Background

Duchenne muscular dystrophy (DMD) is a progressive muscle wasting disorder caused by mutations in the DMD gene encoding the protein dystrophin [1]. While dystrophin is expressed in multiple tissues, loss of dystrophin in the context of DMD is particularly detrimental to skeletal and cardiac muscle function [2]. Dystrophin is a component of the dystrophin-glycoprotein complex (DGC) that stabilizes the muscle fiber sarcolemma and mediates the linkage between the extracellular matrix (ECM) and the intracellular actin cytoskeleton [3,4,5,6]. Loss of dystrophin in DMD results in the absence of the DGC, destabilizing the sarcolemma and rendering it susceptible to contraction-induced damage [7]. Over time, chronic injury caused by asynchronous cycles of myofiber degeneration and regeneration results in failed muscle regeneration, inflammation, and replacement of functional muscle fibers with fibrosis [8]. Patients with DMD present with high plasma levels of muscle creatine kinase at birth, muscle fibrosis and hypertrophy, weakness of the proximal muscles, loss of ambulation, and pulmonary and cardiac dysfunction leading to premature death in the 2nd to 3rd decade of life [9,10,11]. While there are some FDA-approved treatment options for DMD, including corticosteroids and exon-skipping drugs that slow progression of the disease, there is still no cure.

The dystrophin-deficient mdx mouse is the most widely used mouse model for DMD as it exhibits much of the pathology observed in patient skeletal muscle, albeit much milder, including elevated creatine kinase levels, increased levels of degeneration and regeneration, fibrosis, and reduced grip strength and whole-body tension [12, 13]. While many therapeutic genes that improve sarcolemma instability, decrease fibrosis, and increase force in mdx muscle have been identified [14,15,16,17,18,19,20,21], there are gaps in understanding whether their protective mechanisms affect overlapping molecular processes and signaling pathways. Furthermore, there is emerging evidence of an expanded network of DGC-interacting proteins, suggesting that it functions as a central unit to integrate cellular signaling, lateral force transmission, ion channel function, and cytoskeletal organization [22, 23].

Sarcospan (SSPN), a core component of the DGC [24,25,26], prevents muscle degeneration and histopathology in mdx mice in a mechanism that is dependent on increased abundance of utrophin and integrin α7β1 at the sarcolemma [27,28,29,30,31,32]. Together with the sarcoglycans (SG), SSPN forms a tight subcomplex [26] that anchors α-dystroglycan, a receptor for many ECM proteins [33], to the cell membrane to stabilize the cytoskeleton-matrix linkage [34]. SSPN increases SG protein levels and restores integrity of the SG-SSPN subcomplex at the sarcolemma in mdx skeletal and cardiac muscles [27,28,29,30,31, 35]. SSPN overexpression in mdx mice also increased abundance of dystroglycan, enhanced matriglycan glycosylation of α-dystroglycan, and improved laminin binding [27, 30, 35, 36]. Using genetic approaches, we previously demonstrated that SSPN function is dependent on matriglycan modification of α-dystroglycan, which directly interacts with laminin [35]. Restoration of cell-matrix interactions by SSPN improved postexercise activity, protected against eccentric contraction-induced force loss, and prevented declines in pulmonary and cardiac functions [29,30,31].

In the current study, we interrogated the effects of SSPN overexpression on the mdx transcriptome and proteome using transgenic mouse models (Supplemental Table 1). Our findings build on many prior studies including microarray data sets (Omnibus GSE465, GSE1004, and GSE1007) [37] and proteomic analysis [37,38,39,40,41,42,43,44] of mdx muscle and DMD patient muscle biopsies, which revealed signaling pathways that contribute to the dystrophic pathology including anaerobic metabolism, cytoskeleton remodeling, calcium handling, adipogenesis, fibrosis, and endoplasmic reticulum stress. Given that there are many emerging approaches to treating DMD (dystrophin dependent as well as dystrophin independent), determining the effects of such treatments on cellular and molecular processes is important for therapeutic design. Such studies also have the potential to reveal overlapping pathways that are critical for prevention of muscle degeneration as well as pathways that are unaltered by a particular therapy (thereby informing molecular processes that are not major contributors to pathology). For instance, therapeutic restoration of dystrophin expression in mdx muscle by antisense oligonucleotide exon skipping normalizes protein and mRNA expression toward wild-type levels, but does not affect the expression of microRNAs [37]. In our study, we found that SSPN overexpression in mdx muscle does not restore all transcripts and proteins to wild-type levels but ameliorates muscle pathology through compensatory changes in ECM and cytoskeletal composition, including key signaling molecules associated with regulation of cytoskeleton organization and mechanotransduction.

Methods

Mouse models

Wild-type (C57BL/6J) and mdx mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA). We have generated and extensively characterized multiple lines of SSPN-transgenic mdx mice expressing either human SSPN (hSSPN, line 3, approximately 3-fold expression) [35] or murine SSPN (mSSPN, line 28, approximately 30-fold expression) [30] that are outlined in Supplementary Table 1. The SSPN transgenes were all under the control of the human skeletal α-actin promoter. hSSPN and mSSPN exhibit a high degree of identity (> 85%) at the amino acid level, effectively ameliorate mdx pathology, and were included in the study with the rationale that overlapping findings from different transgenic lines would strengthen their relevance. All mouse colonies were granted by the UCLA Animal Welfare Assurance (approval no. A3196-01).

Poly-A-enriched RNA sequencing

Liquid nitrogen frozen tibialis anterior muscles from 12-week-old male wild-type, mdx, and mdxTG (line 28) were pulverized in a mortar and pestle cooled by liquid nitrogen and homogenized into TRIzol (Thermo Fisher Scientific, Waltham, MA, USA) using syringe homogenization with a 21g needle. RNA was extracted using TRIzol phase separation followed by QIAGEN RNeasy column purification using the manufacturer’s instructions (QIAGEN, Hilden, Germany), followed by DNase treatment. RNA concentration and quality were measured using the TapeStation 4200 (Agilent Technologies, Santa Clara, CA, USA). RNA-Seq libraries were prepared from total RNA using the TruSeq Stranded mRNA Library Prep Kit (Illumina, San Diego, CA, USA). High-throughput sequencing was performed at UCLA Technology Center for Genomics & Bioinformatics using the Illumina HiSeq 4000 platform (paired-end 75 bp reads). Demultiplexing was performed with Illumina Bcl2fastq2 v 2.17 program. The total numbers of sequenced reads were 53–67 million per sample. RNA-Seq reads were mapped to the mouse reference genome (mm10) using STAR [45]. For each sample, 86% of reads were uniquely mapped to the genome. Expression levels were quantified for annotated genes (Ensembl v.92), and raw gene counts were normalized to CPM values (counts per million). The analysis was focused on genes with CPM > 1 in at least two samples (13,846 genes). Differential expression analysis was performed using edgeR-QLF [46]. Differentially expressed genes (DEGs) were identified at 1% FDR and fold-change > 2.

Mass spectrometry

The mass spectrometry dataset was first published in part in Stearns-Reider et al. [47]. Quadricep muscles were harvested from 20-week-old male wild-type, mdx, and mdxTG (line 3) mice and snap frozen in liquid nitrogen. Samples were then prepared for mass spectrometry analysis as previously described [48]. In short, samples were pulverized in liquid nitrogen and lyophilized. For each sample, 5 mg (dry weight) of tissue was homogenized in 200 mL/mg high-salt buffer (HS buffer) containing a 1× protease inhibitor [48]. Following three rounds of HS buffer wash, pellets were treated with 6 M guanidine extraction buffer. The remaining pellets from each tissue, representing insoluble ECM proteins, were digested with freshly prepared hydroxylamine buffer, as previously described [49]. A total of 100 μl of the cellular fraction (combined fractions 1, 2, and 3) and 200 μl of the soluble and insoluble ECM fractions were enzymatically digested with trypsin using a filter-aided sample prep approach and C18 tip cleanup. Samples were then analyzed by liquid chromatography-data-dependent acquisition tandem mass spectrometry (LC-MS/MS), as previously described [50]. Samples were analyzed on a Q Exactive HF Orbitrap mass spectrometer (Thermo Fisher Scientific) coupled to an EASY-nanoLC 1000 system through a nanoelectrospray source. The analytical column (100 μm i.d. × 150 mm fused silica capillary packed in house with 4 μm 80 Å Synergi Hydro C18 resin (Phenomenex; Torrance, CA, USA)) was then switched online at 600 nL/min for 10 min to load the sample. The flow rate was adjusted to 400 nL/min, and peptides were separated over a 120-min linear gradient of 2–40% ACN with 0.1% FA. Data acquisition was performed using the instrument supplied Xcalibur (Thermo Fisher Scientific, San Jose, CA, USA) software in positive ion mode. MS/MS spectra were extracted from raw data files and converted into mgf files using a PAVA script (University of California, San Francisco, MSF, San Francisco, CA, USA). These mgf files were then independently searched against mouse Swiss-Prot database using an in-house Mascot server (version 2.2.06; Matrix Science, London, UK). Mass tolerances were +/−10 ppm for MS peaks and +/−0.5 D for MS/MS fragment ions. Trypsin specificity was used allowing for one missed cleavage. Met oxidation, pro-oxidation, protein N-terminal acetylation, and peptide N-terminal pyroglutamic acid formation were allowed for variable modifications, whereas carbamidomethyl of Cys was set as a fixed modification. Following Mascot searches, data was directly loaded into Scaffold™ (Proteome Software Inc.). Peptide spectral matches were directly exported with a 99% confidence in protein identifications and at least 2 unique peptides per protein, resulting in a false discovery rate of 0.54%. Two-group comparisons were done by two-tailed Student’s t-tests. Partial least squares-discriminant analysis (PLSDA) was performed using MetaboAnalyst (version 3.0) with sum and range scaling normalizations.

Indirect immunofluorescence staining of muscle sections

Transverse cryosections (10 μm) from the quadriceps muscles of wild-type, mdx, and mdxTG (line 3) mice at 3–5 months of age were incubated in blocking buffer (3% BSA in phosphate-buffered saline (PBS)) for 30 min at room temperature. Avidin/biotin blocking kit (Vector Laboratories, Newark, CA, USA) was used according to manufacturer’s instructions. Sections were then incubated with the following primary antibodies diluted 1:200 in PBS overnight at 4 °C: β-spectrin non-erythrocyte (PA5-52970, Thermo Fisher Scientific, Waltham, MA, USA), collagen type II (ab34712, Abcam, Waltham, MA, USA), and collagen type V (ab7046, Abcam, Waltham, MA, USA). Yap1 primary antibody (nb110-58358, Novus Biologicals, Centennial, CO, USA) was diluted 1:250. Sox9 antibody (AB5535, Sigma, St. Louis, MO, USA) was diluted 1:100. Sections were washed in PBS for 3 × 30 min at room temperature. Primary antibodies were detected by biotinylated anti-rabbit (BA-1000; 1:500; Vector Laboratories, Newark, CA, USA). Fluorescein-conjugated avidin D (A-2001; 1:500; Vector Laboratories) was used to detect secondary antibodies. Sections were mounted in Vectashield (Vector Laboratories, Newark, CA, USA), and imaging was performed using a Zeiss Axio Imager M2 (Carl Zeiss Inc., Thornwood, NY, USA) with a Hamamatsu ORCA-Flash 4.0 V3 digital complementary metal oxide semiconductor camera and a plan-Apochromat 20×/0.8 M27 objective. Percent nuclear Yap1 was quantified by counting Yap1-positive nuclei, with clear Yap1/DAPI overlap, divided by total DAPI content in five 20× fields per view per biological sample. Quantification of immunofluorescence analysis was performed on 20× images using ImageJ software (NIH, version 1.50i) for collagen II, collagen V, and β-spectrin by line scan analysis and Sox9 by measuring the integrated intensity in 4–12 images per genotype. For line scan measurements, using the line drawing tool on ImageJ, 40 areas of interest per biological replicate were quantified. Lines were drawn along or perpendicular to positive-stained areas, and peak intensity values were plotted denoted as “max” measurement on ImageJ.

Gene ontology and ingenuity pathway analysis

Gene ontology (GO) enrichment analysis for RNA sequencing and proteomic data sets was performed using the PANTHER classification system online user interface [51] and the statistical overrepresentation test. Fold enrichment values and p-values from the overrepresentation test were plotted using GraphPad Prism. RNA sequencing and proteomic data sets were also analyzed through the use of ingenuity pathway analysis (QIAGEN) for generating networks and upstream regulators [52]. Upstream regulators were sorted by activation score, and the top five inhibited and activated regulators were reported with corresponding target genes identified in the data sets.

Integrated gene set enrichment analysis

Significantly differentially expressed transcripts and proteins between mdx and mdxTG were pooled as input for integrated gene set enrichment analysis using Database for Annotation, Visualization and Integrated Discovery (DAVID, version 2021). Each molecule was categorized as unrestored (significantly different from wild type, not significantly different from mdx), restored (significantly different from mdx, not significantly different from wild type), or compensatory (significantly different from both wild type and mdx). Data were visualized on Cytoscape version 3.9.1.

Supplementary figure methods

For a systems level network analysis, DEGs from the three identified categories (restored, unrestored, compensatory) were analyzed using the Database for Annotation, Visualization and Integrated Discovery platform (DAVID, version 2021). This analysis was also performed with the proteomic data set. Gene Ontology (GO) terms with corresponding parameter thresholds of p < 0.05 and minimum gene count = 2 were used as input and visualization in Cytoscape with the EnrichmentMap plugin.

Results

Gene expression and proteomic analyses reveal compensatory and restored pathways

In the current study, our goal was to interrogate the effects of SSPN overexpression on gene and protein expression in mdx muscle relative to wild-type and mdx (non-transgenic) controls. Gene expression analysis was performed using traditional poly-A-enriched RNA sequencing of tibialis anterior muscles isolated from 12-week-old wild-type, mdx, and mdxTG mice. Principal component analysis of the sequencing data reveals clear clustering of each genotype (Fig. 1a). In comparing paired genotypes, we identified 1073 DEGs (853 upregulated, 220 downregulated) in wild-type versus mdx muscle and 748 DEGs (471 upregulated, 277 downregulated) in mdx relative to mdxTG samples. The largest difference in the transcriptomic profile was evident between wild type and mdxTG with 1857 DEGs (1306 upregulated, 551 downregulated). Heat map analysis of DEGs in mdxTG relative to controls (Fig. 1b) reveals patterns that can be categorized as follows: (1) unrestored (significantly different from wild type, not significantly different from mdx), (2) restored (significantly different from mdx, not significantly different from wild type), and (3) compensatory (significantly different from both wild type and mdx).

Fig. 1
figure 1

Overview of RNA sequencing and mass spectrometry reveals distinct transcriptomic and proteomic profiles of SSPN overexpression rescue. a Principal component analysis (PCA) of RNA sequencing data in wild type (WT, n = 5), mdx (n = 4), and mSSPN transgenic (mdxTG, n = 4) tibialis anterior muscle at 12 weeks of age. b Heat map of DEGs. c PCA of mass spectrometry data from WT (n = 5), mdx (n = 5), and hSSPN transgenic (mdxTG, n = 5) quadriceps muscle at 20 weeks of age. d Heat map of differentially expressed proteins (DEPs). e Overlap of Gene Ontology (GO) terms enriched in WT vs mdxTG in transcriptomics vs proteomics data using PANTHER GO analysis platform

Full size image

Given the effects of SSPN overexpression on DEGs, we next investigated the influence of SSPN on protein expression utilizing a mass spectrometry approach that improves capture of typically insoluble ECM proteins [47, 53]. From our proteomics analysis of wild-type, mdx, and mdxTG quadriceps muscle, we identified a total of 1679 proteins in all three genotypes. Principal component analysis revealed distinct clustering of the wild-type, mdx, and mdxTG samples (Fig. 1c). Similar to the RNA sequencing data, we found sets of proteins in restored, unrestored, and compensatory categories (Fig. 1d).

Pathway enrichment analysis identifies compensatory changes in fibrillar collagens and components of the actin cytoskeleton

Comparative analysis of enriched GO pathways in the transcriptomic and proteomics data sets revealed seven shared pathways, including those associated with ECM organization, extracellular structure organization, and cytoskeleton organization (Fig. 1e). By separating enriched GO pathways into the compensatory, restored, and unrestored subcategories for both transcriptomic and proteomic data sets, we observed additional pathways of interest such as those associated with calcium ion binding, NAD/NADP binding, ubiquitin ligase binding, and oxidoreductase binding (Supplementary Fig. 1). Given the identification of the broader categories of ECM organization and cytoskeleton organization identified in the mdxTG model through GO pathway analysis, we curated lists of actin cytoskeleton- and ECM-associated genes in the RNA sequencing data set based on GO categories and subsequently visualized this data using traditional heat maps. The actin cytoskeleton and ECM genes clustered into restored, unrestored, and compensatory categories. It is noteworthy that the compensatory genes represent the largest category of DEGs in this comparison (Fig. 2a–b, Supplementary Tables 2, 3).

Fig. 2
figure 2

SSPN overexpression results in compensatory upregulation of ECM and actin cytoskeleton genes. Heat maps of GO-term curated ECM genes (a) and actin cytoskeleton genes (b) from RNA sequencing data each with restored, unrestored, and compensatory expression patterns in the mdxTG muscle. Expanded heat map insets emphasize that compensatory expression patterns are overwhelmingly from upregulated genes in the mdxTG muscle

Full size image

We next analyzed the expression of ECM and cytoskeletal proteins from the mass spectrometry data set. ECM proteins were classified into 8 functional categories, including (1) structural ECM, (2) other ECM, (3) network collagen, (4) matricellular, (5) fibrillar collagen, (6) fibril-associated collagens with interrupted triple helices (FACIT), (7) ECM regulator, and (8) basement membrane proteins (Fig. 3a). Cytoskeletal proteins were classified into 9 functional categories, including (1) actins and microfilaments, (2) actin-associated, (3) tubulins, (4) microtubule associated, (5) annexins, (6) intermediate filaments, (7) spectrins, (8) myosins, and (9) sarcomere-associated proteins (Fig. 3b). ECM regulators and fibrillar collagens were upregulated in mdxTG compared to both wild-type and mdx muscles (Fig. 3a). For cytoskeletal proteins, we observed significant downregulation in actins, microfilaments, and actin-associated proteins, as well as decreased abundance of myosins and other sarcomere-associated proteins such as tropomyosins and troponin I and T (Fig. 3b). Abundance of microtubule-associated proteins and spectrins was increased in mdxTG samples relative to wild-type controls. Within each functional classification, we additionally identified several proteins with increased expression in mdxTG muscle compared to both wild-type and mdx muscle including cathepsins and integrins (Fig. 4a), fibrillar collagens II, V, and XI (Fig. 4b), and non-erythrocyte β-spectrin (Sptbn1, spectrins, Fig. 4c). Interestingly, some actin isoforms and actin-associated proteins were decreased in mdxTG muscle compared to both wild-type and mdx muscle including skeletal and cardiac α-actins (Acta1, Actc1) and α-actinin (Actn3, Fig. 4d).

Fig. 3
figure 3

Compensatory changes in functional classes of ECM and cytoskeletal proteins in mdxTG muscle. a Graph of the abundance of ECM proteins in 8 primary categories relative to WT. b Graph of the abundance of cytoskeletal proteins in 9 primary categories relative to WT (*p < 0.05 compared to WT, #p < 0.05 compared to mdx by unpaired t-test. By category, mdxTG muscle had compensatory expression (both significantly different from WT and mdx) of matricellular, ECM regulator, basement membrane, actins/microfilaments, actin-associated, myosins, and sarcomere-associated proteins

Full size image
Fig. 4
figure 4

Upregulation of ECM regulators, fibrillar collagens, and spectrins in mdxTG muscle. Relative protein expression of ECM regulators (a), fibrillar collagens (b), spectrins (c), and actins/microfilaments (d) (*p < 0.05 compared to WT, #p < 0.05 compared to mdx by unpaired t-test)

Full size image

Proteomic analysis revealed compensatory upregulation of many collagens, including collagen type II (Col2) and collagen type V (Col5), along with upregulation of Sptbn1. Importantly, recent findings in the golden retriever model of DMD (GRMD) revealed spectrin as a candidate protein responsible for sparing of the cranial sartorius muscle from dystrophic pathology [54]. Spectrin is a mechanosensitive protein that functions as a scaffolding protein at the cell membrane to provide structural and mechanical stability [55]. Additionally, spectrin is a key linker protein in the response to mechanical stimuli, coupling changes in cellular tension to downstream signaling networks [56, 57]. The compensatory upregulation of spectrin along with fibrillar collagens suggests that SSPN overexpression may be rescuing the dystrophic phenotype through the alteration of mechanotransduction signaling networks.

Ingenuity pathway and gene set enrichment analyses support compensation through mechanotransduction pathways

To identify candidate networks and upstream regulators that drive compensatory changes in mdxTG muscle, we analyzed the transcriptomic and proteomic data sets using the ingenuity pathway analysis (IPA) application. Based on the IPA upstream regulator algorithm and corresponding activation z-scores and p-values, we examined the top ten activated and inhibited upstream regulators for each comparison from the RNA sequencing (Fig. 5a) and mass spectrometry (Fig. 5b) data sets. Upon closer analysis of the top five activated and inhibited regulators, we observed many cytoskeletal and ECM target molecules (Table 1, Supplementary Tables 4–6, targets in bold red). Our analyses reveal that SSPN induces large-scale changes in the composition of the ECM and cytoskeleton as well as their associated signaling networks. Thus, we sought to integrate data from the transcript and protein analysis to develop a more comprehensive model reflecting the overlap of signaling networks. We performed integrative gene set enrichment analysis on the significantly differentially expressed transcripts and proteins between mdx and mdxTG (Fig. 6). This comprises the proteins and genes in the compensatory and restored categories. We observed enrichment of gene sets associated with signal transduction pathways affecting a broad spectrum of cellular and biological processes important in myogenesis, regeneration, and homeostasis of skeletal muscle involving the Ras, Rho, and Wnt signaling, actin-cytoskeleton remodeling and organization, and ECM signaling and organization (Fig. 6). These include signaling by Rho GTPases, regulation of cytoskeleton organization and cytoskeletal regulation by Rho GTPases, integrin signaling, focal adhesions formation, and response to mechanical stimulus (Fig. 6). We observed compensatory upregulation of Rho family of GTPases involved in mediating actin dynamics and actomyosin contractility including Ras homolog gene family members (RhoA and RhoC), Rac family small GTPase 1 (Rac1), and cell division cycle 42 (Cdc42). There are compensatory and restorative changes in several factors that regulate Rho GTPase activity including the GTPase activating proteins (Arhgap 6, 22, and 44), all of which are increased in mdxTG muscle. Wnt proteins, including Wnt4 and Wnt5a, play key roles as part of the muscle regeneration program, cytoskeletal remodeling, and myoblast fusion [58, 59]. Wnt4 regulation of RhoA activity is important for maintaining satellite cell quiescence [59]. In addition, Wnt5a and Fzd4 (rescued in mdxTG) have been shown to regulate osteogenic differentiation after mechanical stretch [60]. R-spondins family members 1 and 4 (Rspo1 and Rspo4) positively regulate the canonical Wnt signaling pathway and are important for muscle repair and regeneration [61,62,63]. RhoA and Rac1 can also be activated by and respond to mechanical stress and stimuli. At focal adhesion sites of integrin signaling, these proteins orchestrate actin cytoskeleton remodeling important for mechanical force generation and transmission in concert with talin (Tln2) and vinculin (Vcl) [64]. Several integrin and laminin subunits are also alternatively expressed in mdxTG relative to mdx. As we have shown previously, integrin alpha 7 (Itga7) is increased in mdxTG muscle [27]. The cysteine and glycine-rich protein 3 (Csrp3, also known as muscle LIM protein, MLP) is localized at costameres and involved in promoting mechanical- and stress-induced myocyte differentiation and remodeling [65, 66]. Interestingly, transcription factors JunB and Sox9 are both upregulated in mdxTG muscle. JunB is also important for maintaining muscle growth and hypertrophy [67]. Sox9, important for bone and cartilage formation, is expressed in muscle progenitor cells during development and may play a role in musculoskeletal development [68]. Sox9 has been shown to be a key regulator of ECM deposition, most specifically of collagen II [69], which is highly upregulated in mdxTG muscle (Fig. 4).

Fig. 5
figure 5

Activated and inhibited upstream regulators identified in wild-type, mdx and mdxTG muscle through ingenuity pathway analysis. Ingenuity pathway analysis (IPA) of RNA sequencing data (a) and mass spectrometry data (b) identifying the top 10 upstream regulators that are activated (z-score, red bars) or inhibited (z-score, blue bars) with −log p-values overlaid in black. Comparisons include WT vs mdx (left graphs), WT vs mdxTG (middle graphs), and mdx vs mdxTG (right graphs)

Full size image
Table 1 Ingenuity pathway analysis summary — upstream regulators
Full size table
Fig. 6
figure 6

Integrated gene set enrichment analysis identifies differential expression of mechanosignaling pathways in mdxTG. To develop an integrated model of overlapping signaling networks driving rescue of the mdxTG skeletal muscle, we combined transcripts (diamonds) and proteins (ovals) that were differentially expressed between mdx and mdxTG tissue and performed gene set enrichment analysis. Compensatory changes are in pink, restored changes are in green, and uncategorized changes are in gray. The gene set enrichment analysis highlights changes in Rho, Rac, Wnt, and integrin signaling in addition to cell adhesion and response to mechanical stimulus

Full size image

To validate our findings from the mass spectrometry data and gene set enrichment analysis, we performed immunofluorescence analysis of muscle sections using antibodies against collagen II, collagen V, Sptbn1, yes-associated protein 1 (Yap1), and Sox9 (Fig. 7). Immunofluorescence analysis and quantification validate the mass spectrometry data showing increased collagens II and V and Sptbn1 in mdxTG compared to WT and mdx muscle (Fig. 7a–b). The Hippo-Yap1 network also plays a significant role in the response to mechanical stimuli with potential implications in muscle development and homeostasis following Yap1 translocation to the nucleus [57, 70]. To gain further insight on this downstream mechanotransduction signaling pathway, we quantified the percentage of nuclear Yap1 translocation following indirect immunofluorescence and found an increase in mdxTG muscle relative to WT and mdx muscle (Fig. 7c, Supplemental Fig. 3). Interestingly, our results also indicated profound sarcolemmal localization of Yap1 in mdxTG muscle. Furthermore, immunofluorescence analysis of muscle sections reveals increased Sox9 protein in mdxTG muscle, and that in addition to nuclear localization, Sox9 is abundant at the sarcolemma and neuromuscular junction where it co-localizes with α-bungarotoxin (Fig. 7c). While Yap1 and Sox9 typically localize to the cytoplasm and/or nucleus, previous reports have also described sarcolemmal and connective tissue localization in muscle [70,71,72]. Overall, our analyses indicated that SSPN overexpression in dystrophin-deficiency results in a rewiring of signaling networks associated with cell-matrix communication and mechanotransduction.

Fig. 7
figure 7

Validation of signaling and mechanosensitive pathways increased in mdxTG muscle. Indirect immunofluorescence analysis and quantification of 12-week-old mouse quadriceps using antibodies against (a) collagen II (Col II) and collagen V (Col V), (b)β-spectrin (Sptbn1), (c) yes-associated protein 1 (Yap1) and SRY-Box transcription factor 9 (Sox9) showing increased abundance in mdxTG relative to both WT and mdx. Sox9 is expressed at the neuromuscular junction shown by co-localization with α-bungarotoxin (α-BgTx). Statistical analyses were performed by one-way ANOVA with Tukey’s multiple comparison tests, n = 3–4 biological replicates per genotype, data represented as +/− SEM (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Scale bar = 100 μm

Full size image

To visualize the differential expression pattern of the key components of the ECM and cortical cytoskeleton, we prepared a schematic from both transcriptomic and proteomic perspectives using a color key (Fig. 8). As shown in the summary figure, the differential expression pattern of many of the key players of mechanotransduction at the cell-matrix interface reveals the compensatory effects of SSPN overexpression.

Fig. 8
figure 8

SSPN overexpression rewires ECM-cell communication. Schematic summarizing ECM and cytoskeletal changes in RNA sequencing and proteomic data sets. Changes in gene or protein expression are represented by color according to the key, indicating compensatory upregulation of many ECM and cytoskeletal molecules with downstream signaling effects (in purple). The x- and y-axes in the color key correspond to mdxTG relative to WT and mdxTG relative to mdx, respectively

Full size image

Discussion

In our multi-omics approach to identifying mechanisms of SSPN amelioration of dystrophic pathology, we built on previous comparative transcriptomic and proteomic studies that have identified the molecular changes in mdx muscle throughout progression of disease in muscle groups that are differentially affected by disease [37, 40, 42, 73]. Our approach combines RNA sequencing transcriptomics with ECM-optimized capture techniques to identify SSPN-induced cellular pathways. Findings from our proteomic analysis reveal compensatory alterations in the abundance of collagens in mdxTG muscle, especially the highly glycosylated collagens II, V, and XI. Collagen is the most abundant protein in the ECM, and its high tensile strength allows transmission of forces generated by skeletal muscle fibers. The most abundant collagens in skeletal muscle are collagens I, III, IV, and VI. While collagens were upregulated in the mdxTG, they were primarily collagens not typically observed in skeletal muscle, or those of low abundance, including collagens II, V, and XI [74]. Collagen II is typically found in articular cartilage and intervertebral discs and is highly glycosylated [75, 76]. Collagen XI is also a component of cartilage extracellular matrix [77]. The presence of these bulky disaccharide groups is believed to hinder the formation of highly ordered fibrils and has been shown to increase substrate compliance in co-cultures with collagen I [78]. Collagens V and XI are typically observed only during skeletal muscle development; however, the function of collagen XI in skeletal muscle is unknown. Baghdadi and colleagues [79] recently reported that satellite cells produce collagen V that is critical for the calcitonin receptor and notch signaling cascade, which maintains satellite cells in a quiescent state. Taken together, the changes observed in the mdxTG ECM suggest that upregulation of these developmental and cartilaginous collagens may be beneficial to muscle.

The identification of β-spectrin upregulation as another potential compensatory protein is supported by data from the GRMD dog model of DMD. Nghiem and colleagues [54] identified spectrin as a candidate protein in a study of the cranial sartorius muscle that is spared from dystrophic pathology through compensatory hypertrophy via myostatin signaling. Spectrin is a scaffolding protein that functions to localize and stabilize surface proteins in nonerythroid cells, anchoring them to the cytoskeletal actin network [80]. The repeated triple helical units of the spectrin rod domain would support this function by stabilizing surface proteins and mediating cell-matrix and cell-cell interactions. Interestingly, dystrophin contains similar spectrin-like repeats in its central rod domain and has been reported to bind spectrin [81,82,83]. Data from the IPA and gene enrichment analysis also identify spectrin as a target molecule of signaling pathways associated with a reorganization of the actin cytoskeleton. Together, the upregulation of collagens and scaffolding cytoskeletal components such as spectrin suggests that SSPN overexpression and the increase of adhesion complexes at the membrane may induce a rewiring of signaling between the ECM and the cytoskeleton to stabilize the sarcolemma during contraction and enhance mechanotransduction in order to improve force transmission.

Mechanotransduction has been described in the mdx mouse model in many studies that highlight the role of microtubules, adhesion complexes, mechanosensitive ion channels, and YAP/TAZ signaling in both skeletal and cardiac muscle pathology [70, 84,85,86,87,88,89]. Our study corroborates some of these findings at the transcript level when comparing wild-type and mdx muscle (Supplementary Fig. 2). At the protein level, we observed increased Yap1 localization in the nucleus of mdxTG muscle (Fig. 7). Spectrin has also been shown to be a linker protein that couples mechanical stimuli to downstream signaling networks, including Hippo-YAP/TAZ pathway [56, 57]. Together, these data suggest that SSPN is improving force transmission in part through spectrin-Yap1 mechanotransduction pathways. Our gene set enrichment analysis revealed potential candidate alternative pathways involved in mechanotransduction through regulation of cytoskeleton, focal adhesion, and integrin signaling (Fig. 6). Notably, we identified compensatory changes in RhoA, Rac, and Wnt signaling molecules in the mdxTG muscle. Reciprocal cross talk between the Wnt, Rho GTPase, and the integrin complex mediates cell adhesion signaling cascades and regulation of the cytoskeleton [90,91,92]. Integrins are bidirectional mechano-transducers that detect mechanical cues and translate these signals to affect intracellular and extracellular behavior and response. Through its interaction with adaptor proteins at sites of focal adhesion, the integrin complex regulates the dynamic, cyclic, spatial, and temporal activation of RhoA, Rac1, or Cdc42 leading to actin polymerization and depolymerization which drives force propagation, cell motility, and contractility [90, 93, 94]. RhoA, through downstream effectors, stabilizes actin filaments and regulates the activity of cofilin 2 (CFL2) which together with WD repeat-containing protein 1 (Wdr1) engages in depolymerization and severing of actin filaments, important to replenishing the actin monomer pool, while Rac1 and cdc42 are involved with activation of Wiskott-Aldrich syndrome protein family member 1 (Wasf1) and Abl interactor 2 (ABI2), members of the WAVE complex, to promote actin nucleation and branching [95]. Obscurin (Obscn), a sarcomeric rho-guanine nucleotide exchange factor [96], has been shown to activate RhoA in skeletal muscle [97]. Adducin 2 (Add 2) functions to cap the barbed end of actin filaments and is involved with recruiting spectrin tetramers to actin filaments [98, 99]. Rac1 and cdc42 have also been reported to signal through the DGC complex and may be downregulated during muscle atrophy [100]. Furthermore, RhoA, in addition to Rac1 and Cdc42, regulate transcription by SRF [101]. Interestingly, the transcription factor SRF is in one of the top five significantly inhibited/affected upstream regulators in our IPA analysis between mdx and mdxTG (Table 1). Additionally, we identified compensatory changes in Sox9 in mdxTG muscle with the gene set enrichment analysis, and we validated this at the protein level. Sox9 has been identified as a key transcription factor involved in ECM deposition in cartilage tissue, specifically collagen II [69]. Our findings of compensatory changes in muscle of other cartilage-related proteins including collagen II and collagen XI are novel and suggest a possible new mechanism of improved force transmission in dystrophic muscle.

Conclusions

We generated a data-driven schematic of the mdx muscle rescued by SSPN overexpression that highlights compensatory changes in the ECM and cytoskeleton at both the transcript and protein level (Fig. 8). Our multi-omics data suggest that SSPN rescues dystrophin deficiency partially through a rewiring of cell-matrix interactions that may enhance mechanotransduction signaling cascades and improve lateral force transmission.