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

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 LAP2b 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, a-amanitin. Since a-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 torso Deg 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.

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.
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 Figure 1. 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 mm. 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.
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.

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 gclbcd-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. 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).
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 maternaltubulin-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.

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, Sxl 7BO , which deletes the Sxl gene. In this experiment, we mated Sxl 7BO /Bin; gcl/gcl mothers to Sxl 7BO /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 Sxl 7BO /Bin, while the other half would be Sxl 7BO / Sxl 7BO . The former (Sxl 7BO /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 (Sxl 7BO / Sxl 7BO ) 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 Sxl 7BO /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, Sxl 7BO /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 Sxl 7BO /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 Sxl 7BO / Sxl 7BO females or Sxl 7BO /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 (Sxl 7BO /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 + (Sxl 7BO /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 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 Figure 3 continued on next page embryos), the mean number of PGCs is 12. Embryos in this group are presumably Sxlmales 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.
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 (Sxl RNAi ) partially suppresses the effects of the gcl mutation on PGCs. While all the embryos in this 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). The online version of this article includes the following source data for figure 3: Source data 1. Primordial germ cell (PGC) counts. 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). 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.
A plausible explanation for this bimodal distribution is that the efficiency of rescue reflects sexspecific 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, Figure 5. 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. The online version of this article includes the following source data for figure 5: Source data 1. Primordial germ cell (PGC) counts. Figure 6. 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 mm. 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;Sxl RNAi (B). In 66% of gcl;Sxl RNAi embryos, few or no pole cells are observed, comparable to gcl. However, in 34% of gcl;Sxl RNAi embryos, germ cell count is substantially elevated, indicating partial rescue of the gcl phenotype. The online version of this article includes the following source data for figure 6: Source data 1. Primordial germ cell (PGC) counts.
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).

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.) 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, torso Deg , 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-torso Deg 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 torso Deg embryos, with frequencies less than those Table 2. Rescue of PGC numbers in gcl;Sxl RNAi embryos only occurs in female embryos.

Male Female
No rescue 20 26 Rescue 0 13 Figure 8. 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 mm. (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. The online version of this article includes the following source data for figure 8: Source data 1. Primordial germ cell (PGC) counts.
observed in gcl embryos but significantly more than control embryos (27% of torso Deg embryos express sis-b (n = 16, p=0.043382) and 28% of torso Deg 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 torso Deg embryos. 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 torso Deg 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 torso Deg 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 torso Deg 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 torso Deg embryos resemble WT, we observed scattered nuclei in which Sxl-Pe is active in 43% of torso Deg 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).

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 (MEK E203K and MEK F53S 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 MEK E203K or MEK F53S could not activate Sxl-Pe transcription in PBs or PGCs (not shown, see Discussion). Nevertheless, we found that, as was observed for torso Deg , Sxl-Pe expression is activated in male somatic nuclei by the GOF MEK proteins ( Figure 10D, 46% of MEK E203K 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 Torso Deg 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. 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.

Genotype
No transcription Transcription p-value 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/microtubuledependent 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). Interestingly, as was the case for transcriptional activation, the effects of torso Deg on the sequestration of the pole plasm constituents are quite similar to those observed in gcl embryos. In early torso Deg 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) Figure 9 continued 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 torso Deg embryos. Scale bar represents 10 mm. Figure 10. Sxl is expressed in the male soma in torso Deg 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 torso Deg (C and C') and 46% (n = 13, p=0.019079) of MEK E203K (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 mm. mRNAs spread into the territories of somatic nuclei located along the posterior lateral cortex of torso Deg embryos. In addition, gcl mRNA ( Figure 12) is not properly restricted in torso Deg 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.

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 torso Deg 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, MEK E203K or MEK F53S protein induces the inappropriate dispersal of gcl and pgc mRNAs into the surrounding soma in a pattern very similar to that observed in torso Deg and gcl embryos. Thus, this gcl phenotype would appear to depend upon the canonical terminal signal transduction cascade. Figure 11. Vasa is mislocalized from the posterior in gcl and torso Deg embryos. 0-3 hr old paraformaldehyde-fixed embryos collected from wildtype (WT), gcl, or torso Deg mothers were stained with anti-Vasa antibody to assess whether pole plasm is properly localized in gcl and torso Deg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 mm. Below, plot profiles show mislocalization of pole plasm (visualized using Vasa) away from posterior cap in gcl and torso Deg 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 torso Deg , 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 Figure 12. gcl RNA is mislocalized from the posterior in torso Deg 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 torso Deg 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 mm. Below, plot profiles show mislocalization of pole plasm (visualized using gcl) away from posterior cap in torso Deg 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 torso Deg , respectively. 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 Figure 13. pgc RNA is mislocalized from the posterior in gcl and torso Deg 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 torso Deg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 mm. Below, plot profiles show mislocalization of pole plasm (visualized using pgc) away from posterior cap in gcl and torso Deg 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 torso Deg , respectively.
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 (torso Deg ) 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 Torso Deg 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- Figure 14. nos RNA is mislocalized from the posterior in gcl and torso Deg 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 torso Deg embryos. On top, images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 mm. Below plot profiles show mislocalization of pole plasm (visualized using nos) away from posterior cap in gcl and torso Deg 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 torso Deg , respectively. 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 torso Deg 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 Torsodependent 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 torso Deg , we found that MEK E203K induces Sxl-Pe expression in male somatic Figure 15. Before pole buds develop, pole plasm distribution is unaltered in gcl and torso Deg 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 torso Deg embryos. Images are representative maximum intensity projections of the posterior pole of each indicated genotype. Scale bar represents 10 mm. Below, plot profiles show proper anchoring and localization of pole plasm (visualized using pgc or gcl) at the posterior cap in gcl and torso Deg 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 torso Deg , respectively. For the gcl smFISH experiment, n = 14, 9, and 8 for WT, gcl, and torso Deg , respectively.
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 torso Deg and the two GOF MEK proteins. For these reasons, we cannot unambiguously determine if it is the Figure 16. 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 MEK E203K or MEK F53S 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 mm. 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), MEK E203K , and MEK F53S , respectively. 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 torso Deg embryos but also in MEK E203K and MEK F53S 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 torso Deg 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 torso Deg and GOF MEK embryos. While further experiments will be required to demonstrate microtubule and centrosomal aberrations in torso Deg and GOF MEK embryos, a role for a receptor-dependent MEK/ERK signaling cascade in promoting centrosome accumulation of g-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.

Statistical analysis
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;Sxl RNAi 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
A Nikon-Microphot-SA microscope was used to capture images of DAB-stained embryos (40Â). Images for the Sxl RNAi rescue experiment were acquired using a 100Â, 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 mm 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.
center for different fly lines. We gratefully acknowledge Liz Gavis and Ruth Lehmann for advice and reagents during the course of this work.