A genetic screen for regulators of muscle morphogenesis in Drosophila

Abstract

The mechanisms that determine the final topology of skeletal muscles remain largely unknown. We have been developing Drosophila body wall musculature as a model to identify and characterize the pathways that control muscle size, shape, and orientation during embryogenesis. Our working model argues muscle morphogenesis is regulated by (1) extracellular guidance cues that direct muscle cells toward muscle attachment sites, and (2) contact-dependent interactions between muscles and tendon cells. While we have identified several pathways that regulate muscle morphogenesis, our understanding is far from complete. Here, we report the results of a recent EMS-based forward genetic screen that identified a myriad of loci not previously associated with muscle morphogenesis. We recovered new alleles of known muscle morphogenesis genes, including back seat driver, kon-tiki, thisbe, and tumbleweed, arguing our screen had the depth and precision to uncover myogenic genes. We also identified new alleles of spalt-major, barren, and patched that presumably disrupt independent muscle morphogenesis pathways. Equally as important, our screen shows that at least 11 morphogenetic loci remain to be mapped and characterized. Our screen has developed exciting new tools to study muscle morphogenesis, which may provide future insights into the mechanisms that regulate skeletal muscle topology.

Introduction

The mechanisms that regulate skeletal muscle morphogenesis have been remarkably understudied across Metazoa. Body wall muscles in the Drosophila embryo form a stereotyped pattern with 30 distinct muscles arranged in a spectacular compilation of longitudinal, acute, and oblique orientations. Remarkably, the final muscle pattern shows great diversity along the dorsal-ventral axis, but the pattern is invariant from segment to segment along the anterior–posterior axis (Bate 1990). A single embryonic segment must therefore house the essential morphogenetic information required to orchestrate the unique arrangement and functional morphogenesis of 30 individual muscles.

Mesodermal muscle precursors, known as myotubes, extend asymmetrical, amoeboid projections that navigate across the embryonic segment to contact pre-determined muscle attachment sites on ectodermal tendon precursors at the segment border. The mesenchymal myotube then establishes a myotendinous junction, regains symmetry, and acquires its final, functional orientation (Figure 1, A and B). Myotube guidance refers to the combined cellular processes of leading edge navigation and attachment site selection. During myotube guidance, morphogenetic information is transmitted to the muscle precursors through a bipartite system (Yang et al. 2020b). Short-range secreted signals from the ectoderm provide navigational cues, which direct membrane projections toward specific muscle attachment sites. After the projections arrive at the segment border, a second type of morphogenetic information, which is presumed to reside in the tendon precursors, ensures the myotube selects the correct muscle attachment site. Incredibly, this bipartite information system ensures that individual cells from two distinct deterministic cell populations, which are specified in separate germ layers, locate one another with high fidelity during embryogenesis.

Figure 1.

A genetic screen for regulators of muscle morphogenesis. (A) Diagram of Drosophila embryonic body wall muscle development. (Left) After cell fate specification, founder myoblasts break symmetry and become founder myotubes. (Right) Myotube leading edges elongate toward tenocytes (tendon precursors) that house the muscle attachment sites. Fusion competent myoblasts will fuse with founder myotubes during elongation. Each myotube leading edge ultimately selects the correct muscle attachment site. Myotube guidance encompasses both leading edge extension/navigation and attachment site selection. (B) Live imaging of myotube guidance. Two founder LO1 myotubes in neighboring segments expressed cytoplasmic eGFP (green) and nuclear RFP (violet) under the control of slou. Gal4. Arrows point to myotube leading edges during elongation (white) and during attachment site selection (blue). (#:##) hr: min. (C) Screening strategy. EMS-induced embryonic lethal alleles were identified and crossed with slou>tdTomato to visualize the morphology of the longitudinal oblique 1 (LO1), dorsal transverse 1 (DT1), ventral transverse 1 (VT1), ventral acute 2 (VA2), and ventral acute 3 (VA3) muscles in homozygous mutant embryos.

Our previous studies suggested key molecules that provide or interpret morphogenetic information during myotube guidance remain to be discovered, so we screened for novel regulators of muscle morphogenesis. To extend previous muscle screens (Chen et al. 2008; Johnson et al. 2013), we used the shape and orientation of just 5 of the 30 muscles per segment to identify and classify muscle phenotypes at single cell resolution. By focusing on a subset of muscles, we successfully obtained new alleles in 25 genes that were not previously known to regulate embryonic muscle morphogenesis (Figure 1C). We mapped 8 alleles to causative point mutations in 4 distinct loci that encode the zinc-finger transcription factor Spalt major (Salm), the chromatin binding protein Barren (Barr), the Hedgehog (Hh) signaling regulator Patched (Ptc), and the serine/threonine (ser/thr) kinase back seat driver (bsd). The reagents generated from this screen will provide novel inroads toward understanding the molecular mechanisms that direct muscle morphogenesis and myotube guidance.

Materials and methods

Primary screen

To generate screening stocks, flies were isogenized on chromosome II and males were treated with 25 mM EMS, which should induce approximately 3 lethal mutations per second chromosome (Asburner et al. 2005). Mutagenized males were bred as shown:

The breeding scheme utilizes two dominant markers on the second chromosome (wg^Sp1 and Cyo), which were used to follow the mutagenized chromosome (marked by “*”) through the generations. hs-hid is a dominant lethal transgene. Larva were heat shocked for 1 hour a day for 3 consecutive days as indicated to remove unwanted F₂ larva with the wg^Sp1 hs-hid chromosome from the breeding scheme.

For the primary screen, 0–24 hours F₅ embryos were collected on grape plates, dechorionated, transferred to microscope slides, and slou>tdTom expressing muscles were imaged in live embryos. slou>tdTom is visible in 5 distinct muscles per segment (Figure 1C). The presence and morphology of individual slou>tdTom muscles were used to assign mutations to phenotypic classes (see Table 1). At least 5 homozygous embryos were scored per line. The myogenic phenotypes were scored as completely penetrant, in which every embryo showed the phenotype in a majority of hemisegments, or incompletely penetrant, in which only a subset of embryos showed a phenotype in a subset of hemisegments. In total 58 lines with myogenic phenotypes were recovered from the primary screen.

Table 1.

Phenotypic classes of myogenic alleles

Allele	Primary phenotypea	Secondary phenotypea
barrsj266	Morphogenesis-5 muscles (C)	—
barrsj440	Morphogenesis-5 muscles (C)	Thin muscles (C)
bsdsj286	Morphogenesis-5 muscles (C)	—
bsdsj362	Morphogenesis-5 muscles (C)	—
konsj204	Morphogenesis-5 muscles (I)	—
konsj525	Morphogenesis-5 muscles (C)	—
ptcsj287	Morphogenesis-5 muscles (C)	Oversized muscles (C)
ptcsj461	Morphogenesis-5 muscles (C)	Oversized muscles (C)
tumsj316	Morphogenesis-5 muscles (I)	—
sj168	Morphogenesis-5 muscles (C)	—
sj245	Morphogenesis-5 muscles (I)	—
sj246	Morphogenesis-5 muscles (I)	—
sj264	Morphogenesis-5 muscles (C)	—
sj356	Morphogenesis-5 muscles (I)	—
salmsj154	LO1 elongation and attachment (C)	DT1 elongation (C)
salmsj340	LO1 DT1 elongation and attachment (C)	VA2, VA3 elongation and attachment (I)
sj9	LO1 elongation and attachment (I)	—
sj62	LO1 attachment (I)	—
sj81	LO1 DT1 elongation and attachment (I)	—
sj53	VA2, VA3 elongation and branching (C)	—
sj109	VA2/VA3 branching	DT1 missing or short (I)
sj130	VA3 elongation (C)	—
sj194	VA2, VA3 elongation and branching (I)	—
sj455	VA3 branching (I)	—
sj160	VA3 elongation (I)	LO1 VA2 elongation (I)
sj146	LO1, VA2, VA3 elongation and attachment (I)	—
sj185	LO1, VA2, VA3 elongation and attachment (I)	—
sj189	LO1, VA3 elongation and attachment (I)	LO1 elongation and attachment (I)
sj203	VA2 elongation and branching (C) VA3 missing (C)	LO1 elongation and attachment (I)
sj214	LO1, VA1, VA3 elongation and attachment (I)	—
sj237	VA2 elongation and attachment (I)	LO1 elongation and attachment (I) DT1 missing (I)
sj238	LO1 VA3 elongation and attachment (I)	—
konsj11	Missing DT1 LO1 VA3 (I)	—
thssj102	Missing VA1 VA3 (C)	LO1 elongation and attachment (I)
sj103	Missing VA2 (C)
sj348	Missing VA2 (I)	LO1 VA3 elongation and attachment (I)
sj278	Missing VA3 (C)	VA2 elongation and attachment (C)
sj371	Missing VA2 VA3 (C)	LO1 elongation and attachment (I)
sj405	Missing DT1 VA2 VA3 (C)
sj425	Missing VA2 VA3 (C)	LO1 elongation and attachment (I)
sj15	Thin muscles (C)	—
sj302	Thin muscles (C)	—
sj314	Thin muscles (C)	—
sj326	Thin muscles (I)	—
sj359	Thin muscles (C)	—

^a Only DT1, LO1, and VA1-3 muscles were analyzed.

(C) Complete penetrance: all hemisegments in every embryo imaged showed the phenotype.

(I) Incomplete penetrance: a subset of embryos showed a phenotype with a subset of hemisegments affected.

Allele nomenclature

At the conclusion of the primary screen, each line was assigned a unique allele identifier sj###. For mapped alleles, the unique identifier is superscripted with the gene symbol (e.g., salm^sj154). Unmapped alleles are referenced by the unique allele identifier in figures and tables (e.g., sj168).

Complementation analysis

The 58 lines with muscle phenotypes were inter-crossed, and transheterozygote adult viability was used for initial complementation analysis. Because the expected background mutation rate was greater than one, the lines that failed to complement adult viability were further analyzed for transheterozygous embryonic phenotypes to confirm the noncomplementing alleles disrupted myogenesis.

The 58 lines with muscle phenotypes were also crossed to lethal alleles of known regulators of muscle morphogenesis on chromosome II, or to deficiencies that uncovered known regulators if lethal alleles were not publicly available. Screen lines that failed to complement adult viability were further analyzed for transheterozygous embryonic phenotypes. Only those lines that failed to complement embryonic phenotypes were classified as allelic to known regulators of muscle morphogenesis (see Table 2).

Table 2.

Complementation groups with multiple alleles

Allele	Phenotype a
barrsj266	Morphogenesis-5 muscles
barrsj440	Morphogenesis-5 muscles, thin muscles
bsdsj286	Morphogenesis-5 muscles
bsdsj348	VA2 missing
bsdsj362	Morphogenesis-5 muscles
konsj11	DT1 LO1 VA3 missing
konsj204	Morphogenesis-5 muscles
konsj525	Morphogenesis-5 muscles
ptcsj287	Morphogenesis-5 muscles, oversized
ptcsj461	Morphogenesis-5 muscles, oversized
salmsj154	LO1 elongation and attachment
salmsj340	LO1 DT1 elongation and attachment
snssj302	Morphogenesis-5 muscles, thin muscles
snssj359	Morphogenesis-5 muscles, thin muscles
thssj102	VA1 VA3 missing
thssj237	VA2 elongation and attachment
sj109	Morphogenesis-5 muscles
sj245	Morphogenesis-5 muscles
sj189	VA3 elongation and attachment
sj203	VA2 elongation and branching, VA3 missing
sj278	VA3 missing
sj264	Morphogenesis-5 muscles
sj455	VA3 branching
sj326	Thin muscles
sj461	Thin muscles

^aOnly DT1, LO1, and VA1-3 muscles were analyzed.

Complementation groups with 2–3 alleles are shown. Thirty-three single-allele complementation groups were also recovered.

Stocks for complementation analysis included bsd¹ (Yang et al. 2020a), hoip¹ (Johnson et al. 2013), kon²⁹⁸⁶ (Johnson et al. 2013), Df(2R)pyr36 (Kadam et al. 2009), Df(2R)ths238 (Kadam et al. 2009), arr², barr^L305, Df(2R)Egfr5, Df(2L)BSC354 [eya], ptc^tuf-D, Rca1², robo¹, S^IIN, Sin3a^e64, Sli², smo³, Df(2R)BSC269 [sns], tsh⁰⁴³¹⁹, tum^DH15, and zip¹. Stocks were obtained from the Bloomington Stock Center unless otherwise referenced.

Additional Drosophila stocks

Muscle morphology was visualized with P{GMR40D04-GAL4}attP2 (slou.Gal4), P{UAS-tdTom}, and P{UAS-eGFP}. Myoblast fusion was assayed with P{kirre^rP298.lacZ} (Nose et al. 1998). Df(2R)Exel7098 was used for complementation analysis of putative ptc alleles. Cyo, P{Gal4-Twi}, P{2X-UAS.eGFP} was used to genotype embryos for screening, histology, and nucleic acid collection.

Allele sequencing

For RNA deep sequencing (RNAseq), total RNA was collected from 7 to 12 hours embryos per manufacturer’s specification (RNeasy kit, Qiagen), and submitted to Novogene (Sacramento, CA) for deep sequencing. cDNA libraries were prepped and sequenced using 150 bp paired-end reads on the Illumina NovaSeq 6000 system. Two biological replicates were prepared and sequenced for 6 independent lines. Raw reads were quality and adapter trimmed using cutadapt (v2.4) as described (Martin 2011). Trimmed reads were aligned using STAR (v2.5.4b) to the Drosophila reference genome BDGP6 with Ensemble Gene Annotation (v95) (Dobin et al. 2013). Variants for each line were called using aligned reads with the bcftools (v1.9) mpileup function followed by the bcftools call function (Li 2011). Low-quality variants were filtered using bcftools filter. Variants not found in both replicates from a line were discarded. Variants present in any complementary lines were discarded. Remaining variants were mapped to genes and prioritized using VEP (ensembl.org/Tools/VEP) (McLaren et al. 2016). RNAseq data and analysis are available at the Gene Expression Omnibus, accession number GSE164398. Causative variants were confirmed by complementation analysis with known alleles.

For Sanger sequencing, genomic DNA was collected from homozygous mutant 12–24 hours embryos using the Quick Fly Genomic DNA prep method (BDGP). Overlapping PCR amplicons were generated across the coding sequences, and TOPO cloned into PCR2.1 (Thermofisher). M13R and T7 primers were used to sequence inserts from both the 5’ and 3’ ends. Point mutations were validated by sequencing multiple PCR products.

Immunohistochemistry and imaging

_α-_βgal (1:100, Promega, Z3781) was used to visualize rP298.lacZ. Embryo antibody staining was performed as described (Johnson et al. 2013); primary antibodies were detected with HRP-conjugated secondary antibodies and the TSA system (Molecular Probes). Embryos from the primary screen were live imaged on an Axio Observer 5 compound fluorescent microscope; follow-up imaging was used a Zeiss LSM800 confocal microscope. Imaging parameters were maintained between genotypes where possible.

Data availability

All data necessary for confirming the conclusions of this article are represented fully within the article and its tables and figures. All fly stocks are available upon request.

Results and discussion

An invariant number of mesodermal founder myoblasts and ectodermal tendon precursors (tenocytes) are specified in each abdominal segment (A2–A8) of the Drosophila embryo (Frommer et al. 1996; de Joussineau et al. 2012). By the end of myogenesis, the two deterministic cell populations (myoblasts and tendon cells) interact with a precise spatiotemporal stoichiometry, ostensibly through the process of myotube guidance (Figure 1, A and B) (Yang et al. 2020b). To develop new tools to study myotube guidance, we screened 564 EMS-induced lethal alleles on chromosome II for embryonic muscle phenotypes (Figure 1C). Thirty distinct muscles develop in embryonic hemisegments A2–A8, and each muscle has a unique position and morphology (Figure 1C). We visualized muscle morphology in live homozygous mutant embryos with slou>tdTomato, which is expressed in the founder myoblasts that give rise to the Longitudinal Oblique 1 (LO1), Dorsal Transverse 1 (DT1), Ventral Transverse 1 (VT1), Ventral Acute 2 (VA2), and Ventral Acute 3 (VA3) muscles. In total, our screen identified 58 alleles that disrupt muscle morphogenesis.

The orientation of each muscle is defined as acute, oblique, or transverse, and we used muscle orientation to group our alleles into phenotypic classes. For a given allele, we identified the muscles that failed to acquire the correct orientation, and further categorized the orientation phenotypes as defects in myotube elongation (round muscles), in muscle attachment site selection, or in inappropriate branching (Table 1). In addition to orientation phenotypes, the myogenic alleles caused individual muscles to be absent from every hemisegment, produced a thin muscle phenotype throughout the embryo and, in rare occasion, produced muscles that were larger than wild-type controls. The phenotypes we observed were not mutually exclusive, so we grouped the alleles by the primary phenotype, which was the most frequent phenotype we observed (Table 1). Our classification strategy produced 6 phenotypic classes: (1) all 5 slou>tdTomato expressing muscles showed orientation defects (Figure 2A), (2) LO1/DT1 orientation defects (Figure 2B), (3) VA2/VA3 orientation defects (Figure 2C), (4) orientation defects in 3–4 slou>tdTomato expressing muscles (Figure 2D), (5) thin muscles, and (6) missing muscles. We also assigned alleles to a seventh phenotypic class in which the muscle phenotypes were severe and seemingly nonspecific, perhaps due to defects in anterior–posterior patterning, germ band retraction, or dorsal closure.

Figure 2.

Phenotypic classes. Live Stage 16 embryos that expressed slou>tdTomato. In lateral views, LO1 (blue arrows) and VA2 (magenta arrows) muscles are indicated. In ventral views, the ventral midline (dotted line) and the expected positions of VA3 attachment sites (asterisks) are highlighted. (A) The orientation and morphology of all 5 muscles (LO1, DT1, VT1, VA2, and VA3) were affected. (B) LO1 morphogenesis defects. Oblique morphology often transformed to longitudinal morphology (e.g., sj154). (C) VA2/VA3 morphogenesis defects. VA3 muscles frequently failed to attach near the ventral midline (red asterisks). (D) LO1/VA2/VA3 morphogenesis defects. Muscle phenotypes similar to (B, C). Notice VA3 crosses the ventral midline in sj146.

Complementation analysis identified 25 uncharacterized morphogenetic loci

Our phenotypic classes could reflect relative allele strength, so we performed complementation analysis among all 58 alleles, and identified 11 complementation groups with multiple alleles and 33 complementation groups with just a single allele (Table 2). Complementation groups with multiple alleles had representatives from a single phenotypic class (e.g., sj109 and sj245; see Materials and Methods for allele nomenclature) and from multiple phenotypic classes (e.g., sj264 and sj455). These data argue mutations in a single locus can cause overlapping as well as distinct morphogenetic phenotypes. In total, we identified 44 loci that regulate muscle morphogenesis. Alleles in 32 loci appeared to cause specific muscle phenotypes (classes 1–6), and alleles of the remaining 12 loci caused severe, and likely nonspecific, phenotypes (class 7).

To map our myogenic alleles, we first assembled a collection of stocks that carried lethal mutations in known regulators of muscle morphogenesis for complementation analysis. We have characterized 4 genes on the second chromosome that regulate myotube guidance including hoi polloi (hoip)(Johnson et al. 2013; Williams et al. 2015), pyramus (pyr) and thisbe (ths) (Yang et al. 2020b), and back seat driver (bsd) (Yang et al. 2020a). kon tiki (kon) (Schnorrer et al. 2007), slit, robo (Kramer et al. 2001), and tumbleweed (tum) (Guerin and Kramer 2009; Yang et al. 2020a) direct myotube guidance and are also located on the second chromosome. A previous screen for regulators of muscle morphogenesis on the second chromosome identified arrow (arr), barren (barr), Epidermal growth factor receptor (Egfr), eyes absent (eya), patched (ptc), Regulator of cyclin A1 (Rca1), Star (S), Sin3a, smoothened (smo), sticks and stones (sns), teashirt (tsh), and zipper (zip) (Chen et al. 2008). A total of 19 muscle morphogenesis loci have thus been mapped to the second chromosome.

Complementation analysis between our screen alleles and the known regulators of muscle morphogenesis showed that 13 of our mutations were allelic to 6 morphogenetic loci. sj286 and sj362 mapped to bsd, which encodes a ser/thr kinase (Figure 3A). We Sanger sequenced the bsd alleles, and found bsd^sj286 is an A140V missense mutation in the ATP binding pocket and bsd^sj362 is a D590G missense mutation in the kinase domain. Our previous study suggested bsd kinase activity is required for myotube guidance (Yang et al. 2020a), and the identification of kinase domain alleles in the current screen argues bsd is an essential kinase for muscle morphogenesis.

Figure 3.

New alleles in known regulators of myotube guidance. Live Stage 16 embryos that expressed slou>tdTomato. (A) bsd^sj286 and bsd^sj362 failed to complement bsd¹. LO1 muscles (blue arrows) oriented laterally and VA2 muscles (magenta arrows) were rounded or lateral, indicative of myotube guidance defects. A schematic of the bsd protein is shown; bsd^sj286 is an A140V missense mutation in the ATP binding domain and bsd^sj362 is a D590G missense mutation in the kinase domain (KD). (B) ths^sj102 and ths^sj237 failed to complement Df(2R)ths238. In ventral views, the ventral midline (dotted line) and the expected positions of VA3 attachment sites (asterisks) are highlighted. LO1 muscles often oriented laterally. VA3 muscles were missing or failed attach near the ventral midline (red asterisks). (C) kon^sj11, kon^sj204, and kon^sj525 failed to complement kon²⁹⁸⁶. LO1 and VA2 muscles were often rounded or laterally positioned. The molecular lesions associated with ths^sj102, ths^sj237, kon^sj11, kon^sj204, and kon^sj525 were not determined. High and low magnification images are of different embryos.

Ths is a fibroblast growth factor (FGF) ligand and is required for LO1 morphogenesis and VA3 founder cell fate specification (Yang et al. 2020b). Two of our alleles mapped to ths. ths^sj102 and ths^sj237 embryos showed muscle phenotypes consistent with defects in myotube guidance (Figure 3B), and VA3 muscles were missing from some segments in ths^sj102/Df(2R)ths238 and ths^sj237/Df(2R)ths238 transheterozygous embryos (Figure 3B). Overall, ths^sj237 appeared to cause more severe muscle phenotypes than ths^sj102. Ths is comprised of an N-terminus FGF signaling domain and an inhibitory C-terminal domain that is cleaved for full signaling activity (Tulin and Stathopoulos 2010). Although we did not sequence the ths alleles, it is possible that ths^sj102 and ths^sj237 affect different functional domains of the protein or the ability of Ths to be cleaved.

Kon is an orphan transmembrane receptor that directs myotube guidance (Schnorrer et al. 2007), and three of our alleles mapped to kon. kon^sj11 and kon^sj204 caused some of the strongest phenotypes in our screen, whereas kon^sj525 embryos showed more modest phenotypes (Figure 3C). Tum is a GTPase activating protein that directs microtubule reorganization during myotube guidance (Guerin and Kramer 2009), and one of our alleles mapped to tum (tum^sj316, Table 1). We also identified new alleles in barr, ptc, and sns, which were identified in a previous muscle morphogenesis screen (Chen et al. 2008).

Surprisingly, we did not recover alleles in slit or robo, which are known regulators of myotube guidance (Kramer et al. 2001), or in hoip, which encodes an RNA binding protein that is required for myotube guidance (Johnson et al. 2013), which suggests our screen did not completely saturate morphogenetic loci on the second chromosome. However, we recovered 29 alleles in 25 loci that complemented all known regulators of muscle morphogenesis. Thus, 25 muscle morphogenesis genes from our screen remain to be identified.

Salm directs embryonic muscle morphogenesis

Thirty muscles develop in each embryonic hemisegment at a unique position and with a unique morphology. FGF signaling regulates the morphology of just a subset of muscles, presumably due to restricted expression of the FGF ligands Pyr and Ths (Yang et al. 2020b), and we wanted to identify additional pathways that regulate the morphogenesis of a subset of muscles. LO1 muscles in sj154 and sj340 embryos showed a longitudinal orientation instead of an oblique orientation, but the remaining slou>tdTom expressing muscles looked normal (Figure 2B). RNA deep sequencing (RNAseq) of sj154 and sj340 embryo lysates identified two independent nonsense mutations in spalt-major (salm), and salm¹ failed to complement the LO1 muscle phenotype in sj154 and sj340 embryos (Figure 4A). salm encodes a zing finger transcription factor that controls fiber type identity during adult myogenesis (Schönbauer et al. 2011; Bryantsev et al. 2012), but a role for salm in embryonic muscle morphogenesis has not been characterized. Our data suggest Salm could be a transcriptional regulator of myotube guidance.

Figure 4.

Salm, Barr, and Arr direct muscle morphogenesis. (A–C) Live Stage 16 embryos that expressed slou>tdTomato labeled as in Figure 2. (A) salm^sj154 and salm^sj340 failed to complement salm¹. Note the longitudinal morphology of LO1 muscles (blue arrows) in transheterozygous and salm^sj340 homozygous embryos. A schematic of the Salm protein is shown. salm^sj154 is a Q503* nonsense mutation; salm^sj340 is a R355* nonsense mutation. Yellow boxes represent the 7 C2H2-type zinc finger domains (C2H2). (B) barr^sj266 and barr^sj440 failed to complement barr^L305. LO1 (blue arrows) and VA2 (magenta arrows) muscles showed attachment site defects in transheterozygous embryos. barr^sj440 muscles were also thin (high magnification view). A schematic of the Barr protein is shown. barr^sj266 is Q155* nonsense mutation; barr^sj440 is a K631* nonsense mutation. (C) Myoblast fusion assay. Stages 12 and 16 embryos labeled for rP298.nlacZ (green) and Tropomyosin (violet). Robust Tropomyosin expression is not detectable until after Stage 12. Stage 12 barr^sj440 and sns^sj302 embryos showed fewer lacZ+ nuclei after the first round of myoblast fusion than control embryos. Stage 16 barr^sj440 and sns^sj302 muscles had fewer myonuclei, particularly the VA2 muscle. High and low magnification images are of different embryos in (A, B).

Epigenetic regulation of myoblast fusion and myotube guidance

The cellular mechanisms of myoblast fusion and myotube guidance involve cytoskeletal dynamics, so distinguishing fusion phenotypes from guidance phenotypes can be difficult. Our complementation analysis with known regulators of muscle morphogenesis showed sj266 and sj440 failed to complement barr^L305. All 5 slou>tdTom expressing muscles showed morphology defects in barr^sj266/barr^sj440 embryos (Figure 4B), but barr^sj440 muscles were thin while barr^sj266 muscles were normal size (Figures 2A and 4B). Thin muscles are indicative of myoblast fusion defects, so we assayed myoblast fusion in barr^sj440 embryos. sj302 and sj359 embryos also had thin muscles, and these alleles mapped to a known regulator of myoblast fusion sns. Each of the 30 embryonic muscles begins development as a founder cell that fuses with neighboring fusion competent myoblasts. The first round of myoblast fusion causes an increase in the number of nuclei that express founder cell markers including the transcription factors Nautilus and Kruppel, or the transgene rp298.lacZ, and founder cell markers have been used to assay myoblast fusion in Stage 12 embryos (Chen and Olson 2001; Chen et al. 2003; Johnson et al. 2013). The number of rp298.lacZ nuclei in Stage 12 barr^sj440 embryos was less than control embryos and largely equivalent to sns^sj302 embryos (Figure 4C). At the end of embryogenesis, the number of myonuclei per muscle was also reduced in barr^sj440 and sns^sj302 embryos compared to controls (Figure 4C). These data argue that Barr regulates myoblast fusion and myotube guidance, and highlight the exciting possibility that the barr^sj266 and barr^sj440 alleles can be used to genetically separate the role of Barr in myoblast fusion from its role in myotube guidance.

Although barr was identified in a previous muscle screen (Chen et al. 2008), the role of Barr during myogenesis is unknown. Barr is a chromatin binding protein and is part of the condensin complex that directs chromosome condensation, which is a landmark of early mitosis (Herzog et al. 2013). Because founder cells and fusion competent myoblasts exit the cell cycle prior to myoblast fusion and myotube guidance, it is possible that Barr regulates chromatin for a mitotic-independent function during muscle morphogenesis. For example, the Polycomb group proteins repress transcription by binding to specialized cis-regulatory regions, known as polycomb response elements (PREs), and condensing chromatin into a repressed heterochromatic state (Ringrose and Paro, 2007). Barr binds to PREs in the Hox gene complex and is required for Polycomb-induced gene silencing (Lupo et al. 2001). These data suggest that epigenetic mechanisms regulate myoblast fusion and myotube guidance.

Hh signaling dictates muscle size

The alleles sj287 and sj461 produced a rare and unusual phenotype in which the embryonic muscles were larger than wild-type controls, often with the incorrect orientation (Figure 5A). sj287 and sj461 complimented all of the alleles in our collection of known regulators of muscle morphogenesis, so we used RNAseq to map the alleles. Surprisingly, sj287 mapped a W782* nonsense allele in ptc and sj461 mapped to a C313S missense mutation in ptc, yet both alleles complemented ptc^tuf-D in our initial analysis. We extended our complementation analysis and assayed muscle morphogenesis in embryos transheterozygous for Df(2R)Exel7098 which uncovers ptc. ptc^sj287/Df(2R)Exel7098 and ptc^sj461/Df(2R)Exel7098 embryos phenocopied sj287 and sj461 embryos (Figure 5A), confirming the ptc coding region mutations are causative of muscle morphogenesis phenotypes.

Figure 5.

Hh signaling regulates muscle size. (A) Live Stage 16 embryos that expressed slou>tdTomato. ptc^sj287/ptc^sj461 transheterozygous embryos phenocopied ptc^sj287/Df(2R)Exel7098 and ptc^sj461/Df(2R)Exel7098 embryos. Muscles in ptc^sj287 and ptc^sj461 embryos were oversized, (particularly VA2, red arrows), rounded (blue arrows), and showed attachment site defects (white arrows). High and low magnification images are of different embryos. A schematic of the Ptc protein is shown. ptc^sj461 is a C313S missense mutation in the first extracellular domain; ptc^sj287 is a nonsense mutation in the fourth extracellular domain. (B) Screen summary. Fifty-eight alleles in 44 loci were identified that affect muscle morphogenesis. Greater than half of the alleles were mapped or found to be nonspecific (e.g., phenotypes consistent anterior–posterior patterning, germ band retraction, and dorsal closure defects). Unmapped alleles with specific phenotypes are shaded blue. Six alleles with thin muscles had myoblast fusion defects; 7 alleles that caused missing muscles are presumed to affect cell fate specification. The remaining 14 unmapped alleles (in 11 loci) disrupt muscle morphogenesis (dark blue).

The mechanisms that regulate muscle size in neonatal, adolescent, and adult vertebrates have been studied in great detail (Sartori et al. 2021). What is less appreciated is that muscle size must also be regulated in the developing embryo, and those unknown mechanisms are in place to limit muscle overgrowth. To our knowledge, the ptc^sj287 and ptc^sj461 embryos are the first examples of oversized embryonic muscles, and these tools may provide a starting point for understanding how muscle size is limited during development.

In vertebrates, Hh signaling promotes myoblast differentiation in the somite and myoblast migration in the limb (Münsterberg et al. 1995; Hu et al. 2012). In Drosophila, Hh signaling patterns the visceral mesoderm that gives rise to smooth muscle (Bilder and Scott 1998), but the only body wall muscle known to require Hh signaling for proper development is the segment border muscle (SBM) (Jagla et al. 1998). The SBM is not one of the slou>tdTom expressing muscles, but the proximity of the SBM to tendon cells suggests Hh may play a role in attachment site selection at the segment boundary. Consistent with this hypothesis, ptc^sj287 and ptc^sj461 myotubes showed phenotypes consistent with myotube guidance defects (Figure 5A).

Toward a molecular understanding of muscle topography

Our working hypothesis is that muscle topography is established through the actions of short-range secreted signals that provide navigational cues to growing myotubes, and tenocyte expressed cell recognition molecules that direct myotubes to select the correct muscle attachment site. The results from this screen warrant further investigation into Hh signals as navigational cues, and provide new clues into the gene regulatory events, controlled by Salm and Barr, that coordinate cellular guidance and attachment site selection. We successfully used RNAseq to map 6 new alleles without laborious recombination mapping, which brought the total mapped alleles from our screen to 18. However, the mapped alleles represent just a fraction of the mutations with specific phenotypes that we recovered (Figure 5B). Continued genetic mapping should identify unknown loci with important roles in regulating myotube elongation and muscle attachment site selection.