Introduction
Rheumatoid arthritis (RA) is an autoimmune disease characterized by a chronic relapsing–remitting inflammation primarily of peripheral joints. The tissue-destructive inflammation affects all joint structures, leading eventually to total joint destruction. The pathology of the disease is characterized by disturbed immune regulation with predominance of inflammatory cells on the one hand and defective peripheral immune tolerance on the other [
1]. These immunopathologic events occur predominantly within the joints and particularly in the synovial membrane (SM) as the main site of mononuclear cell infiltration. CD4
+ T cells comprise a large proportion of the inflammatory cells recruited into the RA synovium and contribute significantly to synovial inflammation [
2]. In contrast to these inflammatory CD4
+ T cells, another T-cell subset, known as naturally occurring regulatory T cells (Tregs), has been shown to play an essential role in establishing the balance between pro- and anti-inflammatory mechanisms in the periphery and maintaining self-tolerance, both in rodents and in humans [
1,
3‐
5]. The ability of Tregs to suppress T-cell responses, and thereby to regulate immune reactions, ascribes to them a key role in the pathophysiology of autoimmune diseases and makes them an interesting target for treatment [
2,
6,
7]. The suppressive capacity of human CD4
+ T cells was first shown to be contained within CD4
+ cells expressing high levels of CD25, the α chain of the IL-2 receptor [
8]. Previous research has demonstrated that low expression of the IL-7 receptor α chain (CD127), in combination with CD25, allows the consistent and specific identification of viable CD4
+CD25
+/highCD127
low/- Tregs and facilitates their discrimination from activated CD4
+CD25
+CD127
+ effector T cells [
9‐
11]. Researchers in a number of studies have analysed the presence of Tregs in RA [
12‐
21]. Apart from contradicting results regarding the analysis of peripheral blood (PB), accumulating evidence indicates that Tregs are enriched in the synovial fluid (SF) of RA joints [
12‐
15,
17,
18,
20]. In contrast, the presence and phenotype of Tregs in the immunological organ of the joint, the SM, remains largely unstudied. This is of special interest because Tregs develop their suppressive capacity in a contact-dependent manner [
8,
9,
14]. Researchers in only two studies have analysed synovial biopsies and suggested a much lower frequency of Tregs in synovial tissue than in SF [
18,
20]. These immunohistochemical and quantitative PCR analyses provide preliminary insight into Treg distribution, but the techniques utilized do not allow further phenotypic analysis of synovial Tregs and limit comparability to SF and PB quantitative flow cytometry data. Furthermore, the comparison to equivalent samples from non-autoimmune-driven joint disease is missing. These data are essential to understanding RA-specific pathophysiology.
In our present study, we examined the frequency and phenotypes of CD4+CD25+/highCD127low/- Tregs in the three compartments (PB, SF and SM) of RA patients in order to map the distribution of Tregs between the periphery and the target organ. The data derived from RA patient samples were compared to those from age- and sex-matched samples of patients with OA as a nonautoimmune control group. To the best of our knowledge, this study is the first to show that Treg enrichment into the joint is not specific to inflammatory arthritis. RA pathophysiology is probably not due to a lack of Tregs, because Treg concentration in the SM of RA was significantly higher than in OA. Furthermore, we show that peripheral and synovial Tregs show significant differences regarding activation status and markers associated with Treg function.
Methods
Study population
The characteristics of the study population are summarised in Table
1. RA and OA were determined according to the 1987 criteria of the American Rheumatism Association [
22]. None of the OA patients had signs of a systemic inflammatory disease based on analysis of CRP and leukocyte counts. All patients were scheduled for knee replacement surgery. The ethics committee of the University of Heidelberg approved this study. Informed consent for participation was obtained from all patients.
Table 1
Characteristics of the study population
a
Number of patients (M/F) | 18 (11/7) | 22 (13/9) |
Age (yr) | 67.5 ± 8 | 66.7 ± 8.3 |
Duration of disease (yr) | 14 ± 7 | 15 ± 6 |
DAS28 score | 4.5 ± 1.2 | n.a. |
RF-positive, n (%) | 10 (55.6) | n.a. |
CRP (mg/L) | 17 ± 9 | 5.1 ± 3 |
ESR (mm/h) | 23 ± 12 | 4 ± 2 |
DMARD, n (%) | 14 (77.8) | 0 |
Anti-TNF treatment, n (%) | 5 (27.8) | 0 |
Glucocorticoids, n (%) | 1 (5.6) | 0 |
NSAID, n (%) | 16 (88.9) | 20 (90.1) |
Sample collection
SF and SM samples from knee joints were collected during knee surgery. SF was removed prior to arthrotomy by needle aspiration into ethylenediaminetetraacetic acid (EDTA)–containing tubes (Sarstedt, Nümbrecht, Germany). SM was taken from the suprapatellar pouch intraoperatively. PB samples were obtained concurrently and collected in EDTA-containing tubes prior to surgery.
Cell preparation
SF samples were treated with bovine testicular hyaluronidase (10 mg/ml; Sigma-Aldrich, St Louis, MO, USA) for 30 minutes at 37°C, and cells were washed twice with phosphate-buffered saline (PBS). Single SM samples were freed from other tissue components and rinsed twice with PBS. SM was then minced finely with scissors and digested with collagenase B (1 mg/ml; Roche Applied Science, Indianapolis, IN, USA) and bovine testicular hyaluronidase IV (2 mg/ml) at 37°C for 2 hours in RPMI 1640 culture medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10 μg/ml penicillin/streptomycin (Invitrogen) 10% foetal calf serum (Biochrom AG Biotechnologie, Berlin, Germany). The cell suspension was filtered through a 100-μm filter (BD Biosciences, San Diego, CA, USA) and a 40-μm-pore-size sieve (EMD Millipore, Billerica, MA, USA) to remove any undigested tissue. The filtered cell suspension was washed twice with PBS. Mononuclear cells (PBMCs, SFMCs and SMMCs) were isolated from EDTA anticoagulated whole blood, SF and SM cell suspension samples using Ficoll-Paque™ PLUS density gradient centrifugation (GE Healthcare Life Sciences, Pittsburgh, PA, USA). T cells were isolated from PB, SF and SM mononuclear cells by MACS bead separation (CD3 MicroBeads; Miltenyi Biotec, San Diego, CA, USA) based on previously described protocols [
13]. The volume and weight of all samples were measured, and total cell numbers were determined after mononuclear cell separation and CD3 isolation.
Flow cytometry analysis
Multicolour flow cytometry was used to identify the T-cell subsets and expression of cell surface markers. All monoclonal antibodies (mAbs) were obtained from BD Biosciences unless stated otherwise. In brief, MACS bead separation–isolated T cells were washed twice in staining buffer, blocked with FcR blocking reagent (Miltenyi Biotec) and then stained for 30 minutes at 4°C with fluorescein isothiocyanate (FITC)–labelled mAb against CD4 (clone RPA-T4), phycoerythrin (PE)-labelled mAb against CD25 (clone M-A251) and peridinin chlorophyll protein complex cychrome 5.5 (PerCP-Cy5.5)–labelled mAb against CD127 (clone RDR5; eBioscience, San Diego, CA, USA). For intracellular staining, cells were stained for cell surface markers with CD4-FITC, CD25-allophycocyanin (CD25-APC) (clone 2A3) and CD127-PerCP-Cy5.5, then fixed and permeabilised with the human FoxP3 buffer set (BD Biosciences) according to the manufacturer’s instructions and stained with PE-labelled mAb against FoxP3 (clone 259D/C7). The analysis of activation markers was performed with PE-labelled mAb specific for one of the following cell surface markers: CD45RA (clone HI100; BioLegend, San Diego, CA, USA), CD45R0 (clone UCHL1; BioLegend), CD69 (clone FN50), CD152 (cytotoxic T-lymphocyte antigen 4 (CTLA-4), clone BN13), CD154 (CD40L; clone 89-76), CD274 (programmed cell death receptor 1 ligand (PD-L1), clone 29E.2A3; BioLegend), CD279 (programmed death receptor 1 (PD-1), clone MIH4) and CD62L (L-selectin, clone DREG-56). After being stained for surface molecules, the cells were washed again and analysed with a FACSCalibur flow cytometer (BD Biosciences). A total of 105 events were collected and analysed using the FlowJo software program (TreeStar, Ashland, OR, USA).
Gating strategy and definition of cell populations
Cells were gated for lymphocytes on the basis of forward and side scatter profiles and further gated for CD4 expression. By using the cell surface markers CD25 and CD127, the Treg population was identified as CD4
+CD25
+/highCD127
-/low Tregs. The cutoff for all cell surface markers was established based on isotype controls. The CD4
+ cells with the highest level of CD25 staining were defined as CD4
+CD25
high cells and appeared as a tail to the right from the major CD4
+ cell population, as previously shown [
8]. The CD4
+CD25
+/highCD127
-/low Treg population was distinct and clearly separable from other cells as previously described [
9‐
11]. For the analysis of activation markers, the CD4
+CD25
+/highCD127
-/low Treg population was gated as described previously and further analysed for each PE-labelled cell surface marker. Expression of activation marker and mean fluorescence intensity (MFI) was determined.
Statistical analysis
The demographic parameters of the OA and RA groups were compared using an unpaired t-test for parametric data (age and body mass index) and the χ2 test for proportions (sex). The unpaired t-test was used for analysis of Treg frequency between OA and RA groups, and the paired t-test was used for comparisons between concurrent PB, SF and SM samples. Differences in the expression of activation markers were analysed by performing the Mann–Whitney U test. Correlation analysis was performed using the Pearson correlation coefficient. All P-values reported herein are two-tailed. A P-value <0.05 was considered to show a statistically significant difference. Statistical analysis was performed using GraphPad Prism version 5 software (GraphPad Software, La Jolla, CA, USA).
Discussion
Data about the frequency of Tregs in PB of RA vary throughout the literature, in which it is described as decreased [
13,
14,
16,
17,
21], similar [
12,
15] and increased [
14,
19] compared to healthy controls. This controversy is partly due to the lack of one specific Treg marker and the resulting differences in the identification of Tregs (CD25
+ vs. CD25
high). The transcription factor FoxP3, which is required for Treg development, has improved the specificity of Treg detection, but its intracellular location does not allow separation of viable cells [
23]. The establishment of CD127 as an additional surface marker, in combination with CD25, has been found to facilitate a consistent quantitative identification of viable CD4
+CD25
+/highCD127
low/- Tregs, which are highly positive for FoxP3 [
9‐
11]. Only one previous study has provided quantitative data regarding CD4
+CD25
+/highCD127
low/- Tregs in the PB of RA patients. The investigators showed lower frequency in active RA and similar Treg frequency in RA patients in remission compared to controls [
21]. In our present study, the mean frequency of CD4
+CD25
+/highCD127
low/- Tregs in PB samples was comparable between RA and OA patients. It seems that the analysis of circulating Tregs has not described the disease-specific parameters consistently enough to draw conclusions. This raises the question whether the analysis of Treg frequency at the site of inflammation might provide a clearer pattern of RA pathology. Researchers in a few studies have analysed the SF of affected RA joints and showed an increase in Treg frequency compared to PB samples taken concurrently from those patients and healthy controls [
12‐
15,
17,
20,
24]. This accumulation into the affected RA joints ran counter the hypothesis of a reduced Treg presence in RA pathophysiology. But due to the lack of SF control samples in these studies, it remained unresolved if the increase of Treg frequencies in RA SF might turn out as a reduced frequency when compared to non-RA SF controls. The inclusion of SF from OA patients in our study shows for the first time that Treg enrichment into the joints is not a feature specific to RA, but is also present in the joints of nonautoimmune OA patients. Furthermore, our data disprove the hypothesis of a lack of Tregs in RA because we show that Treg frequency in SF is comparable between RA and OA patients. The fact that Tregs develop their suppressive activity in a contact-dependent manner, as well as that soluble factors alone are unable to exert suppression, highlights the relevance of further investigation of the SM [
8,
9,
14]. Only two studies have analysed the SM for the presence of FoxP3 by immunohistochemistry and qPCR [
18,
20]. Those two studies have suggested a much lower frequency of Tregs in SM than in SF and PB, favouring the hypothesis that, in RA patients, the SM lacks sufficient Treg numbers and thus these patients would benefit therapeutically from an increase in Tregs. In order to further evaluate this hypothesis, we analysed concurrent samples of SM from RA and OA patients and provide for the first time flow cytometry–derived quantitative data. Our results are contradictory to the common assumption that Treg frequency is lower in SM than in PB. Both groups in our study exhibited significant enrichment of Tregs in the SF and SM compared to concurrent PB. When analysed for absolute Treg numbers, the SM showed a significantly higher Treg concentration than SF in OA and RA patients. Additionally, the comparison to OA samples revealed that Treg concentration is significantly higher in the SM of RA patients. Our results indicate that Treg deficiency is not the underlying mechanism of RA development. Instead, the significantly higher CD4
+ T-cell infiltration of RA joints suggests that Tregs are either counterbalanced by effector T cells in the joint or functionally impaired. Herrath
et al. recently showed that the inflammatory milieu in RA joints reduces the ability of Tregs to cope with the overwhelming number of inflammatory cells [
24]. This favours the hypothesis that the suppressive activity of Tregs can be altered and counterbalanced by activated responder T cells in the joint, which are less susceptible to suppression as their counterparts from PB [
14].
The discrepancy between our results and those reported in the previous two studies might be due to factors such as study population, sample collection and utilization of different techniques. Apart from the older age of our study population, the demographics and clinical parameters of our study are comparable to those of these previous studies [
18,
20]. We received all samples at the time of joint surgery, which differs from the study of Raghavan
et al., who collected SF samples mainly during RA flares [
20]. How this can affect Treg frequency and status remains unknown. In studies comprising longitudinal samples taken from patients with psoriatic arthritis, spondyloarthropathy and juvenile idiopathic arthritis, the frequency of Tregs was found to be relatively stable between flares and remissions [
10,
12,
13]. The two-dimensional character of immunohistochemistry and the analysis of tissue slices provides information about a rather limited cell number. The distribution of infiltrating cell populations appears not to be homogeneous throughout the joint. Regions with high inflammatory infiltrates might lead to an underestimation of Tregs due to an overrepresentation of total CD4
+ T cells. In our present study, flow cytometry was the method of choice because it allows for determination of mean frequency of distinct cell populations and the simultaneous analysis of the expression of multiple surface markers in individual cells. Even qPCR does not allow examination of individual cells and a combination of markers.
One underlying hypothesis about Treg enrichment into the affected joints of RA and other inflammatory diseases is that Tregs are either attracted or generated because of ongoing inflammation [
25]. Because our results demonstrate that Treg enrichment is also evident in OA joints, this hypothesis needs reevaluation. It could be argued that OA pathology is also accompanied by low-level chronic synovial inflammation [
26]. Taking this into account, the accumulated Treg presence in the joints of both patient groups can be understood as an attempt of the immune system to control the inflammatory responses. The faster and much more severe progression of joint destruction in RA could be due to the complex inflammatory environment that might lead to impaired Treg function.
Another hypothesis is that compartmentalization rather than inflammation is the driving force. The physiology of the joint compartment differs from the periphery, and Treg compartmentalization has been shown to be tissue- and organ-specific [
17,
27]. The low correlation between PB, SF and SM Treg frequency in our study suggests that distinct mechanisms contribute to selective retention and trafficking of Tregs and that the SF and SM cannot be expected to mirror the PB. We hypothesize that the joint compartment contains a higher polarized CD4
+ T-cell population and, as such, also a higher Treg population compared to PB, where the majority of T cells are naïve and not yet polarized into subsets. Assuming this is true, future studies need to include the simultaneous analysis of inflammatory T-cell subsets in order to map the compartmentalization of effector and regulatory T cells and analyse their balance in the periphery and the joint.
The fact that the joint compartment differs significantly from the periphery became even more evident when the phenotypic analysis of Tregs was performed. In accord with the results of previous studies, our data confirm that peripheral Tregs display a CD45RO
+CD45RA
- memory phenotype [
12,
28]. This was further increased in SM Tregs, where almost all Tregs presented a memory phenotype. Significant differences between the periphery and the joint were also observed when analysed for activation markers CD69 and CD62L. SM Tregs were CD62L
-CD69
+ activated effector memory cells compared to CD62L
+CD69
- resting central memory Tregs in PB. Significant differences were also seen for markers associated with Treg function. CD152 is thought to be essential for the suppressive ability of Tregs [
29,
30]. Peripheral Tregs constitutively express low surface levels of CD152, with increased and long-lasting expression after stimulation [
6,
7,
14,
15,
30‐
33]. In accord with these findings, our data display low expression of surface CD152 in PB, which was significantly increased in the SM, confirming the activated state of synovium-derived Tregs [
14]. Murine and human studies have attributed a relevant role for GITR in Treg function, where antibody binding abrogates suppression [
34,
35]. Our data are in accord with those of previous studies in which researchers have shown low expression of GITR in the periphery [
14] and a significant increase in synovial Tregs [
6,
7,
31‐
34]. Since the intensity of GITR expression has been correlated with the suppressive capacity of the Treg cells [
33], the higher expression on synovium-derived Tregs could indicate a higher suppressive capacity of synovial Tregs. The programmed death receptor 1 (PD-1) CD279 is involved in the mechanism of cell-cycle arrest [
5,
30,
36,
37]. Its ligand CD274 (PD-L1) was recently shown to deliver a negative signal through the PD-1 receptor, which downregulates T-cell proliferation and cytokine production [
38,
39]. When analysed in PB, the expression level of CD279 or CD274 did not show any significant difference between RA and OA patients, with low expression levels for both. CD274 showed a significant increase in SM Tregs in both groups, whereas CD279 increased only in the SM of RA patients. Further analyses are necessary to evaluate how these expression levels translate into functional differences between Tregs from different compartments and between RA and OA patients.
Competing interests
The University of Heidelberg funded this study. The funding source did not have any involvement in the study design, data collection, analysis and interpretation of data, writing of the manuscript or the decision to submit the manuscript for publication. No benefits in any form have been or will be received from any commercial party related directly or indirectly to the subject of this manuscript. The authors declare that they have no competing interests.
Authors’ contribution
BM and TT were responsible for study conception and design, data acquisition, data analysis and interpretation and manuscript writing. PS was responsible for data acquisition, data analysis and interpretation and critical revision of the manuscript. SH was responsible for data acquisition, data analysis and manuscript writing. NR and TG were responsible for data acquisition and critical revision of the manuscript. JPK and NT were responsible for data analysis and interpretation and critical revision of the manuscript. HML was responsible for study conception and design, data analysis and interpretation and manuscript writing. FZ was responsible for study conception and design, data analysis and interpretation and manuscript writing. All authors read and approved the final manuscript.