Microphysiological system for studying contractile differences in young, active, and old, sedentary adult derived skeletal muscle cells

Abstract Microphysiological systems (MPS), also referred to as tissue chips, incorporating 3D skeletal myobundles are a novel approach for physiological and pharmacological studies to uncover new medical treatments for sarcopenia. We characterize a MPS in which engineered skeletal muscle myobundles derived from donor‐specific satellite cells that model aged phenotypes are encapsulated in a perfused tissue chip platform containing platinum electrodes. Our myobundles were derived from CD56+ myogenic cells obtained via percutaneous biopsy of the vastus lateralis from adults phenotyped by age and physical activity. Following 17 days differentiation including 5 days of a 3 V, 2 Hz electrical stimulation regime, the myobundles exhibited fused myotube alignment and upregulation of myogenic, myofiber assembly, signaling and contractile genes as demonstrated by gene array profiling and localization of key components of the sarcomere. Our results demonstrate that myobundles derived from the young, active (YA) group showed high intensity immunofluorescent staining of α‐actinin proteins and responded to electrical stimuli with a ~1 μm displacement magnitude compared with non‐stimulated myobundles. Myobundles derived from older sedentary group (OS) did not display a synchronous contraction response. Hypertrophic potential is increased in YA‐derived myobundles in response to stimulation as shown by upregulation of insulin growth factor (IGF‐1), α‐actinin (ACTN3, ACTA1) and fast twitch troponin protein (TNNI2) compared with OS‐derived myobundles. Our MPS mimics disease states of muscle decline and thus provides an aged system and experimental platform to investigate electrical stimulation mimicking exercise regimes and may be adapted to long duration studies of compound efficacy and toxicity for therapeutic evaluation against sarcopenia.


| INTRODUC TI ON
The segment of the population aged 65 and older is rapidly expanding and (United Nations, Department of Economic and Social Affairs, Population Division, 2019) represents an enormous healthcare challenge as old age is a risk factor for many chronic diseases (Atella et al., 2019). Sarcopenia is a progressive pathology characterized by the loss of muscle mass and strength, and is prevalent in those over 60 years old (Papadopoulou, 2020). Sarcopenia eventually leads to loss of independence and elevated risk of morbidity and mortality and is a massive economic burden with direct annual healthcare costs estimated to be over $19 billion (Goates et al., 2019). Despite these medical and socioeconomic costs, there is a lack of effective therapeutic options available in part because the mechanisms underlying sarcopenia are challenging to study over many years in the same patient.
Satellite cells are the predominant stem cell population in adult skeletal muscle and are thought to play a role in the progressive pathology of sarcopenia (Alway, Myers, & Mohamed, 2014). Satellite cell content is decreased in muscles of older people humans (Verdijk et al., 2014), and there is evidence of functional impairment due in part to altered systemic factors that impact satellite cell activity and differentiation (Conboy & Rando, 2005). Cell-autonomous alterations also contribute to the functional deficit of old satellite cells (Brack & Muñoz-Cánoves, 2016). In addition, age and elevated levels of physical activity have been shown to reprogram muscle satellite cells such that they retain key aspects of the donor's phenotype once in culture. For example, muscle cells from physically active or trained donors retain high oxidative capacity (Bourlier et al., 2013;Lund, Helle, et al., 2018a;Lund, S Tangen, et al., 2018b;Pino et al., 2019) and for older adults, muscle cells have lower oxidative capacity and fewer cells with lower capacity for proliferation and differentiation (Aas, Thoresen, Rustan, & Lund, 2020;Chen, Datzkiw, & Rudnicki, 2020;Sousa-Victor et al., 2014). As such, primary human muscle satellite cells represent a valuable in vitro model to study aging and the impact of exercise.
Human skeletal muscle 3D microphysiological systems (MPS), also referred to as tissue chips, that mimic muscle morphology and function on a tissue level, hold promise as therapeutic testing platforms. There have been significant advancements in development of MPS devices incorporating myogenic cells reviewed in the literature (Madden, Juhas, Kraus, Truskey, & Bursac, 2015). Muscle myobundles have likewise been generated to polymerize around deformable micro posts (Agrawal, Aung, & Varghese, 2017;Mills et al., 2019;Vandenburgh et al., 2008). In these systems, skeletal muscle cells have been embedded in various naturally derived hydrogels to mimic extracellular matrix (ECM) to enhance maintenance and repair of skeletal muscle and aid in myobundle force transmission (Csapo, Gumpenberger, & Wessner, 2020;Hinds, Bian, Dennis, & Bursac, 2011). Introduction of electrode wires and microelectrode or optical sensors allows for electrical stimulation of the muscle cells and recording of the contractile behavior via time-lapsed microscopy or force calculations (Truskey, 2018). Electrochemical stimulation coupled with mechanical contraction of skeletal muscle bundles, aimed at mimicking the action of the motor neurons, has a pivotal role in inducing changes in cell morphology, hypertrophy, and maturation (Guo, Cheung, Yeung, Zhang, & Yeung, 2012;Jaatinen et al., 2016;Langelaan et al., 2011).
These 3D MPS have the advantage to be more physiologically relevant in studying biomechanics than 2D muscle cultures and hold the promise to understand disease mechanisms and tissue remodeling under controlled conditions. However, few studies till date have been published that use human primary cells in a microfluidic environment or use donor-specific cells, to model a disease state, such as from older, sedentary adults who are at higher risk for developing age-related muscle diseases. Moreover, contractile characteristics of muscle cells from populations of old and young donors have also not been extensively studied. Physiology and pharmacological studies in MPS using donor-specific cells may result in a greater likelihood of successful clinical trials of drug candidates. In this respect, the development of tissue chips incorporating satellite cells from younger and older adult donors capable of accurately modeling both healthy and atrophied states would accelerate the pace of pharmacological studies and contribute to the advancement of tissue engineering.
Herein, we report on validation studies of a microfluidic MPS that demonstrates contractile bioengineered myobundles derived from muscle biopsies from young, athletic (YA) and older, sedentary (OS) adults. The goal of our study is to develop a model of muscle aging to investigate the cellular mechanisms underlying sarcopenia, to use for pre-clinical drug evaluation and to inform our studies being conducted on the International Space Station to study effects of microgravity-induced muscle atrophy that may mimic the physiological effects of aging on a faster timescale than on Earth (Sharma et al., 2022). We reasoned that isolating muscle cells from these two distinct human populations maximized our chance of working with phenotypically distinct muscle cells. Thus, incorporating muscle precursor cells isolated from adults phenotyped based on age and physical activity as an autonomous cell model to measure intrinsic effects adapted to long duration studies of compound efficacy and toxicity for therapeutic evaluation against sarcopenia.

K E Y W O R D S
bioengineered skeletal muscle, cellular electrical stimulation, human CD56 + primary cells, muscle myogenesis gene expression, sarcomere immunofluorescence of cellular stress offers the possibility of studying innate characteristics of the donor and more accurately reflects human physiology disease states. In this respect, our MPS seeks to be a more effective bridge between discovery and clinical research.
2 | RE SULTS 2.1 | Human skeletal muscle cell young active (YA) and old sedentary (OS) phenotypes Ten male participants were screened and enrolled to the following groups: young active (YA; 21-40 years, n = 5) and older sedentary (OS, 65-90 years, n = 5). Participants were considered active if they engaged in endurance exercise (running, cycling, or swimming) at least 3 days a week without extensive lay off over the previous 6 months. Participants were considered sedentary if they completed one or fewer structured exercise sessions a week. The characteristics of the study groups are shown in Table 1. As per the study design, the YA group participants were younger, leaner, had a greater cardiorespiratory fitness, and lower BMI compared with the OS group indicating higher fitness and lower adiposity typically associated with an endurance-trained physically active lifestyle. Muscle progenitor cells were isolated from a 50-100 mg biopsy specimen from these volunteers using a pre-plate technique previously described (Sparks et al., 2011). The cells from each donor were expanded for two passages, counted, and pooled in equal ratios to provide mean YA and mean OS myoblast stocks referred to as YA cohort and OS cohort.
The pooled myoblasts were grown to confluence, aliquoted, and frozen. CD56 is an important cell surface marker known to be expressed by satellite cells and other pro-myogenic cells within skeletal muscle (Vauchez et al., 2009). Prior to cell seeding into tissue chips, thawed aliquots were purified to enrich for CD56 + myogenic cells.
The percent recovery of CD56 + cells compared with the total cell count prior to purification was calculated as 66 ± 15% for YA (n = 5 aliquots) and 50 ± 10% for OS (n = 5 aliquots) cohorts. Enrichment of the CD56 + myogenic cell pools was confirmed by flow cytometry and determined to be >99% CD56 + ( Figure S1).

| YA-and OS-derived myobundles display varying multinucleation
The CD56 + enriched cells were encapsulated into a collagen-Matrigel 3D scaffold to provide a suitable tissue-mimicking microenvironment for the bioengineered skeletal muscle myobundles. The cell-laden hydrogel was seeded into an enclosed microfluidic chip fabricated from PDMS and bonded to glass. Platinum electrodes were embedded along the media channel and extended 3.2 mm outside the PDMS for attachment to connectors and a pulse generator during electrical stimulation (Figure 1a, upper). The PDMS chip featured an inner channel 3.2 mm wide aligned with tapered surfacetension pins set 0.3 mm apart and two PDMS posts of 1 mm diameter spaced 5 mm apart (center to center) to allow cells to condense around the posts and form a single myobundle per chip. The outer media channel allowed perfusion of the condensed myobundle.
This 1:5 diameter to distance ratio described previously (Agrawal et al., 2017) proved to be optimal to form approximately 0.8-1 mm thick, stretched myobundles after 48 h when YA-derived cells were seeded at 15 × 10 6 /mL and OS-derived cells were seeded at 20 × 10 6 / mL as visualized by phase contrast microscopy (Figure 1a, lower). A higher density of YA cells caused too much strain and resulted in myobundles falling off the posts, whereas the OS cells at the lower density were too thin and did not condense well.
Myobundles were differentiated over an additional 12 days (Day 14) by a two-step differentiation protocol and an intermittent flow rate to result in 3D myobundles 7.6 mm in length (Figure 1b

| Direct current simulations confirm uniform electric field applied to myobundles
To assist in selecting an optimal range of applied voltage, based on electrodes located along the media channels parallel to the myobundle, finite element analyses of electric currents within the chip were Note: Values are presented as mean ± SD.
Abbreviations: BMI, body mass index; VO 2 peak, maximum rate of oxygen consumption. *Group differences were determined using a non-paired students t-test. spectroscopy (EIS) assessments of the tissue chips immersed with differentiation media were used to obtain the conductivity of the fluid and engineered tissue present between the electrodes. This predicted that a homogeneous distribution of electric field intensity of 9.0 ± 0.9 V/cm occurred at the horizontal centerline of the gel chamber between the tissue posts when a 2 V potential is applied ( Figure S2a). When a 3 V potential is applied, a moderate increase in electric field strength of 13.3 ± 0.1 V/cm was predicted with a similar distribution of currents as 2 V ( Figure S2b). Simulating at a higher voltage of 10 V resulted in a 41.3-59.8 V/cm range of electric field intensity at the centerline which was not optimal ( Figure S2c). Upon closer inspection, a change in the applied voltage from 2 V to 3 V revealed that the presence of the porous hydrogel (triangles, electrical conductivity: σ = 2.98e-8 S/m) had a larger effect in electric field changes than the electrolytic fluid alone (diamonds, electrical conductivity: σ = 0.346 S/m) ( Figure S2d). These simulations helped guide the optimal range of applied voltage to achieve the desired electric field intensity at the hydrogel chamber. the muscle bundle was observed in YA-derived chips when 3 V, 2 Hz, 2 ms parameters were applied at the parallel electrodes. The same stimulation parameters were selected for both YA-and OS-derived myobundles because our objective was to observe differences in contractile response of the OS-derived myobundles compared with the healthy controls (YA-derived myobundles) and not to confound their responses to differences in stimuli.

| YA-and OS-derived myobundles differ in contraction rate and displacement signal
We first differentiated the tissue chips for 13 days followed by an additional 5 days of differentiation in the presence of an applied electrical stimulation for a total of 17 days of culture. During the 5 days of electrical stimulation, we applied the 3 V, 2 Hz, with 2 ms pulse width sequence to the tissue chips for 30 min each day to mimic daily physical activity. Skeletal 3D myobundles have been reported to be stimulated for a maximum of 60 min with 7-12 h rest in between stimulation sequences (Khodabukus et al., 2019). A control group of tissue chips from each cohort did not receive electrical stimulation over the 5 days. At the end of the experiment on Day 17, image sequences of all tissue chips were recorded for a total of 120 s divided into the following phases: 40 s before (resting phase), during (E-stim phase), and after (recovery phase) an applied electrical stimulation.
Digital image correlation (DIC) analysis of these image sequences as described in Figure

YA-and OS-derived myobundles
The significant difference in contraction displacement response F I G U R E 2 Contraction measurements of skeletal myobundles. (a-f) Displacement measurements were obtained using a digital image correlation algorithm on videos taken of (a-c) the YA (red lines) and (d-f) OS (blue lines) during resting phase (a, d), E-Stim phase after applying a 3 V, 2 Hz, 2 ms electrical stimulation (b, e) and recovery phase (c, f) and compared with YA-and OS-derived myobundles that did not undergo electrical stimulation (black dotted lines). Standard deviations are shown in light and dark gray shading on each graph for stimulated and non-stimulated chips, respectively. (g-j) A Fast Fourier Transform analysis and standard signal processing techniques were applied to the displacement measurements to extract the average dominant frequency (g), mean displacement signal (h), average difference between displacement peaks and the sample resting mean displacement signal (i) and average difference between valleys or local minima and the sample resting mean signal (j). Bars, p < 0.05 between stimulation phases of cells within the same age group, and brackets between cells age groups at the same phase F I G U R E 3 Characterization of 3D myobundles sarcomere by α-actinin. YA-derived myobundles non-electrically stimulated for 5 days (a) and electrically stimulated (b) or OS-derived myobundles non-electrically stimulated for 5 days (c) and electrically stimulated (d) were stained with an anti-sarcomeric α-actinin antibody (red) and counterstained with the Dapi nuclear marker (blue) revealing the presence or absence of striated structure. Scale bars = 100 μm (a-d) and 20 μm (enlarged region of interest shown at right side in each figure shows crossstriated appearance of skeletal muscle myobundles) genes were downregulated DEGs in YA-and OS-derived myobundles, respectively, after differentiation as illustrated in the volcano plots ( Figure S6a, b). The complete list of genes, fold-change, and p-values for each treatment group are shown in Figure S7. Twentynine of the DEGs upregulated and 2 downregulated were shared across all four categories (YA no E-Stim, YA E-Stim, OS no E-Stim, and OS E-Stim) as depicted in the Venn diagrams ( Figure S8a, b).
The 29 up-regulated DEGs in both differentiated YA and OS with or without E-Stim were processed using the STRING v10 WEB based bioinformatics tool (Szklarczyk et al., 2011) which generated protein interaction network with three clusters using highest confidence interaction scores (≥0.9) and hiding disconnected nodes from the network ( Figure S9). The 29 DEGs were annotated according to the STRING database and the top 5 gene ontology (GO) terms based on the false discovery rate scores of biological processes, molecular functions, and cellular components, as summarized in Figure S10.

| DISCUSS ION
We describe the development of an electrode embedded microfluidic tissue chip to study phenotypic and physiological differences between human skeletal muscle myobundles generated from pooled primary cells as a model to study age-related muscle atrophy. We provide a comprehensive analysis of the difference in contraction displacement rates, gene expression profiling, and immunostaining, and validate this model to be used as ground studies for spaceflight and for therapeutic discovery. It should be kept in mind that for this study, the reported differences may not be related to just an age difference, but to a different level of muscle activity as the satellite Although the OS-derived myobundles exhibit the same 3D architecture as their YA counterparts, these aged myobundles induce residual, non-synchronic contraction in response to an acute electrical stimulation regime. The differences we observe in muscle contractile characteristics between the YA-and OS-derived myobundles are in line with reports that donor age can impact other key phenotypes that are retained in primary myotubes, including substrate oxidation (Aas et al., 2020). Thus, our MPS provides a physiologically relevant platform to study mechanisms involving molecular and functional response differences in aging muscle.
We profiled the expression of 55 genes relating to myogenesis and myopathy, a majority of which were upregulated similarly in YA-and OS-derived myobundles during the two-step differentiation process compared with their respective myoblast stage. Key transcriptional activators that promote transcription of muscle-specific target genes involved in muscle cell fate commitment were also upregu- Titin and associated proteins are also important components of the myofilament system (Granzier & Labeit, 2005). In the YA-derived myobundles, titin (TTN), troponin T1 (TNNT1), troponin I2 (TNNI2), and T3 (TNNT3) genes are upregulated by 500-, 300-, 2500-, and 600-fold, respectively. With electrical stimulation, genes increased in expression by an additional 1.5-, 2-, and 2-fold for TNNT1, TNNI2, and TNNT3, respectively. These genes highlight the myofiber mix between fast and slow twitch muscle type characteristic of vastus lateralis and induced by electrical pulse stimulation in cultured human myotubes (Marš et al., 2021;Staron et al., 2000). Alphaactinin immunofluorescence staining of the YA-derived myobundles confirmed well-defined z-line architecture in the fused myotubes.
No genes were specifically induced by electrical stimulation in the OS-derived myobundles. Nonetheless, non-stimulated OS-derived myobundles such as non-stimulated YA-derived myobundles exhib- In addition, in our model, insulin growth factor 1 (IGF-1) and myostatin, the negative regulator of muscle growth, were induced upon electrical stimulation in the YA-derived myobundles but attenuated in electrically stimulated OS-derived myobundles. In addition to decreased muscle growth compared with younger individuals, aging is associated with a variable hypertrophic response after acute exercise training (Welle, Totterman, & Thornton, 1996). Under normal conditions, gene expression analysis determined in muscle of young and old subjects indicated that both IGF-1 and myostatin are attenuated in adults >80 years old (Naro et al., 2019). Our results highlight the variable hypertrophic response and the balance between proand anti-atrophy gene expression in the older cohort under an acute electrical stimulation regime.
An advantage of our contraction analysis is that we determine a displacement signal over time from the average displacement magnitude of the bundle in the horizontal and vertical direction across a sizable region of interest (described in Figure S2). Our results indicate that the YA-derived myobundles showed mean displacement magnitude of contraction 4× the mean displacement value of OSderived myobundles (Figures 2h and S5). Our analysis also detected regions of the myobundle that moved in a preferentially axial ori- In addition, the YA-derived myobundles contracted on average 5-8 times faster than during their resting or recovery phases. The synchronous signal of YA-derived tissue chips under electrical stimulation followed by a period of negligible frequency of low intensity contractions in the recovery phase indicates that the YA-derived myobundles undergo relaxation after electrical stimulation is removed. In OS-derived myobundles, a higher frequency twitch pattern of low intensity contractions was detected compared with the resting phase, suggesting that OS-derived myobundles may possess insufficient myotube density or sarcomere structures to generate a synchronous contraction response. Our immunofluorescence sarcomere staining supports this hypothesis.
Native muscle contraction is a three-dimensional phenomenon, and this nature influences its mechanical output and is dependent on variations in muscle shape change (Roberts et al., 2019). Our in vitro system that mimics the multi-scale nature of muscle mechanics can be enhanced with different electrical stimulation "exercise" regimes and may provide a link between structure and functional changes that occur during aging. Furthermore, by studying the physiological effects of myobundles derived from myoblasts from participants phenotyped based on age and physical activity and comparing effects on muscle biology in microgravity, our data stand to highlight the distinct impact that aging and physical activity combined have on human muscle energetics.

| Chemicals
Dow SYLGARD™ 184 Silicone Elastomer Clear was purchased from

| Design and chip fabrication
Chip designs were created using Solidworks 2020 (MLC CAD Systems, LLC, Austin, TX). Chip molds were 3D printed by Accura were prepared by mixing elastomer base and encapsulant (10:1) and pouring into the 3D platinum-containing molds and placing under vacuum for 15 min to remove air. Platinum wires (24 mm) were secured into the molds, and the molds were cured at 50°C for 6 h followed by 1 h at ambient temperature. The cured PDMS was removed from mold and 1 mm biopsy punch was used to generate ports prior to bonding the PDMS to glass slides of various sizes by plasma cleaning PDMS and glass for 60 s under vacuum using a Harrick plasma cleaner. Bonded chips were sterilized with ethanol and heated at 150°C for 30 min prior to cell seeding.

| Study design and participants
Volunteers were recruited to participate in this study from the Orlando, FL area. Volunteers were eligible to participate if they were weight stable (±4.5 kg in preceding 6 months), had a body mass index (BMI) between 20 and 35 kg/m 2 , and were in good general health.
Volunteers were excluded if they were taking medications known to influence muscle metabolism, had a chronic medical condition (diabetes, cardiovascular disease, and cancer), had any contraindications to exercise, had high resting blood pressure (≤150 mmHg systolic, ≤90 mmHg diastolic). Screening and clinical phenotyping were completed over four study visits at the Translational Research Institute (TRI) at AdventHealth, Orlando. The screening visit consisted of a fasting blood draw, physical measurements, medical history/physical activity questionnaires, and resting electrocardiography (ECG).
On Visit #2, participants completed a VO 2 peak test with ECG to determine cardiorespiratory fitness (Distefano et al., 2018). On Visit #3, participants completed magnetic resonance imaging (MRI), and a dual energy X-ray absorptiometry (DXA) scan to assess body composition and quadriceps contractile and physical function testing, as previously described (Distefano et al., 2018). All participants provided written informed consent, and the study protocol

| Percutaneous muscle biopsy and skeletal muscle cell isolation
On the final study visit, participants arrived in a fasting state.
Participants then consumed a small low glycemic index meal   Figure S1).

| Fabrication of myobundles and differentiation
Enriched myoblasts were carefully mixed with a hydrogel composed

| Immunofluorescence staining of skeletal muscle myobundles
Differentiated muscle bundles were fixed and immunostained di-

| Image analysis of myobundles
Immunofluorescence images were acquired using a Biotek Cytation5 Imager (Agilent). Myosin heavy chain images were obtained using a

| Electrical field simulation in tissue chip
While the electric field cannot be measured directly, the electric potential difference or voltage change between two points can be obtained from direct measurements. In the case of the tissue chip, the electrodes exposed to conductive media formed an electrochemical cell at the interface between the liquid electrolyte-rich media and the solid metallic electrodes. An electrochemical impedance spectroscopy (EIS) technique commonly used to evaluate commercial batteries (Wang et al., 2021)

| Statistical analysis
All experimental measurements report error as standard deviation and the statistical significance between test groups (i.e., # YA chip replicates vs. OS chip replicates) was determined by paired t-tests. For clinical data presented in Table 1, group differences were determined using a non-paired students t-test. The statistical significance of the differences in contraction parameters between the OS-and YA-derived chips at the different stimulation phases was determined using a linear mixed-effects model in R-Studio (Bates, Mächler, Bolker, & Walker, 2015). The stimulation phase, cell type, and interactions were considered as fixed effects,

ACK N OWLED G M ENTS
The authors acknowledge that this study was supported by the National Institutes of Health National Center for Advancement of Translational Sciences (5UG3TR002598 to S.M.). We thank Tushar Shenoy for help with the cell enrichment process and Austin Hinkle for help with creating microfluidic devices during this study.