Stromal cells and osteoclasts are responsible for exacerbated collagen-induced arthritis in interferon- deficient mice.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 52, No. 12, December 2005, pp 3739–3748 DOI 10.1002/art.21496 © 2005, American College of Rheumatology Stromal Cells and Osteoclasts Are Responsible for Exacerbated Collagen-Induced Arthritis in Interferon-␤–Deficient Mice Alexandra P. Treschow, Ingrid Teige, Kutty S. Nandakumar, Rikard Holmdahl, and Shohreh Issazadeh-Navikas Objective. Clinical trials using interferon-␤ (IFN␤) in the treatment of rheumatoid arthritis have shown conflicting results. We undertook this study to understand the mechanisms of IFN␤ in arthritis at a physiologic level. Methods. Collagen-induced arthritis (CIA) was induced in IFN␤-deficient and control mice. The role of IFN␤ was investigated in both the priming and effector phases of the disease. The effect of IFN␤ deficiency on synovial cells, macrophages, and fibroblasts from preimmunized mice was analyzed by flow cytometry, immunohistochemistry, and enzyme-linked immunosorbent assay. Differences in osteoclast maturation were determined in situ by histology of arthritic and naive paws and by in vitro maturation studies of naive bone marrow cells. The importance of IFN␤-producing fibroblasts was determined by transfering fibroblasts into mice at the time of CIA immunization. Results. Mice lacking IFN␤ had a prolonged disease with a higher incidence compared with control mice. IFN␤ deficiency was found to influence the effector phase, but not the priming phase, of arthritis. Compared with control mice, IFN␤-deficient mice had greater infiltration of CD11bⴙ cells and greater production of tumor necrosis factor ␣ in vivo, and their macrophages and fibroblasts were both more activated in vitro. Moreover, IFN␤-deficient mice generated a greater number of osteoclasts in vitro, and mice immunized to induce arthritis, but not naive mice, had a greater number of osteoclasts in vivo compared with control mice. Importantly, IFN␤-competent fibroblasts were able to ameliorate arthritis in IFN␤-deficient recipients. Conclusion. Our data indicate that IFN␤ is involved in regulating the activation state of osteoclasts and stromal cells, including macrophages and fibroblasts, but that it has little effect on T cells. Interferons (IFNs) are potent cytokines that are classified as either type I or type II. IFN␣ and IFN␤ are type I IFNs and are considered to be antiinflammatory. Type I IFNs bind to the same receptor complex, which consists of 2 transmembrane proteins (1,2). The binding of IFN␣ and IFN␤ to the receptor has been shown to have distinct biologic functions (3–6). Apart from its antiviral properties, IFN␤ has been shown to have a wide variety of developmental and immunomodulating effects. IFN␤ has been shown to be involved in the development of B cells, neutrophils, and osteoclasts as well as in inhibition of apoptosis of leukocytes (7–10). However, the immunomodulating effects of IFN␤ have been of greatest interest in terms of its therapeutic use. IFN␤ has been shown to downregulate proinflammatory cytokines, such as tumor necrosis factor ␣ (TNF␣) and interleukin-1␤ (IL-1␤), and to increase the secretion of antiinflammatory mediators, such as IL-10 and IL-1 receptor antagonist (11). IFN␤ has also been implicated in reducing T cell proliferation as well as in down-regulation of class II major histocompatibility complex (MHC) on antigen-presenting cells (APCs) (12,13). Due to its antiinflammatory properties, this cytokine has been studied and used in various human immune disorders such as multiple sclerosis (MS) and cancer (14,15). There has been a positive Supported by the Swedish Research Council–Natural Science, the Swedish Research Council–Medicine, The Swedish Rheumatism Association, the Alfred Österlund Foundation, the Tore Nilson Foundation, King Gustaf V’s 80-Year Foundation, the Royal Physiographic Society in Lund, the M. Bergvalls Foundation, the Åke Wiberg Foundation, the Börje Dahlin Foundation, and the Crafoord Foundation. Alexandra P. Treschow, MSc, Ingrid Teige, PhD, Kutty S. Nandakumar, PhD, Rikard Holmdahl, MD, PhD, Shohreh IssazadehNavikas, PhD: University of Lund, Lund, Sweden. Address correspondence and reprint requests to Alexandra P. Treschow, MSc, Section for Medical Inflammation Research, Institute for Cell and Molecular Biology, University of Lund, BMC I11, S-22184 Lund, Sweden. E-mail: Alexandra.Treschow@med.lu.se. Submitted for publication April 25, 2005; accepted in revised form September 13, 2005. 3739 3740 response with IFN␤ treatment of MS patients, and it is currently the most effective treatment used for this autoimmune disease. In recent years, there has been an interest in determining the possible beneficial effects of IFN␤ in rheumatoid arthritis (RA). IFN␤ has been investigated in mice and monkeys with promising results (15,16), and IFN␤ has been shown to regulate osteoclastogenesis in mice (9,13,17). Moreover, in a small clinical trial, administration of IFN␤ led to a significant reduction in the expression of IL-1␤, matrix metalloproteinase 1 (MMP1), and tissue inhibitor of metalloproteinases 1 in the synovial lining and also to a reduction in CD3⫹ T cell infiltration (18). That study also showed that in vitro RA fibroblast-like synoviocytes also had decreased expression of MMP-1 when treated with IFN␤. Together, these findings indicate that IFN␤ may have a protective effect against joint destruction. In line with this, a recent report showed that IFN␤ is highly expressed in the synovium of RA patients compared with that of patients with osteoarthritis or reactive arthritis (19). Despite the early success of IFN␤ treatment in animal models as well as in small clinical trials, recent clinical studies have shown limited effect (13,18,20,21). These inconsistent results may, however, have several explanations. First, animal models of RA differ from human RA in several aspects, such as in the level of T cell infiltration into arthritic joints. Second, treatment protocols used in animal models of RA, including administration of very high amounts of IFN␤ and administration of retrovirus or transformed fibroblasts that produce IFN␤, are difficult to extrapolate to humans. Consequently, there is a need to gain an enhanced understanding of the mechanism of action of IFN␤ at a physiologic level. The present study utilized IFN␤deficient (IFN␤⫺/⫺) mice in comparison with control heterozygous (IFN␤⫺/⫹) mice in order to determine the effect of IFN␤ in collagen-induced arthritis (CIA), the murine model for RA. CIA is one of the most commonly used animal models for RA, and it is induced in mice by injecting heterologous type II collagen (CII) in an adjuvant, leading to a disease resembling RA (22). T cells have been shown to play an important role in the pathogenesis of CIA (23), possibly by the production of proinflammatory cytokines and by providing help to B cells. Although CIA is T cell dependent, T cells are primarily involved in the priming phase of the disease, whereas the effector phase is driven by B cells producing anti-CII antibodies that crossreact with mouse CII (24). Histologic changes associated with CIA involve the infiltration TRESCHOW ET AL of neutrophils, macrophages, and T cells into the synovia. There is also pannus formation and activation of stromal cells, such as fibroblasts and macrophages. These histologic changes are all believed to contribute to the pathogenesis of CIA. In the present study, we used IFN␤⫺/⫺ mice in order to determine the role of IFN␤ in the CIA model. We found that while the T cell compartment appeared unaffected by IFN␤ deficiency, fibroblasts, macrophages, and osteoclasts located in the joints were more activated compared with those in IFN␤⫺/⫹ mice. Therefore, we postulate that IFN␤ deficiency leads to severe arthritis in which stromal cells and osteoclasts perpetuate the disease. Pinpointing the mechanism of IFN␤ is of importance in determining whether IFN␤ can be used as a treatment for arthritis and, if so, which patient group would benefit most from this treatment. MATERIALS AND METHODS Mice. The generation of IFN␤-deficient mice has been described previously (6). The mice were screened for the deletion of IFN␤ by polymerase chain reaction from tissue (tail or toe of mice) as previously described (25). Mice were backcrossed to the B10.RIII strain for 7 and 12 generations by crossing IFN␤⫺/⫹ mice with B10.RIII mice. Mice were bred and kept at the conventional animal facility at the Section for Medical Inflammation Research, University of Lund, and all experiments had animal ethics committee approval. Unless stated otherwise, 8–16-week-old male mice were used. Antigens. CII was prepared from calf cartilage by pepsin digestion as described previously (26). Peptides were synthesized as described previously (27). CII was denatured by heating at 65°C for 20 minutes before use in in vitro proliferation assays. Immunization, scoring, and anti-CII antibody enzymelinked immunosorbent assay (ELISA). For arthritis experiments, mice (8–14 per group) were immunized intradermally in the base of the tail with 100 g CII emulsified 1:1 in Freund’s incomplete adjuvant (IFA; Difco, Detroit, MI). For in vitro lymphocyte assays, mice were immunized in the base of the tail and in each hind footpad with 60 g of CII in IFA at each location. Arthritis was evaluated by visual scoring using an extended scoring protocol (28); scores ranged from 1 to 15 for each paw with a maximum score of 60 per mouse. Each arthritic toe and knuckle was scored as 1, with a maximum of 10 per paw. An arthritic ankle or midpaw was given a score of 5. The anti-CII antibody response was determined by measuring the level of CII-specific antibodies in serum collected 121 days postimmunization. The amounts of total anti-CII IgG as well as the amounts of the IgG1 and IgG2a isotypes were determined through quantitative ELISA as previously described (29). Collagen antibody–induced arthritis (CAIA). Arthritis in 4-month-old mice (9–12 per group) was induced by injecting intravenously an anti-CII monoclonal antibody cocktail of CIIC1 and M2139 (9 mg/mouse) as previously described (30) PHYSIOLOGY OF IFN␤ IN ARTHRITIS without a lipopolysaccharide (LPS) booster. After 48 hours, clinical signs of arthritis were observed, and the arthritis was monitored for 72 days using the scoring protocol described above. Proliferation and cytokine production assays. Ten days after immunization, cells from the draining inguinal and popliteal lymph nodes were prepared and restimulated in vitro in order to determine antigen-specific cell proliferation and the IFN␥ response as described previously (31,32). Six to 10 mice per group were used. For determination of anti-CD3 T cell responses, spleen cells from naive or immunized (10 days prior) mice were seeded at a concentration of 1 ⫻ 106/well in plates precoated with anti-CD3 (clone 145-2C11; from our hybridoma collection) and incubated for 48 hours before pulsing with 1 Ci 3 H-thymidine (Amersham International, Amsterdam, The Netherlands) for 15–18 hours. Five mice per group were used. Macrophage preparation and culture. Spleens were removed 10 days postimmunization, and macrophages were enriched and stimulated as previously described (25); 8–10 mice per group were used. Supernatants were collected after 36 hours of incubation and assayed for cytokine content using ELISA. The production of TNF␣ was determined using the recommended paired antibodies and the protocol of BD PharMingen (Franklin Lakes, NJ). The plates were read using a fluorometer (Wallac, Boston, MA). Synovia preparation and culture. Mice (24–25 per group) were immunized to induce arthritis as described above. Thirty days postimmunization, mice were killed, hind legs were removed, and the synovia of the knees were dissected out, pooled, and placed in 1.6 mg/ml type IV collagenase (Worthington, Lakewood, NJ) and 0.1% DNase I (Sigma-Aldrich, St. Louis, MO) in Dulbecco’s modified Eagle’s medium (DMEM) and incubated for 1 hour at 37°C. The cells were left untreated or first primed with 10 units/ml of IFN␥ for 60 minutes, then incubated for 36 hours with 50 ng/ml of LPS. The expression of surface markers on the synovial cells was determined by flow cytometry using the following conjugated antibodies: fluorescein isothiocyanate (FITC)– conjugated anti–intercellular adhesion molecule (anti-ICAM) (clone 3E2; BD PharMingen), biotinylated anti–vascular cell adhesion molecule (anti-VCAM) (clone 429; BD PharMingen), phycoerythrin (PE)–conjugated anti-CD11b (clone M1/70; BD PharMingen), allophycocyanin-conjugated anti– Ly6-G (clone Rb6-8C5; BD PharMingen), FITC-conjugated anti–class II MHC (clone Y3P; from our hybridoma collection), and biotinylated anti–macrophage F4/80 antigen (clone F4/80; from our hybridoma collection). In order to determine TNF␣ production, monosine (3 M/ml; Sigma-Aldrich) was added 6 hours prior to staining. Intracellular staining was then performed using BD Cytofix/Cytoperm solution and protocol (Becton Dickinson, Franklin Lakes, NJ) using an unconjugated anti-TNF␣ antibody (clone XT22; BD PharMingen) followed by a biotinylated secondary goat anti-rat antibody (Jackson ImmunoResearch, West Grove, PA). Immunohistochemistry. IFN␤⫺/⫺ and IFN␤⫺/⫹ mice (5 per group) were immunized to induce arthritis; on day 40, mice were killed, and paws were dissected and decalcified with EDTA (for 2–3 weeks). The paws were then embedded in OCT compound (Sakura Finetek Europe, Zoeterwoude, The 3741 Netherlands) and snap-frozen in isopentane on dry ice. Staining of slides was performed as previously described (25) Diaminobenzidine (50 mg/ml; Saveen Biotech, Ideon, Sweden) was used for detection, and slides were counterstained with hematoxylin. In all studies, the numbers of positive cells were determined in a blinded manner by calculating the mean count of 5 distinct areas per section. Fibroblast preparation and culture. Fibroblasts were prepared from IFN␤⫺/⫺ and IFN␤⫺/⫹ mice that had shown clinical signs of arthritis for at least 1 week. The fibroblasts were prepared by removing the skin and muscle from the hind legs (8–10 mice per group) and grinding them in a mortar in a solution of 0.25% trypsin in phosphate buffered saline (PBS). Cells were then incubated for 30 minutes at 37°C, washed with DMEM containing 10% fetal calf serum (FCS), and incubated for an additional 90 minutes in 0.1% collagenase. The fibroblasts were subjected to a minimum of 6 passages (detachment by 0.5% trypsin in 5 mM EDTA) to obtain a pure culture. To analyze the phenotype of the fibroblasts, cells (1 ⫻ 104/well) were seeded into 48-well plates in DMEM containing 10% FCS and cultured for 4 days before being detached with cell dissociation media (Sigma-Aldrich). The IL-6 content in the supernatant of the cultured fibroblasts was determined using the recommended paired antibodies and protocol of BD PharMingen. The expression of cell surface markers on fibroblasts was determined by flow cytometry using the following antibodies: FITC-conjugated anti-ICAM, biotinylated antiVCAM, FITC-conjugated anti–class II MHC, PE-conjugated anti-CD40 (clone 3/23; BD PharMingen), biotinylated antiCD44 (clone IM7.8.1; BD PharMingen), PE-conjugated antiCD71 (clone C2; BD PharMingen), and biotinylated anti– IFN␥ receptor ␣-chain (clone GR20; BD PharMingen). Transfer of fibroblasts. Fibroblasts were detached from culture bottles using EDTA/trypsin and washed with PBS, and a single-cell suspension was obtained by passing the fibroblasts through a 23G needle. The fibroblasts were then injected periarticularly into the joints of mice (total of 2 ⫻ 106 fibroblasts/mouse) at 6 injection sites in the metacarpal, metatarsal, and ankle joints. At the same time, the mice (8–10 per group) were immunized to induce CIA as described earlier. In vitro generation of osteoclasts. Bone marrow cells were obtained from the tibias of 4 IFN␤⫺/⫺ and 4 IFN␤⫺/⫹ mice by removing the bone ends and flushing with ␣-minimum essential medium (Gibco BRL Life Technologies, Gaithersburg, MD). Nonadherent cells were washed, 2.5 ⫻ 105/well were seeded into a 48-well plate, and 10 ng/ml macrophage colony-stimulating factor (M-CSF; R&D Systems, Minneapolis, MN) was added. Three days later, media were removed, fresh media containing M-CSF (10 ng/ml) plus recombinant murine RANKL (100 ng/ml; PeproTech, London, UK) were added, and cells were cultured for an additional 3–4 days. Bone marrow cells incubated with M-CSF alone were used as negative control. The osteoclasts were visualized using tartrate-resistant acid phosphatase (TRAP) staining according to Becton Dickinson Technical Bulletin no. 445 (“Tartrate Resistant Acid Phosphatase staining of osteoclasts”). Cells were counterstained with hematoxylin. Osteoclasts were classified as multinucleated and TRAP positive. In all studies, the numbers of positive cells were determined by calculating the mean count from 5 fields of view per well. 3742 TRESCHOW ET AL In situ determination of osteoclasts using TRAP staining. Naive IFN␤⫺/⫺ and IFN␤⫺/⫹ mice ages 8–16 weeks or ⬎1.5 years as well as preimmunized (40 days prior) IFN␤⫺/⫺ and IFN␤⫺/⫹ mice (4–6 per group) were killed and paws were dissected. The paws were fixed in 4% phosphate buffered formaldehyde for 24 hours at 4°C, decalcified with EDTA (for 2–3 weeks), embedded in paraffin, and sectioned at a thickness of 5 m. The sections were rehydrated and stained for TRAP as described above. All joints in the section were counted, and joints that contained ⱖ1 osteoclast were counted as affected. Thereafter, the number of affected joints per total number of counted joints was determined individually in order to compare the 2 groups of mice. Statistical analysis. The frequency of arthritis was analyzed by chi-square test. The Mann-Whitney U test was used in all other statistical analyses. RESULTS Exacerbation of CIA in IFN␤-deficient mice in the chronic phase of the disease. There has recently been an interest in addressing whether IFN␤ has an ameliorating effect on arthritis, but the results of these investigations have been conflicting. We therefore decided to investigate the effect of IFN␤ deficiency on arthritis in the CIA mouse model. IFN␤⫺/⫺ mice were backcrossed to the B10.RIII background for 7 generations and then investigated for arthritis susceptibility. There was no difference in the incidence, day of onset, or severity of arthritis between mice heterozygous for IFN␤ deficiency (IFN␤⫺/⫹ mice) and IFN␤ wild-type littermates (IFN␤⫹/⫹ mice) (data not shown); therefore, both groups were pooled for subsequent comparison with the group of IFN␤⫺/⫺ mice. Although there was no difference in day of onset, IFN␤⫺/⫺ mice were more susceptible to CIA and developed an exacerbated disease compared with control mice. Relapses of arthritis were also observed in IFN␤⫺/⫺ mice, and they had a tendency toward a higher anti-CII antibody response (Figure 1A and Table 1). A similar disease profile was also observed when mice that had been backcrossed for 12 generations were used in the CIA model (Table 1). T cell response to CII not affected by a lack of endogenous IFN␤. Since the CIA model is T cell dependent, it was feasible that the IFN␤⫺/⫺ mice had an exacerbated arthritis due to an increased T cell response to CII. Furthermore, it has been previously reported that IFN␤ affects the proliferative response of T cells (8,33). Concordant with a recent report (8), naive IFN␤⫺/⫺ spleen cells (and lymph node cells [data not shown]) were found to have a significantly greater proliferative response than those of control mice when stimulated with anti-CD3 (Figure 1B). To evaluate whether Figure 1. Interferon-␤ (IFN␤) deficiency leads to augmented collagen-induced arthritis without altering the antigen-specific T cell response. A, Arthritis index, calculated as the mean arthritis score in IFN␤-deficient mice and control (IFN␤⫺/⫹ and IFN␤⫹/⫹) mice. Mice were scored twice weekly, starting from day 14. Data include a total of 8–14 mice per group. ⴱ ⫽ P ⱕ 0.05; ⴱⴱ ⫽ P ⱕ 0.01 versus control mice. B, Anti-CD3 (␣CD3) response of splenocytes from naive IFN␤⫺/⫺ mice and control IFN␤⫺/⫹ mice. ⴱ ⫽ P ⱕ 0.05 versus IFN␤⫺/⫺ mice. C, Anti-CD3 and bovine type II collagen (bCII) response of splenocytes from preimmunized IFN␤⫺/⫺ mice and control IFN␤⫺/⫹ mice. Cells in B and C were incubated for 48 hours before measurement of 3 H-thymidine (3H-TdR) incorporation. D, Antigen-specific response of lymphocytes from IFN␤⫺/⫺ and control IFN␤⫺/⫹ mice immunized 10 days prior to in vitro cultivation. Lymphocytes were restimulated with 50 g/ml of whole bCII, with 50 g/ml or 10 g/ml of the immunodominant peptide sequence 607–621 of CII (p607), or with 50 g/ml of the immunodominant peptide sequence 442–456 of CII (p442). Cells were cultivated for 72 hours before measurement of 3 H-TdR incorporation. Values in B–D are the mean ⫾ SD. Results shown are from 1 of 3 representative experiments. SI ⫽ stimulation index. IFN␤⫺/⫺ mice also had an increased antigen-specific proliferative response, mice were immunized with CII and cells were subsequently restimulated in vitro with bovine CII and the immunodominant peptides (607–621 PHYSIOLOGY OF IFN␤ IN ARTHRITIS 3743 Table 1. Arthritis parameters and anti-CII antibody response in IFN␤⫺/⫺ and control (IFN␤⫺/⫹ and IFN␤⫹/⫹) mice* IgG total, units/ml† IgG1, units/ml IgG2, units/ml Incidence, %‡ Duration§ Maximum score, 0–60¶ Incidence in 12th generation, %‡ CAIA average score, 0–60# CAIA incidence, %# CAIA duration¶# IFN␤⫺/⫺ mice Control mice P 56 ⫾ 11 33 ⫾ 5 88 ⫾ 54 100 63 ⫾ 6 23 ⫾ 3 90 10.2 ⫾ 2.1 90 25.9 ⫾ 1 37 ⫾ 8 20 ⫾ 3 26 ⫾ 5 43 27 ⫾ 6 12 ⫾ 2 40 4.2 ⫾ 1.7 40 19.8 ⫾ 2.1 0.15 0.08 0.08 0.02 0.003 0.01 0.05 0.05 0.09 0.04 * Except where indicated otherwise, values are the mean ⫾ SD. CII ⫽ type II collagen; IFN␤⫺/⫺ ⫽ interferon-␤ deficient; CAIA ⫽ collagen antibody–induced arthritis. † Mean total CII IgG as well as IgG subclass levels in sera collected on day 121 were calculated as arbitrary units/ml using a polyclonal serum. ‡ Calculated from day 65 after immunization. § The average duration of disease was calculated from the first day of evident clinical signs until the end of the experiment (unaffected mice were assigned a value of 0). ¶ The maximum score of affected mice was calculated as the average of the highest scores obtained in mice showing clinical signs of arthritis. # Calculated on day 21 after transfer of CII-specific antibodies. and 442–456). However, there was no significant difference between the IFN␤⫺/⫺ and control mice in their proliferative responses (Figures 1C and D) or in production of IFN␥ (data not shown). Similar results were also observed in mice immunized 3 weeks prior (data not shown). Furthermore, the detected difference in antiCD3 stimulation in naive mice (Figure 1B) was not observed in immunized mice (Figure 1C). Effect of lack of IFN␤ on the effector phase of the disease. Data so far indicated that IFN␤ deficiency did not have an effect in the priming phase of CIA. Instead, the exacerbated disease of IFN␤⫺/⫺ mice may be explained by events occurring in the effector phase. The effector phase of CIA is mediated via arthritogenic anti-CII antibodies, which can be mimicked by using the acute and T cell–independent CAIA model (34). IFN␤⫺/⫺ and control mice were subjected to passive transfer of the disease using 2 collagen-specific monoclonal antibodies. As shown in Table 1, IFN␤⫺/⫺ mice were indeed found to have developed a more severe and prolonged disease compared with that in control mice, indicating that IFN␤ operates in the inflammatory phase in the joints. Augmented activation in vitro of peripheral and synovial APCs in IFN␤-deficient mice. Since the enhanced arthritis in the IFN␤⫺/⫺ mice was not due to an increase in T cell proliferation and could not be explained by an increase in anti-CII antibody production, the exacerbation had to be due to other cells. We therefore investigated cells that could be activated by class II MHC–restricted T cells. Initial investigation of the spleen macrophage population showed a significant increase in TNF␣ production (Figure 2A) following 48 hours of culture in vitro in the presence of both IFN␥ and LPS, but there was no alteration in the level of IL-1␤ or IL-10 production (data not shown). However, since splenic macrophages are distant from the site of joint inflammation, we aimed to investigate whether the macrophages as well as other cells in the synovia were more activated in the IFN␤⫺/⫺ mice. Mice were immunized, and 30 days later the synovia were extracted and stimulated in vitro with IFN␥ and LPS. Following activation, flow cytometric analyses of synoviocytes from IFN␤⫺/⫺ and IFN␤⫺/⫹ mice revealed that the IFN␤⫺/⫺ synoviocytes included a greater number of macrophages; numbers of these increased slightly after stimulation (Table 2). To our surprise, we did not see an increase in TNF␣ intracellular staining when the synovial cells were stimulated with IFN␥ and LPS. This result was in contrast to that for the macrophages derived from the spleen (Figure 2A). This could have been due to a kinetic problem. However, in the synovial population, there was a greater intracellular expression of TNF␣ in IFN␤⫺/⫺ synoviocytes compared with that in control cells, both before and after stimulation. There was also an increase in ICAM-1⫹ and in ICAM- 3744 TRESCHOW ET AL Figure 2. Preimmunized mice deficient in interferon-␤ (IFN␤) have greater production of tumor necrosis factor ␣ (TNF␣) in spleen-derived macrophages and in joints. A, Change in production of TNF␣ (delta TNF␣; calculated as cytokine production by macrophages in the presence of lipopolysaccharide [LPS], IFN␥, or LPS and IFN␥ minus cytokine production by macrophages cultured in media alone). Results shown are from 1 of 3 representative experiments. B, Number of cells staining positive for CD11b and TNF␣ in hind or front paws of IFN␤⫺/⫺ (n ⫽ 5) and control IFN␤⫺/⫹ (n ⫽ 5) mice, all having similar symptoms of arthritis. For each mouse, 5 distinct fields on a single sample slide were counted, and the average number of cells was determined. Histologic findings were obtained from 1 collagen-induced arthritis experiment. Values in A and B are the mean ⫾ SEM. ⴱ ⫽ P ⱕ 0.05 versus IFN␤⫺/⫺ mice. C and D, Immunohistochemistry 40 days after immunization of 1 representative IFN␤⫺/⫺ mouse and 1 control IFN␤⫺/⫹ mouse, respectively, with comparable clinical symptoms of arthritis at the time of analyses. Sections were stained for TNF␣, with positive cells staining brown (original magnification ⫻ 200). 1⫹,VCAM-1⫹ cells after stimulation, compared with stimulated control cells (Table 2). Interestingly, after 36 hours of cultivation, there was a greater number of neutrophils (Ly6-G⫹ and CD11b⫹ cells) in the IFN␤⫺/⫺ synovial population, which increased after stimulation (Table 2). However, fresh synoviocytes originating from IFN␤⫺/⫹ mice had a greater Table 2. mice* number of neutrophils compared with synoviocytes from IFN␤⫺/⫺ mice (34.7% versus 25.8%). This result was expected, since IFN␤⫺/⫺ mice have previously been shown to have a reduced number of neutrophils (8). Increased infiltration and fibroblast activation in synovia of IFN␤ⴚ/ⴚ mice. To investigate the phenotype of joint inflammation during the effector phase of CIA, Expression levels of cell surface markers on synovial cells from IFN␤⫺/⫺ and control (IFN␤⫺/⫹) IFN␤⫺/⫺ mice F4/80 TNF␣ CD11b Ly6-G ⫹ CD11b ICAM-1 VCAM-1 ICAM-1 ⫹ VCAM-1 Control mice Media LPS ⫹ IFN␥ ⌬† Media LPS ⫹ IFN␥ ⌬† 20.3 32.8 24.5 8.4 17.8 43.5 11.4 22.1 29.5 27.7 11.9 27.9 42.0 19.6 1.8 ⫺3.3 3.2 3.5 10.1 ⫺1.5 8.2 17.6 23.9 16.1 2.1 31.6 36.1 22.8 16.6 18.3 19.0 2.4 35.0 28.5 23.4 ⫺1.0 ⫺5.6 2.9 0.3 3.4 ⫺7.6 0.6 * Values are arbitrary units. Cells from 24–25 mice per group were pooled and stimulated. The experiment was performed twice with similar results, and data shown are from 1 representative experiment. IFN␤⫺/⫺ ⫽ interferon-␤ deficient; LPS ⫽ lipopolysaccharide; TNF␣ ⫽ tumor necrosis factor ␣; ICAM-1 ⫽ intercellular adhesion molecule 1; VCAM-1 ⫽ vascular cell adhesion molecule 1. † Change in expression was calculated by subtracting the expression level of cells cultured in media from that of cells cultured with LPS ⫹ IFN␥. PHYSIOLOGY OF IFN␤ IN ARTHRITIS hind and fore paws were harvested 40 days postimmunization and were evaluated by immunohistochemistry. Concordant with the above data (Table 2 and Figure 2A), infiltrated areas of IFN␤⫺/⫺ mice contained more CD11b⫹ cells and a greater amount of TNF␣, compared with control mice (Figures 2B–D). To determine whether fibroblasts were also affected by IFN␤ deficiency, fibroblasts were prepared from mice with clinical signs of arthritis lasting for at least 1 week and were then subjected to 6 passages of trypsinization in vitro before analyses. Fibroblasts from the IFN␤⫺/⫺ mice produced 10-fold higher amounts of IL-6 compared with fibroblasts from control mice when both were cultured for 4 additional days (Figure 3A). Flow cytometric analyses of the fibroblasts also showed that the IFN␤⫺/⫺ fibroblasts were more activated than the control fibroblasts in terms of expression of CD44 and ICAM and had a slightly increased expression of CD40 (Figure 3A). Ability of IFN␤-competent fibroblasts to protect against CIA induction in IFN␤-deficient mice. To investigate whether the increased activation status in vitro of IFN␤⫺/⫺ fibroblasts would also have an impact in vivo during an inflammatory attack directed to the joints, we next conducted fibroblast transfer experiments. Neither IFN␤⫺/⫺ fibroblasts nor control fibroblasts could induce arthritis when injected into the knee (1 ⫻ 105/knee) of irradiated or nonirradiated B10.RIII mice (data not shown). However, transfer of control fibroblasts (6 joint injection sites, with a total of 2 ⫻ 106/mouse) into IFN␤⫺/⫺ mice resulted in significant protection from subsequent induction of CIA, compared with IFN␤⫺/⫺ mice injected with IFN␤⫺/⫺ fibroblasts (Figures 3B and C). The IFN␤⫺/⫺ mice that had received control fibroblasts had an arthritis profile similar to that of control IFN␤⫺/⫹ mice that had received control fibroblasts, demonstrating the importance of fibroblasts in this model, possibly via production of IFN␤. Difference in osteoclast generation in vitro and in vivo in IFN␤-deficient mice. Another important synovial cell that contributes to the arthritis process is the osteoclast, which degrades cartilage and bone. It has previously been shown that IFN␤-deficient and IFN␤ receptor–knockout mice have a greater capacity to generate osteoclasts in vitro and that these mice have an intrinsic bone erosion disorder in vivo (9). We therefore analyzed osteoclastogenesis in IFN␤⫺/⫺ mice both in vitro and in vivo. Bone marrow from IFN␤⫺/⫺ mice generated more osteoclasts than did bone marrow from control mice in vitro (Figures 4A–C). However, TRAP staining analyses did not suggest an increase in osteoclastogenesis in vivo either in 4-month-old or in 16-month-old naive IFN␤⫺/⫺ mice compared with age-matched naive 3745 Figure 3. Amelioration of arthritis by reconstitution of interferon-␤ (IFN␤)–deficient mice with IFN␤-competent fibroblasts. A, Percent of fibroblasts staining positive for different cell surface markers (y-axis at left) and interleukin-6 (IL-6) production following in vitro culture of fibroblasts from IFN␤⫺/⫺ and control mice (y-axis at right). Error bars represent the SD. B and C, Arthritis index (mean score) and incidence of arthritis, respectively, in IFN␤⫺/⫺ mice that had received 2 ⫻ 106 fibroblasts (FB) either from IFN␤⫺/⫺ mice (IFN␤⫺/⫺ ⫹ IFN␤⫺/⫺ FB) or from control mice (IFN␤⫺/⫺ ⫹ IFN␤⫺/⫹ FB), and in control IFN␤⫺/⫹ mice injected with control IFN␤⫺/⫹ fibroblasts (IFN␤⫺/⫹ ⫹ IFN␤⫺/⫹ FB). All mice were injected with fibroblasts and immunized with type II collagen on day 0. Mice were scored twice weekly, starting from day 14. Data are from a total of 8–10 mice per group pooled from 2 separate experiments, each with balanced groups. ⴱ ⫽ P ⱕ 0.05 for IFN␤⫺/⫺ fibroblasts versus control IFN␤⫺/⫹ fibroblasts. ICAM-1 ⫽ intercellular adhesion molecule 1; VCAM-1 ⫽ vascular cell adhesion molecule 1; IFN␥R␣ ⫽ IFN␥ receptor ␣-chain. control mice (data not shown). Therefore, in contrast to the previous report (9), the IFN␤⫺/⫺ mice on the B10.RIII background did not show signs of an intrinsic bone erosion disorder. Analyses of arthritic joints showed a clear in- 3746 TRESCHOW ET AL Figure 4. Enhanced generation of osteoclasts in interferon-␤ (IFN␤)– deficient mice. A and B, In vitro generation of osteoclasts from bone marrow (BM) cells from IFN␤⫺/⫺ or control IFN␤⫺/⫹ mice, respectively. Cells were cultured in vitro with macrophage colony-stimulating factor (M-CSF) and recombinant murine RANKL (RL) for 6–7 days in 48-well plates (original magnification ⫻ 100). C, Average number of osteoclasts generated from bone marrow from 4 IFN␤⫺/⫺ and 4 control IFN␤⫺/⫹ mice. Cells were seeded in duplicate, and 5 fields of view of each well were counted under 200⫻ magnification. Bone marrow cells from IFN␤⫺/⫺ or control IFN␤⫺/⫹ mice cultured in M-CSF alone did not support the generation of osteoclasts and were used as negative control for the experiment (data not shown). D, In vivo staining of osteoclasts (tartrate-resistant acid phosphatase [TRAP] positive and multinucleated). Paws from preimmunized mice (30 days prior) were removed and prepared for paraffin sectioning and stained for TRAP. All joints in the section were counted, and joints that contained ⱖ1 osteoclast were counted as affected. Consequently, the graph represents the number of affected joints divided by the total number of joints counted in both hind and front paws. Values are the mean ⫾ SD (4–6 mice per group). ⴱ ⫽ P ⱕ 0.05 versus IFN␤⫺/⫺ mice. crease in the number of osteoclasts in IFN␤⫺/⫺ mice compared with control mice (Figure 4D), suggesting that IFN␤ plays an important role in down-regulating inflammation-mediated osteoclastogenesis. DISCUSSION IFN␤ has been used in the therapy of various diseases, such as cancer and viral infections, and it is one of the few available treatments for MS (14,35). There has also been a great deal of interest in the possible therapeutic effects of IFN␤ in RA, but there is a need to gain a better understanding of the mechanisms of IFN␤ in arthritis. We have previously shown in experimental autoimmune en- cephalomyelitis, a murine model of MS, that IFN␤ has the greatest effects in reducing the activation of macrophages and microglia, with little effect on T cells. Moreover, IFN␤ deficiency did not cause a shift in the T helper phenotype (25,36). In accordance with this, in the present study we found that a lack of IFN␤ in arthritis led to a greater activation of stromal cells such as macrophages and fibroblasts as well as to an enhanced generation of osteoclasts in the arthritic joints, and we found that IFN␤ had little effect on antigen-specific T cell responses. IFN␤ has not only been shown to have antiinflammatory effects, but it has also been suggested to be involved in development, homeostasis, and apoptosis of several cell populations, such as osteoclasts, T cells, neutrophils, and B cells (8). Osteoclasts are cells that degrade bone, and they are vital in maintaining bone homeostasis; however, excessive osteoclastogenesis in the arthritic joints leads to a net loss of cartilage and bone. Takayanagi et al elegantly showed the importance of IFN␤ in osteoclastogenesis, with a clear increase in osteoclastogenesis both in vitro and in vivo in naive mice lacking the IFN␤ receptor and in IFN␤⫺/⫺ mice (37). In the current study, we also show in vitro that IFN␤⫺/⫺ mice have enhanced osteoclastogenesis. However, we did not see an increase in osteoclastogenesis in vivo in naive mice, but instead we found that immunized IFN␤⫺/⫺ mice have a significant increase in the number of osteoclasts, indicating that IFN␤ in an arthritic joint would be beneficial in reducing the amount of joint destruction. IFN␤ has been shown to be involved in apoptosis of T cells and neutrophils (7,10). Together with stromal cell–derived factor 1 (CXCL12), IFN␤ was shown to inhibit apoptosis of T cells located in the joint synovia of human RA patients and was therefore believed to maintain T cells within the joint (7). However, the role of T cells in RA synovia is not known, and it is possible that T cells are involved in the priming phase of arthritis but have little role in the effector phase of the disease, since T cells have been shown to proliferate poorly and secrete few cytokines (38). In addition, anti–T cell agents seem to have little influence on ongoing arthritis (39,40), whereas anti–B cell and antimonokine reagents such as anti-TNF␣ have had considerable therapeutic effects (41,42). In the present study, the IFN␤⫺/⫺ mice showed no difference in T cell numbers compared with control mice, indicating that there was no increase in apoptosis in the T cell compartment. Furthermore, there was no difference in antigen-specific T cell proliferation upon restimulation in vitro. In addition, there was no difference in the degree of T cell infiltration in the synovia of these mice (data not shown). Neutrophils have been suggested to be involved PHYSIOLOGY OF IFN␤ IN ARTHRITIS both in RA and in CIA (10,43). In line with this, IFN␤ has been shown to prevent apoptosis of neutrophils (10), and IFN␤⫺/⫺ mice have previously been shown to have a reduced number of neutrophils (8). Indeed, in the current study, we observed reduced numbers of neutrophils in both spleen and lymph nodes (data not shown). However, a prominent role of neutrophils was not found, since IFN␤⫺/⫺ mice developed a more exacerbated arthritis. Interestingly, the neutrophil populations in the spleen and synovia of IFN␤-deficient mice were able to withstand (and even proliferate in) in vitro culture (36 hours), which was in contrast to IFN␤⫺/⫹ neutrophils (data not shown and Table 2). This indicated that the influence of IFN␤ on neutrophil apoptosis in vitro differs from that in vivo. In the present study, lack of IFN␤ expression seemed to have a prominent effect on stromal cells, including macrophages and fibroblasts. Flow cytometric analyses of synoviocytes showed that, upon stimulation with LPS and IFN␥, there was an up-regulation of CD11b⫹ cells and an increase in macrophages as well as an up-regulation of the adhesion molecule ICAM-1. This indicates that the stromal cells in the synovia are more readily activated in IFN␤-deficient mice, potentially leading to a more chronic arthritis profile. Fibroblasts have been implicated in both the priming and effector phases of RA (7,44); therefore, targeting fibroblasts should be of therapeutic benefit. In the present study, a lack of IFN␤ was associated with fibroblasts having a more active phenotype with increased IL-6 production. There was also an upregulation in the cell surface molecules CD44, CD40, and ICAM-1, all of which are believed to be involved in the pathogenesis of RA (44). However, since fibroblasts require IFN␤ to produce IFN␣ (6), it is possible that the phenotype of the IFN␤-deficient fibroblast is partially due to IFN␣ deficiency. Nevertheless, the current study determined that local injection of IFN␤-producing fibroblasts into the joints of IFN␤⫺/⫺ mice completely reverted the augmented arthritis phenotype to that of control mice. This is an important finding, since it shows that physiologic levels of IFN␤ produced by nontransfected fibroblasts can have a beneficial effect in vivo. In summary, the present study shows that mice deficient in IFN␤ display a chronic arthritis with a high incidence. These results may have been expected from previous animal studies using IFN␤ treatments. However, we were able to demonstrate that the mechanism of action of IFN␤ is not mediated through T cells but rather results from an increased activation of resident cells of the joint (i.e., fibroblasts, macrophages, and osteoclasts). It is likely that IFN␤ serves to control the activation state of fibro- 3747 blasts, and in the absence (or in the presence of low levels) of IFN␤, fibroblasts become more prone to produce cytokines, chemokines, and growth factors that in turn enhance infiltration of inflammatory cells. In fact, it has previously been shown that treatment of RA-derived fibroblasts in vitro with IFN␤ leads to decreased production of chemokines, such as MMP-1 and MMP-2, as well as to decreased production of prostaglandin E2 (45). The mouse model used in this study thus supported the conclusion that fibroblasts have an important role as producers of IFN␤. We were able to show via transfer of IFN␤-competent fibroblasts that increasing the amount of IFN␤ indeed has a profound effect on arthritis. This is an important finding, since it has recently been proposed that naturally produced IFN␤ plays an antiinflammatory role in RA patients (19). There is no doubt that IFN␤ has potent antiinflammatory properties. However, it remains unclear which cell type and which phase of arthritis is most receptive to IFN␤ treatment, and this may explain the limited success of clinical trials. Animal models can aid in pinpointing mechanism(s) of this cytokine in order to determine the most effective treatment protocol and the group of patients that would derive the most benefit. ACKNOWLEDGMENTS We express our gratitude to Dr. J. Bäcklund and Caroline Parsons for critical reading of the manuscript. REFERENCES 1. Uze G, Lutfalla G, Gresser I. Genetic transfer of a functional human interferon ␣ receptor into mouse cells: cloning and expression of its cDNA. Cell 1990;60:225–34. 2. Novick D, Cohen B, Rubinstein M. The human interferon ␣/␤ receptor: characterization and molecular cloning. Cell 1994;77: 391–400. 3. Van den Broek MF, Muller U, Huang S, Zinkernagel RM, Aguet M. Immune defence in mice lacking type I and/or type II interferon receptors. Immunol Rev 1995;148:5–18. 4. Shen H, Zhang M, Minuk GY, Gong Y. Different effects of rat interferon ␣, ␤ and ␥ on rat hepatic stellate cell proliferation and activation. BMC Cell Biol 2002;3:9–17. 5. Deonarain R, Alcami A, Alexiou M, Dallman MJ, Gewert DR, Porter AC. Impaired antiviral response and ␣/␤ interferon induction in mice lacking ␤ interferon. J Virol 2000;74:3404–9. 6. Erlandsson L, Blumenthal R, Eloranta ML, Engel H, Alm G, Weiss S, et al. Interferon-␤ is required for interferon-␣ production in mouse fibroblasts. Curr Biol 1998;8:223–6. 7. Buckley CD. Michael Mason prize essay 2003: why do leucocytes accumulate within chronically inflamed joints? Rheumatology (Oxford) 2003;42:1433–44. 8. Deonarain R, Verma A, Porter AC, Gewert DR, Platanias LC, Fish EN. Critical roles for IFN-␤ in lymphoid development, myelopoiesis, and tumor development: links to tumor necrosis factor ␣. Proc Natl Acad Sci U S A 2003;100:13453–8. 9. Takayanagi H, Kim S, Taniguchi T. Signaling crosstalk between 3748 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. RANKL and interferons in osteoclast differentiation. Arthritis Res 2002;4 Suppl 3:S227–32. Wang K, Scheel-Toellner D, Wong SH, Craddock R, Caamano J, Akbar AN, et al. Inhibition of neutrophil apoptosis by type 1 IFN depends on cross-talk between phosphoinositol 3-kinase, protein kinase C-delta, and NF-B signaling pathways. J Immunol 2003; 171:1035–41. Jungo F, Dayer JM, Modoux C, Hyka N, Burger D. IFN-␤ inhibits the ability of T lymphocytes to induce TNF-␣ and IL-1␤ production in monocytes upon direct cell-cell contact. Cytokine 2001;14:272–82. Rep MH, Hintzen RQ, Polman CH, van Lier RA. Recombinant interferon-␤ blocks proliferation but enhances interleukin-10 secretion by activated human T-cells. J Neuroimmunol 1996;67:111–8. Van Holten J, Reedquist K, Sattonet-Roche P, Smeets TJ, PlaterZyberk C, Vervoordeldonk MJ, et al. Treatment with recombinant interferon-␤ reduces inflammation and slows cartilage destruction in the collagen-induced arthritis model of rheumatoid arthritis. Arthritis Res Ther 2004;6:R239–49. Abdul-Ahad AK, Galazka AR, Revel M, Biffoni M, Borden EC. Incidence of antibodies to interferon-␤ in patients treated with recombinant human interferon-␤ 1a from mammalian cells. Cytokines Cell Mol Ther 1997;3:27–32. Tak PP, Hart BA, Kraan MC, Jonker M, Smeets TJ, Breedveld FC. The effects of interferon ␤ treatment on arthritis. Rheumatology (Oxford) 1999;38:362–9. Triantaphyllopoulos KA, Williams RO, Tailor H, Chernajovsky Y. Amelioration of collagen-induced arthritis and suppression of interferon-␥, interleukin-12, and tumor necrosis factor ␣ production by interferon-␤ gene therapy. Arthritis Rheum 1999;42:90–9. Van Holten J, Plater-Zyberk C, Tak PP. Interferon-␤ for treatment of rheumatoid arthritis? Arthritis Res 2002;4:346–52. Smeets TJ, Dayer JM, Kraan MC, Versendaal J, Chicheportiche R, Breedveld FC, et al. The effects of interferon-␤ treatment of synovial inflammation and expression of metalloproteinases in patients with rheumatoid arthritis. Arthritis Rheum 2000;43:270–4. Holten JV, Smeets TJ, Blankert P, Tak PP. Expression of interferon ␤ (IFN-␤) in synovial tissue from rheumatoid arthritis patients compared to osteoarthritis and reactive arthritis patients. Ann Rheum Dis 2005. E-pub ahead of print. Tak PP. IFN-␤ in rheumatoid arthritis. Front Biosci 2004;9: 3242–7. Van Holten J, Pavelka K, Vencovsky J, Stahl H, Rozman B, Genovese M, et al. A multicentre, randomised, double-blind, placebo controlled phase II study of subcutaneously administered interferon ␤1a in the treatment of patients with active rheumatoid arthritis. Ann Rheum Dis 2004;64:64–9. Courtenay JS, Dallman MJ, Dayan AD, Martin A, Mosedale B. Immunisation against heterologous type II collagen induces arthritis in mice. Nature 1980;283:666–8. Corthay A, Johansson A, Vestberg M, Holmdahl R. Collageninduced arthritis development requires ␣ ␤ T cells but not ␥ delta T cells: studies with T cell-deficient (TCR mutant) mice. Int Immunol 1999;11:1065–73. Mo JA, Holmdahl R. The B cell response to autologous type II collagen: biased V gene repertoire with V gene sharing and epitope shift. J Immunol 1996;157:2440–8. Teige I, Treschow A, Teige A, Mattsson R, Navikas V, Leanderson T, et al. IFN-␤ gene deletion leads to augmented and chronic demyelinating experimental autoimmune encephalomyelitis. J Immunol 2003;170:4776–84. Miller EJ. Structural studies on cartilage collagen employing limited cleavage and solubilization with pepsin. Biochemistry 1972;11:4903–9. Michaelsson E, Andersson M, Engstrom A, Holmdahl R. Identi- TRESCHOW ET AL 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. fication of an immunodominant type-II collagen peptide recognized by T cells in H-2q mice: self tolerance at the level of determinant selection. Eur J Immunol 1992;22:1819–25. Holmdahl R, Carlsen S, Mikulowska A, Vestberg M, Brunsberg U, Hansson A, et al. Genetic analysis of mouse models for rheumatoid arthritis. In: Adolph KW, editor. Human genome methods. New York: CRC Press; 1998. p. 215–38. Holmdahl R, Klareskog L, Andersson M, Hansen C. High antibody response to autologous type II collagen is restricted to H-2q. Immunogenetics 1986;24:84–9. Nandakumar KS, Svensson L, Holmdahl R. Collagen type IIspecific monoclonal antibody-induced arthritis in mice: description of the disease and the influence of age, sex, and genes. Am J Pathol 2003;163:1827–37. Svensson L, Jirholt J, Holmdahl R, Jansson L. B cell-deficient mice do not develop type II collagen-induced arthritis (CIA). Clin Exp Immunol 1998;111:521–6. Malmstrom V, Michaelsson E, Burkhardt H, Mattsson R, Vuorio E, Holmdahl R. Systemic versus cartilage-specific expression of a type II collagen-specific T-cell epitope determines the level of tolerance and susceptibility to arthritis. Proc Natl Acad Sci U S A 1996;93:4480–5. Noronha A, Toscas A, Jensen MA. Interferon ␤ decreases T cell activation and interferon ␥ production in multiple sclerosis. J Neuroimmunol 1993;46:145–53. Nandakumar KS, Backlund J, Vestberg M, Holmdahl R. Collagen type II (CII)-specific antibodies induce arthritis in the absence of T or B cells but the arthritis progression is enhanced by CIIreactive T cells. Arthritis Res Ther 2004;6:R544–50. Giovannoni G, Miller DH. Multiple sclerosis and its treatment. J R Coll Physicians Lond 1999;33:315–22. Matheu V, Treschow A, Navikas V, Issazadeh-Navikas S. Upregulation of B7 molecules (CD80 and CD86) and exacerbated eosinophilic pulmonary inflammatory response in mice lacking the IFN-␤ gene. J Allergy Clin Immunol 2003;111:550–7. Takayanagi H, Ogasawara K, Hida S, Chiba T, Murata S, Sato K, et al. T-cell-mediated regulation of osteoclastogenesis by signalling cross-talk between RANKL and IFN-␥. Nature 2000;408:600–5. Korganow AS, Ji H, Mangialaio S, Duchatelle V, Pelanda R, Martin T, et al. From systemic T cell self-reactivity to organspecific autoimmune disease via immunoglobulins. Immunity 1999;10:451–61. Firestein GS, Zvaifler NJ. How important are T cells in chronic rheumatoid synovitis? [editorial]. Arthritis Rheum 1990;33:768–73. Fox DA. The role of T cells in the immunopathogenesis of rheumatoid arthritis: new perspectives [review]. Arthritis Rheum 1997;40:598–609. Kazkaz H, Isenberg D. Anti B cell therapy (rituximab) in the treatment of autoimmune diseases. Curr Opin Pharmacol 2004;4: 398–402. Ehrenstein MR, Evans JG, Singh A, Moore S, Warnes G, Isenberg DA, et al. Compromised function of regulatory T cells in rheumatoid arthritis and reversal by anti-TNF␣ therapy. J Exp Med 2004;200:277–85. Suzuki M, Uetsuka K, Shinozuka J, Nakayama H, Doi K. Immunohistochemical study on type II collagen-induced arthritis in DBA/1J mice. Exp Anim 1997;46:259–67. Ritchlin C. Fibroblast biology: effector signals released by the synovial fibroblast in arthritis. Arthritis Res 2000;2:356–60. Taylor PC, Peters AM, Paleolog E, Chapman PT, Elliott MJ, McCloskey R, et al. Reduction of chemokine levels and leukocyte traffic to joints by tumor necrosis factor ␣ blockade in patients with rheumatoid arthritis. Arthritis Rheum 2000;43:38–47.