The etiology and pathogenesis of human inflammatory myopathies remain unclear. Findings of several studies suggest that the degree of inflammation does not correlate consistently with the severity of clinical disease or of structural changes in the muscle fibers, indicating that nonimmune pathways may contribute to the pathogenesis of myositis. This study was undertaken to investigate these pathways in myositis patients and in a class I major histocompatibility complex (MHC)–transgenic mouse model of myositis.
We examined muscle tissue from human myositis patients and from class I MHC–transgenic mice for nonimmune pathways, using biochemical, immunohistochemical, and gene expression profiling assays.
Up-regulation of class I MHC in skeletal muscle fibers was an early and consistent feature of human inflammatory myopathies. Class I MHC staining in muscle fibers of myositis patients showed both cell surface and a reticular pattern of internal reactivity. The pathways of endoplasmic reticulum (ER) stress response, the unfolded protein response (glucose-regulated protein 78 pathway), and the ER overload response (NF-κB pathway) were significantly activated in muscle tissue of human myositis patients and in the mouse model. Ectopic expression of wild-type mouse class I MHC (H-2Kb) but not degradable glycosylation mutants of H-2Kb induced ER stress response in C2C12 skeletal muscle cells.
These results indicate that the ER stress response may be a major nonimmune mechanism responsible for skeletal muscle damage and dysfunction in autoimmune myositis. Strategies to interfere with this pathway may have therapeutic value in patients with this disease.
The human inflammatory myopathies are muscle diseases in which myofiber damage appears to be the consequence of an immune process (1–4). The precise pathogenesis of the immune process leading to muscle damage, however, is unknown. The degree of inflammation does not consistently correlate with severity of the structural changes in the muscle fibers or the severity of clinical disease (5), suggesting that nonimmune processes also play a role in the pathogenesis. Several observations provide further evidence of this. First, marked structural changes in the muscle fibers occur in the absence of any inflammatory cells (6, 7). Second, there is a lack of correlation between the degree of inflammation and the degree of muscle weakness (8). Third, in some myositis patients the disease does not respond even to powerful antiinflammatory therapy (9, 10). Fourth, steroid treatment may eliminate inflammatory cells in myositis muscle tissue, but this removal alone may not substantially improve clinical disease (11), suggesting that immunosuppressive therapies modulate disease activity but do not change other mediators of the disease process. Finally, clinical disease may progress even when identifiable inflammation has subsided (12).
To investigate these and other pathogenic processes, we have generated a conditional transgenic mouse model in which class I major histocompatibility complex (MHC) is overexpressed in skeletal muscle. These mice develop clinical, biochemical, histologic, and immunologic features similar to those of human myositis, providing a close model of the human disease. The disease is inflammatory, limited to skeletal muscles, self-sustaining, more severe in females, and is often accompanied by autoantibodies (13).
The following observations in human myositis patients and in the mouse model of myositis suggest that class I MHC molecules themselves may potentially mediate muscle fiber damage and dysfunction in the absence of lymphocytes: 1) in human myositis, induction of class I MHC antigen in muscle fibers occurs early, preceding inflammatory cell infiltration (14, 15); 2) class I MHC staining of biopsy specimens from patients with myositis reveals both a cell surface and a sarcoplasmic reticulum pattern of internal reactivity, demonstrating that some of the class I MHC may be retained in the endoplasmic reticulum (ER) of these fibers (7, 16, 17); 3) overexpression of class I MHC in muscle fibers may persist in the absence of an inflammatory infiltrate (12); 4) the controlled induction of class I MHC in the murine model was followed by muscle weakness occurring before mononuclear cell infiltration (13); and 5) it has recently been shown that in vivo gene transfer of class I MHC plasmids attenuates muscle regeneration and differentiation (18). Together, these observations—particularly the obvious retention of class I MHC within the cell in both the human and murine diseases—indicate that the muscle fiber damage seen in myositis may be mediated not solely by immune attack (e.g., cytotoxic T lymphocytes and autoantibodies), but may also be mediated through nonimmunologic mechanisms such as the ER stress response. Since class I MHC up-regulation in myositis muscle fibers is widespread even in the absence of visible inflammatory infiltrates, we propose that ER stress may play a potential role in the muscle fiber damage and dysfunction in human myositis.
The ER has several critical roles in the folding, export, and processing of newly synthesized proteins. When there is an imbalance between the load of proteins in the ER and the cell's ability to process that load, a collection of signaling pathways that adapt cells to ER stress will commence. The ER stress response can be provoked by a variety of pathophysiologic conditions, including ischemia, hyperhomocysteinemia, viral infections, mutations that impair protein folding, and excess accumulation of proteins in the ER (19, 20). Cells self-protect against ER stress by initiating several responses. At least 4 functionally distinct responses have been identified: 1) responses involving up-regulation of the NF-κB pathway (ER overload response), 2) up-regulation of genes encoding ER chaperone proteins such as BiP/Grp78 and Grp94, to increase protein folding activity and to prevent protein aggregation, 3) translational attenuation to reduce the load of protein synthesis and to prevent further accumulation of unfolded proteins (unfolded protein response), and 4) cell death, which occurs when functions of the ER are severely impaired. This cell death event is mediated by transcriptional activation of the gene for cholesterol oxidase-peroxidase C/EBP homologous protein (CHOP)/growth arrest and DNA damage 153 (GADD153), a member of the CCAAT/enhancer binding protein family of transcription factors (21), and by activation of ER-associated caspase 12 (22).
Results of the present study revealed evidence for activation of the ER stress response pathway in human myositis. Furthermore, it was demonstrated that the ER stress response pathway is also activated in the conditional class I MHC–transgenic mouse model of myositis, suggesting a potential role for this pathway in both human inflammatory myopathies and the murine model of myositis.
MATERIALS AND METHODS
Muscle biopsy specimens from 10 patients with myositis (5 with polymyositis and 5 with dermatomyositis) and 4 healthy controls were used for biochemical and immunohistochemical studies. The myositis patients were treated with prednisone either alone or in combination with methotrexate. Part of the tissue was fixed in formalin, processed in paraffin, and stained with hematoxylin and eosin. Histologic analysis showed a variable degree of inflammation. The remaining tissue was snap-frozen in isopentane chilled in liquid nitrogen, cut in 8-mm sections, and processed for immunohistochemistry analysis. Muscle biopsy samples obtained from a separate group of 5 untreated dermatomyositis patients were used for gene expression profiling studies. These profiles were compared with those from normal controls who had been profiled previously (23). Patients with myositis met the Bohan and Peter criteria for probable or definite disease (24).
Conditional class I MHC–transgenic (HT double-transgenic) mice were generated as described previously (13). Single-transgenic (H or T) littermates were used as controls. Animals were cared for in accordance with institutional guidelines. Gene induction was performed as described previously (13).
Messenger RNA (mRNA) profiling.
Total RNA was extracted from each muscle biopsy sample or tissue piece using TRIzol reagent (Life Technologies, Gaithersburg, MD), and 10 μg of each total RNA sample was processed for microarray analysis as previously described (23). A single complementary RNA sample from each of the muscles was applied to U74A or U133A microarrays (Affymetrix, Santa Clara, CA). Arrays were stained with phycoerythrin/streptavidin, and the signal intensity was amplified by treatment with a biotin-conjugated antistreptavidin antibody followed by a second staining with phycoerythrin/streptavidin. Second-stained arrays were scanned on a G2500A Gene Array Scanner (Hewlett-Packard, McMinnville, OR) with the photomultiplier tube set at 1,800V.
Quality control and normalization of mRNA profiles.
The profiles were normalized for interchip intensity variation by scaling the overall intensity of each profile to 800, and absolute analysis for each profile was generated with Affymetrix Microarray Suite 5.0 software. Each profile was subjected to a stringent series of quality controls: scaling factor <4, present calls >30%, internal probe set controls 5′/3′ ratios >0.6. All profiles passed quality control measures (http://microarray.cnmcresearch.org/pgaoutline-qcofsamples.asp). It should be noted that there are a number of alternate methods for probe set analyses and normalizations for Affymetrix arrays (such as Probe Profiler, dCHIP, RMA), and that different results can be obtained with different analysis methods. We have recently shown that an analysis using MAS 5.0, with a 10% present call filter, provides excellent signal-to-noise ratios for delineation of diagnostic groups in human muscle biopsy specimens (25), and this was the method used for the current analysis.
U133A profiles in samples from the 5 patients with untreated myositis were compared with profiles in 18 normal muscle specimens. MU74A version 2 profiles in 3 class I MHC–transgenic mice were compared with profiles in 3 age-matched control-strain mice. The MAS 5.0 and 10% present call filter were as described previously (23, 26, 27). U133A arrays have ∼22,000 probe sets encoding mostly confirmed genes, while the MU74A v2 array is composed of ∼12,500 probe sets encoding mostly confirmed genes with some exposed sequence tags included. Changes in expression changes were initially assessed using the Welch 2-sample t-test without correction for multiple testing, using GeneSpring software (Silicon Genetics, Redwood City, CA). Each gene was normalized to the median of all muscle profiles, and expression in the myositis (human or mouse) group was compared with that in controls with the Welch 2-sample t-test. Genes with significantly altered expression (P < 0.05 versus normal profiles) were identified. Genes were selected according to their presumed function, based on information available in public databases or in the literature.
Muscle cell lysates were prepared from muscle obtained from controls and myositis patients as described previously (28). The lysates were electrophoresed on sodium dodecyl sulfate–polyacrylamide gels, transferred to nitrocellulose membranes, and probed with antibodies against class I MHC (clone HC-10, which recognizes invariant regions of HLA–A, B, and C proteins) (a kind gift from Dr. Hidde Ploegh, Harvard Medical School, Boston, MA), anti-Grp78 (Stressgen, Victoria, British Columbia, Canada), and vinculin (Sigma, St. Louis, MO) as a loading control. The autoradiograms were scanned using an Arcus II scanner (Agfa, Mortsel, Belgium), and volume analysis was carried out using Quantity One software (Bio-Rad discovery series; Bio-Rad, Richmond, CA). The ratios of class I MHC (HLA–ABC), Grp78, and vinculin in myositis patients and normal controls were calculated. Statistical significance was calculated using Student's t-test. Mouse muscle tissue lysates were blotted with rabbit anti-mouse caspase 12 antibodies (Cell Signaling Technology, Beverly, MA) and developed with horseradish peroxidase (HRP)–conjugated anti-rabbit IgG.
Immunohistochemistry analysis was performed as described previously (17), using mouse anti–HLA–ABC antibody (W6/32) and HRP-conjugated anti-mouse IgG (Dako, Carpinteria, CA) as primary and secondary antibodies, respectively. As negative controls, isotype-matched mouse immunoglobulins were used instead of primary antibodies.
Frozen muscle biopsy specimens from myositis patients and controls were stained with rabbit anti–NF-κB p65 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), developed with Texas Red–conjugated anti-rabbit IgG, and counterstained with 4′6-diamidino-2-phenylindole (Molecular Probes, Eugene, OR) to visualize the nuclei (blue staining). The specificity of the staining was confirmed by blocking antibody reactivity with an NF-κB–specific peptide. For colocalization experiments, HLA–ABC, Texas Red–conjugated anti-mouse IgG (Santa Cruz Biotechnology), rabbit anti-human calnexin antibodies, and fluorescein isothiocyanate–conjugated anti-rabbit IgG were used. Digital pictures (AxioVision version 3.1 software) of stained sections were taken using an Axioscope microscope (Zeiss, Wetzlar, Germany).
NF-κB levels were assayed by enzyme-linked immunosorbent assay (ELISA) according to the instructions of the manufacturer (Trans-AM NF-κB p65; Active Motif, Carlsbad, CA), but with several modifications. Briefly, nuclei from muscle tissue were isolated and nuclear extracts from transgenic and nontransgenic mice were lysed in lysis buffer. The lysates were incubated with wild-type (WT) or mutated consensus oligonucleotides along with positive and negative controls in 96-well plates. The assays were performed in triplicate. The wells were washed, incubated with anti–NF-κB antibody, and developed with HRP-conjugated secondary antibody, and absorbance at 460 nm was read.
Site-directed mutagenesis and transfections with H-2Kb constructs.
Full-length complementary DNA coding for H-2Kb was cloned into a pcDNA 3.1 Myc/His vector. Putative glycosylation sites at Asn107 and Asn197 (N) were mutated to Gln (Q) using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Mutations were verified by DNA sequencing (H-2Kb N197Q forward CCTGAAGAACGGGCAAGCGACGCTGCTGCG, reverse CGCAGCAGCGTCGCTTGCCCGTTCTTCAGG, H-2Kb N107Q forward GCTCGGCTACTACCAACAGAGCAAGGGCGGC, reverse CGAGCCGATGATGGTTGTCTCGTTCCCGCCG) (nucleotides shown in boldface indicate changed amino acids). Transient transfections with WT H-2Kb construct, H-2Kb Q107, and H-2Kb Q197 in C2C12 mouse muscle cells were performed using lipofectamine. H-2Kb expression was verified by Western blotting using anti-His antibodies. The ER stress response in H-2Kb (WT and mutant)–transfected C2C12 cells was assessed by immunoblotting using anti-Grp78 and anti-His antibodies. Vinculin levels in these samples were also analyzed by immunoblotting, as loading controls.
Quantitative findings were expressed as the mean ± SEM or mean ± SD. The statistical significance of the differences between the patient and control groups was calculated using Student's t-test.
Striking up-regulation of class I MHC in skeletal muscle fibers from patients with myositis.
Several investigators, including our group, have previously demonstrated striking up-regulation of class I MHC in biopsy samples from human myositis patients, although the extent and the level of expression varied among muscle fibers (6, 17, 29). Many muscle fibers showed increased expression at the cell surface, along with increased expression on infiltrating mononuclear and endothelial cells (Figure 1A). Of note, the up-regulation was particularly striking in apparently normal-appearing non-necrotic muscle fibers in the absence of any visible inflammatory infiltrate, and these strongly reactive fibers also showed a reticular pattern of internal reactivity in addition to the usual cell surface staining (Figure 1B). In polymyositis patients, class I MHC–reactive fibers were present all over the fascicle, whereas in patients with dermatomyositis, they were highly restricted to perifascicular areas of the biopsy sample. In normal muscle, the class I reactivity was restricted predominantly to capillary endothelial cells (Figure 1C).
The biochemical levels of class I MHC proteins in lysates prepared from muscle biopsy specimens were quantified by immunoblotting with a monoclonal antibody (clone HC-10) recognizing the invariant region of all 3 human class I MHC proteins (HLA–A, B, and C). Class I MHC expression varied among myositis patients, but was strikingly elevated (∼10-fold) compared with expression in normal controls (mean ± SEM class I MHC:vinculin ratio 12.3 ± 3.6 in patients [n = 10] versus 1.2 ± 0.4 in controls [n = 4]; P < 0.01) (Figures 1D and E). Further, the reticular pattern of class I MHC internal reactivity in muscle fibers of myositis patients was confirmed by colocalizing HLA–ABC with the known ER marker calnexin (Figures 2D–F). Muscle fibers from control subjects exhibited reactivity with calnexin, but not with class I MHC (Figures 2A–C).
Increased expression of Grp78 in muscle from myositis patients and in the mouse model.
The unfolded protein response and ER overload response in the ER lumen lead to up-regulated expression of the unfolded protein response target genes encoding ER resident chaperones such as Grp78 (30). Consistent with this, we found that the levels of Grp78 protein assayed by immunoblotting were strikingly elevated (5.2-fold) in myositis patient biopsy samples compared with control samples (P < 0.0001) (Figures 3A and C). Similarly, Grp78 expression was significantly increased (7.2-fold) in transgenic mice that overexpressed class I MHC in muscle (P < 0.0001) (Figures 3B and D), indicating activation of the ER stress response pathway in this mouse model.
Induction of Grp78 expression in C2C12 muscle cells by wild-type H-2Kb, but not H-2Kb glycosylation mutants.
Prevention of N-glycosylation results in misfolding and proteasomal degradation (31). Therefore, we assessed the ability of H-2Kb and glycosylation mutants of H-2Kb to induce ER stress response in muscle cells. We transiently transfected with WT and glycosylation mutants of H-2Kb (H-2Kb N107Q and H-2Kb N197Q) and assessed their ability to express Grp78. WT H-2Kb, but none of the H-2Kb mutants, induced Grp78 expression (Figure 4A), suggesting that WT, but not degradable, forms of H-2Kb induce ER stress response in C2C12 muscle cells.
Confirmation that several members of the ER stress response pathway are also highly up-regulated in biopsy specimens from human myositis patients.
We further examined myositis biopsy specimens for other members of the ER stress response pathway, by global gene expression profiling. Myositis human U133A (Affymetrix) profiles were compared with normal muscle profiles. The analysis was performed using GeneSpring software (Silicon Genetics). The expression levels of several genes, including those for GADD153, double-stranded RNA–activated protein kinase–like endoplasmic reticulum kinase (PERK), and activating transcription factor 3 (ATF-3), were significantly activated in human myositis specimens (Figure 4B). These data show that in addition to increased protein expression of Grp78, other ER stress response pathway members are up-regulated at the level of gene expression in muscle obtained from myositis patients.
Up-regulation of the NF-κB pathway in human myositis patients and in transgenic mice.
The ER overload response is known to activate NF-κB (32). Therefore, we examined human myositis muscle biopsy specimens for NF-κB activation (translocation of NF-κB p65 from cytosol to nucleus), by immunofluorescence. NF-κB p65 stained the nuclear/subsarcolemmal regions of skeletal muscle fibers from myositis patients (Figures 5A and B) but not from controls (results not shown), and this staining was abolished in the presence of specific NF-κB peptide (results not shown). The levels of activated NF-κB p65 in skeletal muscle cell lysates of HT double-transgenic and control mice were quantitated using a sensitive ELISA. HT double-transgenic mice showed significantly elevated levels of activated NF-κB p65 in skeletal muscle compared with age-matched single-transgenic control mice (mean ± SD absorbance at 460 nm 0.14 ± 0.02 in HT mice versus 0.001 ± 0.0 in H or T mice; P < 0.001) (Figure 5C), indicating a possible role of the NF-κB pathway in the pathogenesis of myositis in this mouse model.
It is known that NF-κB transcription factor promotes the expression of many genes that participate in a variety of biologic processes, including induction of a number of cytokines, chemokines, MHC proteins, and adhesion molecules (33). We and others have previously demonstrated, by immunohistochemistry, up-regulation of several classic NF-κB target genes, such as intercellular adhesion molecule 1 (ICAM-1), HLA, vascular cell adhesion molecule (VCAM), and β2-microglobulin (β2m), at the protein level in myositis (16, 17, 34). In this study, we tested whether other known NF-κB target genes were up-regulated in myositis muscle samples, using gene expression profiling. The analysis showed enhanced expression of several classic NF-κB target genes (e.g., β2m), including class I MHC alleles (Figure 5D). These data show that the NF-κB pathway is highly activated in muscle fibers of myositis patients.
Activation of caspase 12, a critical mediator of ER stress–induced cell death, in transgenic mice.
The main caspase associated with the ER apoptosis pathway is caspase 12, an ER-resident caspase, which, upon activation by ER stress and calcium release, can promote caspase 3–mediated apoptosis. To further characterize its role, we examined the activation of ER-associated caspase 12 in the muscle tissues of HT double-transgenic mice and control littermates, using an antibody that recognizes both intact and cleaved caspase 12. It has been shown previously that caspase 12 is highly expressed in mouse skeletal muscle tissues (35). Double-transgenic mice, but not control mice, exhibited the cleaved 28-kd caspase 12 fragment (Figure 6), suggesting activation of downstream apoptotic pathways in the muscle tissues. The specificity of the caspase 12 antibody for detection of cleaved caspase 12 fragments was further confirmed by immunoblotting of C2C12 muscle lysates that were treated with classic ER stress–inducing chemicals, such as thapsigargin and tunicamycin (results not shown).
In this study we demonstrated that 2 major components of the ER stress response pathway, ER overload response and unfolded protein response, are highly activated in muscle both in human myositis and in the mouse model. We further showed that overexpression of mouse class I MHC induces an ER stress response in transiently transfected C2C12 skeletal muscle cells.
The expression of class I MHC molecules on the surface of a cell is the end point of a complicated process of assembly and transport initiated in the ER. Class I MHC molecules are trimeric, consisting of a transmembrane glycoprotein which is the product of the MHC-linked gene, a small protein called β2m, and a short peptide of 8–10 amino acids. Assembly of class I MHC molecules has much in common with the folding and assembly of other multimeric glycoproteins in the ER. Housekeeping chaperones (calnexin, calreticulin, ERp57, tapasin, and Grp78) assist the assembly process. When certain murine class I MHC alleles are expressed in β2m-deficient cells, misfolding is associated with the formation of heavy chain homodimers in the ER via an unpaired Cys in the cytoplasmic tail. In those studies heavy chain dimerization was also observed in cells expressing β2m, but only after β2m had dissociated from previously assembled complexes (36).
In HLA–B27 (human class I heavy chain)–transgenic mice, spontaneous arthritis develops in the absence of β2m, indicating that cell surface expression of classic trimeric HLA–B27 (heavy chain–peptide–β2m) complexes is not required (37). Interestingly, the same spontaneous arthritis phenotype has been found in the absence of HLA–B27, in mice with a mixed genetic background that were deficient for either β2m or transporter-associated protein (38). These findings raise the possibility that class I heavy chain misfolding and the intracellular sequelae may be a critical component of the disease mechanism in myositis, perhaps through the generation of ER stress (39).
Skeletal muscle cells do not constitutively express or display class I MHC molecules, although they can be induced to do so by proinflammatory cytokines such as interferon-α and tumor necrosis factor α (TNFα) (6, 17, 40, 41). In human myositis, the early and widespread appearance of class I MHC in non-necrotic muscle cells, even distant from lymphocytic infiltration, is a striking feature (6, 7, 42). The pathologic significance of these observations has not been fully studied. We propose that class I MHC molecules themselves may potentially mediate muscle fiber damage and dysfunction in the absence of lymphocytes, via the ER stress response pathway. In particular, we propose that transgenic overexpression of class I MHC in muscle fibers may initiate 2 overlapping ER stress response pathways: the unfolded protein response and ER overload response. Under normal transient stress conditions, the unfolded protein response activates 3 proximal ER sensors, Ire-1, PERK, and ATF-6. These, in turn, control the rate of protein translation, with prolonged unfolded protein response activation ultimately initiating death pathways through CHOP and caspase 12. ER overload response activates NF-κB, and activated NF-κB induces NF-κB target genes (including up-regulation of endogenous class I MHC), which in turn will induce the ER stress response, thus initiating a self-sustaining loop.
It has recently been suggested that Grp78 is the master regulator of the ER stress response, the concert of signaling pathways that alters gene transcription in response to an accumulation of unfolded or misfolded proteins in the ER (30). Grp78 regulates the activation of at least 3 transducers of ER stress, ATF-6, Ire-1, and PERK (43). Grp78, which negatively regulates the unfolded protein response, interacts with all 3 sensors under nonstressed conditions and probably plays an essential regulatory role in unfolded protein response activation (43). In stressed cells, Grp78 binds to accumulating unfolded or misfolded proteins, releasing these 3 transducers and thus allowing activation of downstream nuclear events (44). We have shown that Grp78 is highly up-regulated both in human myositis and in the mouse model. Using global gene expression profiling, we have further demonstrated that several members of the ER stress response pathway are greatly up-regulated in myositis patients compared with healthy controls. These data suggest that the ER stress response pathway is highly activated in myositis. Activation of ER stress and unfolded protein response pathways has also been demonstrated in sporadic inclusion body myositis (45).
To further test the effects of class I MHC induction, we transiently transfected C2C12 muscle cells with WT and mutant H-2Kb constructs and showed that WT, but not mutant, H-2Kb induced Grp78 expression. This may be due to their inability to fold and to interact properly with chaperones such as calnexin and calreticulin in the ER. These mutants may be translocated from the ER lumen to cytosol for degradation by the proteosome (46). The data demonstrate that class I MHC overexpression in muscle cells can lead to ER stress.
We have previously demonstrated that muscle disease in class I MHC–transgenic mice is self-sustaining (13). The disease in HT double-transgenic mice could not be reversed by readministering doxycycline after the onset of clinical disease, probably because of the activation of transgene-independent genes such as endogenous H-2Kb, H-2Db, ICAM-1, and the proinflammatory chemokines and cytokines macrophage inflammatory protein 1α, monocyte chemotactic protein 1 (MCP-1), and interleukin-15 (IL-15) (13). We tested the possibility that the initial transgenic class I MHC overexpression in the sarcoplasmic reticulum of muscle fibers may trigger activation of the NF-κB pathway through ER overload response, and NF-κB would in turn activate several proinflammatory target genes (cytokines, cytokine receptors, cell adhesion molecules, chemoattractant proteins, and growth regulators) including endogenous class I MHC, thus initiating a self-sustaining loop. The ER overload response partly overlaps with, but is distinct from, the unfolded protein response (47). Some, but not all, of the conditions that evoke the unfolded protein response will also trigger the ER overload response. In contrast to the unfolded protein response, the ER overload response activates the nuclear transcription factor NF-κB (32). The promoter sequences of the ER chaperones Grp78 and Grp94 do not contain NF-κB binding sites; accordingly, the ER overload response–associated NF-κB induction does not increase their expression levels.
Since the ER overload response can be triggered by high levels of normal or mutant proteins processed through or retained in the ER, it is believed to have a broad role in maintaining cellular homeostasis under stress conditions (32). It is possible that in human myositis, NF-κB may activate both classical (proinflammatory cytokines IL-1 and TNFα) and nonclassical (ER stress response) pathways (48). The downstream target genes (e.g., class I MHC, ICAM, VCAM, E-selectin, MCP-1) regulated by the NF-κB pathway are highly up-regulated in myositis patients (16). These results further confirm the previous finding of NF-κB activation in inflammatory myopathies (49). Thus, we have shown that NF-κB p65 is activated both in human myositis and in the mouse model, suggesting that this pathway may be directly involved in muscle fiber damage or dysfunction.
Mechanisms underlying the class I MHC–induced muscle damage in myositis may be proposed based on earlier findings and the results obtained in the present study. Class I MHC overexpression (e.g., transgenic or infection or nerve injury in an autoimmune-prone genetic background) in muscle fibers may initiate 2 overlapping ER stress response pathways: ER overload and unfolded protein response. The ER overload response will activate NF-κB, and activated NF-κB will induce NF-κB target genes including endogenous class I MHC up-regulation, which in turn will induce the ER stress response, thus initiating a self-sustaining loop. NF-κB is known to suppress myoblast differentiation by inhibiting MyoD (50). It is also known to induce proinflammatory cytokine expression and consequently, muscle fiber damage. Under normal transient stress conditions, the unfolded protein response will activate 3 proximal ER sensors, Ire-1, PERK, and ATF-6, and these in turn control the rate of protein translation. Prolonged unfolded protein response activation will ultimately initiate death pathways through CHOP and caspase 12.
Therefore, we propose that class I MHC–initiated ER stress response may be one of the nonimmune mechanisms of muscle fiber damage in myositis (Figure 7). Interestingly, we have observed significant muscle wasting and reduction in body weight in the self-sustaining phase of the disease in our mouse model (13). It therefore appears that skeletal muscle cell death in class I MHC–expressing muscle fibers can be induced partly by cytotoxic T lymphocytes, which initiate the “extrinsic” death pathway, and partly by perturbation of intracellular homeostasis (ER stress response), the “intrinsic” damage pathway. These results suggest that therapeutic interventions that target inflammatory and ER stress pathways may be useful in myositis. These effector pathogenic mechanisms may also be relevant in other muscle diseases in which overexpression of class I MHC, or possibly other molecules, occurs.