1. Stem Cells and Regenerative Medicine
Download icon

Ovaries absent links dLsd1 to HP1a for local H3K4 demethylation required for heterochromatic gene silencing

  1. Fu Yang  Is a corresponding author
  2. Zhenghui Quan
  3. Huanwei Huang
  4. Minghui He
  5. Xicheng Liu
  6. Tao Cai
  7. Rongwen Xi  Is a corresponding author
  1. National Institute of Biological Sciences, China
  2. Tsinghua University, China
Research Communication
  • Cited 1
  • Views 1,011
  • Annotations
Cite this article as: eLife 2019;8:e40806 doi: 10.7554/eLife.40806

Abstract

Heterochromatin Protein 1 (HP1) is a conserved chromosomal protein in eukaryotic cells that has a major role in directing heterochromatin formation, a process that requires co-transcriptional gene silencing mediated by small RNAs and their associated argonaute proteins. Heterochromatin formation requires erasing the active epigenetic mark, such as H3K4me2, but the molecular link between HP1 and H3K4 demethylation remains unclear. In a fertility screen in female Drosophila, we identified ovaries absent (ova), which functions in the stem cell niche, downstream of Piwi, to support germline stem cell differentiation. Moreover, ova acts as a suppressor of position effect variegation, and is required for silencing telomeric transposons in the germline. Biochemically, Ova acts to link the H3K4 demethylase dLsd1 to HP1a for local histone modifications. Therefore, our study provides a molecular connection between HP1a and local H3K4 demethylation during HP1a-mediated gene silencing that is required for ovary development, transposon silencing, and heterochromatin formation.

Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed (see decision letter).

https://doi.org/10.7554/eLife.40806.001

eLife digest

The complete set of genetic material within a cell is known as a genome. The genomes of human and other animal cells have regions of active genes interspersed with ‘dark’ regions known as heterochromatin, which contain genes and other types of genetic material that have been inactivated.

Heterochromatin commonly contains sections of genetic material known as transposons. When a transposon is active it is able to move around the genome, therefore, inactivating (or ‘silencing’) transposons helps to maintain the integrity of the genetic material in a cell. It is particularly important to silence transposons in the stem cells that produce sperm and egg cells – known as germline stem cells – to ensure genetic information is faithfully passed on to the next generation.

A protein called HP1a plays a major role in directing where heterochromatin forms in the genome. This process requires an enzyme called dLsd1 to remove a small tag from the genetic material but it is not clear how HP1a regulates the activity of dLsd1. To address this question, Yang et al. studied how egg cells form in fruit flies, which are often used as models of animal biology in experiments.

The team screened a population of fruit flies that carried mutations in many different genes to identify genes that affect the fertility of female flies. This revealed a gene named as ovaries absent (or ova for short) is required for egg cells to form. In germline stem cells ova silences transposons and in the surrounding tissue it represses a specific signal that usually maintains stem cells to allow the stem cells to divide to make egg cells. Further experiments using biochemical techniques found that the protein encoded by ova acts as a bridge to bring HP1a and dLsd1 together to silence genes in heterochromatin.

The next step would be to identify the functional counterpart of the ova gene in mammals, including humans, which may help to discover causes of infertility and develop new fertility treatment.

https://doi.org/10.7554/eLife.40806.002

Introduction

In eukaryotic genomes, heterochromatin is mainly composed of repetitive sequences such as transposons that require active silencing (Slotkin and Martienssen, 2007). Heterochromatin is defined by the presence of repressive epigenetic methylation of histone H3 at lysine 9 (H3K9me) and by heterochromatin protein 1 (HP1), which binds to H3K9me sites (Lachner et al., 2001; Bannister et al., 2001). Heterochromatin formation is mediated by co-transcriptional gene silencing, a process that requires small RNAs and their associated argonaute proteins (Martienssen and Moazed, 2015). In Drosophila, the argonaute protein Piwi and Piwi-interacting RNAs (piRNAs) use base-pairing to target nascent transcripts to the corresponding transposon regions. The Piwi/piRNAs then recruit gene silencing machinery, including HP1a and the H3K9 methyltransferase Egg to form heterochromatin (Yang and Xi, 2017; Czech and Hannon, 2016; Brower-Toland et al., 2007). The formation of heterochromatin also requires erasing of active epigenetic mark by the H3K4 demethylase dLsd1 (Rudolph et al., 2007), but the molecular link between HP1a and local H3k4 demethylation remains elusive.

Piwi, a founding member of the piRNA pathway in Drosophila, was initially identified as a fertility factor; its mutation results in germline degeneration and sterility (Cox et al., 1998; Lin and Spradling, 1997). To identify new genes involved in Piwi/piRNA-mediated gene silencing, we here conducted a female fertility screen by EMS mutagenesis and identified a novel recessive mutation on the second chromosome. The homozygous mutant males are semi-lethal (Supplementary file 1, Table 1), but females are viable but do not lay any eggs; other than sterility, these females do not have other notable defects. Dissection revealed that these females had rudimentary ovaries: rather than a normal ovary, each oviduct in these mutant females was connected to only a tiny mass of cells (Figure 1a,b). Given this nearly ‘ovaryless’ phenotype, we named the gene associated with this mutation as ovaries absent (ova) and named this mutant allele ova1 .

Figure 1 with 2 supplements see all
Ova is a niche factor for GSCs and ovary development in Drosophila.

(a) Phase contrast images of dissected ovaries from flies of indicated genotypes. Scale bar, 500 μm. (b) A graph shows the total offspring number of indicated females (n = 20, 20, 16, 16, 20 respectively). (c) Representative image of germaria from indicated genotypes labeled by α-Spectrin (red), Vasa (green), and DAPI (blue). ova1/1 and ova1/4 ovaries have numerous spherical-shaped spectrosome-containing cells (tumorous) or are empty of germline cells (germless), indicated by lack of germline cell marker Vasa (green). A wild-type (WT) germarium is usually 2 GSCs localized to the anterior tip. Scale bar, 10 μm. (d) A graph shows the percentage of normal, germless, and tumorous germaria of indicated genotypes (n = 14, 20, 30, 117, 57 respectively). (e) c587ts > ova RNAi germarium accumulated GSC-like cells after shift to 29°C for 7 and 14 days. Scale bars, 10 μm. (f) Escort cell-specific expression of ova rescued oogenesis and GSC differentiation defect of ova1/4 females. Red, α-Spectrin; Green, Vasa (g) Quantification of GSC-like cell number in germaria of indicated genotypes (n = 25, 20, 30, 117, 57, 30, 21, 31, 23 respectively). (h) Confocal sections of germaria stained by indicated antibodies or reporter. Scale bars, 10 µm. (i) Quantitative results of pMad and Dad-lacZ positive cell numbers from germaria of indicated genotypes. Values are mean ± SEM.; n > 20. P values by two-tailed Student t-test.

https://doi.org/10.7554/eLife.40806.003

Complementation mapping with deficiency lines, followed by sequencing of candidate genes led us to identify a single nucleotide deletion in an exon of CG5694, which results in a truncated protein of 387 amino acids (aa) rather than the predicted 623 aa full length protein (Figure 1—figure supplement 1a,b). CG5694 encodes a protein with no obvious sequence similarity to any existing proteins in the NCBI database, but does have a conserved nuclear respiratory factor-1 (NRF-1)- like domain at its N-terminus (15–105 aa); this DNA-binding domain was initially identified in the mammalian transcription factor NRF-1 and are known to occur in at least one other Drosophila transcription factor, Erect Wing (Ewg) (Figure 1—figure supplement 1c,d). We used CRISPR-Cas9 to generate a knock-out allele in which the entire coding region of CG5694 was deleted (Figure 1—figure supplement 1e). Homozygous knock-out allele females are sterile and exhibit virtually identical ‘ovaryless’ phenotypes as the ova1 females (Figure 1a). Additionally, transgenic expression of a genomic DNA fragment containing the ova gene region was able to effectively rescue the ovary defect and restore fertility of ova1 homozygous or CG5694 null females (Figure 1a,b). Therefore, ova is allelic to CG5694.

Normally, oogenesis initiates in the germarium, an anterior part of the ovariole where germline stem cells (GSCs) reside. Each germarium normally harbors 2–3 GSCs that can be distinguished by spherically-shaped spectrosome and by their direct contact with the cap cell niche (Figure 1c). The decedents of GSCs move posteriorly as they differentiate into germline cyst, and then bud off from the germarium to form egg chambers (Xie, 2013; Spradling, 1993). Immunostaining of ova1 homozygous and ova1/4 trans-heterozygous ovaries revealed that the mutant ovaries completely lacked vitellaria, and the germaria were either full of GSC-like cells [77% (n = 117) of ova1 germaria] (Figure 1c,d,g) or entirely germless (lacking Vasa expression (Figure 1c,d,g), suggesting that ova is required for GSC differentiation and for germline survival.

To determine whether ova functions cell-autonomously in the germline and/or non-cell-autonomously in somatic supporting cells to regulate GSCs, we conducted mosaic analysis by inducing mitotic clones using a FLP-FRT system (Xu and Rubin, 1993). Similar to wild-type control clones, ova1 mutant GSC clones behaved normally: the mutant GSCs were properly maintained in the niche, and their descendant cells were properly differentiated into germline cysts and egg chambers with properly specified oocytes (Figure 1—figure supplement 2a–d), although germline mutant eggs failed to hatch (Figure 1b). Similarly, germline-specific knocking down ova by UAS-Dcr2; nos-GAL4 (thereafter referred as ova GLKD) also did not cause any obvious defects in ovary morphology, and the number of GSCs and their immediate daughter cystoblasts (collectively referred to as GSC-like cells) per germarium remained largely normal (Figure 1—figure supplement 2g,h). Collectively, these data demonstrate that ova is not cell-autonomously required for the early stages of GSC differentiation. We next used a temperature sensitive GAL4/UAS system (Brand and Perrimon, 1993; McGuire et al., 2004) to specifically deplete ova in somatic escort cells with c587-GAL4 (c587 >ova RNAi) (Song et al., 2004). The somatic escort cells, which usually send out long protrusions that encapsulate the germline, are known to provide the niche environment required for germline cyst differentiation (Kirilly et al., 2011; Morris and Spradling, 2011). After treatment at the restrictive temperature, c587 >ova RNAi germaria began to exhibit a significantly increased number of spectrosome-containing GSC-like cells in a time-dependent manner (Figure 1e,g). The mutant escort cells were still able to send protrusions to the encapsulate the germline cells (Figure 1e), indicating that the GSC differentiation defects is likely not caused by defects in escort cell morphology. The requirement for ova i in somatic escort cells for proper GSC differentiation was further supported by the observation that escort cell-specific expression of an ova transgene was sufficient to rescue the ovary defects of ova mutant females (hereafter referred to as ova germline mutants) (Figure 1f). Therefore, ova functions in somatic escort cells and regulates germline differentiation in the germarium in a non-cell-autonomous manner.

The ova mutant phenotype is reminiscent of the piwi mutant phenotypes: piwi mutant flies also have rudimentary ovaries that contain an abnormal number of differentiation-blocked GSC-like cells, and piwi also functions primarily in somatic escort cells to regulate GSC differentiation (Jin et al., 2013; Ma et al., 2014). Loss of piwi in escort cells causes de-repression of decapentaplegic (dpp), a major self-renewal signal for GSCs, leading to GSC-like cell accumulation in the germarium (Jin et al., 2013; Ma et al., 2014). Interestingly, we found that the ova phenotype was also associated with increased dpp signaling. The extra GSC-like cells in c587 >ova RNAi germaria had dramatically increased expression of Dad-lacZ and phosphorylated Mad (pMad) (Figure 1h,i), both of which are reporters of BMP pathway activity, and decreased expression of bam (Figure 1h), a gene whose expression is normally suppressed by BMP signaling (Song et al., 2004; Chen and McKearin, 2003). Collectively, these data suggest that loss of ova in escort cells leads to ectopic dpp signaling that blocks further GSC differentiation, leading to GSC-like cell accumulation in the germarium.

The phenotypic and molecular similarities between the ova and piwi mutants led us to further test whether Ova and Piwi act via the same genetic pathway to regulate GSCs. As expected, we observed that homozygous piwi mutant females had rudimentary ovaries. As a positive control, escort cell-specific expression of a piwi transgene was sufficient to rescue the piwi mutant ovary phenotype (Figure 2a,b). On the one hand, transgenic expression of ova in escort cells of piwi mutants also partially rescued the ovary morphology phenotype with the frequent appearance of developing germline cysts, including late stages of egg chambers, although oogenesis was still abnormal, GSC-like tumor still remained, and no mature eggs were produced (Figure 2a,b,c). On the other hand, transgenic expression of piwi in escort cells of ova mutants could not rescue any ovary phenotypes (Figure 2a,b). Consistent with previous reports of piwi phenotypes, escort cell-specific knocking down of other Piwi/piRNA pathway effectors, such as panx, also showed a similar GSC-l accumulation phenotype (Figure 2d,e), further supporting the idea that Ova may participate in the same Piwi/piRNA pathway. Next, we tested whether overexpression of ova could rescue the germline TE upregulation phenotype caused by piwi mutation. As a control, ubiquitous expression of piwi, but not soma-only expression of piwi was able to effectively rescue the TE upregulation phenotype. However, neither ubiquitous nor soma-specific expression of ova could rescue the TE upregulation phenotype in piwi mutants (Figure 2f). These observations suggest that, genetically, ova acts downstream of piwi, but there must be additional factors downstream of piwi that cooperatively function with ova to regulate GSC differentiation and transposon silencing.

Ova acts downstream of Piwi genetically.

(a) Phase contrast images of dissected ovaries from flies of indicated genotypes. Scale bar, 500 µm. (b) A graph shows the percentages of germaria with rudimentary or rescued ovaries (n = 62, 51, 143, 198 respectively). (c) Confocal sections of piwi2/3 and ova-rescued ovaries. Arrow indicates the GSC-like tumor; asterisk indicates the developing germline cyst. Red, α-Spectrin. Scale bars, 10 µm. (d) Confocal sections of piwi RNAi and panx RNAi germaria. Red, α-Spectrin. Scale bars, 10 µm. (e) Quantification of GSC-like cell number in germaria of indicated genotypes. (n = 55, 43 respectively). f, qPCR result of TE levels in total ovarian RNA from indicated genotypes (normalized to actin5c). Values are means ± SEM.; n = 3. P values by two-tailed t-test (*, p<0.05; **, p<0.01; ***, p<0.001).

https://doi.org/10.7554/eLife.40806.006

Given that Piwi is associated with a number of chromatin factors that are known to regulate heterochromatin formation and germline transposon silencing, and considering that dpp silencing in escort cells can be attributed to Piwi-dependent gene silencing, we asked whether Ova is also associated with these silencing machinery components and somehow participates in these processes. We performed a yeast two-hybrid (Y2H) screen for potential physical interactions among Ova and other known essential components of the heterochromatin silencing machinery (Yu et al., 2015; Sienski et al., 2015; Sienski et al., 2012), including: Panoramix (Panx), Arx, and Mael, which participates in Piwi/piRNA-mediated gene silencing (Yu et al., 2015; Sienski et al., 2015; Sienski et al., 2012; Muerdter et al., 2013; Dönertas et al., 2013; Ohtani et al., 2013); HP1a, the H3K9me3 methyltransferase Eggless (Egg), and the Egg cofactor Windei (Wde) (Seum et al., 2007; Tzeng et al., 2007; Koch et al., 2009); the H3K4me2 demethylase dLsd1 and its cofactor CoREST (Rudolph et al., 2007); and Piwi. The Y2H screen identified two positive interactions: Ova and HP1a, and Ova and dLsd1 (Figure 3a and Figure 3—figure supplement 1). Notably, the previously reported interaction between HP1a and Piwi was not observed in our screen here (Figure 3—figure supplement 1) (Brower-Toland et al., 2007), possibly due to different expression systems used in the studies. Co-immunoprecipitation experiments also showed positive interactions between Ova and HP1a and between Ova and dLsd1 in ovary lysates (Figure 3b,c). Collectively, these results indicate that Ova is physically associated with the co-transcriptional silencing machinery and directly interacts with HP1a and with dLsd1. Previous studies have reported that HP1a and dLsd1 function in the escort cell niche to restrict dpp signaling and to facilitate GSC differentiation (Wang et al., 2011; Eliazer et al., 2011). These reports, considered alongside the known role of Piwi-dependent gene silencing of the dpp gene locus in normal escort cells, further supporting the notion that these three Piwi-associated factors (Ova, HP1a, dLsd1) function in a shared pathway in escort cells to establish a repressive chromatin state for the dpp gene locus.

Figure 3 with 3 supplements see all
Ova interacts with the heterochromatin machinery.

(a) Y2H assay for protein interaction between Ova and proteins as indicated. (b–c) Western blots showing reciprocal co-IP between Ova and HP1a, and between Ova and dLsd1. The RFP-HP1a transgene was driven by the endogenous promoter. The dLsd1-GFP transgene was driven by a ubiquitous promoter. The Flag-ova transgene was driven by nos-GAL4. (d) Heat map displaying steady state mRNA levels as reads per million (rpm) for the top 60 detected transposons in nosGAL4 driven ova-RNAi, EGFP-RNAi, and w1118 ovaries. The average of three replicates is shown. The most upregulated transposons are highlighted in bold. (e) Correlation scatter plot of log10 transposon mRNA-seq reads between ova GLKD and dLsd1 GLKD ovaries. R = 0.9463, p<2.2×10−16 by Pearson’s correlation coefficient. The most upregulated transposons in both genotypes are highlighted in red dots.

https://doi.org/10.7554/eLife.40806.007

We next tested whether Ova, similar to HP1a and dLsd1 (Wang and Elgin, 2011; Czech et al., 2013), is required for heterochromatin formation and germline transposon silencing. The white locus of In(l)wm4h chromosomal reversion flies is relocated to a position next to a heterochromatin region, and this relocation often causes heterochromatin-based silencing of this gene, resulting from a genomic phenomenon referred to as position effect variegation (PEV), these flies typically display mosaic eyes with red and white facets as a result of this relocation based silencing (Wallrath and Elgin, 1995; Schotta et al., 2003) (Figure 3—figure supplement 2a). Interestingly, removing one functional copy of ova from the In(l)wm4h background was sufficient cause fully-pigmented eyes (Figure 3—figure supplement 2a). Analysis using several additional PEV reporter fly lines (118E-10, 118E-15, 39 C-72, and 6 M-193), each of which has its white gene locus relocated (inserted) into the heterochromatin rich fourth chromosome, showed that ova acts as a suppressor of PEV: the ova transheterozygous flies had fully-pigmented eyes with increased pigment level whereas the ova heterozygous flies from all three of the reporter lines had mosaic eyes (P values by two-tailed Student t-test, Figure 3—figure supplement 2b,c). It thus appears that ova has a functional role in heterochromatic gene silencing.

To test whether or not ova functions in germline transposon silencing, we performed germline-specific knock-down of ova using the UAS-Dcr2; nos-GAL4 driver (ova GLKD), followed by RNA-seq analysis. Interestingly, nos > ova RNAi ovaries had dramatically up-regulated transcripts of a subset of transposons that included the LTR element 412 and the telomeric non-LTR repeats Het-A, TAHRE, Tart (Figure 3d). By comparison, the expression of protein-coding genes and piRNAs was largely un-altered (Figure 3—figure supplement 3a,b,e). Somatic cell-specific knock-down of ova (tj-GAL4 >ova RNAi) only caused mild, if any, TE upregulation (Figure 3—figure supplement 3f). Consistent with a role in germline transposon silencing, a previously reported genetic screen for genes involved in germline transposon silencing identified ova (CG5694) as one of the top hits (Muerdter et al., 2013; Czech et al., 2013). Notably, germline-specifc knock-down of either ova (ova GLKD) or dLsd1 (dLsd1 GLKD) exhibited de-repression of a similar subset of transposons (R = 0.9463 by Pearson’s correlation coefficient, Figure 3e); this subset is distinguished by enrichment for bivalent histone marks (both H3K9me3 and H3K4me2) (Czech et al., 2013; Klenov et al., 2014). In mutants deficient in piRNA biogenesis, the inability to form Piwi/piRNA complexes typically results in retention of Piwi in the cytoplasm (Wang and Elgin, 2011; Malone et al., 2009; Olivieri et al., 2010). The fact that nuclear Piwi localization was largely unaffected in the ova mutant germline (Figure 3—figure supplement 3c) further supports our conclusion that ova is not required for piRNA biogenesis, but may function at the chromatin to mediate Piwi/piRNAs- induced transcriptional gene silencing, a phase that has been referred to as ‘effector step’ (Czech et al., 2013).

To explore the biochemical mechanisms underlying Ova function in greater detail, we used Y2H assays to identify the Ova domains required for its interactions with HP1a and/or dLsd1. We constructed multiple truncated forms of Ova (Figure 4a), and found that the Ova 250–486 fragment and the Ova 388–623 fragment were both able to interact with the chromo shadow domain (CSD) of HP1a (Figure 4b); neither of these Ova fragments could interact with the chromodomain (CD) of HP1a (Figure 4b). We next constructed an Ova fragment composed of the overlapped 388–486 region and confirmed that this fragment was sufficient for interaction with the CSD domain of HP1a (Figure 4b). Mapping the interaction domains of Ova with dLsd1 revealed that both Ova 1–388 and Ova 250–486 fragments, but not Ova 388–486 fragment, could interact with dLsd1 (Figure 4c). Interestingly, transgene expression of the Ova 250–486 fragment, which is able to interact with both HP1a and dLsd1, was sufficient to rescue both the ovary development defect and transposon silencing defect of ova mutant females, similar to the effect produced by transgene expression of a full length ova (Figure 4d,f). In contrast, no rescue effect was observed with the transgenic expression of the Ova 388–623 fragment, which interacts with HP1a only, or with expression of the Ova 1–388 fragment, which interacts with dLsd1 only (Figure 4d). Therefore, the domain that is sufficient to interact with both HP1a and dLsd1 is sufficient for Ova function in ovary development and transposon silencing. These biochemical and genetic experiments indicate that Ova may serve as a protein adaptor that links HP1a and dLsd1. To functionally test this putative adaptor function in vivo, we generated a transgene expressing HP1a::dLsd1 fusion protein. If Ova merely functions as an adapter that bridges the two proteins, the HP1a::dLsd1 transgene should render Ova dispensable and therefore should be able to rescue the ova mutant phenotypes. Strikingly, transgenic expression of HP1a::dLsd1 in escort cells was sufficient to rescue the rudimentary ovary phenotype of ova mutants (Figure 4e). Eighty percent of the HP1a::dLsd1 rescued germaria contained 2–5 GSC-l (n = 41) and all the germaria had properly differentiating cysts. Moreover, ubiquitous expression of HP1a::dLsd1 also significantly rescued the transposon silencing defects of ova mutants and partially restored female fertility (Figure 4f). Given that the genomic fragment transgene of ova (ova-g), which includes the cis-elements of ova, could fully restore fertility (Figure 1a), the incomplete rescue of fertility by the HP1a::dLsd1 fusion could be due to non-physiological levels of the transgene expression. Alternatively, ova could have additional roles beyond the adaptor role that are important for female fertility. In addition to increased expression of transposons, ova germline mutant ovaries also showed moderate upregulation of many protein-coding genes (Figure 4—figure supplement 1). Interestingly, this transgene expression also effectively brought the expression of many protein-coding genes back to wild-type levels (Figure 4—figure supplement 1). These observations indicate that HP1a and Ova may participate in transcriptional silencing of many regular protein-coding genes, in addition to transposons. We conclude that Ova acts as a protein adaptor to link HP1a and dLsd1 to promote HP1a-mediated gene silencing.

Figure 4 with 1 supplement see all
Ova acts as a protein adaptor to link dLsd1 with HP1a.

(a) Schematic drawings of full length and truncated forms of Ova. (b) Mapping the reciprocal-binding regions between HP1a and Ova by Y2H assay. (c) Mapping the reciprocal-binding regions between HP1a and dLsd1 by Y2H assay. (d) Ovaries from flies of indicated genotypes. Escort cell-specific expression of ova full length, ova250-486 or HP1a::dLsd1 rescued ova1/4 ovary defect. Scale bar, 500 µm. (e) A representative image of ova1/4 germarium rescued by escort cell-specific expression of HP1a::dLsd1. Red, α-spectrin; Blue, DAPI. Scale bar, 10 µm. (f) A graph shows the total offspring number of indicated females (n = 11, 23 respectively). (g) A graph shows fold changes of TEs in total ovarian RNA from indicated genotypes (normalized to actin5c). Values are means ± SEM.; n > 4. P values by two-tailed t-test.

https://doi.org/10.7554/eLife.40806.011

Since dLsd1 catalyzes H3K4me2 demethylation, Ova may function to link dLsd1 and HP1a for local H3K4 demethylation during heterochromatic gene silencing. Indeed, ChIP-seq analysis revealed that the H3K4me2 density was specifically increased at Het-A and TAHRE loci but not other TE loci (Figure 5—figure supplement 1). Further analysis revealed that there was a significant increase in H3K4me2 levels and in RNA Pol II occupancy at the 3’UTR of the Het-A and TAHRE transposons in ova GLKD ovarian germline cells (Figure 5a,b). Note that these telomeric transposons are arranged in a head-to-tail fashion; therefore, the 3’ UTR of one element likely directs the transcription of its downstream neighbor (Danilevskaya et al., 1997). To further test this potential role of Ova in linking H3K4 demethylation during HP1a-mediated gene silencing in vivo, we used a clean lacI/lacO reporter system to tether lacI-HP1a to the promoter of a lacO-GFP reporter (Sienski et al., 2015). We found that 26% of ovarioles examined (n = 131) had reduced GFP signal in their germline upon lacI-HP1a induction (Figure 5c,d), although there was no significant reduced in the overall level of GFP mRNA (P value by two-tailed Student t-test, Figure 5e). Importantly, co-expression of ova in the germline caused a significant increase in the number of ovarioles with reduced or abolished GFP signal [86% (n = 138)], and the overall GFP mRNA level was also significantly reduced in these samples (P value by two-tailed Student t-test, Figure 5c–e). ChIP-seq analysis showed that this reduction in reporter expression was accompanied by significantly reduced H3K4m2 levels near the promoter region of the GFP gene reporter (Figure 5f,h).

Figure 5 with 2 supplements see all
Ova regulates HP1a-induced local H3K4 demethylation.

(a) Graphs showing H3K4me2 and Pol II ChIP-seq profiles mapped to indicated transposon loci in control versus ova GLKD ovaries. Dashed boxes, enhancer regions of transposons. RPM, reads per million. Bin, 100 bp. (b) Quantitative comparison of H3K4me2 and Pol II densities in the indicated enhancer regions in 5a (dashed boxes). P values by two-tailed Student’s t-test. (c), Ovarioles with indicated genotypes expressing ubiquitous lacO-GFP reporter in the germline cells. GFP was visualized by antibody staining. Scale bar, 50 µm. (d) A graph showing the percentage of ovarioles of indicated genotypes with normal, reduced, or abolished GFP signals. (e) Quantitative RT-PCR results of GFP mRNA from ovaries of indicated genotypes. Values are means ±SEM.; n > 4. P values by two-tailed Student’s t-test. (f) Graphs showing normalized H3K4me2 density mapped to lacO-GFP reporter region from ovaries of indicated genotypes. Grey box, lacO-binding sites; Purple box, nanos promoter. (g) Graphs showing normalized H3K4me2 density mapped to lacO-terminator-GFP reporter region from ovaries of indicated genotypes. Blue box, VASA terminator. (h) Quantitative comparison of H3K4me2 density in regions indicated by the dashed boxes in f or g. P values by two-tailed Student’s t-test. (i) Quantitative RT-PCR results of GFP mRNA from lacO-GFP and lacO-terminator-GFP reporter ovaries. Values are means ± SEM.; n = 4. P value by Student t-test. (j) A schematic model for Ova function: Ova functions as a protein adaptor to link HP1a with dLsd1 for local H3K4 demethylation during HP1a-induced transcriptional gene silencing.

https://doi.org/10.7554/eLife.40806.013

To further confirm that the alteration of H3K4 deposition is a consequence of Ova recruitment, rather than a secondary effect following altered gene transcription, we performed a similar set of experiments, but with a modified lacO-terminator-GFP reporter that has a transcriptional terminator immediately following the promoter (Figure 5g). This should result in blocked transcription no matter whether a transcriptional activator/repressor is present or not. As expected, this reporter showed a significant reduction of baseline transcription (down to approximately 3.8%) (P value by two-tailed Student t-test, Figure 5i). We found that tethering lacI-HP1a to the promoter failed to alter the H3K4me2 level proximal to the tethering site. Co-expression of Ova, however, almost erased entirely the H3K4me2 marks in the proximal region (Figure 5g,h). These observations further support the notion that Ova links HP1a and dLsd1 for local erasing of H3K4me2 marks.

We also tested whether Ova itself can induce gene silencing by tethering Ova directly to DNA and to mRNA using in vivo reporter systems in the ovarian germline. We used a lacI-Ova and lacO-GFP binary system to tether Ova to genomic DNA (Sienski et al., 2015) and found that such tethering did not have any obvious effect on GFP expression (Figure 5—figure supplement 2a–c). Similarly, tethering Ova to mRNA using a λN-Ova and GFP-boxB binary system did not cause any obvious effect on GFP expression (Figure 5—figure supplement 2d–f). These results are consistent with the idea that Ova acts downstream of HP1a in heterochromatic gene silencing.

Similar to other ‘effector step’ mutations, the loss of ova or dlsd1 only causes de-repression of a subset of transposons; this is in contrast with the widespread transposon de-repression that is common in mutations affecting piRNA biogenesis (Czech et al., 2013). This disparity can possibly be explained by the existence of different silencing mechanisms for particular subsets of transposons. Illustrating this idea, our work suggests that transposons with bivalent histone marks may be preferential targets for Ova and dLsd1. A bivalent pattern of histone methylation may help to regulate the expression of transposons that require a delicate On/Off balance, for example with the expression of telomeric repeats known to be required for normal telomere function (e.g., Het-A, TAHRE, and Tart). The results of our study establishes that Ova has an indispensable role in facilitating dLsd1’s H3K4 demethylation activity during HP1a-induced heterochromatic gene silencing and demonstrates that this Ova function is essential for germline development, heterochromatin formation, and Piwi/piRNA-mediated co-transcriptional gene silencing. Our study suggests that the Piwi/piRNA pathway may adapt a similar effector machinery to repress regular genes, such as the dpp gene in escort cells, in addition to TEs. A study in S. pombe reported a mechanism in which a RNAi protein complex links the activity of the H3K9 methyltransferase Clr4 with H3K4 demethylation by the H3K4 demethylase Lid2 (Li et al., 2008), indicating an evolutionarily conserved interplay of epigenetic marks during transcriptional gene silencing. Given that the mechanisms underlying heterochromatic gene silencing are known to be strongly conserved from Drosophila to mammals, an equivalent functional module that links HP1a with H3K4 demethylation likely exists in mammals as well.

Materials and methods

Key resources table
Reagent type (species) or
resource
DesignationSource or referenceIdentifiersAdditional information
Genetic reagent
(Drosophila
melanogaster)
ova[1]This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
ova[4]This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
c587-GAL4(Song et al., 2004)
(DOI: 10.1242/dev.01026)
RRID:BDSC_67747
Genetic reagent
(Drosophila
melanogaster)
Dad-lacZ(Tsuneizumi et al., 1997)
(DOI: 10.1038/39362)
RRID:DGGR_118114
Genetic reagent
(Drosophila
melanogaster)
bam-GFP(Chen and McKearin, 2003)RRID:DGGR_118177
Genetic reagent
(Drosophila
melanogaster)
piwi[2](Lin and Spradling, 1997)RRID:BDSC_43319
Genetic reagent
(Drosophila
melanogaster)
piwi[3](Lin and Spradling, 1997)RRID:BDSC_12225
Genetic reagent
(Drosophila
melanogaster)
GFP-piwiKatalin Toth
(California Institute of
Tchnology)
Genetic reagent
(Drosophila
melanogaster)
118E-10Lori Wallrath
(University of Iowa)
Genetic reagent
(Drosophila
melanogaster)
118E-15Lori Wallrath
(University of Iowa)
Genetic reagent
(Drosophila
melanogaster)
6 M-193Lori Wallrath
(University of Iowa)
Genetic reagent
(Drosophila
melanogaster)
39C.72Lori Wallrath
(University of Iowa)
Genetic reagent
(Drosophila
melanogaster)
dLsd1-GFPYu Yang (Institute of
Biophysics, Chinese
Academy of Science)
Genetic reagent
(Drosophila
melanogaster)
EGFP-RNAiBloomington Drosophila
Stock Center
(#41553)
Genetic reagent
(Drosophila
melanogaster)
RFP-HP1aBloomington Drosophila
Stock Center
(#30562)
Genetic reagent
(Drosophila
melanogaster)
UAS-Dcr2; nos-GAL4Bloomington Drosophila
Stock Center
(#25751)
Genetic reagent
(Drosophila
melanogaster)
tub-GAL4Bloomington Drosophila
Stock Center
(#5138)
Genetic reagent
(Drosophila
melanogaster)
tub-GAL80tsBloomington Drosophila
Stock Center
(#7016, #7018)
Genetic reagent
(Drosophila
melanogaster)
Df(2L)BSC144Bloomington Drosophila
Stock Center
(#9504)
Genetic reagent
(Drosophila
melanogaster)
attP2Bloomington Drosophila
Stock Center
(#25710)
Genetic reagent
(Drosophila
melanogaster)
In(1)wm4hKyoto Stock Center(#101652)
Genetic reagent
(Drosophila
melanogaster)
Df(2L)ED737Kyoto Stock Center(#150520)
Genetic reagent
(Drosophila
melanogaster)
ova-RNAiVienna Drosophila
Research Center
(#102156)
Genetic reagent
(Drosophila
melanogaster)
piwi-RNAiVienna Drosophila
Research Center
(#101658)
Genetic reagent
(Drosophila
melanogaster)
panx-RNAiVienna Drosophila
Research Center
(#102702)
Genetic reagent
(Drosophila
melanogaster)
EGFP-5xBoxBVienna Drosophila
Research Center
(#313408)
Genetic reagent
(Drosophila
melanogaster)
lacO-GFP-PiwiVienna Drosophila
Research Center
(#313394)
Genetic reagent
(Drosophila
melanogaster)
lacI-HP1a; lacO-
GFP-Piwi
Vienna Drosophila
Research Center
(#313409)
Genetic reagent
(Drosophila
melanogaster)
nos-Cas9Jianquan Ni
(Tsinghua University)
Genetic reagent
(Drosophila
melanogaster)
pCasper4-ova-gThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
GFP-ovaThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-ovaThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-piwiThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-ova1-249This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-ova250-486This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-ova388-623This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-ova1-388This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-HP1a::dLsd1This paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-λN-ovaThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
UASP-lacI-ovaThis paperSee Materials and
methods
Genetic reagent
(Drosophila
melanogaster)
lacO-terminator-GFP-PiwiThis paperSee Materials and
methods
Recombinant DNA
reagent
UASP-λNJulius Brennecke
(Institute of Molecular
Biotechnology)
Recombinant DNA
reagent
UASP-lacI
Recombinant DNA
reagent
lacO-GFP-Piwi
Recombinant DNA
reagent
pGBKT7Clontech (Cat#630443)
Recombinant DNA
reagent
pGADClontech (Cat#630442)
Antibodyrabbit polyclonal
anti-pMad
Ed Laufer (Columbia
Universtity Medical
Center)
RRID:AB_2617125IHC(1:1000)
Antibodyrabbit polyclonal
anti-β-galactosidase
MP Biologicals
(Cat#0855976)
RRID:AB_2687418IHC(1:3000)
Antibodymouse monoclonal
anti-α-Spectrin
Developmental Studies
Hybridoma Bank
IHC(1:50)
Antibodymouse monoclonal
anti-Tubulin
Developmental Studies
Hybridoma Bank
RRID:AB_1157911WB(1:2000)
Antibodyrabbit polyclonal
anti-mCherry
BioVision (cat#5993)RRID:AB_1975001WB (1:2000)
Antibodyrabbit polyclonal
anti-GFP
Life (cat#A11122)RRID:AB_221569IHC(1:1000)
WB(1:10000)
Antibodypolyclonal
anti-rabbit IgG-HRP
ZSJQ-BIO (cat#ZB2301)WB(1:10000)
Antibodyrabbit polyclonal
anti-H3K4me2
Abcam (cat#ab7766)RRID:AB_732924
Antibodymouse monoclonal
anti-RNA polymerase II
Abcam (cat#ab817)RRID:AB_306327
Antibodymouse monoclonal
anti-Flag
Sigma (cat#F1804)RRID:AB_439685IHC(1:300)
WB(1:6000)
Chemical
compound, drug
4’,6’-diamidino-2-
phenylindole
Sigma
(cat#10236276001)
Commercial
assay or kit
anti-Flag resinSigma (cat#A2220)RRID:AB_10063035
Commercial
assay or kit
GFP-Trap agaroseChromoteck
(cat#gta-10)
Commercial
assay or kit
RFP-Trap agaroseChromoteck
(cat#rta-10)
Commercial
assay or kit
Qiagen Plasmid
Midi Kit
Qiagen (#12145)
Commercial
assay or kit
Immobilon Western
Chemiluminescent
HRP Substrate Kit
Millipore
(cat#WBKLS0500)
Commercial
assay or kit
HiScript II Q RT
SuperMix
Vazyme Biotech
(cat#R223-01)
Commercial
assay or kit
ChamQ SYBR qPCR
master Mix
Vazyme Biotech
(cat#Q331)
Commercial
assay or kit
Oligo d(T)25 Magnetic
beads
NEB (cat#S1419S)
Commercial
assay or kit
NEBNext Ultra IIDNA
Library Prep Kits for
Illumina
NEB (cat# E7645S)
Commercial
assay or kit
VAHTS Small RNA
Library Prep Kit
for Illumina
Vazyme Biotech
(cat#NR801)
Commercial
assay or kit
VAHTS Universal DNA
Library Prep Kit
Vazyme Biotech
(cat#ND607)
Commercial
assay or kit
TruePrep Index KitVazyme Biotech
(cat#TD202)
Software,
algorithm
GraphPad PrismGraphPad Prism
(https://graphpad.com)
RRID:SCR_002798
Sequenced-based
reagent
RT-qPCR primersThis paperSee Supplementary file 1,
Table 2

Drosophila strains

Request a detailed protocol

Flies were cultured on standard media with yeast paste added to the food surface. The culture temperature was 25°C unless otherwise noted. Strains used in this study were as follows: ova1 is nucleotide loss allele (A1045) generated in this study. ova4 is a knock-out allele generated in this study by CRISPR-Cas9 (Ren et al., 2013). c587-GAL4 (Song et al., 2004); Dad-lacZ (Tsuneizumi et al., 1997); bam-GFP (Chen and McKearin, 2003); piwi (Lachner et al., 2001) and piwi (Bannister et al., 2001) (Lin and Spradling, 1997); GFP-piwi (gift from Katalin Toth, California Institute of Technology); 118E-10, 118E-15, 6 M-193, and 39C.72 (gift from Lori Wallrath, University of Iowa); dLsd1-GFP (gift from Yang Yu, Institute of Biophysics IBP, Chinese Academy of Sciences); from Bloomington Drosophila Stock Center (BDSC):EGFP-RNAi (#41553) RFP-HP1a (#30562);; UAS-Dcr2; nos-GAL4 (#25751); tub-GAL4 (#5138); tub-GAL80ts (#7016, #7018); Df(2L)BSC144 (#9504); attP2 (#25710); from Kyoto Stock Center: In(1)wm4h (#101652); Df(2L)ED737 (#150520); from Vienna Drosophila Research Center: ova-RNAi (#102156); piwi-RNAi (#101658); panx-RNAi (#102702); EGFP-5xBoxB (#313408); lacO-GFP-Piwi (#313394); lacI-HP1a; lacO-GFP-Piwi (#313409).

Generation of knock-out and transgenic flies

Request a detailed protocol

To obtain ova knock-out allele, two gRNAs (gRNA1: aagtctttacagccttgatc and gRNA2: cgttgggttgaggtacatac) were designed that target ova 5’UTR and 3’UTR respectively and cloned into U6b vector. The plasmids were introduced into nos-Cas9 embryos (Ren et al., 2013). Obtained flies were backcrossed with w1118 for at least three generations to eliminate potential off-target events. For ova-g transgenic fly, w1118 genomic region (2L: 10226867–10234857) was cloned intro pCasper4 vector. The attP-UASP vector was used to generate UASP-Flag-ova, UASP-Flag-ova1-388, UASP-Flag-ova1-249, UASP-Flag-ova250-486, UASP-Flag-ova-388–623, UASP-ova, UASP-piwi, and UASP-HP1a::dLsd1. The GFP-ova construct was obtained using Gateway cloning technology (Invitrogen) and pUGW (DGRC1283) vector. Ova cDNA was cloned into UASP-λN and UASP-lacI (gifts from Julius Brennecke, Institute of Molecular Biotechnology) to generate the UASP-λN-ova and UASP-lacI-ova transgenes respectively. For the lacO-terminator-GFP reporter, 555 bp VASA terminator was injected immediately following start codon of GFP in the lacO-GFP reporter. All the plasmids were purified using a Qiagen Plasmid Midi Kit (#12145) and the DNA sequencing verified plasmids were introduced into embryos using either P-element or nos-phiC31 system to generate transgenic flies according to a standard procedure.

Immunostaining

Request a detailed protocol

Drosophila ovaries were dissected and immunostained as described previously (Yang et al., 2015). Briefly, ovaries were fixed in 4% paraformaldehyde for 15 min, and blocked in 5% normal goat serum in PBT (10 mM NaH2PO4, 175 mM NaCl, pH 7.4, 0.1% Triton X-100). The following primary antibodies were used: rabbit anti-pMad (1:1000, gift from Ed Laufer, Columbia University Medical Center, New York), rabbit anti-β-galactosidase (1:3000; MP Biologicals, 0855976), mouse anti-α-Spectrin (1:50; DSHB), rabbit anti-GFP (1:1000; Life, A11122), mouse anti-Flag (1:300; Sigma, F1804). Secondary antibodies, including goat anti-rabbit, anti-mouse IgGs, conjugated to Alexa (488 or 568) (Molecular Probes) were used at a dilution of 1:300 and tissues were also stained with 0.1 mg/ml DAPI (4’,6’-diamidino-2-phenylindole; Sigma) for 5 min. Images were collected using either a Zeiss LSM510/LSM 800 or Nikon A1 confocal microscope system. All acquired images were processed in Adobe Photoshop and Illustrator.

Fertility test

Request a detailed protocol

To test female fertility, for each vial, three newly enclosed females were collected and mated with three 5–7 days old w1118 males in cornmeal food with yeast paste for two days, then the flies were transferred to a cornmeal food vial without yeast paste. After another three days, the flies were dumped out. The number of offspring was accounted until 16 days after eclosion. Mean values are reported as SEM.

Drosophila eye pigmentation assay

Request a detailed protocol

To measure eye pigmentation, the heads of ten 5–7 days old flies of each genotype were manually dissected. The isolated heads were homogenized in 0.2 ml of methanol, acidified with 0.1% HCl and warmed at 50°C for 5 min; The homogenate was clarified by centrifugation, and the OD at 480 nm of 0.15 ml supernatant was recorded. Mean values are reported with SEM.

Yeast two-hybrid experiment

Request a detailed protocol

Yeast Two-hybrid experiment was performed as described previously (Yang et al., 2015). Briefly, cDNA encoding interesting genes were amplified from w1118 ovary cDNA and cloned into either pGBKT7 bait vector or pGAD prey vector (Clontech). The pGBKT7 and pGAD plasmid carrying interesting genes were co-transformed into AH109 yeast cells according to a standard procedure. Colonies appearing on media lacking tryptophan and leucine (SC-WL) were picked onto selection plate lacking tryptophan, leucine and histidine (SC-WLH) or tryptophan, leucine, histidine and adenine (SC-WLHA) to determine proteins interaction.

Co-immunoprecipitation

Request a detailed protocol

Co-IP was done as previously described (Yang et al., 2015), with minor modifications. Female flies of appropriate genotypes were dissected in ice cold PBS. Ovaries were lysed in lysis buffer (10 mM Hepes pH 7.0, 150 mM NaCl, 5 mM MgCl2, 10% glycerol, 1% Triton X-100, 1x complete protease inhibitor (Roche), 1 mM DTT, 1 mM EDTA, 0.1 mM PMSF) at 4°C for 30 min and spun for 10 min at max speed in a table top centrifuge at 4°C. The supernatant was incubated with tag-recognizing beads including anti-Flag resin (Sigma), GFP-Trap agarose beads (Chromoteck) and RFP-Trap agarose beads (Chromoteck). After incubation, the beads were washed three times with lysis buffer and eluted by boiling in SDS loading buffer, loaded onto SDS-PAGE gels, and analyzed by immunoblotting with indicated antibodies. The following primary antibodies were used: anti-Flag (Sigma, 1:6000), anti-GFP (Life, 1:10000), anti-mCherry (BioVision, 1:2000), anti-Tubulin (DSHB, 1:2000). Secondary antibodies, including: anti-mouse and anti-rabbit IgG-HRP (ZSJQ-BIO, 1:10000). The membrane was developed by Immobilon Western Chemiluminescent HRP Substrate Kit (Millipore) according to the manufacturer’s instructions.

RNA purification and real-time quantitative PCR (RT-qPCR)

Request a detailed protocol

Total RNA from 10 to 20 ovaries was extracted using TRIzol reagent (TaKaRa). After DNase treatment, complementary DNA (cDNA) was synthesized using HiScript II Q RT SuperMix (Vazyme Biotech, R223-01). RT-qPCR was performed in three duplicates using ChamQ SYBR qPCR master Mix (Vazyme Biotech, Q331) on an ABI PRISM 7500 fast real-time PCR system (Applied Biosystems). Fold changes for mRNA were calculated using the △△Ct method (Livak and Schmittgen, 2001). Primers used were shown in Supplementary file 1, Table 2.

RNA sequencing and computational analysis

Request a detailed protocol

Total RNA from ovaries was isolated using TRIzol reagent (TaKaRa). 10 μg of total RNA from each sample used for library preparation after poly(A)-containing mRNA molecule purification (NEB, #S1419S), RNA amplification, double-strand cDNA synthesis, and adaptor ligation (NEB, #E7645S). For the small RNA sequencing, 10 μg enriched small RNA were separated on a 15% denaturing polyacrylamide gel and 18- to 30-nt RNAs were purified according to RNA oligo markers. All the libraries were prepared by following the manufacturer’s instructions and subsequent sequencing on the Illumina GAII instrument (Vazyme, NR801). For CDS gene expression analysis, all the sequencing reads were mapped to the D.mel genome (BDGP6) using STAR program (options: --outFilterMultimapNmax 20 --alignIntronMin 20 --alignIntronMax 500000). The mapped reads were used for expression analysis via Cufflinks package with reference gene annotation from Ensembl. And Cuffdiff was used to perform differential expression. For transposon expression analysis, sequencing reads were mapped to the transposon sequences which download from flybase website using STAR program with default parameters. Then alignment reads were used for calculating the expression level of transposons. Different transposons were combined together if they belong to the same one. The expression levels were normalized to reads per million (RPM). For small RNA analysis, Cutadapt package was used to remove adapter from 3’ end. The reads were aligned to the genome sequence by Bowtie. The reads were discarded which mapped to rRNA, tRNA, snoRNA sequences. And retained reads were aligned to miRNA (pre-miRNA sequences download from miRBase) and whole genome sequences (r5.42) with one mismatch and unique hit. Sequences in the 25–32 nt size range, not annotated as a previously known RNA were classified as candidate piRNAs. The expression levels of small RNA were normalized to RPM according to the total mapped reads number.

ChIP-seq analysis

Request a detailed protocol

ChIP was performed as previously described (Sienski et al., 2012). Briefly, about 200 pairs of ovaries were dissected into cold PBS and washed once. Ovaries were cross-linked in 1.8% paraformaldehyde for 10 min at room temperature then quenched with glycine. Ovaries were homogenized by douncing. Pellet was resuspended in lysis buffer and incubated 10 min on ice. Chromatin was sonicated for immunoprecipitation and followed by reverse crosslink and DNA purification. Recovered DNA fragment was used to prepare libraries using VAHTS Universal DNA Library Prep Kit (Vazyme, ND607) and TruePrep Index Kit (Vazyme Biotech, TD202) sequencing was done on HiSeq2500 (Illumina). Antibodies: polyclonal rabbit anti-H3K4me2 (Abcam, ab7766) and monoclonal mouse anti-RNA polymerase II (Abcam, ab817). ChIP-seq reads were aligned using Bowtie (version 1.1.2) to build version BDGP6 of the Drosophila melanogaster genome. MACS (version 1.4.1) was used to identify regions of ChIP-seq enrichment. The density of reads in each region was normalized to 10 million reads library size. For lacO-GFP ChIP-seq, normalized reads were removed w1118 ChIP reads as the reporter unique mapped reads due to lacO-GFP reporter shared common sequences in fly genome. BigWig files were generated for visualization using Homer package. For transposons, all raw reads were mapped to the transposon database using Bowtie (version 1.1.2) with –v 3 –-best parameters. The sum of the number reads that mapped to genome and transposon was used as a normalization factor for all samples, reporting all feature abundances as RPM mapped.

References

  1. 1
  2. 2
    Targeted gene expression as a means of altering cell fates and generating dominant phenotypes
    1. AH Brand
    2. N Perrimon
    (1993)
    Development 118:401–415.
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
    A novel group of pumilio mutations affects the asymmetric division of germline stem cells in the Drosophila ovary
    1. H Lin
    2. AC Spradling
    (1997)
    Development 124:2463–2476.
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
    The Development of Drosophila Melanogaster
    1. AC Spradling
    (1993)
    M Bate, A Martinez Arias, editors. Cold Spring Harbor Laboratory Press.
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
    Analysis of genetic mosaics in developing and adult Drosophila tissues
    1. T Xu
    2. GM Rubin
    (1993)
    Development 117:1223–1237.
  43. 43
  44. 44
  45. 45

Decision letter

  1. Yukiko M Yamashita
    Reviewing Editor; University of Michigan, United States
  2. Kevin Struhl
    Senior Editor; Harvard Medical School, United States
  3. Zhao Zhang
    Reviewer; Carnegie Institute, United States
  4. Felipe Karam Teixeira
    Reviewer; University of Cambridge, United Kingdom

In the interests of transparency, eLife includes the editorial decision letter, peer reviews, and accompanying author responses.

[Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed.]

Thank you for submitting your article "Ovaries absent links dLsd1 to HP1a for local H3K4 demethylation required for heterochromatic gene silencing" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Kevin Struhl as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Zhao Zhang (Reviewer #1); Felipe Karam Teixeira (Reviewer #2). Reviewer #3 remains anonymous.

The Reviewing Editor has highlighted the concerns that require revision and/or responses, and we have included the separate reviews below for your consideration. If you have any questions, please do not hesitate to contact us.

Summary:

In this manuscript, Yang et al. describes the role of new gene Ova required for female fertility. Whereas individual experiments are well-done overall, all reviewers are concerned that the results do not necessarily support a cohesive story at this point. Specifically, the effects in the soma vs. germline on overall sterility remains unclear, and accordingly, the impact/relevance of transposon activation in ova- on sterility is unclear. (The detailed, specific individual comments are provided below.)

Whereas all the reviewers appreciate the potential impact of this manuscript, these major concerns would be normally deemed 'not within the scope of straightforward revision' per eLife's policy. Thus, we encourage the authors to thoroughly address the major points raised by reviewers in revising this manuscript.

Separate reviews (please respond to each point):

Reviewer #1:

In this manuscript, Yang and colleagues identified a new factor, Ovaries absent (Ova), which is required for maintaining the fertility of female fruit fly. Molecularly, this factor bridges H3K4 demethylase dLsd1 and HP1a together for heterochromatic silencing. At cellular level, the authors propose: 1) Ova functions in somatic escort cells to maintain the niche environment for germline stem cell differentiation by suppressing dpp signaling; 2) in germ cells, Ova promotes transposon silencing (only a subgroup of transposons) during oogenesis. The authors use a broad range of techniques (from genetics, biochemistry, cell biology), and the conclusions are overall convincing and interesting. I particularly think that the "rescuing" experiment by HP1a::dLsd1-fusion is elegant and compelling.

Here I suggest a few further experiments/explanations that can further strengthen this story:

1) I still do not understand in germ cells, how Ova is required for fertility. Is this exclusively caused by the activation of telomeric transposons? Or there are additional triggers? Since the HP1a::dLsd1 fusion can rescue the transposon de-repression, the authors should look and report the fertility of these "rescued" flies.

2) I feel the wm4h section is distracting and unnecessary in this story. Wm4h model is sensitive to genetic background, and likely reflects epigenetic regulation during eye development. I think it would be wise to focus on oogenesis for this story. With wm4h data, the readers (like me) would ask questions regarding how epigenetic marks maintained/passed through generations in somatic cells.

3) The qPCR results (panel T of Figure 1) on dpp are not conclusive and (again) unnecessary. The panels N-S of Figure 1 already provide compelling data at protein level in early germarium. I am assuming the qPCR data (panel T) are from the RNAs extracted from whole ovaries, which are mainly from the late stage egg chambers. Therefore, the RNA data (from late stage egg chambers) could not explain the change of protein level in germarium. Also, I am confused what is ova-RNAi in panel T. Is this c578 driven RNAi or nanos driven? The authors need to explain in figure legend.

4) The authors mentioned that Ova was identified from a genetic screen. But there is no information provided on this screen.

5) The ChIP-seq data currently only reported the K4me2 deposition on two transposons. How about the K4me2 level on other transposons and other genomic regions?

6) How about the HP1a and dLsd1 localization upon Ova depletion?

7) In Figure 4—figure supplement 1, what is the genotype of ova germline mutant (again, the figure legends are not informative, need to revise)? Are ovaries from these flies normal or rudimentary? If rudimentary, given the dramatic difference of ovary structure, it does not make sense to compare them with WT ovaries.

Minor Comments:

Typos in main text paragraph seven: "of in" and "Figure A".

Main text paragraph eleven: "80.43%". I think the number has to be written as words at the beginning of a sentence.

Reviewer #2:

In this manuscript, Yang and colleagues described a new factor required for adult ovary development in flies. Through an EMS screen, they uncovered a recessive mutation in the CG5694/ova gene, which led to drastic ovarian phenotype reminiscent of that observed in piwi mutants. Similar to Piwi, Ova is required in both ovarian somatic and germ cells, and somatic tissues depleted for Ova fail to support germline stem cell (GSC) differentiation. Importantly, the authors provided evidence that over-expression of Ova in ovarian somatic cells was able to partially rescue the phenotypic defects induced by piwi mutants, indicating that both genes genetically interact. By using two-hybrid experiments, the authors provided convincing evidence that Ova can physically interact with HP1a and Lsd1, two chromatin factors involved in establishing silence chromatin. In agreement, they showed that Ova is required for the spreading of heterochromatin in somatic tissues. Finally, by using transgenic reporters and tethering experiments, the authors provided evidence that Ova recruitment to an HP1a-bound reporter locus is sufficient to locally induce H3K4 demethylation – and activity that is mediated by Lsd1.

The antagonism between H3K4 and H3K9 methylation is pervasive throughout evolution, and the balance between activities promoting each of these marks is essential for proper genome regulation. The manuscript – which is clearly presented and well written – provides strong evidence that Ova can act as a linker between the H3K9me2/3 reader HP1a and the enzyme responsible for the removal of H3K4 methylation, proposing an interesting molecular mechanism for heterochromatic establishment.

Major comment:

Transposable elements are a major target of H3K9me2/3-mediated repression mechanisms in both soma and germline. However, in the current version, the experiment testing the effect of loss of ova on transposon repression is poorly documented (Figure 3E-F). Given the importance of such analysis for the conclusion of the manuscript, I believe this needs to be better characterized.

Basically, the evidence that Ova mediates transposon silencing is provided by germline KD experiments (Figure 3E), in which the nanos-GAL4 driver is used in combination with an UAS-RNAi line for the ova gene. RNA-seq results obtained from nanos-GAL4>ova-RNAi ovaries were compared to w[1118] ovaries, suggesting the upregulation of a limited number of TE families (basically the telomeric-associated Het-A, TAHRE, and TART families). First, it is well-established in the literature that germline KD using UAS-RNAi lines originated from the VDRC collection (such as the ova-RNAi line) requires the concomitant over-expression of Drc2 (using the UAS-Dcr2 transgene; Wang and Elgin, 2011; Handler et al., 2011; Czech et al., 2013; Handler et al., 2013; Yan et al., 2014; Sanchez et al., 2016; among others). Despite that, no mention to the use of the UAS-Drc2 transgene was seen in the main text or in the Materials and methods section – the authors need to clarify this point. Second, given the existing variation on transposon content in different stock backgrounds (especially true for the telomeric transposons), it is important to compare the results from ova-RNAi ovaries to appropriate control-RNAi samples (this is also valid for the lsd1-RNAi analysis). Finally, given the requirement of ova in somatic tissues, it would be important to determine whether the same specific regulation (restricted to telomeric-associated transposons) is observed in the soma (by using tj-Gal4 or C587-Gal4 drivers to induce somatic KD).

Along the same line, it would be very important to determine whether TE-family-specific changes in piRNA accumulation can be detected in Ova germline KD experiments (Figure 3—figure supplement 3B). As it is currently presented in Figure 3—figure supplement 3B, the results indicate that Ova is not required for global piRNA production, but it is nonetheless possible that it is involved in the accumulation of piRNAs for telomeric-associated transposons. Similar to the analysis presented in Figure 3E (RNA-seq), it would be good if the authors could expand the analysis on piRNA accumulation to each individual TE family (using the existing data).

Minor Comments:

1) While introducing the process leading to the Piwi/piRNA-mediated recruitment of the silencing machinery in the first paragraph of the main text, the authors mentioned that HP1 binds to Su(var)3-9. While it has been shown that HP1 and Su(var)3-9 physically interact and cooperate in the maintenance and spreading of heterochromatic domains (mostly centromeric heterochromatin) in somatic tissues (Schotta et al., 2002), there's no evidence to date that Su(var)3-9 is involved in Piwi-mediated silencing. To the contrary, Sienski et al., 2015, has recently provided substantial data indicating that SetDB1, but not Su(var)3-9, is involved in Piwi-mediated establishment of silent chromatin. In this context, I suggest the authors revise the text to make this distinction clear to readers.

2) Given the origin of the ova[1] allele (EMS-screen) and the rather strong and specific effect on male viability, it would be useful if the authors could include the viability data for ova[1]/ova[4] trans-heterozygous in Table S1. As it is, it is not clear whether the viability defect is due to ova or to other second-site mutations induced by the EMS-mutagenesis.

3) Main text paragraph eight, in the first sentence, it should be noticed that Piwi mediates transposon silencing in both soma and germline (and not only in the germline). Most important however, while it has been shown that loss of piwi in somatic cells is associated with ectopic dpp signalling (Jin et al., 2013; Ma et al., 2014), I am under the impression that the formal demonstration that this effect is direct – and that it involves gene silencing mediated by Piwi or piRNAs – is still missing. Could the authors clarify this point?

4) In paragraph ten, the authors state that "ova and lsd1 mutants exhibited de-repression of a similar subset of transposons" (Figure 3F). However, the analysis on Figure 3 seems to concern germline KDs, and not mutants – please clarify.

5) In the Materials and methods section, please verify the source of the ova-RNAi line: it is listed as a BDSC stock, but it seems like it was in fact obtained from the VDRC collection. Please also include the source of the lsd1-RNAi line, as it is not currently listed in the Materials and methods section. Finally, clarify whether or not the UAS-Dcr2 transgene was used in the germline KD analysis.

6) In subsection “RNA sequencing and computational analysis”, please indicate the manufacture and kits used to generate the small RNA- and RNA-seq libraries.

Reviewer #3:

In Yang et al., the authors generate mutants for ovaries absent (Ova) a previously identified gene in the piRNA pathway and characterize its function. They show (1) Ova is required in the somatic escort cells for germline stem cell differentiation and in the germ line for fertility, (2) Ova acts downstream of Piwi, (3) Interacts with HP1 and dLsd1 regulators of heterochromatin formation and promotes association of HP1a and dLsd1 and (4) Can induce HP1a induced demethylation. I have some major concerns about this paper: (1) The authors switch between germ line function and somatic function without explanation, (2) the phenotypes are not clearly characterized, (3) Some of the controls for the experiments they have carried out are not correct in that they compare ovaries that have late stage egg chambers to ovaries that accumulate undifferentiated cells and (4) changes in ChIPSeq are modest at the best and are not quantitated, and I am not sure are statistically significant.

Lastly, the model is confusing. The current dogma is that demethylation of H3K4 results loss of K9ac. This then promotes methylation of K9 and which then recruits HP1. In Yang et al's model, how do they propose HP1 can get to its targets in the first place? This needs to clearly explained. I suspect, Ova should affect heterochromatin spreading not formation of heterochromatin. This is something they have not tested nor propose.

Major concerns:

Figure 1A-D: The authors have used Ova1/4 in all the rescues including Figure 1 and 2. They should include the phenotypic characterization of this allelic combination in this figure panel.

Figure 1F-G: This panel needs better pictures as even the "germless" germaria usually stain of anti-spectrin antibody. All the pictures need to be at the same magnification for easy comparison.

Figure 1J-K: The controls should include the non-restrictive temperature and well as C587 at the restrictive temperature at 14 days. 14 days is a long time for restrictive temperature. What happens in 5 -7 days? Also, what happens to protrusions? The idea about protrusions were introduced but then not followed up on.

Figure 1M: Again, please include controls for ova4

Figure 1N-S: Please include quantitation for all these phenotypes. As they show a link to piwi and piwi mutant phenotype is quite complicated, they should categorize Ova phenotype in detail.

Figure 1T: They should not use C587 as a control as this is comparing completely morphologically different ovaries. One enriches for the earlier stages and the other for later stages. They should use bam mutants as a control. Additionally, they should show the RNA seq data and tracks for these experiments for Dpp levels – these data are available to them.

They also report that loss of ova in germ line results in eggs that do not hatch as data not shown. They should report the data.

Figure 2: While the results that over expression of ova in piwi mutant ovaries is exciting they should better characterize what aspect of the piwi mutant does this protein rescue? Are the transposons levels in piwi rescued? What about protrusions? What about Dpp levels? What about Ova germ line expression in piwi mutants? Does it rescue loss of fertility in piwi mutants as well? This is important as they switch from somatic effects to germ line effects.

Figure 3: What happens to heterochromatin levels in ova mutants during oogenesis? Can they stain for H3K9me3 marks?

Figure 3E: Why are the authors looking for only upregulation of transposons in the germ line when the phenotype is mostly from the soma? They should report transposon levels in somatic loss of function of ova. As the major phenotype of piwi, dlsd1 and ova are from the escort cells. They should carry out an RNAseq for this and report the transposons upregulated in the soma.

Figure 4: In this figure, the authors switch back and forth between the somatic effects and germ line. This is not right. They should be direct that for expediency they use germ line as a read out. They should set up the paper that way as well. The paper reads as if one is discovering the role of Ova in escort cells whereas most of the functional data comes from the germ line. For example 4D vs 4F, in D they are using somatic drivers and in F they use a ubiquitous driver and measure germ line effects.

Figure 4F: Again, using w1118 as control is not right. One does not know if the undifferentiated stages express Het-A and TAHRE at higher levels compared to control. Again, the control and experiment are comparing two different morphologically different ovaries.

Figure 5A: Is this germ line or somatic KD? The authors should clearly state this. It is not available in text of figure legends. I assume it is germ line KD. How does this change compare with the genes that are not affected? Is there any quantification for these changes? I am not sure these changes 0-0.75 are real?

There should be quantification for panels A, E and F.

https://doi.org/10.7554/eLife.40806.021

Author response

Whereas all the reviewers appreciate the potential impact of this manuscript, these major concerns would be normally deemed 'not within the scope of straightforward revision' per eLife's policy. Thus, we encourage the authors to thoroughly address the major points raised by reviewers in revising this manuscript.

First, we would like to thank all of the reviewers for their careful reading of our manuscript and for their insightful and constructive comments and suggestions. In this revised manuscript, we have made intensive efforts to address each of the comments. Kindly note, however, that we have not yet finished the ChIP-seq analysis for H3K9me owing to ongoing technical challenges that we have encountered. Other than that, we feel confident that the reviewers will find we have adequately addressed the rest of the comments. As we trust you'll agree, both our study and the manuscript have been significantly improved in the course of our revision process. We believe that the current version should now be suitable for publication in eLife. Again, many thanks for your ongoing work on our behalf.

Separate reviews (please respond to each point):

Reviewer #1:

[…] Here I suggest a few further experiments/explanations that can further strengthen this story:

1) I still do not understand in germ cells, how Ova is required for fertility. Is this exclusively caused by the activation of telomeric transposons? Or there are additional triggers? Since the HP1a::dLsd1 fusion can rescue the transposon de-repression, the authors should look and report the fertility of these "rescued" flies.

We have conducted the fertility test of these “rescued” flies, and the results have been included in updated Figure 4F. Female flies of ova1/ova4 are completely sterile (Figure 1B). As a positive control, ubiquitous overexpression of Ova with tubGAL4 could restore approximately 50% fertility of ova1/ova4 females compared to WT and heterozygous controls (Figure 1B). Ubiquitous overexpression of HP1a::dLsd1 fusion protein could restore approximately 20% fertility, suggesting that this adaptor role of Ova is important for female fertility. Given that the genomic fragment transgene of ova (ova-g), which includes the cis-elements of ova, could fully restore fertility, the incomplete rescue of fertility by the HP1a::dLsd1 fusion could be due non-physiological levels of the transgene expression. Alternatively, ova could have additional roles beyond the adaptor role that are important for female fertility.

2) I feel the wm4h section is distracting and unnecessary in this story. Wm4h model is sensitive to genetic background, and likely reflects epigenetic regulation during eye development. I think it would be wise to focus on oogenesis for this story. With wm4h data, the readers (like me) would ask questions regarding how epigenetic marks maintained/passed through generations in somatic cells.

We appreciate the reviewer’s concern about this Wm4h model, so we have moved the PEV results to the Figure 3—figure supplement 2. Please note that we have tested Wm4h assay in at least two different genetic backgrounds: ova1 and ova4 heterozygous males, and they all show similar results. The effect on PEV is also found in several other PEV models in which the white gene is inserted on the fourth chromosome (Figure 3—figure supplement 2). It is thus clear that Ova is a bona fide suppressor of PEV. We think this could be a useful piece of data for the understanding of ova function, and have therefore decided to include them in the supplement of this manuscript.

3) The qPCR results (panel T of Figure 1) on dpp are not conclusive and (again) unnecessary. The panels N-S of Figure 1 already provide compelling data at protein level in early germarium. I am assuming the qPCR data (panel T) are from the RNAs extracted from whole ovaries, which are mainly from the late stage egg chambers. Therefore, the RNA data (from late stage egg chambers) could not explain the change of protein level in germarium. Also, I am confused what is ova-RNAi in panel T. Is this c578 driven RNAi or nanos driven? The authors need to explain in figure legend.

We agreed that there is abundant dpp expression in late stage egg chambers, which makes the qPCR analysis unnecessary. We have removed the data from the Figure 1. The ova-RNAi in panel T is driven by the c587-GAL4 driver, so ova was specifically knocked down in escort cells. We have included the driver information in the corresponding figures and legends throughout the manuscript to help avoid confusion.

4) The authors mentioned that Ova was identified from a genetic screen. But there is no information provided on this screen.

We have generated about 3000 EMS mutagenized lines for the FRT42B chromosome (the right arm of the second chromosome) to be used for other purposes (mosaic genetic screens). About 500 lines from this collection were homozygous viable, and were subsequently tested for female fertility. The ova1 line was initially identified from this small-scale fertility screen. The FRT42B site in the ova1 was later removed by meiotic recombination. Through deficiency map and other genetic analyses as described in the paper, we found that ova1 is allelic to CG5694, which is actually localized on the left arm of the second chromosome. We used a standard EMS mutagenesis protocol, and will describe the details elsewhere.

5) The ChIP-seq data currently only reported the K4me2 deposition on two transposons. How about the K4me2 level on other transposons and other genomic regions?

We have re-analyzed the ChIP-seq data on all the transposons (see the newly added Figure 4—figure supplement 1). H3K4me2 density is increased in Het-A and TAHRE loci, but not in the majority of other transposons. This result is consistent with the idea that Ova preferably regulates transposons at the telomeric transposons by regulating H3K4m2 levels.

6) How about the HP1a and dLsd1 localization upon Ova depletion?

We have done immunostaining against HP1a and dLsd1 in control and ova GLKD ovaries, and the results are shown in Author response image 1. There is no obvious difference in terms of general expression levels and subcellular localization between wild type and ova mutant ovaries. As HP1a and dLsd1 have thousands of genomic target sites, the more precise ChIP-seq experiments, which are technically challenging, need to be done in the future to address this question.

Author response image 1
dLsd1 and HP1a expression and subcellular localization in control and ova GLKD germaria

7) In Figure 4—figure supplement 1, what is the genotype of ova germline mutant (again, the figure legends are not informative, need to revise)? Are ovaries from these flies normal or rudimentary? If rudimentary, given the dramatic difference of ovary structure, it does not make sense to compare them with WT ovaries.

Thanks for catching this point; we indeed did not describe the genotypes clearly, which likely caused unnecessary confusion. The genotype of the germline mutant flies in that figure is c587/+; ova1/ova4; UAS-ova/+, and the ovaries from these mutant females are morphologically normal (see Figure 4D). Please note that the germline-specific ova mutants have morphologically normal ovaries, and this is also true for ova GLKD mutants (ova was specifically knocked down in the germline, see the updated Figure 1—figure supplement 2G). We have included the genotypes in the corresponding figure legends

Minor Comments:

Typos in main text paragraph seven: "of in" and "Figure A".

Corrected.

Main text paragraph eleven: "80.43%". I think the number has to be written as words at the beginning of a sentence.

Corrected.

Reviewer #2:

[…] The antagonism between H3K4 and H3K9 methylation is pervasive throughout evolution, and the balance between activities promoting each of these marks is essential for proper genome regulation. The manuscript – which is clearly presented and well written – provides strong evidence that Ova can act as a linker between the H3K9me2/3 reader HP1a and the enzyme responsible for the removal of H3K4 methylation, proposing an interesting molecular mechanism for heterochromatic establishment.

Major comment:

Transposable elements are a major target of H3K9me2/3-mediated repression mechanisms in both soma and germline. However, in the current version, the experiment testing the effect of loss of ova on transposon repression is poorly documented (Figure 3E-F). Given the importance of such analysis for the conclusion of the manuscript, I believe this needs to be better characterized.

This is an excellent point. We have tried several times for ChIP-seq analysis of H3K9me2/3 from wild type and germline mutant ovaries, but unfortunately we failed to get any high quality data. This part of the analysis has therefore been hampered by the technical difficulties that we have encountered. We immunostained H3K9me3 in ovary tissues to see if there is a global change in H3K9me3 levels, and we failed to observe any obvious alterations in H3K9me3 expression in the germline cells with ova GLKD.

Basically, the evidence that Ova mediates transposon silencing is provided by germline KD experiments (Figure 3E), in which the nanos-GAL4 driver is used in combination with an UAS-RNAi line for the ova gene. RNA-seq results obtained from nanos-GAL4>ova-RNAi ovaries were compared to w[1118] ovaries, suggesting the upregulation of a limited number of TE families (basically the telomeric-associated Het-A, TAHRE, and TART families). First, it is well-established in the literature that germline KD using UAS-RNAi lines originated from the VDRC collection (such as the ova-RNAi line) requires the concomitant over-expression of Drc2 (using the UAS-Dcr2 transgene; Wang and Elgin, 2011; Handler et al., 2011; Czech et al., 2013; Handler et al., 2013; Yan et al., 2014; Sanchez et al., 2016; among others). Despite that, no mention to the use of the UAS-Drc2 transgene was seen in the main text or in the Materials and methods section – the authors need to clarify this point. Second, given the existing variation on transposon content in different stock backgrounds (especially true for the telomeric transposons), it is important to compare the results from ova-RNAi ovaries to appropriate control-RNAi samples (this is also valid for the lsd1-RNAi analysis). Finally, given the requirement of ova in somatic tissues, it would be important to determine whether the same specific regulation (restricted to telomeric-associated transposons) is observed in the soma (by using tj-Gal4 or C587-Gal4 drivers to induce somatic KD).

We appreciate the reviewer’s points and apologize for not clearly describing our methods. We indeed used the BSC#25751 fly, whose genotype is UAS-Dcr2;nosGAL4 (referred as "GLKD" throughout the updated manuscript). We compared the w1118, ova RNAi alone, and UAS-Dcr2;nosGAL4 ovaries by qPCR and found no obvious difference among them, as shown in Author response image 2. We used tjGAL4 to knock down ova and found that TE mRNA levels only mildly increased (not statistically significant) compared to the controls, as shown in Figure 3—figure supplement 3F. It is possible that the upregulation has been underestimated because we used the whole ovary for RNA analysis, but it appears to us that the transposon derepression phenotype is much more profound in the germline than in the soma following ova depletion.

Author response image 2
qPCR to detect TE levels in controls and ova GLKD ovaries.

Along the same line, it would be very important to determine whether TE-family-specific changes in piRNA accumulation can be detected in Ova germline KD experiments (Figure 3—figure supplement 3B). As it is currently presented in Figure 3—figure supplement 3B, the results indicate that Ova is not required for global piRNA production, but it is nonetheless possible that it is involved in the accumulation of piRNAs for telomeric-associated transposons. Similar to the analysis presented in Figure 3E (RNA-seq), it would be good if the authors could expand the analysis on piRNA accumulation to each individual TE family (using the existing data).

Good point. Guided by this suggestion, we performed the analysis of the piRNA accumulation in each TE-specific piRNAs, and found that most TE-specific piRNAs are also unchanged in ova RNAi ovaries compared to control ones, excepting a mild reduction in Tart-specific piRNAs. This new data is included in the updated Figure 3—figure supplement 3E.

Minor Comments:

1) While introducing the process leading to the Piwi/piRNA-mediated recruitment of the silencing machinery in the first paragraph of the main text, the authors mentioned that HP1 binds to Su(var)3-9. While it has been shown that HP1 and Su(var)3-9 physically interact and cooperate in the maintenance and spreading of heterochromatic domains (mostly centromeric heterochromatin) in somatic tissues (Schotta et al., 2002), there's no evidence to date that Su(var)3-9 is involved in Piwi-mediated silencing. To the contrary, Sienski et al., 2015, has recently provided substantial data indicating that SetDB1, but not Su(var)3-9, is involved in Piwi-mediated establishment of silent chromatin. In this context, I suggest the authors revise the text to make this distinction clear to readers.

Thanks for this input. We have revised the introduction accordingly.

2) Given the origin of the ova[1] allele (EMS-screen) and the rather strong and specific effect on male viability, it would be useful if the authors could include the viability data for ova[1]/ova[4] trans-heterozygous in Table S1. As it is, it is not clear whether the viability defect is due to ova or to other second-site mutations induced by the EMS-mutagenesis.

We have included the ova[1]/ova[4] viability in updated Table S1. Similar to ova[1] homozygous flies, ova[1]/ova[4] is semi-lethal for male.

3) Main text paragraph eight, in the first sentence, it should be noticed that Piwi mediates transposon silencing in both soma and germline (and not only in the germline). Most important however, while it has been shown that loss of piwi in somatic cells is associated with ectopic dpp signalling (Jin et al., 2013; Ma et al., 2014), I am under the impression that the formal demonstration that this effect is direct – and that it involves gene silencing mediated by Piwi or piRNAs – is still missing. Could the authors clarify this point?

Great point. Consistent with previous reports, we have knocked down two well-known piRNA pathway genes, piwi and panx, in escort cells and found that they have similar phenotypes with ova (Figure 2D, E). Together with the finding that HP1a::dLsd1 fusion protein could render ova dispensable for germline differentiation, we believe that the Piwi pathway adapts a similar effector machinery to repress dpp and TEs in escort cells. We agree that although many observations are consistent with the idea that dpp in escort cells is silenced by piRNA-mediated gene silencing machinery, the direct evidence, such as direct binding of Piwi or Ova to the dpp locus, is still lacking.

4) In paragraph ten, the authors state that "ova and lsd1 mutants exhibited de-repression of a similar subset of transposons" (Figure 3F). However, the analysis on Figure 3 seems to concern germline KDs, and not mutants – please clarify.

Both of them are germline-specific RNAi using UAS-Dcr2; nosGAL4 (referred as GLKD). We have modified our statements as “germline specific knock-down of ova and dlsd1 exhibited de-repression of a similar subset of transposons”.

5) In the Materials and methods section, please verify the source of the ova-RNAi line: it is listed as a BDSC stock, but it seems like it was in fact obtained from the VDRC collection. Please also include the source of the lsd1-RNAi line, as it is not currently listed in the Materials and methods section. Finally, clarify whether or not the UAS-Dcr2 transgene was used in the germline KD analysis.

We have added all the missing information in the Materials and methods section. Note that all the germline-specific RNAi were performed using UAS-Dcr2; nosGAL4 (referred as GLKD throughout the manuscript), and we have updated this information in the manuscript.

6) In subsection “RNA sequencing and computational analysis”, please indicate the manufacture and kits used to generate the small RNA- and RNA-seq libraries.

We used Oligo d(T)25 magnetic beads (NEB, #S1419S) to purify mRNA and NEBNext ultra DNA library prep kits for Illumina (NEB, #E7645S) to generate the library. For small RNA library, we used a VAHTS small RNA library prep kit for Illumina (Vazyme, NR801). We have updated this information in the Materials and methods section.

Reviewer #3:

In Yang et al., the authors generate mutants for ovaries absent (Ova) a previously identified gene in the piRNA pathway and characterize its function. They show (1) Ova is required in the somatic escort cells for germline stem cell differentiation and in the germ line for fertility, (2) Ova acts downstream of Piwi, (3) Interacts with HP1 and dLsd1 regulators of heterochromatin formation and promotes association of HP1a and dLsd1 and (4) Can induce HP1a induced demethylation. I have some major concerns about this paper: (1) The authors switch between germ line function and somatic function without explanation, (2) the phenotypes are not clearly characterized, (3) Some of the controls for the experiments they have carried out are not correct in that they compare ovaries that have late stage egg chambers to ovaries that accumulate undifferentiated cells and (4) changes in ChIPSeq are modest at the best and are not quantitated, and I am not sure are statistically significant.

We thank this reviewer for the thoughtful comments. We will give a brief response here, and detailed responses below in our point-by-point responses. We admit that we mixed descriptions of both germline and somatic functions of ova in our study, and this could potentially cause confusion. However, we believe that our phenotype analysis from both somatic and germline cells is a reasonable approach to reveal the basic mechanism of Piwi/piRNA pathway genes. Piwi, the founding member of the Piwi/piRNA pathway, has been demonstrated to have different functions in somatic cells and germline cells. The phenotypes caused by soma and germline-specific depletion of piwi are very similar to those caused by the ova depletion that we report here. Our mechanistic study reveals that the molecular function of Ova is to link HP1a and dLsd1; consider that the HP1a::dLsd1 fusion protein could rescue both TE and GSC defects. We believe that our study provides a first solid step towards a unified explanation of soma and germline phenotypes caused by the Piwi/piRNA pathway mutations: specifically, Piwi/piRNAs, through recruiting heterochromatic silencing machinery, repress TEs in the germline, and represses regular genes (especially dpp) in the somatic escort cells.

Lastly, the model is confusing. The current dogma is that demethylation of H3K4 results loss of K9ac. This then promotes methylation of K9 and which then recruits HP1. In Yang et al's model, how do they propose HP1 can get to its targets in the first place? This needs to clearly explained. I suspect, Ova should affect heterochromatin spreading not formation of heterochromatin. This is something they have not tested nor propose.

We believe what the reviewer is here indicating a general mechanism for heterochromatin establishment (Rudolph et al., 2007). We would note that other studies have shown that bivalent markers (active H3K4me3 and repressive H3K27me3) can be simultaneously present in small genomic regions (e.g., some TEs as in Bernastein et al., (2006); Voigt et al., 2012). In 2015, two studies showed that TEs are co-transcriptionally silenced, and Piwi/piRNA can recruit dLsd1 and Egg for H3K4 demethylation in the promoter region and spread H3K9 methyl downstream of TE body region (Yu et al., 2015; Sienski et al., 2015). In our study, we show that Ova is required for heterochromatic silencing. Meanwhile, ova GLKD leads to an increase in the H3K4me2 level at the promoter regions of TEs. In a lacO-GFP reporter assay, we observed that Ova recruitment alone fails to silence the GFP reporter. However, with pre-deposited HP1a to the lacO sites, ova overexpression could lead to local H3K4me2 reduction and silencing of the GFP reporter. Results from our co-IP, Y2H, and truncated/ chimeric transgenes experiments suggest that Ova acts as a linker between HP1a and dLsd1. Collectively, our data suggest that Ova acts downstream of HP1a to recruit dLsd1 for local H3K4 demethylation and heterochromatic silencing. This epigenetic connection seems to be evolutionarily conserved, as physical and functional links between H3K4 demethylation and H3K9 methylation have also been reported in yeast (Li et al., 2008).

Major concerns:

Figure 1A-D: The authors have used Ova1/4 in all the rescues including figure 1 and 2. They should include the phenotypic characterization of this allelic combination in this figure panel.

Agreed. Consequently, we have now included the ova[1/4] phenotypes in Figure 1C, D, G.

Figure 1F-G: This panel needs better pictures as even the "germless" germaria usually stain of anti-spectrin antibody. All the pictures need to be at the same magnification for easy comparison.

We have re-captured the images at the same magnification as in Figure 1C. In the right panel of Figure 1C, the whole germarium is largely germless, except a few leftover germline cells indicated by anti-spectrin and anti-Vasa staining.

Figure 1J-K: The controls should include the non-restrictive temperature and well as C587 at the restrictive temperature at 14 days. 14 days is a long time for restrictive temperature. What happens in 5 -7 days? Also, what happens to protrusions? The idea about protrusions were introduced but then not followed up on.

Thank you for these excellent suggestions. We have included the more appropriate control as well as the image of c587ts>ova RNAi at 7 days. As expected, the GSC-like cells gradually accumulate with over time. We did not observe any obvious protrusion phenotype, as GSC-l cells are still wrapped by c587>GFP positive cell processes (Figure1E).

Figure 1M: Again, please include controls for ova4

We have included the ova4/+ control in updated Figure 1G.

Figure 1N-S: Please include quantitation for all these phenotypes. As they show a link to piwi and piwi mutant phenotype is quite complicated, they should categorize Ova phenotype in detail.

We have quantified the pMad+ and Dad-lacZ+ cell number per germarium and included the results in Figure 1I.

Figure 1T: They should not use C587 as a control as this is comparing completely morphologically different ovaries. One enriches for the earlier stages and the other for later stages. They should use bam mutants as a control. Additionally, they should show the RNA seq data and tracks for these experiments for Dpp levels – these data are available to them.

Agreed; we have now removed this piece of qPCR data.

They also report that loss of ova in germ line results in eggs that do not hatch as data not shown. They should report the data.

We have included the fertility test result for ova germline mutant females in Figure 1B. The escort cell-specific expression ova in ova[1/4] mutant females (referred as ova germline mutant) has very limited fertility. The occasional escapers could be due to leaky expression of the UAS-ova transgene.

Figure 2: While the results that over expression of ova in piwi mutant ovaries is exciting they should better characterize what aspect of the piwi mutant does this protein rescue? Are the transposons levels in piwi rescued? What about protrusions? What about Dpp levels? What about Ova germ line expression in piwi mutants? Does it rescue loss of fertility in piwi mutants as well? This is important as they switch from somatic effects to germ line effects.

Ova expression in piwi mutants (both c587-GAL4 and tub-GAL4) could moderately rescue the oogenesis defects of piwi mutants. As shown in the added Figure 2C, escort cell overexpression of ova in piwi mutant ovaries could bring frequent appearance of developing germline cysts that can be developed into late stage egg chambers. However, the GSC-like tumors still exit (see arrow in Figure 2C). In addition, ubiquitous overexpression of ova fails to rescue the TE depression phenotype in piwi germline mutants. Therefore, although ova appears to be genetically downstream of piwi, there must be additional factors downstream of piwi that cooperatively function with ova to mediate heterochromatic gene silencing.

Figure 3: What happens to heterochromatin levels in ova mutants during oogenesis? Can they stain for H3K9me3 marks?

Good idea. We stained H3K9me3 in control and ova GLKD and found no obvious difference (see Author response image 3). Given H3K9me3 is required for both somatic and germline TE silencing and Ova mainly functions to repress germline TE, we speculate Ova is not required for H3K9me3 deposition. However, genome-wide H3K9me3 ChIP-seq experiments will need to be done in the future to confirmatively draw this conclusion.

Author response image 3
H3K9me3 expression in control and ova GLKD germaria.

Figure 3E: Why are the authors looking for only upregulation of transposons in the germ line when the phenotype is mostly from the soma? They should report transposon levels in somatic loss of function of ova. As the major phenotype of piwi, dlsd1 and ova are from the escort cells. They should carry out an RNAseq for this and report the transposons upregulated in the soma.

We found that somatic loss of ova only leads to mild TE upregulation, if any (see Figure 3—figure supplement 3F). Therefore, in somatic escort cells, it appears to us that the most robust change is the transcriptional depression of dpp. TEs can also be depressed, but not as robustly. We also used the whole ovary for RNA analysis because of technical feasibility; this could further dampen the depression phenotype, as the most abundant RNAs are from the germline.

Figure 4: In this figure, the authors switch back and forth between the somatic effects and germ line. This is not right. They should be direct that for expediency they use germ line as a read out. They should set up the paper that way as well. The paper reads as if one is discovering the role of Ova in escort cells whereas most of the functional data comes from the germ line. For example 4D vs 4F, in D they are using somatic drivers and in F they use a ubiquitous driver and measure germ line effects.

We admit that we combined both germline and somatic functions of ova in our study, and this could potentially cause confusion. However, we believe that phenotype analysis from both somatic and germline cells should be a reasonable approach to reveal the basic mechanism of Piwi/piRNA pathway genes. Similar to other Piwi/piRNA pathway genes, Piwi, the founding member of the Piwi/piRNA pathway, has been demonstrated to have different functions in somatic cells and germline cells. Depletion of piwi in somatic escort cells gives rise to GSC-like tumors because of dpp derepression; conditional depletion of piwi in adult germline causes robust transposon derepression phenotype without alteration of general ovary morphology.

These soma and germline-specific phenotypes are very similar to that caused by ova mutation reported here. Our mechanistic study reveals that the molecular function of Ova is to link HP1a and dLsd1, and the HP1a::dLsd1 fusion protein could rescue both TE and GSC defects. We believe that our study provides a first solid step towards a unified explanation of soma and germline phenotypes caused by the Piwi/piRNA pathway mutations: specifically, Piwi/piRNAs, through recruiting heterochromatic silencing machinery, represses TEs in the germline, and represses regular genes (especially dpp) in somatic escort cells. Increasing evidence also suggests that the Piwi/piRNA pathway can target many mRNAs in addition to TEs (Shen et al., 2018).

We used different drivers in different cases for specific purposes. In Figure 4, we used the somatic escort cell driver c587 for the purpose of looking at the ovary morphology (germline cyst differentiation) phenotype, as this phenotype is solely caused by the somatic function of ova. We used the ubiquitous driver tubGAL4 in cases such as fertility test, because both the somatic and germline functions of ova contribute to female fertility.

Figure 4F: Again, using w1118 as control is not right. One does not know if the undifferentiated stages express Het-A and TAHRE at higher levels compared to control. Again, the control and experiment are comparing two different morphologically different ovaries.

As also replied to reviewer 2, we have added more controls, including ova RNAi alone, and UAS-Dcr2; nosGAL4 flies, and the TE levels are similar to that in w1118 flies. Please note that the ova GLKD ovary has normal morphology (similar to the wild type ovary) (Figure 1A, Figure 1—figure supplement 2G), and the GSC/ CB (or refer to as GSC-l) cell number per germarium is also normal (Figure 1—figure supplement 2H). This supports the idea that Ova functions non-autonomously to restrict GSC proliferation. Germline-specific loss of ova is sufficient to cause upregulation of Het-A and TAHRE, without changing the ovary morphology.

Figure 5A: Is this germ line or somatic KD? The authors should clearly state this. It is not available in text of figure legends. I assume it is germ line KD. How does this change compare with the genes that are not affected? Is there any quantification for these changes? I am not sure these changes 0-0.75 are real?

It is germline-specific KD (referred as GLKD). We have added the information in the figures and the corresponding figure legends. We have re-analyzed the ChIP-seq data on all the transposons, and H3K4me2 density is significantly increased in Het-A and TAHRE loci but not other loci (see Figure 5—figure supplement 1); We have added the statistical analysis to Figure 5A (see Figure 5B). The Student's t-test showed that the observed increases are statistically significant.

There should be quantification for panels A, E and F.

We have included the quantifications as the new Figure 5G.

https://doi.org/10.7554/eLife.40806.022

Article and author information

Author details

  1. Fu Yang

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    For correspondence
    yangfu@nibs.ac.cn
    Competing interests
    No competing interests declared
  2. Zhenghui Quan

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Visualization, Writing—original draft
    Competing interests
    No competing interests declared
  3. Huanwei Huang

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Data curation, Software, Formal analysis, Methodology
    Competing interests
    No competing interests declared
  4. Minghui He

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Data curation, Formal analysis, Investigation
    Competing interests
    No competing interests declared
  5. Xicheng Liu

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7257-1396
  6. Tao Cai

    National Institute of Biological Sciences, Beijing, China
    Contribution
    Software, Formal analysis, Supervision
    Competing interests
    No competing interests declared
  7. Rongwen Xi

    1. National Institute of Biological Sciences, Beijing, China
    2. Tsinghua Institute of Multidisciplinary Biomedical Research, Tsinghua University, Beijing, China
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    xirongwen@nibs.ac.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5543-1236

Funding

Ministry of Science and Technology of the People's Republic of China (2017YFA0103602, 2014CB850002)

  • Rongwen Xi

National Natural Science Foundation of China (31601059)

  • Fu Yang

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

Acknowledgements

We thank the members of the fly community as cited in the Materials and Methods for generously providing fly stocks and antibodies, the Bloomington Drosophila Stock Center, Vienna Drosophila Stock Center, Tsinghua Fly Center, and Developmental Studies Hybridoma Bank (DSHB) for reagents, Dr. Bing Zhu for reading of the manuscript, and Drs. Julius Brennecke, Allan Spradling, and Yang Yu for scientific discussions. This work was supported by the National Key Research and Development Program of China (2017YFA0103602 to RX), National Basic Research Program of China (2014CB850002 to RX), and National Natural Science Foundation of China (31601059 to FY).

Senior Editor

  1. Kevin Struhl, Harvard Medical School, United States

Reviewing Editor

  1. Yukiko M Yamashita, University of Michigan, United States

Reviewers

  1. Zhao Zhang, Carnegie Institute, United States
  2. Felipe Karam Teixeira, University of Cambridge, United Kingdom

Publication history

  1. Received: August 5, 2018
  2. Accepted: December 20, 2018
  3. Version of Record published: January 16, 2019 (version 1)

Copyright

© 2019, Yang et al.

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

Metrics

  • 1,011
    Page views
  • 168
    Downloads
  • 1
    Citations

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

Download links

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

Downloads (link to download the article as PDF)

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

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

Further reading

    1. Chromosomes and Gene Expression
    2. Stem Cells and Regenerative Medicine
    Rosa María Marión et al.
    Research Article
    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Zhihao Jia et al.
    Research Article