Juvenile dermatomyositis (JDM), the most common pediatric inflammatory myopathy, is a systemic vasculopathy affecting young children. Epidemiology studies documenting an antecedent illness in the 3 mo before the first definite symptom (rash and/or weakness) of JDM are supported by immunologic data that suggest that the disease pathophysiology is Ag driven. The purpose of this study was to compare the gene expression profiles in muscle biopsies of four untreated DQA1*0501+ JDM children with profiles from children with a known necrotizing myopathy (Duchenne muscular dystrophy), as well as an in vitro antiviral model (NF90), and healthy pediatric controls. Nearly half (47%) of the dysregulated genes in JDM were associated with the immune response. In particular, increased expression of IFN-αβ-inducible genes 6-16, myxovirus resistance protein p78, latent cytosolic transcription factor, LMP2, and TAP1 was observed. This profile is consistent with an IFN-αβ transcription cascade seen in the in vitro viral resistance model. The IFN-αβ-inducible profile was superimposed on transcription profiles reflective of myofiber necrosis and regeneration shared with Duchenne muscular dystrophy. Expressed genes were confirmed by quantitative real-time PCR (6-16), immunofluorescence (thrombospondin 4), and immunolocalization (IFN-γ, p21). We hypothesize that these data support a model of Ag (?viral) induction of an apparent autoimmune disease based on dynamic interaction between the muscle, vascular, and immune systems in the genetically susceptible (DQA1*0501+) child.

Juvenile dermatomyositis (JDM)3 is the most common pediatric inflammatory myopathy. Recognition of the first definite symptom, rash and/or muscle weakness (disease onset), can precede diagnosis and treatment by months to years (1). We interviewed parents or caretakers of over 375 children with newly diagnosed JDM and found that the first definite symptom of JDM was preceded frequently by a systemic illness, often with upper respiratory symptoms requiring antibiotics (1, 2, 3). Genetic risk factors are also associated with JDM, for the HLA class II Ag DQA1*0501 (4) shows a stronger association than DR3 (5) and the class I Ag HLA-B8 (6). Over 85% of white and minority children with JDM are positive for DQA1*0501 (vs 25% in controls), suggesting that this allele might be a factor in disease susceptibility (7). Despite this evidence, examination of diagnostic muscle biopsies from DQA1*0501+ JDM children obtained within 3 mo of disease onset did not identify persistent viral RNA nor bacterial DNA, suggesting either that an Ag might initiate the disease, but not perpetuate the process, and/or that the (viral/antimicrobial) message is rapidly degraded (8).

The pathophysiology of JDM is not known, but evidence suggests that both humoral and cell-mediated components may participate in vascular and muscle damage (9). For example, Ig is deposited on muscle fibers in association with the membrane attack complex (10). Although a majority of JDM sera contain a speckled pattern antinuclear Ab of unknown specificity (11), and a subset of these sera have Ab to a 56-kDa nuclear protein (12), which is more frequent in DQA1*0501-positive children with JDM (13), Ab-dependent cell-mediated cytotoxicity has not been demonstrated (14). Peripheral blood from untreated JDM demonstrates a lymphopenia, with a selective decrease in the CD8+ subset as well as ICAM-I-positive non-CD19+ cells (presumed T cells) (15). Muscle biopsies from untreated children with JDM contained proportionally more lymphocytes positive for CD8 and CD56 than concurrently obtained peripheral blood (16). NK cells (CD56) are the primary response element to viral infections and release IFN (17). TCR studies of muscle biopsies from untreated DQA1*0501+ children with a history of preceding infection shortly before the diagnosis of JDM show clonal expansion of T cells (18) containing specific CDR3 Ag-combining regions, suggesting that the immune response is Ag driven (16). Children with JDM and a G-to-A polymorphism in the −308 promoter position of the TNF-α locus have both a statistically significant increased cellular production of TNF-α and a prolonged JDM clinical course (19).

To further describe the pathogenesis of JDM, we used Affymetrix GeneChip microarrays (Affymetrix, Santa Clara, CA) to analyze the genes expressed in muscle of untreated children with JDM. We present expression profile data from four DQA1*0501+ JDM children, using a ∼5600 full-length gene Affymetrix chip, with duplicate chips tested. First, we derived a set of gene expression changes shared by all JDM patients compared with normal control muscle, then we identified subsets of the JDM expression profile that show striking similarity to profiles of an antiviral response (NF90-transfected cells) compared with muscular dystrophy profiles. The information from these expression profiles led to the conceptualization of a novel model of the pathogenesis of JDM in which the IFN-αβ-induced response plays a major role.

We obtained muscle biopsies from four Caucasian girls (age 5–16 years) who fulfilled the criteria for definite JDM (20), were negative for myositis-specific Ags as well as myositis-associated Ags, and were examined by the senior author at the Children’s Memorial Hospital (Chicago, IL). Appropriate informed consent was obtained from all parents as well as children over the age of 12. All four patients were matched for the DQA1*0501+ marker but had different TNF-α-308 alleles and duration of disease before muscle biopsy. This selection was done to ensure that all four of these children probably shared the same DQA-matched disease characteristics. Two patients were heterozygous for TNF-α-308 GA (the A allele is associated with greater production of TNF-α by JDM PBMCs) (19) as well as JDM muscle fibers (21) and had a short disease course (mean of 0.5 years from disease onset to muscle biopsy), and two of the children were homozygous for TNF-α-308 GG with a long period (1.3 years) of untreated disease before a diagnostic muscle biopsy was obtained (Table I). Immediately before anesthesia for the muscle biopsy, blood was taken from each child for subsequent analysis of sera and Ficoll-Hypaque isolation of PBMCs. Magnetic resonance imaging-directed muscle biopsies were obtained from affected regions of the muscle and rapidly frozen at −80°C before shipping to the Children’s National Medical Center (Washington, DC) on dry ice for expression profiling.

Table I.

Demographics of JDM children studied by expression profiling

PatientDQA1*0501TNF-α-308 AlleleAge at Onset (years)Age at Biopsy (years)
133 GA 9.70 10.13 
188 GA 6.07 6.61 
136 GG 14.74 16.02 
129 GG 3.88 5.21 
PatientDQA1*0501TNF-α-308 AlleleAge at Onset (years)Age at Biopsy (years)
133 GA 9.70 10.13 
188 GA 6.07 6.61 
136 GG 14.74 16.02 
129 GG 3.88 5.21 

Muscle biopsies and expression profiles from normal age-matched controls and children with a necrotizing myopathy (Duchenne muscular dystrophy (DMD)) were obtained as described (22). Expression profiles were done on duplicate HuFL GeneChips, and expression changes surviving four pairwise comparisons were retained (22). These results are available on the Journal of Cell Biology web site (http://www.jcb.org/cgi/content/full/151/6/1321/DC1) and on the Children’s National Medical Center Microarray web site (http://microarray.CNMCResearch.org/pga.htm). Included are raw image files for each of the four microarrays, text files containing absolute analyses of each chip (“present” calls; GeneChip software output), and comparison analyses between different chips (“difference” calls; GeneChip software output).

We analyzed samples on duplicate Affymetrix HuFL (5600 gene) GeneChips, with four pairwise comparisons (22). Total RNA was extracted from four JDM muscle biopsies by using TRIzol reagent (Life Technologies, Gaithersburg, MD). Total RNA from two JDM patients who had TNF-α-308 GA was pooled into two parallel pools with equal amounts of RNA (5 μg of each RNA in both pools for a total of 10 μg RNA/assay). Similarly, total RNA from two JDM patients who were TNF-α-308 GG was pooled in two parallel pools with equal amounts of RNA. Ten micrograms of total RNA from each group (four parallel groups) were further processed. Only expression changes that survived eight iterative comparisons were retained for further analysis.

GeneChip software (version 3.3; Affymetrix) was used to analyze Affymetrix microarrays (22). To compare different data sets (e.g., JDM with TNF-α-308 GA vs normal control), each probe pair in an experimental GeneChip assay was compared with control groups, and four matrices were used to determine the “difference calls” that indicate whether the transcription level of a gene is different. A tilde (∼) is assigned by Affymetrix software when the denominator approaches zero (e.g., in absent calls), leading to a possible exaggeration of the resulting fold change. Iterative comparisons of different datasets were analyzed using Microsoft Excel (Microsoft, Redmond, WA). For example, each TNF-α-308 GA JDM chip (n = 2) and TNF-α-308 GG JDM chip (n = 2) was compared with each control chip (n = 2) to determine the expression difference between each JDM allele and the control, resulting in eight pairwise comparisons. In addition, each JDM chip (two TNF-α-308 GA, two TNF-α-308 GG) was compared with each of two DMD expression profiles, again leading to eight pairwise comparisons. Only the difference calls that showed consistently more or fewer calls in all eight pairwise comparisons for each disease were extracted for further analysis.

To confirm the differential regulation of specific genes, separate aliquots of the same muscle biopsies were analyzed in the Chicago laboratory at the Children’s Memorial Institute for Education and Research. Total RNA was isolated from muscle using the RNeasy Mini kit (Qiagen, Valencia, CA) and treated with DNase I (1 U/μg) at 25°C for 30 min, followed by heat inactivation at 75°C for 10 min. QRT-PCR was performed using primers and probes designed by using Primer Express software (PE Applied Biosystems, Foster City, CA).

Primers used were human IFN-inducible peptide (6-16) gene IFN-stimulatable response element (ISRE) (5′-TAAGAAAAAGTGCTCGGAGAGCTC-3′ and 5′-CCGACGGCCATGAAGGT-3′) and human β-actin (5′-TCACCCACACTGTGCCCATCTACGA-3′ and 5′-CAGCGGAACCGCTCATTGCCAATGG-3′).

Probes used were human IFN-inducible peptide (6-16) gene ISRE (5′-6Fam ACAGCGGCTCCGGGTTCTGGA TAMRA-3′) and human β-actin (5′-6Fam ATGCCCTCCCCCATGCCATCCTGCGT TAMRA-3′).

Standard methods were used for RT-PCR on an ABI 7700 sequencer (PE Applied Biosystems), and followed a four-step PCR at 48°C for 30 min and 95°C for 15 min, followed by 40 cycles of 95°C for 15 s and 59°C for 1 min.

We used expression profiles of the intracellular response to viral infection identified by our laboratory (I. Krasnoselska-Riz, unpublished observations). These expression profiles were generated from stably transfected GHOST(3) CXCR4 cells, obtained from the National Institutes of Health AIDS Research and Reference Program (23). The GHOST(3) CXCR4 cells were transduced either with vector expressing a dsRNA-binding protein, NF90, or the empty vector. NF90-transfected cells were HIV infected and shown to inhibit viral replication. Expression profiling analysis has documented that >50% of genes displaying 4-fold or larger changes between NF90 and vector-transfected cells were genes known to be elicited by IFN-αβ, but not by IFN-γ (24), and were associated with viral resistance.

Serial 8-μm-thick muscle sections were flash frozen and processed. We used a 1/500 dilution of thrombospondin-4 polyclonal Ab (provided by J. Lawler, Beth Israel Deaconess Medical Center at Harvard Medical School, Boston, MA), a 1/150 dilution of p21 mAb, and a 1/50 dilution of IFN-γ mAb (both supplied by Santa Cruz Biotechnology, Santa Cruz, CA) in conjunction with an avidin biotin indicator technique (ABC kit; Vector Laboratories, Burlingame, CA). All secondary Abs were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

We defined a stringent set of gene expression differences in DQA1*0501-positive JDM patients that were consistent and independent of disease duration or TNF-α allele differences. We selected four same-sex (female) patients who were matched for DQA1*0501 markers but differed in disease duration (0.5 vs 1.5 years) and in TNF-α alleles (GG or GA genotypes). We then used a duplicate mixed sample experimental design to normalize genetic noise (polymorphic heterogeneity), as previously described (22). Expression profiles for the JDM muscle biopsies were generated using Affymetrix HuGeneFL GeneChips (∼5600 full-length genes) and then compared with the same GeneChips from two separate normal, age-matched controls. Only gene expression changes showing similar fold changes (at least 2-fold difference) in all eight of the following comparisons were retained for further analysis: JDM-GA 1 vs control 1; JDM-GA 1 vs control 2; JDM-GA 2 vs control 1; JDM-GA 2 vs control 2; JDM-GG 1 vs control 1; JDM-GG 1 vs control 2; JDM-GG 2 vs control 1; JDM-GG 2 vs control 2.

We studied four different patient muscle biopsies and used this highly stringent data selection to normalize genetic polymorphic variation in expression patterns between different individuals; gene expression changes correlating with the disease were retained after all eight comparisons. The description of genes tested is listed on our website (http://microarray.CNMCResearch.org/pga.htm; see “muscle, human” under research data). Among the 7,095 probe sets (∼280,000 oligonucleotide features) on the Affymetrix HuGeneFL microarray, we found a consistent number of present calls for each of the four JDM cRNAs tested (JDM-GA, 42 and 41%; JDM-GG, 44 and 41%), and the reproducibility of data between duplicate chips was excellent (Fig. 1,A), with an increase in variability for genes expressed at low levels. There was considerably more variability in expression profiles between JDM samples and control samples (Fig. 1,B). From the eight pairwise comparisons of JDM-GG and JDM-GA profiles with normal control profiles, 178 differentially regulated genes survived all comparisons (Tables II and III). Thus, ∼40% of the 7,095 probe sets tested were expressed in muscle, and ∼6% of them showed differential regulation in JDM; 91 genes showed >2-fold higher expression, and 87 genes showed >2-fold lower expression.

FIGURE 1.

Expression profiling of 5600 genes in JDM muscle and controls. Shown are scatterplots of expression-profiling comparative data, with axes showing relative expression levels of probe sets for genes studied. Only present calls for the individual profiles are shown, representing ∼40% of all genes tested. A, The comparison of two different JDM expression profiles, in which the fold changes of most genes are within the 2-fold difference range (parallel lines). Lack of concordance between JDM 1 (GA allele) and JDM 2 (AA allele) profiles represents combined sample and experimental variability. B, The comparison of JDM with a mixed control muscle profile and the marked differences in gene expression between the two profiles; many genes have fold changes greater than the 2-fold cutoff.

FIGURE 1.

Expression profiling of 5600 genes in JDM muscle and controls. Shown are scatterplots of expression-profiling comparative data, with axes showing relative expression levels of probe sets for genes studied. Only present calls for the individual profiles are shown, representing ∼40% of all genes tested. A, The comparison of two different JDM expression profiles, in which the fold changes of most genes are within the 2-fold difference range (parallel lines). Lack of concordance between JDM 1 (GA allele) and JDM 2 (AA allele) profiles represents combined sample and experimental variability. B, The comparison of JDM with a mixed control muscle profile and the marked differences in gene expression between the two profiles; many genes have fold changes greater than the 2-fold cutoff.

Close modal
Table II.

Gene expression changes in JDM: similarities to a viral response profile (NF90)

Accession No.JDMaNF90DMDImmune Response
M33882 96.4a 35.9 NCb p78 protein (MxA) 
U22970 ∼76.8 48.6 NC 6-16 gene (IFN-inducible peptide precursor) 
Z14982 ∼41.7 NC NC MHC-encoded proteasome subunit gene LMP7 
U52513 38.9 20.0 NC RIG-G 
X57522 ∼36.8 5.0 NC RING4 
U53830 ∼31.1 NC NC IFN regulatory factor 7A 
X02530 27.8 NC NC IP-10 
J03909 ∼27.1 NC 13.0 IP-30 
X67325 22.5 133.7 NC p27 
U72882 18.6 5.7 NC IFN-induced leucine zipper protein (IFP35) 
M14660 ∼16.9 NC NC ISG-54K (IFN-stimulated gene) encoding a 54-kDa protein 
M87789 15.5 NC NC Anti-hepatitis A IgG 
AB000115 15.5 10.0 NC AB000115 mRNA (similar to p44 protein Ag) 
M97935 13.2 NC NC ISGF-3 
X02874 13.0 23.5 NC (2′,5′) oligo(A) synthetase E (1.6-kb RNA) 
X02875 11.2 6.0 NC (2′,5′) oligo(A) synthetase E (1.8-kb RNA) 
M14058 10.6 NC 5.4 Complement C1r 
M30818 10.2 10.0 NC IFN-induced cellular resistance mediator protein (MxB) 
K03430 10.2 NC 5.7 Complement C1q B-chain 
M87434 8.4 NC NC 71-kDa 2′,5′ oligoadenylate synthetase (p69 2-5A synthetase) 
M59815 7.8 NC NC Complement component C4A 
X64364 6.5 NC NC M6 Ag 
X66401 6.4 4.6 NC LMP2 gene 
X82200 5.7 13.0 NC Staf50 
M59830 5.0 NC NC MHC class III HSP70-2 gene (HLA) 
M87284 4.7 NC NC 69-kDa 2′,5′ oligoadenylate synthetase (P69 OAS) 
D32129 4.5 NC 3.4 HLA class-I (HLA-A26) H chain 
U50648 4.5 NC NC IFN-inducible RNA-dependent protein kinase (Pkr) gene 
HG3597- 4.0 NC NC MHC class I 
U37518 3.9 NC NC TRAIL 
J04080 3.9 NC 5.2 Complement component C1r 
M24594 3.8 17.6 NC IFN-inducible 56-kDa protein 
Accession No.JDMaNF90DMDImmune Response
M33882 96.4a 35.9 NCb p78 protein (MxA) 
U22970 ∼76.8 48.6 NC 6-16 gene (IFN-inducible peptide precursor) 
Z14982 ∼41.7 NC NC MHC-encoded proteasome subunit gene LMP7 
U52513 38.9 20.0 NC RIG-G 
X57522 ∼36.8 5.0 NC RING4 
U53830 ∼31.1 NC NC IFN regulatory factor 7A 
X02530 27.8 NC NC IP-10 
J03909 ∼27.1 NC 13.0 IP-30 
X67325 22.5 133.7 NC p27 
U72882 18.6 5.7 NC IFN-induced leucine zipper protein (IFP35) 
M14660 ∼16.9 NC NC ISG-54K (IFN-stimulated gene) encoding a 54-kDa protein 
M87789 15.5 NC NC Anti-hepatitis A IgG 
AB000115 15.5 10.0 NC AB000115 mRNA (similar to p44 protein Ag) 
M97935 13.2 NC NC ISGF-3 
X02874 13.0 23.5 NC (2′,5′) oligo(A) synthetase E (1.6-kb RNA) 
X02875 11.2 6.0 NC (2′,5′) oligo(A) synthetase E (1.8-kb RNA) 
M14058 10.6 NC 5.4 Complement C1r 
M30818 10.2 10.0 NC IFN-induced cellular resistance mediator protein (MxB) 
K03430 10.2 NC 5.7 Complement C1q B-chain 
M87434 8.4 NC NC 71-kDa 2′,5′ oligoadenylate synthetase (p69 2-5A synthetase) 
M59815 7.8 NC NC Complement component C4A 
X64364 6.5 NC NC M6 Ag 
X66401 6.4 4.6 NC LMP2 gene 
X82200 5.7 13.0 NC Staf50 
M59830 5.0 NC NC MHC class III HSP70-2 gene (HLA) 
M87284 4.7 NC NC 69-kDa 2′,5′ oligoadenylate synthetase (P69 OAS) 
D32129 4.5 NC 3.4 HLA class-I (HLA-A26) H chain 
U50648 4.5 NC NC IFN-inducible RNA-dependent protein kinase (Pkr) gene 
HG3597- 4.0 NC NC MHC class I 
U37518 3.9 NC NC TRAIL 
J04080 3.9 NC 5.2 Complement component C1r 
M24594 3.8 17.6 NC IFN-inducible 56-kDa protein 
a

Average fold-change in JDM vs control samples.

b

NC, No change.

Table III.

Gene expression in muscle from untreated children with JDM or DMD: a myofiber necrosis profile

Accession No.JDMDMD
Muscle structure, extracellular matrix, and developmental genes    
L05606 ∼71.3 32.0 Myosin binding protein H 
X13100 ∼52.5 140.0 Embryonic myosin H chain 
U28811 19.4 NCa Cysteine-rich fibroblast growth factor receptor (CFR-1) 
Z38133 13.9 51.1 Skeletal muscle perinatal myosin H chain (MYH8) 
S77094 12.2 19.4 Nicotinic acetylcholine receptor α subunit|AChR 
Z19585 8.2 15.3 Thrombospondin-4 
Z74616 7.4 11.4 Prepro-α(I) collagen 
X78565 5.9 NC Tenascin-C 
U16306 ∼5.0 8.0 Chondroitin sulfate proteoglycan versican 
Z54367 4.2 NC Plectin 
U21128 3.9 4.7 Lumican 
L02950 −4.8 −22.0 μ-Crystallin 
HG2259- −3.3 −8.3 Tubulin, α1, isoform 44 
Signaling, transport, and cell cycle    
X78992 ∼46.7 NC FRF-2 
U09579 ∼31.7 16.0 Melanoma differentiation associated (mda-6, p21) 
X02404 22.7 NC Fragment for second calcitonin gene related peptide (CGRP) 
X67325 22.5 NC p27 
U39412 13.3 NC Platelet α SNAP 
M55542 ∼12.2 NC Guanylate binding protein isoform I (GBP-2) 
D13666 ∼9.5 4.0 Osteoblast specific factor 2 (OSF-2os) 
M55543 7.2 NC Guanylate binding protein isoform II (GBP-2) 
L22342 6.6 NC Nuclear phosphoprotein 
U70322 6.4 NC Transportin (TRN) 
M28211 ∼−19.2 −2.2 GTP-binding protein (RAB4) 
Z31695 ∼−15.2 NC 43-kDa inositol polyphosphate 5-phosphatase 
X87843 −11.1 NC Cyclin H assembly factor 
X77794 −5.6 NC Cyclin G1 
L08666 −2.8 NC Porin 
U48707 −2.8 −3.1 Protein phosphatase-1 inhibitor 
M14199 −2.6 NC Laminin receptor 
S72008 −2.6 NC CDC10 homolog (hCDC10) 
Ribosomal structure and function    
U65581 −4.5 −3.3 Ribosomal protein L3-like 
M36072 −4.5 NC Ribosomal protein L7a (surf3) large subunit 
D87735 −3.9 NC Ribosomal protein L14 
S80343 −3.8 NC (ArgRS)-Arginyl-tRNA synthetase 
U14971 −3.7 NC Ribosomal protein S9 
M60854 −3.5 NC Ribosomal protein S16 
U15008 −3.4 NC SnRNP core protein Sm D2 
U14972 −3.3 NC Ribosomal protein S10 
U14970 −3.3 NC Ribosomal protein S5 
X06617 −3.1 NC Ribosomal protein S11 
D23660 −2.9 NC Ribosomal protein 
HG4542- −2.9 NC Ribosomal protein L10 
HG311- −2.8 NC Ribosomal protein L30 
Z25749 −2.9 NC Ribosomal protein S7 
U58682 −2.8 NC Ribosomal protein S28 
L11566 −2.7 NC Ribosomal protein L18 (RPL18) 
X15940 −2.7 NC Ribosomal protein L31 
HG4319- −2.6 NC Ribosomal protein L5 
Mitochondrial proteins and energy metabolism    
HG3141- ∼−31.4 NC NADH-ubiquinone oxidoreductase 
D26308 −8.2 NC NADPH-flavin reductase 
D16480 −7.7 NC Mitochondrial enoyl-CoA hydratase/3-hydroxyacyl-CoA 
S69232 −5.4 −3.0 Electron transfer flavoprotein-ubiquinone oxidoreductase 
D17400 −4.6 −2.5 6-Pyruvoyl-tetrahydropterin synthase 
X56997 −4.4 NC UbA52 gene coding for ubiquitin-52 amino acid fusion protein 
X99728 −4.2 −3.3 NDUFV3 gene, NADH-ubiquinone oxidoreductase flavoprotein 3 
U17886 −3.7 −2.0 Succinate dehydrogenase iron-protein subunit (sdhB) 
L33842 −3.6 NC Type II inosine monophosphate dehydrogenase (IMPDH2) 
J04823 −3.6 NC Cytochrome c oxidase subunit VIII (COX8) 
M19961 −3.6 NC Cytochrome c oxidase subunit Vb (coxVb) 
X69908 −3.6 NC P2 gene for c subunit of mitochondrial ATP synthase gene 
AC002115 −3.5 NC COX6B gene (COXG) 
HG2279- −3.5 NC Triosephosphate isomerase 
J05073 −3.5 −4.5 Phosphoglycerate mutase (PGAM-M) 
U16660 −3.5 NC Peroxisomal enoyl-CoA hydratase-like protein (HPXEL) 
J05401 −3.4 NC Sarcomeric mitochondrial creatine kinase (MtCK) 
X65867 −3.4 NC Adenylosuccinate lyase 
   (Table continues
Accession No.JDMDMD
Muscle structure, extracellular matrix, and developmental genes    
L05606 ∼71.3 32.0 Myosin binding protein H 
X13100 ∼52.5 140.0 Embryonic myosin H chain 
U28811 19.4 NCa Cysteine-rich fibroblast growth factor receptor (CFR-1) 
Z38133 13.9 51.1 Skeletal muscle perinatal myosin H chain (MYH8) 
S77094 12.2 19.4 Nicotinic acetylcholine receptor α subunit|AChR 
Z19585 8.2 15.3 Thrombospondin-4 
Z74616 7.4 11.4 Prepro-α(I) collagen 
X78565 5.9 NC Tenascin-C 
U16306 ∼5.0 8.0 Chondroitin sulfate proteoglycan versican 
Z54367 4.2 NC Plectin 
U21128 3.9 4.7 Lumican 
L02950 −4.8 −22.0 μ-Crystallin 
HG2259- −3.3 −8.3 Tubulin, α1, isoform 44 
Signaling, transport, and cell cycle    
X78992 ∼46.7 NC FRF-2 
U09579 ∼31.7 16.0 Melanoma differentiation associated (mda-6, p21) 
X02404 22.7 NC Fragment for second calcitonin gene related peptide (CGRP) 
X67325 22.5 NC p27 
U39412 13.3 NC Platelet α SNAP 
M55542 ∼12.2 NC Guanylate binding protein isoform I (GBP-2) 
D13666 ∼9.5 4.0 Osteoblast specific factor 2 (OSF-2os) 
M55543 7.2 NC Guanylate binding protein isoform II (GBP-2) 
L22342 6.6 NC Nuclear phosphoprotein 
U70322 6.4 NC Transportin (TRN) 
M28211 ∼−19.2 −2.2 GTP-binding protein (RAB4) 
Z31695 ∼−15.2 NC 43-kDa inositol polyphosphate 5-phosphatase 
X87843 −11.1 NC Cyclin H assembly factor 
X77794 −5.6 NC Cyclin G1 
L08666 −2.8 NC Porin 
U48707 −2.8 −3.1 Protein phosphatase-1 inhibitor 
M14199 −2.6 NC Laminin receptor 
S72008 −2.6 NC CDC10 homolog (hCDC10) 
Ribosomal structure and function    
U65581 −4.5 −3.3 Ribosomal protein L3-like 
M36072 −4.5 NC Ribosomal protein L7a (surf3) large subunit 
D87735 −3.9 NC Ribosomal protein L14 
S80343 −3.8 NC (ArgRS)-Arginyl-tRNA synthetase 
U14971 −3.7 NC Ribosomal protein S9 
M60854 −3.5 NC Ribosomal protein S16 
U15008 −3.4 NC SnRNP core protein Sm D2 
U14972 −3.3 NC Ribosomal protein S10 
U14970 −3.3 NC Ribosomal protein S5 
X06617 −3.1 NC Ribosomal protein S11 
D23660 −2.9 NC Ribosomal protein 
HG4542- −2.9 NC Ribosomal protein L10 
HG311- −2.8 NC Ribosomal protein L30 
Z25749 −2.9 NC Ribosomal protein S7 
U58682 −2.8 NC Ribosomal protein S28 
L11566 −2.7 NC Ribosomal protein L18 (RPL18) 
X15940 −2.7 NC Ribosomal protein L31 
HG4319- −2.6 NC Ribosomal protein L5 
Mitochondrial proteins and energy metabolism    
HG3141- ∼−31.4 NC NADH-ubiquinone oxidoreductase 
D26308 −8.2 NC NADPH-flavin reductase 
D16480 −7.7 NC Mitochondrial enoyl-CoA hydratase/3-hydroxyacyl-CoA 
S69232 −5.4 −3.0 Electron transfer flavoprotein-ubiquinone oxidoreductase 
D17400 −4.6 −2.5 6-Pyruvoyl-tetrahydropterin synthase 
X56997 −4.4 NC UbA52 gene coding for ubiquitin-52 amino acid fusion protein 
X99728 −4.2 −3.3 NDUFV3 gene, NADH-ubiquinone oxidoreductase flavoprotein 3 
U17886 −3.7 −2.0 Succinate dehydrogenase iron-protein subunit (sdhB) 
L33842 −3.6 NC Type II inosine monophosphate dehydrogenase (IMPDH2) 
J04823 −3.6 NC Cytochrome c oxidase subunit VIII (COX8) 
M19961 −3.6 NC Cytochrome c oxidase subunit Vb (coxVb) 
X69908 −3.6 NC P2 gene for c subunit of mitochondrial ATP synthase gene 
AC002115 −3.5 NC COX6B gene (COXG) 
HG2279- −3.5 NC Triosephosphate isomerase 
J05073 −3.5 −4.5 Phosphoglycerate mutase (PGAM-M) 
U16660 −3.5 NC Peroxisomal enoyl-CoA hydratase-like protein (HPXEL) 
J05401 −3.4 NC Sarcomeric mitochondrial creatine kinase (MtCK) 
X65867 −3.4 NC Adenylosuccinate lyase 
   (Table continues
a

NC, No change.

We clustered misregulated genes in JDM according to pathologic processes and cellular localization (Tables II and III). These results showed that genes involved in immune responses were the largest group (47%) of up-regulated genes in JDM (Table II). IFN-αβ triggers transcription of most of these genes (see Table II and Fig. 4) and can facilitate the expression of IFN-γ-induced genes (17). We hypothesized that one possible stimulus of the IFN-responsive genes might be viral. We then compared the results obtained by expression profiling of JDM samples with an in vitro model for the intracellular antiviral response previously obtained in our laboratory (described in Materials and Methods; I. Krasnoselska-Riz, unpublished observations). An antiviral response was defined by transfection of NF90, a constitutive inducer of a response, which inhibits HIV replication, and the expression patterns were obtained from RNA isolated from isogenic cell lines differing only in the expression of NF90. NF90 expression induced IFN-αβ-stimulated genes to levels comparable with those achieved by specific IFN-β exposure (24). The results of the comparison of the genes expressed in the muscle of untreated children with JDM and the in vitro NF90 model were similar (Table II). We found that 12 up-regulated genes were shared between JDM and NF90 profiles (Table II). Ten of the 12 genes were up-regulated >10-fold in JDM (ranging from 10- to 97-fold), while the same genes were up-regulated >5-fold in the viral profile model (ranging from 5- to 130-fold). The high degree of sharing of JDM and NF90 profiles suggests that JDM pathophysiology is associated with an IFN-stimulated response. The exact cellular origin of this response remains to be determined.

FIGURE 4.

A model for the pathophysiology of JDM. The data show a persistent IFN-induced response in JDM muscle. This response may be due to a persistent, unidentified infection of cells or a hit-and-run mechanism in which the agent initiates antiinfectious/ischemic/necrotic pathways that become self-perpetuating. Specific down- and up-regulated genes shown by expression profiling are indicated. The genes in red were verified at the protein level, and those in green were verified by QRT-PCR. These models incorporated evidence for intracellular IFN-induced gene profiles found in patient muscle biopsy (lymphocyte, endothelial, and/or vascular smooth muscle). Gene expression changes localized to smooth muscle are shown (p21, IFN-γ), with other p21/cell cycle-associated gene transcription changes hypothetically placed in vascular smooth muscle. Degeneration/regeneration cascades in myofibers were also observed, as indicated. Our model proposes that TNF-α exacerbates the protective response via the NF-κB pathway, as indicated in the lower parts of the diagram. (NF-κB binding site is shown upstream of the same genes as ISRE, although most commonly it is located in separate promoters from ISRE.) We propose that HLA-DQA*0501 is an essential site for Ag/?viral attachment or internalization and/or Ag presentation (top of diagram).

FIGURE 4.

A model for the pathophysiology of JDM. The data show a persistent IFN-induced response in JDM muscle. This response may be due to a persistent, unidentified infection of cells or a hit-and-run mechanism in which the agent initiates antiinfectious/ischemic/necrotic pathways that become self-perpetuating. Specific down- and up-regulated genes shown by expression profiling are indicated. The genes in red were verified at the protein level, and those in green were verified by QRT-PCR. These models incorporated evidence for intracellular IFN-induced gene profiles found in patient muscle biopsy (lymphocyte, endothelial, and/or vascular smooth muscle). Gene expression changes localized to smooth muscle are shown (p21, IFN-γ), with other p21/cell cycle-associated gene transcription changes hypothetically placed in vascular smooth muscle. Degeneration/regeneration cascades in myofibers were also observed, as indicated. Our model proposes that TNF-α exacerbates the protective response via the NF-κB pathway, as indicated in the lower parts of the diagram. (NF-κB binding site is shown upstream of the same genes as ISRE, although most commonly it is located in separate promoters from ISRE.) We propose that HLA-DQA*0501 is an essential site for Ag/?viral attachment or internalization and/or Ag presentation (top of diagram).

Close modal

Because JDM is a vasculitic autoimmune disease characterized by small vessel occlusion with primary disease expression in muscle and skin, we expected many pathologic processes involving expression of genes of several different pathways to be involved in JDM. These pathways should include muscle regeneration/degeneration cascades and genes participating in myofiber necrosis, which had been previously identified in the muscle of children with DMD (22). Therefore, we compared the expression profiles of JDM and previously reported profiles for DMD.

We found 37 genes showing >3-fold misregulation shared by both JDM and DMD (Table III). These genes reflect myofiber degeneration and regeneration processes and might indicate gene clusters responsive to functional ischemia, as DMD is known for loss of microvasculature and failed responses to vascular perfusion signals (22). The two largest groups of down-regulated genes in JDM (Table III) were mitochondrial genes involved in energy metabolism (many shared with DMD) and ribosomal proteins (specific to JDM); the reduction in ribosomal proteins may also reflect an antiviral state.

We confirmed four highly differentially regulated genes (IFN-induced 6-16 peptide, cyclin-dependent kinase (CDK) inhibitor p21, thrombospondin 4, and IFN-γ) by QRT-PCR and/or immunolocalization in tissue sections. The 6-16 peptide is an ISRE-containing gene, which was undetected in normal and DMD controls but was highly expressed in all JDM profiles (∼80-fold increase) (Fig. 2). QRT-PCR of individual biopsies showed a 12-cycle difference between control and JDM muscle, indicating a very large difference in RNA levels (>100-fold) (Fig. 3,A). Analysis of peripheral blood RNA by the same method showed elevated levels of 6-16 gene RNA in JDM blood compared with controls; however, 6-16 RNA was undetected in control blood, making fold-change assessments impossible (Fig. 3 B).

FIGURE 2.

Highly redundant expression profile data using Affymetrix GeneChips. Shown are raw data from probe sets showing up-regulation of ISRE of human 6-16 gene in muscle biopsies of TNF-α-GA and TNF-α-GG JDM patients. Twenty probe pairs (40 oligonucleotide features) corresponding to the 6-16 gene on the GeneChip HuGeneFL array are shown. The top feature is a perfect match (PM) 25-mer, paired with a lower mismatch (MM) probe that serves as a control for cross-hybridization. The PM probes in both control 1 and control 2 do not show a significant increase in hybridization relative to the MM control probes. This gene was evaluated as absent by the analysis software. All JDM profiles show a marked increase in the PM probe hybridization relative to MM controls, indicating greatly increased expression of this gene. Relative quantitation of averaged features is provided, as well as the average difference call (i.e., increase or decrease) when comparing JDM TNF-α-GA or JDM TNF-α-GG with the control 1 and 2 dataset. This analysis shows an ∼90- and ∼65-fold induction of 6-16 mRNA in JDM. The color bar indicates intensity range of 500–20,000.

FIGURE 2.

Highly redundant expression profile data using Affymetrix GeneChips. Shown are raw data from probe sets showing up-regulation of ISRE of human 6-16 gene in muscle biopsies of TNF-α-GA and TNF-α-GG JDM patients. Twenty probe pairs (40 oligonucleotide features) corresponding to the 6-16 gene on the GeneChip HuGeneFL array are shown. The top feature is a perfect match (PM) 25-mer, paired with a lower mismatch (MM) probe that serves as a control for cross-hybridization. The PM probes in both control 1 and control 2 do not show a significant increase in hybridization relative to the MM control probes. This gene was evaluated as absent by the analysis software. All JDM profiles show a marked increase in the PM probe hybridization relative to MM controls, indicating greatly increased expression of this gene. Relative quantitation of averaged features is provided, as well as the average difference call (i.e., increase or decrease) when comparing JDM TNF-α-GA or JDM TNF-α-GG with the control 1 and 2 dataset. This analysis shows an ∼90- and ∼65-fold induction of 6-16 mRNA in JDM. The color bar indicates intensity range of 500–20,000.

Close modal
FIGURE 3.

Confirmation of gene expression changes by QRT-PCR and immunolocalization. A, The confirmation of IFN-inducible gene 6-16 by QRT-PCR, in muscle biopsy (A) and in patient peripheral blood (B). The mRNA corresponding to 6-16 was compared with an internal β-actin mRNA control. The fluorescence intensity (ΔRn) on the y-axis is normalized, reporter signal corrected for initial reporter signal, and the x-axis represents the number of PCR cycles. A baseline of 0.02 ΔRn was set for this experiment. The PCR cycle at which the sample Rn crosses the baseline (CT) indicates a higher initial mRNA copy number. JDM muscle had a >1000-fold increase in 6-16 expression relative to normal muscle (A). In peripheral blood, 6-16 expression was undetectable by this assay but was detectable in JDM patient blood. C, Immunostaining for CDK inhibitor p21 in JDM muscle, in vascular smooth muscle (left panel), and in regions showing adjacent inflammatory changes (right panel). Lower-level immunostaining was also seen in some interstitial cells in the perimysial connective tissue. D, Immunostaining for IFN-γ in JDM muscle. Although mRNA corresponding to IFN-γ was not detectable by expression profiling or QRT-PCR, increased protein expression is seen at a relatively high level in vascular smooth muscle arterioles, capillaries, and endothelium (left panel) and macrophages (right panel). Normal muscle did not display detectable IFN-γ protein (data not shown).

FIGURE 3.

Confirmation of gene expression changes by QRT-PCR and immunolocalization. A, The confirmation of IFN-inducible gene 6-16 by QRT-PCR, in muscle biopsy (A) and in patient peripheral blood (B). The mRNA corresponding to 6-16 was compared with an internal β-actin mRNA control. The fluorescence intensity (ΔRn) on the y-axis is normalized, reporter signal corrected for initial reporter signal, and the x-axis represents the number of PCR cycles. A baseline of 0.02 ΔRn was set for this experiment. The PCR cycle at which the sample Rn crosses the baseline (CT) indicates a higher initial mRNA copy number. JDM muscle had a >1000-fold increase in 6-16 expression relative to normal muscle (A). In peripheral blood, 6-16 expression was undetectable by this assay but was detectable in JDM patient blood. C, Immunostaining for CDK inhibitor p21 in JDM muscle, in vascular smooth muscle (left panel), and in regions showing adjacent inflammatory changes (right panel). Lower-level immunostaining was also seen in some interstitial cells in the perimysial connective tissue. D, Immunostaining for IFN-γ in JDM muscle. Although mRNA corresponding to IFN-γ was not detectable by expression profiling or QRT-PCR, increased protein expression is seen at a relatively high level in vascular smooth muscle arterioles, capillaries, and endothelium (left panel) and macrophages (right panel). Normal muscle did not display detectable IFN-γ protein (data not shown).

Close modal

CDK inhibitor p21 is a negative regulator of mitosis. This protein was highly up-regulated in both JDM (32-fold) and DMD (16-fold) and could be involved in degeneration/regeneration pathways or as a response to functional ischemia. We used mAbs to immunolocalize this protein in JDM muscle biopsies (Fig. 3,C). p21 was clearly localized to vascular smooth muscle in areas of the biopsy showing inflammation (Fig. 3 C, right panel), although it was undetected in normal muscle (data not shown). Lower level expression of p21 was also seen in the JDM biopsies, localized to inflammatory cells in connective tissue and degenerating myofibers. p21 expression in vascular smooth muscle might be a response to ischemia-induced pathways, and additional experiments must be done to validate this possibility.

Thrombospondin-4 gene up-regulation was common to both JDM (8-fold) and DMD (20-fold), and expression changes were verified by immunostaining of muscle biopsies. Thrombospondin 4 is an extracellular matrix calcium-binding protein particularly abundant in tendon and early osteogenic tissues. We recently showed thrombospondin-4 immunolocalization to regions of myofiber degeneration in DMD muscle (22). We found an identical pattern of immunostaining in JDM muscle; thrombospondin-4 was localized to interstitial cells in relatively large regions surrounding macrophage-infiltrated myofibers (data not shown).

We did not find a significant increase in the expression of IFN-γ mRNA by expression profiling, presumably because of the low level or unstable nature of the message, which may be facilitated by the high expression of 2′,5′-oligo(A) synthetase. However, the many IFN-regulated genes in JDM profiles suggested that IFN-γ protein affects JDM muscle function. We immunostained the JDM muscle for IFN-γ and found high level expression in vascular smooth muscle, macrophages, and some capillaries (Fig. 3 D). IFN-γ protein was not found in normal control muscle. These data suggest that there is an increase in IFN-γ protein that is consistent with the increased expression of many IFN-γ-induced genes detected by expression profiling in the muscle from children with JDM, but not the controls.

There is evidence of genetic predisposition for JDM disease susceptibility, class II Ag DQA1*0501 (7), in linkage disequilibrium with DQA1*0301 (25), as well as for JDM disease chronicity, and the TNF-α-308 A allele (19). We performed expression profiling of 5600 genes by using highly redundant (40 probes/gene; ∼240,000 features) Affymetrix GeneChips to determine the pathophysiologic pathways that contribute to the muscle pathology. Our method involved the careful clinical characterization and genotyping of four JDM children matched for DQA1*0501 genotype, but with different TNF-α alleles. We then compared JDM expression profiles with previously characterized populations of normal age-matched controls, children with DMD, and profiles reflective of the intracellular antiviral response (NF90 transfections). We found that JDM muscle displayed a complex interplay of at least three cell-specific pathologic cascades: IFN-induced (antiviral/antimicrobial?) cascades in the vasculature and infiltrating immune cells resulting in a vasculopathy in which there is also activation of coagulation (26), ischemic (angiogenic) cascades in smooth muscle and myofibers induced by coagulopathy, and degeneration/regeneration cascades expressed by necrotic myofibers. We hypothesize that the disease process in JDM may develop as a result of positive and negative synergism between those cascades, withcritical molecules, such as TNF-α, that can mediate between cascades, thereby resulting in deleterious cross talk between cascades.

We found high expression of many IFN-induced genes in muscle biopsies from children with untreated JDM, in contrast to biopsies from normal and DMD controls, which further supports the hypothesis that the pathogenesis of this disease is a response to an infectious agent. Transcription of IFN-inducible genes is considered a hallmark of the host defense mechanism against infection by a range of organisms (27, 28), consistent with most changes being shared with an in vitro model of the intracellular response (NF90 transfection; Table II). The profiles suggested that both IFN-αβ and IFN-γ might be active in the JDM muscle. The profile for a protein often used as a marker for viral infection, the IFN-αβ-induced GTPase myxovirus resistance protein p78 (29), had the highest expression level (96-fold increase) in JDM muscle compared with controls. High quantities of myxovirus resistance protein p78 were detected in acute viral skin lesions and in lupus (30), although no active virus has been identified in affected regions of tissue for lupus or JDM (8, 30, 31). We also found up-regulated expression of other IFN markers, most of which were also up-regulated in the NF90 antiviral model (Table II), including 2′,5′-oligoadenylate synthetase, p27, 6-16 gene, latent cytosolic transcription factor, IFN-γ-inducible protein (IP)-10, RIG-G, IFP35, ISG-54, RNA-dependent protein kinase, ISG-56, cellular resistance protein MxB, LMP2, and other MHC class I- and II-encoded proteins. Some of these genes (e.g., 6-16 and ISG-54) contain similar ISRE elements and are regulated by the Janus tyrosine kinase-STAT pathway (Fig. 4). A previous study found high STAT1 expression in perifascicular atrophic muscle fibers in dermatomyositis biopsies (32). We hypothesize that the Janus tyrosine kinase-STAT pathway contributes to the persistence of the IFN-induced response in muscle tissue from JDM patients.

The expression profiles do not distinguish between an active infection in the biopsy and a hit-and-run mechanism in which an infection initiates a cascade that persists after the active organism has been cleared. Genes involved in Ag presentation may hold clues to this issue, because the majority of JDM patients have consistent evidence of a very limited spectrum of autoantibodies (11, 12, 13). Genes controlling the early phase of Ag presentation were up-regulated in JDM; they include MHC class II proteosome elements LMP2 (6-fold higher) and LMP7 (∼42-fold higher) and Ag transporter TAP1 (also called RING4, up ∼37-fold), which presents Ags to CD8+ T cells (common in JDM muscle). TAP activity is dependent on continuing protein translation (33), suggesting that Ags presented to MHC are mainly peptides from newly synthesized proteins. Data showing that there is diminished class II-associated peptide Ii binding to the DQA1*0501/DQA1*0301 molecule suggest possible access to novel peptides early in the processing pathway, which might induce autoimmunity (34). Virally infected cells typically shut down endogenous protein translation (consistent with the down-regulation of many ribosomal genes; Table III); this response ensures presentation of viral Ags. However, we hypothesize that the promotion of new translation via angiogenic and myofiber regeneration cascades may result in the presentation of proteins involved in normal development and maturation of these cell types instead of viral proteins. Thus, feedback between cascades may explain why a limited spectrum of circulating autoantibodies as well as circulating antimicrobial titers (1, 35) are detected in serum from children with JDM.

T cells are common in JDM muscle, but B cells are not. It is not clear whether the relative decrease in B cells in the JDM is a consequence of the B cell TNF-related apoptosis-inducing ligand, which was 4-fold higher in muscle of children with JDM (Table II and Fig. 4). The up-regulation of MHC genes and TNF-related apoptosis-inducing ligand in JDM muscle profiles suggests high Ag processing, proteolysis, and NK cell-mediated degradation, but inhibited B cell maturation.

Antiviral cascades typically induce increased endothelial adhesion and arrest growth of vascular endothelium and smooth muscle, with capillary drop out. We found evidence of both processes in JDM muscle profiles. Platelet (soluble N-ethylmaleimide-sensitive factor attachment protein receptor), S-nitrosoacetylpenicillamine receptor (up-regulated 13-fold in JDM) interacts with the cytoskeleton and provides specificity for vesicle docking and membrane fusion (36), and may contribute to vascular adhesion.

There is evidence that IFN-γ also plays a role in the pathophysiology of JDM, for both IP-30 (up-regulated 27-fold) and the IFN-γ-inducible early response gene IP-10 (up-regulated 28-fold) are increased. IP-10 is a T lymphocyte-specific CXC that shows homology to platelet factors and promotes adhesion and inhibits neovascularization (37). The IFN-γ-induced expression of IP-10 in endothelial cells, macrophages, and smooth muscle cells is potentiated by TNF-α and suppressed by NO (38). High expression of IP-10 in keratinocytes is seen in psoriasis and dermatitis (39), suggesting that IP-10 might contribute to the skin component of JDM.

We also found increased expression of potential negative regulators of the cell cycle in JDM muscle biopsy profiles, namely, proteins p27 and p21, and heat shock protein 70, cyclin H assembly factor, and cyclin G1 (Tables II and III). We localized expression of p21 to vascular smooth muscle in JDM biopsies (Fig. 3 C). p21 and p27 are G1-phase CDK inhibitors and may prevent entry of vascular smooth muscle into the S phase in children with active JDM. Another role of p21 is that it may provide protection against apoptosis (40). The regulation of the cell cycle is complex, and each up-regulated regulatory protein must be colocalized to specific cell types before a clear model of stimulus and suppression can be drawn.

We hypothesize a vicious circle in which both IFN-αβ and IFN-γ cascades lead to functional ischemia of muscle, which then exacerbates the immune response cascades (Fig. 4). Expression profiles of ischemic muscle are not available for comparison, but TNF-α protein levels and NO production are much higher in ischemic muscle (41, 42). We speculate that TNF-α and NO from ischemic muscle interact with the immune response cascade in endothelium and infiltrating T and NK cells, which exacerbates the IFN-induced process.

Localized, grouped myofiber necrosis is characteristic of affected regions of JDM muscle. The necrosis is most evident in areas of inflammation, often localized to a perifascicular pattern suggestive of vascular (ischemic) injury. Because myofibers can regenerate, we expected to find expression profiles characteristic of myofiber degeneration/regeneration in JDM muscle. We found many parallels between the profiles of JDM and DMD (22) (Tables II and III). Muscle structure and cell surface and extracellular cytoskeleton up-regulated genes found in both JDM and DMD, including embryonic myosin heavy chain (52.5-fold), myosin-binding protein H (71.3-fold), skeletal muscle perinatal myosin heavy chain MYH8 (13.9-fold), nicotinic acetylcholine receptor α subunit (12.2-fold), thrombospondin-4 (8.2-fold), prepro-α2 (I) collagen (7.4-fold), and chondroitin sulfate proteoglycan versican (∼5-fold). TNF-α release is a feature of necrotic muscle, and TNF-α itself damages skeletal muscle (43).

Perifascicular atrophy is a specific pathologic feature of JDM muscle (20), suggesting that regeneration is inhibited or blocked in localized, perhaps more ischemic, regions of the muscle. We hypothesize that IFN-induced response cascades, which inhibit mitosis and protein synthesis cascades, thereby inhibit regeneration of necrotic myofibers. TNF-α is an important protein shared among the ischemic, necrotic, and antiviral cascades; this protein may be a critical interpathway communicating protein. TNF-α and IFN-γ are synergistic in the induction of myofiber injury (44), and the association of disease chronicity with TNF-α gene polymorphisms supports our model (Fig. 4).

We used gene expression profiling to document an IFN-mediated immune response in the muscle from untreated JDM children who carry the DQA1*0501 allele. We suggest that the pathophysiology of JDM may result from aberrant cross talk between different cell response pathways (Fig. 4). Our data indicate that functional ischemia leads to degeneration and regeneration of myofibers and angiogenesis, both of which normally require mitosis and new protein synthesis in myogenic and vascular cells. However, IFN-induced cascades suppress those pathways by inhibition of mitosis and protein synthesis via both IFN-αβ- and IFN-γ-induced gene clusters. Concurrently, TNF-α and other signaling molecules involved in ischemic responses and degeneration of myofibers feed back on the IFN-induced cascade, perhaps promoting these signals in endothelium and infiltrating T and NK cells. In JDM, increased TNF-α production is associated with the TNF-α-308A allele and disease chronicity. This model may provide a working hypothesis for defining the association between genetic markers and disease pathophysiology. The genetic association of HLA-DQA1*0501 with disease susceptibility may be related to the strength and type of the child’s response to the original infectious agent, whereas the genetic association of TNF-α (and related) polymorphisms and disease persistence might be associated with deleterious cross talk between cellular response cascades (Fig. 4).

Our model also predicts the effect of conflicting cascades on Ag processing. We hypothesize that intracellular Ag-processing mechanisms induced by the IFN-induced cascade process newly synthesized proteins and present them to the cell surface and infiltrating immune cells. However, instead of presenting only new proteins of foreign origin, the cells also present newly synthesized proteins from myofiber regeneration and angiogenesis cascades. From this model, we predict that most autoantigens represent proteins induced during regenerative or compensatory profiles; thus, the autoantigens may be a consequence of a bystander effect from conflicting expression profiles without a direct relationship to the instigating agent.

Table 3A.

Continued

Accession No.JDMDMD
U65579 −3.2 −4.8 Mitochondrial NADH dehydrogenase-ubiquinone Fe-S protein 8 
D10511 −3.3 NC Gene for mitochondrial acetoacetyl-CoA thiolase 
Y09267 −3.3 NC Flavin-containing monooxygenase 2/gb=Y09267/ntype=RNA 
U05861 −3.1 −3.1 Hepatic dihydrodiol dehydrogenase gene 
Others and unknown    
U16954_at ∼45.3 NC AF1q 
U51010t ∼43.3 22.0 Nicotinamide N-methyltransferase 
D28137 11.3 NC BST-2 
J00123 11.2 NC Enkephalin gene 
D38583 10.3 NC Calgizzarin 
J03242 8.7 NC Insulin-like growth factor II 
U08021 8.5 20.4 Nicotinamide N-methyltransferase (NNMT) 
U61374 8.3 7.0 Sushi-repeat-containing protein (SRPX) 
U88964 6.7 NC HEM45 
U32849 6.6 NC Hou 
U90552 5.7 NC Butyrophilin (BTF5) 
X15525 5.6 NC Lysosomal acid phosphatase gene (EC 3.1.3.2) 
M31013 5.5 NC Nonmuscle myosin H chain (NMHC) 
Y00264 5.4 NC Amyloid A4 precursor of Alzheimer disease 
X55740 ∼5.2 NC Placental cDNA coding for 5′ nucleotidase (EC 3.1.3.5) 
D28589 −35.0 NC KIAA00167 
D25304 ∼−22.5 NC KIAA0006 
M25079 −14.5 NC Sickle cell β-globin 
U79253 −7.2 −7.0 Clone 23893 
HG1428- −6.7 NC Globin, β 
X97324 −6.6 NC AdipophilinA 
L18972 −5.7 NC Anonymous gene 
Accession No.JDMDMD
U65579 −3.2 −4.8 Mitochondrial NADH dehydrogenase-ubiquinone Fe-S protein 8 
D10511 −3.3 NC Gene for mitochondrial acetoacetyl-CoA thiolase 
Y09267 −3.3 NC Flavin-containing monooxygenase 2/gb=Y09267/ntype=RNA 
U05861 −3.1 −3.1 Hepatic dihydrodiol dehydrogenase gene 
Others and unknown    
U16954_at ∼45.3 NC AF1q 
U51010t ∼43.3 22.0 Nicotinamide N-methyltransferase 
D28137 11.3 NC BST-2 
J00123 11.2 NC Enkephalin gene 
D38583 10.3 NC Calgizzarin 
J03242 8.7 NC Insulin-like growth factor II 
U08021 8.5 20.4 Nicotinamide N-methyltransferase (NNMT) 
U61374 8.3 7.0 Sushi-repeat-containing protein (SRPX) 
U88964 6.7 NC HEM45 
U32849 6.6 NC Hou 
U90552 5.7 NC Butyrophilin (BTF5) 
X15525 5.6 NC Lysosomal acid phosphatase gene (EC 3.1.3.2) 
M31013 5.5 NC Nonmuscle myosin H chain (NMHC) 
Y00264 5.4 NC Amyloid A4 precursor of Alzheimer disease 
X55740 ∼5.2 NC Placental cDNA coding for 5′ nucleotidase (EC 3.1.3.5) 
D28589 −35.0 NC KIAA00167 
D25304 ∼−22.5 NC KIAA0006 
M25079 −14.5 NC Sickle cell β-globin 
U79253 −7.2 −7.0 Clone 23893 
HG1428- −6.7 NC Globin, β 
X97324 −6.6 NC AdipophilinA 
L18972 −5.7 NC Anonymous gene 

We are indebted to the children and families who participated in this study. We thank Dr. Jack Lawler for providing thrombospondin 4 Abs. We also thank Dr. Kanneboyina Nagaraju for his thoughtful comments.

1

This work was supported by National Institutes of Health Grants RO1 AR43978 and P60 AR30692 (to L.M.P.) and 3RO1 NS29525-09 (to E.P.H.).

3

Abbreviations used in this paper: JDM, juvenile dermatomyositis; CDK, cyclin-dependent kinase; DMD, Duchenne muscular dystrophy; IP, IFN-γ-inducible protein; ISRE, IFN-stimulatable response element; QRT-PCR, quantitative real-time PCR.

1
Pachman, L. M., J. G. Hayford, M. C. Hochberg, M. A. Pallansch, A. Chung, C. D. Daugherty, B. H. Athreya, S. L. Bowyer, C. A. Fink, H. L. Gewanter, et al
1997
. New-onset juvenile dermatomyositis: comparisons with a healthy cohort and children with juvenile rheumatoid arthritis.
Arthritis Rheum.
40
:
1526
2
Pachman, L. M., J. R. Hayford, A. Chung, C. A. Daugherty, M. A. Pallansch, C. W. Fink, H. L. Gewanter, R. Jerath, B. A. Lang, J. Sinacore, et al
1998
. Juvenile dermatomyositis at diagnosis: clinical characteristics of 79 children.
J. Rheumatol.
25
:
1198
3
Pachman, L. M., E. Mendez, H. Lou, M. Pallansch, R. Lipton, R. Ramsey-Goldman, A. Chung, A. R. Dyer.
1999
. Parent report of antecedent illness and environmental factors before onset of juvenile dermatomyositis (JDM) children: NIAMS JDM research registry data.
Arthritis Rheum.
42
:
395
4
Reed, A. M., L. M. Pachman, C. Ober.
1991
. Molecular genetic studies of major histocompatibility complex genes in children with juvenile dermatomyositis: increased risk associated with HLA-DQA1*0501.
Hum. Immunol.
32
:
235
5
Friedman, J. M., L. M. Pachman, M. C. Maryjowski, R. M. Radvany, W. E. Crowe, V. Hanson, J. E. Levinson, C. H. Spencer.
1983
. Immunogenetic studies of juvenile dermatomyositis. I. HLA-DR antigen frequencies.
Arthritis Rheum.
26
:
214
6
Pachman, L. M., O. Jonasson, R. A. Cannon, L. M. Pachman.
1977
. Increased frequency of HLA-B8 in patients with juvenile dermatomyositis.
Lancet
2
:
567
7
Reed, A. M., J. D. Stirling.
1995
. Association of the HLA-DQA1*0501 allele in multiple racial groups with juvenile dermatomyositis.
Hum. Immunol.
44
:
131
8
Pachman, L. M., D. L. Litt, A. H. Rowley, J. R. Hayford, J. Caliendo, S. Heller, B. S. Ticho, M. Christensen, B. Patterson, S. R. Ytterberg, M. Pallansch.
1995
. Lack of detection of enteroviral RNA or bacterial DNA in magnetic resonance imaging-directed muscle biopsies from twenty children with active untreated juvenile dermatomyositis.
Arthritis Rheum.
38
:
1513
9
De Visser, M., A. M. Emslie-Smith, A. G. Engel.
1989
. Early ultrastructural alterations in adult dermatomyositis: capillary abnormalities precede other structural changes in muscle.
J. Neurol. Sci.
94
:
181
10
Kissel, J. T., R. K. Halterman, K. W. Rammohan, J. R. Mendell.
1991
. The relationship of complement-mediated microvasculopathy to the histologic features and clinical duration of disease in dermatomyositis.
Arch. Neurol.
48
:
26
11
Pachman, L. M., J. M. Friedman, M. C. Maryjowski, O. Jonasson, R. M. Radvany, G. C. Sharp, M. A. Cobb, N. D. Battles, W. E. Crowe, C. W. Fink, et al
1985
. Immunogenetic studies in juvenile dermatomyositis. III. Study of antibody to organ-specific and nuclear antigens.
Arthritis Rheum.
28
:
151
12
Cambridge, G., E. Ovadia, D. A. Isenberg, V. Dubowitz, J. Sperling, R. Sperling.
1994
. Juvenile dermatomyositis: serial studies of circulating autoantibodies to a 56 kD nuclear protein.
Clin. Exp. Rheumatol.
12
:
451
13
Lutz, J. L., T. O. Fedczyna, D. Isenberg, R. Sperling, L. M. Pachman.
2001
. The association of the DQA1*0501 allele and antibody to a 56 KD nuclear protein in sera from patients with juvenile dermatomyositis (JDM).
Arthritis Rheum.
44
:
S293
14
Miller, M. L., R. Lantner, L. M. Pachman.
1983
. Natural and antibody-dependent cellular cytotoxicity in children with systemic lupus erythematosus and juvenile dermatomyositis.
J. Rheumatol.
10
:
640
15
O’Gorman, M. R. G., L. Bianchi, D. Zass, V. Corrochano, L. M. Pachman.
2000
. Decreased CD54 (ICAM-1) positive non-CD19+ lymphocytes in the peripheral blood of untreated children with active symptoms of juvenile dermatomyositis.
Clin. Diagn. Lab. Immunol.
7
:
693
16
Pachman, L. M., M. R. G. O’Gorman, T. Lawton, M. Liotta, R. M. Pope, M. Greene, T. T. Wu.
1998
. Evidence of a TCR Vβ8 motif and increased CD56+ NK cells in muscle biopsies (Mbx) from DQA1*0501 positive untreated children with juvenile dermatomyositis (JDM) very early in their disease course.
Pediatr. Res.
43
:
338
17
Biron, C. A., K. B. Nguyen, G. C. Pien, L. P. Cousens, T. P. Salazar-Mather.
1999
. Natural killer cells in antiviral defense: function and regulation by innate cytokines.
Annu. Rev. Immunol.
17
:
189
18
Reed, A. M., S. Cragoe, M. Strausbach, L. M. Pachman, P. Wettstein.
2001
. Skewed T-cell receptor repertoire in children with juvenile dermatomyositis.
Arthritis Rheum.
44
:
S292
19
Pachman, L. M., M. R. Liotta-Davis, K. D. Hong, T. R. Kinsella, E. P. Mendez, J. M. Kinder, E. H. Chen.
2000
. TNFα-308A allele in juvenile dermatomyositis: association with increased production of tumor necrosis factor α, disease duration, and pathologic calcifications.
Arthritis Rheum.
43
:
2368
20
Bohan, A. P., J. B. Peter.
1975
. Polymyositis and dermatomyositis (parts 1 and 2).
N. Engl. J. Med.
292
:
344
21
Fedczyna, T. O., J. L. Lutz, L. M. Pachman.
2001
. Expression of TNFα by muscle fibers in biopsies from children with untreated juvenile dermatomyositis (JDM): association with the TNF-α 308A allele.
Clin. Immunol.
100
:
236
22
Chen, Y. W., P. Zhao, R. Borup, E. P. Hoffman.
2000
. Expression profiling in the muscular dystrophies: identification of novel aspects of molecular pathophysiology.
J. Cell Biol.
151
:
1321
23
Cecilia, D., V. N. KewalRamani, J. O’Leary, B. Volsky, P. Nyambi, S. Burda, S. Xu, D. R. Littman, S. Zolla-Pazner.
1998
. Neutralization profiles of primary human immunodeficiency virus type I isolates in the context of coreceptor usage.
J. Virol.
72
:
6988
24
Der, S. D., A. Zhou, B. R. G. Williams, R. H. Silverman.
1998
. Identification of genes differentially regulated by interferon α, β, and γ using oligonucleotide arrays.
Proc. Natl. Acad. Sci. USA
95
:
15623
25
Rider, L. G., R. C. Gurley, J. P. Pandey, I. Garcia de la Torre, A. E. Kalovidouris, T. P. O’Hanlon, L. A. Love, R. C. Hennekam, L. L. Baumbach, H. E. Neville, et al
1998
. Clinical, serologic, and immunogenetic features of familial idiopathic inflammatory myopathy.
Arthritis Rheum.
41
:
710
26
Scott, J. P., C. Arroyave.
1987
. Activation of complement and coagulation in juvenile dermatomyositis.
Arthritis Rheum.
30
:
572
27
Shtrichman, R., C. E. Samuel.
2001
. Role of γ interferon in antimicrobial immunity.
Curr. Opin. Microbiol.
4
:
251
28
Bogdan, C..
2000
. The function of type 1 interferons in antimicrobial immunity.
Curr. Opin. Immunol.
12
:
419
29
Chieux, V., W. Chehadeh, J. Harvey, O. Haller, P. Wattré, D. Hober.
2001
. Inhibition of Coxsackievirus B4 replication in stably transfected cells expressing human MxA protein.
Virology
283
:
84
30
Bengtsson, A. A., G. Sturfelt, L. Truedsson, J. Blomberg, G. Alm, H. Vallin, L. Ronnblom.
2000
. Activation of type I interferon system in systemic lupus erythematosus correlates with disease activity but not with antiretroviral antibodies.
Lupus
9
:
664
31
Denman, A. M..
2000
. Systemic lupus erythematosus: is a viral aetiology a credible hypothesis?.
J. Infect.
40
:
229
32
Illa, I., E. Gallardo, R. Gimeno, C. Serrano, I. Ferrer, C. Juarez, F. Iannone.
1997
. Signal transducer and activator of transcription 1 in human muscle: implications in inflammatory myopathies.
Am. J. Pathol.
151
:
81
33
Reits, E. A., J. C. Vos, M. Gromme, J. Neefjes.
2000
. The major substrates for TAP in vivo are derived from newly synthesized proteins.
Nature
404
:
774
34
Reed, A. M., E. J. Collins, L. P. Shock, D. G. Klapper, J. A. Frelinger.
1997
. Diminished class II-associated Ii peptide binding to the juvenile dermatomyositis HLA DQα1*0501/DQβ1*0301 molecule.
J. Immunol.
159
:
6260
35
Christensen, M. L., L. M. Pachman, R. Schneiderman, D. C. Patel, J. M. Friedman.
1986
. Prevalence of Coxsackie B virus antibodies in patients with juvenile dermatomyositis.
Arthritis Rheum.
29
:
1365
36
Foster, L. J., A. Klip.
2000
. Mechanism and regulation of GLUT-4 vesicle fusion in muscle and fat cells.
Am. J. Physiol.
279
:
C877
37
Mach, F., A. Sauty, A. S. Iarossi, G. K. Sukhova, K. Neote, P. Libby, A. D. Luster.
1999
. Differential expression of three T lymphocyte-activating CXC chemokines by human atheroma-associated cells.
J. Clin. Invest.
104
:
1041
38
Grange, R., E. Isotani, K. Lau, K. Kamm, P. Huang, J. Stull.
2001
. Nitric oxide contributes to vascular smooth muscle relaxation in contracting fast-twitch muscles.
Physiol. Genomics
5
:
35
39
Smoller, B. R., M. S. McNutt, M. H. Gray, J. Krueger, A. Hsu, A. B. Gottlieb.
1990
. Detection of the interferon-γ-induced protein 10 in psoriasiform dermatitis of acquired immunodeficiency syndrome.
Arch. Dermatol.
126
:
1457
40
Vattemi, G., P. Tonin, M. Filosto, M. Spagnolo, N. Rizzuto, G. Tomelleri.
2000
. T-cell anti-apoptotic mechanisms in inflammatory myopathies.
J. Neuroimmunol.
111
:
146
41
Sternbergh, W. C., III, T. M. Tuttle, R. G. Makhoul, H. D. Bear, M. Sobel, A. A. Fowler, III..
1994
. Postischemic extremities exhibit immediate release of tumor necrosis factor.
J. Vasc. Surg.
20
:
474
42
Papanastasiou, S., S. E. Estdale, S. Homer-Vanniasinkam, R. T. Mathie.
1999
. Protective effect of preconditioning and adenosine pretreatment in experimental skeletal muscle reperfusion injury.
Br. J. Surg.
86
:
916
43
Goodman, M. N..
1991
. Tumor necrosis factor induces skeletal muscle protein breakdown in rats.
Am. J. Physiol.
260
:
E727
44
Guttridge, D. C., M. W. Mayo, L. V. Madrid, C. Y. Wang, A. S. Baldwin, Jr.
2000
. NF-κB-induced loss of MyoD messenger RNA: possible role in muscle decay and cachexia.
Science
289
:
363