Identification and functional characterization of muscle satellite cells in Drosophila

Work on genetic model systems such as Drosophila and mouse has shown that the fundamental mechanisms of myogenesis are remarkably similar in vertebrates and invertebrates. Strikingly, however, satellite cells, the adult muscle stem cells that are essential for the regeneration of damaged muscles in vertebrates, have not been reported in invertebrates. In this study, we show that lineal descendants of muscle stem cells are present in adult muscle of Drosophila as small, unfused cells observed at the surface and in close proximity to the mature muscle fibers. Normally quiescent, following muscle fiber injury, we show that these cells express Zfh1 and engage in Notch-Delta-dependent proliferative activity and generate lineal descendant populations, which fuse with the injured muscle fiber. In view of strikingly similar morphological and functional features, we consider these novel cells to be the Drosophila equivalent of vertebrate muscle satellite cells.


Introduction
A great deal of insight into the cellular and molecular mechanisms of muscle development has been obtained in two powerful genetic model systems, the mouse and Drosophila. Despite numerous differences in the specific ways in which muscles are formed in these two organisms, there are also remarkable similarities in the fundamental developmental processes that underlie myogenesis (Roy and VijayRaghavan, 1999;Rai et al., 2014;Bothe and Baylies, 2016;Schnorrer et al., 2010). These similarities are most clearly evident when the mechanisms of myogenesis of the large multifibrillar indirect flight muscles of Drosophila are compared to vertebrate skeletal muscles. In both cases, muscle stem cells generated during embryogenesis give rise to a large pool of muscle precursor cells called myoblasts that subsequently fuse and differentiate to produce the multinucleated syncytial cells of the mature muscle. These mechanistic similarities of myogenesis are reflected at the molecular genetic level, in that many of the key genes involved in Drosophila muscle development have served as a basis for the identification of comparable genes in vertebrate muscle development (e.g. (Srinivas et al., 2007;Abmayr and Pavlath, 2012).
In vertebrates, mature skeletal muscle cells can manifest regenerative responses to insults due to injury or degenerative disease. These regenerative events require the action of a small population of tissue specific stem cells referred to as satellite cells (Bothe and Baylies, 2016;Mauro, 1961;Brack and Rando, 2012;Relaix and Zammit, 2012). Muscle satellite cells are located between the sarcolemma and the basal lamina of muscle fibers. Although normally quiescent, satellite cells respond to muscle damage by proliferating and producing myoblasts, which differentiate and fuse with the injured muscle cells. Myoblasts generated by satellite cells are also involved in the growth of adult vertebrate muscle. Given the numerous fundamental aspects of muscle stem cell biology and myogenesis that are similar in flies and vertebrates, it is surprising that muscle satellite cells have not been reported in Drosophila. Indeed, due to the apparent absence of satellite cells in adult fly muscles, it is unclear if muscle regeneration in response to injury can take place in Drosophila.
In a previous study, we showed that a small set of embryonically generated muscle-specific stem cells known as AMPs (adult muscle progenitors) give rise post-embryonically to the numerous myoblasts which fuse to form the indirect flight muscles of adult Drosophila (Gunage et al., 2014). Here, we show that muscle stem cell lineal descendants are present in the adult as unfused cells which have all the anatomical features of muscle satellite cells. In adult muscle, these unfused cells are located in close proximity to mature muscle fibers and are surrounded by the basal lamina of the fibers. Moreover, although normally quiescent, following muscle injury they undergo Notch signaling-dependent proliferation to generate fusion-competent lineal descendants. In view of these remarkable developmental, morphological and functional features, we consider these cells to be the Drosophila equivalent of vertebrate muscle satellite cells. Thus, in flies and vertebrates the muscle stem cell lineage that generates the adult-specific muscles during normal development is also available for adult myogenesis in muscle tissue in response to damage. This finding further opens adult Drosophila muscle for the understanding of muscle maintenance, wasting, damage and repair.

Two different types of cells are present in adult flight muscle
During normal postembryonic development of the indirect flight muscles, a set of approximately 250 mitotically active adult muscle precursors (AMPs) located on the epithelial surface of the wing imaginal disc generates a large number of postmitotic myoblast progeny. These myoblasts subsequently migrate and fuse to form the indirect flight muscles (IFMs) of the adult. The IFMs are composed of the dorso-ventral muscles (DVMs) formed by the de novo fusion of myoblast and the dorsal longitudinal muscles (DLMs) which are formed using remnant larval muscles as templates (Gunage et al., 2014;Fernandes et al., 1991;Dhanyasi et al., 2015).
Consistent with their developmental origin, which results in a large myoblast pool, the IFMs are large multinucleated cells formed by myoblast fusion. For convenience we focus here on the DLMs. The multinucleate nature of these muscles is evident in confocal optical sections through adult flight muscle fibers labeled by TOPRO (marks all nuclei) and myosin heavy chain (MHC) immunostaining, which marks myofibers. As expected, numerous nuclei, clearly located intra-cellular between individual myofibrils, are seen throughout the muscle fiber ( Figure 1A, A', white arrowheads). Interestingly, however, these optical sections also reveal nuclei located peripherally in close proximity to the muscle fiber surface ( Figure 1A, A', green arrowheads).
Additional co-labeling of these adult muscle fibers with Dmef2-Gal4 driving mCD8GFP (marking muscle fiber membranes) indicates that these peripherally located nuclei belong to cells at the muscle fiber surface, which are apparently not fused with their associated muscle fiber cell ( Figure 1B, B', white arrowheads). Figure 1C, D show some these nuclei at higher magnification revealing that they are located at the muscle fiber surface and are surrounded by membrane-specific mCD8GFP label implying that these nuclei are not situated inside the muscle fiber. This observation is confirmed by co-staining these adult muscle fibers for expression of either Act88F, an indirect flight muscle-specific isoform of actin, or Tropomyosin ( Figure 1-figure supplement 1).
Scans along the z-axis through co-labeled optical sections of muscle fibers indicate a small number of GFP-positive cells located closely associated with the surface muscle of fibers but remain unfused. To determine the relative numbers of peripherally located unfused cells versus fused differentiated myoblasts, all optical sections (along the z-axis) of the co-labeled adult DLM muscle fibers were scored for cells associated with the surface of the muscle fibers versus cell nuclei located within the flight muscle fibers. These experiments (n = 12) indicate that a DLM muscle fiber has an average of 20 ± 4 unfused cells versus an average of 700 ± 50 fused myoblasts. Hence the ratio of unfused cells to differentiated fused myoblasts is 1:30 implying that the population of surface-associated cells is markedly smaller than the population of fused myoblasts.
Taken together, these findings indicate that two different types of cells are present in adult muscle. The first comprises the well-characterized population of differentiated myoblasts that have fused B.

B' .
A' . A. Figure 1. Unfused muscle associated cells are present at the surface of adult flight muscles. (A) Dorsal longitudinal muscles (DLMs) stained for Myosin Heavy Chain (MHC) (red) to delineate muscle fibrils and TOPRO (Blue) marking nuclei. White arrowhead marks one example nucleus, surrounded by MHC-labeled myofibrils showing it is inside a myofiber. Green arrowhead, white circle, marks one example nucleus located at the peripheral surface of MHC Figure 1 continued on next page to generate the large multinuclear muscle fibers. The second comprises a novel population of small, apparently unfused cells located at the surface of the muscle fibers. In the following, we will refer to these small, unfused muscle fiber-associated cells as Drosophila satellite cells.

Ultrastructure of satellite cells in adult flight muscle
To characterize the morphological features of the close association of satellite cells with the large multinucleated muscle fibers at the ultrastructural level, an electron-microscopic analysis of adult DLM fibers was carried out. In electron micrographs, the mature muscle fibers are large cells containing multiple prominent nuclei, numerous organelles, as well as extensive sets of elongated myofibrils, and are surrounded by a prominent extracellular matrix ( Figure 2A). In addition to these typical multinucleated muscle cells, the ultrastructural analysis also shows satellite cells as small, wedge-shaped cells closely apposed to the large multinucleated muscle fibers ( Figure 2B, Figure 2-figure supplement 1). These satellite cells have compact nuclei and small cytoplasmic domains with few organelles. The intact cell membrane of the satellite cells is directly adjacent to the intact muscle cell membrane demonstrating unequivocally that they are not fused with the muscle cells. They do, however, appear to be embedded in the same contiguous extracellular matrix of their adjoining muscle fiber.
In terms of their ultrastructural morphology, the satellite cells in adult flight muscle share significant characteristics with satellite cells of vertebrate muscle. In both cases, the cells are small, mononucleated and intercalated between the cell membrane and the extracellular matrix of mature muscle fibers.

Drosophila satellite cells are lineal descendants of adult muscle precursors
Previous work has shown that the myoblasts which fuse to generate adult muscle derive from a small set of stem cell-like AMPs (Gunage et al., 2014). Proliferating AMPs located on the larval wing disc can be identified by clonal MARCM labeling experiments using a Dmef2-Gal4 driver (Lee and Luo, 2001;Wu and Luo, 2006); ( Figure 3A).
Given the fact that Drosophila satellite cells, like myoblasts, are labeled by the Dmef2-Gal4 driver and considering their close ultrastructural association with muscle fibers, we wondered if these satellite cells might also be lineal descendants of AMPs ( Figure 3B). To investigate this, we induced MARCM clones in late larval stages and recovered labeled clones in the adult muscle. In these experiments, Dmef2-Gal4 was used to drive a GFP reporter label, muscle cells were co-labeled using MHC immunostaining, and cell nuclei were co-labeled with TOPRO. Labeled cells in the adult were visualized using confocal microscopy and analyzed in serial stacks of optical sections.
These clonal labeling experiments reveal the presence of a small number of GFP labeled satellite cell nuclei closely apposed to the surface of the adult muscle fibers. A reconstructed 3D view of optical sections shows that these GFP-positive nuclei are distributed along the entire surface of the muscle fibers and located both at the interface between different muscle fibers and at the surface of individual muscle fibers but not within the muscle fibers ( Figure 3C,D). Note that while the nuclei of differentiated myoblasts within the muscle fiber are also targeted in this MARCM experiment, no clonal UAS-GFP labeling is visible due to the persistent expression of the Gal80 repressor by the unlabeled nuclei in the multinuclear muscle cell. Hence, in these experiments, cell nuclei can only be labeled if their cells are unfused and remain outside of the muscle fiber. These clonal MARCM labeling experiments are in accordance with the notion that Drosophila satellite cells are lineal descendants of AMPs, which, in contrast to myoblasts, do not fuse with the mature muscle cells but rather persist as unfused cells in the adult albeit closely associated with the mature muscle fibers. To control for a possible contribution of hemocytes lineages to the unfused cell population, we used the e33c-Gal4 line to show that hemocytes are not seen in muscles (Figure 3-figure supplement 1; for expression pattern of the e33c-Gal4 in hemocytes see Fossett et al. (2003) and Matova and Anderson (2006).

The Zfh1 transcription factor is a marker for satellite cells in adult muscle
To facilitate the analysis of satellite cells, we searched for a protein specifically expressed in adult DLM satellite cells by examining the expression of a set of transcription factors implicated in embryonic muscle specification and myoblast fusion competence. Among these, we identified Zfh1 as a specific marker for adult muscle satellite cells. Zfh1is a zinc finger transcription factor that regulates somatic myogenesis from embryonic stages onward (Postigo and Dean, 1999;Sellin et al., 2009). In larval stages, Zfh1 is expressed in all of the AMP lineage cells on the wing discs; both AMPs and myoblasts are labeled ( Figure 4A). In contrast, in adults, Zfh1 is specifically and exclusively expressed in the unfused muscle satellite cells. Thus, in adult muscle, cells labeled by a Zfh1 antibody are clearly unfused and located closely apposed to the mature muscle fibers ( Figure 4B-D). Similar highly specific labeling of unfused muscle satellite cells is observed in adult muscle with either a GFP-tagged Zfh1 protein or with a Zfh1fused Gal4 driver and a UAS-RedStinger reporter ( Figure 4E,F) (Puretskaia et al., 2017). In all cases, Zfh1 expression is limited to unfused cells and is never seen inside intact adult muscle fibers.
Since Zfh1 is expressed in all AMP lineage cells in larval stages but is restricted to unfused satellite cells in the adult, its expression pattern must change dramatically in pupal stages. To document this, we carried out an analysis of Zfh1 expression at different time points during DLM development in representative pupal stages.
Indirect flight muscle development during pupal stages has been characterized in detail (Fernandes et al., 1991;Roy and VijayRaghavan, 1998;Bate et al., 1991). During the earliest pupal stages at the onset of metamorphosis, the AMP-descendent myoblasts are still located on the third instar wing disc notum. Subsequently, these myoblasts migrate from the wing disc and begin to swarm over a set of persistent larval muscles referred to as DLM templates that act as scaffolds for the developing DLMs. By 20 hr after puparium formation (APF) numerous myoblasts are present around and between the six DLM templates, fusion of myoblasts with these transformed DLM templates is ongoing, and myoblast nuclei are observed inside the developing DLMs. By 30 hr APF, most myoblasts have fused with the DLMs and by 36 hr APF, myogenesis of the adult DLM muscle is essentially complete.
Immunolabeling experiments show that at 20 hr APF all of the nuclei of myoblasts around and in between the DLM templates express Zfh1 ( Figures 5A and 6A). Moreover, the nuclei that are located inside the DLM templates due to the fusion of their myoblasts with the templates also express Zfh1. Thus, at this pupal stage, most if not all the nuclei of the AMP lineage myoblasts, be they fused within DLM templates or unfused outside of these templates, continue to express Zfh1 comparable to the situation in larval stages (see Figure 4A). In contrast, by 30 hr APF a dramatic change in the number and location of Zfh1 expressing cell nuclei has occurred. Zfh1 expression is only seen in a very small number of unfused cell nuclei located outside, albeit closely apposed to, the DLM muscle fiber templates ( Figures 5B and 6B). None of the nuclei located inside of the developing DLM muscle fibers express Zfh1. Thus, by 30 hr APF, when myoblast fusion is largely complete and myofibrils become visible in the muscle, Zfh1 expression has become restricted to a small number of unfused AMP lineal cells comparable to the situation in the adult, in which the only AMP lineal descendants that continue to express Zfh1 are the unfused satellite cells.
Taken together, these findings establish Zfh1 expression as a specific marker for satellite cells in adult muscle. Moreover, they characterize the dramatic change in Zfh1 expression that occurs in DLM muscle development between 20 hr and 30 hr APF, and provide further support for the lineal origin of satellite cells from the muscle stem cell-like AMPs. cell lies in close proximity to wing disc epithelium. Phalloidin (red) marks F-actin and TOPRO (blue) marks all the nuclei. (B). Simplified schematic describing lineal origin of adult DLM fibers and satellite cells. AMPs (green circles) on the third instar wing disc notum give rise to myoblasts (beige circles) located distal to the epithelium. Cells from the AMP lineage either fuse to muscle templates and give rise to adult DLMs or remain unfused as mononucleate cells closely apposed to the DLM surface. (C). MARCM clones with mef2-Gal4 driving UAS mCD8:: GFP induced in the third instar (~120 hr AEL) and recovered in the adults stage. 3D reconstruction of adult muscle with mononucleate GFP-labeled cells (green arrows) located on the surface of mature DLM fibers. Phalloidin (red) marks F-actin and TOPRO (blue) marks all the nuclei. (D) Orthogonal view of the same preparation as in (C). mef2-Gal4-labeled mononucleate MARCM clones (GFP positive) indicated with green arrowheads clearly seen on the surface of adult DLMs. (E) Simplified schematic of (D) Red checkered ovals containing blue ovals indicate mature DLMs. Cells with red membranes and green nuclei represent satellite cells. Scale bars in (A, C, D) 50 mm. N = 12. DOI: https://doi.org/10.7554/eLife.30107.006 The following figure supplement is available for figure 3:  Vertebrate satellite cells are essential for muscle regeneration and repair in that muscle damage results in proliferative activity of satellite cells and the production of myoblasts that help rebuild compromised muscle tissue (Mauro, 1961;Brack and Rando, 2012;Relaix and Zammit, 2012). To investigate if satellite cells in Drosophila can also respond to muscle injury by increased proliferate activity, we induced physical damage in adult flight muscles mechanically and subsequently probed the damaged muscle for proliferative activity in the satellite cell population.
To induce muscle damage, localized stab injury of DLMs was carried out in adult flies using a small needle; care was taken to restrict damage such that only 1 or 2 muscle fibers were affected (Figure 7). DLMs damaged in this way can regenerate. While damage is still clearly evident 2 days after injury, significant morphological regeneration is manifest after 5 days, and after 10 days regeneration has progressed such that only small remnants of the injury are apparent (Figure 7) To determine if muscle damage results in proliferative expansion of satellite cells, we compared the number of Zfh1-Gal4 labeled cell nuclei in uninjured control muscles versus injured muscles 24 hr after damage using UAS-nlsRedStinger. A dramatic increase in the number of Zfh1-labeled nuclei was seen in the damaged muscle as compared to controls ( Figure 8A, B). This increase in Zfh1-positive nuclei number was strongest in the damaged muscle fibers and less pronounced in neighboring undamaged fibers. In contrast, in the damaged muscle fiber, increases in the number of Zfh1-labeled nuclei were seen along the entire extent of the fiber length. As expected, the Zfh1 labeled nuclei in the damaged muscle, as in controls, were largely located at the surface of muscle fibers; few, if any, of the Zfh1 labeled nuclei observed in these experiments were located within the injured muscle fiber. Interestingly, and in contrast to the situation in uninjured controls, many of the numerous Zfh1labeled nuclei associated with the injured muscle appear to be manifest as spatially adjacent couplets ( Figure 8C, D).
Taken together, these findings indicate that physical damage leads to a marked proliferative expansion in the satellite cell population associated with the injured flight muscle fiber. Given that the satellite cell population undergoes proliferative expansion following injury of adult muscle fibers, might some of their lineage correspond to cells that can fuse with the damaged muscle?
To investigate this, we used the Zfh1-Gal4, Gal80ts driver in G-trace experiments. G-trace (Gal4 technique for real-time and clonal expression) is a dual color genetic labeling technique based on Gal4 activity (Evans et al., 2009). The reporters used are RFP (red fluorescent protein) and GFP (green fluorescent protein), where RFP expression is strictly dependent on ongoing real-time Gal4 activity, while GFP expression is lineally dependent on previous Gal4 activity but independent of ongoing Gal4 activity.
G-trace labeling was induced in 1-to 3-day-old adults for 72 hr before muscle injury and recovered at various times (24 hr, 48 hr, 1 week) after muscle injury (Figure 9). At 24 hr after muscle injury, most of the labeled cell nuclei were both RFP positive and GFP positive (i.e. yellow), signifying both real-time and lineage-dependent previous activity of the Zfh1-Gal4 driver in these cells. Moreover, as expected for Zfh1-labeled cells in adult muscle, all of these were located at the muscle fiber surface (see above). Similar findings were obtained at 48 hr after muscle injury; both RFP-positive and GFP-positive cell nuclei were manifest at the muscle surface, while none were seen inside the bulk of DLMs. In contrast, 1 week after injury, labeled cells that appeared to be located inside muscle fibers were observed. Thus, in addition to RFP-positive and GFP-positive cell nuclei located at the surface of the muscle, GFP-positive nuclei indicative of lineage-specific previous activity of the Zfh1-Gal4 driver appeared to be positioned inside muscle fibers, albeit very close to their surface. Interestingly, the shape of these nuclei appeared to be flattened or disc-like in contrast to the round shape of the nuclei located outside of the muscle cell surface.  In uninjured adult muscle, nuclei of the Zfh1 lineage are always located at the muscle cell surface and are never seen inside the muscle fiber (see above). Hence, the possibility that nuclei of the Zfh1 lineage might be located inside the muscle 1 week after injury implies that the corresponding Zfh1 lineal cells have fused with the muscle fiber. To investigate the possibility that Zfh1 lineal progeny might have fused with mature DLMs after injury, we repeated these G-trace experiments in the background of alpha Spectrin immunolabeling. Alpha Spectrin is a plasma-membrane-associated protein that marks muscle cell boundaries and can also be seen at low levels in the cytoplasm of DLMs (LaBeau-DiMenna et al., 2012).
In age matched, uninjured animals with G-trace induced in the adult stage, Zfh1 lineage cell nuclei that are both RFP positive and GFP positive can be seen closely associated with surface of the DLM as delimited by alpha Spectrin labeling but clearly located outside of the muscle cell ( Figure 10A). In contrast, in animals 1 week after muscle injury, nuclei of Zfh1 lineage cells that are GFP labeled are located within the muscle fiber albeit near its surface ( Figure 10B). Remarkably, these nuclei have a flattened, disc-like shape and appear to be markedly larger than those of uninjured controls. Similar findings are obtained in comparable G-trace experiments in which Vinculin immunolabeling is used to demarcate the muscle fiber surface. In uninjured control muscle fibers, RFP-and GFP-positive Zfh1 lineage nuclei are located at the outer surface of the muscle fiber ( Figure 10C). In muscle fibers 1 week after injury, GFP labeled Zfh1 lineage cell nuclei are positioned inside the muscle cell but remain near the muscle cell surface ( Figure 10D).
Taken together, these findings imply that following muscle injury, cells of the Zfh1 lineage, that is, lineal descendants of the Zfh1-expressing satellite cells, fuse with the damaged muscle fibers. This fusion process, which is preceded by a proliferative expansion of the normally quiescent satellite cell population, may contribute to the promotion of muscle fiber repair and, hence, represent a remarkable functional similarity in the role of satellite cells in response to muscle injury in flies and vertebrates.

Proliferative activity of satellite cells in response to muscle injury requires Notch expression in satellite cells and Delta expression in muscle fibers
It has previously been shown that proliferative activity of AMPs during development requires Notch signaling (Gunage et al., 2014). Might the AMP lineal descendant satellite cells in adult muscle also require Notch signaling for injury-induced proliferative activity? To investigate this, we first determined if satellite cells express Notch. For this, we used a Notch-Gal4 driver in G-trace experiments. In uninjured controls, all the nuclei within the muscle fiber were GFP-positive, in accordance with their lineal origin from Notch expressing AMPs, but none were RFP positive ( Figure 11A). In contrast the muscle surface associated nuclei were RFP positive due to real-time activity of the Notch-Gal4 driver in implying that the nuclei of satellite cells express Notch. To establish that the RFP-positive nuclei were indeed the nuclei of satellite cells, we combined these G-trace experiments with Zfh1immunolabeling. In these experiments the RFP-positive nuclei at the muscle fiber surface were always Zfh1-positive, while the GFP-positive nuclei within the muscle fiber were Zfh1-negative ( Figure 11B). These findings indicate that satellite cells in intact muscle fibers express Notch.
Following muscle injury, an expansion of the satellite cell lineage takes place (see above). To determine if the cells in this expanded lineage continue to express Notch, we repeated the G-trace experiments 24 hr after muscle injury. In these experiments, as in the uninjured control, all of the muscle surface associated nuclei were RFP positive due to real-time activity of the Notch-Gal4 driver implying that the nuclei of the expanded satellite cell population continue to express Notch ( Figure 11C). Moreover, quantification of the number of Notch-Gal4 expressing cell nuclei in injured muscle fibers versus uninjured controls indicates that an approximately twofold expansion in the number of Notch expressing satellite cells occurs 24 hr after muscle injury ( Figure 11D).
We next investigated if muscle fibers might express the Notch ligand Delta. Immunolabeling of uninjured flight muscles revealed a significant albeit low level of Delta expression ( Figure 12A). In contrast, a dramatic increase in Delta expression was observed in injured flight muscles ( Figure 12B). Indeed, a quantification of the intensity of immunolabeling in control versus injured muscles indicates that a fourfold increase in Delta expression occurs in response to injury ( Figure 12C). A comparable upregulation of Neuralized, an E3-ubiquitin ligase required in the Delta-Notch signal transduction process for Delta endocytosis, was also observed in damaged muscle versus controls ( Figure 12D, E). Analysis of a Neuralized-LacZ reporter line indicates that the muscle fiber-specific expression of Neuralized is significantly up-regulated following muscle injury ( Figure 12F). Taken together these findings indicate that following muscle injury, Notch is expressed throughout the expanding satellite cell population. Moreover they indicate that Delta expression is upregulated in the injured muscle fibers. Might signaling between muscle fiber associated Delta ligand and satellite-cell-associated Notch receptor be required for the proliferative mitotic activity of satellite cells in response to muscle injury?
To investigate this possibility, we used the mitotic marker phosphohistone-H3 (PH-3) on injured flight muscle. PH-3 labeling was carried out 12 hr after muscle injury. In injured wild-type controls, numerous satellite cells were PH-3 positive, indicative of the extensive mitotic activity associated with injury-induced satellite cell expansion ( Figure 13A).
In contrast, in Act88F-driven (muscle-specific) Delta-RNAi knockdown experiments, limited to adult stages by Gal80-ts, a dramatic reduction in the number of PH-3 labeled satellite cells was observed in injured muscles as compared to controls ( Figure 13B). A quantification of this reduction in PH-3-labeled satellite is shown in Figure 13C. Comparable results were obtained when a dominant negative form of Delta was expressed using the Act88F driver in injured muscle fibers ( Figure 13C). These findings indicate that Delta expression in muscle fibers is required for the induction of injury-dependent mitotic activity of satellite cells.
To determine if Notch expression in satellite cells is similarly required for injury induced mitotic activity of satellite cells; comparable PH-3 labeling experiments were carried out using a temperature sensitive Notch allele. A quantification of the number of PH-3-labeled cells in the injured muscle of Notch temperature-sensitive allele flies at permissive (17˚C) versus restrictive (29˚C) temperature is shown in Figure 11D. While numerous satellite cells were PH-3 positive at the permissive temperature, at the restrictive temperature only few cells were PH-3-positive. (Similar results were obtained by using the chemical inhibitor DAPT, a gamma-Secretase inhibitor, to block Notch pathway activity; data not shown). This finding was confirmed in Dmef2-driven Notch-RNAi knockdown experiments, in which the knockdown was restricted to adult stages via Gal80-ts repressor; a dramatic reduction of PH-3-labeled cell number in injured muscle fibers as compared to controls was observed ( Figure 13E). These findings imply that functional Notch expression in the satellite cells is indeed required for injury-induced proliferation of the satellite cells population Taken together with the previously mentioned experiments, our findings are in accordance with a model in which lineal descendants of muscle stem cell-like AMPs are present in adult muscle as muscle fiber apposed satellite cells. Although normally quiescent, following muscle fiber injury these satellite cells become mitotically active, engage in Notch-Delta signaling-dependent proliferative activity and generate lineal descendant cell populations, which can fuse with the injured muscle fiber.

Discussion
The identification and characterization of satellite cells in Drosophila indicates that muscle stem cell lineages act not only in the development of flight muscle as reported previously (Gunage et al., 2014), but also have a role in the mature muscle of the adult. Thus, comparable to the situation in vertebrates, the Drosophila satellite cells are lineal descendants of the muscle-specific stem cell-like AMPs generated during embryogenesis, become intimately associated with adult muscle fibers and  remain quiescent under normal circumstances. Following muscle injury, these Zfh1-expressing cells engage in Notch-Delta signaling-dependent proliferative activity and generate lineal descendant progeny that can fuse with the injured fibers. With improved immune-EM staining protocols on Drosophila DLMs, visualizing Zfh1 expression in these cells at the ultrastructural level will prove valuable. Previous work has shown a role of Zfh1 in other developmental and maintenance processes in Drosophila. Thus, loss of Zfh1 function leads to defects in Drosophila somatic muscle, heart and gonad development (Sellin et al., 2009;Lai et al., 1993). Adult germline stem cells in Drosophila testes, require Zfh1 for stem cell maintenance (Leatherman and Dinardo, 2008). Zfh1 is also known to control neural lineages in a Notch-dependent manner (Lee and Lundell, 2007;Garces and Thor, 2006;Su et al., 1999). In addition, Zfh1 acts near the top of a transcription factor cascade that influences hematopoesis beginning in embryos and continuing in larvae (Frandsen et al., 2008). Evidently, Zfh1 expression is associated with the development, maintenace and differentiation of stem cells in multiple tissues across the lifespan of Drosophila. A single-cell nucleus located within the muscle fiber albeit close to the fiber's surface (white arrow), as delineated by the Vinculin expression border (dotted line) is GFP labeled (implying lineal origin from a Zfh1-positive cell) indicating that it corresponds to a satellite cell lineal progeny. The labeled nucleus has a flattened disc-like shape. Note that the second, apparently adjacent cell nucleus (green arrow) which is RFP labeled is not located within the muscle fiber. N = 8 per group. Scale bars 10 mm. DOI: https://doi.org/10.7554/eLife.30107.014 The mammalian homologs of Zfh1 known as ZEB1 and ZEB2 have been largely studied for their role in cancer progression (Vandewalle et al., 2009). Moreover, the ZEB family protein ZEB2, controls adult hematopoetic differentiation in adults (Li et al., 2017;Postigo and Dean, 2000). Direct functional analyses of ZEB, in mammalian muscles are restricted to C2C12 cells (Postigo and Dean, 1997;Fontemaggi et al., 2001). In all cases, ZEB expression was found to inhibit differentiation of myoblasts. In view of these results, our findings in Drosophila motivate a deeper look into ZEB function in muscle development in mammals. The remarkable similarities in lineage, structure and function of satellite cells in flies and vertebrates imply that the role of these adult-specific muscle stem cells is evolutionarily conserved and, hence, is likely to be manifest in other animals as well. Recently, satellite cells have been identified in a crustacean (Parhyale hawaiensis) during limb regeneration (Konstantinides and Averof, 2014). It will now be interesting to determine if comparable satellite cells are also present in adult Quantification of PH-3 labeled cells in injured muscle of Notch temperature sensitive allele flies at permissive (17˚C) versus restrictive temperature (29˚C). n = 12. Data presented are mean ± standard error. Student's t test: Figure 13 continued on next page musculature of other key protostome and deuterostome invertebrate phyla such as molluscs, annelids and echinoderms.
In vertebrates, satellite cells can undergo symmetric divisions which expand the stem cell pool and asymmetric divisions in which they self-renew and also generate daughter cells that differentiate into the fusion-competent myoblasts required for muscle regeneration and repair (Abmayr and Pavlath, 2012;Relaix and Zammit, 2012). In Drosophila, symmetric and asymmetric division modes are seen during development in the muscle stem cell-like AMPs. Notch signaling controls the initial amplification of AMPs through symmetric divisions, the switch to asymmetric divisions is mediated by Wingless regulated Numb expression in the AMP lineage, and in both cases, the wing imaginal disc acting as a niche provides critical ligands for these signaling events (Gunage et al., 2014). It will important to determine if fly satellite cells, as lineal descendants of AMPs, manifest similar cellular and molecular features in their proliferative response to muscle injury and, thus, recapitulate myogenic developmental mechanisms in the regenerative response of adult muscle. It will also be important to investigate if the mature muscle acts as a niche in this process.
We show that proliferation of Zfh1 expressing cells in DLMs is rapid and extensive in response to stab wounds. We also show that first instances of fusion are seen removed in time from this burst of proliferation. Such a delay between satellite cell proliferation and fusion is puzzling and is also seen in vertebrates. We speculate that satellite cell fusion in this paradigm may serve as a later step in a regenerative process initiated soon after injury, possibly through signaling between injured muscle fibers and satellite cells. Further investigations characterizing the molecular and physical events in DLM repair are likely to clarify our understanding of this process and the system we have developed allows such directions to be explored.
Drosophila has proven to be a powerful genetic model system for unraveling the fundamental mechanisms of muscle development and stem cell biology, and in both respects many of the findings obtained in the fly have been important for the analysis of corresponding mechanisms in vertebrates (Abmayr and Pavlath, 2012;Roy and VijayRaghavan, 1998;Egger et al., 2008;Homem and Knoblich, 2012;Jiang and Reichert, 2013). With the identification of satellite cells in Drosophila, the wealth of classical and molecular genetic tools available in this model system can now be applied to the mechanistic analysis adult-specific stem cell action in myogenic homeostasis and repair. Given the understanding of various fusion molecules involved in early stages of myogenesis, it will also be interesting to investigate a possible conservation of the fusion molecular machinery for regeneration and repair in the adult (Dhanyasi et al., 2015;Haralalka and Abmayr, 2010). Finally, in view of the evidence for age and disease-related decline in satellite cell number and function in humans (e.g. Chang and Rudnicki, 2014), this type of analysis in Drosophila may provide useful information for insight into human muscle pathology.

Materials and methods
Fly strains, genetics and MARCM Fly stocks were obtained from the Bloomington Drosophila Stock Centre (Indiana, USA) and were grown on standard cornmeal medium at 25˚C.
For MARCM experiments mentioned in flies of genotype Hsflp/Hsflp; FRT 42B, Tub Gal80 were crossed to +; FRT 42B, UAS GFP/CyO Act-GFP; Dmef2-Gal4. For MARCM experiments, two heat shocks of 1 hr each separated by 1 hr were given to either late third instar larvae or young adults for clonal induction. Clones were either recovered in the late larval stage for wing disc analysis or in adult stages, which were dissected and processed for flight muscles.

Statistical analysis
Statistical significance of differences in cell counts between control and treatment groups were calculated by the Student's t test on Microsoft Excel. In all quantitations, the mean and standard error for each group are presented.