We previously described that fibroblast-like cells from the synovium of rheumatoid arthritis patients (RASFib) constitutively express intracellular and surface IL-15, which induces activation of cocultured T cells. Our objective was to study the effect of RASFib IL-15 expression on the function of human CD4+CD25+ regulatory T cells (Treg). RASFib, through their constitutive IL-15 expression, were able to induce the proliferation of human Tregs stimulated through their TCR, and at the same time potentiated their suppressive action on the cytokine secretion of CD4+CD25− responder T cells (Tresp). In parallel, constitutive RASFib IL-15 expression mediated an up-regulated response of Tresp. Subsequently, total CD4+ T cells, containing natural proportions of Treg and Tresp, secreted an increased amount of pathogenic cytokines when cocultured with RASFib despite the presence of proliferating Treg with superior regulatory potency. In summary, RASFib IL-15 exerts a dual action on the equilibrium between Treg and Tresp by potentiating the suppressive effect of Treg while augmenting the proinflammatory action of Tresp; the result is a shift of the Treg/Tresp balance toward a proinflammatory state. This alteration of the Treg/Tresp equilibrium is not observed in the presence of osteoarthritis synovial fibroblasts or dermal fibroblasts, which do not constitutively express surface IL-15. Additionally, Treg with superior suppressive potency were present in the peripheral blood and the synovial fluid of RA patients, but this enhanced immunoregulatory activity was not able to overcome the increased secretion of pathogenic cytokines by RA-Tresp, indicating that rheumatoid arthritis patients demonstrate an altered Treg/Tresp equilibrium in vivo.
The hallmark of rheumatoid arthritis (RA),3 a systemic autoimmune disorder, is a chronic joint inflammation. RA synovitis is characterized by hyperplasia of synovial fibroblasts (RASFib), which plays a pivotal role in the pathogenesis of the disease (1). RASFib actively participate in the erosion of cartilage and bone (1). Additionally, RASFib contribute to creating a tissue microenvironment that favors the initiation and perpetuation of the immune response (2). Direct interactions of RASFib with T cells are important contributors to the chronic RA inflammation: RASFib prevent apoptotic cell death of cocultured T cells (3, 4), up-regulate T cell cytokine and adhesion molecule expression (5), and induce proliferation of T cells upon long-term coculture (6).
Naturally occurring regulatory CD4+CD25+ T cells (Treg) are essential for maintaining peripheral tolerance and controlling autoimmunity (7, 8). Treg are suppressive and anergic to TCR stimulation (7, 8). Suppression by Treg results in decreased proliferation, decreased IL-2, and decreased proinflammatory cytokine production of responder CD4+CD25− T cells (Tresp) (9, 10). The mechanism responsible for the suppressor effect of Treg is not clear and seems to be mediated by direct cell contact with Tresp (7, 8).
A tight control of Treg function is crucial for the immune system. However, the anergy and/or suppressive capacity of Treg can be modified by cytokines (10, 11, 12, 13, 14), by coculture with activated dendritic cells (12, 13, 14, 15, 16, 17), and by TLR ligands (18, 19).
RASFib are a known source of pathogenic cytokines in RA (1). We previously reported that constitutively expressed IL-15 on the surface of RASFib is biologically active on cocultured T lymphocytes through direct cell contact (5). IL-15, a potent T cell growth factor, has been described to drive the proliferation of human Treg (20, 21), to increase human Treg suppressor potency (21), and to interfere with human Treg function (14); additionally, it has been reported that human Treg cocultured with dermal fibroblasts proliferate in the presence but not in the absence of exogenous recombinant IL-15 (22). Therefore, our objective was to investigate the role of RASFib IL-15 expression on the properties of Treg.
Additionally, we were interested in analyzing the characteristics of Treg and Tresp present in the peripheral blood of early RA patients (RAPB) and the synovial fluid of RA patients (RASF). In fact, T cells present in vivo in the synovial fluid of RA patients (RASFTL) are permanently in contact with RASFib lining the intimal layer of the hyperplastic synovial membrane (23), and they demonstrate phenotypical and functional signs of activation (24). T cells circulating in the peripheral blood of RA patients do not show such clear signs of activation, and controversial data exist regarding their phenotypical and functional features (24). Likewise, additional data are needed to clarify the nature of the altered regulatory capacity of RA Treg (25).
Materials and Methods
Synovial membranes were obtained from 10 RA patients undergoing synovectomy or arthroplasty and from 10 osteoarthritis (OA) patients undergoing arthroplasty. Synovial fluid was obtained from 30 patients with established RA who were receiving treatment with oral methotrexate (MTX) and low-dose prednisone. Peripheral blood was obtained from 30 healthy controls and from 30 early RA patients fulfilling at least four American College of Rheumatology criteria (26), who had never received disease-modifying drugs or corticosteroids and with a disease duration of <6 mo. La Paz University Hospital in Madrid, Spain, has a monographic clinic that takes care of early arthritis patients referred from a wide primary care area. This opportunity facilitated recruitment of untreated early RA patients for the present study. Among early RA patients there were 9 men and 21 women, 23 (76%) of whom tested positive for IgM rheumatoid factor; their ages were 19–82 years (mean, 52.72; SD, 17.23; median, 49 years), duration of symptoms at first evaluation was from 2 to 24 wk (mean, 10.35; SD, 6.5; median, 10 wk), and disease activity score 28 (DAS28) (27) at first evaluation was from 4.75 to 7.64 (mean, 5.93; SD, 0.75; median, 5.72).
Ten patients with early RA donated blood for a second time, once disease activity had been controlled. These patients were receiving 15 mg of oral MTX weekly, except for one patient who was taking 20 mg per week and one patient who was taking 25 mg per week. Additionally, all 10 patients except for 1 were taking prednisone, 2.5 mg daily. IgM rheumatoid factor was positive at initial evaluation in 8 of these 10 patients; there were 2 men and 8 women, with ages of 40.4 ± 12.54 years (mean ± SD). DAS28 before initiation of treatment was 5.76 ± 0.81 (mean ± SD) and at the time of second blood drawing it was 1.93 ± 0.5. These patients were in remission as defined by a DAS28 score <2.6 (28). The study was approved by the Hospital Ethics Committee.
Culture of human fibroblasts
RASFib and OASFib were obtained by collagenase digestion (type I; Worthington Biochemical) of human synovial tissue obtained at arthroplasty or synovectomy. Dermal fibroblasts were obtained by collagenase digestion of normal skin obtained from punch biopsies of five healthy volunteers. Cells were plated in 75-cm2 flasks (Corning Costar) and grown in DMEM (Lonza) supplemented with 10% FCS (Lonza), 2 mM l-glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin (Lonza). Cells were passaged at 1/2 dilution when reaching 95% confluence, by gentle trypsinization (0.05% trypsin/0.53 mM EDTA; Invitrogen). Fibroblasts were used between the third and fifth passages. At this time, they appear to be a homogeneous population of fibroblast-like cells that stain positive with anti-Thy-1 (CD90) Ab (29) and are negative for the expression of CD1, CD3, CD19, CD14, HLA class II, CD80, and CD86, as determined by flow cytometry and fluorochrome-conjugated mAbs (BD Pharmingen).
Isolation of CD4+ T cells and purification of CD4+CD25− and CD4+CD25+ T cell fractions
Mononuclear cells were isolated from human blood by density centrifugation over Ficoll-Paque Plus (GE Healthcare). CD4+ T cells were subsequently purified by negative selection in an autoMACS (Miltenyi Biotec) using a CD4+ T cell isolation kit II from Miltenyi Biotec, containing a cocktail of biotin-conjugated Abs against CD8, CD14, CD16, CD19, CD36, CD56, CD123, TCRγ/δ, and Glycophorin A, followed by anti-biotin microbeads. Isolated CD4+ T cells were 98% pure and free of detectable CD14+ monocytes, CD56+ NK cells, and CD19+ B cells. Subsequently, CD4+CD25+-bright T cells were positively selected using a limiting amount of anti-CD25 microbeads (5 μl/107 CD4+ cells) (Miltenyi Biotec) in an autoMACS. CD4+CD25− T cells were subsequently depleted of any contaminating CD25+ T cells following a second round of incubation with CD25+ microbeads (20 μl/107 CD4+ T cells). T cell subpopulations of CD4+CD25+ T cells and CD4+CD25− T cells were used immediately after isolation. In separate experiments, the CD4+CD25−CD127+ Tresp and CD4+CD25+CD127− Treg populations were isolated from total CD4+ T cells by FACS sorting on a BD Vantage SE (BD Biosciences). These cells were stained with CD4-PerCP-Cy5, CD25-FITC, and CD127-PE (all from BD Pharmingen). An Alexa Fluor 647-labeled Ab against the intracellular Ag FoxP3 (clone 236A/E; eBioscience) was used to assess purity of the isolated Treg population.
T cell stimulation
T cells were stimulated in flat-bottom 96-well plates (105 cells/well) using two different anti-CD3 mAbs in two different settings: a plate-bound anti-CD3 IgG subclass mAb (UCTH1; BD Pharmingen) and a soluble anti-CD3 IgE subclass (T3/4.E; Sanquin, formerly CLB). It is known that anti-CD3 Abs, when bound to plastic culture surfaces, facilitate crosslinking that induces T cell proliferation. In contrast, soluble anti-CD3 Abs do not induce T cell proliferation in the absence of APCs or anti-CD28 Abs. A remarkable exception is an IgE anti-CD3 Ab that was developed at the CLB as a switch variant originating from an IgG1 anti-CD3 MoAb (CLB-T3/4.E). Soluble CLB-T3/4.E is able to stimulate T cells directly, without the need for a second signal (30), and it allows the comparison of stimulated T cell responses in the absence or presence of RASFib. No accessory cells and no costimulatory Abs were used in an attempt to induce a mild stimulation of T cells and to avoid the use of agents that may interfere with Treg function.
T cell proliferation assays
T cell proliferation was assessed by [3H]thymidine incorporation and by CFSE dilution. For thymidine incorporation assays, 18 h before the termination of the cultures, the plates were pulsed with 0.5 μCi/well [3H]thymidine (GE Healthcare). The cells were harvested on paper filters and [3H]thymidine uptake was measured in a liquid scintillation counter. Additionally, and to track cell division by flow cytometry, cells were labeled before initiating culture with CFSE (Molecular Probes/Invitrogen), at a final concentration of 8 μM.
Coculture of T cells with RASFib
Coculture experiments were performed with confluent fibroblast cultures prepared 48 h before contact. All experimental conditions were performed in triplicate and variation between replicates was <5%. Fibroblasts were seeded in 96-well plates (Corning) at 1 × 104 cells per well. Forty-eight hours later, CFSE-labeled total CD4+ T cells (TCD4), Tresp, or Treg (105 cells/well) were added and the medium was supplemented with anti-CD3IgE mAb (0.25 μg/ml). Cocultures were maintained for 5 days.
A 0.4-μm Transwell system (Corning) was used to conduct some coculture experiments. The system consists of two compartments: a top well, with a porous matrix (0.4 μm), and a bottom well. This setup allows coculture of two types of cells to grow in the same medium with soluble factors exchanged through the pores, while preventing direct contact between them. RASFib were grown to confluence in the bottom well, and T cells were added either to the same well, allowing contact, or in the top well, avoiding contact.
Assessment of Treg function
The function of naturally occurring CD4+CD25+ Treg present at physiological rates in human peripheral blood was assessed by comparing the proliferation and IFN-γ and TNF-α secretion of anti-CD3-stimulated total CD4+ T cells (TCD4T) against CD25+-depleted CD4+ T cells (Tresp), in the presence or absence of RASFib. The potency of inhibition exerted by Treg was calculated as: [1 − (proliferation or cytokine secretion of total CD4+ T cells/proliferation or cytokine secretion of CD25− T cells)]. In some experiments, and to determine the per cell suppressive potency of Treg, isolated Treg were cocultured with Tresp at different Treg-to-Tresp ratios (1:1, 1:2, 1:5, 1:10, and 1:20); potency of suppression was then calculated as: [1 − (proliferation or cytokine secretion of Treg + Tresp coculture/proliferation or cytokine secretion of Tresp)].
Inhibition by neutralizing mAbs
Parallel experiments included the addition of IL-15 blocking agents: a neutralizing anti-IL-15 mAb (10 μg/ml) (mab247, mouse IgG1; R&D Systems), a recombinant IL-15Rα/human Fcγ1 chimera (100 ng/ml; R&D Systems), and an IL-15 mutant/murine Fcγ2a chimera (10 μg/ml; Chimerigen); the latter is a high-affinity receptor site-specific antagonist for IL-15Rα (31). Alternatively, the binding control anti-HLA class I (clone W6/32, 10 μg/ml; Sigma-Aldrich), an anti-human MHC class II (HLA-DR, -DP, -DQ, 10 μg/ml; BD Biosciences), or appropriate isotype controls (murine IgG1, murine IgG2a, or human IgG1; all from R&D Systems) were used. These Abs were incubated with RASFib for 30 min at 37°C; T cells were subsequently added without washing.
Intracellular cytokine staining, surface staining, and flow cytometry
For intracellular FoxP3 staining, T cells were washed with PBS/2% FCS/0.01% NaN3, permeabilized for 10 min with FACS permeabilizing solution 2 (BD Pharmingen), washed again, and incubated on ice for 1 h with an Alexa Fluor 647-labeled anti-FoxP3 mAb (clone 236A/E7; eBioscience) or an irrelevant isotype control mAb. After washing once with PBS/2% FCS/0.01% NaN3 and once with PBS, cells were resuspended in 1% paraformaldehyde and analyzed in a FACSCalibur flow cytometer using CellQuest software (BD Biosciences). For staining of surface molecules, the permeabilization step was omitted. Surface IL-15Rα was detected with anti-IL-15Rα mAb (R&D Systems), followed by an Alexa Fluor 488-goat anti-mouse Ab (Molecular Probes/Invitrogen). Fluorochrome-conjugated mAbs from BD Pharmingen were used to examine the expression of the phenotypic markers CD3, CD4, CD25, CD127, and CD69. Mean fluorescence intensity (MFI) is calculated as the difference between the MFI of tested cells and the MFI of background staining.
Soluble cytokine detection
ELISAs for IFN-γ and TNF-α were performed in cell-free supernatants using kits from BD Pharmingen.
Total cellular RNA was isolated using the RNAqueous kit (Ambion/Applied Biosystems) with DNase treatment. For each sample, 1 μg of total RNA was subjected to reverse transcription using an Advantage RT for PCR kit (Clontech). Aliquots (1 μl) of the reverse transcription products were used for quantitative PCR in the LightCycler PCR and detection system (Roche Molecular Biochemicals) with the FastStart DNA Master SYBR Green I kit (Roche Diagnostics). The PCR reactions were set up in microcapillary tubes in a volume of 20 μl. The following sense and antisense primers were used: FoxP3 sense, 5′-CAC CTA CGC CAC GCT CATC-3′ and antisense, 5′-ACT CAG GTT GTG GCG GAT G-3′; IL-2Rα (CD25) sense, 5′-TCA GTG CGT CCA GGG ATA CAG G-3′ and antisense, 5′-TCA GGA GGA GGA CGC TGA TCA G-3′; IL-7Rα (CD127) sense, 5′-TCT GGA GAA AGT GGC TAT GC-3′ and antisense, 5′-CCT GGC GGT AAG CTA CAT C-3′. IL-15Rα primers are located in exon 2 (sense) and exon 5 (antisense) to amplify mRNA encoding all IL-15Rα variants that contain exon 2, and thus, bind IL-15: IL-15Rα sense, 5′-GGA ATT CAT CAC GTG CCC TCC CCC CAT G-3′ and antisense, 5′-CGG GAT CCT CAA GTG GTG TCG CTG TGG CCC TG-3′ (product size, 543/444). Amplification of the PCR products was monitored by measuring SYBR Green I dye fluorescence. As an external standard, the transcript of 18S rRNA was amplified from the same cDNA samples using primers manufactured by Ambion. Each sample was run in triplicate. Results were analyzed with LightCycler version 3.5.3 software (Roche Diagnostics). Quantities of specific mRNA in the sample were measured according to the corresponding gene-specific standard curve.
Comparison between groups was by a Mann-Whitney U test. Paired samples were compared using a Wilcoxon matched pairs signed rank sum test. When appropriate, Bonferroni correction for multiple comparisons was applied.
Phenotype and function of human peripheral blood CD4+CD25+ Treg and CD4+CD25− Tresp
Flow cytometry and quantitative RT-PCR demonstrated that CD4+CD25+ Treg isolated with limiting amounts of magnetic beads expressed high levels of FoxP3 in sharp contrast with isolated CD4+CD25− Tresp (Fig. 1, A and B). Additionally, a remarkable CD127 (IL-7Rα) expression was observed on Tresp but not on Treg (Fig. 1, A and B), further confirming the previously described characteristic phenotype of both populations (32, 33). Furthermore, we observed for the first time that IL-15Rα expression is significantly higher on Treg when compared with Tresp (Fig. 1, A and B), suggesting a role for IL-15 on human Treg biology.
The gold standard for Treg identification is functional proof of their inhibitory capacity, since all of the as yet available phenotypical markers can be up-regulated upon T cell activation (34). Therefore, we next sought to examine the function of naturally occurring CD4+CD25+ Treg present at physiological rates in human peripheral blood. To this end, the proliferation and IFN-γ and TNF-α secretion of anti-CD3-stimulated TCD4T were compared against CD25+-depleted CD4+ T cells (Tresp) (Fig. 1,C). In the presence of plate-bound anti-CD3 IgG (UCHT1 at 0.25 μg/ml; BD Pharmingen) the proliferation and cytokine secretion of Tresp was significantly higher when compared with TCD4T (Fig. 1,C). This indicates that the physiological proportion of Treg present in human peripheral blood has a significant regulatory activity. Comparable results were obtained when cells were stimulated with soluble anti-CD3 IgE (CLB-T3/4.E, 0.25 μg/ml) (Fig. 1,C), consistent with previous observations by van Lier et al. (30). As expected, magnetically sorted CD25+ Treg did not secrete IFN- γ or TNF-α and were anergic to TCR stimulation, that is, they did not proliferate when stimulated with anti-CD3 mAbs (Fig. 1 C).
Additionally, the per cell potency of Treg isolated using limiting amounts of magnetic beads was tested (Fig. 1,D). To this end, Treg were cocultured with Tresp at different Treg/Tresp ratios in the presence of a submaximal stimulus (soluble anti-CD3 IgE, CLB-T3/4.E, 0.25 μg/ml). An optimal suppressive capacity and a dose-response curve could be observed (Fig. 1 D). Preliminary experiments demonstrated that the inhibitory function of Treg that had been isolated by flow cytometry was not significantly different from that of Treg isolated with limiting amounts of magnetic beads.
In the presence of RASFib, Treg-mediated suppression of Tresp proliferation is not apparent, while suppression of IFN-γ and TNF-α secretion is enhanced
We then sought to establish whether RASFib are able to modify the equilibrium between Treg and Tresp. To this end, CFSE-labeled TCD4+ T cells, Tresp, or Treg were cultured in the absence or presence of RASFib. When compared against pure T cell cultures, coculture with RASFib resulted in increased proliferative responses to soluble anti-CD3 IgE mAb for both TCD4T and isolated Tresp (Fig. 2,A); in this setting, the proliferation rate of Tresp was not significantly different from TCD4+ T cells (Fig. 2,A): in other words, the suppressive effect of natural proportions of Treg on Tresp proliferation was not apparent. Remarkably, proliferation of anti-CD3-stimulated Treg was observed in the presence of RASFib (Fig. 2 A).
We then examined the effect of RASFib on anti-CD3-stimulated IFN-γ and TNF-α secretion. In the presence of RASFib, IFN-γ and TNF-α secretion were significantly increased when compared with pure T cell cultures for both Tresp and TCD4T (Fig. 2,B). At the same time, IFN-γ and TNF-α secretion of Tresp remained significantly higher when compared with TCD4+ T cells (Fig. 2,B), indicating that the suppressive action of Treg on Tresp cytokine secretion is indeed maintained. Furthermore, the calculated potency of inhibition on cytokine secretion was significantly higher in the presence of RASFib (Fig. 2 C).
Importantly, the above-reported effects were not observed in the presence of OA synovial fibroblasts or dermal fibroblasts. Furthermore, these effects were not apparent when cocultures were established in the presence of 0.4 μM Transwell inserts, which do not allow direct contact between RASFib and T cells. Additionally, these effects were not seen when Treg, Tresp, or TCD4T were cultured in the presence of supernatants taken from cocultures of RASFib with Treg, Tresp, or TCD4T.
Of note, T cells did not proliferate and did not secrete IFN-γ or TNF-α when stimulated with soluble IgG anti-CD3 mAb (clone UCTH1; BD Pharmingen) in the presence of RASFib, indicating that RASFib are not expressing molecules with functional costimulatory capacity.
FoxP3 expression is up-regulated upon T cell activation
It has been reported that activated human CD4+ Tresp may undergo an up-regulation of FoxP3 expression that is not sufficient to induce Treg activity (35, 36, 37, 38). The functional studies we described in the previous section indicate that Treg are not induced de novo in our system since activated Tresp produced high levels of effector cytokines TNF-α and IFN-γ when compared with TCD4T and as opposed to activated Treg, which did not secrete pathogenic cytokines. Nevertheless, we were interested in examining FoxP3 expression in these T cells.
Dot plots of CFSE-labeled TCD4T, Tresp, or Treg double-stained with an allophycocyanin-labeled anti-FoxP3 MoAb confirmed that FoxP3 expression is up-regulated upon activation (Fig. 3). This up-regulation was transient, reached a maximum on the fifth day, and then progressively declined to return to basal expression levels around the ninth day. In the presence of RASFib, up-regulation was observed not only on activated TCD4T and Tresp but also on proliferating Treg (Fig. 3,B). Subsequently, the MFI of FoxP3 expression remained higher among proliferating isolated Treg cocultured with RASFib when compared with proliferating Tresp (Fig. 3,B), which is consistent with previously reported observations (35). Remarkably, two populations of Foxp3-expressing cells were observed among stimulated TCD4T in the presence, but not in the absence, of RASFib: the majority of cells expressed an up-regulated FoxP3 level with an intensity that was comparable to that observed in stimulated Tresp; a minor population (∼10%) demonstrated a higher Foxp3 expression, similar to that observed in stimulated Treg (Fig. 3,B). In contrast, FoxP3 expression was homogeneous among proliferating Tresp (Fig. 3 B). These data suggest that the Treg population present at physiological proportions in human peripheral blood proliferates when TCD4T are stimulated with anti-CD3 in the presence but not in the absence of RASFib.
Effect of RASFib on the per cell suppressive action of Treg
The increased Treg suppressive potency on cytokine secretion observed in the presence of RASFib may be attributable, at least in part, to increased Treg proportions secondary to Treg proliferation. Terefore, we next sought to examine the effect of RASFib on the per cell suppressive potency of Treg.
To this end, TCD4T from six healthy donors were cultured in the presence or absence of RASFib as described above, harvested on the fifth day of culture, and sorted by flow cytometry to isolate CD25+/bright cells. The regulatory potency of these CD25+/bright cells was tested in suppression assays free of RASFib, using freshly isolated autologous CD4+CD25− cells as Tresp and following the protocol described in the preceding sections. Freshly isolated autologous CD4+CD25+ Treg served as controls. Accordingly, the following autologous combinations were assayed: (1) Treg precultured in the presence of RASFib with freshly isolated Tresp; (2) Treg precultured in the absence of RASFib with freshly isolated Tresp; and (3) freshly isolated Treg with freshly isolated Tresp. Proliferation was determined by [3H]thymidine uptake, and cytokine secretion was assayed by ELISA.
The suppressive potency of Treg that had been precultured in the presence of RASFib was significantly higher when compared with Treg precultured in the absence of RASFib and with freshly isolated Tregs (Fig. 4). The potency of the two latter groups of Treg was comparable (Fig. 4). Remarkably, the increased suppressive potency induced by RASFib was apparent not only on Tresp cytokine secretion but also on Tresp proliferation (Fig. 4). That is, although Treg suppression of Tresp proliferation is not apparent in the presence of RASFib, these Treg remain potent suppressors of proliferation when harvested and re-tested in the absence of RASFib. This phenomenon is comparable to the reported effect of dendritic cells on Treg function (15).
As an additional control, CD25bright cells were isolated from CD4+CD25− Tresp cultured for 5 days in the presence or absence of RASFib. When tested in suppression assays with freshly isolated autologous CD25− Tresp, these CD25bright T cells did not demonstrate any regulatory capacity, indicating that Treg are not generated de novo in this system.
Neutralization of IL-15 significantly attenuates the effect of RASFib on the equilibrium between Treg and Tresp
We have previously reported that constitutively expressed IL-15 on the surface of RASFib is biologically active on cocultured T lymphocytes through direct cell contact (5). Therefore, it was important to investigate the role of IL-15 as a mediator of the observed action of RASFib on Treg function. To this end, a panel of IL-15-neutralizing agents was tested in cocultures of fibroblasts with TCD4T, Treg, or Tresp.
The up-regulated responses of TCD4T and of Tresp observed in coculture with RASFib were significantly attenuated by IL-15 blocking agents (Fig. 5, A and B) but not by isotype control or binding control anti-HLA class I (W6/32) or by anti-HLA class II (DR, DP, DQ) Abs. Furthermore, in this setting the proliferation rate of Tresp was significantly higher when compared with TCD4T (Fig. 5, A and B). That is, in coculture with RASFib and in the presence of IL-15-neutralizing agents, the inhibitory effect of Treg on Tresp proliferation was readily observed. Additionally, neutralization of IL-15 abrogated the preproliferative effect of RASFib on Treg (Fig. 5,B). In parallel, neutralization of IL-15 reversed the effect of RASFib on TCD4T and Tresp IFN-γ and TNF-α secretion (Fig. 5,B) and restored the potency of cytokine inhibition to that observed in the absence of RASFib (Fig. 5 C). This indicates that constitutive RASFib IL-15 expression plays an important role in driving the Tresp and Treg proliferation observed in T cell/RASFib cocultures, and in augmenting the inhibitory potency of Treg on Tresp cytokine secretion.
Increased proportions of CD4+CD25+CD127−FoxP3+ T cells are found in the peripheral blood of early RA patients and in the synovial fluid of RA patients
We were next interested in analyzing the characteristics of Treg and Tresp present in the peripheral blood of early RA (RAPB) and the synovial fluid of RA patients (RASF), and studying their behavior when stimulated through their TCR in the presence or absence of RASFib. Data were compared against Treg and Tresp from the peripheral blood of healthy controls (HCPB). Additionally, Treg and Tresp from early RA patients who achieved disease remission after follow-up were also studied (RAPB-R).
When compared with HCPB CD4+ T cells (n = 30), the proportion of CD25+CD127− T cells was significantly increased among RAPB CD4+ T cells (n = 30) (mean ± SD, 9.8 ± 2.7 vs 6.4 ± 2.1, p < 0,05; range, 8.9–16.3 vs 4.1–8.6) (Fig. 6,A). These RAPB CD4+CD25+CD127− T cells expressed FoxP3 together with high levels of IL-15Rα, but they did not express the activation marker CD69 (Fig. 6, B and C). There was no difference in the amount of FoxP3 expression between RAPB CD4+CD25+CD127− and HCPB CD4+CD25+CD127− T cells, as determined by flow cytometry (Fig. 6,B). Importantly, reexamination of 10 early RA patients at 12 mo, once remission of disease activity had been achieved with treatment (methotrexate and low-dose prednisone), demonstrated that the proportion of circulating CD4+CD25+CD127− T cells was no longer different from that of healthy controls (Fig. 6 A).
The proportion of CD25+CD127− T cells was even higher among synovial fluid CD4+ T cells of RA patients (n = 30) (20.5 ± 5.4; p < 0.01 vs RAPB and vs HCPB) (Fig. 6,A). Additionally, the expression level of FoxP3 among these cells was significantly higher when compared with HCPB and RAPB CD4+CD25−CD127+ T cells (p < 0.05) (Fig. 6,B). Interestingly, most of the RASF CD4+CD25+CD127− T cells (>95%) were CD69-negative, indicating that these cells are probably not activated Tresp (Fig. 6 C).
Treg from the synovial fluid of RA and the peripheral blood of early RA patients demonstrate an increased suppressive capacity
Because the gold standard for Treg identification is demonstration of their inhibitory capacity (34), we next examined the inhibitory function of Treg in RA. The results were compared among the above mentioned four groups: HCPB, RAPB, RAPB-R, and RASF T cells. Inhibition assays were performed following two different approaches: (1) the regulatory function of natural proportions of Treg in RA was inferred by comparing the responses of anti-CD3-stimulated total CD4+ vs CD25+ depleted CD4+ T cells, and (2) additionally, the per cell potency of RATreg was assessed in cocultures of magnetically sorted CD4+CD25+ Treg with CD4+CD25− Tresp, established at different Treg/Tresp ratios.
We thought it interesting to first test the suppressive function of the naturally increased proportion of CD4+CD25+FoxP3+ T cells present among RAPB and RASF CD4+ T cells. Of note, TCD4T from RAPB and RASF showed a lower proliferative response to anti-CD3 mAbs when compared with TCD4T from HCPB (Fig. 7,A). In contrast, RAPB and RASF CD4+CD25− Tresp demonstrated higher proliferation rates when compared with HCPB Tresp (Fig. 7,A). Subsequently, the calculated inhibitory potency of Treg on Tresp proliferation was significantly higher for both RAPB and RASF cells when compared with HCPB (Fig. 7 D). This suggests that the low proliferative responses of RAPB and RASF TCD4T in response to anti-CD3 mAbs are attributable to the superior suppressive potency of RA Treg.
In parallel, IFN-γ and TNF-α secretion were higher for TCD4T and also for Tresp isolated from both RAPB and RASF, when compared with the corresponding populations isolated from HCPB (Fig. 7, B and C). That is, despite their low proliferative response, stimulated RAPB and RASF TCD4T secrete more IFN-γ and TNF-α than do HCPB TCD4T. At the same time, the potency of inhibition attributable to the action of Treg on Tresp cytokine secretion was significantly increased for both RAPB- and RASF-derived T cells, when compared with HCPB (Fig. 7 D).
When early RA patients were reexamined after disease activity had been controlled with treatment (n = 10), the functional behavior of total and CD25+-depleted CD4 T cells together with the calculated potency of Treg suppression were not different from healthy controls (Fig. 7).
The observed increased suppressive action of Treg in RA may be related, at least in part, to the higher proportion of CD4+CD25+FoxP3+ T cells present in the peripheral blood and the synovial fluid of RA patients. Therefore, we next sought to determine the suppressive potency of RA Treg on a per cell basis. To this end, isolated CD4+CD25+FoxP3+ T cells (Treg) were cocultured with CD4+CD25− T cells (Tresp) at different Treg-to-Tresp ratios (1:1, 1:2, 1:5, 1:10, and 1:20). At the highest tested ratio (1:1), Treg from all four studied groups were highly suppressive of Tresp proliferation and IFN-γ and TNF-α secretion, and no significant differences were detected among the groups (Fig. 8,A). At a 1:2 and 1:5 ratio, RASF Treg were significantly more suppressive than RAPB Treg, HCPB Treg, and RAPB-R Treg, while no differences were observed among the latter three sets of subjects (Fig. 8,A). At 1:10 and 1:20 ratios, the suppressive potency of RAPB Treg was significantly higher when compared with HCPB Treg, as was the suppresive potency of RASF Treg (Fig. 8 A). This indicates that not only RASF but also RAPB Treg have an increased suppressive capacity on a per cell basis when compared with HCPB Treg. We can only speculate that in vivo activation of RA Treg is responsible for this increased suppressive capacity. The properties of RAPB-R Treg were not different from HCPB Tregs.
Next, isolated RA Treg were cocultured with isolated RA Tresp at a 1:10 Treg-to-Tresp ratio in the presence of RASFib. The effect of RASFib on RAPB and RASF Treg function was similar to that exerted on HCPB Treg properties. In this setting, inhibition of RA Tresp proliferation was not apparent, while inhibition of RA Tresp IFN-γ and TNF-α secretion were maintained with superior potency (Fig. 8,B). Neutralization of IL-15 significantly down-modulated the effect of RASFib (Fig. 8 B).
We have herein shown that in the presence of RASFib, the anergic state of human CD4+CD25+ Treg is abrogated, and their suppressive action on CD4+CD25− Tresp proliferation is not apparent; at the same time, Treg inhibition of Tresp proinflammatory cytokine secretion is not only preserved but potentiated. In parallel, RASFib mediate an up-regulated TNF-α and IFN-γ response of CD4+CD25− Tresp. Subsequently, total CD4+ T cells, containing natural proportions of Treg and Tresp, secrete an increased amount of pathogenic cytokines when cocultured with RASFib, despite the presence of proliferating Treg with superior regulatory potency. In summary, RASFib exert a dual action on the functional equilibrium between Treg and Tresp, by potentiating the suppressive effect of Treg while augmenting the proinflammatory action of Tresp; the result is a shift of the Treg/Tresp balance toward a proinflammatory state.
Experiments with IL-15-neutralizing agents and with Transwell inserts indicate that surface RASFib IL-15 plays a pivotal role in this effect. Of note, this alteration of the Treg/Tresp equilibrium is not observed in the presence of OASFib or dermal fibroblasts, which do not constitutively express surface IL-15 (5). Experiments with neutralizing anti-MHC class I and class II Abs ruled out an implication of Ag presentation in this effect. Consistent with previous reports (35, 36, 37, 38), we observed an induced FoxP3 expression in stimulated Tresp that did not seem to be associated with acquisition of regulatory properties. Additionally, proliferating Tregs in the presence of RASFib demonstrated an up-regulated FoxP3 expression together with a superior regulatory potency. This is in agreement with published data indicating that stimulated Treg show an enhanced suppressive capacity, and that their increased FoxP3 expression remains higher when compared with induced FoxP3 levels in stimulated Tresp (35). Importantly, when Treg that had been cocultured with RASFib were harvested and retested in suppression assays free of RASFib, they demonstrated a superior per cell suppressive potency not only on Tresp cytokine secretion but also on Tresp proliferation.
Other cell lineages have been described to modify the properties of Tregs in a comparable manner (12, 13, 15, 16, 17). Direct interaction with mature, activated bone marrow-derived dendritic cells, but not with splenic dendritic cells expressing equivalent levels of costimulatory molecules, reverts the anergic state of Treg (15, 16) and interferes with Treg inhibition of Tresp proliferation while preserving their suppressive action on Tresp IL-2 secretion (15). However, when Treg that have been cocultured with dendritic cells are harvested and tested in suppression assays free of dendritic cells, they demonstrate a potent suppressor action on Tresp proliferation (15). Of note, B cells are unable to expand Treg despite their potent APC function (15, 16). In fact, the role of costimulation on the bone marrow-derived dendritic cell-induced proliferation of Treg is controversial; alternatively, proinflammatory cytokines have been presented as strong candidates (12, 13, 17). Additionally, cytokines derived from activated monocytes present in the RA joint have been reported to interfere with Treg function (39).
IL-15, initially described as a T cell growth factor (40, 41), acts through a heterotrimeric receptor consisting of a specific high-affinity binding α-chain (IL-15Rα) plus the IL-2 receptor signaling subunits β- and common γ-chain (42, 43). Subsequently, IL-2 and IL-15 share several effects. Both stimulate the proliferation of activated CD4+ and CD8+ T cells, of B cells stimulated through their BCR, and of NK cells (42, 43). Additionally, IL-2 and IL-15 have some divergent effects: IL-2 promotes the elimination of activated T cells (44, 45, 46), whereas IL-15 suppresses T cell apoptosis (47). In mice, the development and survival of CD4+CD25+ Tregs is highly dependent on IL-2 but is independent of IL-15 (48).
CD4+CD25+ Treg potently suppress both proliferation and cytokine secretion of CD4+CD25− Tresp (7, 8, 9). A hallmark of CD4+CD25+ Tregs is their inability to proliferate themselves when stimulated through the TCR (7, 8, 9), because they lack the capacity to transcribe IL-2 (11). Nevertheless, murine CD4+CD25+ Treg do readily proliferate when stimulated via their TCR in the presence of the exogenous cytokines IL-2 or IL-4, but not in the presence of IL-6, IL-7, IL-9, IL-10, or IL-15 (48). Treg proliferation triggered by IL-2 or IL-4 interferes with the inhibition of Tresp proliferation (11); interestingly, at the same time, Treg suppression of IL-2 production is not only spared but significantly enhanced (11). In contrast with murine Treg (48), human CD4+CD25+ Treg proliferate and acquire an enhanced suppressor capacity when stimulated via their TCR in the presence of exogenous IL-15 (20, 21); in fact, Koenen et al. reported that IL-15 is a superior growth factor for human Treg (21). Additionally, we have herein described for the first time that IL-15Rα is highly expressed on human Treg, suggesting a role for IL-15 in Treg biology.
IL-15 is expressed intracellularly by nonlymphoid cells such as monocyte-macrophages (42, 43), by dendritic cells (42, 43), and also by fibroblasts (43, 49). Additionally, whereas IL-2 acts in a soluble form, IL-15 is expressed on the cell surface and acts through direct cell contact (43). Surface IL-15 is constitutively present and physiologically active on fibroblasts from some locations. IL-15 on fibroblasts from human spleen regulates NK cell differentiation from blood CD34+ progenitors (50), and IL-15 on bone marrow fibroblast-like stromal cells contributes to T cell recruitment and expansion in aplastic anemia (51). We previously described that constitutively expressed IL-15 on the surface of RASFib induces T cell activation and cytokine secretion through direct cell contact (5); in contrast, resting OASFib or dermal fibroblasts do not activate T cells, consistent with their lack of constitutive surface IL-15 expression (5). This illustrates how, as described by C. D. Buckley, synovial fibroblasts are key players in favoring the persistence of the inflammatory joint infiltrate (52) and orchestrate the switch from acute resolving to chronic persistent arthritis. At the same time, it stresses the fact that “fibroblasts from different anatomical sites have diverse phenotype and function, secrete distinct patterns of cytokines and chemokines, and condition the intrinsic susceptibility of the organs to inflammatory insults” (2, 52). In the present study we have extended our previous observations by showing that RASFib IL-15 expression is able to break the anergic state of CD4+CD25+ Treg and modify the Treg/Tresp equilibrium. This effect is not apparent in the presence of OASFib or dermal fibroblasts that do not express surface IL-15. In agreement with these findings, Clark and Kupper have described that human Treg cocultured with dermal fibroblasts do not proliferate unless exogenous IL-15 is added to the medium (22).
Additionally, our data indicate that increased proportions of CD4+CD25+CD127− Treg with superior suppressive potency are present in the peripheral blood and the synovial fluid of RA patients. However, this enhanced regulatory activity is not able to overcome the increased secretion of proinflammatory mediators released by preactivated CD4+CD25− Tresp. That is, RA patients demonstrate an altered Treg/Tresp equilibrium in vivo.
It is widely accepted that the frequency and the suppressor potency of CD4+CD25+ Treg found in the synovial fluid of RA patients are significantly elevated when compared with peripheral blood (25, 53, 54, 55, 56), and our results are consistent with previous observations. A question that has been asked by several investigators (25, 53, 54) is why an active joint inflammation persists in RA despite the presence of elevated frequencies of CD4+CD25+ T cells with an enhanced suppressive capacity. Our experimental results help to shed some light on this matter. In vivo, RASFTL are in contact with the lining layer of the synovial membrane where hyperplastic synovial fibroblasts are abundant (23). The proliferation of the RA synovial membrane leads to the formation of villi, and this enormously augments the contact surface between the lining layer of the synovial membrane on one side and the synovial fluid and its components on the opposite side (23). Salmon et al. (4) demonstrated that decreased apoptosis of RA synovial fluid T cells in vivo is due to direct contact with RA synovial fibroblasts within the joint space. Likewise, the data presented herein suggest that in vivo, direct contact of T cells with RASFib may induce an alteration of the Treg/Tresp equilibrium similar to that observed in our ex vivo experiments, and are important contributors to the increased potency of Treg present in RA synovial fluid.
In contrast, previous studies on the frequency and on the functional capacity of circulating CD4+CD25+ Treg in RA have yielded conflicting results (10, 53, 57, 58). In comparison with healthy controls, increased (53) as well as decreased (58) or equivalent (57) PB Treg frequencies have been reported by different authors. Additionally, comparable (58) as well as decreased (10, 57) regulatory activities relative to HCPB Tregs, have been described in RA patients; furthermore, reported data on the nature of the PB Treg dysfunction in RA are conflicting: both an isolated alteration of cytokine suppression (57) and a combined impairment of Tresp proliferation and cytokine suppression have been observed (10).
Several factors may account for these divergent results, including criteria and methods chosen to define and isolate Treg, experimental strategies used for assessing Treg function, potency of the stimulating conditions, clinical activity of RA patients, duration of the disease, and treatment of patients with various drugs. In summary, the criteria and the experimental conditions required to analyze the frequency and function of human CD4+CD25+ Treg have not been standardized, and therefore results may vary among different studies.
In fact, phenotypical characterization of human CD4+CD25+FoxP3+ Treg is difficult to achieve (34). In contrast with murine cells, the line between human CD25+ and CD25− CD4+ T lmphocytes is not easy to draw since humans demonstrate a smooth gradation of intensity from negative to highly positive CD25+ T cells (59). Also, human CD4+ T cells do not display an optimal correlation between the expression of CD25 and FoxP3: most of the human CD4+CD25high cells are Foxp3+, but in addition a variable percentage of human CD25int cells expresses Foxp3 at lower although substantial levels (34). Subsequently, the gold standard for Treg cell characterization is demonstration of their inhibitory action. Importantly, results of suppression assays are not only related to the purity of isolated Treg but also depend on the potency of the stimulus and on the Treg/Tresp ratio examined. Stronger stimuli tend to mask the effect of Treg and require high Treg/Tresp ratios to detect Treg-mediated suppression. Conversely, less potent stimuli allow the detection of inhibition at lower Treg/Tresp ratios and facilitate the study of Tregs with lower suppressive potencies. The superior function of RASFTreg can be appreciated under less stringent conditions and using stronger stimuli when compared with RAPB or HCPB Treg; this may explain why more consistent results have been published in the literature on this matter.
We observed by flow cytometry that an increased frequency of CD4+CD25+CD127−FoxP3+ T cells is present in the peripheral blood of early, untreated RA patients when compared with healthy controls, consistent with data reported by Van Amelsfort et al. on established RA patients (53). Using magnetic beads, we carefully isolated a CD25+ population that was CD127−, FoxP3+, and expressed high levels of IL-15Rα. Suppression tests with submaximal stimuli confirmed an optimal regulatory capacity that was apparent for Treg isolated from all three patient groups studied: RAPB, RAPB-R, and RASF. Furthermore, we observed that the suppressive potency of RAPB Treg was significantly superior to that of HCPB Treg, whereas the suppressive strength of RASF Treg was maximal. Of note, when high Treg/Tresp ratios were used in the presence of submaximal stimuli, differences in suppressor capacity among the tested Treg populations were not appreciated. With decreasing Treg/Tresp ratios, however, differences among HCPB Treg, RAPB Treg, and RASF Treg became apparent. Interestingly, Tregs isolated from the PB of RA patients that achieved remission were functionally comparable with Tregs from the PB of HC, indicating that control of disease activity is associated with a reversal of the in vivo altered Treg/Tresp equilibrium. Discrepancies of our data with a previous report on the frequency and function of PB Treg from early, untreated, RA patients (58) may be related to the technical matters discussed above.
In summary, RASFib IL-15 exerts a dual action on the equilibrium between Treg and Tresp, by potentiating the suppressive effect of Treg while augmenting the proinflammatory action of Tresp, and this results in a shift of the Treg/Tresp balance toward a proinflammatory state. Additionally, Treg with superior suppressive potency are present in the peripheral blood and the synovial fluid of RA patients, but this enhanced immunoregulatory activity is not able to overcome the increased secretion of pathogenic cytokines by RA Tresp, indicating that RA patients demonstrate an altered Treg/Tresp equilibrium in vivo that is comparable to that induced in healthy controls upon coculture with RASFib. Therapeutic strategies directed to restoring this altered equilibrium and/or modifying the action of local IL-15 in the joint may be useful for patients with RA.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Ministerio de Educación Grant SAF 2006-01670 (to M.-E.M.C., M.B.M.) and a Hospital La Paz Research grant (to Y.G.C.).
Abbreviations used in this paper: RA, rheumatoid arthritis; HCPB, peripheral blood T cells of healthy controls; HCPBTL, peripheral blood T lymphocytes from healthy controls; MFI, mean fluorescence intensity; MTX, methotrexate; RAPB, peripheral blood T cells from untreated early rheumatoid arthritis patients; RAPBTL, peripheral blood T lymphocytes from early rheumatoid arthritis patients; RAPB-R, peripheral blood T cells from early rheumatoid arthritis patients who had achieved disease remission with treatment; RASF, rheumatoid arthritis synovial fluid; RASFib, rheumatoid arthritis synovial fibroblast; RASFTL, synovial fluid T lymphocytes from rheumatoid arthritis patients; SFib, synovial fibroblast; TCD4T, total CD4+ T cell; TL, T lymphocyte; Treg, regulatory T cell; Tresp, responder T cell.