Introduction

Epigenetic regulation is a fundamental mechanism required for proper gene expression, cell proliferation, and survival during development1. Histone modifications, such as methylation and acetylation, play important roles in epigenetic regulation through global transcriptional changes2. For example, Histone 3 methylated at lysine9 (H3K9me) recruits Heterochromatin Protein 1a (HP1a), to establish the formation of heterochromatin3,4. Thus, H3K9me was generally considered to be involved in transcriptional repression5 but it was also reported that H3K9me was associated with transcriptionally active genes6, suggesting that the effects of H3K9me may be largely context-dependent.

Histone methyltransferases play critical roles in developmental processes at the epigenetic level and are known to be involved in a diverse range of diseases2. In Drosophila there are three different H3K9 methyltransferases, Su(var)3–9, G9a and Eggless (Egg, also known as dSETDB1). Among these, Egg, the Drosophila orthologue of the mammalian histone methyltransferase SETDB1, is the only H3K9 methyltransferase essential for viability and fertility7,8,9,10. Strong egg mutants show reduced levels of histone 3 lysine 9 trimethylation (H3K9me3) and concomitant oogenesis defects at the earliest stages in the germarium7,10. More specifically, egg mutant ovaries show defects in the survival of germline cells and surrounding somatic cells at very early stages of Drosophila oogenesis, resulting in rudimentary ovaries associated with apoptosis7,10. More recently, mosaic analyses revealed that Egg is required intrinsically for maintenance of both germline stem cells (GSCs) and follicle stem cells (FSCs)11. The same study also showed that Egg plays not only cell-autonomous but also non-cell-autonomous roles in somatic cells that affect the early differentiation of GSCs in the germarium11.

An interesting insight into the link between Egg function and P-element-induced wimpy testis (Piwi)-interacting small RNA (piRNA) production has also been proposed12. Rangan et al.12 showed that Egg is required to regulate piRNA cluster transcription through the deposition of H3K9me3, and for the repression of transposable elements (TEs). TEs are DNA sequences that can change their location in the genome and it has therefore been considered that TEs can generate DNA damage and genomic instability13. Meanwhile, piRNAs are mainly derived from TEs and other repeated elements and can silence target TEs to preserve genome integrity in animal germ cells14. Given the increased occurrence of apoptosis at early stages of oogenesis in egg mutant germline and somatic cells7,10,11, activation of TEs by a reduction of piRNAs in egg mutants may result in an increase in DNA damage, which in turn may trigger cell death. However, direct evidence is required to clarify whether the activation of TEs in egg mutants is indeed sufficient to cause DNA damage-induced apoptosis.

Despite the above-mentioned findings, the underlying mechanisms behind these observations remain to be determined. In particular, identification of target genes regulated by Egg may be crucial. To better understand the role of Egg in oogenesis, we adopted high-throughput RNA sequencing (RNA-seq). It has been shown that RNA-seq linked to bioinformatic analysis is a powerful tool to reveal unknown details of transcriptomic and epigenomic changes in an unbiased manner15,16,17. In the present study, we used RNA-seq to investigate egg mutation-induced transcriptional changes that may well be missed by conventional methods, thereby uncovering unprecedented insights into the role of Egg in oogenesis.

Results

To examine the gene expression changes caused by loss of Egg, we produced RNA-seq data from the anterior portion of wild-type (w1118) ovaries and heterozygous egg mutant (egg2138/Df(2R)Dll-Mp) ovaries carrying an early premature stop codon (egg2138) and a deficiency for the gene region, Df(2R)Dll-Mp11. As an internal control, we initially examined the change in egg gene expression (Fig. 1a). The expression of egg gene was strongly reduced in egg mutant ovaries compared with wild-type ovaries, which was further confirmed by quantitative Real Time-PCRs (qRT-PCR) (Fig. 1a) as well as Western blots (Supplementary Fig. S2). A total of 2627 upregulated genes and 1897 downregulated genes resulting from loss of Egg were identified from the RNA-seq data using DESeq with the criteria: normalized value ≥ 100 in either sample, and fold change ≥ 2. Genes were examined for enrichments in Gene Ontology biological processes18. The full gene list is available in the Gene Expression Omnibus (GEO) database (see Methods). To confirm the RNA-seq data, we also performed validation tests by qRT-PCR using separate RNA obtained from wild-type and egg mutant ovaries (the gene specific primers used for qRT-PCR are listed in Supplementary Table S1).

Figure 1
figure 1

Egg regulates the expression of particular piRNA pathway genes and some lncRNAs. Differentially expressed genes by loss of Egg were identified from RNA-seq data (black bar) using the criteria of fold change ≥2. Fold changes are shown on the log2 scale (egg mutant/wild-type). qRT-PCR assays (white bar) were performed at least three times to confirm the RNA-seq data. Error bars represent standard deviation. (a) Expression level change of egg gene. egg expression was significantly reduced in egg mutant (egg2138/Df(2R)Dll-Mp) ovaries compared with wild-type ovaries. rp49 was used as a control. (b) Representative expression profiles of piRNA pathway genes. Expression levels of ago3, krimp, mael, and zuc were downregulated in egg mutant ovaries. Consistent with zuc downregulation, FasIII expression was upregulated in egg mutant ovaries. (c) Expression level changes of Ago family members. Unlike ago3, expression of ago1 and ago2 was not significantly changed in egg mutant ovaries. (d) Expression level changes of genes for genic piRNAs. Expression levels of tj, brat, and klp10A genes which contain 3′-UTR piRNA clusters were downregulated in egg mutant ovaries. (e) Representative expression profiles of lncRNAs. Some well-known Drosophila lncRNAs, such as roX2, αγ-element, bxd, and hsr-ω were upregulated whereas some putative lncRNA genes, such as pncr011:3L and pncr012:2L, were downregulated in egg mutant ovaries. (f) Summary of expression profiles of putative lncRNA genes. Differentially expressed putative lncRNA genes are shown with the number of genes. The fold change on the log2 scale is placed on the vertical axis. Eighty-three genes were upregulated (≥ + 1) whereas 17 genes were downregulated (−1≥) in egg mutant ovaries. See also Supplementary Table S2.

Full size image

Egg regulates the expression of particular piRNA pathway genes

Egg is required for transcription of all major piRNA clusters and for repression of TEs12. To better understand the underlying mechanism of Egg-associated piRNA production, we examined whether loss of Egg affects the expression levels of any known genes involved in the piRNA pathway by analyzing the RNA-seq data. Our results showed that argonaute-3 (ago3), krimper (krimp), maelstrom (mael), and zucchini (zuc) were downregulated, but that expression levels of piwi, aubergine (aub), vasa, tejas, spindle E, armitage, and FS(1)YB genes were not significantly changed by loss of Egg (Fig. 1b). Consistently, expression levels of Ago3, Krimp, and Mael proteins appeared reduced in egg mutant ovaries (Supplementary Figs S1 and S2). Zuc has been suggested as a key player in traffic jam (tj)-derived piRNA production, and FasIII is considered as a primary target of the tj-derived piRNA-Piwi complex19. Consistent with our finding of zuc downregulation in egg mutant ovaries, FasIII was upregulated in egg mutant ovaries (Fig. 1b). The Drosophila Ago family comprises five members: Ago1, Ago2, Ago3, Piwi, and Aub. Among these five members, the Piwi protein subfamily (Piwi, Ago3, and Aub) is essential for the production of piRNAs and transposon silencing to protect germline cells20. In Drosophila ovaries, piRNAs bind to the Piwi subfamily proteins to form the piRNA-induced silencing complex (piRISC). Meanwhile, Ago1 is mainly involved in repressing the translation of mRNAs by binding to microRNAs (miRNAs) to form the miRISC, and Ago2 associates with small interfering RNAs (siRNAs) to form the siRISC21. Because ago3 expression was downregulated but the expression levels of piwi and aub were not significantly changed in egg mutant ovaries, we also examined expression level changes of the other Drosophila ago family members, Ago1 and Ago2. We found that the expression levels of ago1 and ago2 genes were not significantly changed in egg mutant ovaries (Fig. 1c). In addition to major piRNA clusters, such as the 42AB and flamenco clusters, the 3′-UTRs of tj, brain tumor (brat), and klp10A (a kinesin-13 of Drosophila) genes were also determined as new piRNA clusters that produce somatic piRNAs19. The role of Egg in these new piRNA clusters has not been investigated; therefore, we examined expression level changes of these genes in egg mutant ovaries. Consistent with the previously suggested notion that Egg plays a positive role in major piRNA cluster transcription12, the expression levels of tj, brat, and klp10A genes were all downregulated in egg mutant ovaries (Fig. 1d). Taken together, our results demonstrate that Egg not only promotes piRNA cluster transcription, but also plays a positive role in the expression of particular components of the piRNA machinery.

Egg is involved in regulating expression of some long non-coding RNAs (lncRNAs)

On analysis of the RNA-seq data, we noticed that some well-known Drosophila lncRNAs were upregulated in egg mutant ovaries, prompting us to investigate whether Egg is involved in the regulation of lncRNA expression. In general, lncRNAs are transcripts longer than 200 nucleotides with little or no protein-coding capacity22. In addition to small non-coding RNAs such as piRNAs, miRNAs, and siRNAs, accumulating evidence suggest that lncRNAs also play key regulatory roles in diverse biological processes22.

We found that the expression levels of about 100 potential lncRNAs were significantly changed in egg mutant ovaries. A selected subset of well-known differentially expressed Drosophila lncRNAs is shown in Fig. 1e (the entire list is in Supplementary Table S2). Most of them (83%), including roX2, αγ-element, bxd, hsr-ω, pncr003:2L, pncr004:X, and pncr009:3L, were upregulated, whereas 17% of them, including pncr011:3L and pncr012:2L, were downregulated by loss of Egg (Fig. 1f). Drosophila pncr (putative noncoding RNA) genes are a set of putative lncRNA genes identified by positional curation23. Overall, our data suggest that Egg is involved in regulating expression of some lncRNA genes and that Egg generally plays a repressive role in the expression of these lncRNA genes.

Two pro-apoptotic genes, reaper (rpr) and head involution defective (hid), are derepressed in egg mutant ovaries

Loss of piRNAs in germ cells leads to activation of TEs and consequently to an increase in DNA damage, which in turn can trigger apoptotic cell death13,14. Indeed, egg mutant females show rudimentary ovaries, resulting from apoptosis in the germaria at early stages of oogenesis7,10,11. We confirmed whether apoptotic events indeed happened in egg mutant ovaries using cleaved Caspase-3 antibody. As shown in Fig. 2c, we observed apoptotic germline and somatic cells in egg mutant ovaries, but not in wild-type ovaries. Thus, we compared expression level changes of apoptosis-related genes between wild-type and egg mutant ovaries using the RNA-seq data. As shown in Fig. 2a, the overall expression of apoptotic genes was seemingly derepressed in egg mutant ovaries. In particular, two pro-apoptotic genes, rpr and hid were upregulated in egg mutant ovaries (Fig. 2b). In the case of grim, the normalized value was too low in both wild-type and egg mutant ovaries to consider statistically significant. However, the expression levels of p53 and meiotic 41 (mei-41/ATR, the Drosophila ATM/ATR homologue) genes were not significantly changed and the expression level of the maternal nuclear kinase (mnk/chk-2, the Drosophila Chk-2 homologue) was downregulated in egg mutant ovaries (Fig. 2b). In Drosophila, it is well established that three pro-apoptotic genes, rpr, hid, and grim play key roles in the activation of apoptotic cell death, and each of the three pro-apoptotic genes is sufficient to induce apoptotic cell death in a Caspase-dependent manner24: During apoptosis, Rpr, Hid, and Grim proteins bind to the caspase inhibitor Diap1, thereby releasing activated caspases24. It has been known that most developmental apoptosis in Drosophila is initiated by the activation of the rpr, hid, and grim genes, and the protein activity of these genes is controlled mainly at the transcriptional level24.

Figure 2
figure 2

Apoptotic genes are generally derepressed in egg mutant ovaries. (a) Transcriptome analysis of apoptosis related genes using RNA-seq data. Overall, expression of apoptotic genes appears derepressed in egg mutant ovaries compared with wild-type ovaries. Fold changes are shown on the log2 scale (egg mutant/wild-type). Red boxes indicate representative apoptotic genes further analyzed by qRT-PCR. (b) RNA-seq (black bar) and qRT-PCR (white bar) analyses of the expression of five representative genes (rpr, hid, p53, mei-41, and chk-2). Expression levels of rpr and hid were upregulated whereas chk-2 expression was downregulated in egg mutant ovaries compared with wild-type ovaries. Expression levels of p53 and mei-41 were not significantly changed. Fold changes are shown on the log2 scale (egg mutant/wild-type). qRT-PCR assays were performed at least three times. Error bars represent standard deviation. (c) Ovaries labeled with anti-cleaved Caspase-3 antibody. No apoptotic cells were observed in wild-type ovaries (left panel). Apoptotic germline (arrowhead) and somatic (arrow) cells were detected in egg mutant ovaries (right panel). Scale bars represent 20 µm.

Full size image

Previously, immunostaining analyses revealed that Bag of marbles (Bam)-negative undifferentiated (i.e. GSC-like) cells were accumulated in egg mutant germaria, suggesting that Egg is required for Bam-dependent cystoblast differentiation12,25. However, egg mutant females show small ovaries, suggesting that the overall size of Bam-negative cells in egg mutant ovaries does not increase, probably because of apoptotic cell death in egg mutant germaria7,10,11. Our data suggest that the upregulation of rpr and hid genes in egg mutant ovaries may, at least partly, account for the previously observed apoptotic cell death in egg mutant germaria7,10,11.

Expression levels of decapentaplegic (dpp) and division abnormally delayed (dally) are upregulated in egg mutant ovaries

In the Drosophila ovary, two or three GSCs reside in the GSC niche at the anterior tip of the germarium26. Following GSC division, one daughter cell moves out of the niche and changes its identity to a differentiating cystoblast by upregulating Bam, a key differentiation factor26. Maintenance of GSC identity depends on Dpp, the Drosophila orthologue of Bone morphogenetic protein 2/4 (BMP2/4), together with Dally, a cell-surface glypican that enhances Dpp signaling27. Thus, Dpp should be strictly regulated. In the GSC niche, somatic cap cells produce the majority of the Dpp signal whereas escort cells (ECs), which lie close to the cap cells, do not express the Dpp signal27. Dpp acts through the canonical signal transduction pathway involving the type I receptors Thickveins (Tkv) and Saxophone (Sax), and the type II receptor Punt27. In GSCs, Dpp signaling transcriptionally represses bam, which results in derepression of the essential downstream translational repressor Nanos (Nos)28. Meanwhile, in cystoblasts, bam expression is derepressed and Bam inhibits Nos/Pumilio (Pum) activity to initiate the differentiation process28.

We investigated whether loss of Egg leads to transcriptional changes of the genes involved in Dpp signaling. Consistent with previous reports12,25, our data show that bam was downregulated in egg mutant ovaries (Fig. 3a), which was further confirmed by immunostaining of ovaries with anti-Bam (Supplementary Fig. S1). Moreover, we found that dpp and dally were upregulated, but the expression levels of tkv, sax, punt, mother against dpp (mad), and medea (med) were not significantly changed in egg mutant ovaries (Fig. 3a), suggesting that the increased expression levels of dpp and dally genes by loss of Egg may be responsible for Bam-dependent cystoblast differentiation defects in egg mutant germaria12,25. Consistently, we detected a significantly increased expression level of Dpp protein in egg mutant ovaries by Western blots (Supplementary Fig. S2) and immunostaining (Fig. 3b): We detected quite increased and diffused signals of Dpp protein from the anterior tip of egg mutant germaria compared with a specific signal of Dpp protein at the anterior tip of wild-type germaria (Fig. 3b). Notably, however, we also found that nos was downregulated in egg mutant ovaries (Fig. 3a). These results suggest that some of the Bam-negative cells in egg mutant germaria12,25 may also be Nos-negative cells, or the typical identity of the accumulated germ cells in egg mutant germaria12,25 may not be simply Bam-negative but also Nos-negative (i.e. GSC-like + cystoblast-like). Indeed, consistent with our results, germline clone analysis showed that egg-negative germline cell clones developed normally up to stage 5 of oogenesis but died through apoptosis11.

Figure 3
figure 3

RNAi-mediated dpp knockdown in egg mutant germarium causes an increase in ovary size. (a) Representative expression profiles of genes involved in Dpp signaling. RNA-seq data analysis (black bar) revealed that expression levels of dpp and dally were upregulated whereas those of nos and mei-P26 were downregulated in egg mutant ovaries compared with wild-type ovaries. Consistent with dpp and dally upregulation, bam was downregulated. Expression levels of tkv, sax, punt, mad, med, and pum were not significantly changed. Fold changes are shown on the log2 scale (egg mutant/wild-type). qRT-PCR assays (white bar) of the expression of seven representative genes (dpp, dally, mad, mei-P26, bam, nos, and pum) were performed at least three times. Fold changes are shown on the log2 scale (egg mutant/wild-type). Error bars represent standard deviation. (b) Germaria stained with anti-Dpp antibody. Dpp was detected specifically at the anterior tip (arrow) of wild-type germaria (left panels). Dpp appears increased and diffused from the anterior tip of egg mutant germaria (right panels). Scale bar represents 20 µm. (c) Representative light microscopic image of ovaries. RNAi-mediated dpp knockdown in egg mutant germarium causes an increase in ovary size. Ovaries of egg mutants (middle) are severely reduced in size compared with wild-type ovaries (left). 15–20% (n = 630) of egg mutant ovaries also carrying c587-Gal4 and dpp-RNAi transgenes (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+, right) showed an increase in size compared to egg mutant ovaries (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; +/+, middle). (d) qRT-PCR assays. The increased expression levels of rpr and hid in egg mutant ovaries (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; +/+) were significantly reduced in egg mutant ovaries also carrying c587-Gal4 and dpp-RNAi transgenes (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+) (one-way ANOVA with Bonferroni; **P < 0.01, ***P < 0.001; error bars represent standard deviation of three biological replicate experiments). RNAi-mediated dpp knockdown in egg mutant germarium did not change the expression levels of chk-2 and mei-41 in egg mutant ovaries.

Full size image

RNAi-mediated dpp knockdown in egg mutant germarium causes an increase in ovary size

To evaluate the significance of dpp upregulation to egg mutant ovary phenotypes, we knocked down dpp expression using RNAi in egg mutant background. We first tested the effect of the previously tested dpp-RNAi transgene29 in egg mutant background under the control of the actin5C-Gal4 driver, a highly and ubiquitously active driver. However, as expression of the dpp-RNAi transgene by the actin5C-Gal4 driver resulted in pupal lethality, we could not evaluate the effect of dpp knockdown by the act5C-Gal4 driver on egg mutant ovaries. Given previous results that RNAi-mediated reduction of dpp expression by the c587-Gal4 driver, a driver active in ECs and early follicle cells, suppressed the ovarian tumor phenotype induced by H3K4 demethylase 1-RNAi29, we next evaluated the effect of dpp knockdown in egg mutant background using the dpp-RNAi transgene and the c587-Gal4 driver. We crossed flies carrying the dpp-RNAi transgene and Df(2R)Dll-Mp deficiency, which uncovers the egg gene, (+/Y; Df(2R)Dll-Mp/CyO; dpp-RNAi/dpp-RNAi) with flies carrying the egg mutation and c587-Gal4. (c587-Gal4/c587-Gal4; egg2138/CyO; +/+). We found that a small but consistent percentage (17.33 ± 1.81%, n = 630) of the egg mutant ovaries also carrying c587-Gal4 and dpp-RNAi transgenes (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+) showed an increase in size (Fig. 3c). Ovaries of the genotype (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+) contained an accumulation of disorganized and somewhat bloated ovarioles and exhibited differentiation defects (data not shown) similar to those described previously12,25, raising the possibility that, unlike egg mutant germline and surrounding somatic cells (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; +/+), the defective cells in the dpp-knockdown + egg mutant ovaries (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+) could be maintained for a longer time by escaping apoptosis. Indeed, our qRT-PCR results showed that the increased expression levels of rpr and hid genes in egg mutant ovaries (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; +/+) were significantly reduced in the dpp-knockdown + egg mutant ovaries (c587-Gal4/+; egg2138/Df(2R)Dll-Mp; dpp-RNAi/+) (Fig. 3d).

dpp overexpression in Drosophila ovarian somatic cells (OSCs) induces cell death and activates the expression of rpr and hid

To test whether dpp upregulation in ovarian somatic cells can induce cell death directly, we transfected a Dpp expression construct into OSCs19, and evaluated cell viability and changes in gene expression following dpp overexpression (Fig. 4a). Drosophila OSCs are composed of mitotically active early somatic cells derived from the parental female germline stem/ovarian somatic sheath (fGS/OSS) cell line19. Our qRT-PCR results showed that expression levels of rpr and hid were significantly elevated following dpp overexpression in OSCs (Fig. 4b). Consistent with the qRT-PCR results, MTS assays showed that dpp overexpression significantly decreased the viability of OSCs (Fig. 4c). Collectively, these data suggest that ectopic upregulation of dpp in ovarian somatic cells can induce cell death through activation of rpr and hid. We also examined the impact of loss of Egg on H3K9me3 enrichment levels in dpp, dally, rpr, and hid genes. Our results showed that certain levels of H3K9me3 were present in dpp and rpr genes, but not in dally and hid genes (Fig. 4d). Moreover, H3K9me3 levels in dpp and rpr genes were significantly reduced in the absence of Egg, suggesting that H3K9me3 could directly regulate the expression of dpp and rpr at least in part (Fig. 4d).

Figure 4
figure 4

dpp overexpression induces apoptotic cell death through activation of rpr and hid in Drosophila ovarian somatic cells (OSCs). (a) Schematic illustration of Dpp overexpression experiments in OSCs. (b) qRT-PCR assays. dpp overexpression activated the expression of rpr and hid in OSCs. qRT-PCR assays show changes in expression levels of seven representative genes (egg, bam, dally, p53, chk-2, rpr, and hid) following dpp overexpression in OSCs (unpaired t-test; *P < 0.05; error bars represent standard deviation of three biological replicate experiments). (c) MTS assays. dpp overexpression induced cell death in OSCs. MTS assays show that the viability of dpp-overexpressing OSCs is significantly decreased (unpaired t-test; *P < 0.05; error bars represent standard deviation of three biological replicate experiments). (d) Chromatin immunoprecipitation (ChIP)-qPCR assays. ChIP was performed using H3K9me3 antibody to measure enrichment levels of H3K9me3 in the putative target genes of Egg. The recovered DNA was analyzed by qPCRs. Certain levels of H3K9me3 were present in some of the putative target genes such as dpp, rpr, chk-2, ago3, zuc, mael, tj, and klp-10A genes. H3K9me3 levels in these genes were significantly reduced in egg mutant ovaries. The enrichment levels are relative to act5C. Error bars represent standard deviation of three biological replicate experiments.

Full size image

Discussion

In the present study, we used RNA-seq data analysis and qRT-PCR validation to demonstrate that Egg plays diverse roles in the regulation of piRNA production, lncRNA expression, apoptosis-related gene expression, and Dpp signaling during Drosophila oogenesis. Furthermore, using genetic and cell biological approaches, we demonstrated that ectopic upregulation of dpp caused by loss of Egg in the germarium can trigger apoptosis in vivo.

Regarding piRNA production, our results revealed that among the known piRNA machinery components, ago3, krimp, mael, and zuc genes are the major piRNA-related targets of Egg (Fig. 1b). Consistent with the previously suggested role of Egg for promoting piRNA production, all of the putative target genes were downregulated in egg mutant ovaries (Fig. 1b,d). In Drosophila, two piRNA processing pathways, primary processing and secondary processing, have been proposed14; the primary processing pathway functions in both germline and somatic cells by processing precursor piRNAs into piRNAs whereas the secondary processing pathway functions only in germline cells in which piRNAs are amplified by the ping-pong cycle14,30,31. Zuc plays a central role in the primary processing pathway19. Given the involvement of Egg in both germline and somatic piRNA production12, we tested whether decreased expression of zuc is responsible for the ovarian phenotypes caused by loss of Egg. We investigated phenotypic changes after introduction of wild-type zuc transgenes32 under the control of the actin5C-Gal4 driver into egg mutant background, but the ovaries of the genotypes (egg2138/Df(2R)Dll-Mp; HA-tagged zuc (or EGFP-tagged zuc)/actin5C-Gal4) did not show any significant phenotypic changes compared with those of egg mutants (data not shown). Ago3 is essentially involved in the secondary processing pathway along with Aub14. The nuage, which surrounds the nuclei of nurse cells, has been proposed as a site for the ping-pong cycle30,31. Various types of proteins, including Ago3, Krimp, and Mael, have been identified as nuage components14. The decreased expression levels of the particular nuage components in egg mutant ovaries (Fig. 1b and Supplementary Figs S1 and S2) suggest that the reduction of germline piRNAs in egg mutants may result from not only a reduction of precursor piRNA transcription, as proposed previously12, but also from a failure of the piRNA amplification pathway.

Our data also revealed a previously unknown role of Egg as an lncRNA regulator (Fig. 1e,f). In Drosophila, the existence of lncRNAs has long been known, but only a few lncRNAs have been investigated33. Here, we revealed that Egg is involved in regulating the expression of 100 potential lncRNA genes, and appears to play a repressive role in the expression of these lncRNAs (Fig. 1e,f). Among the upregulated lncRNAs in egg mutant ovaries, two well-known heat-inducible lncRNAs, αγ-element and hsr-ω are located in genomic regions where numerous TEs are found34,35. Given the involvement of Egg in regulating piRNA production and thus TE mobilization, this raises an intriguing possibility of a link between the upregulation of αγ-element and hsr-ω in egg mutant ovaries and their genomic locations as TE hotspots. A strong upregulation of pncr003:2L and pncr004:X by loss of Egg is noteworthy, but the functional significance of these lncRNAs in the Drosophila ovary has not been determined. Although we tried to knock down pncr003:2L and pncr004:X using transgenic flies containing dsRNA for RNAi of pncr003:2L or pncr004:X under the control of the act5C-Gal4 driver, we could not detect any phenotypic changes during oogenesis. Interestingly, pncr003:2L was originally annotated as an lncRNA23, but it was recently reported to encode two small functional peptides that are involved in regulating calcium transport in the Drosophila heart36. Determination of the function of pncr003:2L in the Drosophila ovary is an interesting issue that needs to be addressed.

In this study, we demonstrated that dpp and dally were strongly upregulated in egg mutant ovaries (Fig. 3a) and that dpp knockdown in ECs and early follicle cells in egg mutant background resulted in an increase in the size of the ovaries (Fig. 3c). Previously, by analysis of egg-RNAi knockdown in ECs, an EC-specific requirement for egg in controlling germ cell differentiation was suggested25. Moreover, germ cell differentiation defects caused by egg knockdown in ECs was attributed to an increase in Dpp signaling because removal of one copy of dpp partially suppressed the tumorous phenotype caused by egg knockdown in ECs25. The expression level of dally, an enhancer of Dpp signaling, may be maintained at a high level in the dpp knockdown in egg mutant background (Fig. 3c); therefore, the enhancement of Dpp signaling caused by loss of Egg may be maintained in the dpp knockdown in egg mutant germarium, thereby exhibiting similar differentiation defects as those observed in the egg mutant germarium12,25. However, the defective cells in the dpp-knockdown in egg mutant background could be maintained for a longer time, which may be attributed to a reduction of the enhanced apoptosis in egg mutant ovaries because the increased expression levels of rpr and hid in egg mutant ovaries were significantly reduced in the dpp-knockdown in egg mutant background (Fig. 3d). Given that Egg is broadly expressed in germ cells and somatic cells in the ovary during oogenesis7 (Supplementary Fig. S1), dpp knockdown only in ECs and early follicle cells may not be sufficient to consistently counteract the overall apoptosis-promoting effect caused by dpp upregulation in egg mutant ovaries, which may explain the relatively low occurrence of the effect of the dpp knockdown in ECs and early follicle cells on the increase in ovary size.

We propose a model in which loss of Egg may initially cause a relatively mild level of ectopic dpp and dally overexpression that may be sufficient to cause germ cell differentiation defects through repression of bam. Apoptosis may then be initiated when the level of ectopic dpp overexpression reaches a certain threshold level capable of inducing rpr and hid upregulation perhaps through a dpp positive-feedback loop37. Alternatively, we cannot rule out the possibility that Egg represses dpp, dally, rpr and hid separately although the possibilities are not necessarily exclusive. Further studies are needed to determine the exact molecular mechanism by which Dpp signaling is increased in egg mutant ovaries. Given the role of BMPs in many mammalian stem cell systems27 and the existence of mammalian homologues of Egg and Dpp, the role of Egg in regulating Dpp signaling may provide important insights into their potential roles in mammalian stem cells.

Methods

Drosophila strains and genetics

All flies were maintained at room temperature (24–26 °C) on cornmeal-molasses medium. w1118 was used as a wild-type control. egg2138/Df(2R)Dll-Mp females were used for the isolation of ovarian tissues and phenotypic analysis. egg2138/GFPCyO and Df(2R)Dll-Mp/GFPCyO were provided by T. Hazelrigg (Columbia University, New York, NY, USA). c587-Gal4 was provided by A. Spradling (Carnegie Institute for Science, Baltimore, MD, USA). Transgenic lines expressing N-terminally triple-Hemaglutanin (HA)-tagged zuc and EGFP-tagged zuc were provided by T. Schüpbach (Princeton University, Princeton, NJ, USA). act5C-Gal4 and dpp-RNAi (BL-31531) lines were obtained from the Bloomington Stock Center. To knock down the expression of dpp, c587-Gal4 and dpp-RNAi transgenes or act5C-Gal4 and dpp-RNAi transgenes were introduced into the egg2138/Df(2R)Dll-Mp background by standard genetic crosses. To test the effect of wild-type zuc transgenes on the ovarian phenotype caused by loss of Egg, HA-tagged zuc and act5C-Gal4 transgenes or EGFP-tagged zuc and act5C-Gal4 transgenes were crossed into the egg2138/Df(2R)Dll-Mp background by standard genetic crosses. To knock down pncr003:2L or pncr004:X, transgenic flies containing dsRNA for RNAi of pncr003:2L (Bloomington Drosophila Stock Center, Stock Number 28957) or pncr004:X (Bloomington Drosophila Stock Center, Stock Number 28547) were used under the control of the actin5C-Gal4 driver.

RNA sequencing sample preparation and data analysis

Ovaries were dissected from 2- to 4-day-old Drosophila females and frozen in liquid nitrogen. egg2138/Df(2R)Dll-Mp females have rudimentary ovaries; therefore, mid- (stages 8–9) to late- (stages 10–14) stage wild-type egg chambers were removed manually to match the developmental stages and size. A representative image of egg chambers used for our RNA-seq and qPCRs is shown in Supplementary Fig. S1 as dashed lines. Total RNAs were extracted using TRIzol® reagent (Invitrogen). Isolated RNAs were treated with DNAse I (Takara) and further purified by acid phenol:chloroform extraction. cDNAs were synthesized from 1 μg of DNAse-treated total RNAs by reverse transcription (RT) with SuperScript III (Invitrogen). Twenty-microliter RT reactions containing 2.5 µM oligo- (dT)20 primers, 2.5 µM random hexamers, 0.5 mM each dNTP, 10X reverse transcription buffer, 5 mM MgCl2, 0.01 M DTT, 40 units/µl RNase-OUT™ and 200 units/µl SuperScript® III reverse transcriptase were incubated at 50 °C for 60 min, at 25 °C for 10 min and at 85 °C for 5 min. Raw sequences were aligned to the Drosophila genome (Drosophila genome version R5.42) using Burrows-Wheeler Aligner (BWA) software. Numbers of reads aligned within each gene were counted using the HTSeq package of Python and normalized with the DESeq package of R. Differentially expressed genes between w1118 and egg2138/Df(2R)Dll-Mp ovaries were identified using the DESeq criteria: normalized value ≥100 in either sample and fold change ≥2. All sequencing data has been deposited with the GEO database (Accession GSE87492).

Quantitative Real-Time PCR

Relative mRNA levels were quantified using a LightCycler® 480 Real-Time PCR System (Roche). Each PCR reaction was carried out in 10 µl containing 5 µl LightCycler® 480 SYBR Green I Master (Roche), 0.25 µM gene specific primers and 20 ng cDNA. Reactions were run at 95 °C for 5 min followed by 45 cycles at 95 °C for 10 sec, 58 °C for 10 sec and 72 °C for 10 sec. Results were normalized to ribosomal protein 49 (rp49). Gene expression levels were calculated using the comparative Ct method. All experiments were carried out at least three times for each of three biological replicates. The gene specific primers used for qRT-PCR are listed in Supplementary Table S1.

Immunostaining

Ovaries were dissected in PBS, fixed for 9 min in a 6:1 mix of heptane:fixative (6% formaldehyde, 16.7 mM KPO4, 75 mM KCl, 25 mM NaCl, and 3.3 mM MgCl2), washed 3 × 5 min in PBS with 0.1% Triton-X100 (PBTX), and blocked for 4 hr in 1% BSA in PBTX. Primary and secondary antibodies were diluted in 1% BSA in PBTX. The ovaries were incubated overnight at 4 °C in primary antibody diluted in 1% BSA in PBTX, washed 3 × 10 min in PBTX, incubated for 3 hr at room temperature in secondary antibody (anti-rabbit Alexa Fluor 488 or anti-mouse Alexa Fluor 555, Abcam) at a dilution of 1:500, and washed 3 × 10 min in PBTX. Primary antibodies included rabbit anti-Cleaved Caspase-3 (1:500, Cell Signaling), mouse anti-Dpp (1:50, Santa Cruz Biotechnology), mouse anti-Bam-C (1:50, Developmental Studies Hybridoma Bank), mouse anti-Ago3 (1:250, H. Siomi), mouse anti-Krimp (1:250, H. Siomi), and rabbit anti-Egg (1:500). Anti-Egg polyclonal antiserum was raised against amino acids 79–315 of Egg in rabbits as described previously7. Leica SP8X confocal microscope was used to acquire images and images were processed with LAS X software.

Chromatin immunoprecipitation

Approximately 300 ovary pairs of wild-type or egg2138/Df(2R)Dll-Mp females were used in each experiment. Chromatin preparation was carried out as described previously38. Sonication used Sonics Vibra-Cell VC130 Ultrasonic Processor at power 20, using 4 pulses of 10 sec with 50 sec intervals on ice. Immunoprecipitation was carried out using H3K9me3 antibody (3 µg, Millipore), then immunoprecipitated DNA fragments were quantified by qPCR. Enrichment of H3K9me3 is determined by normalization to a constitutively expressed actin5C gene or egg gene. The mean of the normalized value is reported using three biological replicates.

Ovarian somatic cell (OSC) culture and overexpression of Dpp

The OSC line was a gift from H. Siomi (Keio University, Japan). Cells were grown at 26 °C in Shields and Sang M3 Insect Medium (Sigma) supplemented with 0.6 mg/ml glutathione, 10% FBS, 10 mU/ml insulin and 10% fly extract. A full-length dpp cDNA was PCR-amplified from Drosophila ovarian total cDNAs with gene specific primers. This dpp cDNA was tagged by the Flag epitope (MDYKDDDDK) at the N-terminus and was inserted between the EcoRI and XhoI sites of pAc5.1/V5-HisA (Invitrogen), to generate pAc5.1-Dpp, a Dpp expression vector. The primers used were as follows (5′-3′):

Forward: GAAGAATTCATGGACTACAAAGACGATGACGATAAAATGCGCGCATGGCTTCTACTCCT

Reverse: TTCCTCGAGCTATCGACAGCCACAGCCCACCAC

For Dpp overexpression in OSCs, 3 × 106 trypsinized OSCs were resuspended in 100 μl of Solution V of the Cell Line Nucleofector Kit V (Amaxa Biosystems) and mixed with 5 μg of pAc5.1/V5-HisA or pAc5.1-Dpp. Transfection was conducted in an electroporation cuvette using a Nucleofector instrument (Amaxa Biosystems). The transfected cells were transferred to fresh OSC medium and incubated at 26 °C. After incubation for 48 hr, the cells were treated with TRIzol and total RNAs were extracted as described above.

Cell viability assay (MTS assay)

Cell viability was measured using a CellTiter 96 Aqueous One Solution kit (Promega). 3 × 106 trypsinized OSCs were transfected with 5 μg of pAc5.1/V5-HisA or pAc5.1-Dpp as described above. The transfected cells were seeded in 96-well plates and incubated at 26 °C. After incubation for 48 hr, the medium was changed to 10% M3/FBS medium. Then 20 µl of MTS solution was added to each well and cells were incubated at 26 °C for 3 hrs. Absorbance at 490 nm was measured using a Microplate Reader (Tecan). All experiments were carried out using three biological replicates.