Bioinstructive scaffolds enhance stem cell engraftment for functional tissue regeneration

Abstract

Stem cell therapy is a promising approach for tissue regeneration after traumatic injury, yet current applications are limited by inadequate control over the fate of stem cells after transplantation. Here we introduce a bioconstruct engineered for the staged release of growth factors, tailored to direct different phases of muscle regeneration. The bioconstruct is composed of decellularized extracellular matrix containing polymeric nanocapsules sequentially releasing basic fibroblast growth factor (bFGF) and insulin-like growth factor 1 (IGF-1), which promote the proliferation and differentiation of muscle stem cells (MuSCs), respectively. When applied to a volumetric muscle loss (VML) defect in an animal model, the bioconstruct enhances myofiber formation, angiogenesis, innervation, and functional restoration. Further, it promotes functional muscle formation with human or aged murine MuSCs, highlighting the translational potential of this bioconstruct. Overall, these results highlight the potential of bioconstructs with orchestrated growth factor release for stem cell therapies in traumatic injury.

Stem cells play a pivotal role in tissue repair. Their behavior during the regeneration process is orchestrated by biochemical cues¹. Muscle stem cells (MuSCs) stand as the essential cellular mediators of muscle tissue repair^2,3. In response to injuries, MuSCs are directed by a sequence of niche factors to regenerate damaged muscle⁴. The regeneration capability of MuSCs, however, can be overwhelmed in traumatic injuries such as volumetric muscle loss (VML), which results in permanent tissue deformity and functional deficits⁵. Transplantation of MuSCs thus emerges as an appealing strategy to promote tissue regeneration in such settings. However, the potential of MuSC therapy is hindered by poor efficiency of engraftment and lack of control of the fate of cells after transplantation⁵.

Biomaterial approaches have garnered significant attention to bolster stem cell transplantation^6,7. Transplantation of MuSCs via biomaterials, such as decellularized extracellular matrix (dECM), promotes localization and retention of the donor cells by providing scaffolds for cell attachment⁸. Beyond cell delivery, instructive cues are also critical to direct the fate of transplanted cells. Upon injury, endogenous niches provide biochemical cues to MuSCs under a precise temporal control at different stages, from activation and proliferation to differentiation⁹. Delivery of growth factors (GFs) has been explored to provide instructive cues for transplanted cells⁵. Due to the poor stability of GFs, direct administration is often associated with a burst release of GFs at supraphysiological doses, potentially leading to overdose-induced pathology and transient responses¹⁰. Incorporation of GFs in hydrogels via physical absorption or covalent immobilization prolongs the retention time but may fail to provide appropriate kinetics to guide tissue repair¹¹. Synthesis of heterogenous hydrogels, encompassing multiple compartments, has been utilized to deliver GFs with distinct rates of release^11–13. Nevertheless, precisely manipulating the staged release kinetics of multiple GFs to recapitulate their native dynamics following injury remains challenging¹⁴.

In this study, we report a bioconstruct that enables the recapitulation of the endogenous temporal dynamics of GF release, providing an appropriate regenerative niche for transplanted MuSCs. Basic fibroblast growth factor (bFGF) and insulin-like growth factor 1 (IGF-1), which stimulate the proliferation and differentiation stages, respectively, of MuSCs during muscle regeneration^15,16, were selected as model therapeutics. We encapsulate bFGF and IGF-1 within thin layers of polymers, forming “nanocapsules” (NCs) of GFs (NC^GF), which enhance the stability of the growth factors and enable programmable release kinetics. By conjugating these NC^GF constructs to dECM scaffolds, MuSCs transplanted with this bioconstruct are directed to promote regeneration in a VML model. Notably, this strategy restores the therapeutic efficacy of old MuSCs for VML therapy in aged mice. Furthermore, the bioconstruct demonstrates a promising potential to enhance therapeutic efficacy of human MuSCs (hMuSCs) in VML treatment. Overall, this strategy not only presents an effective approach for VML therapy with translational potential but also offers a versatile platform for directing stem cell behavior by providing orchestrated growth factor signals.

Design of bioconstruct for stage-specific release of GFs

Previous studies have transplanted MuSCs with 3D scaffolds into VML defects to promote cell engraftment¹⁷. However, limited muscle formation has been observed despite the production of GFs, such as bFGF and IGF-1, by the damaged muscle^9,17. To determine the importance of exogenous GF delivery, we first sought to assess the concentrations of bFGF and IGF-1 in transplanted scaffolds and in damaged muscles following VML injuries. Murine VML defects were induced by surgically ablating 30% of tibialis anterior (TA) muscles (Fig. 1a, Extended Data Fig. 1a, b)¹⁸, resulting in persistent loss of muscle weight and function (Extended Data Fig. 1c, d). We utilized dECM derived from muscle as the scaffold for transplantation (Extended Data Fig. 1e, f), to provide the bioactive composition and physical structure to promote MuSC activity⁵. We transplanted dECM scaffolds into TA muscles with VML lesions and harvested the scaffolds and injured muscles separately to determine the concentrations of bFGF and IGF-1 in each. The level of bFGF in the muscle increased immediately after injury and peaked at day 4 post-injury (Fig. 1b), whereas IGF-1 exhibited a delayed increase, starting at day 3 and peaking around day 7 (Fig. 1c). The results are consistent with previous studies showing that muscles produce bFGF and IGF-1 with distinct kinetics in response to VML injuries, corresponding to phases of proliferation and differentiation during the regenerative process^19–21. The concentrations of bFGF and IGF-1 in the dECM scaffolds, on the other hand, remained negligible up to 10 days post-injury (Fig. 1b, c). The limited diffusion of the endogenous GFs into transplanted scaffolds would result in a lack of GF guidance for stem cells within scaffolds. We thus hypothesized that bioconstructs that could deliver bFGF and IGF-1 with staged release kinetics to recapitulate the native regeneration kinetics could enhance the regenerative potential of transplanted MuSCs.

Figure 1|. Fabrication and characterization of the bioconstructs for staged release of GFs.

a, Schematic of the experimental setup for the surgical ablation to create a VML defect and the transplantation of the bioconstruct. b,c, Concentrations of bFGF (b) and IGF-1 (c) in TA muscles and transplanted dECM scaffolds following injury (n=4 biologically independent experiments). d, Schematic of the dual-layer protein NC synthesis via in situ polymerization: (I) enriching of neutral monomer PEGMEA (green molecules), anionic monomer MA (red molecules), and degradable PLA-based crosslinkers (yellow molecules); (II) in situ polymerization of the monomers and crosslinkers to form a negatively charged layer of polymer around the protein core; (III) in situ polymerization of PEGMEA and APM monomers (monomers for conjugation, purple molecules) to form an antifouling surface with reactive groups for further modification. e, Surface charge of native Lysozyme and Lysozyme NCs with different ratios of anionic monomers to total monomer (see text for descriptions of “Low”, “Medium”, and “High”) (n=7 and 8 independent synthesis experiments for native Lysozyme and NCs, respectively). f, A representative TEM image of NC^Ly(Low) constructs. Each Lysozyme protein was labeled by a 5-nm single gold nanoparticle. Scale bar, 25 nm. g, The cumulative release profiles of Lysozyme from NCs in mouse serum at 37°C (n=6 independent synthesis experiments). h, Fluorescence spectra of fluorescein-labelled scaffolds and TAMRA-labelled NC^Ly(Low) constructs before and after conjugation with different concentrations of NC^Ly(Low) constructs. i, Representative fluorescence images (left) and fluorescence intensity quantification (right) over time of muscles transplanted with dECM scaffolds which were conjugated with fluorophore-labelled Lysozyme that was unencapsulated or encapsulated with different NCs (n=6 biologically independent experiments). Data in b, c, e, g, i are presented as the mean ± SEM.

To recapitulate the native kinetics of bFGF and IGF-1, we designed a NC platform to encapsulate GFs within bilayer polymer networks (Fig. 1d). Both bFGF and IGF-1 have a positive charge under physiological conditions. For optimizing NC formulation, Lysozyme was adopted as a model protein due to its similarity of physiochemical characters with the GFs. Lysozyme proteins were first encapsulated within a thin layer of anionic polymer network by in situ polymerization using methacrylic acid (MA, anionic monomer), poly (ethylene glycol) methyl ether acrylate (PEGMEA, neutral monomer), and a degradable polylactic acid-based crosslinker. The anionic polymer creates a strong electrostatic interaction with the protein. To control the strength of the electrostatic force, we synthesized a series of Lysozyme NCs (NC^Ly), denoted as NC^Ly(Low), NC^Ly(Medium), and NC^Ly(High), with different ratios of MA to the total monomers (i.e. MA and PEGMEA). Compared with native Lysozyme (zeta potential, 12 mV), the NC^Ly(Low), NC^Ly(Medium), and NC^Ly(High) constructs demonstrated negatively charged surfaces with zeta potentials of −8 mV, −14 mV, and −18 mV, respectively (Fig. 1e). Subsequent polymerization of PEGMEA and N-(3-aminopropyl) methacrylamide (APM, monomer providing reactive amine groups for further modification) formed a secondary layer of polymer (Fig. 1d), which confers a neutral-charged surface (Fig. 1e). The synthesized bilayer NC^Ly constructs exhibited a uniform morphology and size, consisting of a single Lysozyme protein within each NC (Fig. 1f, Extended Data Fig. 1g, h). No cytotoxicity was found in a cell viability assay of myoblasts treated with different concentrations of the NC^Ly constructs (Extended Data Fig. 1i).

The encapsulated proteins can be released from the NCs via degradation of the crosslinkers and removal of the polymer shells. We postulated that the release kinetics of encapsulated proteins could be controlled by tuning the electrostatic interactions between the protein and polymer shell. To test this hypothesis, we monitored the release kinetics of NCs with different ratios of anionic monomers to total monomers. NC^Ly(Low) constructs exhibited a release without delay, slowing gradually over the first few days (Fig. 1g). By contrast, NC^Ly(Medium) and NC^Ly(High) constructs, each with stronger electrostatic interactions, exhibited delayed release after about 2 and 4 days, respectively (Fig. 1g).

We next conjugated NCs to dECM scaffolds to achieve localized delivery at the injury site. NCs were covalently conjugated to dECM scaffolds with precise dose control (Fig. 1h, Extended Data Fig. 1j, and Supplementary Note 1). To assess the ability of NC-dECM composites to achieve staged release of protein cargos in vivo, fluorophore-labeled NC^Ly(Low), NC^Ly(Medium), and NC^Ly(High) constructs were conjugated to dECM scaffolds and transplanted into VML defects. Direct administration of native Lysozyme via physical adsorption to scaffolds exhibited an initial burst release (Fig 1i). All NCs, on the other hand, exhibited a delayed, and also sustained, release post-transplantation. NCs with the higher MA ratios (NC^Ly(Medium) and NC^Ly(High)) exhibited a greater delay of release over the first two weeks post-transplantation (Fig. 1i). These results illustrate that the bioconstructs formed by NC-dECM composites enable a staged release of protein cargos within VML defects.

Sustained and staged release of bFGF and IGF-1

As the release profiles of NC^Ly(Low) and NC^Ly(Medium) constructs recapitulate the early release of bFGF and the delayed release of IGF-1 in injured muscles following VML injuries, we adopted the protocol to synthesize the NCs of bFGF (NC^FGF) and IGF-1 (NC^IGF). NC^FGF and NC^IGF appeared similar in size and morphology to NC^Ly constructs (Fig. 2a, b) and revealed immediate and 3-day delayed release of GFs, respectively, with both showing sustained release compared to unencapsulated GFs (Fig. 2c, d). The release profiles aligned with the temporal dynamics of bFGF and IGF-1 production during native muscle regeneration^19,20,22. We hereinafter denoted the NC^GF constructs that release GFs in the early or late stages as NC^GF(E) or NC^GF(L), respectively. Encapsulation significantly increased the stability of GFs by reducing serum absorption (Extended Data Fig. 2a, Supplementary Note 2), minimizing macrophage uptake (Extended Data Fig. 2b–e, Supplementary Note 3), and resisting enzyme degradation (Extended Data Fig. 2f, Supplementary Note 4).

Figure 2|. Staged release of GFs promotes MuSC proliferation and differentiation in vitro.

a,b, Representative TEM images of NC^FGF (a) and NC^IGF (b) constructs. Scale bar, 50 nm. c,d, Levels of unencapsulated and encapsulated bFGF (c) (570 ng/mL) and IGF-1 (d) (1683 ng/mL) over time during incubation with mouse serum at 37°C (n=3 independent synthesis experiments). e, Fluorescence images (left) and cell number quantification (right) of MuSCs incubated without bFGF (Ctrl) or with unencapsulated bFGF, bFGF encapsulated with NCs (NC^FGF), or bFGF encapsulated with non-degradable NCs (NC^FGF(ND)) (n=6 independent experiments with MuSCs pooled from four mice). Scale bar, 20 μm. f, Fluorescence images (left) and fusion index quantification (right) of MuSCs incubated without IGF-1 (Ctrl) or with unencapsulated IGF, IGF encapsulated with NCs (NC^IGF), or IGF encapsulated with non-degradable NCs (NC^IGF(ND)) (n=6 independent experiments with MuSCs pooled from four mice). Scale bar, 20 μm. g, Representative fluorescence images of MuSCs incubated without GFs or with GFs encapsulated for timed release, as indicated. Scale bar, 200 μm. h-j, Quantification of the cell counts (h), the percentage of MyHC+ cells (i), and the fusion index (j) of MuSCs treated as in panel (g) (n=5 independent experiments with MuSCs pooled from four mice). Data in c-f, h-j are presented as the mean ± SEM. P values in e, f, h-j were determined by one-way ANOVA. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

We next explored the benefits of sustained GF activity in promoting the proliferation and differentiation of MuSCs compared to the burst release of GFs (Extended Data Fig. 2g, h, Supplementary Note 5). We used the minimum dose of NC^FGF constructs that induced maximal MuSC proliferation (Extended Data Fig. 2i). Nondegradable NC^FGF constructs, denoted as NC^FGF(ND), were synthesized using a similar protocol as for NC^FGF constructs but using non-degradable crosslinkers to eliminate the contribution of NCs to cell proliferation independent of GF release. MuSCs treated with PBS and NC^FGF(ND) constructs demonstrated the same level of cell proliferation (Fig. 2e), suggesting that unreleased bFGF couldn’t stimulate the proliferation of MuSCs. While both bFGF and NC^FGF constructs improved the proliferation of MuSCs, NC^FGF-treated MuSCs exhibited a two-fold increase in cell counts compared to bFGF treatment with a burst release (Fig. 2e). These results demonstrate that NC^FGF constructs, which enable sustained release of bFGF, can significantly promote the proliferation of MuSCs.

Similar results were found in characterizing the NC^IGF potency of inducing myoblast differentiation. Myoblasts were treated with native IGF-1, NC^IGF constructs, and NC^IGF(ND) constructs at the minimum dose that induced maximum effects on differentiation (Extended Data Fig. 2j). The myoblasts treated with NC^IGF constructs exhibited a higher fusion index compared with PBS-treated samples, whereas IGF-1 and NC^IGF(ND) treatment demonstrated no increase (Fig. 2f). Collectively, these results suggest that sustained release of bFGF and IGF-1 via NCs effectively promote MuSC proliferation and differentiation.

To substantiate the advantages of combinational and sequential release of bFGF and IGF-1, we compared the efficacy of each single GF alone with either simultaneous or sequential exposure of both GFs. Compared with the PBS-treated cells, incubation with NC^FGF(E) constructs substantially increased MuSC proliferation (Fig. 2g, h), but inhibited the differentiation and cell fusion (Fig. 2i, j). Co-delivery of NC^FGF(E) and NC^IGF(E) constructs, which released IGF-1 simultaneously with bFGF, achieved the same increase of cell number as NC^FGF(E) treatment and restored the differentiation and fusion to the same level as the PBS-treated cells (Fig. 2g–j). These results demonstrated that simultaneous supplement of IGF-1 with bFGF limited the effectiveness of IGF-1 in directing myoblast differentiation and fusion. The sequential delivery of NC^FGF(E) constructs followed by NC^IGF(L) constructs allowed stimulation of MuSC proliferation in the early stage and induction of differentiation and cell fusion later (Fig. 2g–j). The results illustrate the advantages of stage-releasing bFGF and IGF-1 for muscle regeneration.

Muscle tissue regeneration in the murine VML model

We assessed the potential of staged release of GFs in directing the fate of donor MuSCs in vivo. The NC^FGF(E)-modified dECM scaffolds were employed to elucidate the impact of enhanced proliferation stage on the therapeutic efficacy of MuSC transplantation. Additionally, we tested the effect of enhancing both proliferation and differentiation stages on MuSC therapy by utilizing NC^FGF(E)/NC^IGF(L)-modified dECM scaffolds (Fig. 3a). The dECM scaffolds without NCs were used as the control. We seeded luciferase-expressing MuSCs into these scaffolds (Extended Data Fig. 3a) and transplanted them into acute murine VML defects. The muscles transplanted with NC^FGF(E)/NC^IGF(L) scaffolds exhibited a remarkable increase in bioluminescence intensity (Fig. 3b), demonstrating that staged release of bFGF and IGF-1 promotes the engraftment and expansion of transplanted MuSCs.

Figure 3|. Staged release of GFs improves muscle regeneration and reduces fibrosis.

a. Schematic of NC-dECM composite for staged release of bFGF and IGF-1 to direct the fate of MuSCs. NCs were conjugated to dECM scaffolds to form NC-dECM composites. MuSCs seeded in the bioconstructs were transplanted into VML lesions. b, Representative bioluminescent images (left) and quantification (right) following transplantation of luciferase-expressing MuSCs seeded onto different bioconstructs into TA muscles following VML injuries (n=5 biologically independent experiments). c, The weight of TA muscles after the transplantation of bioconstructs as in panel (b) and analyzed 6 weeks after transplantation (n=12, 11, and 13 biologically independent experiments for Ctrl, NC^FGF(E), and NC^FGF(E)/NC^IGF(L), respectively). d-f, Muscle formation after the transplantation of bioconstructs as in panel (b) and analyzed 6 weeks after transplantation. (d) Representative immunofluorescence images of cross and longitudinal sections. The unrepaired region is demarcated by the white line. Scale bar, 500 μm. (e) Measurements of areas of TA muscles (n=5 biologically independent experiments). (f) The unrepaired areas as percentages of the total muscle areas (n=5 biologically independent experiments). g, (left) Representative immunofluorescence images demonstrating the scar formation and (right) the quantification of fibrotic tissue as determined by Collagen I staining (n=5 biologically independent experiments). Scale bar, 100 μm. h,i, Representative images and quantification of M1(iNOS+CD80+) (h) and M2 (CD206+CD163+) (i) macrophages in injured areas of TA muscles treated as in panel (b) (n=6 biologically independent experiments). Scale bar, 50 μm. Data in b, c, e-i are presented as the mean ± SEM. P values were determined by two-way ANOVA (b, h, i) or one-way ANOVA (c, e-g). *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

The muscle tissue formation was confirmed by histological analysis. Two weeks after MuSC transplantation, the NC^FGF(E)/NC^IGF(L) group demonstrated improved muscle regeneration, diminished scar formation, and less inflammation compared with the NC^FGF(E) and control groups (Extended Data Fig. 3b–d). Six weeks after treatment, muscles transplanted with NC^FGF(E)/NC^IGF(L) scaffolds showed notable increases in weight and area compared to the NC^FGF(E) and control groups (Fig. 3c–e), along with a marked reduction in unrepaired muscle area (Fig. 3f). No change in apoptosis by the delivery of GFs was detected at the injury site (Extended Data Fig. 3e).

The ineffective regeneration after VML lesions is usually accompanied by pathological fibrosis due to prolonged inflammation^23,24. The transplantation of biomaterials, such as dECM scaffolds, can also result in excessive fibrous tissue²⁵. As such, we investigated the scar formation after the bioconstruct transplantation. Compared with uninjured muscle, extensive scar tissue was detected at the edges of the bioconstructs in the control group (Fig. 3g). The NC^FGF(E) treatment significantly reduced scar formation, while the NC^FGF(E)/NC^IGF(L) group exhibited minimal fibrosis (Fig. 3g). To examine a potential determinant of fibrosis, we assessed macrophage polarization. Within NC^FGF(E)/NC^IGF(L) scaffolds, the number of pro-inflammatory (M1) macrophages was reduced in both the early and late stages compared to the NC^FGF(E) and control groups (Fig. 3h). By contrast, a substantial number of pro-regenerative (M2) macrophages were found within NC^FGF(E)/NC^IGF(L) scaffolds in the early stage, decreasing in the late stage (Fig. 3i). Further testing showed that macrophages treated with NC^IGF(L) and NC^FGF(E)/NC^IGF(L) constructs demonstrated a significant increase in the ratio of M2 to M1 macrophages, suggesting the treatment of NC^IGF constructs promotes polarization of macrophages into M2 phenotype (Extended Data Fig. 3f). Taken together, the results suggest that NC^FGF(E)/NC^IGF(L) scaffolds increase polarization into M2 macrophages via NC^IGF treatment, thereby promoting a pro-regenerative microenvironment for the transplanted MuSCs and potentially limiting the inflammatory phase after VML injuries.

De novo myofiber formation with innervation and angiogenesis

Having established improved muscle formation with the NC^FGF(E)/NC^IGF(L) treatment, we aimed to investigate the myofiber formation from the donor MuSCs. MuSCs expressing RFP were transplanted with NC^FGF(E) or NC^FGF(E)/NC^IGF(L) scaffolds. The dECM scaffolds without GFs were the control. Compared with the control group, both NC^FGF(E) and NC^FGF(E)/NC^IGF(L) treatment resulted in an enlarged region of RFP+ fibers, an increased number of RFP+ fibers, and RFP+ fibers with larger cross-sectional areas (Fig. 4a–d). Notably, the NC^FGF(E)/NC^IGF(L) group consistently yielded better outcomes in all these parameters in comparison to the NC^FGF(E) group. Overall, the donor-derived myofiber formation was significantly improved by staged release of GFs. In addition, we investigated the effect of GF delivery on self-renewal of transplanted MuSCs. Both NC^FGF(E) and NC^FGF(E)/NC^IGF(L) treatment increased the number of self-renewed MuSCs (Extended Data Fig. 4a). These results suggest the GF delivery not only improves initial regeneration following VML injuries but also contributes to replenishment of the MuSC pool.

Figure 4|. Staged release of GFs improves the reconstitution of donor-derived muscle.

a, Representative cross-sectional images demonstrating donor-derived RFP+ fibers 6 weeks following VML injuries and transplantation of RFP+ MuSCs seeded onto bioconstructs, as indicated. Scale bar, 500 μm. Boxes: areas of RFP+ fibers shown at higher magnification. Scale bar, 100 μm. b-d, Quantification of muscle formation from replicate experiments as shown in panel (a). (b) Total RFP+ fiber area. (c) Number of RFP+ fibers. (d) Morphometric analysis of RFP+ myofiber cross-sectional area (n=6 biologically independent experiments). e, Representative images (left) and quantification (right) of capillaries (CD31+/Isolectin+) within the regions of donor-derived RFP+ fibers (n=6 biologically independent experiments). Scale bar, 10 μm. f, Representative images (left) and quantification (right) of motor nerves by NF staining (n=5 biologically independent experiments) and NMJs by α-BTX staining (n=6 biologically independent experiments) within the regions of donor-derived RFP+ fibers. For motor nerve staining, scale bar, 50 μm; for NMJ assessment, scale bar, 25 μm. The white arrows indicate the motor nerves (top) or NMJs (bottom). Data in b, c, e, f, are presented as the mean ± SEM. P values in b, c, e, f were determined by one-way ANOVA. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

Vascularization is necessary to supply regenerated myofibers with oxygen and nutrients. Therefore, the success of bioconstruct transplantation relies on the formation of functional vasculatures. Both bFGF and IGF-1 have been widely reported as regulators of angiogenesis^26–28. To assess the contribution of GF delivery to angiogenesis, endothelial cells were isolated from muscle (Extended Data Fig. 4b, c) and treated with PBS, NC^FGF(E) constructs, or NC^FGF(E)/NC^IGF(L) constructs. Compared with the PBS and NC^FGF(E) group, NC^FGF(E)/NC^IGF(L) treatment produced longer and lumen-containing tubes (Extended Data Fig. 4d–f). Transplantation of NC^FGF(E)/NC^IGF(L) scaffolds without MuSCs into VML defects significantly increased endothelial cell migration into the lesions compared with the dECM control or NC^FGF(E) group (Extended Data Fig. 4g, h). We therefore investigated whether the NC^FGF(E)/NC^IGF(L) treatment would promote re-vascularization of regenerated muscle mediated by transplanted MuSCs. Both NC^FGF(E) and NC^FGF(E)/NC^IGF(L) groups were effective in promoting the infiltration of endothelial cells (Extended Data Fig. 4i). The NC^FGF(E)/NC^IGF(L) group exhibited more functional capillaries (Fig 4e, Extended Data Fig. 4j, k, Supplementary Note 6) and neovessels covered by pericytes and vascular smooth muscle cells (Extended Data Fig. 4l), suggesting improved stability and maturation. These results suggest that NC^FGF(E)/NC^IGF(L) treatment improves the development of a functional vascular network within VML lesions.

Effective innervation by motor neurons is also critical to restore proper muscle function. Both bFGF and IGF-1 have been reported as effective therapeutics for peripheral nerve regeneration^29,30. To investigate the potential of the bFGF and IGF-1 NCs to promote innervation, we assessed the neurite outgrowth of neuroblasts after treating with PBS, NC^FGF(E) constructs, and NC^FGF(E)/NC^IGF(L) constructs. The NC^FGF(E)/NC^IGF(L)-treated group exhibited the longest neurite outgrowth (Extended Data Fig. 4m). We then examined the innervation of donor-derived myofibers in vivo. More axons were found within the regenerated area in the NC^FGF(E)/NC^IGF(L) group than the dECM control and the NC^FGF(E) group (Fig. 4f). Additionally, the NC^FGF(E)/NC^IGF(L) group exhibited the greatest number of neuromuscular junctions (NMJs) associated with donor-derived fibers (Fig. 4f). Therefore, NC^FGF(E)/NC^IGF(L) scaffolds enhanced the restoration muscle tissue structures, including vascularization and innervation, laying the foundation for the recovery of muscle function.

Recovery of muscle function

After confirming the enhancement of restoration of muscle structure via MuSC transplantation and NC^FGF(E)/NC^IGF(L) treatment, we explored whether this approach was able to restore muscle functionality following VML injuries. Untreated muscles retained 70% of normal tetanic force (Fig. 5a, b). Treatment with dECM scaffolds alone provided limited strength benefit, whereas the NC^FGF(E) and NC^FGF(E)/NC^IGF(L) treatment restored tetanic force to 81% and 97% of uninjured muscle, respectively (Fig. 5a, b), indicating of robust integration of donor myofibers with the surrounding muscle. Muscle function recovery was further investigated by assessing grip strength. While the NC^FGF(E) group showed improved grip strength, NC^FGF(E)/NC^IGF(L) treatment resulted in the greatest recovery, exhibiting grip strength of 92% of that of uninjured muscle (Fig 5c).

Figure 5|. Staged release of GFs restores muscle function after MuSC transplantation.

a,b, Representative force curves (a) and quantification of the maximum tetanic force (b) of uninjured muscles, VML but untreated muscles, and VML muscles treated with Ctrl (dECM only), NC^FGF(E) scaffolds, or NC^FGF(E)/NC^IGF(L) scaffolds 6 weeks following transplantation (n=10 biologically independent experiments). c, Grip strengths of muscles treated with Ctrl (dECM only), NC^FGF(E), or NC^FGF(E)/NC^IGF(L) scaffolds 6 weeks following transplantation (n=11 biologically independent experiments). d-f, Quantification of the gait indices, including the brake time (d), the duration of the stance phase (e), and the ataxia coefficient (f) of the muscles treated with Ctrl (dECM only), NC^FGF(E), or NC^FGF(E)/NC^IGF(L) scaffolds 6 weeks following transplantation (n=12 biologically independent experiments). Data in b-f are presented as the mean ± SEM. P values were determined by one-way ANOVA. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

We also investigated the recovery of motor functionality after GF treatment using a Digigait system (Extended Data Fig. 5a–c). The VML-induced dysfunction of TA muscles impairs dorsiflexion during walking, causing numerous gait abnormalities in the untreated control group (Extended Data Fig. 5d–n, Supplementary Note 7). Compared with the dECM and NC^FGF(E) groups, the NC^FGF(E)/NC^IGF(L) treatment significantly increased the brake time and the duration of the stance phase (Fig. 5d, e). In addition, the NC^FGF(E)/NC^IGF(L) group demonstrated a recovery in the duration and length of strides (Extended Data Fig. 5o–q). Furthermore, the gait variability, as indicated by the Ataxia Coefficient, and paw area in the NC^FGF(E) and NC^FGF(E)/NC^IGF(L) groups also decreased to levels comparable to those in the uninjured control group (Fig. 5f, Extended Data Fig. 5r).

Treatment of VML in aged mice

The activity and function of MuSCs decline with age, leading to impaired tissue repair in aged skeletal muscles^31,32. Given the demonstrated effectiveness of NC^FGF(E)/NC^IGF(L) scaffolds in promoting young MuSC therapy, we explored whether this approach was able to improve the therapeutic efficacy of aged MuSCs for the treatment of VML lesions in aged mice. Donor MuSCs expressing RFP from young and aged mice were transplanted with dECM and NC^FGF(E)/NC^IGF(L) scaffolds into VML defects in aged mice. Aged MuSCs transplanted with dECM scaffolds resulted in a smaller muscle area than the young MuSC group (Fig. 6a). In contrast, the NC^FGF(E)/NC^IGF(L) treatment significantly increased the muscle area of the aged MuSC group to a level comparable to that of the young MuSC group. Examination of RFP+ fibers revealed fewer and smaller myofibers in aged MuSC transplants without GF delivery, while NC^FGF(E)/NC^IGF(L) treatment restored the size and number of RFP+ fibers in the aged MuSC group to levels similar to those observed in the young MuSC group (Fig. 6b, Extended Data Fig. 6a–c).

Figure 6|. Staged release of GFs improves human or aged murine MuSC engraftment.

a,b, Representative cross-sectional images (a) and quantification (b) of donor-derived RFP+ fibers in aged muscles 6 weeks following the transplantation of RFP-expressing young or aged murine MuSCs seeded onto different bioconstructs into TA muscles following VML injuries (n=5 biologically independent experiments). Scale bar, 100 μm. c, Quantification of the recovery of maximum tetanic force of the aged muscles as in panel (a) (n=5 biologically independent experiments). d, Representative images demonstrating longitudinal ultrasound imaging of TA muscles 6 weeks following transplantation of hMuSCs seeded onto Ctrl (dECM only) or NC^FGF(E)/NC^IGF(L) scaffolds compared with uninjured muscle. The unrepaired region is demarcated by the yellow line. Scale bar, 2 mm. e,f Representative images (e) and quantification (f) of the area of hMuSC-derived myofibers, which is demarcated by the dotted yellow line in (e) (n=4 replicates across three donors). Scale bar, 500 μm. g, Quantification of the recovery percentage of maximum tetanic force following transplantation of hMuSCs seeded onto different bioconstructs 6 weeks after VML injuries (n=5 replicates across three donors). Data in b, c, f, g are presented as the mean ± SEM. P values were determined by two-sided t test. *P≤0.05, **P≤0.01; ns, not significant.

In addition to impaired MuSC activity, a decline in angiogenesis and innervation has also been observed in aged tissues^33,34. A deficit of bFGF and IGF-1 expression and function have been implicated in impaired angiogenesis and innervation with age^33,35,36, suggesting that NC^FGF(E)/NC^IGF(L) scaffolds hold promise as an effective therapeutic approach. The treatment of NC^FGF(E)/NC^IGF(L) scaffolds remarkably increased the number of capillaries and NMJs in both young and aged MuSC groups (Extended Data Fig. 6d–g). Additionally, the NC^FGF(E)/NC^IGF(L) treatment effectively restored the muscle function in the aged MuSC group, approaching the level comparable to the young MuSC group (Fig. 6c). These results demonstrate that NC^FGF(E)/NC^IGF(L) scaffolds restore the regeneration capability of aged MuSCs and enhance their therapeutic outcomes of VML injuries to similar levels of young MuSCs.

Transplantation of human MuSCs (hMuSCs)

To explore the translation potential of NC^FGF(E)/NC^IGF(L) scaffolds, we sought to investigate whether NC^FGF(E)/NC^IGF(L) scaffolds would enhance the efficacy of transplanted hMuSCs into mouse VML lesions. We isolated hMuSCs (Extended Data Fig. 6h,i) and transplanted them using either dECM or NC^FGF(E)/NC^IGF(L) scaffold into VML defects. Compared with the dECM group, TA muscles treated with NC^FGF(E)/NC^IGF(L) scaffolds demonstrated diminished unrepaired area (Fig. 6d), robust myofiber formation from hMuSCs (Fig. 6e, f, Extended Data Fig. 6j, k), and reduced fibrosis compared with the dECM group (Extended Data Fig. 6l). The transplanted hMuSCs supported not only myofiber formation but also self-renewal of quiescent MuCS (Extended Data Fig. 6m). Additionally, the NC^FGF(E)/NC^IGF(L) treatment augmented angiogenesis and innervation within the area of hMuSC-derived myofibers (Extended Data Fig. 6n, o). We then investigated whether NC^FGF(E)/NC^IGF(L) scaffolds would also restore muscle function. NC^FGF(E)/NC^IGF(L) scaffold-treated muscles exhibited remarkable improvements in force production compared to the dECM group (Fig. 6g). Taken together, these findings demonstrate NC^FGF(E)/NC^IGF(L) scaffolds provide effective support to enhance hMuSC-mediated muscle regeneration and muscle function restoration, highlighting the translational potential of this approach.

Outlook

Collectively, we demonstrate a bioconstruct capable of staged release of bFGF and IGF-1 to instruct MuSCs using NC technology. The bioconstruct offers multiple advantages for GF delivery to promote muscle regeneration. First, the NC-dECM structure supports distinct functions with time-dependent precision. NCs facilitate the short-term release of GFs essential for cell guidance, while dECM scaffolds serve as long-term templates for cell adhesion throughout the entire regeneration process. Second, the in situ polymerization technique allows flexible and delicate adjustments of the polymer charge to align the release profiles of GFs with their native expression kinetics through precise modulation of electrostatic interactions between polymer and GFs (Supplementary Discussion 1). In addition, given the physiochemical similarity among GFs, this technique is likely to be applicable to various GF therapeutics.

As critical regulators of muscle regeneration, bFGF and IGF-1 have been delivered either individually^37,38 or in combination^39,40 for VML therapy. These studies have employed biomaterials for localized and prolonged release of these GFs. However, the negligible evidence of myofiber formation and marginal recovery of muscle function in these studies underscored the need for refining the GF delivery approach^37–40. Emerging evidence illustrates the benefits of multiple factor delivery in sequential manner for tissue regeneration⁴¹. Our study emphasizes the significance of sequential and staged release of multiple GFs in VML therapy.

Beyond instructing the fate of MuSCs, bFGF and IGF-1 exert positive effects on diverse cell types^42,43. In our studies, GF delivery modulated macrophage polarization and promoted angiogenesis and innervation (Supplementary Discussion 2). Although the pleiotropic functions of the delivered GFs observed in this study support muscle regeneration, these GFs may also induce unexpected effects by activating other cell types, such as fibroblasts⁴⁴, adipocytes⁴⁵, and lymphocytes^46,47. Understanding the role of GFs on these cells will be important to optimize the therapeutic efficacy of NC^GF scaffolds.

To advance clinical translation, one limitation of this study is the scalability of the technology, given the critical size of VML defects. For treatment of VML defects in humans, many more cells will be needed, and this could potentially be achieved by ex vivo expansion of MuSCs⁴⁸ or the generation of myogenic progenitors from induced pluripotent stem cells⁴⁹. With regard to the scaffold itself, approaches to scale-up fabrication approaches are likely to be needed⁵⁰. Combining staged GF release with these kinds of technological advances holds promise for effective therapeutics for VML lesions in humans.

Methods

Animals

The procedures involving animals were in accordance with an animal protocol approved by the Institutional Animal Care and Use Committee (IACUC) at the Veterans Affairs Palo Alto Health Care System and the University of California, Los Angeles. Mice were housed in a 12-h dark-light cycle under 20 to 24°C with 40–60% humidity in cages. Young mice (male and female) used in this study were 4–6 months old and aged mice (male) were 24–26 months old. Health checks were performed for the aged cohort twice per week. Following VML surgery, the injured mice were monitored twice per day for the first three days, daily for the next two weeks, and twice per week thereafter.

Young NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ (NSG, strain no. 005557) and aged C57BL6/J (strain no. 000664) mice were used for VML surgery. The C57BL6/J mice and Pax7^CreER; Rosa26 ^RFP were used for MuSC isolation. The Pax7^CreER; Rosa26 ^RFP were obtained by crossbreeding Rosa26 ^RFP (strain no. 007914) with Pax7^CreER (strain no. 017763). The CreER system was activated by intraperitoneal injections of tamoxifen (20 mg/mL in corn oil) administered every other day for five instances. The sample size was determined to ensure adequate power based on preliminary experiments. Mice were randomly selected and allocated to groups for each experiment. Experiments were carried out blinded unless indicated otherwise.

Synthesis of NCs

To synthesize Lysozyme NCs (NC^Ly) within a layer of anionic polymers, Lysozyme (1 mg/mL in PBS) was mixed with PEGMEA, MA (20% m/v in PBS), degradable crosslinker AI104 (10% m/v in PBS), ammonium persulfate (APS, 10% m/v in PBS), and tetramethylethylenediamine (TEMED) with the ratio listed in Supplementary Table 1. The mixture for polymerization was incubated in an ice bath for 2 h and dialyzed against prechilled PBS at 4°C overnight to remove unreacted small molecules. The unencapsulated Lysozymes were removed using ion-exchange chromatography (CM Sepharose, Millipore Sigma). The purified anionic NCs were then reacted with PEGMEA, APM (10%, m/v in PBS), AI104 (10%, m/v in PBS), APS (10% m/v in PBS), and TEMED (Supplementary Table 1) in an ice bath for 1h. Free monomers and initiators were then removed by dialysis. The NCs of bFGF and IGF-1 were synthesized via the same procedures. The details of NC^FGF and NC^IGF synthesis are listed in Supplementary Table 1. To synthesize gold-labelled Lysozyme, Lysozyme proteins (0.1 mg/mL in PBS) were reacted with Dibenzocyclooctyne-PEG4-N-hydroxysuccinimidyl ester (DBCO-PEG4-NHS ester) at a 1:1.1 molar ratio at 4°C overnight. Unreacted DBCO molecules were removed by dialysis. The DBCO-labeled Lysozymes were then reacted with azido-gold nanoparticles at a 1:0.8 molar ratio. The gold-conjugated-Lysozymes were encapsulated according to the same Lysozyme encapsulation protocol. The concentrations of NCs were measured based on their protein content using bicinchoninic acid (BCA) colorimetric protein assay (ThermoFisher, USA) according to the manufacturer instructions.

Release kinetics of NCs in vitro

The dECM scaffolds conjugated with Alexa Fluor 647 (AF647)-labelled NC^Ly constructs were used to characterize the release kinetics. The scaffolds loaded with AF647-labelled Lysozyme were used as a negative control. The scaffolds were incubated in the releasing buffer (5% mouse serum and 1% Pen/Strep in PBS), and the concentrations of released proteins in the releasing buffer were determined by the fluorescence intensity.

Cell lines

The mouse myoblast cell line C2C12 (CRL-1772), mouse macrophage cell line J774A.1 (TIB-67) and RAW 264.7 (TIB-71), and mouse neuroblast cell line Neuro-2a (CCL-131) were purchased from the American Type Culture Collection (ATCC) and authenticated by ATCC prior to acquisition.

MuSC isolation

The isolation of MuSCs was performed via a surface-antigen-based or RFP-based approach^51,52. The surface-antigen-based isolation was used for MuSCs of C57BL/6 mice. Hindlimb muscles and triceps were dissected, minced, and digested with collagenase type II and dispase. After enzymatic digestion, mononucleated cells were further dissociated from myofibers via aspiration and ejection through a 20-G needle. The cell suspension was then filtered with 40 μm cell strainers and stained with anti-CD31 (BioLegend, 102506; 1:100), anti-CD45 (BioLegend, 103108; 1:100), anti-VCAM (BioLegend, 105720; 1:100), and anti-Sca1 (BioLegend, 108120; 1:100) antibodies. MuSCs (CD31⁻CD45⁻Sca1⁻VCAM⁺) were isolated using a BD-FACS Aria II or III. The data were collected using BD FACSDIVA software.

For Pax7^CreER;Rosa26^RFP mice, the CreER system was activated by intraperitoneal injections of tamoxifen (20 mg/mL in corn oil) administered every other day for ten days. The suspension of dissociated cells was prepared using the same procedures for C57BL/6 mice. The hindlimb muscles and triceps were dissected, minced, and digested with collagenase type II and dispase. After enzymatic digestion, mononucleated cells were further dissociated from myofibers via aspiration and ejection through a 20-G needle. The cell suspensions were then filtered with 40 μm cell strainers. MuSCs were directly sorted by endogenous RFP protein expression. The purity of MuSC isolation for WT and Pax7^CreER; Rosa26 ^RFP was confirmed by Pax7 immunofluorescence staining.

Efficacy evaluation of sustained release of GFs

To investigate the benefits of sustained bFGF release, freshly isolated MuSCs were cultured in wash medium overnight and treated with NC^FGF(E) constructs (20 ng/mL of bFGF in NCs) for 4 days. MuSCs treated with bFGF (20 ng/mL) for 5 h, followed by 4-day culture in wash medium, were used to determine the efficacy of burst release treatment. After 4-day incubations with sustained or burst release of bFGF, the total cell counts were quantified to determine the therapeutic efficacy.

The significance of sustained delivery of IGF-1 in MuSC differentiation was determined by the cell fusion index. Freshly isolated MuSCs were cultured in growth medium (Ham’s F-10 medium, 10% FBS, 1% Pen/Strep, 2.5 ng/mL bFGF) for 3 days before the IGF-1 treatment. The NC^IGF(L) constructs were incubated with wash medium at 37°C for 3 days. MuSCs treated with the resulting NC^IGF(L) constructs (500 ng/mL of IGF-1 in NCs) for 4 days were used to investigate the effect of sustained release of IGF-1. MuSCs treated with IGF-1 (500 ng/mL in wash medium) for 5 h followed by 4-day cell culture in wash medium were used for the burst release investigation. After the 4-day treatment, all cells were fixed and stained with an antibody to MyHC. The cell fusion index was calculated as the proportion of nuclei in MyHC+ cells containing three or more nuclei to the total number of nuclei.

Efficacy evaluation of sequential release of GFs

To investigate the necessity of dual and sequential GF release, MuSCs isolated from C57BL/6 mice were treated with different NCs, including NC^FGF(E), NC^FGF(E)/NC^IGF(E), and NC^FGF(E)/NC^IGF(L) constructs. The optimized doses characterized before as 20 ng/mL for NC^FGF constructs and 500 ng/mL for NC^IGF constructs were used. The treatment with an equal volume of PBS was adopted as the control. MuSC were cultured in wash medium through the process. The early released NCs, such as NC^FGF(E) and NC^IGF(E) constructs, were added to wash medium for the first 3 days. Meanwhile, the late release samples, NC^IGF(L) constructs were incubated in wash medium at 37°C. The resulting late released samples replaced the early released samples on day 4 and incubated with cells for 4 days. Cells were fixed on day 7. The total cell counts, the percentage of MyHC+ cells, and the cell fusion index were investigated to determine the therapeutic efficacy of different groups.

VML surgery

VML injuries were created via surgical ablation⁸. Mice were anesthetized and maintained with 2% isoflurane. The legs were shaved and sterilized with betadine and 70% ethanol 3 times. A longitudinal skin incision was made over the tibialis anterior (TA) muscle. The facia on the top of the TA muscle was removed to expose the muscle. A 7×2×2 mm (length × width × depth) VML defect was created, generating a 15–20 mg defect of the TA muscle. Scaffolds were implanted immediately into the surgery site, followed by closure of the TA muscle and skin with vicryl sutures.

Gait analysis

Gait capture and analysis were performed using a Digigait System (Mouse Specifics) ⁸. Mice were allowed to explore the treadmill compartment for 1 min with the motor speed set to 0 cm/s. Then the belt speed slowly increased from 10 cm/s to 20 cm/s. The gait was immediately recorded after steady walking. For each mouse, videos were recorded for at least five segments of 5-second of uninterrupted running and analyzed by the software provided by the Digigait imaging system.

Assessment of muscle function in vivo

For functional analysis, mice were anesthetized and maintained with 2% isoflurane. The legs were shaved and sterilized with 70% ethanol. The in vivo assessment of muscle function was performed using a 1310A 3-in-1 Whole Animal System (Aurora Scientific Incorporated). To stabilize the hindlimb during measurement, the patella was located in the U clamp and the knee was firmly immobilized. The foot was placed onto the footplate and attached with tape. After appropriate setup of the mouse, the electrodes were subcutaneously positioned directly above the TA muscle to electrically stimulate it. The isometric tetanic torque was then determined at 150 Hz. The data were collected by an Aurora Real-Time Muscle Data Acquisition System and an Aurora Dynamic Muscle Data Acquisition System.

Endothelial cell isolation and in vitro culture

Isolation of muscle endothelial cells was based on a surface-antigen-based approach⁵³. Hindlimb muscles and triceps were dissected, minced, and digested with collagenase type II and dispase. After enzymatic digestion, mononucleated cells were further dissociated from myofibers via aspiration and ejection through a 20-G needle. The cell suspension was then filtered with 40 μm cell strainers and stained with anti-CD31 (BioLegend, 102510; 1:100) and anti-CD45 (BioLegend, 103108; 1:100) antibodies. The endothelial cells (CD31⁺CD45⁻) were isolated using a BD-FACS Aria II or III and cultured on ECM coated culture flasks in endothelial cell growth medium (Endothelial Cell Growth medium 2 (PromoCell), 5% fetal calf serum, 5 ng/mL epidermal growth factor, 0.5 ng/mL vascular endothelial growth factor, 1 μg/mL ascorbic acid, 22.5 μg/mL heparin, 0.2 μg/mL hydrocortisone). The endothelial cell growth medium was changed every 4 days until cells reached about 90% cell confluency. Endothelial cells were passaged onto ECM gel in endothelial cell growth medium with and without NC^FGF(E) constructs (20 ng/mL) and NC^IGF(L) constructs (500 ng/mL) for 7 days.

Human MuSC isolation and transplantation

Human skeletal muscle biopsy specimens were obtained from de-identified clinical tissue discarded from patients who underwent clinical procedures. The research protocol was approved by the Stanford University Institutional Review Board. We used 3 independent freshly isolated populations of hMuSCs for transplantation in this study. The isolation of hMuSCs was performed via a surface-antigen-based approach⁵⁴. Freshly isolated specimens (about 2– 3 g) were minced and digested with 20 mL collagenase type II solution (0.2% in wash medium) at 37°C for 30 min. After digestion, the sample was diluted to 50 mL with cold wash medium, gently mixed, and then centrifuged at 1600g for 5 min. The supernatant was aspirated until a final volume of 7 mL remained in the tube. The pellet was then resuspended and digested by adding 1.5 mL collagenase type II (0.5% in PBS) and 1.5 mL dispase (1% in PBS). The mixture was incubated at 37°C with agitation for 20 min. After the digestion, mononucleated cells were dissociated from myofibers via aspiration and ejection through a 20-G needle. The cell suspension was then filtered with 100 μm and 40 μm cell strainers and stained with anti-CD31 (BioLegend, 303115; 1:100), anti-CD45 (Invitrogen, MHCD4501; 1:100), anti-CD90 (BioLegend, 328109; 1:100), and anti-CD82 (BioLegend, 342109; 1:100) antibodies. The human MuSCs (CD31⁻CD45⁻CD90⁻CD82⁺) were isolated using a BD-FACS Aria II or III. The freshly isolated hMuSCs were seeded in the bioconstructs and transplanted to young NSG mice right after hMuSC isolation.

Ultrasound imaging

For ultrasound imaging, mice were anesthetized and shaved to remove all remaining hair. Ultrasound images of VML-injured TA muscles were taken in longitudinal planes by a Vevo 3100 ultrasound imaging system (Fujifilmvisualsonics). With ultrasonic gel applied to the TA area, the mice were imaged by holding the transducer parallel to the identified region. The data were collected using Vevo Lab software.

Immunofluorescence staining

For post-fixation, tissue sections were rehydrated in PBS and immersed in 4% PFA for 10 min. Following fixation, the sections were rinsed with PBS 3 times and incubated with PBST (PBS containing 0.25% Triton X-100) for 20 min to permeabilize the tissues. The samples were incubated with 1% donkey serum for 30 min to block nonspecific binding of the antibodies. Then the samples were incubated with the primary antibodies diluted in PBST at 4°C overnight. The following primary antibodies were used: anti-MyHC (DSHB, A4.1025; 1:100), anti-laminin (Abcam, Ab11576; 1:250), anti-laminin (Sigma, L9393; 1:100), anti-Pax7 (DSHB, Pax7; 1:100), anti-COL1A1 (Thermo Fisher Scientific, PA5-29569; 1:250), anti-iNOS (Abcam, ab3523; 1:50), anti-CD80 (Biotechne, MAB740-SP; 1:50), anti-CD206 (Thermo Fisher Scientific, 18704-1-AP; 1:100), anti-CD163 (Santa Cruz Biotechnology, sc-58965; 1:100), anti-CD31 (BD, 550274; 1:100), anti-CD31 (Abcam, ab28364; 1:100), anti-CD45 (BioLegend, 103112, 1:50), anti-Neurofilament (Thermo Fisher Scientific, MA5-14981; 1:200), anti-human CD29 (BioLegend, 303018; 1:50), alpha-Bungarotoxin (Fisher scientific, B13422; 1:100), anti-PDGFR beta (Abcam, ab69506; 1:100), anti-α Smooth muscle actin (Cell Signaling Technology, 19245T; 1:100), Anti-Lamin A/C (Thermo Fisher Scientific, MA3-1000; 1:100). The tissue samples were rinsed with PBST 3 times and incubated with the diluted secondary antibody for 1 h at room temperature. The details of the antibodies can be found in Supplementary Table 2. The samples were washed with PBST 3 times and mounted with FluorSave^™ reagent. The mounted samples were stored at −20°C for further imaging via Keyence BZ-X800 and Zeiss ZEN microscopes.

Image analysis

All images were analyzed using Fiji software. To count the number of biological structures, including nuclei, cells, myofibers, blood vessels, and NMJs, the thresholds of images were determined by staining controls lacking fluorophores or primary antibodies. The objects under the selected threshold were then counted using the “Analyze Particles” command. To exclude unspecific background, we defined the ranges of object sizes: nuclei (1–100 μm²), cells (1–1000 μm²), myofibers (10–10000 μm²), blood vessels (1–5000 μm²), and NMJs (1–5000 μm²), based on manual measurements from preliminary samples.

To measure the area with positive signals such as Col I, RFP, NF, CD31, and isolectin, the threshold of images was determined by similar staining controls. The areas with positive signals were then measured by the “Measure” command. To count objects within ROIs, including MyHC+ cells, fusion index, and Pax7+ cells, ROIs were delineated by MyHC+ or Pax7+ areas. The number of nuclei within each ROI was then counted based on the objective counting methods described previously. Nuclei within MyHC+ ROIs were identified as MyHC+ cells and those in ROIs containing three or more nuclei were counted as fused cells. All nuclei within Pax7+ ROIs were counted as Pax7+ cells.

For the quantification of muscle regeneration, the midpoint section in the series of cross-sections throughout the entire muscle was selected to represent the center of VML defects. The muscle area was quantified by the total area of laminin+ myofibers. The unrepaired area was identified as the area without myofibers, which includes the area of undegraded scaffold and the Collagen Type I+ fibrosis area. To quantify the number of centrally nucleated fibers (CNFs), trichrome-stained images were converted to 8-bit greyscale. The ROIs of myofibers were delineated by adjusting the threshold range. Nuclei were then selected by tuning the threshold. The ROIs with positive nuclei signals (mean gray value 5–255) were classified as CNFs. To measure the porosity of dECM scaffolds, dECM sections were stained using H&E staining. The areas of dECM scaffolds (pink) and pores (white) were measured by Fjji software. The porosity was quantified by the percentage of pore area relative to the total area (combined scaffold and the pore areas).

The number of M1 and M2 macrophages, CD45+ cells, in vitro tube formation of endothelial cells, functional vasculature, blood vessels covered by pericytes or SMCs, and the length of neurites was measured manually by experienced researchers under blinded conditions.

Statistics

Data are presented as means ± SEM unless otherwise specified. Statistical analyses were performed with GraphPad Prism^® using the two-sided t test or one way-ANOVA for the comparison of two or multiple groups. No statistical methods were used to pre-determine sample sizes but our sample size is similar to those reported in previous publications⁸. No animals or data points were excluded from the analyses.

Extended Data

Extended Data Figure 1|. Fabrication of bioconstruct for VML therapy.

a, Surgical procedure of the VML model. (i) TA muscles were exposed, (ii) 7×2×2 mm (length × width × depth) defects were created, (iii) scaffolds were implanted, and (iv) muscles were closed. b, Quantification of the weight of TA muscles with and without VML surgery (n=6 biologically independent experiments). c,d, Quantification of TA muscle weight (c) and maximum tetanic force (d) 6 weeks following VML surgery (n=6 and 4 biologically independent experiments for c and d, respectively). e, H&E and immunofluorescence staining of muscles with and without decellularization. Scale bar, 50 μm. f, Porosity of dECM scaffolds. Representative images of H&E staining (i), quantification of porosity (ii), and size distribution of the pores (iii) of dECM scaffold before and after lyophilization. (iv) A representative SEM image of lyophilized dECM scaffold. Scale bar, 25 μm. g, Quantification the average size (i) and size distribution of NC^Ly(Low) (ii), NC^Ly(Medium) (iii), and NC^Ly(High) (iv) constructs with and without PEGylation compared with Lysozyme (n=6 independent synthesis experiments). h, A representative TEM image of Lysozyme, labeled with one 5-nm gold nanoparticle per protein. Scale bar, 25 nm. i, Relative cell viability of C2C12 cells treated with different NCs (n=5 independent synthesis experiments). j, Loading efficiency of NCs on dECM scaffold. (i) The concentration of DBCO moieties per scaffold with varying DBCO-PEG4-NHS ester amounts (left) and the conjugation efficiency (right) (n=3 independent synthesis experiments). (ii) The concentration of NC^Ly(low) constructs per scaffold with the maximum DBCO number (left) and the conjugation efficiency (right) (n=6 independent synthesis experiments). Data in b-d, f, g, i, j are presented as the mean ± SEM. P values in c, d, f, g were determined by two-sided t test. **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

Extended Data Figure 2|. NCs control the release kinetics and improve the stability of GFs.

a, The remaining amount of protein in the supernatant after incubation of NCs (NC^FGF (left); NC^IGF (right)) with mouse serum, compared with their native counterparts (n=6 independent synthesis experiments). b, Representative images (left) and quantification (right) of cell uptake of fluorophore-labeled bFGF, without and with NC encapsulation, by the J774A.1 macrophages (n=3 independent synthesis experiments). Scale bar, 50 μm. c, Representative FACS plots of MuSC isolation. d, Representative images of Pax7 staining of isolated MuSCs. Scale bar, 100 μm. e, Effect of macrophage incubation of bFGF activity with and without encapsulation (n=6 independent synthesis experiments). f, Effect of trypsin degradation on bFGF activity with and without encapsulation (n=6 independent synthesis experiments). g,h, Release kinetics of bFGF (g) and IGF-1 (h) that had been physically adsorbed to dECM scaffolds (n=6 technical replicates). i,j, Dose-response curves of bFGF or IGF with and without encapsulation. Cell proliferation (i) and differentiation (j) of MuSCs treated with a series of concentrations of bFGF and IGF-1 with and without encapsulation (n=3 independent experiments with MuSCs pooled from four mice). Data in a, b, e-j are presented as the mean ± SEM. P values in a, e, f were determined by two-sided t test. **P≤0.01, ****P≤0.0001; ns, not significant.

Extended Data Figure 3|. Characterization of bioconstructs in vitro and in vivo.

a, A representative image (left) and quantification (right) of the distribution of MuSCs within dECM scaffolds and cultured 24 h in vitro after cell seeding. Scale bar, 100 μm. b-e, Characterization of muscle formation, inflammation, and cytotoxicity of muscles subjected to VML lesions two weeks after treatment with Ctrl (dECM only), NC^FGF(E), or NC^{FGF(E)/IGF(L)} scaffolds. b, (left) Representative images of Gomori trichrome staining of muscles subjected to VML lesions. The dashed yellow line indicates the edge between the scaffold (above the line) and host tissue (below the line). Scale bar, 50 μm. (right) Quantification of centrally nucleated fibers (CNFs) based on trichrome stained images (n=4 biologically independent experiments). c, Representative H&E stained images of muscles subjected to VML lesions. The dashed yellow line indicates the edge between the scaffold (above the line) and host tissue (below the line). Blood vessels are highlighted by the arrows. Scale bar, 100 μm. d, Representative images (left) and quantifications (right) of CD45+ cells in muscles subjected to VML lesions (n=6 biologically independent experiments). Scale bar, 25 μm. e, Representative images (left) and quantifications (right) of TUNEL staining of muscles subjected to VML lesions (n=6 biologically independent experiments). Scale bar, 50 μm. f, Representative images (left) and quantifications (right) of Raw264.7 macrophage polarization into M1 (iNOS+) and M2 (CD206+) macrophages following treatment with PBS (Ctrl), NC^FGF(E), NC^IGF(L), or NC^FGF(E)/NC^IGF(L) constructs (n=5 independent synthesis experiments). Scale bar, 50 μm. Data in b, d-f are presented as the mean ± SEM. P values were determined by one-way ANOVA. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

Extended Data Figure 4|. Muscle regeneration with angiogenesis and innervation.

a, Representative images (left) and quantification (right) of self-renewed transplanted cells (RFP+Pax7+) within the region of regenerated myofibers (n=6 biologically independent experiments). Scale bar, 10 μm. b, Representative FACS plots of endothelial cell isolation. c, A representative image demonstrating the purity of isolated endothelial cells. Scale bar, 20 μm. d-f, Representative images (d) and quantification of the number (e) and length (f) of tube formation by endothelial cells treated with PBS, NC^FGF(E), or NC^FGF(E)/NC^IGF(L) constructs (n=4 independent synthesis experiments with endothelial cells pooled from three mice). Scale bar, 100 μm. g,h, Representative images (g) and quantification (h) of endothelial cell infiltration into different scaffolds transplanted into VML defects without MuSCs 2-week post-transplantation (n=4 biologically independent experiments). Dashed yellow line indicates the edge between host tissue (left of the line) and scaffold (right of the line). Scale bar, 100 μm. i-k, Quantification of the density of endothelial cells (i), functional capillaries (j), and the number of capillaries per myofiber within the region of donor-derived fibers 6-week post-transplantation of RFP+ MuSCs seeded onto different bioconstructs (n=6 biologically independent experiments). l, Representative images (left) and quantifications (right) of neovessels surrounded by pericytes (PDGFRβ+) and vascular smooth muscle cells (α-SMA+) within the region of RFP+ regenerated myofibers (n=3 biologically independent experiments). Scale bar, 10 μm. m, Representative images (left) and quantification (right) of Neuro2a neuroblast differentiation after treatment with PBS, NC^FGF(E), or NC^FGF(E)/NC^IGF(L) constructs. The arrows indicate neurites of differentiated Neuro2a cells (n=3 independent synthesis experiments). Scale bar, 25 μm. Data in a, e, f, h-m are presented as the mean ± SEM. P values in a, e, f, h-k, m were determined by one-way ANOVA. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

Extended Data Figure 5|. Gait analysis of mice treated with different bioconstructs.

a, Schematic of the experimental setup during Digigait data collection. The gait of mice was continuously imaged while they walked on a transparent treadmill. b, (top) Representative images of a walking mouse captured in (a), and (bottom) the digital paw prints generated by the Digigait analysis software. Each paw of the mice was recognized and labelled in a different color. c, (left) Schematic of digital paw print representing the indices measured in gait analysis. (right) A complete stride consisted of two phases: a stance phase and a swing phase. The stance phase could be further divided into a brake phase and a propel phase. d-n, Gait indices significantly changed 6 weeks following VML injury. Compared with uninjured muscles, muscles with VML defects demonstrated gait abnormalities in brake time (d), max dA/dt (e), propel time (f), min dA/dt (g), stance time (h), swing time (i), stride time (j), stride frequency (k), stride length (l), paw area (m), and ataxia coefficient (n), which describes the step-to-step gait variability (n=12 biologically independent experiments). o-r, Quantification of gait indices, including the stride time (o), the stride frequency (p), the stride length (q), and the paw area (r) after transplantation of MuSCs using Ctrl (dECM only), NC^FGF(E), or NC^FGF(E)/NC^IGF(L) scaffolds (n=12 biologically independent experiments). Data in d-r are presented as the mean ± SEM. P values were determined by two-sided t test (d-n) or one-way ANOVA (o-r). *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001; ns, not significant.

Extended Data Figure 6|. Engraftment of aged murine MuSCs or human MuSCs.

a-c, Quantification of muscle area (a), the average CSA (b), and the RFP+ fiber number (c) of aged TA muscles transplanted with Ctrl (dECM only) or NC^FGF(E)/NC^IGF(L) scaffolds seeded with young or aged RFP+ MuSCs 6 weeks following transplantation (n=5 biologically independent experiments). d,e, Representative images (d) and quantification (e) of the density of the endothelial cells within the region of donor-derived RFP+ fibers in (a) (n=5 biologically independent experiments). Scale bar, 50 μm. f,g, Representative images (f) and quantification (g) of NMJs within the regions of donor-derived RFP+ fibers in (a) (n=5 biologically independent experiments). Scale bar, 50 μm. h, Representative FACS plots of hMuSC isolation. i, (left) Representative images of Pax7 staining of isolated hMuSCs and (right) quantification of the purity of the sorted populations (n=9 technical replicates from single donor). Scale bar, 50 μm. j,k, Quantification of the CSA of hMuSC-derived myofibers (j) and the muscle area and unrepaired area (k) 6-week post-transplantation of Ctrl (dECM only) or NC^FGF(E)/NC^IGF(L) scaffolds (n=4 replicates across three donors). l, Representative images (left) and quantification (right) of the area of fibrotic tissue in muscles treated with Ctrl (dECM only) or NC^FGF(E)/NC^IGF(L) scaffolds (n=4 replicates across three donors). Scale bar, 50 μm. m, Representative images of self-renewed hMuSCs 6 weeks following the transplantation of dECM (Ctrl) or NC^FGF(E)/ NC^IGF(L) scaffolds. Scale bar, 25 μm. n,o, Representative images (left) and quantifications (right) of the density of the endothelial cells (n) and NMJs (o) within the region of hMuSC-derived fibers (n=4 replicates across three donors). Scale bar, 50 μm. Data in a-c, e, g, i-l, n, and o are presented as the mean ± SEM. P values were determined by multiple unpaired two-sided t test. **P≤0.01; ns, not significant.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. Source data are provided with this paper. Information of the materials and reagents are listed in the key resources table (Supplementary Table 2).