Type 1 interferons and myositis

Recent studies suggest a mechanistic role for molecules induced by type 1 interferons in the pathogenesis of some forms of myositis. For dermatomyositis, evidence that these molecules injure myofibers seems especially strong. In the group of disorders known as polymyositis, the study of blood samples suggests a potential role. It is unknown what drives the sustained presence of type 1 interferon-inducible molecules in these diseases, as the type 1 interferons themselves have not been specifically detected along with their downstream biomarkers. Therapeutic development for blockade of IFNα is in progress aided by the identification of blood genomic biomarkers.


Introduction
Th e infl ammatory myopathies -including dermatomyositis (DM), inclusion body myositis, and polymyositis (PM) -are poorly understood autoimmune diseases aff ect ing skeletal muscle. Evidence regarding the significance of type 1 interferons to these diseases, especially DM, is reviewed in the present article, with much of this material recently discussed elsewhere [1,2].
Th e type 1 interferons are a class of molecules that include IFNα and IFNβ. After binding to the type 1 interferon receptor (IFNAR) on target cells, these cytokines can stimulate the transcription of a set of genes, the type 1 interferon-inducible genes. Proteins abundantly produced from these genes' transcripts -such as myxovirus resistance protein A, interferon-stimulated gene 15 (ISG15), and 2΄,5΄-oligoadenylate synthetase 1 -remain inside cells. Th ey normally function as defenses against viral infections through a variety of means, such as inhibiting viral transcription, translation, or assembly of viral nucleocapsids. It is possible that the chronic intracellular overproduction of these transcripts and proteins might be directly harmful to cells, such as muscle fi bers in myositis [3].

Myositis, pathology, and mechanisms
Th e varied forms of myositis have distinct clinical and pathological features (Figure 1), and probably involve distinct mechanisms of tissue injury. DM, in addition to clinical skin involvement, has two unique pathological features (perifascicular atrophy and endothelial cell tubuloreticular inclusions) that distinguish it from other muscle diseases. Inclusion body myositis has a unique clinical distribution of involvement, with substantial weak ness of the quadriceps and wrist and fi nger fl exors, as well as specifi c suggestive pathological features including rimmed vacuoles. Th e broad category of PM is mainly distinguished by a collection of otherwise individually non specifi c features. Although frequently lumped together, DM and PM probably involve entirely diff erent mechanisms of tissue injury.
Because type 1 interferons may have many eff ects on cells of the immune system, they may have roles in the varied immune responses present across multiple myositis subtypes. Yet studies to date suggest that only in DM are these molecules strongly and directly infl uencing molecular events in muscle.

Why focus on type 1 interferons in dermatomyositis mechanism?
Recognition that cytokines are present in myositis muscle biopsy samples began with immunohistochemical studies for cytokine proteins [4]. Th is approach is confounded by a number of technical and biological limitations [5], includ ing nonspecifi c immunoreactivity, transient expres sion of cytokines, and their low concentration. For these reasons, some investigators turned to examining cytokine mRNA transcripts in muscle homogenates [6]. Initial PCR-based studies of cytokine transcripts including IFNα and IFNγ generally found no myositis subtype-specifi c diff erences compared with nonmyositis muscle -except for granulocyte-macrophage colonystimulating factor, which was detected in 12 out of 15 myositis samples but in none out of 10 controls [6]. Many subsequent studies of cytokine transcripts and proteins (discussed in [7][8][9]) have reported variable and often confl icting results.

Abstract
Recent studies suggest a mechanistic role for molecules induced by type 1 interferons in the pathogenesis of some forms of myositis. For dermatomyositis, evidence that these molecules injure myofi bers seems especially strong. In the group of disorders known as polymyositis, the study of blood samples suggests a potential role. It is unknown what drives the sustained presence of type 1 interferoninducible molecules in these diseases, as the type 1 interferons themselves have not been specifi cally detected along with their downstream biomarkers. Therapeutic development for blockade of IFNα is in progress aided by the identifi cation of blood genomic biomarkers.
Because of the potential for transient expression and the low concentration of cytokines, downstream persisting eff ects of cytokines have been sought. Two types of biomarkers (macromolecular and molecular) provide strong evidence that DM muscle has experienced strong signaling of the type 1 interferon receptor. More than 25 years ago, tubuloreticular inclusions (also known as lupus inclusions) -macromolecular structures commonly visible with electron microscopy in DM muscle endothelial cells [10,11] and rarely seen in other forms of myositis -were recognized as downstream markers of type 1 interferon signaling. Tubuloreticular inclusions in circulating blood cells develop in patients treated with IFNα [12,13] and those in cultured endothelial cells and other cells develop directly in response to IFNα and IFNβ [14][15][16], but not IFNγ [13]. For uncertain reasons, no PubMed indexed publication made a connection between this tubuloreticular inclusion litera ture and DM over 20 years [17].
Over the past 8 years the marked overproduction of type 1 interferon-inducible transcripts and proteins in muscle has been found to be remarkably unique to DM in comparison with all other muscle diseases studied [2,18,19]. Microarray gene expression studies of muscle biopsy specimens measuring approximately 18,000 transcripts in each of 113 muscle biopsy samples from patients with a wide range of myopathies showed that only DM samples with perifascicular atrophy have marked elevation of type 1 interferon-inducible trans cripts ( Figure 2a) [2]. In analyses combining publicly available data, the remarkable specifi city of these trans cript abundances for DM is impressive. For example, the transcript for the type 1 interferon-inducible gene ISG15 was higher in muscle in all 28 biopsies from adults with DM and perifascicular atrophy and from children with juvenile DM than in every one of 199 non-DM biopsy samples from a wide range of neuromuscular diseases (Figure 2b).
Two type 1 interferon-inducible proteins have similarly been shown to be highly specifi c biomarkers of DM muscle. Myxovirus resistance protein A is impressively and uniquely (in comparison with other muscle diseases) abundant in DM myofi bers with perifascicular atrophy and in DM capillaries ( Figure 3) [18]. ISG15, a ubiquitinlike modifi er, is furthermore attached to many other proteins in DM muscle, the identities of which have not been determined (Figure 2c). Exposure of human skeletal muscle cell cultures to IFNα or IFNβ produces a similar picture of ISG15 conjugation present in human DM samples ( Figure 2c) [2]. DM is a systemic disease, involving muscle, skin, and, variably, other tissues. Skin gene expression profi ling, reported only in abstract format to date [20], has similarly shown marked abundance of type 1 interferon-inducible transcripts. Th e topology of keratinocyte injury in DM skin is similar to that of myofi ber injury in DM muscle [21]. Patients with dermatological features of DM who lack signifi cant clinical evidence of muscle involvement have been classifi ed as clinically amyopathic DM. Autoantibodies to a classic type 1 interferon-inducible protein IFIH1 (inter feron induced with helicase C domain; also called MDA-5) have been recently identifi ed [22]. Using optimized cutoff values in an ELISA assay, the presence of anti-IFIH1 antibodies in clinically amyopathic DM among 262 patients with a range of connective tissue diseases was 69% sensitive and 99.6% specifi c. Signifi cant anti-IFIH1 autoantibody levels were present in 22 out of 32 patients with clinically amyopathic DM, but only in one of 35 patients with classic DM and in none of 53 patients with PM. Th ese remarkably strong data both indicate a clinically valuable biomarker of clinically amyopathic DM and provide mechanistic evidence for some abnormality related to type 1 interferons in clinically amyopathic DM. Th e nature of this relationship is uncertain; one speculation is that IFIH1, a nuclear RNA helicase, is overproduced or altered in some way in clinically amyopathic DM, and evokes an autoantibody response.

A blood type 1 interferon-inducible signature
Blood gene expression profi ling has also demonstrated marked abundance of these transcripts in patients with active DM, such as untreated patients, but also in PM (see below) [19]. One study that did not fi nd marked type 1 interferon-inducible transcript abundance in DM blood samples had included almost only treated patients (11 out of 12 patients receiving prednisone; eight of these patients receiving an additional second immuno sup pressant agent) [23]. Microarray experiments measure the abundance of 10,000s of transcripts simultaneously; any set of transcripts may be called a signature, but what is impressive about such experiments in DM and PM blood samples is the dominance of these gene expression patterns by type 1 interferon-inducible genes. In a study of 23 patients with DM and PM, at least 24 of the highest expressed 25 genes among approximately 38,000 genes studied are all known to be highly inducible by type 1 interferons [19].
Th e type 1 interferon-inducible transcript overproduction in DM and PM is highly correlated, within individual patients, with clinical measures of disease activity [19] ( Figure 4 and unpublished data). Although the absolute magnitude of the upregulation of this signature does not highly correlate with disease severity across patients, within individual patients the signature does track disease activity -a situation similar to the commonly used clinical biomarker creatine kinase.

What drives the production of these downstream biomarkers of type 1 interferon signaling in DM muscle?
A fundamental question that has not been answered is what drives the production of type 1 interferon-inducible molecules in DM muscle. Specifi c elevation of type 1 inter feron transcripts or proteins has not been demonstrated in DM muscle. Both microarray and real-time quantitative PCR do not show impressive diff er ences, compared with other myositis samples, in transcripts encoding a range of IFNα subtypes or IFNβ in the same DM muscle samples that have marked upregulation (10-fold to 100-fold) of downstream type 1 interferoninducible transcripts (unpublished data). Immunoblots from myositis samples similarly do not appear to show diff erential presence of IFNα or IFNβ protein in DM muscle (preliminary unpublished data).
Th is situation parallels that seen in systemic lupus erythematosus. Although studies performed almost 30 years ago detected molecules believed to be IFNα in 60 to 76% of systemic lupus erythematosus blood samples using functional antiviral assays sometimes in combination with neutralizing antibodies [24][25][26][27], the literature has been notable for the absence of detection of IFNα in systemic lupus erythematosus blood or tissue samples by direct methods such as ELISA, immunoblot, or mass spectrometry. For example, one study found measurable levels of IFNα protein by ELISA in only two out of 38 patients, while most of these same 38 samples showed marked increases in type 1 interferon-inducible transcripts [28]. Th e lack of direct detection of IFNα protein has been attributed to potential technical limitations, although unexpected results in science have often been assumed to be erroneous. A more recent functional assay looking at type 1 interferon-inducible transcription has similarly detected activity in systemic lupus erythematosus plasma [29].
Th e interpretation of the results of functional assays is complicated by potential type 1 interferon autocrine mechanisms. In mouse cells, autostimulation of the IFNAR by early secreted type 1 interferons results in marked amplifi cation of IFNα production [30][31][32][33]. Anti-IFNα antibodies used in functional assays to neutralize sample IFNα could potentially diminish type 1 inter feroninducible gene transcription through neutralizing early secreted reporter cell IFNα, although this possibility is speculative.
For juvenile DM, the functional assay for gene transcription has been used for detection of serum type 1 interferon-inducing activity and similarly interpreted as indicating the presence of IFNα in some blood samples [34]. What is particularly remarkable about this study was the marked range of interferon-inducing activity of serum from healthy children and adults, which varies by over 100-fold and includes many healthy people whose activity exceeded the mean value for the juvenile DM population.
Th ese data suggest that the marked production of type 1 interferon-inducible transcripts and proteins in Greenberg Arthritis Research & Therapy 2010, 12(Suppl 1):S4 http://arthritis-research.com/content/12/S1/S4 DM muscle, probably by myofi bers, might result from sustained activation of the type 1 interferon receptor IFNAR in the absence of excessive (compared with the wide range of normal) type 1 interferons, or through mechanisms even further downstream that bypass IFNAR. Th e most natural interpretation of the data to date suggests that what may turn out to be most crucial with regard to DM myofi ber injury is not the abundance of a type 1 interferon, but rather sustained abnormal function of the IFNAR or a further downstream process.

Potential role for type 1 interferons in polymyositis and inclusion body myositis
PM is an umbrella term for patients with various forms of myositis that are diffi cult to classify. DM and PM have substantial diff erences with regard to abundance of type 1 interferon-inducible molecules in muscle biopsy samples ( Figure 5), yet marked overexpression of type 1 interferon-inducible genes has also been found in blood in PM [19]. Within PM muscle, infl ammatory cells typically surround, displace, and sometimes invade muscle fi bers. Th ese cells include T cells, myeloid dendritic cells, macrophages, and plasma cells (reviewed in [3]). Type 1 interferons have multiple eff ects on these cell types, and it is possible that through these eff ects the type 1 interferon system is contributing to PM myofi ber injury.
Patients with inclusion body myositis -a highly infl ammatory disorder of muscle, as judged by abundance of immune system cells and transcripts in muscle -do not have high levels of muscle or blood type 1 interferon-inducible transcripts (Figure 5), although a small propor tion of patients may have modest elevation of such transcripts in blood alone. As in PM, the mechanistic interpretation of blood expression of these transcripts is uncertain, and could refl ect less-specifi c eff ects driving immune cell development.

Predicted exacerbations with TNFα inhibition
TNFα appears to have an antagonistic relationship with type 1 interferons [35]. It may directly inhibit the generation of plasmacytoid dendritic cells from progenitor cells and may inhibit plasmacytoid dendritic cell production of type 1 interferons. Studies of etanercept in Sjogren's syndrome indeed showed that this drug increased type 1 interferon activity [36]. Accordingly, models that propose a signifi cant role for type 1 interferons in the pathogenesis of myositis predict that TNFα inhibition might exacerbate myositis. Published experience with TNFα inhibition in patients with myositis appears to support this model. Two open-label studies of infl iximab have been terminated before completion or had substantial dropout rates for reasons that included disease progression [37,38]. Although this class of drugs may prove useful in the management of some patients, currently it appears unlikely to be of more general use for myositis.

Conclusion: diagnostics and therapeutic development
Th e presence of marked overproduction of type 1 interferon-inducible transcripts in blood specimens from patients with active DM and PM has potential for diagnostic use [19]. Th ese biomarkers may be able to distinguish these disorders from inclusion body myositis and other muscle diseases that sometimes present diagnostic uncertainty. Furthermore, they may be useful for therapeutic development. A phase 1b trial of anti-IFNα therapy has been initiated in DM and PM [39]. Entry criteria into this study include the presence of suffi ciently high type 1 interferon-inducible gene expression in blood. Future studies targeting the IFNAR or further down stream events have strong rationale in DM, and perhaps in PM. Competing interests SAG has worked as a consultant regarding clinical trial planning for MedImmune, LLC and has a Sponsored Research Agreement with MedImmune, LLC. SAG is an inventor of intellectual property pertaining to myositis diagnostics. SAG has National Institutes of Health funding and the publication of articles constitutes evidence of productivity that may be used to support future requests for National Institutes of Health funding. Figure 5. Distinct muscle expression of type 1 interferoninducible genes in infl ammatory myopathies. Distinct muscle expression of type 1 interferon-inducible genes in dermatomyositis (DM) compared with polymyositis (PM) and inclusion body myositis (IBM). Muscle microarray data shown for 20 patients (fi ve each with active dermatomyositis (DMA), active polymyositis (PMA), untreated IBM, and normal) with plotted mean values and error bars for mean ± standard error for each group. Highly expressed genes in DM muscle are orders of magnitude greater than in PM and IBM. Adapted from [19] with permission.