3D printing- and cyclic strain-driven engineering of skeletal muscle blocks using enhanced viscoelastic extracellular matrix bioink

Abstract

Development of physiologically functional skeletal muscle constructs is vital for regenerative therapies and drug-screening applications. Herein, we present an integrated strategy that combines viscoelastic muscle-derived extracellular matrix (MdECM) bioinks with a 3D printing-assisted cyclic strain bioreactor. To enhance mechanical performance, an improved decellularization protocol incorporating isopropanol was introduced to effectively remove residual lipids. The resulting bioinks exhibited superior viscoelasticity, structural stability, and thixotropic recovery compared with conventional formulations. The viscoelastic bioinks enabled stable myoblast encapsulation and alignment within the 3D-printed pillar frames. These counterforces guided robust myotube formation, while optimized ECM concentration and cell density further promoted alignment and myogenic differentiation. Spatial regulation of myotube orientation was achieved by tuning the pillar geometry and spacing. Muscle blocks were matured in a custom-designed multi-chamber bioreactor capable of applying cyclic uniaxial strain, yielding constructs with uniaxial cellular alignment, elevated myogenic marker expression, and contractile responses confirmed by calcium imaging and electrical stimulation. This platform supports scalable production of structurally and functionally mature muscle microtissues, offering a promising model for therapeutic implantation and high-throughput drug screening targeting sarcopenia and muscle-wasting diseases.

Graphical abstract

1. Introduction

Engineering functional skeletal muscle tissue in vitro holds significant promise for regenerative medicine and drug discovery [1,2]. From a therapeutic standpoint, tissue-engineered muscle constructs offer the potential for implantation in cases of volumetric muscle loss or degenerative diseases [3]. In the context of drug screening, the recent surge in demand for pharmacological agents targeting sarcopenia or obesity-induced muscle wasting highlights the necessity of physiologically relevant in vitro muscle models [[4], [5], [6]]. In particular, miniaturized and scalable muscle blocks that accurately recapitulate muscle architecture and function can facilitate the high-throughput screening of muscle-targeting therapeutics [7,8].

Recent advancements in 3D bioprinting and decellularized extracellular matrix (dECM) technologies have opened new avenues for fabricating functional skeletal muscle constructs that replicate both the anatomical architecture and biochemical microenvironment of native tissue [[9], [10], [11], [12]]. 3D bioprinting enables high spatial precision and reproducibility in constructing complex geometries, whereas dECM-derived bioinks provide a tissue-specific biochemical niche that closely mimics the native extracellular milieu, supporting cell adhesion, proliferation, differentiation, and long-term functional maturation [[13], [14], [15], [16]]. In particular, skeletal muscle-derived dECM retains key components, such as collagen, laminin, fibronectin, and myogenic growth factors, which are essential for guiding myoblast fusion and myotube formation [14,17]. Despite these advantages, most dECM-based bioinks have been optimized primarily for their biochemical composition, with limited efforts directed toward tailoring their biomechanical performance. The viscoelastic properties of bioinks, especially under dynamic conditions, are critical for maintaining construct shape and mechanical integrity and facilitating cellular mechanotransduction, which drives alignment, differentiation, and contractile function in engineered muscle tissues [[17], [18], [19]].

In muscle tissue engineering, aligned myotube formation is often achieved by seeding cell-laden hydrogels within mechanically constrained geometries, such as pillar frames, where cell traction forces are counterbalanced by external mechanical boundaries [[20], [21], [22]]. However, such alignment strategies rely critically on the ability of hydrogels to resist deformation while transmitting mechanical cues to the embedded cells. In practice, commonly used extracellular matrix (ECM)-derived bioinks often suffer from low mechanical integrity and poor viscoelastic regulation, resulting in inadequate cell anchorage, reduced alignment, and compromised tissue organization [[23], [24], [25]]. In particular, the viscoelastic properties of the ECM act as key regulators of biochemical cues, which are mainly transduced through the integrin–FAK signaling axis and ultimately influence cell proliferation and differentiation [[26], [27], [28]]. Therefore, bioinks for skeletal muscle tissue engineering should not only provide sufficient mechanical robustness but also present appropriately tuned viscoelastic cues capable of orchestrating these biochemical signals.

Given the limitations of current ECM-derived bioinks in supporting mechanically robust muscle tissue formation, we hypothesized that residual lipids in decellularized muscle tissue may impair the structural integrity and mechanical performance of the resulting hydrogels. These lipids, often retained during conventional decellularization, can hinder collagen fiber alignment and crosslinking, ultimately reducing the ability of the material to support cell-mediated force transmission and tissue organization. To address this issue, we propose an enhanced decellularization method that incorporates isopropanol (IPA) treatment to effectively remove lipid contaminants and reinforce the viscoelastic network of skeletal muscle-derived ECM bioinks. Rheological analysis confirmed improved mechanical properties, and cell culture studies demonstrated enhanced myotube formation and alignment. To facilitate functional maturation, we integrated these optimized bioinks into a 3D-printed platform coupled with a custom-designed cyclic-strain bioreactor. This modular system, fabricated using Ecoflex™ and 3D-printed components, enables synchronized uniaxial mechanical stimulation across multiple constructs with high reproducibility. Based on previously reported exercise-mimicking loading regimens, cyclic strain was applied to the muscle constructs to mimic either physiological or excessive exercise conditions [29,30]. Muscle constructs subjected to the physiological exercise–mimicking regimen exhibited improved morphological and functional outcomes, including enhanced alignment, upregulated myogenic markers, and contractile responsiveness, as validated by transient calcium and electrically evoked contractions.

2. Materials and methods

2.1. Decellularization of porcine muscle tissue

Porcine anterior tibial muscle was purchased from a butcher, thinly sliced, frozen, and repeatedly centrifuged at 300×g for 10 min until the fat was completely removed. The slices were immersed in 1% (w/v) sodium dodecyl sulfate at 22 °C for 24–48 h, with the solution replaced every 12 h. Subsequently, the samples were rinsed in phosphate-buffered saline for 12 h, followed by incubation in 500 mL of a solution containing DNase I (50 U/mL), Tris-HCl (10 mM), MgCl2 (5 mM), and CaCl2 (0.1 mM) at 22 °C for 12 h. The samples were then treated with IPA for 6 h, rinsed with PBS for 18 h, sterilized with 0.1% peracetic acid and 4% ethanol for 4 h, and rinsed again with PBS for 24 h. The final tissue was frozen and lyophilized for three days. Five milligrams of each sample were used for genomic DNA extraction using the AccuPrep Genomic DNA Kit (Bioneer Corp., Republic of Korea). The tissue was digested with lysis buffer (TL buffer, Bioneer Corp.), Protease K, and RNase A at 60 °C overnight. Genomic binding buffer (GB buffer, Bioneer Corp.) and ethanol were added, and DNA was purified according to the manufacturer's protocol. DNA was eluted in 200 μL elution buffer (EA buffer, Bioneer Corp.) and quantified using NanoDrop (Thermo Fisher) with EA buffer as the blank. The values were corrected for dilution factor and sample weight.

2.2. Bioink preparation and rheological measurements

Lyophilized muscle-derived extracellular matrix (MdECM) was chopped and dissolved in 0.5 M acetic acid containing pepsin (100 U/mL) at 25 °C to obtain a final concentration of 2% (w/v). The mixture was stirred for 2 days and then filtered through a 100 μm cell strainer before use. Rheological analyses were performed using 2% MdECM. A rheometer (TA Instruments) with a 25-mm parallel plate was used. The viscosity was measured at 20 °C over a shear rate of 100–10³ 1/s. Amplitude sweep tests and three-interval thixotropy tests (3ITTs) were conducted at 20 °C (0.1, 1–1000% strain). The gelation kinetics were assessed from 20 to 37 °C at 1 °C/min (0.1 rad/s, 1% strain). Frequency sweeps (0.1–100 rad/s, 1% strain), creep-recovery tests (at 0/100 Pa and 0/1 Pa), and stress-relaxation tests (displacement: 100 μm, speed: 50 μm/s) were performed at 37 °C. Stress–relaxation curves were fitted to a two-term exponential decay function (Prony-series form), and the slow relaxation time constant (τ₂) was reported.

2.3. Histological staining

Native and decellularized tissues were fixed in 10% neutral buffered formalin (Biosesang Inc., Republic of Korea), dehydrated, and embedded in paraffin. Paraffin blocks were sectioned at a thickness of 5 μm. Commercial hematoxylin and eosin (H&E; ab245880, Abcam, UK) and Masson's trichrome (ab150686, Abcam) staining kits were used to evaluate cellular removal and collagen preservation, respectively, following the manufacturers' protocols. For lipid visualization, Oil Red O staining (ab150678; Abcam) was performed on cryosections. Briefly, native muscle tissue and MdECM samples were embedded in OCT compound, cryosectioned at 10 μm thickness, and fixed in 10% neutral buffered formalin. Sections were stained with Oil Red O according to the manufacturer's instructions to assess residual lipid content. Immunohistochemistry was performed using primary antibodies against fibronectin (ab2413; Abcam) and laminin (ab11575; Abcam), followed by fluorescent secondary antibodies.

2.4. ECM quantification

The elastin content was quantified using a colorimetric Elastin Assay Kit (BioVision, USA). The tissues (5 mg) were treated with oxalic acid, heated, precipitated, dyed, and measured at 513 nm. Glycosaminoglycans (GAGs) content was measured using a Total Glycosaminoglycan Assay Kit (Abcam) according to the manufacturer's protocol, and absorbance was measured at 400 nm. For triglyceride quantification, a commercial triglyceride assay kit (K622; BioVision, USA) was used according to the manufacturer's instructions. Briefly, samples were homogenized in NP-40–containing buffer, subjected to repeated heating at 95 °C to solubilize lipids, and centrifuged to collect the supernatant. Triglycerides were enzymatically hydrolyzed to glycerol using the provided lipase, followed by oxidation with the reaction mix to generate a colorimetric signal. Absorbance was measured at 570 nm, and triglyceride content was calculated using a standard curve.

2.5. Cell culture and encapsulation

C2C12 murine myoblasts were cultured in low-glucose Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS and 1% P/S at 37 °C in 5% CO₂. The cells were passaged at 60–70% confluence using 0.25% trypsin/EDTA. To prepare the bioink, 10 × DMEM was added, and the pH was neutralized with NaOH. C2C12 cells were mixed with the bioink and cross-linked at 37 °C for 1 h. The constructs were cultured in DMEM High Glucose with 10% FBS and 1% P/S for 3 days and then differentiated in media containing 2% horse serum for 14 days. For qPCR analysis, C2C12 were cultured in DMEM high glucose with 10% FBS and 1% P/S for 5 days (Supplementary Fig. S1).

2.6. Cell viability and proliferation

A live/dead cell viability kit (Invitrogen, USA) was used. The samples were stained with calcein AM and EthD-1 and imaged using a fluorescence microscope. Cell viability was quantified using Fiji/ImageJ software. Cell metabolic activity was assessed using Cell Counting Kit-8 (Dojindo Laboratories Co., Republic of Korea). After incubation with the reagent for 2 h, the absorbance was measured at 405 nm.

2.7. Pillar frame fabrication via 3D printing

A commercial 3D printer (U-FAB Activo, Clecell, Republic of Korea) was used to fabricate the polyethylene vinyl acetate- and polycaprolactone (PCL)-based frames. The initial frames (30 mm × 10 mm × 5 mm) with 3-mm-diameter pillars were revised owing to detachment issues. The modified PCL frames (15 mm × 5 mm × 3 mm) were fabricated with pillar diameters of 1 – 3 mm and spacing of 10 – 20 mm.

2.8. F-actin imaging and directionality

The samples were fixed, permeabilized with Triton-X, blocked with 1% bovine serum albumin, and stained with phalloidin and DAPI. Confocal microscopy images were analyzed using Fiji/ImageJ software and visualized in MATLAB as polar histograms.

2.9. Immunostaining

The samples were fixed, permeabilized, and blocked as previously described. The primary antibody against myosin heavy chain (MHC) (20 μg/mL) was incubated overnight, followed by incubation with an Alexa Fluor 488-conjugated secondary antibody. DAPI was used for nuclear staining.

2.10. qRT-PCR

Total RNA was extracted using RNase Plus and chloroform, precipitated with IPA, washed with ethanol, and dissolved in diethyl pyrocarbonate-treated water. RNA was reverse-transcribed using the Maxima First Strand Kit (Sigma-Aldrich, USA) and analyzed by quantitative reverse transcription polymerase chain reaction (qPCR) using SYBR Green Master Mix. For RT–qPCR normalization, either GAPDH or ACTB (β-actin) was used as a reference gene depending on the experiment. Relative expression was calculated using the 2^−ΔΔCt method.

2.11. Finite element analysis

The 3D geometries were modeled using SolidWorks (Dassault Systems, France). Finite element analysis was performed using ANSYS (2024 R1, ANSYS Inc., USA) to evaluate platform deformation under displacement. The material behavior was defined using the Mooney-Rivlin 5-parameter hyper-elastic model. A frictionless support was applied to the zx-plane, and displacement was applied along the x-axis. The displacement ranged from 10% of the platform width to a maximum of 50 mm, which reflected the range of the linear actuator.

The Mooney-Rivlin equation (5 parameter) was used:

W = C₁₀(I₁−3) + C₀₁(I₂−3) + C₂₀(I₁−3)² + C₁₁(I₁−3)(I₂−3) + C₀₂(I₂−3)² + (1/D₁)(J−1)²,

where W is the strain energy density, C₁₀–C₀₂ are material constants, I₁ and I₂ are strain invariants, J is the Jacobian determinant (volume ratio), and D₁ is the dilatational (bulk modulus-related) parameter.

2.12. Fabrication of the bioreactor platform

The negative mold of the platform was fabricated using a PolyJet 3D printer (J35 PRO, Stratasys, USA). Ecoflex 00-31 (Smooth-On) was mixed in a 1:1 ratio (base:curing agent) and degassed by centrifugation at 3000×g for 1 min. The degassed mixture was poured into the mold and cured at 23 °C for 3 days. The cured platform was demolded and sterilized using a 75% ethanol spray and ultraviolet exposure for 1 h. The pillars, grippers, and supporters were fabricated using the same 3D printing method and sterilized in a similar manner. The pillars were inserted into an elastic platform to complete the assembly process. The bioreactor system was constructed using an Arduino UNO R3 and motor driver (L298N) to control a linear actuator (AMLA_50 mm, Foryou Tech, Japan). The actuator was mounted on a 3D-printed supporter and connected to an elastic platform using a 3D printed gripper. The platform was driven using a single motor driver to apply cyclic strains of either 10% or 20%. Briefly, cyclic strain was applied to the engineered muscle constructs under the following stimulation regimen: a strain amplitude of 10% or 20% was applied at an effective frequency of 0.09 Hz. Each stimulation cycle consisted of 10 min of active loading (phase T2) followed by a 50 min rest period (phase T3), resulting in a repeating 60 min stimulation pattern (Supplementary Fig. S2). This cycle was repeated continuously over a total stimulation period of 14 days. MdECM-based, cell-encapsulated bioink was injected into each chamber. Each chamber (20 mm × 5 mm) was filled with 150 μL of bioink to form a muscle block hydrogel with a height of 1.5 mm anchored between the two pillars (Supplementary Fig. S3).

2.13. Mechanical testing and electrical stimulation

The UTM (LLOYD-K) device was equipped with a 20 N load cell, and the hydrogel samples were secured in a Zigger fixture. Tensile tests were conducted using Nextgen software with a reducing factor of 50 and a crosshead speed of 5 mm/min. For hydrogels without cells, samples were prepared with dimensions of 5 mm in width, 15 mm in length, and 2 mm in height. For tensile tests evaluating the effect of pillar distance, hydrogels were cultured for 14 days in the existing culture mold with dimensions of 15 mm in width, 5 mm in length, and 3 mm in height, but with the hydrogel height reduced to 1.6 mm. Three structures were stacked, their total thickness was measured, and the constructs were then subjected to tensile testing. For electrical stimulation tests, cultured muscle blocks (>14 days) were stimulated with 5 V, 1 Hz, 200 ms pulses using a function generator (GW Instek, Taiwan) via platinum wires in Hanks' Balanced Salt Solution (Sigma-Aldrich).

2.14. Western blot

Protein was extracted from samples using RIPA buffer supplemented with Xpert Protease Inhibitor Cocktail Solution (GenDEPOT, USA). Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Scientific, USA). Equal amounts of protein were separated by 10% SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were incubated with primary antibodies against myosin heavy chain (R&D Systems, MAB4470), troponin T-FS (Santa Cruz Biotechnology, sc-365575), α-actinin (Cell Signaling, 3134S), and GAPDH (Cell Signaling, #97166), followed by incubation with HRP-conjugated secondary antibodies (Cell Signaling, 7076S, 7074S). Protein bands were visualized using the iBright 750 Imaging System (Thermo Scientific, USA).

2.15. Statistical analysis

Data were analyzed using GraphPad Prism version 8. One-way or two-way ANOVA was used. The results are shown as mean ± SEM. Significance was reported as: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

3. Results

3.1. Optimization and characterization of skeletal muscle-specific bioink

Porcine tibialis anterior muscle was decellularized using a multistep workflow comprising lysis, solvent defatting with IPA, disinfection, and lyophilization (Fig. 1A). DAPI staining showed almost complete disappearance of nuclear signals in the MdECM compared with native muscle (Fig. 1B). Quantitative assays confirmed effective decellularization, with DNA reduced from 1344 ± 12.5 ng/mg in native tissue to 37.2 ± 5.3 ng/mg in MdECM. This corresponds to approximately 97 percent removal and a decrease by more than thirty-fold (Fig. 1C). In contrast, structural ECM components, including collagen, were well preserved, supporting matrix structural integrity. Histological staining further supported these results. Hematoxylin and eosin staining showed complete removal of cellular nuclei, while Masson's trichrome staining demonstrated preserved collagen structures with minimal residual cytoplasmic material (Fig. 1D). Immunofluorescence analysis revealed strong signals for fibronectin and laminin, with merged images showing their distribution along the extracellular matrix (Fig. 1E). Collectively, these results indicate that the MdECM retained major extracellular matrix components while removing cellular content; we next evaluated whether the IPA step further improves bioink quality by reducing residual lipids that can compromise gelation and rheology.

Fig. 1.

Decellularization process and characterization of MdECM. (A) Representative photographs of the stepwise decellularization process. (B) DAPI staining showing residual nuclear content (scale bars: 100 μm). (C) Quantification of residual DNA content and elastin retention (DAPI, elastin; n = 4). (D) Histological evaluation using hematoxylin and eosin (H&E) and Masson's trichrome staining (scale bars: 100 μm). (E) Immunofluorescent staining for ECM components (scale bars: 100 μm). (F) Oil Red O staining results (scale bars: 100 μm). Quantitative analysis of (G) triglyceride, (H) elastin, and (I) glycosaminoglycan (GAG) contents (n = 3). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

The extent of lipid removal achieved by IPA treatment was confirmed by Oil Red O staining and triglyceride quantification (Fig. 1F and G). These analyses revealed markedly lower residual lipid content in IPA-treated MdECM (w/IPA) compared with the non-IPA group (w/o IPA), which is consistent with the observed rheological differences described below. Quantitative ECM analysis further confirmed substantial elastin preservation (∼75% relative to native tissue) with comparable levels between IPA-treated and untreated groups, whereas GAG content showed partial retention (∼30% of native) in both conditions (Fig. 1H and I).

3.2. Viscoelastic characterization of skeletal muscle-specific bioink

To determine whether IPA treatment during decellularization influences the structural and mechanical integrity of ECM-derived hydrogels, IPA-treated and untreated MdECM bioinks were comparatively evaluated (Fig. 2A). At the macroscopic level, the IPA-treated formulation retained its initial boundaries and adhered to the surface, whereas the untreated bioink displayed pronounced flow behavior and poor shape fidelity, suggesting reduced structural cohesion (Fig. 2B).

Fig. 2.

Development of IPA-treated muscle-derived extracellular matrix bioink and analysis of viscoelastic properties. (A) Schematic overview comparing the IPA-enhanced decellularization protocol to the typical method, highlighting subsequent evaluations of rheological and structural characteristics. (B) Representative images showing improved shape retention and surface adherence in the IPA-treated bioink compared with the flow behavior of the non-IPA-treated bioink. The red dashed circles indicate the initial boundaries of each bioink. (C) Rheological characterization: (i) Shear-thinning behavior indicating viscosity reduction with increasing shear rate. (ii) Amplitude sweep test assessing the storage modulus and linear viscoelastic region. (iii) The 3ITT results evaluating structural recovery following shear-induced breakdown. (D) Schematic illustration of the thermal gelation analysis under physiological conditions. (E) Photographs of hydrogel formation before and after gelation at 37 °C. (F) Dynamic mechanical analysis of MdECM bioinks: (i) Gelation kinetics assessed via temperature ramp test. (ii) Frequency sweep to evaluate the stability under oscillatory stress. (iii) Stress relaxation test under constant strain. (iv) Creep and recovery tests under constant stress. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

To compare rheological behavior in the liquid state, viscosity, amplitude sweep, and three-interval thixotropy (3ITT) tests were performed at 20 °C, where both formulations remain flowable without temperature-induced gelation. These rheological measurements corroborated the macroscopic observations: both bioinks exhibited shear-thinning behavior; however, the IPA-treated group maintained higher viscosity values across all shear rates (Fig. 2C–i). Amplitude sweep analysis further revealed an increased storage modulus (G′), an extended linear viscoelastic region, and elevated resistance to deformation in the IPA-treated bioink, collectively indicating enhanced mechanical strength (Fig. 2C–ii). In three-interval thixotropy tests (3ITT), the IPA-treated bioink demonstrated more complete structural recovery and delayed network breakdown under repeated shear, reflecting a denser and more resilient hydrogel microstructure (Fig. 2C–iii). The quantitative 3ITT recovery ratio was significantly higher in the IPA-treated bioink (94.9 ± 3.07%) than in the untreated group (71.2 ± 5.43%) (Supplementary Fig. S4A), indicating improved thixotropic recovery and structural rebuildability after shear. This behavior may help the matrix better tolerate repeated cell-generated stresses during tissue formation [31,32].

To monitor changes in rheological properties in real time during gelation, we continuously measured them using a rheometer. Gelation behavior under physiological pH and temperature conditions was investigated (Fig. 2D). Only the IPA-treated bioink underwent stable thermal gelation at 37 °C, while the untreated counterpart failed to retain its shape and collapsed (Fig. 2E). Temperature ramp tests further confirmed this difference: the complex modulus of the IPA-treated group increased more rapidly and reached higher final values, indicating accelerated and more efficient gelation kinetics (Fig. 2F–i). Small amplitude oscillatory shear (SAOS) measurements showed that although both groups maintained G′ greater than G″ across the tested frequency range, the IPA-treated bioink consistently displayed higher absolute moduli, demonstrating superior stability against oscillatory deformation (Fig. 2F–ii). This was corroborated by quantitative SAOS analysis (Supplementary Fig. S4B–i): gel-state G′ of the IPA-treated bioink (12,704 ± 1372 Pa) was approximately two orders of magnitude higher than that of the untreated group (115 ± 121 Pa). Tan δ (G″/G′) remained similarly low in both groups (Supplementary Fig. S4B-ii), indicating solid-like behavior despite the large stiffness difference. The higher G′ suggests a more cohesive ECM network that may better transmit cell-generated tension during alignment [26]. Time-dependent mechanical behavior was assessed through stress-relaxation and creep–recovery tests. In stress relaxation, the IPA-treated hydrogel exhibited a slower decline in stress under constant strain, suggesting stronger elastic resistance and structural retention, whereas the untreated hydrogel dissipated stress more rapidly, indicating a viscosity-dominated response (Fig. 2F–iii). Quantitative fitting of the stress-relaxation curves supported that the relaxation time constant (τ) of the IPA-treated bioink (315.8 ± 12.98 s) was significantly greater than that of the untreated group (87.0 ± 14.83 s), reflecting slower stress decay and a more robust elastic network (Supplementary Fig. S4C). The prolonged relaxation time suggests that the IPA-treated hydrogel sustains mechanical cues longer, a feature known to promote effective myogenic mechanotransduction [33,34]. In the creep–recovery test, the IPA-treated hydrogel displayed partial strain recovery following removal of applied stress, confirming the presence of viscoelastic rebound. In contrast, the untreated hydrogel showed minimal recovery and irreversible deformation, further supporting its viscous-dominated character even at a low applied stress (1 Pa) (Fig. 2F–iv). Consistent with this qualitative observation, the quantitatively extracted creep-recovery ratio revealed that the IPA-treated bioink recovered 91.04 ± 6.89% of its induced strain, while the untreated group recovered only 29.3 ± 5.49%, demonstrating a markedly higher elastic-dominant response in the IPA-treated hydrogel (Supplementary Fig. S4D). The superior elastic recovery further indicates that the IPA-treated ECM can repeatedly withstand and rebound from cell-generated loads, helping maintain the boundary constraints required for uniaxial alignment [35].

Taken together, these findings indicate that IPA incorporation during decellularization substantially enhances the viscoelastic properties and structural integrity of MdECM bioinks. While untreated formulations may offer advantages for applications requiring greater fluidity or injectability, the IPA-treated bioink is better suited for constructing stable three-dimensional muscle-mimetic scaffolds capable of withstanding dynamic mechanical loading in skeletal muscle tissue engineering. Moreover, matrix viscoelasticity has been identified as a key regulator of cell fate and differentiation [[36], [37], [38]]. Thus, the enhanced viscoelastic properties of our MdECM bioink are expected to provide a mechanically robust yet biophysically favorable environment for supporting myogenic differentiation.

3.3. Tuning bioink composition and cell density to enhance myotube alignment and differentiation in a pillar-frame chamber

To determine the appropriate MdECM bioink concentration for skeletal muscle tissue engineering, MdECM was formulated into bioinks of varying concentrations using a pepsin-digestion–based preparation method. Bioinks exceeding 5% (w/v) were poorly soluble, whereas the 1% formulation lacked sufficient rheological integrity to maintain structural fidelity (Supplementary Fig. S5A and B). Therefore, 2%, 3%, and 4% MdECM bioinks were selected for further evaluation.

C2C12 myoblasts encapsulated in each bioink were cultured for 3 days to assess initial cytocompatibility (Fig. 3A–i). Live/dead staining confirmed cell viability above 85% across all conditions (Fig. 3A–ii). Despite comparable viability, cells in lower-concentration bioinks, particularly the 2% formulation, displayed enhanced spreading and cytoskeletal extension, which may be attributable to reduced matrix density and greater pore space [33]. Proliferation assays further supported this observation, showing the highest proliferation in the 2% group, likely facilitated by improved nutrient diffusivity (Fig. 3A–iii; Supplementary Fig. S5C). Based on these findings, the 2% MdECM bioink was chosen for subsequent experiments as it combined favorable cytocompatibility with adequate mechanical stability.

Fig. 3.

Determination of MdECM bioink concentration and cell density to enhance cytocompatibility, contraction, and alignment in pillar-constrained muscle constructs. (A) Biocompatibility assessment: (i) Live/dead fluorescence staining of C2C12 myoblasts encapsulated in MdECM bioinks with varying concentrations (scale bars: 200 μm). (ii) Quantification of cell viability. n.s. indicates not significant (n = 3). (iii) Quantification of cell proliferation (n = 3). (B) Representative images showing the contraction of C2C12-laden constructs in a pillar-frame system. (C) Quantification of the reduction in the construct area over 17 days. (D) Immunofluorescence staining of F-actin at day 17, demonstrating cytoskeletal alignment. Yellow arrows indicate aligned myotubes (scale bars: 200 μm). (E) Directionality analysis using circular histograms and angular distribution plots for constructs with: (i) 1 × 10⁵ cells/mL, (ii) 1 × 10⁶ cells/mL, and (iii) 1 × 10⁷ cells/mL. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

To promote uniaxial alignment, C2C12-laden 2% MdECM bioinks were printed into a pillar-constrained frame system. Over time, constructs exhibited isotropic contraction driven by cellular traction forces, but the surrounding pillars imposed spatial constraints that induced directional tension, resulting in alignment along the longitudinal axis (Fig. 3B; Supplementary Fig. S6). The effect of cell density on contraction and alignment was then investigated using constructs containing 1 × 10⁵, 1 × 10⁶, or 1 × 10⁷ cells/mL (Fig. 3C). After 17 days, constructs with 1 × 10⁷ cells/mL showed the greatest area reduction (∼39%), followed by 1 × 10⁶ cells/mL (∼30%), whereas the 1 × 10⁵ group contracted only minimally (∼25%). The limited compaction at low density indicates insufficient traction force generation. These results suggest that intermediate to high cell densities are necessary to drive effective matrix remodeling. Cytoskeletal organization was examined by F-actin staining at day 17. Constructs with 1 × 10⁶ and 1 × 10⁷ cells/mL exhibited robust uniaxial alignment of myotubes, whereas those with 1 × 10⁵ cells/mL showed disorganized orientation (Fig. 3D). Quantitative directionality analysis confirmed higher alignment indices in the 1 × 10⁶ and 1 × 10⁷ groups, with little difference between them (Fig. 3E; Supplementary Fig. S7).

3.4. Spatial regulation of muscle fiber orientation and mechanical maturation via 3D-printed pillar-frame systems

To examine how mechanical boundary conditions regulate the structural organization of engineered muscle, we developed a 3D printing–assisted chamber with tunable pillar thickness and interpillar spacing (Fig. 4A). This system was designed to direct myotube alignment through the balance between cell traction forces within the construct and the reactive forces applied by the pillars. We hypothesized that when pillars are too closely spaced, traction forces cannot be effectively transmitted across the construct, resulting in poor alignment, whereas excessive spacing would leave the central region under low tensile stress, also impairing alignment (Fig. 4B–i). To test this hypothesis, muscle blocks were cultured between pillars spaced 10, 15, or 20 mm apart and analyzed after 14 days of differentiation (Fig. 4B–ii).

Fig. 4.

Spatial regulation of myotube alignment using a 3D-printing-assisted pillar chamber. (A) Schematic of pillar configurations used to tune boundary conditions. Two parameters were varied: pillar thickness (thin, 1 mm; thick, 3 mm) and interpillar distance (10, 15, 20 mm). (B) Conceptual stress maps of muscle blocks formed between pillars with different spacings, and the uniaxial tensile testing setup used for mechanical evaluation. (C) Immunofluorescence for myosin heavy chain (MHC) showing myotube alignment for each spacing condition (scale bars: 100 μm). (D) Uniaxial tension results: (i) representative stress versus strain curves, (ii) comparison of ultimate tensile strength (n = 3). (E) Photographs of engineered muscle blocks cultured for 17 days (3 days in growth medium + 14 days in differentiation medium) in chambers with thin pillars and thick pillars. (F) Spatial analysis of fiber orientation at the construct center and near the pillars. (i) thin-pillar group, (ii) thick-pillar group. Panels show MHC staining with directionality plots (scale bars: 500 μm). (G) Demonstration of spatial patterning of muscle-fiber orientation using a three-pillar configuration positioned at a right angle (scale bars: 500 μm). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Immunofluorescence staining for MHC revealed that constructs at 15 mm spacing contained densely packed and uniformly aligned myotubes, indicative of advanced maturation. By contrast, the 20 mm group displayed sparse and disorganized fiber morphologies (Fig. 4C). Uniaxial tensile testing corroborated these observations: constructs with 15 mm spacing achieved the highest ultimate tensile strength, whereas both 10- and 20-mm groups exhibited reduced mechanical integrity despite comparable strain capacities (Fig. 4D–i, ii). These findings suggest that moderate tensile loading optimally reinforces structural organization, while either insufficient or excessive strain compromises mechanical development.

We next assessed whether pillar thickness modulates local alignment patterns. Chambers fabricated with thin (1 mm) or thick (3 mm) pillars were used to culture muscle blocks for 17 days (3 days in growth medium + 14 days in differentiation medium) (Fig. 4E). MHC staining showed uniaxial alignment in the central regions of both groups; however, orientation near the pillars was more dispersed in the thick-pillar group, consistent with asymmetrical strain fields imposed by broader constraint surfaces (Fig. 4F). These results indicate that pillar thickness influences localized mechanical cues, thereby modulating regional fiber organization even when global alignment is preserved.

Finally, a three-pillar chamber arranged at a right angle successfully induced multidirectional myotube alignment within a single construct (Fig. 4G). This demonstrates the feasibility of spatially patterning fiber orientation by structural design, which could be advantageous for engineering complex muscle architectures requiring region-specific contractile directions. Such strategies may also inspire cardiac tissue engineering, where multilayered myocardial fibers generate torsional contraction of the heart [[39], [40], [41]].

Collectively, these results highlight the importance of tunable boundary conditions, specifically pillar geometry and spacing, in guiding both the structural organization and mechanical maturation of engineered muscle blocks.

3.5. Engineering a stretchable multi-chamber bioreactor to enhance the myogenic maturation of muscle blocks

To promote the in vitro maturation of engineered skeletal muscle, we developed a stretchable bioreactor capable of applying synchronized cyclic strain to multiple constructs simultaneously. Candidate platform materials, polydimethylsiloxane (PDMS) and Ecoflex, were first compared by uniaxial tensile testing. Both materials exhibited hyperelastic behavior, but Ecoflex demonstrated markedly higher elongation at break and required lower stress to achieve comparable strain (Supplementary Fig. S8A). Consequently, Ecoflex was selected as the platform material. Hyperelasticity parameters were extracted using a Mooney–Rivlin five-parameter model, which was incorporated into finite element simulations to predict strain distribution under physiological loading (Supplementary Fig. S8B). To experimentally assess these predictions, strain distribution was evaluated using digital image correlation (DIC) under equivalent Ecoflex material and deformation conditions, showing good agreement with the FEA results (Supplementary Fig. S9).

The bioreactor system consisted of a multi-chamber Ecoflex platform fabricated by combined 3D printing and mold casting (Supplementary Fig. S10A). This platform was integrated with a linear actuator and Arduino-based control system to deliver cyclic uniaxial strain. Finite element analysis (FEA) guided optimization of the platform geometry by varying gripper-hole number and connection schemes. Among the tested designs, the “4 holes and 2 connections” (4H–2C) configuration produced the most uniform strain across chambers. Additional optimization of lateral (5 mm) and vertical (3 mm) chamber spacing further minimized strain disparities (Supplementary Fig. S10B–C). This optimized design was adopted for subsequent experiments (Fig. 5A and Supplementary Video S1). Prior to cell culture experiments, the Ecoflex/PCL platform was thoroughly washed, and its biocompatibility was validated using a leachate-based cytotoxicity assay showing no detectable cytotoxic effects (Supplementary Fig. S11).

Fig. 5.

Engineering uniaxially aligned muscle blocks using a stretchable multi-chamber bioreactor. (A) Schematic of the bioreactor system showing the stretchable multi-chamber platform, actuator, and chamber design. (B) Photographs of the stretchable platform containing engineered muscle blocks under unstretched and stretched conditions. (C) Immunofluorescence staining for myosin heavy chain (MHC) after 14 days of differentiation, showing enhanced myotube alignment and maturation under cyclic strain (scale bars: 100 μm). (D) Calcium imaging of muscle blocks using Fluo-4 AM during electrical stimulation, confirming synchronized calcium oscillations. (E) Western blot analysis of myosin heavy chain (MHC), α-actinin, and troponin T demonstrating enhanced myogenic protein expression under cyclic strain. (F) Normalized protein expression levels of MHC, α-actinin, and troponin T relative to GAPDH. (G) Quantitative gene expression analysis of myogenic markers under different cyclic strain conditions. qRT-PCR data were normalized to β-actin. Data were presented as mean ± SEM (n = 3). ∗p < 0.05, ∗∗p < 0.01.

To mimic exercise-like mechanical stimulation, engineered muscle blocks were subjected to both eccentric and isometric stimuli at 0.09 Hz for 10 min every hour [[42], [43], [44]]. In addition to comparing static and stimulated groups, we further classified the stimulated constructs by strain amplitude—moderate (10%) and excessive (20%)—to examine how mechanical stress affects the maturation of engineered muscle tissue. Engineered muscle blocks were differentiated for 14 days under cyclic strains of 0%, 10%, or 20%. Constructs subjected to strain elongated axially and thinned in cross-section compared to static controls (Fig. 5B). Immunofluorescence staining revealed that 10% strain produced robust MHC expression and organization (Fig. 5C). By contrast, unstretched constructs showed sparse and disorganized fibers, while 20% strain resulted in partially aggregated and misaligned structures, suggesting that excessive loading disrupts myotube maturation. These data indicate that an intermediate strain magnitude of 10% provides the most favorable environment for structural and molecular development.

Functional maturation was further assessed by electrical stimulation on day 14 (after 14 days of differentiation). Constructs contracted rhythmically in response to square-wave pulses (5 V, 500 mHz [0.5 Hz]), matching the stimulation frequency (Supplementary Fig. S12A–B and Supplementary Video S2). Calcium imaging with Fluo-4 AM demonstrated synchronous oscillations in fluorescence intensity aligned with stimulation pulses, confirming active calcium handling (Fig. 5D; Supplementary Fig. S12C; Supplementary Video S3).

At the molecular level, Western blot analysis on day 7 demonstrated that cyclic strain increased the expression of key myogenic proteins including MHC, α-actinin and troponin T (Fig. 5E and F). To verify the robustness of gene expression trends, analyses were first performed using GAPDH as a reference gene (Supplementary Fig. S13). Consistent with these changes, subsequent gene expression analysis normalized to β-actin further revealed that myogenic regulatory and structural markers including MyoG, MYF6, MHCI, MHCII and MHCIV were broadly upregulated in response to strain stimulation. However, the magnitude of this upregulation was attenuated under the 20% strain condition compared with the 10% strain condition (Fig. 5G). Consistent across protein- and transcript-level readouts, the 10% strain condition produced the strongest myogenic response whereas the 20% strain condition attenuated these effects. Collectively, these results indicate that moderate cyclic strain at 10% promotes myogenic differentiation and maturation, while the 20% strain condition attenuates these maturation-associated readouts (Fig. 5E–G).

Together, these findings validate the stretchable multi-chamber bioreactor as an effective platform for enhancing the alignment, molecular maturation, and contractile function of engineered muscle tissues. Importantly, they underscore the necessity of finely tuned mechanical inputs to optimize muscle development.

4. Discussion

The multistep decellularization protocol efficiently removed cellular material from porcine tibialis anterior muscle, as shown by DAPI imaging and by residual DNA below 50 ng per mg dry tissue, which is a frequently cited threshold for reducing immunogenic risk [45]. Although decellularization substantially reduces cellular remnants, xenogeneic MdECM may still elicit host responses. Therefore, translational development would require stringent quality controls (e.g., residual DNA, endotoxin, and process residuals) and systematic in vivo immunological evaluation of the final bioink formulation. Compared with detergent-only muscle protocols that rely on prolonged exposure to sodium dodecyl sulfate or Triton X-100, our workflow integrated a deliberate delipidation step. Prior studies report that detergent-only schemes often leave intramuscular lipids that hinder downstream gelation and may require harsher processing that compromises matrix proteins [46]. In contrast, the IPA step in our sequence addressed the high lipid fraction characteristic of skeletal muscle and coincided with additional reduction in DNA, consistent with reports that organic solvents facilitate membrane disruption and improve removal of nuclear remnants in lipid-rich tissues [47]. These comparisons position the present protocol as a muscle-tailored compromise between efficacy and matrix preservation: it reduces immunogenic content to commonly accepted levels, mitigates lipid-related barriers to hydrogel formation, and maintains structural proteins that contribute to elasticity and printability [48,49]. Beyond these translational considerations, from a scalability and sourcing standpoint, porcine tibialis anterior muscle is broadly available through established food-industry supply chains, enabling scalable procurement. For bulk production, standardized procurement with traceability and predefined batch release criteria (e.g., residual DNA and rheological benchmarks) could help minimize lot-to-lot variability.

The results of this study underscore the critical role of viscoelasticity in skeletal muscle tissue engineering, particularly in systems in which mechanical constraints are used to guide cell alignment. Although previous studies have utilized ECM-derived hydrogels within pillar frames to induce myotube orientation, few have systematically optimized the rheological properties of the bioink, which directly influences mechanical transmission and structural fidelity during tissue maturation [[50], [51], [52], [53]]. Our findings demonstrate that residual lipids present in conventional decellularized ECM significantly compromise fiber crosslinking and mechanical coherence, leading to a loss of shape fidelity under a mechanical load. By introducing a simple yet effective IPA-based delipidation step, we reinforced the viscoelastic integrity of MdECM, resulting in bioinks with improved shear resistance, recoverability, and structural stability. These properties are essential for sustaining cell-mediated traction and promoting uniaxial alignment during myogenesis. Matrix mechanical and viscoelastic properties are also known to regulate cell morphology and differentiation [54,55]. Therefore, refining MdECM bioink mechanics is important for the rational design of tissue-engineered muscle constructs, and future studies should tailor and characterize these properties more systematically in skeletal muscle tissue engineering.

Equally important is the development of a facile, modular strain-bioreactor system designed to deliver cyclic mechanical stimuli to engineered muscle constructs. A pioneering study has demonstrated topographical guidance with cyclic stretch to probe muscle mechanotransduction by employing laser-engraved pseudo-3D patterns on PDMS [56]. In our study, we developed a pillar-frame system to distribute tensile forces evenly across the entire 3D muscle block, incorporated a multi-chamber design that enables the simultaneous maturation of multiple bulk muscle constructs, and implemented a 3D-printed platform with a strain motor to provide a tunable range of strain amplitudes. Parametric studies of interchamber spacing revealed that the optimized configuration (5 mm lateral, 3 mm vertical) reduced strain heterogeneity, enabling more consistent mechanical conditioning. This platform facilitates myotube alignment and enhances the expression of myogenic and sarcomeric proteins under cyclic strain conditions. Compared with pillar-frame confinement alone, constructs exposed to cyclic strain showed more pronounced alignment and higher expression of key myogenic and sarcomeric genes, suggesting that confinement provides a baseline alignment cue, while dynamic loading further promotes maturation. Mechanistically, confinement may predominantly engage integrin–FAK–YAP/TAZ signaling via static anchorage and viscoelastic resistance, whereas cyclic strain may additionally activate stretch-associated Ca²⁺ signaling and load-sensitive mTOR pathways. Although these pathways were not directly tested in this study, they provide plausible mechanotransductive cues underlying the enhanced outcomes under dynamic loading.

In practical terms, our strain-biased culture system supports the scalable generation of miniaturized muscle blocks with relatively uniform strain exposure, which is an important advantage over bulk tissue constructs that often suffer from diffusion limitations and uneven mechanical stimulation [57,58]. In this regard, the system offers a practical foundation for “muscle-on-a-chip” applications, particularly in the context of drug screening for conditions such as sarcopenia, cachexia, and obesity-related muscle atrophy [[59], [60], [61], [62]]. Although the bioreactor effectively demonstrated the ability to modulate myogenic outcomes, our results also support the notion that excessive stimulation—mimicking strenuous exercise—can hinder muscle maturation, in line with previous reports. These studies, together with our findings, suggest that excessive artificial mechanical loading can suppress mTOR signaling, thereby reducing myogenic factor (Myo) expression, and can also disrupt matrix–myocyte interactions, leading to attenuated biochemical signaling [[63], [64], [65], [66]]. While the observed physiological responses suggest potential relevance to overuse-related muscle damage models (e.g., rhabdomyolysis), a systematic investigation of key dynamic parameters—including strain magnitude, frequency, and duration—is essential for further refinement of the system.

From an engineering perspective, our bioreactor is limited to delivering eccentric and isometric stimuli. Accordingly, this design precludes active, neuromuscular-like concentric contractions, which restricts its capacity to fully recapitulate the mechanical behavior of real exercise. In addition, the contractile responses induced by electrical stimulation were modest and declined over time, suggesting incomplete functional maturation. This may reflect the absence of neuromuscular junctions and insufficient support for sustained excitation–contraction signaling. Although these observations suggest excitation–contraction responsiveness, quantitative metrics—including absolute contractile force, frequency-dependent twitch–tetanus responses, pharmacological modulation, and Ca²⁺ transient readouts—remain important benchmarks for assessing the maturity and physiological relevance of engineered muscle blocks in future studies. Similarly, the dense avascular architecture of hydrogel constructs poses a significant barrier to their long-term survival and integration after implantation. Engineered muscle tissues are unlikely to maintain viability in vivo beyond the acute phase without perfusable networks.

In this study, C2C12 myoblasts were used as a reproducible and widely adopted myogenic model for platform optimization and mechanobiological evaluation. To further support general applicability, batch-to-batch reproducibility should be established across independent MdECM bioink preparations. Further validation in human myogenic cells (e.g., primary myoblasts and/or iPSC-derived myogenic progenitors) would strengthen translational relevance. Therefore, future studies should focus on multiple complementary directions. First, integrating neuromuscular junction components, such as neural spheroids, motor neurons, or optogenetic actuators, may improve the fidelity of the functional output. Second, vascularization strategies, including endothelial co-printing, guided angiogenesis, and microfluidic perfusion, should be prioritized to overcome diffusion limitations and support construct viability. Additionally, tailoring the bioink not only for bulk viscoelasticity but also for spatial stiffness gradients may better emulate the biomechanical heterogeneity of native muscles, potentially influencing both alignment and maturation.

In summary, this study demonstrated that combining ECM viscoelastic reinforcement with modular mechanical conditioning provides a viable path toward scalable and functionally competent muscle tissue engineering. Although key challenges remain, particularly in achieving fully contractile, vascularized, and innervated tissues, our findings offer a technically accessible platform and open new avenues for disease modeling and regenerative applications.

CRediT authorship contribution statement

Jae Woo Back: Data curation, Formal analysis, Investigation, Methodology, Resources, Visualization, Writing – original draft. Da-Young Jo: Data curation, Formal analysis, Methodology, Validation, Writing – original draft. Jeongho Lee: Methodology, Validation, Visualization. Jae-Seong Lee: Methodology. Minjun Ahn: Investigation, Methodology, Writing – original draft, Writing – review & editing. Byoung Soo Kim: Funding acquisition, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the supplementary data to this article:

Data availability

Data will be made available on request.