Engineering Murine Adipocytes and Skeletal Muscle Cells in Meat-like Constructs Using Self-Assembled Layer-by-Layer Biofabrication: A Platform for Development of Cultivated Meat

Meat is a food staple that is consumed and enjoyed worldwide. Global meat consumption has grown by 58% in the past 20 years with population growth and increase in per person consumption accounting equally for that increase. Further economic development in Asia along with increased consumption in North America and Europe [Apostolidis and McLeay 2016; Bhat et al., 2019; Lynch and Pierrehumbert, 2019] is expected to increase demand for meat and meat products even more. Increase in livestock production to meet this demand is unsustainable due to high water consumption, greenhouse gas emission (15% of global greenhouse gas emissions and 37% of methane emissions) [Bhat et al., 2019], accelerated soil erosion, and pollution of waterbodies [Bhat et al. 2017]. This industry accounts for significant use of our resources including 70% of land suitable for agriculture (30% of total land) and almost one-third of agricultural water consumption [Stoll-Kleemann and O'Riordan, 2015]. Although solutions such as promoting low-meat diet style or switching to plant-based proteins has been suggested, their implementation has not been widespread [Graca et al., 2015; Macdiarmid et al., 2016] and additives incorporated in these foods are a source for concern. Cultivated meat is an environment-friendly and ethically appealing [Bhat et al., 2019] alternative to traditional meat. It has been estimated that once cultivated meat hits the market, it will result in up to 45% decrease in energy use, 78–96% lower greenhouse gas emissions, 99% lower land use, and 82–96% lower water consumption [Tuomisto and de Mattos, 2011]. Furthermore, issues with pathogenic contamination can be easily prevented, beneficial nutrients including minerals, vitamins, and fibers that normally do not exist in meat can be introduced, and production time can be reduced to weeks from years [Bhat et al., 2019].

Recently, tissue culture methods, developed originally for regenerative medicine and tissue engineering, have been repurposed for growing meat aggregates (minced meat type granules) in the lab as a way to address this environmental challenge. For instance, different materials such as gelatin [Benmeridja et al., 2020] and textured soy protein [Ben-Arye et al., 2020] have been used as scaffolding material for development of cultivated meat. Significant developments have also been made by startup companies such as Mosa Meat, Memphis Meat, Aleph Farms, and Super Meat [Post, 2017] in the development of scaffold-based cultivated meat. Nevertheless, research in this area is in its infancy [Dolgin, 2019], standards are nonexistent, and a number of challenges including ability to produce dense tissues that are thick and fibrous with the full complement of cell types present in animal tissues, at scale, have not been addressed [Allan et al., 2019]. The focus has so far been on growing meat aggregates as opposed to meat-like tissues due to the difficulties regarding coculture and differentiation of fat and muscle progenitor cells, where incorporation of mature fat cells is critical to derive the taste that is associated with meat, as well as problems in vascularization that is necessary for growing large tissue constructs [Post, 2017; Dolgin, 2019]. Such approaches include the use of scaffolds made of natural or synthetic biomaterials or even extracellular matrix (ECM) derived from explants of animal tissues in order to culture and sometimes grow only muscle cells [Van Eelen et al., 1999; Benjaminson et al., 2002; Vein, 2004; MacQueen et al., 2019]. These additives are not ideal as they do not achieve tissue-like cell densities, may be difficult to completely remove unwanted scaffolding materials after culture, and can introduce food safety concerns.

Here, we use a newly developed ECM-free biofabrication-based technique [Shahin-Shamsabadi and Selvaganapathy, 2020b] to form cell sheets as 2D building blocks and rapidly assemble them to thick slabs of meat-like tissue. These cell sheets are composed of differentiated and mature murine adipocyte and muscle cells as shown in Figure 1a. The process starts with the formation of individual cell sheets using progenitor cells that are partially differentiated and primed toward their final mature [Shahin-Shamsabadi and Selvaganapathy, 2020a]. Briefly, the progenitor cells (myoblasts and preadipocytes) are grown separately in low confluency and treated with their specific differentiation media for 2 days. Then they are typsinized before reaching confluency and completing their differentiation (i.e., before the myoblasts start to fuse and preadipocytes start to accumulate fat), and the appropriate ratio of the partially differentiated adipocyte and myoblasts (corresponding to the desired lipid content in the meat) are plated onto another tissue culture plate at high confluency where they will complete their differentiation process. It has been previously shown that trypsinization and replating does not affect the subsequent differentiation of the cells into mature adipocytes and skeletal muscle cells [Shahin-Shamsabadi and Selvaganapathy, 2020a]. Upon reaching full confluency, the skeletal muscle cells fuse with each other and form a contiguous sheet which is loosely attached to the surface and exerts a small traction force on the edges. Exposure of this confluent layer to a slightly acidic medium (medium A) leads to its rapid delamination from the culture plate and subsequent contraction to a stable sheet. The sheet is then exposed to slightly basic medium (medium B) to preserve its flat profile and then transferred into neutral medium (medium N) (Fig. 1a). Cells also produce their own ECM that is preserved in this process and provides additional structure and texture to the sheets. Finally, several individual sheets are assembled on top of one another whereupon they adhere forming a thick meat-like tissue. This layer-by-layer process can be used to form highly dense, multicellular, textured tissues that are difficult to produce in other ways. Using this process, a large number of sheets can be fabricated in normal culture plates without the need for a bioreactor which makes the process scalable, and only for fabrication of tissues with higher thicknesses incorporation of bioreactors is necessary.

Murine myoblasts (C2C12) and preadipocytes (3T3-L1) were used in this study to demonstrate the feasibility of this approach. Different protocols were devised and compared (Fig. 1b) in order to identify conditions that allowed formation of stable sheets as well as production of different amounts of protein and lipid in the constructs. Different ratios of initial cell numbers (C2C12 to 3T3-L1 ratios of 1:0, 1:1, and 1:3) were also used with each of these protocols to produce varying lipid contents in the tissues formed. In the first protocol (P1), these cells were cultured in muscle differentiation medium (M-DM) for 2 days and delamination was performed at day 3. In the second protocol (P2), cells were treated with fat differentiation and maintenance media (F-DM and F-MM), respectively, for 1 and 4 days before delamination was performed at day 5. In the last protocol (P3) after treating with F-DM and F-MM for a total of 5 days, cells were treated with M-DM for 2 more days, and then delamination was performed.

The microstructure of the cell sheets was studied by scanning electron microscopy (SEM) and light microscopy. SEM images for 1:0 and 1:3 ratios in P3 (Fig. 2a) show the difference between the morphology of cells in each group. In the 1:0 group where only C2C12 cells are present, all of the cells have elongated morphologies while in 1:3 group, 3T3-L1 cells that had accumulated lipid are round and bulky. Hematoxylin and eosin staining of the cross-section of the 1:0 group shows the highly dense and compact microstructure of the sheets (Fig. 2a inset) with thicknesses (∼50–100 μm) which are much higher than a single layer of cells in 2D culture. The cellular density and packing also increases significantly after delamination as the sheets shrink and become more compact. Fluorescent staining of the C2C12s with red and 3T3-L1s with green fluorescent dyes in a 1:3 ratio sample cultured using P3 shows the presence of both muscle fibers and lipid droplets after delamination which is important to achieve the meat texture and its taste (Fig. 2b). The addition of fat cells does not seem to affect the differentiation of the muscle cells and all the sheets delaminated properly. Nevertheless, the shrinking process does not damage the fibers or the accumulated fat globules (Fig. 2b). Several cell sheets can be stacked on top of each other and their adhesion results in formation of thicker structures that better resemble meat tissue. Total protein and lipid content of 2-layer stacked sheets with various ratios of C2C12 and 3T3-L1 cells were analyzed a day after stacking. Samples with highest number of initial 3T3-L1 cells and cultured using P3 protocol had the highest lipid and protein content as shown in Figure 2c. It was interesting to find that the protein and lipid content of the tissues can be controlled by changing the differentiation protocol as well as the ratio of initial cell numbers. Weight percentage of fat in different types of meat could range from 2 to 45% depending on the type of meat and its source. In case of beef, lean round can have as low as 6% fat while this amount for lean T-bone is 10% and could go up to 20% for regular beef [Gotto et al., 1984]. In order to compare these samples with actual meat, fresh extra lean beef meat was purchased from local market and samples with similar weights to 1:0 in P1 (∼5 mg) were separated (no fat tissue was observable with naked eye), and the same assays were performed on them. These meat samples had protein and lipid contents that were close to that of 1:0 in P1 sheets (1.5 and 0.9 times, respectively). The lower protein content of sheets could be due to the fact that cells secrete less ECM once cultured in 2D and proper stimulation could increase the protein content of the sheets. Figure 2e shows the increase of lipid content in 1:1 and 1:3 groups compared to 1:0 group in each protocol. Using only M-DM (P1), an ∼18% increase was observed for both groups. With F-DM (P2), an ∼15 and ∼35% increase was observed for 1:1 and 1:3 groups, respectively. With P3, these increases were ∼5 and ∼20%. Based on these results, it can be shown that lipid content of samples is completely tunable (increase by 5–35%) compared to samples with only muscle cells. It should be noted that the lipid measurement is from both cell membrane of skeletal muscle and adipocyte cells as well as lipid droplets accumulated in adipocytes. The measurement from muscle-only cell sheet represents the lipid content of bilipid membrane of the cells and their organelles while the increases in coculture with adipocytes can be attributed mostly to the stored lipids in adipocytes. From the perspective of manufacturing, the ability to rapidly produce high protein and lipid contents as well as the ease of delamination of single sheets to assemble them further is of importance. Different protocols used in the final growth step and the ratio of cells used in the culture has a significant effect on all of these parameters as shown in Figure 2d. When M-DM (P1) was used, higher adipocyte numbers made it more difficult to delaminate the sheets while in the other 2 protocols, the opposite was observed. The increase in adipocytes led to an increase in lipid content, and the lipid droplets produced were robust and did not burst upon delamination. These lipid droplets do not exist in the muscle-only sheets (1:0) while are clearly abundant and are preserved after delamination and digestion of 1:3 groups of P3 (Fig. 2f).

This manufacturing process is scalable and can be used to make large meat-like tissues in 24-well plates, 6-well plates, or 10 cm dishes (Fig. 3). The ability to form thick and meat-like structures was shown by forming sheets in 6-well plates (1:0 ratio and P1) and stacking 18 layers of them. A multistep assembly process was used wherein 3 constructs, each containing 6 individual cell sheets, were formed with 1 h incubation, and later these 3 constructs were assembled on top of each other and incubated further for 24 h (Fig. 3b) (thickness of 1–2 mm can be achieved using this number of layers). Scalability of the technique to form larger structures was further shown by making the cell sheets in 10 cm dishes (1:0 ratio and P1) (Fig. 3c). Individual cell sheets show a significant shrinkage immediately after delamination (from 10 to 3.5 cm in case of 10 cm dishes) which helps with creating thicker and stable structures. Once assembled into multilayer constructs, the amount of shrinkage is dramatically lower even over longer incubation times (a 6-layer stack shrank from 3.5 to 2.5 cm after 24 h of incubation) which is important to create stable structures.

One of the defining features of the cell sheet approach is that it does not require exogenous ECM, and the cells produce their own. In comparison, other approaches for cultivated meat such as the use of bioprinting typically require scaffolds or additional ECM to be added in the bioink that typically is from other animal sources. For instance, bioprinted skeletal muscle [Kim et al., 2018] and adipose [Benmeridja et al., 2020] tissues have been created but require exogenous ECM such as fibrinogen or gelatin methacrylate. In addition, bioprinted constructs can also have low cell densities that makes them inadequate to recreate the natural structure of meat with its high cell density and achieve similar mechanical properties without long durations of growth. Such long durations of growth require some form of perfusion of the tissues which complicates the fabrication process. Implementation of bioreactors for bioprinted constructs to apply different electrical and mechanical stimuli to mimic the effect of exercise that is necessary for structure and taste of the meat also needs to be considered. Despite these shortcomings, the key advantage of bioprinting is its ability to pattern the composition spatially. This ability can be used to create a marbling-like compositional variation that would be of use in cultivated meat applications.

In contrast, layer-by-layer assembly of sheet-like tissue constructs is a scalable approach that can be used to build meat-like tissues of any size and thickness. It can avoid the use of scaffolds that are required in other cultivated meat technologies as the cultured cells themselves secrete ECM that is preserved and makes robust sheets. The growth and construction of these as individual sheets and their subsequent parallel assembly into thicker structures elegantly avoids mass transfer-induced growth limitations and the need for integrated vasculature. Finally, the complex ECM produced in these sheets includes biomolecules such as E-cadherin and Laminin 5 [Yamato et al., 2001] which then lend adhesive properties to the individual sheets and enables them to be stacked to form continuous thick constructs [Iwata et al., 2011]. Unlike current technologies that focus on production of minced meat [Dolgin, 2019], the constructs produced here better resemble natural structure of meat and thicker pieces of lab grown meat with natural texture can be produced. The process described in this paper for manufacture of meat-like tissues is scalable and amenable for automation, which is an important criterion for economic production [Allan et al., 2019]. It does not require any specialized equipment and is the first demonstration, to our knowledge, of incorporating both adipocytes and skeletal muscle cells in meat without the use of any other exogenous scaffolding material or ECM.

Despite its promises, the use of cell sheets for cultivated meat is new, and several advancements are required to translate it commercially. Although this process produces robust individual sheets that can handle significant mechanical forces during pipetting from one vessel to another or handling with tweezers, larger sheets (10 cm or more in diameter) which are anticipated for commercial production will require automated handling of some form. Furthermore, techniques for proper alignment of the sheets during the assembly process needs to be developed as well. Suitable bioreactors to recreate effect of exercise after assembly can also be implemented to further improve the properties such as texture and taste of the product. Such bioreactors and the external stimuli can also promote further maturation of the cells and strengthen interconnectivity between the sheets.

Here, as a proof of concept for the platform, murine cells were used to develop meat-like constructs, but the technique is compatible with any fusing cells and can potentially be used with similar cell types from other animal sources such as cows, pigs and chicken for human consumption. Nevertheless, other advances are required to transform it into a viable product. Of particular importance is the creation of stable cell sources. Primary animal cells including satellite cells (muscle stem cells), mesenchymal, embryonic, or induced pluripotent stem cells from human food-relevant sources can be considered for the muscle progenitor cells, while adipose tissue-derived stem cells or mesenchymal stem cells can be considered for the fat progenitor cells. Cell lines for both of these 2 main cell types in meat as well as other cell types, including endothelial cells comprising blood vessels and cell types forming the connective tissue, can be created through genetic or chemical induction and are probably a better option to reduce variability and contamination risks associated with primary cell harvest [Specht et al., 2018]. It is also suggested that by establishing cell lines that are less dependent on exogenous growth factors, cost of the media and dependency on growth factors could be decreased [Sinacore et al., 2000]. In the design and development of media to be used with the technique introduced here, growth and differentiation of preadipocytes to adipocytes and myoblasts to myofibers and myotubes in coculture should be considered. Serum components are not compatible with food safety standards, and they can be a source of contamination. The dependence of the process on such components makes it difficult to bring it to market [Sinacore et al., 2000], and therefore the development of suitable animal- and xeno-free media that are low cost and can be produced at high volume should be considered. Electrical and mechanical stimuli can also be provided to the assembling tissue to mimic the highly complex structure of the native tissue and accelerate its growth and maturation [Rangarajan et al., 2014; Maleiner et al., 2018]. Visual appearance of the tissues formed can be adjusted from its current yellowish-pink tinge which is due to the absence of blood and reduced myoglobin production in the culture condition with high oxygen levels [Post and Hocquette, 2017] to a natural pinkish tone by culturing cells in low oxygen conditions [Kanatous and Mammen, 2010] or by using plant-based heme or addition of extra iron [Bhat et al., 2019]. Once the improvements suggested here are implemented, the technique introduced here will be a viable approach for scalable production of meat tissues with natural textures and complexities.

C2C12 and 3T3-L1 cells were cultured in their growth medium up to 80% confluent. For partial differentiation, culture started with confluency of 40–60% in each cells’ growth medium which was then switched to the differentiation medium. Cells were kept in this medium for 2 days before they were trypsinized and used for sheet formation. A total cell number of 0.235 × 10⁶ was used in 24-well plates, 1.15 × 10⁶ cells in 6-well plates, and 7 × 10⁶ cells in 10 cm dishes. C2C12s were cultured in high glucose DMEM supplemented with 10 v/v% heat inactivated (HI) fetal bovine serum (FBS). In their differentiation medium HI-FBS was replaced with 2% horse serum and 1% Insulin-Transferrin-Selenium solution (Gibco, 100X, 41400045). 3T3-L1 cells were cultured in low glucose DMEM with 10 v/v% FBS. Their differentiation medium was high glucose DMEM with 10 v/v% FBS, 1 mg/mL insulin, 0.5 mM IBMX, 0.25 μM Dexamethasone, and 2 µM Rosiglitazone while their maintenance medium was high glucose DMEM with 10 v/v% FBS and 1 mg/mL insulin. All media had 1% penicillin-streptomycin (Gibco, 10,000 U/mL, 15140148).

Partially differentiated cells were replated at specified numbers for each well size and maturated using any of the 3 defined protocols (P1, P2, or P3). Delamination process started by treating cells with acidic medium (Medium A, differentiation medium containing 0.1% V/V acetic acid, final pH of ∼6). This starts the delamination of sheets which can be accelerated by gentle shaking of the well plate or by applying shear force using gentle pipetting of medium to the edges of the wells. Once delamination was completed, acidic medium was replaced with basic medium (Medium B, DM with 0.1% V/V 0.1 M sodium hydroxide in deionized water, final pH of ∼8) in order to prevent samples from forming clumps. After 5 min of treatment with medium B, samples were transferred to neutral medium (Medium N, differentiation medium, pH of 7.4).

Cells were stained with long-term fluorescent trackers – C2C12 with red DiI and 3T3-L1 with green DiO following the supplier’s protocol (Thermofisher, catalog numbers D282 and D275, respectively) – before being partially differentiated. Fluorescent images were taken using an inverted fluorescent microscope (Olympus, USA) using FITC (475-485/485-536 nm) and TXRED (542-582/582-644 nm) filters. For SEM purposes, samples were fixed with 2% formaldehyde for 30 min and were critically point dried (Leica Microsystems, Wetzlar, Germany), cut in half, coated with gold, and imaged in cross-section using TESCAN VP. SEM at 10 kV. For histological staining, fixed samples were dehydrated step wise in 40, 60, and 80% ethanol in water, and after paraffin embedding and sectioning, staining was done with hematoxylin and eosin and images were taken using an inverted microscope.

Sheets were digested using 500 μL of 2 mg/mL collagenase/dispase (Sigma-Aldrich, catalog number 10269638001) in PBS for 2 h. Lipid content was measured using two 25 μL aliquots of this digest solution in 96 black well plate by addition of 200 μL 0.2% V/V Nile red dye (10 mg/mL in acetone, Thermofisher, catalog number N-1142) in PBS, following a 15 min incubation, the fluorescent intensity was measured at excitation/emission of 560/640 nm using a platereader (Infinite® M200, Tecan, Männedorf, Switzerland). In parallel, 100 μL of digest solution was lysed using 100 μL of lysis solution (0.5% Triton X-100 in PBS). After a 15 min incubation with lysis solution, two 25 μL aliquots were transferred to new 96-well plates and 200 μL of Pierce^TM BCA Protein assay (Thermofisher, catalog number 23227) kit solution (50:1 ratio mixture of solutions A and B of the kit) was added to each well. Absorbance was measured at 562 nm after 30 min incubation. A total of 5 samples were used for each assay (n = 5).

In vitro studies were performed at the Biointerfaces Institute at McMaster University. C2C12 and 3T3-L1 cells were kindly provided by Dr. Sandeep Raha from the Department of Pediatrics at McMaster University.

Ethical approval is not required for this type of research.

The authors have no conflicts of interest to declare.

This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canadian Institute for Health Research (CIHR). P.R.S. wishes to acknowledge support from the Canada Research Chairs Program.

Alireza Shahin-Shamsabadi and P.Ravi Selvaganapathy planned the research and experimental methodology. Alireza Shahin-Shamsabadi conducted the experiments, obtained and organized the data, and wrote the draft manuscript. Alireza Shahin-Shamsabadi and P.Ravi Selvaganapathy reviewed the data and edited the manuscript.