Woven dECM yarn scaffolds enable volumetric muscle loss regeneration via immunomodulation

Abstract

Volumetric muscle loss (VML) represents a significant unmet clinical need in current medical practice. Decellularized extracellular matrix (dECM) offers a promising clinical approach. However, its inherent architecture impedes adaptive topological guidance for rapid cellular infiltration, spatial organization, and coordinated immune response, limiting functional restoration. Here, we report a dECM yarn scaffold (YS) fabricated by 3D weaving of rotary-cut yarns, achieving precise structural control, full interconnectivity and high porosity. In murine VML models, YS significantly improved vascularization, innervation, muscle mass, and strength restoration. Single-nucleus RNA-sequencing identified decreased SPP1⁺ neutrophil infiltration alongside increased CD206⁺/IGF-1⁺ macrophages, whose enhanced IGF-1 secretion stimulated PAX7⁺ muscle cell growth via amplified IGF-1R signaling. The robust regenerative efficacy and translational potential of YS was further confirmed in a canine VML model. Our study shows that restructured dECM scaffolds address structural constraints to enable effective in situ muscle regeneration, while simultaneously establishing a novel scaffold platform for regenerative medicine.

Volumetric muscle loss remains a significant unmet clinical need. Herein, the authors report a yarn scaffold based on decellularized extracellular matrix fabricated via 3D weaving of rotary-cut yarns, which achieves precise structural control, full interconnectivity, and high porosity.

Introduction

Skeletal muscle constitutes over 40% of human body mass and is primarily responsible for mobility, posture maintenance, thermoregulation, organ protection, and metabolic functions¹. Skeletal muscle possesses innate regenerative capacity, which enables self-repair after minor injuries². VML is defined as a loss exceeding 20% of the original muscle mass and can result from trauma, tumor resection, fractures, or degenerative diseases. VML compromises the regenerative capability of muscle tissue and can lead to scar tissue formation as opposed to functional muscle tissue. To date, clinical VML treatments remain limited to fibrotic tissue debridement and autologous muscle transplantation. However, these approaches have their own constraints, such as donor-site morbidity, necrosis, infection, limited functional recovery, and inadequate tissue mass availability^3,4. Consequently, the development of novel therapeutic approaches for VML is imperative to address these critical limitations⁵.

Compared to cell therapy, in vitro tissue engineering, and gene therapy, biomaterial-mediated in situ tissue regeneration offers advantages in procedural simplicity and clinical translation⁶. The biochemical cues conveyed by dECM have demonstrated special efficacy to both natural and synthetic materials in reprogramming immune microenvironments and facilitating constructive tissue remodeling⁷. Achieving in situ regeneration of VML with implanted biomaterial scaffolds necessitates a comprehensive balance of pore structure (encompassing inter-connectivity, alignment, and size), degradation behavior, and mechanical performance, a combination that has long remained a technical challenge^6,8. Consequently, researchers have developed diverse dECM fabrication strategies aimed at reconstructing muscle structure and restoring function in VML^9,10. Electrospun dECM-oriented nanofibers were shown to enhance myocyte growth and myotube formation in vitro, yet cell infiltration into the scaffold remained challenging¹¹. Muscle-derived dECM without further processing exhibited limited cell infiltration due to low porosity and small pore size³. Incorporation of pores using various techniques yielded high-porosity, and large-pore dECM materials accelerated vascularization to establish a pro-regenerative microenvironment, but ultimately lacked oriented guidance structures¹². Printed tissue-specific dECM bioinks were found to enable structural control over pores but struggled to overcome degradation and mechanical limitations¹³. Other efforts to improve dECM-directed regeneration include the incorporation of additional cytokines. Muscle dECM scaffolds with controlled bFGF/IGF-1 release promoted functional recovery and strength in murine models of VML¹⁴. The true clinical translational potential of diverse dECM scaffolds from these strategies remains unclear and necessitates validation in larger animal models⁵. Commercial dECM scaffolds, such as decellularized small intestinal submucosa (dSIS), urinary bladder matrix, and acellular dermal matrix, have demonstrated muscle regeneration potential in clinical trials; however, their reliance on microporosity from decellularization for cell migration and nutrient transport appears to restrict implantable scaffold dimensions and regenerative efficacy^15,16. Thus, proposed strategies to enhance pore size and porosity have garnered increasing attention, including methods such as freeze-thaw cycling, microchannel introduction, and hydrogel printing, but their overall efficacy remains suboptimal^9,17–19. Consequently, overcoming such technical hurdles that impede the development of structurally adapted dECM scaffolds for VML therapy presents a significant bottleneck^4,10.

Fiber-based textile platforms represent viable strategies for engineering three-dimensional (3D) porous scaffolds, capitalizing on their inherent interconnected pore networks and fiber-mediated contact guidance to support cell migration and alignment²⁰. Although cell sheets^21,22, amniotic membranes²³, and small intestine submucosa (SIS)²⁴ can be processed into fibrous formats suitable for sutures, patches, and vascular grafts, the limited dimensions of donor tissues constrain the fabrication of continuous yarns at clinically relevant lengths. In contrast to conventional weaving techniques, which lack the capability for three-dimensional weaving, the emergence of 3D weaving has enabled high-precision spatial control over scaffold architecture across both macro- and micro-scales^25,26.

By leveraging omics approaches, researchers are able to decode the regulatory role of novel biomaterials in tissue regeneration. These techniques help uncover distinct cellular heterogeneity critical to regeneration, identify key target cells influenced by the biomaterial, reconstruct cellular differentiation trajectories, and elucidate the temporal dynamics of the regenerative process²⁷. Such insights provide a foundation for the rational design and optimization of biomaterials, thereby accelerating their clinical translation^28,29.

Here, we developed a rotary-cutting technique to fabricate continuous porcine SIS-dECM yarns. By integrating 3D printing with textile engineering, we achieved precise control over the macro- and micro-architectural features of dECM yarn scaffolds (YS), which was readily adaptable to Good Manufacturing Practice (GMP) conditions. In vitro evaluations quantified YS-mediated effects on cellular migration, proliferation, and differentiation. In vivo assessments via subcutaneous implantation measured tissue ingrowth, vascularization, and macrophage polarization. Single-nucleus RNA-sequencing elucidated the underlying pro-regenerative mechanisms. Efficacy in enabling in situ regeneration was validated through both small- and large-animal VML models. Collectively, our findings establish YS as a highly efficient, tissue-inducible biomaterial with attractive clinical-translation potential for VML therapy.

Results

Design, fabrication, and physical characterization of 3D weaving YS

Continuous dECM yarns were fabricated from dSIS through a combination of rotary-cutting and twisting. Three-dimensional YS were then constructed from the dECM yarns using 3D weaving technology (Fig. 1a). The porcine SIS was decellularized through integrated physical, chemical, and biological methods to yield dSIS. H&E staining confirmed retention of abundant ECM components post-decellularization, while DAPI staining demonstrated the effective removal of nuclear material. Quantitative analysis revealed elimination of 98.6% of nuclei in dSIS (Fig. 1b, c). DNA quantification assays showed DNA content reduced from 724.6 ± 55.5 ng/mg to 37.9 ± 7.2 ng/mg, meeting the decellularization threshold (<50 ng/mg)³⁰ (Fig. 1d). The presence of epitopes of α-galactosidase (α-Gal) was determined by α-Gal activity assay, which showed significant decreases post-processing (Fig. 1e). Proteomic analysis identified preservation of key ECM proteins including collagens, fibronectin, glycoproteins, laminins, proteoglycans and others (Fig. 1f, Supplementary Fig. 1a–d, Supplementary Data 1). Subsequent analyses revealed that some proteins associated with chemotaxis, regeneration, and immunomodulation were preserved (Supplementary Fig. 1e–g). Enrichment analysis revealed that the differential proteins were mainly decellularization-related cytoplasmic and nuclear proteins (Supplementary Fig. 1h).

Fig. 1. Fabrication and characterization of biochemical/physical properties of the 3D weaving YS and MS.

a Schematic of YS preparation from tubular dSIS via sequential rotary-cutting, twisting, and weaving. b H&E and DAPI staining of porcine SIS pre- and post-decellularization. c–e Quantification of cell number, DNA content, and α-galactosidase activity pre-/post-decellularization (n = 9 biologically independent samples). f Proteomic analysis of ECM ratio in dSIS (n = 3 biologically independent samples). g–i Photo of custom-assembled rotary-cutting, twisting and weaving device. j, Tunable ribbon widths achieved by adjusting rotary-cutting speed. k SEM images showing the structure of yarn made of 5-mm ribbons twisted at 2.5, 5.0, and 7.5 revolutions per centimeter (r/cm). l, Length quantification of yarns produced from 2-mm, 5-mm, and 10-mm width ribbons cut from 20-cm tubular SIS (n = 9 biologically independent samples). m Breaking stress of yarns produced from 5-mm width ribbons under different twisting rates (n = 9 biologically independent samples). n Porosity of YS and MS (n = 3 biologically independent samples). o Macroscopic photo of lateral and top view of the YS and MS. p Toluidine blue assay detecting YS and MS permeability with corresponding SolidWorks fluid dynamics simulations modeling flow distribution. q Shape deformation and mechanical properties of YS and MS compressed by 30% strain. r Woven scaffolds in customizable geometries including letters, polygons, tubes, kidney, liver, lung shapes. s Customized fabrication of defect-matched YS: i) creation of irregular leg defect in rat; ii) MRI scanning; and iii) implantation of 3D weaving YS. Bar heights and error bars represent the mean ± s.d. Statistical analysis: two-tailed unpaired t-test (c, d, e, n), one-way ANOVA and Tukey’s multiple comparisons (l, m), ns, no significance, *P ≤ 0.05, ***P ≤ 0.001, ****P ≤ 0.0001.

The continuous dSIS ribbons obtained by rotary-cutting of 20-cm tubular dSIS mounted on an expanded polytetrafluoroethylene (ePTFE) rod (Fig. 1g, Supplementary Movie 1) at widths of 2 mm, 5 mm, and 10 mm yielded ribbons with lengths of 176.0 ± 16.4 cm, 354.6 ± 27.3 cm, and 877.1 ± 38.3 cm, respectively (Fig. 1j, l). The 2-mm-wide ribbons were too narrow, exhibiting frequent fracturing during cutting, and the 10-mm-wide ribbons constrained the achievable ribbon length. Consequently, 5-mm-wide ribbons were selected and twisted for yarn formation and to avoid technical limitations. Twisting of the ribbons produced yarns with helical surface grooves (Supplementary Movie 2; Fig. 1h). Scanning electron microscopy (SEM) imaging revealed an increase in groove counts within three kinds of yarns (twist rates: 2.5, 5.0, and 7.5 r/cm) (Fig. 1k, Supplementary Fig. 2a). The yarn diameters varied according to twist rates (2.5 r/cm = 384.3 ± 21.5 μm, 5.0 r/cm = 300.4 ± 32.7 μm, and 7.5 r/cm = 323.4 ± 11.8 μm) (Supplementary Fig. 2c). As the twisting rate increased, the failure stress increased from 7.8 ± 3.0 MPa to 14.8 ± 2.3 MPa, while the failure strain increased from 71.1 ± 12.5% to 98.9 ± 17.4%. (Fig. 1m, Supplementary Fig. 2d). Yarns displayed structural looseness at 2.5 r/cm and exhibited heterogeneous kink formation at 7.5 r/cm. Based on these parametric evaluations, 5 r/cm was selected for yarn production (Supplementary Fig. 2b).

Next, we fabricated structurally tailored YS through a 3D weaving approach, achieving precise architectural control (Fig. 1i, Supplementary Movie 3). The multilayered dSIS membrane scaffolds (MS) control group was prepared by flattening and folding dSIS sheets into a multi-layered 3D structure, which was subsequently trimmed to the desired dimensions. Porosity of YS was significantly higher than that of MS (88.5 ± 1.7% vs 52.7 ± 2.0%) (Fig. 1n). Stereomicroscopy revealed honeycomb-like lateral structures and highly aligned yarns in frontal views of cubic YS. Multilayered membranes exhibited parallel lamellar stacking in the lateral section and a dense structure on the top surface (Fig. 1o, Supplementary Movies 4, 5). Statistical analysis revealed that the lateral pore size of the YS was 293.5 ± 54.6 µm. In contrast, the lateral pore size of the MS was 20.4 ± 8.4 µm (Supplementary Fig. 3d). The YS exhibited 100% interconnectivity in all directions, whereas the MS displayed a lack of interconnectivity perpendicular to the membrane plane but partial interconnectivity parallel to the membrane plane. The toluidine blue permeation assay demonstrated rapid permeation in YS, whereas the liquid failed to permeate into MS. Computational fluid dynamics (CFD) simulations performed in SolidWorks demonstrated uniform liquid perfusion throughout the YS, in contrast to heterogeneous surface dispersion observed in MS (Fig. 1p). We conducted repeated compression experiments on YS and MS under wet conditions to observe their morphological changes and structural retention. Mechanical testing revealed shape recovery with stable compressive force retention in YS after cyclic loading, whereas MS exhibited interlayer slippage accompanied by diminished compressive force (Fig. 1q).

Beyond cubic constructs, the platform generated customized geometries including alphabetic characters, polyhedrons, tubes, and organ-mimetic shapes (kidney, liver, lung), as demonstrated by stereomicroscopy and micro-CT imaging (Fig. 1r, Supplementary Fig. 3c). To validate the customizability of the fabrication process, an irregular VML defect was created in rat hindlimbs, in which three-dimensional geometric data of the defect was acquired via magnetic resonance imaging (MRI), enabling the fabrication of shape-matched YS (Fig. 1s). These data demonstrate that our fabrication strategy enabled the engineering of highly porous, interconnected YS with precise architectural control at both macro- and micro-scales.

Cell behaviors regulated by YS in vitro

The guiding effect of YS on rat skeletal muscle cells (L6 cells) was investigated in vitro. DiD-labeled L6 cells were seeded on YS and exhibited progressive infiltration throughout the YS interior on day 1. Fluorescence intensity increased over time, indicating active cellular migration into the scaffold. In contrast, the majority of cells accumulated on the MS scaffold periphery, and no cellular distribution could be observed within the scaffold interior (Fig. 2a). Fluorescence intensities in the YS were higher than in the MS on days 1, 3, and 5 (Fig. 2b). DNA content quantification confirmed proliferative activity in both groups. However, the DNA content of YS was significantly higher than that of MS on days 3 and 5, suggesting the YS supported cell growth (Fig. 2c).

Fig. 2. Guiding effects of YS and MS scaffolds on L6 cells.

a Distribution of 1,1’-Dioctadecyl-3,3,3’,3’-Tetramethylindodicarbocyanine,4-Chlorobenzenesulfonate Salt (DiD)-labeled L6 cells (red) on scaffolds (white). White arrows indicate MS boundaries. b Quantification of fluorescence intensity of (a) (n = 3 biologically independent samples). c DNA content quantification for cell proliferation (n = 6 biologically independent samples). d Live/dead staining: green (live), red (dead). e Macroscopic cell arrangement at 5 days. f Cell viability quantification (n = 9 biologically independent samples). g F-actin (phalloidin-FITC, green) and nuclei (DAPI, blue) staining showing cell arrangement. h Alignment angle quantification between cell and yarn (n = 3 biologically independent samples, 11 cells per sample). Bar heights and error bars represent the mean ± s.d. Statistical analysis: two-way ANOVA and Tukey’s multiple comparisons (b, c, f), ns no significance, ****P ≤ 0.0001.

Live/dead staining revealed that L6 cells were predominantly viable (green) on both YS and MS scaffolds on all days, with few dead cells (red) observable in each condition and on each day (Fig. 2d), demonstrating that the scaffolds were cytocompatible. Low-magnification images demonstrated scaffold yarn-guided cell orientation in the YS group, which contrasted with the random cell distribution in the MS group (Fig. 2e). The quantification confirmed the proportion of viable cells was greater than 90% in both groups across all timepoints, with no significant intergroup differences (Fig. 2f).

Phalloidin and DAPI staining revealed progressive increases in cell alignment along the yarn grooves from day 1 to 5, demonstrating scaffold-guided orientation. Conversely, cells exhibited random organization in the MS group, and this remained unchanged across the time course. Quantitative analysis revealed cell-yarn alignment angles on the YS were mostly between 30° and 60°, contrasting with the wide angular distribution observed on the MS (Fig. 2g, h). RT-qPCR analysis confirmed elevated expression of myogenic differentiation genes (Myod1, Myf5, Myh9, Myo6, Pax7, Myog) in YS compared with MS (Supplementary Fig. 4a–f).

Collectively, these data demonstrate that the YS enhanced cell migration, viability, proliferation, functional gene expression; suggested that YS could serve as a conducive microenvironment for orchestrated muscle tissue regeneration.

Performance of YS in vivo

YS and MS scaffolds (1.0 × 1.0 × 0.8 cm) were subcutaneously implanted in rat models to evaluate their performance in vivo. The scaffolds were harvested at 2- and 4-weeks post-implantation to assess cell infiltration, tissue ingrowth, vascularization, and macrophage polarization (Fig. 3a). At the 2 weeks timepoint, both YS and MS were surrounded by vascularized neo tissue. By 4 weeks, the YS exhibited volume reduction while the MS showed no significant changes (Fig. 3b). Quantitative results confirmed that the weight of YS harvested at the 4 weeks timepoint was significantly lower than that of MS (Fig. 3c). DAPI staining revealed that extensive cell infiltration occurred throughout the entire YS at 2 weeks, with a further increase in cell number observed at 4 weeks. However, within the MS scaffolds, only a small number of cells were distributed among multilayer membranes, although the cell number at 4 weeks was higher than that at 2 weeks (Fig. 3d). Statistical analysis indicated that the number of YS cells was significantly higher than that of MS cells at all timepoints (Fig. 3e). H&E staining revealed that at 2 weeks, cells and newly synthesized ECM had nearly filled the entire YS, while substantial areas devoid of cells and tissue distribution remained in the MS (Fig. 3f). Quantification data revealed that the unoccupied area within the YS group was significantly smaller than that in the MS group (Fig. 3g). At the 4 weeks timepoint, the area of the YS implants had decreased due to degradation and remodeling by the infiltrated host cells. In contrast, the area of the MS showed no significant change, but large unoccupied areas devoid of cells and tissue remained in central regions (Fig. 3h). Quantitative results demonstrated that the unoccupied area in the MS was significantly larger than that in the YS (Fig. 3i).

Fig. 3. Subcutaneous implantation of YS and MS to evaluate cellular characteristics and rate of tissue ingrowth.

a Macroscopic morphology of the scaffolds and a schematic of subcutaneous implantation and sample harvesting at different timepoints. b, c Explanted scaffold morphology and weight quantification (n = 3 biologically independent samples). Dashed circle indicates scaffold boundaries. d DAPI staining showing the cell infiltration. e Cell number inside the scaffolds (n = 3 biologically independent samples, 3 random fields per sample). f–i H&E staining showing cell and tissue ingrowth with corresponding quantification (n = 3 biologically independent samples, 2 random fields per sample). j H&E staining showing the formation of capillaries. k The number of capillaries per view in (j) (n = 3 biologically independent samples, 3 random fields per sample). l Immunofluorescence staining of CD31 (green) confirms capillaries. m Capillaries density quantification in (l) (n = 3 biologically independent samples, 3 random fields per sample). n, o Immunofluorescence staining of CD68 antibody(green) and iNOS antibody (red) showing pro-inflammatory macrophage distribution, and corresponding quantification (n = 3 biologically independent samples, 3 random fields per sample). p, q Immunofluorescence staining of CD68 antibody(green) and CD206 antibody (red) showing pro-regenerative macrophage distribution, and corresponding quantification (n = 3 biologically independent samples, 3 random fields per sample). Bar heights and error bars represent the mean ± s.d. Statistical analysis: two-tailed unpaired t-test (g, i), two-way ANOVA and Tukey’s multiple comparisons (c, e, k, m, o, q), ns, no significance, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

H&E staining revealed that numerous microvessels were distributed within the YS group at 2 weeks, and the number had decreased by 4 weeks. In contrast, the MS group exhibited almost no vascular ingrowth over 2 weeks with a small number of vessels appearing within the scaffolds at 4 weeks (Fig. 3j). Quantitative analysis revealed that the number of YS vascular ingrowths was significantly higher than that of the MS group at 2 weeks (Fig. 3k). Immunofluorescence staining for CD31 confirmed microvascular regeneration, and staining was consistent with microvessel numbers (Fig. 3l, m).

Quantitative analysis demonstrated that CD68⁺iNOS⁺ macrophages (M1-like) in the YS group were predominantly localized around the yarns, whereas in the MS group they were primarily distributed on the material periphery (Fig. 3n). At 2 weeks, the number of CD68⁺iNOS⁺ macrophages in YS was significantly higher than in MS (Fig. 3o). By 4 weeks, CD68⁺iNOS⁺ macrophages counts decreased in both groups, with no significant intergroup difference observed. CD68⁺CD206⁺ macrophages (M2-like) were extensively infiltrated within the YS at both 2 and 4 weeks, while only a sparse population of CD68⁺CD206⁺ macrophages could be detected around the MS scaffold (Fig. 3p). Statistical analysis confirmed that CD68⁺CD206⁺ macrophage counts in YS were substantially elevated compared to MS at both timepoints (Fig. 3q). These findings collectively indicated that the YS exhibited superior cellular infiltration, vascular ingrowth, and pro-reparative macrophage recruitment compared with MS scaffolds.

YS promotes regeneration in a rat model of VML

The rat tibialis anterior (TA) VML model (50% defect) was used to evaluate the effect of YS (1.0 × 0.5 × 0.5 cm) on promoting in situ muscle regeneration. At 2 weeks, both YS and MS were covered by tissue, yet the boundary between the scaffolds and the host tissue remained discernible. The TA site in the untreated group exhibited muscle atrophy, while the autograft group was filled with muscle-like tissue (Supplementary Fig. 5a). By 8 weeks, no macroscopically visible YS material remained, and the defect site was filled with newly regenerated muscle tissue, showing no distinct boundary with the surrounding native tissue. In the MS group, while most of the scaffold had degraded, a pronounced defect was still evident. The entire TA site in the untreated group was covered with connective tissue, whereas the regenerated muscle in the autograft group closely resembled native muscle (Fig. 4a). Gravimetric analysis of the harvested whole TA muscle revealed that the ratio of muscle mass (injured side to normal side) in the YS group was higher than in the MS and untreated groups. Although lower than the autograft group, this difference was not statistically significant (Fig. 4b).

Fig. 4. Muscle regeneration of the rat TA muscle defects treated with YS and MS.

a Macroscopic images of untreated defects, YS, MS, and autograft implantation at point of implant (0 day) and explant at 8 weeks. Dashed boxes indicate the boundary of host tissue and implants. b The muscle mass ratio of the injured side/normal side. (n = 3 biologically independent samples). c Masson’s trichrome staining showing collagen deposition and muscle regeneration. Dashed lines indicate the boundaries. d Quantification of collagen density (n = 3 biologically independent samples, 2 random fields per sample). e Immunofluorescence staining for Pax7-positive (green) muscle cells. f Statistical analysis of Pax7 fluorescence intensity in (e) (n = 3 biologically independent samples, 3 random fields per sample). g Immunofluorescence staining for desmin-positive (green) myofibers. h Statistical analysis of desmin fluorescence intensity in (g) (n = 3 biologically independent samples, 3 random fields per sample). i, Immunofluorescence staining for CD31(green) displaying microvasculature. j Statistical results of microvessels (n = 3 biologically independent samples, 3 random fields per sample). k Representative CMAP traces. l Quantification of CMAP at 8 weeks (n = 3 biologically independent samples). m Representative mechanical force curves. n Quantification of tetanic force (n = 3 biologically independent samples). Bar heights and error bars represent the mean ± s.d. Statistical analysis: one-way ANOVA and Tukey’s multiple comparisons (b, l, n), two-way ANOVA and Tukey’s multiple comparisons (d, f, h, j), ns, no significance, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

H&E and Masson’s trichrome staining revealed that at 2 weeks, the pores of the entire YS were filled with cells and their secreted collagenous ECM. In the MS group, cells and collagen-rich ECM were predominantly distributed in non-central scaffold regions, while being substantially reduced in the central region. The autograft group exhibited a scattered myofiber distribution with minimal collagen deposition. The untreated group showed only minor collagen deposition above the defect site, with no significant myofiber regeneration. By 8 weeks, the YS group closely resembled the autograft group, with the defect area filled by newly regenerated myofibers. In contrast, the MS group displayed substantial deposition of high-density collagenous ECM, preventing clear visualization of the scaffold architecture. The Untreated group exhibited sparse collagen deposition, and the defect site remained. The diameter and density of myofibers in the normal group showed no significant changes between 2 and 8 weeks (Fig. 4c, Supplementary Fig. 5b). Collagen quantification, performed via Masson’s trichrome and Picrosirius Red staining, demonstrated that collagen density in the YS group was significantly higher than in all other groups at 2 weeks. At the 8-week timepoint, YS collagen density was significantly lower than in the MS groups, but showed no significant difference compared to the autograft group (Fig. 4d, Supplementary Fig. 5c, d). Pax7 (myosatellite cells marker) staining indicated that at 2 weeks, YS and autograft groups contained an abundance of myosatellite cells distributed throughout the implant site, but few myosatellite cells were observed in the MS and untreated groups. Quantification revealed significantly higher Pax7 fluorescence intensity in the YS group compared to the MS and untreated groups, although the intensity in the YS group remained significantly lower than that of the autograft group. At 8 weeks, the numbers of myosatellite cells in both the YS and autograft groups had increased, whilst few and scattered myosatellite cells were observable in the MS and untreated groups (Fig. 4e). Quantification of fluorescent area confirmed that the YS group was higher than the MS and untreated groups and showed no significant difference from the autograft group (Fig. 4f). Desmin is an intermediate filament protein that serves as a specific marker for the myogenic lineage, and is used here to visualize the structural organization and cytoplasmic maturation of both nascent and mature myofibers. Desmin staining indicated that at the 2-week timepoint, only YS and autograft groups exhibited small myofibers, and no obvious myofiber distribution was observed in the MS or untreated groups. Quantification revealed significantly higher desmin fluorescence intensity in the YS group compared to the MS and untreated groups, although the YS group remained lower than in the autograft group. At 8 weeks, abundant desmin-positive, thick myofibers were present in both the YS and autograft groups. In contrast, only a few scattered, thin myofibers were observed at the defect margins in the MS and untreated groups (Fig. 4g). Quantification of fluorescent area confirmed that the YS group was significantly higher than the MS and untreated groups and showed no significant difference from the autograft group, although the YS group was lower than the normal tissue comparison group (Fig. 4h).

Immunofluorescent staining for CD31 revealed the extent of functional vascularization. At 2 weeks, capillary ingrowth was evident in the YS and autograft groups. Minimal vascularization was observed in the MS group, while the untreated group showed a limited amount of microvasculature, and only at the defect periphery. Quantification data indicated the number of microvessels in the YS group was significantly higher than in other groups, with no significant difference compared to the autograft group. By 8 weeks, microvessels were present in all groups (Fig. 4i). Quantification showed the number of vessels in the YS group was higher than in MS groups, but the differences were not statistically significant (Fig. 4j).

Compound muscle action potential (CMAP) recordings at 8 weeks confirmed that all groups responded to electrical stimulation (Fig. 4k). The CMAP amplitude in the YS group was significantly higher than in the MS and untreated groups, although lower than in the autograft group (Fig. 4l). NF was employed to specifically stain axonal processes, providing evidence of nerve ingrowth into the defect site. Immunofluorescence staining for NF9 showed almost no positive expression at 2 weeks. At 8 weeks, NF9-positive expression was observed only in the YS, autograft, and normal tissue groups (Supplementary Fig. 6a). Quantification confirmed that the number of NF9-positive cells in the YS group at 8 weeks was significantly greater than in the MS and untreated groups, but lower than in the autograft and normal tissue groups (Supplementary Fig. 6b). The newly formed muscle tissue in all groups exhibited varying degrees of tetanic force generation in response to electrical stimulation (Fig. 4m, Supplementary Fig. 6c). Quantitative analysis demonstrated that the tetanic force in the YS group was higher than in the MS and untreated groups yet remained lower than in the autograft and normal tissue groups (Fig. 4n). Taken together, these results demonstrated that compared to MS, YS significantly enhanced cellular infiltration, vascularization, neurotization, myofiber regeneration, and recovery of tetanic force in a rat TA VML model.

Single-nucleus RNA sequencing reveals muscle regeneration mechanisms

RNA sequencing and differential gene expression analysis were performed on samples collected at 2 weeks. Volcano plot analysis revealed 362 significantly upregulated genes and 358 significantly downregulated genes (Supplementary Fig. 7a). KEGG pathway enrichment analysis demonstrated that these differentially expressed genes were predominantly enriched in cytokine-cytokine receptor interactions and immune-related signaling pathways (Supplementary Fig. 7b). Heatmap analysis of muscle regeneration-related genes showed significantly upregulated expression of key genes (Dysf, Dmd, Cav3, Myod1, Myh7, and Myl2) in the YS group compared to the MS group. Conversely, heatmap analysis of immune function-related genes indicated significantly downregulated expression of pro-inflammatory genes (Tnf, Il1β, Il1r2, and Il6) in the YS group relative to the MS group (Supplementary Fig. 7c, d). This may be attributed to the relatively dense structure of MS, which leads to hypoxia or the accumulation of degradation products, thereby triggering a more severe inflammatory response. The intense inflammatory response of MS may result in slow regeneration.

To comprehensively elucidate the mechanism underlying YS-mediated muscle regeneration, we prepared single-nucleus suspensions from the explants of YS and MS at 2 weeks for single-nucleus RNA sequencing. Based on the known conserved marker gene expression profiles, 10 major cell types were annotated (Fig. 5a) and visualized using UMAP (Fig. 5b). The YS group exhibited higher proportions of muscle cells, whereas the MS group showed increased proportions of fibroblasts and myeloid cells. Both groups also contained B cells, endothelial cells, neutrophils, pericytes, satellite cells, T cells, natural killer cells (Fig. 5c, Supplementary Fig. 7e). Further subclustering of myocytes identified Muscle_Pax7⁺ (myocytes expressing Pax7, Myod1, Myf5) and Muscle_Acta2⁺ (myocytes expressing Acta2) subpopulations (Fig. 5d), visualized by UMAP (Fig. 5e). Quantitative analysis revealed a higher proportion of Muscle_Pax7⁺ cells in the YS group compared to the MS group (Fig. 5f). Gene Ontology (GO) enrichment analysis indicated that Muscle_Pax7⁺ cells were predominantly enriched in pathways related to cell-substrate adhesion, myeloid leukocyte migration, and others (Fig. 5g). Myeloid cells were subclustered into Macro_Spp1⁺ Inos⁺ (Spp1, Inos) macrophages, Macro_Cd206⁺ Igf1⁺ macrophages (Cd206, Igf1), Macro_Itgav⁺ macrophages (macrophages that also express Itgav), cDCs (Xcr1), and Macro_Proliferating macrophages (Top2a, Mki67) subpopulations (Fig. 5h) and visualized via UMAP (Fig. 5i). Statistical analysis demonstrated that the YS group had a higher proportion of Macro_Cd206⁺ Igf1⁺ macrophages than the MS group (Fig. 5j). GO enrichment analysis revealed that Macro_Cd206⁺ Igf1⁺ macrophages were enriched in pathways related to the muscle system process and the others (Fig. 5k). Neutrophils were subclustered into two subpopulations: Neu_Il1β⁺ neutrophils (Il1b, Il6) and Neu_Spp1⁺ neutrophils (Spp1 positive, Inos negative) (Fig. 5l), visualized via UMAP (Fig. 5m). Statistical analysis revealed that the proportion of Neu_Spp1⁺ neutrophils was lower in the YS group compared to the MS group (Fig. 5n). GO enrichment analysis indicated that Neu_Spp1⁺ neutrophils were enriched in pathways related to muscle adaptation, muscle cell differentiation and others (Fig. 5o).

Fig. 5. Single-nucleus RNA sequencing reveals the mechanisms of YS directed muscle regeneration at 2 weeks.

a, d, h, l Dot plot of total, muscle, myeloid and neutrophil cells showing the expression of canonical markers of each cell type. b, e, i, m UMAP dimensional reduction representing cells categorized into several main clusters with each cell color-coded based on its cell type. c, f, j, n Bar plots showing cell proportions of each cluster presented by percentages of sequenced cells. g, k, o Bar plot of the enriched GO biological processes of highly expressed genes in each subpopulation. The data were analyzed using over-representation analysis, with the hypergeometric test employed as the statistical test. p Two-dimensional visualization representing the overall strength of incoming/outgoing communication in cell subpopulation based on receptor and ligand expressio7n. q Heatmap depicting the strength of interactions among cell subpopulations. Darker red hues represent an increase relative to the MS. r Chord diagram of cellular communication relationship among seven cell clusters. The line thickness is proportional to the number of ligand-receptor pairs expressed in the connected cell types. Darker blue hues denote a reduction in the number of ligand-receptor pairs in the YS relative to the MS. s Comparison of the overall information flow, highlighting significant signaling differences with dashed boxes. t Dot plot illustrating the upregulated signaling ligand-receptor pairs between cell clusters in YS versus MS. The abbreviations for cell subpopulations are consistent with those used in (r). The differentiation trajectory of macrophages, with each color coded for subpopulations and pseudotime, and density plots showing the number of pseudobulked cells in the relative pseudotime scale. v Percentage of neutrophils in the YS and MS (n = 3 biologically independent samples). w Dot plot depicting expression of chemotaxis-associated genes, glycolysis-associated genes, and anti-apoptosis-associated genes in neutrophils within the YS and MS. x Schematic diagram depicting the mechanism of YS induces muscle regeneration. Images and data represent n = 3 animals per group. Bar heights and error bars represent the mean ± s.d. Statistical analysis: two-tailed unpaired t-test (v), *P ≤ 0.05.

Analysis of intercellular communication and interactions was performed using CellChat. Based on receptor and ligand expression, the overall strength of incoming/outgoing communication in cell subpopulations showed that compared with the MS group, the YS group exhibited increased outgoing and incoming signals in Macro_Cd206⁺ Igf1⁺ cells and increased incoming signals in Muscle_Pax7⁺ cells. Conversely, decreased outgoing and incoming signals were observed in both Macro_Spp1⁺ Inos⁺ and Neu_Spp1⁺ cells (Fig. 5p). Heatmap analysis of cell-cell interactions revealed increased interactions between the Macro_Cd206⁺ Igf1⁺ subpopulation itself, and between Macro_Cd206⁺ Igf1⁺ and Muscle_Pax7⁺ cells in the YS group compared to the MS group. Conversely, interactions between Macro_Spp1⁺ Inos⁺ and Macro_Cd206⁺ Igf1⁺, as well as between Neu_Spp1⁺ and Macro_Cd206⁺ Igf1⁺, were decreased (Fig. 5q). The chord plot revealed that compared to the MS group, cell communication from both Macro_Spp1⁺ Inos⁺ and Neu_Spp1⁺ to Macro_Cd206⁺ Igf1⁺ was significantly weakened in the YS group, while cell communication from Macro_Cd206⁺ Igf1⁺ to both Macro_Cd206⁺ Igf1⁺ and Muscle_Pax7⁺ was significantly enhanced (Fig. 5r). Information flow analysis revealed that inflammation-related genes (Csf, Il1, Spp1) were upregulated in the MS group, while muscle regeneration-related genes (Igf, Sema3) were upregulated in the YS group (Fig. 5s). We compared receptor and ligand expression among several cell subpopulations. Relative to the MS group, expression of the Igf1-Igf1r ligand-receptor pair from Macro_Cd206⁺ Igf1⁺ cells to Muscle_Pax7⁺ cells was elevated in the YS group. Conversely, the expression of the Spp1-Cd44 ligand-receptor pair from Neu_Spp1⁺ cells to Macro_Cd206⁺ Igf1⁺ cells was reduced in the YS group (Fig. 5t).

To delineate the transition states associated with different macrophage types, pseudotemporal analysis was performed. Macro_Spp1⁺ Inos⁺ macrophages dominated early stages, while Macro_Cd206⁺ Igf1⁺ macrophages appeared at later stages in the transition states. Compared with the MS group, the Macro_Cd206⁺ Igf1⁺ subpopulation emerged earlier in the YS group (Fig. 5u). Quantitative data revealed that the proportion of neutrophils was significantly higher in the MS group than in the YS group (Fig. 5v). Compared with the YS group, the MS group exhibited higher expression of genes associated with chemotaxis, glycolysis, and anti-apoptosis in neutrophils (Fig. 5w).

We also performed an in-depth analysis of the mechanism by which YS promoted vascularization. The heatmap depicting intercellular interactions revealed enhanced interactions between Macro_Cd206⁺ Igf1⁺ macrophages and endothelial cells in the YS group compared to the MS group (Supplementary Fig. 8a). We compared receptor and ligand expression levels among Macro_Cd206⁺ Igf1⁺ macrophages and endothelial cells showed that the YS group exhibited enhanced expression of the Igf1-Igf1r ligand-receptor pair from Macro_Cd206⁺ Igf1⁺ macrophages to endothelial cells relative to the MS group (Supplementary Fig. 8b).

We validated the relevant pathways using immunofluorescence staining. At 2 weeks, the ratio of CD206⁺ macrophages was significantly higher in the YS group compared to the MS group, while the ratio of iNOS⁺ macrophages was lower in the YS group than in the MS group (Supplementary Fig. 9a, b). CXCR2 is a neutrophil marker, the number of CXCR2⁺ neutrophils was significantly lower in the YS group than in the MS group (Supplementary Fig. 9c, d). HIF1α is highly expressed under hypoxic conditions, its expression was significantly lower in the YS group than in the MS group (Supplementary Fig. 9e, f). IGF1⁺ expression was significantly higher in the YS group than in the MS group (Supplementary Fig. 10a, b). Injection of IGF1 into the YS implantation site resulted in an observed increase in the number of muscle satellite cells (Supplementary Fig. 10c, d). In contrast, SPP1⁺ expression was significantly lower in the YS group than in the MS group (Supplementary Fig. 10e, f). Injection of SPP1 into the YS implantation site resulted in an observed decrease in the number of CD206⁺ macrophages (Supplementary Fig. 10g, h).

Based on these analyses, we elucidated the mechanism whereby YS promotes muscle regeneration. Compared with MS, YS exhibited reduced Neu_Spp1⁺ neutrophils and upregulated Macro_Cd206⁺ Igf1⁺ macrophages, enhancing IGF1 secretion. This further promoted the proliferation and differentiation of Muscle_Pax7⁺ muscle cells via IGF1-IGF1R signaling (Fig. 5x).

Validating the clinical translational potential of YS implantation in a canine model of VML

Due to the vast difference in muscle volume and mechanical loading between rodents and humans, a larger animal model is essential. We established a VML model in the quadriceps femoris of beagles (4.0 × 2.0 × 2.0 cm) to validate the clinical translational potential of implanted YS in promoting muscle regeneration. Radiological assessments were performed at 4- and 16-weeks (Fig. 6a). The implanted scaffolds were manufactured and packaged in a GMP facility, followed by third-party gamma irradiation sterilization (Fig. 6b). At the 16 weeks timepoint, the harvested YS group exhibited significant muscle regeneration, whereas the MS group displayed more fibrous connective tissue (Fig. 6c). Computed tomography angiography (CTA) at 16 weeks revealed perfused arteriole ingrowth in the YS group compared to the MS group (Fig. 6d). Quantification confirmed higher arteriole density in the YS group versus the MS group (Fig. 6h). Immunofluorescence staining for CD31 supported the robust ingrowth of arterioles in the YS group relative to the MS group (Fig. 6k). Quantitative results indicated a significantly higher arteriole count in the YS group than in the MS group, with no significant difference observed when YS was compared to the normal tissue group (Fig. 6m). MRI revealed a density variation between the scaffold implantation sites and the surrounding tissue in both groups at 4 weeks. By 16 weeks, the YS group exhibited a signal intensity closer to that of the adjacent normal muscle, whereas the MS group still displayed obvious defects and the presence of non-muscle tissue components (Fig. 6e). Histological analysis with H&E and Masson’s trichrome staining showed that the YS group exhibited enhanced muscle fiber regeneration and reduced disorganized collagen deposition relative to the MS group (Fig. 6f, g). Collagen quantification confirmed that the collagen area ratio was significantly lower in the YS group than in the MS group (Fig. 6i). Quantitative analysis of Picrosirius Red staining also revealed that the collagen area was significantly larger in the MS group than in the YS group (Supplementary Fig. 11). Desmin staining indicated denser regeneration of thick muscle fibers in the YS group relative to the MS group (Fig. 6j). Quantitative analysis of the fluorescence density revealed significantly greater levels in the YS group than in the MS group, although the fluorescence densities remained lower than in the normal tissue group (Fig. 6l).

Fig. 6. Muscle regeneration in canine quadriceps femoris VML models treated with YS and MS.

a Schematic and timeline of scaffold implantation, monitoring, and explant times. b The weaving, packaging, and sterilization of scaffolds in GMP facility. c Macroscopic views of scaffold implantation and subsequent explant at 16 weeks. d, h CT angiography showing microvascular ingrowth at 16 weeks with corresponding quantitative analysis (n = 3 biologically independent samples). e MRI at the material implantation site. f H&E staining showing muscle regeneration at 16 weeks. g, i Masson’s trichrome staining showing the collagen deposition with corresponding quantification at 16 weeks (n = 3 biologically independent samples, 2 random fields per sample). j, l Immunofluorescence staining for Desmin (green) revealing myofiber regeneration, and corresponding quantification (n = 3 biologically independent samples, 2 random fields per sample). k, m Immunofluorescence staining for CD31 (green) confirming the presence of microvasculature, and corresponding quantification (n = 3 biologically independent samples, 2 random fields per sample). n, p Representative CMAP traces and peak amplitude quantification at week 16 (n = 3 biologically independent samples). o, q Immunofluorescence staining for β-tubulin (green) showing reinnervation, and corresponding quantification (n = 3 biologically independent samples). Bar heights and error bars represent the mean ± s.d. Statistical analysis: two-tailed unpaired t-test (h), one-way ANOVA and Tukey’s multiple comparisons (i, l, m, p, q), ns, no significance, *P ≤ 0.05, **P ≤ 0.01, ****P ≤ 0.0001.

Next, we performed electrophysiological testing to determine the neural response functions of the neo muscle at the 8-week timepoint. The maximum amplitude of the CMAP in the YS group was higher than that in the MS group, but less than that in the normal tissue group (Fig. 6n, p). Immunofluorescence staining for the neural marker β-tubulin demonstrated higher expression in both the YS group and the normal tissue group compared to the MS group (Fig. 6o). Quantitative results showed that the number of β-tubulin⁺ cells was greater in the YS group than in the MS group, although lower than the β-tubulin⁺ cell presence in the normal tissue group (Fig. 6q). In summary, compared to MS, YS exhibited significantly enhanced vascularization, innervation, and myofiber regeneration in the beagle VML defect repair model.

Discussion

Native tissue-derived dECM exhibits bioactivity and modulates the immune microenvironment, but the absence of an adaptive porous architecture fails to provide cells with directional migration cues or topographical guidance, thereby limiting morphological and functional restoration at implant sites. We addressed these issues by developing rotary-cutting technology to obtain long dECM yarns and then utilized 3D weaving technology to precisely regulate the macroscopic morphology and microstructure of YS. The resulting YS possessed unique anisotropic guiding structures, 100% pore interconnectivity, and high porosity, which are suitable for a pro-regenerative microenvironment for rapid and directed migration and arrangement of endogenous cells, as well as tissue regeneration and integration. In vitro, YS exhibited cytocompatibility for skeletal muscle cells, promoting cell migration, maintaining cell viability, enhancing proliferation, and upregulating the expression of functional genes. In rat subcutaneous and tibialis anterior VML models, YS significantly enhanced cell infiltration, pro-repair macrophage recruitment, vascularization, innervation, and muscle volume and strength recovery. We then used scRNA-Seq to assess mechanistic actions of implanted YS, identifying upregulated presence of Cd206⁺ Igf1⁺ macrophages as being a key feature. These macrophages enhanced IGF1 secretion and consequently promoted Pax7⁺ myosatellite cell proliferation and differentiation. To demonstrate the clinical translation capacity of the developed YS, we used a canine VML model, which showed that YS promoted functional muscle regeneration compared to the conventionally employed dSIS MS.

Rapid cellularization and vascularization are prerequisites for in situ tissue regeneration induced by biomaterial scaffold implantation³¹. Nutrient and oxygen supply to the implant’s core region relies on passive diffusion, and insufficient supply often results in ischemia and tissue necrosis³². Therefore, the scaffold’s porous structure requires high interconnectivity throughout the entire scaffold, pore sizes that match the needs of muscle cell infiltration and tissue regeneration (100–200 μm), and directional guidance architecture to promote inward migration⁹. In comparison to reported cutting strategies²¹, the rotary-cutting process enables the preparation of elongated dECM yarns that reach an approximate 40-fold length of tubular SIS, and are capable of fulfilling the manufacturing demands for large human defects. Unlike processing methods such as dECM reconstitution into hydrogels³³ or electrospun fibers¹¹, the rotary-cutting process preserves the intrinsic bioactivity and microstructure of dECM, whilst resisting rapid degradation post-implantation, and maintaining pore structure and mechanical stability. 3D weaving technology achieves customized fabrication of scaffold macro-shapes, offering greater flexibility in matching defect geometry than traditional decellularized matrix, tissue, and membrane materials. Crucially, this technology precisely controls yarn arrangement, and here we showed that the process endowed the fabricated YS with 100% interconnectivity, guidance structures, high porosity, and compressive resistance. This enabled larger-sized YS implants to facilitate rapid cell migration and organization, nutrient/metabolite transport, vascularization, and in situ muscle regeneration, as validated in rat and beagle dog VML models. Single-nucleus RNA-sequencing analysis revealed that the YS material upregulated the proportion of pro-reparative macrophages and enhanced their IGF-1 secretion, in turn acting on endothelial cells to promote vascularization.

According to the previous studies, dECM accelerated muscle repair through preservation of bioactive factors, which upon release promoted macrophage infiltration and polarization, as well as satellite cell recruitment, proliferation, and differentiation³⁴. In VML models, persistent neutrophil presence (>14 days) was shown to inhibit myosatellite cell differentiation⁴. Various physical and biochemical modulation strategies have been employed to improve muscle functional recovery by reducing neutrophil numbers and suppressing pro-inflammatory factor and chemokine secretion^35,36. The dense structure and low porosity of membrane dECM scaffolds (such as MS) impede oxygen and nutrient transport, thereby promoting the subsequent induction of hypoxia. Such a milieu forces neutrophils to rely primarily on glycolysis for energy, leading to increased expression of glycolytic pathway factors and the significant elevation in the production of survival and anti-apoptotic factors, enhanced SPP-1 (phosphoprotein osteopontin) secretion, the recruitment and activation of pro-inflammatory macrophages, and consequently, inhibition of myofiber regeneration³⁷. SPP-1, binding to cell surface receptors (e.g., CD44), mediates cell migration, adhesion, and proliferation, and is crucial for macrophage recruitment/migration³⁸. Huang et al. demonstrated that the post-injury early release of pro-inflammatory SPP-1 by neutrophils in mice activated pro-inflammatory responses in macrophages and T-cells, constituting one mechanism of exacerbated skin fibrosis³⁹. Conversely, the highly interconnected pore structure of our developed YS accelerated cell infiltration, vascularization, nutrient transport, and metabolite clearance, significantly upregulating the proportion of C206⁺ IGF-1⁺ macrophages and promoting their IGF-1 secretion. This, in turn relayed the enhanced PAX7⁺ myosatellite cell proliferation and differentiation effects observed in the YS treatment groups. Indeed, IGF-1 promotes satellite cell proliferation and differentiation by binding receptors to activate the PI3K/Akt/mTOR and MAPK signaling pathways⁴⁰. Additionally, the YS material’s 5 kPa mechanical properties match the requirements of the initial muscle regeneration phase, which has been suggested to contribute to pro-regenerative mechanisms⁴¹. The mechanistic insights gained in this study provide a new rationale for material design and construction, underscoring the importance of dECM scaffold processing to provide optimal cell niches for enhanced tissue regeneration.

In summary, this study validated the preclinical safety and efficacy of the novel YS, with these pre-clinical investigations laying the foundation for subsequent progression towards clinical trials. Synergistic therapeutic strategies incorporating physical rehabilitation (e.g., exercise⁴², electromagnetic stimulation⁴³) are expected to further enhance the function and integration of regenerated muscle. In clinical practice, muscle defects are often complicated by factors such as fractures, skin loss, aging, and diabetes, posing greater demands on YS design and fabrication⁴⁴. Therefore, in-depth elucidation of muscle regeneration mechanisms combined with spatiotemporally controlled delivery of bioactive substances in a variety of clinically relevant diseases models may offer new solutions for VML with various co-morbidities^45,46.

Methods

Ethical statement

Animal experiments were approved by the Animal Experimental Ethics Committee of Nankai University (Approval No. 2022-SYDWLL-000450) and conducted in compliance with guidelines for the care and use of laboratory animals. A total of 71 male Sprague Dawley rats (8-10 weeks old, weighing 280-320 g) were purchased from SPF Biotechnology (Beijing, China). A total of 12 rats were used for subcutaneous implantation to evaluate scaffolds' biocompatibility; 30 rats were used to construct VML models; 1 rat was used to construct an irregular VML model to validate the customization process; 10 rats were implanted with scaffolds and harvested at 2 weeks for RNA sequencing to evaluate pro-regenerative effects; and 18 rats were used for single-nucleus RNA-sequencing to decode the regeneration mechanism and for in vivo validation of key signaling factors. For in vivo experiments, each group included at least n = 3 rats to obtain numerical data. Rats were housed under specific pathogen-free (SPF) conditions at 23 ± 3 °C, 40-70% humidity, and a 12-h light/dark cycle, with free access to standard chow and tap water. Additionally, six beagles (weighing 10-15 kg), were purchased from Beijing Marshall Biotechnology (Beijing, China), were used for the VML defect model. Beagles received standard kibble and drinkable tap water ad libitum, and pens were cleaned regularly. Each beagle group included n = 3 animals to obtain numerical data. At the time of euthanasia, the entire wound site was harvested. All beagles were sacrificed sixteen weeks post-surgery using overdose anesthesia.

Materials

Porcine small intestines were obtained from Tianjin Ershang Yingbin Meat Food (Tianjin, China). Ribonuclease (RNase), deoxyribonuclease I (DNase I), α-galactosidase (α-GAL) assay kits, 4% paraformaldehyde (PFA) solution, and sodium dodecyl sulfate (SDS) were purchased from Solarbio (Beijing, China). Ethanol and xylene were procured from Tianjin Chemical Reagent Company (Tianjin, China). The 3-0 and 9-0 sutures were supplied by Ningbo Medical Needle (Zhejiang, China). Haematoxylin and eosin (H&E) and Masson’s trichrome staining solutions were sourced from Leagene (Beijing, China).

Preparation of YS

The muscular and serosal layers of the small intestine were scraped away, and the tissue was sterilized in 0.5% peracetic acid. It was then immersed in 1% SDS solution for 72 h to facilitate decellularization, followed by thorough rinsing in sterile phosphate-buffered saline (PBS) for 96 h. Subsequent treatment with DNase and RNase solution at 37 °C for 24 h, followed by another 96-h PBS wash, yielded dSIS. The dSIS was mounted onto polytetrafluoroethylene (PTFE) rods and then subjected to rotary-cutting to fabricate continuous ribbons. By controlling the rotational and translational speeds of the PTFE rods, ribbons of 2.0 mm, 5.0 mm, and 10.0 mm widths were generated. These ribbons were twisted and wound onto spools, and varying the twisting speed produced yarns with different mechanical strength. For YS fabrication, G-code corresponding to the designed architecture was first generated using CIMCO Edit2022 software (CIMCO, USA). Needles were then positioned on a fixed base according to the design. The dECM yarns were drawn from the spools and automatically wound onto the underlying needles following the programmed path. Upon completion of winding, the needles were carefully removed and replaced with the dECM yarns to secure the scaffold structure, resulting in the YS (Supplementary Fig. 2). The fabricated YS materials were packaged and sterilized by 12 kGy gamma irradiation for subsequent experiments in vitro and in vivo.

Preparation of MS

The multilayered dSIS membrane served as the control group. Specifically, the tubular dSIS was incised and flattened into sheets, which were then air-dried at room temperature in an ultra-clean workbench. The dSIS sheets were folded multiple times to the required dimensions and subsequently stacked to form a three-dimensional (3D) structure. This 3D block was then trimmed to the desired size to obtain the MS.

Biochemical characterization of dECM

Histological changes in the ECM and nuclei before and after decellularization were assessed using H&E staining and DAPI fluorescence staining. DNA content was quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Grand Island, NY) according to the manufacturer’s instructions. Residual α-Gal activity was measured using the α-Gal Activity Assay Kit (Solarbio). Proteomic analysis of SIS and dSIS was performed via liquid chromatography-tandem mass spectrometry (LC-MS/MS) (LC-Bio Technologies (Hangzhou) Co., Ltd., China). Proteomics data analysis using the LC-Bio platform (LC-Bio Technologies (Hangzhou) Co., Ltd., China).

Mechanical characterization

The breaking stress and breaking strain of dECM yarns were measured using a universal testing machine (HY0580, Shanghai Hengyi Testing Instruments, China) with a 100 N load cell. Yarn specimens were stretched at a crosshead speed of 10 mm/min until fracture. YS and MS were prepared into 1.0 cm × 1.0 cm × 1.0 cm cubes and subjected to 100 compression cycles at 100 mm/min with a 3 mm displacement; the mechanical response was recorded during testing. All materials were immersed in PBS at room temperature for 12 h prior to mechanical characterization.

Physical characterization of the YS

The structural morphology of the dECM yarns was observed using SEM (ThermoFisher, Eindhoven, Netherlands). The YS architecture was scanned using a Bruker SkyScan Micro-CT system (SkyScan 1276, Allentown, PA, USA), and the acquired data were reconstructed using NRecon software (version 1.7.1.6). Scaffold porosity was measured via the ethanol displacement method. Fluid flow analysis simulations were performed using SolidWorks 2017 (SolidWorks, USA). For permeability assessment, the YS and MS scaffolds were positioned between two glass bottles, and the permeation of a 0.05% toluidine blue solution was observed.

Customized YS shapes to match muscle defect sites

Rats were euthanized to establish a large-volume quadriceps muscle defect model. The defect was scanned using a magnetic resonance rat imaging system (HT/MRSI60-60KY, China) with the following parameters: magnetic field strength: 1.2 T ± 0.05 T; resonance frequency: 51 MHz ± 2 MHz; magnet bore diameter: 300 mm. Based on the macroscopic structural data obtained from the scan, the required G-code was programmed to arrange the needle array for rapid weaving, fabricating the YS. The customized YS was then implanted into the wound, and a follow-up MRI scan was performed to assess shape adaptation.

YS and MS regulation of cell behavior

Cell migration assay

L6 cells labeled with DiD fluorescent dye (D4019, UElandy, China) were seeded onto YS and MS (1.0 × 1.0 × 1.0 cm) at a density of 1.0 × 10⁶ cells/scaffold. Cells were cultured in medium consisting of 90% Dulbecco’s Modified Eagle Medium (DMEM) and 10% FBS, with the medium replaced daily. After 1, 3, and 5 days of culture, scaffolds were harvested, fixed in 4% PFA for 12 h, and sectioned using a cryostat (CM1950, LEICA, Germany). Samples were automatically scanned and stitched using a fluorescence microscope (Thunder, Leica, Germany). Fluorescence intensity was quantified using ImageJ software v1.5.

Cell proliferation assay

L6 cells were seeded onto the materials and cultured for 1, 3, and 5 days. DNA content was quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Grand Island, NY, USA).

Cell viability and adhesion assay

Rat L6 cells were seeded onto scaffolds at a density of 1.0 × 10⁴ cells/scaffold and cultured for 1, 3, and 5 days. Following PBS washes, samples were subjected to Live/Dead staining (Solarbio, China) and phalloidin/DAPI (Sigma-Aldrich, USA) fluorescent staining. Images were captured using a laser scanning confocal microscope (TSC SP8, Leica, Germany). Cell quantification and orientation analysis were performed using ImageJ software v1.5.

Cell differentiation assay

RT-qPCR was performed to assess the regulatory effects of YS on gene expression in L6 cells. Total mRNA was extracted from cells cultured for 5 days, and mRNA concentration was determined using a NanoDrop spectrophotometer (NanoDrop Technologies, USA). RT-qPCR was conducted using the SYBR Green detection method on a CFX96 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The relative expression levels of target mRNAs were calculated using the 2^−ΔΔCT method and normalized to the β-actin gene. Primer sequences are provided in Supplementary Table 1. Primers were ordered from Beijing Tsingke Biotech Co., Ltd (Beijing, China).

Subcutaneous implantation

YS and MS (1.0 × 1.0 × 0.8 cm) were implanted subcutaneously into the dorsal region of rats. Specimens were harvested at 2 and 4 weeks, fixed in 4% PFA, and cryosectioned. Cell and tissue infiltration were characterized using DAPI staining and H&E staining (Solarbio, China). Collagen deposition was assessed via Masson’s trichrome staining.

VML in rat

For the rat VML model, animals were fasted for 12 h preoperatively. Rats were anesthetized through intraperitoneal administration of sodium pentobarbital (60 mg/kg), followed by maintenance using isoflurane in oxygen delivered via a breathing apparatus. Body temperature was maintained using a heating pad throughout surgery. The operative site was shaved and disinfected with povidone-iodine and 75% ethanol. An incision was made to expose the tibialis anterior muscle. A muscle segment measuring 10.0 × 5.0 × 5.0 mm (approximately 50% of the total muscle volume) was excised. Sterile scaffolds of appropriate dimensions were implanted into the defect, secured with sutures, and the skin was closed using absorbable sutures. The incision site was disinfected again with povidone-iodine to complete the procedure. Postoperatively, all animals were placed laterally on a heating blanket (37 °C) for 2 h to ensure smooth respiration and thermoregulation. A quiet environment was maintained until full recovery from anesthesia. Animals were fasted for 12 h post-surgery but can drink freely. During the wound healing period, animals were monitored daily for physical condition, including feeding, excretion, signs of surgical infection, and suture integrity. At 8 weeks, CMAP was assessed using an electrophysiological system (RM-6240, Chengdu Instrument Factory, China). Rats were anesthetized with isoflurane, and recording electrodes were inserted into both ends of the regenerative muscle. Following CMAP testing, muscles were harvested post-euthanasia. Tetanic force was recorded using a mechanical testing apparatus while supramaximal electrical stimulation (50 V) was applied via the electrophysiological system.

VML in Beagles

For the Beagle VML model, animals were fasted for 12 h preoperatively. Anesthesia was induced by intramuscular injection of xylazine (1.5 mg/kg) and maintained with isoflurane gas in oxygen. Core body temperature, electrocardiogram, and arterial oxygen saturation were continuously monitored throughout the procedure, with supplemental heating provided via a thermal pad. The surgical site was shaved and aseptically prepared using povidone-iodine solution and 75% ethanol. A skin incision exposed the quadriceps femoris muscle. Hemostasis was achieved using electrocautery, and a muscle segment measuring 4.0 × 2.0 × 2.0 cm (representing 30-40% of total muscle volume) was excised. Sterile scaffolds of appropriate dimensions were implanted, secured with sutures, and the skin was closed with absorbable sutures. The incision site was disinfected again with povidone-iodine.

Postoperatively, all animals were positioned laterally on a heating blanket (37 °C) for 2 h to maintain thermoregulation and ensure unobstructed respiration. Animals were maintained in a quiet environment until full anaesthetic recovery. Food was withheld for 12 h post-surgery, while water was provided ad libitum. During wound healing, animals underwent daily monitoring of physical condition, including feeding, excretion, signs of surgical infection, and suture integrity. Ceftiofur sodium (5 mg/kg) was administered subcutaneously as a single dose to provide sustained antibiotic coverage over 7 days. The skin wound received daily povidone-iodine disinfection for 7 days. MRI (Siemens, Germany) was performed on the injury site at 4 and 16 weeks postoperatively. Computed tomography (CT) angiography (GE, USA) was conducted at 16 weeks. At the 16-week endpoint, compound muscle action potentials (CMAP) were assessed under isoflurane anaesthesia using an electrophysiological system, with recording electrodes inserted into both ends of the regenerative muscle. Samples were harvested following euthanasia for histological analysis.

Transcriptome analysis

YS and MS were implanted into the rat TA with VML model. 2 weeks post-implantation, the implant sites were harvested for RNA isolation and subsequent RNA sequencing (Gene Denovo Biotechnology Co., Guangzhou, China). Gene abundance was compared between the YS and MS groups to identify differentially expressed genes (DEGs) with a fold change >2 or < -2 and a p-value < 0.05. DEGs were then subjected to enrichment analysis using the OmicShare platform (Genedenovo).

Single-nucleus sequencing analysis and validation

YS and MS were implanted in the rat TA with the VML model, and samples were harvested after 2 weeks. The tissue was disrupted, and nuclei were isolated for single-nucleus sequencing (SeekGene Biotechnology Co., Beijing, China). Sequencing data were analyzed using Seurat (v 5.5.0). Cell trajectory analysis was performed using Monocle (v 2.32.0). Cell-cell interaction analysis was conducted using CellChat (v 2.2.0.9001). For efficacy validation, IGF-1 and SPP1 were injected into the YS implantation site. Daily injections of 100 ng IGF-1 in 100 μL PBS (Novoprotein, P05019) and 20 ng SPP1 in 100 μL PBS (MedChemExpress, P08721) were administered, the control group received saline solution^47,48. Tissues were harvested at 2 weeks post-implantation for cryosectioning and fluorescent staining.

Histological staining

The harvested samples were fixed by immersion in 4% PFA and sectioned using a cryostat (CM1950, Leica, Germany). H&E staining, Picro-Sirius Red stain, and Masson’s trichrome staining were performed according to the respective kit manufacturer’s instructions. Images were captured using an upright microscope (Leica DM3000, Germany), and quantitative analysis of areas was conducted using ImageJ software v1.5j.

Immunofluorescence staining

Frozen sections were washed twice with PBS. Cell membranes were permeabilized using 1% Triton X-100 in PBS. Sections were incubated with goat serum/primary antibodies overnight at 4 °C. After washing with PBS, sections were incubated with PBS-diluted secondary antibodies at room temperature for 2 h. Following another PBS wash, sections were mounted with DAPI. Images were captured using a laser scanning confocal microscope (TSC SP8, Leica, Germany). Quantitative analysis of the sample area was performed using ImageJ software v1.5. Primary antibodies for fluorescence staining are listed below. For secondary antibodies, sections were stained with either goat anti-rabbit IgG Alexa 594 (1:500, a11037, Invitrogen, USA) or goat anti-mouse IgG1 Alexa 488 (1:500, a11029, Invitrogen, USA), as appropriate.

For rat immunofluorescence staining, sections were stained with rabbit polyclonal anti-CD31 (1:100, Cohesion, CQA6562), rabbit polyclonal anti-Desmin (1:50, Abcam, ab15200), rabbit polyclonal anti-CD206 (1:500, Abcam, ab64693), rabbit polyclonal anti-iNOS (1:200, Abcam, ab15323), mouse monoclonal anti-CD68 (1:250, Abcam, ab31630) and mouse monoclonal anti-NF9 (1:200, Abcam, ab7794). For key pathway protein validation, sections were stained with rabbit polyclonal anti-Pax7 (1:100, Cohesion, CQA3315), rabbit polyclonal anti-IGF-1 (1:100, Cohesion, CPA5435), rabbit polyclonal anti-SPP1 (1:100, Abways, CY5333), rabbit polyclonal anti-CXCR2 (1:100, Cohesion, CPA4620), and rabbit polyclonal anti-HIF1α (1:100, Cohesion, CPA3239).

For Beagle dog immunofluorescence staining, sections were stained with rabbit polyclonal anti-β-tubulin (1:100, Cohesion, CPA4302, China), rabbit polyclonal anti-CD31 (1:100, Cohesion, CQA6562), and rabbit polyclonal anti-Desmin (1:100, Cohesion, CPA1337).

Statistical analysis

All quantitative results were obtained from at least three samples and from three independent experiments. Data were presented as the mean ± s.d. All statistical analyses were performed in GraphPad Prism 7 (GraphPad Software, USA). A single comparison was made with an unpaired Student’s t-test. Multiple comparisons were carried out using one-way analysis of variance (ANOVA) and Tukey’s post-hoc analysis, and multiple comparisons across data with multiple variables were carried out using two-way ANOVA followed by Tukey’s post-hoc analysis. The statistical test used for each data set is described in each figure legend. Quantification of the samples was performed using ImageJ software v1.5. For all experiments, statistical significance is denoted as *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001, ns: non-significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.Z. conceived the research; M.Z., X.Z., K.W., D.K., and M.F. designed the experiments; G.S., W.C., H.L., Y.W., and Y.Z. designed, fabricated and characterised scaffolds; G.S., S.W., and S.Z. performed microsurgery of animal experiments. M.Z., G.S., X.Z., K.W., D.K., and M.F. interpreted the data, analysed the data and wrote the manuscript. All authors discussed the data and direction of the project at regular intervals throughout the study.

Peer review

Peer review information

Nature Communications thanks Jea-Hyun Baek, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was financially supported by the National Key Research and Development Program of China (2025YFF1505200), National Natural Science Foundation of China (82272156), Haihe Laboratory of Sustainable Chemical Transformations(25HHWCSS00020), and Tianjin Science and Technology Program (Grant No.23JCZDJC00810).

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information. The proteomics data generated in this study have been deposited in the PRoteomics IDEntifications (PRIDE) database under accession code PXD079406. The RNA sequencing data are available from the Sequence Read Archive (SRA) database, with accession number PRJNA1472589. The single-nucleus sequencing data are available from the Sequence Read Archive (SRA) database, with accession number PRJNA1478168. All raw data from this study are available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77032-y.