gms | German Medical Science

Generation of functional skeletal muscle neo-tissue remains a major challenge in tissue engineering. So far functional skeletal muscle tissue in a clinically relevant size could not be created yet. However, there are multiple options for clinical applications so that the skeletal muscle tissue engineering is still a developing field [1]. In the past, tissue engineering of skeletal muscle first evolved in the 80s and early 90s when Vandenburgh et al. engineered contracting muscle cells in vitro for the first time [2]. Thereafter, the myooids – muscle tissue engineered in vitro with a maximum thickness of 1 mm [3] – introduced by Strohman et al. [4], raised hopes that clinically relevant sizes of functional skeletal muscle may be engineered in the near future [1]. However, one of the remaining obstacles is to develop a suitable scaffold for muscle tissue engineering in vitro and in vivo [5]. The materials and different architectures which have been used for functional skeletal muscle in the past and presence will be discussed in this review. Furthermore, the specific demands for skeletal muscle tissue engineering concerning the cell differentiation challenge as well as the task of axial vascularization for future clinical application will be highlighted.

Tissue engineering of skeletal muscle has been in focus of different studies for the treatment for inherent muscular diseases, in particular Duchenne’s muscular dystrophy (DMD) [1]. Though promising in the beginning, the implantation of muscle precursor cells or engineered skeletal muscle tissue did not meet the expectations. Other applications, such as gene therapy with antisense oligonucleotides (AONs) e.g. could also be promising [6], [7].

In plastic and reconstructive surgery, the transfer of myocutaneous free flaps to cover soft tissue defects is one of the most frequent applications in a clinical setting. Unfortunately, this leads to a functional loss as well as loss of volume at the donor site, known as donor site morbidity. Moreover, the transferred tissue has to be supplied with blood by a microsurgically created anastomosis between the transplant’s (“pedicle”) vessels and the recipient vessels. This comes along with the risk of flap failure due to technical problems at the anastomosis or reduced blood supply at the recipient site. Engineered flaps could be an alternative to cover soft tissue defects without the disadvantage of donor site morbidity [1], [5].

Under specific circumstances the transplanted muscle tissue itself is not only used for covering exposed vulnerable structures, but also as a functional substitute for paralysed, missing or denervated muscles, thus called “free functional muscle transfer”. The most common clinical situation for such demand is facial nerve palsy. The transfer of innervated free flaps, e.g. the gracilis muscle of the thigh as a free transplant or the temporal muscle as a pedicled transfer, is routinely used in clinics to reanimate the facial mimic [8]. Though the results are encouraging, this technique often does not lead to satisfactory results [9]. Thus, in such special situations engineered muscle tissue could be designed to more closely resemble the original muscle in size and shape. Ideally the engineered muscle neo-tissue should also encompass a motor nerve with an adequate calibre to create a microsurgical anastomosis with a motoric nerve at the recipient site [1], [5].

The typical architecture of mature skeletal muscle is based on its highly orientated muscle fibers which are organized in bundles and the latter form muscles with a distinct function. The parallel orientation of the muscle fibers is a pre-requisite to enable a longitudinal force generation which constitutes the exact function of the muscle [1]. Furthermore, skeletal muscle tissue is one of the most vulnerable tissues regarding its tolerance of hypoxia. Hence an inadequate vascularization rapidly leads to extended necrosis of muscle tissue and/or transformation into fibrous scar tissue, which is not contractile anymore. Therefore, vascularization plays an important role for tissue engineering of skeletal muscle in clinically relevant sizes. The vascularization as well as pro-angiogenic factors will be discussed in the following paragraphs [10]. While stability is one of the main points of any extracellular matrix (ECM) in general, the ECM of skeletal muscle tissue in particular should exhibit good elasticity to support and conduct the contraction of the muscle fibers. Thus, a potential scaffold for skeletal muscle tissue should balance stability and elasticity and enable the parallel alignment of muscle precursor cells [10]. Besides the necessary mechanical impact on contractility itself, it was also shown that myogenic differentiation is significantly influenced by the stiffness of the scaffold [11].

In vivo, injured muscle tissue is regenerated by satellite cells [12]. These cells were first described by Mauro et al. and named as satellite cells because of their localization beneath the basal lamina of muscle fibers [13]. Today, the “satellite cell” is identified by its expression of the transcription factor Pairedbox 7 (Pax7) [1], [14]. However, satellite cells enclose two sub-populations of cells depending on their co-expression of MyoD, a myogenic transcription factor. The main population of satellite cells (approximately 90%) also expresses MyoD (also named Myf5), which marks their commitment to the myogenic lineage [15]. Along with this myogenic pre-differentiation goes the disadvantage of a decreased proliferation rate complicating the generation of clinically relevant sizes of muscle tissue in vitro [16], [17]. As an advantage, this commitment to the myogenic line also enables a safe application of satellite cells in a clinical setting without a significant risk of trans- or dedifferentiation [1]. Therefore, satellite cells have been used mainly as cell source for skeletal muscle tissue engineering in the past [18], [19]. Only a minor sub-population of approximately 10 percent does not express the myogenic marker MyoD and shows in turn stem cell properties with the possibility of differentiation into multiple mesenchymal cell populations. This sub-population regenerates the Pax⁺/MyoD⁺-cell population in vivo through asymmetric self-renewal [1], [20]. The regenerative potential of the Pax⁺/MyoD^- sub-population is astonishing, since whole muscle bundles can be regenerated in vivo [21]. Though, isolated satellite cells show a significant loss of their proliferative potential when cultured in vitro [22]. This phenomenon has been explained with the loss of the stem cell niche, which is based on the cell contact to the basal lamina and the ECM [23]. Thus, the generation of a suitable number of muscle-precursor cells for skeletal muscle tissue in vitro remains a challenge. Gilbert and his group demonstrated that the satellite cell niche can be mimicked in vitro by cultivating isolated satellite cells on laminin cross-linked PEG hydrogels with an elasticity of 12 kPa which exactly equals the elasticity of the basal lamina in skeletal muscle tissue [24]. Therefore, these muscle precursor cells isolated from mature muscle tissue seem to be the most suitable cell source for skeletal muscle tissue engineering. Though, the isolation of an adequate number of precursor-cells and the preservation of the proliferation rate in vitro are still an obstacle that has to be overcome before a clinical application could be realized [25].

The use of mesenchymal stem cells (MSC) has often been proposed because of their higher proliferation rate in vitro [26]. Adipose-derived mesenchymal stem cells (ADSCs) in particular promise to offer an easy accessible and highly potential cell source for application in a clinical setting [27]. However, myogenic differentiation of mesenchymal stem cells is challenging in vitro as well as in vivo. Brazelton et al. reported a poor incorporation rate of 5–10% of transplanted MSCs in skeletal muscle tissue in vivo [28]. Though the majority of transplanted MSCs shows no myogenic differentiation in vivo, the transplanted MSCs contribute to myogenic regeneration in injured or dystrophic muscle tissue through paracrine effects [29]. The secretion of anti-inflammatory, anti-apoptotic and angiogenic factors by transplanted MSCs constitute these paracrine effects which support the local regeneration of injured skeletal muscle tissue [30]. The pro-angiogenic effect is also seen when the secretome of MSCs is added in vivo [31]. However, mesenchymal stem cells also fuse with co-cultured with myoblasts in vitro [25] which underlines their versatile contribution to muscle regeneration. As an interesting feature, mesenchymal stem cells can be transplanted allogenically due to their low immunogenicity [32], [33]. Recently, induced pluripotent stem cells (iPSCs) were introduced into tissue engineering, displaying an even higher proliferation rate but also a seriously augmented risk of dedifferentiation and tumorigenicity in vivo [5]. Therefore, this cell source has only rarely been studied for tissue engineering applications in vivo, yet [1].

A plethora of materials have been applied and characterized for different tissue engineering purposes, but only few have met the specific demands for skeletal muscle tissue engineering. In the first place, biocompatibility and the absence of tumorigenicity are crucial issues for application in vivo [5]. Therefore, certain materials which have been widely used for tissue engineering are not suitable for in vivo studies. Matrigel™ for example, a hydrogel extracted from Engelbreth-Holm-Swarm (EHS) mouse sarcoma cells and containing a variety of growth factors, shows good results in vitro but cannot be used in a clinical setting [34], [35]. Natural components of the native (ECM) of skeletal muscle, e.g. collagen I, elastin or Laminin are promising and clinically applicable candidates for tissue engineering of skeletal muscle in vivo [36]. Their use in vivo is non-hazardous and bovine collagen I shows a very low immunogenicity in xenogenous models in vivo [37]. However, their disadvantage lies in their fast degradation in vivo. The stability of fibrin, elastin as well as collagen is completely lost after several weeks in vivo [38]. Since vascularization, neurotization and myogenic differentiation of implanted myoblasts into mature muscle fibers takes several months, these materials have to be modified in order to make their way into any clinical setting. Concerning stability, biodegradable synthetic materials are a cost-effective and easy to handle alternative. Materials like poly-l-lactic acid (PLLA) or poly-e-caprolactone (PCL) are stable over approximately one year in vivo [39], [40]. Though both materials are biocompatible, the acidic degradation products of PLLA can lead to cell toxic effects [41]. Furthermore, the disadvantages of PCL are mostly its high hydrophobicity and low elasticity [36]. Furthermore, PCL can be coated or blended with materials like collagen [42], [43] or gelatine [44] to enhance cell-attachment. The aim of designing scaffolds for skeletal muscle tissue engineering is to create a three-dimensional architecture mimicking the natural ECM with its biologic as well as mechanic properties as closely as possible. Hence, potential scaffolds for tissue engineering of skeletal muscle should reflect the parallel alignment of native myotubes and myofibrils as well as the elasticity of its native surrounding ECM [1]. Following the cell guidance theory, as first postulated by Curtis and Wilkinson [45], the myogenic differentiation and parallel alignment of myogenic cells can thus be enhanced [46]. Hence matrices like fibrin or other hydrogels with random orientation may not render best results. A possibility to gain parallel alignment within a scaffold is the unidirectional freeze-drying of hydrogels. Thus, ice crystals form in a spatially orientated pattern leading to orientated pores afterwards [1]. This method has successfully been used for materials like collagen and silk fibroin [47] and the pore-size can be controlled by the freezing temperature [48]. As a disadvantage, the alignment of the pores remains spatial and the scaffold itself shows a random architecture, though.

The most promising method to create strictly parallel aligned structures could be the electrospinning technique [49]. The formation of fibers by electrical voltage is a complex method and the multiple parameters like concentration of the solution, flow rate and viscosity as well as the voltage and distance to the counter-electrode enables the adjustment of the resulting scaffold’s properties in a wide range [43], [50]. Thus, a plethora of biomaterials can be spun to nano- or microfibers like hyaluronic acid, collagen I, elastin as well as synthetic polymers [1], [51]. In order to enhance the cell-attachment in vitro and in vivo post-spinning modifications like coating [52] or plasma treatment [53] have been developed. However, these methods are limited to few fiber-layers. For three-dimensional nanofibrous scaffolds synthetic polymers can be blended with biopolymers like collagen, hyaluronic acid or elastin. Furthermore, two different polymer solutions can be spun separately into one fiber. The surface of the resulting fibers of this core-shell spinning method is formed solely by the surrounding polymer [1]. Zhang et al. demonstrated that the core-shell spinning technique using PCL as core and collagen I as shell could be superior to a post-spinning coating of PCL-fibers with collagen [54]. However, control of the “pore-size” or better “interspaces between the fibers” remains a major challenge and a significant disadvantage of the electrospinning method in general. The more parallel the fibers are aligned the denser they are packed after spinning. As a consequence the absence of adequate interspaces interferes with cell migration and – more vital for in vivo application – also ingrowth of vessels into the scaffolds [43], [55]. Baker and co-workers have addressed this point by co-spinning a water-soluble polymer solution, e.g. poly(ethylene-oxide) (PEO). The resulting fibers – named sacrificial fibers – can be solved after spinning and leave behind interspaces for enhanced cell migration and vascularization [56]. Taken together, PCL-collagen blend nanofibers are – also based on our own experimental results [36], [43] – promising candidates as scaffolds for muscle tissue engineering, pending the development of sufficient interspaces by sacrificial fibers of other means.

In order to generate functional neo-tissue in clinically relevant size for later clinical application, at least 1 mm in thickness of such neo-muscle has to be achieved. However, thes is usually the limitation for tissue engineering in vitro, based on diffusion characteristics which are generally believed to be limited to a distance of 500 µm. Hence the generation larger neo-tissue constructs needs a sufficient vascularization of the implanted scaffold and thus the engineered tissue in vivo [57]. Usually the scaffolds are implanted first and the cells are added in a second operation after complete vascularization of the matrix or small scaffolds are pre-seeded prior to implantation. With the latter approach apoptosis of implanted cells due to a lack of initial nutrient is likely to occurred, while with the first approach, one may observe less cell death due to prior vascularization [58]. The duration of the pre-vascularization period, i.e. the time until vessels haven sufficiently grown into the center of the scaffold depends on the material and architecture of the scaffold in general and the pore size in particular [1]. Hence a porosity of approximately 90% of a scaffold with high interconnectivity and an adequate pore size enable the migration of precursor and endothelial cells so that vascularization as well as tissue formation is possible not only at the periphery of a matrix but also at the center [59], [60]. Regarding muscle precursor cells, the pore size of the scaffold should ideally range between 50 µm and 200 µm [61]. These pre-requisites are well presented in hydrogels or sponges with high porosity where cells can freely migrate through the matrix by degrading the hydrogel. However, in scaffolds with parallel alignment in general and parallel electrospun scaffolds in particular, the size and interconnectivity of the interspaces remains a challenge despite multiple methods described to increase the interspaces as discussed above [1], [43].

In a rat AV-loop in vivo study we could characterize the vascularization of randomly versus parallel spun PCL-collagen blend nanofiber scaffolds in vivo [43]. In this study, an AV-loop was microsurgically created as first described by Erol and Spira in 1980 in the groin of rats and implanted into the matrix inside an isolation chamber [62]. Though technically challenging and time-consuming, the advantage of this model lies in the strict axial vascularization of the whole matrix. Thus, vessels sprout from the loop vessels inside the chamber only and the engineered tissue inside the chamber can be transplanted with a microsurgical anastomosis of the pedicle to the site where the engineered tissue is needed [43]. Therefore, the AV-loop model is a technique applicable for a potential clinical use in tissue engineering in general. Though the total number of vessels inside the scaffolds was higher within the randomly spun nanofiber scaffolds, the parallel spun group showed a more constant vascularization of the center whereas the vessels in the randomly spun group sprouted in the periphery of the scaffolds without an adequate vascularization of the center even after 8 weeks in vivo [43]. Interestingly, the migration of cells through smaller pore sizes has been observed by Zhang et al. with fibroblasts cultivated on randomly spun PCL/collagen scaffolds before [54]. As an explanation, the dynamic structure of nanofiber scaffolds was postulated: due to the flexible and fibrous structure, cells can push the fibers aside and thus migrate through the scaffold. Herein, parallel spun scaffolds are even more flexible since the fibers are not attached to each other. This could possibly explain the more consistent vascularization of the scaffolds [1].

As a further prerequisite for later clinical application, the newly generated skeletal muscle tissue should be innervated by a motor nerve. In order to enable axial vascularization and neurotization at the same time in vivo, we developed and characterized a new small animal AV-loop model, i.e. the EPI-loop model. In this microsurgical rat model, the AV-loop is based on a different vascular axis (the epigastric vessels) as well as motor branch of the obturator nerve. After myoblasts and MSCs coimplantation in a prevascularized isolation chamber, we could observe areas of myogenic differentiation with alpha-sarcomeric actin and MHC expression in the constructs. Quantitative PCR analysis showed an expression of myogenic markers in all specimens [63]. Thus, neurotization and addition of bFGF and dexamethasone allow myogenic differentiation of MSC in an axially vascularized in vivo model for the first time. These findings are a new step towards clinical applicability of skeletal muscle tissue engineering and display its potential for regenerative medicine [63].

In the future skeletal muscle tissue engineering could be further augmented by means of MSC- and/or ADSC cocultivation with muscle precursor cells (myoblasts). Furthermore the differentiation process could be supported by means of smart matrices, i.e. functionalized nanofiber scaffolds releasing myogenic transcription factors or specific components of the ECM/basal lamina of native skeletal muscle tissue. Possibly the architecture of the fibers themselves might also play a pivotal role in the future for myogenic differentiation.

For in vivo application in general and later clinical application in particular, one has to address the problems outlined above in adequate large animal models first, before possible translation into humans. Herefore the development of a large animal (sheep) model for axial vascularization during the last years could be an important step towards later clinical application: after first development of the sheep AV-loop model itself [64], it was further characterized [65] and applied for tissue engineering of vascularized bone [66], [67], including the autologous MSC implantation [68] as well as incorporation of BMP-2 [69], [70] and an extrinsic vascularization pattern [71]. In conclusion results obtained from cell cultures and small animal models could once not only be translated to this microsurgical large animal vascularization model in the field of bone Tissue Engineering, but also for the even more challenging task of skeletal muscle tissue engineering.

The authors declare that they have no competing interests.