Self epitopes shared between human skeletal myosin and Streptococcus pyogenes M5 protein are targets of immune responses in active juvenile dermatomyositis.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 46, No. 11, November 2002, pp 3015–3025 DOI 10.1002/art.10566 © 2002, American College of Rheumatology Self Epitopes Shared Between Human Skeletal Myosin and Streptococcus pyogenes M5 Protein Are Targets of Immune Responses in Active Juvenile Dermatomyositis Margherita Massa,1 Nick Costouros,2 Federica Mazzoli,1 Fabrizio De Benedetti,1 Antonio La Cava,2 Tho Le,2 Isme de Kleer,2 Angelo Ravelli,3 Margaret Liotta,4 Sarah Roord,2 Charles Berry,2 Lauren M. Pachman,4 Alberto Martini,3 and Salvatore Albani2 Objective. To identify self T cell epitopes associated with proinflammatory immune responses and clinically active juvenile dermatomyositis (juvenile DM). The target of our search for relevant epitopes was represented by amino acid sequences shared between human skeletal myosin and Streptococcus pyogenes M5 protein. The long-term objective of the project is to identify suitable targets for immunotherapy of the disease. Methods. We used computerized algorithms to identify putative agretopes on both the human myosin and Streptococcus M5 proteins. Direct binding assays for homolog peptides were used to confirm such predictions. Antigenicity and functional cross-reactivity were evaluated by cytotoxicity assays and by measurement of cytokine levels. Specific T cells were isolated by T cell capture, and T cell receptor (TCR) V␤ gene usage was identified by reverse transcriptase–polymerase chain reaction. Results. We identified peptides that are targets of disease-specific cytotoxic T cell responses. T cell reac- tivity against the self peptides correlates with clinical signs of early, active myositis. Such reactivity is accompanied by production of proinflammatory cytokines, which may contribute to the damage. T cell crossrecognition of bacterial and human homologs was shown functionally as well as by sorting peptide-specific T cells and identifying oligoclonal and largely overlapping TCR V␤ gene usage. Conclusion. These findings represent the first identification of a self epitope in juvenile DM, providing a potential candidate for antigen-specific immune therapy. Juvenile dermatomyositis (juvenile DM) is an autoimmune systemic disease that involves primarily the skin and muscles (1). The etiopathogenesis is still unknown, but it is believed to be autoimmune in origin. In muscle biopsy samples, the presence of a perivascular mononuclear cell infiltrate, including CD4-positive and CD8-positive T lymphocytes and macrophages, has been documented (2,3). Peripheral blood lymphocytes from patients with inflammatory myopathies, including juvenile DM, have been shown to cause muscle-specific injury in cytotoxicity assays (4–6). Lymphocytes from patients with inflammatory myopathies incubated with autologous muscle were shown to produce a “lymphotoxin” that caused muscle necrosis and impaired protein synthesis (7). More recently, it was shown that upregulation of transgenic major histocompatibility complex class I antigens in the skeletal muscle leads to autoimmune myositis in transgenic mice (8). These observations support the hypothesis that recognition of self epitopes of the muscle triggers activation of T cells, which may induce chronic autoimmune damage. However, the potential muscle targets of such autoreactivity Supported in part by IRCCS Policlinico San Matteo, by grant 390RFM/94/01 from the Italian Ministry of Health, by grants 5P50AR-44850-04 and N01-AR-92241 from the NIH, and by a grant from the Ter Meulen Fund of the Royal Netherlands Academy of Arts and Sciences. 1 Margherita Massa, PhD, Federica Mazzoli, PhD, Fabrizio De Benedetti, MD, PhD: IRCCS Policlinico San Matteo, Pavia, Italy; 2 Nick Costouros, MD, Antonio La Cava, MD, PhD, Tho Le, MS, Isme de Kleer, MD, Sarah Roord, MD, Charles Berry, PhD, Salvatore Albani, MD, PhD: University of California, San Diego, La Jolla, California; 3Angelo Ravelli, MD, Alberto Martini, MD: IRCCS Gaslini, Universita’ di Genova, Genoa, Italy; 4Margaret Liotta, MS, Lauren M. Pachman, MD: Northwestern University, Chicago, Illinois. Address correspondence and reprint requests to Salvatore Albani, MD, PhD, Department of Pediatrics, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0663. E-mail: firstname.lastname@example.org. Submitted for publication January 23, 2002; accepted in revised form July 10, 2002. 3015 3016 MASSA ET AL and the mechanisms which trigger and perpetuate inflammation have not yet been identified. In juvenile DM, as well as in other autoimmune diseases, abnormal responses to microbial antigens sharing homology with putative targets of disease may be proposed as either direct or accessory etiopathogenic mechanisms (9–13). We have previously described a patient with juvenile DM characterized by a cyclic course. In this patient, recurrences were associated with evidence of recent streptococcal infections (14). Our observation and other reports suggesting a possible relationship between streptococcal infection and juvenile DM (15,16) led us to hypothesize that recognition of sequences of streptococcal origin might trigger an autoimmune response to a muscle antigen. We have screened a protein database for microbial proteins and peptides that would share with human skeletal myosin amino acid sequences containing putative anchor residues to the most represented HLA class I alleles. This analysis has enabled us to identify sequences of homology between human skeletal myosin heavy chain and streptococcal M5 protein (14). These sequences of homology are different from those between streptococcal M5 protein and human cardiac myosin, which are relevant in the pathogenesis of acute rheumatic fever (ARF) (17–23). In the present study, we identified epitopes derived from homologous sequences shared between human skeletal myosin and Streptococcus M5 protein as targets of cytotoxic responses in patients with early, active juvenile DM. These peptides were also efficient in stimulating T cells from patients with active juvenile DM, but not those from controls, to produce tumor necrosis factor ␣ (TNF␣), thus suggesting a functionally relevant role for these responses in the pathogenic process. Immune reactivity was strictly epitope- and disease-specific and was significantly elevated in patients with active juvenile DM. T cell recognition of bacterial and human homologs was shown functionally as well as by sorting peptide-specific T cells and identifying T cell receptor (TCR) V␤ gene usage. These results suggest that abnormal responses to epitopes shared between a target of the autoimmune process and an infectious agent may be involved in the pathogenesis of juvenile DM by direct cytotoxicity to muscle fibers and by triggering proinflammatory pathways that lead to a cascade of events whose final outcome is tissue damage. PATIENTS AND METHODS Patients. Sixteen patients with juvenile DM (mean ⫾ SD age 9.5 ⫾ 4.3 years) were included in the study. All met the Table 1. Disease duration, presence of active myositis, and treatment at time of sampling in patients with juvenile dermatomyositis* Patient Disease duration, months Active myositis Treatment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 1 1 1 9 1 15 16 16 23 36 48 56 84 120 144 Yes Yes Yes Yes Yes No Yes No Yes No Yes No No No No No None None None None Pred. Pred. None None None Pred. CSA CSA, pred. Pred. MTX, pred None Pred. * All patients were diagnosed as having juvenile dermatomyositis according to the criteria of Bohan and Peter (24). Pred. ⫽ prednisone; CSA ⫽ cyclosporin A; MTX ⫽ methotrexate. Bohan and Peter criteria for the diagnosis of juvenile DM (24). Eight patients were admitted to the Children’s Memorial Medical Center, Northwestern University (Chicago, IL), and 8 were admitted to the University Hospital of Pavia (Pavia, Italy). Disease duration at the time of sampling was defined as the time from the first disease-related symptom (skin rash or muscle weakness). Active myositis at time of sampling was defined on the basis of decreased muscle strength, determined using a standard muscle strength score (25). All patients with normal muscle strength also had muscle enzyme levels that were less than twice the upper limit of the normal range. Disease duration, clinical evidence of muscle involvement, and current treatment at the time of sampling are summarized in Table 1. Ten healthy subjects served as controls (mean ⫾ SD age 19.2 ⫾ 7.0 years, due to ethical problems in recruiting age-matched children), and 12 children served as disease controls (mean ⫾ SD age 9.0 ⫾ 5.3 years). The disease controls consisted of 6 patients with juvenile idiopathic arthritis (JIA), 3 with Duchenne’s muscular dystrophy (DD), and 3 with poststreptococcal diseases (PSDs; 1 with poststreptococcal arthritis, 1 with poststreptococcal glomerulonephritis, and 1 with ARF), and all had active disease. Antistreptolysin O (ASO) titers in patients with poststreptococcal diseases ranged from 700 IU/ml to 1,200 IU/ml. Blood was obtained by venipuncture. Parental consent was obtained. Sample processing and generation of lymphoblastoid B cell lines. Peripheral blood mononuclear cells (PBMCs) were separated using the standard Ficoll-Hypaque technique, aliquotted, and frozen viable in liquid nitrogen. Samples were transported in a liquid nitrogen dry shipper (Taylor Wharton, Indianapolis, IN) from Chicago and San Diego to Pavia, where the cytotoxic assays were performed. Samples were also sent from Chicago and Pavia to San Diego, where intracellular cytokine staining and T cell capture (TCC) experiments were performed. Lymphoblastoid B–derived cell lines were obtained from all patients and controls using a standard Epstein- IMMUNE RESPONSE TARGETS IN ACTIVE JUVENILE DM 3017 Table 2. Sequences, putative MHC anchor residues, and actual HLA binding of the peptides derived from amino acid (aa) region 364–375 of the streptococcal M5 protein and of their homologs derived from human skeletal myosin heavy chain* Peptide code (aa region) Antigen Sequence Putative anchor residues HLA binding ratio, mean ⫾ SD† M5 (364–372) Myo (111–119) M5 (367–375) Myo (114–122) Streptococcal M5 Human myosin Streptococcal M5 Human myosin ALEKLNKEL EFQKMRRDL KLNKELEES KMRRDLEEA L365 and L372 F112 and L119 L368 and S375 M115 and A112 1.78 ⫾ 0.06 0.68 ⫾ 0.06 1.09 ⫾ 0.03 0.76 ⫾ 0.01 * The first two peptides and the second two peptides were paired, based on their sequence homology. HLA binding assays were performed from affinity-purified HLA class I molecules obtained from lymphoblastoid cell lines derived from 3 unrelated subjects with different major histocompatibility complex (MHC) genotypes. † Pan–HLA class I peptide HA-1 (GILGFVFTL) was used as a standard of known MHC affinity (equivalent to a value of 1). Binding of the tested peptides was expressed relative to this standard as a ratio. Barr virus (EBV) transformation protocol. Briefly, PBMCs were cultured with supernatant from the B958 cell line and 0.1 mg/ml of cyclosporin A (Sandoz, East Hanover, NJ) for ⬃14 days. These lines were used as targets for cytotoxic responses. Antigens. Synthetic peptides encompassing sequences shared between M5 and myosin proteins were purchased from Research Genetics (Huntsville, AL). Peptides were ⬎90% pure, as determined by high-performance liquid chromatography. Sequences are shown in Table 2. Identification of putative epitopes. The amino acid (aa) sequence of the human skeletal muscle myosin heavy chain was compared by computer analysis with protein sequences tabulated in the National Biomedical Research Foundation (NBRF) database using BLAST and FASTA software and default parameters (26). We found that stretches (aa 111–122) of human skeletal muscle myosin heavy chain were homologous to a significant extent with serotype M5 protein, one of the major antigens of group A Streptococcus pyogenes (aa 364–375, 228–235, and 380–391). Subsequently, prediction of HLA class I universal major binding motifs was performed using an algorithm kindly provided by Dr. Sette (Epimmune, San Diego, CA). Scores were confirmed utilizing a computerized prediction of peptide binding motifs to individual HLA alleles based on the BioInformatics and Molecular Analysis Section (BIMAS) site (online at www.bimas.dcrt.nih.gov). Sequence homologies were predominantly focused on common agretopic motifs rather than on absolute identity in sequence. Sequences are shown in Table 2. Using the same approach, we identified a peptide (KGLRRDLDA) of M5 protein that shares a 4–amino acid (RRDL) sequence homology with both human cardiac and skeletal myosin (27). HLA–peptide binding. HLA class I molecules, purified by immunoaffinity chromatography as previously described (28), were added to a solution of degassed phosphate buffered saline (PBS) containing biotinylated peptides of 9 amino acid residues at an HLA class I to biotinylated peptide ratio of 1:10. Fifty microliters of the solution was added to wells of a polystyrene/propylene 96-well plate (Fisher Scientific, Fair Lawn, NJ) and incubated for a minimum of 2 hours at room temperature. After incubation, anti–HLA class I antibody (PharMingen, San Diego, CA) was added at a 1:1 molar ratio of anti–HLA class I to HLA–peptide preparation, and the incubation was allowed to proceed for a minimum of 2 hours at room temperature or overnight at 4°C. A microdialyzer cassette (Cellulose Ester Membrane Frames for Microdialyzers; Fisher Scientific) with a 300-kd molecular weight cutoff membrane was then assembled following the manufacturer’s instructions. The preparation was transferred to the assembly, and the wells were sealed to prevent the evaporation of samples. Dialysis was performed for 24 hours with 3 degassed PBS changes to remove excess unbound antibodies and peptides. After dialysis, Neutra-avidin horseradish peroxidase (HRP) at a 1:1,000 dilution in PBS (streptavidin HRP; Sigma, St. Louis, MO) was added, followed by dialysis in the dark for 24 hours with 3 degassed PBS changes with a Spectra/Por 96-well microdialyzer (150-l volume, 0.028–square inch membrane surface area; Fisher Scientific). Once dialysis was complete, a 50-l test sample was transferred to a 96-well enzymelinked immunosorbent assay plate (Costar, Cambridge, MA) and developed with the appropriate substrate solution. For background level, the following conditions were also tested: HRP only, primary antibody with HRP, biotinylated peptide only with HRP, HLA class I without biotinylated peptide but with HRP, and PBS only. The assay was stopped with the addition of equal volumes of 1M H3PO4, and the result was read at 450–650 nm. Evaluation of cytotoxic T cell responses. PBMCs (2 ⫻ 106/ml) were incubated in 24-well plates and pulsed with 10 g/ml of peptide. Recombinant human interleukin-2 (IL-2) (40 units/ml; Hoffmann-La Roche, Nutley, NJ) was added on days 0 and 3, and the cytotoxic assay was performed on day 5. These cells were incubated for 4 hours with autologous EBVtransformed B cells and preincubated overnight with and without 50 g/ml of peptides at effector to target (E:T) cell ratios of 25:1, 10:1, and 1:1 in duplicate. Cytotoxic responses were evaluated by the lactate dehydrogenase (LDH) release assay (CytoTox 96 nonradioactive cytotoxicity assay; Promega, Madison, WI) (29). We found this assay to be at least as sensitive as the 51Cr assay, which was also performed in some instances, yielding identical results (not shown). LDH release was measured in an enzymatic assay according to the manufacturer’s instructions. The percentage release of target cells, as in a standard 51Cr release assay, was calculated as 100 ⫻ (experimental release ⫺ spontaneous release/maximum release ⫺ spontaneous release). The percentage of specific lysis was calculated by subtracting the percentage release obtained when effector cells were incubated with nonpulsed cells (always ⬍1%) from that obtained in the presence of pulsed target cells. Spontaneous release from target cells was consistently ⬍10%. 3018 Identification of peptide-specific T cells. The experiment was performed by TCC (30). The various steps of the procedure are described as follows. Preparation of HLA class I molecules and peptide loading. Human HLA class I molecules were purified from the lysate of EBV-transformed lymphoblastoid cell lines by immunoaffinity chromatography using anti–HLA class I antibodies produced by hybridoma W6/32 (American Type Culture Collection, Manassas, VA). Purified HLA class I molecules were solubilized in a Tris buffer containing 50 mM diethylamine and 2% ␤-octyl-glucopyranoside (Calbiochem, San Diego, CA). Preparation of peptides. Peptides M5 (aa 367–375) and Myo (aa 114–122), synthesized as C-terminal amides, were N-terminal biotinylated during synthesis (1 biotin/molecule; 100% biotinylation). Preparation of artificial antigen-presenting cells (aAPC). Phosphatidylcholine and cholesterol (Sigma) were combined in a glass tube at a molar ratio of 7 to 2, respectively. For labeled liposomes, N-(fluorescein-5-thiocarbamoyl) 1,2dihexadecanoyl-sn-glycero-3-phosphoethanolamine, triethylammonium salt (fluorescein–DHPE; Molecular Probes, Eugene, OR) was added at a final concentration of 1:1 molar ratio of fluorescein to phosphatidylcholine. The solvent was evaporated under an argon stream for 30 minutes and dispersed at a final concentration of 10 mg/ml in 140 M NaCl and 10 M Tris HCl, pH 8.0 (buffer A) containing 0.5% sodium deoxycholate. The solution was sonicated until clear and stored at ⫺20°C. HLA class I–peptide complexes were added to the lipids in buffer A with deoxycholate at an HLA class I to lipids ratio of 1:7 (weight:weight). Liposomes were formed through dialysis at 4°C against PBS in a 10K Slide-A-Lyzer (Pierce, Rockford, IL) for 48 hours. When biotinylated peptides were used, the formed aAPC were incubated with streptavidin– fluorescein isothiocyanate (PharMingen) for 20 minutes before the aAPC were added to the cells. For each sample, 1 g of HLA class I per 6,000 cells was used. Preparation of effector T cells and staining for fluorescence-activated cell sorting (FACS) analysis. PBMCs (2 ⫻ 106/ml) were incubated in 24-well plates and pulsed with 10 g/ml of M5 (aa 367–375) or Myo (aa 114–122) peptide; 40 units/ml of recombinant human IL-2 was added on days 0 and 3. On day 5, cells were washed twice, stained with labeled monoclonal antibodies for human CD3 (PharMingen) or with isotype control for 20 minutes at 4°C, washed twice, and resuspended in staining buffer. T cell capture. Stained cells were incubated with aAPC for 30 minutes at room temperature. Cells and aAPC were washed twice and resuspended in staining buffer. Subsequently, acquisition was done, and cells were sorted on a FACSVantage (Becton Dickinson, San Jose, CA). T cells bound by aAPC presenting either the streptococcal peptide or its human homolog were identified and sorted during the TCC experiments (500–1,600 cells). Messenger RNA (mRNA) was extracted by using the Micro-Fast Track kit (Invitrogen, Carlsbad, CA). V␤ usage of the TCR of antigen-specific sorted cells. PBMCs were cultured with the different peptides, and antigenspecific T cells were sorted using a FACSVantage. Messenger RNA was extracted from 500–1,600 sorted cells by using the MASSA ET AL Micro-Fast Track kit. The yield of mRNA was ⬃2 ng and was resuspended in 20 l of water. Two microliters of mRNA for each reaction was reverse transcribed into single-stranded DNA with the oligo(dT) primer (cDNA [complementary DNA] Cycle kit; Invitrogen). Single-stranded cDNA (1.5 l) was amplified with constant primer and different V-region primers (V␤1–V␤24). The sequences of the V␤-specific primers used were as follows: V␤1, CTAAACCTGAGCTCTCTGGAG; V␤2, GCTTCTACATCTGCAGTGC; V␤3, CTGGAGTCCGCCAGCACC; V␤4, GCAACATGAGCCCTGAAG; V␤5.1, GATGAATGTGAGCACCTTGGAG; V␤5.3, GCTGAATGTGAACGCCTTGTTG; V␤6.1, GATCCAGCGCACACAGC; V␤6.2, GATCCAGCGCACAGAGC; V␤7, CCTGAATGCCCCAACAGC; V␤8, CCAGCCCTCAGAACCAG; V␤9, CCCTGGAGCTTGGTGACTCTG; V␤10, CCAGYCCACGGAGTCAGG; V␤11, CCCTGGAGTCTGCCAGGC; V␤12, CTCTGGAGTCCGCTACCAG; V␤13, GCTCAGGCTGCTGTCGGCTGC; V␤14, GTCTCTCGAAAAGAGAAGAGG; V ␤ 15, CCCTAGAGTCTGCCATCC; V␤16, GGTGCAGCCTGCAGAAC; V␤17, GGATCCAGCAGGTAGTGCG; V␤18, CCTCCTCAGTGACTCTGGC; V ␤ 19, CACTGTGACATCGGCCCAAAAG; V␤20, CCTGTCCTCAGAACCGGG; V␤21, CCAGCCAGCAGAGCTTGG; V␤22, CTGAACATGAGCTCCTTGG; V␤23, CCGGTCCACAAAGCTGG; and V␤24, CATCCGCTCACCAGGCCTG. Twenty picomoles of each primer and 1.25 units of Taq polymerase (Boehringer, Mannheim, Germany) were used. The total volume was 25 l. The cycling parameters for polymerase chain reaction (PCR) were as follows: the PCR reaction mixture was heated without Taq and dNTPs at 94°C for 4 minutes, followed by 40 cycles of 30 seconds at 94°C, 30 seconds at 58°C, and 30 seconds at 72°C. At the end of 58°C of the first cycle, 0.5 ml of 100 mM dNTPs and 0.25 l of Taq polymerase (1.25 units; Boehringer) were added. The final elongation step was 7 minutes at 72°C. The reaction was repeated a second time using the same conditions and onetenth of the volume of the first PCR as template. The PCR-amplified products were analyzed on a 4% agarose gel. Intracellular cytokine staining. PBMCs (2 ⫻ 106) from 6 patients with juvenile DM or 10 healthy controls were incubated in 24-well plates and pulsed with 10 g/ml of M5 (aa 367–375) or Myo (aa 114–122) peptide. Recombinant human IL-2 (40 units/ml) was added on days 0 and 3. On day 5, 4 hours before beginning the staining procedure, monensin (10 l/ml) (PharMingen) was added to each well. Cells were washed twice with PBS, 0.5% bovine serum albumin, and 0.01% NaN3, and stained with labeled anti-CD3 monoclonal antibody (PharMingen) for 30 minutes at 4°C. The intracellular staining was performed using the Cytofix/Cytoperm Kit (Becton Dickinson), and the anti–human TNF␣–, anti–human interferon-␥ (IFN␥)–, or anti–human IL-2–labeled antibody, or the appropriate isotype control, was added to the cells for 30 minutes at 4°C. Cells were washed, resuspended in staining buffer, and analyzed using a FACSCalibur (Becton Dickinson). Statistical analysis. The statistical analysis comparing patients with active disease with those whose disease was in remission, as well as with healthy controls, JIA patients, DD patients, and PSD patients, was performed by means of a nonparametric test based on rank-sum statistics. The situation here differed from the usual one in which tests are derived IMMUNE RESPONSE TARGETS IN ACTIVE JUVENILE DM 3019 Table 3. Cytotoxic responses to 4 peptides in patients with juvenile DM, healthy controls, and patients with other diseases* Juvenile DM patients Peptide code (aa region) Total (n ⫽ 16) Active disease (n ⫽ 8) In remission (n ⫽ 8) Healthy controls (n ⫽ 10) M5 (364–372) Myo (111–119) M5 (367–375) Myo (114–122) 12.4 ⫾ 4.8 18.8 ⫾ 8.1 24.4 ⫾ 7.6 10.1 ⫾ 3.1 19.9 ⫾ 8.7 12.4 ⫾ 5.9 35 ⫾ 12.8 15.9 ⫾ 4.3 4.9 ⫾ 3.6 25.2 ⫾ 15.9 12.5 ⫾ 5.8† 3.5 ⫾ 3.5# 1.0 ⫾ 0.8 4.0 ⫾ 1.5 1.6 ⫾ 0.9‡ 1.6 ⫾ 1.0‡ Patients with other diseases JIA (n ⫽ 6) DD (n ⫽ 3) PSDs (n ⫽ 3) 4.7 ⫾ 1.5 1.8 ⫾ 1.1 4.9 ⫾ 1.2§ 2.3 ⫾ 1.4** 0.4 ⫾ 0.4 0.2 ⫾ 0.2 4.4 ⫾ 1.6¶ 3.4 ⫾ 3.4¶ 2.4 ⫾ 1.4 10.6 ⫾ 2.0 5.9 ⫾ 4.8¶ 1.0 ⫾ 0.5†† * Values are the mean ⫾ SEM percentage of specific lysis at an effector to target ratio of 25:1. Cytotoxic activity toward autologous Epstein-Barr virus–transformed lymphoblastoid B cell lines pulsed with the streptococcal peptides M5 (amino acids [aa] 364–372 or aa 367–375), or with the human myosin heavy chain peptides Myo (aa 111–119 or aa 114–122), was measured in all patients with juvenile dermatomyositis (juvenile DM), juvenile DM patients with or without active myositis at the time of sampling, and healthy controls, and in patients with juvenile idiopathic arthritis (JIA), Duchenne’s muscular dystrophy (DD), or poststreptococcal diseases (PSDs; 1 with poststreptococcal arthritis, 1 with poststreptococcal glomerulonephritis, and 1 with acute rheumatic fever). P values are versus patients with active juvenile DM (calculated as described in Patients and Methods). † P ⬍ 0.05. ‡ P ⬍ 0.005. § P ⬍ 0.03. ¶ P ⬍ 0.1. # P ⬍ 0.002. ** P ⬍ 0.02. †† P ⬍ 0.06. from applying permutation principles to either paired or unpaired data without ties. When a few ties are present, the usual theory does not apply, but approximate tests are often adequate if the sample size(s) is(are) large. In these data, there were many ties (due to the apparent absence of cytotoxic activity), and the sample sizes were small. To obtain correct, exact P values, standard permutation methods were used, and P values were determined by enumerating the entire permutation distribution for each comparison. For the comparison of patients with active disease with those whose disease was in remission, patients were treated as a matched pair and the allowable permutations were adjusted accordingly. Results presented are for 2-tailed tests. RESULTS Peptide selection. To select appropriate peptides for cytotoxicity assays, sequence homologies among human skeletal myosin and microbial proteins were obtained by scanning nonredundant NBRF databases using BLAST and FASTA software (26). Subsequently, prediction of HLA class I universal major binding motifs was performed utilizing an algorithm kindly provided by Dr. Sette (Epimmune). Scores were confirmed utilizing a computerized prediction of peptide binding motifs to individual HLA alleles based on the BIMAS site (see Patients and Methods for URL). Sequence homologies were predominantly focused on common agretopic motifs rather than on absolute identity in sequence (26). Two overlapping epitopes derived from the 364–375-aa region of M5 type protein, and their counterparts, derived from the human skeletal myosin heavy chain homologous regions (aa 111–119 and aa 114–122), ap- peared to be of interest. Analysis of another stretch of the M5 protein (aa 228–235 and aa 380–391), which was previously described (14) as being homologous to myosin heavy chain, did not result in the identification of putative anchor residues (not shown). Avidity of the peptides for HLA class I molecules was measured by a direct HLA class I/peptide binding assay. Affinity-purified HLA class I molecules from 3 unrelated subjects were incubated with the relevant biotinylated peptides and with biotinylated HA-I. HA-I (GILGFVFTL) is a pan–HLA class I binder that was used as a standard (equivalent to a value of 1). Binding of the tested peptides was expressed relative to this standard as a ratio. Results from these experiments showed that the peptide pairs had binding avidity comparable with that of HLA class I molecules (Table 2). Hence, it may be argued that differences in immunogenicity within and between the pairs could not be ascribed to disparate HLA binding avidity. Elevated cytotoxicity to the Myo (aa 114–122) peptide and to its streptococcal homolog M5 (aa 367– 375) in patients with active juvenile DM compared with controls. Cytotoxicity was evaluated against autologous EBV-transformed lymphoblastoid B cell lines pulsed with the same peptides used to stimulate PBMCs for 5 days. While cell lines from healthy controls showed negligible cytotoxic responses to the 4 peptides, those from juvenile DM patients as a whole showed increased cytotoxic responses to the 4 peptides (Table 3). When we stratified the whole juvenile DM population studied, 3020 Figure 1. Results of cytotoxicity assays of peripheral blood mononuclear cells against autologous targets pulsed with M5 (amino acids [aa] 367–375) and Myo (aa 114–122). Shown are results in 8 patients with active juvenile dermatomyositis (juvenile DM), 8 patients with juvenile DM in remission, 10 healthy controls, and 12 controls with active disease (6 with juvenile idiopathic arthritis [JIA], 3 with Duchenne’s muscular dystrophy [DD], and 3 with poststreptococcal diseases [PSDs]). Cross-reactivity studies were performed in 3 patients with active juvenile DM and in 3 healthy controls. We tested effector cells pulsed with the peptide of streptococcal origin (M5 [aa 367–375]) against target cells pulsed with the homologous peptide derived from human skeletal myosin heavy chain (Myo [aa 114–122]). Results, expressed as percentage of specific lysis of a 25:1 effector:target ratio, are shown as the mean ⫾ SEM. P values were calculated as described in Patients and Methods. * ⫽ P ⬍ 0.005 versus healthy controls; ⫽ P ⬍ 0.002 versus patients with juvenile DM in remission; $ ⫽ P ⬍ 0.05 versus healthy controls. based on clinical correlations, we found that cell lines from juvenile DM patients with active myositis, defined as described in Patients and Methods, had significantly greater cytotoxic responses to the M5 (aa 367–375) and Myo (aa 114–122) peptides (P ⬍ 0.005) than did cell lines from healthy controls (Figure 1 and Table 3). The differences between patients and controls in cytotoxic responses to the peptides Myo (aa 111–119) and M5 (aa 364–372) did not reach statistical significance (P ⫽ 0.09 and P ⫽ 0.79, respectively). To further explore the relationship between cytotoxicity and disease activity, the cytotoxic activity in juvenile DM patients with active myositis was compared with that in patients with remission of disease (those with normal muscle strength). We found that juvenile DM patients with active myositis had significantly greater cytotoxic responses to the Myo (aa 114–122) and M5 (aa 367–375) peptides (P ⬍ 0.002 and P ⬍ 0.05, respectively) than did patients with normal muscle strength (Figure 1 and Table 3). Cytotoxicity to the Myo (aa 114–122) peptide was significantly inversely correlated with disease duration (r ⫽ ⫺0.491, P ⫽ 0.04). It is worth noting that among the patients with disease duration of ⬎1 year, the 3 patients with decreased MASSA ET AL muscle strength (patients 7, 9, and 11 in Table 1) had elevated cytotoxic responses to the Myo (aa 114–122) peptide (mean ⫾ SEM 33.6 ⫾ 18.0% at an E:T ratio of 25:1; P ⫽ 0.030 versus healthy controls), suggesting that the persistence of an active phase of the disease, despite the long disease duration, may be related to a cytotoxic response to a target epitope of the human myosin heavy chain. We also found that cytotoxic responses to the Myo (aa 114–122) peptide were significantly greater (P ⫽ 0.022) in patients not receiving corticosteroids (mean ⫾ SEM 21.6 ⫾ 7.63% at the E:T ratio of 25:1) than in those receiving them (mean ⫾ SEM 5.4 ⫾ 1.9%). Together, these results show that cytotoxic responses to the M5 (aa 367–375) and Myo (aa 114–122) peptide pair are associated with clinically active disease, early phase of the disease, and absence of corticosteroid treatment. These findings support the pathogenic role of these peptides. Juvenile DM–specific cytotoxicity to the Myo (aa 114–122) peptide and to its streptococcal homolog M5 (aa 367–375). We tested an additional group of disease controls consisting of 6 patients with JIA (a different autoimmune disease), 3 with DD (a disease with muscle fiber damage and muscle cellular infiltrate of nonautoimmune origin), and 3 with PSDs. All of these patients had active disease. We found that the cytotoxic activity to the myosin and M5 protein homologous peptides was negligible and comparable with that in healthy controls (Figure 1 and Table 3). Cytotoxicity to the Myo (aa 114–122) and M5 (aa 367–375) peptides was significantly lower in JIA than in active juvenile DM (P ⬍ 0.02 and P ⬍ 0.03, respectively). These data show that T cell reactivity to the myosin Myo (aa 114–122) peptide and its homolog from M5 protein, M5 (aa 367–375), is found in children with clinically active juvenile DM and not in control groups with clinically active disease. ARF-specific cytotoxicity to a control peptide from streptococcal M5 protein (KGLRRDLDA) containing the RRDL sequence homologous with cardiac and skeletal myosin. It is well accepted that PBMCs from patients with ARF recognize an epitope from human cardiac myosin containing the RRDL amino acid sequence (27). We sought to explore whether such recognition was also present in patients with active juvenile DM. By using the same approach described above for peptide selection, we identified a peptide of M5 protein that shares a 4–amino acid (RRDL) sequence homology with both human cardiac and skeletal myosin. Cytotoxicity to this peptide was tested in patients with active juvenile DM as well as in patients with active ARF. We found that cytotoxicity to this peptide IMMUNE RESPONSE TARGETS IN ACTIVE JUVENILE DM Figure 2. Results of cytotoxicity assays of peripheral blood mononuclear cells against autologous targets pulsed with M5 protein peptide KGLRRDLDA. Shown are results in 3 patients with active juvenile dermatomyositis (juvenile DM) and in 3 patients with acute rheumatic fever (ARF). We tested effector cells pulsed with the peptide of streptococcal origin. Results, expressed as percentage of specific lysis of a 25:1 effector:target ratio, are shown as the mean ⫾ SEM. The P value was calculated as described in Patients and Methods. was significantly greater (P ⬍ 0.05) in ARF patients than in juvenile DM patients (Figure 2). These results confirm that T cell responses to this peptide are specific for ARF and do not play a role in juvenile DM. T cells recognizing M5 (aa 367–375) streptococcal peptide related to T cells recognizing myosin-derived peptide Myo (aa 114–122). Functional cross-reactivity was studied by cytotoxicity assay. Effector cells were generated from PBMCs of patients with active juvenile DM by incubation for 5 days with the M5 (aa 367–375) streptococcal peptide, and their cytotoxic activity was tested against autologous target cells pulsed overnight with the human homolog Myo (aa 114–122). As shown in Figure 2, incubation with the M5 (aa 367–375) peptide of PBMCs from 3 patients with active juvenile DM resulted in the activation of presumably few cytotoxic T cells reactive against the Myo (aa 114–122) peptide (P ⬍ 0.05 versus healthy controls at an E:T ratio of 25:1). Moreover, PBMCs from 3 patients with active juvenile DM were cultured with either the streptococcal peptide M5 (aa 367–375) or the human myosin peptide Myo (aa 114–122). T cells from the different cultures were then incubated with aAPC encompassing the HLA class I–streptococcal peptide complex or the HLA class I–human homologous peptide complex. Antigen-specific 3021 T cells for each peptide were identified by TCC and sorted by FACS (Figure 3). Subsequently, reverse transcriptase–PCR was performed on the sorted cells for TCR V␤ usage. Antigen-specific T cells sorted after culture with each peptide of the peptide pair (M5 [aa 367–375] and Myo [aa 114–122]) had largely overlapping V␤ usage (Figure 3). Combined with the cytotoxicity assays showing functional cross-recognition of the two epitopes, these results demonstrate that degenerate recognition of homologous peptides by similar or identical T cells may trigger pathogenic mechanisms based on molecular mimicry. Production of TNF␣ from T cells of patients with active juvenile DM, but not from those of controls, induced by T cell recognition of the peptide pair M5 (aa 367–375) or Myo (aa 114–122) encompassing the shared epitopes. In addition to the induction of cytotoxic responses, we investigated whether the selected peptides were able to induce proinflammatory responses. We incubated PBMCs from 10 healthy controls and 6 patients with active juvenile DM with the peptides M5 (aa 367–375) and Myo (aa 114–122). We then used intracellular staining and FACS analysis to measure the production of IL-2, IFN␥, and TNF␣ by CD3⫹ T cells. Both peptides induced significantly elevated production of TNF␣ by T cells from patients compared with those from controls (Table 4). No IL-2 or IFN␥ production was found (data not shown). DISCUSSION The mechanisms leading to muscle inflammation and muscle fiber damage in juvenile DM are considered to be autoimmune in nature. However, little is known about the triggers and the targets of such abnormal immune responses. In this study, we found that peripheral blood T cells of patients with juvenile DM lyse autologous target cells presenting an HLA class I binding peptide derived from skeletal myosin. This suggests that this myosin heavy chain epitope is a target autoantigen in juvenile DM and that these cytotoxic T lymphocytes contribute to muscle fiber damage. In support of the pathogenic significance of this response, we observed that increased cytotoxicity to the Myo (aa 114– 122) peptide was found in patients with shorter disease duration and clinically evident muscle involvement and who were not receiving corticosteroid treatment. In adult patients with polymyositis, lysis of muscle cells by infiltrating CD4 or CD8 T cells has been suggested by studies demonstrating Fas ligand expression on T cells and Fas expression on muscle fibers, and 3022 MASSA ET AL Figure 3. T cell capture (TCC) analysis of peptide-specific T cells from patients with active juvenile DM. This experiment is representative of a series involving a total of 3 patients. Artificial antigen-presenting cells (aAPC; prepared as described in Patients and Methods and in reference 30) containing HLA class I molecules obtained from a patient’s lymphoblastoid cell line were loaded with biotinylated peptides and incubated with the patient’s peripheral blood mononuclear cells, which were expanded for 5 days with the relevant peptides. Cells were sorted and mRNA was extracted for analysis of T cell receptor V␤ usage by reverse transcriptase–polymerase chain reaction. Gates were set based on nonspecific binding of fluorescein isothiocyanate (FITC)–conjugated streptavidin to T cell–aAPC complexes bearing nonbiotinylated M5 (aa 367–375) peptide (left panel). The middle panel shows T cells bound by aAPC presenting the biotinylated streptococcal peptide M5 (aa 367–375) stained with FITC-conjugated streptavidin. The right panel shows T cells bound by aAPC presenting the biotinylated human myosin peptide Myo (aa 114–122) stained with FITC-conjugated streptavidin. Cells were gated based on CD3 positivity. The y-axis represents side scatter. The table shows that antigen-specific T cells sorted after culture with either peptide of the peptide pair (M5 [aa 367–375] and Myo [aa 114–122]) have largely overlapping V␤ usage. See Figure 1 for other definitions. by additional studies showing expression of perforin, granzyme B, and TIA-1 in infiltrating CD4 and CD8 T cells, as well as in natural killer (NK) cells (31–33). Analysis of muscle infiltrate in juvenile DM has shown activated CD4 and CD8 T lymphocytes (2), as well as CD56 NK cells (3). Moreover, in juvenile DM, a strong expression of HLA class I antigens at the periphery of muscle fibers in biopsy specimens has been described Table 4. Percentage of T cells producing tumor necrosis factor ␣ upon in vitro stimulation with the peptide pair in patients with active juvenile dermatomyositis (juvenile DM) and in healthy controls* Peptide code (aa region) Juvenile DM patients (n ⫽ 6) Healthy controls (n ⫽ 10) P M5 (367–375) Myo (114–122) 0.6 ⫾ 0.1 0.4 ⫾ 0.1 ⬍0.02 ⬍0.02 0.012 0.023 * Values are the mean ⫾ SEM or the mean. aa ⫽ amino acid. (2,34). These two observations are consistent with a pathogenic mechanism involving CD8⫹ cytotoxic T cells. We also tested, for humoral and proliferative responses, recombinant M5 protein and 4 synthetic peptides that were able to bind the HLA class II molecules (according to the algorithm we used). These were different from those used for the cytotoxicity assay, although they encompassed the same core homologies. No differences were found between patients with juvenile DM and controls, suggesting that these approaches were not informative. Our data do not exclude a possible role of CD4 T cells based on the recognition of different epitopes. In cells from juvenile DM patients with active disease, incubation with the M5 (aa 367–375) peptide generated modest, although detectable, cytotoxic activity against the human homolog Myo (aa 114–122). This IMMUNE RESPONSE TARGETS IN ACTIVE JUVENILE DM 3023 Table 5. Epitopes from common pathogens presenting a sequence homology with the Myo (amino acid region 114–122) peptide derived from human myosin heavy chain* Proteins Amino acid region Sequence Myosin heavy chain Borrelia burgdorferi (ErpK protein) Mycoplasma hominis (p50 adhesion) Haemophilus influenzae translation initiation factor 2 (IF-2) Helicobacter pylori (conserved hypothetical protein) Escherichia coli colicin protein Bacillus subtilis Hsp70 (DnaK) 114–122 157–165 211–219 56–63 46–54 148–156 484–492 KMRRDLEEA KRKKELEES KIVSEWEEV KLAQQEAE KEKKELEKE KAFQEAEQR RMVKEAEEN * Boldface type denotes identity. Italic type denotes conservative change (sufficient similarities in size and charge to cause little alteration in protein conformation). finding may be explained by the reactivity of the same T cell populations to homologous peptides of either microbial or self origin (35,36). Further analysis of these T cell populations by means of a novel method, TCC, for the identification of rare antigen-specific T cells (30), showed oligoclonality of these T cell populations with largely overlapping TCR V␤ genes. Although we have not demonstrated direct T cell cross-reactivity at a clonal level, together these results suggest that Myo (aa 114– 122) is the target of disease-specific responses, and that reactivity to exogenous homologous antigens, such as M5 (aa 367–375), contributes to the process supporting a pathogenic mechanism based on molecular mimicry. Incubation of T cells with both peptides (Myo [aa 114–122] and M5 [aa 367–375]) induced production of the proinflammatory cytokine TNF␣ in juvenile DM patients, but not in controls. Production of proinflammatory cytokines in response to T cell stimulation may be another important mechanism for perpetuating and amplifying the autoimmune damage in this context. Our findings do not appear to depend on differences in HLA class I typing between patients and controls, since juvenile DM appears to be associated with class II alleles. In fact, the initially reported association between HLA–B8 and juvenile DM (37) was subsequently found to be weak (38) and due instead to linkage disequilibrium with class II alleles of the DRB1 and DQA1 loci (39–41). More recently, associations of juvenile DM with class II alleles, particularly HLA–DQA1*0501/DQB1*0301 and DQA1*0201, have been described. It has been hypothesized that such associations may have functional relevance. Juvenile DM–associated alleles may have the capability to bind a peculiar, and possibly autoimmune, set of peptides (42). The peptides tested in this study were designed for comparable binding affinity to different HLA alleles, in accordance with recently described HLA class I binding super motifs. This approach should overcome individual genetic differences at HLA loci and emphasize disease-related differences in T cell function instead. In fact, the increased cytotoxic activity was found mainly in the subset of patients with active myositis. No differences between patients and controls were found when HLA class II–restricted responses to M5 protein and class II–restricted peptides were measured. Furthermore, our findings do not appear to be due to a nonspecific increase in the induction of cytotoxic responses secondary to inflammation. Indeed, 1) no significant increase was observed in control patients with other inflammatory diseases; 2) the increase was significant only for the pair Myo (aa 114–122)/M5 (aa 367–375), but not for the other streptococcal or autologous peptides tested; and 3) only the cytotoxic response to Myo (aa 114–122), but not to the other peptides, was significantly associated with the presence of active myositis. Cells from patients with PSDs did not show increased cytotoxicity to the Myo (aa 114–122) peptide, showing that patients with recent streptococcal infection and with a Streptococcus-triggered autoimmune inflammation (not involving skeletal muscles) do not recognize this self-myosin epitope. Therefore, increased cytotoxic activity to the Myo (aa 114–122) peptide is not a generalized consequence of streptococcal infections. Since studies of muscle biopsy samples of patients with DD show muscle fiber lysis (therefore, with probable release of the myosin epitope) as well as mononuclear cell infiltrate and HLA class I expression on muscle fibers (2,3), the absence of cytotoxic response to the Myo (aa 114–122) peptide in DD suggests that the cytotoxic responses found in juvenile DM patients are not a nonspecific consequence of muscle fiber lysis and muscle inflammation. To further demonstrate disease specificity, we designed a peptide from human cardiac myosin 3024 encompassing one of the core epitopes recognized by patients with ARF. When such a peptide was used as a target of cytotoxic responses, only patients with active ARF, but not patients with active juvenile DM, developed T cell responses to this control peptide. Although in some instances clinical evidence relating juvenile DM to an antecedent streptococcal infection has been reported (14), juvenile DM is not usually considered to be a PSD, at least not one with a strict chronological causal association. In fact, we found evidence of possible streptococcal infection antedating disease onset (i.e., increase in ASO titer in two subsequent determinations) in only two of our patients. The attempt to directly associate serologic markers of acute infection with a chronic autoimmune T cell response may be deceiving. The lack of increased ASO titers in association with juvenile DM does not in fact exclude the possibility that abnormal T cell responses to the epitopes described here may be relevant to pathogenesis of juvenile DM. Sydenham’s chorea (43), a PSD with onset months after infection, may serve as an example to corroborate the concept that generation of an abnormal immune response may have clinically detectable consequences several months after Streptococcus infection. To further complicate clinical immunologic correlations for this rare disease, it may be considered also that the onset of juvenile DM is frequently insidious, with a long interval between the first juvenile DM– related symptom and the diagnosis (44). It is therefore often difficult to obtain information on the nature of the possible eliciting infectious agent. A recent case–control study showed that antecedent illness in the 3 months prior to the first symptom of juvenile DM was significantly increased in children with juvenile DM compared with controls (44); in this time frame, the antecedent illness was respiratory in nature in ⬎50% of children with juvenile DM, with 9% having a documented streptococcal infection. This, as well as other erratic epidemiologic findings reported (15,16), may lead to the hypothesis of a role for diverse environmental agents. In support of the concept of a diverse etiology for the same mechanism of cross-recognition, we found that not only Streptococcus, but also other common human pathogens bear sequence homologies with the Myo (aa 114–122) peptide of skeletal myosin (Table 5). Differences in the individual’s susceptibility to diverse triggering microbial agents, all encompassing homologies with putative targets of autoreactive responses, may represent a way to reconcile the rather inconsistent reports found to date in the literature regarding associations between juvenile DM and microbial agents. MASSA ET AL In summary, we would like to propose that at least some of the targets for autoimmune reactivity in juvenile DM are contained in the myosin epitope, which we have identified. Exposure to Streptococcus or to other common environmental agents encompassing homologous epitopes may trigger autoreactive T cells. These autoreactive lymphocytes, together with other potential etiopathogenic factors such as local inflammation leading to cytokine release and efficient presentation of cryptic epitopes, may contribute significantly to chronic muscle damage. Immunomodulation of these abnormal responses will be an interesting avenue to explore as a possible therapeutic approach to this disease, as shown in other systems (45,46). ACKNOWLEDGMENTS We thank Dott. Angela Berardinelli (Neurological Institute Mondino, University of Pavia, Italy) for collecting samples from children with DD. We wish to thank Ms Nicole Lewon for excellent editorial assistance. REFERENCES 1. Cassidy J, Petty R. Textbook of pediatric rheumatology. 3rd ed. Philadelphia: WB Saunders; 1995. 2. McDouall RM, Dunn MJ, Dubowitz V. Nature of the mononuclear infiltrate and the mechanism of muscle damage in juvenile dermatomyositis and Duchenne muscular dystrophy. J Neurol Sci 1990;99:199–217. 3. Pachman LM, O’Gorman MRG, Lawton T, Liotta M, Pope RM, Greene M, et al. Studies of muscle biopsies (MBx) from DQA1*0501⫹ untreated children with juvenile dermatomyositis (JDM) very early in their disease course: evidence of a TCR V␤8 motif and increased CD56⫹ NK cells. Arthritis Rheum 1998;41 Suppl 9:S203. 4. Dawkins RL, Mastaglia FL. Cell mediated cytotoxicity to muscle in polymyositis. N Engl J Med 1973;288:434–8. 5. Haas DC, Arnason BGW. Cell-mediated immunity in polymyositis: creatine phosphokinase release from muscle cultures. Arch Neurol 1974;31:192–6. 6. Iannaccone ST, Bowen DE, Samaha FJ. Cell-mediated cytotoxicity and childhood dermatomyositis. Arch Neurol 1982;39:400–2. 7. Johnson RL, Fink CW, Ziff M. Lymphotoxin formation by lymphocytes and muscle in polymyositis. J Clin Invest 1972;51: 2435–49. 8. Nagaraju K, Raben N, Yan B, Loeffler L, Danning C, Lee E, et al. Upregulation of transgenic MHC class I (H-2Kb) in the skeletal muscle leads to autoimmune myositis in transgenic mice. Arthritis Rheum 1999;42 Suppl 9:S225. 9. Albani S, Carson DA. A multistep molecular mimicry hypothesis for the pathogenesis of rheumatoid arthritis. Immunol Today 1996;17:466–70. 10. Wucherpfennig KW, Strominger JL. Molecular mimicry in T cell mediated autoimmunity: viral peptides activate human T cell clones specific for myelin basic protein. Cell 1995;80:695–705. 11. Atkinson MA, Bowman MA, Campbell L, Darrow BL, Kaufman DL, Maclaren NK. Cellular immunity to a determinant common to glutamate decarboxylase and coxsackie virus in insulin-dependent diabetes. J Clin Invest 1994;94:2125–9. IMMUNE RESPONSE TARGETS IN ACTIVE JUVENILE DM 3025 12. La Cava A, Nelson JL, Ollier WER, MacGregor A, Keystone EC, Thorne JC, et al. Genetic bias in immune responses to a cassette shared by different microorganisms in patients with rheumatoid arthritis. J Clin Invest 1997;100:658–63. 13. Albert LJ, Inman RD. Molecular mimicry and auto immunity. N Engl J Med 1999;341:2068–74. 14. Martini A, Ravelli A, Albani S, Viola S, Scotta MS, Magrini U, et al. Recurrent juvenile dermatomyositis and cutaneous necrotizing arthritis with molecular mimicry between streptococcal type 5M protein and human skeletal myosin. J Pediatr 1992;121:739–42. 15. Rider LG, Okada S, Sherry DD, Wallace CA, Zemel LS, Jacobs JC, et al. Epidemiologic features and environmental exposures associated with illness onset in juvenile idiopathic inflammatory myopathy (JIIM). Arthritis Rheum 1995;38 Suppl g:S362. 16. Koch MJ, Brody JA, Gillespie MM. Childhood polymyositis: a case-control study. Am J Epidemiol 1976;104:627–31. 17. Dale JB, Beachey EH. Epitopes of streptococcal M proteins shared with cardiac myosin. J Exp Med 1985;162:583–91. 18. Dale JB, Beachey EH. Human cytotoxic T lymphocytes evoked by group A streptococcal M proteins. J Exp Med 1987;166:1825–35. 19. Cunningham MW, Swerlick RA. Poly specificity of anti streptococcal murine monoclonal antibodies and their implications in auto immunity. J Exp Med 1986;164:998–1012. 20. Bisno AL, Berrios X, Quesney F, Monroe DM Jr, Dale JB, Beachey EH. Type-specific antibodies to structurally defined fragments of streptococcal M proteins in patients with acute rheumatic fever. Infect Immun 1982;38:573–9. 21. Cunningham MW. Pathogenesis of group A streptococcal infections. Clin Microbiol Rev 2000;13:470–511. 22. Cunningham MW, Antone SM, Smart M, Liu R, Kosanke S. Molecular analysis of human cardiac myosin cross-reactive B- and T cell epitopes of the group A streptococcal M5 protein. Infect Immun 1997;65:3913–23. 23. Guilherme L, Cunha-Neto E, Coelho V, Snitcowsky R, Pomerantzeff PMA, Assis RV, et al. Human heart-filtrating T cell clones from rheumatic heart disease patients recognize both streptococcal and cardiac proteins. Circulation 1995;92:415–20. 24. Bohan A, Peter JB, Bowman RL, Pearson CM. A computer assisted analysis of 153 patients with polymyositis and dermatomyositis. Medicine (Baltimore) 1977;56:255–86. 25. De Benedetti F, De Amici M, Aramini L, Ruperto N, Martini A. Correlation of serum neopterin levels with disease activity in juvenile dermatomyositis. Arch Dis Child 1993;69:232–5. 26. Rammensee HG, Friede T, Stefanovic S. MHC ligands and peptide motifs: first listing. Immunogenetics 1995;41:178–228. 27. Quinn A, Ward K, Fischetti VA, Hemric M, Cunningham MW. Immunological relationship between the class I epitope of streptococcal M protein and myosin. Infect Immun 1998;66:4418–24. 28. Albani S, Keystone EC, Nelson JL, Ollier WER, La Cava A, Montemayor AC, et al. Positive selection in auto immunity: abnormal immune responses to a bacterial dnaJ antigenic determinant in patients with early rheumatoid arthritis. Nat Med 1995;1:448–52. 29. Brander C, Wyss-Coray T, Mauri D, Bettens F, Pichler WJ. Carrier mediated uptake and presentation of a major histocompatibility complex class IB restricted peptide. Eur J Immunol 1993;23: 3217–23. 30. Prakken B, Wauben W, Genini D, Samodal R, Barnett J, Mendivil A, et al. Artificial antigen-presenting cells as a tool to exploit the immune ‘synapse.’ Nat Med 2000;6:1406–10. Sugiura T, Murakawa Y, Nagai A, Kondo M, Kobayashi S. Fas and Fas ligand interaction induces apoptosis in inflammatory myopathies: CD4⫹ T cells cause muscle cell injury directly in polymyositis. Arthritis Rheum 1999;42:291–8. Cherin P, Herson S, Crevon MC, Hauw JJ, Cervera P, Galanaud P, et al. Mechanisms of lysis by activated cytotoxic cells expressing perforin and granzyme-B genes and the protein TIA-1 in muscle biopsies of myositis. J Rheumatol 1996;23:1135–42. Goebels N, Michaelis D, Engelhardt M, Huber S, Bender A, Pongratz D, et al. Differential expression of perforin in muscleinfiltrating T cells in polymyositis and dermatomyositis. J Clin Invest 1996;97:2905–10. Topaloglu H, Muntoni F, Dubowitz V, Sewry C. Expression of HLA class I antigens in skeletal muscle is a diagnostic marker of juvenile dermatomyositis. J Child Neurol 1997;12:60–3. Thompson SD, Murray KJ, Grom AA, Passo MH, Choi E, Glass DN. Comparative sequence analysis of the human T cell receptor ␤ chain in juvenile rheumatoid arthritis and juvenile spondylarthropathies: evidence for antigenic selection of T cells in the synovium. Arthritis Rheum 1998;41:482–97. Nepom BS, Nepom GT, Coleman M, Kwok WW. Critical contribution of beta chain residue 57 in peptide binding ability of both HLA-DR and -DQ molecules. Proc Natl Acad Sci U S A 1996;93: 7202–6. Pachman LM, Johasson O, Cannon RA, Friedman JM. HLA-B8 in juvenile dermatomyositis. Lancet 1977;2:567–8. Friedman JM, Pachman LM, Maryjowski ML, Jonasson O, Battles ND, Crowe WE, et al. Immunogenic studies of juvenile dermatomyositis: HLA antigens in patients and their families. Tissue Antigens 1983;21:45–9. Friedman JM, Pachman LM, Maryjowski ML, Radvany RM, Crowe WE, Hanson V, et al. Immunogenetic studies of juvenile dermatomyositis: HLA-DR antigen frequencies. Arthritis Rheum 1983;26:214–6. Reed AM, Pachman LM, Hayford J, Ober C. Immunogenetic studies in families of children with juvenile dermatomyositis. J Rheumatol 1998;25:1000–2. Reed AM, Pachman LM, Ober C. Molecular genetic studies of major histocompatibility complex genes in children with juvenile dermatomyositis: increased risk associated with HLADQA1*0501. Hum Immunol 1991;32:235–40. Reed AM, Collins EJ, Shock LP, Klapper DG, Frelinger JA. Diminished class II associated Ii peptide binding to the JDM HLA DQA1*0501/DQB1*0301 molecule. J Immunol 1997;159:6260–5. Rullan E, Sigal LH. Rheumatic fever. Curr Rheumatol Rep 2001;3:445–52. Pachman LM, Hayford JR, Hochberg MC, Pallansch MA, Chung A, Daugherty CD, et al. New-onset juvenile dermatomyositis: comparisons with a healthy cohort and children with juvenile rheumatoid arthritis. Arthritis Rheum 1997;40:1526–33. Bonnin D, Albani S. Induction of oral tolerance to heat shock proteins in autoimmune diseases. Biotherapy 1998;10:213–21. Prakken BJ, van der Zee R, Anderton SM, van Kooten PJ, Kuis W, van Eden W. Peptide-induced nasal tolerance for a mycobacterial heat shock protein 60 T cell epitope in rats suppresses both adjuvant arthritis and nonmicrobially induced experimental arthritis. Proc Natl Acad Sci U S A 1997;94:3284–9. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.