A large reverse-genetic screen identifies numerous regulators of testis nascent myotube collective cell migration and collective organ sculpting

Maik C Bischoff; Jenevieve E Norton; Erika A Munguia; Sarah E Clark; Noah J Gurley; Rebecca Korankye; Emmanuel Addai Gyabaah; Taino Encarnacion; Christopher J Serody; Corbin D Jones; Mark Peifer

doi:10.1091/mbc.E24-10-0456

. 2025 Jan 28;36(2):ar21. doi: 10.1091/mbc.E24-10-0456

A large reverse-genetic screen identifies numerous regulators of testis nascent myotube collective cell migration and collective organ sculpting

Maik C Bischoff ^a,^*, Jenevieve E Norton ^a, Erika A Munguia ^a, Sarah E Clark ^a, Noah J Gurley ^a, Rebecca Korankye ^a, Emmanuel Addai Gyabaah ^a, Taino Encarnacion ^a, Christopher J Serody ^a, Corbin D Jones ^a, Mark Peifer ^a,^b,^*

Editor: Richard Fehon^c

PMCID: PMC11809313 PMID: 39745864

Abstract

Collective cell migration is critical for morphogenesis, homeostasis, and wound healing. Migrating mesenchymal cells form tissues that shape the body's organs. We developed a powerful model, exploring how Drosophila nascent myotubes migrate onto the testis during pupal development, forming the muscles ensheathing it and creating its characteristic spiral shape. To define genes regulating this, we used RNA sequencing (RNA-seq) to identify genes expressed in myotubes during migration. Using this dataset, we curated a list of 131 ligands, receptors, and cytoskeletal regulators, including all Rho/Ras/Rap1 regulators, as candidates. We then expressed 279 short hairpin RNAs (shRNAs) targeting these genes and examined adult testes. We identified 29 genes with diverse roles in morphogenesis. Some have phenotypes consistent with defective migration, while others alter testis shape in different ways, revealing the underlying logic of testis morphogenesis. We followed up on the Rho-family GEF dPix in detail. dPix knockdown drastically reduced migration and thus muscle coverage. Our data suggest different isoforms of dPix play distinct roles in this process and reveal a role for its partner Git. We also explored whether dPix regulates Cdc42 activity or cell adhesion. Our RNA-seq dataset and genetic analysis provide an important resource for the community to explore cell migration and organ morphogenesis.

Collective cell migration helps shape organs. Defining genes regulating organogenesis is a key task.
We examine the migration of muscle precursors that ensheathe and thereby sculpt adult Drosophila testes. We used myotube-specific RNA-seq to identify the “parts list” of migrating cells, and knocked down 131 candidate genes, assessing defects in adult testis shape. A total of 29 genes have roles in morphogenesis. Some have phenotypes consistent with migration defects, while others alter testis shape in different ways. We followed up the Rho-family GEF dPix.
Our mutants help define the underlying logic of testis morphogenesis, and our datasets provide an important resource for future research.

INTRODUCTION

Cell migration is critical for normal development, organogenesis, and tissue homeostasis. Some cells, like leukocytes, migrate individually, but many cells migrate collectively (Scarpa and Mayor, 2016; Mishra et al., 2019). Much of what we know about the cell migration machinery comes from studying cells in vitro, either in simplified two-dimensional settings, often using fibroblasts or in more complex three-dimensional models. These systems offer advantages, including relative ease of manipulation and access to high-resolution imaging. However, we ultimately need to understand how cells migrate in their natural settings, inside the bodies of the animals of which they are a part.

To address this gap, researchers have developed a diverse array of in vivo models that illustrate different types of collective cell migration. For example, both zebrafish lateral line development and Drosophila border cell migration provide examples of epithelial cells integrating external signals in a leader-follower manner (reviewed in Bussmann and Raz, 2015; Campanale and Montell, 2023). Similar modes of external attraction or repulsion are used in axon guidance (Aberle, 2019) and in myotube guidance in somatic muscle development (Schnorrer and Dickson, 2004).

Other systems highlight self-regulated collective migration, where cell–cell adhesion molecules synchronize the movement of cells. Examples include Dictyostelium collective motility (Fujimori et al., 2019) or Drosophila follicle cell rotation (Cetera and Horne-Badovinac, 2015). Other cell types remain more loosely linked by cadherin-based junctions and take on more mesenchymal characteristics, as is seen in vertebrate neural crest cells or some collectively migrating cancer cells (Mishra et al., 2019). These diverse modes of migration reflect differences in the underlying cell machinery, both that regulating the cytoskeletal events that mediate motility and the guidance cues that attract or repel cells. One key challenge for our field is to define these underlying mechanisms.

Recently, we developed a new in vivo model for studying collective cell migration (Bischoff et al., 2021). We examine the migration of the muscle precursors that will ensheathe the Drosophila testis in muscle and thereby sculpt the adult testis. During pupal development, these cells, the testis nascent myotubes (TNM), move from the genital disc to the testis (Figure 1A, 30–36 h after puparium formation [APF]). There they migrate through a narrowly confined space between the pigment cells and the germline. They move collectively, loosely connected via N-cadherin-mediated cell-cell junctions, and their motility involves filopodial rather than lamellipodial protrusiveness (Bischoff et al., 2021). Our previous analysis revealed that these cells protrude and thus migrate toward the free cell edge of the sheet. This is a contact-dependent process, with Rho family GTPases regulating the stability of cell-extracellular matrix contacts, destabilizing those at cell-cell edges and stabilizing those at the free edge, thus promoting motility.

FIGURE 1: — Stages of testis morphogenesis and diagrams illustrating the steps used to generate our RNA-seq datasets. (A) Testis morphogenesis begins during pupal development at 30 h APF when nascent myotubes move from the genital disc onto the testis. By 45 h APF, they have covered the entire testis. They then elongate and condense (53 and 66 h APF) enclosing the testis in circumferential muscles. (B–J) The sequential steps in our RNA-seq screen.

This system offers many advantages (Bischoff and Bogdan, 2021). The tissue can be cultured ex vivo, with migration continuing for many hours. This feature, along with the very flat cell morphology, as cells squeeze between the pigment cells and the germline, allows live-cell imaging at very high resolution, approaching what is possible in cultured fibroblasts. In addition, the powerful genetic tools available in Drosophila allow precise manipulations. For example, one can drive gene expression specifically in migrating myotubes using the GAL4-UAS system. Combining this with the virtually genome-wide library of UAS-driven RNA interference (RNAi) reagents allows gene knockdown at will. It also allows expression of fluorescent proteins for visualizing cell protrusions, cell junctions or other subcellular structures. These tools helped power our previous work.

This system also allows us to explore a key issue in organogenesis: how one tissue shapes another. Postmigratory myotubes mature into circumferential muscles that shape the testis by an external “sculpting” process (Figure 1A; 53 h APF). This is reminiscent of mammalian lung morphogenesis, in which smooth muscles shape branching of the airway epithelium (Goodwin et al., 2019; Goodwin et al., 2023). After migration, myotubes elongate perpendicular to the proximodistal axis, and, in a convergent extension-like process, extend the entire testis, resulting in its characteristic spiral shape (Figure 1A; 53 h APF-adulthood; Bischoff and Bogdan, 2023). The external sculpting is fascinating, as it directly links collective migration to morphogenesis, but it is also helpful for performing a genetic screen, as it allows a quick read-out that reveals potential defects in migration and postmigratory shape changes. We can use deviations in adult testis shape (Figure 1A; adulthood) to identify genes leading to defects in migration or postmigration morphogenesis. If the musculature covers only part of the testis due to defects in migration, elongation and ultimately spiral-formation partially fails, and the testis has a dilated tip and the spiral has fewer “revolutions.” If there is almost no coverage or if the genital disc seminal vesicle connection fails, testis retain their pupal ellipsoid shape (Stern, 1941). Cell-cell adhesion defects or failure in active gap closure cause gaps in the muscle sheet, resulting in an uneven surface and kinks in the testis—this can also affect the normal spiral shape of the testis (Rothenbusch-Fender et al., 2017; Bischoff et al., 2021)

Our current goal is to identify the proteins driving contact-regulated directed cell migration, and those involved in shaping the testis after migration. To identify these regulators, we took a two-step approach. First, we used myotube-specific bulk RNA sequencing (RNA-seq) to identify the “parts list” of the migrating cells, that is, the genes expressed during migration. We then used the literature and other resources to identify transmembrane receptors and transmembrane or secreted ligands that might regulate migration. To this list, we added genes known to regulate migration in other systems, and, because of the demonstrated roles of Rho-family GTPases, a set of their regulators. We then obtained available RNAi reagents and used Mef2-GAL4 to drive RNAi specifically in myotubes, knocking down genes one-by-one, and used adult testis phenotypes to identify hits. This revealed a long list of potential regulators. Together our gene expression data and RNAi screen provide the community with resources to further explore this fascinating process of collective cell migration and tissue morphogenesis

RESULTS

Identifying genes expressing in migratory-stage myotubes using RNA-seq

The first step in our effort to identify regulators of myotube collective cell migration was to identify genes expressed in these cells. To do so, we used RNA-seq to identify both genes expressed in migratory stage nascent myotubes, and those expressed in myotubes after migration was completed (Figure 1A, 31 h APF vs. 45 h APF). To identify myotubes in this complex tissue, we used flies in which the muscle specific Mef2-GAL4 driver drives UAS-eGFP. We dissected six sets of 150–200 testis from pupae 31 h APF, a timepoint in the middle of the migratory phase, and six sets of 150–200 testis from pupae 45 h APF, a timepoint after the completion of migration and during intercalation (Figure 1, A and B; Bischoff and Bogdan, 2023). We disassociated cells using Collagenase and Elastase (Figure 1C; Rust et al., 2020). To identify the right population of cells for sorting, we stained them with propidium iodide (PI), which stains DNA in dead cells but is not taken up by living cells, and DyeCycle violet, a cell permeable DNA dye, and sorted them by Fluorescence-activated cell sorting (FACS) (Figure 1D).

We discarded cells with high PI staining (Figure 2A), as they represented dead cells. When analyzing levels of eGFP and DyeCycle violet, we could easily identify three populations of TNM with high GFP-expression, revealing them to be myotubes of the genital tract. On the DyeCycleViolet axis, myotubes formed three distinct populations (1–3)—consistent with differing numbers of nuclei (Figure 2B). Microscopic analysis revealed population 1 to be mononucleated (Figure 2C). The only GFP-positive mononucleated cells in the dissected tissues are myotubes of the ejaculatory duct and the paragonia (Susic-Jung et al., 2012). Micrographs of population 2 and 3 revealed them to have 2 or 3 nuclei, respectively, thus clarifying that these are the multinucleated myotubes that migrate on the testis (Figure 2, D and E). Therefore, we combined population 2 and 3 for the subsequent library preparation.

We next used a TRIzol protocol to obtain RNA from FACS-sorted myotube populations 2 and 3 (Figure 1E). Subsequently, we performed Qbit and Bioanalyzer quality control. cDNA was made and amplified (Figure 1, F and G). We used the Takara SMART-Seq v4 Kit for library preparation, that included Poly-A selection (Figure 1H). For sequencing we used the NextSeq P3 flow cell method with a 2 × 50 bp paired end format (Figure 1, I and J). We worked with mean values of paired ends and lanes. Principal component analysis showed that the 31 h condition and the 45 h condition each cluster together, with 92% variance between them and only 3% variance within conditions (Supplemental Figure S1). For quality control, we used FastQC. FastQC showed that most replicates had high depth of coverage with a minor exception of one replicate from 45 h that had lower reads (average number of reads for 31 h: 109.7M; St. deviation 14.5M; average number of reads for 45 h: 89.0M; St.Dev. 28.7M). All replicates had high quality reads and no problematic reads.

These data provided us with sets of genes expressed at 31 h APF, during migration, thus providing a parts-list for migrating myotubes, as well as the set of genes expressed at 45 h APF, after completion of migration and during intercalation (Supplemental Table S1). We normalized the data using both Fragments Per Kilobase of transcript per Million mapped reads (FPKM) and DESeq2 (v1.40.2) estimates of dispersion parameters both sample- and gene-wise (Supplemental Table S2). As an initial test of whether the gene sets at the two timepoints made sense with what we know is happening at those times, we performed gene set enrichment analysis (Fast gene set enrichment analysis (FGSEA), version 1.28.0) using the genes showing strong differential expression as a “biological quality control.” Consistent with our expectations, the gene ontology (GO) terms “myofibril assembly”, “muscle attachment” and “sarcomere organization” were among the highest up-regulated during the 45 h time step—confirming the validity of the patterns observed in our data set (Figure 3A; Supplemental Table S3). Gene ontology terms related to synaptic development were also enriched, perhaps reflecting the assembly of neuromuscular junctions.

FIGURE 3: — Identifying genes up- and down-regulated at the early and late timepoints. (A) GO terms of genes up-regulated at the 45 h APF timepoint (top, highlighted in gold; equivalent to those down-regulated at the 31 h timepoint) or down-regulated at the 45 h APF timepoint (bottom; equivalent to those up-regulated at the 31 h timepoint). At the right are examples of up-regulated GO terms that reflect muscle differentiation or potential neuromuscular synapse assembly. (B–D) Volcano plots of genes significantly up-regulated at 31 h APF (left, highlighted in teal) or significantly up-regulated 45 h APF (right, highlighted in magenta). Genes which did not reach the significance threshold are depicted in gray (statistical test: Wald test (null model: There is no change between the timepoints). Nominal p-values were adjusted for multiple testing). (B) Highlighted here are selected examples of genes with known roles in muscle differentiation that are significantly up-regulated at the 45 h APF timepoint. (C) Highlighted here are selected examples of genes that encode mesoderm transcription factors or proteins involved in myotube fusion, most of which are significantly up-regulated at the 31 h APF timepoint. (D) Highlighted here are 22 genes that are significantly up-regulated at the 31 h APF timepoint that we chose to include in the RNAi screen.

Next, we inspected the DeSeq2 data manually. This revealed that many of the most up-regulated genes in the postmigratory stage (45 h APF) are directly related to sarcomere and neuromuscular junction formation, such as the muscle myosin heavy and light chains (Mhc and Mlc1), Tropomyosin (Tm2), and Neuromusculin (nrm), a protein involved in synaptic target recognition (Figure 3B). We also created an interactive version of the Volcano plot for others who want to analyze the data looking for different features (https://rpubs.com/Cserody/bischoff_volcano_symbols). During the migratory stage, we expected a higher abundance of early mesodermal transcription factors and of genes that regulate myotube fusion, which occurs directly before migration. Because these GO terms did not exist, we analyzed the same DeSeq2 data manually for typical genes with these functions. Indeed, 14 out of 18 genes we selected are relatively up-regulated during the migratory time step at 31 h APF (≙ down-regulated at 45 h APF), including Lame duck (lmd), a zinc finger transcription factor essential for the specification of fusion competent myotubes and for myotube fusion (Duan et al., 2001), HLH54F, an HLH transcription factor required for the specification and migration of longitudinal gut muscle founders (Ismat et al., 2010), and Rolling pebbles (rols), a protein required for myotube fusion in founder cells (Figure 3C; Menon and Chia, 2001), further validating the data. The full list of differentially expressed genes is in Supplemental Table S2. These data will provide a resource for all wanting to explore the molecular mechanisms underlying myotube migration or testis morphogenesis.

Selecting candidate genes for our RNAi screen

We then used the 31 h APF gene list to identify potential regulators of migration. Our first list of candidate genes was derived by manually examining the genes that were significantly up-regulated during the 31 h timepoint within our DeSeq analysis, relative to the 45 h timepoint. From these we chose a list of 22 genes which, based on their known roles, seemed to be plausible regulators of cell migration (Figure 3D).

To identify additional candidates among genes that were expressed during migration and did not get down-regulated later, we took a second approach. We first identified sets of proteins of interest—for example, transmembrane receptors—using the Flybase-curated “Gene Group” lists. We analyzed multiple gene groups in that way, and normalized the count data to the total number of reads and to gene length (reads per kilobase of transcript per million reads mapped (RPKM); Supplemental Table S4) to look for highly expressed members of each group (Figure 4). Because we were interested in directed cell migration, we focused on transmembrane receptors. Therefore, we selected the 25 most highly expressed genes in the “transmembrane receptor” gene group (Figure 4A). Two of the highly expressed genes were previously found to be functionally important for TNM migration. We excluded the FGF receptor heartless (htl), which was analyzed previously in detail (Rothenbusch-Fender et al., 2017), but retained the Integrin betaPS subunit myospheroid (mys) (Bischoff et al., 2021), reasoning different RNAi lines might reduce its function to different levels. We supplemented our list with the most highly expressed receptor tyrosine kinases (RTKs) that were not already within the top 25 transmembrane receptors but were still among the nine highest expressed RTKs (Figure 4B). Interestingly, genes encoding multiple transmembrane receptors that are involved in axonal pathfinding—a process related to contact-dependent cell migration—are highly expressed during migration (31 h APF, Figure 3D). These include the classical axon guidance factors Plexin A (PlexA), the Plexin coreceptor off-track (otk), the Plexin B ligand Semaphorin 2a (Sema2a), Netrin B (NetB), beaten path IIb (beatIIb), beaten path IIIc (beatIIIc), the Ephrin receptor tyrosine kinase (Eph) and the Latrophilin homologue Calcium-independent receptor for α-latrotoxin (Cirl) (Figure 4, A and B).

FIGURE 4: — Genes selected for our RNAi screen. Four lists derived from the Flybase-curated “Gene Group” lists. In each, genes are ranked by level of expression (reads per kilobase per million mapped reads), and genes included in our RNAi screen are boxed in red. (A) Transmembrane receptors. (B) Receptor tyrosine kinases. (C) Cadherin family members. (D) Integrin subunits.

In parallel with identifying potential transmembrane receptors, we also looked for possible transmembrane or secreted ligands that might be involved. From the “receptor ligand” gene group, we included the three most highly expressed genes, mav, miple1 and Semaphorin2a, in our candidate list. Because Wnt-signaling is known to play a role in TNM development (Kozopas et al., 1998; Rothenbusch-Fender et al., 2017), we added the two most highly expressed Wnt-genes, Wnt4 and Wnt5.

Cell-cell and cell-matrix adhesion also play important roles in cell migration. As signals might be transduced directly via cadherin family receptors, we analyzed the expression of all Cadherins (Figure 4C). Consistent with our previous finding that N-Cadherin (Cad-N) plays important role in TNM migration (Bischoff et al., 2021), it was the second most highly expressed cadherin. We thus included Cals, kug, and shg on our candidate gene list. We also included the three most highly expressed integrin subunits (Figure 4D). Because proteolytic degradation of the extracellular matrix (ECM) environment might be involved in TNM migration we opted to include the three most highly expressed Metzincin matrix metalloproteases AdamTS-A, kuz and AdamTS-B.

In addition to proteins directly involved in cell-cell signaling, we also explored a subset of cytoskeletal regulators. Pharmacological perturbations revealed that Formins play an important role in TNM migration (Bischoff et al., 2021). We thus added most highly expressed Formins DAAM and Frl to our list, and added the actin-regulatory F-Bar protein Cip4. Finally, Rho family GTPases are fundamental for TNM migration (Bischoff et al., 2021). Guanine nucleotide exchange factors (GEFs) activate these GTPases by stimulating the release of GDP to allow GTP binding, while GTPase-activating proteins (GAPs) turn GTPases off by binding to activated G proteins and stimulating their GTPase activity. We thus included all existing Rho-family GEFs and GAPs to the screen. Due to their important role in the regulation of cell adhesion, we also included all Flybase-annotated Rap1 GTPase GAPs and GEFs. Finally, it is worth noting that the expression level cut-offs we used for choosing genes to include in the screen were arbitrary, and many proteins not selected might also have roles in testis morphogenesis—our RNA-seq data could be used to prioritize other genes to follow-up in the future.

The diverse phenotypes observed in our screen reveal the complexity of events that shape this organ

Using this gene set, we examined the collections of UAS-RNAi reagents at the Vienna (Dietzl et al., 2007) and Bloomington Drosophila Stock centers (Ni et al., 2011) and ordered one or more RNAi lines for each gene. Transgenes regulated by UAS sequences can be activated the GAL4 transcription factor. We used a line in which GAL4 is under control of the promotor of the muscle-specific gene Mef2. This allowed us to knockdown gene expression in a tissue-specific manner and assess the potential role of each gene in myotube migration. Females carrying Mef2-GAL4 were crossed to males carrying each UAS-RNAi line. One-to-three-day old adult male progeny were collected and testis dissected from 10 males, replicated three times. Testis were examined under a dissecting scope. For any lines with potential morphological defects, testes were then stained with fluorescently labeled phalloidin, which binds F-actin and thus outlines muscles. These were then imaged by confocal microscopy.

We tested 279 RNAi lines covering 131 genes. A subset of the RNAi lines led to embryonic or larval lethality, precluding analysis of adult phenotypes (Supplemental Table S5). Some led to pharate adult lethality, before adult eclosion—for these genes, testes were dissected from pharate adults. We identified morphological defects for 29 genes (Table 1). These varied widely in the strength, penetrance, and nature of the defects observed (Supplemental Table S5).

A subset of the RNAi lines alters muscle coverage, suggesting defects in cell migration or cell adhesion

Our original goal was to define the factors required for testis nascent myotube migration. The wild-type testis is full covered in circumferential muscle and has a spiral shape with a very gradual reduction in diameter toward the proximal end (Figure 5A). Reduced migration or failure to close gaps in the sheet could lead to defects in muscle coverage. However, as our screen proceeded and phenotypes began to emerge, we realized our mutant collection included genes that are required for diverse processes in testis morphogenesis.

FIGURE 5: — RNAi lines that alter muscle coverage or distal testis shape. Adult testes from wild type or adults expressing a UAS-driven RNAi line targeting the indicated gene under the control of *Mef2-*GAL4. All are stained with fluorescently-labeled phalloidin to reveal F-actin, which highlights muscles. (A) Wild type. The wild-type testis is fully covered in circumferential muscle and has a spiral shape with a very gradual reduction in diameter toward the proximal end. (B–T) Testis from knockdown lines, arranged according to phenotypic class and severity. Detailed descriptions of each are in the text. (B–D) Extreme loss of muscle coverage. The inset in D shows a close-up illustrating the “striated muscle phenotype.” (E–I) Strong loss of distal muscle coverage (arrows). The more proximal arrow in G shows variation in testis diameter. (J–M) Variable loss of distal muscles or gaps in muscle coverage (arrows). (N–P) Strong N or weaker gaps that are not confined to the distal end (arrows). (Q–T) Distal testis is enlarged (arrows). B–P are genes included in our Class 1 (Migration/Adhesion defects). Q–T are a subset of the genes included in our Class 2 (Testis Shaping defects).

The first class of knockdowns (Class 1 = Migration/adhesion defects) had phenotypes resembling those we had anticipated from defects in migration. In these there was a failure to fully cover the testis in muscle, ranging from weaker effects in which the distal tip was uncovered to more severe effects with substantial loss of muscle coverage. The two strongest effects were seen when knocking down dPix (also known as RtGEF; Figure 5B), a Rho-family GEF targeting Rac1 and Cdc42, or the Jak/Stat receptor Dome (Figure 5D; a second RNAi line had a weaker version of this defect). We saw a similar defect when expressing a dominant-negative version of the matrix-metalloprotease Kuz (Figure 5C). For these genes, knockdown led to dramatic testis shape changes, with a highly expanded distal end and shortened proximal distal axis. The line targeting Dome also had an interesting additional effect: the muscles remaining appeared striated in morphology (Figure 5D, inset), suggesting defects in cell fate determination similar to those previously seen in dWnt2 hypomorphic mutants (Rothenbusch-Fender et al., 2017).

Other genes with potential effects on migration and adhesion included those for which RNAi led to strong, moderate or mild loss of distal muscle coverage. Very strong loss of distal muscle coverage was seen after knockdown of the adapter protein Rols (Figure 5E), CG9098, encoding an uncharacterized predicted GEF with an N-terminal SH2 domain whose Nsp family mammalian orthologs are predicted to lack enzymatic activity and instead act as adapters (Figure 5F), or the F-Bar protein Cip4 (Figure 5G). CG9098 knockdown also led to narrowing of the proximal testis, while Cip4 knockdown also led to additional alterations in overall testis shape. Knockdown of the DOCK family RhoGEF Spg (Figure 5H), RhoGEF2 (Figure 5I), or the axon guidance receptor PlexA (two lines had similar phenotypes; Figure 5, J and K) all caused moderate loss of distal muscle coverage. Spg and RhoGEF2 were two of the genes where different RNAi lines caused distinct phenotypes (Table 1)—we mention the other lines below. More mild loss of distal muscle coverage was seen after knockdown of the unconventional bipartite Rac-GEF Mbc (Figure 5L; 1 of 2 lines had this effect). A subset of the testis expressing the dominant-negative version of the matrix-metalloprotease Kuz also had this milder muscle loss phenotype (Figure 5M).

Some RNAi lines led to gaps all along the sheet, perhaps reflecting milder defects in migration, or defects in cell adhesion or the ability to actively close gaps. Knockdown of the G-protein coupled receptor (GPCR) Tre1 led to extensive muscle coverage gaps that extended much more proximally (Figure 5N). One of the RNAi lines knocking down the Jak/STAT receptor Dome has similar but much weaker gaps restricted to the proximal part of the testis (Figure 5O), a phenotype also caused by knockdown of RhoGAPp190 (Figure 5P).

Other RNAi lines altered testis morphogenesis in more complex ways, potentially affecting testis shaping during later morphogenesis

These were only a subset of the RNAi lines that altered the adult testis. More surprising to us were the lines in which knockdowns retained nearly complete or full muscle coverage, but in which the final muscle sheet seemed to have different physical properties, not allowing normal tissue shaping (Class 2 = Testis shaping defects). Some of these had relatively mild effects. One of the lines knocking down Mbc led to broadening of the distal third of the testis (Figure 5Q), while one of the lines knocking down the DOCK family RhoGEF Spg (Figure 5R), the line knocking down the Rho family GAP CDGAPr (Figure 5S), and one line knocking down the cell cycle kinase Polo (Figure 5T) led to mild broadening of the distal tip—other testes affected by this polo RNAi line failed to connect to the seminal vesicle and thus muscle migration never started.

Other knockdowns affected the “shaping” of the testis in more complex ways. In wild type, the testis diameter is fairly uniform, with very gradual and smooth reduction in diameter toward the proximal end. Some RNAi lines affected the consistency of the testis diameter, with a resulting “waviness” of the testis border. Three RNAi lines led to both broadening of the distal end and a waviness of the testis border due to variations in diameter. These included knockdown of the axon guidance ligand NetB (Figure 6B) and two lines knocking down the alpha-tubulin isoform αtub85E (Figure 6, C and D). Knockdown of the Ras/Rap family GAP Gig or the BTB/POZ domain protein Rsh led to variation in testis diameter without distal enlargement (Figure 6, E and F). Knockdown of the RhoGAP Rlip and one of the lines targeting RhoGEF2 broadened and slightly shortened the whole testis, with (Rlip) or without (RhoGEF2) variation in testis diameter (Figure 6, G and H). Knockdown of the RhoGEF Pura had a very curious effect—broadening the distal testis while narrowing the proximal testis (Figure 6I). In contrast, two of the lines targeting the Ras GEF Sos led to narrowing only of the distal testis tip (Figure 6, J and K). Finally, knockdown of the FERM domain-containing protein Cdep or knockdown of RhoGEF64C narrowed and elongated the entire testis (Figure 6, L and M).

A final class of RNAi lines affected muscle alignment

In wild type, the muscles are well aligned into a circumferential pattern (Figure 6A, inset). A subset of the RNAi lines altered this normal parallel circumferential alignment (Figure 6O, inset), perhaps reflecting problems in the parallel alignment/intercalation phase (Class 3 = Muscle alignment and shape altered). In the testis of animals expressing these RNAi lines, spiral formation completely or partially fails. Two of the genes with this phenotype form a ligand/receptor pair: two RNAi lines targeting the gene encoding the Wnt receptor fz2 (Figure 6O, inset, P) and two lines encoding its ligand dWnt4 (Figure 6, Q and R). The RNAi line targeting the gene encoding the Hedgehog pathway receptor Smo altered muscle arrangement and testis shaping distally (Figure 6S) while an RNAi line targeting the GPCR Cirl led to a somewhat similar phenotype, with the distal end of the testis enlarged and some alterations in muscle arrangement (Figure 6T). Knockdown of the Ig-family receptor Beat-IIIc led to a complex phenotype including failure of testis spiraling, extremely variable testis diameter and muscle alignment defects (Figure 6U). Intriguingly, overexpression of Beat-IIIc and loss-of-function of Wnt4 and Fz2 all alter motor neuron synaptic specificity (Inaki et al., 2007).

Finally, a few other RNAi lines had more complex effects. Several RNAi lines led to apparent failure of the genital disc to connect to the testis in some individuals: these included lines targeting the genes encoding polo, dPix, and rols (Supplemental Table S5). Finally, in animals in which the RNAi line targeted the gene encoding RapGAP1, the testis was extremely small (Supplemental Figure S2A vs. S2B), though the remnant was covered in muscle (Supplemental Figure S2C), suggesting a possible non-tissue autonomous effect.

It is important to note that genes that scored negative in our screen may still play important roles in testis morphogenesis. Some RNAi lines do not effectively knockdown the targeted mRNA, and thus may be false negatives. There also was variability in penetrance from animal to animal with many lines. We include in Supplemental Table S5 nine additional genes where the phenotype penetrance fell below our cutoff—this group also included additional RNAi lines for six of the genes where a different line did lead to a phenotype we scored as positive. Follow-up on these nine is likely warranted. Finally, it is interesting to compare our results with those of an earlier screen that systematically analyzed knockdown of genes in muscle, also driving RNAi with Mef2-GAL4 (Schnorrer et al., 2010). They screened for effects on viability, wing posture, locomotion and flight. Of the 29 genes on our list, seven were among those where at least one RNAi line led to lethality—RNAi lines like these would not have been assessed by us unless the lethality was as pharate adults. Nine additional genes were scored as wild type for the phenotypes they assessed, suggesting their defects in other muscles were not substantial enough to lead to defects in viability, wing posture, locomotion or flight—though of course, different RNAi lines for the same gene can differ in penetrance. Only three genes scored in both screens for phenotypes other than lethality: Cdep knockdown led to locomotion defects, Gig knockdown led to a weak flyer phenotype, and Smo knockdown led to a wing posture defect.

dPix RNAi leads to a very strong delay in cell migration

To illustrate the utility of these combined RNA-seq and RNAi datasets, we examined one of our hits in more detail. Among the strongest phenotypes observed in our screen was that caused by knockdown of the dPix protein (also known as RtGEF), the Drosophila homologue of the mammalian Rho-type GEF beta-PIX (Zhou et al., 2016). After dPix^RNAi, many testes were still round and largely uncovered in muscle, while in less severe examples the testis was shortened in the proximal distal axis and the distal end was uncovered (Figure 7, A and B; representative of 22 testes). dPix has known roles in synaptic structure and growth in the nervous system (Parnas et al., 2001; Ho and Treisman, 2020), in maintaining epithelial architecture and collective cell migration of the follicle cells during oogenesis (Dent et al., 2019), and in regulating Hippo signaling in imaginal discs (Dent et al., 2015). The RNAi line we used is a well validated one, which is known to reduce dPix mRNA, and to mimic the effect of the strong dPix¹⁰³⁶ mutant in effects on Hippo signaling in imaginal discs (Dent et al., 2015).

FIGURE 7: — dPix knockdown dramatically slows myotube migration, and reducing N-cadherin partially suppresses adult testis defects. (A and B) Adult testis illustrating the range of phenotypes seen after *dPix^RNAi*. (C) Wild-type myotube migration beginning ∼38 h APF. Green = Lifeact-eGFP, revealing actin. Magenta = mCherry with an added NLS, marking nuclei. (D and E) Representative examples of the delay in migration after *dPix^RNAi*. Some myotubes that are far apart remain interconnected by long processes that stretch over myotubes located between them (arrows). (F) Adult testis after *Ncad^RNAi*. (G) Adult testis after *dPix^RNAi*; *Ncad^RNAi*. (H) Quantification of phenotypic severity, quantified as ratio of width to length. Statistical test: ordinary One-way ANOVA with Šídák's multiple comparisons test. Adjusted p-values: *NCad*^RNAi versus *dPix*^RNAi: <0.0001, *NCad*^RNAi versus *NCad*^RNAi + *dPix*^RNAi 0.0019, *dPix*^RNAi versus *NCad*^RNAi + *dPix*^RNAi: 0.015.

To understand how migration was affected after dPix^RNAi to lead to this severe adult phenotype, we live-imaged explanted pupal testis during migration (31 h APF). We visualized cells and their protrusions using Lifeact-eGFP, revealing actin, and nuclei using mCherry with an added nuclear localization signal (NLS). In wild type, when nascent myotubes from the genital disc encounter the germline, they begin to migrate into the space between the underlying germline cells and the overlying pigment cells (Figure 7C; green = Lifeact-eGFP; magenta = mCherry-NLS). Cells migrate as a mesenchymal cohort, joined by N-cadherin junctions (Bischoff et al., 2021). They have filopodia both at the leading edge and interdigitated filopodia at cell-cell junctions. Cells are stimulated to migrate in the direction where they sense a free edge, with this directionality determined in part by slower turnover of focal adhesions at free edges and faster turnover at cell-cell borders (Bischoff et al., 2021). However, cells remain in contact with neighbors throughout migration (Figure 7C, C’), and gaps between cells are closed by a combination of N-cadherin mediated adhesion, free edge protrusions and actin-based purse strings that form around gaps.

When we live-imaged dPix^RNAi testis, we observed a dramatic delay in migration behavior. Nascent myotubes moved onto the testis from the genital disc, but their progress from the proximal end was exceptionally slow (Figure 7, D and E; representative of eight movies—images of the full set after 420 min are in Supplemental Figure S3). This is consistent with the adult phenotype, where only the most proximal testis is covered in muscle. In fact, the movies we took understate the difference. We could not capture images from wild-type testes in which the myotubes were just starting to enter the testis, as at that stage in wild type the connection between the genital disc and the testis was not fully formed, and they detach upon dissection. This suggests the testes we observed after dPix knockdown were in fact developmentally further along than one would think given the distance the myotubes had migrated, as they remained connected. However, despite the delayed migration the cells did not dramatically differ from wild-type myotubes—they were well spread on the underlying cyst cells and did not detach from their neighbors. While some images at a very early stage had many filopodia (Figure 7E), as we could not capture wild-type cells at this stage, it is hard to determine whether this is abnormal. Later, filopodial number did not seem elevated (Figure 7, D and E; Supplemental Figure S3)—in fact it might be reduced. Future work will be needed to quantitatively analyze protrusive behavior after dPix knockdown. One feature was notable; some myotubes that were far apart remained interconnected by long processes that stretch over myotubes located between (Figure 7, D and E, arrows), suggesting potential defects in cell-cell contact disassembly upon neighbor exchange. Thus, dPix^RNAi dramatically alters migration rate.

One possible explanation for the dPix^RNAi phenotype was an effect on cell-cell adhesion. Perhaps elevated cell adhesion prevented cells from escaping the proximal region. If this was the case, we reasoned that reducing cell–cell adhesion might suppress the dPix^RNAi phenotype. Myotubes are connected by N-cadherin–mediated cell junctions, and adhesion is important for sealing gaps in the muscle sheet. Knockdown of N-cadherin leads to gaps in muscle coverage all along the proximal to distal axis (Figure 7F; Rothenbusch-Fender et al., 2017). We thus asked whether reducing N-cadherin by Ncad^RNAi suppressed the dPix^RNAi phenotype. To avoid potential issues with differences in the number of UAS-transgenes, UAS-GFP was added to single knockdown crosses so all have two UAS lines. As noted above, the phenotype of dPix^RNAi varies from very strong reduction in muscle coverage and an almost round testis to examples where the testis was shortened in the proximal-distal axis but muscle coverage defects were less severe. To quantify suppression, we assessed the degree of shortening of the testis, calculating the ratio of width to length. Ncad^RNAi alone does not substantially shorten the testis (Figure 7H; n = 16 for all treatments). dPix^RNAi leads to a broad range of phenotypes, with some testis severely shortened and rounded and others less shortened (Figure 7H). In contrast, animals in which both were knocked down (dPix^RNAi Ncad^RNAi) had testis that were almost all in the less-severe category (Figure 7, G and H). Thus, reducing cell–cell adhesion alleviates the effect of dPix^RNAi. In the future, it will be of interest to live-image migration after double knockdown, to see how migration changes.

Knocking down different dPix isoforms has different effects on testis shaping

The dPix gene is complex, with eight differentially spliced isoforms (Figure 8A). The catalytic Rho-GEF (Dbl homology (DH)) domain, the N-terminal SH3 domain and the pleckstrin homology (PH) domain are all encoded in the common exons shared by all isoforms (Figure 8A). However, other exons differ in complex ways between the different isoforms. One major difference is between isoforms F, H, and I and the other isoforms. F, H, and I share a long exon just downstream of the catalytic domain that is missing in the other isoforms (Figure 8A). This difference also means splicing of this FHI-specific exon into the next downstream exon is in a different protein reading frame, and thus all three of these also lack the downstream protein region containing the binding site for the Pix partner Git (Figure 8A). Below we refer to these collectively as the FHI isoforms.

Our RNA-seq dataset allowed us to define which isoforms are expressed during myotube collective migration. When we mapped sequence reads to the map of different isoforms, we saw something surprising. At the 31 h mid-migration timepoint, almost all transcription appears to start at the upstream start site (Figure 8B, green vs. magenta arrows), suggesting isoforms B, F, and G are expressed at very low to zero levels. There are, at most, very low levels of expression of the exon shared by isoforms D and G (Figure 8B, black arrow), although these isoforms were detected in the 45 h postmigration sample. This leaves isoforms A, E, H, and I. The long exon shared by the F, H, and I isoforms is also expressed at very low levels (Figure 8B, cyan arrow), and intriguingly the reads only extend part of the way through the predicted exon. Putting these data together, it appears that the predominant dPix isoforms expressed during migration are isoforms A and E, with levels of A about twice as high (Figure 8B, blue arrows). Both of these share the predicted GIT—binding site (Figure 8A).

Previous work revealed that different isoforms of dPix differ functionally, having differential importance in distinct tissues. The RNAi line which we used, shRNA-HMS00741, is the one used by Dent et al. (Dent et al., 2015) and it targets isoforms A, B, D, and F. They found that this RNAi line reduces levels of dPix mRNA and alters Hippo signaling in imaginal discs in ways similar to the effects of the strong allele dPix¹⁰³⁶, suggesting that isoforms A, B, D, and/or F are important in those tissues. Isoform A, one of those targeted by this RNAi line, is sufficient to rescue dPix mutant defects in oogenesis (Dent et al., 2019). Thus, the RNAi line we used would target isoform(s) important for function in those tissues.

However, while examining the role of dPix in neuromuscular synapse growth Ho and Treisman revealed additional complexity (Ho and Treisman, 2020). In this tissue, knocking down isoforms A, B, D, and F, using the RNAi line we used, did not mimic the effects in that tissue seen the strong allele dPix¹⁰³⁶. Instead, in that tissue, an RNAi line targeting the large exon shared by isoforms F, H, and I (shRNA-KK13571), and an allele with a mobile element insertion into this exon dPix^MB10902, mimicked the effect of the strong allele dPix¹⁰³⁶, suggesting that in this tissue the F, H, and I isoforms are predominant. Most intriguing, when they used the RNAi line we used, shRNA-HMS00741, targeting isoforms A, B, and D, it had an effect on synapses opposite that caused by knockdown of isoforms F, H, and I (Ho and Treisman, 2020). From this and other data they concluded that F, H, and I isoforms have an antagonistic relationship with the A, B, and D isoforms and suggested that A, B, and D isoforms sequester F, H, and I isoforms and therefore prevent F, H, and I isoforms from performing their function. In their interpretation, the RNAi line that we observed to cause strong defects in testes eliminates (A, B, D, and F) isoforms, and thus potentially could cause overactivation of H and I isoforms, even though F is affected as well.

To begin to explore the roles of different isoforms of dPix in shaping the testis, we used additional genetic reagents. As we outlined above, the RNAi line we identified in the screen, which knocks down isoforms A, B, D and F (HMS00741), causes a very strong defect in migration, with most of the testis uncovered by muscle (Figure 8, C vs. D)—in the model of Ho and Treissman that could result from H/I overactivation. We next examined animals homozygous for the strong allele dPix¹⁰³⁶, which results from a mobile element insertion in the first intron (Figure 8A) and has strongly reduced levels of protein in mutant embryos (Parnas et al., 2001). In a subset of these animals, the testis had no muscle coverage, suggesting failure of attachment of the genital disc to the testis (7/17 testis). In those with muscle coverage, the testis had dilated tips of variable severity (Figure 8, E and F; 10/17 testes) and the testis was substantially shortened (Figure 8Q). Animals with the strong allele dPix¹⁰³⁶ over a deficiency that removes the dPix gene (Df(2L)Exel6046) had stronger but variable testis shape phenotypes, varying from a dilated tip to more extreme defects in testis shape (Figure 8G; representative of 14 testes)—all were also shortened (Figure 8Q) and had reduced testis coiling. We observed similar defects in testis shaping in dPix¹⁰³⁶/Df(2L) ED1315 testes (Figure 8, H and Q; representative of 13 testes). Finally, an RNAi line that should target all isoforms (shRNA-KK108300) also led to dilated tips (Figure 8I; representative of 4/7 testes) but did not shorten the testis (Figure 8Q). These data support a role for dPix in shaping the testis.

However, intriguingly, knocking down isoforms F, H, and I, using the same RNAi line that gave Ho and Treisman their strongest phenotype (shRNA-KK13571), had only mild effects on testis shape and length (Figure 8J; 2/6 testes had dilated tips; Figure 8Q). dPix^MB10902, the insertional mutant that disrupts the shared exon of these isoforms, had no effect on testis shape and a modest effect on testis length (Figure 8, K and Q; 5/8 testis examined), though a subset of mutants had apparent failure of attachment of the genital disc to the testis (3/8 testis examined). We thus tested the idea that FHI isoforms act antagonistically to the ABD isoforms. To do so, we used UAS-driven constructs that express the F-isoform (Ho and Treisman, 2020). This led to penetrant distal tip expansion and gaps in distal muscle coverage (Figure 8L; 18/23 testes examined) and shortened the testis (Figure 8Q). In contrast, overexpressing the A and B isoform had little or no effect on testis morphology (Figure 8, M and Q; 7 testes examined). Together, these data support the idea that the A, B, and D isoforms play the largest role in the testis, consistent with the fact that isoform A is most highly expressed at the RNA level. They also are consistent with the idea of antagonism between different isoforms. To further explore this, we examined whether combining a mutant affecting the F, H, and I isoforms, dPix^MB10902, with our dPix RNAi line targeting isoforms A, B, D, and F, might suppress the RNAi phenotype. However, while dPix^MB10902 is homozygous viable and expressing that RNAi line alone does not lead to lethality, dPix^MB10902 is lethal in combination with our strong dPix-RNAi, suggesting that balance among the different dPix isoforms is important for additional processes in other tissues.

The dPix binding partner Git plays an important role in testis shaping

Pix proteins are unique among RhoGEFs in that they can form heterodimers with GIT family proteins, a family of Arf GAPs. Each can homodimerize or they can heterodimerize to form the Pix-GIT signaling scaffold. Some Pix roles are shared with GIT and others are not. For example. in Drosophila dPix and GIT work together in Hippo signaling in imaginal discs (Dent et al., 2015) and in regulating follicle cell epithelial architecture during oogenesis (Dent et al., 2019). In contrast, Git does not play a role in neuromuscular synapse growth (Ho and Treisman, 2020). Strikingly, the one tissue where Git does not parallel dPix in function is one in which isoforms F, H, and I, which lack the Git interacting region, are important, whereas tissues that appear to rely on other isoforms like isoform A do require Git.

Because isoform A and most of the other isoforms containing the Git binding site were targeted by the dPix RNAi line that gave our strong phenotype, and since those isoforms were also those most strongly expressed in migrating myotubes, we examined the effect of Git mutants on testis shaping. We examined two different mutants, git^Ex21c, a probable null allele that deletes the first 109 amino acids (Bahri et al., 2009), and git^F03586, a piggyBac transposon insertion in an intron that is reported to be protein null (Podufall et al., 2014). git^Ex21c homozygotes had penetrant defects, with strong enlargement of the distal testis and muscle coverage gaps (Figure 8, N and O; 14/17 testis examined). git^F03586 homozygotes also had distal enlargement and muscle coverage or alignment issues, though these were less penetrant (Figure 8P; 4/8 testis examined). The testis was also shorter and broader in both mutants (Figure 8Q). This suggests the dPix/Git complex regulates testis shaping.

dPix RNAi elevates rather than decreases the activity of a Cdc42 sensor

Biochemical and cell biological studies revealed that mammalian beta-Pix can act a GEF for Rac1 and Cdc42. We thus hypothesized dPix knockdown would reduce levels of Cdc42 activity. We used an EGFP-tagged biosensor for Cdc42 activity, based on the Cdc42-and Rac-interactive binding (CRIB) domain of the Pak family kinase Mbt. Mbt is known to bind Cdc42 in a GTP-dependent manner (Schneeberger and Raabe, 2003). The sensor we used has previously been tested in pulldown assays—it was pulled down by both GDP- and GTP-loaded Cdc42, with only a modest preference for active Cdc42 (Rötte et al., 2024).

We extended this by testing the sensor in testis myotubes, expressing it there using the Mef2-GAL4 driver. We analyzed myotubes on the seminal vesicle at the base of the testis where cortical localization was most easily visualized as the myotubes are closely packed there. The sensor was strongly enriched at the cell cortex (Figure 9A), consistent with the idea that endogenous Cdc42 and/or Rac activity are high there. Cdc42 RNAi substantially reduced the cortical signal (Figure 9B), something we verified using line scans across the plasma membrane (Figure 9C) and which we observed with two different Cdc42 RNAi lines (Figure 9F). Rac2 RNAi also reduced the cortical membrane signal (Figure 9, D and E), though not by as much as Cdc42 RNAi (Figure 9F). To test whether in this tissue the sensor only sensed Cdc42 protein levels rather than active Cdc42, we also expressed a GDP-locked (dominant-negative) version of Cdc42 in our target tissue. If the sensor solely reported Cdc42 levels, this should have elevated sensor activity—instead sensor activity was mildly reduced (Figure 9, G and H).

FIGURE 9: — Knockdown of the AB isoforms of dPix elevates rather than reduces the signal from a Cdc42 biosensor. (A, B, D, G, I, J) Images of the Cdc42 biosensor (the CRIB domain of Mbt fused to EGFP) expressed in the myotubes on the seminal vesicle at the base of the testis, using *Mef2-GAL4*. (A) In wild type, the biosensor signal is found at the plasma membrane. (B) *Cdc42* knockdown reduces cortical biosensor signal. (C) Line scans perpendicular to the membrane reveal the reduction in the cortical signal. Wild type: N = 7, RNAi: N = 8. (D and E) Slightly reduced biosensor signal after Rac2 knockdown. Wild type: N = 6, RNAi: N = 8. (F) Quantification of normalized membrane signal in wild type, two different *Cdc42* RNAi lines, and a *Rac2* RNAi line. Statistical tests: Kruskal Wallis test and Dunn's multiple comparisons test. Wild type versus *Cdc42*^{RNAi 1}: p < 0.0001, wild type versus *Cdc42*^{RNAi 2:} p < 0.0001, wild type versus *Rac2*^RNAi: p = 0.0075. (G and H) Slightly reduced biosensor signal after expression of a GDP-locked Cdc42N17 mutant. Wild type: N = 8, Cdc42N17: N = 8. (I and J). *dPix* knockdown elevates the cortical biosensor signal. Wild type: N = 17, RNAi: N = 30. (K) Line scans perpendicular to the membrane. (L) Quantification of normalized membrane signal. Statistical test: Student's t test, p < 0.0001.

We next knocked down dPix, using the RNAi line that has the strongest effect on testis migration. To our surprise, biosensor signal was not reduced as we had expected. Instead, it was increased somewhat (Figure 9, I vs. J; representative of 17 wild type and 30 dPix^RNAi samples). Once again, this apparent elevation of the sensor at the cortex was verified by line scans across the membrane (Figure 9K), and the difference in membrane signal was statistically significant across our samples (Figure 9L). In the future, it will be important to test this difference with additional sensors for Cc42 and Rac1, and to test whether the increase we observe after dPix knockdown might involve changes in Rho activity, given the known RhoA-Cdc42 antagonism. However, together with the analysis above, this suggests different dPix isoforms may have surprising effects.

DISCUSSION

One key issue for our field is defining the molecular mechanisms underlying organogenesis. The Drosophila testis is shaped by a set of circumferential muscles that arise via the collective cell migration of myotubes from the genital disc onto the testis during pupal development. They migrate to fully enclose the underlying germline and soma (Rothenbusch-Fender et al., 2017; Bischoff and Bogdan, 2021). We previously characterized this migration, revealing that these cells migrate in the confined space between the germline and the overlying pigment cells (Bischoff et al., 2021). These cells are loosely connected by N-cadherin mediated adhesion and move toward free edges, thus driving cells forward. They also close gaps between the cells, preventing later holes in muscle coverage. Finally, they differentiate into aligned circumferential muscles that shape the testis into its stereotypical elongated spiral shape (Figure 10A)

FIGURE 10: — Cartoon illustrating the sequential events of testis morphogenesis and the steps at which we hypothesize each gene may act. (A) Diagrams of the entire testis and closeups of the myotubes at each step. (B) Changes in actin and N-cadherin localization at each step. (C) Stages of morphogenesis and proposed steps at which different genes act.

Intriguing parallels between collective cell migration and axon guidance

Our initial goal in setting up this screen was to identify regulators of collective cell migration and cell-cell adhesion, by identifying adult testis phenotypes we predicted would be caused by defects affecting these mechanisms. Many of our RNAi knockdowns fit the expected profile. These included several with partial coverage of the adult testis in muscle, leading to gaps in the distal region, consistent with defective migration. Others had gaps in muscle coverage all along the proximal-distal axis, suggestive of more severe migration defects or defects in adhesion. Our RNA-seq data and the results of our knockdown screen provide some potentially interesting insights.

As neurons send out axons, they interact with other axons, with other cell types, and with extracellular substrates along their path in a form of directed cell migration. We found that many proteins known for their roles as axon guidance factors in the nervous system are highly expressed during testis nascent myotube collective cell migration. Further, knockdown of four of these cause defects in testis morphogenesis: Plexin A, Netrin B, beat-IIIc, and the Latrophilin homologue Cirl. In principle, axon guidance shares many features with contact-regulated modes of mesenchymal collective cell migration, with related emergent behaviors regulated via direct contact or secretion of guidance factors. Contact-dependent axon repulsion, for example, has features in common with contact inhibition of locomotion (Stramer and Mayor, 2017). During Drosophila follicle cell rotation, Semaphorin/PlexA and Fat2/Lar-signaling are crucial to regulate contact-dependent planar-polarized protrusion formation, thus driving and synchronizing cellular locomotion (Stedden et al., 2019). Substrate-derived guidance cues are another feature that was previously recognized to be highly similar between axon guidance and collective cell migration (Aberle, 2019). Consistently, Netrin-signaling also affects collective cell migration, for example, in cultured mammalian liver cells (Han et al., 2019) and in vivo in Drosophila cardiac cell migration (Raza and Jacobs, 2016). One speculative possibility is that an ancestral function of these proteins is to regulate collective cell migration outside of the nervous system. Consistent with this notion, both Plexins and Netrins appeared early in or even before animal evolution and predate the evolution of the nervous system (Junqueira Alves et al., 2021; Cortes et al., 2023). The results of our screen provide exciting opportunities for future studies to define non-neuronal roles of canonical “axon guidance factors” by understanding their functions in the context of testis myotube migration.

The regulation of Rho family and Ras/Rap GTPases plays important roles in many aspects of testis morphogenesis

The small Rho-family GTPases Cdc42, Rac2 and RhoA all play nonconventional roles in myotube collective cell migration. Cdc42 and Rac2 regulate integrin-adhesion lifetime and hence the ability to migrate, and RhoA stimulates retraction of cell edges and filopodia (Bischoff et al., 2021). The receptor tyrosine kinase Htl also plays a role, suggesting that Ras family small GTPases may be involved (Rothenbusch-Fender et al., 2017). One of our goals was to begin to uncover the regulators of these GTPases. To do so, we analyzed the consequences of knockdown of all Drosophila Rho-family and Ras/Rap GEFs and GAPs. This revealed that multiple GEFs and GAPs are necessary for normal testis morphogenesis and full muscle coverage. This suggests a high level of fine-tuning of GTPase activation and deactivation.

To illustrate this, we began to dissect the role of dPix in more detail. dPix is the fly homologue of mammalian beta-Pix, which is a GEF for Rac1 and Cdc42. beta-Pix regulates diverse cellular processes from synaptogenesis to collective cell migration, both in vitro (Plutoni et al., 2016) and in vivo (Omelchenko et al., 2020). One of the strongest hits in our knockdown screen came when using a well-validated RNAi reagent (Dent et al., 2015) targeting dPix. This led to strong to nearly complete loss of muscle coverage of the adult testis. When we examined pupal testis myotube migration live, we found very strong delays in migration, consistent with this adult defect.

However, as we dug deeper, the story became more complex. Pix has multiple isoforms and its function in different tissues appears to differentially depend on their respective functions. Our initial RNAi line targets a subset of the isoforms, A, B, D, and F. Dent et al. used this same RNAi line and found an important role for these isoforms in regulating Hippo signaling in imaginal discs (Dent et al., 2015). However, Ho and Treisman, studying growth of neuromuscular synapses, found something quite different. In this tissue the FHI isoforms play the key role, while all other isoforms act as antagonists of these isoforms (Ho and Treisman, 2020). Our findings suggest that both sets of isoforms have functions in testis shaping, and are consistent with the idea of antagonism, as overexpressing the F isoform led to defects in muscle coverage. It remains somewhat puzzling that our initial dPix RNAi reagent, HMS00741, has the strongest muscle coverage defects. While we think this likely reflects the isoform antagonism issues, we cannot rule out an off-target effect. In the future, it will be important to see whether this phenotype can be rescued by re-expression of specific wild-type dPix isoforms.

We also found a role for the Pix binding partner Git in testis shaping–-this made sense as Git can bind isoform A, which is the major isoform expressed in the tests. Git cannot bind isoforms FHI. Future studies are needed to define how the dPix-Git interaction relates to the differential isoform-functions, and whether this somehow mediates or modulates dPix function as a GEF. Finally, we did two experiments to begin to assess the mechanism by which dPix knockdown alters migration. One hypothesis was that knockdown leads to too much cell-adhesion. We tested this by knocking down both dPix and N-cadherin, hypothesizing that whether we reduced adhesion, we might ameliorate the dPix^RNAi phenotype. N-cadherin knockdown alleviated the effect of dPix^RNAi, consistent with the idea that dPix^RNAi may cause elevated cell-cell adhesion. In parallel, we tested the hypothesis that dPix knockdown would reduce the activity of a Cdc42 sensor. To our surprise, it did not—in fact activity was elevated. Perhaps the dPix isoform targeted is not simply acting as a Cdc42 GEF, even though all of the isoforms contain the GEF domain. It will be interesting to explore how the different isoforms interact to shape the activity of small GTPases.

Sculpting the testis is a multistep process and our screen provides leads into many different aspects of this

While we initially sought to identify regulators of migration and cell adhesion, the results of our screen opened our eyes to other aspects of testis organ morphogenesis. We found numerous candidates whose knockdown allowed full muscle coverage but still caused severe defects in testis morphology. We can now put these into the context of the testis shape changes we characterized that occur after testis nascent myotubes migration (Bischoff and Bogdan, 2023). In that work, we identified distinct phases of morphogenesis (Figure 10A). The first phase is migration (30–36 h APF) during which myotubes migrate from the genital disc to cover the testis. The next phase is myotube elongation, in which the myotubes elongate perpendicular to the proximal-distal axis, beginning to take on the morphology of muscles (45 h APF). The third phase is myotube condensation, in which the elongated myotubes narrow drastically and are linked by filopodial processes (53 h APF). The final phase is proximodistal spreading and bilateral constriction, in which the muscles take on their mature circumferential arrangement and elongate the testis (66 h APF). These steps involve stereotyped changes in cell-cell junctions and the cytoskeleton (Figure 10B).

Based on that analysis, we are now able to hypothesize which specific step or steps of development are likely affected upon knockdown of our positive candidates (Figure 10C). Candidates that cause partial coverage or gaps in the sheet like PlexA and mbt likely affect the migration phase (Class 1), as this step was affected in earlier mutants we examined with coverage defects (Rothenbusch-Fender et al., 2017; Bischoff et al., 2021). The next crucial step in Drosophila testis-morphogenesis is the formation of the distinct and fascinating spiral-shape. Our findings suggest that this process relies on the correct myotube arrangement and alignment after migration. This seems to be self-regulated, as muscle-specific knockdowns that caused irregularities in the parallel organization of muscles also caused a loss of the spiral-shape (Class 3). Furthermore, in Fz2 and Wnt4, we found a promising potential receptor/ligand pair that may be important for this self-regulation process, potentially via a planar cell polarity mechanism. The profound defects in muscle arrangement caused by their knockdown suggest they might regulate the nematic ordering of myotubes. In addition to these mechanisms, mechanical forces likely contribute to alignment, as they have been previously proposed to drive tension driven self-organization of myofibril formation (Lemke and Schnorrer, 2017).

Subsequently, as myotubes differentiate into muscle, they undergo an intriguing condensation and decondensation process, thinning and then re-expanding the myofibrils (Figure 10B). Knockdown of Cdep caused hypercondensed muscle cells, suggesting a role in the decondensation process from 53 to 66 h APF. During the final decondensation process, muscles become wider in the proximodistal axis (spreading), while constricting perpendicular to this axis, causing the testis to become thinner and even longer, thus increasing the number of revolutions in the spiral. Knockdown of multiple candidates, including RhoGef2, pura, and RhoGEF64, caused the entire testis—or parts of it—to become thinner or wider, suggesting a role of these GTPase regulators in this final phase of morphogenesis (Figure 10C). Some candidates—like gig or rsh—cause defects, that cannot be categorized easily, but that cause shape-irregularities without affecting coverage, suggesting important roles during the sculpting process (grouped together in Class 2).

Together, the suite of different defects that emerged from our screen revealed that after migration the mechanical properties of the organ still must be fine-tuned, in an interplay between the muscles and the underlying tissue, to allow normal testis sculpting-morphogenesis. This is reminiscent of the shaping of the Drosophila oocyte by the follicle epithelium, which modifies its ECM to generate a molecular corset, enabling egg cell-elongation (Cetera and Horne-Badovinac, 2015). The process is also reminiscent of the fascinating sculpting of vertebrate airway epithelia by smooth muscle cells (Goodwin et al., 2019; Goodwin et al., 2023), and the many other cases in which smooth muscle cells sculpt other tissues during vertebrate development (Jaslove and Nelson, 2018) In the testis, the different contributions of the mechanical properties of the underlying ECM, the mechanical properties of the muscle cells themselves, and the role of the correct muscle cell arrangement remain to be elucidated, revealing their individual roles in correct sculpting. We are excited to use this system to explore the underlying mechanisms, revealing how simple self-regulated organization can lead to tissue-sculpting via a process with intriguing morphological complexity.

MATERIALS AND METHODS

Request a protocol through Bio-protocol

Testis dissection and dissociation for RNA-seq

We collected ∼100 timed UAS-eGFP/+; Mef2-GAL4/+ prepupae, either at 31 h APF or 45 h APF. We used 6 replicates per condition. We dissected pupal testes in ice cold M3 medium + BPYE (Shield and Sang) with 10% FCS (Thermo Fisher Scientific), and 1x Penicillin/Streptomycin (Life Technologies) and rinsed samples once in ice cold cell dissociation buffer (Thermo Fisher Scientific). We dissociated testes in 500 µl ice cold cell dissociation buffer with 2.5 mg/ml collagenase (Invitrogen) and 4 mg/mg elastase (Worthington Biochemical) in a glass block dish. Then we transferred the mix into a push cap tube and incubated it at 37°C in a water bath. During this step, we pipetted the mix up and down every 5 min, using a p200 pipette, to effectively dissociate testes cells. We prefiltered cells using a 50 µm filter (Partec) and pelleted cells by centrifuging them 5 min at 700 × g. We replaced the buffer with 1 ml ice cold FACS buffer (1x PBS, 2% BSA, 2.5 mM EDTA) and pelleted cells again by centrifuging them 5 min at 700 × g. We replaced the buffer with 500 µl fresh ice cold FACS buffer. Then we added propidium iodide to a final concentration of 1 µg/ml and DyeCycle Violet to a final concentration of 5 µM. We incubated the samples 12 min at 37°C in a water bath and FACS sorted them after that.

FACS sorting and RNA isolation

FACS sorting was performed at the UNC Flow Cytometry Core Facility on a BD FACS Aria III. We sorted living (PI-negative) testis myotubes (GFP) that are bi- and tri-nucleated (DyeCycle Violet, distinct populations, see Figure 2). We sorted cells directly into 100 µl TRIzol LS and performed standard TRIzol RNA isolation by first adding 100 µl standard TRIzol and incubating the mix 15 min at room temperature. We added 75 µl Chloroform, vortexed the mix for 30 s and incubated it for 3 min at room temperature. We centrifuged the mix 20 min at 4°C and 12,000 × g. We transferred the aqueous phase to a fresh tube and added 180 µl Isopropanol and 1 µl Glycoblue. Then we vortexed the mix 15 s and precipitated the RNA over night at −80°C. To pellet the RNA, we centrifuged the mix at 4°C with 12,000 × g and washed the pellet twice with 75% ethanol. We dried the pellet for 5 min, to then resuspended it in 20 µl RNA-seq free water. RNA-seq was performed at the UNC Advanced Analytics Core.

Bioinformatic analysis

Quality control of the reads was performed with FastQC (v 0.12.1). All runs passed metrics, with no poor-quality reads. The Drosophila reference genome was downloaded from Flybase (dmel-all-aligned-r6.5, including the genome annotation file Drosophila_melanogaster.BDGP6.46.111.gtf). BBMap (v39.08) with default parameters was used to align the fastq files to the reference genome, which were then processed with Samtools (v1.21) to generate aligned .bam files. Read counts for gene models were determined using FeatureCounts (in the Subread v2.0.6) to produce a raw counts table from the .bam files. These counts were then used normalized by reads-per-kilobase-million (RPKM). After normalization, RPKM estimates were used for visualizations and raw counts for downstream differential expression analysis using DESeq2 (v1.40.2).

The raw counts table was read into R (v4.3.1 including packages: data.table 1.15.4, dplyr 1.1.4, reshape2 1.4.4), and the non-protein-coding genes were filtered out. All further analysis was performed using R. DESeq2 compared the expression of the 6 replicate cell sets taken at 31 h to 6 replicate cell sets taken at 45 h. The model was y ∼ timepoint + e. The results were visualized in a principal component analysis plot using the ggplot2 graphing suite (ggplot2 3.5.1). DESeq2 uses a generalized linear modeling approach for significance testing after correcting for overdispersion of read counts using a dispersion estimation procedure based on a negative binomial distribution. Our model simply compared 31 h versus 45 h and estimated the log2 fold change in expression and used this to perform a Wald test (null model that there is no change between the timepoints). Nominal p-values were then adjusted for multiple testing. Gene tables for several classes of functionality were cross-referenced with the results to produce boxplots illustrating the differential expression levels of the key genes. Three-dimensional volcano plots were produced using the Glimma package (v2.12.0) and a dynamic instantiation is hosted here: https://rpubs.com/Cserody/bischoff_volcano_symbols. An FGSEA was performed using the FGSEA package for R (fgsea1.28.0, also EnrichmentBrowser 2.32.0, ggrepel 0.9.5, and dynamicTreeCut 1.63.1), and its native writeGMT and getGenesets functions were used to generate the gmt and genesets files. For FGSEA, we limited to GO terms with over a 10 and under 1000 members.

Fly genetics

We performed crosses at 25°C. Prepupae for subsequent live-cell imaging developed at 26.5°C in a cell incubator. RNAi stocks used in the screen can be found in Supplemental Table S5. For control crosses, we used w¹¹¹⁸ and referred to it as wild type or WT.

Testis staining and microscopy

For the screen, testes from adult flies were dissected 1–3 d posthatching in 1.5x PBS and fixed 20 min in 4% PFA in PBS. Subsequently, we washed them three times in 1.5x PBS and once in 1.5x PBS with 0.1% Tween 20. Overnight phalloidin staining was performed using 1:500 Phalloidin (Thermo Fisher Scientific, Alexa Fluor 488) in 1.5x PBS with 0.1% Tween 20 at 4°C. After washing samples three times using 1.5x PBS, we mounted them in PBS in 35 mm glass bottom dishes. For imaging, we used an LSM Pascal (Zeiss) with a 10x (Zeiss EC Plan-Neofluar 10x/0.3) dry objective. All images of adult testes are projections of large z-stacks using maximum intensity. We modified brightness and contrast in Photoshop (Adobe) and Fiji (ImageJ) to make all structures visible.

Live-cell imaging

We dissected testes from 31 h APF testes in M3 Medium (Shields and Sang, Sigma-Aldrich) with 10% FCS (Thermo Fisher Scientific), and 1x penicillin/streptomycin (Life Technologies) at room temperature (Bischoff and Bogdan, 2023). For mounting them live, we used 0.5% low gelling agarose (Sigma-Aldrich) in M3 Medium. For image acquisition, we used a Nikon Ti2 inverted microscope with a Yokogawa CSU-W1 spinning disc and Hamamatsu ORCA-fusion BT sCMOS camera and a 25x/1.05 Silicone Apochromat objective. Using Fiji (ImageJ), we changed Gamma to 0.5, to make all parts of the tissue visible without oversaturating other parts of the image. We modified brightness and contrast in Photoshop (Adobe) to make all structures visible.

Cdc42/Rac activity sensor quantification

To quantify Cdc42/Rac activity, we used a characterized sensor that has the CRIB domain of the Drosophila Pak2 gene mbt fused to GFP and under control of a UAS promoter (Rötte et al., 2024). To quantify its effects, we expressed it in testis myotubes using the Mef2-GAL4 driver. As it is naturally strongly enriched at cell-cell borders, we used a line scan method to quantify it in wild type or after expressing RNAi constructs. To do so, we wrote a Fiji script. The user must mark cell-cell borders and the script generates a line of defined length perpendicular to that line, with the center exactly on top of the cell-cell border so that multiple profile measurements can later be overlayed. We averaged the measurements within each sample and then normalized them by setting the minimum of the averaged values to 0. To simply show the fluorescence intensity at the membrane in one dimension, we used Fiji to create lines on top of cell-cell contacts and measured the averaged fluorescence intensity.

Detailed author contributions

Maik Bischoff conceived the project, prepared the samples for RNA-seq, analyzed the outcome to determine which genes to include in the screen, directed the team who carried out the screen, and placed the genes in phenotypic categories. Jenevieve Norton carried out much of the screen, organized the data, and trained undergraduates who participated in the screen, Erika Munguia also made substantial contributions to the screen, assisted by the team of Rebecca Korankye, Emmanuel Addai Gyabaah, and Taino Encarnacion. Sarah Clark carried out several experiments regarding dPix. Noah Gurley helped prepare samples for RNA-seq. Christopher Serody and Corbin Jones did the initial bioinformatics analysis, In collaboration with Maik Bischoff. Maik Bischoff, Corbin Jones and Mark Peifer wrote the paper with editorial contributions from the other authors.

Supplementary Material

mbc-36-ar21-s001.pdf^{(603KB, pdf)}

mbc-36-ar21-s002.xlsx^{(1.1MB, xlsx)}

mbc-36-ar21-s003.xlsx^{(1MB, xlsx)}

mbc-36-ar21-s004.xlsx^{(89.6KB, xlsx)}

mbc-36-ar21-s005.xlsx^{(2MB, xlsx)}

mbc-36-ar21-s006.xlsx^{(26.9KB, xlsx)}

mbc-36-ar21-s007.xlsx^{(11KB, xlsx)}

ACKNOWLEDGMENTS

We are very grateful to the Bloomington and Vienna Drosophila Stock Centers for many shipments of RNAi lines, to Jessica Treisman and Sally Horne-Badovinac for Drosophila stocks and helpful discussions, to Nat Prunet and the Biology Imaging Core, Roman Bandy of the UNC Flow Cytometry Core and Gabrielle Cannon of the UNC Advanced Analytics Core for technical support, the Peifer lab and the two anonymous reviewers for helpful feedback on the manuscript, and the Peifer, Bergstralh/Finegan and Williams labs for feedback throughout. M.C.B was supported by the DFG Walter Benjamin Programme (ref. GZ: BI 2384/1-1) and work in the Peifer lab is supported by NIH R35 GM118096.

Abbreviations used:

APF: After pupal formation
CRIB: Cdc42- and Rac-interactive binding
DH: Dbl homology
ECM: extracellular matrix
FACS: Fluorescence-activated cell sorting
FGSEA: Fast gene set enrichment analysis
FPKM: Fragments Per Kilobase of transcript per Million mapped reads
GAP: GTPase-activating protein
GEF: Guanine nucleotide exchange factor
GO: gene ontology
NLS: nuclear localization signal
PH: Pleckstrin homology
PI: propidium iodide
RNAi: RNA interference
RNA-seq: RNA sequencing
RTK: receptor tyrosine kinase
shRNA: short hairpin RNA
TNM: testis nascent myotubes.

Footnotes

This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E24-10-0456) on January 2, 2025.

REFERENCES

Aberle H (2019). Axon guidance and collective cell migration by substrate-derived attractants. Front Mol Neurosci 12, 148. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bahri SM, Choy JM, Manser E, Lim L, Yang X (2009). The Drosophila homologue of Arf-GAP GIT1, dGIT, is required for proper muscle morphogenesis and guidance during embryogenesis. Dev Biol 325, 15–23. [DOI] [PubMed] [Google Scholar]
Bischoff MC, Bogdan S (2021). Collective cell migration driven by filopodia-new insights from the social behavior of myotubes. Bioessays 43, e2100124. [DOI] [PubMed] [Google Scholar]
Bischoff MC, Bogdan S (2023). Dissecting collective cell behavior in migrating testis myotubes in Drosophila. Methods Mol Biol 2608, 117–129. [DOI] [PubMed] [Google Scholar]
Bischoff MC, Lieb S, Renkawitz-Pohl R, Bogdan S (2021). Filopodia-based contact stimulation of cell migration drives tissue morphogenesis. Nat Commun 12, 791. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bussmann J, Raz E (2015). Chemokine-guided cell migration and motility in zebrafish development. EMBO J 34, 1309–1318. [DOI] [PMC free article] [PubMed] [Google Scholar]
Campanale JP, Montell DJ (2023). Who's really in charge: Diverse follower cell behaviors in collective cell migration. Curr Opin Cell Biol 81, 102160. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cetera M, Horne-Badovinac S (2015). Round and round gets you somewhere: Collective cell migration and planar polarity in elongating Drosophila egg chambers. Curr Opin Genet Dev 32, 10–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cortes E, Pak JS, Ozkan E (2023). Structure and evolution of neuronal wiring receptors and ligands. Dev Dyn 252, 27–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dent LG, Manning SA, Kroeger B, Williams AM, Saiful Hilmi AJ, Crea L, Kondo S, Horne-Badovinac S, Harvey KF (2019). The dPix-Git complex is essential to coordinate epithelial morphogenesis and regulate myosin during Drosophila egg chamber development. PLoS Genet 15, e1008083. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dent LG, Poon CL, Zhang X, Degoutin JL, Tipping M, Veraksa A, Harvey KF (2015). The GTPase regulatory proteins Pix and Git control tissue growth via the hippo pathway. Curr Biol 25, 124–130. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dietzl G, Chen D, Schnorrer F, Su KC, Barinova Y, Fellner M, Gasser B, Kinsey K, Oppel S, Scheiblauer S, et al. (2007). A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature 448, 151–156. [DOI] [PubMed] [Google Scholar]
Duan H, Skeath JB, Nguyen HT (2001). Drosophila lame duck, a novel member of the Gli superfamily, acts as a key regulator of myogenesis by controlling fusion-competent myoblast development. Development 128, 4489–4500. [DOI] [PubMed] [Google Scholar]
Fujimori T, Nakajima A, Shimada N, Sawai S (2019). Tissue self-organization based on collective cell migration by contact activation of locomotion and chemotaxis. Proc Natl Acad Sci U S A 116, 4291–4296. [DOI] [PMC free article] [PubMed] [Google Scholar]
Goodwin K, Lemma B, Zhang P, Boukind A, Nelson CM (2023). Plasticity in airway smooth muscle differentiation during mouse lung development. Dev Cell 58, 338–347.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
Goodwin K, Mao S, Guyomar T, Miller E, Radisky DC, Kosmrlj A, Nelson CM (2019). Smooth muscle differentiation shapes domain branches during mouse lung development. Development 146, dev181172. [DOI] [PMC free article] [PubMed] [Google Scholar]
Han P, Liu J, Lei Y, Lin Z, Tian D, Yan W (2019). Netrin-1 promotes the collective cell migration of liver cancer cells in a 3D cell culture model. J Physiol Biochem 75, 489–498. [DOI] [PubMed] [Google Scholar]
Ho CH, Treisman JE (2020). Specific isoforms of the guanine-nucleotide exchange factor dPix couple neuromuscular synapse growth to muscle growth. Dev Cell 54, 117–131.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Inaki M, Yoshikawa S, Thomas JB, Aburatani H, Nose A (2007). Wnt4 is a local repulsive cue that determines synaptic target specificity. Curr Biol 17, 1574–1579. [DOI] [PubMed] [Google Scholar]
Ismat A, Schaub C, Reim I, Kirchner K, Schultheis D, Frasch M (2010). HLH54F is required for the specification and migration of longitudinal gut muscle founders from the caudal mesoderm of Drosophila . Development 137, 3107–3117. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jaslove JM, Nelson CM (2018). Smooth muscle: A stiff sculptor of epithelial shapes. Philos Trans R Soc Lond B Biol Sci 373, 20170318. [DOI] [PMC free article] [PubMed] [Google Scholar]
Junqueira Alves C, Silva Ladeira J, Hannah T, Pedroso Dias RJ, Zabala Capriles PV, Yotoko K, Zou H, Friedel RH (2021). Evolution and diversity of semaphorins and plexins in choanoflagellates. Genome Biol Evol 13, evab035. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kozopas KM, Samos CH, Nusse R (1998). DWnt-2, a Drosophila Wnt gene required for the development of the male reproductive tract, specifies a sexually dimorphic cell fate. Genes Dev 12, 1155–1165. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lemke SB, Schnorrer F (2017). Mechanical forces during muscle development. Mech Dev 144, 92–101. [DOI] [PubMed] [Google Scholar]
Menon SD, Chia W (2001). Drosophila rolling pebbles: A multidomain protein required for myoblast fusion that recruits D-Titin in response to the myoblast attractant dumbfounded. Dev Cell 1, 691–703. [DOI] [PubMed] [Google Scholar]
Mishra AK, Campanale JP, Mondo JA, Montell DJ (2019). Cell interactions in collective cell migration. Development 146, dev172056. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ni JQ, Zhou R, Czech B, Liu LP, Holderbaum L, Yang-Zhou D, Shim HS, Tao R, Handler D, Karpowicz P, et al. (2011). A genome-scale shRNA resource for transgenic RNAi in Drosophila. Nat Methods 8, 405–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
Omelchenko T, Hall A, Anderson KV (2020). β-Pix-dependent cellular protrusions propel collective mesoderm migration in the mouse embryo. Nat Commun 11, 6066. [DOI] [PMC free article] [PubMed] [Google Scholar]
Parnas D, Haghighi AP, Fetter RD, Kim SW, Goodman CS (2001). Regulation of postsynaptic structure and protein localization by the Rho-type guanine nucleotide exchange factor dPix. Neuron 32, 415–424. [DOI] [PubMed] [Google Scholar]
Plutoni C, Bazellieres E, Le Borgne-Rochet M, Comunale F, Brugues A, Seveno M, Planchon D, Thuault S, Morin N, Bodin S, et al. (2016). P-cadherin promotes collective cell migration via a Cdc42-mediated increase in mechanical forces. J Cell Biol 212, 199–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
Podufall J, Tian R, Knoche E, Puchkov D, Walter AM, Rosa S, Quentin C, Vukoja A, Jung N, Lampe A, et al. (2014). A presynaptic role for the cytomatrix protein GIT in synaptic vesicle recycling. Cell Rep 7, 1417–1425. [DOI] [PubMed] [Google Scholar]
Raza Q, Jacobs JR (2016). Guidance signaling regulates leading edge behavior during collective cell migration of cardiac cells in Drosophila. Dev Biol 419, 285–297. [DOI] [PubMed] [Google Scholar]
Rothenbusch-Fender S, Fritzen K, Bischoff MC, Buttgereit D, Oenel SF, Renkawitz-Pohl R (2017). Myotube migration to cover and shape the testis of Drosophila depends on heartless, cadherin/catenin, and myosin II. Biol Open 6, 1876–1888. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rötte M, Höhne MY, Klug D, Ramlow K, Zedler C, Lehne F, Schneider M, Bischoff MC, Bogdan S (2024). CYRI controls epidermal wound closure and cohesion of invasive border cell cluster in Drosophila. J Cell Biol 223, e202310153. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rust K, Byrnes LE, Yu KS, Park JS, Sneddon JB, Tward AD, Nystul TG (2020). A single-cell atlas and lineage analysis of the adult Drosophila ovary. Nat Commun 11, 5628. [DOI] [PMC free article] [PubMed] [Google Scholar]
Scarpa E, Mayor R (2016). Collective cell migration in development. J Cell Biol 212, 143–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schneeberger D, Raabe T (2003). Mbt, a Drosophila PAK protein, combines with Cdc42 to regulate photoreceptor cell morphogenesis. Development 130, 427–437. [DOI] [PubMed] [Google Scholar]
Schnorrer F, Dickson BJ (2004). Muscle building; mechanisms of myotube guidance and attachment site selection. Dev Cell 7, 9–20. [DOI] [PubMed] [Google Scholar]
Schnorrer F, Schonbauer C, Langer CCH, Dietzl G, Novatchkova M, Schernhuber K, Fellner M, Azaryan A, Radolf M, Stark A, et al. (2010). Systematic genetic analysis of muscle morphogenesis and function in Drosophila. Nature 464, 287–291. [DOI] [PubMed] [Google Scholar]
Stedden CG, Menegas W, Zajac AL, Williams AM, Cheng S, Ozkan E, Horne-Badovinac S (2019). Planar-polarized semaphorin-5c and plexin A promote the collective migration of epithelial cells in Drosophila. Curr Biol 29, 908–920.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stern C (1941). The growth of testes in Drosophila. I. The relation between vas deferens and testis within various species. J Exp Zool 87, 113–158. [Google Scholar]
Stramer B, Mayor R (2017). Mechanisms and in vivo functions of contact inhibition of locomotion. Nat Rev Mol Cell Biol 18, 43–55. [DOI] [PubMed] [Google Scholar]
Susic-Jung L, Hornbruch-Freitag C, Kuckwa J, Rexer KH, Lammel U, Renkawitz-Pohl R (2012). Multinucleated smooth muscles and mononucleated as well as multinucleated striated muscles develop during establishment of the male reproductive organs of Drosophila melanogaster. Dev Biol 370, 86–97. [DOI] [PubMed] [Google Scholar]
Zhou W, Li X, Premont RT (2016). Expanding functions of GIT Arf GTPase-activating proteins, PIX Rho guanine nucleotide exchange factors and GIT-PIX complexes. J Cell Sci 129, 1963–1974. [DOI] [PMC free article] [PubMed] [Google Scholar]

Mol Biol Cell.

Reviewer Report

reviewer_reportE24-10-0456.pdf^{(922KB, pdf)}

Open in a new tab

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mbc-36-ar21-s001.pdf^{(603KB, pdf)}

mbc-36-ar21-s002.xlsx^{(1.1MB, xlsx)}

mbc-36-ar21-s003.xlsx^{(1MB, xlsx)}

mbc-36-ar21-s004.xlsx^{(89.6KB, xlsx)}

mbc-36-ar21-s005.xlsx^{(2MB, xlsx)}

mbc-36-ar21-s006.xlsx^{(26.9KB, xlsx)}

mbc-36-ar21-s007.xlsx^{(11KB, xlsx)}

[B1] Aberle H (2019). Axon guidance and collective cell migration by substrate-derived attractants. Front Mol Neurosci 12, 148. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B2] Bahri SM, Choy JM, Manser E, Lim L, Yang X (2009). The Drosophila homologue of Arf-GAP GIT1, dGIT, is required for proper muscle morphogenesis and guidance during embryogenesis. Dev Biol 325, 15–23. [DOI] [PubMed] [Google Scholar]

[B3] Bischoff MC, Bogdan S (2021). Collective cell migration driven by filopodia-new insights from the social behavior of myotubes. Bioessays 43, e2100124. [DOI] [PubMed] [Google Scholar]

[B4] Bischoff MC, Bogdan S (2023). Dissecting collective cell behavior in migrating testis myotubes in Drosophila. Methods Mol Biol 2608, 117–129. [DOI] [PubMed] [Google Scholar]

[B5] Bischoff MC, Lieb S, Renkawitz-Pohl R, Bogdan S (2021). Filopodia-based contact stimulation of cell migration drives tissue morphogenesis. Nat Commun 12, 791. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B6] Bussmann J, Raz E (2015). Chemokine-guided cell migration and motility in zebrafish development. EMBO J 34, 1309–1318. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B7] Campanale JP, Montell DJ (2023). Who's really in charge: Diverse follower cell behaviors in collective cell migration. Curr Opin Cell Biol 81, 102160. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B8] Cetera M, Horne-Badovinac S (2015). Round and round gets you somewhere: Collective cell migration and planar polarity in elongating Drosophila egg chambers. Curr Opin Genet Dev 32, 10–15. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B9] Cortes E, Pak JS, Ozkan E (2023). Structure and evolution of neuronal wiring receptors and ligands. Dev Dyn 252, 27–60. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B10] Dent LG, Manning SA, Kroeger B, Williams AM, Saiful Hilmi AJ, Crea L, Kondo S, Horne-Badovinac S, Harvey KF (2019). The dPix-Git complex is essential to coordinate epithelial morphogenesis and regulate myosin during Drosophila egg chamber development. PLoS Genet 15, e1008083. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B11] Dent LG, Poon CL, Zhang X, Degoutin JL, Tipping M, Veraksa A, Harvey KF (2015). The GTPase regulatory proteins Pix and Git control tissue growth via the hippo pathway. Curr Biol 25, 124–130. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B12] Dietzl G, Chen D, Schnorrer F, Su KC, Barinova Y, Fellner M, Gasser B, Kinsey K, Oppel S, Scheiblauer S, et al. (2007). A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature 448, 151–156. [DOI] [PubMed] [Google Scholar]

[B13] Duan H, Skeath JB, Nguyen HT (2001). Drosophila lame duck, a novel member of the Gli superfamily, acts as a key regulator of myogenesis by controlling fusion-competent myoblast development. Development 128, 4489–4500. [DOI] [PubMed] [Google Scholar]

[B14] Fujimori T, Nakajima A, Shimada N, Sawai S (2019). Tissue self-organization based on collective cell migration by contact activation of locomotion and chemotaxis. Proc Natl Acad Sci U S A 116, 4291–4296. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B15] Goodwin K, Lemma B, Zhang P, Boukind A, Nelson CM (2023). Plasticity in airway smooth muscle differentiation during mouse lung development. Dev Cell 58, 338–347.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B16] Goodwin K, Mao S, Guyomar T, Miller E, Radisky DC, Kosmrlj A, Nelson CM (2019). Smooth muscle differentiation shapes domain branches during mouse lung development. Development 146, dev181172. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B17] Han P, Liu J, Lei Y, Lin Z, Tian D, Yan W (2019). Netrin-1 promotes the collective cell migration of liver cancer cells in a 3D cell culture model. J Physiol Biochem 75, 489–498. [DOI] [PubMed] [Google Scholar]

[B18] Ho CH, Treisman JE (2020). Specific isoforms of the guanine-nucleotide exchange factor dPix couple neuromuscular synapse growth to muscle growth. Dev Cell 54, 117–131.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B19] Inaki M, Yoshikawa S, Thomas JB, Aburatani H, Nose A (2007). Wnt4 is a local repulsive cue that determines synaptic target specificity. Curr Biol 17, 1574–1579. [DOI] [PubMed] [Google Scholar]

[B20] Ismat A, Schaub C, Reim I, Kirchner K, Schultheis D, Frasch M (2010). HLH54F is required for the specification and migration of longitudinal gut muscle founders from the caudal mesoderm of Drosophila . Development 137, 3107–3117. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B21] Jaslove JM, Nelson CM (2018). Smooth muscle: A stiff sculptor of epithelial shapes. Philos Trans R Soc Lond B Biol Sci 373, 20170318. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B22] Junqueira Alves C, Silva Ladeira J, Hannah T, Pedroso Dias RJ, Zabala Capriles PV, Yotoko K, Zou H, Friedel RH (2021). Evolution and diversity of semaphorins and plexins in choanoflagellates. Genome Biol Evol 13, evab035. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B23] Kozopas KM, Samos CH, Nusse R (1998). DWnt-2, a Drosophila Wnt gene required for the development of the male reproductive tract, specifies a sexually dimorphic cell fate. Genes Dev 12, 1155–1165. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B24] Lemke SB, Schnorrer F (2017). Mechanical forces during muscle development. Mech Dev 144, 92–101. [DOI] [PubMed] [Google Scholar]

[B25] Menon SD, Chia W (2001). Drosophila rolling pebbles: A multidomain protein required for myoblast fusion that recruits D-Titin in response to the myoblast attractant dumbfounded. Dev Cell 1, 691–703. [DOI] [PubMed] [Google Scholar]

[B26] Mishra AK, Campanale JP, Mondo JA, Montell DJ (2019). Cell interactions in collective cell migration. Development 146, dev172056. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B27] Ni JQ, Zhou R, Czech B, Liu LP, Holderbaum L, Yang-Zhou D, Shim HS, Tao R, Handler D, Karpowicz P, et al. (2011). A genome-scale shRNA resource for transgenic RNAi in Drosophila. Nat Methods 8, 405–407. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B28] Omelchenko T, Hall A, Anderson KV (2020). β-Pix-dependent cellular protrusions propel collective mesoderm migration in the mouse embryo. Nat Commun 11, 6066. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B29] Parnas D, Haghighi AP, Fetter RD, Kim SW, Goodman CS (2001). Regulation of postsynaptic structure and protein localization by the Rho-type guanine nucleotide exchange factor dPix. Neuron 32, 415–424. [DOI] [PubMed] [Google Scholar]

[B30] Plutoni C, Bazellieres E, Le Borgne-Rochet M, Comunale F, Brugues A, Seveno M, Planchon D, Thuault S, Morin N, Bodin S, et al. (2016). P-cadherin promotes collective cell migration via a Cdc42-mediated increase in mechanical forces. J Cell Biol 212, 199–217. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B31] Podufall J, Tian R, Knoche E, Puchkov D, Walter AM, Rosa S, Quentin C, Vukoja A, Jung N, Lampe A, et al. (2014). A presynaptic role for the cytomatrix protein GIT in synaptic vesicle recycling. Cell Rep 7, 1417–1425. [DOI] [PubMed] [Google Scholar]

[B32] Raza Q, Jacobs JR (2016). Guidance signaling regulates leading edge behavior during collective cell migration of cardiac cells in Drosophila. Dev Biol 419, 285–297. [DOI] [PubMed] [Google Scholar]

[B33] Rothenbusch-Fender S, Fritzen K, Bischoff MC, Buttgereit D, Oenel SF, Renkawitz-Pohl R (2017). Myotube migration to cover and shape the testis of Drosophila depends on heartless, cadherin/catenin, and myosin II. Biol Open 6, 1876–1888. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B34] Rötte M, Höhne MY, Klug D, Ramlow K, Zedler C, Lehne F, Schneider M, Bischoff MC, Bogdan S (2024). CYRI controls epidermal wound closure and cohesion of invasive border cell cluster in Drosophila. J Cell Biol 223, e202310153. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B35] Rust K, Byrnes LE, Yu KS, Park JS, Sneddon JB, Tward AD, Nystul TG (2020). A single-cell atlas and lineage analysis of the adult Drosophila ovary. Nat Commun 11, 5628. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B36] Scarpa E, Mayor R (2016). Collective cell migration in development. J Cell Biol 212, 143–155. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B37] Schneeberger D, Raabe T (2003). Mbt, a Drosophila PAK protein, combines with Cdc42 to regulate photoreceptor cell morphogenesis. Development 130, 427–437. [DOI] [PubMed] [Google Scholar]

[B38] Schnorrer F, Dickson BJ (2004). Muscle building; mechanisms of myotube guidance and attachment site selection. Dev Cell 7, 9–20. [DOI] [PubMed] [Google Scholar]

[B39] Schnorrer F, Schonbauer C, Langer CCH, Dietzl G, Novatchkova M, Schernhuber K, Fellner M, Azaryan A, Radolf M, Stark A, et al. (2010). Systematic genetic analysis of muscle morphogenesis and function in Drosophila. Nature 464, 287–291. [DOI] [PubMed] [Google Scholar]

[B40] Stedden CG, Menegas W, Zajac AL, Williams AM, Cheng S, Ozkan E, Horne-Badovinac S (2019). Planar-polarized semaphorin-5c and plexin A promote the collective migration of epithelial cells in Drosophila. Curr Biol 29, 908–920.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[B41] Stern C (1941). The growth of testes in Drosophila. I. The relation between vas deferens and testis within various species. J Exp Zool 87, 113–158. [Google Scholar]

[B42] Stramer B, Mayor R (2017). Mechanisms and in vivo functions of contact inhibition of locomotion. Nat Rev Mol Cell Biol 18, 43–55. [DOI] [PubMed] [Google Scholar]

[B43] Susic-Jung L, Hornbruch-Freitag C, Kuckwa J, Rexer KH, Lammel U, Renkawitz-Pohl R (2012). Multinucleated smooth muscles and mononucleated as well as multinucleated striated muscles develop during establishment of the male reproductive organs of Drosophila melanogaster. Dev Biol 370, 86–97. [DOI] [PubMed] [Google Scholar]

[B44] Zhou W, Li X, Premont RT (2016). Expanding functions of GIT Arf GTPase-activating proteins, PIX Rho guanine nucleotide exchange factors and GIT-PIX complexes. J Cell Sci 129, 1963–1974. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

A large reverse-genetic screen identifies numerous regulators of testis nascent myotube collective cell migration and collective organ sculpting

Maik C Bischoff

Jenevieve E Norton

Erika A Munguia

Sarah E Clark

Noah J Gurley

Rebecca Korankye

Emmanuel Addai Gyabaah

Taino Encarnacion

Christopher J Serody

Corbin D Jones

Mark Peifer

Roles

Series information

Abstract

INTRODUCTION

FIGURE 1:

RESULTS

Identifying genes expressing in migratory-stage myotubes using RNA-seq

FIGURE 2:

FIGURE 3:

Selecting candidate genes for our RNAi screen

FIGURE 4:

The diverse phenotypes observed in our screen reveal the complexity of events that shape this organ

A subset of the RNAi lines alters muscle coverage, suggesting defects in cell migration or cell adhesion

FIGURE 5:

Other RNAi lines altered testis morphogenesis in more complex ways, potentially affecting testis shaping during later morphogenesis

FIGURE 6:

A final class of RNAi lines affected muscle alignment

dPix RNAi leads to a very strong delay in cell migration

FIGURE 7:

Knocking down different dPix isoforms has different effects on testis shaping

FIGURE 8:

The dPix binding partner Git plays an important role in testis shaping

dPix RNAi elevates rather than decreases the activity of a Cdc42 sensor

FIGURE 9:

DISCUSSION

FIGURE 10:

Intriguing parallels between collective cell migration and axon guidance

The regulation of Rho family and Ras/Rap GTPases plays important roles in many aspects of testis morphogenesis

Sculpting the testis is a multistep process and our screen provides leads into many different aspects of this

MATERIALS AND METHODS

Testis dissection and dissociation for RNA-seq

FACS sorting and RNA isolation

Bioinformatic analysis

Fly genetics

Testis staining and microscopy

Live-cell imaging

Cdc42/Rac activity sensor quantification

Detailed author contributions

Supplementary Material

ACKNOWLEDGMENTS

Abbreviations used:

Footnotes

REFERENCES

Reviewer Report

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases