1. Developmental Biology
Download icon

Antagonism between germ cell-less and Torso receptor regulates transcriptional quiescence underlying germline/soma distinction

  1. Megan M Colonnetta
  2. Lauren R Lym
  3. Lillian Wilkins
  4. Gretchen Kappes
  5. Elias A Castro
  6. Pearl V Ryder
  7. Paul Schedl
  8. Dorothy A Lerit  Is a corresponding author
  9. Girish Deshpande  Is a corresponding author
  1. Department of Molecular Biology, Princeton University, United States
  2. Department of Cell Biology, Emory University School of Medicine, United States
Research Article
  • Cited 0
  • Views 538
  • Annotations
Cite this article as: eLife 2021;10:e54346 doi: 10.7554/eLife.54346

Abstract

Transcriptional quiescence, an evolutionarily conserved trait, distinguishes the embryonic primordial germ cells (PGCs) from their somatic neighbors. In Drosophila melanogaster, PGCs from embryos maternally compromised for germ cell-less (gcl) misexpress somatic genes, possibly resulting in PGC loss. Recent studies documented a requirement for Gcl during proteolytic degradation of the terminal patterning determinant, Torso receptor. Here we demonstrate that the somatic determinant of female fate, Sex-lethal (Sxl), is a biologically relevant transcriptional target of Gcl. Underscoring the significance of transcriptional silencing mediated by Gcl, ectopic expression of a degradation-resistant form of Torso (torsoDeg) can activate Sxl transcription in PGCs, whereas simultaneous loss of torso-like (tsl) reinstates the quiescent status of gcl PGCs. Intriguingly, like gcl mutants, embryos derived from mothers expressing torsoDeg in the germline display aberrant spreading of pole plasm RNAs, suggesting that mutual antagonism between Gcl and Torso ensures the controlled release of germ-plasm underlying the germline/soma distinction.

Introduction

Following fertilization, a Drosophila embryo undergoes 14 consecutive nuclear divisions to give rise to the cellular blastoderm. While the initial nuclear divisions take place in the center of the embryo, the nuclei begin to migrate toward the periphery around nuclear cycle (NC) 4–6 and reach the cortex at NC9/10 (Farrell and O'Farrell, 2014). Even before bulk nuclear migration commences, a few nuclei move toward the posterior of the embryo, enter a specialized, maternally derived cytoplasm known as the pole plasm, and induce the formation of pole buds (PBs) (Williamson and Lehmann, 1996; Wilson and Macdonald, 1993; Wylie, 1999). The centrosomes associated with these nuclei trigger the release of pole plasm constituents from the posterior cortex and orchestrate precocious cellularization to form the primordial germ cells (PGCs), the progenitors of the germline stem cells in adult gonads (Lerit and Gavis, 2011; Raff and Glover, 1989). Unlike pole cell nuclei, somatic nuclei continue synchronous divisions after they reach the surface of the embryo until NC 14 when they cellularize (Blythe and Wieschaus, 2015).

The timing of cellularization is not the only difference between the soma and PGCs. Although newly formed PGCs divide after they are formed, they undergo only one or two asynchronous divisions before exiting the cell cycle. Another key difference is in transcriptional activity (Nakamura and Seydoux, 2008). Transcription commences in the embryo during NC 6–7 when a select number of genes are active (Ali-Murthy et al., 2013). Transcription is more globally upregulated when the nuclei reach the surface, and by the end of NC 14, zygotic genome activation (ZGA) is complete (Ali-Murthy et al., 2013; Harrison and Eisen, 2015). This transition is marked by high levels of phosphorylation of residues Serine 5 (Ser5) and Serine 2 (Ser2) in the C-terminal domain (CTD) of RNA polymerase II (Schaner et al., 2003; Seydoux and Dunn, 1997). By contrast, in newly formed PGCs, transcription is switched off, and PGC nuclei have only residual amounts of Ser5 and Ser2 CTD phosphorylation (Deshpande, 2004; Martinho et al., 2004; Seydoux and Dunn, 1997). Moreover, and consistent with their transcriptionally quiescent status, other changes in chromatin architecture that accompany ZGA are also blocked in PGCs (Schaner et al., 2003).

Three different genes, nanos (nos), polar granule component (pgc), and germ cell-less (gcl), are known to be required for establishing transcriptional quiescence in newly formed PGCs (Deshpande et al., 2005; Deshpande, 2004; Deshpande et al., 1999; Hanyu-Nakamura et al., 2008; Kobayashi et al., 1996; Leatherman et al., 2002; Martinho et al., 2004). The PGCs in embryos derived from mothers carrying mutations in these genes fail to inhibit transcription, and this compromises germ cell specification and disrupts germ cell migration. (As these are maternal effect genes, embryos derived from nos/pgc/gcl mothers display the resulting mutant phenotypes and will be referred to as nos/pgc/gcl here onwards.) Interestingly, these three genes share only a few targets, suggesting overlapping yet distinct mechanisms of action. Nos is a translation factor and thus must block transcription indirectly. Together with the RNA-binding protein Pumilio (Pum), Nos interacts with recognition sequences in the 3’-untranslated regions (3’UTRs) of mRNAs and inhibits their translation (Asaoka et al., 2019; Sonoda and Wharton, 1999; Wharton and Struhl, 1991). Currently, the key mRNA target(s) that Nos-Pum repress to block transcription is unknown; however, in nos and pum mutants, PGC nuclei display high levels of Ser5 and Ser2 CTD phosphorylation and activate transcription of gap and pair-rule patterning genes and the sex determination gene Sex-lethal (Sxl) (Deshpande et al., 2005; Deshpande et al., 1999). pgc encodes a nuclear protein that binds to the transcriptional elongation kinase p-TEFb, blocking Ser5 CTD phosphorylation (Hanyu-Nakamura et al., 2008). In pgc mutant pole cells, Ser5 phosphorylation is enhanced, as is transcription of several somatic genes, including genes involved in terminal patterning (Deshpande, 2004; Martinho et al., 2004).

While the primary function of nos and pgc appears to be blocking ZGA in PGCs, gcl has an earlier function, which is to turn off transcription of genes activated in somatic nuclei prior to nuclear migration (Leatherman et al., 2002). Targets of gcl include two X-chromosome counting elements (XCEs), scute (sc/sis-b) and sisterless-a (sis-a), that function to turn on the sex determination gene, Sxl, in female soma (Cline and Meyer, 1996; Salz and Erickson, 2010). gcl embryos not only fail to shut off sis-a and sis-b transcription in PBs, but also show disrupted PGC formation. In some gcl embryos, PGC formation fails completely, while in other embryos only a few PGCs are formed (Cinalli and Lehmann, 2013; Jongens et al., 1992; Lerit et al., 2017; Robertson et al., 1999). In this respect, gcl differs from nos and pgc, which have no effect on the process of PGC formation, but instead interfere with the specification of PGC identity.

Studies by Leatherman et al., 2002 suggested that the defects in PGC formation in gcl mutant embryos are linked to failing to inhibit somatic transcription. They found that when PBs first form during NC 9 in wild-type (WT) embryos, levels of CTD phosphorylation PB are only marginally less than in nuclei elsewhere in the embryo. However, by NC 10, there was a dramatic reduction in CTD phosphorylation even before PBs cellularize. By contrast, in gcl mutant embryos, about 90% of the NC 10 PB nuclei had CTD phosphorylation levels approaching that of somatic nuclei. Moreover, this number showed an inverse correlation with the number of PGCs in blastoderm stage gcl embryos. Whereas WT blastoderm embryos have >20 PGCs per embryo, gcl embryos had on average just three PGCs under their culturing conditions. Interestingly, expression of the mouse homologue of Gcl protein, mGcl-1, can rescue the gcl phenotype in Drosophila (Leatherman et al., 2000). Supporting the conserved nature of the involvement of Gcl during transcriptional suppression, a protein complex between mGcl-1 and the inner nuclear membrane protein LAP2β is thought to sequester E2F:D1 to reduce transcriptional activity of E2F:D1 (Nili et al., 2001).

The connection Leatherman et al. postulated between failing to turn off ongoing transcription and defects in PGC formation in gcl mutants is controversial and unresolved. This model predicts that a non-specific inhibition of polymerase II should be sufficient to rescue PGC formation in gcl embryos. However, Cinalli and Lehmann, 2013 found that the PGC formation defects seen in gcl embryos were not rescued after injection of the RNA polymerase inhibitor, α-amanitin. Since α-amanitin treatment disrupted somatic cellularization without impacting PGC formation in WT embryos, they concluded that it effectively blocked polymerase transcription. On the other hand, subsequent experiments by Pae et al., 2017 raised the possibility that inhibiting transcription in pole cell nuclei is a critical step in PGC formation. These authors showed that Gcl is a substrate-specific adaptor for a Cullin3-RING ubiquitin ligase that targets the terminal pathway receptor tyrosine kinase, Torso, for degradation. The degradation of Torso would be expected to prevent activation of the terminal signaling cascade in PGCs. In the soma, Torso-dependent signaling activates the transcription of several patterning genes, including tailless, that are important for forming terminal structures at the anterior and posterior of the embryo (Casanova and Struhl, 1989; Klingler et al., 1988; Martinho et al., 2004; Pignoni et al., 1992; Strecker et al., 1989). Thus, by targeting Torso for degradation, Gcl would prevent the transcriptional activation of terminal pathway genes by the MAPK/ERK kinase cascade in PGCs. Consistent with this possibility, simultaneous removal of gcl and either the Torso ligand modifier, torso-like (tsl) or torso resulted in rescue of germ cell loss induced by gcl. Surprisingly, however, Pae et al., 2017 were unable to observe a similar rescue of gcl phenotype when they used RNAi knockdown to compromise components of the MAP kinase cascade known to act downstream of the Torso receptor (Ambrosio et al., 1989; Duffy and Perrimon, 1994; Furriols and Casanova, 2003). Based on these findings, they proposed that activated Torso must inhibit PGC formation via a distinct non-canonical mechanism that is both independent of the standard signal transduction pathway and does not involve transcriptional activation.

In the studies reported here, we have revisited these conflicting claims by examining the role of Gcl in establishment/maintenance of transcriptional quiescence. The studies of Leatherman et al., 2002 indicated that two of the key X chromosomal counting elements, sis-a and sis-b, were inappropriately expressed in gcl PBs and PGCs. Since transcription factors encoded by these two genes function to activate the Sxl establishment promoter, Sxl-Pe, in somatic nuclei of female embryos, their findings raised the possibility that Sxl might be ectopically expressed in PBs/PGCs of gcl embryos. Here we show that in gcl embryos, Sxl transcription is indeed inappropriately activated in PBs and newly formed PGCs. Moreover, ectopic expression of Sxl in early embryos disrupts PGC formation similar to gcl. Supporting the conclusion that Sxl is a biologically relevant transcriptional target of Gcl, PGC formation defects in gcl embryos can be suppressed either by knocking down Sxl expression using RNAi or by loss-of-function mutations. As reported by Pae et al., 2017, we found that loss of torso-like (tsl) in gcl embryos suppresses PGC formation defects. However, consistent with a mechanism that is tied to transcriptional misregulation, rescue is accompanied by the reestablishment of transcriptional silencing in gcl PGCs. Lending further credence to the idea that transcription misregulation plays an important role in disrupting PGC development in gcl embryos, we found that expression of a mutant form of Torso that is resistant to Gcl-dependent degradation (hereafter referred to as torsoDeg Pae et al., 2017) ectopically activates transcription of two Gcl targets, sis-b and Sxl, in PB and PGC nuclei. In addition, stabilization of Torso in early PGCs also mimics another gcl phenotype, the failure to properly sequester key PGC determinants in PBs and newly formed PGCs.

Results

Gcl represses the expression of XCEs in nascent PGCs

To reexamine the role of gcl in transcriptional quiescence reported by Leatherman et al., 2002, we first used single molecule fluorescent in situ hybridization (smFISH) to assess whether sis-b is properly turned off in gcl mutants. As shown in Figure 1a, nuclear sis-b transcripts are not detected in WT PBs or PGCs (n = 16 embryos). In contrast, in 67% of gcl embryos, we observed sis-b transcripts in PB and PGC nuclei (Figure 1b, n = 21 embryos, p=2.1e-05). sis-b transcripts are present most frequently in gcl PBs; however, we can also detect transcripts after PGC cellularization. Leatherman et al. reported that a second XCE, sis-a, is not properly turned off in gcl PBs and PGCs. To determine if gcl is required to repress other XCEs besides sis-a and sis-b, we probed for runt expression in gcl mutants. We found that like sis-b, runt is also expressed in a subset of gcl PB nuclei and PGCs (29% of gcl embryos, n = 11, p=0.009653), while it is never observed in WT PBs or PGCs. Curiously, in experiments where we examined sis-b and runt transcription simultaneously, we observed some PB/PGC nuclei that expressed both XCEs, and some that only expressed one or the other. In this regard, it is important to keep in mind that transcription during early embryogenesis is stochastic, as only a subset of nuclei express the same gene at any given time (Fukaya et al., 2016; Muerdter and Stark, 2016; Zoller et al., 2018). Consequently, we used embryo counts in place of individual pole cell counts to compare between different samples, which likely underrepresents the frequency of the observed ectopic transcription events in PBs/PGCs. Nonetheless, as WT PBs or PGCs never display sis-b transcripts, our data show that Gcl is required to repress the transcription of XCEs in PBs and PGCs.

sis-b and Sxl are transcribed in gcl PBs and PGCs.

smFISH was performed using probes specific for sis-b or Sxl on 0–3 hr old embryos to assess the status of transcription in gcl PBs. Wild-type (WT) embryos of similar age were used as control. Posterior poles of representative pre-syncytial blastoderm embryos are shown with sis-b (a/b) or Sxl (c/d) RNA visualized in red and Hoescht DNA dye in blue. While 0% of control embryos display sis-b (a/a’, n = 16) or Sxl (c/c’, n = 18) transcription in PBs, transcription of both sis-b (b/b’) and Sxl (d/d’) is detected in gcl mutant PBs. We observed sis-b transcription in 67% (n = 21, p=2.10e-05) and Sxl transcription in 42% (n = 31, p=0.001593) of gcl embryos. Scale bar represents 10 µm.

Sxl RNA is detected in gcl PBs and PGCs

We next used smFISH to determine if the transcriptional target of the XCEs, the Sxl-Pe promoter, is active in gcl PBs and/or PGCs. In the soma, we found that the pattern of Sxl-Pe activity was indistinguishable between WT and gcl embryos, as Sxl-Pe transcripts are not detected prior to nuclear migration, nor are they observed in NC 10 somatic nuclei. In approximately half of the embryos, Sxl-Pe transcripts are observed in somatic nuclei from NC 11 until NC 14. Moreover, in these embryos, two nuclear dots of hybridization are detected in most nuclei, indicating that they are female (Erickson and Quintero, 2007; Keyes et al., 1992). In the remaining gcl and WT NC 11–14 embryos, Sxl-Pe transcripts are not observed in somatic nuclei, indicating that these embryos are male.

While the pattern of Sxl-Pe activity in the soma of gcl embryos is the same as WT, this is not true in the germline. As shown in Figure 1d, Sxl-Pe transcripts can be detected in PBs and PGCs in 42% of gcl embryos (n = 31 embryos, p=0.001593), while transcripts are not observed in WT PBs or PGCs (Figure 1c, n = 18 embryos). It is notable that the Sxl-Pe promoter remains active after the PBs cellularize, and nascent Sxl-Pe transcripts can be detected in PGC nuclei of gcl embryos, while they are never observed in the WT PGCs. In gcl embryos, Sxl-Pe transcripts are found not only in female PGCs, as evidenced by Sxl-Pe expression in somatic nuclei, but also in male gcl PGCs, which lack somatic Sxl-Pe. In the NC 11–14 embryos examined, the frequency of female gcl embryos expressing Sxl-Pe transcripts in their PGCs is somewhat higher than that of male gcl embryos (Table 1). Two factors could contribute to this bias. First, Sxl-Pe promoter activity is turned on by XCEs (Sis-A, Sis-B, Runt) in a dose-dependent manner, and these XCEs are also gcl targets. Second, there are two copies of the Sxl gene in females, which could increase the probability that it will be active in gcl mutants.

Table 1
PGC transcription in gcl embryos shows a slight, but not significant, sex bias.

Significance for sex ratios of embryos showing transcription in PBs and PGCs was determined using Fisher’s exact test; p-values are displayed in the right column.

No PGC transcriptionPGC transcriptionp-value
MaleFemaleMaleFemale
WT1418001
gcl12155140.235205

To determine if the Sxl-Pe mRNAs detected in gcl PBs and PGCs are properly processed, exported, and translated, we probed WT and gcl embryos with Sxl antibodies. As Sxl-Pe is not activated in WT female embryos until NC 11, Sxl protein is only readily detectable in somatic nuclei during NC 13/14. It is normally absent in the somatic nuclei of male embryos and in the PGCs of both sexes. While the pattern of Sxl protein accumulation in the soma of gcl embryos is the same as WT, this is not true in PGCs. Sxl protein can be detected in the PGCs of gcl embryos (Figure 2A–D; WT control: n = 40; 2/40 Sxl positive PGC nuclei as opposed to gcl: n = 36; 16/36 Sxl positive, p=7.621e-05). These data indicate that the Sxl-Pe promoter is normally repressed by Gcl in PBs and newly formed PGCs. While the failure to turn off the ongoing transcription of sis-b, sis-a (and possibly runt) likely contributes to the activation of Sxl-Pe in gcl PBs and PGCs, the fact that activation of the promoter is earlier than normal and is subsequently observed in both female and male PGCs suggests that XCE activity may not be the only contributing factor.

Gcl represses Sxl expression in the early embryonic pole cells, and ectopic expression of Gcl is sufficient to repress Sxl in somatic nuclei.

0–4 hr old paraformaldehyde-fixed embryos from mothers of indicated genotype were stained with anti-Sxl antibody to assess whether Sxl expression is upregulated in gcl PGCs (A–D). Posterior of the embryos are oriented to the right in all images. Panels A and B: early syncytial blastoderm stage embryos. Sxl protein is absent in the pole cells from the control (wild-type [WT]) embryo (A) whereas some of the gcl mutant pole cells show presence of Sxl (B). Panels C and D: Syncytial blastoderm stage female embryos from mothers of the indicated genotype were stained using anti-Sxl antibody. Similar to pole buds, only gcl mutant pole cells show Sxl protein (D) as opposed to the control (C). Panels E–F’: To determine whether Gcl is sufficient to repress Sxl expression on its own, embryos derived from females carrying gcl-bcd 3’UTR transgene (F) were stained using anti-Sxl antibodies. WT embryos were used as a control (E). The gcl-bcd 3’UTR transgene consists of genomic sequences of the gcl coding region fused to the 3’UTR of the anterior determinant bcd, resulting in ectopic localization of gcl mRNA to the anterior pole. Anterior poles are oriented to the left in each image. Images on the right in the panels E’ and F’ show just the anterior pole from the same embryos. While Sxl-specific signal is strong and uniform in the control embryo, selective reduction in Sxl in the anterior is readily seen in the gcl-bcd 3’UTR embryo (marked with an asterisk).

Ectopic expression of gcl represses Sxl

The experiments described above indicate that gcl is required to keep Sxl off in PBs and PGCs. We wondered whether gcl is sufficient to downregulate Sxl expression independent of other maternally derived components of the pole plasm, like nos, that are known to be required to keep the Sxl gene off. To address this question, we took advantage of a transgene in which the gcl mRNA protein coding sequence is fused to the bicoid (bcd) 3’UTR (Jongens et al., 1994; Leatherman et al., 2002). Using this transgene, Leatherman et al., 2002 found that expression of Gcl at the anterior of the embryo induced a local reduction in the expression of sis-b, sis-a, as well as terminal patterning genes such as tailless and huckebein. Nuclear accumulation of Sxl protein is uniform across the WT control female embryo, including the anterior (n = 12), while male embryos are completely devoid of Sxl (n = 15). We found that Sxl protein accumulation was diminished in nuclei at the anterior of gcl-bcd-3’UTR female embryos. While reduction in Sxl was observed in all female embryos, it was readily discernible in 9/13 embryos; p=2.23e-04. By contrast, Sxl was absent in gcl-bcd-3’UTR male embryos as in the case of control (n = 15) (Figure 2E and F). This localized disruption of Sxl expression is coincident with the anterior expression of Gcl protein in the gcl-bcd-3’UTR embryos, indicating that Gcl alone is sufficient to repress Sxl.

Premature expression of Sxl in the PGCs leads to germ cell loss and defective germ cell migration

Since our findings indicate that Sxl is inappropriately expressed in gcl PBs and newly formed PGCs, an important question is whether precocious expression of Sxl has detrimental effects on PGC development. To test this possibility, we ectopically expressed Sxl in early embryos. We mated maternal-tubulin-GAL4 (referred to as mat-GAL4) virgin females with males carrying a UAS-Sxl transgene. The maternally deposited GAL4 was able to drive the zygotic expression of Sxl protein in early female and male embryos independent of its normal regulation. We compared the total number of PGCs in late syncytial and early cellular blastoderm (stage 4/5) mat-GAL4/UAS-Sxl with mat-GAL4 embryos. In WT, there are typically about 25 PGCs in stage 4/5 embryos. This number is reduced nearly twofold in mat-GAL4/UAS-Sxl embryos (Figure 3). A reduction of PGCs was also observed when we mated virgins carrying the germ cell-specific nosGAL4-VP16 promoter to UAS-Sxl males to drive expression in the germline (6.5 PGCs per gonad in nosGAL4/UAS-Sxl embryos, n = 15, compared to 10 PGCs per gonad in nosGAL4/+ control, n = 12 embryos). Further, overexpression of Sxl in the germline impaired PGC migration. Figure 4 shows PGC migration defects in nosGAL4-VP16/UAS-Sxl embryos (3/21 UAS Sxl/+ control embryos showed >5 mispositioned PGCs as opposed to 9/17 nosGAL4-VP16/UAS-Sxl embryos; p=0.04). Taken together, these findings demonstrate that precocious expression of Sxl protein has deleterious effects on PGC development and behavior during early embryogenesis.

Precocious expression of Sxl results in reduction in total number of primordial germ cells (PGCs).

Embryos of indicated genotypes were stained for pole cell marker Vasa (panels a and b; imaged in red) to discern the effects of precocious Sxl activity on the early PGCs. UAS-Sxl transgene males were mated with females carrying maternal-tubulin-GAL4 driver (panel b) to assess if precocious Sxl expression adversely influences early PGCs. mat-GAL4 (panel a) and gcl (not shown) embryos served as positive and negative controls, respectively. (c) Quantitation of PGC counts in different genetic backgrounds. The number of pole cells in embryos from mothers of indicated genotypes were counted and compared. Bars represent the mean ± S.D. (n = 23 for gcl, n = 14 for mat-GAL4/UAS-Sxl, n = 12 for mat-GAL4). ****p<0.0001 for gcl and mat-GAL4/UAS-Sxl compared to wild type (WT). Note that *p>0.01 for gcl compared to mat-GAL4/UAS-Sxl (not indicated in the graph).

Germ cell-specific expression of Sxl leads to germ cell migration defects during mid-embryogenesis.

Embryos from mothers of the indicated genotypes were stained for the germ cell marker Vasa. UAS-Sxl transgene males were mated with virgin females carrying the germline-specific driver nos-GAL4-VP16 to assess if precocious Sxl expression can influence PGC migration and survival (panels C–F). Embryos at stage 12 (A, C, E) and stage 13 (B, D, F) are shown as germ cell behavior defects become apparent from stage 12 onwards. UAS-Sxl/+ embryos served as control (A and B). Readily detectable germ cell migration defects were seen in the experimental embryos as opposed to the control. 3/21 UAS Sxl/+ control embryos showed >5 mispositioned PGCs as opposed to 9/17 nosGAL4-VP16/UAS-Sxl embryos; p=0.04 (significance determined using Welch’s two sample t-test).

Simultaneous removal of gcl and Sxl ameliorates the gcl phenotype

The finding that premature ectopic expression of Sxl protein has adverse effects on PGC development supports the idea that one critical function of gcl is repressing Sxl-Pe. If this is correct, then compromising Sxl activity in the early embryo should mitigate the PGC defects seen in gcl embryos. For this purpose, we generated gcl embryos that also carry a small deficiency, Sxl7BO, which deletes the Sxl gene. In this experiment, we mated Sxl7BO/Bin; gcl/gcl mothers to Sxl7BO/Y fathers, and 0–12 hr old progeny were probed with Sxl and Vasa antibodies. While all of the progeny from this cross lack maternally derived Gcl protein, only half of the progeny lack the Sxl gene. For female embryos, one half would be Sxl7BO/Bin, while the other half would be Sxl7BO/ Sxl7BO. The former (Sxl7BO/Bin) have a functional Sxl gene, and, since they are females, they will express Sxl protein in the soma, which can be detected with Sxl antibody. The latter (Sxl7BO/ Sxl7BO) do not have a functional Sxl gene and would not express Sxl protein even though they are female. There are also two classes of male embryos. One half would be Bin/Y, while the other half would be Sxl7BO/Y. The former (Bin/Y) has a functional Sxl gene, but since they are males (with a single X chromosome), Bin/Y embryos would not express Sxl protein. The latter, Sxl7BO/Y lacks a functional Sxl gene and would also not express Sxl protein.

To identify the different classes of embryos, we stained with Sxl antibody. Using this approach, we can unambiguously identify the genotype of the Sxl7BO/Bin, as they express Sxl protein throughout the soma. One quarter of the embryos fall into this class. The remaining three quarters of the embryos do not express Sxl protein, and we cannot unambiguously identify their genotype or sex. However, we know that one third of the embryos that do not stain with Sxl antibody are Bin/Y males and have thus a functional Sxl gene. The remaining embryos (two thirds of the embryos that do not stain with Sxl antibody, or one half of the total embryos in the collection) are either Sxl7BO/ Sxl7BO females or Sxl7BO/Y males, and, in both cases, they lack a functional Sxl gene.

If removal of Sxl ameliorates the gcl defects in PGC formation, then we should observe an increase in the number of PGCs in only one half of the embryos from this cross. Moreover, this increase should be observed in the embryos that do not stain with Sxl antibody. However, within the group of embryos that do not stain with Sxl antibody, only two thirds should show an increased number of PGCs. All of these expectations are met. The graph in Figure 5 shows that the average number of PGCs in Sxl+ gcl (Sxl7BO/Bin) (mean ~3, n = 14) (female) embryos is not too different from that in gcl embryos (mean ~2, n = 24) that are WT for Sxl. In the class of embryos that lack Sxl protein, there are two unequal groups, as expected. In one group, which corresponds to about one third of the unstained embryos, the mean number of PGCs is 3.5. This group matches closely with the number of PGCs in Sxl+ (Sxl7BO/Bin) females and thus presumed to be Sxl+ (Bin/Y) males. In the other group, which corresponds to about two thirds of the unstained embryos (or half the total number of embryos), the mean number of PGCs is 12. Embryos in this group are presumably Sxl- males and females. The combined average of the PGC count for all of the embryos that do not stain with Sxl antibody is ~8.5, which is also significantly higher than gcl (~2; see Figure 5 legend for details). These findings indicate that removing the Sxl gene ameliorates the effects of the gcl mutation on PGCs.

Simultaneous removal of gcl and Sxl mitigates the gcl phenotype.

(A-B) 0–12 hr old embryos (from the cross 7BO/Y;gcl/gcl x7BO/Bin;gcl/gcl) were stained using anti-Sxl antibody and for the germline marker Vasa. 7BO is a small deficiency chromosome that specifically deletes the Sxl gene. Embryos that stained positive for Sxl were disregarded (n = 14) since only embryos lacking Sxl and gcl are relevant in this experiment. Male embryos of genotype Bin/Y; gcl/gcl (A) are compared with embryos believed to be of genotype 7BO/7BO; gcl/gcl or 7BO/Y; gcl/gcl(B). The number of pole cells in embryos from mothers of indicated genotypes were counted and plotted (C). Bars represent the mean ± SD (n = 23 for 7BO/7BO; gcl/gcl, n = 19 for 7BO/Y; gcl/gcl, n = 26 for Bin/Y; gcl/gcl). ****p<0.0001 for 7BO/7BO; gcl/gcl and 7BO/Y; gcl/gcl compared to Bin/Y; gcl/gcl. p=0.03 for 7BO/7BO; gcl/gcl compared to 7BO/Y; gcl/gcl. Significance was determined using Welch’s two sample t-test.

RNAi knockdown of Sxl also ameliorates the PGC formation defects in gcl embryos

To confirm that ectopic activation of Sxl-Pe in gcl mutants has deleterious effects on PGCs, we also used RNAi to knockdown expression of Sxl protein. gcl mothers carrying a mat-GAL4 driver were mated to males carrying UAS-Sxl-RNAi transgene, and the embryos derived from this cross were stained with anti-Vasa antibodies to visualize PGCs. Figure 6 shows that RNAi knockdown of Sxl (SxlRNAi) partially suppresses the effects of the gcl mutation on PGCs. While all the embryos in this experiment were of identical genotype, they fell into two classes: one in which the number of PGCs in syncytial/early cellular blastoderm embryos is nearly WT and another that had few PGCs.

Knockdown of Sxl partially suppresses germ cell loss of gcl embryos.

gcl;mat-GAL4 virgin females were mated with males carrying UAS-Sxl RNAi. Embryos derived from this cross were stained with anti-Vasa antibody to visualize PGCs (A). Scale bar represents 20 µm. Total number of PGCs were counted for each embryo from different genotypes, and a Mann–Whitney U-test was employed to analyze significant differences between wild type (WT), gcl, and gcl;SxlRNAi (B). In 66% of gcl;SxlRNAi embryos, few or no pole cells are observed, comparable to gcl. However, in 34% of gcl;SxlRNAi embryos, germ cell count is substantially elevated, indicating partial rescue of the gcl phenotype.

A plausible explanation for this bimodal distribution is that the efficiency of rescue reflects sex-specific differences in the dose of X-linked sex-determination genes. Females have two copies of not only Sxl but also the XCEs responsible for activating Sxl-Pe, whereas males have only a single copy of these genes. Consistent with gene dose being relevant, there is a modest female-specific bias in the frequency in which we detect Sxl-Pe transcripts in gcl PBs/PGCs (Table 1). To test this directly, we determined the sex of the gcl and control embryos using smFISH with sis-b and Sxl probes. At the syncytial blastoderm stage somatic nuclei in female embryos have two dots of hybridization for both sis-b and Sxl. By contrast, male embryos have one dot of hybridization for sis-b and no signal for Sxl (Figure 7). When we stained embryos derived from the experimental cross, we observed that all embryos showing an increase in PGC formation were females (Sxl+ and two dots of sis-b signal) (Table 2, n = 59, p=0.002456).

Sexing embryos based on transcription puncta from X-chromosomes.

0–3 hr old wild-type (WT) embryos were probed for Sxl (green) and sis-b (red) transcription using smFISH, and these embryos were co-stained with Hoescht to visualize DNA. (A) Embryos with two X-chromosomes (females) show two transcription puncta for both sis-b and Sxl, corresponding to expression from each X. (B) XY embryos (males) transcribe sis-b from the only X chromosome and fail to activate expression of Sxl. (A and B) show merge; (A’ and B’) show smFISH signals. A representative section of somatic nuclei is shown in each panel. Scale bar represents 10 µm.

Table 2
Rescue of PGC numbers in gcl;SxlRNAi embryos only occurs in female embryos.
MaleFemale
No rescue2026
Rescue013

Ectopic transcription is attenuated in gcl;tsl PGCs

In their studies showing that Gcl targets the terminal pathway receptor Torso for proteolysis, Pae et al., 2017 found that mutations in the torso-like (tsl) ligand modifier or RNAi knockdown of torso also suppressed the PGC defects in gcl embryos. We confirmed that simultaneous removal of maternal tsl and gcl resulted in a substantial rescue of the PGC formation defects in gcl embryos (Pae et al., 2017). Figure 8 and Table 3 show that gcl;tsl embryos display a significant increase in the number of PGCs as compared to gcl embryos, and that the rescue is highly penetrant (p<2e-16, Figure 8D). (Note also that the rescue is more substantial than that observed in the Sxl experiments.)

Rescue of PGCs in gcl;tsl embryos.

smFISH using Sxl probes was performed to assess the status of transcription in PBs of wild-type (WT) (A), gcl (B), and gcl;tsl (C) 0–3 hr old embryos. Posterior poles of representative blastoderm embryos are shown with Sxl RNA visualized in green and Hoescht DNA dye in blue. While 0% of control embryos display Sxl transcription in PBs, transcription of Sxl is detected in 67% buds of gcl embryos (indicated with a carrot in the representative embryo). In gcl;tsl embryos, however, 0% display any ectopic transcription (Table 3). n = 28, 23, and 24 for WT, gcl, and gcl;tsl embryos, respectively; by Fisher’s exact test, p=1e-06 and 1 for WT compared to gcl and gcl;tsl, respectively, and p=2e-06 for gcl compared to gcl;tsl. Scale bar represents 10 µm. (D) Pole cell counts from WT, gcl, and gcl;tsl embryos were counted using anti-Vasa staining (n = 17, 25, and 18, respectively). ***p<0.001 for the compared genotypes shown. Significance was determined using a one-Way ANOVA (p=0) with pairwise t-test comparisons (p=0 for WT vs. gcl, p=0.14 for WT vs. gcl;tsl, p=0 for gcl vs. gcl;tsl). These data suggest that rescue of the gcl PGC numbers is highly penetrant in gcl;tsl embryos.

Table 3
Transcription status in PBs and PGCs of wild-type (WT), gcl, and gcl;tsl embryos (assessed using smFISH for sis-b and Sxl).

Significance for proportions of embryos showing transcription in PBs and PGCs was determined using Fisher’s exact test; p-values are displayed in the right column.

GenotypeNo transcriptionTranscriptionp-value
WT280
gcl9141.00e-06
gcl;tsl2401

Leatherman et al., 2002 found that gcl was required for turning off somatic gene transcription in PBs/PGCs, and they suggested that one of the critical functions of gcl in PGC formation is the silencing of transcription. In addition to confirming that gcl is required to turn off transcription in PBs/PGCs, we also identified an important target for gcl mediated repression, the Sxl establishment promoter, Sxl-Pe. Taken together with the fact that removal of tsl gives nearly complete rescue of the PGC formation defects in gcl, these observations would imply that gcl must target Torso for degradation (at least in part) in order to block the terminal pathway from promoting the transcriptional activity of somatic genes (including activation of Sxl-Pe). If this prediction is correct, then the misexpression of Sxl-Pe and other genes should not be observed in embryos from gcl;tsl mothers where the PGC formation defects are rescued. To test this prediction, we performed smFISH on gcl;tsl embryos using Sxl and sis-b probes along with gcl and WT embryos as positive and negative controls, respectively. Table 3 shows that removal of tsl restores transcriptional quiescence in the PBs/PGCs of gcl;tsl embryos (Figure 8 and Table 3, p=1e-06 and 1 for WT compared to gcl and gcl;tsl, respectively, by Fisher’s exact test). Taken together, these data confirm that inactivation of the terminal signaling pathway by Gcl is critical for silencing transcription in PBs and PGCs and that this silencing function plays an important role in PGC formation.

A degradation-resistant form of Torso also activates transcription in PGCs

Our finding that the survival of gcl;tsl PGCs is accompanied by the reestablishment of transcriptional silencing provides strong support for the idea that gcl targets Torso for degradation to block terminal signaling dependent transcription. A prediction of this model is that transcription of gcl targets should be ectopically activated in PBs/PGCs when Gcl-dependent proteolysis of Torso is blocked. To test this prediction, we took advantage of a mutant transgene version of torso, torsoDeg, generated by Pae et al., 2017, that lacks the Gcl interaction domain and is thus resistant to Gcl-dependent proteolysis. Embryos from females carrying both mat-GAL4 and UAS-torsoDeg were probed for sis-b and Sxl-Pe promoter activity. Figure 9 shows that both sis-b and Sxl-Pe transcripts are inappropriately expressed in the PBs and PGCs of torsoDeg embryos, with frequencies less than those observed in gcl embryos but significantly more than control embryos (27% of torsoDeg embryos express sis-b (n = 16, p=0.043382) and 28% of torsoDeg embryos express Sxl (n = 25, p=0.030307)). Thus, ectopic upregulation of Sxl and sis-b transcription observed in gcl pole cells is recapitulated in torsoDeg embryos.

Transcriptional quiescence in pole cells is compromised in torsoDeg embryos.

smFISH using probes specific for sis-b or Sxl in 0–3 hr old embryos was performed to assess the status of transcription in torsoDeg PBs. Posterior poles of representative pre-syncytial blastoderm embryos are shown with sis-b (a/b) or Sxl (c/d) RNA visualized in red and Hoescht DNA dye in blue. While 0% of control embryos display sis-b (a/a’, n = 16) or Sxl (c/c’, n = 18) transcription in PBs, transcription of both sis-b (b/b’) and Sxl (d/d’) is detected in buds of torsoDeg embryos. Note that transcription in wild-type (WT) embryos is only in somatic nuclei (a). We observed sis-b transcription in 27% (n = 15, p=0.043382) and Sxl transcription in 28% (n = 25, p=0.030307) of torsoDeg embryos. Scale bar represents 10 µm.

Taken together with the data reported by Leatherman et al., 2002, our results indicate that ectopic expression of Gcl at the anterior of the embryo downregulates transcription of multiple genes. If the relevant target for gcl in gcl-bcd-3’UTR embryos at the anterior is the Torso receptor, then we would predict that torsoDeg should impact transcription not only in the germline, but also in the soma. Since the X-chromosome counting system, which regulates Sxl-Pe activity, is (at least partially) overridden in torsoDeg PBs and PGCs, it seemed possible that it might also be overridden in the soma. To test this possibility, we examined Sxl-Pe expression in the soma of torsoDeg embryos. In WT females, Sxl-Pe transcripts can be detected in virtually all somatic nuclei, and two dots of hybridization are typically observed (Figure 7). In males, Sxl-Pe is off and their somatic nuclei are completely devoid of the signal. While female torsoDeg embryos resemble WT, we observed scattered nuclei in which Sxl-Pe is active in 43% of torsoDeg male embryos (Figure 10C, n = 14, p=0.023871). This finding is also consistent with earlier studies in which we found that a constitutively active form of the Torso receptor, RL3, turns on Sxl-Pe in males (Deshpande, 2004).

Sxl is expressed in the male soma in torsoDeg and MEK GOF embryos.

0–3 hr old embryos were probed for somatic Sxl transcription using smFISH. While 0% of control male embryos display Sxl expression in the soma (A and A', n = 10), all control females display uniform somatic Sxl expression (B and B', n = 17). However, we observed sporadic somatic Sxl activation in 43% (n = 14, p=0.023871) of torsoDeg (C and C') and 46% (n = 13, p=0.019079) of MEKE203K (D and D') male embryos. A representative section of somatic nuclei is shown in each panel (blue) with Sxl transcripts in red. Scale bar represents 10 µm.

Does Gcl target a non-canonical Torso-dependent signaling pathway?

In the canonical terminal pathway, binding of the Tsl ligand to Torso activates a MAP kinase cascade that ultimately results in the phosphorylation and subsequent degradation of the transcriptional repressor Capicua by the ERK kinase (de las Heras and Casanova, 2006; Grimm et al., 2012). Degradation of Capicua, in turn, results in the transcription of terminal patterning genes, such as tailless. Surprisingly, however, Pae et al., 2017 found that unlike RNAi knockdowns of the torso receptor, RNAi knockdown of two terminal pathway kinases, dsor1 (MEK) and rolled (MAPK) that function downstream of Torso, failed to rescue the PGC defects of gcl embryos. From this finding, the authors concluded that Gcl-mediated degradation of the Torso receptor must disrupt the operation of a novel non-canonical Torso signaling pathway. To test the possibility that this non-canonical pathway might have a transcriptional output like the canonical transduction pathway, we used mat-GAL4 to drive the expression of two activated versions of the MAPK/ERK kinase (MEKE203K and MEKF53S Goyal et al., 2017) in mothers and then assayed Sxl-Pe transcription in PBs and PGCs of their progeny. We found that maternal deposition of MEKE203K or MEKF53S could not activate Sxl-Pe transcription in PBs or PGCs (not shown, see Discussion). Nevertheless, we found that, as was observed for torsoDeg, Sxl-Pe expression is activated in male somatic nuclei by the GOF MEK proteins (Figure 10D, 46% of MEKE203K males showed patchy somatic Sxl expression, n = 13, p=0.019079). Taken together with our previous findings (Deshpande, 2004), this result would argue that the canonical Torso signaling pathway is capable of impacting Sxl-Pe promoter activity. In this context, it is also interesting to note that a key transcriptional target of the terminal signaling pathway, tailless, is not activated in gcl PBs or PGCs. This is also true for embryos expressing TorsoDeg or either of the GOF MEK variants (not shown). Since tailless transcription is ectopically activated in pgc mutant PGCs, it would appear that the canonical terminal signaling pathway is not able to overcome the repressive effects of the Pgc protein in the case of tailless, even in a gcl background.

torsoDeg disrupts the sequestration of germline determinants

One of the more striking phenotypes in gcl mutants is a failure to properly sequester protein and mRNA components of the pole plasm. In WT embryos, nuclei entering the posterior pole trigger the release of the pole plasm from the posterior cortex of the embryo by a centrosome/microtubule-dependent mechanism (Lerit and Gavis, 2011; Raff and Glover, 1989). Once released, the pole plasm constituents are distributed within the growing bud by a microtubule-dependent mechanism. However, spreading is restricted to the growing bud and the pole plasm components are ultimately incorporated into newly formed PGCs when the buds cellularize. In gcl embryos, nuclear entry also triggers the release of the pole plasm from the cortex; however, the pole plasm proteins and mRNAs are not retained in the newly formed PBs after they are released, but instead spread to the cytoplasmic territories of neighboring somatic nuclei along the cortex and also into the interior of the embryo (Lerit et al., 2017). The difference between WT and gcl in the localization of pole plasm constituents is shown for Vasa protein (Figure 11) and gcl (Figure 12), pgc (Figure 13), and nos (Figure 14) mRNAs. As shown in maximum intensity projections and the accompanying distribution graphs, Vasa and the three pole plasm mRNAs are sequestered in the PGCs of WT embryos. In contrast, in gcl embryos, Vasa protein, and pgc and nos mRNAs spread into the territories occupied by nearby somatic nuclei. As evident from the profiles of pole plasm distribution for individual embryos, the extent of spreading varies somewhat from embryo to embryo; however, retention of pole plasm constituents in PGCs is clearly disrupted in gcl embryos. In single sections, we also observe pole plasm constituents spreading into the interior of the embryo as well as along the posterior lateral cortex. We also detected no gcl mRNA in the gcl mutant, as expected (Figure 12).

Vasa is mislocalized from the posterior in gcl and torsoDeg embryos.

0–3 hr old paraformaldehyde-fixed embryos collected from wild-type (WT), gcl, or torsoDeg mothers were stained with anti-Vasa antibody to assess whether pole plasm is properly localized in gcl and torsoDeg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below, plot profiles show mislocalization of pole plasm (visualized using Vasa) away from posterior cap in gcl and torsoDeg embryos (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. n = 12, 13, and 13 for WT, gcl, and torsoDeg, respectively.

gcl RNA is mislocalized from the posterior in torsoDeg embryos.

smFISH using probes specific for gcl was performed in 0–3 hr old embryos to assess whether pole plasm is properly localized in gcl and torsoDeg embryos. gcl embryos lack gcl RNA, as previously reported (Jongens et al., 1992). On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below, plot profiles show mislocalization of pole plasm (visualized using gcl) away from posterior cap in torsoDeg embryos (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. n = 11, 10, and 16 for wild type (WT), gcl, and torsoDeg, respectively.

pgc RNA is mislocalized from the posterior in gcl and torsoDeg embryos.

smFISH using probes specific for pgc was performed in 0–3 hr old embryos to assess whether pole plasm is properly localized in gcl and torsoDeg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below, plot profiles show mislocalization of pole plasm (visualized using pgc) away from posterior cap in gcl and torsoDeg embryos (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. n = 10, 14, and 14 for wild type (WT), gcl, and torsoDeg, respectively.

nos RNA is mislocalized from the posterior in gcl and torsoDeg embryos.

smFISH using probes specific for nos was performed in 0–3 hr old embryos to assess whether pole plasm is properly localized in gcl and torsoDeg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below plot profiles show mislocalization of pole plasm (visualized using nos) away from posterior cap in gcl and torsoDeg embryos (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. n = 4, 6, and 6 for wild type (WT), gcl, and torsoDeg, respectively.

Interestingly, as was the case for transcriptional activation, the effects of torsoDeg on the sequestration of the pole plasm constituents are quite similar to those observed in gcl embryos. In early torsoDeg embryos, pole plasm constituents appear to be localized correctly to the posterior pole (Figure 15). However, after the nuclei migrate to the surface of the embryo, the localization of pole plasm components is disrupted. Vasa protein (Figure 11) and pgc (Figure 13) and nos (Figure 14) mRNAs spread into the territories of somatic nuclei located along the posterior lateral cortex of torsoDeg embryos. In addition, gcl mRNA (Figure 12) is not properly restricted in torsoDeg embryos, and like pgc and nos mRNAs, it is distributed along the lateral cortex. This finding is of special interest as it suggests the existence of an antagonistic relationship between torso and gcl. While gcl negatively regulates the Torso receptor by promoting its degradation, Torso activity likely controls the sequestration of pole plasm—including gcl mRNA—to the PBs and PGCs. Such a mechanism would avoid inappropriate exposure of the neighboring somatic nuclei to gcl RNA (and possibly protein), ultimately ensuring proper germline/soma distinction.

Before pole buds develop, pole plasm distribution is unaltered in gcl and torsoDeg embryos.

smFISH using probes specific for pgc or gcl was performed in 0–3 hr old embryos to assess whether pole plasm is properly localized in young gcl and torsoDeg embryos. Images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below, plot profiles show proper anchoring and localization of pole plasm (visualized using pgc or gcl) at the posterior cap in gcl and torsoDeg embryos (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. For the pgc smFISH experiment, n = 9, 7, and 7 for wild type (WT), gcl, and torsoDeg, respectively. For the gcl smFISH experiment, n = 14, 9, and 8 for WT, gcl, and torsoDeg, respectively.

Sequestration of germline determinants is disrupted by activated MEK

Although we found that ectopically expressed GOF MEK proteins are unable to recapitulate the effects of torsoDeg on transcriptional activity in PBs and PGCs, it was unclear whether this negative result means that a non-canonical Torso-dependent signaling pathway is responsible for activating transcription in gcl PBs and PGCs. To explore this question further, we tested whether ectopic expression of GOF MEK can induce defects in the sequestration of pole plasm components. As shown in Figure 16, MEKE203K or MEKF53S protein induces the inappropriate dispersal of gcl and pgc mRNAs into the surrounding soma in a pattern very similar to that observed in torsoDeg and gcl embryos. Thus, this gcl phenotype would appear to depend upon the canonical terminal signal transduction cascade.

MEK gain of function embryos also display defects in pole plasm localization.

smFISH using probes specific for pgc or gcl was performed in 0–3 hr old embryos to assess whether pole plasm is properly localized in embryos collected from mothers expressing MEKE203K or MEKF53S driven by mat-GAL4. On top, images are representative maximum intensity projections of pgc RNA localization at the posterior pole of each indicated genotype. Scale bar represents 10 µm. Below, plot profiles show mislocalization of pole plasm (visualized using either pgc or gcl) away from posterior cap in embryos expressing one of two MEK gain of function transgenes (E203K and F53S) (see Materials and methods for details of quantification). Each plot shows a representative experiment, with each line depicting pole plasm distribution of an individual embryo. n = 15, 19, and 9 for wild type (WT), MEKE203K, and MEKF53S, respectively.

Discussion

gcl differs from other known maternally deposited germline determinants in that it is required for the formation of PBs and PGCs. gcl PGCs exhibit a variety of defects during the earliest steps in PGC development. Unlike WT, gcl PGCs fail to properly establish transcriptional quiescence. While other genes like nos and pgc are required to keep transcription shut down in PGCs, their functions only come into play after PGC cellularization (Deshpande, 2004; Deshpande et al., 1999; Martinho et al., 2004). By contrast, gcl acts at an earlier stage beginning shortly after nuclei first migrate into the posterior pole plasm and initiate PB formation. In gcl PBs, ongoing transcription of genes that are active beginning around nuclear cycle 5–6 is not properly turned off. This is not the only defect in germline formation and specification. As in WT, the incoming nuclei (and the centrosomes associated with the nuclei) trigger the release of the pole plasm from the posterior cortex. However, instead of sequestering the germline determinants in PBs so that they are incorporated into PGCs during cellularization, the determinants disperse into the soma where they become associated with the cytoplasmic territories of nearby somatic nuclei. There are also defects in bud formation and cellularization. Like the release and sequestration of germline determinants, these defects have been linked to the actin cytoskeleton and centrosomes (Cinalli and Lehmann, 2013; Lerit et al., 2017).

Two models have been proposed to account for the PGC defects in gcl mutants. In the first, Leatherman et al., 2002 attributed the disruptions in PGC development to a failure to turn off ongoing transcription. The second argues that the role of gcl in imposing transcriptional quiescence is irrelevant (Cinalli and Lehmann, 2013; Pae et al., 2017). Instead, the defects are proposed to arise from a failure to degrade the Torso receptor. In the absence of Gcl-dependent proteolysis, high local concentrations of the Tsl ligand modifier at the posterior pole would activate the Torso receptor. According to this model, the ligand–receptor interaction would then trigger a novel, transcription-independent signal transduction pathway in PBs and PGCs that disrupts their development. These conflicting models raise several questions. Does gcl actually have a role in establishing transcriptional quiescence in PBs and PGCs? If so, is this activity relevant for PB and PGC formation? Is the stabilization of Torso in gcl mutants responsible for the failure to shut down transcription in PBs and PGCs? If not, does gcl target a novel, transcription-independent but Torso-dependent signaling pathway? Is the stabilization of Torso responsible for some of the other phenotypes that are observed in gcl mutants? In the studies reported here we have addressed these outstanding questions, leading to a resolved model of Gcl activity and function.

We show that shutting off transcription is, in fact, a critical function of Gcl protein. As previously documented by Leatherman et al., we find that several of the key X-linked transcriptional activators of Sxl-Pe are not repressed in newly formed PBs and early PGC nuclei, and Sxl-Pe transcription is inappropriately activated in the presumptive germline. In previous studies, we found that ectopic expression of Sxl in nos mutants disrupts PGC specification. In this case, the specification defects in nos embryos can be partially rescued by eliminating Sxl activity (Deshpande et al., 1999). The same is true for gcl mutants: elimination or reduction in Sxl function ameliorates the gcl defects in PGC formation/specification. Conversely ectopic expression of Sxl early in embryogenesis mimics the effects of gcl loss on PGC formation. Importantly, the role of Gcl in inhibiting Sxl-Pe transcription is not dependent upon other constituents of the pole plasm. When Gcl is ectopically expressed at the anterior of the embryo, it can repress Sxl. This observation is consistent with the effects of ectopic Gcl on the transcription of other genes reported by Leatherman et al., 2002. Since the rescue of gcl by eliminating the Sxl gene or reducing its activity is not complete, one would expect that there must be other important gcl targets. These targets could correspond to one or more of the other genes that are misexpressed in gcl PB/PGCs. Consistent with this possibility, transcriptional silencing in gcl PBs/PGCs is reestablished when terminal signaling is disrupted by mutations in the tsl gene. On the other hand, it is possible that excessive activity of the terminal signaling pathway also adversely impacts some non-transcriptional targets that are important for PB/PGC formation and that transcriptional silencing in only part of the story (see below).

Pae et al., 2017 showed that mutations in the Gcl interaction domain of Torso (torsoDeg) stabilize the receptor and disrupt PGC formation. Consistent with the notion that Torso receptor is the primary, if not the only, direct target of gcl, they found that mutations in the Torso ligand modifier, tsl, or RNAi knockdown of torso rescued the PGC formation defects in gcl embryos. As would be predicted from their findings and ours, ectopic expression of the TorsoDeg protein induces the inappropriate transcription of sis-b and Sxl-Pe in PBs and newly formed PGCs. Thus, the failure to shut down ongoing transcription in gcl PBs and PGCs must be due (at least in part) to the persistence of the Torso receptor in the absence of Gcl-mediated degradation. Corroborating this idea, the ectopic activation of transcription in gcl PGCs is no longer observed when the terminal signaling pathway is disrupted by the removal of tsl. Taken together, these data strongly suggest that the establishment/maintenance of transcriptional silencing in PBs is a critical function of Gcl.

Since RNAi knockdowns of terminal pathway kinases downstream of torso did not rescue gcl mutants, Pae et al., 2017 postulated that the Tsl-Torso receptor interaction triggered a novel, non-canonical signal transduction pathway that disrupted PGC development. If their suggestion is correct, then the activation of sis-b and Sxl-Pe in PBs/PGCs in gcl and torsoDeg embryos would be mediated by this novel terminal signaling pathway. Here, our results are ambiguous. Consistent with the suggestion of Pae et al., 2017, GOF mutations in MEK, a downstream kinase in the Torso signaling pathway, did not activate Sxl-Pe transcription in pole cells. However, an important caveat is that the GOF activity of MEK variants we tested is likely not equivalent to the activity from the normal Torso-dependent signaling cascade (Goyal et al., 2017). As the pole plasm contains at least two other factors that help impose transcriptional quiescence, the two GOF MEK mutants we tested may simply not be sufficient to overcome their repressive functions. Two observations are consistent with this possibility. First, like torsoDeg, we found that MEKE203K induces Sxl-Pe expression in male somatic nuclei. The same is true for a viable GOF mutation in Torso: it can induce ectopic activation of Sxl-Pe in male somatic nuclei, but is unable to activate Sxl-Pe in PGCs (Deshpande, 2004). Second, a key terminal pathway transcription target tailless is not expressed in gcl mutant PBs/PGSs even though the terminal pathway should be fully active. This is also true for embryos expressing torsoDeg and the two GOF MEK proteins. For these reasons, we cannot unambiguously determine if it is the canonical terminal signaling pathway or another, noncanonical signaling pathway downstream of Torso that is responsible for the expression of sis-b, Sxl-Pe, and other genes in gcl mutant PB/PGCs.

There are also reasons to think that the canonical Torso signal transduction cascade must be inhibited for proper PGC formation. One of the more striking phenotypes in gcl mutants is the dispersal of key germline mRNA and protein determinants into the surrounding soma. A similar disruption in the sequestration of pole plasm components is observed not only in torsoDeg embryos but also in MEKE203K and MEKF53S embryos. Thus, this gcl phenotype would appear to arise from the deployment of the canonical Torso receptor signal transduction cascade, at least up to the MEK kinase. However, this result does not exclude the possibility that the Tsl→Torso→ERK pathway has other non-transcriptional targets that, like Sxl-Pe expression, can also interfere with PB/PGC formation. If this was the case, it could potentially explain why global transcriptional inhibition failed to rescue the PGC defects in gcl embryos (Cinalli and Lehmann, 2013). In this respect, a potential—if not likely—target is the microtubule cytoskeleton. In previous studies, we found that the PB and PGC formation defects as well as the failure to properly sequester critical germline determinants in gcl arise from abnormalities in microtubule/centrosome organization (Lerit et al., 2017). Preliminary imaging experiments indicate that centrosome distribution of torsoDeg PBs is also abnormal, suggesting that inappropriate activation of the terminal signaling pathway perturbs the organization or functioning of the microtubule cytoskeleton and/or centrosomes. Such a mechanism would also be consistent with the dispersal of germline mRNA and protein determinants in torsoDeg and GOF MEK embryos. While further experiments will be required to demonstrate microtubule and centrosomal aberrations in torsoDeg and GOF MEK embryos, a role for a receptor-dependent MEK/ERK signaling cascade in promoting centrosome accumulation of γ-tubulin and microtubule nucleation has been documented in mammalian tissue culture cells (Colello et al., 2012). It is thus conceivable that MEK/ERK signaling has a similar role in Drosophila PB nuclei and PGCs. It will be important to determine if Torso-dependent activation of MEK/ERK can perturb the behavior or organization of centrosomes and/or microtubules in early embryos, and, if so, whether the influence can alter the pole plasm RNA anchoring and/or transmission. Taken together, our data reveal a mutual antagonism between the determinants that specify germline versus somatic identity. Future studies will focus on how and when during early embryogenesis such feedback mechanisms are activated and calibrated to establish and/or maintain germline/soma distinction.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Genetic reagent (D. melanogaster)gclJongens et al., 1994
Genetic reagent (D. melanogaster)gcl-bcd-3’UTRJongens et al., 1994
Genetic reagent (D. melanogaster)Maternal-tubulin-GAL4 (67.15)Eric Wieschaus
Genetic reagent (D. melanogaster)nosGAL4-VP16Bloomington Drosophila Stock CenterBDSC: 7303; RRID:BDSC_7303
Genetic reagent (D. melanogaster)UASp-Sxl (DB106)Helen SalzMaintained in the lab of H. Salz
Genetic reagent (D. melanogaster)Sxl7BOTom Cline
Genetic reagent (D. melanogaster)UAS-Sxl RNAi (VALIUM20)Bloomington Drosophila Stock CenterBDSC: 34393; RRID:BDSC_34393
Genetic reagent (D. melanogaster)tsl4Bloomington Drosophila Stock CenterBDSC: 3289; RRID:BDSC_3289
Genetic reagent (D. melanogaster)UASp-torsoDegPae et al., 2017Maintained in the lab of R. Lehmann
Genetic reagent (D. melanogaster)MEKE203KGoyal et al., 2017Maintained in the lab of S. Shvartsman
Genetic reagent (D. melanogaster)MEKF53SGoyal et al., 2017Maintained in the lab of S. Shvartsman
Genetic reagent (D. melanogaster)UAS-egfp RNAi (VALIUM20)Bloomington Drosophila Stock CenterBDSC: 41552; RRID:BDSC_41552
AntibodyAnti-Vasa (rat polyclonal)Paul LaskoRRID:AB_2568498Used 1:1000
AntibodyAnti-Vasa (mouse monoclonal)Developmental Studies Hybridoma BankDSHB: 46F11; RRID:AB_10571464Used 1:15
AntibodyAnti-Sxl (mouse monoclonal)Developmental Studies Hybridoma BankDSHB: M18; RRID:AB_528464Used 1:10
Sequence-based reagentpgcEagle et al., 2018smFISH probe setExonic probes
Sequence-based reagentgclEagle et al., 2018smFISH probe setExonic probes
Sequence-based reagentnosEagle et al., 2018smFISH probe setExonic probes
Sequence-based reagentSxlThomas GregorsmFISH probe setIntronic probes
Sequence-based reagentsis-bThomas GregorsmFISH probe setIntronic probes
Sequence-based reagentrunThomas GregorsmFISH probe setIntronic probes
Sequence-based reagenttllBiosearch Technologies; this papersmFISH probe setExonic probes; sequences available in Supplementary file 1
OtherHoeschtInvitrogenFisher Scientific: H3570

Fly stocks and genetics

Request a detailed protocol

The following fly stocks were used for the analysis reported in this manuscript. white1 (w1) was used as the WT stock. gcl, a null allele, and gcl-bcd-3’UTR stocks were generous gifts from Jongens et al., 1994; Jongens et al., 1992. tsl4 (BDSC #3289), a loss-of-function mutation, was obtained from Eric Wieschaus. egfp RNAi (BDSC #41552), UAS-Sxl (Helen Salz - DB106), and MEK gain-of-function transgenic stocks MEKE203K and MEKF53S (gift of Stas Shvartsman, Goyal et al., 2017) were driven by maternal-tubulin-GAL4 (67.15) driver stock, which carries four copies of maternal-tubulin-GAL4 (gift from Eric Wieschaus). The nosGAL4-VP16 driver (BDSC #7303) was also used. UAS-torsoDeg flies were kindly provided by Ruth Lehmann (Pae et al., 2017). The Sxl deficiency line, Sxl7BO, was a gift from Tom Cline.

Immunostaining

Request a detailed protocol

Embryos were formaldehyde-fixed, and a standard immunohistochemical protocol was used for DAB staining as described previously (Deshpande et al., 1999). Fluorescent immunostaining employed fluorescently labeled (Alexa) secondary antibodies. The primary antibodies used were mouse anti-Vasa (1:10, DSHB, Iowa City, IA), rat anti-Vasa (1:1000, gift of Paul Lasko), mouse anti-Sex lethal (1:10, DSHB M18, Iowa City, IA), and rabbit anti-Centrosomin (1:500, gift from Thomas Kaufmann). Secondary antibodies used were Alexa Fluor goat anti-rat 488 or 546 (1:500, ThermoFisher Scientific, Waltham, MA) and Alexa Fluor goat anti-rabbit 647 (1:500, ThermoFisher Scientific, Waltham, MA), DAPI (10 ng/mL, ThermoFisher Scientific, Waltham, MA), and Hoescht (3 µg/mL, Invitrogen, Carlsbad, CA). Stained embryos were mounted using Aqua Poly/mount (Polysciences, Warrington, PA) on slides. At least three independent biological replicates were used for each experiment.

Single molecule fluorescent in situ hybridization smFISH was performed as described by Little and Gregor using formaldehyde-fixed embryos (Little et al., 2015; Little and Gregor, 2018). All probe sets were designed using the Stellaris probe designer (20-nucleotide oligonucleotides with 2-nucleotide spacing). pgc, gcl, and nanos smFISH probes (coupled with either atto565 or atto647 dye, Sigma, St. Louis, MO) were a gift from Liz Gavis (Eagle et al., 2018), and Sxl, sis-b, and runt intronic probes (coupled with either atto565 or atto633 dye, Sigma, St. Louis, MO) were a gift from Thomas Gregor. tll probes (coupled with Quasar 570) were produced by Biosearch Technologies (Middlesex, UK). All samples were mounted using Aqua Poly/mount (Polysciences, Warrington, PA) on slides. At least three independent biological replicates were used for each experiment.

Statistical analysis

Request a detailed protocol

For smFISH experiments, total number of embryos expressing sis-b, runt, or Sxl in PBs/PGCs were counted, and pairwise comparisons of the proportion of embryos positive for transcription in PBs/PGCs or proportion of male embryos expressing Sxl in the soma were performed using Fisher’s exact test. Sex bias in gcl and gcl;SxlRNAi embryos was analyzed by comparing proportions also using Fisher’s exact test. To calculate significant differences in number of embryos displaying Sxl expression in pole cells or reduced at the anterior from ectopic gcl expression (based on DAB-visualization), we used Welch’s two sample t-test. Using NC13/14 embryos, PGCs were counted from the first Vasa-positive cell to the last through an entire z-volume captured at 1-micron intervals. Rescue in gcl;tsl embryos was analyzed either using Fisher’s exact test for proportions of embryos showing PGC transcription or a one-way ANOVA with pairwise t-test comparisons for pole cell counts. Data were plotted and statistical analyses were performed using Microsoft Excel, R Project, or GraphPad Prism software. For the Sxl RNAi rescue experiment, data were analyzed by Student’s two-tailed t-test or a nonparametric Mann–Whitney U-test and are displayed as mean ± SD. Data shown are representative results from at least two independent biological replicates.

Microscopy and image analysis

Request a detailed protocol

A Nikon-Microphot-SA microscope was used to capture images of DAB-stained embryos (40×). Images for the Sxl RNAi rescue experiment were acquired using a 100×, 1.49 NA Apo TIRF oil immersion objective on a Nikon Ti-E system fitted with a Yokagawa CSU-X1 spinning disk head, Hamamatsu Orca Flash 4.0 v2 digital CMOS camera, and Nikon LU-N4 solid state laser launch. Imaging for all other smFISH and fluorescent immunostaining experiments was performed on a Nikon A1 inverted laser-scanning confocal microscope.

Images were assembled using ImageJ (NIH) and Adobe Photoshop and Illustrator software to crop regions of interest, adjust brightness and contrast, generate maximum-intensity projections, and separate or merge channels. To assess the spreading of the RNAs or protein in different mutant backgrounds compared to the control we generated plot profiles using ImageJ. The posterior-most 75 µm of each embryo was plotted for comparison, and embryos from a single biological replicate are plotted in figures given that variation between fluorescence between replicates obscured the pole plasm distribution trends if embryos from all replicates were plotted together.

References

    1. Nili E
    2. Cojocaru GS
    3. Kalma Y
    4. Ginsberg D
    5. Copeland NG
    6. Gilbert DJ
    7. Jenkins NA
    8. Berger R
    9. Shaklai S
    10. Amariglio N
    11. Brok-Simoni F
    12. Simon AJ
    13. Rechavi G
    (2001)
    Nuclear membrane protein LAP2β mediates transcriptional repression alone and together with its binding partner GCL (germ-cell-less)
    Journal of Cell Science 114:3297–3307.
    1. Pignoni F
    2. Steingrímsson E
    3. Lengyel JA
    (1992)
    Bicoid and the terminal system activate tailless expression in the early Drosophila embryo
    Development 115:239–251.
    1. Seydoux G
    2. Dunn MA
    (1997)
    Transcriptionally repressed germ cells lack a subpopulation of phosphorylated RNA polymerase II in early embryos of Caenorhabditis elegans and Drosophila melanogaster
    Development 124:2191–2201.

Decision letter

  1. Michael B Eisen
    Senior and Reviewing Editor; University of California, Berkeley, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The paper reconciles conflicting models for the role of the Drosophila protein Germ cell-less (Gcl) in the formation of primordial germ cells. The authors provide compelling evidence that the protein establishes/maintains germ-line cells in a quiescent state by interfering with the ability of Torso receptor signaling to activate transcription of a number of genes, the sex determination gene Sex-lethal being the main focus here. This work also demonstrates a role for Gcl and the inhibition of Torso signaling in the proper localization of Drosophila germ plasm, suggesting a previously unappreciated role for Torso signaling in the organization or function of the cytoskeleton. This paper will be of specific interest to investigators who study germ and stem cell formation, and more broadly, to those concerned with the question of how receptor-mediated signaling, and more specifically, signaling via receptor tyrosine kinases, influences transcription and other cellular processes.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Antagonism between Germ cell-less and Torso regulates transcriptional quiescence underlying germline/soma distinction" for consideration by eLife. Your article has been reviewed by a Senior Editor and two reviewers. The reviewers have opted to remain anonymous. Our decision has been reached after consultation between the reviewers.

Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife. As you will see the reviewers raise issue, some of which I will call out here:

1) They do not feel that the experiments convincingly establish hypothesis set up about the work of Pae et al.

2) They found the subtlety of the quantitative results to be inconsistent with a simple explanation.

3) After detailed review, they judge that the work will be of more interest to specialists than to a more general readership. It may therefore be more appropriate for Development.

4) In several places they (and I) found the manuscript hard to follow. Suggestions to address this are included in the reviews.

Reviewer #1:

This paper helps resolve heretofore confusing data on the roles of pole plasm components and the terminal (torso) signaling pathway on germ cell specification, transcriptional quiescence, and maintenance. Specifically, the work helps elucidate the role of the pole plasm component germ cell less (gcl). The topic is a good contribution to the field, although of most interest to readers interested in Drosophila pole plasm and germ cell specification. One consideration is the method of quantification. The authors score the changes with genotype on a per embryo basis, rather than on a per pole bud/cell basis. However, the number of pole cells formed per embryo varies with the different genotypes assessed. In at least some cases the effect of this means that the effects are likely even stronger than the authors state and the authors are being conservative. However, these points, the rational for the quantification methods used, and whether they might in some other cases overestimate the effects of the genetic tests are not adequately discussed.

1) Abstract: It is hard to grasp the proposed regulatory pathway from the string of contradictions and observations listed. This is made worse by ambiguous statements. For example, does "like gcl" mean loss of function or forced overexpression? The two suggest opposite signs in terms of regulation. A simple statement of the proposed regulatory pathway suggested by the new data would help the reader and could replace the last sentence, which has little information content.

2) Introduction: Please clarify here at which nuclear cycle nuclei migrate into the pole plasm. In that context, is very early transcriptional activation (nuclear cycles 6-7) also turned on in nuclei destined for the germ line? If so is that early transcription turned off in the nuclei once they reach the pole plasm? Alternatively, is "early" transcription always kept off in nuclei that will come to inhabit the germ cells? IE: is transcription "switched off" as stated, or is it always maintained in an off state in the nuclei that will come to inhabit the germ plasm?

3) Introduction, in several places: Should "gcl mutant embryos" instead be "embryos derived from gcl mutant mothers"? The authors should take care with the wording and the labeling of figures throughout in this regard.

4) Introduction: The authors say the "…number correlated well…". However, it is actually reciprocal. Embryos from gcl homozygous mutant mothers had a high percent of pole bud nuclei with phosphorylated CTD but a low number of pole cells compared to wild type.

5) Subsection “Gcl represses the expression of XCEs in nascent PGCs”: Please define XCE at first usage for those not in the immediate field. I suggest that the authors move the first paragraph of the Results into the Introduction. Since Sxl-PE is only expressed in early female embryos, do the effects of having a gcl mutant mother on firing of Sxl-PE only show up in female but not in male embryos?

6) Figure 1 and Subsection “Gcl represses the expression of XCEs in nascent PGCs”: To convince the reader that sisb and Sxl are being newly transcribed the authors should make clear that they are only scoring hybridization signal in dots in the nucleus. Were the data shown and scored from an intronic probe? That would be best. In stating the quantitation of 67% and 42%, the authors say embryos. Was signal only observed in one or two pole cells per embryo, as the figures shown suggest? Should the authors also give numbers for what percent of pole buds and pole cells showed signal (scoring over a number of embryos)? Especially since the actual number of pole buds/pole cells per embryo is different in the different genotypes. This is a point for the quantitation in many of the figures. At least the authors should lay out the rational for counting on a per embryo vs. a per pole cell basis, and discuss how the differences in number of germ cell precursors per embryo in the different genotypes might effect the meaning of the counts.

7) Subsection “Sxl RNA is detected in gcl pole buds and PGCs”: The authors should give the frequencies (on a per cell/bud basis) for male vs female and indicate if the numbers are statistically significantly different.

8) Subsection “Sxl RNA is detected in gcl pole buds and PGCs”: Here the numbers given are for PGC nuclei rather than for embryos. That is good. But the authors should state how many different embryos were assessed and wether all all PGCs in each embryo were counted. Were the counts per section or obtained by focusing up andf down on a whole mount and so counting all PGCs is each embryo.

9) Subsection “Ectopic expression of gcl represses Sxl”: Were only female embryos scored? Why did obly 9/13 embryos show reduction of Sxl at the anterior? How did the authors ascertain on an embryo by embryo basis that reduction of Sxl in the anterior was coincident with expression of gcl from the gCl-bcd 3"UTR transgene? Did the 4 embryos that did not show reduction of anterior Sxl also not express gcl at the anterior? The result would be more convincing if the authors showed co-staining for expression of Gcl protein in the same embryos.

10) Subsection “Premature expression of Sxl in the PGCs leads to germ cell loss and defective germ cell migration”: I thought that Lehmann showed that nosGal4-VP16 did not drive transcription in early pole cells? The authors should show data backing up their statement that the number of PGCs was reduced in nosGal4-VP16; UAS-Sxl early embryos. That reduction is not apparent in the later embryos shown in Figure 4. How early does the ectopic Sxl become expressed under the conditions used?

11) Subsection “Simultaneous removal of gcl and Sxl ameliorates the gcl phenotype” and Figure 5: The data would be more convincing if the authors had some independent way to classify male vs. female embryos, other than expressing or not expressing Sxl. Could they use a balancer with GFP or some other way to distinguish male control with intact Sxl from female embryos with Sxl deleted?

12) Subsection “A degradation-resistant form of Torso also activates transcription in PGCs”, and data in Figure 7: Again, please clarify the quantitation. Does 27% of torsodeg embryos express Sisb meant that all PGCs in that positive embryo have Sisb or at least one Pgc per embryo? The two have very different implications as to the strength of the effect. Might it be better to give counts on the percent of PGCs that score positive (counting at least 10 different embryos). What are the corresponding numbers for the wild type controls in this experiment? Panel A (wt) shows that some PGCs express Sisb. How significant is the difference between control and experimental for each?

13) Discussion: Could the effects on transcription be secondary to a failure to confine pole plasm components to the pole cell buds?

Reviewer #2:

Past work suggests that the role of the Drosophila Germ Cell-Less (GCL) is to maintain primordial germ cells in a transcriptionally quiescent state. Pae et al., 2017 found that GCL mediates degradation of the Torso RTK, whose ectopic activity perturbs pole cell development. Colonetta et al., confirm that some somatic genes are ectopically activated in gcl mutant pole buds/cells and that ectopic GCL can inhibit expression of some genes in somatic cells, as can a mutant version of Torso (from Pae et al.,) that is not degraded by GCL. Contradicting Pae et al., Colonetta et al., argue that Torso's effects on PGCs require MAP Kinase pathway signaling. They also describe a novel phenotype in Torso[deg], gcl mutant, and MEK gain-of-function MEK embryos where pole plasm components get mis-localized. My criticisms are as follows: Several of the experiments contradicting/correcting Pae et al., are not definitive. Moreover, even if correct, the findings refine the model of Pae et al., without providing much new mechanistic insight. While of interest to Drosophila workers, particularly ones with an interest in early patterning and primordial germ cells, this manuscript might be considered esoteric by others who may also find reading difficult, owing to some poor writing.

1) The authors clearly show a number of genes, particularly ones involved in sex determination (sis-b, runt, Sxl), are expressed ectopically in polar buds/primordial germ cells, in the absence of GCL. While the authors add some additional genes to the list, this phenomenon has previously been demonstrated by these and other authors, and as noted in this manuscript, although the use here of smFISH to detect nascent transcripts is a nice touch.

2) The authors express gcl ectopically at the anterior of the embryo, showing that this leads to a local reduction in the expression of Sxl and use this finding to argue that GCL inhibits transcription of specific targets, and is sufficient to do so, "even in the absence of other pole plasm components," presumably including Torso. However, Torso is active in this domain and if Torso does lead to the activate of transcription Sxl via its effects upon sis-a, sis-b, runt, then this might be the expected result. If indeed the effect of Torso on pole cells is transcriptional, then Torso may be the sole target of GCL. The definitive experiment to answer this question has not been carried out by these authors and that is to assess the transcription of sis-a, sis-b, runt, Sxl, in pole buds/cells of embryos that are both null mutants for gcl and torso. If all of those genes failed to be expressed in the double mutant background, this would provide support for the notion that Torso acts to mediate the transcription of those genes, and might be the sole target of GCL. If any of those genes were expressed, that would argue that GCL can act to silence at least some genes, independent of Torso. Given the close chromosomal linkage of gcl and torso, it would be difficult, though not impossible, to obtain double mutants to test. It is certainly possible that with a concerted effort, these double mutants could be isolated and examined within a couple of months. Although the gcl tor/gcl tor double mutant would be ideal, the gcl/gcl; tsl/tsl strain might suffice, although we really do not have a complete picture of Tsl's role in the activation of Torso.

3) With regard to a transcriptional effect of Torso on pole buds/cells, the authors subtly misrepresent Pae et al. to set up a straw man in the statement that Pae et al., "proposed that activated Torso must inhibit PGC via distinct non-canonical mechanism that is both independent of the standard signal transduction and does not involve transcriptional activation." Although Pae et al., do state at several points in their paper that this effect of Torso is not transcriptional in nature, at the end of the manuscript they acknowledge that "additional downstream pathways (to MAPK/ERK) may require ligand-induced Torso receptor activation. Candidates include the JAK/STAT pathway, which is activated in dominant gain-of-function alleles of Torso (Li et al., 2002)." This would clearly be a transcriptional effect. Moreover, there is good reason to suspect that STAT may play a role in Torso's effects on pole buds/cells. First, there are several other contexts in which RTK's activate STAT. More to the point, however, the JAK/STAT pathway has been implicated in the control of Sxl transcription (Sefton et al., 2000), the mRNAs encoding the JAK/STAT pathway components are maternally loaded into the embryo, and Torso-dependent activation of STAT and Ras at an early step in PGC development has been shown to be required for their proliferation and migration (Li et al., 2003). Some of these reports should be of interest to Colonetta et al.

4) The section of the paper on "Simultaneous removal of gcl and Sxl ameliorates the gcl phenotype" is poorly written, probably impenetrable to anyone outside of the community of Drosophila workers studying early patterning events. The discussion of the expected progeny and the interpretation of the results are altogether unclear. Similarly, the experiment in which RNAi directed against Sxl is performed in the absence of GCL reveals an unexpected bimodal distribution of almost normal and pole cell-lacking embryos, for which no potential explanation is offered. In the absence of a credible potential explanation of this result, the conclusion that the loss of SXL is suppressing the lack of GCL is not convincing.

5) To ask the question of whether Map Kinase signaling downstream of Torso activity is required for Torso's effect upon pole buds/cells, Pae et al., performed RNAi against dsor1 (MEK) and rolled (MAPK) in a gcl null background. The failure of the RNAi treatment to suppress the effect of GCL loss on pole cells, was used as evidence that MAPK signaling downstream of Torso was not required for the Torso-dependent effects on pole buds/cells. To address this question, Colonetta el al., expressed activated MEK in mothers and assessed Sxl expression. Consistent with Pae et al., activated MEK failed to drive Sxl expression. However, they did observe Sxl expression in the somatic cells of some males. Given that this is a contrived and very different situation than pole cells, and that Pae's experiments in pole cells tested for requirement, while Colonetta's tested for sufficiency (apples to oranges) the question of MAPK signaling in Torso's effect upon pole buds/cells remains unanswered and it is baffling why Colonetta et al., did not examine sis-a, sis-b, runt, Sxl expression in the pole buds/cells of gcl mutant, dsor1 (MEK) and rolled (MAPK) knockdown embryos.

6) Figure 9, Figure 10, Figure 11, Figure 12, and Figure 13 show plot profiles of various pole plasm components (Vasa, gcl RNA, pgc RNA, nos RNA) in individual wild-type, gcl mutant and Torso-deg embryos. While the photographic images of the pole plasm components are consistent with the interpretations proposed by the authors, the plot profiles, which apparently only show a subset of the examined embryos for each component/genotype are more effective in displaying embryo to embryo variation, that they are in showing consistent patterns of mis-localization associated with the pole plasm components in the three genetic backgrounds.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Antagonism between Germ cell-less and Torso regulates transcriptional quiescence underlying germline/soma distinction" for further consideration by eLife. Your revised article has been evaluated by Michael Eisen as the Senior and Reviewing Editor.

Following your appeal, the manuscript was re-evaluated. We appreciate that you have addressed the earlier critiques, and made the manuscript significantly clearer. We also note the addition of experiments showing that the transcription of Sxl and sis-b elicited in the absence of Gcl is suppressed by the elimination of Torso activation, providing additional support for their conclusions regarding the connection between Gcl and Torso.

It is our judgment now that the manuscript likely warrants publication in eLife if you can address the following remaining concerns raised by one of the reviewers:

1) The distinction between terminal and non-terminal nuclei with respect to the ability of ectopic anterior Gcl to inhibit Sxl transcription should be explained more fully. All of the instances of ectopic Sxl transcription reported here occur in situations in which Torso or MEK signaling are occurring. That being the case it is not possible to conclude unambiguously that Gcl is uniquely able to suppress Sxl transcription. It's ability to suppress Sxl transcription may rely on Sxl transcription having been activated by ectopic Torso or MEK activity. While I would not require the authors to carry out additional experiments in which Gcl is expressed outside of the terminal regions of wild-type female embryos, or even better, in Torso-lacking female embryos, I believe that it is at least essential to discuss this issue and how it affects the conclusion that Gcl is uniquely able to suppress Sxl transcription. On the other hand, the results of examining Sxl transcription in Torso mutant-derived female embryos ectopically expressing Gcl might prove illuminating, if the authors are willing to undertake the effort.

2) The means of expression of the various transgenic construct used should be explained more clearly with respect to whether the constructs are expressed in the female under Gal4-mediated control and deposited in the embryo versus maternal deposition of Gal4 in the embryo and zygotic expression of the gene that has been crossed in. Similarly, the identities of the transgenic constructs, UASt, UASp, or UASother-based, should be noted.

3) The section in the Discussion on the effects of activated MEK, and of the ERK/MAPK cassette on Sxl transcription in PBs/PGCs versus somatic cells should be modified with additional consideration of the possibility that an alternative pathway downstream of Torso does exist, possibly working independent of the MAPK/ERK pathway. In modifying the text, see the more detailed consideration of this possibility in the Public Review, which outlines my reasoning.

https://doi.org/10.7554/eLife.54346.sa1

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1:

This paper helps resolve heretofore confusing data on the roles of pole plasm components and the terminal (torso) signaling pathway on germ cell specification, transcriptional quiescence, and maintenance. Specifically, the work helps elucidate the role of the pole plasm component germ cell less (gcl). The topic is a good contribution to the field, although of most interest to readers interested in Drosophila pole plasm and germ cell specification. One consideration is the method of quantification. The authors score the changes with genotype on a per embryo basis, rather than on a per pole bud/cell basis. However, the number of pole cells formed per embryo varies with the different genotypes assessed. In at least some cases the effect of this means that the effects are likely even stronger than the authors state and the authors are being conservative. However, these points, the rational for the quantification methods used, and whether they might in some other cases overestimate the effects of the genetic tests are not adequately discussed.

We realize that the rationale behind the quantitation was not adequately explained in the earlier version. As the reviewer correctly pointed out, we were concerned about the differences between the number of pole buds/cells per embryo among the different genotypes. It is important to note that smFISH technique reports on the stochastic nature of early embryonic transcription and as a result, it is possible that the effects are underestimated. Nonetheless, our observations are statistically highly significant, and the conclusions are robust in all cases. Please see our response to specific point #6 for a more detailed description of our reasoning. We have also included the rationale for this method of quantification in the text in subsection “Sxl RNA is detected in gcl pole buds and PGCs”.

1) Abstract: It is hard to grasp the proposed regulatory pathway from the string of contradictions and observations listed. This is made worse by ambiguous statements. For example, does "like gcl" mean loss of function or forced overexpression? The two suggest opposite signs in terms of regulation. A simple statement of the proposed regulatory pathway suggested by the new data would help the reader and could replace the last sentence, which has little information content.

We have edited the Abstract suitably to clarify several points. We have specifically used the phrase “embryos maternally compromised for germ cell-less” to explain the maternal effect of gcl mutation. We have also mentioned this explicitly in the text.

2) Introduction: Please clarify here at which nuclear cycle nuclei migrate into the pole plasm. In that context, is very early transcriptional activation (nuclear cycles 6-7) also turned on in nuclei destined for the germ line? If so is that early transcription turned off in the nuclei once they reach the pole plasm? Alternatively, is "early" transcription always kept off in nuclei that will come to inhabit the germ cells? IE: is transcription "switched off" as stated, or is it always maintained in an off state in the nuclei that will come to inhabit the germ plasm?

There is no reason to believe that specific nuclei are determined to acquire germ cell fate. Exposure and sustained association with the germ plasm components are sufficient to confer pole cell identity. Ephrussi and Lehmann, (1992) demonstrated that ectopic localization of Oskar is sufficient to form functional pole cells in the anterior (Osk-Bcd3’UTR). Importantly, Paul Macdonald and colleagues showed that after heat shock an hsp-70 promoter-driven oskar transgene can form germ cells at random ectopic locations (Ha et al., 1992). These observations demonstrate that, in principle, any somatic nucleus can acquire germ cell fate upon sustained exposure to germ plasm.

Nuclei enter the germ plasm around nuclear cycle 9-10. There is a low level of transcription across the embryo at this stage including the X chromosome counting elements (but not Sxl-Pe). Consequently, when they migrate to the germ plasm, nuclei are “transcriptionally” active; however, transcription is then shut down by germ plasm components including gcl, pgc, and nos. Leatherman and Jongens, (2000) have suggested that Gcl activity is especially crucial in this regard. This is clarified in the text.

3) Introduction, in several places: Should "gcl mutant embryos" instead be "embryos derived from gcl mutant mothers"? The authors should take care with the wording and the labeling of figures throughout in this regard.

We refer to the embryos derived from homozygous gcl mothers as “gcl”. Pole cells or pole buds of gcl embryos are also denoted as such. We have edited the text and the legends carefully to correct this. We explicitly state this early in the Introduction.

4) Introduction: The authors say the "…number correlated well…". However, it is actually reciprocal. Embryos from gcl homozygous mutant mothers had a high percent of pole bud nuclei with phosphorylated CTD but a low number of pole cells compared to wild type.

We thank the reviewer for pointing this out and have corrected the text as recommended.

5) Subsection “Gcl represses the expression of XCEs in nascent PGCs”: Please define XCE at first usage for those not in the immediate field. I suggest that the authors move the first paragraph of the Results into the Introduction. Since Sxl-PE is only expressed in early female embryos, do the effects of having a gcl mutant mother on firing of Sxl-PE only show up in female but not in male embryos?

We have tried to provide adequate background about somatic sex determination pathway to make the text readily accessible to the readers.

Sex-specificity: gcl is a maternally deposited RNA, which is translated in a sex-non-specific manner. The mutant phenotype induced by maternal loss of gcl i.e. loss of pole buds/cells displays no obvious sex-specificity. It should be noted that in wild type embryos, Sxl-Pe transcription is not turned on either in male or female PGCs. Sxl-Pe is transcriptionally activated only in female somatic nuclei while it remains off in male somatic nuclei. Our results show that both male and female PGCs showed ectopic activation of Sxl-Pe transcription with a modest female specific bias. We have provided quantitation to document this observation. This is consistent with the fact that several of the known Gcl targets are X-linked, including sis-a, sis-b, runt and Sxl. The difference in gene dose plus the fact that sis-a, sis-b and runt are X-linked “numerators” likely results in higher (or more frequent) Sxl-Pe transcription in female embryos.

6) Figure 1 and Subsection “Gcl represses the expression of XCEs in nascent PGCs”: To convince the reader that sisb and Sxl are being newly transcribed the authors should make clear that they are only scoring hybridization signal in dots in the nucleus. Were the data shown and scored from an intronic probe? That would be best. In stating the quantitation of 67% and 42%, the authors say embryos. Was signal only observed in one or two pole cells per embryo, as the figures shown suggest? Should the authors also give numbers for what percent of pole buds and pole cells showed signal (scoring over a number of embryos)? Especially since the actual number of pole buds/pole cells per embryo is different in the different genotypes. This is a point for the quantitation in many of the figures. At least the authors should lay out the rational for counting on a per embryo vs. a per pole cell basis, and discuss how the differences in number of germ cell precursors per embryo in the different genotypes might effect the meaning of the counts.

This is a good point, and we have edited the text to clarify that we are only scoring embryos based on hybridization signal in PB/PGC nuclei. We have also explained our rationale for scoring embryos rather than PBs/PGCs of each embryo, which we settled on largely due to the stochastic nature of detection of transcription during early embryogenesis.

Somatic transcription during cycle 14, the major wave of ZGA, is characterized by bursting. Even at this stage where transcription is maximally induced, one can observe nuclei in which one or the other copy of a gene, which should be active, is not transcriptional engaged. Earlier, during nuclear cycles 9-11, during PGC formation, levels of transcription are significantly lower than during the major wave of ZGA. At these earlier stages, intervals between transcriptional bursts are expected to much greater, and in static images of ongoing transcription, one expects to see many fewer nuclei in which transcriptionally active genes can be detected. This is the case in WT somatic nuclei during this period, and it is also the case in gcl PBs and nascent PGCs: only a subset of the nuclei is actively engaged in transcription. The effects of bursting are reflected in smFISH experiments in which we probed for two different transcripts. In some cases, both genes are found to be active in the same nuclei, while in other nuclei, only one (or no) gene is active. In the case of sis-b, the transcription unit is only 1.4 kb. This means that there will be only a short delay between the end of the burst, and the disappearance of the transcript. For Sxl-Pe we have intron probes. Introns are spliced out co-transcriptionally, and the signal will disappear even though the gene might still be transcribed. Finally, for PBs, gcl embryos have only 1-3 buds as opposed to 5-6 in WT. This means that we can only score a few cells in each embryo. For these reasons, we scored the percent of embryos (WT, gcl, Torsodeg) in which ongoing transcription is detected. This is a fully robust test as sis-b and Sxl-Pe transcripts are not detected in WT PB or PGCs.

We have included a summary of this rationale in the appropriate section where smFISH experiments are first described (subsection “Sxl RNA is detected in gcl pole buds and PGCs”).

7) Subsection “Sxl RNA is detected in gcl pole buds and PGCs”: The authors should give the frequencies (on a per cell/bud basis) for male vs female and indicate if the numbers are statistically significantly different.

We have repeated this analysis and analyzed the data to assess the sex of the embryos by counting the number of dots of hybridization per nucleus. We were able to sex embryos using number of dots of Sxl-specific hybridization (Avila and Erickson, 2007). We also employed smFISH using sis-B probe as an additional marker of sexual identity for confirmation. A clear visualization of probe puncta in female vs. male somatic nuclei is presented in Figure 7 (newly added).

8) Subsection “Sxl RNA is detected in gcl pole buds and PGCs”: Here the numbers given are for PGC nuclei rather than for embryos. That is good. But the authors should state how many different embryos were assessed and wether all all PGCs in each embryo were counted. Were the counts per section or obtained by focusing up andf down on a whole mount and so counting all PGCs is each embryo.

All the pole cell nuclei per embryo were counted (n~15); we have included this information for each quantification.

9) Subsection “Ectopic expression of gcl represses Sxl”: Were only female embryos scored? Why did obly 9/13 embryos show reduction of Sxl at the anterior? How did the authors ascertain on an embryo by embryo basis that reduction of Sxl in the anterior was coincident with expression of gcl from the gCl-bcd 3"UTR transgene? Did the 4 embryos that did not show reduction of anterior Sxl also not express gcl at the anterior? The result would be more convincing if the authors showed co-staining for expression of Gcl protein in the same embryos.

As males do not express Sxl protein, Sxl staining is observed only in female embryos, so only female embryos have been scored, a point we now clarify in the figure legend/Methods. It should be noted that all female embryos expressing gCl-bcd-3’UTR showed some reduction in Sxl-specific staining. Also, all male embryos are devoid of Sxl and thus provided an internal negative control. We attempted to use the Gcl antiserum that was kindly provided by the Jongens lab. Unfortunately, there was considerable background staining and those data were omitted.

10) Subsection “Premature expression of Sxl in the PGCs leads to germ cell loss and defective germ cell migration”: I thought that Lehmann showed that nosGal4-VP16 did not drive transcription in early pole cells? The authors should show data backing up their statement that the number of PGCs was reduced in nosGal4-VP16; UAS-Sxl early embryos. That reduction is not apparent in the later embryos shown in Figure 4. How early does the ectopic Sxl become expressed under the conditions used?

We apologize if this was unclearly presented, and we have revised the corresponding legend to appropriately emphasize these relevant results. The nos-Gal4-VP16 driver was used to drive Sxl misexpression in the mid-to-late stages of embryogenesis. For the early embryos we used a maternal-tubulin Gal4 driver strain i.e. 67.15 that carries 4 copies of the insert and can drive robust expression. In our hands, substantial deposition of GAL4 protein is more effective than nos-Gal4-VP16 to partially overcome the transcriptional silencing in PGCs.

11) Subsection “Simultaneous removal of gcl and Sxl ameliorates the gcl phenotype” and Figure 5: The data would be more convincing if the authors had some independent way to classify male vs. female embryos, other than expressing or not expressing Sxl. Could they use a balancer with GFP or some other way to distinguish male control with intact Sxl from female embryos with Sxl deleted?

We have examined the rescued embryos using sis-b specific smFISH in addition to Sxl signal. sis-b served as an independent marker to assess if there is a sex-bias and whether female (XX) embryos are preferentially rescued, leading to a bipolar distribution. Since we examined blastoderm stage embryos, Sxl RNA in-situ allowed us to sex the embryos based on presence (female) or absence (male). We further confirmed the sexual identity by the presence of either two dots (2X i.e. female) or a single dot (1X i.e. male) of sis-b specific signal. We have included a representative example each of both a male and a female embryo illustrating this point in Figure 7 (newly added). This has allowed us to ascertain the genotype and the sex of the embryos unambiguously as recommended by the reviewer.

12) Subsection “A degradation-resistant form of Torso also activates transcription in PGCs”, and data in Figure 7: Again, please clarify the quantitation. Does 27% of torsodeg embryos express Sisb meant that all PGCs in that positive embryo have Sisb or at least one Pgc per embryo? The two have very different implications as to the strength of the effect. Might it be better to give counts on the percent of PGCs that score positive (counting at least 10 different embryos). What are the corresponding numbers for the wild type controls in this experiment? Panel A (wt) shows that some PGCs express Sisb. How significant is the difference between control and experimental for each?

Please see the response above for our rationale on quantifying embryos rather than individual buds/cells. We have clarified that only somatic nuclei in the WT embryos express sis-b as the nuclei in Figure 9A (renumbered from Figure 7) exhibiting transcription puncta are posterior somatic nuclei, not PBs.

13) Discussion: Could the effects on transcription be secondary to a failure to confine pole plasm components to the pole cell buds?

This is a good point and is in fact in agreement with our model that pertains to centrosome-mediated transport and sequestration is at the heart of establishment of germline/soma distinction in the early embryo. Future experiments will indeed focus on testing different aspects of the model (please refer to last two paragraphs of Discussion).

Reviewer #2:

Past work suggests that the role of the Drosophila Germ Cell-Less (GCL) is to maintain primordial germ cells in a transcriptionally quiescent state. Pae et al., 2017 found that GCL mediates degradation of the Torso RTK, whose ectopic activity perturbs pole cell development. Colonetta et al., confirm that some somatic genes are ectopically activated in gcl mutant pole buds/cells and that ectopic GCL can inhibit expression of some genes in somatic cells, as can a mutant version of Torso (from Pae et al.,) that is not degraded by GCL. Contradicting Pae et al., Colonetta et al., argue that Torso's effects on PGCs require MAP Kinase pathway signaling. They also describe a novel phenotype in Torso[deg], gcl mutant, and MEK gain-of-function MEK embryos where pole plasm components get mis-localized. My criticisms are as follows: Several of the experiments contradicting/correcting Pae et al., are not definitive. Moreover, even if correct, the findings refine the model of Pae et al., without providing much new mechanistic insight. While of interest to Drosophila workers, particularly ones with an interest in early patterning and primordial germ cells, this manuscript might be considered esoteric by others who may also find reading difficult, owing to some poor writing.

In our estimate the question of ‘significance’ raised by reviewer #2 is the most crucial. Reviewer #2 suggests that elucidating the role of gcl in germline/soma distinction is of little interest outside a subset of the fly community. This seems to us to be an unusually narrow view. We believe that the central question addressed in our manuscript namely “to understand how gcl functions in establishing germline/soma distinction in early Drosophila embryos” is of broad relevance to stem cell biologists and those interested in gene regulation and early development. For example, many other invertebrates and even vertebrates utilize localized determinants for germline specification, and must encounter somewhat similar problems in germline/soma distinction.

In addition, gcl homologs are not only found in mammals, but also the mouse mgCl-1 can rescue gcl mutants. Like the fly protein, mGCl-1 is associated with the nuclear matrix and it interacts directly with the inner nuclear membrane proteins LAP2β, Emerin, and MAN1. The LAP2β:mGCl-1 complex is reported to sequester E2F:D1 to the nuclear envelope, reducing E2F:D1 transcriptional activity. Although the mgCl-1 gene is not required for germline specification, it does have a role in spermatogenesis. It is also required for normal nuclear morphology in liver and endocrine pancreatic cells. Given the functional conservation of mammalian Gcl proteins, their localization to the nuclear envelope, and their role in transcriptional regulation (newly added lines in Introduction), it seems likely that interest in the functions of the Drosophila protein would not be restricted to the fly community.

Also, in this context, the mass spec analysis in Pae et al., of proteins pulled down by WT Gcl is of interest. Near the top of the list of co-immunoprecipitated proteins are two components of the ‘facilitates chromatin transcription’ (FACT) complex. The FACT subunit Spt16 is second on the list with a score of 332. The other FACT subunit, Ssrp, is seventh on the list with a score of 231. Obviously, the presence of two transcription elongation factors is intriguing as sequestration of FACT in the nuclear envelope by Gcl could provide an additional (even if only a backup) mechanism for shutting off ongoing transcription.

Reviewer #2 also suggests that our paper is only a refinement of previously published work from the Lehmann lab. On this point, we also disagree.

One significant issue is transcription. An important contribution of Cinalli and Lehmann is, that contrary to previous studies by the Jongens lab, it “establishes” that transcription is not only irrelevant to gcl function during PGC formation but apparently is also incorrect. In the section of that paper that focuses on transcription, the authors state this point explicitly:

“Thus, we conclude that Gcl does not inhibit Pol II dependent transcription during PGC formation.”

In the Introduction, Pae et al., cite this earlier work as evidence for a non-transcriptional role of Gcl in PGC formation stating:

“…other experiments indicated that the major function of GCL is likely independent of transcriptional regulation (Cinalli, 2012; Cinalli and Lehmann, 2013).”

This same point is reiterated in the Results section, postulating that there is an alternative non-transcriptional pathway downstream of Torso that must be blocked by Gcl. They state:

“This is consistent with previous findings that the PGC formation defect seen in gcl embryos cannot be rescued by global transcriptional inhibition (Cinalli and Lehmann, 2013).”

Our findings clearly contradict the first claim that gcl does not inhibit PolII transcription. Our findings also contradict the second claim that silencing transcription is not an important gcl function.

To begin with, the observation that silencing Sxl is an important function for gcl in PGC formation is clearly not a “refinement” of the conclusions drawn in Cinalli and Lehmann and subsequently in Pae et al. They concluded exactly the opposite—that there were no relevant gene targets that needed to be silenced in PBs/PGCs by gcl.

Second, the studies of Pae et al., suggest that one (if not the) target for gcl in PGC formation/specification is the terminal pathway receptor Torso. They showed that Gcl interacts with Torso to promote Cul3-dependent degradation. In the absence of gcl, Torso is not degraded in newly formed pole buds (PBs), and, by an unspecified mechanism (that is postulated to be distinct from the canonical Ras-Raf-MEK-MAPK signaling cascade), the presence of activated Torso disrupts PB cellularization/specification. If this model is (generally) correct, then a degradation-resistant form of Torso should mimic key gcl phenotypes. One of the gcl phenotypes observed by us and by the Jongens’ lab is transcriptional. gcl PB nuclei fail to shut off ongoing transcription of X-chromosome counting elements and inappropriately turn on transcription of the Sxl establishment promoter, Sxl-Pe. We have shown that PBs/PGCs in torsodeg embryos also express these genes. Thus, proteolysis-resistant Torso activates transcription in PBs/PGCs, including a key gcl transcriptional target. Again, this is not a “refinement” of the main conclusions of Pae et al., as this paper clearly stated that transcription isn’t relevant.

Finally, Pae et al., proposed that gcl mediated degradation of Torso blocked the activation of a non-canonical and non-transcriptional terminal signaling pathway. Since activated MEK didn’t turn-on transcription in PBs/PGCs, we can’t rule out the existence of a “non-canonical” Torso/Torso-like signal transduction cascade that is independent of MEK/ERK. However, in this case, this postulated MEK/ERK independent pathway must be responsible for inappropriately activating transcription in PBs/PGCs in gcl mutants. In particular, we find that removing torso-like not only rescues PGC formation in gcl embryos but also eliminates transcription in the gcl PBs/PGCs. This is also not a “refinement” of Pae et al.

A second relevant gcl phenotype, not discussed in Cinalli and Lehmann, or in Pae et al., is the misdistribution of the germ plasm. In WT, germ plasm (mRNAs and proteins) is initially anchored to the actin cortical cytoskeleton. When nuclei (or more specifically centrosomes) enter the posterior pole, this triggers the release of the germ plasm and its subsequent dynein/MT dependent transport and accumulation around PB nuclei. This process is disrupted in gcl embryos. As in WT, PB nuclei trigger the release of the germ plasm in gcl embryos; however, it is not properly distributed around the PB nuclei. While some germ plasm remains concentrated on the posterior side of the gcl PB nuclei, much of the germ plasm “escapes” into the surrounding soma. Our previously reported findings indicated that abnormalities in centrosome/MT behavior in gcl PBs are responsible not only for the loss of germ plasm but also for defects in PB cellularization. Significantly, we find that Torsodeg elicits a similar spreading of the germ plasm into the soma, suggesting that the failure to degrade Torso also impacts centrosomes/MTs in PBs. This possibility was not considered by Pae et al.

Pae et al., suggest that there is a novel non-canonical pathway downstream of Torso that interferes with PGC formation. While this is a possibility, MEK/ERK have been reported to impact MT function. Consistent with the idea that the canonical kinase cascade needs to be shut down in PBs by gcl, we find that activated MEK induces a similar spreading of germ plasm as is observed in gcl mutants and in Torsodeg. While this effect could be non-transcriptional, it nevertheless requires canonical terminal pathway components.

1) The authors clearly show a number of genes, particularly ones involved in sex determination (sis-b, runt, Sxl), are expressed ectopically in polar buds/primordial germ cells, in the absence of GCL. While the authors add some additional genes to the list, this phenomenon has previously been demonstrated by these and other authors, and as noted in this manuscript, although the use here of smFISH to detect nascent transcripts is a nice touch.

As the reviewer points out, the Jongens lab reported that several genes are inappropriately expressed in PGCs. However, the involvement of Gcl during establishment and/or maintenance of transcriptional quiescence was subsequently discounted by Cinalli and Lehmann, and Pae et al. This raised a critical question underlying the central function of Gcl: does it or does it not contribute to transcriptional quiescence required for germline/soma distinction? For this reason, it was important to determine which of the previously reported findings were correct. Our results confirmed findings from the Jongens lab. We extended these results by examining Sxl-Pe expression. Since we found that disrupting Sxl function partially rescued the PGC formation/specification defects in gcl embryos, an obvious prediction is that Sxl-Pe is inappropriately activated in gcl PB/PGCs. Thus, this experiment needed to be done as well.

2) The authors express gcl ectopically at the anterior of the embryo, showing that this leads to a local reduction in the expression of Sxl and use this finding to argue that GCL inhibits transcription of specific targets, and is sufficient to do so, "even in the absence of other pole plasm components," presumably including Torso. However, Torso is active in this domain and if Torso does lead to the activate of transcription Sxl via its effects upon sis-a, sis-b, runt, then this might be the expected result. If indeed the effect of Torso on pole cells is transcriptional, then Torso may be the sole target of GCL. The definitive experiment to answer this question has not been carried out by these authors and that is to assess the transcription of sis-a, sis-b, runt, Sxl, in pole buds/cells of embryos that are both null mutants for gcl and torso. If all of those genes failed to be expressed in the double mutant background, this would provide support for the notion that Torso acts to mediate the transcription of those genes, and might be the sole target of GCL. If any of those genes were expressed, that would argue that GCL can act to silence at least some genes, independent of Torso. Given the close chromosomal linkage of gcl and torso, it would be difficult, though not impossible, to obtain double mutants to test. It is certainly possible that with a concerted effort, these double mutants could be isolated and examined within a couple of months. Although the gcl tor/gcl tor double mutant would be ideal, the gcl/gcl; tsl/tsl strain might suffice, although we really do not have a complete picture of Tsl's role in the activation of Torso.

This is an excellent and very valuable suggestion. We thank the reviewer for emphasizing the importance of this experiment.

As the reviewer correctly pointed out, generating a recombinant between gcl and torso mutations will be arduous because of their proximity. However, Pae et al., used a gcl/gcl; tsl/tsl fly strain to demonstrate the rescue. We have successfully reproduced the results reported by Pae et al., i.e. embryos derived from gcl/gcl; tsl/tsl females showed a substantial rescue of the gcl PGC formation defects.

Not discussed in Pae et al., was whether the rescue of the PGC formation defects by eliminating torso-like in gcl embryos also results in the reestablishment of transcriptional quiescence. It does. Taken together with the finding that there is at least one gene, Sxl, that needs to be shutoff in PBs/PGCs, this observation provides strong support for the idea that one of the important functions of gcl in PGC formation is silencing transcription. Moreover, this function depends upon gcl promoting the degradation of the Torso receptor.

3) With regard to a transcriptional effect of Torso on pole buds/cells, the authors subtly misrepresent Pae et al. to set up a straw man in the statement that Pae et al., "proposed that activated Torso must inhibit PGC via distinct non-canonical mechanism that is both independent of the standard signal transduction and does not involve transcriptional activation." Although Pae et al., do state at several points in their paper that this effect of Torso is not transcriptional in nature, at the end of the manuscript they acknowledge that "additional downstream pathways (to MAPK/ERK) may require ligand-induced Torso receptor activation. Candidates include the JAK/STAT pathway, which is activated in dominant gain-of-function alleles of Torso (Li et al., 2002)." This would clearly be a transcriptional effect. Moreover, there is good reason to suspect that STAT may play a role in Torso's effects on pole buds/cells. First, there are several other contexts in which RTK's activate STAT. More to the point, however, the JAK/STAT pathway has been implicated in the control of Sxl transcription (Sefton et al., 2000), the mRNAs encoding the JAK/STAT pathway components are maternally loaded into the embryo, and Torso-dependent activation of STAT and Ras at an early step in PGC development has been shown to be required for their proliferation and migration (Li et al., 2003). Some of these reports should be of interest to Colonetta et al.

Our read of Cinalli and Lehmann as well as Pae et al. is somewhat different from that of the reviewers. According to these authors gcl’s role in transcriptional quiescence was, at best, inconsequential. Moreover, in some instances (see first quote above), they go farther and state that gcl has no role in downregulating transcription. Thus, when Pae et al., suggested that the JAK/STAT pathway might be one of the non-canonical pathways activated by Torso, they clearly did not mean that a JAK/STAT-dependent transcriptional output is relevant for PB/PGC formation/specification. Instead, they were likely referring to the reported defects in PGC mitosis, adhesion and migration when STAT activity is manipulated (Li and Li, 2003; Li et al., 2003).

As for the role of JAK/STAT in activating Sxl-Pe, the reviewer is correct: Sxl-Pe is a target of the JAK/STAT pathway. This was shown not only by Sefton et al., but also by Jinks et al., (2000) and Avila and Erickson, (2007). Jinks et al. showed that the effects of compromising the JAK/STAT pathway on Sxl-Pe transcription were only seen in the center of the embryos, while Sxl-Pe activity at the termini appeared to be normal.

The reason for the spatially restricted effects were investigated by Avila and Erickson, (2007). They also found that loss of zygotic upd and maternal hop (JAK) impacted Sxl-Pe in the center of the embryo, not at the termini. A somewhat different result was obtained in embryos lacking maternal mrl (STAT). Unlike upd or hop, a reduction in Sxl-Pe could be detected at the termini of the embryo when there was no maternal mrl (STAT). Based on the studies of Li and Li, and Lie et al., Avila and Erickson concluded that STAT activation by the Torso receptor likely accounted for the relative resistance of terminal somatic nuclei to a disruption in the upd-JAK dependent activation of Sxl-Pe.

However, there are other important findings in Avila and Erickson. First, the JAK/STAT pathway is not a key regulator of Sxl-Pe. To begin with, upd expression is first detected in nuclear cycle 13. By contrast, Sxl-Pe comes on in nuclear cycle 12 (according to Avila and Erickson, while we detected transcripts at nuclear cycle 11 using smFISH). Consistent with its expression pattern, upd mutations only impact Sxl-Pe activity in nuclear cycle 14, while earlier expression of Sxl-Pe resembles wild type. The same temporally restricted effects were observed for embryos lacking maternal hop (JAK) or mrl (STAT). Expression resembled wild type in nuclear cycles 11-13, while mutant embryos show defects in Sxl-Pe transcription only during nuclear cycle 14. Based on these findings, Avila and Erickson concluded that the JAK/STAT pathway is not needed for the initial activation of Sxl-Pe in the soma during nuclear cycles 11-13, but rather is required to sustain high levels of Sxl-Pe expression in nuclear cycle 14. Potentially supporting a role in maintaining transcriptional activity, but not the initial sex-specific activation of Sxl-Pe, Jinks et al., found that constitutively active JAK Kinase (Hop-Tum) does not induce Sxl-Pe in male somatic nuclei.

If the conclusions of Avila and Erickson are correct, activation of STAT in gcl PGCs is unlikely to be responsible for Sxl-Pe transcription in PB/PGC, let alone for the continued expression of other genes such as sis-b, sis-a, or runt. In addition, while Avila and Erickson found that upd/hop (JAK)/mrl (STAT) are only needed for Sxl-Pe activity in nuclear cycle 14, this is much later than the PB/PGC formation defects first manifest themselves in gcl embryos. It is also after the initial activation of Sxl-Pe in PB of gcl mutant embryos.

4) The section of the paper on "Simultaneous removal of gcl and Sxl ameliorates the gcl phenotype" is poorly written, probably impenetrable to anyone outside of the community of Drosophila workers studying early patterning events. The discussion of the expected progeny and the interpretation of the results are altogether unclear. Similarly, the experiment in which RNAi directed against Sxl is performed in the absence of GCL reveals an unexpected bimodal distribution of almost normal and pole cell-lacking embryos, for which no potential explanation is offered. In the absence of a credible potential explanation of this result, the conclusion that the loss of SXL is suppressing the lack of GCL is not convincing.

This section has been revised for clarity. Additionally, we have offered a plausible explanation for the bimodal distribution in the RNAi knockdown experiment. As we observed a modest female-specific bias with respect to ectopic activation of Sxl in early PGCs, we wondered if mitigating Sxl levels preferentially rescued PGCs in gcl female embryos as compared to males. As can be seen in Table 2 this turned out to be indeed correct. Altogether these data document a sex-specific nature of the rescue in the RNAi knockdown experiment and provide a reasonable explanation for its bimodal nature.

5) To ask the question of whether Map Kinase signaling downstream of Torso activity is required for Torso's effect upon pole buds/cells, Pae et al., performed RNAi against dsor1 (MEK) and rolled (MAPK) in a gcl null background. The failure of the RNAi treatment to suppress the effect of GCL loss on pole cells, was used as evidence that MAPK signaling downstream of Torso was not required for the Torso-dependent effects on pole buds/cells. To address this question, Colonetta el al., expressed activated MEK in mothers and assessed Sxl expression. Consistent with Pae et al., activated MEK failed to drive Sxl expression. However, they did observe Sxl expression in the somatic cells of some males. Given that this is a contrived and very different situation than pole cells, and that Pae's experiments in pole cells tested for requirement, while Colonetta's tested for sufficiency (apples to oranges) the question of MAPK signaling in Torso's effect upon pole buds/cells remains unanswered and it is baffling why Colonetta et al., did not examine sis-a, sis-b, runt, Sxl expression in the pole buds/cells of gcl mutant, dsor1 (MEK) and rolled (MAPK) knockdown embryos.

As noted above, the reviewer is correct in stating that activated MEK is unable to activate transcription. This is consistent with the idea that canonical downstream pathway components may not be involved in transcriptionalactivation. On this account, our data would potentially be in agreement with Pae et al. This was explicitly acknowledged in the text. We also pointed out that there were reasons to think that the levels of canonical terminal pathway signaling in PBs/PGCs might not be equivalent to that in torsoDeg or gcl mutants when GOF MEK proteins are ectopically expressed. If that were the case, Torso/Torso-like dependent transcriptional activation in gcl PBs/PGCs might actually be mediated by the canonical signaling pathway. This issue will require additional studies, beyond the scope of the present manuscript.

6) Figure 9, Figure 10, Figure 11, Figure 12, and Figure 13 show plot profiles of various pole plasm components (Vasa, gcl RNA, pgc RNA, nos RNA) in individual wild-type, gcl mutant and Torso-deg embryos. While the photographic images of the pole plasm components are consistent with the interpretations proposed by the authors, the plot profiles, which apparently only show a subset of the examined embryos for each component/genotype are more effective in displaying embryo to embryo variation, that they are in showing consistent patterns of mis-localization associated with the pole plasm components in the three genetic backgrounds.

We agree with the reviewer that these phenotypes are variable. The plots show embryo-to-embryo variability within a single genotype, but more importantly, they also serve to demonstrate between genotypes. For example, Figure 11 shows the distribution of Vasa is similarly skewed in gcl and Torsodeg mutants resulting in higher levels of Vasa far from the posterior pole. This response is absent in WT controls. However, the distinction between the WT and the experimental embryos is clear-cut. Also, these phenotypes for the first time establish that Torso receptor and the downstream components of the pathway can affect RNA/protein transport that is likely mediated by MT centrosomes. We believe that this is an important finding with implications that go far beyond Drosophila germ cell biology.

[Editors’ note: what follows is the authors’ response to the second round of review.]

It is our judgment now that the manuscript likely warrants publication in eLife if you can address the following remaining concerns raised by one of the reviewers:

1) The distinction between terminal and non-terminal nuclei with respect to the ability of ectopic anterior Gcl to inhibit Sxl transcription should be explained more fully. All of the instances of ectopic Sxl transcription reported here occur in situations in which Torso or MEK signaling are occurring. That being the case it is not possible to conclude unambiguously that Gcl is uniquely able to suppress Sxl transcription. It's ability to suppress Sxl transcription may rely on Sxl transcription having been activated by ectopic Torso or MEK activity. While I would not require the authors to carry out additional experiments in which Gcl is expressed outside of the terminal regions of wild-type female embryos, or even better, in Torso-lacking female embryos, I believe that it is at least essential to discuss this issue and how it affects the conclusion that Gcl is uniquely able to suppress Sxl transcription. On the other hand, the results of examining Sxl transcription in Torso mutant-derived female embryos ectopically expressing Gcl might prove illuminating, if the authors are willing to undertake the effort.

We have altered the text to account for both possibilities (please refer to the Discussion). We are not able to unambiguously attribute transcription inhibition to direct action by Gcl with our experiments. However, it is worth noting that in our experiment ectopically localizing gcl to the anterior (gCl-bcd-3’UTR), Sxl expression should be predominantly under the control of the canonical somatic pathway, namely activation by X-linked counting elements (Salz and Erickson, 2010), rather than a Torso-mediated pathway. While Torso is present at the anterior of the embryo, Sxl is activated throughout the soma, even in the absence of Torso in the middle of the embryo. Therefore, ectopic Gcl may be able to repress Sxl transcription activated by a Torso-independent pathway, though we are not able to confirm this since Torso is present in the anterior terminus of the embryo as well as the posterior terminus that we focus on for much of the manuscript. On the other hand, our results indicating that tsl mutations can rescue the PGC transcription phenotype of gcl mutants suggest that the PGC transcription we observe likely results from an inability of Gcl to directly inhibit the Torso pathway in the PGCs.

Unfortunately, torso null mutants are female sterile, making the proposed experiment extremely challenging (Schupbach and Wieschaus, 1989).

2) The means of expression of the various transgenic construct used should be explained more clearly with respect to whether the constructs are expressed in the female under Gal4-mediated control and deposited in the embryo versus maternal deposition of Gal4 in the embryo and zygotic expression of the gene that has been crossed in. Similarly, the identities of the transgenic constructs, UASt, UASp, or UASother-based, should be noted.

We have altered the text to add details of how transgenes were expressed throughout. Please see experimental descriptions in the Results for zygotic expression, combination of maternal and zygotic depletion for gcl and Sxl, respectively, zygotic expression, maternal depletion of both gcl and tsl, maternal expression for clarification on individual experiments. Details of how these genes were expressed are also in the legends of the figures.

Additionally, we have added details of the UAS constructs in the reagents table of Materials and methods. All UAS constructs are expressed in the germline since they are either UASp or VALIUM20 (which the BDSC website specifies works in both soma and germline).

3) The section in the Discussion on the effects of activated MEK, and of the ERK/MAPK cassette on Sxl transcription in PBs/PGCs versus somatic cells should be modified with additional consideration of the possibility that an alternative pathway downstream of Torso does exist, possibly working independent of the MAPK/ERK pathway. In modifying the text, see the more detailed consideration of this possibility in the Public Review, which outlines my reasoning.

We have modified the text, as suggested. Please see the Discussion for our statement considering alternative downstream pathways.

https://doi.org/10.7554/eLife.54346.sa2

Article and author information

Author details

  1. Megan M Colonnetta

    Department of Molecular Biology, Princeton University, Princeton, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation, Visualization, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5685-1670
  2. Lauren R Lym

    Department of Cell Biology, Emory University School of Medicine, Atlanta, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5039-2303
  3. Lillian Wilkins

    Department of Molecular Biology, Princeton University, Princeton, United States
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  4. Gretchen Kappes

    Department of Molecular Biology, Princeton University, Princeton, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Elias A Castro

    Department of Cell Biology, Emory University School of Medicine, Atlanta, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1439-5918
  6. Pearl V Ryder

    Department of Cell Biology, Emory University School of Medicine, Atlanta, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    Additional information
    ORCID: 0000-0003-3699-3633
  7. Paul Schedl

    Department of Molecular Biology, Princeton University, Princeton, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  8. Dorothy A Lerit

    Department of Cell Biology, Emory University School of Medicine, Atlanta, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Investigation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    dlerit@emory.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3362-8078
  9. Girish Deshpande

    Department of Molecular Biology, Princeton University, Princeton, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Investigation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    gdeshpan@princeton.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5200-7090

Funding

National Institute of General Medical Sciences (126975)

  • Paul Schedl

Eunice Kennedy Shriver National Institute of Child Health and Human Development (093913)

  • Paul Schedl
  • Girish Deshpande

National Heart, Lung, and Blood Institute (K22HL126922)

  • Dorothy A Lerit

National Institute of General Medical Sciences (138544)

  • Dorothy A Lerit

National Science Foundation (DGE-1656466)

  • Megan M Colonnetta

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

Acknowledgements

This work was supported by grants from National Institute of Health (NICHD:093913) to PS and GD, and (NIGMS: 126975) to PS. Work in the Lerit lab was supported by K22HL126922 and R01GM138544. MC was supported by NSF Graduate Research Fellowship (DGE-1656466).

We thank Dr. Gary Laevsky and the Molecular Biology Confocal Microscopy Facility which is a Nikon Center of Excellence. Gordon Grey provided fly media. We thank the Bloomington stock center for different fly lines. We gratefully acknowledge Liz Gavis and Ruth Lehmann for advice and reagents during the course of this work.

Senior and Reviewing Editor

  1. Michael B Eisen, University of California, Berkeley, United States

Publication history

  1. Received: December 11, 2019
  2. Accepted: January 15, 2021
  3. Accepted Manuscript published: January 18, 2021 (version 1)
  4. Version of Record published: January 28, 2021 (version 2)

Copyright

© 2021, Colonnetta et al.

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

Metrics

  • 538
    Page views
  • 80
    Downloads
  • 0
    Citations

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

Download links

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

Downloads (link to download the article as PDF)

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

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

Further reading

    1. Developmental Biology
    Feng Wang et al.
    Research Article Updated

    The X-linked gene Rlim plays major roles in female mouse development and reproduction, where it is crucial for the maintenance of imprinted X chromosome inactivation in extraembryonic tissues of embryos. However, while females carrying a systemic Rlim knockout (KO) die around implantation, male Rlim KO mice appear healthy and are fertile. Here, we report an important role for Rlim in testis where it is highly expressed in post-meiotic round spermatids as well as in Sertoli cells. Systemic deletion of the Rlim gene results in lower numbers of mature sperm that contains excess cytoplasm, leading to decreased sperm motility and in vitro fertilization rates. Targeting the conditional Rlim cKO specifically to the spermatogenic cell lineage largely recapitulates this phenotype. These results reveal functions of Rlim in male reproduction specifically in round spermatids during spermiogenesis.

    1. Developmental Biology
    2. Neuroscience
    Hiroki Takechi et al.
    Research Article

    Transmembrane protein Golden goal (Gogo) interacts with atypical cadherin Flamingo to direct R8 photoreceptor axons in the Drosophila visual system. However, the precise mechanisms underlying Gogo regulation during columnar- and layer-specific R8 axon targeting are unknown. Our studies demonstrated that the insulin secreted from surface and cortex glia switches the phosphorylation status of Gogo, thereby regulating its two distinct functions. Non-phosphorylated Gogo mediates the initial recognition of the glial protrusion in the center of the medulla column, whereas phosphorylated Gogo suppresses radial filopodia extension by counteracting Flamingo to maintain a one axon to one column ratio. Later, Gogo expression ceases during the midpupal stage, thus allowing R8 filopodia to extend vertically into the M3 layer. These results demonstrate that the long- and short-range signaling between the glia and R8 axon growth cones regulates growth cone dynamics in a stepwise manner, and thus shape the entire organization of the visual system.