Suppression of autoimmune arthritis in interleukin-1-deficient mice in which T cell activation is impaired due to low levels of CD40 ligand and OX40 expression on T cells

Authors


Abstract

Objective

To elucidate the roles of interleukin-1 (IL-1) in the development of 2 etiologically different rheumatoid arthritis (RA) models: the type II collagen (CII)–induced arthritis (CIA) model and the human T cell leukemia virus type I transgenic (HTLV-I Tg) mouse model.

Methods

For the CIA model, DBA/1J-background IL-1α−/−, IL-1β−/−, IL-1α/β−/−, and wild-type littermate mice were immunized with CII. For the HTLV-I Tg model, BALB/c IL-1β−/− or IL-1α/β−/− mice were crossed with HTLV-I Tg mice. The effects of IL-1 deficiency were assessed as follows: Development of arthritis was assessed both macroscopically and microscopically. Serum antibody titer was measured by enzyme-linked immunosorbent assay. Proliferative response of lymph node cells was assayed by measurement of 3H-thymidine incorporation. Expression of T cell surface molecule CD40 ligand (CD40L) and OX40 was determined by multicolor flow cytometric analysis.

Results

The development of arthritis was markedly suppressed in IL-1α/β−/− mice in both models, although the effect was less prominent in HTLV-I Tg mice. Deficiency of only IL-1α or only IL-1β was also associated with disease suppression. Antibody production after immunization with CII was normal in IL-1α/β−/− mice, while autoantibody production was suppressed in IL-1α/β−/− HTLV-I Tg mice. In IL-1α/β−/− mice, the T cell proliferative response against CII was greatly reduced in both the CIA and the HTLV-I Tg models, suggesting inefficiency of T cell activation. Furthermore, expression of CD40L and OX40 on T cells was greatly reduced in IL-1α/β−/− mice.

Conclusion

These observations suggest that T cell activation by IL-1 is important for the development of autoimmunity and arthritis in these mice.

Rheumatoid arthritis (RA) is a serious medical problem, with ∼1% of the people in the world affected. The disease is autoimmune in nature and characterized by chronic inflammation of the synovial tissue in multiple joints, which leads to joint destruction (1). It is known that expression of various proinflammatory cytokines, such as interleukin-1 (IL-1), IL-6, and tumor necrosis factor α (TNFα), is augmented in the joints of RA patients (2, 3), suggesting involvement of these cytokines in the pathogenesis of the disease. In particular, IL-1 seems to play an important role, because it induces inflammation and acute-phase responses, activation of the immune system including thymocyte maturation and Th2 cell proliferation, enhancement of bone metabolism (by activating osteoclasts to secrete metalloprotease), and febrile response (see review, see ref. 4).

IL-1 is produced by various types of cells, including macrophages, monocytes, and synovial lining cells (5). Three genes are included in the IL-1 family, i.e., the genes for IL-1α, IL-1β, and IL-1 receptor antagonist (IL-1Ra). IL-1α and IL-1β exert similar biologic activities through IL-1 receptor type I (IL-1RI), while IL-1Ra is a naturally occurring inhibitor of IL-1 and competes for the same receptor (6). It has been shown that IL-1 can induce many other genes. For example, we have demonstrated that IL-1β induces the IL-1α gene, and vice versa (7). IL-1 also induces IL-6 and cyclooxygenase 2 genes (8). It is thought that these factors, together with IL-1 or in a cascade process, exhibit pleiotropic effects in animals.

Recently, we found that IL-1Ra-deficient mice spontaneously develop autoimmune arthritis (9). Furthermore, it has been reported that IL-1α causes arthritis when it is injected into rabbit joints (10), and treatment with anti–IL-1α/β as well as with anti–IL-β antibody alone, but not with anti–IL-1α antibody alone, has been shown to ameliorate experimental collagen-induced arthritis (CIA) in mice (11). These observations suggest the importance of IL-1 in the development of arthritis. However, to date, roles of IL-1 have not been elucidated completely. In this study using 2 etiologically different RA models, we analyzed the function of IL-1 in the development of autoimmunity, and, using gene-deficient mice, we investigated the roles of IL-1α and IL-β.

The CIA mouse model is one of the most well-established models for RA (12, 13). CIA can be induced in susceptible rodents by intradermal injections of homologous or heterologous native type II collagen (CII). Susceptibility to the disease is dependent on class II major histocompatibility complex (MHC) haplotypes, and only mice with H-2q and H-2r haplotypes respond to immunization with CII and develop arthritis (14). Since treatment of mice with either anti–T cell receptor monoclonal antibody (mAb)or anti–CD4 mAb before immunization abrogates development of the disease and CD4+ T cell clones reactive with CII transfer the disease to naive mice, it is believed that class II MHC-restricted CD4+ T cell-mediated immune reactions against CII cause the disease. Furthermore, since levels of antibody against CII have been found to correlate with the development of arthritis (14) and transfer of the antibody against CII can induce arthritis (15), most investigators believe that both cellular and humoral immunity to CII are necessary for the full development of CIA. Since antibodies against CII are also detected in humans, this model is considered to represent some aspects of the effector phase of RA (16, 17).

The human T cell leukemia virus type I transgenic (HTLV-I Tg) mouse model is another RA model (18). These mice carry the env-pX region of HTLV-I, the etiologic agent of adult T cell leukemia (for review, see refs. 19 and 20). They spontaneously develop marked synovial and periarticular inflammation with articular erosion caused by invasion of granulation tissue (21), and they produce high levels of autoantibodies including those against IgG, CII, and heat-shock proteins (22). These abnormalities closely resemble those in human RA. Moreover, epidemiologic findings have suggested involvement of this virus in some proportion of the population with RA (23). Thus, the HTLV-I Tg mouse model is a unique RA model in which arthritis develops spontaneously via the same pathogen as one found in RA. It has been suggested that this model involves autoimmune pathogenesis, based on the findings that high levels of autoantibodies were detected in the serum, CII-specific oligoclonal T cells accumulated in the joints, disease onset was suppressed in nude mice, and adoptive transfer of bone marrow cells induced arthritis in nontransgenic mice (22, 24–26). We have also found augmented expression of proinflammatory cytokines, including IL-1, in the joints of HTLV-I Tg mice (22). The pathogenesis of the autoimmunity, however, has not yet been completely elucidated.

In this study, we investigated the roles of IL-1 in the development of HTLV-I-induced arthritis and CIA, using IL-1α–, IL-1β-, and IL-1α/β–deficient mice. We found that the incidence of arthritis in IL-1-deficient mice was greatly suppressed in both models, confirming the importance of IL-1 in the development of arthritis. We found that deficiency of either only IL-1α or only IL-1β was enough to suppress the development of CIA, in contrast to previously reported results (11). HTLV-I–induced arthritis was also suppressed due to deficiency of IL-1, although the effect was less significant than that seen in CIA. Furthermore, we analyzed effects of IL-1 deficiency on the immune system. We found that production of antibodies against CII was not significantly affected in IL-1–deficient mice in the CIA model, but T cell proliferation upon incubation with CII was severely impaired. T cell response against CII was also reduced in HTLV-I Tg mice. We also found that the expression of CD40 ligand (CD40L) and OX40 on T cells was reduced in IL-1–deficient mice, suggesting that this deficiency may cause the low responsiveness of T cells. These observations suggest that IL-1 plays an important role in the development of autoimmunity.

MATERIALS AND METHODS

Mice. IL-1α−/−, IL-1β−/−, and IL-1α/β−/− mice were generated as previously described (7). HTLV-I Tg mice, originally produced by injecting the LTR-env-pX-LTR region of the HTLV-I genome into a (C3H/HeN × C57BL/6J)F1 ovum (18), were backcrossed to BALB/cA mice (Clea, Tokyo, Japan) for 12 generations and then were used for the experiments. These mice start to develop arthritis spontaneously at 4 weeks of age, and 60% and 80% of the mice are affected at 3 months and 6 months of age, respectively.

For the HTLV-I Tg model, IL-1β−/− and IL-1α/β−/− mice were backcrossed to BALB/cA mice for 4 generations, and then crossed with HTLV-I Tg mice on a BALB/cA background. The progenies carrying the transgene were selected and crossed with cytokine-deficient mice again to obtain homozygous-deficient mice. Development of arthritis was then compared between homozygous-deficient mice and wild-type littermates. The pX transgene was detected by dot-blot hybridization of the tail DNA (18).

To examine CIA sensitivity, IL-1α−/−, IL-1β−/−, and IL-1α/β−/− mice were backcrossed to DBA/1J mice (Charles River, Yokohama, Japan) for 4 generations, 5 generations, and 2 generations, respectively. Mice with the H-2q/q haplotype were selected at the second generation by polymerase chain reaction (PCR) using D17MIT 22–specific primers (27). The genotype of each cytokine gene was also examined by PCR using specific primers for each cytokine (7).

All mice were kept under specific pathogen-free conditions in a clean, environmentally controlled room in the Laboratory Animal Research Center, Institute of Medical Science, University of Tokyo. The experiments were conducted according to the institutional ethical guidelines for animal experiments and the safety guidelines for gene manipulation experiments.

Induction of CIA. Mice were immunized intradermally at the base of the tail with 200 μg of bovine CII (Collagen Gyjutsu Kenshukai, Osaka, Japan) in 0.02M Tris HCl-0.15M NaCl with Freund's complete adjuvant (CFA; Difco, Detroit, MI). Twenty-one days after the first immunization, mice were boosted by intradermal injection with 200 μg of CII with Freund's incomplete adjuvant. The mice were inspected for the development of arthritis once a week for 10 weeks.

Clinical assessment of arthritis. Development of arthritis was judged by macroscopic evaluation. Each joint was examined weekly for swelling and redness, and the severity was graded from 0 to 3 for each paw (grade 0 = no changes; grade 1 = mild swelling of the joint and/or redness of the foot pad; grade 2 = obvious swelling of the joint; grade 3 = severe swelling of the joint and ankylotic changes). The severity score correlated well with the histologic findings on microscopy (21). We never observed lymphocyte infiltration or pannus formation in the joints of mice with a severity score of 0, although a slight proliferation of the synovial lining cells could be seen in some of these mice.

Proliferative response of lymph node cells (LNCs). In the CIA model, 8-week-old mice were immunized with 200 μg of CII with CFA, then were killed 6 days after the immunization. Nonimmunized 8-week-old mice were used in the HTLV-I Tg model. The inguinal, axillary, and brachial lymph nodes were collected, and single cell suspensions were prepared in RPMI 1640 (Sigma, St. Louis, MO) supplemented with 10% fetal calf serum, 5 × 10−5M 2-mercaptoethanol, 2 × 105 units/ml penicillin, and 200 μg/ml streptomycin. LNCs (2 × 106/ml) were cultured with either CII (25 μg/ml) or phorbol myristate acetate (PMA; 10 ng/ml) plus ionomycin (200 ng/ml), for 3 days. Then, cells were labeled with 3H-thymidine (1 μCi/well; Amersham, Boston, MA) for 6 hours, and radioactivity was measured using a Micro beta TRILUX system (Wallac, Turku, Finland).

Flow cytometric analysis. LNCs were collected from the inguinal, axillary, and brachial lymph nodes of 8–12-week-old mice. To analyze CD40L (CD154) expression, cells (2 × 106/ml) were cultured with or without CII (25 μg/ml) for 24 hours before flow cytometric analysis. They were then incubated with biotin-conjugated mAb against CD40L for 30 minutes at 4°C, followed by staining with fluorescein isothiocyanate (FITC)-conjugated mAb against CD4 (PharMingen, San Diego, CA) and phycoerythrin (PE)/streptavidin for 30 minutes at 4°C. In the case of OX40, cells were stimulated with CII for 3 days, and then were stained with PE-conjugated anti-OX40 mAb (PharMingen) and FITC-conjugated anti-CD4 mAb. The intensity of fluorescence was quantified with a FacsCaliber flow cytometer (Becton Dickinson, San Diego, CA), as previously described (28).

Antibody titration. Anti-CII antibody, anti-IgG antibody (rheumatoid factor [RF]), total IgG, and total IgM in the serum of HTLV-I Tg mice were measured by enzyme-linked immunosorbent assay, as previously described (22), when the animals were 6 months old. The anti-CII antibody titers were represented as relative values compared with that of CII-immunized DBA/1J mouse serum, and RF titers were relative to that of MRL/Mp-lpr/lpr mouse serum. The unit used is represented as (1/[dilution of the standard reference serum that gives the equivalent optical density value to the test serum]) × 104.

Histopathologic analysis. Limbs were fixed with 10% neutral formalin. After decalcification with 5% formic acid, they were embedded in paraffin, sectioned in 4-μm slices, and stained with hematoxylin and eosin.

Northern blot hybridization. Cytokine messenger RNA (mRNA) was analyzed by Northern blot hybridization, as described previously (22). Mouse IL-1α and IL-1β complementary DNA (CDMmIL-1a and CDMmIL-1b, respectively) were kindly provided by Dr. Tesuo Sudo, and the Xho I fragments (2.0 and 1.3 kb) were used as probes. As a control, β-actin was used. The intensity of the mRNA bands on the autoradiograms was measured with a BAS2000 system (Fuji, Kanagawa, Japan).

RESULTS

Suppression of arthritis development in IL-1−/− mice in both the CIA model and the HTLV-I Tg model. To examine susceptibility to CIA, BALB/c-background IL-1α–, IL-1β-, and IL-1α/β-deficient mice were backcrossed to DBA/1J (H-2q/q) mice, and CIA-sensitive H-2q/q mice were used for the experiments. As shown in Figure 1A, the incidence of arthritis was greatly reduced in IL-1α/β−/− mice: 68% of wild-type mice had developed arthritis by 10 weeks after CII immunization, whereas none of the IL-1α/β−/− mice had developed arthritis at this time (Figure 1B). Some of the IL-1α/β−/− mice did develop arthritis later, but their disease severity scores were reduced compared with those of wild-type mice (data not shown). These observations indicate that IL-1 plays a crucial role in the development of CIA. Interestingly, we found that either IL-1α or IL-1β deficiency was effective in reducing the incidence of disease. The severity score, however, was high in IL-1α−/− mice, whereas it was low in IL-1β−/− mice and IL-1α/β−/− mice. Thus, IL-1α and IL-1β may have differential roles in the development of arthritis.

Figure 1.

Incidence (A and C) and severity (B and D) of arthritis in interleukin-1 (IL-1)–deficient mice, in the collagen-induced arthritis (CIA) mouse model (A and B) and the human T cell leukemia virus type I transgenic (HTLV-I Tg) mouse model (C and D). In the CIA model, mice were immunized with type II collagen on day 0 and day 21 (arrows). The numbers of mice used in this model were as follows: IL-1α−/− 10 females and 11 males, IL-1β−/− 7 females and 10 males, IL-1α/β−/− 8 females and 6 males, wild-type 14 females and 14 males. In the HTLV-I Tg mouse model, development of arthritis was monitored periodically after weaning. The numbers of mice used in this model were as follows: IL-1α/β−/− 22 females and 16 males, IL-1β−/− 25 females and 19 males, wild-type HTLV-I Tg 27 females and 23 males. • = wild-type mice; ▪ = IL-1α/β−/− mice; ▴ = IL-1α−/− mice; □ = IL-1β−/− mice. * = P < 0.05; ** = P < 0.01 versus wild-type mice, by chi-square test. † = P < 0.05 versus wild-type mice, by Mann-Whitney U test.

We also examined effects of IL-1 deficiency in the HTLV-I Tg model. Since arthritis develops efficiently on the BALB/c background (24), IL-1α/β– and IL-1β–deficient mice were backcrossed to BALB/cA mice for 4 generations, and then crossed with BALB/cA–HTLV-I Tg mice to produce homozygous cytokine-deficient animals. Wild-type HTLV-I Tg mice developed arthritis spontaneously after 4 weeks of age. As shown in Figure 1C, the incidence of arthritis was greatly suppressed in IL-1α/β−/− mice, as had been observed in the CIA model. However, in the HTLV-I Tg model, the suppression was only partial, with the incidence in both IL-1α/β−/− and IL-1β−/− mice reduced by approximately half compared with that in wild-type HTLV-I Tg mice. In contrast to findings in the CIA model, arthritis severity in the affected mice in the HTLV-I Tg model was high (Figure 1D). These results indicate that IL-1 plays an important role in the initial phase of the disease, but this pathologic role can be substituted by some other factors in the later stage of the disease.

Histopathologic similarity between the lesion in IL-1−/− mice and that in wild-type mice. Histologic examination of the joints of the CIA mice revealed erosion of the articular bones and cartilage, associated with proliferation of synovial lining cells and infiltration of inflammatory cells, which consisted mostly of neutrophils, into affected tissues (Figure 2B). In contrast, IL-1α/β−/− mice without articular swelling did not exhibit any histologic evidence of inflammation or overproliferation of synovial lining cells (Figure 2C). Nonarthritic IL-1α−/− and IL-1β−/− mice also did not exhibit any histologic changes.

Figure 2.

Histologic findings in the tarsal joint of the hind limb. A–C, Collagen-induced arthritis (CIA) model. D–F, Human T cell leukemia virus type I transgenic (HTLV-I Tg) mouse model. The mice were killed 10 weeks after the first immunization in the CIA model and at 24 weeks of age in the HTLV-I Tg model. A, Wild-type mouse (female, 18 weeks old) without type II collagen (CII) immunization: normal joint. B, Wild-type mouse (female, 18 weeks old): arthritic joint 70 days after immunization with CII. Note the erosion of bone and cartilage with proliferation of synovial lining cells forming pannus-like structures (arrowheads). Infiltration of inflammatory cells, consisting mainly of neutrophils, is prominent. C, Interleukin-1α/β−/− (IL-1α/β−/−) mouse (female, 17 weeks old): suppression of arthritis 70 days after immunization with CII. Only weak proliferation of villous synovial tissues is seen. D, Wild-type mouse (female, 24 weeks old): arthritic joint. Note the very active synovial lining cell proliferation (arrowheads). Multilayered synovial tissues extend over the surface of the tarsal bone, forming pannus associated with infiltration of inflammatory cells. E, IL-1α/β−/− mouse (female, 24 weeks old). The joints appear normal, with the only exception being a mild proliferative change of the synovial lining cells. F, IL-1α/β−/− mouse (female, 24 weeks old): arthritic joint. There is vigorous proliferation of synovial lining cells, some of which have slipped off into the synovial space. Bone has been destroyed by pannus-like tissue, with infiltration of inflammatory cells (arrowhead). (Original magnification × 4 in A, B, C, E, and F; × 20 in D.)

HTLV-I Tg mice developed arthritis spontaneously, and the histopathologic findings were different from those in CIA. Articular erosion caused by invasion of overproliferating granulation tissues was marked in HTLV-I Tg mice (Figure 2D), whereas direct erosion of cartilage and bone, without invasion of synovial tissues, was predominant in the CIA model (Figure 2B). In the HTLV-I Tg model, the histologic findings in the IL-1α/β−/− mice that did not develop arthritis were normal (Figure 2E). However, once they developed arthritis, those mice showed histopathologic features very similar to those found in wild-type HTLV-I Tg mice (Figure 2F). Taken together, these histologic findings provide further evidence that IL-1 deficiency was effective in preventing the onset of arthritis, although the severity became similar to that in wild-type HTLV-I Tg mice once the disease developed.

Lack of effect of IL-1 deficiency on the expression of other cytokines, except IL-6. The effect of IL-1 deficiency on the expression of other cytokines was examined, and Figure 3 shows representative results from 1 of 3 experiments. The expression of the proinflammatory cytokines IL-1β, IL-6, and TNFα, but not IL-1α, was enhanced 2.2-, 6.4-, and 1.3-fold, respectively, in the arthritic joints in the CIA model, compared with nonarthritic joints (Figures 3A and 4A). Similar augmentation was observed in the HTLV-I Tg model, in which IL-1α, IL-1β, IL-6, and TNFα expression was enhanced 2.1-, 8.5-, 16.4-, and 2.8-fold, respectively (Figures 3B and 4B). Interestingly, we found that the expression of these cytokines was similarly augmented in cytokine-deficient mice in both models, although the level of IL-6 expression in IL-1α/β−/− mice in the HTLV-I Tg model was slightly lower (30–100%) than that in wild-type mice. Basal IL-6 expression levels were also reduced in IL-1α/β– and IL-β–deficient nonarthritic mice in the CIA model. We could not measure cytokine expression levels in arthritic IL-1α/β−/− mice in the CIA model because only a few of these mice developed arthritis. Our observations indicate that deficiency of IL-1 does not affect the expression of other cytokines in these models, with the exception of an effect of IL-1β deficiency on IL-6 expression.

Figure 3.

Proinflammatory cytokine expression in IL-1–deficient mice. A, CIA model. Poly(A)+ RNA was prepared from the joints of arthritic (Arth.) or nonarthritic mice 42 days after immunization with CII, and expression of IL-1α, IL-1β, IL-6, and tumor necrosis factor α (TNFα) mRNA was analyzed by Northern blot hybridization. B, HTLV-I Tg mouse model. Poly(A)+ mRNA was prepared from the joints of 8–12-week-old arthritic or nonarthritic mice, and cytokine gene expression was analyzed by Northern blot hybridization. See Figure 2 for other definitions.

Figure 4.

A, Relative levels of IL-1α, IL-1β, IL-6, and tumor necrosis factor α (TNFα) mRNA in arthritic (Arth.) or nonarthritic IL-1–deficient mice in A, the CIA model and B, the HTLV-I Tg model. The IL-1α, IL-1β, IL-6, and TNFα mRNA bands in the autoradiograms shown in Figure 3 were measured using a BAS2000 system, and intensities relative to those obtained with nonimmunized wild-type (WT) or non-Tg mice are shown. Three mice were used for each experiment; values are the mean and SD. See Figure 2 for other definitions.

Normal production of antibodies against CII in IL-1α/β−/− mice in the CIA model, but low autoantibody levels in the HTLV-I Tg model. It is well known that humoral immune response against CII plays a crucial role in the development of CIA (13). Accordingly, we examined the development of specific antibodies against CII in IL-1-deficient mice that had been immunized with CII. As shown in Figure 5A, all of the IL-1α/β−/−, IL-1α−/−, and IL-1β−/− mice developed high levels of antibody against CII similar to those in wild-type mice, although they did not develop arthritis efficiently. The IgG subclasses were also not changed significantly in these mice (Figure 5B). These results indicate that anti-CII antibody production is normal in IL-1–deficient mice, suggesting that high levels of this antibody are not sufficient to cause the development of arthritis in this model.

Figure 5.

Antibody production in IL-1–deficient mice. A and B, CIA model. Mice were immunized with CII on days 0 and 21. Each serum sample was diluted 1:5,000, and antibody levels were measured by enzyme-linked immunosorbent assay (ELISA) on day 70. The number of mice in each group (all males) was as follows: IL-1α/β−/− 17, IL-1β−/− 16, IL-1α−/− 14, wild-type (WT) 13. A, Serum levels of IgG-type anti-CII antibody. B, Subclasses of the antibody. C and D, HTLV-I Tg model. Serum samples were diluted 25-fold, and after ELISA, titers were calculated in comparison with a high-titer standard serum. Each group consisted of 10 female mice. C, Levels of IgG-type rheumatoid factor. D, Levels of anti-CII antibody. Solid circles represent mice that developed arthritis; open circles represent mice that did not. Bars indicate the group means. * = P < 0.05; ** = P < 0.01 versus nonimmunized or non-Tg mice, by F-test. See Figure 2 for other definitions.

In contrast, mean levels of autoantibody against IgG in HTLV-I Tg mice were significantly reduced in IL-1α/β−/− and IL-1β−/− animals compared with wild-type mice (Figure 5C). It is noteworthy that the mice exhibiting high levels of autoantibody against IgG developed arthritis, while autoantibody levels in nonarthritic mice were relatively low. A similar tendency was observed with regard to endogenous anti-CII antibody levels (Figure 5D). These results indicate that autoantibody production is suppressed in IL-1–deficient HTLV-I Tg mice, and the autoantibody levels correlate well with the development of arthritis.

Significantly reduced proliferative response of LNCs to CII in IL-1α/β−/− mice. To further characterize the defects in IL-1α/β−/− mice, we examined the proliferative response of LNCs against CII after primary immunization with CII. As shown in Figure 6A, the proliferative response of the LNCs from IL-1α/β−/− mice was approximately one-tenth that of the LNCs from wild-type mice. This was not a result of developmental defects of the LNCs, because the response to PMA plus ionomycin was normal. As shown in Figure 6B, the proliferative response to CII observed in LNCs from wild-type HTLV-I Tg mice was augmented compared with that of LNCs from non-Tg mice, indicating development of cellular autoimmunity in HTLV-I Tg mice (22). Interestingly, this proliferative response was reduced in IL-1α/β−/− HTLV-1-Tg mice as was observed in the CIA model, although the standard deviation among mice was large. These findings suggest that T cell priming efficiency is reduced in IL-1α/β−/− mice.

Figure 6.

Proliferative responses of lymph node cells (LNCs) from IL-1α/β−/− mice to CII. A, CIA model. IL-1α/β−/− and wild-type (WT) mice were immunized with CII, and LNCs were prepared 6 days after immunization. The proliferative response of the LNCs to CII was quantified by measuring the incorporation of 3H-thymidine into the acid-insoluble fraction. B, HTLV-I model. Mice were killed at 5 months of age, and the proliferative response of LNCs to CII was determined by 3H-thymidine incorporation. The proliferation index was calculated as 3H-thymidine incorporation with stimulator/3H-thymidine incorporation without stimulator. Individual values are shown; bars show the mean and SD. * = P< 0.05; ** = P < 0.01 versus wild-type or non-Tg mice, by F-test. PMA/Iono = phorbol myristate acetate plus ionomycin (see Figure 2 for other definitions).

Reduced expression of CD40L and OX40 on T cells in IL-1α/β−/− mice. It is known that interaction between CD40L and CD40 plays a crucial role in T cell priming, and either CD40 deficiency or CD40L deficiency abrogates efficient IgG antibody production against T-dependent antigens (29–31). It is also known that OX40 ligation with OX40L activates naive T cells to produce Th2 cytokines and differentiate into Th2 cells (32, 33) and promotes B cell production of antibodies to T-dependent antigens (34). Since T cell priming was impaired and autoantibody production was reduced in IL-1α/β−/− HTLV-1 Tg mice, we next examined the levels of CD40L and OX40 antigen expression on T cells. We found that the expression of CD40L and OX40 on CD4+ T cells from IL-1α/β−/− mice after immunization with CII was significantly lower than that on T cells from wild-type mouse T cells (Figures 7A and B). The expression of these antigens on T cells from IL-1α/β−/− HTLV-I Tg mice after in vitro stimulation with CII was also significantly reduced compared with that on T cells from wild-type HTLV-I Tg mice (Figures 7C and D). These observations clearly show that CD40L and OX40 expression on CD4+ T cells is reduced in IL-1α/β−/− mice.

Figure 7.

Expression of CD40 ligand (CD40L) and OX40 on CD4+ T cells from IL-1–deficient mice. A and B, CIA model. A, Expression of CD40L. Lymph node cells (LNCs) were collected 6 days after CII immunization and cultured with or without CII for 24 hours in vitro. They were then incubated with biotin-conjugated monoclonal antibody against CD40L followed by fluorescein isothiocyanate (FITC)–conjugated anti-CD4 antibody and phycoerythrin (PE)/streptavidin, and CD4+ cells were gated. B, Expression of OX40. LNCs were collected 6 days after CII immunization and cultured with or without CII for 3 days. They were then stained with PE-conjugated anti-OX40 antibody and FITC-conjugated anti-CD4 antibody, and CD4+ cells were gated. C and D, HTLV-I Tg model. C, Expression of CD40L. LNCs were collected and cultured with or without CII for 24 hours. D, Expression of OX40. LNCs were collected and cultured with or without CII for 3 days. Shaded profiles represent unstimulated controls. WT = wild-type; arth. = arthritis (see Figure 2 for other definitions).

DISCUSSION

We assessed the role of IL-1 in the development of autoimmune arthritis in 2 models, the CIA and HTLV-I Tg mouse arthritis models, using gene-targeted mice. We found that IL-1 deficiency strongly suppressed development of arthritis in both models, indicating that IL-1 has a critical function in the development of arthritis. We also showed that T cells from IL-1–deficient mice were not fully activated upon stimulation with CII in vitro in both models. Furthermore, levels of expression of CD40L and OX40, which play crucial roles in T cell–antigen-presenting cell as well as in T cell–B cell interactions (34, 35), on CD4+ T cells were low in these mutant mice. These observations suggest that T cell activation is impaired in IL-1–deficient mice due to the failure of CD40L and OX40 induction on T cells, resulting in suppression of arthritis development in these mice. Consistent with this, it was recently reported that anti-OX40 antibody suppressed development of CIA, indicating the importance of OX40–OX40L interaction in the development of arthritis (36). Our observation that the proliferative response of T cells against CII was only marginally affected by anti–IL-1R antibody treatment at the time of secondary stimulation (data not shown) suggests that IL-1 activates T cells at the priming stage.

In this context, we recently showed that IL-1 enhanced T cell–dependent antibody production through induction of CD40L and OX40 expression on CD4+ T cells (37). Our present finding of impaired CD40L and OX40 expression on T cells from IL-1–deficient CII-immunized mice as well as from IL-1–deficient-HTLV-I Tg mice is consistent with this observation. However, in the CIA model, antibody production against CII was not significantly impaired. These observations suggest that even low-level T cell priming (∼10% of the maximum level) is enough for antibody production. In support of this notion, recent studies showed that antigen-specific T cell proliferation was severely impaired in OX40-deficient mice, although the ability for antibody production remained completely normal in these mutant animals (38, 39).

The data also suggest that administration of adjuvant, the immunization method used in the CIA model in this study, may abrogate the necessity for IL-1 for antibody production, because various cytokines such as TNFα or IL-6, which have biologic activities similar to those of IL-1, are induced by the treatment. Actually, specific serum antibody levels were normal in IL-1RI−/− mice when these mice were immunized with trinitrophenyl/keyhole limpet hemocyanin/alum (TNP-KLH/alum) or TNP-KLH/CFA either intraperitoneally or subcutaneously (40, 41), in contrast to the low levels of antibody against sheep red blood cells in IL-1α/β−/− mice immunized without adjuvant (37). Nonetheless, development of arthritis was severely suppressed in the IL-1–deficient mice. These observations suggest that the function of IL-1 in activating cellular immunity, rather than enhancing antibody production, is crucial for the development of arthritis.

The effects of IL-1 deficiency on HTLV-I–induced arthritis were less significant than those observed in the CIA model; the incidence was reduced only by half compared with that in wild-type mice, and it increased with time. Levels of autoantibody against CII and IgG in HTLV-I Tg mice were low on average, but they became as high as those in the wild-type mice after onset of the disease. To explain this, we showed that various inflammatory cytokines were augmented in the joint irrespective of the presence of IL-1, especially after disease onset. This augmentation of cytokine expression is explained by the transcriptional transactivating activity of HTLV-I Tax (19, 42) (Figure 3). Since the activity of TNFα and IL-6 partially overlaps with that of IL-1 (3), these cytokines may compensate for the function of IL-1 in activating T cells. It is also known that both IL-1 and IL-6 promote osteoclast formation through independent pathways (3, 43).

Interestingly, disruption of either the IL-1α or the IL-1β gene caused significant suppression of CIA. This was also the case in the HTLV-I Tg model: although we did not examine the effect of IL-1α deficiency, IL-1β deficiency alone was enough to suppress the onset of the disease to the level observed in IL-1α/β−/− mice. These results were unexpected because IL-1α and IL-1β should complement each other by acting on the same receptor (44). The level of expression of IL-1α in IL-1β−/− mice after immunization with CII was similar to that in wild-type mice, and that of IL-1β in IL-1α−/− mice was also similar. We conclude that IL-1α and IL-1β act synergistically, and deficiency in one of these molecules causes great reduction of the overall activity.

It is interesting that the severity scores of the IL-1α−/− mice with CIA were similar to those of wild-type mice, although the incidence of disease onset was low in those mice. These results contrast with findings in IL-1β−/− mice, in which both the incidence and the severity score were reduced. This indicates that IL-1α affects the pathogenesis of the disease at the trigger point, whereas IL-1β seems to be involved in both the triggering and the progression phases of the disease. In this regard, we found that expression of IL-6 was reduced in IL-1β−/− mice while that in IL-1α−/− mice was not changed. It is possible that the difference in the IL-6 level influenced the disease severity. In contrast to our observations, Joosten et al reported that anti–IL-1α treatment did not cause significant suppression of CIA (11). It is possible, however, that the amount of antibody they used was not enough to suppress arthritis or that intracellular IL-1α played a role (6).

It is well known that IL-1 induces inflammation by activating synovial cells, endothelial cells, lymphocytes, and macrophages to produce various chemokines, cytokines, and inflammation mediators (5). These inflammation mediators could cause infiltration of inflammatory cells into sites of inflammation, increase the permeability of blood vessels, and induce fever (45). Furthermore, IL-1 promotes synovial cell growth and activates synovial cells and osteoclasts to produce metalloproteases and collagenases that cause erosion of the bone and of joint cartilage (3). It is possible that these activities of IL-1 are also involved in the development of arthritis at the effector phase. The current study showed, however, that these activities of IL-1 are not necessarily a prerequisite for the development of inflammation in the HTLV-I model since, among affected mice, the severity score was not different in those with IL-1 deficiency. IL-1α was also not needed in order for inflammation to develop in the CIA model, suggesting that other cytokines may account for the activity.

In conclusion, we have shown that IL-1 has a key function in the development of autoimmunity and arthritis by activating T cells in 2 etiologically different RA models. It is highly probable that this cytokine also plays an important role in the development of RA in humans. Thus, suppression of the production of IL-1 or inhibition of its activity should be beneficial for the treatment of RA.

Acknowledgements

We are deeply grateful to all of the staff at our laboratory and at the Center for Experimental Medicine, IMSUT, for valuable discussions, technical help, and excellent animal care throughout the course of this work.

Ancillary