collagen type II
collagen type II emulsified in CFA
- iNKT cells:
invariant natural killer T cells
Invariant natural killer T (iNKT) cells are a unique lymphocyte subtype implicated in the regulation of autoimmunity and a good source of protective Th2 cytokines. Agonist α-galactosylceramide (α-GalCer) of iNKT cells exert a therapeutical effect in type 1 diabetes. We investigated whether iNKT activation with α-GalCer was protective in collagen-induced arthritis (CIA) in DBA/1 mice, a standard model of rheumatoid arthritis. Here, we have shown that in vivo iNKT cell function was altered in DBA/1 mice since stimulation with α-GalCer led to decreased IL-4 and IFN-γ levels in sera, as compared with C57BL/6 mice. α-GalCer induced a clear-cut diminution of clinical and histological arthritides. An anti-IL-10 receptor antibody abrogated the protective effect of α-GalCer, suggesting a key role for IL-10 in the protection against CIA by activated iNKT cells. Confirming these data, disease protection conferred by α-GalCer correlated with the ability of LN CD4+ cells to secrete larger amounts of IL-10. These findings suggest that in CIA susceptibility to autoimmunity is associated with dysfunctions of iNKT cells. Our demonstration that iNKT cell activation by α-GalCer remains efficient in CIA-prone DBA/1 mice to provide protective IL-10 suggests that this could be used therapeutically to treat autoimmune arthritis.
Rheumatoid arthritis (RA) is an autoimmune chronic systemic inflammatory disease that primarily affects the synovial membranes of multiple joints 1. A cardinal feature of joint inflammation in RA is proliferative inflammation of synovial cells, i.e., synovitis, which results in the destruction of the adjacent cartilage and bone. Although its precise etiology is not clearly understood, it is generally accepted that the proinflammatory cytokines TNF and IL-1 are critical mediators in the inflammatory process of arthritis 2, whereas Th2 cells producing IL-4 and IL-10 suppress the disease 3. The fact that RA is a Th1-dominant autoimmune disease 4 is also supported by findings obtained with cytokine/anti-cytokine treatments. In particular, studies with animal models have demonstrated that systemic or locally administered Th2 cytokines can effectively protect against arthritis in mice. We and others have previously shown the therapeutic effect of IL-4 and IL-13 administration in mouse collagen-induced arthritis (CIA), a typical model for RA 1, 5–10. Similar results were also obtained with the administration of IL-10, a potent immunosuppressive cytokine 11, in various arthritis models 12–15, 16. However, the Th1/Th2 cytokine effects on experimental models of RA are complex since, for example, in some models, IL-10-deficient mice get more severe disease 17, suggesting a protective role of IL-10, but at the same time, IL-4-deficient mice are protected from arthritis 18, suggesting a deleterious role of this Th2 cytokine on the disease.
Considering the pathogenic role played by Th1 cells in RA development along with the fact that Th1/Th2 imbalance also participates in the perpetuation of chronic inflammation associated with this disease 3, it is tempting to exploit immunoregulatory mechanisms to inhibit the autoimmune response in experimental models of RA. In this regard, it is of interest to underline the peculiar capacity of a particular subset of immunoregulatory T cells to influence the Th1/Th2 profile during immune responses, i.e., invariant Vα14 NK T (iNKT) cells.
iNKT cells represent a unique T cell lineage that are potent producers of immunoregulatory cytokines, and have been implicated in several autoimmune diseases in mice and humans, including type 1 diabetes, experimental autoimmune encephalomyelitis (EAE), systemic lupus erythematosus, uveitis, colitis and scleroderma 19–23. In several situations, quantitative and/or qualitative defects in iNKT cells have been reported, and they are believed to be associated with the emergence of autoimmune diseases 24.
iNKT cells co-express NK markers such as CD161 and a highly restricted TCR repertoire, composed of a single invariant α chain (Vα14-Jα18 in mice and Vα24-Jα18 in humans) together with a limited TCR Vβ repertoire 25. They recognize glycolipid antigens in the context of the non-classical MHC class I molecule CD1d that include the exogenous glycosphingolipid antigen α-galactosyl ceramide (α-GalCer), obtained from a marine sponge, but absent in mammalian cells 26, and the glycolipid isoglobotrihexosylceramide (iGβ3) that has been recently identified as an endogenous molecule endowed with the capacity to positively select iNKT cells 27.
Based on the demonstration that iNKT cell stimulation with α-GalCer in vitro and in vivo leads to the rapid secretion of large quantities of the regulatory cytokines IL-4 and IL-10 28, we and others have investigated the possibility of inhibiting Th1 autoimmune diseases by treatment with this cognate iNKT cell ligand. Convincing results showing a therapeutic effect of α-GalCer in animal models have been obtained in type 1 diabetes 22, 29 and in EAE 19, 30.
The role of iNKT cells in RA has not been extensively studied until now. Despite the fact that iNKT cell stimulation by OCH, another synthetic ligand, have been shown to inhibit CIA in C57BL/6 mice 31, no study to date has been performed in the standard, well-characterized model of RA, namely CIA in DBA/1 mice, a genetically predisposed strain for development of CIA 32–35.
By examining the effect of α-GalCer on the development of CIA in DBA/1 mice, we show here a clear cut decrease in clinical and histological signs of arthritis despite a reduced responsiveness of iNKT cells in these mice. Importantly, this protection was not mediated through a Th1/Th2 deviation but rather by an IL-10-dependent suppressive pathway.
Analysis of iNKT cells in CIA-susceptible DBA/1 mice
We first examined whether α-GalCer stimulates cytokine secretion in DBA/1 mice, and compared the levels to that obtained in non-autoimmune prone C57BL/6 mice. Mice received a single injection of the antigen and were killed 2 h later; cytokine concentrations were then determined in the serum. Treatment of DBA/1 mice with α-GalCer led to a rapid appearance of IL-4 and IFN-γ in the serum, but, interestingly, for both cytokines, responses were about twice lower than those in C57BL/6 mice (Fig. 1) (IL-4: 693 ± 154 pg/mL vs. 1557 ± 137 pg/mL, p=0.009; IFN-γ: 2077 ± 378 pg/mL vs. 4005 ± 581 pg/mL, p=0.009).
The lower iNKT cell responses in DBA/1 mice were obtained despite a non-reduced iNKT cell pool in these mice, as referenced to C57BL/6 mice. Indeed, in both spleen and liver, the frequency of iNKT cells that were identified by double staining with a monoclonal antibody against TCR-αβ and CD1d-tetramers loaded with α-GalCer (α-GalCer/CD1d-tetramer) was superimposable in the two strains of mice (Fig. 2A, B). Moreover, the absolute numbers of α-GalCer/CD1d-tetramer+ cells in both the liver and the spleen did not significantly differ between DBA/1 and C57BL/6 mice (Fig. 2C). Taken together, these data clearly indicate that the deficiency of iNKT cells in DBA/1 mice concerns their functional capacities rather than their number.
α-GalCer treatment confers protection against CIA
We next investigated whether stimulation of iNKT cells by α-GalCer affected the course of arthritis. We determined the effect of α-GalCer treatment on CIA in DBA/1 mice immunized with collagen type II (CII). To determine whether the timing of iNKT cell activation (prior, during or after CIA induction) is important in modulating CIA, we performed different sets of experiments. In the first group, mice received α-GalCer i.p. on the day of CII immunization, and 8 and 4 days before immunization. In the second group, α-GalCer was injected i.p. on the day of CII immunization only. Lastly, in the third group, α-GalCer was given on the day of CII immunization, and 4, 7, 12 and 17 days after immunization. Fig. 3 shows the evolution of arthritis assessed by clinical evaluation. All of the α-GalCer-treated groups showed attenuated clinical signs of arthritis in comparison to control mice (see statistics in legend), although the best therapeutic effect was obtained in the group that received only one α-GalCer injection, on day 0. The clinical evaluation also shows that α-GalCer treatment neither influenced the onset of arthritis, nor modified the incidence of the disease (Table 1). However, the maximum score of arthritis, which represents the absolute severity of the disease, is slightly lower in group 1 (α-GalCer injection prior to immunization) and group 2 (α-GalCer injected on day 0). We also performed an histological evaluation of arthritis, showing a statistically significant decrease of inflammation and destruction parameters of arthritis as compared to controls, in mice treated only once with α-GalCer the day of CII immunization (Table 1 and Fig. 4). A statistically significant decrease of inflammation parameters was also shown in mice treated on days 4, 7, 12 and 17 with α-GalCer. A reduction in histological scores was observed between mice treated with α-GalCer prior to CIA induction and control mice, although differences are not statistically significant (Table 1).
|Group||Treatment||Clinical parameters||Histological scores|
|Onset (days)||Arthritis scores maximum||Incidence||Inflammation||Destruction|
|Control (n=8)||Vehicle||31.8 ± 1.4||18.6 ± 1.6||8/8 (100%)||2.08 ± 0.20||1.85 ± 0.19|
|Group 1 (n=10)||α-GalCerdays –8, –4, 0||32.6 ± 1.38||14.3 ± 0.8||10/10 (100%)||1.90 ± 0.12||1.64 ± 0.14|
|Group 2 (n=9)||α-GalCerday 0||30.6 ± 3.29||12.3 ± 1.9||8/9 (88%)||1.66 ± 0.19*||1.19 ± 0.18***|
|Group 3 (n=10)||α-GalCerdays 0, +4, +7, +12, +17||31.1 ± 1.46||14.8 ± 0.9||10/10 (100%)||1.60 ± 0.18**||1.52 ± 0.20|
When α-GalCer was given on day 0, it was administered separately from collagen type II emulsified in CFA (CII/CFA) Considering the importance of the method of injection in other models of autoimmune diseases like EAE 36, we evaluated, in another experiment, the effect of α-GalCer diluted within the CII/CFA given on day 0. Both clinical and histological analysis highlighted a therapeutic effect on arthritis exerted by this treatment, confirming the experiment shown in Table 1 and Fig. 4: the onset was not modified by treatment with α-GalCer diluted in CII/CFA (39.1 ± 1.9 vs. 40.8 ± 3.5 days for α-GalCer- and vehicle-injected group, respectively), and histological evaluation also showed a reduction of inflammation and destruction signs of arthritis in α-GalCer + CII/CFA-treated group (inflammation: 1.79 ± 0.28 vs. 0.84 ± 0.29, p=0.05, and destruction: 1.57 ± 0.26 vs. 0.8 ± 0.27, p=0.05 for vehicle- and α-GalCer-injected group, respectively).
Taken together, our results clearly demonstrate that α-GalCer administration prior and/or during CIA induction is accompanied by a reduction in clinical and histological signs of arthritis, either when α-GalCer was injected within the CII/CFA or separately.
In these experiments, the antibody response to CII was also evaluated by measuring immunoglobulin (total IgG, IgG1 and IgG2a isotypes) directed to CII. No statistically significant differences in antibody levels or in the anti-CII IgG1/IgG2a ratio were observed in any group of α-GalCer-treated animals as compared to controls (data not shown).
α-GalCer treatment promotes IL-10 secretion during CII-specific Th response
The immunological mechanism for arthritis protection by α-GalCer was investigated by measuring CD4 T cell response to the CII-immunizing antigen. DBA/1 mice were immunized with CII/CFA in hind paws and treated or not with α-GalCer within the immunizing CII/CFA. Eleven days later, CD4+ T cells from draining lymph nodes were assayed for cytokines production in response to an in vitro challenge with CII (Fig. 5). Lymph node CD4+ cells from CII-immunized and α-GalCer-treated mice secreted levels of IFN-γ and IL-4 similar to that in their counterparts from vehicle-treated mice upon in vitro re-stimulation with CII (Fig. 5A, B).
In contrast, the α-GalCer-induced protection was associated with a clear cut increase in IL-10 production by lymph node CD4+ cells in response to the in vitro re-stimulation with CII (Fig. 5C).
Neutralization of IL-10 activity reverses the therapeutic effect of α-GalCer in CIA
To further investigate whether the protective role of α-GalCer in CIA was mediated via an IL-10-dependent pathway, neutralizing experiments were performed in CIA mice treated with α-GalCer by co-administering an mAb against the IL-10 receptor (anti-IL-10R). α-GalCer + control IgG-treated mice were used as control and, as expected, a decrease in clinical scores of arthritis was observed in this group as compared to vehicle-injected control mice (Fig. 6). Conversely, treatment of CIA mice with the anti-IL-10R completely abrogated the protective effect of α-GalCer (Fig. 6); the maximum arthritis scores were decreased in the α-GalCer + IgG-treated group as compared with the α-GalCer+anti-IL-10R-treated group (6.8 ± 1.9 vs. 12.8 ± 3.4, respectively). Incidence of arthritis was 75% in α-GalCer + IgG-treated group, 100% in α-GalCer + anti-IL-10R and in vehicle group . Moreover, a reduction in histological scores was observed in mice treated with α-GalCer + IgG as compared with mice treated with α-GalCer + anti-IL-10R (mean ± SEM of destruction and inflammation parameters : 0.52 ± 0.26 vs. 1.01 ± 0.28, respectively).
iNKT cell targeting with α-GalCer has proven to be therapeutic in some autoimmune disease models such as in NOD mice for diabetes or the EAE model for multiple sclerosis 19, 22, 30, 36, 37. In this study, we asked whether this therapeutic strategy was also applicable in a well-characterized model of RA, i.e., CIA in DBA/1 mice. We show here that treatment with α-GalCer prevents CIA since it leads to an amelioration in the course and the severity of the disease, regarding clinical and histological signs of arthritis. This therapeutic effect was shown whatever way (i.p. or within the CII/CFA mixture) the α-GalCer was administered, and whatever injection timing we tested.
Our results differ from those of a previous study by Chiba et al.31, who showed that α-GalCer was ineffective in CIA. However, α-GalCer therapeutic efficiency varies on a strain of mouse used, as attested by a very recent study in a lupus model in SJL/J or BALB/c mice 23. Indeed, genetic susceptibility to autoimmunity is linked to MHC class II in animal models, as well as in humans, and it is likely that α-GalCer preferentially exerts its therapeutic effect in susceptible mouse strains. In this context, it should be underlined that our study was performed with DBA/1 (H-2q) mice that are highly susceptible to CIA 32, 33, 35, 38, 39, while the study from Chiba et al. was performed in C57BL/6 mice (H-2b). The latter are only partly susceptible to the disease, with a lower incidence and less severity 40, and in some studies, they show a resistance to CIA 41. Moreover, mice were aged less than 16 weeks at the end of all CIA experiments, thus eliminating the risk of a psoriasis arthritis-like disease that occurs spontaneously in DBA/1 male mice aged more than 4 months 42 and that could have interfered with CIA.
One important point to elucidate is the mechanism by which α-GalCer exerts its therapeutic properties. In Th1-like autoimmune diseases, such as diabetes and multiple sclerosis, two distinct mechanisms have been proposed: (i) a shift from a Th1 toward a Th2 pattern with IL-4 secretion 19, 22, 30, 36, 37, 43, 44, and (ii) induction of immunosuppressive cytokine production like IL-10 30, 37 (which is also considered as a Th2 cytokine in mouse).
In accordance with an IL-10-dependent mechanism rather than a shift from IFN-γ to IL-4 secretion, our study shows that the therapeutic effect of α-GalCer was completely reversed when it was given together with an antibody-neutralizing IL-10 receptor. Moreover, CII-specific lymph nodes CD4+ T cells from mice immunized with CII and injected with α-GalCer produced in vitro increased levels of IL-10, while IL-4 and IFN-γ levels were not modified. Furthermore, CIA mice treated with α-GalCer did not display any modification in anti-CII IgG1 and IgG2a levels.
Our demonstration that the α-GalCer therapeutic effect against CIA in DBA/1 mice is mediated through the stimulation of IL-10 secretion is in concordance with the known therapeutic effect of IL-10 in RA models. Indeed, it has been repetitively shown that the administration of IL-10 was efficient in DBA/1 mice with CIA, and most of these studies were performed using gene transfer strategies to obtain high and continuous IL-10 in vivo secretion 12, 45, 46. We demonstrated also a therapeutic effect of IL-10 in TNF-α transgenic mice, another model for RA 15.
It is still unclear, however, which cells secrete IL-10 after iNKT stimulation with α-GalCer. IL-10 has been shown to be produced by iNKT cells themselves upon exogenous stimulation 47, but it is also likely that up-regulation of IL-10 production results from an indirect effect of α-GalCer-activated iNKT cells. One possibility could be an IL-10 production by B cell after their activation by iNKT 48 cells through a CD1d-restricted mechanism. CD4+CD25+ regulatory T cells are also able to secrete high quantities of IL-10, but it is unclear whether iNKT cells induce IL-10 secretion by those cells. In fact, several observations in diabetes have shown that iNKT cells did not exert their effects through other regulatory T cells 22. Further investigations are therefore required to elucidate the exact mechanism of the IL-10 therapeutic effect upon iNKT cell stimulation by an exogenous ligand.
Our results show that the most effective treatment with α-GalCer was obtained when it was administered the day of CIA induction, which corresponds to the phase of initiation of the disease. Indeed, the importance of treatment timing in iNKT cell stimulation protocols has been documented in studies on other models of autoimmune diseases, such as EAE. In this model, early immunization with α-GalCer protects against the disease, while a later immunization potentiates it 36. Thus, we could hypothesize that α-GalCer is more efficient during the initiation period of the disease, while it could be deleterious during the amplification phase. According to this hypothesis, a very recent report shows a deleterious effect of iNKT cells in the K/BxN serum transfer model of RA 49. This model is confined to the inflammatory responses induced by the deposition of autoantibody in joint spaces, and thus corresponds to the amplification phase of RA.
These considerations lead to the question of the natural role of iNKT cells, in absence of their exogenous stimulation, in autoimmune diseases. A link between an iNKT cell deficiency and the susceptibility of various mouse strains to autoimmunity has been proposed in various models such as diabetes in NOD mice and EAE in SJL mice, and it has also been evoked in humans suffering from both pathologies 50, 51. Importantly, in patients with RA, iNKT cells were also shown to be depressed in the peripheral blood 52–54, supporting the idea that susceptibility to RA may be associated with iNKT cell defects. In this regard, our study shows a partial iNKT cell deficiency at the functional level in CIA susceptible DBA/1 mice.
Our demonstration that α-GalCer is efficient in CIA-prone mice displaying iNKT cell dysfunctions is an important point regarding the question of its therapeutic use to treat autoimmune arthritis. However, even if α-GalCer has already been used in cancer clinical trials 55, the precise mechanism by which IL-10-dependent iNKT cell activation leads to the protection of CIA should be elucidated before using iNKT cell ligands to treat RA.
Materials and methods
Male mice aged 5–9 weeks and belonging to the DBA/1 strain, which is susceptible to CIA, and male C57BL/6 mice, aged 6–9 weeks were purchased from Harlan Olac (Bicester, UK).
Spleen cell suspensions were prepared using a homogenizer, and red blood cells were lysed in hemolysis buffer (NH4Cl, KHCO3, EDTA). Afferent popliteal and inguinal lymph nodes were dissected out of the hind limbs. Cell suspensions were obtained after teasing of the lymph nodes. Liver was perfused with 1× PBS, and then pressed through a mesh. Liver mononuclear cells (MNC) were separated from parenchymal cells (pellet) by centrifugation at 50 × g for 5 min. MNC were collected and resuspended in a 35% Percoll solution (Amersham Biosciences Europe, Orsay, France) and centrifuged for 25 min at 750 × g. MNC were collected from the pellet and red blood cells were lysed as described above. MNC were then washed and resuspended in PBS/5% heat-inactivated fetal calf serum (FCS, Gibco-BRL). For measurement of collagen-specific T cell responses, lymph node cells were enriched for CD4+ T cells by magnetic-activated cell sorting (MACS, Miltenyi biotech, Bergisch-Gladbach, Germany) using a quadriMACS magnet fitted with a MACS LS+ column. Purity of CD4+ cell-enriched fractions was 85–95% after reanalysis.
Antibodies and flow cytometry analysis
PE-labeled anti-NK1.1 (clone PK136), PE-labeled anti-CD69 (clone H1–2F3), Cy-5.5-labeled anti-CD4 (clone RM4–5) and FITC-labeled anti-TCR-αβ (clone H57–597) were purchased from BD Pharmingen (San Diego, CA). Allophycocyanin-labeled tetramers were prepared in our laboratory from the mCD1d/mβ2 m expression vector constructed by Kronenberg's group 56, then loaded or not with α-GalCer (Pharmaceutical Research Laboratory of Kirin Brewery Co., Tokyo, Japan).
Cells were stained at 4°C in PBS containing 2% heat-inactivated FCS and 0.01 M sodium azide, incubated for 5 min with 2.4G2.3 mAb for blocking Fcγ receptors, then incubated for 30 min with appropriate dilutions of various mAb coupled to PE, fluorescein or allophycocyanin. Flow cytometry was performed on a four-color FACSCalibur (Becton Dickinson, Mountain view, CA). Dead cells were excluded on the basis of forward and side scatter characteristics. Statistics presented are based on at least 1000 events gated on the population of interest. Results were analyzed using Mac CellQuest software.
In vivo challenge with α-GalCer
To explore iNKT cell activation in vivo, DBA/1 and C57BL/6 mice received a single injection of α-GalCer (2 μg/dose, i.p.). They were killed 2 h later and sera were recovered, as previously described 22. Control mice were injected with an identical volume of vehicle solution alone.
CIA induction and evaluation
Arthritis was induced with native bovine CII (Chondrex, Morwell Diagnostics, Zurich, Switzerland) as previously described 32, 57. Male DBA/1 mice were injected subcutaneously at the base of the tail with 100 µg CII emulsified in 2 5µL CFA (Sigma Aldrich, Lyon, France) containing 1 mg/mL Mycobacterium tuberculosis. On day 21, mice were boosted with a subcutaneous injection of CII in IFA (Difco Laboratories, Detroit, MI). Mice were monitored for evidence of arthritis in the four paws using a blind procedure. For each mouse, clinical severity of arthritis was scored (0: normal; 1: erythema; 2: swelling; 3: deformity; and 4: necrosis) in ten joints: three joints of the two hind legs (toes, tarsus, ankle) and two joints of the two forelegs (toes and tarsus). The maximum score reached for each of the ten joints was 4, so the maximum score of clinical arthritis reached for a single mice on a given day was 40. The mean arthritic score on each clinical observation day was calculated in each treatment group. For histological analysis, the animals were killed 62 days after induction of CIA, and their legs were dissected free and processed for histological studies, as described elsewhere 57. Extensive sections were cut for each paw and at least four were examined. The lesions were blindly evaluated for each joint as previously described using a four-point scale (0–3, where 0 is normal and 3 severe). This global histological score reflects both synovitis (synovial proliferation, inflammatory cell infiltration) and joint destruction (bone and cartilage thickness and irregularity and presence of erosions). We also evaluated separately the articular destruction by taking into account the degradation of bone and cartilage regardless of inflammation on a four-point scale 0–3, where 0 is no destruction and 3 the presence of subchondral bone erosions.
Individual sera were tested using a standard ELISA to look for antibodies to papain-solubilized native CII (kindly given by D Herbage, Lyon, France) (total IgG, IgG1, IgG2a) as described elsewhere 58.
Treatment of CIA with α-GalCer
For protection studies against CIA, DBA/1 mice immunized with CII (day 0) received i.p. a 4 μg/dose of α-GalCer or an appropriate amount of control vehicle (polysorbate), both diluted in sterile saline on day 0, or on days –8, –4, 0, or on days 0, +4, +7, +12, +17. In a second set of experiments, mice received 4 μg/dose of α-GalCer added in the CII/CFA immunizing solution or an equal volume of vehicle at day 0. Using this latter regimen, the effect of α-GalCer after neutralization of IL-10 activity was examined. α-GalCer-treated mice received a 200 μg/dose of an anti-IL-10R (hybridoma 1B1.3a) or rat IgG1 (isotype control 22) i.p. at days 0, +3, +7, +9, +14.
Measurement of T cell response
For measurement of collagen-specific T cell responses, lymph node responder CD4+ cells were suspended in complete culture medium composed of RPMI 1640 supplemented with glutamax-1, PS, gentamycin (160 μg/mL), 10% heat-inactivated FCS and 50 μM β-mercaptoethanol (all purchased from Gibco-BRL). Syngeneic splenocytes treated with mitomycin (50 µg/mL) at 37°C for 105 min were prepared as antigen-presenting cells (APC). Cultures were done in triplicate in flat-bottom microplates by mixing 4 × 105 CD4+ cells and 7 × 105 APC with several dilutions of papain-solubilized native CII 59 (50, 100, 200 µg/mL) or medium alone. After a 3-day incubation at 37°C in a 5% CO2 atmosphere, culture supernatants were collected and assessed for the presence of cytokines by ELISA.
Cytokine assays by specific ELISA
IL-4, IL-10 and IFN-γ levels in serum and in culture supernatants were measured using commercially available ELISA kits (Duoset, R&D Systems, Abingdon, UK), according to the manufacturer's instructions. The sensitivity of cytokine assays was 20 pg/mL and 200 pg/mL in culture supernatants and in serum, respectively.
All statistics were done using the StatView version 5.0 Software. For all results (differences in cytokine production, A max, histological scores, and quantitative clinical scores data on a given day) a Mann-Whitney test was used. Differences were considered significant when p<0.05.
We are grateful to Delphine Lemeiter (UPRES EA-3408), Monique Etienne and Simone Beranger (UPRES EA-3410) for their outstanding technical assistance. We thank Maria Leite-de-Moraes (CNRS UMR 8147) for critical review of the manuscript. We are especially indebted to Kirin Brewery Co, Ltd (Gunma, Japan) for providing α-GalCer, and to Mitchel Kronenberg (La Jolla Institute for Allergy and Immunology, San Diego, CA), Peter Van Endert, and Jean-Marie Fourneau (INSERM U580) for providing plasmid containing CD1d and β2m genes and assisting in the preparation of CD1d/α-GalCer-tetramer, respectively. This work was supported by the Association de Recherche sur la polyarthrite rhumatoïde (ARP), and the Société française de Rhumatologie (SFR), grants from Wyeth and from INSERM (PNRD 2004). This work was also supported by personal grants from the Association pour la Recherche sur le Cancer (R.Z.) and from l'Académie de Médecine (R.Z.).