|Home | About | Journals | Submit | Contact Us | Français|
Regulatory T cells (Tregs) exert their anti‐inflammatory activity predominantly by cell contact‐dependent mechanisms. A study was undertaken to investigate the regulatory capacity of autologous peripheral blood Tregs in contact with synovial tissue cell cultures, and to evaluate their presence in peripheral blood, synovial tissue and synovial fluid of patients with rheumatoid arthritis (RA).
44 patients with RA and 5 with osteoarthritis were included in the study. The frequency of interferon (IFN)γ‐secreting cells was quantified in synovial tissue cell cultures, CD3‐depleted synovial tissue cell cultures, synovial tissue cultures co‐cultured with autologous CD4+ and with CD4+CD25+ peripheral blood T cells by ELISPOT. Total CD3+, Th1 polarised and Tregs were quantified by real‐time PCR for CD3ε, T‐bet and FoxP3 mRNA, and by immunohistochemistry for FoxP3 protein.
RA synovial tissue cell cultures exhibited spontaneous expression of IFNγ which was abrogated by depletion of CD3+ T cells and specifically reduced by co‐culture with autologous peripheral blood Treg. The presence of Treg in RA synovitis was indicated by FoxP3 mRNA expression and confirmed by immunohistochemistry. The amount of FoxP3 transcripts, however, was lower in the synovial membrane than in peripheral blood or synovial fluid. The T‐bet/FoxP3 ratio correlated with both a higher grade of synovial tissue lymphocyte infiltration and higher disease activity.
This study has shown, for the first time in human RA, the efficacy of autologous Tregs in reducing the inflammatory activity of synovial tissue cell cultures ex vivo, while in the synovium FoxP3+ Tregs of patients with RA are reduced compared with peripheral blood and synovial fluid. This local imbalance of Th1 and Treg may be responsible for repeated rheumatic flares and thus will be of interest as a target for future treatments.
Rheumatoid arthritis (RA) is characterised by autoimmune phenomena and destructive polyarticular arthritis with hyperplasia of the synovial lining cells, neovascularisation and leucocyte infiltration of the sub‐lining forming distinct anatomical patterns of secondary lymphoid tissues.1 The predominance of a Th1 lymphocyte profile together with defective peripheral immune tolerance appears to be pivotal in the pathogenesis of RA.2 High ratios of interferon γ (IFNγ) to interleukin 4 (IL‐4) have been found in synovial tissue and synovial fluid from patients with RA,3,4 while IFNγ/IL‐4 ratios were comparably low in the peripheral blood of patients with active disease.5
Various regulatory CD4+ T cell (Treg) subsets have been described in human pathology, including natural CD4+CD25+ naïve Treg,6 antigen‐induced CD4+CD25+ effector/memory Treg,7 IL‐10‐producing Tr18 and transforming growth factor β (TGF‐β)‐producing Th3 cells.9 Anti‐inflammatory mediators such as IL‐10 and TGF‐β may be abundantly present in inflamed RA joints, but they obviously fail to control the disease process sufficiently.10,11 Cellular peripheral tolerance mechanisms appear attractive for targeted treatment in autoimmunity.12,13 In comparison to a healthy state, Tregs in the peripheral blood are already reduced in early active RA,14 suggesting compromised Treg function in RA. In line with this observation, transfer of CD4+CD25+ Tregs markedly slowed down disease progression in the collagen‐induced arthritis model15 and in bone marrow transplanted children with juvenile idiopathic arthritis.16 However, high numbers of Tregs have been reported in the synovial fluid of patients with RA,17,18,19 and impaired Treg function in established active RA could be restored by successful blockade with tumour necrosis factor (TNF).20,21 These observations, together with resistance of synovial effector T cells against Treg, put the therapeutic options of Treg expansion and transfer not only in RA into perspective.22 In this investigation we found a potent regulatory effect of autologous peripheral blood Treg on synovial tissue cultures of patients with RA, which led us to search again for Treg in RA joints including inflamed tissue.
Forty‐four patients with RA who fulfilled the revised American College of Rheumatology (ACR) criteria23 were consecutively recruited for the study when undergoing joint surgery. Synovial tissue from five patients with osteoarthritis was used as a control. The study protocol was approved by the Frankfurt University Hospital ethics committee and all patients gave their written informed consent. The characteristics of the patients with RA are summarised in table 11.
Specimens for pairwise analyses were simultaneously obtained by vein puncture needle aspiration of the joint immediately before arthrotomy. Fluids were collected in NH4‐heparin containing S‐Monovettes (Sarstedt, Germany). Peripheral blood mononuclear cells and synovial fluid mononuclear cells were isolated by Ficoll Hypaque (Biochrom, Germany) density gradient centrifugation.
Preparation of synovial tissue for functional assays was done rapidly by injecting 1 mg/ml collagenase (Sigma‐Aldrich, Germany) into synovial tissue samples and incubating for 20 min at ambient temperature. The synovial tissue was subsequently minced and incubated for an additional 50 min at 37°C in collagenase 1%. The cell suspension was strained by 70 μm nylon filters (Falcon, USA), washed twice in phosphate‐buffered saline (PBS) and cultured in HAM F 10 medium (Invitrogen, Germany) containing 5% fetal calf serum and 1% penicillin/streptomycin at 37°C in 5% CO2 atmosphere over night.
Synovial tissue cell cultures were CD3‐depleted using anti‐CD3 monoclonal antibody‐coated magnetic beads (Miltenyi Biotec, Gladbach, Germany) according to the manufacturer's instructions. A rosette kit (StemCell Technologies, USA) was applied for isolation of CD4+ peripheral blood T cells (fig 11)) and CD25+ cells were enriched with an automated magnetic bead activated cell sorting system (autoMACS, Miltenyi Biotec, Germany). T cells were stained with 10 μl anti‐CD4‐PE/CY5, anti‐CD3‐ECD, anti‐CD45‐FITC (all BD Pharmingen, USA) or anti‐CD25‐PE (Miltenyi Biotec, Germany), incubated for 10 min at room temperature and measured by flow cytometry (Beckman Coulter, USA).
Synovial tissue cultures and peripheral blood mononuclear cell fractions were seeded in equal densities of 104 cells per well in sterile polyvinylidene fluoride microtitre plates (Millipore, USA) pre‐coated with anti‐IFNγ or anti‐IL‐10 antibody (Mabtech AB, Sweden) and cultured for 36 h at 37°C in a 5% CO2 atmosphere. Biotin‐labelled secondary monoclonal antibodies were added after removing the cell layer and thorough washing. Plates were dried for 2 h at ambient temperature and subsequently stained for 15 min with streptavidin alkaline phosphatase complexes and 5‐bromine‐4‐chlor‐3‐indoxylphos‐phate/nitroblue‐tetrazoliumchloride solution (BCIP/NBT, Mabtech, Sweden). Cytokine spots were quantified (A•EL•VIS GmbH, Hannover, Germany) and the mean spot number per well from duplicate or triplicate experiments was calculated. Net counts were established after background subtraction.
Acknowledging the relevance of transcription factors for gene expression and differentiation of lymphocytes, T‐bet and FoxP3 transcripts were quantified. FoxP3, the most specific Treg marker,25 is present in naïve natural and memory Tregs.6,7 T‐bet is mainly expressed in differentiated Th1 cells and closely linked to IFNγ expression.26 In brief, synovial tissue was immediately stored in RNAlater (QIAGEN, Hilden, Germany), homogenised after removal from RNAlater, peripheral blood mononuclear cells were disrupted and total RNA was isolated with RNeasy Mini Kit (QIAGEN) or TRIzol Reagent (Invitrogen, USA). First strand cDNA synthesis was performed using 1 μg total RNA by Thermoscript RT (Invitrogen, USA). Quantitative PCR was performed by pre‐designed TaqMan gene expression assays (Applied Biosystems, California, USA) for human T‐bet (Hs00203436), FoxP3 (Hs00203958), CD3ε transcripts (Hs00167894), a constitutively expressed gene in CD3+ T cells, and eukaryotic 18S rRNA (4333760T). Probes were labelled with FAM/MGB fluorescent dye and amplified with Absolute QPCR ROX Mix (ABgene, Epsorn, UK) in an ABI Prism 7700 sequence detector (Applied Biosystems, California, USA) in 25 μl reaction volume. The expression was defined by calculating the mean cycle number of triplicate experiments.
Specimens were immediately embedded in OCT compound (TissueTek TT 4583; Sakura Finetech, Torrance, California, USA) and snap frozen in liquid nitrogen. Sections 7 μm thick were prepared, fixed in acetone, washed in ddH2O and subsequently in PBS. Endogenous peroxidase activity was blocked using 0.1% H2O2. Slides were incubated for 30 min in blocking solution (2% horse serum in PBS) and incubated overnight at 4°C with 10 μg/ml mouse monoclonal anti‐FoxP3 antibody (AbCam, Cambridge, UK) or an isotype‐matched mouse IgG1 antibody (DakoCytomation, Glostrup, Denmark). Sections were incubated for 30 min with biotinylated goat anti‐mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, Pennsylvania, USA) secondary antibody in PBS. Immune complexes were stained with horseradish peroxidase‐conjugated streptavidin complex (Vectastain Elite ABC kit; Vector, Burlingame, California, USA) and aminoethylcarbazole chromogen substrate (DakoCytomation, Glostrup, Denmark). Nuclei were counterstained with haematoxylin.
The results are presented as mean (SE) unless otherwise stated. Group comparisons were calculated by the Mann‐Whitney U test for untailed groups and by the Wilcoxon signed rank test for tailed groups. The significance of correlations was estimated by the Pearson test. Statistical analyses were done with Graph Pad Prism 4.0 Software.
ELISPOT assays allowed detection of IFNγ in synovial tissue cultures without artificial stimulation (fig 1A1A).). The mean (SE) frequency of IFNγ spots was 65 (20)/104 cells (n=13) in RA‐derived synovial tissue cultures and 7 (2.5)/104 cells in OA‐derived synovial tissue cultures (n=5, p=0.03). In contrast, the frequency of IL‐10‐secreting cells was similar in RA and OA synovial tissue cultures (218 (94)/104 cells (n=5) vs 161 (86)/104 cells (n=4), p=0.9). CD3 depletion almost completely abrogated IFNγ production in RA synovial tissue cultures (mean frequency 1 (0.5)/104 cells per well (n=3), p=0.01), indicating that IFNγ in synovial tissue cultures was predominantly released by CD3+ T lymphocytes (fig 1C1C).). CD3+ depletion also strongly reduced IL‐10‐secreting synovial tissue cultures (fig 1D1D).
“Untouched” CD4+ cells were isolated from peripheral blood (fig 1B1B)) and co‐cultured with corresponding synovial tissue cultures. The number of intrinsically IFNγ‐secreting CD4+ peripheral blood T cells was low (8 (3)/104 cells per well), and co‐culture led to substantially reduced IFNγ frequency in autologous synovial tissue cultures (38 (17)/104 cells per well, p=0.002); the regulatory effect was even more pronounced when enriched CD4+CD25+ Tregs were used (IFNγ spots 3.5 (2.5)/104 cells per well (n=16), p<0.05, fig 1C1C).). In contrast, spontaneous IL‐10 secretion in RA synovial tissue cultures was not significantly modified by co‐culture with total CD4+ or CD4+CD25+ cells (218 (94) vs 206 (96) vs 178 (79)/104 cells), while their intrinsic IL‐10 frequency (25 (19)/104 cells) was relatively low in CD4+ peripheral blood T cells (fig 1D1D).
The remarkable regulatory capacity of autologous peripheral blood Tregs on synovial tissue cultures led us to search thoroughly for the presence of Tregs in RA synovial tissue. Immunohistochemistry revealed that Tregs were sparsely distributed throughout the entire sub‐lining tissue. FoxP3‐positive cells could also be detected at the basal layer of the hyperplastic synovial lining, and some clusters of these cells were seen in areas of lymphoid aggregates (fig 22).
Quantitative analysis of transcription factors prototypic for Th1 and Treg showed a close correlation between the T‐bet/FoxP3 mRNA ratio and CD3ε levels in synovial tissue (r=0.95; n=32, p<0.0001). To clarify whether there was a preponderance of Th1 compared with Tregs exclusively in the synovial tissue, we divided the patients into two equivalent groups according to their high or low T‐bet/FoxP3 mRNA ratio in synovial tissue (n=16; p<0.0001). We next looked at the corresponding ratios in synovial fluid and peripheral blood and found that the high T‐bet/FoxP3 ratio in synovial tissue was not seen in peripheral blood and synovial fluid samples (n=8; p>0.05 for both). Confirming this imbalance in the group with a high T‐bet/FoxP3 ratio in the synovial tissue compartment, a significantly lower ratio was measured in the synovial fluid (n=20, p=0.014). The amount of FoxP3 transcripts in synovial tissue was significantly lower than in peripheral blood (p<0.0001) and synovial fluid (p=0.0008) after normalisation to 18S rRNA and CD3ε transcripts (p=0.03 vs peripheral blood, p=0.01 vs synovial fluid; fig 3A, BB).
CD3ε mRNA expression in synovial tissue correlated significantly with DAS 28 (r=0.4, n=31, p=0.045, fig 3C3C).). Indicating that most of these CD3ε positive cells may exhibit an inflammatory profile, DAS 28 also correlated with T‐bet/FoxP3 mRNA ratios (r=0.4, n=31, p=0.02, fig 3D3D).). Further comparison of mRNA levels with clinical and treatment data in our cohort revealed that disease duration had no influence on the expression of FoxP3. In addition, there was no correlation between anti‐TNF treatment or steroid dosage and the expression of FoxP3 or FoxP3/CD3ε in synovial tissue (r=−0.04, p=0.8 for FoxP3; r=−0.01, p=0.9 for FoxP3/CD3ε).
This study has shown that IFNγ, a pro‐inflammatory mediator critically important for innate and adaptive immune processes, is expressed in primary synovial cell cultures predominantly by CD3+ T cells. Their frequency could be precisely quantified using the ELISPOT technique without applying artificial stimuli. The frequency of IFNγ (6/1000) and IL‐10 (20/1000) secreting T cells determined by this method in synovial tissue was in a similar range as was previously found by in situ detection methods,27 and our results in the peripheral blood show good agreement with a previous ELISPOT study.28
We observed reduced IFNγ secretion in synovial tissue cultures following co‐culture with peripheral blood Treg. Ongoing work will try to elucidate whether the naïve or memory phenotype of Treg is responsible for this key finding. In this context, we were able for the first time to show Treg in RA synovial tissue by FoxP3 immunohistochemistry, but found expression of FoxP3 transcripts significantly lower here than in synovial fluid or peripheral blood using quantitative transcript analyses. Our data are in agreement with a previous report of scarce CD25 expression in T cell infiltrates of synovial biopsies but a higher frequency of CD25+ T cells in the synovial fluid in active states of early arthritis.29 Following these data, a quantitative Treg deficiency in the RA synovium, either by restricted access or by impaired persistence in an inflamed milieu, can be postulated. Mechanisms potentially shortening the half‐life of Treg tissue have recently been published,21,30 but the cause of insufficient supplementation remains elusive. Exemplary flow cytometric analyses of synovial tissue culture also failed to reveal appreciable numbers of CD4+/CD25high cells, although these cells could easily be detected in corresponding peripheral blood samples (data not shown). As the staining of surface molecules might have been affected by the tissue preparation, we continued our quantitative analyses with intracellular Th1 and Treg markers. T‐bet expression is tightly correlated with IFNγ expression and migration properties of activated effector T cells.31 We found an increased T‐bet/FoxP3 transcript ratio in RA synovial tissues which could be correlated with the amount of CD3ε mRNA and clinical disease activity. These observations argue in favour of the hypothesis that CD3ε‐positive T cells infiltrating inflamed joint tissue underlie a selective distribution in favour of Th1 versus Tregs in active disease.
Treg‐mediated tolerance against arthritogenic peptides in humans requires cell‐cell contact.32 Prevention of disease in a murine model of type 1 diabetes was associated with Treg localisation in the target tissue or in draining lymph nodes, and their regulatory function was mediated by direct contact with dendritic cells.33,34 Alternative paracrine mechanisms of Treg function would require at least adjacent localisation.35 Tregs are thus required on the site of inflammation or the draining lymphoid tissue to display sufficient suppression, and their relative absence/disappearance in synovial tissue would be a sustainable explanation for the recurrent imbalance of Th1 and Treg activity leading to repeated disease flares in RA. Although the published data are complex and the molecular mechanisms may differ between natural naïve and effector/memory Tregs,36,37,38,39,40,41,42,43 therapeutic targeting of the migratory Treg appears to be an attractive goal.
The authors thank Martina Herrero San Juan for her continuous technical support during the years of this study.
IFNγ - interferon γ
IL - interleukin
RA - rheumatoid arthritis
TNF - tumour necrosis factor
Treg - regulatory T cell
This work was supported by a Tandem grant of the Clinic of the Goethe University (awarded to FB and AH) and by funds of the Dr Hans Schleussner Foundation of Immune Pharmacology awarded to HHR.
Competing interests: None.