Effects of topological constraints on the alignment and maturation of multinucleated myotubes

Microfluidic‐based technologies enable the development of cell culture systems that provide tailored microenvironmental inputs to mammalian cells. Primary myoblasts can be induced to differentiate into multinucleated skeletal muscle cells, myotubes, which are a relevant model system for investigating skeletal muscle metabolism and physiology in vitro. However, it remains challenging to differentiate primary myoblasts into mature myotubes in microfluidics devices. Here we investigated the effects of integrating continuous (solid) and intermittent (dashed) walls in microfluidic channels as topological constraints in devices designed to promote the alignment and maturation of primary myoblast‐derived myotubes. The topological constraints caused alignment of the differentiated myotubes, mimicking the native anisotropic organization of skeletal muscle cells. Interestingly, dashed walls facilitated the maturation of skeletal muscle cells, as measured by quantifying myotube cell area and the number of nuclei per myotube. Together, our results suggest that integrating dashed walls as topographic constraints in microfluidic devices supports the alignment and maturation of primary myoblast‐derived myotubes.

Microfluidics-based approaches offer the possibility to manipulate fluid flow, providing a strategy to control communication between cells that is mediated by factors present in the fluid phase (Maimon et al., 2018;Mills et al., 2018;Osaki et al., 2018;Southam et al., 2013;Uzel et al., 2016). C2C12 cells are widely used in microfluidic devices as they can be easily differentiated into myotubes.
However, C2C12-derived myotubes do not adequately mimic the physiology (Osaki et al., 2018) and metabolic properties (Abdelmoez et al., 2020) of skeletal muscle. Myotubes derived from iPSCs resemble the properties of skeletal muscle cells more closely than C2C12 myotubes but their integration in microfluidic devices is complex, taking up to several months to implement (Osaki et al., 2018). Primary myoblasts differentiate into myotubes that retain many properties of native skeletal muscle in a relatively fast and consistent manner (Ruas et al., 2012;Taylor-Weiner et al., 2020).
However, the use of primary myoblasts in microfluidics devices is hampered by difficulties in achieving successful differentiation on glass substrates and the fact that differentiation is dependent on adequate and uniform initial seeding density of the myoblasts, which can be difficult to achieve within microfluidics channels. In this study, we investigated the effects of topological constraints implemented on microfluidics devices on the alignment and maturity of myotubes derived from primary myoblasts. We aimed to design devices that supported a seeding density of primary myoblasts that was conducive to myotube differentiation and maturation while promoting the alignment of differentiated myotubes. We tested two types of topographic constraints: channels with conventional continuous walls (solid walls) and channels with walls composed of 250 μm long segments separated by 50 μm (dashed walls). The rationale for investigating the effects of dashed walls on myotube alignment and maturation was our observation, using simulations of fluid flow in channels, that dashed walls created pockets of low velocity in the channels, which we hypothesized could lead to improved control of seeding densities of primary myoblasts.
Cells are grown on tissue culture polystyrene (TCPS) substrates in established protocols of primary myoblast differentiation into myotubes (Ruas et al., 2012), whereas glass often acts as the cell culture substrate in microfluidic devices. This is because microfluidics devices are usually made of polydimethylsiloxane (PDMS), which can be easily mounted on the glass to form a tight seal whereas PDMS adhesion to TCPS is generally poor. To facilitate the implementation of primary myoblast differentiation on microfluidic devices, we used a method we recently developed that improves the strength of adhesion between TCPS and PDMS, consisting of an oxygen plasma treatment of TCPS and PDMS surfaces followed by temperature annealing of both surfaces (Song et al., 2018). Using this setup, here we show that topological constraints promote myotube alignment, irrespective of whether the topological constraints are formed by solid or dashed walls. However, dashed walls were more effective at promoting myotube maturity. Together, these results suggest that using dashed walls as topographic constraints can be beneficial for the development of microfluidics devices when control of seeding density is a critical variable.
Polystyrene (PS) dishes were dried before cell seeding. Cells were kept at less than 50% confluency in 20 ml GM in an incubator at 37°C and 5% CO 2 , and the GM was changed every 72 h.

| Microfluidic device fabrication
Microfluidic devices with topological constraints were fabricated by standard soft lithography and replica molding processes (Qin et al., 2010;Xia & Whitesides, 1998). For the molding process, glass slides (75 mm × 50 mm, Sigma-Aldrich) were used as a substrate.
As SU-8 photoresist adheres poorly to glass, a thin layer of the negative photoresist, SU-8 2 (~1 µm in thickness; MicroChem) was first spin-coated as an adhesive layer at 4000 RPM for 30 s (J. Liu, Song, et al., 2014). The photoresist was soft-baked on a hot plate at 65°C for 1 min and 95°C for 2 min before UV light exposure at 10 mW cm −2 for 60 s without a photomask. The photoresist was then post-baked on a hot plate at 65°C for 1 min and 95°C for 2 min.
On top of the adhesive layer, SU-8 2100 (MicroChem) was spincoated at 4500 rpm for 30 s, yielding a layer with a thickness of approximately 50 µm. The SU-8 layer was soft-baked on a hot plate at 65°C for 3 min and 100°C for 15 min, exposed to UV-light for patterning at 10 mW cm −2 for 20 s, and post-baked on a hot plate at 65°C for 5 min and 100°C for 20 min. After being allowed to cool down to room temperature, the SU-8 layer was developed in propylene glycol monomethyl ether acetate (Sigma-Aldrich), rinsed with isopropanol alcohol (Sigma-Aldrich), and dried with compressed nitrogen. The surface of the mold was treated with trimethylchlorosilane (Sigma-Aldrich) in a vacuum chamber overnight to silanize the surface for casting. We cast this SU-8 master with a mixture of polydimethylsiloxane (PDMS; Sylgard 184; Dow Corning) and curing agent at an 11:1 weight ratio. This ratio was chosen to promote the adhesion of the PDMS onto a PS petri dish (Song SONG ET AL. | 2235. The mixture was degassed in a vacuum chamber for 30 min and heated at 65°C for 2 h in an oven. After curing, the PDMS microchannel device was peeled off from the master, and biopsy punches (Kai Medical) were used to create 8 mm medium reservoirs as well as 2 mm inlet ports for cell seeding.
After the cleaning process, the surface of the device was treated with oxygen plasma at 300 mTorr (Harrick Plasma) for 1 min and coated with polyethylene glycol (PEG) solution (PEG; acetone; 1:1 in volume) on parafilm for 1 h at room temperature to maintain the hydrophilicity of the device over time (Kovach et al., 2014). We observed that the devices remained hydrophilic for~48 h after PEG coating.
Then, the device was rinsed with DI water, dried with compressed nitrogen, and attached to a PS cell culture dish. Finally, the device mounted on the dish was incubated at 65°C for 1 h in an oven, which resulted in reversible bonding with a tight seal and good adhesion. This method allows us to keep the device in a dry state for further processing (e.g., UV sterilization) and to use the standard extracellular matrix coating for primary myoblasts culture, avoiding the need for further physical and/or chemical surface modifications (Lee & Ram, 2009;Song et al., 2018;Sunkara et al., 2011;Tang & Lee, 2010).

| Cell culture in the microfluidic device
For sterilization, the device was exposed to UV light in a laminar flow cell culture hood for 20 min. Then, the device was filled with 50 µg/ml collagen solution overnight. After aspirating the collagen solution, we added 100 µl GM in each reservoir and kept the device in an incubator at 37°C until the flow was stabilized in the device. In the meantime, primary myoblasts were trypsinized and prepared in Eppendorf tubes.
Primary myoblasts were suspended in 3.5 µl GM and carefully injected into the device prefilled with GM through the inlet port, which promoted uniform cell seeding. The hydrostatic pressure formed by the cell suspension allowed the cells to flow into the device. We kept the device in an incubator at 37°C and 5% CO 2 overnight to allow primary myoblasts to adhere to the PS substrate on the device. The following day (D1), we exchanged GM with differentiation medium (DM) consisting of high-glucose DMEM, 5% horse serum (HS; Thermo Fisher Scientific; cat #16050-122), and 1% penicillin-streptomycin. DM was changed daily, and the device was kept in the incubator at 37°C and 5% CO 2 for 4 days (D1-D4). Three samples of each device were used for each batch of cells (technical replicates), and the experiment was repeated three times with different batches of cells (biological replicates) from new vials. For each device design, we employed three different densities of seeded primary myoblasts: 5 × 10 3 , 10 × 10 3 , and 20 × 10 3 cells/mm 2 .

| Immunostaining of cells on-chip
On D5, we fixed the cells using 10% formalin for 20 min at room temperature and permeabilized them with 0.1% Triton X-100 for 10 min at room temperature. Then, the nuclei and the cytoskeleton were labeled with 4ʹ,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific) and Alexa Fluor 488 Phalloidin (Thermo Fisher Scientific), respectively, for 40 min at room temperature.

| Imaging and image analysis
For each device, 12 consecutive locations along the topological constraints in the culturing area were imaged using a ×10 objective on a ZEISS inverted fluorescence microscope. Three types of image analysis were carried out with ImageJ: angle of myotube orientation, number of nuclei per myotube, and myotube area. To facilitate image analysis, the 12 consecutive images were aligned and assembled to generate a single image, covering half of the culturing area with constraints, thereby avoiding the areas of the input port and areas without constraints. Imaged cells that were found to overlap with other cells were manually separated by adding black lines at the intersections using Adobe Photoshop. The assembled ×10 images were used for analysis of cell orientation and maturity. The angles of myotube orientation were measured by using the Directionality plugin of ImageJ. Myotubes were identified by Alexa Fluor 488 Phalloidin staining. The number of nuclei in a myotube was determined by counting the number of DAPI positive nuclei contained in an Alexa Fluor 488 Phalloidin labeled cell. The area of a myotube was measured by counting the number of pixels per myotube. The image processing for counting nuclei and sizes was also performed by ImageJ. The processed data were collected and sorted by wall types, wall spacings, and seeding densities. Subsequently, the data for each condition were integrated and analyzed without normalization.

| Microfluidic devices with topological constraints
We designed two types of microfluidic devices with integrated topological constraints formed by solid walls or dashed walls (Figure 1).   | 2237 multinucleated myotubes, which could be observed within a day after changing the medium (Figure 4). On Day 2 (D2), longer myotubes started to be observed, and on Day 3 (D3), thick and long myotubes were observed. Single primary myoblasts that did not differentiate into myotubes were also observed throughout the experiment.

| Topographical constraints induced cell alignment
Immunofluorescence imaging at Day 5 (D5) revealed that topographical constraints induced the alignment of primary myoblasts and myotubes in the devices ( Figure 5). The fraction of aligned cells was calculated as the fraction of cells that had a long axis oriented within 10°of the direction of the walls. At all seeding densities tested (5K: 5 × 10 3 cells/mm 2 , 10K: 10 × 10 3 cells/mm 2 , and 20K: 20 × 10 3 cells/mm 2 ) cell alignment showed a trend of decreasing with the increasing spacing of the walls (Figure 6). The fraction of aligned cells tended to be slightly lower in channels with dashed walls than solid walls, and this difference was statistically significant for wall spacings of 200 µm at 10K (p < .05) and 25 µm at 20K (p < .01). Immunofluorescence images showed that in channels with dashed walls, the cells were able to cross the walls, which contributed to the decrease in cell alignment compared to cells in channels with solid walls.

| Topographical constraints affected myotube maturation
We investigated the effects of the topological constraints on the maturity of primary myoblast-derived myotubes by measuring myotube areas and counting the nuclei numbers per myotube ( Figure 7). For this analysis, we considered only multinucleated myotubes that contained two or more nuclei. The myotube area was generally larger in devices with channels containing dashed walls than in channels with solid walls. In particular, at 5K, a statistically significant increase in myotube area in dashed walls compared to solid walls was observed for wall spacings of 25 µm (p < .001). At 10K, statistically significant increases in myotube area for channels dashed walls compared to solid walls were observed for spacings of 100 and 200 µm (p < .05 and p < .001, respectively) and at 20K, the study of the interactions between skeletal muscle and peripheral neurons at the neuromuscular junction has greatly benefited from the use of microfluidics technologies (Uzel et al., 2014).
However, the cell lines that are widely used to generate myotubes in microfluidics devices, primarily C2C12 cells, do not recapitulate many key aspects of skeletal muscle physiology (Osaki et al., 2018). Therefore, there is a need to improve the compatibility of microfluidics devices with primary myoblast-derived myotubes, while preserving the advantages that microfluidics devices offer, such as providing alignment for skeletal muscle cells to achieve a biomimetic anisotropic morphology as well as control of fluid flow. In this study, we introduced topological constraints in microfluidics channels to obtain aligned myotubes derived from We investigated the effects of introducing dashed walls as topographic constraints in microfluidics channels to improve the uniformity of cell seeding compared to solid walls in the devices with the goal of obtaining enhanced maturity of the resulting myotubes.
We observed that a seeding density of 20K in the devices led to lower myotube area and the number of nuclei per myotube compared to seeding densities of 5K and 10K. However, compared to F I G U R E 7 Area of myotubes with respect to the wall spacing and type of topological constraint, at seeding densities of 5K, 10K, and 20K. S and D in the width indicate solid walls and dashed walls, respectively. *p < .05, **p < .01, and ***p < . Myotubes derived from primary myoblasts were maintained in the microfluidic devices for up to 5 days. This was enabled by the fact that the microfluidic devices were mounted directly on TCPS, which is the standard substrate for myotube differentiation (Ruas et al., 2012), instead of glass that is generally used as a substrate in microfluidic devices. However, the use of TCPS instead of glass hindered analysis by confocal microscopy. Although we observed the formation of multinucleated myotubes through the cultured area of the devices, we also detected mononucleated primary myoblasts in all conditions, signaling a need for further improvements to promote the maturation of myotubes.

| CONCLUSIONS
We investigated the effects of topological constraints integrated in microfluidic models on the alignment and the maturity of multinucleated myotubes derived from primary myoblasts. We demonstrated that topological constraints in the form of dashed and solid walls were able to direct the alignment of myotubes derived from primary myoblasts. Importantly, dashed walls promoted the maturity of myotubes, resulting in larger numbers of nuclei per myotube and larger myotube areas. Together, these data show that dashed walls as topographic constraints integrated with microfluidic devices promote the maturation of myotubes derived from primary myoblasts. All authors contributed to manuscript revision and gave approval to the final version.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.