Establishment and validation of strains carrying GFP-tagged satellite DNA-binding proteins

We performed affinity purification coupled to mass spectrometry (AP-MS) to identify chromocenter-associated proteins in Drosophila embryos, ovaries, and GSC-enriched testes using D1 and Prod as baits (Fig 1A). In nuclei from these tissues, D1 and Prod foci were often found in proximity to each other (Fig 1B), and we expected to identify common interactors between the 2 baits. To minimize variability between baits, we used an affinity purification strategy involving pulldown of GFP-tagged baits using GFP nanobody-coupled beads. For D1, we made use of an available strain where GFP has been inserted upstream of the endogenous locus. To validate this strain, we assessed the number of chromocenters (D1 and Prod foci per cell) in male and female GSCs as well as fertility in both sexes in comparison to a wild-type reference strain (D. melanogaster yw). We observed that the number of D1 and Prod foci per cell in male and female GSCs was identical to the wild type (Fig 1C and 1D), indicating that chromocenter formation was unperturbed in the GFP-D1 strain in these cell types. In addition, we observed that homozygous GFP-D1 males and females were fertile, although the fertility of homozygous GFP-D1 females is reduced compared to the wild type (Fig 1E and 1F). For Prod, we first generated a transgenic strain where a GFP-tagged codon-optimized Prod was placed downstream of a 400 bp promoter sequence and inserted into the attP40 locus (S1A Fig). The GFP-prod transgene fully rescued the larval lethality associated with loss-of-function prod mutations, prod^k01180 and prod^U [33] (S1B Fig). Hereafter, GFP-Prod refers to the GFP-prod transgene in a prod mutant background (p400-GFP-prod, prod^k01180/prod^U) (S1B Fig). We observed that the number of GFP-Prod foci was at wild-type levels in male and female GSCs (Fig 1C and 1D), suggesting that the transgene does not interfere with chromocenter formation in these cells. Like the GFP-D1 strain, GFP-Prod males were fully fertile, while GFP-Prod females were sub-fertile in comparison to the wild type (Fig 1E and 1F). However, ovaries from both the GFP-D1 and GFP-Prod strains were morphologically similar to the control (S1C Fig), suggesting that early stages of oogenesis are likely unaffected in these strains. As a negative control for GFP-D1 and GFP-Prod, we used a strain ubiquitously expressing nucleoplasmic GFP (NLS-GFP). The NLS-GFP did not enrich at chromocenters, and we observed that the number of chromocenters was comparable to the wild type, GFP-D1 and GFP-Prod strains (Fig 1C and 1D). Therefore, the GFP-tagged D1 and Prod strains and the NLS-GFP control strain were used to characterize the chromocenter-associated proteome in intact Drosophila tissues.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Experimental setup for the characterization of the chromocenter-associated proteome.

(A) Experimental scheme for the identification of D1- and Prod-associated proteins across multiple Drosophila tissues using quantitative mass spectrometry. This figure panel was created using Biorender. (B) D1 and Prod are in proximity within the nucleus in multiple cell types including gastrulating embryos, female germ cells, and male germ cells. (C) Bar plot of the total number of D1 foci present in female GSCs and male GSCs in the wild type (yw), NLS-GFP, GFP-D1, and GFP-Prod strains; n indicates the number of cells analyzed. (D) Bar plot of the total number of Prod foci present in female GSCs and male GSCs in the wild type (yw), NLS-GFP, GFP-D1, and GFP-Prod strains; n indicates the number of cells analyzed. (E, F) Bar plot of the number of progeny sired by a single tester male (E) or female (F) when mated with 2 wild-type flies of the opposite sex over the course of 1 week; **** represents p < 0.0001 from an ordinary one-way ANOVA. All error bars are SD and all scale bars are 5 μm. Source data for (C–F) can be found in S1 Data. GSC, germline stem cell.

https://doi.org/10.1371/journal.pbio.3002984.g001

Multi-tissue characterization of the chromocenter-associated proteome

First, we optimized lysis and affinity purification conditions such that our bait proteins were effectively solubilized (S1D Fig) and immunoprecipitated (S1E Fig). Following this, we affinity purified GFP-D1, GFP-Prod (and NLS-GFP as a control) from the embryo, ovary, and testis lysates and identified the associated proteins by quantitative mass spectrometry. While we were unable to analyze Prod pulldowns specifically in GSC-enriched testes due to poor sample quality, we identified 196 proteins that were significantly enriched in the D1 and Prod pulldowns in comparison to the control pulldown of NLS-GFP across all tissues (log₂FC>1 and p_adj<0.1, Figs 2A–2C, S2A, S1–S3 Tables), with these hits likely representing both direct and indirect interactions. In embryos, D1 co-purified known chromocenter-associated or pericentromeric heterochromatin-associated proteins, such as Stwl [34–36], ADD1 [28,29], mei-S332 [37], Mes-4 [28], Hmr [29,38,39], Lhr [29,38], and members of the chromosome passenger complex such as Borealin (Borr) and Aurora B [40] (Fig 2A and S1 Table). Prod pulldown in embryos co-purified chromocenter-associated proteins such as Hcs [41] and Saf-B [42] (S1 Table). Our proteomic data also identified some of the previously reported D1/Prod associations such as D1-Xpc (embryo), D1-Stwl (embryo), D1-Smn (ovary), D1-CG5599 (embryo), D1-CG42232 (embryo), Prod-CG15107 (ovary), and Prod-bocks (embryo) [43,44].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The chromocenter-associated proteome across Drosophila tissues.

(A, B) Volcano plots of the D1-associated proteome in embryos (A) and the Prod-associated proteome in ovaries (B). The dashed lines mark log₂FC>1 and p_adj<0.1; blue points indicate proteins involved in transposon repression, red points indicate proteins involved in DNA repair, and yellow points indicate proteins pulled down by D1 and Prod. For proteins belonging to more than one category, we prioritized the representation as follows, transposon repression > DNA repair > D1/Prod common interactors. Number of biological replicates are indicated in the plot. (C) Intersection plot representing number of overlapping proteins in the D1 and Prod pulldowns from different tissues based on a log₂FC>1 and uncorrected p < 0.05 cutoff. Blue arrowheads indicate number of proteins common to D1 and Prod pulldown in the same tissue while magenta arrowheads indicate number of common proteins pulled down by D1 or Prod across different tissues. (D) Enrichment of the selected proteins pulled down by D1 and Prod from more than one tissue. The size and color of the circles are proportional to the log₂FC while the black outline indicates p_adj<0.1. (E) Hypergeometric test of DNA repair or transposon repression pathway proteins from the multi-tissue D1 and Prod interactome based on a log₂FC>1 and uncorrected p < 0.05 cutoff. Dashed line indicates p = 0.05. Source Data for (A, B, D) can be found in S1 and S2 Tables. Source Data for (C) can be found in S2 Data.

https://doi.org/10.1371/journal.pbio.3002984.g002

We used a less stringent cutoff (log₂FC>1, p < 0.05) to analyze our proteomic data across the different baits and tissues, based on the rationale that identical false positive across independent experiments were extremely unlikely. Through this approach, we found that Prod co-purified components of the replication protein A (RPA complex) (S2B Fig), while D1 co-purified multiple proteins involved in DNA replication (S2C Fig) and chromosome condensation (S2D Fig). Interestingly, we observed a substantial overlap between D1- and Prod-associated proteins (yellow points in Fig 2A and 2B and S1–S3 Tables) from the same tissue, with 58 hits pulled down by both baits (blue arrowheads, Fig 2C) in embryos and ovaries. This observation is consistent with the fact that both D1 and Prod occupy sub-domains within the larger constitutive heterochromatin domain in nuclei [27]. Surprisingly, only 9 proteins were co-purified by the same bait (D1 or Prod) across different tissues (magenta arrowheads; Fig 2C and S1–S3 Tables). In addition, only a few proteins such as an uncharacterized DnaJ-like chaperone, CG5504, were associated with both D1 and Prod in embryos and ovaries (Fig 2D). One interpretation of these results is that the protein composition of chromocenters may be tailored to cell- and tissue-specific functions in Drosophila. However, more targeted experiments are required to validate whether certain proteins are indeed tissue-specific interactors of D1 and Prod.

Finally, we used AlphaFold Multimer (AFM) to predict interaction interfaces for a subset D1 and Prod interactors (log₂FC>2, p < 0.01) from all tissues and potentially distinguish direct versus indirect interactions [45]. AFM-based structural modeling has significantly advanced identification of protein interaction interfaces and has been used across a wide variety of organisms and protein types [46]. Moreover, the AFM-derived interface predicted template modelling (ipTM) score has been used as a standard metric to evaluate the accuracy of the structural models. Based on previous studies [45,47,48], ipTM scores above 0.5 represent a medium confidence model of the interaction interface while ipTM scores above 0.8 represent high confidence models. Despite both D1 and Prod containing large intrinsically disordered regions (IDRs), which hinder the accuracy of structural predictions, we found that 6/78 D1 interactors and 16/79 Prod interactors yielded an ipTM score above 0.5 (S4 Table). Taken together, we have identified several proteins associated with the satellite DNA-binding proteins, D1 and Prod across Drosophila tissues and AFM-based structural modelling suggests that some of these hits could be direct interactors of D1 and Prod.

DNA repair and TE repression proteins are co-purified by D1 and Prod in embryos

Manual inspection of our data revealed that D1 and Prod in embryos pulled down multiple proteins involved in DNA repair, such as POLDIP2 (D1 and Prod), Top2, Xpc, and Elg1 (all D1 only) (Fig 2A and S5 Table). Tandem repeats like satellite DNA are inherently vulnerable to copy number changes [49] and are known to employ specialized mechanisms during DNA repair [50–54]. Therefore, these interactions may contribute to repeat homeostasis, especially during the rapid early embryonic divisions. In addition, we observed that the canonical TE repression protein, Maelstrom (Mael) was co-purified by D1 and Prod in embryos (Fig 2A and S1 Table). In addition, heterochromatin-associated TE repression proteins such as Sov, Hmr, and Lhr [38,55–57] were also co-purified by D1. Consistently, a hypergeometric test for pathway enrichment revealed that DNA repair and TE repression proteins (log₂FC>1, p < 0.05) were enriched in the D1 and Prod pulldowns from embryos and to a lesser extent in other tissues (Fig 2E).

TE repression proteins are associated with chromocenters

TEs are typically viewed as agents of genomic instability and their unchecked mobilization is usually associated with DNA damage, mutation, cell death, and sterility [58]. In germline tissues from flies, as well as other metazoans, genome defense against transposons requires a dedicated small RNA pathway known as the Piwi-interacting RNA (piRNA) pathway [59,60]. In Drosophila, transcription of piRNA precursors is mediated by a protein complex comprised of Rhino, Cutoff, and Deadlock (RDC complex) [61–63], followed by Nxf3-dependent nuclear export of precursor transcripts and processing into mature piRNAs [59,60]. Mature piRNAs are loaded onto a family of RNA-binding Argonaute proteins including Piwi (P-element induced wimpy testis), a canonical member of the piRNA pathway [59,60]. piRNA-bound Piwi enters the nucleus, locates genomic TE loci through sequence complementarity and mediates TE repression through local heterochromatin formation and chromatin compaction. Heterochromatin-associated proteins such as hybrid male rescue (Hmr) [38], lethal hybrid rescue (Lhr) [38], and small ovary (Sov) [55–57] are implicated in heterochromatin formation and transcriptional repression at transposon loci [57]. Of these, Hmr and Sov were co-purified by D1 in embryos in this study. In addition, recent studies in cultured cells of embryonic origin have also identified that both Piwi and Hmr interact with Prod [64]. These data support the notion that satellite DNA-binding proteins may contribute to piRNA pathway function, potentially through regulating heterochromatin stability at TE loci.

In order to validate our proteomics data, we used strains where either Piwi, which is a canonical piRNA pathway Argonaute protein, or Sov, which is implicated in piRNA-dependent heterochromatin formation at TEs [55–57], were tagged with GFP. Although Piwi mediates transcriptional repression at TE loci, it is localized throughout the entire nucleus and the GFP-Piwi signal expectedly overlapped with the heterochromatic D1 and Prod foci in embryonic nuclei (Fig 3A). In contrast, Sov, which is recruited to transposon loci by the piRNA pathway protein Panoramix [57], exhibited a punctate GFP signal in embryonic nuclei, which was frequently present at the boundaries of D1/Prod-containing chromocenters (Fig 3B). We observed a similar pattern for GFP-Piwi and Sov-GFP localization in adult germ cells from males and females (S3A–S3D Fig). Moreover, the D1-interacting heterochromatin protein, Hmr, which contributes to TE repression [38], was also observed at the chromocenter boundary (S3E and S3F Fig), in agreement with recent reports suggesting that Hmr may bridge centromeric and pericentromeric heterochromatin [29,32].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Molecular relationships between proteins involved in transposon repression and chromocenter formation.

(A, B) GFP-Piwi (A) and Sov-GFP (B) embryos were stained with D1/Prod (magenta) and DAPI (blue). Arrowheads indicate Sov-GFP foci in proximity to chromocenter boundaries. (C) Enrichment of the selected proteins pulled down by D1, Prod, and Piwi from embryos. The size and color of the circles are proportional to the log₂FC while the black outline indicates p_adj<0.1. (D) Spectral intensities for the indicated proteins from the GFP control (green) and Piwi (magenta) pulldowns from 3 independent replicates. Adjusted p-values (Benjamini–Hochberg) are indicated for each comparison. (E) ipTM scores of the pairwise protein structural models based on AFM. Gray points indicate low confidence structural models, blue points indicate medium confidence structural models (ipTM >0.5), and red points indicate high confidence structural models (ipTM >0.8). All scale bars are 5 μm. Source Data for (C, E) can be found in S7 and S8 Tables. AFM, AlphaFold Multimer; ipTM, interface predicted template modelling; Sov, small ovary.

https://doi.org/10.1371/journal.pbio.3002984.g003

To further test the potential associations between D1/Prod and Piwi, we performed a reciprocal AP-MS experiment using embryonic lysates from the GFP-Piwi strain (and the NLS-GFP strain as a control) (S7 Table). Piwi co-purified proteins identified as D1/Prod interactors in embryonic lysates (Fig 3C and 3D). Strikingly, 4 proteins (CG14715, Fit, CG10208, and Ugt35D1) were high confidence hits from the Piwi, D1, and Prod pulldowns (Fig 3C and 3D). All these proteins are relatively unstudied. CG14715 is an uncharacterized FK506-binding protein, while Ugt35D1 is a putative UDP-glycosyltransferase. Female independent of transformer (Fit) is a sexually dimorphic small peptide that has been implicated in feeding control [65], while CG10208 is an uncharacterized protein with no known domains or function. To determine how these proteins may potentially interact with each other and with D1, Prod, and Piwi, we used AFM to model interaction interfaces between all possible pairs (S8 Table). Interestingly, 8/28 of the protein pairs had ipTM scores above 0.5 (blue and red points, Fig 3E), with these medium and high confidence scores highlighting interfaces that may give rise to direct interactions. Among these, both D1 and Prod had medium confidence interfaces with Piwi, which were mediated by contacts between disordered regions on D1 and Prod that threaded through a central cleft in the Piwi protein (S4A Fig). At the same time, we were unable to detect a Piwi-D1 interaction by co-immunoprecipitation from embryonic lysates (S4B Fig), suggesting that this interaction may only occur in a small subset of cells (e.g., germ cells) in embryos or that the interaction may be mediated by weak/transient associations between disordered regions, with both possibilities necessitating higher sensitivity approaches like mass spectrometry. Overall, the data from the proteomic, microscopy, and structural modeling experiments strongly hint at a functional relationship between satellite DNA organization and the piRNA pathway.

Satellite DNA clustering into chromocenters does not rely on piRNA pathway proteins

Repressive histone posttranslational modifications, such as H3K9me2/3, are enriched at pericentromeric heterochromatin [13,25] and recent studies identifying piRNAs complementary to complex satellite DNA repeats [66] suggest that the piRNA pathway may promote repressive PTMs at satellite DNA repeats and facilitate chromocenter formation. This idea was supported by prior studies showing that an interaction between Piwi and the heterochromatin protein, HP1, contributes to heterochromatin organization in both somatic and germ cells [67–69]. Since Piwi mutation results in a severe agametic phenotype in ovaries and testes, we instead used mutations of aubergine, which mediates piRNA amplification through the ping-pong mechanism [60]. In the absence of aubergine, total piRNA levels are decreased and piRNA-free Piwi fails to enter the nucleus [70]. We observed that aubergine mutation (aub^HN2/aub^QC42) as well as germline knockdown of aubergine resulted in a mild de-clustering of Prod foci in female GSCs but had no effect on the number of D1 foci in the same cells (S5A–S5C Fig). These findings suggest that the piRNA pathway does not have a prominent role in chromocenter stability in female GSCs.

Satellite DNA-binding proteins make a subtle contribution to TE repression in adult ovaries

For an initial assessment of whether D1 and Prod contribute to TE repression, we used established reporters of transposon activity in the adult germline (GFP-Burdock-lacZ) and adult soma (gypsy-lacZ) of Drosophila ovaries [71,72]. In the gonadal soma, we used the tj-gal4 driver to express RNAi against D1 in a genetic background containing the gypsy-lacZ reporter. Despite effective depletion of D1 (S6A Fig), we did not observe increased expression of the TE reporter (gypsy-lacZ) based on staining for beta-galactosidase (β-Gal) in comparison to the negative control (mCherry RNAi) (S7A Fig). Similar to D1 depletion, Prod depletion did not result in de-repression of gypsy-lacZ in the somatic follicle cells (S6B and S7A Figs). In contrast, RNAi-mediated depletion of Sov, which is required for TE repression in gonadal somatic cells[55,56], resulted in β-Gal signal in the follicle cells across multiple stages of oogenesis, indicating expression of the somatic TE reporter (S7A Fig). Next, we tested whether loss of D1 in the germline resulted in increased expression of the germline TE reporter, GFP-Burdock-lacZ. To do so, we used a previously validated mutant allele (D1^LL03310) in trans to a chromosomal deficiency spanning the D1 locus (Df(3R)^Bsc666) in a background containing GFP-Burdock-lacZ [26]. Here, we did not observe GFP signal in the germline cells of both D1 mutant ovaries and the heterozygous controls (S7B Fig). Similarly, RNAi-mediated Prod depletion (S6C Fig) in a background containing GFP-Burdock-lacZ also did not lead to observable GFP signal in the ovaries (S7B Fig). In contrast, depletion of aubergine, which is required for TE repression in germ cells, resulted in a strong germline GFP signal (S7B Fig). These data suggest that D1 and Prod do not contribute to the repression of gypsy and Burdock during oogenesis. In addition, previous studies have determined that Prod depletion in adult ovaries does not alter the expression of other TEs such as HeT-A, TAHRE, and blood [71,72]. However, we cannot exclude from these data that D1 and Prod contribute to the repression of other TEs.

Therefore, we performed RNA sequencing to determine transcriptome-wide effects on TE expression. Since Chk2 arrests germ cell development in response to TE de-repression and DNA damage [73,74], we used heterozygous and D1 mutant ovaries in a chk2 mutant (chk2⁶⁰⁰⁶) background to prevent developmental arrest of potential TE-expressing germ cells. We observed that both genotypes exhibited similar gonad morphology (Fig 4A). Consistent with our data from the TE reporters and previous studies, expression of most TE families was not elevated in D1 mutant ovaries in comparison to the heterozygous control (Fig 4B and S9 Table). One exception was the LTR retrotransposon tirant, which has recently invaded D. melanogaster populations [75], and was expressed 5- to 6-fold higher in D1 mutant ovaries. However, the magnitude of TE expression in the D1 mutant ovaries is not in the range of what has been observed for ovaries that lack canonical piRNA pathway proteins [63,76]. The mild increase in tirant expression is therefore unlikely due to compromised piRNA pathway function.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The effect of D1 on piRNA biogenesis and TE expression in ovaries.

(A) Brightfield images of ovaries from D1 heterozygous and D1 mutant females in a chk2⁶⁰⁰⁶ background. Scale bars: 100 μm. (B) Volcano plot of differentially expressed transposon families in ovaries from D1 heterozygous and D1 mutant females in a chk2⁶⁰⁰⁶ background (n = 3). Statistically significant changes (log₂FC >1, p_adj <0.05) are indicated with the black outlines. (C, D) Size profiles of Argonaute-bound small RNAs and miRNAs isolated from D1 heterozygous and D1 mutant ovaries. Reads from 3 replicates per genotype were normalized per million miRNA reads. (E) Average FPKM for piRNAs mapping antisense to transposon consensus sequences from D1 heterozygous and D1 mutant ovaries (n = 3). Black outline indicates statistically significant changes in antisense piRNAs (log₂FC >1, p_adj <0.05). (F) Volcano plot of differentially expressed genes in ovaries from D1 heterozygous and D1 mutant females in a chk2⁶⁰⁰⁶ background (n = 3). Statistically significant changes (log₂FC >1, p_adj <0.05) are indicated with the black outlines. Blue points indicate the 347 genes associated with transposon repression (S6 Table). Source data for (B, F) can be found in S9 Table. Source data for (C–E) can be found in S3 Data. piRNA, Piwi-interacting RNA; TE, transposable element.

https://doi.org/10.1371/journal.pbio.3002984.g004

Levels of TE-targeting piRNAs are largely unperturbed in the absence of D1

Since transposon silencing by the piRNA pathway relies on the production of piRNAs and their loading onto the Piwi, Aub, and Ago3 Argonaute proteins [60], we performed small RNA sequencing to determine whether the Argonaute-bound piRNA pool was altered in D1 mutant ovaries in comparison to heterozygous control ovaries. Our data show that total levels of mature piRNAs, normalized to miRNA reads, are not substantially affected by the loss of D1 (Fig 4C and 4D). We also did not observe a marked reduction in antisense piRNAs complementary to TEs in D1 mutant ovaries, including those mapping to tirant (Fig 4E). Our data indicate that the biogenesis of TE-mapping piRNAs is unaffected in the absence of D1. Furthermore, our RNA-seq data also demonstrated that none of the 347 genes that are known to contribute to TE silencing (S6 Table) were differentially expressed in D1 mutant ovaries (blue points; Fig 4F and S9 Table). Altogether, our data suggest that the subtle changes in the expression of a few TEs in D1 mutant ovaries are unlikely due to compromised piRNA biogenesis or altered expression of piRNA pathway components. Rather, we propose that impaired heterochromatin stability or other piRNA-independent factors [77,78] in the D1 mutant could be the underlying cause.

The piRNA biogenesis machinery is mislocalized to satellite DNA in the absence of D1

In contrast to TE transcripts, we observed that transcripts corresponding to the abundant 1.688g/cm³ satellite DNA repeat (denoted as SAR DM and SAR2 DM; Fig 4B) were up-regulated in D1 mutant ovaries in comparison to the heterozygous control. Conversely, transcripts mapping to the Responder (Rsp) satellite DNA repeat were down-regulated in the absence of D1 (Fig 4B). These changes in satellite DNA transcription were correlated with changes in piRNAs originating from these repetitive loci (Fig 4E). We further mapped the small RNA reads to the different monomers that make up the 1.688 g/cm³ satellite DNA and found a specific up-regulation of the 260 bp and 359 bp variants (Fig 4E).

Recent studies have demonstrated that the Rhino-Deadlock-Cutoff complex stimulates transcription of complex satellite DNA repeats, which are then processed into mature piRNAs [66,79]. Interestingly, the absence of a ZAD-zinc finger DNA-binding protein known as Kipferl causes the RDC complex to mislocalize to and transcribe the 359 and Rsp satellite DNA repeats at higher levels in ovaries [79]. Based on our observation that D1 mutant ovaries exhibit up-regulated transcripts and piRNAs corresponding to the 359 bp satellite DNA (Fig 4B and 4E), we tested whether D1 may also regulate the localization of the RDC complex during oogenesis. To do so, we used a strain expressing GFP-tagged Cutoff (EGFP-Cuff) to assess RDC complex localization in ovarian nurse cells, where pericentromeric satellite DNA repeats from all chromosomes are clustered into a single prominent DAPI-dense and D1-enriched chromocenter (arrowhead; Fig 5A). In the EGFP-Cuff strain, we observed that Cuff was localized to the periphery of the nurse cell chromocenter and was also present at many sites across the nucleus, consistent with the presence of piRNA clusters at pericentromeric heterochromatin and chromosome arms (Figs 5A, S8A, and S8B). We next examined Cuff localization in D1 mutant nurse cells in comparison to the heterozygous control. In both the heterozygous control and D1 mutant, DAPI-dense chromocenters were still detected in nurse cells (yellow arrowheads; Fig 5B). However, we observed an approximately 3-fold increase in chromocenter size in nurse cells lacking D1 (Fig 5B and 5C), likely due to impaired satellite DNA clustering. Strikingly, this increase in chromocenter size in D1 mutants was accompanied by a substantial mis-localization of Cuff to satellite DNA (Fig 5B and 5D), where we observed approximately 6-fold more Cuff at D1 mutant chromocenters in comparison to the heterozygous control (Fig 5B and 5D). In addition, we observed that the expanded chromocenters in the D1 mutant nurse cells were also positive for Nxf3 (Fig 5E), which normally colocalizes with Cuff at piRNA clusters and mediates nuclear export of piRNA precursors [80,81]. Consistently, RNA in situ hybridization revealed increased transcription of the 359 satellite DNA repeat at the expanded Cuff-positive chromocenters in D1 mutant nurse cells (Fig 5F and 5G). Finally, the 359 bp probe signal was abolished following RNase A treatment (Fig 5H), which is further indication that the 359 bp satellite DNA repeat is aberrantly transcribed following loss of D1. Our data indicate that D1 inhibits RDC localization to certain complex satellite DNA repeats in Drosophila ovaries. However, unlike in Kipferl mutant ovaries, the RDC mislocalization in D1 mutant ovaries did not have a substantial effect on the levels of piRNAs targeting TEs and TE silencing. Altogether, the TE reporter experiments, gene expression analysis, and small RNA sequencing suggest that TE repression in the germ cells and somatic cells of adult gonads is largely independent of D1 or Prod.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. D1 blocks localization of the piRNA biogenesis machinery to chromocenters.

(A) A single nurse cell from an EGFP-Cuff ovary stained with D1 (magenta), Vasa (blue), and DAPI (gray). The arrowhead points to the DAPI-dense chromocenter. The insets highlight Cuff localization at the periphery of D1-containing chromocenters. (B) Localization of EGFP-Cuff in nurse cell nuclei from control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries. Yellow arrowheads point to the location of the chromocenter. (C) Box-and-whisker plot of the size of nurse cell chromocenters relative to total nuclear size in control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries; **** represents p < 0.0001 from a Student’s t test. (D) Box-and-whisker plot of EGFP-Cuff enrichment at chromocenters relative to the rest of the nurse cell nucleus in control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries; **** represents p < 0.0001 from a Student’s t test. (E) Localization of EGFP-Cuff (green) and Nxf3 (magenta) in nurse cell nuclei from control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries. Yellow arrowheads point to the location of the chromocenter. (F) RNA in situ hybridization against the 359 bp satellite DNA transcripts (plus strand, magenta) in control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries in an EGFP-Cuff (blue) background that were also stained for DAPI (gray). (G) Box-and-whisker plot of chromocenter-localized 359 bp satellite DNA transcripts in nurse cells from control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries in an EGFP-Cuff background; **** represents p < 0.0001 from a Student’s t test. All scale bars are 5 μm. (H) RNA in situ hybridization against the 359 bp satellite DNA transcripts (plus strand, magenta) in control (D1^LL03310/TM6C) and D1 mutant (D1^LL03310/Df(3R)^Bsc666) ovaries in an EGFP-Cuff (green) background that were also stained for DAPI (blue) following RNase A treatment post fixation. Source data for (C, D, G) can be found in S4 Data. piRNA, Piwi-interacting RNA.

https://doi.org/10.1371/journal.pbio.3002984.g005

Maternal and zygotic D1 contribute to TE repression in adult gonads

Although D1 and Prod appear to be dispensable for TE repression in adult gonads, our proteomic data identified that TE repression proteins were mainly co-purified by D1 and Prod in embryos. Thus, testing the functional relevance of these interactions required depleting satellite DNA-binding proteins during embryogenesis. This experiment is challenging since F₁ mutants (the progeny of heterozygous parents) contain maternal deposits during embryogenesis (S9A–S9C Fig) and protein/RNA depletion experiments can be ineffective in early embryos [82]. However, the progeny of F₁ mutants (F₂ mutants) lack both maternal and zygotic protein (S9A–S9C Fig) and allow us to test the role of satellite DNA-binding proteins in TE repression during embryogenesis. Since D1 mutants are viable and fertile (unlike prod mutants, which are larval lethal [33]), we chose to assess TE repression in F₁ D1 mutants, F₂ D1 mutants, and their associated heterozygous controls (S9A–S9C Fig). Briefly, the F₁ D1 mutant lacks zygotically expressed D1 but retains a maternal contribution during embryogenesis and potentially early larval development (S9A and S9C Fig). In comparison, the F₁ D1 heterozygote contains both maternally contributed and zygotically expressed D1 (S9A and S9C Fig). In the F₂ D1 mutants, which are the progeny of F₁ D1 mutant females, both maternal and zygotic D1 are absent during embryogenesis and adulthood (S9B and S9C Fig), while the corresponding F₂ D1 heterozygote lacks maternal D1 in the early embryo but expresses D1 upon zygotic genome activation (S9B and S9C Fig).

We initially assessed female gonad morphology in the above genotypes since unrestricted transposon expression can result in arrested germ cell development and agametic ovaries [4,83]. Strikingly, we observed that F₂ D1 mutant females predominantly exhibited agametic ovaries (Fig 6A), with the vast majority of ovarioles either lacking any discernible Vasa-positive germ cells or containing only undifferentiated Vasa-positive germ cells, which did not develop into cysts (Fig 6B and 6C). Moreover, the failure of germline development and agametic ovaries in the F2 D1 mutant strongly correlated with loss of female fertility (Fig 6D). Interestingly, the presence of either maternal or zygotic D1 was sufficient to rescue this phenotype, as gonad morphology and early germ cell development of both the F₁ D1 mutant females (maternal D1 only) and the F₂ D1 heterozygote females (zygotic D1 only) were mostly unaffected (Fig 6A–6C), even though both genotypes exhibited more variable fertility (Fig 6D). We next tested whether certain transposons were expressed in F₂ D1 mutant ovaries using single molecule RNA in situ hybridization (smFISH). Here, we observed that the copia retrotransposon, but not other TEs such as burdock or gypsy, was highly expressed in approximately 70% of early germ cells from F₂ D1 mutant ovaries (Fig 6E–6G). In comparison, copia expression was not elevated in early germ cells from the yw wild-type strain or in the F₁ D1 mutant (Fig 6A). While approximately 90% of the aubergine mutant germaria (positive control) contained copia expressing germ cells (Fig 6E and 6F), this expression was mostly restricted to differentiated germ cell cysts; undifferentiated aubergine mutant germ cells (GSCs and cystoblasts, arrowheads; Fig 6E) exhibited a markedly lower level of copia expression (Fig 6G). Thus, the high expression of copia, and potentially other TEs, occurring in the early germ cells of F₂ D1 mutants, may contribute to the loss of GSCs or block their differentiation, ultimately leading to agametic gonads. Consistently, we found that chk2 mutation rescues the agametic gonads in the F₂ D1 mutant (Fig 6H and 6I). Together, our data reveal embryogenesis (or potentially early larval development) as a critical period when the presence of maternal and/or zygotic D1 is sufficient for establishing TE silencing, which can then be heritably maintained throughout development (Fig 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Loss of D1 during embryogenesis triggers transposon expression in adult ovaries.

(A) Brightfield images of ovaries from the indicated genotypes. Scale bars: 100 μm. (B) Ovaries from the indicated genotypes were stained with Vasa (green) and DAPI (magenta). Scale bars: 5 μm. (C) Percentage of germaria containing a normal population of germ cells, only germline stem cells/cystoblasts (GSC/CB only) or completely lacking germ cells in yw (n = 48), aubergine mutant (n = 66), F₁ D1 heterozygous (n = 51), F₁ D1 mutant (n = 65), F₂ D1 heterozygous (n = 81), and F₂ D1 mutant (n = 109). (D) Bar plot of the number of progeny sired by a single female of the indicated genotypes when mated with 2 wild-type males over the course of 1 week. CV represents coefficient of variation; **** represents p < 0.0001 from an ordinary one-way ANOVA. (E) RNA in situ hybridization against Copia (green) and total polyadenylated mRNA (blue) was performed in ovaries of the indicated genotypes that were also stained for DAPI (magenta). Yellow and black arrowheads indicate undifferentiated germ cells in the aubergine mutant germarium. Scale bars: 5 μm. (F) Percentage of germaria with germ cells expressing Copia transcripts in yw (n = 62), aubergine mutant (n = 67), F₁ D1 mutant (n = 71), and F₂ D1 mutant (n = 50). (G) Box-and-whisker plot of copia expression normalized to total polyA mRNA in undifferentiated germ cells (GSCs/CBs) from yw (n = 16), aubergine mutant (n = 22), F₁ D1 mutant (n = 18), and F₂ D1 mutant (n = 29); **** represents p < 0.0001 from one-way ANOVA. (H) Brightfield images of ovaries from the indicated genotypes. Scale bars: 100 μm. (I) Percentage of germaria containing a normal population of germ cells, only germline stem cells/cystoblasts (GSC/CB only) or completely lacking germ cells in F₂ D1 heterozygous (n = 62) and F₂ D1 mutant (n = 38) ovaries in a Chk2 mutant background. Source data for (C, D, G, I) can be found in S5 Data. GSC, germline stem cell.

https://doi.org/10.1371/journal.pbio.3002984.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. The role of D1 in heritable TE repression.

An embryonic pool of D1 that originates from either maternal deposition (Mat +) and/or zygotic expression (Zyg +) is important for TE repression in adult gonads. In the absence of embryonic D1 (F₂ D1 mt, Mat–Zyg–), TE expression contributes to Chk2-mediated gonadal atrophy. Ovary image in the top panel was adapted from [66], which is distributed under the terms of the Creative Commons Attribution License and permits unrestricted use and redistribution provided that the original author and source are credited. TE, transposable element.

https://doi.org/10.1371/journal.pbio.3002984.g007

Fly husbandry and strains

All fly stocks were raised on standard Bloomington medium at 25°C unless otherwise indicated. D. melanogaster yw was used as a wild-type stock. prod^U (BDSC42686), GFP-D1 (BDSC50850), Ubi-NLS-GFP (BDSC5826, BDSC5826), Df(3R)^Bsc666 (BDSC26518), aub^QC42 (BDSC4968), aub^HN2 (BDSC8517), piwi¹ (BDSC43637), piwi² (BDSC43319), UAS-Dcr-2 (BDSC24650), mCherry ^RNAi (BDSC35785), and aub ^RNAi (BDSC39026) were obtained from the Bloomington Drosophila stock center. D1^LL03310 (DGRC140754) and FRT42D prod^k08810 (DGRC111248) were obtained from the Kyoto stock center. GFP-Burdock-lacZ (v313212), tj-gal4, gypsy-lacZ (v313222), GFP-Piwi (v313321), EGFP-Cuff (v313269), D1 ^RNAi (v101946), and Prod ^RNAi (v106593) were obtained from the Vienna Drosophila Resource Center. Hmr-HA [38] was a gift from Daniel Barbash. Sov-GFP (sov^def1; Dp(1;2)^FF025056) [55] was a gift from Ferenc Jankovics. nos-gal4 (second chromosome insertion) and chk2⁶⁰⁰⁶ were gifts from Yukiko Yamashita. nos-gal4 (third chromosome insertion) [91] and UAS-Upd [92] have been previously described.

Construction of the GFP-Prod strain

First, a 400 bp promoter upstream of the Prod start site from D. melanogaster was PCR amplified using the following primer pair, GATCAAGCTTCTGTTGTTATGCATATCGTTC and GATCGAATTCCCGGGTATCCTTGCTC and subcloned into the HindIII and EcoRI sites on pUASt-GFP-attB, replacing the UAS sequence. Subsequently, a codon-optimized Prod sequence was inserted downstream of the GFP, and transgenic flies were generated using PhiC31 integrase-mediated transgenesis into the attP40 site (BestGene). The GFP-Prod allele was generated by recombining the p400-GFP-Prod transgene with prod^k08810. Two virgin prod^U females were crossed to single tester male carrying either the p400-GFP-Prod, prod^k08810 allele, or the prod^k08810 allele as a control in vials at 25°C. The percent of trans-heterozygous prod mutant flies was quantified in each replicate.

Fertility assay

For male fertility assays, a single tester male was maintained with 2 virgin yw females in vials for 7 days and the number of resulting progenies was counted until 20 days post setup. For female fertility assays, a single tester female was maintained with 2 yw males in vials for 7 days and the number of resulting progenies was counted until 20 days post setup. Vials that contained any deceased parent flies were omitted from the analysis.

Immunofluorescence staining

Immunofluorescence staining of Drosophila tissues was performed as follows. For fixation, adult tissues were dissected in PBS, transferred to 4% formaldehyde in PBS, and fixed for 30 min. Fixation of Drosophila embryos was performed using a standard protocol. Briefly, collected embryos were moved to glass scintillation vials and washed 3 times with 1× PBS, dechorionated using 3 ml of 50% bleach solution for 90 s by vigorous shaking and washed again in 1× PBS 5 times to remove the bleach. Dechorionated embryos were fixed using a heptane: 4% paraformaldehyde (PFA) (1:1) solution by vigorously shaking for 20 min. Following fixation, the bottom aqueous phase was removed and ice-cold methanol was added to devitellinize the fixed embryos. After this step, embryos were washed 3 times in ice-cold methanol and stored in methanol at −20°C for further use. Prior to immunofluorescence staining, embryos were rehydrated using serial washes of methanol: PBS-T at 4:1, 1:1, and 1:4 volumes and followed by 1 wash in 1× PBS-T for 5 min each. For immunostaining, fixed tissues were washed in 1× PBS-T (PBS containing 0.1% Triton-X) for at least 60 min, followed by incubation with primary antibody in 3% bovine serum albumin (BSA) in PBS-T at 4°C overnight. Samples were washed 3 times in PBS-T, incubated with secondary antibody in 3% BSA in PBS-T at 4°C overnight, washed as above, and mounted in VECTASHIELD with DAPI (Vector Labs). Fluorescent images were taken using a Leica TCS SP8 confocal microscope with 63× oil-immersion objectives (NA = 1.4). Images were processed using Adobe Photoshop or ImageJ.

RNA in situ hybridization

RNA in situ hybridization on ovaries was performed as previously described [93]. Briefly, 3 to 4 ovaries were dissected in RNase-free 1× PBS and fixed in 3.7% formaldehyde in PBS for 30 min on a nutator with gentle shaking. Following fixation, samples were washed with RNase-free 1× PBS 3 times and subsequently incubated in ice-cold ethanol at 4°C overnight on a nutator. The next day, samples were washed with RNase-free wash buffer (2XSSC, 10% formamide) for 3 min at room temperature and subsequently incubated with the hybridization mix (50 to 125 nM probes, 2× SSC, 10% dextran sulfate, 1 g/L E.coli tRNA, 2 mM vanadyl ribonucleoside complex, 0.5% Rnase-free BSA, and 10% deionized formamide) at 37°C overnight. Samples were then washed twice (wash buffer) at 37°C for 30 min each and mounted in VECTASHIELD with DAPI (Vector Labs). The following probe was used to hybridize to the 359 bp satellite plus strand (AGGATTTAGGGAAATTAATTTTTGGATCAATTTTCGCATTTTTTGTAAG). Stellaris RNA FISH probes against the full-length burdock and gypsy consensus sequences were synthesized (LGC Biosearch technologies). Stellaris RNA FISH probes targeting Het-A and copia were gifts from Yukiko Yamashita. An oligo-dT probe was used to hybridize to polyadenylated RNA. Fluorescent images were taken using a Leica TCS SP8 confocal microscope with 63× oil-immersion objectives (NA = 1.4). Images were processed using Adobe Photoshop.

Image analysis

All image analysis was performed using ImageJ. For the quantitation of relative chromocenter size, we demarcated a region of interest (ROI^N) encompassing the entire nucleus as well as second ROI encompassing only the chromocenter (ROI^C) for each nurse cell (stages 4–6) analyzed. ROI^C/ROI^N equals the relative chromocenter size. For enrichment of EGFP-Cuff at chromocenters, we estimated the total EGFP-Cuff signal in ROI^C and divided it by the total EGFP-Cuff signal in ROI^N to give the percentage of EGFP-Cuff at chromocenters. For the normalized estimation of Copia expression, we estimated the average intensity of nuclear Copia (ROI^Copia) and normalized this value to the 3 cytoplasmic ROIs that estimated the average intensity of total mRNA based on the oligo-dT probe. For the normalized estimation of 359 transcripts at chromocenters, we estimated the average intensity of 359 probe signal at chromocenters (ROI^C) for each nurse cell and normalized this value to a cytoplasmic ROI that have background fluorescence.

Isolation of nuclei from embryos

Flies reared in population cages were allowed to lay eggs on fresh yeast paste for approximately 18 h. Embryos were collected using a set of sieves, washed with 1× PBS and dechorionated using 50% bleach. Next, embryos were moved to a dounce homogenizer containing light nuclear buffer (15 mM Tris-HCl (pH 7.4), 60 mM KCl, 5 mM MgCl₂, 15 mM NaCl2, and 1 M sucrose) supplemented with 1× protease inhibitor cocktail (PIC, Roche,11697498001) and 1 mM PMSF. The homogenate was then incubated on ice for 5 min and filtered through 2 layers of Miracloth tissue (Calbiochem). The filtered homogenate was then centrifuged for 20 min at 480 rpm at 4°C to pellet any debris. The supernatant containing nuclei was transferred into a fresh tube and the nuclei were pelleted at 4,000 g for 30 min at 4°C. The pellet was then resuspended in 1 ml light nuclear buffer containing 0.1 mM EDTA and the nuclei were again centrifuged at 16,000 g for 5 min before lysis.

Tissue lysis

Nuclei isolated from embryos were lysed for 90 min at 4°C with extraction buffer (15 mM Tris-HCl (pH 7.4), 60 mM KCl, 5 mM MgCl₂, 450 mM NaCl2, 1% Triton-X [vol/vol], 500 units of benzonase, 1× PIC, and 5 μl/ml PMSF). The resulting lysate was separated from the insoluble chromatin by centrifugation for 20 min at 14,000 g at 4°C. For each ovary replicate, approximately 200 pairs were dissected from <1-day-old flies, flash frozen with liquid nitrogen and stored at −80°C prior to lysis; 500 μl of ice-cold lysis buffer (450 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5% glycerol, 1 mM MgCl2, 1 mM EDTA, 0.2% NP40, 0.5 mM DTT, 1× PIC, 5 μl/ml PMSF) was added to frozen ovaries, which were then homogenized with a motorized pestle until complete breakdown of the tissue. After homogenization, samples were incubated with benzonase (Sigma Aldrich, 2 μl/500 μl of lysate) for 90 min at 4°C on a rotating wheel. Following this, samples were first centrifuged for 3 min at 10,000 g at 4°C to precipitate the debris from the tissues, homogenized again for 30 s and centrifuged for 5 min at approximately 18,700 g at 4°C. The supernatant (without the surface lipid layer) was transferred to a new 1.5 ml tube and centrifuged again for 5 min at approximately 18,700 g prior to use. For each testis replicate, approximately 100 pairs of GSC-enriched (nos-gal4 > UAS-Upd) testis tumors were dissected, flash frozen with liquid nitrogen, and stored at −80°C prior to lysis; 300 μl RIPA lysis buffer [450 mM NaCl, 0.5% NP-40 [vol/vol], 0.5% Sodium deoxycholate [mass/vol], 0.1% SDS, 50 mM Tris (pH 7.4)] containing 1× PIC (Roche,11697498001) and 5 μl/ml PMSF (Sigma-Aldrich, P7626) was added to frozen testes for 10 min. Tissue disruption was performed by applying 20 strokes with the tight pestle in a 1 ml dounce homogenizer, the samples were incubated for another 20 min on ice following which benzonase was added for 1 h at 10°C. Homogenized lysates were centrifuged at 14,000 g for 20 min at 4°C prior to use. GFP-Piwi and NLS-GFP embryos were collected as indicated in the previous section, dechorionated, flash frozen with liquid nitrogen, and stored at −80°C prior to lysis. Embryo lysis for these samples was performed in an identical manner to the ovary lysis above.

Affinity purification and mass spectrometry

Affinity purification was performed using anti-GFP nanobody coupled magnetic beads (approximately 30 μl slurry) (Chromotek AB_2631358). Briefly, beads were collected using a magnetic rack, washed 3 times using the lysis buffer and incubated with the lysate for 90 min on a rotating wheel at 4°C. After the incubation, beads were washed 3 times for 5 min with a wash buffer (450 mM NaCl, 50 mM Tris-HCl (pH 7.5), 1 mM MgCl2, 1 mM EDTA, 0.5 mM DTT) at room temperature. Washed beads containing the protein complexes were subjected to on-beads proteolysis. Samples were transferred into a 10 kDa molecular weight cutoff spin column (Vivacon 500, Sartorious), following the FASP protocol [94]. Beads in solution were dried, denaturated (8M Urea), reduced (5 mM TCEP, 30 min, 37°C) and alkylated (10 mM Iodoacetamide, 30 min, 37°C). Beads were then washed 3 times with 50 mM ammonium bicarbonate (250 μl). During the buffer exchange, samples were centrifuged at 10,000 g. Subsequently, samples were proteolyzed with 0.5 μg of Trypsin (Promega, sequencing grade) for 16 h at 37°C. The proteolysis was quenched with 5% formic acid and peptides were subjected to C18 cleanup (BioPureSPN PROTO 300 C18, Nest group), following the manufacturer’s procedure. The eluted peptides were then dried using a speedvac and resuspended in 20 μl of 2% acetonitrile and 0.1% formic acid.

Mass spectrometry data acquisition

The MS data sets were acquired with 2 different instrumental setups. The first setup (A) was utilized for the D1, Prod, and NLS-GFP samples from embryos, ovaries, and testes. The second setup (B) was employed for acquiring the Piwi and NLS-GFP samples from embryos. In setup A, LC-MS/MS acquisition was performed on an Orbitrap QExactive+ mass spectrometer (Thermo Fisher) coupled to an EASY-nLC-1000 liquid chromatography system (Thermo Fisher). Peptides were separated using a reverse phase column (75 μm ID × 400 mm New Objective, in-house packed with ReproSil Gold 120 C18, 1.9 μm, Dr. Maisch GmbH) across a gradient from 3% to 25% in 170 min and from 25% to 40% in 10 min (buffer A: 0.1% (v/v) formic acid; buffer B: 0.1% (v/v) formic acid, 95% (v/v) acetonitrile). The DDA data acquisition mode was set to perform 1 MS1 scan followed by a maximum of 20 scans for the top 20 most intense peptides with MS1 scans (R = 70’000 at 400 m/z, AGC = 1e6 and maximum IT = 100 ms), HCD fragmentation (NCE = 25%), isolation windows (1.8 m/z), and MS2 scans (R = 17’500 at 400 m/z, AGC = 1e5 and maximum IT = 55 ms). A dynamic exclusion of 30s was applied and charge states lower than 2 and higher than 7 were rejected for the isolation. In setup B, LC-MS/MS acquisition was performed on an Orbitrap Exploris480 mass spectrometer (Thermo Fisher) coupled to a Vanquish Neo liquid chromatography system (Thermo Fisher). Peptides were separated using a reverse phase column (75 μm ID × 400 mm New Objective, in-house packed with ReproSil Gold 120 C18, 1.9 μm, Dr. Maisch GmbH) across a linear gradient from 7% to 35% in 120 min (buffer A: 0.1% (v/v) formic acid; buffer B: 0.1% (v/v) formic acid, 95% (v/v) acetonitrile). The DDA data acquisition mode was set to perform 1 MS1 scan followed by a maximum of 20 scans for the top 20 most intense peptides with MS1 scans (R = 60’000 at 400 m/z, AGC = 1e6 and auto maximum IT), HCD fragmentation (NCE = 28%), isolation windows (1.4 m/z), and MS2 scans (R = 15’000 at 400 m/z, AGC = 2e5 and auto maximum IT). A dynamic exclusion of 25s was applied and charge states lower than 2 and higher than 6 were rejected for the isolation.

Mass spectrometry data analysis

Acquired spectra were analyzed using the MaxQuant software package version 1.5.2.8 embedded with the Andromeda search engine against Drosophila proteome reference data set (http://www.uniprot.org/, downloaded on 18.01.2021, 22’044 proteins including not reviewed proteins) extended with reverse decoy sequences. The search parameters were set to include only full tryptic peptides, maximum 2 missed cleavage, carbamidomethyl as static peptide modification, oxidation (M) and deamidation (N-ter) as variable modification and “match between runs” option. The MS and MS/MS mass tolerance was set to 10 ppm. False discovery rate of <1% was used for protein identification. Protein abundance was determined from the intensity of the top 2 unique peptides and were only considered for downstream analysis if identified in all D1/Prod/Piwi replicates. Log2 intensity values were median normalized and missing values were imputed as follows. In instances where the protein of interest was not identified in all control replicates, we imputed the missing values from a normal distribution generated with an average that is 4 times lower than the mean of the D1/Prod/Piwi sample values and a matching standard deviation. In instances where the protein of interest was identified in only 1 control replicate, we imputed the missing values based on the unique value identified for the control and a standard deviation matching the pulldown samples. In instances with only 1 missing value for the control replicates, we imputed using a normal distribution generated with the average and standard deviation of the control values.

Statistical analysis was performed using unpaired two-sided t test and multiple testing correction was performed using the Benjamini–Hochberg method. Hits identified from the differential analysis between the bait pulldown versus the NLS-GFP control, with log₂FC>1 and p_adj<0.1, were considered as interacting proteins. For cross-bait and cross-tissue comparisons in Figs 2C, 2E, and S2B–S2D and structural modeling in S4 Table, we used a relaxed cutoff with log₂FC>1 and uncorrected p < 0.05. In Prod pulldowns from embryos, we observe both previously identified interactions and heterochromatin-associated proteins suggesting that the surprising absence of detectable Prod spectra is a technical issue. For pathway enrichment analysis, we first generated a list of Drosophila DNA repair proteins (GO:0006281, S5 Table) and a custom list of transposon repression proteins based on the literature (S6 Table) and used a hypergeometric test to calculate the enrichment of proteins belonging to these lists in the D1/Prod pulldowns.

Structural predictions using Alphafold2

To predict the structures of binary protein–protein interactions, we used version 2.3.2 of Alphafold2, which contains updated parameters for multimeric protein–protein interactions [45]. Default values for all parameters and for generating multiple sequence alignments were used. To rank the models according to a confidence metric, we used the ipTM score, which has been show to distinguish interacting from non-interacting proteins [95]. Structural models are available upon request.

Co-immunoprecipitation and western blotting

For the co-immunoprecipitation, 125 to 250 mg of frozen embryos were thawed on ice and lysed in an identical manner to the ovary lysis protocol detailed above. Approximately 50 μg of lysate was collected as input while approximately 970 μg of lysate at a concentration of 2 mg/ml was incubated with anti-GFP nanobody coupled beads (approximately 50 μl slurry) (Chromotek AB_2631358) for 90 min at 4°C. After the incubation, beads were washed 3 times for 5 min with a wash buffer (450 mM NaCl, 50 mM Tris-HCl (pH 7.5), 1 mM MgCl2, 1 mM EDTA, 0.5 mM DTT) at room temperature. Protein complexes bound to the beads were eluted in 80 μl of SDS sample buffer by heating at 95°C for 5 min. Input samples were also brought to a final volume of 80 μl by adding 30 μl of SDS sample buffer. Equal volumes of input and IP samples were loaded on a 4% to 12% SDS gradient gel (Sigma Aldrich). Immunoprecipitated proteins were detected using mouse anti-GFP (1:1,000, Roche 1181 4460001), guinea pig anti-D1 (1:1,000, [26]), and rabbit anti-histone H3 (1:1,000, Abcam ab1791). Secondary antibodies used were Alexa Fluor Plus 800 donkey anti-mouse (1:10,000, Invitrogen) and donkey anti-guinea pig IR dye 680 (1:10,000, Li-Cor). Western blots were imaged using the Odyssey DLx fluorescence imager (Li-Cor).

RNA sequencing and data analysis

For each sample, approximately 5 pairs of ovaries from 4-day-old females were dissected in RNase-free 1× PBS and flash frozen in liquid nitrogen until RNA extraction. RNA extraction for each replicate was performed using the RNeasy RNA extraction kit (Qiagen). Samples were treated with DNase post RNA extraction and purified using an RNA purification kit (Promega). RNA concentrations were assessed using a Nanodrop as well as a Qubit RNA analyzer for sample quality and RIN scores. rRNA-depleted libraries were prepared and sequenced by Novogene on an NovaSeq 6000 (Illumina) using paired-end 150 bp sequencing. The resulting raw reads were cleaned by removing adaptor sequences, low-quality-end trimming, and removal of low-quality reads using BBTools v 38.18 (Bushnell, B. BBMap. Available from: https://sourceforge.net/projects/bbmap/). The exact commands used for quality control can be found on the Methods in Microbiomics webpage (Sunagawa, S. Data Preprocessing—Methods in Microbiomics 0.0.1 documentation. Available from: https://methods-in-microbiomics.readthedocs.io/en/latest/preprocessing/preprocessing.html). The quality-controlled reads were aligned against BDGP6.32 using STAR aligner [96]. Transcript and transposon abundances were quantified using Tetranscripts [97]. Differential gene expression analysis was performed using Bioconductor R package DESeq2 v 1.37.4 [98].

Small RNA sequencing and data analysis

Small RNA sequencing was performed as previously described [79]. In brief, approximately 5 pairs of ovaries from 3 biological replicates per genotype were lysed and Argonaute-sRNA complexes were isolated using TraPR ion exchange spin columns. sRNAs were subsequently purified using acidic phenol; 3′ adaptors containing 6 random nucleotides plus a 5 nt barcode on their 5′ end and 5′ adaptors containing 4 random nucleotides at their 3′ end were subsequently ligated to the small RNAs before reverse transcription, PCR amplification, and sequencing on an Illumina NovaSeqSP instrument. Raw reads were trimmed for linker sequences, barcodes, and the 4/6 random nucleotides before mapping to the D. melanogaster genome (dm6), using Bowtie (version.1.3.0, settings: -f -v 3 -a—best—strata—sam) with 0 mismatches allowed. Genome-mapping reads were crossreferenced with Flybase genome annotations (r6.40) using BEDTools to allow the removal of reads mapping to rRNA, tRNA, snRNA, snoRNA loci, and the mitochondrial genome. For TE mappings, all genome mappers were used allowing no mismatches. Reads mapping to multiple elements were assigned to the best match. Reads mapping equally well to multiple positions were randomly distributed. Libraries were normalized to 1 million sequenced microRNA reads. For calculation of piRNAs mapping to TEs, only antisense piRNAs were considered, and counts were normalized to TE length.