To identify the autoantigen recognized by the autoantibody that is associated with clinically amyopathic dermatomyositis (C-ADM) and rapidly progressive interstitial lung disease (ILD).
To identify the autoantigen recognized by the autoantibody that is associated with clinically amyopathic dermatomyositis (C-ADM) and rapidly progressive interstitial lung disease (ILD).
An anti–CADM-140 antibody–positive prototype serum sample was used to screen a HeLa cell–derived complementary DNA (cDNA) library. Selected cDNA clones were further evaluated by immunoprecipitation of their in vitro–transcribed and in vitro–translated products using anti–CADM-140 antibody–positive and anti-CADM-140 antibody–negative sera. The lysates of COS-7 cells transfected with the putative antigen were similarly tested. An enzyme-linked immunosorbent assay (ELISA) to detect the anti–CADM-140 antibody was established using a recombinant CADM-140 antigen, and its specificity and sensitivity for C-ADM and rapidly progressive ILD were assessed in 294 patients with various connective tissue diseases.
By cDNA library screening and immunoprecipitation of in vitro–transcribed and in vitro–translated products, we obtained a cDNA clone encoding melanoma differentiation–associated gene 5 (MDA-5). The anti–CADM-140 antibodies in patients' sera specifically reacted with MDA-5 protein expressed in cells transfected with full-length MDA-5 cDNA, confirming the identity of MDA-5 as the CADM-140 autoantigen. The ELISA, using recombinant MDA-5 protein as the antigen, showed an analytical sensitivity of 85% and analytical specificity of 100%, in comparison with the “gold standard” immunoprecipitation assay, and was useful for identifying patients with C-ADM and/or rapidly progressive ILD.
Given that RNA helicase encoded by MDA-5 is a critical molecule involved in the innate immune defense against viruses, viral infection may play an important role in the pathogenesis of C-ADM and rapidly progressive ILD. Moreover, our ELISA using recombinant MDA-5 protein makes detection of the anti–CADM-140 antibody routinely available.
Polymyositis/dermatomyositis (PM/DM) is a chronic inflammatory disorder that mainly involves muscle and skin lesions. Clinical features of the disease are heterogeneous, with varying degrees of skin manifestations, myositis, and pulmonary involvement (1). A unique subgroup of patients with DM has typical DM rashes but little or no evidence of myositis. This condition is termed “clinically amyopathic DM” (C-ADM) and is a distinct subset within the PM/DM spectrum (2). In patients with C-ADM, especially those in eastern Asia, rapidly progressive interstitial lung disease (ILD) sometimes develops; this condition is resistant to immunosuppressive treatment and is associated with poor outcomes (3, 4).
A variety of serum autoantibodies are specifically detected in patients with PM/DM, including antibodies reactive with aminoacyl–transfer RNA synthetase (aaRS) (5), signal recognition particle (SRP) (6), Mi-2 (7), and p155 (8, 9). These autoantibodies are associated with distinct clinical subsets of PM/DM. For example, anti–aaRS is associated with ILD, arthritis, Raynaud's phenomenon, and mechanic's hands (antisynthetase syndrome); anti-SRP is associated with severe refractory PM; anti–Mi-2 is associated with typical DM; and anti-p155 is associated with malignancy-associated DM. Therefore, these antibodies are useful in the diagnosis as well as classification of the disease (10).
We recently reported a novel PM/DM-associated autoantibody, termed “anti–CADM-140 antibody,” which is strongly associated with C-ADM and rapidly progressive ILD (4). The anti–CADM-140 antigen is a cytoplasmic protein of ∼140 kd; its expression is ubiquitous, but its molecular identity was not known. In this study, we have identified an RNA helicase encoded by melanoma differentiation–associated gene 5 (MDA-5) as the CADM-140 antigen, using a series of molecular and immunologic techniques. This information provides a clue about the pathogenesis of C-ADM and rapidly progressive ILD and should be useful for establishing convenient assays for detecting the anti–CADM-140 antibody, in contrast to the only technique available currently, an immunoprecipitation (IP) assay, which is complicated and requires the use of a radioisotope and cultured cells.
Serum samples were obtained from 294 randomly selected patients with connective tissue diseases or idiopathic pulmonary fibrosis (IPF) whose cases were followed at Keio University Hospital and collaborating medical centers. These patients included 35 with “classic” DM, 32 with C-ADM, 53 with PM, 69 with systemic lupus erythematosus (SLE), 68 with systemic sclerosis (SSc), and 37 with IPF. Thirty-two healthy volunteers were also included as healthy control subjects. All sera were collected after the subjects gave their written informed consent, as approved by the individual institutional review boards.
The diagnosis of C-ADM was based on the criteria proposed by Sontheimer (2), as follows: clinical skin symptoms typical of DM but minimal or no clinical features of myositis for >2 years after the onset of skin manifestations. The diagnoses of PM, DM, SLE, and SSc were made on the basis of corresponding criteria proposed by Bohan and Peter (11) or the American College of Rheumatology (12, 13). IPF was diagnosed according to the consensus classification of idiopathic interstitial pneumonias (14). Clinical information was collected retrospectively from all patients by reviewing their clinical charts. Rapidly progressive ILD was defined as a condition of worsening radiologic interstitial change with progressive dyspnea and hypoxemia within 1 month of the onset of respiratory symptoms.
One anti–CADM-140 antibody–positive prototype serum described in our original study (4) was used to screen a complementary DNA (cDNA) expression library. Further analyses were performed to confirm the identity of the CADM-140 antigen, using a set of randomly selected serum samples from our sample library: 9 positive for the anti–CADM-140 antibody (all from patients with C-ADM), 8 negative for the anti–CADM-140 antibody (2 each from patients with C-ADM or PM, and 1 each from a patient with classic DM, SLE, SSc, or IPF), and 7 from healthy control subjects.
The serum anti–CADM-140 antibody was measured by IP using 35S-labeled HeLa cell extract, as described previously (4). Sera that immunoprecipitated a protein with a molecular weight identical to that precipitated by the anti–CADM-140 antibody–positive prototype serum were defined as positive for the anti–CADM-140 antibody.
A cDNA expression library was constructed from HeLa cell–derived messenger RNA, using the λZAP-cDNA synthesis kit (Stratagene, La Jolla, CA) with an oligo(dT) primer (Oligotex-dT30; Takara, Kusatsu, Japan). A total of 9.4 × 106 plaques of λZAP phages were screened with the anti–CADM-140 antibody–positive serum, as described previously (15). After a second screening, the selected clones were mixed with an equal number of preidentified negative clones and subjected to the same phage expression procedure, probing with a set of anti–CADM-140 antibody–positive and anti–CADM-140 antibody–negative sera. The reactivity of individual serum samples with the candidate clones was assessed independently by 3 experienced investigators (KH, TS, and TF) and regarded as positive when it was judged positive by all 3 investigators.
The cDNA inserts encoded by candidate clones were subcloned into the pcDNA3.1/V5-His-TOPO vector (Invitrogen, Carlsbad, CA) and subjected to in vitro transcription and translation using the Single Tube Protein System 3 (Novagen, Darmstadt, Germany) in the presence of 35S-methionine (ICN Biomedicals, Irvine, CA), according to the manufacturer's procedure. The 35S-labeled products were subjected to IP with a set of anti–CADM-140 antibody–positive and anti–CADM-140 antibody–negative sera (4).
The cDNA inserts amplified from recombinant phages by polymerase chain reaction (PCR) were sequenced using the BigDye Terminator Cycle Sequencing Ready Reaction Kit on an ABI Prism 310 genetic analyzer (Applied Biosystems, Carlsbad, CA), as described previously (15). The nucleotide sequences were then applied to a BLAST search of genetic databases at the National Center for Biotechnology Information (NBCI) at http://www.ncbi.nlm.nih.gov.
First-strand cDNA was synthesized from HeLa cell total RNA, using random primers and AMV Reverse Transcriptase XL (Takara, Otsu, Japan), and used as a template for amplification of the 5′ portion of MDA-5 cDNA (nucleotides 153–2176) by PCR. The primers used were 5′-ATTTCACCTGTCCCGCAGACAA-3′ (sense) and 5′-ATCACTACTCCCACCACTACTAC-3′ (antisense). The cDNA fragment was then ligated in frame into the 5′ end of the MDA-5 cDNA, which was isolated from the cDNA library screening, and subcloned into the pcDNA3.1/V5-His-TOPO vector to generate a full-length cDNA for MDA-5. The plasmid harboring full-length MDA-5 or a control plasmid without insert was transfected into cultured COS-7 cells, using the Lipofectamine 2000 CD Transfection Reagent kit (Life Technologies, Rockville, MD). The total cell extracts, prepared by sonication, were used for IP with a set of anti–CADM-140 antibody–positive and anti–CADM-140 antibody–negative sera. The immunoprecipitated materials were applied to immunoblots probed with a goat anti–MDA-5 polyclonal antibody (Everest, Oxfordshire, UK) (16). In some experiments, COS-7 cell extracts were used, instead of immunoprecipitates, as an antigen source in immunoblots.
MDA-5 is a member of the retinoic acid–inducible gene I (RIG-I) family of RNA helicases, which also includes RIG-I and LGP-2. Recombinant MDA-5, recombinant RIG-I (rRIG-I), and recombinant LGP-2 (rLGP-2) were produced and purified as FLAG fusion proteins using a baculovirus expression system, as described previously (17, 18). The individual RIG-I family proteins were used in immunoblots probed with a mouse anti-FLAG monoclonal antibody (clone M2; Sigma-Aldrich, St. Louis, MO), anti–CADM-140 antibody–positive sera, or healthy control sera.
An ELISA system using rMDA-5 as an antigen source was developed as described previously (16), with some modifications. Briefly, 96-well polyvinyl plates (Sumilon multiwell plates, H type; Sumitomo Bakelite, Tokyo, Japan) were coated with purified rMDA-5 (0.5 μg/ml) at 4°C for 12 hours, followed by incubation with patients' sera diluted 1:250. All samples were examined in duplicate, and the antibody units were calculated from the optical density at 450 nm results, using a standard curve obtained from serial concentrations of a serum sample containing a high titer of the anti–CADM-140 antibody. The cutoff level was set at 8.0 units, based on 10 SDs above the mean value obtained from 32 healthy control sera.
Comparisons between groups were made using the chi-square test with Yates' correction or the Mann-Whitney U test, when appropriate.
Nine candidate clones (clones 1–9) were obtained in the initial screening. These clones were further examined for their specific reactivity with the anti–CADM-140 antibody by a phage expression system, using a set of serum samples: 9 that were anti–CADM-140 antibody positive, 8 that were anti–CADM-140 antibody negative, and 7 from healthy control subjects. Two clones were selected based on a statistically significant difference in positive frequency between the anti–CADM-140 antibody–positive and anti–CADM-140 antibody–negative sera: clone 1 (78% versus 20%; P = 0.004) and clone 8 (100% versus 13%; P = 0.0004). We next examined the reactivities of the anti–CADM-140 antibody–positive or anti–CADM-140 antibody–negative sera with the in vitro–transcribed and in vitro–translated products of the cDNA encoded by clones 1 and 8 (Figure 1). None of the anti–CADM-140 antibody–positive or anti–CADM-140 antibody–negative sera reacted with the clone 1 product. In contrast, the clone 8 product was recognized by all 9 anti–CADM-140 antibody–positive sera but was not recognized by any of the anti–CADM-140 antibody–negative or healthy control sera. Thus, clone 8 was selected as a potential target of the anti–CADM-140 antibody.
The nucleotide sequencing of clone 8 revealed that it completely matched the sequence of the carboxyl terminal portion (nucleotides 1743–3449) of MDA-5 (accession no. NM022168). According to the gene databases at the NCBI, the alternative names of MDA-5 include interferon (IFN) induced with helicase C domain 1 (IFIH1), helicase with 2 CARD domains (Helicard), RNA helicase-DEAD box protein 116 (RH116), and murabutide down-regulated protein. MDA-5 is a cytoplasmic protein with an estimated molecular weight of 140 kd (19); these characteristics were consistent with those of the CADM-140 antigen.
To confirm the identity of the CADM-140 antigen as MDA-5, the full-length MDA-5 cDNA was transfected into COS-7 cells, and the cellular lysates were subjected to IP using a set of anti–CADM-140 antibody–positive and anti–CADM-140 antibody–negative sera. The immunoprecipitates were then applied to immunoblots probed with an anti–MDA-5 polyclonal antibody (Figure 2). The transfection of MDA-5 cDNA successfully induced the expression of MDA-5, which was recognized by the anti–MDA-5 antibody. All the anti–CADM-140 antibody–positive sera immunoprecipitated MDA-5 from the transfected cell lysates but not from control cell lysates. In contrast, we failed to detect any reactivity by anti–CADM-140 antibody–negative sera, irrespective of the transfection of MDA-5 or control plasmid. Taken together, these findings indicate that the CADM-140 antigen was identical to MDA-5.
MDA-5 is one of the RIG-I family proteins, which sense intracellular viral infections and are involved in the innate immune response, including the production of type I IFN (20). RIG-I family proteins are highly homologous, each containing a DexD/H-box helicase domain. To evaluate the potential cross-reactivity of the anti–CADM-140 antibody in patients' sera with other RIG-I family proteins, immunoblotting using rRIG-I, rMDA-5, and rLGP-2 was performed (Figure 3). The anti-FLAG antibody recognized all 3 recombinant proteins, which had a tag at their amino-terminal end. The representative anti–CADM-140 antibody–positive sera reacted with rMDA-5, but not with rRIG-I or rLGP-2. All 9 sera positive for the anti–CADM-140 antibody showed the same reactivity pattern, indicating that MDA-5 was specifically recognized by the anti–CADM-140 antibodies in the patients' sera.
We used rMDA-5 as the antigen in an ELISA and screened a total of 294 serum samples obtained from patients with various connective tissue diseases or IPF (Figure 4). Of these samples, 27 were positive for the anti–CADM-140 antibody by the “gold standard” IP assay. The level of anti–CADM-140 antibody shown by the ELISA was significantly greater in the samples from patients with C-ADM than in those from patients with classic DM, PM, SSc, SLE, or IPF (P < 0.001 for all comparisons). When the serum samples were grouped according to whether they were positive or negative for anti–CADM-140 antibodies based on the cutoff level, 23 (85%) of the 27 that were positive for the anti–CADM-140 antibody by the IP assay were also positive by ELISA (Table 1). In contrast, all the sera that were negative for the anti–CADM-140 antibody by IP assay were negative by ELISA. Thus, the analytical sensitivity and specificity of the anti–CADM-140 antibody ELISA were 85% and 100%, respectively.
We also evaluated the clinical sensitivity and specificity of the ELISA for the clinical diagnosis of C-ADM. Of 262 samples obtained from patients with classic DM, PM, SSc, SLE, or IPF, only 1 sample from a patient with classic DM gave a positive result by ELISA. Significantly more samples from patients with C-ADM than from the disease control patients were positive by the ELISA (69% versus 0.4%; P < 0.0001), indicating a clinical sensitivity of 69% and clinical specificity of 99.6%. In our cohort, 17 patients (16 with C-ADM and 1 with classic DM) had rapidly progressive ILD during the course of the disease. The frequency of anti–CADM-140 antibody detected by ELISA was significantly higher in the patients with rapidly progressive ILD than in those without it (82% versus 3%; P < 0.0001). Thus, the sensitivity and specificity of the ELISA with respect to rapidly progressive ILD were 82% and 97%, respectively.
In the present study, we identified MDA-5 as the autoantigen recognized by the anti–CADM-140 antibody. MDA-5 is involved in innate immune defense against viruses, cellular growth suppression, and apoptosis (20, 21). It senses intracellular viral infection and triggers innate antiviral responses, including the production of type I IFNs, which further up-regulate MDA-5 to suppress viral replication and modulate the subsequent adaptive immunity (22). MDA-5 is unique among the autoantigens targeted by DM sera in terms of its cellular localization and function, because the other DM-specific autoantigens, such as Mi-2, p155, MJ, and small ubiquitin-like modifier 1 activating enzyme, are nuclear proteins involved in transcriptional or translational regulation (23–28).
Viral infection has long been thought to be one of the potential etiologies for idiopathic inflammatory myopathies (29–31). Patients with C-ADM and/or rapidly progressive ILD usually experience an acute onset and a course that is resistant to immunosuppressive treatment; however, the recurrence rate is low if and when the patient recovers, suggesting a possible association between this disease subset and infection. Interestingly, this clinical profile is clearly different from those of other forms of PM/DM, which often show chronic and recurrent clinical courses.
Given the critical role of MDA-5 in the innate immune defense against viruses, one hypothesis is that the production of anti–MDA-5 antibodies is an epiphenomenon during virus infection that is associated with the onset of C-ADM and rapidly progressive ILD; namely, infection of the skin and lung epithelium by a certain virus may up-regulate the expression of MDA-5 in the infected tissues. A subsequent antiviral immune response would induce apoptosis of the infected cells through cytotoxicity, resulting in the release of a large amount of proteolytic fragments of MDA-5 and/or a complex of virus and MDA-5. This could trigger the autoimmune response to MDA-5. A similar scenario has been shown to occur for the proteolytic fragments of histidyl transfer RNA synthetase generated by granzyme B during lymphocyte cytotoxicity, which induces an autoantibody response (32). In addition, this autoimmune response to MDA-5 may accelerate ongoing tissue damage through autoantibody-dependent or independent cytotoxic mechanisms. However, this hypothesis remains speculative, and further investigation is necessary.
The autoantibodies in sera from patients with C-ADM recognized MDA-5 but failed to bind RIG-I or LGP-2, indicating that their autoimmune response to MDA-5 was highly specific. MDA-5 and RIG-I recognize distinct viral RNAs, the distinguishing features of which are the presence or absence of a 5-triphosphate at the end of the viral RNA and the length of the viral double-stranded RNA (17). Because picornavirus is one of the viruses specifically recognized by MDA-5, it is possible that this virus is involved in the anti–MDA-5 antibody response as well as the pathogenesis of C-ADM and rapidly progressive ILD. In this regard, Christensen et al have shown that antibodies to coxsackievirus B, one of the picornavirus species, are significantly more frequent in patients with juvenile DM than in those with juvenile idiopathic arthritis or healthy children (30).
Our ELISA using rMDA-5 was specific for the detection of the anti–CADM-140 antibody and thus makes it possible to measure this antibody easily and rapidly in clinical settings. Detection of the anti–CADM-140 antibody by ELISA would be useful for diagnosing C-ADM as well as for predicting rapidly progressive ILD. This emphasizes the importance of measuring the anti–CADM-140 antibody routinely to improve the outcomes in patients with C-ADM. However, the sensitivity of our ELISA was somewhat low in comparison with the gold standard IP assay. Sera that gave a false-negative result in our ELISA reacted with rMDA-5 in immunoblots and showed a faint reactivity in IP-based assays (data not shown). Therefore, the autoantibodies in these sera are likely to preferentially recognize an epitope(s) on denatured MDA-5 rather than on the native molecule. It would be possible to improve the sensitivity of our ELISA by enhancing the antigenicity of the MDA-5 preparation.
In summary, this study is the first to identify MDA-5 as the CADM-140 antigen, which is a major autoantigen targeted by the serum antibodies of patients with C-ADM and/or rapidly progressive ILD. This information is useful for elucidating the pathogenesis of C-ADM and rapidly progressive ILD and for developing convenient assays for detecting the anti–CADM-140 antibody.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kuwana had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Sato, Takashi Fujita, Kuwana.
Acquisition of data. Sato, Hoshino, Satoh, Tomonobu Fujita, Kawakami, Takashi Fujita.
Analysis and interpretation of data. Sato, Hoshino, Kuwana.
We thank Drs. Shinichi Inada, Kazuo Takahashi, and Yukie Yamaguchi for providing the anti–CADM-140–positive sera, Mutsuko Ishida for assisting in the cell preparations and IP assays, and Kazuhiko Yamamoto for assisting in the cDNA library screening.