Asymmetrical paravertebral muscles fibrosis causes progression of adolescent idiopathic scoliosis via myostatin signalling in fibro-adipogenic progenitors

Abstract

Background

Adolescent idiopathic scoliosis (AIS) is characterized by progressive spinal deformity; however, its underlying mechanisms are poorly understood. While asymmetry of the paravertebral muscles (PVMs) has been linked to AIS progression, its molecular basis remains unclear.

Methods

PVMs biopsies from the concave and convex sides of 10 patients with AIS (Cobb angle >45°, aged 14–17 years) were collected during corrective surgery for histological and gene expression analyses. Bulk RNA sequencing data (GSE254300) from five paired PVMs samples were reanalysed to identify differentially expressed genes, followed by Gene Ontology enrichment and Gene Set Enrichment Analysis. Single-cell RNA sequencing data (PRJNA722100) were used to examine the pathway activation in fibro-adipogenic progenitors (FAPs). In vitro, mouse skeletal muscle-derived FAPs were cultured with or without myostatin and assessed for fibrogenic and adipogenic differentiation. In vivo, unilateral myostatin injections were administered to the PVMs of bipedal mice to induce scoliotic deformities, which were evaluated using radiography and histological analysis.

Results

Histological and transcriptomic analyses revealed increased collagen deposition and extracellular matrix (ECM) remodelling on the concave side of PVMs. Single-cell RNA sequencing identified FAPs with enhanced myostatin pathway activation on the concave side of PVMs. In vitro, myostatin promoted FAPs proliferation and fibro-differentiation via SMAD3 signalling. In vivo, unilateral myostatin overexpression induced asymmetric PVMs fibrosis and spinal curvature in bipedal mice, which were alleviated by pharmacological inhibition of myostatin or SMAD3.

Conclusion

This study revealed increased ECM fibrosis was more pronounced on the concave side of PVMs than on the convex side in patients with AIS. Asymmetrical myostatin-driven fibrogenesis in FAPs was a significant mechanism underlying asymmetrical PVMs fibrosis and scoliosis progression, highlighting the therapeutic potential of targeting the myostatin-SMAD3 axis in AIS.

The translational potential of this article

This study identified asymmetric, myostatin-driven fibrosis in PVMs as a key contributor to AIS pathogenesis. Therapeutic inhibition of myostatin or SMAD3 significantly reduced spinal deformity and muscle fibrosis in bipedal mouse models, suggesting the potential for using myostatin-targeting agents to slow or prevent scoliosis progression in patients with AIS.

Graphical abstract

1. Introduction

Adolescent idiopathic scoliosis (AIS) is a complex, three-dimensional spinal deformity affecting approximately 1–3 % of adolescents, which poses a significant clinical challenge due to its unclear aetiology and potential for progression [1]. Severe cases can lead to cardiopulmonary compromise and adversely affect both physical and psychological well-being [2], underscoring the urgent need for effective therapeutic strategies. While genetic [[3], [4], [5]], environmental [6], hormonal [7], and biomechanical [8,9] factors have been implicated in AIS pathogenesis, the mechanisms underlying AIS development and progression remain unclear.

The paravertebral muscles (PVMs) constitute a critical subsystem for maintaining spinal stability, which is essential for ensuring balanced spinal loading and proper alignment [10]. PVMs asymmetry has been detected by imaging [11], biomechanical tests [12], and muscle histology [13] in patients with AIS. Furthermore, asymmetrical PVMs function is strongly associated with curve severity in AIS [12], although its causal role in initiating or driving the progression of scoliosis remains unclear. The extracellular matrix (ECM) is essential for skeletal muscle function, maintaining tissue architecture [14], transmitting contractile force [15], and forming a specialized niche that regulates muscle stem cells quiescence, activation, and migration [[16], [17], [18], [19]]. While previous studies have reported increased collagen accumulation on the concave side of PVMs in AIS [20,21], there is a lack of direct evidence linking ECM fibrosis defects to AIS.

Fibro-adipogenic progenitors (FAPs), a population of mesenchymal stromal cells resident in skeletal muscle, are central regulators of ECM homeostasis [22,23]. Under physiological conditions, FAPs promote muscle repair by transient expansion and assistance of muscle stem cells [24,25]. However, under chronic injury or pathological conditions, dysregulated FAPs activity results in muscle fibrosis [26,27]. Despite the established role of FAPs in fibrotic conditions, their contribution to PVMs pathology in AIS remains poorly characterized. Myostatin, a member of the transforming growth factor-beta (TGF-β) superfamily, is a potent negative regulator of muscle growth and a key driver of fibrogenesis [28,29]. Previous studies have identified asymmetric myostatin expression in the PVMs of patients with AIS [30]. However, the functional role of myostatin in driving PVMs asymmetry and spinal deformity remains unclear.

In this study, we identified asymmetric ECM fibrosis in the PVMs of patients with AIS, with the concave side exhibiting more pronounced changes. Using single-cell RNA sequencing and histological analyses, we found that FAPs on the concave side displayed enhanced myostatin activity, which promoted fibro-differentiation of FAPs in vitro. In a bipedal mouse model, unilateral overexpression of myostatin in PVMs induced spinal curvature and muscle fibrosis. Notably, pharmacological inhibition of the myostatin/SMAD3 signalling pathway mitigated fibrosis and improved spinal deformity. These findings reveal a novel molecular mechanism driving AIS pathogenesis and identified the myostatin/SMAD3 axis as a promising target for treatment.

2. Materials and methods

2.1. Human samples

All procedures involving human participants were approved by the Institutional Review Board of Peking Union Medical College Hospital (Approval No. I-24PJ1534) and followed the ethical principles of the Declaration of Helsinki. PVMs specimens were collected intraoperatively from 10 patients with AIS (Cobb angle >45°, aged 14–17 years) undergoing posterior spinal correction. The inclusion criteria were confirmed AIS with no prior surgical intervention. The exclusion criteria were neuromuscular or congenital scoliosis and systemic diseases. The characteristics of patients were presented in Supplementary Table 1. Bilateral muscle specimens were obtained from the concave and convex sides of the apical vertebral body. Written informed consent was obtained from each participant and their legal guardians.

2.2. Bulk RNA sequencing of PVMs

RNA sequencing data (GSE254300) from paired PVMs samples (n = 5) were obtained from our previous study [31]. Differentially expressed genes (DEGs) were identified by applying thresholds of | log2 fold change (log2FC) | > 1.5 and P-value <0.05. Gene Ontology enrichment analysis was performed to identify overrepresented biological processes, molecular functions, and cellular components. Gene set enrichment analysis was performed to identify the enriched gene sets.

2.3. Animal models

The animal study protocol was approved by Ethics Committee of the Peking Union Medical College Hospital (D-25YSB0846/BLARC-LAWER-202412009). Animal care and use were performed in accordance with the guidelines of the institution. Bipedal mouse models were established using female C57BL/6 mice (SPF Biotechnology Co., Ltd., Beijing, China) following previously published protocols [32]. Briefly, the forelimbs and tails of mice were removed at 3 weeks of age. With progressively increasing water and food intakes, the mice gradually adopted a standing posture. Compared to traditional quadrupedal models, the animal model more accurately simulates human upright posture and axial spinal loading and is widely used in scoliosis research [33]. Mice were randomly categorized into two groups (n = 7 per group) after surgery: (1) the Unilateral Myostatin group in which 100 nM recombinant myostatin (80 μL, HY-P72632, MCE) was administered by injection into the left PVMs twice weekly for 4 weeks, with 80 μL 0.1 % dimethyl sulfoxide (DMSO, ST038, Beyotime) was injected contralaterally. (2) the Bilateral DMSO group in which 80 μL 0.1 % DMSO was bilaterally injected into the PVMs.

The mouse scoliosis model was induced in bipedal mice (C57BL/6, 3 weeks old) via unilateral injection of recombinant myostatin, as described above. Then, the mice were randomly assigned to three groups (n = 6 per group): myostatin inhibitory protein (MIP) treatment group, SMAD3 inhibitor small molecule 3 (SIS3) treatment group, and control group. The MIP/SIS3 group received injections of 80 μL MIP (10 mg/kg, HY-P10476, MCE) or SIS3 (5 mg/kg, HY-13013, MCE) into the left side of the PVMs every 3.5 days for two weeks, with 80 μL 0.1 % DMSO injected contralaterally. The control group received 80 μL of 0.1 % DMSO, which was bilaterally into the PVMs. The PVMs were harvested exclusively from the region between the 10th thoracic vertebra (T10) and the 1st lumbar vertebra (L1) two weeks after MIP or SIS3 final injection, targeting both the injection site and the apical vertebra of the scoliosis curve in mice.

2.4. Histological and immunofluorescence analysis

PVMs tissues from patients with AIS or bipedal mice were fixed in 4 % paraformaldehyde (#P0099, Beyotime) for 24 h, paraffin-embedded, and sectioned at 3 μm. For haematoxylin and eosin (H&E) staining, sections were deparaffinized, rehydrated, and stained with haematoxylin (#C0107, Beyotime) for 5 min and eosin (#C0109, Beyotime) for 2 min. Fibrosis was quantified using Sirius Red staining (#Y026186, Beyotime) in three random fields per section using ImageJ (NIH, Maryland, USA) by independent double-blinded investigators. For immunofluorescence, sections underwent antigen retrieval in sodium citrate buffer (#HY-B1610N, MCE) at 37 °C for 60 min, peroxidase quenching, blocking with 5 % bovine serum albumin (BSA, #ST2254, Beyotime). Sections were incubated overnight at 4 °C with antibodies: Mouse anti-PDGFRA (1:200, #14-1401-82, Thermo Fisher Scientific), Rabbit anti-laminin (1:200, #ab11575, Abcam), Mouse anti-collagen 1 (1:200, #67288-1-Ig, ProteinTech), Rabbit anti-myostatin (1:200, #19142-1-AP, ProteinTech), and Rabbit anti-p-SMAD3 (1:250, #44-246G, Thermo Fisher Scientific). After washing with phosphate buffer saline, sections were treated with horse radish peroxidase-conjugated secondary antibodies, tyramide amplification, and DAPI counterstaining. FAPs (1.0 × 10^5 cells/well) in 24-well plates were fixed, permeabilized, blocked, and stained with rabbit anti-Ki-67 (1:200, #ab15580, Abcam), rabbit anti-perilipin (1:200, #ab3526, Abcam), and rabbit anti-α-smooth muscle actin (SMA) (1:200, #ab5694, Abcam) following the tissue immunofluorescence protocol.

2.5. X-ray imaging

The mice were imaged using a digital X-ray system (Faxitron UltraFocus; Arizona, USA). Mice were anaesthetized with 2 % isoflurane and positioned prone on the imaging platform to ensure consistent alignment. Radiographs were acquired in the anteroposterior and lateral planes at 40 kVp and 0.2 mA with a 3-s exposure time. Spinal curvature was quantified by measuring the Cobb angle in the anteroposterior plane and the kyphotic angle in the lateral plane by independent double-blinded spinal surgeons.

2.6. Micro-computed tomography imaging

The mice were imaged using a high-resolution micro-computed tomography (micro-CT) scanner (SkyScan 1276, Kontich, Belgium). The animals were euthanized and positioned in a custom holder to ensure consistent alignment. Scans were acquired at 50 kVp and 200 μA using a 0.5 mm aluminum filter, achieving a voxel resolution of 10 μm. Images were reconstructed using NRecon software (v1.7.4.2, Antwerp, Belgium) with beam-hardening correction.

2.7. FAPs isolation and culture

FAPs (PDGFRα–positive cells negative for CD31, CD45, and integrin-α7) were isolated as previously described [26] using fluorescence-activated cell sorting at the Baylor Cell Sorting Core Facility. Isolated FAPs were cultured in growth medium (GM, high-glucose Dulbecco's modified Eagle's medium [DMEM], #11965092, Gibco] supplemented with 2.5 ng/mL basic fibroblast growth factor [#450-33-10UG, Thermo Fisher Scientific], 20 % fetal bovine serum [FBS, #16000044, Gibco], and 10 % heat-inactivated horse serum [#26050070, Thermo Fisher Scientific]). The growth medium was then switched to differentiation medium (DMEM with 10 % FBS, 11.5 mg/mL isobutyl methylxanthine [#HY-12318, MCE], and 1 mg/mL insulin [#HY-P0035, MCE]) supplemented with 1 mM dexamethasone (Dex, #HY-14648, MCE) or 100 nM recombinant myostatin (#HY-P72632, MCE).

2.8. Single-cell RNA sequencing analysis of PVMs

Single-cell RNA sequencing (scRNA-seq) database (PRJNA722100) was analysed, which performed on PVMs samples from the concave and convex sides of the apical vertebral body in two patients with AIS [34]. The Seurat package (version 4.0.6) in R software (version 4.0.5) was used to analyze the scRNA-seq data. Cells with <200 or >6000 detected genes or >10 % mitochondrial gene content were excluded, resulting in the retention of 15,282 cells for further analysis. Gene expression was normalized to total unique molecular identifiers using Seurat's “NormalizeData” function, followed by scaling with “ScaleData” for dimensional reduction. Canonical correlation analysis mitigated batch effects by projecting data into a subspace and computing mutual nearest neighbours as anchors for correction. Dimensionality reduction was performed using “RunUMAP,” followed by clustering with the “FindClusters” function at a resolution of 0.1, which identified seven cell clusters. Cell types were assigned based on canonical marker gene expression, with markers identified using “FindAllMarkers” (p < 0.05, |log2FC| > 0.25). Differential gene expression was assessed using the “FindMarkers” function and Wilcoxon rank-sum test (p < 0.05, |log2FC| > 0.25). DEGs in FAPs were extracted for pathway enrichment analysis to identify the molecular differences in AIS.

2.9. Quantitative Real-Time PCR

Total tissue or cellular RNA was extracted using total RNA extraction kit (#RN001, ES Science). RNA was reverse-transcribed to complementary deoxyribonucleic acid (cDNA) using the High-Capacity cDNA Reverse Transcription Kit (#4368814, Applied Biosystems). cDNA in a 20 μL SYBR Green qPCR kit (#Q411-02, Vazyme Biotech) was used according to the protocol and was performed on a QuantStudio 3 Real-Time PCR System (Applied Biosystems). Primer pairs were designed for COL1A1, COL6A1, ACTA2, GAPDH, Col1a1, Col6a1, Acta2, and Gapdh (Supplementary Table 2). Melt curve analysis confirmed primer specificity. Gene expression was quantified using the 2^{^−ΔΔCt} method.

2.10. Western blot analysis

Equal amounts of protein (20 μg) were separated by 10 % sodium dodecyl sulphate polyacrylamide gel electrophoresis (EpiZyme) and transferred to polyvinylidene fluoride membranes (Bio-Rad). Membranes were blocked with 5 % BSA for 1 h at room temperature, then incubated overnight at 4 °C with primary antibodies: Rabbit anti-collagen 1 (1:1000, #14695-1-AP, ProteinTech), Rabbit anti-αSMA (1:1000, #ab5694, Abcam), Rabbit anti-SMAD3 (1:1000, #51–1500, Thermo Fisher Scientific), and Rabbit anti-p-SMAD3 (1:1000, #44-246G, Thermo Fisher Scientific). Rabbit anti-GAPDH (#8884, Cell Signaling Technology) was used as a loading control. After washing with Tris-buffered saline with Tween-20, the membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody for 1 h at room temperature. Protein bands were visualized using enhanced chemiluminescence (ECL, #P0018S, Beyotime) and imaged using the ChemiDoc Imaging System (Tanon, Shanghai, China). Band intensities were quantified using ImageJ and normalized to GAPDH levels.

2.11. Smad3 knockdown and overexpression in FAPs in vitro

Small interfering ribonucleic acid (siRNA) targeting mouse Smad3 were obtained from Shanghai GeneChem Co. Ltd., (Shanghai, China). The Smad3 siRNA sense sequence was 5′- GGATTGAGCTGCACCTGAA-3'. Scrambled siRNA with no homology to mouse genes was used as a negative control. FAPs were cultured in GM at 37 °C in 5 % CO₂, seeded in six-well plates at 70 % confluence, and transfected with 50 nM si-Smad3 or scrambled siRNA using Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol. The SMAD3 overexpression and negative control plasmid was synthesized by Shanghai GeneChem Co. Ltd., (Shanghai, China). FAPs were subjected to plasmid transfection according to the manufacturer's instructions.

2.12. Statistical analysis

All data are expressed as the mean ± standard deviation. Paired t-tests were used to compare the outcomes on the concave and convex sides of the PVMs and Unpaired t-tests were used to compare the outcomes in two independent groups. One-way analysis of variance with Tukey's post-hoc test was used for three or more groups. Statistical significance was defined as p < 0.05 (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001). All analyses were performed using GraphPad Prism 9.0.

3. Results

3.1. AIS is associated with asymmetrical ECM fibrosis of PVMs

To investigate whether AIS is associated with ECM fibrosis between the concave and convex sides of the PVMs, we reanalysed the RNA sequencing data of PVMs samples from both sides of the apical vertebral body, sourced from our previous study (GSE254300). Differential genes were substantially enriched in biological processes and cellular components related to ECM turnover and remodelling on the concave side of the PVMs (Fig. 1A). GSEA supported the differential ECM activity and highlighted that fibrosis was enhanced on the concave side of the PVMs compared to the convex side (Fig. 1B). These findings supported those of earlier studies reporting that AIS is associated with unbalanced PVMs [20,21]. Next, we extended our analysis to tissue sections from PVMs biopsies obtained from patients with AIS. These results also revealed significantly increased fibrosis and deposition of collagen1 (COL1) on the concave side of the PVMs compared to the convex side (Fig. 1C–E). These results were supported by quantitative PCR (qPCR), which confirmed the upregulation of fibrosis-associated genes (Fig. 1F) on the concave side of the PVMs. Collectively, our findings supported the hypothesis that AIS was associated with unbalanced ECM fibrosis on both sides of the PVMs.

Fig. 1.

Adolescent idiopathic scoliosis (AIS) is associated with asymmetrical extracellular matrix (ECM) fibrosis of paravertebral muscles (PVMs) (A) Gene Ontology enrichment analysis revealing ECM activity and fibrosis were enhanced on the concave side of the PVMs. (B) Gene Set Enrichment Analysis (GSEA) revealing collagen-containing ECM and ECM structural constituents were enhanced on concave side of the PVMs. (C) Representative histological images of Haematoxylin and Eosin (H&E) and Sirius Red staining of the PVMs from patients with AIS. Scale bars: 200 μm. (D) Immunofluorescence staining for Laminin (green) and COL1 (red) in PVMs from the concave and convex sides. Scale bars: 100 μm. (E) Quantification of fibrosis area according to Sirius Red staining data, n = 10. Paired t-tests, ∗∗p < 0.01. (F) COL1A1, COL6A1, and ACTA2 mRNA expression on the concave and convex sides of the PVMs in patients with AIS. n = 10. Paired t-tests, ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

3.2. Asymmetrical myostatin activation of FAPs in the PVMs

To investigate the mechanism underlying asymmetric ECM fibrosis on the concave and convex sides of the PVMs, we analysed an scRNA-seq dataset (PRJNA722100) [34]. The cells were classified into seven distinct clusters based on their transcriptional profiles (Fig. 2A and B). The marker gene characteristics and proportions of each cell type are presented in Fig. 2C and D. Given that skeletal muscle fibrosis is primarily driven by FAPs, we focused on this cell population, which was identified by the high expression of PDGFRA and DCN [35] (Fig. 2E). Pathway analysis of FAPs revealed upregulated pathways, including transmembrane receptor protein serine/threonine kinase signalling pathway, TGF-β receptor superfamily signalling pathway, and regulation of cellular responses to growth factor stimuli (Fig. 2F). Notably, differential gene expression analysis identified significantly increased expression of MSTN, a member of the TGF-β superfamily and a ligand for receptor protein serine/threonine kinases [36], on the concave side of the PVMs than on the convex side (Fig. 2G and H). To validate these findings, we examined tissue sections from PVMs biopsies of patients with AIS. Immunofluorescence analysis confirmed a significantly higher percentage of PDGFRa- and myostatin-positive cells on the concave side of the PVMs than on the convex side (Fig. 2I and J), which were spatially close to the COL1 positive area as shown by Laminin/COL1 staining of serial adjacent sections (Supplemental Fig. 1). These results suggested that asymmetric myostatin expression in PVMs may promote FAPs differentiation into fibrocytes, thereby contributing to enhanced ECM fibrosis on the concave side. These results were supported by western blot analysis, which confirmed the upregulation of myostatin protein on the concave side of the PVMs (Supplemental Fig. 2). We hypothesized that myostatin asymmetry underlies the observed ECM fibrosis imbalance between the concave and convex sides of PVMs in AIS.

Fig. 2.

Asymmetrical myostatin activation of fibro-adipogenic progenitors (FAPs) in PVMs (A) The UMAP plots of main cell types for the PVMs. FAPs = fibro-adipogenic progenitors; SMCs = smooth muscle cells; MuSCs = muscle stem cells. (B) The UMAP plots of main cell types for the PVMs from the concave and convex side. (C) Dot plot of highly expressed genes in each cell type. (D) Percentage of each cell type. (E) UMAP plots of genes marked FAPs. (F) Pathway enrichment analysis of differentially expressed genes in FAPs. (G) Hierarchical clustering heatmap illustrating differentially expressed genes (DEGs) between the concave and convex side of PVMs, with MSTN (myostatin) marked (red arrow). (H)Volcano plot illustrating DEGs between the concave and convex side of the PVMs. (I) Co-localization of Laminin (red), PDGFRα (green), and Myostatin (yellow) on the concave and convex side of the PVMs in patients with AIS. Scale bar: 50 μm. (J) Quantification of data I, n = 5. Paired t-tests, ∗∗p < 0.01.

3.3. Myostatin induces the differentiation of FAPs into fibrocytes in vitro

To determine whether myostatin promotes the differentiation of FAPs into fibrocytes in vitro, we isolated FAPs from mouse skeletal muscle tissue using fluorescence-activated cell sorting based on established markers [26] (Fig. 3A). To explore whether myostatin promotes the proliferation of FAPs, isolated FAPs were treated with 100 nM recombinant myostatin, which significantly increased their proliferation rate as indicated by a higher proportion of Ki-67-positive cells compared to the control group (Fig. 3B and C). Previous studies have shown that FAPs differentiate into adipocytes when cultured in a differentiation medium containing dexamethasone (Dex), a known inducer of myostatin expression in muscles [37]. To investigate whether myostatin mimics Dex in promoting adipogenesis, we cultured FAPs in a differentiation medium lacking Dex but supplemented with 100 nM myostatin. In contrast to the Dex-containing medium, which induced adipocyte formation, myostatin supplementation did not promote adipogenesis (Fig. 3D and E). Instead, myostatin-treated FAPs predominantly differentiated into α-SMA-positive cells, a hallmark of fibrotic differentiation (Fig. 3F and G). To further characterize these effects, we analysed the protein and gene expression levels of fibrotic markers. After 1 week in a differentiation medium supplemented with myostatin instead of Dex, the protein levels of fibrosis markers were significantly increased (Fig. 3H and I). Consistent with these findings, the mRNA levels of fibrotic markers, assessed after 2 days of myostatin treatment, were also upregulated (Fig. 3J). Collectively, these results demonstrated that myostatin induces fibro-differentiation in FAPs rather than adipogenesis in vitro. These findings suggested that myostatin plays a critical role in driving fibrotic remodelling, potentially contributing to the asymmetric fibrosis observed in PVMs and the progression of AIS.

Fig. 3.

Myostatin promotes fibro-differentiation of FAPs in vitro. (A) Flow cytometry analysis for isolating FAPs from mouse skeletal muscle using fluorescence-activated cell sorting based on PDGFRα, CD31, CD45, and integrin-α7 expression. The P4 gates correspond to FAPs populations. (B) Immunofluorescence staining for ki-67 (red) in FAPs with or without 100 nM myostatin treatment for 2 days. Scale bar: 50 μm. (C) Quantification of data B. n = 3. Unpaired t-tests, ∗∗p < 0.01. (D) Oil Red O staining (red) and immunofluorescence staining for Perilipin (green) in FAPs cultured in differentiation medium with or without dexamethasone (Dex) or myostatin for 1 week. Scale bar: 50 μm. (E) Quantification of data D. n = 3. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗∗∗p < 0.001. (F) Immunofluorescence staining for α-SMA (red) in FAPs cultured in differentiation medium with or without Dex or myostatin for 1 week. Scale bar: 50 μm. (G) Quantification of data F, n = 3. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗∗p < 0.01. (H) Western blot analysis of α-SMA and COL1 in FAPs treated with Dex or myostatin for 1 week. (I) Quantification of data H, n = 3. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗∗∗p < 0.001. (J) qPCR analysis of fibrotic gene expression (Col1a1, Col6a1, Acta2) in FAPs after 2 days of Dex or myostatin treatment. n = 3. One-way ANOVA with Tukey's post-hoc test, ns = not significant; ∗∗∗p < 0.001.

3.4. Asymmetric activation of myostatin in PVMs leads to scoliosis

To investigate whether asymmetric activation of the myostatin signalling pathway promotes asymmetric ECM fibrosis of PVMs and induces scoliosis in vivo, we used a mouse model. A bipedal mouse model was generated using 3-week-old female mice. Recombinant myostatin (100 nM) was injected into the left side of PVMs and 0.1 % dimethyl sulfoxide (DMSO) was injected into the right side (Fig. 4A). Injections were administered twice weekly for 4 weeks to simulate unilateral elevation of myostatin signalling. The control group received bilateral injections of 0.1 % DMSO. Spinal alignment was assessed using X-ray and micro-CT imaging 2 weeks after the final injection (Fig. 4A). Compared to the bilateral DMSO group, mice injected with unilateral myostatin exhibited significantly greater spinal deformities in both the coronal and sagittal planes (Fig. 4BD, E). Additionally, the unilateral myostatin injection group displayed an asymmetric thoracic cage, indicative of scoliosis-like deformities (Fig. 4B). These results indicated that asymmetric myostatin expression on the two sides of the PVMs aggravated scoliosis in mice. After euthanasia, PVMs were harvested from both sides for histological analyses. As expected, the myostatin-injected side showed a significantly larger fibrotic area than the DMSO-injected side in the unilateral myostatin group (Fig. 4C and F). Consistent with these findings, the protein and mRNA levels of fibrotic markers were also significantly upregulated on the myostatin-injected side of the PVMs (Fig. 4G–J). These findings indicated that unilateral activation of myostatin signalling in PVMs enhances ECM fibrosis and promotes scoliosis-like phenotypes in vivo, supporting the role of myostatin in AIS progression.

Fig. 4.

Asymmetric activation of myostatin in the PVMs leads to scoliosis in a bipedal mouse model. (A) Schematic of the experimental design. Three-week-old bipedal female mice received either bilateral injections of dimethyl sulfoxide (DMSO) (Bilateral DMSO group) or injections of recombinant myostatin (left side) and DMSO (right side) (Unilateral Myostatin group) into the PVMs, administered twice weekly for 4 weeks. Spine evaluation and tissue harvested were performed 2 weeks after the final injection. (B) Representative X-ray and micro-CT images in the bilateral DMSO group and the unilateral myostatin group. (C) Representative H&E and Sirius Red staining of the PVMs in the bilateral DMSO group and the unilateral myostatin group. Scale bar: 200 μm. (D) Quantification of coronal spinal curvature in the two groups. n = 7. Unpaired t-tests, ∗∗∗p < 0.001. (E) Quantification of sagittal spinal curvature in the two groups. n = 7. Unpaired t-tests, ∗∗P < 0.01. (F) Quantification of fibrotic area based on Sirius Red staining, shown in data (C) above. n = 7. One-way ANOVA with Tukey's post-hoc test, ∗∗P < 0.01. (G) Western blot analysis of COL1 and α-SMA in the left and right side of the PVMs of the Unilateral Myostatin group. (H) Laminin (red) and COL1 (green) co-immunofluorescence in the left and right side of PVMs in the Unilateral Myostatin group. Scale bar: 100 μm. (I) Quantification of data G, n = 7. Paired t-tests, ∗∗p < 0.01. (J) qPCR analysis of Col1a1, Col6a1 and Acta2 in the left and right side of the PVMs of the two groups. n = 7. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗P < 0.05; ∗∗∗p < 0.001.

3.5. Myostatin promotes fibro-differentiation of FAPs through SMAD3 activation

To elucidate the mechanism by which myostatin induces FAPs differentiation into fibrocytes, we focused on SMAD3, a key intracellular mediator of TGF-β family signalling that regulates gene expression involved in cell proliferation, differentiation, and fibrosis [38]. We first analysed tissue sections of PVMs biopsies from AIS patients and found that concave side of PVMs exhibited a greater proportion of myostatin- and p-SMAD3-positive cells than convex side (Fig. 5A). Similarly, in unilateral myostatin group of bipedal mouse model, the myostatin-injection side of PVM showed a higher proportion of myostatin- and p-SMAD3-positive cells compared to the DMSO injection side (Supplemental Fig. 3). To confirm these findings in vitro, FAPs isolated from mouse skeletal muscles were treated with 100 nM recombinant myostatin for two days. Protein levels of p-SMAD3 were significantly upregulated in myostatin-treated FAPs compared with the control group (Fig. 5B and C), indicating that myostatin enhanced SMAD3 activation in FAPs. To investigate whether myostatin promotes fibro-differentiation of FAPs via SMAD3 activation, Smad3 expression was silenced using siRNA and overexpressed by plasmid transfection. Transfection with Smad3 siRNA significantly reduced total SMAD3 and p-SMAD3 levels in FAPs, both with and without myostatin treatment (Fig. 5D–F). Smad3 knockdown markedly reduced the expression of fibrotic markers COL1 and α-SMA, even in the presence of myostatin (Fig. 5D–F). Conversely, the expression of fibrosis-related proteins increased when Smad3 was overexpressed (Fig. 5G and H). These results demonstrated that myostatin promotes FAPs fibro-differentiation by activating SMAD3 signalling.

Fig. 5.

Myostatin promotes FAPs differentiation into fibrocytes via SMAD3 activation. (A) Myostatin (yellow) and p-SMAD3 (cyan) co-immunofluorescence in the concave and convex sides of PVMs from AIS Patients. Scale bars: 200 μm. Right: Quantification of Myostatin and p-SMAD3 positive cells. n = 10. Paired t-tests, ∗∗∗p < 0.001. (B) Western blot analysis of SMAD3 and p-SMAD3 in FAPs treated without or with 100 nM recombinant myostatin for 48 h. (C) Quantitation of data B, n = 3. Unpaired t-tests, ns = no significant; ∗∗∗p < 0.001. (D) Western blot analysis of SMAD3, p-SMAD3, COL1, aSMA in FAPs transfected with control siRNA or siRNA-Smad3, with or without myostatin treatment for 1 week. (E) Immunofluorescence staining for aSMA in FAPs transfected with control siRNA or siRNA-Smad3, with or without myostatin treatment for 1 week. (F) Quantification of data D, n = 3. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗∗P < 0.01; ∗∗∗p < 0.001. (G) Western blot analysis of SMAD3, p-SMAD3, COL1 and aSMA in FAPs transfected with overexpress plasmid-Smad3 with or not with myostatin for 1 week. (H) Quantification of data G, n = 3. One-way ANOVA with Tukey's post-hoc test, ∗p < 0.05; ∗∗p < 0.01;∗∗p < 0.001.

3.6. Inhibition of myostatin–SMAD3 signalling attenuates PVMs fibrosis and scoliosis progression in a mouse model

Given that the myostatin/SMAD3 signalling pathway promotes FAPs differentiation into fibrocytes, contributing to muscle fibrosis both in vitro and in vivo, we investigated whether inhibiting this pathway could mitigate scoliosis in mouse models. A mouse model of scoliosis was established via left side myostatin injection into the PVMs, as previously described. To inhibit myostatin/SMAD3 signalling, MIP or SIS3, an inhibitor of SMAD3, was injected into the concave (left) side of the PVMs once every 3.5 days for two weeks, while DMSO was injected into the convex (right) side. The control group received bilateral DMSO injections according to the same regimen (Fig. 6A). Following euthanasia, the bilateral PVMs were harvested for histological analysis. In the MIP- or SIS3-treated left PVMs, the fibrotic area and COL1 deposition were significantly reduced compared to those in the left PVMs of the DMSO-injected control group (Fig. 6B–D). Notably, MIP or SIS3 treatment improved spinal alignment, as assessed by X-ray, and attenuated the progression of coronal curvature and kyphosis compared to the control group (Fig. 6E and F). These findings indicated that inhibiting the myostatin/SMAD3 signalling pathway, which drives FAPs fibro-differentiation and muscle fibrosis, ameliorates fibrosis of PVMs and scoliosis severity in vivo, suggesting that the myostatin/SMAD3 signalling pathway is a potential therapeutic target for AIS.

Fig. 6.

Inhibition of myostatin–SMAD3 signalling attenuates paravertebral muscles (PVMs) fibrosis and scoliosis progression in a mouse model (A): Schematic of the experimental design. Three-week-old bipedal female mice received left-side injections of recombinant myostatin and right-side injections of DMSO into the PVMs twice weekly for 4 weeks. Mice were randomly assigned to three groups (n = 6 per group) after 2 weeks: the myostatin inhibitory protein treatment group (MIP group), the SMAD3 inhibitor small molecule 3 treatment group (SIS3 group), and the control group. The MIP/SIS3 group received injections of MIP or SIS3 into the left side of the PVMs every 3.5 days for 2 weeks. The control group received bilateral injections of DMSO into the PVMs. Spine evaluation and tissue harvesting were conducted 2 weeks after the final injection. (B) Representative HE and Sirius Red staining of the left side PVMs in the control group, MIP group, and SIS3 group. Scale bar: 200 μm. (C) Quantification of fibrotic area based on Sirius Red staining in data (B) above. n = 6. One-way ANOVA with Tukey's post-hoc test, ns = no significant; ∗∗∗P < 0.001. (D) Laminin (red) and COL1 (green) co-immunofluorescence in the left side PVMs in the control group, MIP group and SIS3 group. Scale bar: 100 μm. (E) Representative X-ray images in the control group, MIP group and SIS3 group at 6 weeks and 10 weeks. (F) Quantification of the coronal and sagittal spinal curvature in the three groups at 6 weeks and 10 weeks. n = 6. One-way ANOVA with Tukey's post-hoc test, ns = not significant; ∗P < 0.05; ∗∗P < 0.01.

4. Discussion

This study revealed a novel pathogenic mechanism underlying AIS, identifying asymmetric fibrosis of the PVMs as a key driver of disease onset and progression. Although the aetiology of AIS is widely acknowledged as multifactorial [6]–involving genetic predisposition [[3], [4], [5]], mechanical loading [8,9], and environmental influences [6]–our data identified muscle pathology as a potentially initiating factor. Specifically, we demonstrated that elevated myostatin expression on the concave side of PVMs induces ECM fibrosis via fibro-differentiation of FAPs mediated by SMAD3 signalling (Fig. 7).

Fig. 7.

Schematic diagram of asymmetrical myostatin-SMAD3 activation in FAPs results in paravertebral muscle fibrosis and progression of AIS.

Previous studies have reported imbalances in PVMs during AIS progression [39,40]. However, the association between asymmetric PVMs and AIS is supported only by clinical correlations and indirect evidence. Previous studies have reported a reduced cross-sectional area of muscle fibres on the concave side of PVMs in patients with AIS, with decreased oestrogen receptor [34] or Tent5a [41] expression in muscle stem cells, which compromises muscle regeneration. Additionally, the ECM is essential for skeletal muscle function, which maintains tissue architecture [14], transmits contractile force [15], and establishes a specialized niche that regulates the quiescence, activation, and migration of muscle stem cells [[16], [17], [18]]. However, the role of ECM remodelling in PVMs asymmetry in AIS has been largely overlooked, as most studies have focused primarily on imaging-based assessments [42,43] or histological changes [13] rather than on the underlying molecular mechanisms. This study provides the first direct evidence linking ECM fibrosis caused by dysregulated myostatin signalling in FAPs to the initiation and progression of AIS. We demonstrated increased fibrosis on the concave side of the PVMs, which may impair muscle function and regenerative capacity, thereby contributing to the progression of AIS.

FAPs, which are key regulators of ECM production and remodelling [22,23], support muscle repair under physiological conditions by maintaining tissue homeostasis and promoting muscle stem cells function [24]. However, insufficient clearance of FAPs following injury can result in excessive ECM deposition and fibrosis, as observed in dystrophic muscle conditions [26,27]. Despite their critical role in muscle homeostasis, FAPs have been largely unexplored in previous studies on PVMs in AIS. In this study, although FAPs abundance did not differ between the concave and convex sides, FAPs on the concave side exhibited elevated myostatin signalling. Elevated myostatin levels promoted fibro-differentiation of FAPs, thereby contributing to the initiation and progression of AIS. These findings provided the first evidence that FAPs-driven asymmetric fibrosis of PVMs may represent a key pathogenic mechanism underlying AIS.

Although previous studies have reported asymmetric myostatin expression in the PVMs of AIS [30], its functional role remains unclear. Myostatin has historically been regarded as a negative regulator of skeletal muscle homeostasis by inhibiting muscle stem cells myogenesis [44]. However, its role in promoting fibroblast differentiation cannot be overlooked. Previous studies have demonstrated reduced muscle fibrosis in mice lacking myostatin [45] or treated with myostatin antagonists [46]. Our study integrated single-cell and bulk RNA sequencing, histological evaluation, and functional in vivo and in vitro experiments to provide compelling evidence for the causal role of myostatin in AIS pathogenesis. In AIS patients of this study, correlation analysis revealed a moderate positive relationship (R² = 0.3240, p = 0.0859) between the myostatin expression ratio (concave/convex) and the main curve Cobb angle, indicating a potential association between myostatin asymmetry and scoliosis severity (Supplemental Fig. 4). In a bipedal mouse model, unilateral overexpression of myostatin in PVMs induced scoliotic deformity and asymmetric muscle fibrosis. These findings demonstrated that localized myostatin dysregulation in PVMs is sufficient to initiate spinal curvature and drive progression. In future studies, we will establish an animal model with unilateral overexpression of myostatin in PVMs to investigate whether spinal curvature is driven solely by localized myostatin activity.

Critically, pharmacological inhibition of the myostatin/SMAD3 axis significantly attenuated PVMs fibrosis and delayed progression of spinal deformity in bipedal mice. These results highlighted the therapeutic potential of targeting the myostatin/SMAD3 pathway in AIS. Current conservative management, such as bracing, focuses primarily on mechanical stabilization and does not address the underlying molecular pathology [47]. Our findings suggested that modulating myostatin signalling may represent a novel and complementary therapeutic approach to halt or reverse the progression of AIS. However, systemic inhibition of myostatin in muscle tissue may lead to off-target effects such as muscle hypertrophy [48]. Developing targeted delivery strategies to minimize systemic exposure represents a highly promising therapeutic approach. A previous study showed that lipid-coated mesoporous silica nanoparticles enhance delivery to FAPs [49]. This approach may enable FAPs-specific inhibition while reducing the unintended effects on other cells.

Except for SMAD3 signalling, myostatin potentially activates non-canonical TGF-β pathways, such as YAP/TAZ and Akt/mTOR, which were implicated in muscle fibrosis and remodelling [50,51]. YAP/TAZ signalling has been demonstrated to promote fibroblast activation and extracellular matrix (ECM) deposition in fibrotic conditions [52], whereas Akt/mTOR signalling has been shown to be involved in regulating muscle stem cells proliferation and hypertrophy [53]. Although the roles of non-canonical TGF-β pathways remain underexplored, these pathways may interact with myostatin/SMAD3 signalling to contribute to PVMs asymmetry in AIS. In future study, we will explore the interplay between myostatin/SMAD3 and non-canonical TGF-β pathways to further elucidate the molecular complexity of AIS and identify additional therapeutic targets.

This study also has several directions for future research. First, limited cohort size reflects the challenge of obtaining paired PVMs biopsies from AIS patients. The preliminary findings warrant validation in larger cohorts to establish generalisability. Second, we did not assess potential neuromuscular confounders, such as altered neural input or changes in muscle contractility, which may contribute to PVMs asymmetry in AIS and require further investigation. Moreover, although pharmacological inhibition of myostatin or SMAD3 mitigated fibrosis and scoliosis in our mouse model, the long-term safety of myostatin/SMAD3 blockade remains unknown.

In conclusion, our study demonstrated that asymmetric myostatin expression induces asymmetric FAPs-mediated fibrosis in the PVMs of patients with AIS, contributing to scoliosis progression. By elucidating this mechanism, we identified myostatin-SMAD3 signalling as a potential therapeutic target for AIS. These findings advance our understanding of AIS aetiology and establish a foundation for developing novel treatment strategies to attenuate the progression of this common spinal deformity.

Author contributions

All authors have critically reviewed and approved the final manuscript to be published. Conceptualization: HS and JS; Data curation: HS, YH and HC; Funding acquisition: JS; Investigation: HZ, JD; Methodology: HS, HC; Project administration: JS; Resources: HZ, JZ, XH, and HC; Software: HS; Supervision: JS; Writing - original draft: HS; Writing - review & editing: HS and JS.

Declaration of generative AI in scientific writing

The authors declare that no generative artificial intelligence (AI) tools were used in the writing, editing, or data analysis of this manuscript.

Funding statement

This study was supported by the Major Program of the National Natural Science Foundation of China (82230083).

Declaration of competing interest

The research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

The bulk RNA sequencing data analysed in this study are publicly available in the Gene Expression Omnibus (GEO) under accession number GSE254300. The single-cell RNA sequencing data are accessible in the NCBI BioProject database under accession number PRJNA722100. All other data are available in the main text or the supplementary materials.