Main

Cell-based meat (CBM)—also termed cultured meat, cultivated meat or clean meat—is an evolving technology with the aim of producing meat from cell cultures1,2. The development of this technology is motivated by sustainability issues related to traditional animal-based meat production methods, including the overuse of antibiotics, food and water security, food safety, environmental footprints and poor animal welfare conditions3. CBM technology may also enable new fields of culinary meat innovation3,4.

Tissue-engineering techniques, such as seeding cells inside an environment that recapitulates the native tissue conditions, are required to replicate muscle development in vitro. Vascularized skeletal muscle tissues can be generated inside three-dimensional (3D) scaffolds by co-culture of muscle cells, endothelial cells and supporting cells5,6,7,8,9. Skeletal muscle progenitors—the primary cell type required for this technique10—differentiate into muscle fibres under starvation conditions in the presence of insulin-like growth factor 1 (IGF-1)11,12. Their immediate native environment is composed of a honeycomb-like structure called the perimysium, which is filled with a soft gel called the endomysium13. The endomysium can be recapitulated using simple biological gels, such as fibrin and collagen, and extracellular matrix (ECM)-producing supporting cells, such as fibroblasts. During tissue development, the gels and supporting cells reorganize and a complex tissue-related ECM can be deposited14. While it is feasible for supporting cells to produce an amorphous endomysium-like ECM during tissue maturation in vitro, the perimysium honeycomb structure should be added externally as a porous, edible scaffold that also provides mechanical support to the developing tissue. Since the scaffold constitutes a main component of the final product, it should resemble the composition and properties of meat. Muscle tissues are abundant in aligned microvascular networks formed by endothelial cells15,16. While endothelial cells might not have a direct effect on meat quality, they regulate skeletal muscle tissue behaviour via angiocrine signalling17,18,19.

Textured soy protein (TSP) is a porous, food-grade, inexpensive byproduct of soybean oil processing that was invented in 196020. It is frequently used as a raw material in meat substitutes, due to its texture and high protein content (>50%)21, which both improve its nutritional value and provide anchor points for cell adherence; therefore, it can be used as a scaffold. Its porosity is an important scaffolding feature that facilitates tissue development throughout the volume of the 3D scaffold22. TSP can be tailored to various sizes and shapes, which will be useful for scale-up processes (for example, adjustment to bioreactor geometry) in future CBM production. These properties render TSP a suitable candidate for processes aimed at engineering tissues for human consumption. Therefore, we aimed to create a bovine muscle tissue on an edible scaffold made of TSP.

Results

TSP scaffold characterization

TSP is a dry, porous, protein-based material (Fig. 1a) that expands and becomes softer when soaked in water (Supplementary Fig. 1). Two TSP scaffolds were tested, TSP-1 and TSP-2, containing 69% and 53% protein, respectively. Scanning electron microscopy (SEM) images showed that the fundamental structure of the TSP-1 scaffold, on the dimensions of cell-surface interactions23, contained spheres of 36 ± 7 nm (Fig. 1b). Analysis of the TSP-1 pore size distribution (Fig. 1c) showed that the vast majority of pores were small (88% <50 µm), several were medium sized (11% in the range 50–400 µm) and a small percentage were large (0.5% >400 µm) (Supplementary Fig. 2). The small, medium and large pores accounted for 15, 54 and 32% of the total pore area, respectively (Fig. 1d). Even though there were only a few large pores, they accounted for a considerable proportion of the scaffold. 3D quantification of TSP-1 and TSP-2 micro-computed tomography (micro-CT) scans (Fig. 1e) showed that the TSP pores were highly interconnected, with 56 ± 2 and 42 ± 10% porosity and 181 ± 11 and 148 ± 17 cm2 g−1 surface area, respectively (Fig. 1f and Extended Data Fig. 1). In addition, the TSP showed pore directionality (Supplementary Video 1), which is crucial for myotube alignment24,25,26.

Fig. 1: TSP scaffold characterization.
figure 1

a, TSP-1 scaffold. Scale bar, 500 μm. b, SEM image of the basic structure of the TSP scaffold. Scale bar, 200 nm. c, Confocal image of TSP-1 scaffold pores. Scale bar, 500 μm. d, Percentage of pore area as function of pore diameter, as determined by confocal measurements. The experiment was repeated three times. e, Micro-CT 3D reconstruction of TSP-1 scaffold pores. Yellow marks the largest interconnected pore in the scaffold. Scale bar, 1 mm. f, Porosity, connectivity and surface area analyses of TSP-1 and TSP-2 from micro-CT measurements. The data points represent individual values. Bar heights and error bars represent mean ± s.d. (n = 3 for all measurements). Statistical significance was determined by t-test. No significant differences were observed.

Full size image

To assess the ability of the TSP scaffolds to support cell proliferation, TSP scaffolds were seeded with 5 × 104 fibroblasts per scaffold and observed for 3 weeks. The fibroblasts attached to and proliferated on and throughout the thickness of the TSP scaffold, covering the TSP surface (Supplementary Fig. 3a,b and Supplementary Video 2). Bovine satellite cell (BSC) monoculture and multicellular seeding efficiency proved to be >80%, showing that the majority of the cells remained on the TSP scaffold following cell seeding (Supplementary Fig. 4a,b). The cell penetration capacity was shown qualitatively on a 1-cm unmodified TSP chunk, where cells in the middle versus outer sections were found to be at similar densities (Supplementary Fig. 5).

BSC myogenesis on 3D scaffolds

BSC proliferation (LLM1) and differentiation media (LLM2) formulations were optimized for 3D myogenesis. Poly-l-lactic acid (PLLA) and polylactic glycolic acid (PLGA) scaffolds were used for initial optimization of the proliferation and differentiation protocols, building on our previous work9,27,28. The optimized protocol was tested on TSP scaffolds. LLM1 proliferation medium was optimized for improved cell coverage, which is crucial for myotube fusion and maintenance of BSC stemness. The results from two-dimensional cultures showed that LLM1 medium prevented BSC differentiation (as demonstrated by the absence of myotubes; Supplementary Fig. 6a) while improving cell coverage and viability compared with the control proliferation medium (Supplementary Fig. 6b). When tested on TSP scaffolds, the LLM1 medium significantly improved BSC coverage from 18–71% (Fig. 2a,c; P = 0.000061). Phalloidin staining showed that in LLM1 medium the cells exhibited an elongated fibre-like morphology, whereas in the control medium the cells had a rounded morphology (Fig. 2a).

Fig. 2: BSC proliferation and differentiation on TSP scaffolds.
figure 2

a, BSCs stained with DiI (red; top) or phalloidin (red) plus DAPI (blue) (bottom) after proliferation in control medium (left) or LLM1 medium (right). Scale bars, 300 μm (top); 10 μm (bottom). The experiment was repeated twice. b, BSCs stained with DiI (red; top) or desmin (green; bottom) after differentiation on TSP-1 (left) or TSP-2 scaffolds (right) for 4 d. Scale bars, 50 μm (top); 30 μm (bottom). The experiment was repeated twice. c, Quantification of BSC coverage after the proliferation stage, as indicated by DiI staining (n = 5 for control and n = 8 for LLM1; ****P = 0.000061). d, Quantification of BSC coverage after the differentiation stage, as indicated by DiI staining (n = 4 for each TSP type). The experiment was repeated once. e, Quantification of myotube coverage after differentiation (n = 3 for each TSP type). In c–e, the data points represent individual values, bar heights and error bars represent mean ± s.d., and statistical significance was determined by t-test. f, Representative confocal images of BSC differentiation on TSP after 4 d. The cells were isolated from four different bovine subjects (from left to right): BSCs, BSC-1, BSC-2 and BSC-3. Cells were stained with desmin (red) and nuclei were stained with DAPI (blue). Scale bar, 50 μm.

Full size image

Differentiation media supplemented with IGF-1 and epidermal growth factor (EGF) supported the formation of elongated myotubes versus small myotubes in the control treatment (Supplementary Fig. 7a). Quantification of desmin staining (Supplementary Fig. 7b) showed that LLM1 + LLM2 treatment improved myotube coverage 3.9-fold (from 3.14% to 12.4%), average myotube area 2.4-fold and myotube shape complexity 2.9-fold compared with the control proliferation and control differentiation media treatment. A full factorial model (2 × 23) of three main effects and their interactions was fit to an experiment with randomized samples (Supplementary Fig. 7b). The main factors were the proliferation medium (control or LLM1), the addition of IGF-1 to the differentiation medium and the addition of EGF to the differentiation medium. The model showed that IGF-1 addition, the proliferation medium and their interaction (proliferation medium × IGF-1) significantly improved the total myotube coverage (P = 0.0003, P = 0.0007 and P = 0.0004, respectively), average myotube area (P = 0.0004, P = 0.0008 and P = 0.0004, respectively) and myotube shape complexity (P = 0.0130, P = 0.0002 and P = 0.0053, respectively) (Supplementary Table 1). EGF did not significantly impact myotube coverage; however, it had a significant effect on the average myotube area (P = 0.0157) and myotube shape complexity (P = 0.0023) and had a synergistic effect with LLM1 (proliferation medium × EGF) on myotube shape complexity (P = 0.0166; that is, both IGF-1 and EGF are important in the differentiation medium LLM2). When evaluating the responses of TSP-1 and TSP-2 scaffolds to the new differentiation protocol (Fig. 2b), a substantial proportion of the cell population remained on the scaffold after differentiation under starvation conditions (Fig. 2d). In addition, multinucleated, elongated myotubes were observed on both TSP scaffold types, albeit with low myotube coverage on both (Fig. 2e). Myotubes formed with BSCs from different isolations (Fig. 2f) on TSP.

Myogenesis of multicellular cultures on TSP scaffolds

The effect of multicellular seeding with bovine smooth muscle cells (BSMCs) and/or bovine endothelial cells (BECs) on myogenesis progression was investigated on TSP scaffolds (Fig. 3a). Abundant myogenin-positive nuclei were observed in the BSC/BSMC co-culture and BSC/BSMC/BEC tri-culture. Higher magnification of BSC/BSMC co-culture and BSC/BSMC/BEC tri-culture revealed that cells exhibited an elongated morphology (Fig. 3b). Quantification of myogenin-positive nuclei (Fig. 3c) showed significantly higher counts on TSP-1 scaffolds, with the highest values recorded for BSC/BSMC and tri-culture (5.6- and 4.6-fold higher than for the BSC monoculture with P values of P = 0.0011 and P = 0.0107, respectively). In line with these findings, western blot analyses showed a 3.6- and 4.6-fold increase in myogenin levels in BSC/BSMC and the tri-culture, respectively, compared with the monoculture (Fig. 3d,e). Quantitative PCR (qPCR) analysis of myogenesis-related genes showed that BSC/BSMC had significantly higher myogenin expression compared with the tri-culture (Fig. 3f; P = 0.01).

Fig. 3: Myogenesis of multicellular cultures on TSP scaffolds.
figure 3

a, Immunofluoresence staining of BSCs (DiI; red), nuclei (DAPI: blue) and myotube differentiation (myogenin; white) after 7 d in differentiation medium. Scale bars, 300 μm. The myogenesis experiments were repeated four times. b, High-magnification images of BSC/BSMC co-culture and BSC/BSMC/BEC tri-culture. Scale bars, 30 μm. c, Image analysis of myogenin-positive nuclei. Respective P values (top to bottom) were *P = 0.0107, **P = 0.0062, ***P = 0.0007 and **P = 0.0011. d, Western blot analysis of myogenin and GAPDH expression in empty scaffolds and scaffolds with BSC, BSC/BSMC co-culture and tri-culture (Tri) after differentiation. e, Quantification of myogenin expression normalized to GAPDH from western blot gel. Respective P values (top to bottom) were **P = 0.0057, **P = 0.0072, ****P = 0.000021 and ***P = 0.0004. f, qPCR analysis of MYH, MyoD, MyoG and PAX3 in BSC/BSMC co-cultures and tri-cultures. **P = 0.01. In c, e and f, the data points represent individual values, whereas bar heights and error bars represent mean ± s.d. (n = 3 for all groups). A two-way ANOVA was performed with Šidák’s post-hoc multiple comparison.

Full size image

Tissue development on TSP scaffolds

The effect of multicellular seeding on ECM deposition and scaffold weight was investigated after tissue development (Fig. 4). To estimate ECM protein deposition, trichrome staining was performed on histological sections of TSP scaffolds following tissue development. The scaffolds were stained dark blue and protein depositions were stained light blue (Fig. 4a), allowing the area fraction to be quantified (Fig. 4b). There was no significant effect of TSP type on protein deposition. However, the BSC/BSMC co-culture notably increased protein deposition compared with BSC monoculture (1.9- and 1.7-fold change for TSP-1 and TSP-2, respectively). Scaffold weights were measured and normalized to empty scaffolds incubated under the same conditions, to reveal the cell contribution to scaffold weight (Fig. 4c). Analysis of cell combinations and TSP type showed that BSC monoculture substantially reduced the scaffold weight compared with the other cell combinations, while the scaffold weight in BSC/BSMC co-culture was 1.6-fold higher than in BSC monoculture for both TSP-1 and TSP-2, providing further evidence for the contribution of BSMCs to tissue development. Significant collagen type I deposition in BSC/BSMC co-culture and tri-culture was observed on TSP-1 scaffolds (Fig. 4d); this was 3.2-fold higher in BSC/BSMC co-culture and 3.8-fold higher in tri-culture versus BSC monoculture (P = 0.000042 and P = 0.000002, respectively), with no significant difference between BSC/BSMC co-culture and tri-culture (Fig. 4e).

Fig. 4: ECM deposition of multicellular cultures on TSP scaffolds after BSC differentiation.
figure 4

a, Trichrome staining of cross-sections from TSP-2. b, Quantification of protein deposition using trichrome staining (n = 2 for BSC/BEC on TSP-1 and BSC on TSP-2; n = 3 for all other conditions; *P = 0.0153). c, Scaffold weight normalized to empty scaffold (n = 3 for all conditions). The experiment was repeated three times. Respective P values (top to bottom) were *P = 0.0143, **P = 0.0064 and *P = 0.0282. d, Western blot gel of collagen and GAPDH expression. The experiment was repeated once. e, Quantification of collagen expression normalized to GAPDH from western blot gel (n = 3 for all conditions). Respective P values (top to bottom) were ****P = 0.000002, ****P = 0.000042, *P = 0.0106 and *P = 0.0179. In b, c and e, the data points represent individual values, whereas bar heights and error bars represent mean ± s.d. A two-way ANOVA was performed with Šidák’s post-hoc multiple comparison.

Full size image

Protein expression in engineered muscle tissues

To investigate the contribution of BSMCs to engineered muscle tissue, the proteome of BSC and BSC/BSMC cultures on PLLA/PLGA scaffolds was analysed using mass spectrometry. PLLA/PLGA scaffolds were used in these experiments because they are non-protein scaffolds, whereas TSP scaffolds contain 50–70% proteins and are composed of thousands of different types of protein. In total, 3,196 proteins were measured, of which 976 were differentially expressed (P < 0.05). Of the differentially expressed proteins, 85 showed elevated expression (log2[fold change] > 0.8) and 119 showed decreased expression (log2[fold change] < −0.8) (Fig. 5a) in BSC/BSMC co-culture compared with BSC monoculture. A gene interaction plot of the elevated proteins showed collagen and laminin clusters and a complex network of ECM-related and muscle maturation proteins (Fig. 5b,c and Extended Data Fig. 2) in the co-culture combination. Enriched Gene Ontology terms showed that most changes were attributable to ECM composition, organization and function (Extended Data Fig. 3). The Gene Ontology terms were organized into four functional clusters and plotted over the differentially expressed proteins (Fig. 6), which showed that cluster A (ECM gene sets) had an average fold change of 3.5, with 29 elevated proteins and two proteins with decreased expression (Fig. 6a and Extended Data Fig. 4). Cluster B (collagen gene sets; Fig. 6b) had an average fold change of 3.4, including 16 elevated proteins and no proteins with decreased expression. Clusters C (Fig. 6c) and D (Fig. 6d) showed similar but weaker trends.

Fig. 5: Proteomic analysis of BSC/BSMC co-culture compared with BSC monoculture on PLLA/PLGA scaffolds after differentiation.
figure 5

a, Volcano plot showing elevated (red) and downregulated (blue) proteins in co-culture compared with monoculture. b, Heat map showing the clustering of differentially expressed proteins. BSC-1, -2 and -3 indicate replicates of BSC monoculture samples, whereas BSC/BSMC-1 and -2 indicate replicates of BSC/BSMC co-culture samples (n = 3 for BSC monoculture; n = 2 for BSC/BSMC). c, Interaction of the genes that encode the proteins that are elevated in the co-culture. The red and purple circles denote collagen and laminin gene clusters, respectively. The experiment was repeated once.

Full size image
Fig. 6: Distribution of differentially expressed proteins representing selected Gene Ontology terms.
figure 6

a, ECM. b, Collagen. c, Sarcolemma, basement membrane and basal lamina. d, Actin filament binding and stress fibres.

Full size image

Mechanical and texture evaluation

We conducted mechanical and tasting tests with TSP seeded with BSC and BSC/BSMC. Macroscopic observation showed that scaffolds with BSC/BSMC were less fragile and were covered with an additional shiny layer compared with scaffolds with BSC, which we assume is the secreted ECM (Fig. 7a). Mechanical property measurements showed that the different constructs displayed a Young’s modulus in the same range as the native bovine muscle (Fig. 7b). BSC/BSMC co-culture displayed higher ultimate tensile strength (UTS) values, and were hence less brittle and had similar mechanical properties to native bovine muscle. The fried samples were prepared using an iron-cast frying pan that was pre-heated to 160 °C. The products were pan fried for 1 min on each side until a brown colour was created following the Maillard reaction. The baked samples were placed into a pre-heated oven at 180 °C for 10 min. We performed sequential comparative testing tests. Three volunteers tasted the product after cooking and unanimously noted that the cell-based meat products had a pleasant meaty flavour and sensorial attributes, achieving a typical meat bite and texture. There was also a notable difference compared with scaffold that did not contain any cells (Fig. 7c).

Fig. 7: Mechanical and texture properties of TSP scaffolds.
figure 7

a, Representative images of 6-mm cylinder TSP scaffolds seeded with BSCs only (left) or with BSC/BSMC co-culture (right). b, Young’s modulus and UTS of the various scaffold types and of native bovine muscle from the literature24,25. For the Young’s modulus, respective P values (top to bottom) were ****P = 0.000004, ***P = 0.0002 and *P = 0.0283. For the UTS, respective P values (top to bottom) were ****P = 0.00000008, ****P = 0.000022786 and **P = 0.0018. The data points represent individual values, whereas the bar heights and error bars represent mean ± s.d. (n = 4 for empty scaffolds; n = 5 for BSC scaffolds; n = 6 for BSC/BSMC scaffolds). A two-way ANOVA was performed with Šidák’s post-hoc multiple comparison. c, Representative images of baked and fried TSP scaffolds with and without cells.

Full size image

Discussion

CBM development is an emerging sub-discipline of cellular agriculture, which aims to utilize state-of-the-art biotechnological solutions to generate food. CBM research utilizes skeletal muscle tissue-engineering technology, replacing human cells (which are used for tissue regeneration) with farm animal cells1,3,4,29. TSP—a mass-produced, edible, porous byproduct of soybean oil production—was utilized to enhance cell adhesion and 3D growth in CBM-targeted cultures (Supplementary Fig. 3). The scaffold material is inexpensive and does not require further modifications, rendering it highly applicable for the mass production of food products. While these properties are useful for CBM, they are not unique to TSP. Other possible scaffolding materials include seitan or texturized proteins from other organisms20,21. While TSP shows great potential, optimization of its structural and mechanical properties to mimic the perimysium connective tissue in the skeletal muscle is needed3.

We show that TSP contains a network of interconnected pores (Fig. 1f and Supplementary Video 1), with the majority of pore sizes proving suitable for cell adhesion, growth, spread and maturation (Fig. 1d)30. TSP’s porous structure also contains larger pores, which support nutritional and oxygen supply to the developing muscle tissue. The utilization of such pores for media perfusion for scale-up purposes is yet to be explored. In addition, the TSP sponge-like structure facilitates efficient cell seeding (Supplementary Fig. 4) and cell penetration into the thick scaffold samples (Supplementary Fig. 5) and its composition also facilitates cell adherence (Supplementary Fig. 3), which is crucial for scaling up production processes. TSP partially degraded during culture, which was beneficial to co-cultures with BSMCs, due to the ECM deposition, which maintained the scaffold’s mechanical properties and thus allowed higher cellular content. In contrast, in the BSC monoculture, scaffold degradation was less beneficial and caused a breakdown of the construct, as represented by its lower UTS compared with co-culture. This effect was noticeable during the differentiation phase and may have been a result of myotube contraction.

Muscle tissues contain a number of cell types with complex cell–cell interactions required for proper tissue development, including those of ECM-producing support cells and endothelial cells3. Therefore, BSCs were seeded with BSMCs, BECs or both. BSCs were used as myogenesis precursors. BSCs are formed at later stages of skeletal muscle development31 and undergo senescence in vitro32. Their main role is muscle repair31. Optimization of BSC cultures for scale-up purposes of CBM production was previously described32,33. However, earlier muscle stem cells that are active during foetal muscle development should also be considered as potential candidates34,35. BSMCs were employed here as ECM-producing supporting cells to provide additional support in microvascular development. Skeletal muscle-derived fibroblasts can serve a similar purpose, due to their central role in skeletal muscle ECM production and simple isolation and culture in vitro36. Co-culture with fibroblasts and myofibroblasts for CBM purposes were recently shown to enhance myogenesis progression in mouse models37. Fibroblast/adipogenic precursors could also be considered due to their additional potential to develop adipose tissues35.

The BSC myogenesis protocol includes proliferation and differentiation phases, which were optimized here using growth factors that affect myogenesis progression and BSC proliferation, and by removing non-edible components such as Matrigel and carcinogenic cytosine arabinoside. In the proliferation phase, seeded cells were allowed to adhere and proliferate until they reached high confluence, which facilitated cell fusion. The cells were then incubated in starvation (low serum) medium, which induced fusion into myotubes. IGF-1 promotes BSC proliferation and myotube formation; EGF contributes to BSC proliferation; and basic fibroblast growth factor (FGF) maintains BSC stemness3. LLM1 proliferation medium was optimized to improve cell coverage (which is crucial for myotube fusion) and to maintain BSC stemness (which is crucial for coordinated myotube formation). Minimization of the differentiation duration using starvation conditions was beneficial (data not shown). LLM2 differentiation medium was optimized for improved myotube size, shape and total coverage, and yielded elongated but sparse myotubes. While media formulation greatly improved myogenesis progression, an extensive list of growth factors was required, which exacerbated the cost of CBM production29. This may be overcome by using cost-effective growth factor sources, alternative induction of the growth factor signalling pathways, or conditioned medium from cells that produce these growth factors. In addition, serum alternatives will be required for CBM purposes4.

When cultured with BSMCs or in tri-culture, BSC differentiation improved, collagen type I disposition was increased (as indicated by the light blue staining in Masson’s trichrome38,39) and the scaffolds were less fragile and had higher weights compared with those embedded with BSCs only. In the case of co-culture with BSMCs, the TSP-1 scaffold weights were found to be higher than 100%, meaning the scaffold weight increased due to cell seeding compared with empty scaffold. Proteome analysis showed the contribution of BSMCs to both complex ECM deposition and skeletal muscle maturation. The proteome measurements tracked the two main biological processes that are the focus of this research: myogenesis and ECM deposition. Verbruggen et al.33 showed that it is possible to grow a large mass of BSCs on micro-carriers and to preserve the cells in their stemness stage, thereby enhancing their proliferation and differentiation potential to muscle cells32. Together with the development of adipose tissue40, these results might lead to a mass production scale-up of CBM.

Our mechanical measurements of TSP scaffolds seeded with BSCs and BSCs/BSMCs showed a Young’s modulus in the same range as that of native muscle, and the BSC/BSMC combination had the same UTS as native muscle. Moreover, the final product was tasted by volunteers, who reported that there was a significant difference between the TSP scaffolds with and without cells. TSP scaffolds with cells had a meaty flavour and a typical meat bite and texture. Taste studies examine the sensory attractiveness of a product and allow us to understand the sensory profile of the food.

This work provides evidence for the use of TSP as a novel CBM scaffold. The TSP structure showed pore sizes suitable for cell culture. Cells seeded on thick TSP samples attached and proliferated on the TSP surface. After optimizing the media composition for 3D scaffolds, BSCs covered a large proportion of the TSP scaffold, exhibited improved morphology and differentiated into elongated myotubes. Co-culture with BSMCs and tri-culture with BSMCs and BECs improved ECM deposition and myogenic differentiation, and provided evidence for complex ECM development. The results presented here represent the potential for CBM to be scaled up, forming new protein sources for human consumption. This would reduce our reliance on animal agriculture and contribute to more sustainable food security.

Methods

Isolation of BSCs

The BSCs and their growth and differentiation protocols were provided by M. R. Hathaway (Minnesota University)41. BSCs were isolated from the semi-membranosus muscle of a bovine carcase using pronase enzymatic digestion, then stored in liquid nitrogen41.

BSC-1/BSC-2 cells and bovine embryonic satellite cells (BSC-3) were isolated from the semi-membranosus muscle tissue of a 1-year-old Holstein Friesian cattle carcase and from the embryo of a Holstein Friesian cattle carcase, respectively. The resected tissues were soaked immediately in sterile phosphate-buffered saline (PBS) supplemented with 3% penicillin-streptomycin-nystatin solution (PSN; Biological Industries). The tissues were washed 3× with PBS/PSN, cleaned of residual fat and connective tissue and minced using sterile tweezer/scissors. For digestion, the tissues were transferred to a sterile 50-ml centrifuge tube (tube 1) containing 60 mg collagenase (type 1; Gibco) in 20 ml Dulbecco’s PBS (PBS −Ca/−Mg) and filtered through a 0.22-µm filter. The tube was placed in a water bath at 37 °C for 1.5 h and shaken vigorously every 10 min. Tube 1 was centrifuged at 200g for 10 min and the supernatant was transferred to a new 50-ml tube (tube 2). Pre-warmed 0.25% trypsin-EDTA was added to tube 1, which was placed in a water bath at 37 °C for 20 min. In parallel, tube 2 was centrifuged for 5 min at 300g, the supernatant was discarded and the cell pellet was resuspended in 10 ml BIO-AMF-2 medium (Biological Industries). Once the trypsinization was finished, tube 1 was centrifuged for 5 min at 300g, the supernatant was discarded and 5 ml BIO-AMF-2 medium was added. Tube 1 was centrifuged again for 5 min at 200g. The middle phase of tube 1 was collected and transferred into tube 2. Finally, the cell pool (tube 2) was centrifuged at 500g for 10 min to pellet all of the collected cells. The supernatant was discarded and the cell pellet was resuspended in 10 ml BIO-AMF-2 medium. The suspension was passed through a 40-µm strainer into 100-mm dishes and placed in an incubator at 37 °C under 5% CO2 for 2 h to remove fibroblasts. Then, the dish was rinsed gently with warmed fresh BIO-AMF-2 medium and the cells were seeded in a new 0.1% gelatin-coated plate.

Cell culture protocols

BSC passages 2–4 were sub-cultured in 0.1% w/v gelatin-coated flasks. The sub-culture medium included BSC growth medium (described below) supplemented with 50 µM ZnCl2 (Millipore; 108816), 10 ng ml−1 recombinant human leukaemia inhibitory factor (R&D Systems; 7734-LF), 62 ng ml−1 recombinant human heparin-binding epidermal growth factor-like growth factor (rhHB-EGF; R&D Systems; 259-HE), 100 ng ml−1 recombinant human IGF-1 (rhIGF-1; R&D Systems; 291-G1) and 10 ng ml−1 recombinant basic FGF (R&D Systems; 2099-FB). The BSC growth medium consisted of Dulbecco’s modified Eagle’s medium (DMEM)/HEPES (Gibco; 42430-025) and Ham’s F-10 Nutrient Mix (Gibco; 22390-025) at a 1:1 ratio, supplemented with 10% foetal bovine serum (FBS; Hyclone; SH30071.03), 1% MEM non-essential amino acids (Gibco; 11140-035), 1% GlutaMAX (Gibco; 35050-038) and 1% penicillin-streptomycin-amphotericin (antibiotics(Ab)/amphotericin(Am)) (Biological Industries; 03-033-1B).

Skeletal muscle microvascular BEC (AngioProteomie; cAP-b0009) passages 4–8 were sub-cultured in pre-coated flasks (quick coating solution (AngioProteomie; cAP-01)) in BEC medium, which included bovine endothelial growth medium (AngioProteomie; cAP-02) supplemented with 10% FBS.

Aortic BSMC (Cell Applications; B354-05) passages 4–8 were cultured in BSMC growth medium (Cell Applications; B311-500) according to the manufacturer’s recommendations.

Cells were maintained in a 5% CO2 humidified incubator at 37 °C. Culture medium was replaced (100%) every other day.

Two-dimensional myogenesis

BSCs were seeded in a 24-well plate at a density of 263 cells per mm, then cultured for 4 d in proliferation medium and 7 d in differentiation medium, unless stated otherwise. The control proliferation medium consisted of DMEM (Gibco; 41965-039) supplemented with 10% FBS and 1% Ab/Am. Optimized proliferation medium LLM1 comprised control proliferation medium supplemented with 50 µM ZnCl2, 100 ng ml−1 rhIGF-1, 62 ng ml−1 rhHB-EGF and 10 ng ml−1 recombinant basic FGF. The control differentiation medium included DMEM supplemented with 2% donor horse serum (Biological Industries; 04-004-1A) and 1% Ab/Am. Optimized differentiation medium LLM2 included control differentiation medium supplemented with 100 ng ml−1 rhIGF-1 and 62 ng ml−1 rhHB-EGF.

TSP scaffold preparation

TSP flakes (TSP-1 (Arcon-T-U-172; ADM; 158172), which contains 69% protein, 19% dietary fibres, 9% moisture and 3% fat) and TSP-2 (TVP-U-173; ADM; 165173), which contains 53% protein, 32% carbohydrates, 17% dietary fibres, 9% moisture and 3% fat) were used to generate the TSP scaffolds. TSP flakes were sterilized at Sorvan Radiation using medical-grade gamma treatment at 25–40 kGy for 3.5 h. One day before cell seeding, TSP flakes were soaked in ultra-pure water (BI; 01-866-1A) overnight at 37 °C, then cut into 6-mm cylinders with a height of 1–2 mm, using a biopsy punch. The scaffold weight was 40 ± 10 mg.

Synthetic scaffold fabrication

Porous PLLA (Polysciences) and PLGA scaffolds (Boehringer Ingelheim), with a pore size of 200–600 µm, were prepared using a salt-leaching technique. Briefly, 0.4 g of 200- to 600-µm NaCl2 particles were evenly distributed on an 18-mm-diameter cylindrical Teflon mould. A chloroform solution with 2.5% PLLA and 2.5% PLGA was prepared, 0.24 ml of which was placed on the salt particles. The chloroform was allowed to evaporate overnight, generating a solid residue composed of a mixture of PLLA, PLGA and salt particles. The construct was then washed for 8 h in distilled water to dissolve the salt. PLLA/PLGA scaffolds were cut into 6-mm cylinders using a biopsy punch and sterilized for 15 min in 70% ethanol. Scaffolds were then washed three times in PBS before cell seeding.

DiI staining

BSCs were stained with 5 µg ml−1 DiI (Thermo Fisher Scientific; D282), diluted in control proliferation medium, for 20 min at 37 °C. The DiI solution was then aspirated and the plate was washed twice for 10 min with control proliferation medium at 37 °C.

Cell seeding on TSP scaffolds

Four different cell combinations were suspended in 7 µl thrombin (20 NIH units per ml; Sigma–Aldrich; T1063) with the following cell counts and ratios per scaffold:

  1. (1)

    BSC monoculture: 7.5 × 105 BSCs

  2. (2)

    2:1 BSC/BSMC co-culture: 7.5 × 105 BSCs and 3.75 × 105 BSMCs

  3. (3)

    2:1 BSC/BEC co-culture: 7.5 × 105 BSCs and 3.75 × 105 BECs

  4. (4)

    2:1:1 tri-culture: 7.5 × 105 BSCs, 3.75 × 105 BSMCs and 3.75 × 105 BECs

Then, 7 µl fibrinogen (15 mg ml−1; Sigma–Aldrich; F3879) was added immediately before seeding the cells onto the scaffolds. The scaffolds were then incubated in a 5% CO2 humidified incubator at 37 °C for 30 min to allow for gelation of the fibrin. Culture medium was then added to the scaffolds and replaced (100%) every 2 d. Scaffolds were cultured for 7 d in LLM1 or LLM1/BEC-containing (1:1) medium and then for 4 or 7 d in LLM2 differentiation medium, as indicated.

Whole-mount staining

Scaffolds were fixed in 4% paraformaldehyde (Electron Microscopy Sciences; 15714), diluted in PBS for 20 min and subsequently washed three times in PBS for 5 min. The cell membrane was then permeabilized with 0.3% Triton X-100 (DeaJung; 8566-4400) and diluted in PBS for 10 min. The scaffolds were then washed with PBS. For actin staining, the scaffolds were then incubated with phalloidin-TRITC (1:100 diluted in PBS; Sigma–Aldrich) and DAPI (diluted 1:1,000 in PBS) for 20 min and washed three times with PBS and imaged. For the other stainings, the scaffolds were immersed in blocking serum (5% BSA in PBS) for at least 1 h at room temperature or 4 °C overnight. Scaffolds were then incubated with primary antibody in the concentrations described below (in 5% BSA in PBS) for 3 h at room temperature or overnight at 4 °C. The scaffolds were washed four times with PBS and incubated with secondary antibody (diluted in PBS) for 3 h. The scaffolds were then washed three times and stored at 4 °C until imaging, using the Zeiss LSM 700 inverted confocal microscope (Carl Zeiss) and the ZEN software (Carl Zeiss). Further image analysis was conducted using FIJI software.

Desmin staining was performed using goat α-desmin (1:50 diluted with 5% BSA; Santa Cruz Biotechnology) as a primary antibody and Alexa Fluor 488 donkey α-goat IgG (1:400; Invitrogen) as a secondary antibody. Myogenin staining was performed using mouse α-myogenin (1:200; Santa Cruz Biotechnology; sc-52903) as a primary antibody and Alexa Fluor 488 donkey α-mouse IgG (1:400; Molecular Probes) as a secondary antibody.

Immunohistochemical staining

Scaffolds were fixed in 10% neutral buffered formalin (Sigma–Aldrich) and embedded in paraffin using standard fixation and embedding procedures. The paraffin-embedded sections (5 µm) were then deparaffinized by immersion in 100% xylene and rehydrated by serial immersions in decreasing concentrations of ethanol. Standard protocols were used for haematoxylin and eosin and Masson’s trichrome staining of the paraffin-embedded sections. The stained slides were imaged using the Pannoramic MIDI automatic digital slide scanner (3DHISTECH). Images were analysed with the Pannoramic Viewer software (3DHISTECH).

BSC and myotube coverage analysis

Scaffolds were imaged using confocal microscopy with 5× magnification and 2 × 2 tile scan. Both the scaffold and the DiI fluorescence were imaged. Images were analysed using FIJI software. Maximum intensity projection was performed for each channel. For the scaffold channel, a constant threshold was maintained for all samples. Since the DiI fluorescence intensity depends on cell proliferation, images were randomized and a threshold was manually determined for each scaffold in a blinded manner. Then, the coverage was determined as the ratio between the area of the Dil-stained BSCs and the area of the scaffold.

Myotube coverage was quantified in a similar fashion, but with a constant threshold for all samples.

Average myotube size and shape complexity quantification

Desmin staining was quantified using the ‘analyse particles’ module, yielding the myotube area and perimeter. Myotube shape complexity (a gauge for myotube elongation) was calculated as: (myotube perimeter)2/(4π × myotube area).

Western blots

Western blots were performed to assess the expression of several proteins: collagen, myogenin and GAPDH (control). Total protein extracts of multicellular cultures on TSP scaffolds (in BSC, BSC/BSMC or BSC/BSMC/BEC culture) were prepared by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Proteins were then transferred to a nitrocellulose membrane (0.45 µm; Invitrogen). The blots were incubated in 7.5% non-fat dry milk in PBS containing 0.1% Tween-20 (PBS-T) for 1 h at room temperature, rinsed three times with PBS-T, incubated for 1 h at room temperature with primary antibodies (mouse anti-GAPDH (1:400; Santa Cruz Biotechnology), rabbit anti-collagen I (1:1,000; Abcam) or mouse anti-myogenin (1:200; Santa Cruz Biotechnology)), rinsed again three times with PBS-T and incubated for 1 h at room temperature with secondary antibodies (horseradish peroxidase-conjugated goat anti-mouse IgG (1:2,500) or horseradish peroxidase-conjugated donkey anti-rabbit IgG (1:2,500); both from Jackson ImmunoResearch). After four washes with PBS-T, the bound antibodies were visualized with a horseradish peroxidase chemiluminescence detection kit (EZ-ECL; Biological Industries). Protein quantification was performed by densitometric analysis (Multi Gauge version 3.0 software; FujiFilm). GAPDH served as an internal control.

RNA extraction

RNA was extracted using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. Briefly, scaffolds were washed with PBS and minced using scissors. Then, they were homogenized in a tube containing buffer RLT plus β-mercaptoethanol, after which they were centrifuged for 3 min at 14,100g before supernatants were extracted into a new 2-ml tube containing 70% ethanol at a 1:1 ratio. Samples were then transferred into a column and centrifuged for 15 s at 8,000g. The flow through was removed and the rest of the sample was reloaded and centrifuged for 15 s at 8,000g. RW1 buffer was added into the column, which was then centrifuged at 8,000g. RPE buffer was then added to the column, which was centrifuged for 15 s at 8,000g. This was repeated and the columns were centrifuged for 2 min at 8,000g, then for 1 min at maximum speed. DNase RNase-Free double distilled water (40 µl) was added to the column, which was centrifuged for 1 min at 8,000g, and the RNA concentration was measured using a NanoDrop (Thermo Fisher Scientific). Complementary DNA (cDNA) was then prepared using the High-Capacity cDNA Reverse Transcription Kit (ABI; 4374966), according to the manufacturer’s instructions.

qPCR

qPCR was performed using the RNA TaqMan Fast Universal PCR Master Mix (2×), with no AmpErase uracil N-glycosylase (ABI; 4352042). The following primers were used: MyoG (Bt03258929; Thermo Fisher Scientific), MYH (Bt03273061; Thermo Fisher Scientific), 18S (Hs03003631; Thermo Fisher Scientific), PAX3 (Bt04303789; Thermo Fisher Scientific) and MyoD (Bt04282788; Thermo Fisher Scientific). A total of 10 ng of cDNA was used for each reaction, according to the manufacturer’s instructions. The reaction was run in the 7300 ABI system (according to ABI 7300 fast program). Data were analysed using 7500 (software version 2.3). The results are presented as 2−∆ct and normalized to the 18S housekeeping gene.

Proteome analysis

PLLA/PLGA scaffolds were seeded with either 2.5 × 105 BSCs in monoculture or 2.5 × 105 BSCs and 1.25 × 105 BSMCs in co-culture, in 5 µl fibrin. Scaffolds were incubated for 7 d in LLM1 medium and for 4 d more in LLM2 medium. Scaffolds were then washed three times in PBS. Samples were sent to Technion’s Smoler Proteomics Center where they were digested with trypsin and analysed by liquid chromatography–tandem mass spectrometry on a Q-Exactive plus (Thermo Fisher Scientific), and their peptides were identified by MaxQuant software (Mathias Mann’s laboratory at the Max Planck Institute) against the bovine proteome part of the UniProt database and a decoy database. A t-test was performed on the normalized intensities (label-free quantification (LFQ)) of BSC monoculture and BSC/BSMC co-culture proteins, with at least two tandem mass spectrometry counts and one razor-positive unique peptide, which were not identified as possible contaminants. Proteins with a P value < 0.05 were defined as differentially expressed. Differentially expressed proteins with a log2[fold change] higher than 0.8 or lower than −0.8 were defined as having increased or decreased expression, respectively. Gene interaction plots were generated using the STRING database (string-db.org) and Gene Ontology annotation was obtained using DAVID Bioinformatics Resource 6.8 (https://david.ncifcrf.gov/tools.jsp).

SEM

TSP samples were freeze-fractured in liquid nitrogen, carbon-coated using a Polaron carbon coater (Quorum Technologies) and scanned using a high-resolution scanning electron microscope (Zeiss Ultra-Plus FEG SEM) in Technion’s Electron Microscopy Center (MIKA).

TSP pore size measurement

TSP-1 flakes were soaked in water overnight and then imaged with a Zeiss LSM 700 inverted confocal microscope (Carl Zeiss) equipped with a Neofluar 10×/0.30 M27 objective and ZEN software (Carl Zeiss), with the scaffold pore direction parallel to the microscope light trajectory. Laser intensities (405, 551 and 555 nm) were adjusted to obtain a clear contrast between the TSP walls and the hollow pores. Pore areas were quantified in FIJI.

Micro-CT

Micro-CT images of TSP-1, TSP-2 and PLLA/PLGA scaffolds were acquired using a SkyScan 1276 micro-CT (Bruker). To enhance the scaffold contrast, TSP-1 and TSP-2 flakes were soaked in 1% iodine solution for 3 d and then scanned while wet. TSP flakes were scanned using an Al 0.25-mm filter at 55 kV/72 μA with an exposure time of 1.1 s. Images were acquired at an isotropic resolution of 8 μm using the 360° step and shoot mode with a 0.3° rotation step and two averages. PLLA/PLGA scaffolds were imaged with no filter, 40 kV/100 μA and an exposure time of 500 ms. Images were acquired at an isotropic resolution of 8 μm using the 360° step and shoot mode with a 0.2° rotation step and three averages. Images were reconstructed using SkyScan NRecon software. Reconstruction settings included automatic misalignment compensation, level 10 ring artefact correction and 25% beam hardening artefact correction. The volume and surface area of the pores were quantified using the surface tool of the Imaris 9.0.2 software (Bitplane). A volume of interest was defined from the whole 3D image stack and grey levels were inverted to enable surface analysis.

Cell seeding efficiency

Scaffolds were seeded as described above. Cells that adhered to the bottom of the well or that were found in the medium were collected and counted. The seeding efficiency was calculated as the percentage of total seeded cells.

Mechanical properties

The mechanical properties of empty TSP scaffolds, TSP scaffolds seeded with BSCs only or TSP scaffolds seeded with BSC/BSMC co-culture were measured with a biodynamic test instrument (ElectroForce; Bose). Briefly, samples were stretched uniaxially until failure. The strain was calculated by deviation of the length from the initial length and stress was calculated by dividing the measured force by the scaffold’s cross-sectional area. The slope within the linear region of the stress–strain curve was determined as the Young’s modulus and the maximum stress in the plot was determined as the UTS. The calculated UTS and Young’s modulus were then compared with the mechanical properties of native bovine muscle taken from the literature42,43.

Statistical analysis

Statistical analysis was performed using computerized statistical programs. Factorial analysis was performed using JMP Statistical Discovery 14.1.0 (SAS Institute) and two-way ANOVA was performed using Prism 6 (GraphPad Software) with Tukey’s post-hoc multiple comparison. The data are presented as bar graphs with error bars representing standard deviation (s.d.).

Reporting Summary

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