Histologic studies of the muscles suggest that cytokines are involved in inflammatory myopathy. The therapeutic effects of cytokine blockade are controversial, with anecdotal reports of clinical efficacy. The aim of this study was to discern the significance of interleukin-1 (IL-1) and tumor necrosis factor α (TNFα) as therapeutic targets in polymyositis (PM) by studying their involvement and the effects of their blockade in C protein–induced myositis (CIM), a murine model of PM.
C57BL/6 mice were immunized with recombinant skeletal C protein fragments to induce CIM. The expression of IL-1 and TNFα in the muscles of mice with CIM was detected using immunohistochemical and real-time polymerase chain reaction analyses. After the onset of myositis, the mice with CIM were treated with recombinant IL-1 receptor antagonist (IL-1Ra), anti–IL-1R monoclonal antibody, recombinant TNF receptor (p75)–fusion protein (TNFR-Fc), or anti-TNFα monoclonal antibody. The muscles were examined histologically for the severity of myositis.
IL-1α– and TNFα-positive macrophages were observed in the muscle tissue of mice with CIM as early as 7 days after immunization. IL-1α, IL-1β, and TNFα expression in the muscles increased as the severity of myositis peaked, at both the messenger RNA and protein levels. Continuous subcutaneous delivery of IL-1Ra resulted in suppression of established CIM. Intermittent delivery (1-day intervals) of anti–IL-1R monoclonal antibody suppressed myositis, while intermittent delivery of IL-1Ra did not suppress myositis. Treatment with anti-TNFα monoclonal antibody and with TNFR-Fc also reduced the severity of CIM.
IL-1 and TNF blockade ameliorated CIM after disease onset and should potentially be a new strategy for the treatment of inflammatory myopathy. As IL-1 blockade, treatment with anti–IL-1R monoclonal antibody appeared more feasible than the other approaches.
Polymyositis (PM) is a chronic autoimmune inflammatory myopathy affecting striated muscles (1). An accumulated body of evidence supports the notion that the pathology of PM is driven by cytotoxic CD8 T cells (2–7), but the event that initiates the inflammatory processes has not been identified. Currently, patients with PM are treated primarily with nonspecific immunosuppressants, including high-dose corticosteroids, methotrexate, and/or other small-molecule immunosuppressants. Because the administration of therapeutic agents can elicit a wide variety of adverse reactions, treatments that address the specific pathology of PM need to be developed.
In the development of new therapeutic approaches to human diseases, animal models have served as a means with which to identify therapeutic targets and to test the effect of new treatments (8–11). Despite the known limitations, experiments in animals with collagen-induced arthritis (CIA) have facilitated development of new treatments for rheumatoid arthritis (RA). Treatment approaches such as blockade of interleukin-1 (IL-1), tumor necrosis factor α (TNFα), and IL-6 have had an enormous effect in modulating the disease course of RA (12–15).
However, in myositis research, no appropriate animal model of PM had been available until the murine model of C protein–induced myositis (CIM) was developed (16). Unlike the classic model of experimental autoimmune myositis (EAM), which is induced by repeated immunizations to desferlin gene–mutated mice (SJL/J strain) with crude myosin, CIM can be induced in C57BL/6 (B6) mice by a single immunization with recombinant human fast-type skeletal muscle C protein fragments. Although serum can be used to transfer EAM to naive mice (17), CIM is primarily mediated by cytotoxic CD8 T cells (18). Among the available models, CIM mimics human PM best in terms of pathology (16, 18).
Because of the availability of biologic anticytokine reagents for clinical use, these reagents have been anecdotally tested for the treatment of patients with PM and patients with dermatomyositis (DM) who did not respond to conventional treatment (19–21). In this regard, results of animal experiments using anticytokine reagents will represent a rationale for conducting controlled clinical studies in humans. We recently observed that anti–IL-6 receptor (anti–IL-6R) antibodies were effective for the treatment of CIM (22).
In mice with CIM and in the muscles of patients with PM (23, 24), both IL-1 and TNFα are expressed by infiltrating mononuclear cells. Previously, we observed that the development of CIM was suppressed in IL-1α/β–null mutant mice but not in TNFα-null mutant mice (16). This observation suggested the differential requirement of inflammatory cytokines for CIM induction. However, it is unclear whether blockade of IL-1 or TNFα after disease onset can suppress CIM. Also, genetically mutated mice may undergo skewed development of the immune system and respond aberrantly to immunogens. In the present study, we examined the therapeutic effects of IL-1 blockade and TNFα blockade in mice with established CIM. In addition, antibodies and soluble decoy molecules were compared for the in vivo effects of IL-1 blockade.
MATERIALS AND METHODS
Induction of CIM.
Female B6 mice (Charles River), ages 8–10 weeks, were immunized by intradermal injection of recombinant human fast-type skeletal muscle C protein fragments emulsified in Freund's complete adjuvant (CFA) together with intraperitoneal injection of pertussis toxin (16). All experiments were carried out under specific pathogen–free conditions in accordance with the ethics and safety guidelines for animal experiments of Tokyo Medical and Dental University and RIKEN.
Murine IL-1R antagonist (IL-1Ra), hamster anti-mouse IL-1R chimeric IgG1 monoclonal antibody (clone M147), and murine recombinant TNF receptor (p75)–fusion protein (TNFR-Fc) were provided by Amgen. Anti-TNFα chimeric (rat and murine) IgG2a monoclonal antibody (clone cV1q) and control chimeric IgG2a monoclonal antibody with unknown antigen specificity (clone cVaM) were provided by Centocor. IL-1Ra was administered continuously with subcutaneously implanted osmotic minipumps (Durect), while the other treatments were administered intraperitoneally 3 times weekly.
Histologic severity of inflammation.
The system for scoring the histologic severity of inflammation in muscles affected by experimental myositis was originally established using a Lewis rat model of myositis (25, 26). This system was applied successfully to the evaluation of several treatments of CIM (16, 22).
For each mouse, 2 sections of bilateral muscles (hamstrings or quadriceps) were evaluated. Myositis was defined as mononuclear cell infiltration spreading around muscle fibers, including at least 1 necrotic muscle fiber. The histologic score for myositis severity was determined according to the number of muscle fibers associated with cellular infiltration (grade 1 = lesions with <5 muscle fibers involved, grade 2 = lesions involving 5–30 muscle fibers, grade 3 = lesions involving an entire muscle fasciculus, and grade 4 = extensive lesions across muscle fasciculi). When multiple lesions with the same grade were observed in the 2 sections of a muscle block, 0.5 point was added to the grade. The mean score of bilateral muscles was calculated and used as the score for each mouse. The muscle sections were evaluated in a blinded manner by at least 2 independent observers, who reported comparable results.
Quantification of cytokine messenger RNA (mRNA) expression.
Total RNA was isolated from hindleg muscle tissue with Isogen reagent (Nippon Gene). Quantitative polymerase chain reaction (PCR) was performed on a PerkinElmer 7700 sequence detector using sets of primers and FAM-labeled TaqMan probes specific for IL-1α, IL-1β, TNFα, and GAPDH complementary DNA (Assays-on-Demand; Applied Biosystems).
Enzyme-linked immunsorbent assay (ELISA) of muscle homogenate.
The muscles were homogenized in phosphate buffered saline containing protease inhibitors (cOmplete Mini tablets; Roche Diagnostics) and radioimmunoprecipitation assay buffer (Millipore), 3 times for 20 seconds with homogenizer (Mini-BeadBeater; BioSpec) at 2500 revolutions per minute. The supernatants were collected, and the concentrations of IL-1β and active transforming growth factor β1 (TGFβ1) were measured with Quantikine ELISA kits (R&D Systems) according to the manufacturer's directions.
The expression of CD11b, CD68, IL-1α, and TNFα in muscle sections was examined immunohistochemically, as previously described (16). To quantify the stained cells, 10 low-power (200×) fields in each section were selected to include myositis lesions. The numbers of CD11b-, IL-1α–, and TNFα-positive cells in these fields were then determined.
Histology scores and quantitative analysis of immunohistochemical studies were analyzed using the Mann-Whitney U test.
IL-1 and TNFα expression in mice with CIM.
In the murine model of CIM, which was induced by immunizing B6 mice with recombinant C protein fragments, necrotic muscle fibers with surrounding mononuclear cell infiltration appeared as early as 7 days and peaked 14–21 days after immunization (16). Real-time quantitative PCR analysis of mRNA in the muscles revealed that IL-1α, IL-1β, and TNFα mRNA were up-regulated 7 days after immunization, with increased expression observed at the peak of inflammation (Figures 1A–C).
Similar results were observed at the protein level. ELISA showed that IL-1β expression was significantly increased in the muscles of mice with CIM (Figure 1D). Immunohistochemical studies showed that IL-1– and TNFα-positive cells appeared in the muscles 7 days after the C protein immunization and become more abundant 21 days after immunization (Figures 2A–D and G). In contrast, active TGFβ1 was barely detectable 7 days after immunization and stayed at a constant level even at the peak of inflammation. (Figure 1D).
Resident macrophages were present around nonnecrotic muscle fibers, since CD11b-positive cells were present in the muscle tissue of unimmunized mice (Figures 2E and G). Immunohistochemical studies showed that the resident macrophages expressed no detectable inflammatory cytokines (Figure 2G). Seven days after the C protein/CFA immunization, the number of macrophages in the muscles increased significantly (Figures 2F). Thus, macrophages were recruited and activated to produce inflammatory cytokines in the muscles beginning in the early stage of myositis.
Therapeutic effects of IL-1 blockade on established CIM.
IL-1–null mutant mice have been shown to be resistant to CIM induction. In the present study, IL-1 was blocked to discern whether this approach can alleviate the severity of CIM after disease onset. Blockade was carried out using IL-1Ra or anti–IL-1R monoclonal antibodies. Treatment was started 7 days after immunization. IL-1Ra was administered continually (0.24, 0.8, and 2.4 mg/day/mouse) for 7 days with minipumps implanted under the back dermis. The same amount of bovine serum albumin acted as a control. Histologic scoring of the muscles from the treated mice showed that IL-1Ra successfully suppressed CIM, in a dose-dependent manner (Figure 3A).
Although continuous administration of IL-1Ra was effective, this mode of infusion would not be practical for the treatment of patients. Thus, we next treated the mice with CIM intermittently (7, 9, and 11 days after immunization) and examined the muscles 3 days after treatment completion. Although a total of 7.2 mg of IL-1Ra was administered, intermittent administration of IL-1Ra failed to ameliorate the disease (Figure 3B). Continuous administration of IL-1Ra (0.8 mg/day), starting on the same day as the intermittent injections, was effective. The advantage of continuous treatment over intermittent treatment was maintained in a 2-week treatment protocol (Figure 3B).
Generally, monoclonal antibodies bind more stably to cell surface receptors than do soluble forms of the corresponding ligands. Indeed, when recombinant TNF receptor (p75)–fusion protein (TNFR-Fc) (100 μg/day and 50 μg/day) was injected intraperitoneally according to the same 1-week intermittent administration protocol, treatment with both doses ameliorated CIM (Figure 3B).
Therapeutic effects of TNFα blockade on ongoing CIM.
Mice with CIM were also treated with TNFα-blocking reagents, including anti-TNFα monoclonal antibody and TNFR-Fc. Different doses of the anti-TNFα monoclonal antibody (50 μg, 200 μg, and 500 μg) and TNFR-Fc (100 μg) were administered 3 times weekly for 2 weeks, starting 7 days after immunization. The muscles of the treated mice were assessed 3 days after treatment completion. The histologic scores of the mice treated with either type TNFα (500 μg) or TNFR-Fc were significantly lower than those of the control mice (Figures 4A and B).
The expression of IL-1 and TNFα in the muscles of mice with CIM was observed beginning in the early phase of disease (day 7) and increased as the severity of inflammation peaked. Blockade of either cytokine after disease onset suppressed CIM. These results suggest that both IL-1 and TNFα are potential therapeutic targets in the treatment of PM.
Both IL-1 blockade and TNFα blockade reduced the severity of CIM. Previous histologic studies of the muscles of patients with PM showed that IL-1 expression by mononuclear cells accompanied up-regulation of class I major histocompatibility complex (MHC) molecules on the muscle fibers (27), and that IL-1R expression on muscle fibers was most pronounced in the vicinity of IL-1–expressing cells (28). TNFα-positive mononuclear cells have also been observed in the muscles of patients with PM (24). Like IL-1, TNFα increased class I MHC expression on human myoblasts in vitro (29, 30). Also, TNF can damage muscle fiber directly (31). These findings suggest that IL-1 and TNFα expression of activated macrophages in muscles may contribute to both up-regulation of class I MHC molecules on muscle fibers and direct muscle damage.
IL-1 is involved in antigen-specific T cell differentiation. T cell proliferative responses to type II collagen were impaired in IL-1α/β–double-null mutant mice immunized with type II collagen for induction of CIA, which is also a model of induced autoimmune disease. In vitro experiments suggested that dendritic cells cannot activate T cells fully if they are not activated by IL-1 (32–34). Recently, it was shown that IL-1, together with TGFβ and IL-6, is involved in the differentiation of Th17 cells (35). In addition, IL-1 promoted Th17 cell differentiation in mice with experimental autoimmune encephalomyelitis (EAE) (36) and also in IL-1Ra–deficient mice with destructive arthritis (37). However, IL-17A was dispensable in the development of CIM (22). Actually, CD3 cells from the inguinal lymph nodes of mice with CIM that had or had not received IL-1Ra treatment proliferated equally in response to C protein–pulsed dendritic cells (data not shown). Thus, we did not see attenuation of pathogenic T cell responses in the IL-1Ra–treated mice.
The therapeutic effects of cytokine blockade could not be tested until CIM had developed. The classic EAM model not only is mediated by CD4 humoral immune responses (17) but also requires continual administration of CFA throughout the disease course. This makes it difficult to discern whether any treatment blocked the adjuvant effects of CFA or the pathologic processes of myositis per se. In contrast, CIM can be induced with a single immunization and can be treated after disease onset. However, it is still difficult to initiate treatment at the very peak of the disease, because regression occurs spontaneously.
We performed ELISA of active TGFβ as an antiinflammatory cytokine. Previous studies showed that TGFβ was expressed in the muscles of patients with PM (23) and suggested that TGFβ from macrophages may contribute to muscle regeneration after muscle injury (38). This is consistent with the marginal elevation of TGFβ expression observed in the muscles of mice with CIM.
Because a major clinical manifestation of myositis is muscle weakness, rotarod testing was used to measure muscle function as the clinical outcome in our previous study (16). We actually used the same rotarod test in the anti-TNFα antibody treatment experiment and observed that the mean running time of the treated mice was longer than that of control mice. However, the difference did not reach statistical significance. According to our experience, the rotarod test is not as sensitive as histologic analyses, because mice become accustomed to the device and thus are able to avoid falling off, and the efforts of mice are inconsistent. We have used different techniques to measure muscle weakness in mice with CIM, including measurement of walking time, walking distance, and rearing time in open-field tests but have not yet identified an appropriate technique for use in mice with CIM.
In a clinical trial of the treatment of multiple sclerosis, TNF inhibitors occasionally exacerbated the disease (39). In accordance with this finding, TNF-null mice developed severe EAE (40). TNF inhibitors sometimes induce a lupus-like syndrome in the clinical setting (41). In an animal model, TNF-null (NZB × NZW)F1 mice developed lupus nephritis (42), and injection of high doses of TNF delayed disease onset (43). In contrast, TNF inhibitors were proven to be useful in the treatment of both RA and CIA (8, 12). Nevertheless, TNFα-null mice are fully susceptible to CIA (44). The current study showed that TNFα inhibitors are effective in treating a murine model of PM, although TNFα-null mice are fully susceptible to CIM induction. Thus, this is the second instance in which the inducibility of autoimmune diseases is different between TNFα-null mice and mice treated with TNFα inhibitors. It has been proposed that increased numbers of memory CD4 T cells and augmented interferon-γ production from CD4 T cells are responsible for exacerbated disease activity of CIA in TNFα-null mice (44). Although CD4 T cells and IL-17–producing lymphocytes play a critical role in CIA (45), both are dispensable for muscle injury in CIM. The genetic absence of TNFα appears to have extensive effects on the effector function of lymphocytes. The efficacy of TNF inhibitors in the treatment of PM and DM is controversial (19–21). According to anecdotal case reports, myositis occurred in patients with RA even after initiation of treatment with TNF inhibitors (46, 47). It would be of particular interest to know the clinical effects of TNF inhibitors in patients with myositis.
In the US, IL-1Ra (anakinra) has been approved for the treatment of RA. Because its terminal half-life ranges from 4 hours to 6 hours, IL-1Ra should be injected subcutaneously every day. The discrepant efficacy of IL-1Ra and anti–IL-1R monoclonal antibody in the present studies may be explained by differences in the half-life as well as the affinity to IL-1R.
DM is another inflammatory myopathy that is driven by autoimmunity. It has been proposed that PM is mediated by cytotoxic CD8 T cells, while DM is mediated by humoral responses. However, the accumulated body of evidence suggests that PM and DM are similar in terms of muscle pathology as well as responses to various treatments (48). Sontheimer proposed that both PM and DM are within a single disease spectrum (49). With the CIM model, we believe that by using the CIM model in this study, we provide support for clinical trials of IL-1 and TNF blockade to treat PM and DM.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kohsaka had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Sugihara, Okiyama, Miyasaka, Kohsaka.
Acquisition of data. Sugihara, Okiyama, Watanabe, Miyasaka, Kohsaka.
Analysis and interpretation of data. Sugihara, Okiyama, Watanabe, Miyasaka, Kohsaka.
We thank Drs. Wayne Tsuji and Takashi Nakae for their critical advice and Eri Yoshimoto for her technical assistance.