Transcripts immunoprecipitated with Sxl protein in primordial germ cells of Drosophila embryos

Abstract

In Drosophila, Sex lethal (Sxl), an RNA binding protein, is required for induction of female sexual identity in both somatic and germline cells. Although the Sxl‐dependent feminizing pathway in the soma was previously elucidated, the downstream targets for Sxl in the germline remained elusive. To identify these target genes, we selected transcripts associated with Sxl in primordial germ cells (PGCs) of embryos using RNA immunoprecipitation coupled to sequencing (RIP‐seq) analysis. A total of 308 transcripts encoded by 282 genes were obtained. Seven of these genes, expressed at higher levels in PGCs as determined by microarray and in situ hybridization analyses, were subjected to RNAi‐mediated functional analyses. Knockdown of Neos, Kap‐alpha3, and CG32075 throughout germline development caused gonadal dysgenesis in a sex‐dependent manner, and Su(var)2‐10 knockdown caused gonadal dysgenesis in both sexes. Moreover, as with knockdown of Sxl, knockdown of Su(var)2‐10 in PGCs gave rise to a tumorous phenotype of germline cells in ovaries. Because this phenotype indicates loss of female identity of germline cells, we consider Su(var)2‐10 to be a strong candidate target of Sxl in PGCs. Our results represent a first step toward elucidating the Sxl‐dependent feminizing pathway in the germline.

Introduction

Sex determination of the germline, which gives rise to sexually dimorphic and functionally distinct gametes (sperms and eggs), is a crucial step in the propagation of a species. It has been proposed that the sexual fate of the germline is regulated by both cell‐autonomous cues and the surrounding soma (Murray et al. 2010). For example, in Drosophila, primordial germ cells (PGCs) initiate male fate in response to JAK/STAT signaling from the gonadal soma (Wawersik et al. 2005; Casper & Van Doren 2006, 2009), whereas the female fate is induced cell‐autonomously by the function of Sex‐lethal (Sxl) in PGCs (Hashiyama et al. 2011). However, it remains unclear how Sxl initiates female germline development.

Sxl encodes an RNA binding protein that regulates sex‐biased translation and splicing (Penalva & Sánchez 2003). In the soma, Sxl is activated at the blastodermal stage in a female‐specific manner (Bopp et al. 1991; Keyes et al. 1992). One of the targets for Sxl is male‐specific lethal 2 (msl2) RNA (Gebauer et al. 1998). Sxl protein binds to 5′ and 3′ untranslated regions (UTRs) of msl2 RNA and represses its translation to block dosage compensation in female soma (Bashaw & Baker 1997; Kelly et al. 1997; Beckmann et al. 2005; Lucchesi & Kuroda 2015). Another target of Sxl is transformer (tra) RNA (Inoue et al. 1990). Sxl protein regulates alternative splicing of tra RNA to produce functional Tra protein only in the female, which in turn causes the expression of a female‐specific form of Double‐sex (Dsx), along with transformer‐2 (tra‐2) function (Penalva & Sánchez 2003; Camara et al. 2008). In male soma, a male‐specific form of Dsx is produced by default in the absence of Sxl function (Camara et al. 2008). The resulting sex‐specific forms of Dsx determine the sexual dimorphism of the soma (Penalva & Sánchez 2003; Camara et al. 2008). By contrast, in the germline, tra, tra2, and dsx functions are dispensable for sex determination (Marsh & Wieschaus 1978; Schüpbach 1982; Oliver 2002). Accordingly, it has been postulated that distinct pathways act downstream of Sxl to determine germline and somatic sexes.

In this study, we obtained a strong candidate target of Sxl in PGCs by RNA ImmunoPrecipitation coupled to sequencing (RIP‐seq), in situ hybridization, and knockdown analyses.

Materials and methods

Fly stocks

Flies were maintained on standard Drosophila medium at 25°C. nos‐Gal4‐VP16 (nos‐Gal4) (Van Doren et al. 1998) was used to express the following UAS constructs in the germline: UASp‐Flag‐EGFP‐Sxl and UASp‐Flag‐EGFP (referred to as UASp‐EGFP‐Sxl and UASp‐EGFP); from the Bloomington Drosophila Stock Center (BDSC), UAS‐Redstinger (8545), UAS‐Sxl^RNAi (34393), UAS‐Su(var)2‐10^RNAi‐1 (32915), UAS‐Su(var)2‐10^RNAi‐2 (32956), UAS‐Pp2B‐14D^RNA‐1 (25929), UAS‐Pp2B‐14D^RNAi‐2 (40872), UAS‐Neos^RNAi‐1 (55960), UAS‐Kap‐alpha3^RNAi (27535), UAS‐CG6364^RNAi‐1 (35351), UAS‐CG6364^RNAi‐2 (57244), and UAS‐CG10077^RNAi (32388); and from the Vienna Drosophila Resource Center (VDRC), UAS‐Dicer2 (60008), UAS‐Neos^RNAi‐2 (42145), UAS‐Neos^RNAi‐3 (108950), UAS‐CG32075^RNAi‐1 (19255), and UAS‐CG32075^RNAi‐2 (105415). da‐Gal4 (55851) from BDSC was used to express UASp‐Flag‐EGFP‐Sxl and UASp‐Flag‐EGFP. In maternal and zygotic knockdown analysis, we induced expression of multiple dsRNAs targeting distinct regions of each gene, except in the cases of Kap‐alpha3 and CG10077.

Transgenes

For construction of UASp‐Flag‐EGFP‐Sxl, the open reading frames (ORF) of early transcripts of Sxl (Sxl‐Pe) and EGFP were amplified from an embryonic cDNA library (Brown & Kafatos 1988) and pEGFP‐N1 vector (Clontech), respectively. Primer pairs Sxl‐Pe‐NotI‐Fw/Sxl‐Pe‐BamHI‐Rv and EGFP‐NotI‐Fw/EGFP‐NotI‐Rv (Table S1) were used to amplify Sxl‐Pe and EGFP, respectively. The FLAG tag was constructed using a pair of complementary oligonucleotides, Flag‐KpnINotI‐Fw/Flag‐KpnINotI‐Rv (Table S1). The ORF of Sxl‐Pe and the FLAG tag were digested with NotI/BamHI and KpnI/NotI, respectively, and ligated into pUASp vector (Rørth 1998). NotI‐digested EGFP was inserted into the NotI site of the resultant plasmid, and its direction of insertion was confirmed by sequencing on an ABI3100 instrument (Applied Biosystems).

For construction of UASp‐Flag‐EGFP, EGFP was amplified using primer pair EGFP‐BamHI‐Fw/EGFP‐XbaI‐Rv (Table S1) from pEGFP‐N1 vector. The FLAG tag was constructed using a pair of complementary oligonucleotides, Flag‐KpnIBamHI‐Fw/Flag‐KpnIBamHI‐Rv (Table S1). EGFP and the FLAG tag were digested with BamHI/XbaI and KpnI/BamHI, respectively, and then ligated into the cloning sites of pUASp vector.

Germline transformation was performed using y w embryos as recipients. w⁺ transformants were crossed to y w females to establish homozygous stocks of UASp‐Flag‐EGFP‐Sxl and UASp‐Flag‐EGFP.

RNA immunoprecipitation and RNA sequencing (RIP‐seq)

Stage 10–15 embryos [4 h 20 min – 13 h 20 min after egg laying (AEL)] derived from nos‐Gal4/nos‐Gal4 females mated with UASp‐Flag‐EGFP‐Sxl/UASp‐Flag‐EGFP‐Sxl or UASp‐Flag‐EGFP/UASp‐Flag‐EGFP (control) males were homogenized in ice‐cold extraction buffer [EB: PBS containing 0.5% Triton X‐100, cOmplete Protease Inhibitor Cocktail (Roche), and 100 U/mL Protector RNase inhibitor (Roche)] and centrifuged at 20 000 g for 10 min at 4°C. The supernatant was collected and incubated with anti‐GFP antibody (A11122; Thermo Fisher Scientific) in the presence of protein A–Sepharose beads (GE Healthcare) for 6 h at 4°C. After incubation, RNAs were extracted from supernatants and immunoprecipitates using ISOGEN (Nippon Gene). Amplified cDNAs were prepared using the Ovation RNA‐Seq System V2 kit (NuGEN) and subjected to RNA sequencing (RNA‐seq). RNA‐seq reads were aligned to the transcript models of Drosophila melanogaster (dmel‐all‐transcript‐r5.53) with Bowtie2 (Langmead & Salzberg 2012), and eXpress (Roberts & Pachter 2013) was used to calculate FPKM (fragments per kilobase of transcripts per million mapped reads) from the alignment results. The raw fastq files were deposited in DDBJ Bio Project database under Accession No. DRA006010.

RNA‐immunoprecipitation and RT‐PCR

Stage 1–15 embryos (0–13 h 20 min AEL) derived from da‐Gal4/da‐Gal4 females mated with UASp‐Flag‐EGFP‐Sxl/UASp‐Flag‐EGFP‐Sxl males were homogenized in ice‐cold EB and centrifuged at 20 000 g for 10 min at 4°C. The supernatant was collected and incubated with or without anti‐GFP antibody in the presence of protein A–Sepharose beads for 6 h at 4°C. After incubation, RNAs were extracted from the immunoprecipitates using ISOGEN. Total RNA was also extracted before immunoprecipitation. cDNAs were synthesized using SuperScript III (Thermo Fisher Scientific), and PCR was performed using primer pairs msl2‐IPRT‐Fw/msl2‐IPRT‐Rv and vasa‐IPRT‐Fw/vasa‐IPRT‐Rv (Table S2).

Microarray analysis

To identify PGC‐enriched genes, gene expression was compared between PGCs and whole embryos by microarray analysis. PGCs were isolated from EGFP‐vasa embryos at 11 different stages [Stage 4: 1 h 20 min – 2 h 20 min AEL; Stage 5–7: 2 h 20 min – 3 h 20 min AEL; Stage 8–9: 3 h 20 min – 4 h 20 min AEL; Stage 10: 4 h 20 min – 5 h 20 min AEL; Stage 11: 5 h 20 min – 7 h 20 min AEL; Stage 12: 7 h 20 min – 9 h 20 min AEL; Stage 13–14: 9 h 20 min – 11 h 20 min AEL; Stage 15: 11 h 20 min – 13 h 20 min AEL; Stage 16: 13 h 20 min – 16 h 20 min AEL; Stage 17 (early): 16 h 20 min – 19 h 20 min AEL; Stage 17 (late): 19 h 20 min – 20 h AEL] by fluorescence‐activated cell sorting (FACS) as described (Shigenobu et al. 2006). Whole EGFP‐vasa embryos were also collected at 11 different stages. Total RNA extraction, RNA amplification, and microarray analysis were carried out as described (Hayashi et al. 2017). The resultant data from PGCs and whole embryos were deposited in GEO under Accession No. GSE83460 and No. GSE102307, respectively. Log₂ ratios from two‐color microarray experiments were normalized across samples from each developmental stage, and estimated expression values of each gene at each developmental stage were obtained using the limma package in R (https://bioconductor.org/packages/release/bioc/html/limma.html). Using the lmscFit function, the following model was fitted to the normalized log₂ expression values:

where E is log₂ signal intensity of the normalized expression value; C and D are cell types [PGCs and whole embryos (soma) at 11 developmental stages] and dye effects, respectively; and ε is the residual. The sums of the intercept and C_i for each probe in PGCs and soma were used as estimated expression values.

In situ hybridization

Whole‐mount in situ hybridization of embryos was performed as described (Hayashi et al. 2004). Total RNAs were isolated from embryos at various stages (0–8 h AEL) using ISOGEN, and cDNAs were synthesized using SuperScript III. cDNAs corresponding to ben, bol, Kap‐alpha3, Neos, Pp2B‐14D, Su(var)2‐10, CG6364, CG7791, CG9153, CG10077, and CG32075 were amplified by RT‐PCR using the following primer pairs: ben‐Fw/ben‐Rv, bol‐Fw/bol‐Rv, Kap‐alpha3‐Fw/Kap‐alpha3‐Rv, Neos‐Fw/Neos‐Rv, Pp2B‐14D‐Fw/Pp2B‐14D‐Rv, Su(var)2‐10‐Fw/Su(var)2‐10‐Rv, CG6364‐Fw/CG6364‐Rv, CG7791‐Fw/CG7791‐Rv, CG9153‐Fw/CG9153‐Rv, CG10077‐Fw/CG10077‐Rv, and CG32075‐Fw/CG32075‐Rv (Table S1), respectively. Amplified cDNAs were cloned into pGEM‐T Easy vector (Promega). Templates for RNA probes were amplified from these plasmids using the T7 and SP6 primers. Digoxigenin (DIG)‐labeled RNA probes were synthesized from the fragments using T7 or SP6 RNA polymerase (Roche).

Signals were detected using a horseradish peroxidase‐conjugated anti‐DIG antibody (11207733910; Roche) and amplified using the TSA Biotin System and streptavidin‐fluorescein (FITC; PerkinElmer). In situ hybridization combined with immunostaining was performed as described (Hayashi et al. 2004). The embryos were stained with chick anti‐Vasa antibody (1:100) and rabbit anti‐RFP antibody (1:1000, R10367; Thermo Fisher Scientific). Alexa Fluor 546‐conjugated goat anti‐rabbit antibody (1:500, A11035; Thermo Fisher Scientific) and Alexa Fluor 633‐conjugated goat anti‐chick antibody (1:500, A21103; Thermo Fisher Scientific) were used as secondary antibodies.

The embryos used for whole‐mount in situ hybridization were obtained from nos‐Gal4/nos‐Gal4 females mated with UAS‐Redstinger/Y males. The resulting female, but not male, embryos expressed UAS‐Redstinger (nuclear RFP) (Barolo et al. 2004). Thus, sexes of the embryos could be determined by immunostaining with anti‐RFP antibody.

Immunostaining

Immunofluorescence staining of adult gonads was performed as described (Hayashi et al. 2004). Gonads were dissected from adult flies 3–5 days after eclosion. The following primary antibodies were used at the indicated dilutions: rabbit anti‐Vasa antibody (1:500), mouse anti‐Hts antibody [1:10, 1B1; Developmental Studies Hybridoma Bank (DSHB)], and mouse anti‐Fas3 antibody (1:10, 7G10 anti‐Fasciclin III; DSHB). Primary antibodies were detected using the following secondary antibodies: Alexa Fluor 488–conjugated goat anti‐rabbit antibody (1:500, A11034; Thermo Fisher Scientific) and Alexa Fluor 546–conjugated goat anti‐mouse antibody (1:500, A11030; Thermo Fisher Scientific).

Isolation of PGCs and RT‐PCR

Primordial germ cells were isolated by FACS, as described (Shigenobu et al. 2006), from stage 12–16 embryos (8 h 20 min – 16 h 20 min AEL) derived from EGFP‐vasa/EGFP‐vasa; nos‐GAL4/nos‐GAL4 females mated with UAS‐Redstinger/Y males. GFP and RFP double‐positive cells and GFP‐positive cells were isolated as female and male PGCs, respectively. Total RNAs were extracted from the isolated PGCs (the number of PGCs = 1000) using the RNeasy Mini kit (QIAGEN), and cDNAs were synthesized using SuperScript III. PCR was performed using primer pairs Su(var)2‐10‐RT‐Fw/Su(var)2‐10‐RT‐Rv1, Su(var)2‐10‐RT‐Fw/Su(var)2‐10‐RT‐Rv2, GFP‐RT‐Fw/GFP‐RT‐Rv, RFP‐RT‐Fw/RFP‐RT‐Rv, and rp49‐RT‐Fw/rp49‐RT‐Rv (Table S2).

Results and Discussion

Isolation of RNAs immunoprecipitated with Sxl protein in PGCs

Sxl exhibits sex‐biased expression in PGCs during embryogenesis (Hashiyama et al. 2011), and RNAi knockdown of Sxl in PGCs causes a tumorous phenotype in adult ovaries (Hashiyama et al. 2011) (cf. Fig. 3D). Furthermore, when Sxl is induced in male PGCs during embryonic stages 10–15, and the resulting PGCs are transplanted into female embryos, they follow a female fate and differentiate into mature oocytes in the host ovaries (Hashiyama et al. 2011). These observations indicate that female germline fate is induced in PGCs by Sxl during stages 10–15.

To identify the downstream targets of Sxl in the pathway leading to female germline development, we isolated the transcripts that co‐immunoprecipitated with Sxl protein from PGCs at embryonic stages 10–15. Because Sxl is also expressed in somatic tissues during these stages (Penalva & Sánchez 2003), we hesitated to use an anti‐Sxl antibody for immunoprecipitating germline transcripts from whole embryos due to the risk of contamination by somatic transcripts. To solve this problem, we used flies expressing Sxl protein tagged with enhanced green fluorescent protein (Sxl‐EGFP) exclusively in PGCs, and then precipitated Sxl‐binding transcripts from whole embryos using anti‐GFP antibody.

Prior to these experiments, we tested whether Sxl‐EGFP recapitulated the activity of endogenous Sxl. A gain‐of‐function Sxl mutation that produces functional Sxl both in male and female soma causes male‐specific lethality at the larval stage due to a failure in dosage compensation caused by misregulation of msl2 mRNA translation (Maine et al. 1985; Bashaw & Baker 1997; Kelly et al. 1997). We expressed Sxl‐EGFP in the soma of both male and female embryos under the control of da‐Gal4 driver, and examined the percentage of males in adulthood. A significant decrease in the percentage of males among adult flies was evident, whereas the percentage of males that expressed EGFP under the control of the da‐Gal4 driver was approximately 50% (Fig. S1A). Furthermore, msl2 mRNA was immunoprecipitated with an anti‐GFP antibody from embryos expressing Sxl‐EGFP under the control of da‐Gal4 (Fig. S1B). Although we cannot exclude the possibility that Sxl‐EGFP expression results in male to female sex‐reversal, these results suggest that Sxl‐EGFP is functional in the soma.

To express Sxl‐EGFP in PGCs, we used nos‐Gal4‐VP16 (nos‐Gal4), which produces Gal4‐VP16 under the control of the germline‐specific nos promoter (Van Doren et al. 1998). In embryos produced by females carrying the nos‐Gal4 gene (maternal nos‐Gal4), maternally supplied nos‐Gal4 RNA is partitioned into PGCs, and its protein product activates UAS‐dependent gene expression from stage 9 until at least the end of embryogenesis (Van Doren et al. 1998; Hayashi et al. 2017). The embryos expressing Sxl‐EGFP in PGCs at stages 10–15 were collected for RIP‐seq analysis with anti‐EGFP antibody. As a control, the embryos expressing EGFP in PGCs were processed as above (see Materials and Methods). Three groups of transcripts precipitated with Sxl‐EGFP (Sxl‐EGFP‐IP) and EGFP (EGFP‐IP), as well as the transcripts not precipitated with Sxl‐EGFP (Sxl‐EGFP‐sup), were obtained for sequencing analysis. We selected 595 transcripts that were 32‐fold more abundant in Sxl‐EGFP‐IP than in both EGFP‐IP and Sxl‐EGFP‐sup. Because Sxl protein binds seven or more nucleotides of poly(U) tracts in its target transcripts (Samuels et al. 1994; Wang et al. 1997; Penalva & Sánchez 2003), we then selected 308 transcripts containing such poly(U) tracts as candidate Sxl‐binding transcripts (Table S3),

One of these 308 transcripts was Sxl RNA itself (No. 199 in Table S3). In the soma, Sxl RNA is transcribed from an early Sxl promoter (Sxl‐Pe) in a female‐specific manner, and the transcript is translated to produce functional Sxl protein. Sxl protein then binds to the intron of pre‐mRNA transcribed from late Sxl promoter (Sxl‐Pm), thereby inhibiting its splicing. Consequently, a female‐specific form of Sxl RNA is produced (Bell et al. 1988; Penalva & Sánchez 2003; Camara et al. 2008). Our sequencing analysis revealed that the obtained Sxl transcript is the female‐specific form produced from Sxl‐Pm, probably because its splicing complex contains Sxl protein. This suggests that early Sxl protein is already functional in PGCs before and during stages 10–15. This is consistent with our previous observation that Sxl transcripts from Sxl‐Pe is observed in PGCs from stage 9 or 10 to stage 14, whereas late Sxl transcripts from Sxl‐Pm is detectable from stage 14 onward (Hashiyama et al. 2011). Furthermore, in the soma, Sxl protein binds to 3′ UTR of mature Sxl mRNA and represses its translation to regulate Sxl protein level (Yanowitz et al. 1999). Therefore, Sxl may also regulate its own translation in PGCs through binding to mature mRNA transcribed from Sxl‐Pm.

Candidate genes enriched in PGCs during embryogenesis

Because the genes that act downstream of Sxl in somatic sex determination are dispensable for germline sex determination (Marsh & Wieschaus 1978; Schüpbach 1982; Oliver 2002), we speculated that downstream targets of Sxl in germline sex determination would be expressed predominantly in PGCs. Based on the transcriptome data of whole embryos and PGCs (Materials and methods), we selected 11 genes from among the 282 genes encoding the 308 candidate Sxl target transcripts. Table 1 shows that these genes were expressed in PGCs at high or moderate levels (log₂ expression value ≥10) and enriched in PGCs (log₂ expression ratio of PGC to whole embryo ≥1). To confirm their expression in PGCs, we performed fluorescence in situ hybridization (FISH). CG9153 expression was undetectable in PGCs, and bendless (ben), boule (bol), and CG7791 expression were not enriched in PGCs during stages 11–14 (Fig. S2). Among the remaining seven, Suppressor of variegation 2‐10 [Su(var)2‐10], Protein phosphatase 2B at 14D (Pp2B‐14D), Neosin (Neos), Kap‐alpha3, and CG10077 were predominantly expressed in both male and female PGCs during stages 11–14 (Fig. 1A–E). Expression of CG6364 and CG32075 was higher in male than in female PGCs at stage 14 (Fig. 1F and G).

Table 1. List of candidate genes enriched in primordial germ cells (PGCs), as determined by microarray analysis

Gene ID	Gene namea^a Genes for which log2 expression values in PGCs ≥10 and log2 fold change values (PGCs/whole embryos) ≥1 were selected as PGC‐enriched genes. The highest values of probes of each gene are shown.	log2 expression values in PGCsb^b Using a linear model, log2 expression values were calculated for each probe in PGCs from stage 8/9 until stage 16.							log2 fold change values (PGCs/whole embryos)c^c log2 fold change values were calculated by comparing expression values between PGCs and whole embryos from stage 8/9 until stage 16.
		St. 8/9	St. 10	St. 11	St. 12	St. 13/14	St. 15	St. 16	St. 8/9	St. 10	St. 11	St. 12	St. 13/14	St. 15	St. 16
FBgn0000173	ben	10.974	10.889	10.962	10.777	10.379	10.087	10.058	0.831	1.078	1.643	1.181	0.697	0.219	0.468
FBgn0003612	Su(var)2‐10	10.859	10.623	10.165	9.442	8.795	8.647	8.234	0.279	0.361	0.338	0.673	0.834	0.805	1.026
FBgn0011206	bol	8.484	9.090	9.503	9.863	10.106	10.007	9.757	0.131	0.222	0.533	0.905	1.088	0.499	0.300
FBgn0011826	Pp2B‐14D	10.261	10.063	9.954	9.343	9.026	8.879	8.582	2.622	2.991	3.162	2.040	1.107	0.829	0.538
FBgn0024542	Neos	10.655	10.747	10.724	10.345	10.258	10.229	10.341	0.680	0.518	1.093	0.623	0.450	0.081	0.445
FBgn0027338	Kap‐alpha3	12.377	12.207	11.702	10.915	10.673	10.691	10.417	0.661	0.739	0.909	1.223	1.768	1.877	1.930
FBgn0033038	CG7791	9.225	9.469	9.670	9.954	9.924	10.192	10.299	0.479	0.939	0.809	0.911	0.463	0.883	1.069
FBgn0035207	CG9153	10.585	10.773	10.888	10.783	10.834	10.786	10.878	1.114	0.814	1.139	0.865	0.669	0.682	0.980
FBgn0035720	CG10077	11.702	11.487	11.352	11.078	11.164	10.922	10.832	1.736	1.727	1.604	1.035	1.164	1.031	1.591
FBgn0052075	CG32075	9.636	9.886	10.133	9.929	9.914	9.637	9.518	0.632	0.729	1.253	1.136	1.251	0.905	1.487
FBgn0263398	CG6364	10.399	10.590	10.818	11.041	10.952	10.885	11.096	0.786	1.110	1.018	1.272	1.010	1.300	1.921

• ^a Genes for which log₂ expression values in PGCs ≥10 and log₂ fold change values (PGCs/whole embryos) ≥1 were selected as PGC‐enriched genes. The highest values of probes of each gene are shown.
• ^b Using a linear model, log₂ expression values were calculated for each probe in PGCs from stage 8/9 until stage 16.
• ^c log₂ fold change values were calculated by comparing expression values between PGCs and whole embryos from stage 8/9 until stage 16.

Figure 1

mRNA expression from Su(var)2‐10, Pp2B‐14D, Neos, Kap‐alpha3, CG10077, CG6364, and CG32075 in male and female PGCs during embryogenesis. (A–G) mRNA expression from Su(var)2‐10 (A), Pp2B‐14D (B), Neos (C), Kap‐alpha3 (D), CG10077 (E), CG6364 (F), and CG32075 (G) in male (upper panels) and female (lower panels) primordial germ cells (PGCs) from embryonic stages 11–14. For each stage, only mRNA signals (left panels) and merged images (right panels) are represented. Embryos were derived from nos‐Gal4/nos‐Gal4 females mated with UAS‐Redstinger/Y males. Embryos were in situ hybridized with probes for each gene (green) and immunostained for Vasa (magenta), a marker of PGCs, and Redstinger (RFP, not shown). Female, but not male, embryos expressed RFP in PGC nuclei. Probes were designed to detect all RNA variants identified in each gene region. In situ hybridization with sense probes for each gene is represented in Figure S3 Scale bars: 10 μm.

Knockdown analysis of candidate genes in the germline

In addition to maternal nos‐Gal4, we used zygotically expressed nos‐Gal4 (zygotic nos‐Gal4) to induce expression of dsRNA against each transcript in PGCs from embryonic stage 9 onward (RNAi‐dependent maternal and zygotic knockdown; mzKD) (Van Doren et al. 1998; Hayashi et al. 2017). Among the seven aforementioned genes, four were required for germline development. Adult flies with mzKD of Su(var)2‐10 exhibited gonadal dysgenesis in both sexes (Table 2) due to a severe reduction in the number of the germline cells in testes and ovaries (Fig. 2C and D). Adults with mzKD of Neos and Kap‐alpha3 exhibited female‐biased gonadal dysgenesis (Table 2). Neos mzKD also caused a failure of oocyte growth (Fig. 2F). mzKD of Kap‐alpha3 decreased the number of the germline cells in ovaries, and consequently empty germaria were evident (Fig. 2H). By contrast, CG32075 mzKD caused a reduction in germline number in both sexes (Fig. 2I and J), but its penetrance was higher in testes than in ovaries (Table 2).

Table 2. Number of dysgenic gonads in adult flies with mzKD of candidate genes

RNAi straina^a Ovaries and testes from adults with mzKD of each gene were observed. Mating scheme, genotype of flies observed, and induction temperature of dsRNAs are shown in Table S4 Control testis and ovary were obtained from nos‐Gal4/nos‐Gal4 adults. Gonads were dissected from adults 3–5 days after eclosion.	Ovaries			Testes
	No. ovaries observed	No. dysgenic ovaries (%)b^b Ovaries with no mature eggs and regressed testes with length <1 mm were considered dysgenic gonads.	Significance versus control ovariesc^c *P > 0.05, **P < 0.05: Significance versus control was calculated using Fisher's exact test.	No. testes observed	No. dysgenic testes (%)b^b Ovaries with no mature eggs and regressed testes with length <1 mm were considered dysgenic gonads.	Significance versus control testesc^c *P > 0.05, **P < 0.05: Significance versus control was calculated using Fisher's exact test.	Significance of ovaries versus testesd^d #P > 0.05, ##P < 0.01: Significance of difference between the number of dysgenic ovaries and testes was calculated using Fisher's exact test.
‐ (Control)	60	0 (0)		50	0 (0)
Su(var)2‐10 ^RNAi‐1	99	99 (100)	**	62	62 (100)	**	#
Su(var)2‐10 ^RNAi‐2	36	36 (100)	**	33	33 (100)	**	#
Pp2B‐14D ^RNAi‐1	50	0 (0)	*	50	0 (0)	*	#
Pp2B‐14D ^RNAi‐2	50	0 (0)	*	50	0 (0)	*	#
Neos ^RNAi‐1	68	49 (72.1)	**	50	3 (6.0)	*	##
Neos ^RNAi‐2	56	8 (14.3)	**	50	0 (0)	*	##
Neos ^RNAi‐3	58	8 (13.8)	**	50	2 (4.0)	*	#
Kap‐alpha3 ^RNAi	76	12 (15.8)	**	50	2 (4.0)	*	##
CG10077 ^RNAi	50	2 (4.0)	*	50	0 (0)	*	#
CG32075 ^RNAi‐1	50	29 (58)	**	50	50 (100)	**	##
CG32075 ^RNAi‐2	50	0 (0)	*	50	0 (0)	*	#
CG6364 ^RNAi‐1	50	0 (0)	*	50	2 (4.0)	*	#
CG6364 ^RNAi‐2	50	2 (4.0)	*	50	0 (0)	*	#

• ^a Ovaries and testes from adults with mzKD of each gene were observed. Mating scheme, genotype of flies observed, and induction temperature of dsRNAs are shown in Table S4 Control testis and ovary were obtained from nos‐Gal4/nos‐Gal4 adults. Gonads were dissected from adults 3–5 days after eclosion.
• ^b Ovaries with no mature eggs and regressed testes with length <1 mm were considered dysgenic gonads.
• ^c *P > 0.05, **P < 0.05: Significance versus control was calculated using Fisher's exact test.
• ^d #P > 0.05, ##P < 0.01: Significance of difference between the number of dysgenic ovaries and testes was calculated using Fisher's exact test.

Figure 2

Phenotypes caused by mzKD of Su(var)2‐10, Neos, Kap‐alpha3, and CG32075 in adult testes and ovaries. (A and B) Control testis and ovary were obtained from nos‐Gal4/nos‐Gal4 adults. (C and D) Testis and ovary of non‐Sb adults derived from nos‐Gal4/nos‐Gal4 females mated with UAS‐Su(var)2‐10^RNAi‐2/TM3, Sb (BDSC32956) males. (E and F) Testis and ovary of adults derived from nos‐Gal4/nos‐Gal4 females mated with UAS‐Neos^RNAi‐1/UAS‐Neos^RNAi‐1 (BDSC55960) males. (G–J) Testes and ovarian germaria of adult progeny derived from UAS‐dicer2/UAS‐dicer2; nos‐Gal4/nos‐Gal4 females mated with UAS‐Kap‐alpha3^RNAi/UAS‐Kap‐alpha3^RNAi (BDSC27535) (G and H) and UAS‐CG32075^RNAi‐1/UAS‐CG32075^RNAi‐1 (VDRC19255) (I and J) males. Testes and ovaries were stained for Vasa (green), Hts [a marker for spectrosomes and fusomes (magenta)], and Fasciclin III [marker for a subset of somatic cells in germaria and follicles (magenta)]. Scale bars in A–F, G, and I: 50 μm; H and J: 10 μm.

Next, we induced expression of dsRNA against Su(var)2‐10, Neos, Kap‐alpha3, and CG32075 by maternal nos‐Gal4 from stage 9 until the end of embryogenesis (RNAi‐dependent maternal knockdown; mKD), and observed gonadal phenotypes in the adult stage. mKD of Neos, Kap‐alpha3, and CG32075 did not exhibit obvious abnormalities in adult gonads (Table S5). By contrast, mKD of Su(var)2‐10 caused a tumorous phenotype of germline cells in adult ovaries, but not in testes (Fig. 3A, B, E, and Fig. S4). A very similar phenotype was observed in ovaries when mKD of Sxl was performed in the germline (Fig. 3B and D). This tumorous phenotype indicates loss of female sexual identity in the germline (Steinmann‐Zwicky et al. 1989; Steinmann‐Zwicky 1994; Nagoshi et al. 1995). Thus, Su(var)2‐10 is a strong candidate for an Sxl target gene in the pathway leading to female sexual identity during embryogenesis. However, the penetrance of the phenotype was lower for mKD of Su(var)2‐10 (16.5%; No. ovarioles counted = 462) than for mKD of Sxl (87.7%; No. ovarioles counted = 508) (Fig. 3E), suggesting that another target gene of Sxl may be active in feminization of the germline. Alternatively, it is also possible that Su(var)2‐10 activity is not reduced by RNAi to a level that is insufficient for feminization of the germline.

Figure 3

Phenotypes of mKD of Su(var)2‐10 and Sxl in adult testes and ovaries. (A–D) Testes (A and C) and ovaries (B and D) of Ser adults derived from nos‐Gal4/TM3, Sb Ser females mated with UAS‐Su(var)2‐10^RNAi‐2/TM3, Sb (BDSC32956) [Su(var)2‐10 mKD] (A and B) and UAS‐Sxl^RNAi/UAS‐Sxl^RNAi (BDSC34393) (Sxl mKD) (C and D) males. Testes and ovaries were stained for Vasa (green), Hts (magenta), and Fasciclin III (magenta). Arrowheads show egg chambers with tumorous germline cells. Scale bars: 10 μm. (E) Percentage of normal (black) and tumorous (white) gonads in nos‐Gal4/nos‐Gal4 (Control), Su(var)2‐10 mKD, and Sxl mKD adults (■, Normal; □, Tumorous). N: number of gonads observed. Significance was calculated by Fisher's exact test, *P < 0.05.

Although mKD of Neos, Kap‐alpha3, and CG32075 did not result in apparent abnormalities in adult gonadal morphologies, mzKD of these genes caused sex‐biased phenotypes (Table 2), suggesting that they contribute to the mechanisms regulating sex‐specific germline development. Furthermore, mzKD of Su(var)2‐10 caused germline loss in both sexes (Table 2 and Fig. 2C and D). Thus, in addition to its role in Sxl‐dependent feminizing pathway, Su(var)2‐10 is also required for germline maintenance in both sexes.

Possible mechanism regulating Su(var)2‐10 function by Sxl

Su(var)2‐10 contains two putative Sxl‐binding sites (SBS) consisting of seven nucleotide poly(U) tracts; one is located in an intronic region, 4–11 bp upstream of the 3′ splice site (SBS1 in Fig. 4A), and the other is in the 3′ UTR region or the intronic region adjacent to the 5′ splice site (SBS2 in Fig. 4A). In the soma, Sxl regulates alternative splicing of tra pre‐mRNA through binding to SBS, 4–12 bp upstream of the 3′ splice site within its first intron (Boggs et al. 1987; Penalva & Sánchez 2003). Therefore, we speculated that Sxl may regulate alternative splicing of the Su(var)2‐10 intron containing SBS1 to produce a female‐specific form. To address this issue, we subjected RNA extracted from male and female PGCs at stages 10–15 to RT‐PCR using primers Fw and Rv1 (Fig. 4A). However, we could not detect a sex‐specific variant of Su(var)2‐10 in PGCs [Su(var)2‐10 fragment 1 in Fig. 4B].

Figure 4

Poly(U) tracts of seven or more nucleotides located in Su(var)2‐10 isoforms, and the expression of exons containing poly(U) tracts in male and female primordial germ cells (PGCs). (A) Schematic representation of Su(var)2‐10 RNA variants (FlyBase: http://flybase.org). Protein coding and UTR regions are indicated by black and white boxes, respectively. Intron regions are indicated by black lines. The sites of poly(U) tracts of seven or more nucleotides are indicated by orange lines (putative Sxl‐binding site: SBS1 and 2). Primers used for RT‐PCR analysis are indicated by magenta triangles (Fw, Rv1, and Rv2). (B) Su(var)2‐10 fragments 1 and 2 were amplified from cDNA using primer pairs Fw/Rv1 and Fw/Rv2 (A), respectively. cDNA synthesis reactions were performed with (RT+) or without (RT‐) reverse transcriptase, using mRNA extracted from male and female PGCs as the template. PGCs were isolated from stage 12–16 embryos derived from EGFP‐vasa/EGFP‐vasa; nos‐Gal4/nos‐Gal4 females mated with UAS‐Redstinger/Y males. Female and male PGCs were collected as GFP and RFP double‐positive and GFP‐positive cells, respectively. Male and female PGCs were isolated in triplicate. Ribosomal protein L32 (rp49) was amplified as an internal control.

Alternatively, it is also possible that Sxl protein regulates translation of Su(var)2‐10 RNA in PGCs by binding to SBS2 in its 3′ UTR region. RT‐PCR using another pair of primers, Fw and Rv2 (Fig. 4B), followed by sequencing, revealed that the Su(var)2‐10 3′ UTR region encompassing SBS2 is expressed in PGCs. Although Sxl protein binds to 5′ and 3′ UTRs of msl2 RNA to repress its translation (Bashaw & Baker 1997; Kelly et al. 1997; Beckmann et al. 2005; Lucchesi & Kuroda 2015), we speculate that Sxl enhances translation of Su(var)2‐10 transcripts in female PGCs. This is because mKD of Su(var)2‐10 activity resulted in a phenotype very similar to that caused by mKD of Sxl. However, because no antibody against Su(var)2‐10 protein is currently available, this idea remains to be tested.

Possible function of Su(var)2‐10 in the pathway leading to female sexual identity in the germline

Su(var)2‐10 encodes Drosophila protein inhibitor of activated STAT (dPIAS) (Mohr & Boswell 1999; Betz et al. 2001; Hari et al. 2001; Shuai 2006), which represses stat92E activity to suppress JAK/STAT signaling (Betz et al. 2001). JAK/STAT signaling is required in PGCs to induce male sexual development within the gonads (Wawersik et al. 2005). Activation of JAK/STAT signaling in female PGCs by overexpressing a JAK/STAT ligand, unpaired (upd), leads to ectopic expression of male‐specific genes (Wawersik et al. 2005). Considering that reduction in Sxl activity by the snf¹⁴⁸ mutation causes aberrant activation of JAK/STAT signaling and ectopic expression of male‐specific genes in ovarian germline cells (Shapiro‐Kulnane et al. 2015), we speculate that Sxl represses JAK/STAT signaling by enhancing Su(var)2‐10 production in female PGCs in order to prevent male sexual fate. Future studies are needed to determine whether Sxl increases the production of Su(var)2‐10 protein in PGCs. Our data represent a first step toward elucidation of the Sxl‐dependent pathway that leads to feminization of the Drosophila germline.

Acknowledgments

We thank the Genomics Center of University of Minnesota for Illumina Next‐generation Sequencing. We also thank the Bloomington Drosophila Stock Center and Vienna Drosophila Resource Center for providing us with fly stocks, and the Developmental Studies Hybridoma Bank for antibodies. This work was supported in part by Grants‐in‐Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS) (KAKENHI Grant Numbers: 25114002 and 24247011).

Author contributions

R. O. and S. K. designed the experiments. R. O., S. M., and M. H. performed the experiments, and M. S. and S. S. performed bioinformatics analyses. R. O. and S. K. wrote the paper. All authors reviewed the manuscript.