It has been reported that polymyositis (PM) is driven by CD8+ cytotoxic T lymphocytes. The C protein–induced myositis (CIM) model we have established is similar to PM in pathology except that it undergoes spontaneous remission. We undertook the present study to delineate the roles of innate and acquired immunity in myositis.
C57BL/6 mice were immunized with recombinant C protein fragments together with Freund's complete adjuvant (CFA) and Toll-like receptor (TLR) ligands at hind leg footpads and tail bases. CIM mediated by adoptive transfer of T cells to naive mice was treated with cytokine antagonists.
Second immunization with C protein fragments revealed no induction of tolerance. Injection of CFA and TLR ligands at the hind leg footpads reinduced myositis in the same legs. Interestingly, initial myositis was observed only in the CFA-treated forelegs. Transfer of C protein fragment–specific T cells from mice with CIM induced myositis in CFA- and TLR ligand–treated legs of recipient mice. CFA treatment resulted in the recruitment of macrophages producing inflammatory cytokines. Induction of myositis was inhibited by blocking interleukin-1 receptor or tumor necrosis factor α.
Myositis development requires activation of autoaggressive T cells and conditioning of muscle tissue. CIM regression is due to attenuation of local CFA-induced immune activation. These results are in accordance with a “seed and soil” model of disease development and might offer clues to decipher clinical aspects of PM.
Polymyositis (PM) is a chronic inflammatory myopathy of unknown etiology. It affects striated muscles and induces varying degrees of muscle weakness, especially in the proximal muscles (1). Dysphagia and respiratory muscle weakness with choking episodes and/or recurrent aspiration pneumonia can lead to premature death of patients. Current standard treatment is administration of high-dose glucocorticoids and/or immunosuppressants, which do not address the specific pathology of PM. Some patients have unwanted side effects, while other patients' disease is refractory to these drugs.
Immunohistochemical analysis of biopsy specimens of muscle tissue in PM suggested that muscle injury is driven primarily by CD8+ cytotoxic T lymphocytes (CTLs) (2). Thus, the disease process derives from systemic autoimmune reactions to muscle autoantigen(s), which could incite systemic muscle inflammation. However, the entire muscle system is not necessarily subject to inflammation in PM. This is evident especially when the muscles of the whole body are scanned with magnetic resonance imaging (MRI). The affected muscles are often separated clearly from unaffected muscles (3). Moreover, peripheral blood of PM patients in clinical remission still contained expanding CD8+ cytotoxic T cell clones (4, 5). These clones appeared autoreactive since they were found in the inflamed muscles prior to the remission induction (5).
In order to gain mechanistic insights into the pathology of PM, we have established a murine PM model, skeletal muscle C protein–induced myositis (CIM) (6). CIM can be induced by single immunization with recombinant mouse or human skeletal muscle C protein fragments in Freund's complete adjuvant (CFA). Our previous studies demonstrated that CIM is primarily mediated by CD8+ CTLs (6, 7). This is in sharp contrast with the classic myosin-induced experimental autoimmune myositis model, which is driven by CD4+ T cells and humoral immunity (8, 9). While experimental autoimmune myositis is induced in SJL/J mice carrying a mutated dysferlin gene that causes spontaneous muscle necrosis and secondary inflammation (10), CIM can be induced in C57BL/6 (B6) and other strains of mice.
As with most other models of autoimmune disease that are inducible by immunization with autoantigens, CIM resolves spontaneously. The myositis peaks 2 to 3 weeks after the immunization with C protein fragments and regresses in 10 weeks (6, 11). Since muscle fibers regenerate rapidly, no histologic abnormality is observed after the resolution of disease.
Spontaneous regression of inducible autoimmune disease models has been intensively investigated because the results might grant insight into the development of new treatment modalities. For example, disease regression in experimental autoimmune encephalomyelitis (EAE; a model of multiple sclerosis) in rats induced by immunization with myelin basic protein is associated with the appearance of immunosuppressive cytokines including transforming growth factor β and interleukin-4 (IL-4) (12). Recently, accumulating evidence suggests that naturally arising CD25+CD4+ Treg cells actively act to maintain peripheral self tolerance (13). CD25+ cell–depleted mice had significantly more severe diseases in murine models of multiple sclerosis, rheumatoid arthritis (RA), myasthenia gravis, and Hashimoto thyroiditis (14–18).
In the present studies, extensive investigation was directed at addressing the mechanistic processes involved in spontaneous regression of CIM. The results show that attenuated activity of autoaggressive T cells was not exclusively responsible for the spontaneous regression of CIM. Instead, attenuation of innate immunity in the muscles also contributed to disease regression. These facts led us to propose a “seed and soil” theory of autoimmune tissue damage, in which autoaggressive T cells and activation of innate immunity act as seed and soil, respectively. Combination therapies that address pathomechanisms of both aspects involved in autoimmune disease damage may be an effective approach to explore in the treatment of disease.
MATERIALS AND METHODS
B6 mice were purchased from Charles River Japan. 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.
CIM induction and recurrence.
To induce CIM, female B6 mice ages 8–10 weeks were immunized intradermally with 200 μl of an emulsion consisting of 200 μg of murine C protein fragments emulsified in CFA containing 100 μg of heat-killed Mycobacterium butyricum (Difco) (11). The immunogens were injected at the hind leg footpads and tail bases. At the same time, 0.2–2 μg of pertussis toxin (Seikagaku Kogyo) in phosphate buffered saline (PBS) was injected intraperitoneally. Some mice were subjected to additional intradermal injections at the front paws of 100 μl of an emulsion consisting of PBS/CFA or PBS/Freund's incomplete adjuvant (IFA). In a modified protocol, mice were immunized intradermally with 200 μl of the C protein fragment/CFA emulsion only at the left hind leg footpads and tail bases together with intraperitoneal injection of 0.2–2 μg of pertussis toxin. These mice were treated after 42 days with an intradermal injection of 100 μl of a C protein fragment/CFA emulsion, a CFA emulsion containing PBS vehicle, a C protein fragment/CFA emulsion containing 1 mg of polymyxin B (Sigma-Aldrich), an IFA emulsion containing PBS vehicle, an IFA emulsion containing 100 μg of poly(I-C) sodium salt (Sigma-Aldrich), or an IFA emulsion containing 100 μg of lipopolysaccharide (LPS; Sigma-Aldrich) at the contralateral hind leg foot pads and tail bases.
Hematoxylin and eosin–stained 10-μm sections of the hamstrings and quadriceps and the brachial triceps were examined histologically for the presence of mononuclear cell infiltration and degeneration of muscle fibers. Histologic severity of myositis in each muscle block was graded as follows (6): grade 1 = involvement of a single muscle fiber or <5 muscle fibers; grade 2 = a lesion involving 5–30 muscle fibers; grade 3 = a lesion involving a muscle fasciculus; grade 4 = diffuse, extensive lesions. When multiple lesions with the same grade were found in a single block, 0.5 points was added to the grade. The stained sections were evaluated by 2 independent observers (NO and TS), who reported comparable results.
Cell proliferation assay and interferon-γ (IFNγ) detection in culture supernatants.
Bone marrow cells from B6 mice were treated with granulocyte–macrophage colony-stimulating factor to prepare bone marrow–derived dendritic cells (BMDCs) (19). More than 70% of the treated cells were positive for CD11c staining. Lymph node (LN) cells were prepared from the inguinal and popliteal LNs from the mice with CIM 21 days after the immunization. One hundred thousand LN cells and 1 × 104 C protein fragment–pulsed or untreated mature BMDCs were cultured for 3 days. Proliferation was evaluated with incorporation of 3H-thymidine during the last 8 hours of the incubation. The culture supernatants were examined for the concentration of IFNγ with an enzyme-linked immunosorbent assay (ELISA) kit (Mouse IFN-gamma DuoSet; R&D Systems).
Adoptive transfer of CIM.
LN cells were prepared from the inguinal and popliteal LNs of the mice with CIM 21 days after the immunization. Three million LN cells and 1.5 × 106 C protein fragment–pulsed mature BMDCs were cultured with 100 IU/ml of recombinant human IL-2 (Shionogi Pharmaceuticals) for 3 days. Eight million LN cells were adoptively transferred to naive mice that were treated simultaneously with subcutaneous hind leg footpad injection of 50 μl of CFA emulsion, IFA emulsion, or IFA emulsion containing 100 μg of poly(I-C), LPS, or CpG-containing oligonucleotide (CpG ODN) 1826 (InvivoGen). Their muscles were examined histologically after 14 days.
Cryostat frozen sections (6 μm) fixed in cold acetone were stained with anti-CD68 monoclonal antibodies (mAb) (FA-11; AbD Serotec). Nonspecific staining was blocked with 4% Blockace (DS Pharma Biomedical). Bound antibodies were visualized with peroxidase- labeled anti-rat IgG antibodies and associated substrates (Histofine Simple Stain Max PO; Nichirei Biosciences). Isotype controls were used as a negative control.
Antiinflammatory cytokine treatment.
Hamster/mouse chimeric anti-murine IL-1 receptor (IL-1R) IgG1 mAb (M147) (20), rat/mouse chimeric anti–tumor necrosis factor α (anti-TNFα) IgG2a mAb (cV1q) (21), and rat anti–IL-6R IgG1 mAb (MR16-1) (22) were provided by Amgen, Centocor, and Chugai Pharmaceutical, respectively. Bovine serum albumin (Sigma-Aldrich) or rat antidinitrophenol IgG1 mAb (KH-5; Chugai Pharmaceutical) was used as a control.
Histologic scores were analyzed statistically using the Mann-Whitney U test.
No C protein–specific tolerance induction after spontaneous regression of CIM.
As is the case in most inducible models of autoimmune disease, our model of CIM regresses spontaneously (6, 11). To study whether immunologic tolerance to C protein fragments had been established after the regression, we rechallenged mice with CIM with C protein fragments emulsified in CFA. First, B6 mice were immunized with a C protein fragment/CFA emulsion at only the left hind leg footpads and tail bases with intraperitoneal injections of pertussis toxin. This treatment resulted in myositis development in the muscles of the ipsilateral hind legs 3 weeks after initial immunization. Then, after disease regression, these mice were reimmunized with the same antigen/CFA emulsion in the footpads and tail bases. Contralateral legs were used for repeat immunization due to skin damage in previously immunized ipsilateral legs. IFA alone was injected as a vehicle control. Histologic evaluation of the muscles of the contralateral hind legs 14 days after the repeat C protein fragment/CFA immunization revealed that repeat immunization with C protein fragment/CFA had reinduced the myositis, while control IFA treatment had not (Figure 1A). These results showed that tolerance to the C protein fragment was not established following disease regression.
Notably, CFA injection without C protein fragments, but not IFA, at hind leg footpads and tail bases induced myositis after disease regression. Myositis reinduced with CFA alone was less severe than that reinduced with C protein fragment/CFA reimmunization, suggesting that attenuated activity of the autoaggressive T cells may also partly account for disease regression. As a control, CFA alone was used in the first and second treatments, and we found no inflammation that damaged muscles (Figure 1A).
To study the mechanism of CFA-induced recurrence more carefully, we tried to inhibit the recurrence by adding polymyxin B, an inhibitor of LPS (a Toll-like receptor 4 [TLR-4] ligand), in CFA emulsion. Polymyxin B partially inhibited myositis recurrence (Figure 1B). The partial inhibition was likely attributable to activators of innate immunity other than LPS contained within CFA (23). The effect of TLR ligands was examined directly with injection of LPS- and poly(I-C) (a TLR-3 ligand)–containing emulsions without C protein fragments as the second treatment. Injection of these TLR ligands induced recurrence of myositis (Figure 1B). Thus, redevelopment of myositis requires activation of local innate immunity, but not necessarily T cell reactivation.
Requirement of local CFA treatment for the development of myositis.
Previously, we immunized mice at their hind leg footpads and tail bases and examined their femoral muscles for histologic changes. The results of the reimmunization experiments prompted us to examine the brachial muscles of the immunized animals. We found that they did not develop myositis at the brachial muscles (Table 1). We then injected CFA at the right front paws and IFA at the left front paws in mice at the same time that they were immunized in the conventional way to induce CIM. These mice developed myositis only in the CFA-treated forelegs and hind legs. The brachial muscles of the IFA-treated forelegs had no myositis (Table 1). No myositis was observed in the forelegs of naive mice treated in the same way. These findings reinforced the fact that effector T cell development and local activation of innate immunity are both critical for myositis development.
Table 1. Requirement of local CFA treatment for myositis induction*
Foreleg status, muscle, footpad treatment
Histologic score, mean ± SD
Four mice that were treated to induce C protein–induced myositis (CIM) at the hind leg footpads and tail bases (foreleg untreated) did not develop myositis of the brachial muscles. In 6 mice, at the same time as treatment to induce CIM, Freund's complete adjuvant (CFA) and Freund's incomplete adjuvant (IFA) emulsions were injected additionally into the right and left footpads, respectively, of forelegs (foreleg treated), and myositis developed in the brachial muscles of the CFA-treated legs. Myositis of the femoral and the right/left brachial muscles was histologically assessed 21 days after the immunization. CP = C protein fragments.
1.88 ± 0.75
1.00 ± 0.42
0.83 ± 1.03
Requirement of local activation of innate immunity for adoptive transfer of CIM.
To explore the mechanism of effector T cell activation (a component of acquired immunity) in muscle cells engaged in local innate immunity, we used an adoptive transfer CIM model (7). Prior to the transfer to naive mice, the LN cells from mice with CIM were cocultured for 5 days with C protein fragment–pulsed BMDCs together with recombinant IL-2. Upon coculture with C protein fragment–pulsed BMDCs, LN cells showed enhanced proliferation and IFNγ induction compared with LN cells cocultured with untreated BMDCs (Figure 2A). Thus, transferred LN cells contained T cells reactive specifically to C protein fragments. Because we had confirmed that intradermal injection of CFA at leg footpads did not cause myositis (Figure 2B), we injected CFA intradermally at the right hind leg footpads and IFA intradermally at the left hind leg footpads simultaneously upon transfer of LN cells. Two weeks later, we examined the muscles of the hind legs histologically and found myositis only in the hind legs that were injected with CFA, but not in the hind legs that were injected with IFA (Figure 2B).
As in the aforementioned studies on CFA-induced recurrence, TLR ligands, including poly(I-C), LPS, and CpG ODN, successfully primed muscle tissues as well as CFA and facilitated adoptive transfer of myositis (Figure 2C). Thus, it was confirmed that not only effector T cells, but also conditioning of the muscle tissues via activation of innate immunity, is essential for the development of myositis.
Local activation of innate immunity as a therapeutic target in CIM.
Activation of local innate immunity with CFA or other TLR ligands was crucial to facilitate autoaggressive T cell attack of muscle fibers. To elucidate the distal effects of CFA treatment, muscles were histologically examined 14 days after CFA treatment. In comparison with normal muscles, muscle tissue 14 days after CFA treatment contained more mononuclear cells but showed no signs of damage. Pathologic scores remained grade 0 since no muscle damage was observed. Mononuclear cells were positive for CD68 (Figure 3A), but not for CD4, CD8, or B220 (data not shown). Thus, macrophages were recruited into the muscles of the CFA-treated limbs without tissue damage.
Triggering of macrophage TLRs induces production of type I IFNs and inflammatory cytokines including IL-1, TNFα, and IL-6 (24). Since a separate experiment showed that IFNα/β/ω receptor 1–null B6 mice, lacking type I IFN receptors, were as susceptible to CIM induction as wild-type mice (data not shown), it is unlikely that macrophage type I IFN contributed to autoaggressive T cell infiltration into muscle tissue. We thus assumed that alternative cytokines, including IL-1, TNFα, and IL-6, were crucial to recruit autoaggressive T cells.
To study the role of IL-1, TNFα, and IL-6, anti–IL-1R, anti-TNFα, and anti–IL-6R mAb were employed to block individual cytokines in the CIM adoptive transfer model. Intraperitoneal administration of these reagents was initiated when activated LN cells from mice with CIM were transferred to naive mice that were treated simultaneously with CFA at the hind leg footpads. Anti–IL-1R, anti-TNFα, and anti–IL-6R mAb were administered until 14 days following transfer of LN cells from mice with CIM, when muscles of the recipient mice were histologically evaluated. Blockade of both IL-1R and TNFα suppressed the transfer, while anti–IL-6R mAb, a dose of which suppressed development of conventional CIM (11), failed to suppress induction of myositis (Figure 3B). Thus, IL-1 and TNFα from macrophages may be responsible for the conditioning of muscle tissues that mediates attack by autoaggressive T cells.
By investigating spontaneous regression of mononuclear cell infiltration and muscle injury in CIM, we found that activation of autoaggressive effector T cells and that of innate immunity at the target tissues are required for autoimmune tissue injury. TNFα and IL-1 are essential in the activation of innate immunity. Presumably, local macrophages are the source of TNFα and IL-1 in muscles that are activated by CFA. CFA may have dual roles since it should provoke systemic T cell responses specific to C protein fragments by activating DCs. Once T cell responses against muscles are established, local injection of CFA and TLR ligands at the footpads can cause myositis to recur in legs otherwise free from myositis. The effect of CFA in activating local innate immunity appeared more potent than individual TLR ligands and could not be inhibited fully by a TLR-4 inhibitor. This is most likely because CFA contains activators of innate immunity other than TLR-4 ligands (23, 25–27). Thus, CIM depends on systemic T cell autoimmunity as well as activation of innate immunity, especially that of TLR signaling in the muscles.
In CIM, TNFα and IL-1 production from macrophages stimulated with CFA or TLR ligands is a key event to activate innate immunity. Activation of innate immunity per se is insufficient for development of myositis, because single and repeat CFA injections increased local macrophages in number while not inducing muscle damage. The other essential event is activation of C protein fragment–reactive T cells. In contrast to LN cells from mice with CIM cocultured with C protein fragment–pulsed BMDCs, LN cells cocultured with untreated BMDCs did not proliferate or produce IFNγ. Moreover, they failed to transfer myositis to a naive recipient mouse with footpad CFA injection (data not shown). Histologic scores were set at zero to indicate no muscle damage. Involvement of activated T cells is crucial for muscle damage; therefore, histologic scores greater than zero suggested that CIM and its recurrence were mediated by C protein fragment–reactive T cells inducing muscle damage.
These observations echo the “seed and soil” model, which had been proposed in metastatic processes of tumor cells. Metastasis depends on tumor cells liberated from tumor mass as seeds and also depends on target tissues as soil that accepts the tumor cells readily for their local growth. Analogously, autoreactive CD8+ T cells act as seeds while muscle tissues act as soil. Requirement of the 2 factors was demonstrated for the first time in animal models of autoimmunity.
Molecular events of the “soil” activation in muscle tissues in CIM may include up-regulation of adhesion molecules on endothelial cells, chemokine release from macrophages, and costimulatory molecule expression by myofibers, which can be promoted by TNFα and/or IL-1 (28). To address this point, we used immunohistochemistry to study the expression of intercellular adhesion molecule 1 and class I major histocompatibility complex in the muscles from the CFA- and IFA-treated legs, and we found no difference in their expression levels (data not shown). Although more intensive studies will be required to draw definitive conclusions, other molecules should be of importance for recruitment of autoaggressive T cells.
It was reported previously that collagen-induced arthritis, a murine model of RA, could be reactivated by oral administration of LPS (29). Since the reactivation accompanied increased levels of antibodies against the immunizing antigen (type II collagen), the authors concluded that B cell activation by LPS might be responsible for the reactivation. EAE was also reactivated by intravenous injection of LPS (30). It appeared to be mediated by proliferation and cytokine production in a fraction of effector/memory CD4+ T cells. While these studies implied that systemic delivery of LPS stimulated pathogenic lymphocytes in the systemic pool, our studies revealed the contribution of local innate immunity activated by LPS.
It is still an open question whether the same “seed and soil” model applies to human PM. Apparently, CFA is not involved in PM. Since the recurrent myositis and the T cell adoptive transfer in the present studies showed that TLR ligands were sufficient to prime the muscles, stimulation by endogenous TLR ligands might account for production of IL-1 and TNFα, both of which were reportedly expressed by infiltrating cells in muscles in PM/dermatomyositis (DM) and CIM (6, 11, 31). Alternatively, aberrant production of inflammatory cytokines and chemokines might be responsible for attraction of autoaggressive T cells. It is known that muscle fibers can produce these cytokines under physiologic conditions, especially during exercise (32–35). Etiologic studies revealed heavy muscular exertion as a risk factor for development of PM/DM (36). As was reviewed earlier, MRI scans typically show that some muscle fascicles are affected and that others remain unaffected (3). Cytotoxic CD8+ T cells, which appeared specific to the muscles, were present for a long time after successful treatment of the disease (4). These facts could be explained by the 2 requirements: activation of effector T cells and that of the muscle tissues.
Experiments of repeat antigen/CFA immunization and CFA treatment have demonstrated that tolerance was not induced during spontaneous regression of disease and that activation of local innate immunity alone can cause recurrence of myositis. We also noted that reimmunization of repeat antigen/CFA provoked more severe myositis than CFA treatment alone. Thus, the T cell activation level following regression must have waned, which could also contribute to the disease regression. In this regard, contribution of Treg cells was studied with antibody-mediated CD25+ T cell depletion, which did not alter the disease course of CIM (data not shown). Genetic absence of IL-10, which is one of the key effector molecules of Treg cells, did not interfere with the regression (data not shown). It was thus suggested that no active suppression of autoreactive T cells operates in the regression phase.
Waning T cell activation might partially account for the fact that myositis was modest in the transfer model, which was assessed histologically 2 weeks after the T cell transfer. Incidence of myositis transfer varied among experiments (25–80%), which might depend on the frequency and activation levels of C protein fragment– specific T cells. We had to compare the severity and incidence of transferred myositis in the same set of experiments. Nonetheless, the adoptive transfer model has enhanced the value of CIM since it allows us to evaluate the effector phase of myositis. Together with conventional CIM, we can dissect the pathologic processes involved in the induction and effector phases of myositis. In this regard, IL-6 inhibition was effective for treating conventional CIM (11), while IL-6 inhibition in the recipient animals did not prevent adoptive transfer of CIM. It is likely that IL-6 plays a more prominent role in the induction phase of myositis. Of special interest are interventions targeting the myositis effector phase, since we always have to treat patients after disease has been established.
At the moment, histologic evaluation is the only reliable way to assess severity of rodent models of PM, a problem shared by all murine myositis models. Serum levels of creatinine kinase or other muscle-derived proteins are unreliable. They are often high in normal mice, presumably because of their physical activity. The rotarod test is difficult because some mice become accustomed to it and avoid falling off. Other mice are willing to drop off from the device. We believe that we need devices that directly quantify muscle strength of rodents to follow clinical disease course.
Glucocorticoids, which are a first-line medication in the current therapeutic approaches to PM, should suppress activation of T cells as well as innate immune cells. They are effective, but have quite a few side effects. On the other hand, most small-molecule immunosuppressants target primarily lymphocyte activation. The results of the present study suggest that combinatorial approaches that address activation of T cells and innate immunity could be optimal for treating autoimmune myositis and suggest that glucocorticoids could become dispensable if activation of innate immunity in the muscles is suppressed by alternative treatment. This implication appears interesting for designing future clinical trials to treat PM.
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. Okiyama, Yokozeki, Miyasaka, Kohsaka.
Acquisition of data. Okiyama, Sugihara, Oida, Ohata.
Analysis and interpretation of data. Okiyama, Sugihara, Oida, Ohata, Kohsaka.
Author Ohata is an employee of Chugai Pharmaceutical.
We thank Eri Yoshimoto for her technical assistance, and Centocor R&D (Radnor, PA), Amgen (Seattle, WA), and Chugai Pharmaceutical (Tokyo, Japan) for providing monoclonal antibodies.