Human Skeletal Muscle Cells (HSMCs), Isolation and Culture

Skeletal muscle samples were obtained from two JDM donors during muscle biopsy. The control muscle samples were obtained from patients with bone tumors. The discarded paratumoral muscle tissues were used as subsequent experiments (control HSMCs isolation and immunohistochemical staining). Primary human skeletal muscle was isolated and expanded by tissue explant adherent method. After washing these samples with PBS, non-muscle tissues were removed and the samples were sectioned into 1-mm3 columnar tissue blocks. Utilizing ophthalmic forceps, these blocks were methodically spaced at 0.5-cm intervals in a culture flask (NEST, Jiangsu, China) and covered with a specific growth medium blend. This medium, composed of high-glucose Dulbecco’s modified Eagle’s medium (DMEM) (GIBCO, USA), 20% fetal bovine serum (ExCell Bio, Jiangsu, China), and 2% Penicillin–Streptomycin Solution (beyotime, Shanghai, China), nourished the cells incubated at 37 °C in a 5% CO2 environment. The growth medium was replenished as necessary. For long-term storage, HSMCs were cryopreserved in a 90% growth medium and 10% dimethylsulfoxide mixture. Generally, cultured cells were maintained in DMEM enhanced with 10% fetal bovine serum and 1 × penicillin/streptomycin at standard conditions. Only cells from passages 1–3 were used for experiments.

Preparation for Single-Cell RNA Sequencing

To initiate the single-cell RNA sequencing, cells were first prepared by passing them through a 40-µm cell strainer to ensure a uniform single-cell suspension. Each cell suspension, 10 × barcode gel magnetic beads, and an oil mix were meticulously pipetted into the chambers of the chromium chip G to produce Gel Beads-in-emulsion (GEM). Each GEM was then subjected to a reverse transcription, performed at 53 °C for 45 min in a specialized PCR instrument (Thermo Fisher Scientific, Waltham, MA, USA). Subsequent complementary deoxyribonucleic acid (cDNA) purification and amplification ensured the quality and quantity of starting material. Enzymatic fragmentation was adjusted to yield fragments primarily between 200 and 300 base pairs. These fragments then underwent magnetic bead selection, end-repair, A-tailing, and Read2 sequencing primer ligation. Adapter-ligated cDNA was finally indexed, and the library was constructed. Prior to sequencing, libraries were quantified using the Qubit 2.0 Fluorometer and validated for size distribution using the Agilent 2100 Bioanalyzer. The validated libraries were then sequenced on the NovaSeq6000 (Illumina, San Diego, CA, USA) platform with a 150-bp paired-end configuration.

10× Genomics single-nucleus RNA-Seq (snRNA-seq) experiment

The 10× Genomics library preparation and snRNA-seq were carried out by NovelBio Co., Ltd (Shanghai, China). The process employed the 10× Genomics Chromium Controller Instrument and the Chromium Single-Cell 3′ V3 Reagent Kits, sourced from 10× Genomics (Pleasanton, CA, USA). Initially, the nuclei were washed twice using PBS. Subsequently, they were left to incubate for 30 min at room temperature. With precision, approximately 1000 nuclei/μl were then loaded onto the 10× Chromium chips. These chips were pre-loaded with the 10× single-cell 3′ V3 chemistry and barcode, facilitating the creation of single-cell Gel Bead-In-Emulsions (GEMs). Post-GEMs creation, the synthesis of cDNA commenced. Once this step was completed, the GEMs were broken to release the barcoded cDNA. This cDNA then underwent an amplification process consisting of 14 cycles post library construction. The process continued with fragmentation and index PCR amplification. The resulting final libraries were then assessed quantitatively using the Qubit High Sensitivity DNA assay (Thermo Fisher Scientific, Waltham, MA, USA). Additionally, the size distribution of these libraries was ascertained with the assistance of a High Sensitivity DNA chip, utilized on a Bioanalyzer 2200 (Agilent Technologies, Santa Clara, CA, USA). The entire sequencing process was undertaken on the Novaseq. 6000 platform (Illumina, San Diego, CA, USA), deploying a 2 × 150-bp paired-end sequencing protocol.

Preprocessing and quality control of the snRNA-seq data

In our endeavor to quantify both exonic and intronic reads captured via snRNA-seq, we constructed a tailored "pre-mRNA" reference package. This was built upon the human reference genome dataset, specifically the version refdata-gex-GRCh38-2020-A, in alignment with the 10× Genomics protocol. The raw data from the sequencing was matched to this "pre-mRNA" reference employing the Cell Ranger toolkit (version v4.0.0). For the ensuing data preprocessing steps, the resulting nucleus-gene expression matrix was assimilated into the "Seurat" software package, version v3.2.3. We applied strict criteria during the preprocessing phase. Genes that manifested counts in less than three nuclei were pruned from the dataset. This filtration process was vital in nullifying the genes that might have been detected due to potential random noise. The quality control of the nuclei was paramount. They were subjected to a rigorous screening process based on unique molecular identifier (UMI) counts, ensuring a range between 500 and 50,000. Similarly, the gene count range was established between 300 and 7000. Any nuclei that surpassed a 0.05 proportion in mitochondrial genes or ribosomal genes were disregarded. These boundaries were essential to eliminate inferior-quality nuclei, which might have been indicative of doublets or perturbations from technical discrepancies. An additional layer of quality control was the exclusion of nuclei with doublet scores exceeding 0.35, as predicted by the Scrublet tool. Furthermore, any nuclei exhibiting enhanced expression across marker genes spanning multiple lineages were meticulously omitted to ensure data integrity in subsequent analyses.

Data Download and Processing

For our analysis, we sourced muscle cell datasets from healthy individuals as a comparative benchmark. These datasets were procured from the renowned Gene Expression Omnibus (GEO) repository, which can be accessed at https://www.ncbi.nlm.nih.gov/geo/. The control muscle tissues used in genetic analysis were distinct from the JDM samples in terms of its origin and processing. The healthy control individuals were adults aged 18–50 years who were recruited through advertisements at the university and local community based on information provided in the original text [8]. These samples were notably from individuals without any diagnosed muscular anomalies. Specifically, we utilized data under the accession numbers GSM6611295, GSM6611297, and GSM6611299. Notably, all three samples pertained to muscle tissue and were collected prior to any form of exercise. In our endeavor to maintain uniformity in our analysis, the filtering criteria employed for the snRNA-seq data of JDM were replicated for this control group. It is worth mentioning that the JDM samples originated in our laboratory, derived from biopsy specimens of patients diagnosed with the condition. For a comprehensive analysis, we integrated the JDM sample data with the snRNA-seq data of the aforementioned control samples. Given the multifaceted nature of the combined data, it was imperative to address and rectify potential batch effects. To achieve this, we harnessed the capabilities of the "harmony" package. This package boasts an algorithm specifically tailored to streamline and align intricate multi-sample datasets, ensuring uniformity and coherence across the board.

Normalization, Feature Selection, Integration, Scaling, and Clustering of the SnRNA-Seq Data

For each sample, the sum of the UMI counts for each nucleus was normalized to 10,000 followed by log-transformation. We used the FindVariableFeatures function of Seurat to select 2000 highly variable genes for each sample based on mean variability plot. Nuclei of all samples were integrated via canonical correlation analysis implemented in Seurat, which was set to identify the top 30 canonical correlations to correct for potential batch effects and identify shared cell states across samples. Unwanted sources of variation were mitigated by regressing out specific factors: the proportion of mitochondrial genes, UMI count, gene number, the proportion of mitochondrial genes, and the proportion of ribosomal genes with linear models using ScaleData. The data were then centered for each gene by subtracting the mean expression and scaled by dividing the centered expression by the standard deviation. Principal component analysis (PCA) was applied and the first 20 PCA components were used to compute a shared nearest-neighbor (SNN) graph of the nuclei. The SNN graph was visualized using uniform manifold approximation and projection (UMAP). Clustering was performed using the Louvain algorithm with the clustering parameter resolution set to 0.5, specifically to identify the subpopulation structures within the samples.

Gene Set Variation Analysis (GSVA)

Cell types were analyzed for genomic variation using the GSVA software package, with the objective to explore and interpret the biological pathways activated in each cell type. Gene sets for GSVA were obtained from the MSigDB database (c2. Cp. Kegg. V7.1. Symbols. gmt), including more than 5000 gene sets that cover various biological processes and pathways. A GSVA score was generated for each cell using the "GSVA" software package, with kcdf set to Poisson, reflecting the underlying distribution of the gene expression data. The “complexHeatmap” and “ggplot2”software packages were used to visualize the activated pathways, providing an interactive and informative view of the data.

Detection of Serum Helper T cell (Th) 1/Th2 Cytokine Levels by Flow Cytometry

The serum levels of interleukin (IL) -2, IL-4, IL-6, IL-10, tumor necrosis factor-α (TNF-α), and IFN-γ from patients with JDM (pre and post treatment) were determined by using the Cytometric Bead Array Human Th1/Th2 Cytokine Kit II, purchased from BD Pharmingen (San Jose, CA) (cat. no. 551809). Cytokine titers were expressed as pg/ml, as calculated by reference to standard curves based on known amounts of recombinant cytokines.

The diagnosis of JDM was based on the 2017 EULAR/ACR classification criteria [9] and disease remission should fulfil the Pediatric Rheumatology International Trials Organization (PRINTO) criteria for clinically inactive disease in JDM, meet at least three out of four of the following criteria on or off therapy: (i) physician global assessment of overall disease activity (PhyGloVAS) ≤ 0.2, (ii) creatine kinase (CK) ≤ 150 U/l, (iii) CMAS score ≥ 48, (iiii) MMT-8 score ≥ 78 [10]. Patients with complete cytokine information on pre-treatment and remission state were included in this study.

Immunohistochemical (IHC) Staining

IHC was applied to assess the expression of p-STAT3 in muscle tissues from both JDM and controls. Antibody specific to p-STAT3 was purchased from Cell Signaling Technology (Beverly, MA, USA), using a working dilution as 1:500. The slides were scanned by a Pannoramic MIDI scanner (3DHISTECH, Hungary) and viewed using CaseViewer software v2.2.

ROS Measurement

Primarily isolated human skeletal muscle cells from JDM and HSMC cell line were stimulated with IL-6 (25 ng/ml, Novoprotein, Jiangsu, China; suspended in PBS) or PBS (served as control) for 24 h and cells were then collected into tubes. Cells were stained with DCFH-DA (1 mM in 1 × PBS) for 30 min at 37 °C. ROS analysis of HSMCs was completed by using Reactive Oxygen Species Assay Kit (beyotime, S0033S, Shanghai, China) by a microplate reader (TECAN, Switzerland). HSMC treated with humanized recombinant anti-IL-6R antibody (Tocilizumab, ACTEMRA®, Roche, working concentration 100 ng/ml) was listed as intervention group. The concentration of IL-6 and tocilizumab was set based on previous studies and our pre-experiments performed by different concentration gradients [11, 12]. We decided on a final IL-6 stimulation concentration of 25 ng/ml, and tocilizumab treatment concentration of 100 ng/ml.

Quantitative Reverse Transcription PCR

Total RNA was extracted after HSMC (from JDM and normal cell line as control) were stimulated with IL-6 (25 ng/ml) using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). RNA was reverse transcribed into cDNA using the PrimeScript RT Master Mix (Takara, Shiga, Japan) according to the manufacturer’s protocol. Quantitative PCR was performed on the CFX Opus 96 Real-Time PCR System (Bio-Rad, Hercules, CA, USA) using SYBR Green reagents (Monad Biotech, Jiangsu, China); the corresponding primers are listed in Supplementary Table 1. The expression of target genes was calculated using the 2−ΔΔt method after normalizing to the housekeeping gene GAPDH and control group.

Western Blot

The protein extracted from both JDM and control muscle tissues as well as HSMCs (isolated from JDM or normal cell line) stimulated with IL-6 (25 ng/ml) were degenerated, electrophoresed, and transferred to the polyvinylidene fluoride membrane (Bioss, Beijing, China). Subsequently, the membrane was blocked using 5% non-fat milk and incubated with primary antibodies-rabbit stat3 (1:1000, ab68153, Abcam) and p-stat3 (1:1000, ab32143, Abcam) and mouse anti-β-actin (Bioss, Beijing, China) overnight at 4 °C. On the following day, the membranes were washed and incubated with the corresponding horseradish peroxidase-conjugated secondary antibodies (Bioss, Beijing, China) and detected with an enhanced chemiluminescence reagent (Bioss, Beijing, China) using the chemiluminescence system (GBOX-CHEMI-XRQ, Syngene, USA). The grayscale of the detected bands were assessed by ImageJ software (National Institutes of Health), and the results were expressed as fold changes after normalizing to β-actin.