Serum samples from 94 Japanese patients with adult DM (67 with classical DM and 27 with CADM) who were seen at Keio University Hospital (2001-2008) and Tokai University Hospital (2010-2013) were screened for autoantibodies using immunoprecipitation (IP) assays. Blood samples were obtained after the patients had provided written informed consent approved by the Keio and Tokai University Institutional Review Boards.

The IP assay with K562 or HeLa cell extracts was performed as previously described [9] . For analysis of RNAs, antibodies bound to protein A-Sepharose CL-4B beads (Pharmacia Biotech AB, Uppsala, Sweden) were incubated with extracts of K562 or HeLa cells. They were then washed with NET-2 buffer (50 mM Tris-HCL, pH 7.5, 150 mM NaCl (Wako Pure Chemical Industries, Osaka, Japan), 0.05% Nonidet P-40 (Nacalai tesque, Kyoto, Japan). After ethanol precipitation, RNAs were dissolved in electrophoresis sample buffer composed of 10 M urea (Bio-Rad Laboratories, Hercules, CA), 0.025% bromophenol blue (Wako Pure Chemical Industries, Osaka, Japan), and 0.025% xylene cyanol-FF (Wako Pure Chemical Industries, Osaka, Japan) in TBE buffer (90 mM Tris-HCl, pH 8.6, 90 mM boric acid (Wako Pure Chemical Industries, Osaka, Japan), and 1 mM EDTA (Wako Pure Chemical Industries, Osaka, Japan)). The RNA samples were resolved in 7M urea-10% polyacrylamide gels, which were then silver stained (Bio-Rad Laboratories, Hercules, CA). For protein studies, antibody-coated Sepharose beads were mixed with [³⁵S] methionine-labeled K562 or HeLa extracts. After washing, the Sepharose beads were resuspended in SDS-sample buffer (2% SDS (Sigma, St. Louis, USA), 10% glycerol (Wako Pure Chemical Industries, Osaka, Japan), 62.5 mM Tris-HCl, pH 6.8, 0.005% bromophenol blue). The proteins were then fractionated by 10% SDS-PAGE and dried. Radiolabeled protein components were analyzed by autora- diography. Anti-ARS, anti-SRP, anti-Mi-2, known myositis-specific antibodies, as well as anti-CADM-140/MDA5 and anti-p155/TIF1-γ are detectable with these assays by comparison with corresponding standard sera [10] .

The clinical symptoms, clinical course, and laboratory findings were retrospectively assessed from medical records in all patients including those positive for anti-ARS, anti-Mi-2, anti-CADM-140/MDA5 and anti-p155/ TIF1-γ autoantibodies.

The diagnosis of DM was based on the diagnostic criteria proposed by Bohan and Peter [11] . The assessment of muscle weakness was performed according to a manual muscle test (MMT) [12] . The diagnosis of CADM was based on classification criteria proposed by Sontheimer et al. [13] , i.e. DM patients with no clinical muscle symptoms for more than two years after the onset of skin manifestations. The remission of myositis was defined as both improvement of muscle strength on MMT and normalization of the serum creatine kinase (CK) value.

ILD was defined based on findings of chest radiography, chest CT, and lung function testing. Rapidly progressive ILD (RP-ILD) was defined as a subset of patients presenting progressive dyspnea along with hypoxemia and worsening of interstitial changes on the chest X-ray within one month of the onset of respiratory symptoms.

All comparisons between each patient group were performed using Fisher’s 2-tailed exact test or Student’s t-test. Two-sided p-values of less than 0.05 were considered statistically significant.

Using IP, we screened 94 DM patients´sera including 67 patients with classical DM and 27 with CADM. Myositis-specific autoantibodies (anti-ARS and anti-Mi-2) and myositis-associated autoantibodies (anti-SSA/Ro, anti-U1RNP) as well as recently identified autoantibodies (anti-CADM-140/MDA5, anti-p155/TIF1-γ) were studied. These proteins were easily distinguishable from those immunoprecipitated by sera reactive with other known antigens [14] .

Sera from 27 of the 94 DM patients (29%, 25 with classical DM and 2 with CADM) were found to contain

DM: dermatomyositis, CADM; clinically amyopathic dermatomyositis, ARS: aminoacyl tRNA sythetase, SRP: signal recognition particle.

anti-ARS antibodies (anti-Jo-1: 10; anti-EJ: 7; anti-PL-7: 5; anti-PL-12: 4; anti-OJ: 1). Five (5%) with classical DM had anti-Mi-2, 19 (20%, 2 with classical DM and 17 with CADM) had anti-CADM-140/MDA5, and 8 (9%, 7 with classical DM and 1 with CADM) had anti-p155/TIF1-γ antibodies. Interestingly, one patient with classical DM had both anti-Jo-1 and anti-p155/TIF1-γ antibodies. Anti-MJ/NXP-2 was not found in our series of adult DM; neither was anti-SRP, or PM-specific antibodies detected. Thus, overall, 58 of the 94 (62%) DM patients had myositis-specific antibodies.

The frequency of V-sign or shawl-sign rash in patients with anti-p155/TIF1-γ was significantly higher than in patients with anti-ARS (15% vs. 63%: P = 0.015). The frequency of ILD was significantly higher in patients with anti-ARS or anti-CADM-140/MDA5 in comparison with anti-Mi-2-positive (81% vs. 0%: P = 0.0013, 89% vs. 0%: P = 0.0005, respectively) and anti-TIF1-γ positive patients (81% vs. 13%: P = 0.0008, 89% vs. 13%: P = 0.003, respectively). Moreover, it should be noted that anti-CADM-140/MDA5-positive patients suffered significantly more RP-ILD compared to patients with anti-ARS (7% vs. 84%, P < 0.0001). Interestingly, anti-p155/TIF1-γ positive patients had a significantly higher rate of malignancy than anti-ARS-, anti-CADM- 140/MDA5- or anti-Mi-2-positive patients (75% vs. 7%: P = 0.0004, 5%: P = 0.0006, 0%: P = 0.02, respectively).