hiPSC-derived 3D Bioprinted Skeletal Muscle Tissue Implants Regenerate Skeletal Muscle Following Volumetric Muscle Loss

Engineering of biomimetic tissue implants provides an opportunity for repairing volumetric muscle loss (VML), beyond a tissue’s innate repair capacity. Here, we present thick, suturable, and pre-vascularized 3D muscle implants containing human induced pluripotent stem cell-derived myogenic precursor cells (hiPSC-MPCs), which can differentiate into skeletal muscle cells while maintaining a self-renewing pool. The formation of contractile myotubes and millimeter-long bers from hiPSC-MPCs is achieved in chemically, mechanically, and structurally tailored extracellular matrix-based hydrogels, which can serve as scaffolds to ultimately organize the linear fusion of myoblasts. Embedded multi-material bioprinting is used to deposit complex patterns of perfusable vasculatures and aligned hiPSC-MPC channels within an endomysium-like supporting gel to recapitulate muscle architectural integrity in a facile yet highly rapid manner. Moreover, we demonstrate successful graft-host integration and de novo muscle formation upon in vivo implantation of pre-vascularized constructs within a VML model. This work pioneers the engineering of large pre-vascularized hiPSC-derived muscle tissues toward next generation VML regenerative therapies.


Introduction
Volumetric muscle loss (VML), often caused by traumatic events, involves the en bloc loss of skeletal muscle and results in poor muscle function. These injuries often overwhelm the intrinsic ability for muscle to repair itself and instead lead to extensive brosis. 1,2 Functional muscle transfer, which involves the transplantation of vascularized muscle from a donor, is a standard approach for VML treatment. 3 However, shortages of donor tissue, donor-site morbidity, and suboptimal reconstitution of muscle defects impair functional recovery. [3][4][5] Myogenic implants obviate some of these concerns, but also fail to fully reconstitute muscle function following VML injury. Impaired muscle regenerative capacity in implant aps has been linked to the lack of engraftable muscle stem (i.e., satellite) cells and the limited functional differentiation of these cells into de novo myo bers. 6 As such, developing clinically relevant, functional tissue implants using autologous cells, which are capable of muscle regeneration, has emerged as a potential solution.
Bioengineering clinically relevant muscle tissue implants with the capacity for muscle regeneration requires identifying and embedding skeletal muscle stem cells, as well as controlling the cellular alignment and differentiation within the engineered 3D scaffolds to reconstitute the architecture of skeletal muscle. 3D bioprinted skeletal muscle constructs have shown increased myo ber differentiation and promising in vivo results. 7 However, the majority of these constructs rely on primary myogenic (progenitor) cells that possess limited proliferation capacity and lose their myogenic potential during serial in vitro passaging, 8 thereby hampering the clinical translation of VML therapies. In contrast, human induced pluripotent stem cells (hiPSCs) exhibit excellent proliferation potential, 9,10 and are capable of being differentiated into myogenic (progenitor) cells and regenerating skeletal muscle, similar to satellite cells. 11,12 The striking ability of hiPSC-derived muscle precursor cells (MPCs) to self-organize by forming primary and secondary muscle bers in a 2D in vitro culture system is similar to the muscle organization process observed in vivo. 13 To implement this ability of hiPSC-MPCs in 3D implants, a biomaterial must provide adequate stem cell niches that guide myogenic differentiation and maturation by providing the right instructive cues and inducing cell alignment. 14- 16 Furthermore, functional vascular networks should be interwoven into the volumetric implants to minimize diffusion limitations and to serve as guides for the ingrowth of host endothelial cells (ECs) and hematopoietic cells, thereby achieving timely angiogenesis and preventing necrosis in the core of thick (> 1 mm) engineered tissues. 17,18 The rapid vascularization of volumetric muscle constructs based on hiPSC-MPCs is of paramount importance for the cells' survival, myogenic differentiation, and maturation into functional myo bers, as these cells are metabolically active. [19][20][21][22] A common approach for building 3D implants has been the mold fabrication of hydrogel encapsulated isogenic 3D hiPSC-derived arti cial skeletal muscle constructs, which gives limited control over architecture. 23 Here, we propose a strategy of combining 3D bioprinting and hiPSC-MPC culture technologies to engineer pre-vascularized muscle tissue implants that emulate the intricate structural complexity and multi-dimensional hierarchy of native muscle, thus forming the basis for a new generation of VML repair strategies ( Figure 1). An embedded multi-material bioprinting (EMB) paradigm was selected, enabling the rapid and high-resolution bioprinting of aligned muscle bers and vascular free-forms within a liquid support matrix which could be solidi ed on demand. [24][25][26] A low viscosity and thermally reversible gelatin-methacryloyl (GelMA)-based bioink was used, as it would mitigate the lethal shear stresses on cells that typically induce cell membrane damage or even cell death during the rapid extrusion of bioinks through nozzles. [27][28][29][30][31] To create perfusable channels, a sacri cial gelatin bioink was printed between the engineered muscle bers and then seeded with human umbilical vein endothelial cells (HUVECs). To study the levels of anastomosis and graft-host integration on a cellular level, the prevascularized constructs were subcutaneously implanted in a murine model. Furthermore, a nano brous suturable scaffold was integrated onto the printed construct to achieve a prolonged retention of the implants in the injury site. Finally, we established an in vivo VML repair proof-of-concept by implanting the hiPSC-MPC engineered muscle ber bundles within a VML-injured mouse model. We thereby introduce a bioprinting strategy to e ciently engineer clinically relevant skeletal muscle tissues comprising millimeter-long striated and contractile muscle bers from differentiated and matured hiPSC-MPCs.

Results And Discussion
EMB for fabricating muscle ber bundle architectures The EMB strategy consists of directly printing a hiPSC-MCPs-laden bioink within a pre-gelled GelMA supporting matrix, unlike conventional embedded printing methods in which inks are typically printed within microgels and slurries ( Figure 1). 27,28 To form millimeter-long myo bers, hiPSC-MPCs must be anchored via abundant Arginyl-glycyl-aspartic acid (RGD) binding sites on the engineered ECM matrix, which ultimately organize the linear fusion of myoblasts, similar to the native ECM. 32,33 The bioink for muscle bers consists of gelatin and GelMA, both of which contain denatured collagen, a major ECM component of skeletal muscle tissue with abundant RGD binding sites. By adjusting the concentration of GelMA and the crosslinking density, we can easily tune the porosity and degradation of the hydrogels 34 in order to achieve adequate mechanical support and degradation behaviors for the engineering of myo bers with hiPSC-MPCs. The role of the GelMA supporting matrix is comparable to the hierarchical ECM structure of the native endomysium, which is mainly composed of type I and type III collagens. 35 This surrounding connective tissue physically supports densely bound aligned myo ber bundles and perfusable vessels, 36 preventing structural collapse during muscle contractions and movements. In addition, the incorporation of GelMA pre-polymers in both the supporting matrix and the bioink prevents the delamination of the bioprinted layers upon photo-crosslinking.
The thermally reversible gelation properties of both gelatin and GelMA enable a straightforward temperature-controlled printing process. 37,38 Complex multi-layered patterns can be fabricated by harnessing the reversible thermal-healing properties of the GelMA supporting matrix just below the gelation temperature (between ~8 °C and ~20 °C, sol-gel state) (Figure 2 and S1). To achieve an integrated crack-free embedded structure, ad hoc heating of the bioink-containing nozzle (up to 37 °C) can be used to partially melt the interfacing areas of the sol-gel supporting matrix ( Figure S2). This thermal-healing behavior was corroborated by visibly reduced shape delity and the formation of microsized cracks when printing at a lower support matrix temperature (~6 °C). To maintain shape delity, the bioink stiffness could be increased through increasing polymer concentration (Figure 2b, c). However, higher mechanical stiffness is a tradeoff that reduces biocompatibility with soft tissues and laden cells.
Alternatively, gelatin can be added to the GelMA bioink to temporarily increase its stiffness and viscosity for bioprinting followed by a post-crosslinking removal step at ~37 o C, 39 creating non-merged parallel channels ( Figure 2d). By adjusting ink ow rates and nozzle translational speed, the diameter of the bioprinted channels could be tuned (160±8 to 220±16 µm) ( Figure S3). Changing the center-to-center inter-channel distance did not signi cantly affect the shape delity and the distance could be reduced to the length of almost one channel diameter before the channels started to merge ( Figure S4). This distance is comparable to the diffusion limit of the native skeletal muscle environment and thus helps with the engineering of large viable tissues. 40 To obtain aligned channels resembling muscle architecture, we printed vertically aligned micro bers within the supporting matrix with this minimal spacing, thus increasing the density of the muscle tissue. We set out to use a photo-crosslinking strategy to transform the embedding bath, including the printed patterns, into a solid hydrogel scaffold that was robust enough to enable in vitro culture, in vivo implantation, and ultimately, muscle tissue repair (Figure 2e,f). To this end, various concentrations of GelMA and photo-crosslinking conditions (i.e., UV exposure time) were assessed to obtain a solid construct with mechanical properties optimized for muscle stem cell differentiation and maturation. This screening was also based on our previous studies that focused on identifying the necessary mechanical properties for a hydrogel to mimic the native skeletal tissue ECM (i.e., 10-50 kPa and >30% elastic strain). 41 Using a 5% (w/v) GelMA matrix, we investigated stiffness as a function of UV exposure duration (Figure 2g and S5a). The optimum UV condition (55 s, 5.5 J/cm 2 ), with a matrix stiffness of ~8 kPa after 21 days of in vitro culture ( Figure S5b), resulted in a stable construct that prevented the collapse of the printed patterned geometries during long-term tissue culture.
The EMB of the recurring micropatterns was demonstrated by rapidly printing various longitudinally aligned patterns with high printing density within a relatively large-sized supporting bath (1 cm 3 ), thus abstractly representing the vessels and muscle bers in a native muscle ber bundle (Figure 2h, i and S6).
Longitudinal patterns could also be combined with the printing of transversal patterns (Figure 2j). In addition, coaxial bioinks could be easily incorporated into the EMB technique to fabricate patterned arrays of core-shell micro bers in the supporting bath ( Figure 2k). The thicknesses of the core and sheath could be tuned by adjusting the diameter of the coaxial nozzles ( Figure S7). Complex 3D shapes such as fractal-like branched networks containing aligned micro bers could also be printed in this supporting gel to recapitulate the architecture of four different types of natural vascularized skeletal muscle aps ( Figure 2l-o, < 5 min print time per construct). The proposed EMB technology can facilitate the fabrication of large-sized vascularized tissue implants through modulating the size and geometry of the supporting matrix. Alternatively, centimeter-scale building blocks could be assembled into even larger constructs using 3% (w/v) GelMA as UV-curable glue. Gluing together nine uniquely patterned building blocks readily resulted in the formation of large, complex, and perfusable 3D constructs with volumes of almost 10 cm 3 ( Figure 2p).

Perfusable construct engineering
To endow engineered tissue constructs with perfusable vascular-like networks, a thermo-reversible gelatin ink was formulated, which could be removed by warming up the photocured GelMA supporting matrix to 37 °C in PBS or cell culture media ( Figure 3a). The method readily enabled the fabrication of a bifurcating channel network within a ~1 cm 3 supporting matrix ( Figure 3b) and enabled the controlled screening of bioengineering parameters such as vessel diameter, number, and density. Speci cally, printing 5% (w/v) gelatin within a 5% GelMA matrix at 2 mm/s formed open and stable channels (diameter = 336.5±34.9 µm) (Figure 3c). Although the fabricated channels were larger than the native blood vessels in skeletal muscle, we anticipated that the addition of such perfusable vessels inside a large implant could accelerate the host's blood invasion post implantation, facilitate the formation of micro-capillaries, and ultimately, promote implant integration. Scanning electron microscopy (SEM) imaging of freeze-dried constructs con rmed printed channel patency (Figure 3d-f), which was corroborated by successfully perfusing the channels with uorophores that also represented the perfusion and diffusion of nutrients, growth factors, and waste products in these constructs (Figure 3g,h and Movie S1). Time-lapse confocal uorescence imaging revealed a uniform diffusion gradient of the uorescent-labeled dextran (~20 kDa) within the supporting matrix, which was characterized by a relatively high diffusional permeability (P d = 7.25 ± 0.675 x 10 -5 cm s -1 ), thus favoring the exchange of cellular nutrients and waste products and corresponding to previously reported values (Figure 3i,j). 42,43 In order to explore the vascularization of the perfusable channels, we incorporated HUVECs into the constructs (Figure 3k). In two weeks, the HUVECs proliferated and migrated along the channel wall, thereby forming a tubular cellular network that covered the entire channel's interior throughout the construct (Figure 3l-n). The complete endothelialization of the perfusable channels was con rmed by the positive staining of platelet endothelial cell adhesion molecules (PECAM/CD31), a biomarker found in the intercellular junctions of ECs. To facilitate the formation of vascular networks throughout the construct, HUVECs were also embedded in the crosslinkable GelMA supporting matrix. In some areas of these constructs, small CD31-positive sprouts connected to the endothelialized perfusable channels were found (Figure 3o,p). It was observed that HUVECs embedded in 5% (w/v) GelMA started to proliferate after three days of in vitro culture, while HUVECs within 7% (w/v) GelMA did not ( Figure S8). Furthermore, we observed that embedding HUEVCs in the supporting bath can promote the diffusion of biomolecules across the thick bath ( Figure S9), potentially due to the increased porosity. HUVEC-laden baths can therefore compensate for the delayed biomolecule diffusion from the endothelial-seeded perfusable channels into the supporting matrix, an observation previously shown to occur due to the formation of vascular junctions and endothelial barrier layers. 44,45 hiPSC-based muscle ber bundle engineering Next, we set out to explore the optimal bioink conditions for hiPSC-MPCs to mature and differentiate inside perfusable bioprinted constructs. Among the printable gel mixtures inside a 5% GelMA supporting matrix (previously discussed in Figure 2d), 7.5% (w/v) GelMA bioink (stiffness = 40±8 kPa, further reduced to 22±2 kPa after 3 weeks of incubation in standard culture medium) provided a 3D microenvironment that was mechanically ideal for the myogenic induction of hiPSC-MPCs (i.e., 10-50 kPa and >30% elastic strain) ( Figure S10). 41, 46 To this end, 3-to 4-week old hiPSC-derived progenitor cells were obtained following our previously described serum-free differentiation protocol ( Figure S11) 11,47 and were encapsulated in a bulk supporting matrix or bioink for 3D culture studies. Further steps of myogenic induction were carried out by delivering differentiation factors to the hiPSC-MPC-laden constructs. Initially, we encapsulated hiPSC-MPCs in 7.5% (w/v) GelMA ( Figure 4a) and cultured the constructs for 4 weeks to investigate cellular behaviors modulated by UV crosslinking and cell density ( Figure S12). 48-50 hiPSC-MPCs exhibited increased cell spreading and elongation with shorter UV exposure times and higher cell concentrations. The 7.5% (w/v) GelMA hydrogels were most compatible with the long-term 3D culture of hiPSC-MPCs and the development of striated myo bers with wellorganized sarcomeres, while also containing a pool of satellite cells, indicated by positive Actn2, Pax7, MyoG, Titin, and MyHC staining (Figure 4b-e and S13). Additionally, spontaneous contractile activity was observed in encapsulated hiPSC-MPCs in 7.5% (w/v) GelMA after 15 days of in vitro culture (Figure 4f and Movie S2).
The optimized muscle bioink was used in conjunction with the optimized EMB to obtain vertically aligned hiPSC-MPC-laden micro bers at a density of ~7 bers/mm 2 . (Figure 4g). Speci cally, a pattern of straight bers (diameter = 184.4±22.3 µm) containing hiPSC-MPCs (20×10 6 cell/mL) in gelatin/GelMA bioink was printed. The differentiated cells remained viable for at least four weeks of in vitro culture (Figure 4h and S14). We successfully fabricated striated, millimeter-long muscle bers within the printed micro bers using hiPSC-MPCs (Figure 4i). These results highlight the capacity of hiPSC-laden printed bers to create mesoscale functional muscle tissue. However, when printing the optimized hiPSC-laden bioink within larger constructs (i.e., 5 x 5 x 5 mm 3 ) without perfusion channels, the cells expressed notably less satellite and mature muscle differentiation markers (Pax7, MyoG, and MyHC) after three weeks, as compared to the same cells cultured in smaller hydrogel constructs ( Figure 4j vs Figure 4b,c). It was hypothesized that the large non-perfused printed constructs (i.e., 5 x 5 x 5 mm 3 ) hampered the diffusion of nutrients and waste products to sustain hiPSC-MPC maturation and differentiation. Indeed, comparing the viability and metabolic activity of C2C12 myoblast cells in perfusable vs non-perfusable GelMA constructs corroborated this hypothesis ( Figure S15).
Perfusable channels were printed at various distances and in parallel with the hiPSC-MPC-laden bers ( Figure S16) to de ne the maximum perfusion distance for maintaining high cell viability (~450 µm). Ã 1:4 channel-to-ber ratio rescued the engineered myogenic phenotype, as con rmed by increased biomarker expression (Figure 4k and S17). hiPSC-MPCs in both perfused and non-perfused constructs demonstrated high cell growth, alignment, and spatial con nement ( Figure 4j,k (F-actin row)). A fast Fourier transform (FFT) was performed to assess the degree of myo ber alignment in both conditions. Interestingly, the alignment index of hiPSC-MPCs printed in constructs with and without perfusable channels after 10 days in vitro was not signi cantly different ( Figure 4l). However, we observed differences in the clustered alignment of myo bers between the two groups ( Figure S18). In non-perfused constructs, three angle peaks were observed (-45º, 0º, and 50º), indicating relatively scattered myo ber alignment along the printed patterns. In contrast, a single peak at approximately 0º was observed in perfused constructs, as well as an average alignment index of 1.5, indicating improved myo ber alignment and fusion along the printed pattern. This data supports the formation of a stronger myo ber fusion along a single axis, which con rms improved muscle ber maturation in perfused patterned constructs.
The renewal of satellite cells and the differentiation to newly formed myo bers follows three stages, beginning with high Pax7 but negligible MyoD expression in the quiescent satellite cells. 51 Upon muscle injury, the transient in ammatory response, and speci cally the recruitment of M2 macrophages, induces the activation and proliferation of satellite cells, marked by an increased expression of MyoD and continued Pax7 expression. Finally, as satellite cells differentiate into myoblasts and fuse to form new myo bers, Pax7 expression declines followed by an increased expression of mature muscle markers including MyoG (a characteristic indication of intermediate myogenic muscle maturation) and MyHC. In our non-perfused in vitro engineered muscle constructs, we observed reduced Pax7 expression from day 10 to 20 ( Figure 4m), arrested or delayed myogenic differentiation marked by moderate MyoG (Figure 4n), and reduced MyHC expression compared to the perfused constructs ( Figure 4o). This could be attributed to the ischemic necrosis caused by ine cient transfer of nutrients and oxygen in these constructs. The expression dynamics in perfused engineered constructs con rmed this hypothesis, as Pax7 expression sharply increased between days 10 and 20 (Figure 4m), indicating high levels of satellite cell activation and proliferation. Moreover, the e cient vascularization and nutrient transfer in the perfused constructs led to increased MyHC expression ( Figure S19) and a ~5-10 fold higher fusion index (Figure 4o Figure S20), while the expression of an earlier marker, MyoG, was decreased tõ 20%. Detailed SEM analysis of acellular printed muscle bioink revealed an aligned and nano brous structure, which might have contributed to the successful con nement, alignment, and maturation of hiPSC-MPCs ( Figure 4r). 52 Interestingly, the pore size of the engineered endomysium (i.e., supporting matrix) was much bigger than that of the muscle bioink. Together, these results indicate that the incorporation of perfusable channels can facilitate the diffusion of nutrients and waste products, thereby enabling the engineering of relatively large functional tissues that are required for VML repair. Furthermore, the abundant presence of Pax7-positive muscle stem cells, which provide long-term regenerative capacity, is an important feat, one which is also present in newly regenerated skeletal muscle tissue. 53 To predict and potentially prevent the damaging of in vivo implanted constructs, which undergo incessant physical stress and friction at the implant-host tissue interface during body movements, we studied the mechanical properties of the printed constructs. The constructs that contained printed bers and channels remained intact during 30 days of in vitro culture, and the stiffness and elongation at break were not signi cantly different than those of the bulk supporting matrix without printed channels and bers (i.e., non-printed construct) ( Figure S21). Furthermore, the printed constructs showed persistent elastic behavior during cyclic mechanical loading up to 40% strain and did not demonstrate any fatigue (Figure 4t), a behavior similar to native skeletal muscle that tolerates cyclic contraction-relaxation stress with excellent elasticity during muscle activity. 54, 55 These results indicate that the mechanical properties of the printed constructs could aid in preventing the collapse of perfusable vessels and could physically support the aligned myo bers against cyclic stresses during the regeneration process after in vivo implantation.
Pre-vascularized grafts facilitate in vivo cell survival and successfully integrate with host We explored the subcutaneous in vivo grafting of the vascularized printed constructs with different designs ( Figure 5). The rst implant design ( Figure 5a) contained a bifurcating network coated with HUVECs (~1 cm 3 ), and exhibited angiogenesis stimulation after 3 weeks along with a substantial number of de novo lumens (white arrows) throughout the thick construct ( Figure 5b). HUVEC-laden bulk GelMA constructs (i.e., non-printed) of the same size ( Figure 5c) showed fewer vascular lumens in the areas where the hydrogel interfaced with the native tissue ( Figure 5d). These ndings suggest that the formation of micro-vessels in our hydrogels was triggered by the implanted cells, and not only through the host's blood invasion. This nding was previously observed in organized human EC constructs patterned into cords via microfabrication which acted as templates for guided vascularization upon implantation in vivo. 56 Next, to quantitatively investigate the impact of 3D patterned perfusion and pre-endothelialization on both printed constructs and host cell in ltration, four groups (G1-4) were prepared: 3D bioprinted perfusable channels seeded with HUVECs were printed within a supporting matrix of acellular (G1) (Figure 5e,f) or HUVEC-laden (G2) (Figure 5g,h) constructs. As control groups, 3D bioprinted nonendothelialized perfusable channels in supporting matrix constructs (G3), and non-printed bulk HUVECladen supporting matrix (G4) constructs were prepared ( Figure S22). After one week of in vivo culture, the perfusable channels (G1-3) were in ltrated by red blood cells and showed clear signs of endothelialization (Figure 5e-h and Figure S23). In these constructs, relatively fast in ltration and a high accumulation of murine vascular networks near the graft-host interface were observed. Together with the observation that the scaffolds completely integrated with the host tissue ( Figure S23), this data con rmed that the engineered muscle constructs were biocompatible. Both murine CD31-positive cells (i.e., from the host) and human CD31-positive cells (i.e., from the implanted HUVECs) were observed inside the perfusable constructs (Figure 5f(i), h(i)). Moreover, embedding HUVECs in the engineered endomysium-like matrix promoted sprouting near the edges of the printed channels (Figure 5h). At the surrounding tissue site, the migration and outgrowth of transplanted HUVECs was found (Figure 5f (iii)).
In the control groups ( Figure S22 a(iii) and b(iii)), murine ECs were mostly found in the surrounding tissues. In G2 speci cally, several vessels consisting of concentrically patterned murine and human endothelial cells were observed (Figure 5h (iii)), corroborating with the excellent graft-host integration.
At the three-week harvest timepoint, the perfusable constructs were overgrown by vasculature, including within the channels, indicating anastomosis with the host (Figure 5f (ii) and S22). Human and murinepositive CD31 cells inside G1 and G2 demonstrated full in ltration of the murine vascular network into the proliferative HUVEC networks (Figure 5f (ii) and 5h (ii)). In the tissue surrounding G2, a prevalent presence of migrated HUVECs from the pre-vascularized channels to the host-tissue interface was observed in comparison with G1, possibly due to larger HUVEC density and the formation of sprouting networks at a higher rate (Figure 5f (iv) and 5h (iv)). G3 and G4 constructs did not show such clear grafthost integration ( Figure S22).
G2 exhibited the highest degradation rate and the smallest remaining hydrogel area after harvest ( Figure  5i). Presumably due to their perfusable geometry, G2 constructs allowed for accelerated biomaterial degradation through the accumulation of host cytokines including TNF-α, which would trigger foreign body response and an increased expression of metalloproteinase1/3 (MMP1/3), leading to gelatin degradation. 57 We compared the level of construct grafting temporally and spatially by quantifying human and murine CD31-positive vessels within G1 and G2 perfusable channels and the surrounding tissue (Figure 5j,k). Although initially (week 1) lower, the number of murine vessels in the channels were increased after 3 weeks to match the number of human vessels (Figure 5j), which in turn increased by ~5-10 fold over time. The total number of vessels inside the perfusable channels increased by ~2.5 fold over 3 weeks (Figure 5k), indicating a gradual formation of vascular networks between the implanted and host's ECs, similar to previous studies. [58][59][60] The gradual decline in the number of vessels inside the tissue surrounding G2 was pertained to the pre-existence of ECs and vascular networks embedded inside the supporting matrix, which had induced a much higher number of vessels in the hydrogel-host interface than the channels at the initial stages of in ltration. Moreover, the presence of perfusable channels signi cantly increased the survival of cells in the transplanted constructs ( Figure 5l). Overall, the perfusable bioprinted constructs exhibited excellent biocompatibility and interaction with the host vasculature, and pre-endothelized channels and supporting matrices showed signi cant impact on host in ltration, and implant integration.
Pre-vascularized hiPSC-MPC laden constructs enhance engraftment in volumetric muscle injury model Finally, we aimed to prove the concept of repairing VML in the quadriceps femoris muscle of a rat. To prevent the sample dislocation and to facilitate integration with the host, we integrated the implants with suturable nano brous scaffolds made of a mixture of polyglycerol sebacate and polycaprolactone (PGS/PCL) (Figure 6a). Previously, we demonstrated the biocompatibility and biodegradability of these electrospun scaffolds for in vivo wound repair and stimuli-responsive drug delivery. 61 Here, we used a photo-crosslinkable glue composed of 5% (w/v) GelMA pre-polymer solution to attach the 3D bioprinted constructs to the scaffolds using UV irradiation (Figure 6b). Successful adhesion was revealed by SEM imaging of the hydrogel-nano brous scaffold interface (Figure 6c,d). The mechanical characterization of the sutured nano brous electrospun scaffolds revealed an elastic modulus of 16.4±5.8 MPa and an ultimate tensile strength (UTS) of 1.5±0.7 MPa, which were comparable to reported values (Figure 6eg). 62 Stretching the composite scaffold did not result in fracture at the suture point, suggesting that the suture could tolerate high mechanical stress and large dynamic movement post-surgery.
To create a VML injury model, a rectangular incision was made into a mouse's quadriceps femoris, as previously reported. 63 Approximately 2x3x6 mm 3 of murine muscle volume (corresponding to 0.19 ± 0.005 % of the mouse's total weight) was removed to represent VML injury models. 64-66 3D vascularized muscle tissue constructs (i.e., V3) glued to PGS/PCL membranes were placed into the muscle wound with the membrane facing up (Figure 6h). Additional groups of acellular GelMA constructs with perfusable channels (i.e., V2) and VML incision with no treatment (i.e., V1) were studied. A layer of brotic capsule was observed around the V3 implant four weeks post implantation (Figure 6i, V3) with proliferation, and later with the migration of host cells into the construct. At the graft-host interface, newly regenerated muscle tissue was identi ed (Figure 6j) by the presence of centered (as opposed to marginal) nuclei with smaller nucleus diameters across several myo bers. 67,68 Moreover, a strong expression of embryonic skeletal muscle myosin heavy chain (MYH3) was found near V3 scaffolds (Figure 6i, MYH3/DAPI) and in myo bers with centered nuclei in both cross-sectional and longitudinal muscle tissues (Figure 6k), suggesting that the incorporation of vascularized hiPSC-MPC laden scaffolds would increase de novo muscle regeneration. V2 scaffolds were signi cantly less incorporated and showed lower de novo muscle formation during the same four-week period (Figure 6i, V2). High proliferation and some MYH3 expression were notable at the graft-host interface, which was higher than the V1 scaffolds that underwent minimal change after six weeks (Figure 6i, V1). The grafting of the implanted hiPSC-derived muscle cells and the stemness of the satellite cells were con rmed by the presence of Pax-positive in the V3 constructs ( Figure 6l) at a relatively higher level than V2 and V1 ( Figure S24a, b(i-ii)) and with numerous Pax-positive nuclei on the edges between the myo bers. Although Pax expression, thus regeneration capacity, was found to be higher in the hydrogel implants (V2-3) than the untreated group (V1), it is unclear whether the Pax-positive cells are of human origin. Further characterization with human speci c markers can help to isolate the regenerative effect of hiPSCs from native muscle regeneration. However, we observed high expression levels of human spectrin and lamin ac (Figure 6m) near the grafthost interface, which suggests that the regenerative potency of the implanted hiPSC-MPC laden constructs contributed to the promoted muscle regeneration in V3.
Our study showed angiogenesis and vascular invasion of the implanted vascularized constructs into the host tissue. As shown in Figure 6n (same cross-sectional area shown in 6j-2) and Figure 6o, human vasculature (expressing CD31) was prevalently found among murine vasculature near the scaffold interface and the surrounding host tissue, as well as in the V2 control group ( Figure S24b(iii)). This is in line with the subcutaneous study ( Figure 5i) where both the human and the murine ECs were observed at the implant site. Consequently, the expression of ECs from both human and murine species at the injury site after four weeks of in vivo culture suggests that the fabricated muscle constructs are capable of homing in host tissue cells. Moreover, strong expression of βIII-tubulin (TUJ1) at the host-scaffold interface ( Figure 6p) and the surrounding tissue (Figure 6q) suggests that the constructs could support integration with host motor or sensory neurons, which, as previously reported, can enhance myocyte fusion and can support the formation of functional neuromuscular junctions. 3 TUJ1 expression was relatively lower in the V2 control group (Figure S24b(iv)). These results were conclusively observed in bilateral scaffold implants in both legs of the mice ( Figure S24c). hiPSC-MPCs secrete vascular endothelial growth factor (VEGF), and cytokines which contribute to angiogenesis and accelerate the host's response to injury. 69, 70 We measured VEGF secretion in a medium supernatant of in vitro cultured hiPSC-MPC-laden 3D bioprinted constructs (Figure 6r). These constructs contained hollow channels but were not coated by HUVECs. We observed a 3-fold increase in VEGF secretion over the course of the second to third week after the induction of hiPSC-MPC differentiation, which contributed to axonal growth, angiogenesis and later muscle regeneration over the course of the in vivo study. [71][72][73][74][75] Overall, the vascularized hiPSC-MPC laden muscle constructs showed increased muscle regeneration, angiogenesis and maintenance of potent stem cells compared to control groups (V2 and V1). Together, these results provide a VML repair strategy based on pre-vascularized large-scale hiPSC-derived skeletal muscle implants.

Conclusion
This work pioneered a combination of biofabrication and stem cell technologies for the engineering of large pre-vascularized muscle ber bundles. Speci cally, we used EMB with sacri cial and photocrosslinkable materials within a self-healing supporting gel matrix, which provided an interesting opportunity to fabricate complex features. The muscle ber bioink was optimized for the 3D culture of iPSC-MPCs, a unique feat which was further supported by the capacity of this bioink to differentiate and mature these cells into skeletal myo ber bundles. Endowing the construct with endothelialized perfusable channels was of vital importance for sustaining the viability and function of muscle (stem) cells in vitro and in vivo, while also promoting graft-host integration. The 3D pre-vascularized tissue construct with hiPSC-MPCs was successfully implanted in a VML-injured animal model and demonstrated signi cantly higher cell survival in pre-vascularized bioprinted constructs compared with bulk constructs. By combining hiPSC-MPCs and a novel variant of EMB technology, this work provides a concrete step towards the engineering of large cell-laden muscle tissue constructs for VML repair strategies.

Embedded multi-material (EMB) bioprinting of 3D constructs
To prepare the supporting matrix, molds were fabricated with polydimethylsiloxane (PDMS) and sterilized using 70% ethanol, overnight UV exposure in the biosafety cabinet, and sterile PBS washing. Next, 0.25 wt% PI and 5 wt% freeze-dried GelMA were mixed in 40% HBSS and 60% hiPSC-MPC proliferation medium and the solution was incubated for 1 hour at 37 °C (GelMA pre-polymer solution). Next, the solution was poured into the mold, and was allowed to gelate (sol-gel state) at 4 °C for ~5-8 minutes.
The muscle bioink was prepared with 7.5% GelMA, 2 wt% gelatin and 0.25% PI dissolved in HBSS and hiPSC-MPC proliferation medium (2:3) and was followed by a 1-hour incubation step at 37 °C. At the time of bioprinting, a pellet of hiPSC-MPCs was uniformly dissolved in 600 µl bioink prepolymer solution to make the bioink, which was then transferred to a 1ml syringe with a printhead needle (30G, blunt, BD). 3D bioprinting was performed using a Cellink Inkredible bioprinter that was calibrated prior to each run with respect to the positional accuracy of the needle and a reference mold in XYZ directions. In the case of dual ink printing, the second nozzle (perfusable channel ink) was also calibrated in reference to the muscle ink nozzle. The muscle ink was extruded at a 4 μl/min rate using an ad hoc syringe pump (New Era Pump Systems). To create perfusable channels, a soluble 5% gelatin ink was prepared and printed inside the bath in the form of highly aligned channels. To ensure the printability of the ink, gelatin was thermally gelated and printed at 19-22 °C. Next, the GelMA construct with bioprinted hiPSC and gelatin channels were exposed to a UV light source (~100 mW cm -2 , OmniCure S2000) to be chemically crosslinked. Constructs were then taken out from the molds, and were washed (PBS) and collected onto a 24-well plate pre-lled with warm hiPSC-MPC proliferation medium, followed by incubation at 37 °C and 5% CO 2 . The media was changed after 1 hour to remove the dissolved gelatin from the constructs. The medium was refreshed on day 1 after printing and regularly every 2 days afterwards, until the differentiation medium was changed, or until xation or implantation of the constructs. By submerging the constructs in medium and incubating the construct at 37°C, the gelatin inside the printed channels gradually dissolved into the solution, leaving a fully hollow channel inside the crosslinked GelMA bath.
For bioprinting the muscle ap geometries, a CAD drawing of each geometry was used to laser-cut and create customized PDMS molds. Next, a 3D gcode was created using the Cellink Heartware software. An appropriate angle of printing was chosen such that the ratio of Z travel to X travel (and Z travel to Y travel) remained larger than 1. The mold was cut open and the nozzle traveled in the Z direction to bioprint the structure. The construct was then bioprinted and crosslinked under UV light using the parameters mentioned above.
To extrude multi-material coaxial laments, a coaxial nozzle was fabricated by adhering concentric 25G inner and 18G outer needles using hot glue. To bioprint the sheath, 5% GelMA and 2% alginate was fed into the outer needle at 16 μl/min. To bioprint the core, 5% GelMA and 2% gelatin supplemented with 0.25% PI and 1% CaCl 2 was fed to the inner needle at 8 μl/min. The core-sheath crosslinking was achieved through the Ca 2+ crosslinking of alginate as well as GelMA covalent gelation upon UV irradiation.

Characterization of mechanical and rheological properties
To perform compression mechanical tests on the constructs (cylindrical samples, diameter = 0.9 cm, height = 0.5 cm), a parallel plate platform was used to compress the samples at room temperature until rupture (ADMET, MTESTQuattro). The linear portion of the resulted stress-strain curve (i.e., <20% strain) was used to calculate the Young's modulus.
Cyclic compression tests were performed and recorded on an Instron 5966 Universal Testing System with 500 N load cell. To ensure symmetry, the constructs were 3D bioprinted in the same cylindrical molds as mentioned above and were cut in half (perpendicular to the bioprinted channels). 2 mm sample cuts showed nearly identical results. Five cycles (0-10% strain) were performed sequentially to ensure good contact and the absence of hysteresis. Then, incremental strain ranges were used to perform sequential cyclic tests (starting at 0% strain, going up to 10%, 20%, 30%, and 40%), returning to 0% strain in between the increments. To ensure reproducibility, the strain incremental measurement was repeated for one sample and an excellent agreement was obtained.
To measure the rheological properties of the bioinks (Anton Paar MCR 302 rheometer), a 10-millimeter diameter concentric cylinder Couette DIN bob with a 10.84 mm diameter cup (~1 mL sample volume) was used for all measurements, giving a geometry gap between the inner and outer cylinders of 0.42 mm.
This geometry was selected to minimize the impact of drying and to increase the surface area to obtain a better signal than possible via a parallel plate or cone and plate geometry. A C-PTD200 module (bottom), chiller (Julabo circulating chiller) and solvent trap attachments (Anton Paar) were used to maintain temperature control. Experimental data recording and analysis were performed with Anton Paar RheoCompass. Temperature sweeps were done between 4 and 37 ºC; all samples were allowed to equilibrate for at least 10 minutes at 4 ºC prior to measurement. For oscillatory temperature ramps, the ramp rate was 1 ºC/min; the strain amplitude and frequency were γ 0 = 0.3% and ω=10 rad/s, respectively, chosen to be in the linear viscoelastic regime (LVE) at 4 ºC determined by amplitude sweeps. The sweep was then reversed to examine rheological hysteresis. Before each sample measurement, calibrations were done to the measuring system inertia of the upper plate, as well as the motor that compensated for residual friction. Next, after gap recalibration, the constructs were introduced to the machine and were allowed to reach an equilibrium before running the test (10 minutes).

Characterization of microporous structures of hydrogels
To perform SEM imaging, the constructs were rst incubated in PBS overnight at 37 ºC. The constructs were then collected and freeze-dried inside 1 ml Eppendorf tubes that were punctured on the cap for air release. Next, the constructs were cut at different horizontal and vertical cross-sections parallel or perpendicular to the bioprinted channels. SEM imaging was performed at the Harvard center for nanoscale systems.

Characterization of perfusable microchannels
To characterize the perfusability of the hollow channels, a perfusion test was performed using FITC-Dextran dye. Brie y, 5% FITC-Dextran powder was dissolved in PBS and covered by aluminum foil. The constructs were rst placed horizontally (channels were oriented parallel to the ground) on a petri dish on a confocal microscope and a stacked image (image 0) was captured from the channels using the bright eld channel and the green uorescent laser. Next, the constructs were gently ipped vertically (channels were now oriented vertically) and 10 µl of the FITC solution was added to the channels from the top. The constructs were immediately ipped over to the initial orientation, and the diffusion was recorded using the time series function of the confocal microscope in a 20-minute span. Stacked images of the green uorescent laser and bright eld were recorded every 10 minutes. The images were then analyzed using ImageJ to calculate uorescence intensity near the channels. To calculate the diffusional permeability, the following formula was used following a previous protocol 45 : Where P d represented the diffusional permeability, l 1 was the average intensity at image 0, l 2 was the average intensity in the later timepoints (e.g. t ~10 min), l b was the background intensity before FITC addition, and d was the diameter of the printer channel (~345 µm). The portion of the intensity curve on the left side of the peak was used to calculate the uorescent intensity.
Cell culture and differentiation Culture of HUVECs: Human Umbilical Vein Endothelial Cells (HUVECs) with prescreened angiogenesis were purchased from Lonza. Cells were cultured in EBM TM -2 Basal medium supplemented with EGM TM -2 pre and post printing.
Derivation and culture of hiPSC-MPCs: The NCRM1 hiPSCs were cultured according to previous procotols. 33 Brie y, Matrigel-coated plates (BD) were used to seed and culture the cells using mTesR1 medium (Stem Cell Technologies). Aggregate and single-cell passaging could both be used. The cells were veri ed free of mycoplasma. hiPSC-derived myogenic progenitors were generated following a serum-free differentiation method. 11,47 Three-to four-week-old wild-type hiPSC-derived primary myogenic populations were detached to replate (cell density = 3.5-4 x 10 4 cm 2 ) myogenic progenitors on Matrigelcoated plates (Corning) and the cells were then cultured in skeletal muscle growth medium (SKGM-2, Lonza) supplemented with 10 μM ROCK inhibitor for 24 hours. Afterwards, the ROCK inhibitor was removed and the culture was continued with SKGM-2 for ~2 days until they were dissociated using Accutase (Stemcell Technologies) and were frozen or prepared for bioprinting.
Culture and differentiation of C2C12s: Mouse C2C12 cells were purchased from ATCC (CRL-1772) and were cultured and maintained following the vendor protocol. To differentiate C2C12s into skeletal muscle cells, the application of proliferation medium (Dulbecco's modi ed Eagle medium (DMEM), 10% FBS, 1% P/S) was stopped and DMEM supplemented with 2% horse serum (Gibco) and 1% P/S was used to induce differentiation and continue culture by refreshing the medium every other day.
Fabrication of cell-laden bulk hydrogels HUVEC-laden GelMA constructs: To fabricate bulk hydrogels containing embedded HUVECs, 5% GelMA bath pre-polymer solution was prepared as explained in the previous sections. Next, HUVECs were trypsinized and the pellet was resuspended in the pre-polymer solution to reach a nal cell density of 7.5 x 10 6 cells/ml. Next, the pre-polymer solution was poured into PDMS molds as explained above and incubated in 4 ºC for 5-8 minutes. The constructs were then crosslinked under UV light (100 mW/cm 2 ) for 55 seconds, and were extracted from the molds, washed with PBS and placed in 24-well plates supplemented with HUVEC culture medium. The medium was refreshed 1 hour after crosslinking, the day after, and after that regularly every 2 days.
iPSC-MPC-laden GelMA: To fabricate bulk hydrogels containing embedded hiPSC-MPCs, 7.5% GelMA bath pre-polymer solution containing 0.25% PI was prepared in a solution containing 40% HBSS and 60% SKGM-2+Ri medium. hiPSC-MPCs were trypsinized and embedded in this pre-polymer solution at a density of 7.5 million cells/ml. Next, the pre-polymer solution containing cells was poured into PDMS molds and incubated at 4 ºC for 5-8 minutes. The constructs were next crosslinked under UV light (100 mW/cm 2 ) for 55 seconds, and were extracted from the molds, washed with PBS and placed in 24-well plates supplemented with SKGM-2+Ri culture medium. The medium was refreshed 1 hour after crosslinking, the day after, and then regularly every 2 days.
Fabrication of 3D vascularized (HUVEC-seeded) constructs 5% soluble gelatin ink was rst prepared according to the soluble ink preparation instructions. Next, cells were trypsinized and centrifuged for 5 minutes at 1000 rpm. Afterwards, HUVECs were added to the ink at a density of 4 x 10 7 cells/ml. Meanwhile, the media of the wells containing constructs was aspirated by pipette. Constructs were placed vertically using a small spatula in 48-well plates. 10 μL of HUVEC-laden bioink (4 x 10 5 cells) were injected two times on top of each construct through the cross section of the hollow channels using a pipette. Constructs were immediately ipped on the side. After adding cells to all constructs, the well plate was returned to the incubator for 45 minutes to allow cells to loosely attach to the wall. At this time, no media was added to the constructs. After this incubation period, the constructs were placed vertically again and another two rounds of 10 μL HUVEC-laden bioink were injected inside all channels in each construct. Constructs were again ipped horizontally and incubated for 45 minutes at 37°C. After this time, constructs were checked under the microscope to con rm the presence of cells inside the channels. HUVEC media was gently added to each construct and the constructs were incubated overnight. The media was changed the next day to avoid possible contaminations. The constructs were replated in new 48-well plates after 2-3 days to avoid consumption of media by the cells that might have migrated to the bottom of the well plate.

Characterization of in vitro cultured samples
Fixation of cell-laden constructs was done at different timepoints (e.g. day 10, day 20, day 30).

Fast Fourier transform (FFT) analysis
To quantify the extent of muscle ber alignment, FFT analysis was performed using a previously described method 77 . Brie y, the images were stored and analyzed as uncompressed .TIFF les and were converted to grayscale 8-bit images. ImageJ was used to conduct 2D FFT analysis on four images per print condition. The obtained image was then analyzed using a MATLAB program we developed to plot the FFT alignment intensity versus angle or orientation in degrees. From each orientation histogram, the angle at which most bers were aligned was determined. Alignment index was calculated as the fraction of cells aligned within 20º of the peak angle. 78 The larger the alignment index was, the higher was the fraction of bers aligned near a peak angle. A completely randomly aligned matrix was indicated by an alignment index of 0.
In vivo subcutaneous study HUVEC-laden constructs cultured for 3-4 weeks were selected for implantation. Four groups of constructs containing HUVECs were fabricated as below: G1 -Endothelialized perfusable channels in the supporting bath: perfusable channels were fabricated in a 5% GelMA supporting bath as described in the previous sections. The channels were later seeded with HUVECs and cultured for 2 weeks before surgery.
G2 -Endothelialized perfusable channels in the HUVEC-laden supporting bath: HUVEC-laden bulk GelMA pre-polymer solution was prepared as previously described. This bath was used to 3D bioprint perfusable channels according to the protocol described previously.
G3 -Perfusable channels in the supporting bath. This group did not contain any cells and was bioprinted as a control group for G1. Brie y, perfusable channels were bioprinted in a 5% GelMA bath as described previously. The constructs were cultured for 2 weeks before surgery.
G4 -HUVEC-laden supporting bath. This group contained a bulk cell-laden hydrogel with no 3D bioprinting involved. The constructs were prepared according to the protocol described above for fabricating bulk hydrogels. The constructs were cultured for 2 weeks before surgery. This group served as a control group for G2, investigating the effect of incorporating bioprinted perfusable channels in the constructs.
All animal procedures were approved by the Institutional Animal Use and Care Committee of Harvard Medical School (Protocol number: 2017N000114). 12-week-old nude rats (Charles River) were acclimatized for a week prior to surgery in the vivarium of Brigham and Women's Hospital. Subcutaneous embedding test was performed on nude rats anesthetized with Iso urane anesthetic. Brie y, subcutaneous space was accessed through dorsal incision with the length of 2 cm. Subsequent subcutaneous pockets were created by blunt separation. Next, three hydrogel constructs were introduced into the subcutaneous pockets on bilateral sides of the cut. Total of 6 constructs were implanted for each group. To alleviate post-surgery pain of rats, carprofen (5 mg/kg BW/day s.c. as a single injection) was postoperatively administered for 72 hours and the rats were monitored hourly for 6 hours post-surgery and twice per day for 5 days. Samples of each group were harvested at two timepoints (weeks 1 and 4).
The implanted gels and surrounding tissue were removed for further staining and studying angiogenic properties. Next, the harvested cell-laden printed constructs were washed with PBS and xed in 4% paraformaldehyde for 24 hours before staining with Haemotoxylin and Eosin (H&E) and human-speci c and rat-speci c CD31 antibodies (iHisto.io).
Fabrication and characterization of suturable PGS/PCL scaffolds PGS Synthesis: Poly(glycerol sebacate) (PGS) was synthesized using an established protocol. 79 Equimolar quantities of sebacic acid (Sigma-Aldrich) and glycerol (Fisher Scienti c) were mixed using a magnetic stirrer. The mixture was reacted under microwave radiation in 1 minute intervals for a total of 7 minutes at 500 W (Hamilton Beach, VA). After every minute interval, the microwave door was opened for 10 seconds to allow vapors to be removed. The synthesized PGS was allowed to cool to room temperature.
PGS/PCL Scaffold fabrication: Polycaprolactone (PCL) Mn 80,000 (Sigma-Aldrich) was dissolved at 20% (w/v) in hexa uoroisopropanol (HFIP) (CovaChem, IL) using a magnetic stirrer in a covered container to prevent evaporation. In a separate covered container, the synthesized PGS was dissolved in HFIP at 20% (w/v). Equal volumes of the PGS and PCL solutions were added together and mixed on a magnetic stirring for 3 hours at room temperature. The solution was transferred to a syringe and electrospun at 18 kV using a 22 G blunt-tipped needle. The bers were collected on a rotating drum coated with aluminum foil. The drum was rotated at a speed of 50 rpm and set at a distance of 22 cm from the tip of the needle. The electrospun mats and aluminum foil were removed from the drum and desiccated for 24 hours.
Preparation and sterilization of scaffolds for surgery: On the day before surgery, PGC/PCL scaffolds were sterilized with 70% ethanol and were left inside the biosafety cabinet overnight. On the day of the surgery, the scaffolds were detached from the foil gently using two forceps inside the biosafety cabinet.
The bioprinted constructs were attached to the scaffolds by adding a drop (~15µL) of GelMA prepolymer solution (7.5% GelMA+ 0.5% PI) to the mesh and placing the constructs on top of the pre-polymer solution. The constructs were crosslinked under UV light (100 mW/cm 2 ) for 5 seconds. The adhesion of construct-mesh makeup was tested by applying tensile forces to the construct with forceps. Next, the constructs attached to the mesh were placed in 48-well plates, supplied with SKGM-2 medium and incubated at 37 ºC until surgery.
Mechanical characterization of PGS/PCL scaffolds: Using a surgical suture, 20mm x 6mm x 0.1mm rectangular shaped specimens of the PGS/PCL scaffold were sutured. The tensile strength of both the sutured and non-sutured scaffolds were studied using a uniaxial material testing system (Instron 3342). Rectangular-shaped specimens (40mm x 6mm x 0.1mm) were stretched at a strain rate of 7mm per minute, in compliance with the ASTM D822 standard. Five samples were tested for each group. The Young's modulus 'E' of the bers was calculated from the linear region of the stress-strain curve corresponding to 0%-30% of the tensile strain.
Morphological characterization of PGS/PCL scaffolds: Electrospun PGS/PCL sheets were imaged using scanning electron microscopy. First, desiccated sheets were coated in platinum and palladium for 10 seconds using a Sputter Coater 108 Auto (Cressington, UK). The coated sheets were imaged on a high vacuum at 2 kV using a FEI Quanta 200 ESEM (FEI Company, Hillsboro, OR). To image scaffold-hydrogel interfaces, a 1cm 2 piece of the PGS/PCL scaffold was photo-crosslinked on a polymerized GelMA hydrogel cube. The scaffold was pre-wet with liquid GelMA before putting on the hydrogel cube and crosslinked with a UV light for 20 seconds operating at 850 mW/cm 2 intensity. The PGS/PCL-GelMA construct was freeze-dried for 3 days. The samples were sputter-coated with a 5nm thin layer of Pt/Pd alloy (80:20) for increased conductivity and imaged using JEOL JSM-IT100 InTouchScope™ scanning electron microscope operated at an accelerated voltage of 5-10 kV. Fiber diameters of the scaffold and pore sizes of the GelMA hydrogel were measured using Image J software. Three samples were analyzed, and 10 images taken from different areas of the cross-section.

In vivo VML study
All animal procedures were approved by the Institutional Animal Use and Care Committee of Harvard Medical School (Protocol number: 2016N000375). 8-week-old mice (Jackson Laboratories; Bar Harbor, ME, USA) were acclimatized for a week prior to surgery in the vivarium of Brigham and Women's Hospital.
Four NOD-scid IL2Rgamma null mice and two C57BL6J WT mice were used in this study for the iPSCladen and acellular hydrogel implants, respectively. Mice were anaesthetized using a 2-4% iso urane vaporizer induction chamber and the surgical site was disinfected with chlorhexidine and 70% isopropanol (Contec, Spartanburg, SC).
The mice were divided into three equal groups: VML treated with 3D vascularized muscle tissue (V3), VML treated with acellular bioprinted constructs (V2) and untreated VML (V1). Bilaterally, a longitudinal incision at quadriceps muscle was created followed by a sharp resection of the quadriceps muscle in parallel to the femur length. Following a prior protocol, in all study groups, approximately 34.7 ± 0.7 mg of the muscle was cut, leaving only a base of the skeletal muscle. 80 In the groups receiving treatment (V3 and V2), the construct (acellular bioprinted or vascularized muscle tissue) was attached to a PGS/PCL scaffold and placed with the scaffold side up bilaterally into the VML defects. The skin incision was closed in all groups using 4-0 Prolene suture (Ethicon Inc, Somerville, NJ). To represent the remodeling phase of muscle healing, the hind limb quadriceps muscles of the mice were harvested four weeks postsurgery and xed in 10% formalin followed by 70% ethanol for histological analysis.

Figures
We set out to use a photocrosslinking strategy to transform the embedding bath including the printed patterns into a solid hydrogel scaffold that was robust enough to enable in vitro culture, in vivo implantation, and eventually muscle tissue repair. To this end, various GelMA concentrations and photocrosslinking conditions (i.e., UV exposure time) were assessed to obtain a solid construct with mechanical properties optimized for muscle stem cell differentiation and maturation. 5% (w/v) GelMA and a UV exposure time of 55 seconds was used for further experimentation (dashed line). UV exposure times below 30 seconds (3 J/cm2) failed to completely solidify the embedding bath regardless of the GelMA concentration, but longer irradiation provided the required shape delity. (h-j) The optimized hydrogel formulations, printing parameters, and photocrosslinking method enabled the 3D printing of various well-aligned, micropatterned, multi-material 3D constructs, as visualized using uorescence confocal microscopic cross sections (using inks spiked with red and green polystyrene nanoparticles). (k) Combining a coaxial nozzle with EMB enabled the fabrication of core-shell micro bers in the GelMA supporting matrix, which was con rmed by the cross-section view of uorescence image. (l-o) Further combining the optimized 3D printing parameters with more advanced computer-aided designs readily enabled the printing of biomimetic engineered muscle tissue aps. (p) Using GelMA ink as UV-glue enabled the assembly of 9 micropatterned blocks (~1 cm3) into a large complex construct.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.