1. Developmental Biology
Download icon

Identification and functional characterization of muscle satellite cells in Drosophila

  1. Dhananjay Chaturvedi
  2. Heinrich Reichert
  3. Rajesh D Gunage  Is a corresponding author
  4. K VijayRaghavan  Is a corresponding author
  1. National Center for Biological Sciences, Tata Institute of Fundamental Research, India
  2. Biozentrum, Switzerland
Research Article
  • Cited 6
  • Views 2,715
  • Annotations
Cite this article as: eLife 2017;6:e30107 doi: 10.7554/eLife.30107

Abstract

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.

https://doi.org/10.7554/eLife.30107.001

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.

Results

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).

Figure 1 with 1 supplement see all
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 labeled myofiber. (A’) Orthogonal view of (A). (B) DLMs stained for Dmef2 Gal4 > UAS mCD8::GFP marking muscle membrane (green) MHC (red) and TOPRO (blue). Unfused nuclei are enveloped in GFP-labeled membrane (white arrowhead), B’ Orthogonal view of (B) at the muscle fiber surface. (C, D) Magnified views of nucleus indicated in (A’) and (B’), respectively. N = 15. Scale bar 10 μm.

https://doi.org/10.7554/eLife.30107.002

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 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.

Figure 2 with 1 supplement see all
Unfused muscle-associated cells have ultrastructural features of satellite cells.

(A, B) Transmission electron micrographs of adult flight muscle. (A) Nuclei inside DLM fibers are large round structures surrounded by nuclear membranes (white dotted lines). A. Distinct sarcomeres in the cytoplasm of the muscle syncytium (marked as S) and mitochondria (marked as M). (B) Mononucleate cell apposed to mature muscle surface. Cell membrane (marked by a green double-headed arrow) seen distinctly apposed to mature muscle membrane (red arrow) and beneath the basement membrane (yellow arrow) of the muscle fiber. Organelles and wedged shaped nucleus (white dotted line) are visible in the cytoplasm of this cell. N = 8 Scale bar 1 μm.

https://doi.org/10.7554/eLife.30107.004

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).

Figure 3 with 1 supplement see all
Unfused cells of the AMP lineage persist in adult muscle.

(A) Single-cell MARCM clones of AMP lineage (mef2-Gal4 driver) induced in the third instar (120AEL) and recovered from a single 15 m heat shock at 37°C, clones from the notum of the wing disc (induced in the late third instar (120AEL). A labeled single cell clone (green) is indicated by a white arrow. Right panel shows the same cell (white arrow) in orthogonal view. The AMP 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 μm. N = 12.

https://doi.org/10.7554/eLife.30107.006

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 adult muscle Zfh1 is a specific marker for unfused satellite cells.

(A) Zfh1 immunolabeling (green) of third instar wing disc notum of wild-type flies. Zfh1 expression can be seen in myoblast nuclei located on the disc epithelium revealed by Phalloidin staining (red), as seen in XZ and YZ orthogonal views. TOPRO stains all nuclei (blue). Scale bar 50 um. N = 10. (B) Zfh1 and β−3-tubulin co-immunolabeling of adult DLMs in wild-type flies. Zfh1 expressing nuclei (green) co-stained with Hoechst (white) are located between DLM fibers labeled with Phalloidin (magenta). Two Zfh1 expressing nuclei are marked with green arrowheads. Nuclei inside DLMs do not express Zfh1 (one example indicated with white arrowhead). The cell boundaries of the DLMs and the Zfh1 expressing cells are delimited by β−3-tubulin; the Zfh1 expressing cell with its cytoskeleton is clearly separate from the adjacent DLM fibers. N = 25. (C) Three orthogonal views of the same preparation as in (B) taken at planes 1, 2 and 3 (dotted lines in B) document the positions of Zfh1 expressing cells outside the muscle fiber (green arrows). Their position contrasts with that of the fused Zfh1-negative DLM nuclei near the surface, one of which is indicated (white arrow). N = 15. (D) Same preparation as in (B) with montage showing the individual confocal channels for Hoechst staining (top left), Zfh1 immunolabeling (top right), β−3-tubulin immunolabeling (bottom left) and Phalloidin staining (bottom right). Scale bar 10 um. (E) Expression of GFP-tagged Zfh1 protein (green) in adult muscle of Zfh1::GFP/TM3 flies co-stained with Phalloidin (red) and TOPRO (blue). Top three panels show individual confocal channels; bottom panel is a superposition of the individual channels. Zfh1 protein expression is limited to unfused cells and is not seen inside muscle fibers. Scale bar 20 um. N = 15. (F) Expression of Zfh1-Gal4 (green) in adult muscle of UAS-RedStinger/+; Zfh1:Gal4/+flies co-labeled with Phalloidin (red). Top two panels show individual confocal channels; bottom panel is a superposition of the individual channels. Zfh1-Gal4 expression is limited to unfused cells and is not seen inside muscle fibers. N = 15. Scale bar 20 um.

https://doi.org/10.7554/eLife.30107.008

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 Zfh1-fused 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.

Pattern of Zfh1 expression in AMP lineal cells at 20 hr and 30 hr APF.

Zfh1 immunolabeling (green) of unfused myoblasts and of myoblasts that have fused with DLM templates, co-labeled with Phalloidin (red) and TOPRO (blue). Zfh1 is expressed in all AMP lineal myoblasts at 20 hr APF but is restricted to a small set of unfused AMP lineal cells at 30 hr APF. (A) At 20 hr APF, Zfh1 expression is seen in all unfused myoblast nuclei and in all nuclei inside the DLM templates (outlined by white dotted line). N = 8. (B) At 30 hr APF, Zfh1 expression is limited to a few nuclei located in between DLM templates and is no longer seen in fused nuclei. Single example shown here. N = 10. Scale bars 10 um.

https://doi.org/10.7554/eLife.30107.009
Pattern of Zfh1 expression in AMP lineal cells at 20 hr and 30 hr APF.

Zfh1 immunolabeling (green) of unfused myoblasts and of myoblasts that have fused with DLM templates, co-labeled with Phalloidin (red) and TOPRO (blue). Zfh1 is expressed in all AMP lineage myoblasts at 20 hr APF but is restricted to a small set of unfused AMP lineage cells at 30 hr APF. (A, B) As in Figure 5 but at lower magnification with images showing whole templates at respective time points. Scale bars 50 μm.

https://doi.org/10.7554/eLife.30107.010

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.

Muscle injury results in the proliferative expansion of the satellite cell population and the generation of fusion competent myoblasts

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)

DLM fibers regenerate following induced physical damage.

Representative images of injured flight muscles (right) and time matched controls (left) at day 2, 5 and 10 after localized stab injury. Adult DLMs stained with Phalloidin (green) and TOPRO (red). At day 2 following injury, breaks in actin filaments, and corresponding disruptions in distribution of nuclei at the site of the injury wound (indicated by white dotted line) are seen. At day 5 following injury, the wound is reduced in size and the actin filament arrangement and myonuclei distribution is more has recovered. At day 10 following injury, regeneration is virtually complete and only small remnants of the wound are apparent. N = 10/group per time point. Scale bar 15 um.

https://doi.org/10.7554/eLife.30107.011

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 Zfh1-labeled nuclei associated with the injured muscle appear to be manifest as spatially adjacent couplets (Figure 8C, D).

Zfh1-positive satellite cells located between DLMs proliferate in response to physical injury.

(A) Uninjured control DLMs. Zfh1-Gal4 driving UAS RedStinger in adult DLMs. A small number of Zfh1-Gal4-positive mononucleate satellite cells (green) are located between DLM fibers. Anti-DsRed (green) co-stained with Phalloidin (magenta). Scale bar 50 μm. (B) Injured DLMs. Zfh1-Gal4 driving UAS RedStinger in adult DLMs injured by stab wound (* denotes injured fiber). At 24 hr after stab wound, numerous Zfh1-Gal4-positive mononucleate satellite cells (green) are seen between DLM fibers at the site of injury but also away from the site of injury. Anti-DsRed (green) co-stained with Phalloidin (magenta). Scale bar 50 μm. (C) Single Zfh1-Gal4-labeled nucleus located between uninjured DLM fibers in area delineated by white square in A. (D) Multiple doublets and a few singlets of Zfh1-Gal4-positive nuclei located near site of injury in area delineated by white square in (B). N = 15 Scale bar 50 μm.

https://doi.org/10.7554/eLife.30107.012

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.

Following muscle injury, satellite cell lineal progeny localize to the surface and the interior of DLM fibers.

Localization of satellite cell lineal progeny examined with Zfh1-Gal4 driven G-trace labeling in uninjured control and in DLM muscles at 24 hr, 48 hr and 1 week after injury. In all cases, G-trace was induced in the adult stage and in the injured animals, 24 hr before infliction of a stab wound. In the uninjured control, only a few cell nuclei located at the DLM surface are labeled as expected for Zfh1-positive satellite cells (top left). At 24 hr and 48 hr after injury more cell nuclei located at the DLM surface are labeled indicating proliferative expansion of the Zfh1-positive satellite cell lineage (top right, bottom left). At 1 week after injury, labeled cell nuclei are located both at the surface and in the interior of DLM fibers implying that some of the Zfh1-positive satellite cell lineal descendants have now fused with the injured DLMs (bottom right). N = 6 per group Scale bar 50 um.

https://doi.org/10.7554/eLife.30107.013

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).

G-trace labeled lineal progeny of Zfh1 expressing satellite cells fuse with damaged DLM fibers after injury.

Localization of satellite cell lineal progeny examined with Zfh1-Gal4 driven G-trace in uninjured controls (A, C) and in experimental animals 1 week after injury (B, D). G-trace was induced in the adult stage and in the injured animals. This induction was 72 hr before infliction of a stab wound. In (A–D), the left four panels are individual confocal channels for Hoechst staining (top left), alpha Spectrin or Vinculin staining (top right), G-trace driven RFP expression (bottom left) and G-trace driven GFP expression (bottom right); the right panel is a superposition of the four channels and viewed from an orthogonal YZ perspective. (A) Control. A single-cell nucleus closely associated with the outside of the DLM surface, as delineated by the alpha Spectrin expression border (dotted line), is both RFP labeled (implying real-time Zfh1 expression) and GFP labeled (implying lineal origin from a Zfh1-positive cell) indicating that it corresponds to a Zfh1-expressing satellite cell. (B) Injured. A single-cell nucleus located within the muscle fiber albeit close to the fiber’s surface, as delineated by the alpha Spectrin 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. (C) Control. A single-cell nucleus closely associated with the outside of the DLM surface as is delineated by the Vinculin expression border (dotted line), is both RFP labeled (implying real-time Zfh1 expression) and GFP labeled (implying lineal origin from Zfh1-positive cells) indicating that it corresponds to a Zfh1-expressing satellite cell. (D) Injured. 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 μm.

https://doi.org/10.7554/eLife.30107.014

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 Zfh1-immunolabeling. 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.

Notch-Gal4 driven G-trace labeling reveals real time Notch expression in muscle satellite cells.

(A) Notch-Gal4 driven G-trace labeling of uninjured adult DLM co-stained by TOPRO. The cell nuclei located on the surface of the DLM fibers are RFP positive (red, anti-RFP labeled) indicating that they correspond to satellite cells that are actively expressing Notch. In contrast, the numerous myonuclei within the muscle fiber are GFP positive (green, anti-GFP labeled) confirming the fact that they are lineal descendants of Notch expressing AMPs. (B) Notch-Gal4 driven G-trace labeling of uninjured adult DLM co-labeled by Zfh1 immunostaining. The cell nuclei located on the surface of the DLM fibers show strong expression of RFP indicating real-time Notch-Gal4 expression and of Zfh1 indicating that they are satellite cells. The top left panel is a superposition of individual channels for Zfh1, RFP, and GFP expression; the remaining panels are the corresponding single channels. (C) Notch-Gal4 driven G-trace labeling of adult DLM 24 hr after injury, co-stained by TOPRO. The cell nuclei located on the surface of the injured DLM fibers have increased in number but are still RFP positive (red, anti-RFP labeled) indicating real-time expression of Notch-Gal4 in these nuclei of the expanded satellite cell lineage. As in (A), myonuclei within the muscle fiber are GFP positive (green, anti-GFP labeled). (D) Quantification of the number of real-time Notch-Gal4 expressing nuclei in control versus injured muscle fibers in these G-trace experiments. Twice as many RFP-positive cells are observed in injured versus control DLMs. n = 10 Data presented are mean ± standard error Student's t test: p-value<0.001 ***. Scale bar 10 μm.

https://doi.org/10.7554/eLife.30107.015

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).

Delta and Neuralized are upregulated in injured muscle fibers.

(A, B) Delta-GFP (anti-GFP, red) expression in DLMs co-labeled with Phalloidin (blue) in control (A) versus injured (B) muscle fibers reveals a marked upregulation of Delta-GFP expression upon injury. (C) Quantification of signal intensity of Delta-GFP in control versus injured DLM fibers; injured muscles show significant upregulation of Delta expression in comparison to uninjured muscle (quantification in arbitrary intensity units). n = 12 Data presented are mean ± standard error. Student's t test: p-value<0.001 ***. (D, E) Neuralized-LacZ expression (green, anti-LacZ immunolabeling) co-labeled by Phalloidin (red) in control versus injured muscle fibers. In comparison to controls (D), injured muscles show elevated Neuralized-lacZ levels in myonuclei, some of which are indicated by white arrows in (E). (F) Quantitation of Neuralized-LacZ expression in control versus injured muscle. For quantification the number of lac-Z-positive nuclei were counted. n = 7. Data presented are mean ± standard error. Student's t test: p-value<0.001 ***. Scale bars 30 μm.

https://doi.org/10.7554/eLife.30107.016

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).

Notch-Delta signaling is required for satellite cell proliferative activity in injured muscle.

(A) Injured control muscle. Mitotic activity assayed by PH-3 expression (green, anti-PH-3 immunolabeling) in DLMs co-labeled for Delta expression (red, anti-Delta immunolabeling) and TOPRO. Numerous satellite cells (three indicated by white arrows) are PH-3-positive indicative of the mitotic activity required for injury-induced expansion of the satellite cell population. (B) Injured muscle with adult-specific Delta downregulation (via Act88F-Gal4, TubGal80ts driving UAS Delta RNAi). Mitotic activity of satellite cells assayed by PH-3 expression as in (A) is absent. (C) Quantification of the number of PH-3 expressing cells in control versus Delta downregulated flies; Delta downregulation is achieved by targeted Delta-RNAi knockdown as well as by targeted dominant negative Delta (DN Delta) expression. n = 9 Data presented are mean ± standard error. Student's t test: p-value<0.001***. (D) 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: p-value<0.001***. (E) Quantification of mitotically active PH-3-labeled satellite cells in control versus Notch downregulated flies. n = 12. Data presented are mean ± standard error. Student's t test: p-value<0.001***.

https://doi.org/10.7554/eLife.30107.017

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.

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 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.

In knockdown and overexpression experiments the following lines were used: +; +; Dmef2-Gal4, Gal80ts, Act 88 F-Gal4, Gal80ts, UAS Notch RNAi (Bloom, 35213), UAS Neur RNAi (Bloom, 26023), UAS DN Delta (Bloom, 26697), UAS Delta RNAi (VDRC, 37288 and GD3720).

Other stocks used- Dl-GFP (Bloom, 59819), Neur-LacZ (Bloom, 12124), Bloom-55121 and 55122.

G-trace analysis-

Notch Gal4 (Bloom 49528) (Dey et al., 2016) was crossed to GTRACE stock(Bloom, 28280). F1-Progenies, N > GTRACE, from this cross were used for the experimental analysis. The following strains were a kind gift from Christian Böekel (Technische Universität Dresden, Germany): +;+;Zfh1-T2A-Gal4, tub-Gal80ts/TM3 and +;+;Zfh1::GFP/TM3. Crosses were set with +;+;Zfh1-T2A-Gal4, tub-Gal80ts/TM3 and GTRACE/Cyo; Gal80ts (Lolitika Mandal, IISCER, Mohali) animals at 18°C. Eclosed animals between 1 to 3 days of age were shifted to 29°C for 72 hr before injury. Injured animals and uninjured controls were incubated at 29°C until dissection.

Immunohistochemistry and confocal microscopy

Flight muscles were dissected from specifically staged flies were dissected and then fixed in 4% paraformaldehyde diluted in phosphate buffered saline (PBS pH-7.5). For each experiment group in this manuscript, replicates ranged from 6 to 25. Immunostaining was performed according to (Weitkunat and Schnorrer, 2014) with few modifications. In brief, samples were then subjected to two washes of 0.3% PTX (PBS + 0.3% Triton-X) and 0.3% PBTX (PBS + 0.3% Triton-X+0.1 %BSA) for 6 hr each. Primary antibody staining was performed for overnight on a shaker and secondary antibodies were added following four washes of 0.3% PTX 2 hr each. Excess of unbound secondary antibodies was removed at the end of 12 hr by two washes of 0.3% PTX 2 hr each following which samples were mounted in Vectashield mounting media. For immunostaining, anti-GFP (Chick, 1:500, Abcam, RRID: AB_300798), anti-Delta (monoclonal mouse, 1:50, Hybridoma bank C594.9B RRID:AB_528194), anti-MHC (Mouse, 1:100, kind gift from Dr. Richard Cripps), TOPRO-3-Iodide (1:1000, Invitrogen), Hoechst 33342 (1:500, ThermoFisher) anti-Neuralized (1:50, Rabbit) (Lai et al., 2001), Anti-Zfh-1 (1:1000, rabbit, gift from Prof. Ruth Lehmann lab), Anti-Dmef2 (rabbit 1:3000, gift from Bruce Patterson RRID:AB_2568604) Phalloidin (Alexa-488/647/568 conjugate, 1:500, ThermoFisher), anti-phosphohistone-H3 (Rabbit, 1:100, Millipore), anti-Spectrin (3A9 mouse monoclonal, DSHB 1:5, RRID:AB_528473), anti-Vinculin (1:200, mouse Abcam RRID:AB_444215) antibodies were used. Secondary antibodies (1:500) from Invitrogen conjugated with Alexa fluor-488, 568 and 647 were used in immunostaining procedures.

Confocal and electron microscopy

For confocal experiments, an Olympus FV 1000 confocal point scanning microscope and Zeiss LSM 780 were used for image acquisition. Images were processed using ImageJ software (Rasband WS, ImageJ U S. National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/, 1997–2012). Quantification of number of actively dividing cells in PH-3 labeling experiments was performed as described in Gunage et al., 2014).

For electron microscopic analysis, the muscles were processed according to (Garcia-Murillas et al., 2006). In brief, flight muscles were dissected in ice-cold fixative (2.5% glutaraldehyde in 0.1 M PIPES buffer at pH 7.4). After 10 hr of fixation at 4°C, samples were washed with 0.1M PIPES, post-fixed in 1% OsO4 (30 min), and stained in 2% uranyl acetate (1 hr). Samples were dehydrated in an ethanol series (50%, 70%, 100%) and embedded in epoxy. Ultrathin sections (50 nm) were cut and viewed on a Tecnai G2 Spirit Bio-TWIN electron microscope. Results presented in this manuscript were replicated eight times.

Muscle injury

To induce regeneration response in the flight muscle, flies were injured through the cold stab method. Flies were CO2 anesthetized and a single stab injury was performed manually with dissection pin or tungsten needle dipped in liquid nitrogen (Fine Scientific Tools, Item no-26002–10, Minutien Pins-Stainless Steel/0.1 mm Diameter). Care was taken so that the tungsten needle tip did not cross the hemithorax so that the damage was restricted to a minimum. Although completely effective, the method has limited precision. Control flies were age matched adult flies but with no injury to muscles. Injured animals recovered on corn meal Drosophila food. The flies were then processed for immunostaining of flight muscles as mentioned in the immunohistochemistry procedure. The detailed protocol of muscle injury can be found at Bio-protocol (Chakraborty et al., 2018).

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.

References

  1. 1
  2. 2
    Cells with persistent twist expression are the embryonic precursors of adult muscles in Drosophila
    1. M Bate
    2. E Rushton
    3. DA Currie
    (1991)
    Development 113:79–89.
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Insights into neural stem cell biology from flies
    1. B Egger
    2. JM Chell
    3. AH Brand
    (2008)
    Philosophical Transactions of the Royal Society B: Biological Sciences 363:39–56.
    https://doi.org/10.1098/rstb.2006.2011
  10. 10
  11. 11
    Development of the indirect flight muscles of Drosophila
    1. J Fernandes
    2. M Bate
    3. K Vijayraghavan
    (1991)
    Development 113:67–77.
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
    The Drosophila homeobox genes zfh-1 and even-skipped are required for cardiac-specific differentiation of a numb-dependent lineage decision
    1. MT Su
    2. M Fujioka
    3. T Goto
    4. R Bodmer
    (1999)
    Development 126:3241–3251.
  43. 43
  44. 44
  45. 45

Decision letter

  1. Fiona M Watt
    Reviewing Editor; King's College London, United Kingdom

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Newly identified satellite cells respond to damage through Notch-Delta signaling to fuse with adult Drosophila muscles" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

All reviewers agree that the identification of satellite cells in Drosophila is potentially exciting. However, they feel that the work is preliminary and not well substantiated by the quality of the data presented. Please see reviewers' comments for specific suggestions.

Reviewer #1:

In this paper, the authors extended their previous characterization of muscle-specific stem cells in fly development by presenting evidence that the descendants of these stem cells in adult flies similarly as vertebrate muscle satellite cells. This conclusion would be highly significant if substantiated. While the authors have presented some evidence to show that these cells undergo mitosis upon injury, whether their progeny contribute to adult muscle was hard to evaluate, at least based on the GFP data alone. Additional experiments to clearly demarcate these GFP-positive cells as part of the muscle fibres would strengthen the conclusion.

Reviewer #2:

In their paper the authors attempt to extend their previous findings published in eLife that a set of cells known as AMPs, which give rise to muscle during early development, are also capable of acting as satellite cells in the adult.

As the paper currently stands the data presented are not convincing. No evidence presented for regeneration, let alone regeneration of new tissue that leads to functional tissue. Without more rigorous data, in particular better clonal data, I would not recommend publication. In fact, at this point the data, at best, suggest that the AMP remnants stick around and can divide but do not give rise to any new cells, as has been shown to be the case by Fox and Spradling with the adult hindgut imaginal ring. Below are experiments that would go a long way to proving the author's points. I would strongly encourage the authors to address most of the comments with new, better, and more rigorous experiments rather than arguing them away. Ultimately, either there are or there aren't satellite cells that give rise to new muscle following injury AND are necessary for regeneration. And a publication claiming there are satellite stem cells based on poor quality data, which is later shown to be incorrect would reflect poorly on the authors and eLife.

Major issues:

1) How many satellite cells do the authors think there are? What is the ratio between satellite and differentiated cells?

2) Figure 1: Without better markers I cannot tell where the GFP labeled cells are.

3) The EdU data, including methodology, should be included.

4) How close is panel C to the big hole shown in A?

5) Why is the GFP so weak/strange looking in E in comparison to Figure 1D-E? And the merge in H? There is a lot of GFP that is of equal or more intense than said clones and not associated with anything.

6) Figure 3D: Over what area and what percentage of satellite cells? How was the region chosen? Do all satellite cells become mitotically active following injury? 50%? Also, I'm surprised the difference here is only 2 stars.

7) Why are most of the GFP descendents PH3 positive 24 hours after injury? I would expect only half to be positive since half of the progeny would not be satellite cells, but daughters of satellite cells. Do the authors think the daughters can divide? Or perhaps do the authors think the response by satellite cells takes 24 hours? Does PH3 number go down with time? If so what are the kinetics? Knowing this would make testable predictions about when one should see labeled cells outside and inside the muscle.

8) The quality of the PH3 staining is poor and therefore not convincing. It essentially looks like background. The presence of mitotic spindles would help the author's case. Along these lines what does the equivalent of 3E-3H PH3 staining look like in uninjured tissue? They could help with background green levels.

9) Can the authors rule out dividing cells as being hemocytes?

10) Figure 3I-3N: Why do the authors induce clones 6hr prior to injury whereas in earlier experiments they induced clones in the larval period?

11) Figure 3I-3N: Nothing in this figure leads me to believe that there are now GFP+ cells in the muscle. The position of the nucleus appearing to "intercalate" could be based solely on optical sectioning. The satellite cells should both self-renew and give rise to a myoblast that differentiate and fuse, so there should always be two labeled cells, right? Where is the second cell? Better markers, close–up views, and 3-D reconstruction would bolster the author's assertion. Furthermore, I would like to get a quantification of how many new muscle nuclei there are.

12) Ultimately, the authors need to use a reporter that would be capable of labeling either the muscle membrane or cytoplasm so we can see the new cell and the extent of repair. I understand the author's argue that the CD8::GFP gets diluted. But with enough damage it should be possible for enough new nuclei to be made to produce an identifiable signal. Or to use more sensitive membrane reporter, for example see "ultrafast tissue staining with chemical tags" published in PNAS by Kohl et al., (2014).

13) The presence of a muscle fibre that is GFP positive cannot rule out fusion of a "satellite cell" directly with a fibre versus the progeny of a "satellite cell" fusing with a muscle fibre. The use of twin spot clonal systems (those in which each progeny is labeled with a different color following mitotic recombination) could help resolve this issue.

14) Notch staining in 4A-C looks strongly nuclear. In fact, Notch staining in flies tends to be membrane and in endocytic vesicles and never seen in the nucleus. In addition, there is no control using N RNAi to show staining goes away. This is true of Delta and neutralized with respect to lack of controls for immune-reactivity.

15) Figure 4N: I assume the area examined is the same for PH3 experiments. If so why are 5% of cells NRE+. In other words, there are approx. 175 dividing cells in Figure 4D but 10-12 NRE positive cells in Figure 4N.

16) Figure 4O-P: Where is Delta? Is it nuclear again, like N staining? Should be either membrane or vesicular. These low magnification images are not informative at all. Also, it appears that Dl staining is present in the Delta-RNAi image?

17) The neuralized staining looks like background and could benefit from a control (neutralized knockdown followed by staining). Also – are the levels adjusted the same? R inset looks like it has less background than S inset.

18) Does neur knockdown in muscle (using Act88F) also block PH3? What happens with Notch knockdown using Act88F?

Reviewer #3:

The article by Gunage et al. describes a MARCM clonal analysis in Drosophila aimed at identifying adult muscle satellite cells. That is, non-differentiated myoblasts associated with the mature adult muscles that might contribute to adult muscle repair upon activation. The significance of this finding, if proven, is extremely high: there has been no prior evidence that satellite cells exist in Drosophila, yet the existence of such cells would enable the use of a powerful system to investigate the specification and biology of these cells. Such a finding would be a major advance.

In their paper, Gunage et al. identify clones of cells expressing dMEF2-Gal4 that appear to satisfy some of the criteria of satellite cells. They are located at or near the adult flight muscles; there is some evidence that they proliferate in response to injury; and aspects of the biology of these cells is affected by manipulation of the Notch pathway, that regulates the specification of Drosophila adult myoblasts, and plays a major role in satellite cell activation in vertebrates.

Despite these promising findings, I feel that the depth of analysis presented here does not go far enough to confirm the existence of these cells in Drosophila. Given the significance of the proposed conclusions, I think that the analyses should be far more rigorous than shown here. Specific concerns are as follows:

1) The stains and documentation in numerous figures do not effectively support the authors claims about the cells. For example, in Figure 1 panels C and D, it is not apparent that the GFP-labeled cells are located at the periphery of the muscle, because there are no markers used to label the periphery. Also, the red dotted lines are in slightly different locations comparing panel D with E, further raising doubt as to the precise location of the boundaries of the muscle cells. I suggest co-staining with a membrane marker to more clearly localize the cells. In addition, Figure 1 panels F and G are so weakly stained that it is not possible to see what is intended.

2) Figure 2 shows electron micrographs of flight muscle cells and a cell associated with the flight muscles. There is no evidence that this cell either corresponds to a satellite cell or shares any of the markers of satellite cells. I think this figure should be removed.

3) Since dMEF2-Gal4 is expressed in cells other than adult muscle precursor cells associated with the wing discs, the authors cannot claim that the GFP-positive cells observed in the adult arise from the wing disc myoblasts without more detailed lineage tracing carried out during the pupal stage.

4) There are no markers that are specific for the cells identified as satellite cells. An analysis of a number of candidate markers would improve confidence that the identified cells are myoblast-like, such as Twist, dMEF2, and others, and would improve the ability of the authors to follow the cells as they proliferate. Otherwise, one might argue that the infiltrating cells are non-muscle cells, perhaps responding to injury or sepsis, and unrelated to a repair mechanism.

5) Figure 3. It is difficult to reconcile some of the data presented in this figure. Firstly, can the authors describe (or ideally show) the proximity of the stained areas to the muscle injury? This applies to all subsequent figures where the muscle is injured. It would be good to confirm that the band of PH3-positive cells is close to the site of damage, and does not occur at a location away from the area of damage. Otherwise, the PH3 stain might instead represent some kind of cellular response. In panels C and F there appear to be a relatively large number of PH3-positive cells, whereas in panels I-N the number of cells is quite sparse. While I understand that the GFP marking system in I-N only labels a subset of cells, it would be nice to know how many such satellite cells exist, and whether the number of marked cells in C and F is consistent with the numbers of precursors. This also relates to comment 4 above, where there is a paucity of markers for these important cells. Also, the PH3 and GFP stains in panels E and F are too faint to be considered reliable. Finally, in panels J-N, the authors show GFP cells associated with an injury site, but the location of the injury is not shown, and the relative locations of the GFP-positive cells relative to the cell membrane cannot be determined, thus it is not clear if the cells are really infiltrating.

6) Figure 4. The Notch staining in panel B is not convincing. Why is the DNA stain in panels H-J fainter than the same stain in panels K-M?

7) For the Delta knockdown experiment and for the Neuralized expression levels, there must be quantitation of the degree of knockdown and the degree of Neur over-expression. Comparison of stained sections is not satisfactory, and instead the authors should carry out quantitative RT-PCR or western blotting to confirm their stains. Also, why is the Phallodin intensity greater in panel S than R? Accurate quantitation as suggested here would protect the authors from concerns that differential efficiency of staining or imaging is the cause of the observed changes in intensity.

8) There is no evidence that the adult muscles repair, that the identified cells have any role in the repair, or that the repair depends upon Notch pathway members.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your work entitled "Identification and Functional Characterization of Muscle Satellite Cells in Drosophila" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

While both reviewers appreciate the improvement over the last submission, they remain unconvinced that the study definitively shows the origin, existence and activity of these cells similar to the mammalian satellite cells. They are especially concerned about the quality of the lineage and functional data implicating these cells in muscle repair.

Reviewer #1:

In their resubmission, the authors argue that Drosophila flight muscles contain satellite cells that proliferate in response to damage and give rise to progeny that fuse with muscle to lead to functional repair. While I agree that there has been an improvement in the quality of data since their last submission, in particular the presence of superficial cells, I am not convinced that these cells are a stem cell population that participates in muscle repair. The lineage and functional data presented is poor quality, not only from the level of cellular resolution but with respect to the assays used. I think part of the problem is the syncytial nature of the tissue, which makes it hard to analyze. Definitive proof will probably require live imaging in intact flies, something that I would thing feasible with 2 photon microscopy.

Major issues:

1) Are the muscle nuclei polyploid? Are the green arrow nuclei diploid? If so it would help to DAPI quantify.

2) In Figure 1A there is only one superficial nuclei. In 1C there are three. The scale bar suggests the picture is the same size but 1C looks smaller?

3) "(In control experiments, MARCM clones were also triggered in pupal stages and recovered in the adult to confirm that the labeled cells were not infiltrating cells that might derive from unknown proliferating cells located external to the wing disc; see Figure 3—figure supplement 1A-C)”. I don't understand this. Is the reasoning that during pupation there are no cell divisions in this lineage and therefore you would expect not to recover clones? Why is it parentheses? Also, throughout the paper there are examples of text placed in parentheses. Is this a convention the authors use to indicate an afterthought? It's unnecessary.

4) "Note that, while differentiated myoblasts are also targeted in this MARCM experiment, no GFP labeled cells are visible within the adult muscle fibres since their membrane tethered GFP becomes diffuse due to incorporation into the extensive muscle fibre membrane following cell fusion (see schematic in Figure 3D)." Wouldn't the argument used by the authors later, that gal80 is present, explain why there is no membrane GFP?

5) "Care was taken to restrict damage such that only 1 or 2 muscle fibres were affected and that fibres were not severed by the injury (Figure 4A, B)." Out of how many fibres? How much area would this represent?

6) "DLMs damaged in this way can regenerate and in morphological respects appear normal after approximately 3 weeks (Figure 4—figure supplement 1)."

7) How do they appear after 1 day? 2 days? 3 days? etc. It's confusing why one would look after 3 weeks, especially when data presented later implies that muscle function returns after 2 days! By looking at many early time points one might get an idea of how the proposed repair occurs. In fact, maybe the area of damage is repaired by fusion of remaining tissue or hypertrophy of adjacent nuclei. There is no way to rule these alternatives out based on the resolution (spatial and temporally) provided. Also, it's unclear where the original injury might have been. There is no presentation of muscles that have no/poor repair (e.g. DmefGal4 X Notch RNAi) to compare with.

8) "For both PH-3 and EdU labeling, evidence for increased mitotic activity was largely restricted to the damaged muscle fibres and rarely observed in undamaged muscle fibres”. Could you show a damaged fibre next an undamaged fibre with PH3 staining.

9) In contrast, within the damaged muscle fibre, evidence for increased mitotic activity was seen in nuclei along the entire extent of the fibre length. It surprises me that mitosis would be seen along the entire fibre. How many nuclei would that be? Is that the case with vertebrate satellite muscle cells? How many of new progeny fuse with the fibre?

10) "In order to label only cells that were generated by mitotic activity in the adult, clones were induced in the adult 6h prior to physical injury and recovered 24h later." According to the authors divisions are rare, so how were clones recovered at all? The better way is follows right after in the parentheses.

11) "This implies that some of the daughter cells generated by satellite cells during injury-induced proliferative activity remained at the muscle cell surface while others appear to have entered muscle fibre's interior and may have fused with the injured muscle cell." This was not convincing. They may have not entered the interior but rather intermixed in the space which has been disrupted by injury. Or that injury leads to non-specific fusion caused by injury.

12) The FUCCI system does not provide information about lineage. And the quantity and quality of the data was not better than PH3. Along these lines, could the authors show examples of metaphase, anaphase and telophase? This would be much more convincing that the surprisingly numerous PH3 positive cells. Also, the long chain of "proliferative" cells shown using the FUCCI system was never seen or described in earlier experiments and makes me wonder if it's an artifact.

13) Nts affects the entire animal. DmefGal4 X Notch RNAi affects all muscle. MARCM clones of Notch mutants would be more useful since one could see they are PH3 negative in a sea of PH3 positive WT cells.

14) Was DAPT fed to the flies, for how long?

15) The Notch responsive element is not a GFP construct of E(spl). It's, I believe, Su(H) binding sites fused to Gbe binding sites and based on the original reporter made by Sarah Bray's lab.

16) "Targeted Notch downregulation led to significant perturbation in the flight initiation response, indicating the importance of satellite cell lineage proliferation and fusion to restore muscle function after injury." Actually, this is merely a strict correlation. That is, PH3 falls and so does flight initiation. But whether they are directly connected cannot be concluded. Notch may be required in Dmef2 positive cells for other things. I am also concerned and surprised that repair occurs in 2 days. And along these lines why in Figure 4—figure supplement 1 did the authors wait until 3 weeks for evidence of repair? Furthermore, what about Notch RNAi after 4 days, 5 days? (They only look at Notch RNAi after 2 days). Given the huge variation even among each genotype it's hard to see how one can conclude statistical significance. Can it recover? I would not expect it to given how well PH3 levels fall. Therefore, if it recovered it would argue something else is going on. And why only do 3 trials? Given the ease of the experiment you could do 300 trials.

17) It's unclear why the authors chose a red box next to the area of damage as opposed to a box with the area of damage in the middle. And why a 100uM and not more? At what distance does PH3 fall? Or is it an entire fibre becomes activated?

Reviewer #2:

The manuscript by Gunage et al. is a significantly improved version of a manuscript submitted earlier this year, that attempts to describe and characterize for the first time satellite cells in Drosophila. This appears a particularly difficult task, given a paucity of suitable markers for these cells and no established muscle injury model in this organism. Nevertheless, the significance of this work, if proven, would be high.

The prior version of the manuscript suffered from a lack of suitable counterstains and marker analysis for the satellite cells, and this has been improved. It is clear that a separate population of cells exists on the flight muscle surface, and that upon injury there is an increase in the number of proliferating cells in and around the flight muscles. However, I still remain a little unconvinced that this revised work still definitively shows the origin, existence and activity of these cells.

Firstly, there is still no good data supporting that the satellite cells arise from the wing imaginal discs, versus any other discs or any other part of the body that expresses Dmef2-gal4. The authors indicate they studied pupal time points in a supplementary figure, however that data is poorly described and does not appear to represent a time course.

Secondly, the authors report that they were not able to identify any markers of the satellite cells from a candidate list. It is unclear if they have tested Dmef2 itself. The satellite cells should be positive for this protein because the Dmef2-gal4 driver is active in the cells.

Thirdly, the authors are still struggling to definitively demonstrate that the satellite cells and their daughter cells correspond to the nuclei that are thought to infiltrate the injured muscle. In Figure 4D, for example, it looks as though essentially every nucleus becomes PH3 positive. Do the endogenous muscle nuclei (i.e. those that were present in the muscle prior to injury) also accumulate PH3? A clarification of not only the number of PH3-poistive nuclei in each sample, but also the total number of nuclei, would help resolve this. This point is not well supported by the data in Figure 4 H-K, where the PH3 stain is so weak it could be argued to be either present in all nuclei or absent from all nuclei.

In addition, the notion that some of the activated satellite cells remain on the surface of the muscle is attractive, but is not well supported by the existing data. Here, there is no proof that the surface nuclei are not fused to the muscle (and no support from the data in panel 4S-U), nor that the activated cells do not fuse to the muscle in the following day or so. Later time points following injury and MARCM induction might resolve this issue.

Finally, it is still not clear what is the level of repair taking place. The rescue of flight defects is modest and not necessarily due to muscle defects (it could be some form of shock or inflammatory response that causes a reduction in flight after injury); the authors do not provide any data suggesting that new myofibrillar proteins are being made; and there is no indication that there is an overall increase in the number of muscle nuclei following repair. Thus, the roles of these cells are not defined.

[Editors’ note: what follows is the decision letter after the authors resubmitted for further consideration.]

Thank you for submitting your article "Identification and Functional Characterization of Muscle Satellite Cells in Drosophila" for consideration by eLife. I do apologise for the very long time it took to review your manuscript. The senior editors who saw the original version have stepped down as editors, and none of the three in-depth reviewers was available to comment. This, compounded with your August submission, when many editors were on vacation, explains – but of course does not excuse – the delay. Your article has been reviewed by a single peer reviewer, and the evaluation has been overseen by Fiona Watt as the Senior Editor. The reviewer opted to remain anonymous.

Our normal practice is to have a discussion between multiple reviewers to achieve a consensus position on each manuscript. However, in this case there was only one in-depth review and so we have provided it in full below. Both the reviewer and the editor are supportive of publication, provided that you address each of the points raised.

Reviewer #1:

The manuscript by Gunage et al. provides an analysis of adult flight muscles in Drosophila and presents evidence consistent with the existence of cells within these muscles that possess properties that are highly similar to those that have been described for mammalian satellite cells. These properties include localization at the periphery of muscle fibres, derivation from developmentally active muscle precursors, proliferation in response to muscle damage, contribution of nuclei into regenerated muscle fibres, and dependence on Notch signaling for myogenic activation. The authors also identify a new marker of these cells Zfh-1 and establish a new muscle injury/repair model in the fly. While much of the data presented are descriptive in nature, the clonal analyses and lineage tracing are compelling in support of the authors' hypothesis. The manuscript thus represents an important first step, and although clearly there is much more work to be done to fully understand the nature and activities of these cells this work is likely to impact the field by advancing the relevance of the fly system for studies of muscle regenerative biology and providing a new model system for genetic and mechanistic studies. For these reasons, I am supportive of publication in eLife, assuming the authors can address the few points below:

1) Subsection “Two different types of cells are present in adult flight muscle”: "This observation is confirmed by co-staining these adult muscle fibres for expression of either Act88F, an indirect flight muscle specific isoform of actin, or Tropomoysin." Is this data contained in the manuscript? If so, it should be properly called out and if not, it should be added.

2) The authors should provide in the Discussion section more information about their newly defined marker Zfh-1, including its other known functions in the fly (it appears to be involved in muscle development and in self-renewal of germline stem cells) as well as information about whether its homolog is expressed in mammalian satellite cells (for which multiple publicly available gene expression data sets are available).

3) Statistical tests used should be identified in the figure legends and methods should describe statistical approaches and randomization (if used).

4) If technically feasible, immunoEM to demonstrate that the "satellite cells" identified by ultrastructural analysis are the same cells as those identified by confocal microscopy and Zfh-1 expression would greatly enhance the authors' conclusions.

https://doi.org/10.7554/eLife.30107.020

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

In this paper, the authors extended their previous characterization of muscle-specific stem cells in fly development by presenting evidence that the descendants of these stem cells in adult flies similarly as vertebrate muscle satellite cells. This conclusion would be highly significant if substantiated. While the authors have presented some evidence to show that these cells undergo mitosis upon injury, whether their progeny contribute to adult muscle was hard to evaluate, at least based on the GFP data alone. Additional experiments to clearly demarcate these GFP-positive cells as part of the muscle fibres would strengthen the conclusion.

We have addressed above concerns by several additional experiments using methods such as EdU labeling, Fly-FUCCI and 3D reconstruction. Further, the role of these cells and of the Notch pathway in repair after injury is now strengthened behavioral assays.

Reviewer #2:

In their paper the authors attempt to extend their previous findings published in eLife that a set of cells known as AMPs, which give rise to muscle during early development, are also capable of acting as satellite cells in the adult.

As the paper currently stands the data presented are not convincing. No evidence presented for regeneration, let alone regeneration of new tissue that leads to functional tissue. Without more rigorous data, in particular better clonal data, I would not recommend publication. In fact, at this point the data, at best, suggest that the AMP remnants stick around and can divide but do not give rise to any new cells, as has been shown to be the case by Fox and Spradling with the adult hindgut imaginal ring. Below are experiments that would go a long way to proving the author's points. I would strongly encourage the authors to address most of the comments with new, better, and more rigorous experiments rather than arguing them away. Ultimately, either there are or there aren't satellite cells that give rise to new muscle following injury AND are necessary for regeneration. And a publication claiming there are satellite stem cells based on poor quality data, which is later shown to be incorrect would reflect poorly on the authors and eLife.

Major issues:

1) How many satellite cells do the authors think there are? What is the ratio between satellite and differentiated cells?

We have counted the numbers of satellite cells and differentiated cells in the DLMs. There are 20 ± 4 satellite cells for 700 ± 50 differentiated cells for one (of 6) DLMs. This has been incorporated in the text.

2) Figure 1: Without better markers I cannot tell where the GFP labeled cells are.

In addition to the experiments described in this figure with superimposed stacks, we have carried out new labeling experiments using different markers. We present this labeling of cells in a new figure panel, which also has a scan in the z axis that allows the reader a 3D view of the labeled cells in relation to the muscle fibres. In addition (Figure 1—figure supplement 1) we show that these cells do not express flight muscle-specific Actin88F or Tropomyosin.

3) The EdU data, including methodology, should be included.

We present the EdU data in a new figure panel (Figure 4F, G). We also include the methodology in a new paragraph of the Materials and methods section.

4) How close is panel C to the big hole shown in A?

We mention that the panel C is at about 100 micrometers from the site of injury in the corresponding Results section (Figure 4A).

5) Why is the GFP so weak/strange looking in E in comparison to Figure 1D-E? And the merge in H? There is a lot of GFP that is of equal or more intense than said clones and not associated with anything.

We have repeated the corresponding experiments and now present new panels corresponding to Figure 4H-K which show unfused GFP labeled cells more clearly with much less background (which is likely to be due to fusion of labeled cells). The difference in GFP staining in these panels to that shown in Figure 3E-G is less in the case illustrated, although differences are to be expected since the clone induction time is different (late third instar versus adult) for wild type (Figure 3E-G) compared to damaged muscle (Figure 4H-K).

6) Figure 3D: Over what area and what percentage of satellite cells? How was the region chosen? Do all satellite cells become mitotically active following injury? 50%? Also, I'm surprised the difference here is only 2 stars.

The area being considered is now shown in the panel in low magnification of the injured muscle in Figure 4A shown as a red dotted rectangle (a corresponding area was taken from uninjured control). We have redone the experiment and analysis and corresponding results are now part of Figure 4.

7) Why are most of the GFP descendents PH3 positive 24 hours after injury? I would expect only half to be positive since half of the progeny would not be satellite cells, but daughters of satellite cells. Do the authors think the daughters can divide? Or perhaps do the authors think the response by satellite cells takes 24 hours? Does PH3 number go down with time? If so what are the kinetics? Knowing this would make testable predictions about when one should see labeled cells outside and inside the muscle.

New experiments added, now Figure 4H-K (magnified inset in Figure 4K) clearly shows example of lineage trace and PH3 labeling demonstrating mitotically active as well as inactive satellite cells. Our results based on PH3, Edu labeling (Short pulse and chase) and Fly-FUCCI show less than 50% of unfused cells are mitotically active.

For these experiments clonal labeling was performed during larval stages and clones were recovered post injury 12h. Mitotic activity of activated stem cells decreases with time after injury and our analysis of clonal analysis has clearly revealed the fusion is active around 24h as shown in Figure 4L-Q and Figure 4S-U.

8) The quality of the PH3 staining is poor and therefore not convincing. It essentially looks like background. The presence of mitotic spindles would help the author's case. Along these lines what does the equivalent of 3E-3H PH3 staining look like in uninjured tissue? They could help with background green levels.

We have repeated the corresponding out experiments and provided new figures (Figure 4H-K and also Figure 4—figure supplement 2A, B) in which the staining quality is substantially improved. Also, we have used Fly-FUCCI to demonstrate the same and the results are now included in the Figure 4—figure supplement 2C, D.

9) Can the authors rule out dividing cells as being hemocytes?

To rule this out, we have carried out labeling experiments with the e33c-Gal4 line that marks hemocytes (Fossett et al., 2003, Matova and Anderson, 2006). The dividing cells were not labeled indicating that they are not hemocytes. This experiment is now incorporated into the Results section in Figure 3—figure supplement 1E, F.

10) Figure 3I-3N: Why do the authors induce clones 6hr prior to injury whereas in earlier experiments they induced clones in the larval period?

Clones were induced in the adult and not in the larva in order to label only those lineages that had proliferated specifically in the adult in response to injury and not those lineages, which had proliferated earlier (Figure 3I-3N of earlier MS is now become Figure 4L-Q). The 6 h time period of clone induction prior to injury was chosen to ensure that adequate FLP-ase protein is present at the time of injury. We now mention this in the Results section to clarify this.

11) Figure 3I-3N: Nothing in this figure leads me to believe that there are now GFP+ cells in the muscle. The position of the nucleus appearing to "intercalate" could be based solely on optical sectioning. The satellite cells should both self-renew and give rise to a myoblast that differentiate and fuse, so there should always be two labeled cells, right? Where is the second cell? Better markers, close up views, and 3-D reconstruction would bolster the author's assertion. Furthermore, I would like to get a quantification of how many new muscle nuclei there are.

We have carried out a 3-D reconstruction and present this together with close–up view and a visualization of all optical sections in a movie in new figure panels (Video 5 and Video 6, 7); this clearly shows that there are GFP positive cells in the muscle. We have also carried out a quantification of how many new muscle nuclei there are in induced clones and show this in a new figure panel (Figure 5O). While we understand the logic of the referee in assuming that there should always be two labeled cells, we cannot assert that this must the case for two reasons. First, we have not studied the mode of cellular proliferation of the satellite cell lineages following injury and thus cannot say if it is asymmetric or symmetric and if it includes transit- amplifying cells. Second, given the amount of time between clone induction and recovery, the satellite cells could divide more than once thus producing more than one round progeny (even if there are no transit-amplifying cells). Our results with Fly-FUCCI experiments showed a clear lineage and hints towards possible mechanisms referees are suggesting (Figure 4—figure supplement 2C, D). As we mention in the Discussion section, a detailed characterization of the proliferation properties of satellite cells in response to injury is an important topic for further investigation.

12) Ultimately, the authors need to use a reporter that would be capable of labeling either the muscle membrane or cytoplasm so we can see the new cell and the extent of repair. I understand the author's argue that the CD8::GFP gets diluted. But with enough damage it should be possible for enough new nuclei to be made to produce an identifiable signal. Or to use more sensitive membrane reporter, for example see "ultrafast tissue staining with chemical tags" published in PNAS by Kohl et al., (2014).

We have repeated these experiments and the results are included in the Figure 4S-U.

13) The presence of a muscle fibre that is GFP positive cannot rule out fusion of a "satellite cell" directly with a fibre versus the progeny of a "satellite cell" fusing with a muscle fibre. The use of twin spot clonal systems (those in which each progeny is labeled with a different color following mitotic recombination) could help resolve this issue.

We cannot claim that only the satellite cell progeny but not the satellite cell can fuse directly with a muscle fibre. It is indeed possible that both occur. Our experiments show that following the injury induced proliferation, some of the cells in the satellite cell lineage fuse with the muscle fibres and other do not (Figure 4F-R). We have now mentioned this explicitly in the results (Figure 4—figure supplement 2C, D) and Discussion section.

14) Notch staining in 4A-C looks strongly nuclear. In fact, Notch staining in flies tends to be membrane and in endocytic vesicles and never seen in the nucleus. In addition, there is no control using N RNAi to show staining goes away. This is true of Delta and neutralized with respect to lack of controls for immune-reactivity.

The referee is correct in that staining for the N intracellular domain is definitely not nuclear. The impression from the figure mentioned, that it is nuclear is due to the fact that the nucleus of the satellite cell is relatively large compared to the rest of the cytoplasm (see our EM figure). To show that N intracellular domain is not in the nucleus, we have carried out high resolution imaging of the satellite cells co-labeled for N intracellular, cell membrane and nucleus. We present this in a new figure panel (Figure 5A-D). We have also carried out the control using N RNAi; this is presented in a new figure panel. Similarly, we have carried out the controls for Delta (Figure 6 A, B) and Neuralised (Figure 6 F, G); this is presented in new figure panels.

15) Figure 4N: I assume the area examined is the same for PH3 experiments. If so why are 5% of cells NRE+. In other words, there are approx. 175 dividing cells in Figure 4D but 10-12 NRE positive cells in Figure 4N.

The area examined is the same, and the numbers are as presented. Thus, under control conditions we count 175 dividing cells and only around 10 NRE positive cells. We do not claim to investigate the mechanisms responsible for this difference in numbers, rather we simply show that in both cases (division and NRE expression) there is a marked increase in the number of implicated cells following injury.

16) Figure 4O-P: Where is Delta? Is it nuclear again, like N staining? Should be either membrane or vesicular. These low magnification images are not informative at all. Also, it appears that Dl staining is present in the Delta-RNAi image?

We have carried out new experiments and present new magnifications that clearly show that Delta is not nuclear and that it is not present in the RNAi image. These data are shown in new figure panels (Figure 6A, B).

17) The neuralized staining looks like background and could benefit from a control (neutralized knockdown followed by staining). Also – are the levels adjusted the same? R inset looks like it has less background than S inset.

We have carried out new experiments including controls that now clearly document the neuralized staining. These data are shown in new figure panels (Figure 6F, G).

18) Does neur knockdown in muscle (using Act88F) also block PH3? What happens with Notch knockdown using Act88F?

We have carried out new experiments that show that both Neur and N knockdown downregulate proliferation. These data are shown in new figure panels (Figure 6 IK).

Reviewer #3:

[…] Specific concerns are as follows:

1) The stains and documentation in numerous figures do not effectively support the authors claims about the cells. For example, in Figure 1 panels C and D, it is not apparent that the GFP-labeled cells are located at the periphery of the muscle, because there are no markers used to label the periphery. Also, the red dotted lines are in slightly different locations comparing panel D with E, further raising doubt as to the precise location of the boundaries of the muscle cells. I suggest co-staining with a membrane marker to more clearly localize the cells. In addition, Figure 1 panels F and G are so weakly stained that it is not possible to see what is intended.

We have repeated the corresponding experiments and now present new panels which show the unfused GFP labeled cells more clearly with much less background (Figure 1C-D, Figure 3—figure supplement 1). In addition, we have carried out new labeling experiments using different markers including a membrane marker. We present this labeling of cells in new figure panels, which also has a scan in the z axis that allows the reader a 3D view of the labeled cells in relation to the muscle fibres (3D video showing two types of cells see Video1).

2) Figure 2 shows electron micrographs of flight muscle cells and a cell associated with the flight muscles. There is no evidence that this cell either corresponds to a satellite cell or shares any of the markers of satellite cells. I think this figure should be removed.

We include this micrograph to show unequivocally that there are small cells in the adult muscle that are not fused with the muscle fibres but are very closely associated with muscle and even share the muscle cell ECM (previously shown in Figure 1D-F, 3D reconstruction Video 4, 5 and also demonstrated in Figure 3—figure supplement 1). Cells of this morphological type are novel and have not been previously described. This clearly confirms the notion that there are two different types of cells in adult muscle, one type has fused its nucleus with the muscle cell and the other novel type, while closely associated with the muscle, has not. We clarify this in the revised Results section.

3) Since dMEF2-Gal4 is expressed in cells other than adult muscle precursor cells associated with the wing discs, the authors cannot claim that the GFP-positive cells observed in the adult arise from the wing disc myoblasts without more detailed lineage tracing carried out during the pupal stage.

We have traced the dMEF2-Gal4 expressing cells associated with the larval wing discs through pupal stages and into the adult to confirm that the labeled cells in the adult do arise from the wing disc muscle precursors. Representative staining results from pupal stages are shown in a new figure (Figure 3—figure supplement A-C).

4) There are no markers that are specific for the cells identified as satellite cells. An analysis of a number of candidate markers would improve confidence that the identified cells are myoblast-like, such as Twist, dMEF2, and others, and would improve the ability of the authors to follow the cells as they proliferate. Otherwise, one might argue that the infiltrating cells are non-muscle cells, perhaps responding to injury or sepsis, and unrelated to a repair mechanism.

We have screened for other candidate markers, but have not found any that are specific for these satellite cells. (This could be a reason why these cells have not been identified previously in Drosophila). To rule out that these cells might be infiltrating hemocytes, we have carried out labeling experiments with the Gal4e33 line which marks hemocytes. The dividing cells were not labeled indicating that they are not hemocytes. This experiment is now incorporated into the Results section Figure 3—figure supplement 1E, F.

5) Figure 3. It is difficult to reconcile some of the data presented in this figure. Firstly, can the authors describe (or ideally show) the proximity of the stained areas to the muscle injury? This applies to all subsequent figures where the muscle is injured. It would be good to confirm that the band of PH3-positive cells is close to the site of damage, and does not occur at a location away from the area of damage. Otherwise, the PH3 stain might instead represent some kind of cellular response.

We now show the site of the labeled cells (Figure 4A shown in red dotted rectangle) relative to the site of injury in a new panel, which presents low magnification of the injured muscle (Figure 4A white dotted circle).

In panels C and F there appear to be a relatively large number of PH3-positive cells, whereas in panels I-N the number of cells is quite sparse. While I understand that the GFP marking system in I-N only labels a subset of cells, it would be nice to know how many such satellite cells exist, and whether the number of marked cells in C and F is consistent with the numbers of precursors. This also relates to comment 4 above, where there is a paucity of markers for these important cells.

We have determined the number of unfused cells (mentioned in main text result) and have also determined that 50% percent of the unfused cells are Ph3 positive, indicating that less than 50% of the satellite cells have become active following injury (Mitotic activity is seen primarily in injured area). This is now mentioned in the corresponding Results section (Figure 4C-E and H-K). We have also carried out a quantification of how many new muscle nuclei there are in induced clones and show this in a new figure panel (Figure 5O).

Also, the PH3 and GFP stains in panels E and F are too faint to be considered reliable. Finally, in panels J-N, the authors show GFP cells associated with an injury site, but the location of the injury is not shown, and the relative locations of the GFP-positive cells relative to the cell membrane cannot be determined, thus it is not clear if the cells are really infiltrating.

We have repeated the corresponding out experiments and provided new figures in which the staining quality is substantially improved. We provide new figure panels in which the location of the injury relative to the stained cells is shown (see above). We have carried out a 3-D reconstruction and present this together with close–up view and a visualization of all optical sections in a movie in new figure panels; this clearly shows that there are GFP positive cells in the muscle. Moreover, in new experiments (Figure 4S-U) we show fused membrane GFP cells (white dotted circle) in the vicinity of unfused GFP positive cells (red arrow). These findings are presented in new figure panels incorporated into the Results section.

6) Figure 4. The Notch staining in panel B is not convincing. Why is the DNA stain in panels H-J fainter than the same stain in panels K-M?

We have carried out new labeling experiments and present new magnifications that more clearly show the Notch (and Delta) staining. These data are shown in Figure 5A-D and Figure 6A, B.

7) For the Delta knockdown experiment and for the Neuralized expression levels, there must be quantitation of the degree of knockdown and the degree of Neur over-expression. Comparison of stained sections is not satisfactory, and instead the authors should carry out quantitative RT-PCR or western blotting to confirm their stains. Also, why is the Phallodin intensity greater in panel S than R? Accurate quantitation as suggested here would protect the authors from concerns that differential efficiency of staining or imaging is the cause of the observed changes in intensity.

We have repeated all the experiments and the new panels are included in the result section Figure 6B, H and Figure 6—figure supplement 1 and explained in the Materials and methods section.

8) There is no evidence that the adult muscles repair, that the identified cells have any role in the repair, or that the repair depends upon Notch pathway members.

We present new data showing that adult muscles do repair in anatomical terms, and that this repair depends on the N pathway in a new figure. To address the role of Notch pathway in repair, a functional assay is now included in the Figure 5 panel W, X and Figure 4—figure supplement1.

[Editors' note: the author responses to the second round of review follow.]

We thank both reviewers for their very useful comments, all of which we have addressed in the extensively revised manuscript. We request that this be considered a new submission as in the revised manuscript virtually all of the figures have been replaced by new experiments and figures and text which, we feel have substantially improved the manuscript.

Central among these are new experiments that identify and utilize Zfh1 as a highly specific marker for satellite cells in adult muscle. With this marker we also perform new lineage tracing experiments to confirm the origin of satellite cells and demonstrate that their lineal progeny fuse with muscle fibers after injury.

Reviewer #1:

In their resubmission, the authors argue that Drosophila flight muscles contain satellite cells that proliferate in response to damage and give rise to progeny that fuse with muscle to lead to functional repair. While I agree that there has been an improvement in the quality of data since their last submission, in particular the presence of superficial cells, I am not convinced that these cells are a stem cell population that participates in muscle repair. The lineage and functional data presented is poor quality, not only from the level of cellular resolution but with respect to the assays used. I think part of the problem is the syncytial nature of the tissue, which makes it hard to analyze. Definitive proof will probably require live imaging in intact flies, something that I would thing feasible with 2 photon microscopy.

We provide new lineage tracing data based on the highly specific satellite cell

marker Zfh1 as well as G-trace techniques that shows both the lineal origin of the

satellite cells from AMP lineages and demonstrates that in response to injury the

labeled satellite cells proliferate and generate lineal progeny that fuse with

damaged muscle fibers. These new findings are presented in several new figures

and are described in new text.

Major issues:

1) Are the muscle nuclei polyploid? Are the green arrow nuclei diploid? If so it would help to DAPI quantify.

Staining with DAPI is presented in multiple panels though out manuscript and

they are not polyploid. The cells indicated using green arrows are indeed diploid

in nature.

2) In Figure 1A there is only one superficial nuclei. In 1C there are three. The scale bar suggests the picture is the same size but 1C looks smaller?

We present new data in a new Figure 1 as well as in new subsequent figures,

which clearly show multiple superficial nuclei.

3) "(In control experiments, MARCM clones were also triggered in pupal stages and recovered in the adult to confirm that the labeled cells were not infiltrating cells that might derive from unknown proliferating cells located external to the wing disc; see Figure 3—figure supplement 1A-C)”. I don't understand this. Is the reasoning that during pupation there are no cell divisions in this lineage and therefore you would expect not to recover clones? Why is it parentheses? Also, throughout the paper there are examples of text placed in parentheses. Is this a convention the authors use to indicate an afterthought? It's unnecessary.

We have carried out new experiments and have removed this part of the text. With the new specific maker for satellite cells permitting the demarcation of the lineal origin of stem cells through pupal stages we now not only establish the lineage but also rule out that satellite cells might derive from other proliferating cells.

4) "Note that, while differentiated myoblasts are also targeted in this MARCM experiment, no GFP labeled cells are visible within the adult muscle fibres since their membrane tethered GFP becomes diffuse due to incorporation into the extensive muscle fibre membrane following cell fusion (see schematic in Figure 3D)." Wouldn't the argument used by the authors later, that gal80 is present, explain why there is no membrane GFP?

We have removed this figure and replaced it with new data and figures in the revised version. Moreover, we agree with reviewer on the fact that gal80 present in the muscle prevents labeling of fused cell nuclei in MARCM experiments. Indeed, this is further evidence for the unfused nature of labeled satellite cells.

5) "Care was taken to restrict damage such that only 1 or 2 muscle fibres were affected and that fibres were not severed by the injury (Figure 4A, B)." Out of how many fibres? How much area would this represent?

This is now much clearly presented multiple panels though out manuscript and it is always 1-2 out 6 DLM muscles. This is much clearer in low zoom images and this represents about one tenth of muscle area.

6) "DLMs damaged in this way can regenerate and in morphological respects appear normal after approximately 3 weeks (Figure 4—figure supplement 1)."

We now present new data on the nature of the injury and the timeline of regeneration at 2, 5 and 10 days in a new Results section and note that at 10 days the regeneration is largely complete.

7) How do they appear after 1 day? 2days? 3 days? etc. It's confusing why one would look after 3 weeks, especially when data presented later implies that muscle function returns after 2 days! By looking at many early time points one might get an idea of how the proposed repair occurs. In fact, maybe the area of damage is repaired by fusion of remaining tissue or hypertrophy of adjacent nuclei. There is no way to rule these alternatives out based on the resolution (spatial and temporally) provided. Also, it's unclear where the original injury might have been. There is no presentation of muscles that have no/poor repair (e.g. DmefGal4 X Notch RNAi) to compare with.

See response above to point 6).

8) "For both PH-3 and EdU labeling, evidence for increased mitotic activity was largely restricted to the damaged muscle fibres and rarely observed in undamaged muscle fibres”. Could you show a damaged fibre next an undamaged fibre with PH3 staining.

We have eliminated the corresponding figure in the revised version and now characterize satellite cell proliferation due to injury using the satellite cell specific Zhf1 label. Moreover, we also use G-trace to label the satellite cells after injury. Both new experiments show both damaged and undamaged fibres which support the above mentioned spatial distribution of proliferation.

9) In contrast, within the damaged muscle fibre, evidence for increased mitotic activity was seen in nuclei along the entire extent of the fibre length. It surprises me that mitosis would be seen along the entire fibre. How many nuclei would that be? Is that the case with vertebrate satellite muscle cells? How many of new progeny fuse with the fibre?

See response to point 8).

10) "In order to label only cells that were generated by mitotic activity in the adult, clones were induced in the adult 6h prior to physical injury and recovered 24h later." According to the authors divisions are rare, so how were clones recovered at all? The better way is follows right after in the parentheses.

The corresponding experiment has been removed in the revised version and replaced by Zfh1 based G-trace lineage tracing methods performed in the injured adult.

11) "This implies that some of the daughter cells generated by satellite cells during injury-induced proliferative activity remained at the muscle cell surface while others appear to have entered muscle fibre's interior and may have fused with the injured muscle cell." This was not convincing. They may have not entered the interior but rather intermixed in the space which has been disrupted by injury. Or that injury leads to non-specific fusion caused by injury.

The corresponding experiment has been removed in the revised version and replaced by Zfh1 based G-trace lineage tracing methods performed in the injured adult.

12) The FUCCI system does not provide information about lineage. And the quantity and quality of the data was not better than PH3. Along these lines, could the authors show examples of metaphase, anaphase and telophase? This would be much more convincing that the surprisingly numerous PH3 positive cells. Also, the long chain of "proliferative " cells shown using the FUCCI system was never seen or described in earlier experiments and makes me wonder if it's an artifact.

The corresponding experiment has been removed in the revised version and replaced by Zfh1 based G-trace lineage tracing methods performed in the injured adult.

13) Nts affects the entire animal. DmefGal4 X Notch RNAi affects all muscle. MARCM clones of Notch mutants would be more useful since one could see they are PH3 negative in a sea of PH3 positive WT cells.

Due to the highly restricted nature of the injury, localized effects can be clearly discerned and differentiated from possible global effects by the use of the new highly specific markers.

14) Was DAPT fed to the flies, for how long?

DAPT was fed for 3days after eclosion and injury was performed at the end of 48h and muscles were processed at the end of 72h.

15) The Notch responsive element is not a GFP construct of E(spl). It's, I believe, Su(H) binding sites fused to Gbe binding sites and based on the original reporter made by Sarah Bray's lab.

The corresponding experiment has been removed in the revised version.

16) "Targeted Notch downregulation led to significant perturbation in the flight initiation response, indicating the importance of satellite cell lineage proliferation and fusion to restore muscle function after injury." Actually, this is merely a strict correlation. That is, PH3 falls and so does flight initiation. But whether they are directly connected cannot be concluded. Notch may be required in Dmef2 positive cells for other things. I am also concerned and surprised that repair occurs in 2 days. And along these lines why in Figure 4—figure supplement 1 did the authors wait until 3 weeks for evidence of repair? Furthermore, what about Notch RNAi after 4 days, 5 days? (They only look at Notch RNAi after 2 days). Given the huge variation even among each genotype it's hard to see how one can conclude statistical significance. Can it recover? I would not expect it to given how well PH3 levels fall. Therefore, if it recovered it would argue something else is going on. And why only do 3 trials? Given the ease of the experiment you could do 300 trials.

The corresponding experiment has been removed in the revised version.

17) It's unclear why the authors chose a red box next to the area of damage as opposed to a box with the area of damage in the middle. And why a 100uM and not more? At what distance does PH3 fall? Or is it an entire fibre becomes activated?

The corresponding figure has been removed in the revised version.

Reviewer #2:

The manuscript by Gunage et al. is a significantly improved version of a manuscript submitted earlier this year, that attempts to describe and characterize for the first time satellite cells in Drosophila. This appears a particularly difficult task, given a paucity of suitable markers for these cells and no established muscle injury model in this organism. Nevertheless, the significance of this work, if proven, would be high.

The prior version of the manuscript suffered from a lack of suitable counterstains and marker analysis for the satellite cells, and this has been improved. It is clear that a separate population of cells exists on the flight muscle surface, and that upon injury there is an increase in the number of proliferating cells in and around the flight muscles. However, I still remain a little unconvinced that this revised work still definitively shows the origin, existence and activity of these cells.

Firstly, there is still no good data supporting that the satellite cells arise from the wing imaginal discs, versus any other discs or any other part of the body that expresses Dmef2-gal4. The authors indicate they studied pupal time points in a supplementary figure, however that data is poorly described and does not appear to represent a time course.

We thank reviewer for these valuable and point specific comment. We now present extensive spatiotemporal data on satellite cell origin from larval through pupal stages which clearly show that satellite cells are lineal descendants of AMPs in the wing disc. This work is based on the newly identified Zfh1 marker.

Secondly, the authors report that they were not able to identify any markers of the satellite cells from a candidate list. It is unclear if they have tested Dmef2 itself. The satellite cells should be positive for this protein because the Dmef2-gal4 driver is active in the cells.

We thank the reviewer for this important comment. As a consequence, we have screened for such a marker and have discovered that Zfh-1 is a highly specific marker for satellite cells in the adult musculature. Data based on this maker are included throughout the extensively revised manuscript.

Thirdly, the authors are still struggling to definitively demonstrate that the satellite cells and their daughter cells correspond to the nuclei that are thought to infiltrate the injured muscle. In Figure 4D, for example, it looks as though essentially every nucleus becomes PH3 positive. Do the endogenous muscle nuclei (i.e. those that were present in the muscle prior to injury) also accumulate PH3? A clarification of not only the number of PH3-poistive nuclei in each sample, but also the total number of nuclei, would help resolve this. This point is not well supported by the data in Figure 4H-K, where the PH3 stain is so weak it could be argued to be either present in all nuclei or absent from all nuclei.

In addition, the notion that some of the activated satellite cells remain on the surface of the muscle is attractive, but is not well supported by the existing data. Here, there is no proof that the surface nuclei are not fused to the muscle (and no support from the data in panel 4S-U), nor that the activated cells do not fuse to the muscle in the following day or so. Later time points following injury and MARCM induction might resolve this issue.

We have replaced these experiments by new Zfh1 driven G-trace experiments which clearly demonstrate that satellite cell lineal descendants fuse with muscle following injury.

Finally, it is still not clear what is the level of repair taking place. The rescue of flight defects is modest and not necessarily due to muscle defects (it could be some form of shock or inflammatory response that causes a reduction in flight after injury); the authors do not provide any data suggesting that new myofibrillar proteins are being made; and there is no indication that there is an overall increase in the number of muscle nuclei following repair. Thus, the roles of these cells are not defined.

We have eliminated the flight data in the revised manuscript. Moreover, we present a spatiotemporal characterization of the injury repair process. We feel that an in–depth characterization of the molecular mechanisms of the repair process goes beyond the scope of this manuscript which is focused on identification and characterization of satellite cells in normal and injured muscles.

[Editors' note: the author responses to third round of review follow.]

Reviewer #1

[…] 1) Subsection “Two different types of cells are present in adult flight muscle”: "This observation is confirmed by co-staining these adult muscle fibres for expression of either Act88F, an indirect flight muscle specific isoform of actin, or Tropomoysin." Is this data contained in the manuscript? If so, it should be properly called out and if not, it should be added.

We are grateful to the reviewer for highlighting this gap. As you have suggested, we have added a supplement to Figure 1 (Figure 1—figure supplement 1) where we show unfused muscle associated mononucleate cells located at the surface of, and between, adult myofibres within whom A) DLM specific Actin (Act88F) and B) Tropomyosin localize to myofibrils. These stainings reiterate the localization of satellite cells illustrated in Figure 1.

2) The authors should provide in the Discussion section more information about their newly defined marker Zfh-1, including its other known functions in the fly (it appears to be involved in muscle development and in self-renewal of germline stem cells) as well as information about whether its homolog is expressed in mammalian satellite cells (for which multiple publicly available gene expression data sets are available).

We thank the reviewer for this important suggestion. Clearly it is essential to describe Zfh1 function related knowledge, from the perspective of this study in this manuscript. In the discussion, we have summarized in vivoinvestigations into the role of Zfh1 and its mammalian homolog ZEB in development. The summary can be found in the Discussion section of this manuscript. This is a valuable addition to this manuscript.

3) Statistical tests used should be identified in the figure legends and methods should describe statistical approaches and randomization (if used).

We appreciate this suggestion from our reviewer. Accordingly, descriptions of statistical analyses and parameters have been added to the figure legends to Figure 11D; Figure 12C and F; Figure 13C and E. A section on the statistics we employed has been added to the Materials and methods section as well.

4) If technically feasible, immunoEM to demonstrate that the "satellite cells" identified by ultrastructural analysis are the same cells as those identified by confocal microscopy and Zfh-1 expression would greatly enhance the authors' conclusions.

We agree with the reviewer on the value of demonstrating Zfh1 expression in satellite cell nuclei, delienating apposition to muscles, retaining their own cell membranes, with the resolution of electron microscopy. A clean immunoEM image of Zfh1 in satellite cells would robustly consolidate our observations made through immunohistochemistry. In the revision, we now readily acknowledge this in the Discussion section: “With improved immune-EM staining protocols on Drosophila DLMs, visualizing Zfh1 expression in these cells will prove valuable.” However, at this time, the best available protocols for immunoEM in DLMs (Reedy et al., 2000.) show heavy background. While we could try to optimize the protocols for our system and setup, all conceivable avenues of doing so require the generation of new reagents e.g. antibodies which would take months, and at a minimum a few more months for optimization of the protocol. Variability in quality of images is inherent to the combination of antibodies and EM for immunoEM in DLMs. The lack of guaranteed, unquestionably clean immunoEM images after the prospective effort and time, makes this experiment technically unfeasible at this time. This experiment will remain among our priorities for further studies, of which many are ongoing.

https://doi.org/10.7554/eLife.30107.021

Article and author information

Author details

  1. Dhananjay Chaturvedi

    Department of Developmental Biology and Genetics, National Center for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Identified Zfh1 as a satellite cell marker and characterized their ability to fuse post injury
    Contributed equally with
    Rajesh D Gunage
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3957-1236
  2. Heinrich Reichert

    The Centre for Molecular Life Sciences, Biozentrum, Basel, Switzerland
    Contribution
    Conceptualization, Formal analysis, Supervision, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Rajesh D Gunage

    Department of Developmental Biology and Genetics, National Center for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Present address
    Stem Cell Program and Division of Haematology/Oncology, Children’s Hospital, Howard Hughes Medical Institute, Boston, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Identified the muscle lineage of satellite cells and demonstrated their relation to notch signaling
    Contributed equally with
    Dhananjay Chaturvedi
    For correspondence
    rajeshgunage@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5694-4658
  4. K VijayRaghavan

    Department of Developmental Biology and Genetics, National Center for Biological Sciences, Tata Institute of Fundamental Research, Bangalore, India
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Investigation, Writing—original draft, Writing—review and editing
    For correspondence
    vijay@ncbs.res.in
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4705-5629

Funding

Department of Science and Technology, Ministry of Science and Technology (JC Bose Fellowship)

  • K VijayRaghavan

Science and Engineering Research Board

  • Dhananjay Chaturvedi

Department of Science and Technology, Ministry of Science and Technology

  • Rajesh D Gunage

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was possible due to the generous support of the National Centre for Biological Sciences, Tata Institute of Fundamental Research and the J C Bose Fellowship of the Government of India. We thank the Central Imaging and Flow Facilities (NCBS) for the microscopes used in this study. The NCBS Fly facility provided indispensible help with managing fly stocks. The authors would also like to thank Rajan Thakur for all the help provided for electron microscopy related experiments. We are deeply thankful to Ruth Lehmann (NYU, New York, USA) for the Zfh1 antiserum. Special thanks to Christian Böekel (CRT, Dresden, Germany) for Zfh1::Gal4 and Zfh1::GFP lines.

Reviewing Editor

  1. Fiona M Watt, King's College London, United Kingdom

Publication history

  1. Received: July 4, 2017
  2. Accepted: October 24, 2017
  3. Accepted Manuscript published: October 26, 2017 (version 1)
  4. Version of Record published: November 10, 2017 (version 2)
  5. Version of Record updated: June 21, 2018 (version 3)

Copyright

© 2017, Chaturvedi et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,715
    Page views
  • 489
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Physics of Living Systems
    Silas Boye Nissen et al.
    Research Article Updated
    1. Developmental Biology
    Debadrita Bhattacharya et al.
    Research Article