Bone marrow–derived and synovium-derived mesenchymal cells promote Th17 cell expansion and activation through caspase 1 activation: Contribution to the chronicity of rheumatoid arthritis

Authors


  • Presented in part at the 31st European Workshop for Rheumatology Research, Amsterdam, The Netherlands, March 2011.

Abstract

Objective

Th17 cells have been implicated in rheumatoid arthritis (RA). We hypothesized that the interaction of T cells with bone marrow–derived mesenchymal stem cells (BM-MSCs) or with fibroblast- like synoviocytes (FLS) might, with the help of T cell–secreted inflammatory cytokines (i.e., interleukin-17A [IL-17A], tumor necrosis factor α [TNFα], and/or interferon-γ [IFNγ]), promote Th17 cell expansion and activation.

Methods

Peripheral blood mononuclear cells (PBMCs) from healthy blood donors were cocultured with BM-MSCs or FLS from RA patients or osteoarthritis (OA) patients. Cocultures were exposed to phytohemagglutinin with or without IL-17A, TNFα, or IFNγ. Quantitative reverse transcription–polymerase chain reaction analysis, enzyme-linked immunosorbent assay, and cytofluorometry were used to measure IL-17A production.

Results

Interaction of PBMCs with BM-MSCs inhibited Th1 and Th2 responses, but promoted Th17 cell expansion, as early as 24 hours, as demonstrated by increases in retinoic acid receptor–related orphan nuclear receptor γ or IL-17A messenger RNA (mRNA) levels, IL-17A secretion levels, and IL-17A–secreting cell frequency, as well as by T cell switching to the Th17 pathway after 2 rounds of stimulation with MSCs. IL-17A production was also increased in PBMCs stimulated with anti-CD3 plus anti-CD28 or in isolated CD3+ or CD45RO+ T cells, thus demonstrating the role of T cell activation. Levels of mRNA for IL-6, IL-8, and IL-1β were further amplified when T cell–secreted inflammatory cytokines were added. Interestingly, OA FLS or RA FLS also enhanced IL-17A and IL-6 production, but only RA FLS enhanced IFNγ and IL-1β production. We further demonstrated that MSC-mediated Th17 promotion requires caspase 1 activation by using an inhibitory peptide and measuring its activity.

Conclusion

We found that the interaction of MSCs or FLS with T cells promotes the activation and expansion of Th17 cells through caspase 1 activation. Since proinflammatory and T cell–secreted inflammatory cytokines are also amplified, this mechanism may participate in the chronicity of RA.

Th17 cells have recently been identified as a subset of T helper cells that is distinct from Th1 and Th2 cells (1–5). In addition to their role in host protection against extracellular bacteria and fungi (6–8), Th17 cells have been implicated in inflammatory and autoimmune diseases such as rheumatoid arthritis (RA), multiple sclerosis, inflammatory bowel disease, and psoriasis (refs.9–12 and, for review, see ref.13). The crucial role of interleukin-17 (IL-17) in the development of autoimmune diseases has been demonstrated in the SKG mouse model, where self-reactive IL-17–secreting T cells were shown to mediate an autoimmune arthritis resembling RA (14). Consistent with the effect of IL-17A on the initiation of inflammatory arthritis, a single injection of IL-17A was shown to induce cartilage damage in a mouse model (15).

A recent study suggested that IL-17A may drive autoimmune responses by promoting the formation of spontaneous germinal centers (16). In addition to their initiating role in RA, Th17 cells also play a role in the propagation of inflammation, through enhancement of formation of antibodies, secretion of proinflammatory cytokines such as IL-1β, IL-6, and tumor necrosis factor α (TNFα), and synthesis of metalloproteinases, cyclooxygenases, or other proinflammatory mediators, by cells residing in the synovium, such as fibroblast-like synoviocytes (FLS), monocytes, macrophages (17–20). Th17 cells also participate in bone degradation by increasing the expression of RANKL on osteoblasts, which in turn, leads to enhancement of osteoclastogenesis, through the interaction of RANKL with RANK on osteoclasts (21, 22).

While a role of Th17 cells has been implicated in RA (for review, see ref.23), the mechanisms that lead to their maintenance in the RA synovium remain elusive. Indeed, even though inflammatory cytokines, such as IL-1, IL-23, IL-6, and transforming growth factor β (TGFβ), which promote Th17 differentiation (24), are found in the RA synovium, cytokines such as interferon-γ (IFNγ) or IL-12, which counteract their differentiation, are also expressed. Treg cells, plasmacytoid dendritic cells, and mesenchymal stem cells (MSCs), which contribute to the immune regulation of T cell functions, have also been demonstrated (refs.25–27 and, for review, see ref.28). To address this paradox, some previous studies have proposed that Treg cells may be functionally deficient in RA, due in part to a defect in the CTLA-4 molecule (29–30), whereas others have proposed that the immunomodulatory activity of MSCs could be defective in an inflammatory environment (31). Studies of the effects of bone marrow–derived MSCs (BM-MSCs) in collagen-induced arthritis (CIA) have shown that BM-MSCs do not affect the development of arthritis and may even accelerate CIA through enhancement of IL-6 production (32, 33). A more recent study implicated the interaction of synovial fibroblasts with T cells in the autocrine IL-17A production and subsequent matrix metalloproteinases induction in early RA (34).

Because BM-MSCs differentiate into FLS, the preponderant cells in the RA synovium, we analyzed the capacity of BM-MSCs or FLS from RA patients to regulate the functions of Th1, Th2, as well as Th17 in the presence or absence of a combination of T cell–secreted inflammatory cytokines, such as IL-17A, IFNγ, and TNFα, which are known to influence the functions of FLS and MSCs in the inflamed synovium (20). We also investigated the mechanisms involved in the regulation of IL-17A production.

MATERIALS AND METHODS

Generation and culture of human BM-MSCs.

Bone marrow aspirates were obtained from 3 healthy donors after obtaining their consent and BM-MSCs were derived. Briefly, peripheral blood mononuclear cells (PBMCs; 2.5 × 105/ml) from the bone marrow were seeded in α-minimum essential medium (Invitrogen) containing 10% fetal calf serum (FCS), 2 mM L-glutamine, and 100 units/ml of penicillin/streptomycin (complete medium). Cells were allowed to adhere for 72 hours, and the nonadherent cells were removed. Half of the culture medium was changed every 3 days. After 21 days, or sooner if the cells were confluent, the cells were harvested and passaged. Cells were used between passages 4 and 9. MSCs were identified in 4 different ways: by their immunophenotype based on the expression of CD73, CD90, CD105 and the absence of expression of CD45 and CD34; by colony-forming unit–fibroblast enumeration; by the aldehyde dehydrogenase assay (AldeFluor kit; StemCell Technologies), which evaluated the percentages of stem cells in cultures; and by their capacity to differentiate into osteoblasts after culturing in osteogenic medium (data not shown).

Expansion of FLS.

FLS were grown from synovial tissue samples obtained from the hips or knees of 6 patients with RA and from 3 patients with OA who were undergoing joint surgery. The RA patients fulfilled the American College of Rheumatology criteria for RA (35). OA was diagnosed according to the typical clinical signs and symptoms according to the ACR criteria (36, 37). FLS were isolated by enzyme digestion and cultured in Dulbecco's modified Eagle's medium/10% FCS/complete medium. RA FLS beyond the third passage were >99% negative for CD45 and CD34, and were positive for CD73, CD90, and CD105. However, unlike MSCs, they were not able to differentiate into osteoblasts, as assessed by alizarin red staining. Aldehyde dehydrogenase activity was observed in <5% of the cells (data not shown). FLS were used between passages 4 and 9.

Synovial T cell expansion.

Synovial T cells were expanded from synovial tissues cultured for 14 days in RPMI 1640/10% FCS/complete medium, in the presence of recombinant IL-2 (20 IU/ml; R&D Systems). Cultures were fed every 3 days. T cells were then collected and analyzed by fluorescence-activated cell sorting. More than 95% of the cells expressed CD3.

Coculture assays.

PBMCs from healthy blood donors were isolated by Ficoll-Hypaque (1.077 gm/ml) density-gradient centrifugation. Cocultures were initiated by overnight seeding of MSCs or FLS in 96-well plates at a concentration of 2 × 104 cells/well containing RPMI 1640/10% human AB serum (Blood Bank Center in Lyon). The next day, a total of 1 × 105/well of PBMCs or synovial T cells were seeded on top of the MSCs layer for 24–72 hours in the presence of either phytohemagglutinin (PHA; 5 μg/ml), anti-CD3 plus anti-CD28 monoclonal antibodies (mAb; 5 μg/ml of each), or 1 × 105 irradiated allogeneic mononuclear cells (MNCs) from another blood donor. In some experiments, neutralizing anti–IL-6 mAb, anti–IL-1β mAb (both from R&D Systems), CTLA-4Ig fusion molecules (Bristol-Myers Squibb), or IL-18 binding protein (R&D Systems) were added to cultures at the concentrations indicated below.

Isolation of CD3+ and CD45+ T cells.

In some experiments, PBMCs were sorted by magnetic-activated cell sorting (MACS), according to the manufacturer's instructions (Miltenyi Biotec), using anti-CD3–coated or anti-CD45RO–coated magnetic beads. MACS sorting results were validated by staining of CD3 and CD45RO. Results were considered efficient if CD3+ or CD45RO+ cells were >85%.

Flow cytometry.

Fluorescein isothiocyanate (FITC)–conjugated or phycoerythrin (PE)–conjugated mouse anti-human CD73, CD90, CD105, CD45, CD34, or CD3 (all from BD Biosciences) were used to stain cultured cells. IL-17A production was revealed after permeabilization of cells by intracellular staining with Alexa Fluor 488–conjugated anti-human anti–IL-17A mAb (BioLegend). Analyses were performed using a FACSCanto II cytometer and Diva software (both from BD Biosciences).

T cell proliferation assay.

Cells were given an 18-hour pulse with 1 μCi of 3H-thymidine. 3H-thymidine incorporation was measured on day 3 of activation with PHA or with anti-CD3 plus anti-CD28 mAb, or on day 5 of stimulation with irradiated allogeneic MNCs.

Assay for IL-17A–secreting cells.

T cells secreting IL-17A were sorted with an IL-17A magnetic bead secretion assay (Miltenyi Biotec), according to the manufacturer's instructions. Briefly, this kit uses a CD45/IL-17A bispecific antibody affinity matrix (IL-17 catch reagent) to immobilize IL-17 proximal to the outer surface of secreting cells. Cells were subsequently double-labeled with PE-conjugated anti–IL-17 mAb and FITC-conjugated CD4 mAb, then analyzed by flow cytometry (FACSCanto II).

Enzyme-linked immunosorbent assays (ELISAs).

Concentrations of IL-17A, IL-1β, IL-6, and IFNγ were evaluated with commercially available ELISA kits, according to the manufacturer's instructions (R&D Systems).

Secondary and tertiary stimulation of PBMCs or T cells.

For secondary experiments, PBMCs or CD3+ MACS-sorted T cells were left untreated (control) or were activated with PHA, with or without MSCs. After 3 days of stimulation, cells were washed, allowed to rest for 24 hours, and stimulated with anti-CD3 plus anti-CD28 mAb in the presence or absence of MSCs. IL-17A secretion was measured 24 hours later by ELISA. For tertiary experiments, T cells from the secondary stimulation step were MACS sorted according to the presence/absence of CD45RO expression and expanded for an additional 10 days with recombinant IL-2 (10 IU/ml) and recombinant IL-23 (50 ng/ml). IL-17A secretion was measured following an additional 24 hours of stimulation with anti-CD3 plus anti-CD28 mAb.

Quantitative reverse transcription–polymerase chain reaction (RT-PCR) analysis of cytokine expression.

Total RNA was extracted using an RNeasy Plus Mini kit (Qiagen). A total of 200 ng of RNA was reverse transcribed using a QuantiTect RT kit, and PCR amplification was performed using a LightCycler instrument with a QuantiFast Green PCR kit (all from Qiagen). Specific amplification of genes of interest was performed using LightCycler primer set kits, according to the manufacturer's instructions. Thermocycling was initiated for 5 minutes at 95°C, followed by 40 cycles of amplification (10 seconds at 95°C and 30 seconds at 60°C). Levels of expression of messenger RNA (mRNA) were determined using LightCycler computer software. Results were calculated as the ratio of mRNA for the genes of interest to mRNA for CD3ε or cyclophilin B (or, peptidylpropyl isomerase B), a housekeeping gene, to take into account genes specifically expressed or specifically not expressed by T cells, as previously reported (38).

Caspase 1 activity.

To measure the activity of caspase 1, PBMCs or CD3+ T cells were left untreated or were activated with PHA in the presence or absence of MSCs for 18 hours. Cells were then lysed with lysis buffer (Abcam) according to the manufacturer's instructions, except that cytosolic fractions were isolated after 30 minutes of centrifugation at 14,000 revolutions per minute at 4°C, as previously described (39). Cleavage of paranitroanilide from the YVAD substrate following caspase 1 treatment for 10–30 minutes at 37°C was measured using a photometer at 405 nm wavelength. Results are expressed as arbitrary units.

Inhibition of caspase 1 activity.

Cell cultures were initiated in the presence of the fluoromethylketone peptide Z-YVAD-FMK (MBL CliniSciences), which inhibits caspase 1 activity by competing with YVAD for the specific recognition sequence for IL-1β–converting enzyme/caspase 1. The concentrations used were 2, 4, and 10 μM.

Statistical analysis.

Data are presented as the mean ± SD percentage positivity. Except where indicated otherwise, statistical significance of the data was determined by 2-tailed nonparametric t-test. P values less than or equal to 0.05 were considered significant.

RESULTS

BM-MSCs inhibit Th1 and Th2 cell responses independently of the presence of IL-17A, TNFα, and/or IFNγ.

BM-MSCs are known to modulate T cell immune responses. As shown in Figure 1A, whatever the mode of T cell activation, whether allogeneic antigen-presenting cells, mitogen, PHA, or anti-CD3/anti-CD28 mAb, a clearcut dose-dependent inhibition of T cell proliferation was observed. Because the best ratio of MSCs to T cells was 1:5, we used this cell ratio for subsequent experiments. We then analyzed the expression of mRNA markers for Th1 and Th2 cells following stimulation of PBMCs with PHA. As expected, a profound decrease in IL-2, IFNγ, TNFα, IL-4, and IL-10 mRNA levels was observed (Figure 1B). However, FoxP3 mRNA expression was not significantly modified in these cocultures. Addition of exogenous IL-17A, TNFα, or IFNγ did not reverse these cytokine mRNA patterns, even when the 3 agents were combined and applied together. This suggested that the T cell–secreted inflammatory environment does not influence the inhibitory effects of MSCs on Th1 and Th2 cell activation.

Figure 1.

Mesenchymal stem cell (MSC)–mediated inhibition of Th1 and Th2 cell responses is not reversed by T cell–secreted inflammatory cytokines. A, MSCs were added to peripheral blood mononuclear cell (PBMC; 105/well) cultures at different cell ratios or PBMCs were cultured alone. Cocultured cells were stimulated with irradiated allogeneic antigen-presenting cells (Allo-Ag), phytohemagglutinin (PHA), or anti-CD3 plus anti-CD28 monoclonal antibodies (A/CD3+CD28) (5 μg/ml each). T cell proliferation was measured by 3H-thymidine (3HT) incorporation on day 5 or day 3 of culture. Results are representative of 3 experiments. Values are the mean ± SD. B, MSCs and PBMCs were cultured alone or were cocultured at a 1:5 ratio for 24 hours in the presence or absence of PHA (5 μg/ml), and exogenous interleukin-17A (IL-17A; 50 ng/ml), tumor necrosis factor α (TNFα; 1 ng/ml), or interferon-γ (IFNγ; 50 ng/ml) was added either alone or in combination. Expression of the indicated genes was calculated as the ratio of mRNA for the genes of interest to mRNA for CD3ε or cyclophilin B (CPB). Values are the mean ± SD of 4 experiments. = P ≤ 0.05 versus controls.

BM-MSCs up-regulate IL-17A transcription and secretion levels, as well as the frequency of cells secreting IL-17A.

Among various newly discovered Th cell subsets (40), the Th17 pathway has been described as one that leads to the generation of inflammatory T cells (1–5). We therefore investigated the effects of MSCs on Th17 cell activation. We observed a strong up-regulation of the expression of mRNA for IL-17A and retinoic acid receptor–related orphan nuclear receptor γ following stimulation of PBMCs with PHA (Figure 2A). Of note, IL-17A mRNA was not expressed by PHA-activated MSCs.

Figure 2.

Bone marrow–derived MSCs (BM-MSCs) up-regulate IL-17A transcription and secretion levels, as well as the frequency of IL-17A–secreting cells. MSCs and PBMCs were cultured alone or were cocultured at a 1:5 ratio for 24 hours (A and B) or 48 hours (B) in the presence or absence of PHA (5 μg/ml), and exogenous IL-17A (50 ng/ml), TNFα (1 ng/ml), or IFNγ (50 ng/ml) was added either alone or in combination. A, Expression of mRNA for IL-17A and retinoic acid receptor–related orphan nuclear receptor γ (RORC). B, Production of IL-17A protein in cell supernatants under the indicated conditions. In addition to MSC cocultures (left and middle), PBMCs were cultured with conditioned medium from MSC cultures that had been stimulated for 24 hours with PHA or were cocultured with MSCs from which the conditioned medium had been washed out (right). Values in A and B are the mean ± SD of 3–5 experiments. = P ≤ 0.05 versus controls. C, Percentage of T cells secreting IL-17A. T cells were isolated with an IL-17A secretion assay (see Materials and Methods) and assessed by cytofluorometry. D, Percentage of intracellular staining for IL-17A, as analyzed by cytofluorometry. See Figure 1 for other definitions.

Interestingly, a strong increase in IL-17A protein was also observed as early as 24 hours poststimulation in the presence of MSCs and was further amplified at 48 hours (Figure 2B). Induction of IL-17A secretion by MSCs definitely required cell–cell contact, since conditioned medium harvested from PHA-activated MSCs did not induce any IL-17A production by PHA-activated T cells. The inverse was also true. When conditioned medium was washed out from the MSCs before their interaction with T cells, the levels of IL-17A secretion were reduced by half following coculture with PHA-stimulated T cells, indicating that both soluble factors and cell–cell contact may play a role in Th17 promotion.

An increase in the frequency of IL-17A–secreting cells was also observed, as assessed by both IL-17A secretion assay and IL-17A intracellular staining (Figures 2C and D). Most of the cells secreting IL-17A were CCR6+ and CD4+. Therefore, these results clearly demonstrated that MSCs up-regulate IL-17A transcription and secretion levels, as well as the frequency of IL-17A–secreting cells. The influence of the T cell–secreted inflammatory environment resulted in the amplification of IL-17A mRNA expression, but this difference did not achieve statistical significance (Figure 2A).

BM-MSCs induce a switch of activated T cells to the Th17 pathway.

We next sought to determine whether this increase in IL-17A production was due to T cell activation. We performed 2 different sets of experiments. In the first, T cells were activated through their T cell receptor (TCR)/CD3 complex with anti-CD3 plus anti-CD28 mAb. The results showed that MSCs induced an increase in IL-17A, which was almost as high as that induced by PHA (Figure 3A). In the second experiment, T cells were isolated or were left unmodified relative to their CD3 expression. The results showed that when MSCs were present, PHA activation strongly enhanced IL-17A secretion in the two populations (Figure 3B). Interestingly, no monocytes were present in the CD3+ T cell fraction, thus demonstrating that MSCs played the role of accessory cells in this coculture assay.

Figure 3.

Bone marrow–derived MSCs (BM-MSCs) induce a switch of activated T cells to the Th17 pathway. A, PBMCs were cultured in the presence or absence of MSCs at a 1:5 ratio for 24 hours with or without PHA (5 μg/ml) or anti-CD3 plus anti-CD28 (5 μg/ml each), and IL-17A protein production in supernatants was measured by enzyme-linked immunosorbent assay (ELISA). B, CD3+ T cells were sorted with the use of magnetic beads and activated with PHA in the presence or absence of MSCs. IL-17A production was compared with that of PBMCs. C, After 3 days of activation, PBMCs or CD3+ T cells were washed, allowed to rest for 24 hours, and subjected to secondary stimulation for 24 hours with anti-CD3 plus anti-CD28 monoclonal antibodies (mAb). IFNγ and IL-17A protein production in supernatants was measured by ELISA. D, Secondarily activated T cells were sorted according to their expression of CD45RO, expanded for 10 additional days, and were left unstimulated (NO) or were stimulated with anti-CD3 plus anti-CD28 mAb in the presence or absence of MSCs. IL-17A protein production in supernatants was measured at 24 hours. Values in A and B are the mean ± SD of 3 experiments. Values in C and D are the mean and are representative of 2 experiments. = P ≤ 0.05 versus controls, except as indicated by the horizontal bars. See Figure 1 for other definitions.

Next, we investigated whether T cells cultured in the presence of MSCs could switch to the Th17 pathway. PBMCs were washed or CD3+ T cells were isolated after 1 round of stimulation, then the cells were allowed to rest for 24 hours, and reactivated with anti-CD3 plus anti-CD28 mAb in the presence or absence (control) of MSCs. As shown in Figure 3C, a switch from IFNγ-producing T cells to IL-17A–producing T cells was observed if MSCs were present during the 2 rounds of activation. We also showed that Th17 cells were mostly derived from the CD45RO+ population by MACS sorting for their expression of CD45RO and stimulating them again with anti-CD3 plus anti-CD28 mAb. Interestingly, we observed that third-round–stimulated T cells no longer required the presence of MSCs in order to overproduce IL-17A (Figure 3D).

Up-regulation of proinflammatory cytokines by MSCs is potentiated in the presence of T cell–secreted inflammatory cytokines.

Because IL-6, IL-1β, and IL-8 cytokines are also involved in the pathogenesis of RA, we addressed whether they were increased in cocultures with MSCs. As shown in Figure 4A, we found that MSCs indeed significantly enhanced IL-6 mRNA expression levels. IL-1β and IL-8 mRNA levels were also increased, albeit to a lesser degree. The levels of the 3 cytokines were increased even higher in the presence of IL-17A alone, with or without TNFα and/or IFNγ. Since IL-17A is known to enhance IL-6 secretion by MSCs (20), we investigated whether the up-regulation of IL-17A secretion mediated by MSCs (Figure 2B) could in turn enhance IL-6, IL-1β, and IL-8 mRNA secretion levels. With this aim, we cultured MSCs in the absence of other cells and in the presence or absence of recombinant IL-17A. As shown in Figure 4B, exposure of MSCs to IL-17A alone at the cell concentration we used resulted in significantly increased expression of mRNA for IL-8 and IL-1β, but not IL-6. Thus, the increased IL-6 expression in cocultures was likely to be mostly related to monocytes rather than MSCs.

Figure 4.

Up-regulation of IL-1β, IL-6, and IL-8 mRNA expression by MSCs is potentiated by T cell–secreted inflammatory cytokines. A, MSCs and PBMCs were cultured alone or were cocultured at a 1:5 ratio for 24 hours in the presence or absence of PHA (5 μg/ml), and exogenous IL-17A (50 ng/ml), TNFα (1 ng/ml), or IFNγ (50 ng/ml) was added either alone or in combination. Expression of the indicated genes was calculated as the ratio of mRNA for the genes of interest to mRNA for CPB. B, MSCs were cultured for 24 hours without PBMCs or PHA and were stimulated with IL-17A (50 ng/ml), TNFα (1 ng/ml), and/or IFNγ (50 ng/ml). Expression of the indicated genes was calculated as the ratio of mRNA for the genes of interest to mRNA for CPB. Values are the mean ± SD of 3–4 experiments. = P ≤ 0.05 versus controls, except as indicated by the horizontal bars. In B, 1-tailed nonparametric t-tests were used because of the assumption that inflammatory cytokine mRNA levels should increase and not decrease. See Figure 1 for definitions.

Considering the positive effects of the combined action of IL-17A, IFNγ, and TNFα on proinflammatory cytokine secretion (Figure 4A), various combinations of the 3 cytokines were added to MSC cultures. As shown in Figure 4B, TNFα at 1 ng/ml induced a weak increase in IL-8 and IL-1β mRNA expression, but strongly potentiated the effects of IL-17A on IL-6, IL-1β, and IL-8 mRNA expression. Surprisingly, IFNγ further increased the effects of IL-17A and TNFα on the mRNA levels of these 3 proinflammatory cytokines. This demonstrated that under certain circumstances, IFNγ could potentiate, but not counteract, the effects of IL-17A.

Caspase 1 activation is required for the enhancement of IL-17A production by MSCs.

IL-6 and IL-1β are known to contribute to Th17 differentiation (4). Therefore, we investigated their contribution by adding anti–IL-6 or anti–IL-1β mAb, either alone or in combination, to our coculture assays. As shown in Figure 5A, only partial inhibition of IL-17A secretion, with no blocking effect, was observed. We therefore used a CTLA-4Ig fusion molecule to inhibit the interactions of accessory cells with T cells. This resulted in a weak decrease in IL-17A secretion and no potentiating effects when combined with anti–IL-1 and anti–IL-6. Because caspase 1 activation is involved in the mechanisms that lead to inflammation by allowing IL-18 and IL-1β to become bioactive (41), we tested its role in our model. The inhibition of caspase 1 activity abrogated IL-17A production (Figure 5D) as well as IL-1β and IL-18 secretion (results not shown). However, only partial inhibition of IL-17A levels was observed when both IL-1β and IL-18 were blocked by their specific inhibitors (Figure 5D). Measurement of the enzymatic activity of caspase 1 revealed that its increase required that activated PBMCs be in contact with MSCs (Figure 5E).

Figure 5.

Caspase 1 activation is required for the enhancement of IL-17A production by MSCs. A–D, MSCs and PBMCs were cultured alone or were cocultured at a 1:5 ratio for 24 hours in the presence or absence of PHA (5 μg/ml). IL-17A protein production in supernatants was measured by enzyme-linked immunosorbent assay (ELISA) and compared to that in supernatants from primary stimulation cultures. Cultures were performed in the presence or absence of anti–IL-6 (A/IL-6; 30 μg/ml), anti–IL-1β (A/IL-1β; 30 μg/ml), or both monoclonal antibodies (mAb) (A), in the presence or absence of 50 μg/ml or 100 μg/ml of CTLA-4Ig fusion protein (B), in the presence or absence of anti–IL-1β plus anti–IL-6 mAb (30 μg/ml) with or without CTLA-4Ig (100 μg/ml) (C), and in the presence or absence of IL-18 binding protein (IL-18BPa; 50 μg/ml) to inhibit IL-18, with or without anti–IL-1β mAb (D). To the latter cultures, the Z-YVAD-FMK peptide (A/caspase 1) was added (10 μM) to inhibit caspase 1 activation in cocultures of MSCs with PBMCs (last bar). = P ≤ 0.05 versus controls. E, MSCs and PBMCs were cultured alone or were cocultured at a 1:5 ratio for 18 hours in the presence or absence of PHA (5 μg/ml). Caspase 1 activity in supernatants was measured by ELISA. Values are the mean ± SD of 3 experiments. See Figure 1 for other definitions.

The ability of MSCs to increase IL-17A production is shared by cells of mesenchymal origin.

MSCs contribute to cell–cell interactions in the bone marrow, whereas FLS, which derive from MSCs, do so inside the RA joint. We investigated whether FLS could also enhance the production of IL-17A. Thus, we added RA FLS at the same ratio as MSCs to PHA-activated PBMCs and measured the levels of IL-17A released in the supernatants. As shown in Figure 6A, a strong increase in the levels of expression of IL-17A mRNA was induced in the presence of RA FLS. Unexpectedly, when FLS from OA patients were used (control), a similar increase in IL-17A was observed, suggesting that the IL-17A mRNA–enhancing property of the FLS was not related to their inflammatory state. When the production of IFNγ, IL-6, and IL-1β was evaluated, however, a clear differential effect depending on the origin of the FLS was observed. Indeed, RA and OA FLS up-regulated IL-6 to almost the same levels, whereas RA, but not OA, FLS significantly up-regulated IFNγ and IL-1β production (Figure 6B).

Figure 6.

The ability of MSCs to increase IL-17A production is shared by cells of mesenchymal origin. A, Healthy PBMCs were cultured alone or were cocultured with fibroblast-like synoviocytes (FLS) from osteoarthritis (OA) patients or rheumatoid arthritis (RA) patients at a 1:5 ratio for 24 hours in the presence of PHA (5 μg/ml), and levels of IL-17A in the supernatants were measured by enzyme-linked immunosorbent assay (ELISA). B, Healthy PBMCs were cultured alone or were cocultured with FLS from OA patients or RA patients at a 1:5 ratio for 24 hours in the presence of PHA (5 μg/ml), and levels of IFNγ, IL-1β, and IL-6 in the supernatants were measured by ELISA. C, Healthy PBMCs were cultured alone or were cocultured with FLS from RA patients at a 1:5 ratio for 24 hours in the presence or absence of anti-CD3 plus anti-CD28 monoclonal antibodies or TNFα, and the levels of IL-17A in the supernatants were measured by ELISA. D and E, RA PBMCs (D) or RA synovial T cells (E) were cultured alone or were cocultured with autologous FLS at a 1:5 ratio for 24 hours in the presence of PHA (5 μg/ml), and levels of IL-17A in the supernatants were measured by ELISA. Values in A and B are the mean ± SD of 3 experiments. Values in C–E are individual results. = P ≤ 0.05 versus controls. See Figure 1 for other definitions.

Moreover, we observed that IL-17A production was also enhanced in the presence of anti-CD3 plus anti-CD28, but not of TNFα, thus reinforcing the idea that the promotion of Th17 cells is dependent on T cell, but not stromal cell, activation (Figure 6C). We demonstrated that RA FLS could also increase IL-17A production when autologous cells were used. As shown in Figures 6D and E, in the presence of either autologous RA PBMCs or autologous RA synovial T cells, IL-17A production increased. Thus, we demonstrated that the interaction of RA FLS with activated synovial T cells delivers a signal that engages T cells to switch to the Th17 pathway.

DISCUSSION

BM-MSCs have been widely described as cells that inhibit the responses of both Th1 cells and Th2 cells, either directly or through modulation of antigen-presenting cell functions (42–44). We show herein that BM-MSCs also increase T cell inflammatory responses through amplification of IL-17A production at both the mRNA and protein levels (Figures 2A and B). This increase appeared as early as 24 hours following stimulation, improved further at 48 hours, and required both cell–cell contact and soluble factors. Accordingly, we observed that the irradiation of MSCs at 30 Gy decreased the levels of IL-17A mRNA by half (data not shown). The increase in IL-17A production was also observed at the cellular level, as assessed by frequency of cells secreting IL-17A (Figures 2C and D). This was related to T cell activation, since MSCs were able to amplify IL-17A production even when T cells were isolated or when PBMCs were activated with anti-CD3 plus anti-CD28 mAb (Figures 3A and B).

A recent study showed that MSCs are able to enhance IL-17A production by lipopolysaccharide-primed splenocytes following 3 days of coculture (33). But, we observed in the present study that this amplification occurred as soon as 24 hours poststimulation. Such a rapid expansion of Th17 cells was quite surprising, because Th17 cells are not easy to generate in vitro. Our results thus suggested that IL-17–producing cells were likely to have expanded from memory T cells, rather than having been generated from naive T cells. Consistent with this, we did not observe any significant increase in IL-17A production by CD45RA/CD4+ naive T cells in the presence of MSCs (data not shown), and we found that Th17 cells were mostly derived from the CD45RO+ population (Figure 3D). Moreover, Th17 cells were unlikely to derive from regulatory T cells, since the presence of MSCs did not lead to any significant modification of FoxP3 mRNA expression (Figure 1B).

Consistent with our findings, Ghannam et al recently showed that MSCs inhibit the differentiation of naive T cells into Th17 cells (45), and Van Hamburg et al (34) showed that memory, but not naive, Th17 cells can enhance their IL-17A production in the presence of stromal cells. Another possible explanation for the rapid expansion of Th17 cells is the activation by MSCs of γ/δ T cells, which are potent and rapid producers of IL-17A and precede the development of classic Th17 cells (46). We did not observe any significant MSC-mediated increase in T cells expressing γ/δ TCR, however (data not shown).

The increase in IL-17A in our model (i.e., in the presence of MSCs) was associated with a significant increase in IL-6 (Figure 4A). Because IL-17A has been reported to enhance IL-6 secretion by MSCs (20), we investigated whether the up-regulation of IL-6 secretion could be mediated by MSCs. But, no significant increase in IL-6 secretion was observed in the presence of IL-17A (Figure 4B). This finding was unexpected, but it might be related to the weak concentrations of MSCs used in our coculture assays. However, a synergistic effect on IL-6, IL-1β, and IL-8 mRNA expression was observed when TNFα and IFNγ were combined with IL-17A, suggesting that the T cell–mediated inflammatory environment may play a role in the secretion of proinflammatory cytokines by MSCs. The positive role of IFNγ on IL-17A–mediated proinflammatory cytokine production was quite surprising, since IFNγ has been reported to negatively regulate the development of Th17 cells (24). However, IL-17A and IFNγ have also been shown to potentiate neutrophil attraction in hapten-challenged skin (47), supporting the idea that under certain circumstances, IL-17A and IFNγ can act in concert.

Taken together, our results show that even though MSCs may exert an immunomodulatory activity on Th1 and Th2 cells (Figure 1), they may, in parallel, propagate inflammation by driving T cells toward the Th17 pathway and, with the help of T cell–secreted inflammatory cytokines, amplifying proinflammatory cytokine secretion. In support of these results, animal studies have shown that injecting MSCs in order to modulate collagen-induced arthritis resulted instead in amplification of the inflammation (32, 33).

Among the proinflammatory cytokines that were enhanced in our model, IL-6 and IL-1β are known to contribute to the generation of Th17 cells. We observed only their partial involvement in the present studies (Figures 5A–C). Addition of the CTLA-4Ig fusion molecule, which blocks the binding of B7-1 and B7-2 to CD28 molecules, did not further enhance the inhibitory effects of anti–IL-1β and anti–IL-6 on IL-17A secretion (Figure 5C). Because IL-1β requires processing by caspase 1 to become bioactive, we then examined whether there was any involvement of this enzyme. We clearly demonstrated the requirement of caspase 1 by inhibiting its activity (Figure 5D). Interestingly, measurement of caspase 1 activity demonstrated that this enzyme was activated only when activated PBMCs were in contact with MSCs (Figure 5E).

Consistent with our results, a recent study has shown the critical role of caspase 1 activation in driving IL-17A production in a model of experimental autoimmune encephalitis (48). Those investigators demonstrated the involvement of both IL-1β and IL-18 in the mechanisms leading to IL-17 production by autoimmune T cells. But we observed only a partial inhibition of IL-17A production when both IL-18 and IL-1β were inhibited (Figure 5D), suggesting that an additive molecule might be involved in our model. However, this cytokine was not IL-33, since its secretion was not inhibited when caspase 1 was blocked (data not shown). Whether activation of caspase 1 may be related to the activation by uric acid of the NALP3 protein, a member of the inflammasome complex, as demonstrated in OA (49), remains to be investigated. A phase II clinical trial in RA using a caspase 1 inhibitor demonstrated some efficacy, but the trial was stopped because of toxicity (50). Developing nontoxic drugs that target caspase 1 could therefore be a promising treatment in RA.

Finally, to approach RA pathogenesis, we investigated whether FLS, which originate from MSCs, could also promote Th17 expansion. Our results showed that not only RA FLS, but also OA FLS were able to up-regulate IL-17A mRNA levels by PBMCs (Figure 6A). Van Hamburg et al (34) recently reported that IL-17A production is enhanced in the presence of RA FLS (34). Thus, our results further extend their findings by demonstrating that the capacity of FLS to promote Th17 cell expansion is not dependent on the inflammatory state of these cells, but is a property shared by stromal cells. This latter finding is reinforced by other studies demonstrating that human fibroblasts support the expansion of Th17 cells in various models (51, 52). However, a clear differential effect of RA FLS and MSCs on inflammatory cytokine secretion was observed. Indeed, RA FLS were found to be able to enhance IL-17A, IL-6, IL-1β, and IFNγ secretion (Figure 6B), whereas MSCs significantly increased only IL-17A and IL-6 (Figures 1B, 2A, and 4A). Interestingly enough, FLS from OA patients behaved like MSCs rather than RA FLS (Figure 6B), suggesting that the inflammatory state of RA FLS may contribute to the activation of both Th1 and Th17 cell subsets. Our results also suggested that such a mechanism may occur inside the inflamed synovium, since cocultures of RA FLS with autologous PBMCs or synovial T cells enhanced IL-17A production (Figures 6D and E).

In conclusion, we have shown that when they encounter T cells, BM-MSCs or FLS trigger inflammation by promoting Th17 cells through caspase 1 activation. We suggest that this may contribute to chronicity in RA, since IL-17A production helped by IFNγ and/or TNFα leads to further activation of FLS and MSCs through amplification of IL-6, IL-1β, and IL-8 proinflammatory cytokines.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Eljaafari had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Eljaafari, Chevrel, Miossec.

Acquisition of data. Eljaafari, Tartelin, Aissaoui, Osta, Lavocat.

Analysis and interpretation of data. Eljaafari, Tartelin, Aissaoui, Miossec.

Acknowledgements

We wish to thank the Centre Commun d'Imagerie Laennec (CECIL) platform (INSERM, Laennec Medical Faculty, UCBL Lyon 1, Lyon, France) for allowing us to perform our cytofluorometric analyses and for excellent technical assistance.

Ancillary