Modelling Skeletal Muscle Ageing and Repair In Vitro

Healthy skeletal muscle can regenerate after ischaemic, mechanical, or toxin-induced injury, but ageing impairs that regeneration potential. Tis has been largely attributed to dysfunctional satellite cells and reduced myogenic capacity. Understanding which signalling pathways are associated with reduced myogenesis and impaired muscle regeneration can provide valuable information about the mechanisms driving muscle ageing and prompt the development of new therapies. To investigate this, we developed a high-throughput in vitro model to assess muscle regeneration in chemically injured C2C12 and human myotube-derived young and aged myoblast cultures. We observed a reduced regeneration capacity of aged cells, as indicated by an attenuated recovery towards preinjury myotube size and myogenic fusion index at the end of the regeneration period, in comparison with younger muscle cells that were fully recovered. RNA-sequencing data showed signifcant enrichment of KEGG signalling pathways, PI3K-Akt, and downregulation of GO processes associated with muscle development, diferentiation, and contraction in aged but not in young muscle cells. Data presented here suggest that repair in response to in vitro injury is impaired in aged vs. young muscle cells. Our study establishes a framework that enables further understanding of the factors underlying impaired muscle regeneration in older age.


Introduction
Skeletal muscle regeneration is a complex and fnely regulated biological process that shares molecular and cellular aspects with embryonic development [1][2][3]. Tis homeostatic process is made possible by the coordinated cooperation between diferent cell populations, such as satellite cells (mitotically quiescent stem cell population residing between the sarcolemma and basal lamina), fbroblasts, infammatory cells, and the myofber microenvironment, all decisive to the adequate regeneration of the muscle tissue [4][5][6][7]. Te role of satellite cells in muscle regeneration has been extensively studied. Upon injury, they exit their quiescent and initiate proliferation. Tis is mediated by a rapid activation of the Notch signalling pathway, upregulation of cyclin D1, and suppression of the TGF-β-Smad3 signalling pathway [8]. After generating enough progeny, the cells exit the proliferative state to initiate diferentiation where they fuse into the injury site [9].
Although muscle exhibits a robust regeneration capacity, this becomes impaired with ageing contributing to the decline of musculoskeletal health, which, in turn, is a contributory factor in the development of sarcopenia [10]. Tis impairment is partially due to the limited proliferative capacity of satellite cells on account of Notch and Cyclin D1 downregulation and a suboptimal myofber niche created by chronic infammation-a hallmark of the ageing muscle [11][12][13].
It is vital to understand the muscle regenerative response in the context of ageing since it has been suggested that an accumulation of contraction-induced microdamage that is not adequately repaired may be one of the causes of muscle wasting, sarcopenia, and anabolic resistance typically observed in older people [14]. Yet, there exists no in vitro human model to enable such investigations. Terefore, the aim of this study was to develop an in vitro human muscle regeneration model and characterise the changes in morphology, cell cycle, and mRNA transcription during recovery from damage in young and aged muscle.
To generate this model, we derived myotubes (diferentiated muscle cells) from myoblasts isolated from young and older individuals and provoked an injury using the muscle toxin barium chloride (BaCl 2 ) that is commonly used to induce injury to examine muscle repair and regeneration [15]. BaCl 2 destroys the cytoskeleton without abrogating the cell's regeneration capacity [16][17][18] and thus can be used to enhance our understanding of myotube regeneration and underlying nuclear messaging occurring during this process.
Te proposed muscle regeneration model could be easily transformed into a reproducible high-throughput application. If successful, this high-throughput model will have signifcant application in studies investigating muscle regeneration and ageing, sarcopenia, metabolic and genetic diseases, as well as for initial drug discoveries.

Experimental Design.
Te experimental design is illustrated in Figure 1(a). Te acute efects of BaCl 2 exposure on skeletal muscle regeneration were studied using a modifed experimental design previously described by Fleming et al. [19]. After human or mouse myoblasts became confuent and diferentiated into myotubes, they were injured by exposure to the toxin BaCl 2 . After injury, cells underwent a proliferation and diferentiation period that replicated the preinjury protocol. Tree-four independent experiments were obtained from each cell line.

Mouse Myoblast
Culture. C2C12 murine myoblasts (LGC standards/ATCC) were cultured in 6-well plates at a density of 3000 cells/cm 2 and incubated in a humidifed 5% CO 2 atmosphere at 37°C in the growth medium (GM) composed of high-glucose Dulbecco's Modifed Eagle Medium (DMEM, Sigma) supplemented with 1% penicillinstreptomycin solution (Invitrogen) and 10% fetal bovine serum (FBS, Gibco). Myoblasts were cultured to ∼90% confuence before GM was replaced with the low-serum diferentiation medium (DM), composed of high-glucose DMEM plus 2% horse serum and 1% penicillinstreptomycin solution. Te total culture period was 2 days in GM followed by a further 4 days in DM for myotube formation. All experiments were performed at passages below 10.

Human Myoblast
Culture. Human skeletal muscle cells (aged donor (68 yrs, male) purchased from Promocell and young donor (20 yrs, male), purchased from Lonza were seeded at 3500 cells/cm 2 onto 35 mm 6-well plates and grown in GM (low-glucose DMEM with 20% FBS and 1% penicillin-streptomycin solution). At 90% confuence, GM was replaced with DMEM containing 2% horse serum and 10 μg/mL insulin (Sigma) to induce myogenic diferentiation. Te total culture period was 3 days in GM followed by a further 6 days in DM. All experiments were performed at passages 3-5.

Cell Injury.
Myotubes were diferentiated from human and mouse myoblasts as detailed above until reaching confuency. Fresh DM was added to all cell lines before inducing injury. Ten, 50 μL/mL of 12% w/w BaCl 2 solution was added to the medium, followed by a 2 h, 4 h, or 6 h incubation for C2C12, young, and aged human myotubes, respectively, to cause injury. Myotubes derived from the C2C12 cell line were exposed to BaCl 2 for 6 h, as detailed in the literature [19]. Initially, we also exposed human myotubes for 6 h to BaCl 2 , but this caused complete removal of the nuclei, in addition to the cytoskeleton injury, and resulted in an inability to repair the injury (Figure 1(b)). Tis prompted us to identify the optimal injury time, defned as the period of BaCl 2 exposure that caused damage to the cytoskeleton while maintaining the number of nuclei constant (Figure 1(c)), for each cell line. To identify this optimal period, we stained the cytoskeleton with phalloidin and the nuclei with DAPI and quantifed the nuclei number (Figures 1(b) and 1(c)). We showed that the optimal exposure duration was 6 h for C2C12 and 4 h and 2 h for young and aged human-derived myotubes. After incubation with BaCl 2 , cultures were washed with phosphate-bufered saline (PBS) (Sigma) to remove residual BaCl 2 containing media. Preinjury (CTRL) and directly postinjury (0 h) samples were collected at the end of BaCl 2 incubation. Injury was followed by a regenerative period that follows the exact protocol used to diferentiate cultures. During regeneration, cells for morphological and transcriptome analysis were collected: on day 2 (END PROL) and day 7 (END DIFF) for C2C12 and on day 3 (END PROL) and day 9 (END DIFF) for human cells.  Figure 1: Human muscle cells are more sensitive to barium chloride-induced injury than mouse muscle cells. Experimental design: myotubes derived from mouse (C2C12) and young and aged human myoblasts were chemically injured with 12% BaCl for 6, 4 and 2 h, respectively and then allowed to regenerate (a). Immunofuorescence images showing F-actin and nuclei labelling with phalloidin and DAPI of C2C12, young, and aged human donors before injury (CTRL) and when incubated for 6 h with BaCl 2 . Scale bars represent 100 μm (b). No signifcant diference in the total nuclei number was found before (CTRL) and after injuring the mouse (C2C12) and young and aged human myotubes with 12% BaCl 2 for 6, 4 or 2 h, respectively; Circles indicate individual data points of technical replicates (c). and incubated for 4 h under optimal growth conditions (humidifed incubator 37°C, 5% CO 2 ). After incubation, cells were washed 2x with PBS followed by immunostaining protocol.

Cell
2.6. Immunostaining. Cells were fxed using 10% formaldehyde and permeabilised with 0.25% Triton X-100 (Sigma-Aldrich) solution. Te actin cytoskeleton of cells was identifed using Alexa Fluor 568 phalloidin conjugate (1 : 40, Invitrogen), and Prolong Gold Antifade Mountant with DAPI (Invitrogen) was used to counterstain myonuclei. To further assess the cell dynamics postinjury, EdU assay (Click-iT Plus EdU Imaging kit, Invitrogen) was applied to cultures marking the proliferative cells. In brief, 30-minute incubation with Click-It Plus reaction cocktail consisting of 1x reaction bufer, copper protectant, reaction bufer additive, and Alexa Fluor Picolyl Azide was performed at room temperature in the dark. Cells were then washed with PBS. Myonuclei were labeled with Hoechst staining for 30 min. All fuorescence staining was performed before injury (CTRL), postinjury (0 h), at the end of proliferation (END PROL), and at the end of diferentiation (END DIFF) timepoints during the regenerative period.
2.7. Image Acquisition, Processing, and Analysis. Cells were visualised under a fuorescence microscope (Olympus), and image acquisition performed using Leica Microscope software. At least 10 felds within each well were captured and approximately 180 myotubes from each condition were included in the analysis. Myogenic fusion index was calculated as the ratio of the nuclei number in myotubes with two or more nuclei versus the total number of nuclei. To quantify the percentage of cells synthesising DNA, the ratio between EdU + cells and total myonuclei was calculated. General image analysis was performed using ImageJ software v2.3.0 (NIH, U.S).

RNA Extraction.
Myoblasts/myotubes were collected at three timepoints across the experiment: at control (CTRL), at the end of proliferation (END PROL), at the and end of diferentiation (END DIFF) during the regenerative period and transferred to low binding/RNase free microcentrifuge tubes containing 350 μL RNeasy lysis bufer (Qiagen) supplemented with ß-mercaptoethanol (Sigma-Aldrich) to isolate total RNA using RNeasy mini kit (Qiagen). Following lysis, samples were homogenised and loaded onto RNeasy silica membrane for RNA binding. Concentrated and pure RNAs were eluted in RNase-free water. Concentrations were quantifed using spectrophotometry NanoDrop 2000, Termo Fisher Scientifc refer to the details of an equipment to assess RNA quality. It is not a citation and should be mantained as it is.
2.9. RNA Sequencing. RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Messenger RNA was purifed from total RNA using poly-T oligo-attached magnetic beads. To select cDNA fragments of preferentially 370∼420 bp in length, the library fragments were purifed with the AMPure XP system (Beckman Coulter, Beverly, USA). Ten, PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers, and Index (X) Primer. At last, PCR products were purifed (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system. After cluster generation (cBot Cluster Generation System), the library preparations were sequenced on an Illumina Novaseq platform and 150 bp paired-end reads were generated. Diferential expression analysis of two groups was performed using the DESeq2 R package (1.20.0). Te resulting p values were adjusted using Benjamini and Hochberg's approach for controlling the false discovery rate. Genes with an adjusted P < 0.05 found by DESeq2 were assigned as diferentially expressed. Four independent experiments were performed.

Pathway
Analysis. GO enrichment analysis was calculated using GOseq R package, and KOBAS software was implemented for enrichment analysis in KEGG pathways. Terms with adjusted p values less than 0.05 were considered signifcantly enriched by diferentially expressed genes. GO and KEGG pathway images were generated using the same software.

Statistical Analysis.
Te data were generally presented as mean ± the standard error of the mean (SEM). Statistical analysis was conducted using Graph Pad Prism for Mac Version 9.04. One-way ANOVA with Dunnet post hoc tests was used to analyse the total myonuclei number and EdU + cell data. Fusion index and myotube diameter were compared to preinjury values by using two-sided unpaired ttests. p values <0.05 were considered statistically signifcant and were indicated within fgures as * p < 0.05, * * p < 0.01, and * * * p < 0.001.

Human Muscle Cells Are More Sensitive to Barium Chloride-Induced Injury than Mouse Muscle Cells.
Previous studies reported the development of injury models using 12% BaCl 2 for 6 h in myotubes derived from the C2C12 mouse cell line. Since it is well established [18][19][20], we decided to utilise the C2C12 cell line in our experiments for comparative purposes. We were able to confrm that 12% BaCl 2 for 6 h adequately removed the cytoskeleton without afecting the number of nuclei in the myotubes. However, when we incubated human myotubes under the same conditions, we discovered that such exposure time was causing excessive destruction of the cell structure and cell death (Figure 1(b)). Reducing the exposure time, we identifed the optimal incubation period −2 h for myotubes derived from old and 4 h for cells derived from young myoblasts. Tese exposure times were deemed optimal as they caused cytoskeletal removal without signifcantly changing the total myonuclei number (Figure 1(c)).

Young and Aged Muscle Cells Show a Similar Number of
EdU + Cells during the Proliferation Phase. Te proliferation activity of muscle cells was assessed by EdU (5-ethynyl-2′deoxyuridine) incorporation, which detects the cells entering the S-phase (Figure 2(a)). Te number of EdU + cells increased signifcantly during the proliferation phase in muscle cells derived from both young and aged donors, referred to henceforth as young and aged muscle cells (Figure 2(b)). By the end of the regenerative phase, the number of EdU + cells declined in young and aged cells.
Total myonuclei number increased in human cells during proliferation and remained signifcantly elevated in young cells by the end of the diferentiation period in relation to preinjury (Figure 2(c)).
Mouse myotubes derived from C2C12 showed signifcant increase in EdU + cells during the proliferation phase ( Figure 2(b)) from baseline and a signifcant increase in myonuclei number during proliferation and after the differentiation phase (Figure 2(c)).

Aged Human Cells Show Impaired Diferentiation during
Regeneration. Ten, we assessed myotube width and fusion which are important functional indicators of muscle cell diferentiation (Figure 3(a)). We showed that young muscle cells recovered myotube width (Figure 3(b)) and fusion index (Figure 3(c)) to preinjury values. In contrast, aged cells did not exhibit the same levels of recovery, as evidenced by smaller myotube diameter (Figure 3(b)) and fusion index (Figure 3(c)). Mouse muscle cells showed a slight increase in myotube width and fusion index by the end of the self-repair process compared to preinjury (Figures 3(b) and 3(c)).

Cell Cycle and PI3k-Akt Signalling Pathways Are Significantly Enriched in Both Young and Aged Muscle Cells during
the Proliferative Stage of Regeneration. RNA-sequencing (RNA-seq) data were obtained from young and aged human muscle cells at preinjury (control) and during the proliferation and diferentiation phases. Data obtained during proliferation or diferentiation phases were compared against the baseline. Volcano plots were generated for the young and aged muscle cells. Young cells displayed 895 upregulated and 1187 downregulated genes during proliferation (Figure 4(a)), while 1447 transcripts were identifed as upregulated and 2284 downregulated in aged cells (Figure 4(b)). At the end of the regeneration phase, 253 genes were signifcantly upregulated and 257 downregulated in young muscle cells (Figure 4(c)) and 853 upregulated and 1326 downregulated in aged cells muscle cells (Figure 4(d)) when compared with preinjury. Te Venn diagram of young muscle cells during proliferation shows 10,715 gene transcripts common to baseline and 529 exclusive to proliferation ( Figure 5(a)), while old muscle cells exhibit 10,643 common transcripts and 611 exclusive to proliferation ( Figure 5(b)).
Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of diferentially expressed genes (DEGs) identifed PI3-Akt signalling, cytokine-cytokine receptor interaction, neuroactive ligand-receptor interaction, and cell cycle pathways as the most enriched in both young and older muscle cells at the end of the proliferation compared with baseline ( Figures 5(c) and 5(d)).
Te Venn diagram of young muscle cells during differentiation revealed 11,041 gene transcripts common between baseline and diferentiation and 328 gene transcripts exclusive to diferentiation, whereas the old cells exhibited 10,807 common gene transcripts and 449 exclusive to differentiation (Figures 5(e) and 5(f )).
Cytokine receptor interaction, protein digestion, and absorption together with pathways related to immune response were the most overrepresented KEGG pathways in young at the end of the diferentiation period ( Figure 5(g)).
Aged muscle cells continued to show signifcant enrichment of PI3-Akt signalling pathway as well as cytokine receptor interaction and focal adhesion pathways ( Figure 5(h)).

Aged Myotubes Depict Downregulation of Muscle Cell
Diferentiation and Development-Related Processes. Gene Ontology (GO) analysis of DEGs in young myotubes identifed the cell cycle as the most overrepresented molecular function during the proliferative period ( Figure 6(a)). In aged ones, genes involved in muscle circulation and muscle system processes showed the highest enrichment ( Figure 6(b)). At the end of the regeneration, however, extracellular matrix-related processes were the most enriched in young cells (Figure 6(c)), whereas skeletal muscle processes (i.e., muscle adaptation, contraction, hypertrophy, and relaxation) remained overrepresented in aged cells (Figure 6(d)). A closer analysis of GO enrichment demonstrates the downregulation of these biological processes in aged (Figure 6(f )) but not in young cells (Figure 6(e)), which could explain the impaired muscle cell regeneration in aged muscle shown in the morphological analysis.

Discussion
Herein, we report the successful development of a human in vitro model which is used to investigate the muscle regeneration response across the lifespan. Tis was achieved using muscle cell lines derived from both young and old donors which were then diferentiated into myotubes and injured. We exposed myotubes to BaCl 2 to induce injury and characterise the follow-on repair process with a particular focus on nuclear proliferation, muscle morphology, and the transcriptome. Te transcriptome analysis was of pivotal importance since muscle regeneration relies on specifc mRNA profles for muscle reconstruction [21]. BaCl 2 is a skeletal muscle toxin previously used to study muscle injury and regeneration in vivo [17,18,22,23] and in vitro in monolayer [19] or tissue-engineered muscles [20,24]. While the majority of the studies used animal models [8,25,26], only a few used human-derived muscle cells [27,28] and none investigated the transcriptome which is fundamental to further understand muscle regeneration.
Te most common method to quantify cell proliferation in response to stimuli is the BrdU assay [29,30]. We    Journal of Tissue Engineering and Regenerative Medicine discovered that our muscle model retains the ability to reenter the S-phase (proliferative) of the cell cycle as demonstrated by EdU incorporation in the repaired myotubes. Although the traditional view has been that postmitotic nuclei-like those in diferentiated myocytes-do not proliferate, recent fndings show otherwise. For example, a recent study where mice were labelled with green fuorescent protein and deuterium water provided evidence for proliferative activity of a myocyte [31]. Another mouse study, utilising a 15N thymidine stable isotope tracer, showed that cardiomyocyte nuclei proliferated in both normal and injured heart [32]. Te same study confrmed that the microRNA, miR-222 was involved in the myocyte nuclei proliferation. Our fndings are in line with these studies, but we appreciate that nonfused muscle cells may also have contributed to the observed proliferation and repair. Tis is an inherent limitation of cellular models-the maturation-and hence fusion index is never absolute. Despite this limitation, we feel that a cell model with human genetic makeup (unlike animal models) and compatible with high-throughput testing paramount, e.g., in nutrient, and drug testing ofsets such as limitation.
Aged muscle cells were unable to rescue the diameter and fusion index to preinjury levels, which contrasts with young cells where full recovery was shown. Tese fndings are in agreement with human studies examining muscle growth capacity postexercise (as a form of injury) showing an impaired hypertrophic responses in aged vs. young adults [33,34]. Such impairment can be partly attributed to a proinfammatory systemic environment [11,35] that may cause dysregulation of signalling pathway cascades such as FGF2, Notch, and Cdkn2a in muscle progenitor cells [25,36,37], resulting in a diminished diferentiation capacity, evidenced herein by thinner and less fused myotubes.
We successfully mapped gene transcripts before the injury and during regeneration in young and aged human muscle cells using RNA-seq. Tis enabled us to identify signifcantly enriched KEGG signalling pathways, and GO processes evoked during muscle repair or regeneration. Te KEGG analysis revealed that the cell cycle signalling pathway was signifcantly enriched in both young and aged muscle cells during the proliferative phase which appears to have prompted both young and aged cells to enter the proliferative S-phase after injury as discussed above.     Te PI3k-Akt and downstream mTOR regulate muscle cell proliferation, survival, and diferentiation [28,38,39]. PI3k-Akt pathway was signifcantly enriched in young and aged cells during the proliferation phase. When cells exited the S-phase and entered diferentiation, PI3k-Akt remained enriched in aged, but not in young cells. Te role of PI3k-Akt in aged cells during the diferentiation phase remains somehow unclear since this pathway has many downstream efectors leading to diferent cell fates [39][40][41][42].
Tis notion is further supported by the GO analysis, which suggests that aged cells during the diferentiation and proliferation exhibit signifcant downregulation of processes involved in muscle tissue development indicating diminished maturation capacity. Interestingly, the Venn diagram of aged muscle cells revealed that 10,807 gene transcripts overlapped between diferentiation and baseline, whereas young muscle cells shared 11,041 gene transcripts between timepoints, further highlighting the diference between aged and young cells in their ability to repair to preinjury levels.
Te cytokine-cytokine signalling pathway remained enriched in both young and aged cells during the proliferation and diferentiation phases. Cytokines constitute a major class of regulators of skeletal myogenesis [43,44], so its permanent overrepresentation during the recovery from injury is not surprising. Taken together, the KEGG analysis strongly suggests that factors that stimulate cell cycling played an important role during proliferation while cytokines were central across the regeneration process.
Tis study deserves some important considerations. For example, to reach the similar injury levels across cell lines, exposure time to BaCl 2 in aged muscle cells had to be shorter than that in young muscle cells (2 h vs. 4 h). Although this can be considered as one of the limitations as the injury time is diferent across the cell lines, it also suggests that ageing diminishes the tolerance to injury and this is an interesting fact that is worth further investigations.
To summarise, we successfully developed an in vitro model that provides a high-throughput platform enabling cellular and molecular investigations of muscle regeneration across the life course. As expected, aged myotubes showed impaired regeneration as evidenced by a reduced myofusion index and myotube width after repair. We postulate that this is due to downregulation of genes involved in muscle development and function. We anticipate the use of our model as a high-throughput platform to further examine regeneration across the lifespan, as well as to investigate potential drugs or metabolic/genetic diseases.  Figure 6: GO enrichment analysis of young and aged muscle cells show dissimilar responses. Dot plot shows overrepresented GO terms of biological processes, molecular functions, and cellular locations at the baseline (CTRL) vs. at the end of the proliferation during the regeneration period (END PROL) in young (a) and aged (b) and CTRL vs. at the end of the regeneration period (END DIFF) in young (c) and aged (d) muscle cells. Te size of the dot is based on the gene count enriched in the pathway, and the color of the dot denotes pathway enrichment signifcance. GO enrichment bar chart of biological processes shows most enriched GO terms in young (e) and aged (f ) muscle cells at the END DIFF. n � 4.