Abstract
One of the major limitations to advancing the development of soft robots is the absence of lightweight, effective soft actuators. While synthetic systems, such as pneumatics and shape memory alloys, have created important breakthroughs in soft actuation, they typically rely on large external power sources and some rigid components. Muscles provide an ideal actuator for soft constructs, as they are lightweight, deformable, biodegradable, silent, and powered by energy-dense hydrocarbons such as glucose. Vertebrate cell lines and embryonic cultures have allowed critical foundational work to this end, but progress there is limited by the difficulty of identifying individual pathways in embryonic development, and the divergence of immortal cell lines from these normal developmental programs. An alternative to culturing muscles from embryonic cells is to exploit the advantages of species with metamorphic stages. In these animals, muscles develop from a predefined pool of myoblasts with well-characterized contacts to other tissues. In addition, the endocrine triggers for development into adult muscles are often known and tractable for experimental manipulation. This is particularly true for metamorphic muscle development in holometabolous insects, which provide exciting new avenues for tissue engineering. Using insect tissues for actuator development confers additional benefits; insect muscles are more robust to varying pH, temperature, and oxygenation than are vertebrate cells. Given that biohybrid robots are likely to be used in ambient conditions and changing environments, this sort of hardiness is likely to be required for practical use. In this study, we summarize key processes and signals in metamorphic muscle development, drawing attention to those pathways that offer entry points for manipulation. By focusing on lessons learned from in vivo insect development, we propose that future culture designs will be able to use more systematic, hypothesis-driven approaches to optimizing engineered muscle.
Impact statement
This review summarizes our current understanding of metamorphic muscle development in insects. It provides a framework for engineering muscle-based actuators that can be used in robotic applications in a wide range of ambient conditions. The focus is on identifying key processes that might be manipulated to solve current challenges in controlling tissue development such as myoblast proliferation, myotube formation and fusion, cytoskeletal alignment, myotendinous attachment and full differentiation. An important goal is to gather findings that cross disciplinary boundaries and to promote the development of better bioactuators for nonclinical applications.
Background
There is currently great interest in growing fully organized muscle tissue in vitro because it promises to benefit many applications, from health care to robotics. Key topics include muscle repair therapies 1 and the development of new approaches to manufacturing cultured meat, which is currently only available in a disorganized “ground” form. These technologies can benefit from research on the fundamental pathways involved in muscle organization and from insights derived from studies on different muscle types and species.
In robotics, whole-muscle engineering will be especially pivotal, as the advancement of soft machines is currently limited by an absence of lightweight, flexible actuators with onboard power sources. 2 Engineered muscle tissue offers an attractive option for soft actuation, as it is strong, compliant, self-healing, chemically fueled, environmentally and biologically compatible, and silent.3,4 Studies using whole excised muscles have demonstrated that muscle tissue is an effective actuator for synthetic systems and can be successfully kept functioning for more than 24 h 5 (vertebrate muscles), and in some cases as long as 90 days6,7 (invertebrate muscles) without a biological system for supplying oxygen and nutrients.
Because excised muscles must be manually dissected from live animals and fixed into artificial constructs, they do not represent a viable option for large-scale robot fabrication. However, findings from these studies have helped to inspire the development of contractile muscle cultures, which have enjoyed considerable success as actuators.8–13 Existing cultured muscle actuators may use cell lines, homogenized muscle tissue, or homogenized whole embryos to seed thin, typically sheet-like structures.8,14 This design allows muscle cells to be grown as a monolayer to maintain contact with the bathing solution, and maximizes the displacement produced by low forces.
Because the maximum force produced by a muscle is proportional to its cross-sectional area, 15 applications using higher forces may require engineered muscle cells be grown as three-dimensional (3D) cultures, rather than as monolayers. Additionally, most existing cell cultures have not been able to align myotubes in a life-like manner or design them to contract synchronously, as muscle tissue does in vivo. These limitations result in muscle constructs that produce only a small fraction of the forces that muscles can produce in vivo.16–18 Future efforts to design bioactuators will need to consider not only maximum force production4,19 but also how to engineer passive and active compliance, 20 work production and absorption, and maximum or sustainable power output. 21
Some insights into developing whole muscle tissues are likely to be found in organ engineering work. Many of these studies have identified effective strategies for encouraging self-assembly of complex tissues, such as heart and liver, from stem cells. One particularly promising strategy that has come out of this work involves the use of realistic scaffold materials. Although many tested scaffold have improved 3D tissue construction, decellularized donor scaffolds have been the most effective, likely due to the maintenance of appropriate vascularization, stiffness, and protein conformation.22–24
For further development of muscle tissue constructs specifically, holometabolous insects represent a promising study system. Already, insect cultures have proven more robust to changes in temperature, pH, and acidity than existing vertebrate culture systems. In one case, embryonic insect cultures remained contractile for more than 2 months in ambient conditions without media changes. 25 Additionally, holometabolous insects undergo complete metamorphosis, which offers numerous benefits over embryonic tissue for this type of work. Whereas studying organization in embryonic development is complicated by small tissue size and a rapidly changing microenvironment, metamorphic muscles develop in a fully formed animal from a prespecified pool of myoblasts interacting with reliably positioned nerves and tendon cells.26–29
In this study, we cover the major known factors in each stage of adult insect myogenesis, with an emphasis on environmental factors and developmental pathways that might be manipulated in vitro. We propose that an understanding of metamorphic muscle development in vivo will inform both novel approaches to muscle engineering and targeted changes to existing culture systems.
Embryonic Muscle Specification and Development
The cells that form adult muscles in holometabolous insects are first specified in the embryo. In all insects, as in vertebrates, a wide array of tissue types, including fat body, cardiac and visceral muscles, and somatic muscles originate from the mesoderm. Within each segment, the somatic muscles are developed from a region specified by both intrinsic mesoderm expression and inductive signaling from the ectoderm.30–32
This process has been best described in Drosophila, where this region is designated by the segmentation gene sloppy paired (slp).33,34 In combination with wingless, slp expression from the ectoderm induces the underlying mesodermal cells to express slp and high levels of Twist, creating a “competence domain” that can form somatic muscles (Fig. 1). Twist levels, in particular, are a decisive factor in myogenesis; previous work has shown that ectopic Twist can prevent the formation of visceral tissues or even reassign ectoderm cells to form muscle. 35

AMP selection in the Drosophila embryo. Myogenic cells can be found adjacent to the slp-expressing regions of the ectoderm. Muscle progenitors are selected from a competence domain of mesodermal cells that express l'sc. Within each competence domain, l'sc expression is restricted to a single cell by Notch-mediated lateral inhibition. The progenitor then divides asymmetrically, restricting numb to a single daughter cell. A subset of those daughter cells that do not receive numb become AMPs. AMP, adult muscle precursor; l'sc, lethal of scute; slp, sloppy paired. Color images are available online.
Within the high-Twist domain, cells must be further divided into progenitors and fusion-competent myoblasts (FCMs). Locally, Wingless and Decapentaplegic (Dpp) signaling from the ectoderm define “equivalence groups” of myoblasts in the mesoderm. These cells share an equal potential to become muscle progenitor cells. 36 In Drosophila, the cells of the equivalence groups express the proneural gene lethal of scute (l'sc). 37 Lateral inhibition, mediated by Notch signaling, restricts l'sc expression to a single cell in each equivalence group, called the progenitor. Each progenitor then divides asymmetrically to form two daughter cells with unique identities, which will become either two founder cells for different larval muscles or one larval founder cell and one adult muscle precursor (AMP). These cells are differentiated from one another by the cytoplasmic protein Numb, which is restricted to a single side of the progenitor cell and segregated to one of the daughter cells (Fig. 1).26,38 The resulting founder cells express identity-specific genes like S59 and vestigial and will eventually confer identity information to the resulting muscle.
Those slp domain cells that do not form progenitors become FCMs, which are thought to be naive, meaning that they cannot form complete muscles on their own and adopt the expression pattern of any founder cell they fuse with. This view is supported by experiments where myoblast fusion is blocked: in Drosophila myoblast city mutants, founder cells elongate and form correct attachment sites, eventually differentiating into tiny, mononucleate muscles that can contract and are appropriately innervated. 39 FCMs, meanwhile, remain rounded and mostly undifferentiated, and many eventually degenerate.
Although most of what is currently known in insect development has been learned from Drosophila, the principal programs are likely similar across Holometabola. In Tribolium beetles and even crickets, the Wingless homolog appears to perform the same patterning function,40,41 and asymmetric Numb distribution appears to be important for differentiating myogenic lineages even in vertebrates. 42
AMP specification
Adult muscle precursors (AMPs) are specified during embryonic development, at the same time as larval founder cells are formed. During progenitor division in Drosophila, the protein Numb is asymmetrically distributed within the cell by a pathway that appears to depend on autonomous intracellular signaling. 43 Numb inhibits Notch signaling in a highly conserved pathway, and the daughter cell that receives Numb maintains this inhibition, and thus the same expression pattern as the progenitor cell.26,44–46 By contrast, in the daughter cell that does not receive Numb, Notch signaling becomes active and the expression pattern changes. Of those cells that lose progenitor expression, a subset will become AMPs. 26
The distribution of AMPs is segment specific and seems to be controlled by Hox genes in the mesodermal cells themselves, in combination with EGF signaling.47–50 In Drosophila, Spitz, an activator of EGF signaling, is provided by the neighboring epidermal cells and larval founder cells. Spitz appears to be required to trigger transcription of AMP identity genes, and may also help to maintain AMPs through larval life. 50 Therefore, attempts to isolate and maintain AMPs in vitro may need to consider maintenance of Spitz/EGF signaling to maintain AMP identity.
Work in Drosophila has also shown that after AMPs are specified in their appropriate locations, they send out long processes along peripheral nerves until all AMPs are connected. 50 When these connections are experimentally severed, the AMPs become mobile and fail to maintain appropriate positioning or reestablish the connection. Currently, little is known about how these processes find their targets. However, they may help to understand problems with AMP positioning and may eventually prove important for actuator patterning.
All AMPs in Drosophila also express three transcription factors—Cut, Zfh1, and Twist—into larval life. 50 Cut levels have been shown previously to help specify different populations of AMPs. 51 Zfh1 represses mef2 transcription, blocking somatic myogenesis. 52 Twist expression is widespread throughout the embryo but is restricted to AMPs by the end of embryogenesis. In combination with active Notch signaling, it allows the cells to remain undifferentiated.53,54 Taken together, these factors maintain AMP potentiality and allow them to continue proliferating throughout larval life, eventually forming pools of identical AMPs. These features may prove useful for assaying AMP health and function in cultured systems, and healthy muscle development will likely require that they are maintained. The transcription factors required for adult myogenesis are less well understood outside of Drosophila, but further investigation on this question will enable a better understanding of both muscle metamorphosis and the required conditions for engineering these tissues.
Secondary Myogenesis and Metamorphosis
Imaginal pioneer specification
As in larval development, adult muscles gain their identity from specialized cells selected from the AMP pool in the last larval instar. 55 These new founders, which have been termed Imaginal Pioneers (IPs), 56 specify the identity of the eventual muscle, while the remaining AMPs differentiate into FCMs that will add to muscle size. In these respects, IPs appear to be precisely analogous to embryonic founder cells.
IPs are distinguished from the rest of the AMP pool by a larger size and distinct morphology, and their number corresponds to the number of adult muscle fibers to be formed. 56 IPs in Drosophila express the embryonic founder cell marker Dumbfounded (Duf), which is initially expressed by the entire population of AMPs but is restricted to IPs and a set of small, FCM-like myoblasts by early metamorphosis (Fig. 2).55,57 Interestingly, this selection process differs from embryogenesis in that it does not require Notch-mediated lateral inhibition. 55 Instead, IPs are selected by a pathway involving fibroblast growth factor (FGF) signaling. 58 This selection is likely directed by a combination of homeotic identity genes in the AMPs themselves and inductive signaling from the epidermis.48,55 Although we expect that IPs are selected in the manner for all Holometabolans, little work has been done on these cells outside of Drosophila. However, Duf has a conserved role in muscle fusion, and its function as a marker is likely to be useful in identifying these cells. 59

Secondary myogenesis. AMPs proliferate throughout larval life, creating a pool of cells by early metamorphosis. Duf is expressed in all AMPs during larval life but is restricted to a small number of IPs and surrogate IPs through FGF signaling by early metamorphosis. Swarming FCMs mutually inhibit differentiation by activating Notch signaling. As AMPs approach their fusion targets, Notch signaling degrades, allowing differentiation to initiate. The myotube sends out filopodia, which help to recruit fusion proteins. The AMP sends finger-like protrusions toward the myotube, and Sns from the AMP creates stable complexes with Duf from the myotube, initiating fusion pore formation. Duf, dumbfounded; FCMs, fusion-competent myoblasts; FGF, fibroblast growth factor; IPs, imaginal pioneers; Sns, Sticks and Stones. Color images are available online.
The other population of Duf-expressing cells, which we term “surrogate IPs,” is morphologically similar to FCMs and appears not to seed myotube formation in healthy animals. However, when the IPs are ablated, many more muscle fibers form in a manner that is dependent on Duf expression.60–62 It is therefore likely that the primary role of IPs is to control fiber number during muscle development by suppressing fusion with the surrogate IPs and seeding myotubes themselves.
Migration
Before proceeding with development, AMPs must migrate to the sites of their corresponding adult muscles, typically from a nearby peripheral nerve54,63 or imaginal disc. 64 This migration appears to be strongly dependent on cues from the nervous system and epidermis, rather than autonomous identity: experimentally manipulating the homeotic genes of AMPs in the thorax and abdomen of Drosophila does not alter their migration pattern.49,65 Conversely, manipulating the segmental identity gene Ubx in the ectoderm of the mesothorax, so that the nerves and epidermis resemble the metathorax, does change migration of AMPs there.66,67
In Manduca sexta, severing a motoneuron early in pupal development leads the adult muscle it innervates to be smaller, and sometimes completely absent.27,68 Denervation later in pupal development does not disrupt muscle size, suggesting that there is a narrow window during which the nerve mediates AMP migration. Further experiments denervating only a subset of the dorsolongitudinal flight muscles (DLMs) in Manduca show that denervated DLM myoblasts will accumulate around the adjacent, intact nerve terminals and contribute to their muscle fibers. 28
Nerves likely direct migration by maintaining IPs or maintaining Duf expression or its equivalent in IPs. 57 In Drosophila, this may be mediated by FGF signaling, as overexpression of the FGF receptor Heartless (Htl) causes supernumerary Duf-expressing IPs to emerge. 58 In the Drosophila embryo, Duf acts as an attractant to FCMs and is required for successful myoblast fusion, 69 and it appears to play a similar role in adult development. 55 While we are not aware of any other chemical messengers known to direct migration, AMPs send out their processes along peripheral nerves, so it is possible that they receive short-distance signals from the axons. 50 Further investigation into migration-directing signals and the influence of neural input on developing muscle across species will be highly beneficial to muscle tissue engineering ventures.
Proliferation
AMP proliferation is a multifaceted process controlled by larval muscles, ecdysteroids, neurons, and imaginal disc epithelia. Soon after AMP specification, proliferation is initiated by signals from adjacent larval muscles. In Drosophila, muscles release the Drosophila insulin-like peptide dIlp6 to AMPs that are closely associated with and extend filopodia to them. DIlp6 activates the insulin pathway, which in turn activates Notch signaling to provoke AMPs to re-enter the cell cycle. 70
In addition to the basal proliferation rate, a sharp increase in proliferation occurs during early pupal development and continues for much of metamorphosis. 71 This elevated proliferation rate appears to be controlled by circulating ecdysteroids, innervating motoneurons, and disc epithelia.
Ecdysteroids have long been known as key regulators of insect metamorphosis. This is well studied in M. sexta, where ecdysteroid levels rise sharply from the beginning of metamorphosis until the ninth day of pupation, and then drop off sharply until adulthood. 72 Artificially elevating these levels after the ninth day results in hypertrophied flight muscles, suggesting a role for ecdysteroids in muscle growth. 73 Conversely, in a system of cultured AMPs from the developing Manduca leg, experimentally removing ecdysteroid results in decreased proliferation, suggesting that ecdysteroids are necessary to maintain normal levels of AMP proliferation. 74
Proliferation also appears to depend in part on input from the innervating motoneuron. Studies using 5-Bromodeoxyuridine (BrdU) to quantify AMP proliferation rate have found that denervation after migration significantly decreases proliferation in both Drosophila 57 and Manduca.27,68 This results in adult muscles that are both smaller and less numerous than controls, and occasionally blocks muscle development altogether.68,71,75 In these cases, the AMPs maintain their initial, low proliferation rate, suggesting that there may be an intrinsic rate of proliferation that is modulated by neural input.
Neural input has similarly been shown to benefit artificial muscle cultures. In one study, AMPs from the Manduca head, but not the developing flight muscle, could be successfully cultured. 76 The authors posited that this was due to a difference in cell purity: cultures from the head are inherently heterogeneous, as it is tightly packed with many tissue types, including nervous tissue. By contrast, the DLM is a large and relatively homogenous muscle, and it is therefore possible to harvest only AMPs from that region.
This view is supported by experiments showing that cultured AMPs from the developing Manduca leg proliferate more when grown in direct contact with neural cultures. 74 Interestingly, this was true even with neurons taken from the brain, suggesting that the proliferative effect is not segment specific or unique to motoneurons. Additionally, neural activity is not required to promote proliferation, whereas neural viability and proximity appeared necessary. In their review, Consoulas et al. suggested that the effect might be mediated by a short-distance diffusible signal, such as Hedgehog protein. 77 This possibility seems promising, as Hedgehog has been shown to promote proliferation of other metamorphic tissues in Tribolium castaneum beetles. 78
Recent work has additionally shown that imaginal disc epithelia promote proliferation. In Drosophila, the wing disc epithelium releases FGF, which acts through Heartless (Htl) to activate the Wingless pathway. 79 This activation is required to achieve normal levels of AMP proliferation, and constitutively active Htl receptors result in an enlarged myoblast pool, at least in the wing disc.
Although there are many questions left to be answered with respect to the control of AMP proliferation, there appear to be many potential inputs, and future ventures will likely benefit from utilizing or mimicking multiple nonmuscle tissue types.
Myoblast fusion
Each mature muscle fiber is formed from a collection of myoblasts, which fuse together to form a syncytium. When embryonic fusion is blocked experimentally, each founder cell will form a thin, mononucleate, mature muscle fiber, but the FCMs will remain undifferentiated and many will die. 39 The same appears to be true in adult Drosophila development; IPs can form small adult muscles, but fusion with FCMs is required for large, multinucleate muscles. 60 Control of myoblast fusion will help to engineer true myofibers, rather than disconnected muscle cells that cannot produce life-like forces.
During adult muscle development, before FCMs are near their fusion targets, they are prevented from fully differentiating by Notch signaling.80,81 The swarming FCMs mutually activate one another's Notch signaling by providing the ligand Delta to their neighbors. 80 Notch upregulates Him, which in turn represses mef2, an important differentiation gene. However, when FCMs approach their fusion targets, Notch signaling decays, allowing mef2 to be expressed. This leads to termination of Twist expression and enables the production of proteins that are necessary for fusion, such as Sticks and Stones (Sns) and D-WIP. 80 The decay in Notch signaling is at least partially dependent on autonomous Vestigial expression in the myoblasts themselves, but the pathway by which it acts is unknown. 82
Once FCMs are in close proximity to their targets, they extend protrusions into the founder cell.83–85 This appears to be a general requirement for fusion, as it has also been observed in Drosophila embryos, 83 cultured Drosophila cells, 84 and even mammalian satellite cells. 86 Although the mechanisms underlying these structures remain unclear, experiments in vitro have revealed that cell/cell and cell/matrix adhesion molecules can trigger localized actin polymerization, creating these protrusions and promoting fusion. 84
During development of the DLMs of Drosophila, FCMs fuse with the remnant larval scaffold. This appears to be a conserved process, as it also occurs in the DLMs of the Lepidopteran M. sexta and flight muscles of the Coleopteran Leptinotarsa decemlineata.87,88 In Drosophila, the scaffold also sends out filopodia to the FCMs. 89 Once the FCMs are in range, filopodia facilitate adhesion and fusion by contacting FCMs and recruiting fusion proteins (Fig. 2). FCMs express the fusion protein Sns, which forms a complex with Duf from the AMP to initiate fusion. This process is conserved in beetles, where knocking down duf and sns homologs results in thinner muscles, likely due to blocked myoblast fusion. 59 Such myotube filopodia and Duf/Sns interactions are likely to be general to all AMP fusion, as similar features are required in vertebrate myoblast fusion.90,91
Much remains to be learned about the process of myoblast fusion, but what is known points to the importance of functional cytoskeletal proteins, adhesion proteins, and Notch signaling. This may provide useful insights for improving culture media or expressing constitutively active receptors in cell lines where fusion is reduced or absent.
Differentiation
Terminal differentiation begins with the end of Notch activation and the onset of mef2 expression as the myoblasts approach their targets. 81 After fusion, FCMs adopt the expression pattern of their founder cell.92–94 Together with external signals from adjacent tissues and the hemolymph, these expression patterns determine the overall character of the adult muscle, including features such as actin isoform and segmental identity.
Intrinsic patterning genes in the target AMPs control the expression of downstream differentiation genes. For example, in Drosophila, misexpressing an abdominal Hox gene in the developing flight muscles prevents them from expressing the appropriate, thorax-specific actin gene. 49 Additionally, partial loss of function of the same gene in abdominal muscles causes them to express the thorax-specific actin gene inappropriately. In an extreme case, expressing an identity gene for direct flight muscles, ap, in indirect flight muscles causes their degeneration. 95
Although cell source and intrinsic expression will be critical to obtaining desirable traits, some features can be controlled through the environment. In vertebrate muscle cultures, exogenous electrical stimulation has been shown to improve sarcomere organization, contraction synchrony, and myotube length.96,97 This likely points to the importance of signaling from the motoneuron, although the role of neurons in differentiation of adult insect muscles remains unclear.
While some muscles seem able to differentiate normally in vivo without innervation, 71 others show a disrupted expression pattern. 75 In a system of cultured insect cells, contractile fibers formed in cocultures of neurons and AMPs but not in cultures of AMPs alone, perhaps suggesting some redundancy in vivo. 74
Proper differentiation, particularly myofibrillogenesis, also seems to depend on myotendinous attachment and the resulting tension. 98 These attachments depend on integrins, proteins that mediate cell adhesion, at the myotendinous junction (MTJ) that bind to extracellular matrix material and to the actin cytoskeleton of the myotube, providing tension directly to the myofibrils during their assembly. This suggests that any program to control differentiation should also include improvements to muscle attachment.
Attachment
Developing interfaces that will neither fail mechanically nor damage the tissue is a perennial problem in muscle engineering. 18 One promising avenue to fixing this problem is to utilize the normal process of MTJ formation.
MTJ formation happens in parallel with myofibrillogenesis and requires input from both developing myotubes and epidermal cells.99,100 In Drosophila, epidermal cells that will form adult muscle attachment sites seem to be specified before the onset of metamorphosis. 101 These cells are specified by the gene stripe, and during adult myogenesis, developing fibers contact them using filapodial extensions.102,103 Following contact, the myofiber expresses the integrin PS2, while the epidermal cells begin to express PS1. 101 These integrins bind to protein complexes on the opposing cells, forming secure attachments. In the indirect flight muscles of Drosophila, correct attachment requires Broad Complex genes, which form a key part of the ecdysteroid regulatory network. 104 Therefore, it is possible that physiological ecdysteroid levels will facilitate MTJ formation in vitro.
The muscle cell is active in organizing attachment, forming a modified, terminal z-band between its myofibrils and each attachment point. 103 The cell also heavily modifies its membranes, creating a series of folds at the attachment site, where the integrin complexes from the tendon cells will attach. Crucially, tendon cells secrete the protein Slowdown, which prevents the muscle from accumulating integrin prematurely. In the absence of Slowdown, the morphology of the muscle's leading edge is disrupted and MTJ architecture is aberrant. 105 Animals mutant for slowdown show reduced movement and sometimes rupture of the muscle or tendons.
Clearly, designing appropriate scaffold materials and enabling attachment will be a major obstacle to developing useful muscle constructs. Useful approaches are likely to include cell adhesion proteins, hormones, and novel artificial tendons.
Strategies for Engineering Insect Muscle
Taken together, the current body of knowledge about metamorphic muscle development offers many potential approaches to improving engineered muscle tissue (see Table 1 for a summary of strategies).
Strategies for Engineering Insect Muscle
The normal developmental programs of insect muscles can be used to design manipulations for each desired tissue culture outcome. In this study, we suggest some of the pathways which may be manipulated to approach common goals.
AMP, adult muscle precursor; Duf, dumbfounded; FGF, fibroblast growth factor; Htl, heartless; MTJ, myotendinous junction; Sns, Sticks and Stones.
For existing protocols, referencing normal developmental pathways may provide insights into the health and function of developing cultures. In addition to standard measures such as alignment, force production, and viability, it will be helpful to leverage known expression patterns to compare cultured cells with their counterparts in vivo. In heterogeneous culture systems, assaying for expression of as the transcription factors Zfh1, Twist, and Cut in Drosophila systems, or their equivalents in the species of interest, will help to identify AMPs. Where systems fail at an identifiable stage of myogenesis, the normal expression patterns can serve as guidance for finding solutions (e.g., assaying Duf and Sns for fusion problems). Heterogeneous expression of these factors in culture systems could also be used as a strategy for directing precursor cells into appropriate lineages. For example, cocultured Twist-expressing cells could be used to initiate the formation of myoblasts from Drosophila precursor cells. However, the effectiveness of this strategy will depend on the species specificity of Twist (and other signaling molecules) if it is to be applied to cells from other sources.
For new systems, the most straightforward manipulations will likely be those involving circulating factors; already, growth medium supplemented with the ecdysteroid mimic 20-HE has been used to promote proliferation in cultured insect myoblasts.25,74 Extending this approach to mimic physiological levels of hormones and other circulating factors, such as insulin-like peptides, is likely to improve proliferation and help to control muscle size.72,106–109
The biological underpinnings of muscle development described in this study can also be used to inform broader aspects of in vitro muscle design, such as cell source, assay selection, and fate determination. It would be ideal to develop and maintain a stable line of AMPs, and the source species and location of such a line will ultimately determine the character of resulting muscle constructs. Once a source is selected, its characteristics can be optimized by genetic modification. Promising candidates for modification include cell surface receptors, which may bypass the need for external signals, and structural proteins that influence the force production and attachment strength of resultant constructs. Previous work in this field has already demonstrated that transfection with a constitutively active form of the FGF receptor Heartless is known to increase AMP proliferation in vivo, and transfection with the Caenorhabditis elegans fusogenic protein Eff-1 can initiate fusion in a nonfusing cell line, especially with overexpression of Sns.79,84 In addition, it might be possible to define factors controlling the identities of specific IPs and to use these cells to direct the eventual fate of the muscles as they develop in vitro. Eventually, it may also be possible to express specific identity genes, allowing a single stable cell line to produce many muscle types.
Any strategy to form organized muscle constructs will need to account for mechanical interactions and signaling between the myoblasts themselves and surrounding tissues. To this end, we expect that many advancements will take the form of improved scaffolds. Already, several scaffold designs have offered significant benefits to cultured muscle. For example, providing tension through the scaffold material improves myotube and myofibril alignment, 110 and micropatterning the growth substrate with proteins and textural features promotes myotube alignment and myofibrillogenesis.16,111 For example, in vitro muscle networks grown from Manduca embryonic tissues can be organized into linear muscle blocks attached to silk sutures at their ends and into muscle loops or rings. 17 These shapes are produced by growing cells in PDMS channels that confine and guide the elongating fibers, or by aligning them on micropatterned silk.
We propose that scaffolds will be further improved by coating surfaces with short-distance diffusible signals, such as growth factors, as well as contact-dependent signals, such as Duf and integrins. In cases where these messengers are chemically unstable, it may even be desirable to embed scaffolds with other cell types that will provide these signals naturalistically. Cell-embedded scaffolds with high viability have previously been fabricated using electrospinning and 3D printing approaches,112,113 and this year, this approach was used successfully to coculture human endothelial cells with myoblasts (C2C12 cells). 114
Metamorphosis offers many benefits for developing bioactuators, chief of which is the relative ease of studying muscles' natural development. While work in embryonic development has provided pivotal insights into muscle culture, metamorphic muscles allow us to identify tissue/tissue interactions and individual signaling pathways. By harnessing the work that has been done in various holometabolous insects, we will be able to make targeted, hypothesis-driven manipulations to engineered muscle systems and drastically improve bioactuation.
Footnotes
Disclosure Statement
The authors declare that there is no conflict of interest.
Funding Information
Preparation of this review was supported by the National Science Foundation grants IOS-1557672 and DGE-IGERT-1144591 to B.A.T.
