|Home | About | Journals | Submit | Contact Us | Français|
From the muscles that control the blink of your eye to those that allow you to walk, the basic architecture of muscle is the same: muscles consist of bundles of the unit muscle cell, the muscle fiber. The unique morphology of the individual muscle fiber is dictated by the functional demands necessary to generate and withstand the forces of contraction, which in turn leads to movement. Contractile muscle fibers are elongated, syncytial cells, which interact with both the nervous and skeletal systems to govern body motion. In this review, we focus on three key cell–cell and cell–matrix contact processes, that are necessary to create this exquisitely specialized cell: cell fusion, cell elongation, and establishment of a myotendinous junction. We address these processes by highlighting recent findings from the Drosophila model system.
The model system, Drosophila melanogaster, has been used to great effect to study fundamental issues of muscle development [1–7]. This model organism offers the cell biologist an in vivo system, coupled to a long-established genetic tradition to study muscle morphogenesis. In addition, application of genomics and varied imaging approaches makes this model a highly tractable system for the study of the cell biology of muscle.
Body wall muscles in Drosophila are generated twice during the life of the fly: first, during embryogenesis to form the larval muscles , the process that is the main focus of this review; and subsequently during metamorphosis, in which cells set aside during embryonic myogenesis are used to generate the adult muscles . In the embryo, a single fiber is considered a single muscle, whereas in the adult, multiple fibers constitute a single muscle. A similar mechanism, however, governs fiber formation in both situations: each fiber is seeded by a specialized myoblast, called a founder cell (FC), which fuses repeatedly with neighboring fusion competent myoblasts (FCMs) to generate a multi-nucleated myotube (Figure 1) . Upon fusion, the newly incorporated FCM-derived nuclei adopt the transcriptional profile of the FC/myotube. By virtue of a complex developmental specification process, individual FCs/myotubes express different combinations of cell identity regulators, which endow them with unique morphological characteristics, including size (i.e. the number of fusions with FCMs), shape, and spatial orientation [9,10••].
Fusion is accompanied by elongation of the growing myofiber, which navigates towards tendon cells that arise in the overlying epidermis. Through interactions with these tendon cells, a stable attachment forms between muscle, epidermis, and cuticle (the exoskeleton) (Figure 1). These initial, crucial myogenic processes – fusion, elongation, and attachment – are at the heart of this review. Innervation of each muscle fiber occurs after fusion and tendon attachment, while the stereotypical arrangement of the fiber contractile apparatus (sarcomeres) appears late during muscle morphogenesis, just before hatching [1,10]. The contractile properties and inter-cellular associations provide body wall muscles with the capacity to execute and govern larval motility.
A series of recent reviews provide a comprehensive, updated description of myoblast fusion in Drosophila and in vertebrates [11•,12,13,14•]. Here we discuss recent insights to fusion, emphasizing the remaining gaps in our knowledge, with particular emphasis on the contribution of the actin-based cytoskeleton.
Myoblast fusion, as in all cases of cell–cell fusion, requires several distinct cellular behaviors . Initially, FCs/myotubes and FCMs must recognize and adhere to each other. The result of recognition and adhesion is a series of cellular events that are necessary to bring the cell membranes of the two cells in close proximity to one another. Subsequently, the tethering of the plasma membranes, the formation of pores between the membranes and the expansion of these pores leads to the merging of the two myoblasts (Figure 1d).
In the Drosophila embryo, myoblast recognition and adhesion is mediated by a set of immunoglobulin (Ig) domain transmembrane proteins: Dumbfounded (Duf, a.k.a Kirre) and Roughest (Rst a.k.a IrreC), which are found primarily in FCs, and the FCM elements Sticks and stones (Sns) and Hibris (Hbs) [12–14,16•]. Fusion-related roles for vertebrate homologs of these proteins have been recently revealed [17,18], implying conservation of this system of cell recognition and adhesion. However, several significant issues pertaining to the underlying molecular mechanisms remain unclear, and require further investigation. Prominent among these are the initiation of contact between distant myoblasts (is a diffusible factor involved?), selection of fusion partners (‘first come, first serve’ vs. ‘specificity of partnering’), and generation of tightly apposed myoblast membranes, primed for fusion. Recent evidence suggests that the Ig domain recognition receptors form a ring-like structure, termed the FuRMAS [13,19•] at the fusion site. These data have been used to suggest that the recognition receptors, upon engagement, are cleared from the site of actual membrane fusion. The FuRMAS, by the nature of a ring-like structure, could also limit the site of membrane fusion to within this ring. Time-lapse imaging of tagged receptors would provide a significant step forward in verifying this model.
Downstream of the receptors, genetic, cell, and biochemical approaches have implicated a set of actin regulatory proteins as crucial for myoblast fusion. Specifically, mutations in genes including Rac, Myoblast city (Dock 180), Kette (Nap1), Scar (WAVE), Vrp1 (Solitary; D-WIP), WASp, and Arp2/3, lead to defects in myoblast fusion [11•,12,13,14•]. While clarification is needed as to how particular actin regulatory proteins are recruited specifically to the recognition receptors and the fusion site [20,21•,22,23,24••,25••,26••,27,28,29••,30], it is clear that one crucial result of their function is a branched actin structure, termed the actin focus [19•,25••,29••]. Time-lapse imaging of Drosophila myoblasts proved conclusively that this structure marks the fusion site and revealed the dynamics of this structure. The actin focus is on average short-lived (11.9 min, range 5.7–29 min), measuring 2 μm2 in size (range 0.7–4.5 μm2); this structure disappears as cytoplasmic continuity between the fusing cells is achieved (Figure 2). Genetic analysis reveals that actin focus formation requires recognition receptor function for its formation [29••]. Consistent with these genetic experiments is the observation that the FuRMAS surrounds the actin focus [13,19•]. While double labeling experiments using actin and membrane reporter constructs suggest that the actin focus can be found on both sides of the fusion event (myotube and FCM), the distribution of actin within the focus may be biased to one cell or the other [25••,29••] (B. Richardson, I. Bothe and MKB unpub.). Whether this indicates the existence of different actin structures or reveals novel aspects of actin focus maturation remains to be investigated.
With genetics implicating SCAR/WAVE and WASp regulation of Arp2/3 as crucial to myoblast fusion, a simple model would be that these actin regulatory pathways are required for actin focus formation. However, contrary to expectation, analysis of single and double mutants between different pathway members reveals persistent and even enlarged actin foci at fusion sites  (B. Richardson and MKB unpub). The genetic data indicate, therefore, that these pathways are not required for formation of the foci, but may ut may influence actin rearrangements leading to dissolution of these structures. While some debate continues, these data also illustrate that the proteins required for formation of the actin focus remain to be identified. Additional Arp2/3 regulators have been uncovered in other systems , providing new genes to investigate for this key role in actin regulation during myoblast fusion.
An essential question arising from these studies is the function of the actin focus at the fusion site. Time-lapse imaging correlates formation and, importantly, removal of the actin focus, with cytoplasmic continuity and ultimately cell–cell fusion. With the exception of the recognition receptors and a protein implicated in receptor recycling, mutants of which do not form actin foci, all other known fusion gene mutants display actin foci that perdure at the fusion site. Of these, some mutants have enlarged foci and others have wild-type sized foci [14,29••,32]. Testing is needed to determine whether the fusion block in any of these mutants is due to the inability of the cell to remove the focus specifically, or whether the long-lived focus is just a consequence of other events gone awry. Nevertheless, several models for the cellular function of the actin focus have been put forward. These include targeting vesicles to the site of fusion [25••], supplying scaffolding to maintain cellular integrity while fusion pores are created [29••], and/or providing a force for fusion pore enlargement . To provide the necessary data to evaluate these models, a combination of approaches are needed: identification of new genes and mechanisms via genetic and biochemical experiments; improved imaging, employing time-lapse videography and transmission electron microscopy; determination of composition and dynamics of cell membranes and membrane-bound vesicles associated with the fusion site; and finally, detailed description and analysis of fusion pore formation and expansion. While the Drosophila system has provided new views of a prominent and highly regulated form of cell–cell fusion, much remains to be investigated, particularly with regard to the cell-biological mechanism underlying the fusion process.
In parallel to their growth via fusion with FCMs, myotubes in the Drosophila embryo elongate towards the epidermis and attach to tendon cells at both ends (Figure 3). The highly stereotypic pattern of muscle–tendon attachments implies that regulatory mechanisms are at play, ensuring specificity and proper execution of the myotube targeting process. It is well-established that tendon cells produce spatial cues influencing the direction towards which myotubes extend. The muscles, in turn, seek out their attachment sites by sending out filopodia-like extensions that survey their environment for these instructive cues [1,33].
Perhaps the sole bona fide identification of a tendon-derived guidance signal has been described in the ventral-longitudinal (VL) muscles, a repeating set of four parallel fibers that stretch between epidermal attachments sites at the adjacent embryonic segment border. Segment-border tendon cells secrete the ECM glycoprotein Slit, which directs myotube targeting by binding to the Roundabout (Robo) receptor, expressed at the edges of developing VL muscle fibers [34•]. Therefore, although the Slit–Robo ligand–receptor interaction is best known as the basis for repulsion of both axons and muscle fibers away from the ventral midline of the Drosophila embryo , it also functions as an attractive cue, guiding myotubes to specific epidermal target sites. Mutations in which properly fused myotubes extend randomly and fail to connect to their epidermal target sites have recently uncovered a second myotube targeting mechanism, apparently acting in parallel to the Slit/Robo pathway [36••,37•,38••,39••]. Importantly, mutant myotubes of this class are capable of forming stable attachments, but do so with incorrect tendon target cells, or even with adjacent muscles, implying specific impairment of target recognition capacity, while other aspects of myotube differentiation remain unaffected. A key player that has emerged from this group is D-Grip, the sole fly homolog of the Grip (Glutamate receptor interacting protein) adaptor protein family. Grip proteins harbor multiple PDZ domains, and have been functionally implicated in various aspects of synaptic protein trafficking and localization [40–42]. In Drosophila embryos, D-Grip is specifically expressed in developing muscles, and the D-Grip protein accumulates within endosomes at the ends of extending myofibers .
While D-Grip probably acts as a mediator of intracellular trafficking, it is becoming apparent that this multi-PDZ-domain adaptor serves a key myogenic role as a common platform for multiple muscle factors contributing to myotube targeting. Prominent among these is Kon-tiki (Kon; a.k.a Perdido), a transmembrane protein and potential receptor mediating recognition of tendon-cell guidance signals by growing embryonic myotubes [36••,38••]. Kon associates with D-Grip through a C-terminal PDZ binding domain, localizes to the ends of extending VL myotubes, and is required for D-Grip localization to these sites. Kon contributes to additional aspects of muscle–tendon attachment. One of these is to regulate the extent of motile exploratory activity by myotube ends [38••]. Interestingly, a similar function has been recently proposed for the tendon cell transmembrane protein LRT [43•], which can physically associate with Robo receptors. LRT may therefore restrict myotube migratory activity by signaling through Robo, or indirectly, by competing with Slit for Robo receptor binding, thereby stifling its stimulatory effect on myotube dynamics [43,44]. Such utilization of both muscle and tendon-based programs underscores the importance of setting limits on guidance-seeking mechanisms, once initial target recognition is achieved.
An additional member of the D-Grip-based guidance machinery is the Ig superfamily transmembrane protein Echinoid (Ed) [45•]. Structurally related to the vertebrate Nectins, Ed functions as a homotypic cell adhesion molecule, mediating aspects of Drosophila neurogenesis and epithelial morphogenesis [46–48]. Similar to Kon, Ed associates with D-Grip PDZ domains via its extreme C-terminal region. Furthermore, Ed colocalizes with D-Grip at the ends of VL muscles, and the two elements display strong genetic interactions with respect to VL myotube extension defects . Targeting and guidance errors rather than defective adhesion are the main characteristic of embryos in which the Ed/D-Grip system is impaired, implying that Ed and D-Grip influence myotube behavior by affecting cortical and cytoplasmic features of these cells. Significantly, Ed affects the outcome of cell–cell interactions through communication with cortical micro-filaments [48–50], and therefore serves as a good candidate for a link between the D-Grip platform and the actin-based cytoskeleton.
This series of studies has therefore revealed a molecular complex, positioned at the tips of growing muscle fibers, which integrates a number of inputs aimed at achieving proper myotube targeting. Remarkably, a highly similar Grip-based platform mediates glia–neuron interactions during mammalian CNS development . In this instance, the PDZ domains of Grip bring together several glial cell-surface proteins, including NG2, a mammalian homolog of Kon [52,53], which is thought to interact with a neuronal receptor. Cortical complexes of this type therefore represent a conserved machinery for instructive communication between distinct cell types, leading to functional maturation of a differentiating tissue.
While involvement of the Grip-based complex in myotube targeting implies functional reliance on the actin-based cytoskeleton, mutations resulting in defective muscle targeting have now revealed that growing myotubes also require properly polarized microtubule (MT) arrays, in order to reach their correct attachment sites [37•]. Key elements involved in this system are Tumbleweed (Tum) and Pavarotti (Pav), established components of the cytokinesis-mediating complex centralspindlin [54–56]. In Drosophila embryonic muscles, Tum and Pav are jointly responsible for localization of non-centrosomal, γ-tubulin-based microtubule organizing centers to the vicinity of myotube nuclei. These nuclei commonly cluster in the center of growing multi-nucleated fibers [37•]. The resulting microtubule array that forms extends from the myotube interior towards the periphery, and is polarized, with the microtubule plus-ends pointing towards the extending tips of the growing fibers. A similarly oriented microtubule array, nucleated from non-centrosomal organizing centers, forms in cultured mammalian myotubes , suggesting that reorganization of the microtubule cytoskeleton in this fashion reflects a conserved requirement during muscle differentiation. From a functional standpoint, the axial microtubule array may serve as a structural framework enabling elongation of myotube ends [37•,57], or alternatively, may provide polarized tracks for trafficking and localization of crucial guidance and targeting signals that ensure proper selection of myotube attachment sites.
Contact between myotubes and tendon cells is closely followed by establishment of a myotendinous junction, so that muscle and epidermis maintain a strong physical connection that will be able to withstand the considerable forces imposed on it once muscle contractions initiate . Integrins play a major, conserved role in this process, and indeed, the Drosophila embryonic myotendinous junction now serves as a prominent example and setting for study of integrin-based adhesion between distinct cell types . Both muscle and tendon cells express αβ integrin heterodimers, which fortify cell attachment to the junction by serving as transmembrane links between the ECM and the internal actin cytoskeleton. These heterodimers are composed of a common β subunit (βPS) and distinct α subunits—αPS1 for the tendon cell integrin and αPS2 in muscles. Genetic disruption of the muscle αPS2βPS integrin does not interfere with construction, elongation, and attachment of muscle fibers. However, once the muscles of such mutant embryos begin to contract, they disconnect from their attachment sites and retract into ball-shaped structures [60–63]. This classic ‘myospheroid’ phenotype now serves as a diagnostic tool for identification of additional elements contributing to formation and consolidation of myotendinous junctions.
Several ECM components act as ligands for the Drosophila αPS2βPS integrin in the context of the myotendinous junction, including the Laminin α-chain protein Wing Blister , and Tiggrin [65,66]. Recent studies have now identified Drosophila Thrombospondin (Tsp) as a major αPS2βPS integrin ligand at muscle attachment sites [67••,68••]. Vertebrate Thrombospondins comprise a family of multimeric ECM glycoproteins, which have been implicated in diverse functional settings, such as cell aggregation and attachment, angiogenesis and synaptogenesis [69–71]. Drosophila Tsp is produced and secreted by tendon cells at stages corresponding to establishment of muscle–tendon contacts. The muscle detachment phenotypes of Tsp mutant embryos, genetic interactions of Tsp with integrin encoding genes, and biochemical binding assays, combine to identify Tsp as a ligand for αPS2βPS. The integrin-binding capacity and localization patterns of Tsp suggest a dynamic sequence [67••,68••], in which Tsp first contributes to the initial association of myotube ends with the tendon cell and its ECM. Subsequently, muscle–tendon interactions set in motion a general program of tendon cell differentiation , leading to increased expression and secretion of Tsp, which in turn fortifies the integrin-mediated myotube attachment to the myotendinous junction.
Further insight to the regulation of Tsp activity comes from the study of Slowmotion (Slow), a new player in myotendinous junction construction [73•]. Slow, the Drosophila homolog of vertebrate EGFL7 [74,75], is secreted from tendon cells in parallel to Tsp, and attenuates the Tsp-αPS2βPS integrin interaction by forming a complex with Tsp. Streamlining of muscle integrin activity in this manner turns out to be crucial for proper morphogenesis of the myotendinous junction into a structure capable of withstanding the wear-and-tear of intense larval muscle activity .
Progress in the understanding of integrin–ECM interactions in the context of myotendinous junction formation has been matched by refined characterization of the manner by which integrins are linked to cytoplasmic and cytoskeletal components. Thus the Drosophila homolog of ZASP (Z-band alternatively spliced PDZ-motif protein), a member of the PDZ-LIM protein family [76,77], acts as an adapter that strengthens the integrin–actin connection following its initial establishment via Talin, a ubiquitous mediator of integrin-based processes [78••]. Zasp performs this role repeatedly during myogenesis, initially to bolster the myotendinous junction, and later to mediate integrin-dependent assembly of muscle fiber sarcomeres .
A second adaptor of this type is Wech [79••], a well-conserved member of the growing, multi-domain TRIM protein family, the functional attributes of which are only beginning to emerge [80,81,82]. At Drosophila myotendinous junctions, Wech acts within tendon cells and myotubes to bridge a key interaction between Talin, which associates with the βPS integrin subunit, and the Integrin-linked Kinase (ILK) complex, a crucial mediator of the integrin-microfilament link . This specific adaptor capacity of Wech appears to be conserved in mammals , suggesting that a fundamental aspect of integrin-mediated adhesion has now been uncovered.
In this review we have focused on a restricted set of myogenic processes, to demonstrate the power of the Drosophila embryo as a model system for studying muscle cell properties and behaviors. A wide variety of issues, such as cell size, shape, polarity, migration, and adhesion can all be addressed using this versatile system. The relative simplicity and segmentally repeated nature of the embryonic musculature, amenable to study via classical and modern genetic approaches coupled to imaging and biochemical analysis, provide a unique, in vivo experimental setting. The recent introduction of comprehensive RNAi-based genetic methods to the Drosophila field [84,85••], along with the application of large-scale genomic techniques [36••,86–88] now promises to significantly enhance these studies. For example, the capacity to control RNAi expression in a temporal and tissue-specific manner helps overcome a variety of obstacles confronting more conventional genetic approaches. Importantly, this tool now enables comprehensive exploration of adult fly myogenesis, and we expect that considerable insight to issues of muscle cell biology will emerge from this ‘new frontier’ in the years to come.
We would like to thank members of the Baylies lab, particularly K. Dobi for critical reading, and S. Yu, I. Bothe, and T. Metzger for figures. We also thank T. Volk for kindly providing Figure 3c. We acknowledge our funding agencies: NIH (GM078318 and GM056989) and MDA to MB, ISF, and MDA to B. Shilo and EDS, and the Mary L. Ralph Fund to MB, B Shilo and EDS.
Papers of particular interest, published within the annual period of review, have been highlighted as:
• of special interest
•• of outstanding interest