Functional regeneration of tissue engineered skeletal muscle in vitro is dependent on the inclusion of basement membrane proteins

Abstract Skeletal muscle has a high regenerative capacity, injuries trigger a regenerative program which restores tissue function to a level indistinguishable to the pre‐injury state. However, in some cases where significant trauma occurs, such as injuries seen in military populations, the regenerative process is overwhelmed and cannot restore full function. Limited clinical interventions exist which can be used to promote regeneration and prevent the formation of non‐regenerative defects following severe skeletal muscle trauma. Robust and reproducible techniques for modelling complex tissue responses are essential to promote the discovery of effective clinical interventions. Tissue engineering has been highlighted as an alternative method, allowing the generation of three‐dimensional in vivo like tissues without laboratory animals. Reducing the requirement for animal models promotes rapid screening of potential clinical interventions, as these models are more easily manipulated, genetically and pharmacologically, and reduce the associated cost and complexity, whilst increasing access to models for laboratories without animal facilities. In this study, an in vitro chemical injury using barium chloride is validated using the C2C12 myoblast cell line, and is shown to selectively remove multinucleated myotubes, whilst retaining a regenerative mononuclear cell population. Monolayer cultures showed limited regenerative capacity, with basement membrane supplementation or extended regenerative time incapable of improving the regenerative response. Conversely tissue engineered skeletal muscles, supplemented with basement membrane proteins, showed full functional regeneration, and a broader in vivo like inflammatory response. This work outlines a freely available and open access methodology to produce a cell line‐based tissue engineered model of skeletal muscle regeneration.

Mature adult skeletal muscle consists of aligned myofibres, encased within a basement membrane and a predominantly type I collagen extracellular matrix (ECM). These multinucleated cells contain the contractile apparatus of skeletal muscle, with the force generated transmitted through the ECM to the tendons (Cooke, 2004;Szent-Györgyi, 2004). Located between the membrane of these multinucleated cells (sarcolemma) and the basement membrane resides a population of stem cells, termed satellite cells, which are required for muscle hypertrophy and regeneration (Hurme & Kalimo, 1992;Kuang, Kuroda, Le Grand, & Rudnicki, 2007;Mauro, 1961;Moss & Leblond, 1971). Following injury, satellite cells activated by the post injury environment proliferate rapidly producing large numbers of progeny (Doumit, Cook, & Merkel, 1993;Haugk, Roeder, Garber, & Schelling, 1995;Hurme & Kalimo, 1992), which express early markers of myogenic commitment (Kuang et al., 2007). It is these progeny which commit to the myogenic lineage fusing to regenerate the damaged myofibres and return full functionality (Cornelison et al., 2004;Seale et al., 2000). A number of studies have shown the importance of inflammatory mediators on muscle progenitor proliferation and differentiation, and so the role of inflammation in regeneration is seen as an essential coordinating event following injury (Arnold et al., 2007;Lu et al., 2010;Segawa et al., 2008;Summan et al., 2006).
Currently, limited pharmacological interventions are available, in clinical practice, to increase regenerative repair within injured and/or diseased populations. As such, individuals that fall into these groups are left with skeletal muscle which never recovers full function, and in many cases can cause long term pain and disability. The processes governing skeletal muscle regeneration are highly complex, relying upon the coordinated effects of multiple cell types in vivo. To date, studies of muscle regeneration have relied upon the use of animal models. However, attempts to reduce the reliance upon animals models, as per the 3Rs directive (EU, 2010;Russell, Burch, & Hume, 1959), in addition to difficulties translating animal data to human physiology (Boldrin, Muntoni, & Morgan, 2010;Gerry & Leake, 2014;Shanks, Greek, & Greek, 2009), demonstrates a requirement for new approaches. Recreating complex biological processes, such as skeletal muscle regeneration in in vitro tissue culture systems, requires sophisticated models of tissues which go beyond simple monolayer cultures.
Examination of cellular responses typically relies upon monolayer cell culture methods; however, these monolayer culture methods lack the advanced tissue hierarchy seen in vivo. The lack of spatial organisation in monolayer models may limit the capacity of cellular models to undergo complex physiological events, such as muscle regeneration.
Tissue engineering has been highlighted as an alternative method, allowing the generation of three-dimensional in vivo like tissues without the need for laboratory animals. Reducing the requirement for animal models promotes rapid screening of potential clinical interventions, as these models are easily manipulated genetically and pharmacologically and therefore are associated with reduced cost and complexity.
The culture system employed is based upon a bespoke 3D culture mould and uses primary rat myogenic precursor cells (requiring the sacrifice of small laboratory animals), negating some of the advantages associated with using engineered tissues as models. Therefore, a system that employs a freely available and open source 3D printed mould, to allow other investigators to accurately and rapidly reproduce the system, and does not use any primary animal tissue, represents a significant advance to the tissue engineering field.
Here, we present a comparative analysis of injury and regeneration in both monolayer and tissue engineered 3D culture systems using the murine skeletal myoblast cell line C2C12. Morphological analysis, gene expression and functional output are used to assess the differences and similarities between monolayer and 3D model systems and identify a system which most closely mimics in vivo skeletal muscle regeneration.

| Monolayer (2D) culture
Cells were plated into six well plates containing 0.2% gelatin coated glass cover slips at a density of 1 × 10 4 cells/cm 2 . To examine the effect of surface matrix, experimental plates and coverslips were coated with a 1.5 mg/mL Matrigel ® solution as per manufacturer's instructions (Corning, UK). Myoblasts were cultured to confluence in GM before being cultured in low serum differentiation medium (DM); composed of 97% DMEM, 2% Horse Serum (HS, Sigma Aldrich, UK) and 1% P/S. The total culture period was standardised at 2 days GM, followed by a further 3 days DM.

| 3D tissue engineered constructs
Collagen constructs were generated using C2C12 myoblasts, as previously published . Type I collagen only hydrogels were formed by the addition of 85% vol/vol type I rat tail collagen (First Link, UK; dissolved in 0.1 M acetic acid, protein at 2.035 mg/mL), with 10% vol/vol of 10X minimal essential medium (MEM, Gibco, UK). This solution was neutralised by the addition of 5 M and then 1 M sodium hydroxide (NaOH) dropwise, until a colour change to cirrus pink was observed. Collagen/Matrigel ® constructs were generated by the addition of 65% vol/vol type I rat tail collagen, with 10% vol/vol of 10X minimal essential medium.
This solution was neutralised as above. This was followed by the addition of 20% vol/vol Matrigel ® (Corning ® , Germany). Myoblasts were added to the neutralised collagen or collagen/Matrigel ® solution at a density of 4 × 10 6 cells/mL in a 5% vol/vol GM solution, before being transferred to the pre-sterilised biocompatible polylactic acid (PLA) 3D printed inserts (Rimington, Capel, Christie, & Lewis, 2017) to set for 10-15 min in an incubator. Collagen only gels were set in 500 μL inserts (Rimington et al., 2018a), whilst Matrigel ® /Collagen gels were set in 50 μL inserts (Rimington et al., 2018b(Rimington et al., , 2018c. All moulds used in this manuscript are freely available to download at the following URL: https://figshare.com/ projects/3D_Printed_Tissue_Engineering_Scaffolds/36494. GM was added for 4 days and changed daily, before being changed to DM, refreshed every 2 days, for a further 10 days in culture. Figure S5 contains a cross sectional image and macroscopic image of deformed hydrogels to illustrate the morphology and appearance of mature control constructs.

| Barium chloride injury and regeneration
Barium chloride (BaCl 2 ) was chosen as an injurious stimulus due to previous in vivo publications and its high water solubility, allowing easy and reproducible in vitro application (Hardy et al., 2016;Mueller et al., 2016). Once cultures had reached maturity, as defined above, they were exposed to chemical injury by BaCl 2 . Prior to inducing injury, fresh DM was added to all conditions. Precisely 50 μl/mL of 12% wt/wt BaCl 2 solution was then added to the medium for injury culture conditions, followed by a 6 hr incubation to cause injury.

| Image collection and analysis
Images were captured using a Leica DM2500 (monolayer) or a Zeiss LSM average myotube width were all conducted manually. Total nuclei were calculated using an in-house macro implemented in ImageJ (Schindelin et al., 2012). Analysis was conducted from nine monolayer images taken across three coverslips per biological repeat, or from a 21-image tile scan of a 3D construct for every condition, derived from n ≥ 3 biological repeats.

| Assessment of muscle function by electrical stimulation
Electric field stimulation was used to assess the functional capacity (force generation) of tissue engineered constructs. Constructs were washed twice in PBS, and one end of the construct removed from the supporting mould pin. The loose end of the construct was then attached to the force transducer (403A Aurora force transducer, Aurora Scientific, Canada) using the eyelet present in the construct.
The construct was positioned to ensure its length was equal to that before removal from the pin and covered (3 mL RT-PCR procedure was: 50 C, 10 min (for cDNA synthesis), 95 C, 5 min (reverse transcriptase inactivation), followed by 40 cycles of 95 C, 10 s (denaturation), 60 C, 30 s (annealing/extension). Melt analysis was then carried out using standard ViiA protocol. Relative gene expressions were calculated using the comparative CT ( ΔΔ CT) method giving normalised expression ratios (Schmittgen & Livak, 2008). RPIIβ was the designated housekeeping gene in all RT-PCR assays and no sample controls for each primer set were included on every plate. damage to the mononuclear cell population. The chemical BaCl 2 was selected, demonstrating a dose dependent ability to specifically remove myotubes ( Figure S1). Furthermore, BaC1 2 has been documented to induce injury in animal models of skeletal muscle regeneration (Hardy et al., 2016).

| Statistical analysis
BaCl 2 insult in monolayer cultures caused total removal of myotubes, demonstrated by the significant ablation of fused nuclei (p < .001), however no reduction in total nuclei number indicates the retention of mononuclear cells following injury (Figure 1a-c). Following a further 2 days culture in GM, nuclei number was significantly F I G U R E 1 Treatment of differentiated C2C12 cultures with BaCl 2 specifically removes myotubes from culture and initiates a regenerative response. (a) ×20 widefield micrographs of recovery time points in C2C12s. Stained for actin (phalloidin, red) and nuclei (DAPI, blue), scale bars represent 50 μm. (b) Fusion index, (c) total nuclei, (d,e) measures of myotube maturity. Values for 0 hr and 2 days post injury have been omitted as too few myotubes were present in these conditions to accurately measure these variables (b-e) Graphs express mean ± SD, asterisks above bars denotes significance from control, ***denotes significance p < .001, (f-i) RT-PCR analysis of cellular developmental and inflammatory markers. All graphs display mean ± SD, *denotes significance p < .05, **denotes significance p < .01, ***denotes significance p < .001. Following insult, significant reductions in myotube density were observed (p < .001, Figure 2a,b), accompanied by decreased total nuclei (Figure 2c). During the regenerative period, recovery of myotube density was completely inhibited (Figure 2a), with reductions in nuclei number also evident across time (Figure 2c). This data suggests that hydrogels composed solely of type I collagen do not support regeneration following injury, and so are not a viable ex vivo model of skeletal muscle regeneration.

| Inclusion of basement membrane proteins in the form of Matrigel ® is sufficient to support regeneration in 3D
The potential lack of a stem cell like niche between the sarcolemma and basement membrane in hydrogels consisting of type I collagen only, could explain the inability to regenerate. To confirm the requirement for a basement membrane, hydrogels containing the basement membrane supplement Matrigel ® were injured and allowed to regenerate. Hydrogels composed of collagen/Matrigel ® were cultured in 50 μl inserts compared to 500 μl for collagen only hydrogels, to increase experimental throughput. The comparison of the mould sizes has been made previously and shown to be consistent .
Immediately following injury in collagen/Matrigel ® hydrogels, myotube density and coverage were significantly reduced (p < .001, Although not statistically significant, force generation was enhanced at End DM. Twitch force outputs were on average 2.5-fold higher and tetanus 2.8-fold higher than control. Figure 3f shows representative force traces at two frequencies for control gels, no difference in force trace shape was observed during recovery. To examine the molecular mechanisms of regeneration, gene expression analysis was carried out on myogenic (Figure 4a (adipogenic) and Runx2 (osteogenic) were also quantified, although no significant variation from control expression was observed (Figure 4c).
To ensure that the addition of basement membrane components and growth factors contained within Matrigel ® was not the sole cause of regeneration seen in tissue engineered constructs, further 2D experiments using Matrigel ® coated coverslips were carried out. BaCl 2 injury caused a significant reduction in fusion index (p < .001), that despite recovering significantly (p = .016), remained significantly (p < .001) below control at experimental termination ( Figure S3). A proliferative phase was evident at End GM time-point ( Figure S3), in addition to consistent myotube widths pre and post injury. Significant reductions in nuclei per myotube were observed (p = .031); however, this was the only variation from the trends observed with gelatin as a substrate. Increased fusion index at control and following injury could potentially affect the regenerative process; however, correlation analysis shows no relationship between fusion, before or after injury, and the level of regeneration ( Figure S4). This data confirmed the requirement for a 3D environment for full regeneration.

| DISCUSSION
In this study, we present a model of skeletal muscle regeneration which requires no laboratory animal sacrifice or donor tissue and is designed to be straightforward to replicate between laboratories. In addition, we have made a direct comparison of monolayer to 3D F I G U R E 2 Type I collagen hydrogels lack regenerative capacity following injury. This blockade is most likely preventing the fusion of myoblasts into myotubes (Bentzinger, Wang, & Rudnicki, 2012). In addition to the potential myogenic blockade an increase in the non-myogenic adipogenic transcription factor Pparg (Memon et al., 2000;Siersbaek, Nielsen, & Mandrup, 2010) (Gorza, Sartore, Triban, & Schiaffino, 1983;Hindi & Kumar, 2016;Schiaffino, Rossi, Smerdu, Leinwand, & Reggiani, 2015).
This response to switch to embryonic myosin may protect remaining myotubes from further injury and perhaps increase the ability of these myotubes to regenerate and support myogenesis. Throughout regen- Collagen I based tissue engineered skeletal muscles showed no regenerative capacity; however, the addition of Matrigel ® was sufficient to allow regeneration following injury, showing the importance of basement membrane proteins in regenerative processes. Furthermore, the 3D organisation of these proteins must be a requirement for regeneration as basement membrane supplementation in monolayer was shown to be insufficient to support full regeneration. The presence of a niche with 3D organisation is key for defining stem cell proliferation, lineage commitment and self-renewal in in vivo muscle and the 3D environment supplied by tissue engineered muscles is able to reproduce this in a way that is impossible in monolayer culture. The presence of this regenerative niche within collagen/Matrigel ® engineered muscles allows potential future work to examine how different injury mechanisms, which may disrupt this niche, affect regeneration.
The full morphological regeneration of myotube number and full functional regeneration matches the response to BaCl 2 insult in in vivo experiments (Hardy et al., 2016). The ability to measure functional output, something not possible in monolayer cultures, is a major advantage of 3D systems allowing direct and rapid assessment of tissue function which is the primary clinical measure of recovery from injury. The inflammatory response of 3D cultures was also shown to be more biomimetic than monolayer. Both Il6 and Mcp1 showed increased expression, compared to Il6 alone in monolayer, this increased number of inflammatory cytokines adds complexity to the regenerative environment post injury. As immune cells play an important role in directing and regulating skeletal muscle regeneration an inflammatory response as close to in vivo as possible will be required to understand the interaction between muscle and immune cells. As such we see the tissue engineered model presented as superior to the monolayer equivalent as it not only provides a fully regenerative response and functional output but expresses a fuller inflammatory response to injury important for the recruitment and activation of immune cells which support regeneration in vivo.
This study, which compares directly the use of the same injury mode in both monolayer and 3D cultures, demonstrates the usefulness of tissue engineered models for disease, and highlights where these models can be used to produce results superior to monolayer culture. We also recognise that monolayer culture is still widely used for preclinical interventions and have shown that a monolayer model of skeletal muscle regeneration is perhaps of use for some applications which require only limited biological relevance. Previous work demonstrating regeneration in tissue engineered muscles relies upon isolation of myogenic precursors from rat tissue (Juhas et al., 2014(Juhas et al., , 2018, whereas this study presents a cell line based model contained within a 3D printed mould making it suitable from high throughput work often required in therapeutic development. We show that regeneration of tissue engineered skeletal muscles containing basement membrane proteins provides a system which is representative of the in vivo response to chemical injury, providing a platform for the investigation of the myogenic events which underpin regeneration.
This model can be used as a tool for researchers to obtain in vivo like results without requiring laboratory animals, improving the relevance of preclinical screening and potentially reducing failure rates of novel clinical interventions.