patients exhibit a distinct phenotype characterized by increased expression of CD161, RORγt and ICOS. Moreover, we identify the importance of Foxp3 expression by Tregs, with a novel role for Foxp3 int Tregs with a heightened capacity to produce IL-17A.