Gelatin methacryloyl was synthesized following a previously published protocol (Costantini et al., 2016). Briefly, gelatin (type A3, ~300 Bloom from porcine skin) was dissolved at 10% (w/v) in PBS at 60°C. Methacrylic anhydride (MA, 0.08 mL per gram of gelatin) was then added to the gelatin solution dropwise under vigorous stirring and the mixture was allowed to react for 2 h (Figure 1A). After a five-fold dilution with additional warm PBS, the GelMA solution was dialyzed against deionized water using 12–14 kDa cutoff dialysis tubes (Spectrum Laboratories) for 6 days at 50°C to remove unreacted MA and additional by-products. GelMA was lyophilized and stored at −20°C until use.

¹H-NMR (Bruker AVANCE AQS 600 MHz) operating at 600.13 MHz was used to determine the DS of the synthesized GelMA according to a previously published work (Ovsianikov et al., 2011). Briefly, GelMA was dissolved in warm D₂O, and the ¹H-NMR spectrum was acquired when the sample reached an equilibrium temperature of 40°C. The DS could be calculated by applying the following formula: DS(%)=MolVal,Leu,Ile×I5.7ppmI1.2ppm×100MolLys,Hly where Mol_Val,Leu,Ile is the total number of moles of valine, leucine, and isoleucine amino acids present in gelatin; Mol_Lys,Hly is the total number of moles of lysine and hydroxylysine present in gelatin; I_1.2 ppm is the integration of the peak ascribed to the resonance of hydrophobic alkyl side chains of valine, leucine and isoleucine while I_5.7 ppm is the integration of the peak ascribed to the vinyl proton of methacrylamide substituent groups. On the basis of the known amino acid composition of gelatin, the total number of moles for each amino acid could be calculated.

Soft molds were fabricated in PDMS by replica molding of a PMMA (Plexiglas) master produced by CNC milling. Briefly, we designed the model of the mold with a CAD software (AutoCAD, Autodesk). The mold was composed of three microchannels with the same length (L = 16.5 mm) but different cross sections: 2,000 μm × 2,000 μm, 1,000 μm × 1,000 μm, and 500 μm × 500 μm. To enhance the mechanical stability and to facilitate the handling of the hydrogels, we added an additional U-shaped channel as to create a frame to the three microchannels. Then, we imported the 3D model in a CAM software (SolidCAM, Dassault Systèmes) and generated a tool path code to obtain a positive master, which was precision-milled out of a 5-mm-thick Plexiglas sheet using a CNC milling machine (Proxxon MF70 with Mach3 software, Newfangled Solutions). Finally, PDMS (Sylgard 184, Dow Corning, 10:1 pre-polymer to catalyst ratio) was poured on top of the Plexiglas master, degassed, and cross-linked at 60°C for 3 h. Prior to use, PDMS molds were sterilized by thorough washing in pure ethanol (99.8%) followed by washing in sterile PBS.

Gels were characterized in terms of their compressive stiffness under unconfined compression using an Instron 3365 universal tester (Norwood, MA, USA). Cylindrical specimens (8 mm in diameter, 4 mm in height) of formulated GelMA were tested at room temperature (RT) up to 50% final strain with the following parameters: 0.01 N preload force, 0.125/min strain rate. Values for the compressive modulus were calculated from the initial linear region (0–20% of strain) of the obtained stress–strain curves. Each measurement was performed in triplicate and results are reported as the mean ± SD.

Cells encapsulated within GelMA hydrogels were fixed in PFA 2% and processed for fluorescence immunocytochemistry as previously described (Scardigli et al., 2008). Briefly, samples were re-hydrated with PBS and cells were permeabilized with 0.2% v/v Triton X-100 in PBS for 30 min. Subsequently, samples were incubated with mouse anti-myosin heavy chain (MHC, Clone MF20 DSHB, 1:2) primary antibody diluted in blocking buffer (0.2% Triton X-100 and 20% heat-inactivated goat serum in PBS) for 20 min at RT. After several washes with buffer, sections were incubated with anti-mouse FITC (Chemicon, 1:500) secondary antibody. Samples were counterstained with DAPI to detect nuclei, washed three times with wash buffer, and mounted on Vectashield antifade mounting medium (Vector Laboratories). Samples were imaged with a Nikon Eclipse 2000-TE microscope equipped with a CoolSNAP MYO CCD camera (Photometrix) and MetaMorph software.

Image analysis was performed using ImageJ software. The distributions of myotube orientation within the hydrogel structures following in vitro culture was obtained by analyzing IF micrographs (MHC signal) with OrientationJ plugin (Püspöki et al., 2016). The plugin evaluates the orientation for every pixel of the image based on the structure tensor and provides an orientation distribution plot as an output.

One of the main features that influences the network density and stiffness of GelMA hydrogels is the DS of ε-amino groups of lysine and hydroxylysine with methacrylic moieties. Despite the complexity of the ¹H-NMR spectrum of GelMA, an accurate and quantitative determination of DS can be carried out through a ¹H-NMR-based relative quantitation method. This method can be easily applied to GelMA since its amino acid composition is well known (Eastoe, 1955). The peak around 1.2 ppm (Figure 1B) can be ascribed to the resonance of the hydrophobic alkyl side chains of valine (Val), leucine (Leu), and isoleucine (Ile). These hydrophobic side chains can be considered chemically inert and thus they do not participate in the reaction with MA. On the other hand, the peaks at 5.7 and 6.0 ppm can be ascribed to the vinyl proton of methacrylamide substituent groups of lysine and hydroxylysine amino acids. By using these two peaks as reference peaks and by applying the following formula (see Section “Materials and Methods” for more details): DS(%)=MolVal,Leu,Ile×I5.7ppmI1.2ppm×100MolLys,Hly=0.384×103.49/21000.02×1000.0385=51.6% a DS of 51.6% for the synthesized GelMA was obtained.

Following its synthesis, purification and characterization, GelMA was used for the preparation of a set of hydrogels. Since our first goal was to evaluate the influence of hydrogel stiffness on myoblast differentiation, before encapsulating the myogenic precursor, we explored two different variables that influence hydrogel stiffness: (i) the concentration of GelMA in the precursor hydrogel solutions and (ii) the UV crosslinking time. First, we performed few experiments to determine the lowest limits for the two parameters—i.e., the minimum concentration and the shortest UV cross-linking time that would eventually give rise to a gel. It turned out that for concentrations below 3% w/v—independently of the UV-light cross-linking time—it was not possible to obtain a stable hydrogel. On the contrary, for a GelMA concentration of 3% w/v, it was possible to obtain stable hydrogels for a minimum cross-linking time of 4 min (UV dose ≈ 310 mJ/cm²). These values of GelMA concentration and UV light cross-linking time were thus taken as lowest levels for the two parameters. Then, we prepared other three solutions with GelMA concentrations up to 8% w/v—namely 4, 6, and 8% w/v. The four GelMA solutions were UV cross-linked for either 4 or 5 min, and the resulting hydrogels were then mechanically tested to determine their compressive behavior. The stress–strain plots for the formulated hydrogels are reported in Figure 2, while Table 1 summarizes the values of compressive modulus calculated from the stress–strain data. As it can be noticed, both GelMA concentration and UV cross-linking time have an impact on the compressive modulus, which increases with increasing either of the two parameters. Since it is well known that myoblasts, when cultured on top of hydrogel systems, sense the substrate stiffness and, as demonstrated in previously published works (Engler et al., 2004), undergo a better differentiation on substrates having a stiffness around 8 ÷ 12 kPa, we decided not to considerably exceed this range and thus we did not prepare hydrogels with GelMA concentration higher than 8% w/v. Furthermore, since the effect of UV irradiation time for a given composition (% GelMA) was minor, we decided to limit cell experiments to the 4-min irradiation group, as to keep the UV dose to cells as low as possible.

Cell-laden hydrogels were timely observed by phase contrast microscopy up to 14 days, and the results are summarized in Figure 3. At lower concentrations (3–4% w/v), C2C12 rapidly spread within the hydrogel matrices revealing the formation of short myotubes already 48 h after polymerization, whereas at higher concentrations, cells presented limited spreading (6% w/v) or even almost exclusively round shape (8% w/v). At later time points—7/14 days after polymerization—C2C12 encapsulated into 3% GelMA hydrogel displayed a remarkable amount of myotubes forming a 3D-entangled network homogeneously distributed in the hydrogel volume. Similarly, 4% GelMA hydrogel supported a pronounced formation of myotubes, comparable to the one obtained with 3% GelMA, albeit with a slightly less homogeneous distribution. On the contrary, at higher GelMA concentrations (6–8%), C2C12 presented a remarkable detriment in myogenic differentiation, with minor myotube formation generally localized in clusters.

The first propaedeutic experiment evidenced a clear correlation between matrix stiffness and the ability of myoblasts to undergo efficiently myogenesis. Results showed that the best candidates were those at lower GelMA concentration (3 and 4%)—i.e., with lower compressive moduli. Despite the performances of 3% GelMA were slightly better, we decided to use 4% GelMA as these hydrogels were easier to handle. After determining the most suitable GelMA concentration for myogenic differentiation, a micromolding approach was established to investigate myotube organization under different confinement regimes, as sketched in Figure 4. The mold was designed to produce cell-laden thin hydrogel strings at different cross sections—namely 2,000 μm × 2,000 μm, 1,000 μm × 1,000 μm, and 500 μm × 500 μm—connected to a U-shaped supporting frame, further reinforced by a modeled glass capillary at its core. This solution endowed the structure with enhanced rigidity to support the ensuing myostructures and restraining hydrogel contraction due to myotube tensions.

C2C12 cells were successfully retained within the three different hydrogel strings as confirmed by phase contrast micrographs showing a regular round shape (Figures 5A–C). After 3 days of culture, the structures revealed a remarkable compaction (Figures 5D–F) most likely due to C2C12 differentiation and myotube formation, particularly highlighted in the hydrogel with the smallest cross section (arrow in Figure 5F). Fourteen days after cell encapsulation, IF analyses were performed to evaluate the organization of MHC positive myotubes (Figures 5G–L). Micrographs showed a considerable parallel orientation for the myostructure cultured in the smallest (500 μm × 500 μm) and middle (1,000 μm × 1,000 μm) hydrogel structures (Figures 5H,I), while in the largest ones (2,000 μm × 2,000 μm), parallel organization was just partially achieved (Figure 5G), revealing the significant importance of proper geometrical confinement of myogenic precursors during the SM differentiation process. This can be better appreciated from the myotube orientation distribution plots (Figures 5M–O). In all cases, myotube orientation distribution was peaked along the hydrogel string axis. However, by reducing the hydrogel cross section, the distribution plots appeared increasingly sharp (Figures 5N,O), confirming that the geometrical confinement plays a significant role in myotube alignment.