https://orcid.org/0000-0003-0664-9751
Abstract
Dermatomyositis is an acquired auto-immune disease characterized by skin lesions and muscle-specific pathological features such as perifascicular muscle fibre atrophy and vasculopathy. Dermatomyositis patients display an upregulation of type I interferon-inducible genes in muscle fibres, endothelial cells, skin and peripheral blood. However, the effect of type I interferon on muscle tissue has not yet been determined. Our aim was to study the pathogenicity of type I interferon in vitro and to evaluate the efficacy of the type I interferon pathway blockade for therapeutic purposes. The activation of type I interferon in differentiating myoblasts abolished myotube formation with reduced myogenin expression while in differentiated myotubes, we observed a reduction in surface area and an upregulation of atrophy-associated genes. In vitro endothelial cells exposure to type I interferon disrupted vascular network organization. All the pathogenic effects observed in vitro were abolished by ruxolitinib. Finally, four refractory dermatomyositis patients were treated with ruxolitinib and improvement ensued in skin lesions, muscle weakness and a reduced serum type I interferon levels and interferon-inducbile genes scores. We propose JAK inhibition as a mechanism-based treatment for dermatomyositis, a finding that is relevant for the design of future clinical trials targeting dermatomyositis.
Introduction
Idiopathic inflammatory myopathies are a heterogeneous group of disabling auto-immune diseases (Dalakas, 2015). Whereas the underlying pathogenic mechanism are different (Greenberg et al., 2002; Greenberg and Amato, 2004), they are usually treated with corticosteroids in combination with immunosuppressants (Dalakas, 1991, 2011; Dalakas and Hohlfeld, 2003; Ernste and Reed, 2013; Luo and Mastaglia, 2015). While receiving this conventional treatment, patients frequently relapse, sometimes developing a refractory disease course, which may lead to muscle and skin damages (Dalakas et al., 1993; Mastaglia and Zilko, 2003; Dalakas, 2010). Therefore, specific mechanism-based treatments are needed.
Among the group of idiopathic inflammatory myopathies, dermatomyositis is defined by a specific skin rash and characteristic pathological changes in skeletal muscle (Zong and Lundberg, 2011). The morphological features are perifascicular atrophic muscle fibres and overexpression of the major histocompatibility complex class I, with a vasculopathy characterized by deposition of complement C5b-9 on capillaries, endothelial cell hyperplasia with tubuloreticular inclusions and focal loss of capillaries (Lahoria et al., 2016). Although the precise relation between atrophy and vasculopathy is not fully understood, it has been hypothesized that the perifascicular atrophy may be related to local hypoxia (Pestronk et al., 2010; Preuße et al., 2016; De Luna et al., 2017) suggesting that common factors may participate to both of these features.
Transcriptomic studies carried out on skeletal muscle biopsies from dermatomyositis patients have shown a specific upregulation of multiple interferon-stimulated genes suggesting that type I interferons (IFN-I) play an important role (Greenberg et al., 2005; Suárez-Calvet et al., 2014). The expression of some interferon-stimulated genes (Schneider et al., 2014), such as MXA, ISG15 and DDX58 (also known as RIG-I), has been confirmed at the protein level in perifascicular regions and on the capillaries of the muscle biopsies (Salajegheh et al., 2010; Suárez-Calvet et al., 2014, 2017; Uruha et al., 2017). Dermatomyositis patients also have high levels of circulating IFN-I cytokines including IFN-β (Liao et al., 2011) and IFN-α (Niewold et al., 2009), and the disease activity is correlated with interferon-stimulated genes transcript levels in the blood (Greenberg et al., 2012).
In humans, there are five different types of IFN-I: IFN-α, IFN-β, IFN-ɛ, IFN-κ and IFN-ω (Trinchieri, 2010). They are recognized by heterodimeric receptor complexes, comprising IFN-α receptor (IFNAR1 and IFNAR2) subunits that transduce signals to the nucleus by the JAK/STAT complex resulting in the upregulation of hundreds of different interferon-stimulated genes, including IFN-I cytokines, involved in anti-viral defence (Ivashkiv and Donlin, 2014). While the IFN-I pathway has been implicated in the pathophysiology of dermatomyositis, its role in muscle and skin damage remains unknown. This point is crucial in the light of new developments for targeted treatment in dermatomyositis, such as Janus kinase (JAK) 1 and 2 inhibitors.
Materials and methods
Myoblasts, myotubes and HMEC-1 culture
All of the procedures were carried out in accordance with the French legislation on ethical rules. Human primary skeletal muscle cells were isolated from the quadriceps muscle of adult donor (53 years old) without muscular disorders, as previously reported (Edom et al., 1994). In all experiments the cells were first expanded and enriched in myoblasts by bead sorting (CD-56/NCAM, Miltenyi Biotec) to reach a myogenic purity of more than 95% as confirmed by desmin immunostaining (Dako). Cultures of human primary skeletal muscle cells were analysed during myoblasts differentiation and differentiated myotubes. Proliferating myoblasts (5 × 104 cells) were cultured in medium consisting of 199 Medium and Dulbecco’s modified Eagle medium (DMEM) in a 1:4 ratio (Life Technologies) supplemented with 20% foetal calf serum (Life Technologies), 5 ng/ml human epithelial growth factor (Life Technologies), 0.5 ng/ml bFGF, 50 mg/ml fetuin (Life Technologies) and 5 mg/ml insulin (Life Technologies) for 24 h in 5% CO2 at 37°C until 80% confluence according to Bertrand et al. (2014). To induce myoblast differentiation and myotube formation the medium were switched to serum-free DMEM supplemented with 50 µg/ml gentamycin and incubated at 37°C in 5% CO2 for 72 h. To evaluate the effect of IFN-I on differentiated myotubes, differentiated cells were stimulated for 48 h with IFN pathway activators or suppressors (see below). Human dermal microvascular endothelial cells-1 (HMEC-1; ATCC) were cultured in MCDB-131 (Coring Life Science) containing 10% FBS, 10 ng/ml EGF, 1 µg/ml hydrocortisone and 2 mM l-glutamine. We performed tube formation assays (in vitro angiogenesis) with HMEC-1 in the presence of IFN-I. HMEC-1 cells were seeded at 6.5 × 104 cells/cm2 on Millicell® angiogenesis slides (Millipore) containing ECMatrix™ gel following the manufacturer’s instructions.
In vitro IFN-I pathway activation and inhibition
For IFN-I pathway activation myoblasts and/or myotubes were either: treated once with the TLR3 ligand Poly(I:C) high molecular weight at 50 µg/ml (InvivoGen) which is a strong stimulant of IFN-I pathway, considered as a synthetic analogue of double-stranded RNA (Junt and Barchet, 2015), or treated daily with recombinant human type I interferon (IFN-I, 103 U/ml, PBL Interferon Source), recombinant human interferon beta 1a (IFN-β, 103 U/ml, PBL), or recombinant human interferon alpha 2a (IFN-2α, 103 U/ml, Roche). As a positive control for muscle atrophy we used dexamethasone (10 µg/ml) (Menconi et al., 2008). To analyse the effect of IFN-I on angiogenesis, IFN-I 1000 IU/ml was used in the tube formation assay. As a control, we used d, l-sulforaphane at 20 µM, an inhibitor of the tube formation ability of endothelial cells (Bertl et al., 2006). For IFN-I pathway inhibition before adding IFN-I stimuli, myoblasts and myotubes were preincubated with anti-human interferon alpha/beta receptor chain 2 (IFNAR, Clone MMHAR-2, 5 µg/ml, PBL), or with anti-human interferon alpha (anti-IFN-α, Clone MMHA-11, 5 µg/ml, PBL), or with anti-human interferon beta (anti-IFN-β, Clone MMHB-3, 5 µg/ml, PBL). As a control, purified mouse IgG1 isotype control (IgG1, 5 µg/ml, BD Pharmigen) or mouse IgG2a isotype control (IgG2, 5 µg/ml, BD Pharmigen) were used. We preincubated IFN-activated muscle cells or HMEC-1 cells with ruxolitinib (1 mM or 5 mM in DMSO, respectively) (Selleckchem) to inhibit JAK1/2 signalling pathway (Zhou et al., 2014).
IFN-I level measurements
To measure IFN concentrations in the culture supernatant a bioassay was used which measures its protective action towards viral-induced cytopathic effect using Madin-Darby bovine kidney (MDBK) cells and vesicular stomatitis virus as the challenge virus as described previously (Palmer et al., 2000). Myoblasts and myotubes were stimulated for 30 min, then new medium was added to wells and cells were incubated overnight. IFN-α was detected with an assay using 500 µl of supernatant. Titres are expressed in international units (IU/ml) as the reciprocal dilution that inhibited the formation of 50% of the plaques. Results are expressed relative to international reference samples titrated in the same conditions. The minimal titre of IFN-α detectable with this assay was 2 IU/ml.
The SIMOA assay was performed to measure IFN concentration in the patients serum as recently described (Rodero et al., 2017). Two IFN-α specific antibodies [isolated and cloned from two patients with APS1/APECED (autoimmune polyglandular syndrome type 1/autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy) described previously (Meyer et al., 2016)] were used. The 8H1 antibody clone was used as a capture antibody after coating on paramagnetic beads (0.3 mg/ml), and the 12H5 was biotinylated (biotin/antibody ratio = 30/1) and used as the detector. The limit of detection was calculated by the mean value of all blank runs + 3 standard deviations (SDs) and was 0.23 fg/ml. The results are expressed in fg/ml.
Immunofluorescence
Myotubes were fixed with absolute ethanol, then the wells were washed and kept 4°C in phosphate-buffered saline (PBS) until the staining. To perform the staining, cells were permeabilized (Triton 0.5%), washed and incubated with blocking buffer (2% FBS/PBS) for 30 min at room temperature. After that the cells were incubated with the specific primary antibody (1:50) for anti-myosin heavy chain (hybridome, clone MF20), anti-myogenin (Dako), anti-HLA-ABC (Dako) and desmin (Dako), for 1 h at room temperature. The primary antibody was revealed using conjugated anti-secondary antibody and the nuclei were stained with DAPI (Sigma). The 25 sequentional images were obtained using an inverse fluorescent microscopy Axio Observer.A1 and AxioCam Camera (Carl Zeiss).
Image analysis
To assess the effect of IFN-I pathway activation on myoblast fusion and differentiation, the fusion index was measured by counting the number of nuclei inside the myotubes as a percentage of the total number of nuclei in the image. In addition, myogenin expression was measured by counting the positive nuclei for myogenin as a percentage of the total number of nuclei in the image. In both cases, the percentages were determined after counting more than 1500 nuclei and 16 images per condition. To assess the effect of IFN-I pathway activation on myotube surface area all 25 pictures obtained from myotubes immunostained with MF20 were measured for the total surface area (Arouche-Delaperche et al., 2017). The counting was performed using the Cell Profiler® software and ImageJ software. To analyse HMEC-1 tube formations, five pictures/well were taken every hour during 10 h with an inverted Nikon microscope. The total segment length, the number of meshes and the number of junctions were automatically measured using the Macro ‘Angiogenesis Analyzer’ (Gilles Carpentier. Contribution: Angiogenesis Analyzer, ImageJ News, 5 October 2012) for NIH ImageJ 1.47v (DeCicco-Skinner et al., 2014).
Real-time PCR analysis
Myotubes and HMEC were processed and RNA was extracted using the RNeasy® Plus Mini Kit following the manufacturer’s instructions (Qiagen). RNA concentration was determined using a nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific), and cDNAs were synthesized from 1 µg of total RNA using GoScript™ reverse transcription system and random primers (Promega). The synthesized cDNAs were used as templates for quantitative real-time PCR (LightCycler 480 II, Roche) using human MX1/MxA, OAS1, HLA-ABC, FBXO32/atrogin and TRIM63/MURF-1 primers for muscle cells (Supplementary Table 3). The RNAs expression was normalized with emerin expression. The mRNA quantification on HMEC-1 cells was performed using TaqMan® fluorogenic probes for RIG-I/DDX58 (Hs00204833_m1), ISG15 (Hs00192713_m1), TLR3 (Hs01551078_m1), VEGF (Hs00900055_m1), MX1 (Hs00895608_m1) and TGFB1/TGFβ (Hs00998129_m1) and GAPDH (Hs99999905_m1) as endogenous control (Applied Biosystems). The change in the relative expression of each gene was calculated using ΔΔCt formula using non-stimulated cells as a control.
RNA was extracted from peripheral blood mononuclear cells (PBMCs) with the RNAqueous-Micro kit (Life Technologies) in accordance with manufacturer’s guidelines. Reverse transcription was performed with a High-Capacity cDNA Reverse Transcription Kits (Thermo Fisher Scientific). Reactions were set-up in duplicate using TaqMan® Fast Universal PCR (AB Thermo). Samples were normalized 18S (Hs99999901_s1) expression level. The levels of transcript from the following genes were studied: IFI44L (Hs00199115_m1), IFIT1 (Hs00199115_m1), ISG15 (Hs00192713_m1), IFI27 (Hs01086370_m1), SIGLEC1 (Hs00988063_m1) and RSAD2 (Hs01057264_m1).
Histological and immunohistochemistry in human muscle biopsies
Frozen samples from six patients defined on the basis of ENMC 2003 criteria were analysed in the Department of Neuropathology (Charité-Universititätsmedizin, Berlin, Germany), following standardized procedures. Cryostat sections (8-µm thickness) of muscle biopsy specimens were used. Immunohistochemical analysis included the following antibodies: anti-FBXO32 (Abcam), anti-MXA (Santa Cruz), anti VEGF (Santa Cruz), and anti-ISG15 (Abcam). An automated slide staining system (BenchMark XT; Ventana Medical Systems) was used. Double immunofluorescence was performed as previously described (Preuße et al., 2016).
Patient’s characteristics and treatment with ruxolitinib
Four dermatomyositis patients with refractory disease were treated with ruxolitinib (Jakavi®, Novartis®), up to 40 mg per day for 3 months. dermatomyositis was diagnosed according to European Neuromuscular Centre (ENMC) criteria (Hoogendijk et al., 2004) and was considered refractory because the disease remained active after initiating two different immunosuppressive therapy combined with corticosteroids with or without intravenous immunoglobulins (IVIgs). Disease activity was assessed using the Medical Research Council scale (MRC-5) (Compston, 2010) and the Cutaneous Dermatomyositis Disease Area and Severity Index (CDASI) (Rider et al., 2011; Anyanwu et al., 2015). Written informed consent was obtained from patients to obtain blood samples, analyse the disease activity and evaluate skin damage.
Statistical analysis
All experiments were performed at least three times. Statistical analyses were performed with GraphPad Software (Graphpad Software, La Jolla, CA, USA). Values are expressed as means ± standard error of the mean (SEM) of triplicate. Comparisons were carried out using one-way ANOVA followed by Bonferroni tests, t-test and Spearman correlations. Differences were considered significant when P ≤ 0.05.
Results
Transcriptomic and proteomic analysis in dermatomyositis muscle showed a correlation between IFN-I pathway and vascular remodeling
We hypothesized that IFN-I pathway activation could be deleterious for both muscle and vascular cells. Microarray datasets of dermatomyositis patients were analysed, looking for related angiogenesis genes and links with interferon-stimulated genes. We observed a significant upregulation of genes involved in the cellular response to hypoxia and angiogenesis (Supplementary Table 1). The expression of interferon-stimulated genes and angiogenesis-related genes are represented in the heatmaps (Supplementary Fig. 1A and B). These results were confirmed by proteomic studies performed on muscle biopsies from dermatomyositis patients (Supplementary Fig. 1C and Supplementary Table 2). We observed a strong correlation between the level of transcripts encoding for interferon-stimulated genes, with angiogenesis proteins and vascular damage (Supplementary Fig. 1D and E) suggesting that IFN-I may induce vascular damage.
The activation of IFN-I pathway in muscle cells induces myotubes atrophy and IFN-α production
In light of the previous results, the pathogenic effect of IFN-I was first tested on muscle cells. We performed IFN-I pathway activation on differentiated human myotubes using polyinosinic-polycytidylic acid [poly(I:C)] (PIC) and different recombinant IFN-I (IFN-I, IFN-2α and IFN-β). We observed a reduction in the size of the myotubes incubated with all of the IFN-I inducers compared to control myotubes (Fig. 1A). The quantification of myotube surface area confirmed the significant reduction in myotube size (Fig. 1B). These differences were due to a real process of muscle atrophy as confirmed by the significant upregulation of the two key muscle atrophy genes TRIM63/MurF-1 and FBXO32/atrogin in the myotubes treated with IFN-I pathway activators (Fig. 1C and D). The overexpression of TRIM63/MurF-1 and FBXO32/atrogin was similar to that observed in dexamethasone-treated myotubes. In addition, these atrophic human myotubes also harbored a significant upregulation of HLA-ABC molecules similarly to muscle fibres in dermatomyositis muscle biopsies (Fig. 1E and Supplementary Fig. 2). Of note, in vitro, the myotube response to IFN-I pathway stimuli, was confirmed by the upregulation of interferon-stimulated genes such as MX1/MxA and OAS1 (Fig. 1F and G).
The activation of the IFN-I pathway induces muscle atrophy in vitro. (A) Differentiated myotubes non-stimulated (control) or stimulated with PIC, IFN-I, IFN-2α, IFN-β or dexamethasone (DEX) were stained with anti-myosin heavy chain antibody MF20 (green) and Hoechst (blue). (B) Myotube surface was quantified by CellProfiler and ImageJ. Quantitative PCR analysis of (C) TRIM63/MURF-1, (D) FBXO32/atrogin, (E) HLA-ABC, (F) MX1/MxA and (G) OAS1 mRNA were performed and normalized with emerin (EMD). (H) IFN-α secretion was analysed by bioassay on myotube supernatants. Data shown as mean ± SEM of triplicate wells and are representative of three different experiments. *P ≤ 0.05, **P ≤ 0.001, ***P ≤ 0.0001, ****P ≤ 0.0001.
The ability of myotubes to produce IFN-I after IFN-I pathway activation was also tested. PIC, IFN-I and IFN-2α induced a significant secretion of IFN-α (Fig. 1H).
Next, we tested the effect of IFN-I pathway activation on muscle repair and differentiation. We observed a dramatic reduction in myotube formation and myogenin expression (myogenin being a transcription factor important for myogenic commitment) after PIC exposure, as well as a mild reduction after IFN-I, IFN-2α and IFN-β stimulation (Supplementary Fig. 3).
To confirm the central role of IFN-I, we observed that IFNAR blocking abolished the pathogenic effect of IFN-I activation in myotubes and myoblasts (Supplementary Fig. 4A and 5A). Similarly, anti-IFN-α and anti-IFN-β neutralizing antibodies decreased these pathogenic effects (Supplementary Fig. 4B and 5B). Together, these results demonstrate the pathogenicity of IFN-I pathway activation that recapitulates in vitro the dermatomyositis muscle fibre damage. Moreover, IFN-I pathway activation impairs muscle repair.
The activation of IFN-I pathway impairs endothelial cells angiogenesis
Next, the effect of IFN-I pathway activation on angiogenesis was assessed. We observed that IFN-I reduced in vitro the formation of pseudocapillaries compared to untreated endothelial cells (Fig. 2A). Time-lapse tube formation analysis showed a sustained reduction in the total segment length, in the number of meshes and in the number of junctions in endothelial cells treated with IFN-I compared to untreated cells after 10 h of culture (Fig. 2B–D).
The effect of IFN-I on angiogenesis in HMEC-1 cells. (A) Time-lapse tube formation assay was performed using HMEC-1 cells cultured on matrix gel-coated wells without or with the presence of IFN-I. (B) The total segment length, (C) the number of junctions and (D) the number of meshes were analysed for up to 10 h of experiment. HMEC-1 cells were treated with PIC, IFN-I, IFN-2α, IFN-β and the quantification of (E) MX1/MxA, (F) ISG15, (G) TLR3, (H) DDX58 (RIG-I), (I) VEGF and (J) TGFB1 mRNA was performed by qPCR. Data shown as mean ± SEM of triplicate wells and are representative of three different experiments. *P ≤ 0.05, **P ≤ 0.001, ***P ≤ 0.0001, ****P ≤ 0.00001.
A significant upregulation of interferon-stimulated genes (MX1/MxA, ISG15, TLR3 and DDX58/RIG-I) was observed in endothelial cells treated with the different inducers of the IFN-I pathway (Fig. 2E–H). To investigate the mechanisms involved in angiogenesis further, we analysed the expression of VEGF and TGFB1, both known to trigger vascular endothelial cell growth (Massagué et al., 2000; Ferrara, 2004; Presta et al., 2005). Both genes were significantly upregulated by PIC but not by IFN-I, IFN-2α or IFN-β treatment (Fig. 2I and J). These results demonstrate the pathogenic effect of IFN-I at the vascular level in vitro reproducing the same pathogenic findings observed in both dermatomyositis muscle and skin.
In vivo, FBXO32/atrogin are expressed in MX1/MxA-positive muscle fibres and vessels co-express VEGF and MX1/MxA
As the in vitro studies showed a link between IFN-I and muscle atrophy and between IFN-I and vasculopathy, we aimed to confirm these observations in vivo. We found that FBXO32/atrogin was indeed expressed in the perifascicular atrophic muscle fibres that were also MxA-positive (Fig. 3A and B). We also confirmed that the atrophic fibres in the perifasicular area that were also positive for MxA and were also the smallest diameter fibres. At the vascular level we found muscle capillaries positive for VEGF that were also positive for MxA. (Fig. 3C). Overall our results reinforce the idea of the pathogenicity of IFN-I in dermatomyositis and opens the opportunity for the use of IFN-I based-mechanism treatments for dermatomyositis patients.
Immunohistochemical analysis in muscle biopsies of dermatomyositis patients. (A) Atrogin is expressed in the same muscle fibres expressing (B) MxA located in perifascicular areas in serial sections of a representative dermatomyositis biopsy. (C) Double immunofluorescence showing the co-localization of MxA (green) and VEGF (red) in intramuscular capillaries in a dermatomyositis biopsy. Data are representative of three different patients.
In vitro JAK inhibition using ruxolitinib prevents the pathogenic effects of IFN-I in muscle and endothelial cells
Based on our observations that: (i) muscle cells are able to produce IFN-I; (ii) IFNAR plays a key role in IFN-I-induced damage; and (iii) that IFNAR activation induces the upregulation of interferon-stimulated genes, we tested in vitro the effect of JAK1/2 inhibition using ruxolitinib. Ruxolitinib treatment prevented IFN-β induced myotube atrophy as confirmed by the quantification of the myotube surface (Fig. 4A and B). Similarly, ruxolutinib-treated myoblasts previously exposed to IFN-β showed a significant increase in differentiation (Fig. 4C) as demonstrated by fusion index measurements (Fig. 4D). We also showed that ruxolutinib prevented the IFN-I induced damage of endothelial cells (Fig. 4E). Indeed the total segment length, the number of meshes and the number of junctions was normalized in the presence of ruxolitinib (Fig. 4F–H). These results demonstrate that in vitro ruxolutinib can reverse the pathogenic effect of IFN induced damage in both muscle and endothelial cells.
Ruxolutinib prevents IFN-I negative effects in myoblasts, myotubes and HMEC-1. (A) Myotubes preincubated with ruxolitinib (RUXO) were then incubated with IFN-β and stained for MF20 (green) and Hoechst (blue) to quantify (B) the myotubes surface area. (C) Myoblasts preincubated with ruxolitinib were then stimulated during differentiation with IFN-β and stained for MF20 (green) and Hoechst (blue) to quantify (D) the fusion index. Data shown as mean ± SEM of triplicate wells from one experiment. (E) Tube formation assay with HMEC-1 cells was performed in the presence of IFN-I or ruxolitinib+IFN-I to quantify (F) the total segment length, (G) the number of meshes and (H) the number of junctions at 10 h of culture using the angiogenesis analyser on ImageJ. Data shown as mean ± SEM of triplicate wells and are representative of three different experiments. All the controls were treated with DMSO, the vehicle for ruxolitinib. **P ≤ 0.001,***P ≤ 0.0001, ****P ≤ 0.00001.
Ruxolitinib treatment improves patients with refractory dermatomyositis
Based on our in vitro results, four refractory dermatomyositis patients were treated with ruxolitinib for 3 months (Table 1). They had previously received several immunosupressants/immunomodulatory agents (Table 1). All were considered refractory cases with, at least, persistent invalidating severe skin lesions. Patients 3 and 4 also presented muscle weakness with and without high creatine kinase level (only for Patient 3). Ruxolitinib was administered while maintaining low-dose prednisone ± IVIg. Following the initiation of ruxolitinib therapy, facial skin rash improved in all four patients (Fig. 5A–H), as did the CDASI scores. In the two patients presenting with clear muscle weakness (Table 1) muscle strength improved and the creatine kinase levels also decreased significantly in Patient 3. When performed, all patients demonstrated an improvement in their quality of life score (Table 1, health assessment questionnaire). Efficacy of ruxolitinib was also demonstrated biologically by reductions in serum IFN levels (Fig. 5I) and interferon-stimulated genes score in PBMCs (Fig. 5J).
Clinical, biological and histological descriptions of three patients before treatment and the improvement of all patients after 3 months treatment with ruxolitinib for clinical scores
| Patient | Age (years) | Sex | Disease duration (years) | Myostis- specific antibody | Previous treatments | CDASI activity score | HAQ score | MRC-5 (psoas) | CK level (IU/l) | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | ||||||
| 1 | 59 | F | 5 | Anti-TIF1γ | CS, HCQ, IVIg, PLEX, MTX, AZA | 26 | 15 | 0.875 | 0.75 | 4+ | 4+ | 65 | 183 |
| 2 | 79 | F | 4 | Anti-SAE | CS, MTX, IVIg, MMF | 27 | 7 | 0.875 | 0.375 | 4+ | 4+ | 79 | 181 |
| 3 | 84 | F | 1 | Anti-TIF1γ | CS, MTX, AZA, IVIg | 44 | 14 | 2.125 | 0.75 | 3 | 4 | 2559 | 226 |
| 4 | 45 | F | 6 | Anti-TIF1γ | CS, HCQ, MTX, AZA, MMF, RTX, CYC, IVIg | 40 | 15 | ND | ND | 3 | 4 | 41 | 21 |
| Patient | Age (years) | Sex | Disease duration (years) | Myostis- specific antibody | Previous treatments | CDASI activity score | HAQ score | MRC-5 (psoas) | CK level (IU/l) | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | ||||||
| 1 | 59 | F | 5 | Anti-TIF1γ | CS, HCQ, IVIg, PLEX, MTX, AZA | 26 | 15 | 0.875 | 0.75 | 4+ | 4+ | 65 | 183 |
| 2 | 79 | F | 4 | Anti-SAE | CS, MTX, IVIg, MMF | 27 | 7 | 0.875 | 0.375 | 4+ | 4+ | 79 | 181 |
| 3 | 84 | F | 1 | Anti-TIF1γ | CS, MTX, AZA, IVIg | 44 | 14 | 2.125 | 0.75 | 3 | 4 | 2559 | 226 |
| 4 | 45 | F | 6 | Anti-TIF1γ | CS, HCQ, MTX, AZA, MMF, RTX, CYC, IVIg | 40 | 15 | ND | ND | 3 | 4 | 41 | 21 |
AZA = azathioprine; CDASI = cutaneous Dermatomyositis Disease Area and Severity Index; CS = corticosteroids; CYC = cyclophosphamide; HAQ = health assessment questionnaire, HCQ = hydroxychloroquine; IVIg = intravenous immunoglobulin; MMF = mycophenolate mofetil; MRC-5 = medical Research Council scale graded on 5; MTX = methotrexate; ND = not done; PLEX = plasma exchange; RTX = rituximab.
Clinical, biological and histological descriptions of three patients before treatment and the improvement of all patients after 3 months treatment with ruxolitinib for clinical scores
| Patient | Age (years) | Sex | Disease duration (years) | Myostis- specific antibody | Previous treatments | CDASI activity score | HAQ score | MRC-5 (psoas) | CK level (IU/l) | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | ||||||
| 1 | 59 | F | 5 | Anti-TIF1γ | CS, HCQ, IVIg, PLEX, MTX, AZA | 26 | 15 | 0.875 | 0.75 | 4+ | 4+ | 65 | 183 |
| 2 | 79 | F | 4 | Anti-SAE | CS, MTX, IVIg, MMF | 27 | 7 | 0.875 | 0.375 | 4+ | 4+ | 79 | 181 |
| 3 | 84 | F | 1 | Anti-TIF1γ | CS, MTX, AZA, IVIg | 44 | 14 | 2.125 | 0.75 | 3 | 4 | 2559 | 226 |
| 4 | 45 | F | 6 | Anti-TIF1γ | CS, HCQ, MTX, AZA, MMF, RTX, CYC, IVIg | 40 | 15 | ND | ND | 3 | 4 | 41 | 21 |
| Patient | Age (years) | Sex | Disease duration (years) | Myostis- specific antibody | Previous treatments | CDASI activity score | HAQ score | MRC-5 (psoas) | CK level (IU/l) | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | Before treatment | After treatment | ||||||
| 1 | 59 | F | 5 | Anti-TIF1γ | CS, HCQ, IVIg, PLEX, MTX, AZA | 26 | 15 | 0.875 | 0.75 | 4+ | 4+ | 65 | 183 |
| 2 | 79 | F | 4 | Anti-SAE | CS, MTX, IVIg, MMF | 27 | 7 | 0.875 | 0.375 | 4+ | 4+ | 79 | 181 |
| 3 | 84 | F | 1 | Anti-TIF1γ | CS, MTX, AZA, IVIg | 44 | 14 | 2.125 | 0.75 | 3 | 4 | 2559 | 226 |
| 4 | 45 | F | 6 | Anti-TIF1γ | CS, HCQ, MTX, AZA, MMF, RTX, CYC, IVIg | 40 | 15 | ND | ND | 3 | 4 | 41 | 21 |
AZA = azathioprine; CDASI = cutaneous Dermatomyositis Disease Area and Severity Index; CS = corticosteroids; CYC = cyclophosphamide; HAQ = health assessment questionnaire, HCQ = hydroxychloroquine; IVIg = intravenous immunoglobulin; MMF = mycophenolate mofetil; MRC-5 = medical Research Council scale graded on 5; MTX = methotrexate; ND = not done; PLEX = plasma exchange; RTX = rituximab.
Ruxolitinib improves clinical and biological outcome in refractory patients with dermatomyositis. (A–H) Patients with refractory dermatomyositis before (left) and after initiating ruxolitinib treatment (30 mg/day) (right) demonstrated clear skin improvement. (I) IFN-α concentration in the patient’s serum was measured by single molecule array (Simoa) assay before and after treatment initiation. (J) IFN score measured from patients’ PBMCs before and after treatment initiation. *P ≤ 0.05.
Discussion
In this study, we have demonstrated in vitro that IFN-I reproduces the main dermatomyositis pathological findings, including muscle atrophy and vasculopathy. Importantly, we have demonstrated that these pathogenic effects are prevented using JAK inhibitor ruxolitinib in vitro and that this molecule also improved clinically, refractory dermatomyositis patients.
The presence of IFN-I in blood and muscle of dermatomyositis patients has previously been reported as well as the correlation between IFN-I activation and the disease activity (Greenberg et al., 2005; Baechler et al., 2007; Walsh et al., 2007; Niewold et al., 2009; Salajegheh et al., 2010; Cappelletti et al., 2011; Liao et al., 2011; Suárez-Calvet et al., 2014; Allenbach et al., 2016; Uruha et al., 2017). However, the pathogenic effect of IFN-I activation and its underlying pathomechanisms has not been yet demonstrated.
Here we report a new link between IFN-I and muscle atrophy with an upregulation of atrogenes (TRIM63/MURF-1 and FBXO32/atrogin) after IFN-I exposure. This link is supported by both in vitro and experiments on patient biopsies including immunohistochemical analysis showing atrogin in ISG15-positive perifascicular atrophic muscle fibres. These results indicate that muscle atrophy in dermatomyositis may be subsequent to IFN-I pathway activation. The impairment of myogenesis observed herein is in line with a previous study in a mouse muscle cell line (C2C12) (Franzi et al., 2013). However, in our study, we have used primary human skeletal muscle cells, which is a more physiological model. Moreover, the activation of the IFN-I pathway in muscle cells induces the secretion of IFN-I creating an autocrine amplifying loop and establishing the muscle as a source of IFN-I (Tournadre et al., 2012; Greenberg, 2014; Suárez-Calvet et al., 2014) in addition to other sources of IFN in the muscle tissue, such as the plasmacytoid dendritic cells (Greenberg et al., 2005; López de Padilla et al., 2007).
We also studied the effect of IFN-I on HMEC-1 cells, which are established from dermal microvascular endothelial cells (Arnaoutova and Kleinman, 2010), and retain the morphological, phenotypic and functional characteristics of human microvascular endothelial cells (Ades et al., 1992). Our tube formation assays demonstrate that an IFN-I enriched environment is pathogenic for capillary network formation. These results are consistent with previous results showing the correlation of IFN-I with the cutaneous manifestation in dermatomyositis patients (Huard et al., 2017). In addition, activation of the IFN-I pathway reproduces the upregulation of proteins found in capillaries in dermatomyositis muscle biopsies such as ISG15, MxA, TLR3 and RIG-I (Greenberg et al., 2005; Salajegheh et al., 2010; Cappelletti et al., 2011; Suárez-Calvet et al., 2014; Li et al., 2015). We also found that the activation of IFN-I pathway with PIC induced the upregulation of VEGF and TGF-β, both found in the capillaries of muscle biopsies (Lundberg et al., 1997; Grundtman et al., 2008). We only observed this upregulation with PIC but not with recombinant IFN-I. In all our in vitro experiments we also observed a stronger effect when the cells were cultured with PIC compared to recombinant IFN-I. This could be explained because TLR3, which recognizes PIC (Junt and Barchet, 2015), can induce other pathways in addition to IFN-I such as NFkB suggesting that the presence of VEGF and TGF-β in dermatomyositis muscle is not only a consequence of IFN-I. The expression of both proteins can be secondary to the vessel damage as a compensatory mechanism to enhance angiogenesis (Ochoa et al., 2007) and muscle regeneration (Arsic et al., 2004).
In conclusion, we show that IFN-I can induce both endothelial and muscle fibre injuries. However, we did not address what is the first hit, either muscle or the vessels.
The most important finding of our study is that IFN-I plays a key role in the pathogenesis of the disease and that ruxolitinib inhibits the pathogenic effects of IFN-I in both muscle and in endothelial cells. Although we cannot exclude that ruxolitinib can also modulate other cytokines whose signaling is mediated by the JAK pathway (Schindler et al., 2007; Heine et al., 2013; Febvre-James et al., 2018). Several options are addressable: neutralization of IFN-I cytokines, blocking IFN-I receptors, or blocking downstream signal transduction (such as JAK1/2 inhibition). Monoclonal antibodies against IFN-I cytokines or receptors are a targeted treatment but require a parenteral administration. In addition, since there are numerous IFN-I cytokines, and those involved in the dermatomyositis pathophysiology are not known yet, this approach may be difficult.
Ruxolitinib is a JAK1/2 inhibitor developed for the treatment of myelofibrosis (linked with a JAK2 gain of function mutation) (Furumoto and Gadina, 2013; Mascarenhas and Hoffman, 2013). It was shown to be efficient in the treatment of autoimmune diseases, such as systemic lupus erythematosus, another acquired interferonopathy (Baker and Isaacs, 2018). Moreover, JAK molecules are transducers for numerous membrane receptors including cytokines such as IL-2, IL-6, IL12/23, in addition to IFN. Consequently, JAK inhibitors have been developed to treat rheumatoid arthritis with impressive results (Taylor et al., 2017). Moreover, clinical trials using JAK1/2 inhibitors showed their good tolerance (Taylor et al., 2017).
JAK1/2 activation induces the phosphorylation of STAT-1, a key transcription factor that mediates IFN-I signalling (Ivashkiv and Donlin, 2014). In the present study, we have demonstrated a rapid clinical improvement in four patients with refractory dermatomyositis. In all, skin rash dramatically improved, and patients with muscle weakness started to show improved muscle strength. The effect of ruxolitinib on muscle was less obvious (compared to skin) but patients were assessed after only 3 months. Further prospective studies are needed to clarify the full benefit of ruxolitinib on skeletal muscle. Previous case reports have reported the efficacy of JAK inhibitors in five other patients with dermatomyositis. One patient treated with ruxolitinib (JAK1/2) demonstrated skin improvement and regained muscle strength (Hornung et al., 2014), and skin disease improved in four others treated with tofacitinib (JAK 1/3) (Kurtzman et al., 2016; Paik and Christopher-Stine, 2017). Interestingly, two of the latter reported patients described subjective improved muscle strength, and another patient muscle strength improvement was demonstrated objectively with the use of a hand-held dynamometer. Our study is the first to show a decrease in the interferon-stimulated gene signature in patients after treatment, that further supports the future use of JAK inhibitors in refractory dermatomyositis patients. Nevertheless, results of prospective clinical trial is now needed to confirm these results.
In conclusion, we propose JAK inhibition as a new dermatomyositis mechanism-based treatment, a finding that is relevant for the design of future clinical trials targeting specifically this form of myositis.
Acknowledgement
The authors thank Petra Matylewski for outstanding technical assistance.
Funding
The CAPES-COFECUB, Fiocruz/Inserm, Sorbonne Université/Faperj French/Brazilian joint programs (L.L.). INSERM, UPMC, The Myositis Association (X.S.C.). Association Française contre les Myopathies (AFM) and Instituto de Salud Carlos III (PI 15/1597 to E.G.). This work was developed in the context of the CNPq/Inserm/Fiocruz/UPMC International Associated Laboratory on Cell Therapy and Immunotherapy. The authors declare no conflict of interest.
Supplementary material
Supplementary material is available at Brain online.
Abbreviations
- HMEC-1
human dermal microvascular endothelial cells-1
- IFN-I
type I interferon
- PIC
polyinosinic-polycytidylic acid (poly(I:C))
References
Author notes
Leandro Ladislau, Xavier Suárez-Calvet, Ségolène Toquet and Océane Landon-Cardinal contributed equally to this work.
