The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition

Following fertilization of a mature oocyte, the formation of a diploid zygote involves a series of coordinated cellular events that ends with the first embryonic mitosis. In animals, this complex developmental transition is almost entirely controlled by maternal gene products. How such a crucial transcriptional program is established during oogenesis remains poorly understood. Here, we have performed an shRNA-based genetic screen in Drosophila to identify genes required to form a diploid zygote. We found that the Lid/KDM5 histone demethylase and its partner, the Sin3A-HDAC1 deacetylase complex, are necessary for sperm nuclear decompaction and karyogamy. Surprisingly, transcriptomic analyses revealed that these histone modifiers are required for the massive transcriptional activation of deadhead (dhd), which encodes a maternal thioredoxin involved in sperm chromatin remodeling. Unexpectedly, while lid knock-down tends to slightly favor the accumulation of its target, H3K4me3, on the genome, this mark was lost at the dhd locus. We propose that Lid/KDM5 and Sin3A cooperate to establish a local chromatin environment facilitating the unusually high expression of dhd, a key effector of the oocyte-to-zygote transition.


Author summary
Nuclear enzymes that add or remove epigenetic marks on histone tails potentially control gene expression by affecting chromatin structure and DNA accessibility. For instance, members of the KDM5 family of histone demethylases specifically remove methyl groups on the lysine 4 of histone H3, a mark generally correlated with gene expression. Lid (Little imaginal discs), the Drosophila KDM5, is essential for viability but is also required for female fertility. In this paper, we have found that the specific removal of Lid in developing oocytes perturbs the decompaction of the sperm nucleus at fertilization and the integration of paternal chromosomes in the zygote. Sperm nuclear decompaction normally requires the presence of a small redox protein called Deadhead (Dhd), which is massively

Introduction
In sexually reproducing animals, fertilization allows the formation of a diploid zygote through the association of two haploid gametes of highly different origins and structures. Generally, the spermatozoon delivers its compact nucleus within the egg cytoplasm, along with a pair of centrioles, while the egg provides one haploid meiotic product and all resources to sustain zygote formation [1]. In some species, this maternal control extends to early embryo development, as in Drosophila melanogaster, where the initial amplification of embryo cleavage nuclei occurs without significant zygotic transcription [2]. Instead, the bulk of transcriptional activity takes place in the fifteen interconnected large polyploid germline nurse cells that deposit gene products in the cytoplasm of the growing oocyte [3]. The developmental potential of the egg is thus initially dependent on the establishment of a highly complex transcriptional program in female germ cells.
One of the earliest events of the oocyte-to-zygote transition is the rapid transformation of the fertilizing sperm nucleus into a functional male pronucleus. In Drosophila, the needleshaped, highly compact sperm nucleus is indeed almost entirely organized with non-histone, Sperm Nuclear Basic Proteins (SNBPs) of the protamine-like type [4,5]. Male pronucleus formation begins with the genome-wide replacement of SNBPs with maternally supplied histones, a process called sperm chromatin remodeling, which is followed by extensive pronuclear decondensation [1]. Finally, zygote formation involves the coordinated migration and apposition of male and female pronuclei and the switch from meiotic to mitotic division within the same cytoplasm.
Here, we report the results of a genetic screen specifically designed to find new genes required for the oocyte-to-zygote transition in Drosophila. Our screen identified two histone modifiers, the Lid/KDM5 histone H3K4me3 demethylase and the Sin3A-HDAC1 histone deacetylase complex, which are both required for the integration of paternal chromosomes into the zygote. These interacting epigenetic factors are known to regulate the expression of hundreds of genes in somatic tissues but their role in the establishment of the ovarian transcriptome is unknown. Strikingly, RNA-Sequencing analyses revealed that, despite the modest impact of their depletion on ovarian transcripts, Lid and Sin3A are critically required for the massive expression of deadhead (dhd), a key effector of the oocyte-to-zygote transition [6,7]. Furthermore, we demonstrate that germline knock-down of these histone modifiers specifically prevent sperm chromatin remodeling through a mechanism that depends on the DHD thioredoxin.
form a male pronucleus following fertilization is generally associated with the development of haploid embryos that possess only maternally-derived chromosomes (gynohaploid embryos) and that never hatch [8]. We chose to screen transgenic lines from the TRiP collection that express small hairpin RNAs (shRNAs) under the control of the Gal4 activator [9]. We selected shRNA lines that targeted genes with known or predicted chromatin-related function and that show adult ovarian expression (Flybase). Among the 374 tested TRiP lines, 157 (41.9%) induced female sterility or severely reduced fertility when induced with the germline-specific P {GAL4::VP16-nos.UTR} MVD1 Gal4 driver (nos-Gal4), thus underlying the importance of chromatin regulation for oogenesis and early embryo development (S1 Table). We then specifically searched for shRNAs that induced a maternal effect embryonic lethal phenotype associated with gynohaploid development (Fig 1A). Gynohaploid embryos can be efficiently identified by scoring the zygotic expression of a paternally-transmitted P[g-GFP::cid] transgene at the gastrulation stage or beyond [10]. Among the sterile lines with late developing embryos (class 4 in Fig 1A and S1 Table) that were identified, four shRNA lines (GLV21071, GL00612, HMS00359 and HMS00607) produced embryos that were negative for GFP::CID ( Fig 1B). Note that none of these shRNAs induced complete female sterility and about 1 to 4% of embryos hatched and were thus diploid ( Table 1).
Two of the identified lines (GLV21071 and GL00612) express the same shRNA against the little imaginal disc (lid) gene, which encodes the single fly member of the KDM5/JARID1A family of histone demethylases [11,12]. KDM5 demethylases specifically target the trimethylation of lysine 4 of histone H3 (H3K4me3), a mark typically enriched around the Transcriptional Start Site (TSS) of transcriptionally active genes [13,14]. The two other shRNAs (HMS00359 and HMS00607) target the Sin3A and HDAC1/rpd3 genes, respectively. The conserved Sin3A protein scaffold interacts with the histone lysine deacetylase HDAC1 to form the core SIN3 histone deacetylase complex, which is generally considered as a transcriptional repressor [15]. The SIN3 complex regulates the expression of genes involved in a number of metabolic and developmental processes [16][17][18][19]. Interestingly, Lid and the largest Sin3A isoform were previously shown to physically and functionally interact [16,17,20,21], thus opening the possibility that these histone modifiers could control the same pathway leading to the formation of a diploid zygote.

Lid and SIN3 are required for sperm chromatin remodeling at fertilization
The Lid demethylase has been previously shown to be required in the female germline for embryo viability [22,23]. Both studies reported a dual phenotype for eggs produced by lid KD females (hereafter called lid KD eggs): while a majority of eggs fail to initiate development, a variable but significant fraction developed but died at later stages. Our own observations confirmed that about 15% of lid KD embryos reach or develop beyond the blastoderm stage (S1 Fig). Furthermore, our analysis of P[g-GFP::cid] expression ( Fig 1B) indicates that most of these late, non viable KD embryos are haploid and develop with maternal chromosomes. To follow the fate of paternal chromosomes in lid KD eggs, we crossed lid KD females with males expressing the sperm chromatin marker Mst35Ba::GFP (ProtA::GFP) [24]. In Drosophila, protamine-like proteins such as Mst35Ba are rapidly removed from sperm chromatin at fertilization [1] and, accordingly, ProtA::GFP is never observed in the male nucleus of control eggs. In striking contrast, the vast majority of fertilized lid KD eggs contained a needle-shaped sperm nucleus that was still positive for ProtA::GFP, indicating that sperm chromatin remodeling was compromised (Fig 1C and 1D). Anti-histone immunostaining indeed revealed that the replacement of SNBPs with maternally supplied histones was variable in KD eggs, ranging from complete absence of histones in the sperm nucleus to the coexistence of variable amounts The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition of histones and ProtA::GFP ( Fig 1D). Notably, we noticed that partially decondensed sperm nuclei in lid KD eggs were systematically positive for histones. Taken together, our observations indicate that sperm chromatin remodeling is severely impaired in lid KD eggs, thus explaining the absence of paternal chromosomes in most developing embryos.
Germline depletion of Lid was previously shown to affect karyosome morphology and chromosome positioning in metaphase I oocytes [25]. In contrast, another study had previously reported that meiotic progression was not affected in lid KD oocytes [22]. Interestingly, although our own observations indeed confirmed the aberrant karyosome structure in lid KD oocytes, we observed that the second meiotic division appeared to resume normally in a vast majority of lid KD eggs (S2 Fig). We noted, however, that, following meiosis, the female pronucleus frequently (62%, n = 76) failed to appose to the sperm nucleus in lid KD eggs ( Fig 1C). Thus, we propose that defective pronuclear migration largely accounts for the previously reported failure of lid KD embryos to initiate cleavage divisions [23]. Like Navarro-Costa The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition et al., we noted that the rosette of polar body chromosomes was frequently abnormal in morphology, but this phenotype appeared independent of meiosis per se.
Remarkably, we found that germline KD of Sin3A also induced a highly penetrant sperm nuclear phenotype with all scored fertilizing sperm nuclei retaining a needle-like shape (100%, n = 19) (S3 Fig). Finally, a similar but less penetrant phenotype was also observed in rpd3 KD eggs (S3 Fig). As this low penetrance could result from less efficient gene knock-down, we chose to mainly focus on lid and Sin3A in the rest of this study.

Transcriptomic analysis of lid KD and Sin3A KD ovaries identifies deadhead as a common and major target gene
In eggs from wild-type females, anti-Lid immunostaining failed to detect Lid protein in the male or female pronucleus, thus suggesting that its implication in sperm chromatin remodeling was indirect. In fact, Lid was not detected in embryos before the blastoderm stage (S4 Fig).
We thus turned to RNA sequencing (RNA Seq) to analyze the respective impact of lid KD and Sin3A KD on the ovarian transcriptome.
We compared transcriptomes obtained from lid KD or Sin3A KD ovaries (using the maternal triple driver MTD-Gal4) with control transcriptomes (MTD-Gal4 only). Globally, our analyses revealed a relatively modest impact of lid KD and Sin3A KD on ovarian gene expression, with more genes downregulated in both cases ( Fig 2B and S5 Fig; S2 Table). Note that these changes are expected to reflect the activity of Lid and Sin3A in germ cells, as ovarian somatic cells (see Fig 2A) do not express the targeting shRNAs. In their transcriptome analysis (based on microarrays) of lid KD wing imaginal discs, Azorin and colleagues found a similar number of differentially-expressed genes, most of them being downregulated [14]. However, RNA Seq analyses recently published by Drelon et al. in contrast found 1630 genes (FDR<0.05) dysregulated in wing discs from a null lid mutant [26]. Moreover, Liu and Secombe [27] found 8,056 genes differentially expressed (60% were down-regulated) in lid adult mutant flies (FDR<0.05), a number which could reflect the greater cell type complexity involved in this analysis.
We found that only 29% (139) of the 473 differentially-expressed genes in lid KD were also differentially-expressed in Sin3A KD, and only 100 genes (21%) were dysregulated in the same direction in both KD ( Fig 2B). As a matter of comparison, Gajan et al. found a 65% overlap in Drosophila S2 cells [17]. As the Sin3A shRNA that we used targets all predicted alternatively spliced mRNAs, this suggests that the knock-down affects Sin3A isoforms with Lid-independent functions [16].
Remarkably, however, we noticed that the deadhead (dhd) gene was by far the most severely impacted gene, downregulated by more than two orders of magnitude in both lid KD and Sin3A KD transcriptomes (Fig 2C, S6 Fig). The implication of dhd appeared particularly interesting because we and others have recently established that this germline-specific gene is critically required for sperm nuclear decompaction at fertilization [6,28]. dhd indeed encodes a specialized thioredoxin that cleaves disufide bonds on SNBPs, thus facilitating their removal from sperm chromatin [6,28]. RT-PCR and Western-blot analyses confirmed the severe down-regulation of dhd in lid KD and Sin3A KD (S5 Fig). dhd is a small, intronless gene located in the middle of a cluster of fifteen densely packed genes spanning about 40 kb of genomic DNA. Interestingly, the dhd gene lies within a 1.4 kb region that is immediately flanked by two genes with testis-specific expression (Trx-T and CG4198) (Fig 2D). Despite this apparently unfavorable genomic environment, dhd is one of the most highly-expressed genes in ovaries [29], as confirmed by our RNA Seq profiles (Fig 2C  and S6 Fig). Interestingly, although this 40 kb region contains six additional genes expressed in ovaries, dhd is the only one affected by lid KD or Sin3A KD (Fig 2D). Thus, Lid and the SIN3 complex exert a critical and surprisingly specific control on the transcriptional activation of dhd in female germ cells.

Impact of Lid depletion on the distribution of H3K4me3 in ovaries
To evaluate the impact of lid KD on the distribution of its target histone mark in the female germline, we performed chromatin immunoprecipitation and sequencing (ChIP-Seq) analyses of H3K4me3 in ovaries from control and lid KD females. Consistent with earlier reports of a global increase of H3K4me3 in lid mutant tissues [14,23,25,30], we observed that H3K4me3 ChIP peaks in lid KD ovaries were globally more pronounced compared to control ovaries ( Fig 3A). Our analysis actually revealed that about 10% (1528) of H3K4me3 identified peaks in control ovaries were significantly increased in lid KD ovaries (S3 Table; FDR<0.05). For those peaks that were associated with genes, the relative enrichment of H3K4me3 in lid KD ovaries mainly affected the promoter region and gene body (Fig 3B). A similar effect was previously observed in lid depleted wing imaginal discs, with H3K4me3 abundance specifically increased at the TSS of Lid direct target genes [14]. We nevertheless found 46 H3K4me3 peaks that were significantly decreased in lid KD ovaries compared to control ovaries (S3 Table; FDR<0.05). Among these, the H3K4me3 peak on the dhd gene was the second most severely affected (S3 Table and Fig 3C). Furthermore, only ten of the negatively affected peaks covered genes that were downregulated in lid KD ovaries, including dhd. Remarkably, the prominent H3K4me3 peak on dhd was almost completely lost in lid KD ovaries while other peaks within the dhd region remained essentially unchanged. CG4198, which lies immediately downstream of dhd is a notable exception, as this gene also shows a decrease of H3K4me3 ( Fig 3C). In this case, however, it is interesting to note that the presence of the mark is not correlated with transcriptional activity.
At first, the paradoxical loss of H3K4me3 enrichment on the dhd gene upon Lid depletion suggests that the demethylase activity of Lid is not locally responsible for this regulation. In fact, it has been established that lid mutant females with a catalytic dead JmjC � lid rescue transgene are viable and at least partially fertile [23,26,31], thus suggesting that the catalytic activity of Lid is not absolutely required to form a viable zygote. To directly test the implication of the Lid demethylase domain in dhd regulation, we measured dhd expression in JmjC � rescued females. Surprisingly, dhd transcripts were severely reduced in these females compared to control females rescued with a wild-type lid transgene (S7 Fig). We thus conclude that the demethylase activity of Lid is important for dhd expression. At least two hypotheses could reconcile this conclusion with our ChIP-Seq results. First, it is possible that the demethylase activity of Lid is only transiently required at the dhd locus, perhaps to switch on its transcription at mid-oogenesis, while later, massive dhd expression would no longer require this activity. Alternatively, Lid could exert a control on dhd transcription by restricting the level of H3K4me3 at an enhancer element, similarly to what was previously reported for KDM5 demethylases in different model organisms [32][33][34]. Although our ChIP-Seq analyses failed to identify any The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition obvious candidate enhancer element in the vicinity of the dhd gene region, we cannot exclude this possibility.
Besides its JmjC demethylase domain, Lid/KDM5 possesses a conserved C-terminal PHD motif capable of binding H3K4me2/3. This binding motif is required for the recruitment of Lid at the promoter of target genes, where it could promote their activation [27]. The local recruitment of Lid, either through its C-terminal PHD motif or through its DNA binding ARID (AT-rich interaction domain) motif, or both, could thus establish a chromatin The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition environment permissive to dhd massive expression in late oogenesis. In this context, the role of the Sin3A/HDAC1 complex also remains to be clarified. The Sin3A histone deacetylase complex is generally considered as a transcriptional repressor [15], but it also functions as a transcriptional activator in Drosophila S2 cells [17]. Lacking an intrinsic DNA binding ability [15], the recruitment of this complex to chromatin requires an additional factor. It is thus tempting to propose that Lid itself could recruit Sin3A/HDAC1 locally to activate dhd expression in female germ cells.

Forced dhd expression in lid KD ovaries partially restores sperm chromatin remodeling at fertilization
Taken together, our cytological and transcriptomic analyses strongly suggest that the loss of dhd expression in lid KD ovaries at least contributes to the observed fertilization phenotype. To directly test this possibility, we attempted to restore dhd expression in lid KD female germ cells through the use of transgenic constructs. A genomic transgene (P[dhd]) that fully rescued the fertility of dhd mutant females [6] only had a very limited impact on the hatching rate of lid KD embryos (Table 1) but quantitative RT-PCR analyses revealed that P[dhd] remained essentially silent in lid KD ovaries (Fig 4A). This result indicates that the 4.3 kb genomic region present in this transgene is sufficient to recapitulate the endogenous control exerted by Lid on dhd transcription. We then designed another transgene expressing dhd under the control of the giant nuclei (gnu) regulatory sequences. Like dhd, gnu is specifically expressed during oogenesis and is functionally required during zygote formation [35]. In addition, our RNA Seq data indicated that its expression is not controlled by Lid. We observed that the P[gnu-dhd] transgene indeed restored fertility to dhd homozygous mutant females albeit to modest level (about 7% embryo hatching rate; Table 1). In fact, rescued females only produced about 10% of the normal amount of dhd mRNA in their ovaries and the DHD protein remained almost undetectable in Westernblot ( Fig 4A). Interestingly, when introduced into lid KD females, the P[gnu-dhd] transgene also slightly increased embryo hatching rate (Table 1). Furthermore, cytological examination of eggs laid by these females revealed a limited but clear improvement of sperm nuclear decondensation (Fig 4B and 4C). These results thus indicate that forced expression of dhd can improve the survival of lid KD eggs through its positive impact on sperm chromatin remodeling. Finally, we tried to further increase the level of expression of dhd by using a Gal4 inducible transgene, P [UAS-dhd WT ]. Indeed, induction of this transgene in the germline of dhd J5 mutant females fully restored their fertility (Table 1). We also observed a strong effect on the hatching rate of embryos laid by lid KD, P[UAS-dhd WT ] females (about 28%; Table 1). However, a P[UASdhd sxxs ] transgene expressing a catalytic mutant DHD with no rescuing potential also improved the fertility of lid KD females. This effect suggests that Gal4 becomes limiting in the presence of two UAS transgenes, with a negative impact on knock-down efficiency. The fertility of P[UASdhd WT ] rescued females was nevertheless doubled compared to P[UAS-dhd sxxs ] control females (Table 1), thus supporting the idea that partial dhd re-expression in lid KD ovaries significantly improved the probability of these eggs to form a viable, diploid zygote. Besides its already established roles in controlling the oocyte epigenome and the architecture of meiotic chromosomes, we show that the transcriptional regulation of dhd (and possibly additional early acting genes) is indeed a critical function of Lid and associated factors in female germ cells.

Trr controls sperm chromatin remodeling through a DHD-independent pathway
Intriguingly, germline KD of the Trithorax group protein Trithorax-related (Trr, also known as dMLL3/4), a histone methyltransferase responsible for monomethylation of H3K4 [36], was recently shown to induce a sperm decondensation defect at fertilization similar to the one reported here for lid, Sin3a and rpd3 KD [37]. In their study, however, Prudêncio et al. did not find any significant change in dhd mRNA level in trr KD early embryos. To more directly exclude any implication of DHD in the trr KD phenotypes, we stained control and KD eggs with an anti-DHD antibody. At fertilization, maternally-expressed DHD is abundant throughout the egg cytoplasm (100%, n = 41) but is rapidly degraded after pronuclear apposition. As expected, DHD protein remained undetectable in most lid KD eggs (92%, n = 50), including those that were fixed before the end of meiosis II (S8 Fig). In sharp contrast, DHD protein was normally detected in a majority of trr KD eggs, even though these eggs indeed contained a needle-shaped sperm nucleus still packaged with SNBPs ( S8 Fig). This result thus confirms that Trr/dMLL3/4 controls sperm nuclear remodeling through a yet unknown, DHD-independent mechanism. Conversely, Trr was shown to control meiosis progression through the activation of Idgf4 [37], a gene that is not affected by lid or Sin3a KD (this study). Thus Trr and Lid/ Sin3A respectively activate a distinct repertoire of genes important for the oocyte-to-zygote transition and sperm chromatin remodeling.

Conclusion
Our maternal germline genetic screen has unveiled a complex and remarkably specific transcriptional regulation of the dhd gene by Lid/KDM5 and the Sin3A/HDAC1 complex. In addition to its crucial role in sperm protamine removal at fertilization, DHD was recently involved in the establishment of a redox state balance at the oocyte-to-zygote transition with a number of identified target proteins [7]. This important DHD-dependent thiol proteome remodeling is thus ultimately controlled by Lid and the SIN3 complex, underlying the critical contribution of these transcriptional regulators to this delicate developmental transition. Future work will aim at dissecting the chromatin mechanisms at play in setting up dhd specific activation in female germ cells.

Drosophila strains
Flies were raised at 25˚C on standard medium. The w 1118 strain was used as a wild-type control. shRNAs lines used in this study (see S1 Table) were established by the Transgenic RNAi Project (TRiP) at Harvard Medical School and were obtained from the Bloomington Drosophila Stock Center at Indiana University. The lid and Sin3A shRNA lines target all predicted isoforms of their respective target genes. Additional stocks were EGFP-Cid [38], P{otu-GAL4:: VP16.R}1; P{GAL4-nos.NGT}40; P{GAL4::VP16-nos.UTR}MVD1 ("MTD-Gal4"), P{GAL4:: VP16-nos.UTR}MVD1 ("nos-Gal4"), P[Mst35Ba-EGFP] [24] and Df(1)J5/FM7c [39].   rp49 and relative to expression in w 1118 ). Data are presented as mean ± SD of 2 biological replicates. P values indicate one-way ANOVA with Dunnett's multiple comparisons test to a control ( ���� P < 0.0001; �� P < 0.01, � P < 0.03; n.s = not significant). Right: Western blot analysis of DHD expression in ovaries of indicated genotypes. Alpha-tubulin detection is used as a loading control in Western-blotting. B-Quantification of sperm nuclear phenotype in eggs laid by females of indicated genotypes. C-Confocal images of a MTD>lid22, P[gnu-dhd] haploid embryo during the third nuclear division with incompletely remodeled male nucleus. Karyogamy has failed and the embryo contains four haploid nuclei of presumably maternal origin. The paternal nucleus (inset) still contains a region packaged with ProtA::GFP (arrow). Note that DNA positive dots at the spindle poles are Wolbachia endosymbionts. Scale bar: 10 μm. https://doi.org/10.1371/journal.pgen.1008543.g004 The Lid/KDM5 histone demethylase complex activates a critical effector of the oocyte-to-zygote transition

Germline knock-down and fertility tests
To obtain KD females, virgin shRNA transgenic females were mass crossed with transgenic Gal4 males at 25˚C and females of the desired genotype were recovered in the F1 progeny. To measure fertility, virgin females of different genotypes were aged for 2 days at 25˚C in the presence of males and were then allowed to lay eggs on standard medium for 24 hours. Embryos were counted and then let to develop for at least 36 hours at 25˚C. Unhatched embryos were counted to determine hatching rates.

Immunofluorescence and imaging
Early (0-30 min) and late (about 6 hours) embryos laid by randomly selected females were collected on agar plates. Embryos were dechorionated in bleach, fixed in a 1:1 heptane:methanol mixture and stored at -20˚C. Embryos were washed three times (10 min each) with PBS1X 0.1%, Triton X-100 and were then incubated with primary antibodies in the same buffer on a wheel overnight at 4˚C. They were then washed three times (20 min each) with PBS 0.1%, Triton X-100. Incubations with secondary antibodies were performed identically. Embryos were mounted in Dako mounting medium containing DAPI.
Ovaries were dissected in PBS-Triton 0.1% and fixed at room temperature in 4% formaldehyde in PBS for 25 minutes. Immunofluorescence was performed as for embryos except for secondary antibodies that were incubated four hours at room temperature. Ovaries were then mounted as described above.

Gene expression analysis by RT-QPCR
Total RNA was extracted from ovaries of 3-day-old females using the NucleoSpin1 RNA isolation kit (Macherey-Nagel), following the instructions of the manufacturer. Duplicates were processed for each genotype. cDNAs were generated from 1μg of purified RNA with oligo (dT) primers using the SuperScript TM II Reverse Trancriptase kit (Invitrogen). Generated cDNAs were diluted to 1/5 and were used as template in a real time quantitative PCR assay using SYBR1Premix Ex TaqTM II (Tli RNaseH Plus) (Takara). All qRT-PCR reactions were performed in duplicate using Bio-Rad CFX-96 Connect system with the following conditions: 95˚C for 1 min followed by 40 cycles of denaturation at 95˚C for 10 s, annealing at 59˚C for 30 s and extension at 72˚C for 30 s. Relative fold change in gene expression was determined by the comparative quantification ΔΔCT method of analysis [40]. The housekeeping gene rp49 was used to normalize cDNA amounts in the comparative analysis. The primer sets used in the PCR reactions were: dhd-forward 5'-TCTATGCGACATGGTGTGGT -3' and dhd-reverse 5'-TCCACATCGATCTTGAGCAC -3'; lid-forward 5'-ATTGGTTTCACGAGGATTGC-3' and lid-reverse 5'-CATAGCCACTTGGGTCGATT -3'; Rp49-forward 5'-AAGATCGTGAAGAA GCGCAC -3' and Rp49-reverse 5'-GATACTGTCCCTTGAAGCGG -3'. Statistical tests were performed using GraphPad Prism version 6.00 for Mac OS X (GraphPad Software).

Ovarian RNA sequencing and analysis
For each samples, 8 pairs of ovaries were dissected from 6 day-old virgin females and total RNA were extracted using the NucleoSpin1 RNA isolation kit (Macherey-Nagel), following the instructions of the manufacturer. Extracted RNAs were treated with Turbo TM DNAse (Ambion #AM2238). After DNase inactivation, RNAs were purified using the NucleoSpin1 RNA Clean-up XS kit (Macherey-Nagel) according to manufacturer's instructions. Sequencing was completed on two biological replicates of each genotype: Control Sequencing libraries for each sample were synthesized using TruSeq Stranded mRNA kit (Illumina) following supplier recommendations (Sample Preparation Guide-PN 15031047, version Rev.E Oct 2013) and were sequenced on Illumina Hiseq 4000 sequencer as Single-Reads 50 base reads following Illumina's instructions (GenomEast platform, IGBM, Strasbourg, France). Image analysis and base calling were performed using RTA 2.7.3 and bcl2fastq 2.17.1.14. Adapter dimer reads were removed using DimerRemover. Sequenced reads were mapped to the Drosophila melanogaster genome assembly dm6 using TopHat (version 2.1.1) with default option. The aligned reads were assigned to genes by FeatureCounts, run with default options on the dmel-all-r6.15 version of the Drosophila melanogaster genome annotation. Differentially expressed genes were identified using the R-package DESeq2 (version 1.14.1). The annotated genes exhibiting an adjusted-P < 0.001 were considered differentially expressed compared to Control. each biological replicate, eighty ovary pairs were dissected from 2 day-old females and flash frozen. Dissected ovaries were fixed in 1.8% formaldehyde at room temperature for 10 minutes. Chromatin was sonicated using a Diagenod Bioruptor (18 cycles, high intensity, 30s on/ 30s off) to generate random DNA fragments from 100 to 800 base pairs. Sheared chromatin was incubated overnight at 4˚C with H3K4me3 antibody (ab8580 Abcam). Immunoprecipitated samples were treated with RNAse A, proteinase K and DNA purified using the ChIP DNA Purification kit (Active Motif #58002) following the manufacturer's instructions. Quantification assessment of purified DNA was done using Qbit dsDNA HS Assay on the Qbit fluorometer (Invitrogen). Immunoprecipited DNA quality was evaluated on a Bioanalyzer 2100 (Agilent).
Sequencing libraries for each sample were synthesized using Diagenode MicroPlex Library Preparation kit according to supplier recommendations (version 2.02.15) and were sequenced on Illumina Hiseq 4000 sequencer as Paired-End 50 base reads following Illumina's instructions (GenomEast platform, IGBM, Strasbourg, France). Image analysis and base calling were performed using RTA 2.7.3 and bcl2fastq 2.17.1.14. Adapter dimer reads were removed using DimerRemover. Sequenced reads were mapped to the Drosophila melanogaster genome assembly dm6 using Bowtie (version 2.3.3) with default option. Only uniquely aligned reads have been retained for further analyses. Duplicated reads were removed using picard-tools (version 2.17.10). Peak calling was performed for each individual samples and on merged biological replicates using MACS algorithm (version 2.1.1) with default option and a relaxed qvalue cut-off of 0.1. Consistent peaks between biological replicates were identified using irreproducible discovery rate (IDR version 2.0.3) with a 0.05 cut-off. Differentially modified H3K4me3 peaks between Control and lid Knock-down ovaries were identified using the Rpackage DiffBind (version 2.2.12) with a 0.05 FDR cut-off.

Data visualization
The Deeptools software was used to convert alignment files to bigwig (bamCoverage) and to generate H3K4me3 heatmap and density profiles (computeMatrix and plotHeatmap). The generated bigwig files were visualized using IGV software.