Rapid disassembly and Piwi-independent reassembly of Drosophila piRNA cluster heterochromatin following acute heat shock

Abstract

Environmental stress activates transposons and is proposed to generate genetic diversity that facilitates adaptive evolution. piRNAs guide germline transposon silencing, but the impact of stress on the piRNA pathway is not well understood. In Drosophila, the Rhino-Deadlock-Cuff complex (RDC) drives transcription of clusters composed of nested transposon fragments, generating precursors that are processed into mature piRNAs in the cytoplasm. We show that acute heat shock triggers rapid, reversible loss of RDC localization and cluster transcript expression with coordinate changes in the cytoplasmic processing machinery. Maternal piRNAs bound to Piwi are proposed to guide Rhino localization to clusters during early embryogenesis. However, RDC relocalization after heat shock is accelerated in piwi mutants and delayed in thoc7 mutants, which disrupt piRNA precursor binding to THO complex, and we show that maternally deposited piRNAs are dispensable for RDC localization to the major 42AB cluster. Cluster specification is reconsidered in light of these findings.

INTRODUCTION

Environmental stress activates transposons in evolutionarily diverse systems (Grandbastien et al. 1997; Van Meter et al. 2014; Fouche et al. 2020). Although transposable element mobilization can induce deleterious mutations linked to disease (Belancio et al. 2009; Payer and Burns 2019), McClintock proposed that stress-induced transposon mobilization could generate beneficial genetic diversity, facilitating adaptive evolution (McClintock 1984). It is now clear that transposons have been co-opted for many essential host functions (Feschotte and Pritham 2007). However, transposition must occur in the germline to alter the inherited genome and drive evolution, and the germline response to stress is poorly understood (Cappucci et al. 2019; Kelleher et al. 2020; Mombach et al. 2022).

piRNAs control transposon silencing in the germline, and this pathway is best understood in the Drosophila female germline. In Drosophila, 23–30 nt small silencing piRNAs are derived from genomic loci composed of nested transposon fragments called clusters and from a subset of isolated transposon insertions bound by the HP1 homolog Rhino (Rhi) (Bergman et al. 2006; Brennecke et al. 2007; Klattenhoff et al. 2009; Mohn et al. 2014; Shpiz et al. 2014). Rhino anchors the RDC complex, consisting of Rhino, Deadlock, and Cuff, which recruits Moonshiner and transcription initiation factors that drive transcription from both genomic strands, producing long unspliced transcripts (Andersen et al. 2017). These piRNA precursors are bound by the conserved THO complex and UAP56, which are components of the transcription and export (TREX) complex (Mohn et al. 2014; Zhang et al. 2014, 2018; Chen et al. 2016; Hur and Chung 2016; Hur et al. 2016). piRNA cluster transcripts are transferred from TREX to a specialized noncanonical export system that includes the Drosophila-specific Nxf3 protein and CRM1 (Drosophila melanogaster gene embargoed) (ElMaghraby et al. 2019; Kneuss et al. 2019; Mendel and Pillai 2019).

Following export, these transcripts undergo ping-pong and phased processing. The ping-pong machinery localizes to perinuclear nuage granules and includes multiple Tudor domain proteins, the DEAD box RNA-binding protein Vasa, and the PIWI clade Argonaute proteins Aubergine (Aub) and Ago3 (Brennecke et al. 2007; Lim and Kai 2007; Xiol et al. 2014; Webster et al. 2015). Within the ping-pong cycle, piRNAs bound to Aub guide cleavage of transposon transcripts, leading to posttranscriptional silencing and production of precursors for piRNA that bind Ago3 and guide cluster transcript cleavage, generating the precursors for Aub-bound piRNAs (Brennecke et al. 2007). The ping-pong cycle also generates the precursors for phased processing, which generates head-to-tail arrays of piRNAs bound to Piwi, the founding member of the PIWI clade (Han et al. 2015; Mohn et al. 2015; Wang et al. 2015b). The RNA helicase Armi localizes to nuage and mitochondria and is proposed to shuttle precursors between these two compartments to facilitate phased processing with the mitochondrial nuclease Zuc (Cook et al. 2004; Pandey et al. 2017; Ge et al. 2019). piRNA binding by Piwi triggers localization to the nucleus, where the complexes mediate transcriptional silencing with the interaction partner Panoramix/Silencio (Sienski et al. 2015; Yu et al. 2015).

The piRNA response to heat stress has not been extensively characterized, but “heavy” heat shock, triggered by cycling between 4°C and 37°C, is reported to upregulate some transposable element families and lead to Ago3 localization to lysosomes (Cappucci et al. 2019). In contrast, a short 30 min heat shock at 38°C has been reported to trigger a modest change in piRNA levels mapping to transposons, but the target transposons are not upregulated, and piRNA levels were not reduced for the elements that were upregulated (Funikov et al. 2015). Additionally, long-term maintenance of flies at 30°C is reported to lower piRNA expression while increasing ping-pong processing of some piRNAs (Fast and Rosenkranz 2018). These studies suggest that heat stress alters germline transposon expression, but the impact of heat shock on organization and function of the piRNA biogenesis machinery has not been systematically analyzed.

Here we show that acute heat shock leads to rapid, reversible loss of RDC localization and cluster transcript expression. Time course studies indicate that loss of TREX localization follows RDC dissociation and that TREX localization recovers after the RDC. We also show that heat stress leads to reversible localization of Vasa to the nucleus and Aub to mitochondria, which may reflect trapping of these factors at normally transient compartments in the biogenesis cycle.

Maternally deposited piRNAs bound to Piwi have been proposed to guide RDC localization to clusters during oogenesis, while maintenance of Rhino localization to clusters is proposed to be Piwi and piRNA independent (Brennecke et al. 2008; Akkouche et al. 2017). However, here we show that piwi mutations accelerate recovery of Rhino localization following heat shock, and that genetically eliminating maternal piRNAs mapping the major piRNA cluster at 42AB has no significant impact on Rhino binding or piRNA production from a paternally supplied copy of this cluster. Heat shock thus provides novel insights into piRNA cluster chromatin assembly and the function of piRNAs and Piwi in this process.

RESULTS

Heat shock disrupts piRNA cluster heterochromatin

Heat shock increases transposon expression in plants and animal somatic cells (Grandbastien et al. 1997; Belgnaoui et al. 2006; Raje et al. 2018; Garrigues et al. 2019; Sun et al. 2020; Roquis et al. 2021; Takehira et al. 2021), and stress-induced transposition could facilitate adaptive evolution by generating beneficial genetic diversity. However, adaptive mutations must be transmitted through the germline to impact reproductive fitness, and in Drosophila, the germline response to heat shock has not been extensively characterized (Jardim et al. 2015; Fanti et al. 2017; Cappucci et al. 2019). To define the effects of heat shock on transposon expression in the D. melanogaster germline, we incubated w¹ flies at 37°C for 1 h and performed RNA-seq at 1, 12, 24, 48, and 72 h after shifting back to room temperature. At 1 h post-heat shock, R1_DM, Copia2, and Invader1 transposon families were upregulated over 10-fold relative to controls (Fig. 1A). However, expression of all other transposons families was suppressed relative to controls, and reduced expression persisted for 24 h (scatter plots, Fig. 1A). Transposon expression is restored to control levels by 48 h and at 72 h post-heat shock (Fig. 1A). The three elements that were initially overexpressed are progressively silenced during recovery.

FIGURE 1.

Acute heat shock downregulates all but a few transposons and the major piRNA cluster 42AB. (A) Representative scatter plots comparing all mapping transposon family expression (rpkm, log₁₀ scale) in w¹ no heat shock control versus samples heat shocked at 37°C for 1 h and allowed to recover for 1, 12, 24, 48, and 72 h. Transposon family expression was measured using Illumina long RNA sequencing (n ≥ 2), and each point represents a transposon family; families with an expression difference of fivefold or greater are labeled and colored red; dotted line is twofold difference. (B) Signal tracks on the left show all mapping reads from long RNA sequencing against the hsp70/αγ element genomic region on Chromosome 3R. This region contains numerous transposon insertions, including multiple Invader1 and Copia2 insertions. The tracks are in order: the reference gene sequence track (navy blue), repeat masker (black), transposon annotations (pink), w¹ no heat shock controls, and w¹ heat shocked for 1 h at 37°C allowed to recover for 1, 12, and 24 h. Heat shock induces expression of hsp70B and Invader1 and Copia2, which decrease expression coordinately. As all mapping sequences do not correspond to one location, we also included tracks showing unique mapping long sequencing signal in the smaller insert from the region highlighted by the orange box directly around the αγ element on the right. Tracks are in the same order, with different scales from above. Heat shock induces unique RNA expression from the αγ element and the surrounding region. This unique expression indicates that at least some Invader1 and Copia2 expression originates from this genomic region after heat shock. (C) Unique mapping long RNA signal tracks for cluster 42AB in w¹ control and w¹ heat shocked for 1 h at 37°C, allowed to recover for 1, 12, 24, 42, and 72 h. Heat-shocked samples show lower expression across cluster 42AB at 1 and 12 h of recovery time; this expression increases with time and recovers to wild-type levels by 48 and 72 h post-heat shock. P-values were calculated for 42AB expression levels (rpkm) comparing each heat shock with recovery time point to no heat shock controls and are displayed at the bottom left corner of each graph, (n.s.) not significant.

Heat shock suppresses expression of most protein coding genes (Jamrich et al. 1977), but DESeq2 comparison of no heat shock controls and heat shock with 1 h recovery indicated that <1% of genes were differentially expressed (adjusted P-value < 0.05), with the differently expressed genes enriched for heat shock and unfolded protein response pathways (HSPs; Supplemental Fig. S1A). A significant fraction of the transcriptome is deposited in developing oocytes for use during embryogenesis, and stable maternal RNAs are likely to mask any change in mRNA abundance due to reduced transcription.

In adult ovaries, acute heat shock thus leads to transient overexpression of a very limited number of transposon families. Two of these families, Copia2 and Invader1, are present in multiple copies within the 42AB piRNA cluster (Supplemental Fig. S1B). To determine if increased expression of these elements is associated with reduced piRNA levels, we sequenced 18–30 nt long RNAs 1 h after heat shock treatment, when R1_DM, Copia2, and Invader1 have reached maximal expression and many other transposon families show reduced expression. However, we did not detect a significant change to piRNAs mapping to the suppressed or overexpressed transposon families, or to piRNAs mapping to clusters (Supplemental Fig. S1C). Significantly, multiple copies of Copia2 and Invader1 are adjacent to the αγ element, which lies between divergently transcribed hsp70 genes (Fig. 1B; Maside et al. 2002; Kuan et al. 2009; Brewer-Jensen et al. 2016), and expression of these families is highly correlated with expression of flanking hsp70 genes (Fig. 1B). We therefore speculate that increased expression of these transposon families is driven by heat shock induced transcription, not loss of piRNA silencing.

Heat shock significantly reduces expression of long RNAs that map uniquely to both strands of the 42AB piRNA cluster (Fig. 1C). Expression from both strands is partially restored at 24 h post-heat shock, and fully restored by 48 and 72 h post-heat shock. This pattern of loss and recovery extends to most of the piRNA clusters (Supplemental Fig. S1D). The HP1 homolog Rhino anchors a complex that promotes transcription from both genomic strands (Volpe et al. 2001; Klattenhoff et al. 2009). We therefore assayed Rhino localization by indirect immunofluorescence and confocal microscopy, using a mono-specific antibody, and localized the 42AB cluster by FISH using probe to unique DNA sequences immediately adjacent to the cluster. In control nurse cell nuclei, Rhino localized to multiple foci, including prominent foci adjacent to 42AB (Fig. 2A). In contrast, at 1 h postheat shock, all Rhino foci were lost, including the major foci adjacent to 42AB (Fig. 2B). Twenty-four hours after heat shock, Rhino foci, including foci adjacent to 42AB, were restored (Fig. 2C).

FIGURE 2.

Heat shock causes the loss of Rhino localization at 42AB cluster genomic region. Rhino partially recovers after 72 h. (A–C) Confocal images of nurse cell nuclei in w¹ control (A) versus w¹ heat shocked at 37°C for 1 h and allowed to recover for 1 h (B) and 24 h (C) labeled for Rhino (anti-Rhino, green), DNA FISH for unique sequence directly adjacent to cluster 42AB (42AB probe, red), and nuclear envelope (blue). Inset in the merged image magnifies a focus pointed out by the hollow arrows. Rhino foci are next to 42AB probe foci before heat shock and 24 h recovery post-heat shock, whereas Rhino foci are lost following heat shock and 1 h recovery. Scale bar, 2 μm. (D) Quantification of the spatial relationship between Rhino and 42AB DNA probe. The percentage of 42AB probe foci next to a Rhino foci was measured for each z-section (see Materials and Methods for how “next to” was defined). Foci are defined as 5 SD above the background. For each condition, z-sections from more than eight nuclei were acquired, and each dot in the graph represents a single z-section. 100% indicates that all 42AB probe foci were associated with Rhino foci, and 0% indicates that all 42AB probe foci were not adjacent to Rhino foci. P-values for the indicated comparisons are above the bars and not significant is marked as (n.s.). (E) CUT&RUN signal tracks showing reads uniquely mapping to 42AB piRNA cluster for anti-Rhino (red) and anti-IgG (black) in w¹ flies subjected to the following conditions: no heat shock control and heat shock at 37°C for 1 h with 1, 24, or 72 h of recovery. Flies heat shocked at 37°C for 1 h and allowed to recover for 1 and 24 h showed no Rhino enrichment at 42AB relative to IgG control. Flies allowed to recover 72 h post-heat shock show some Rhino enrichment at 42AB; however, the degree of enrichment is less than no heat shock controls.

To quantify these observations, we used laser scanning confocal microscopy to optically section nurse cell nuclei and computationally defined 42AB and Rhino foci, using a signal threshold of 5 SD above the average for the field. Rhino foci “adjacent” to 42AB/FISH foci were identified based on signal overlap (see Materials and Methods). Acute heat shock thus triggers rapid, reversible loss of Rhino localization to the 42AB cluster (Fig. 2D).

To further characterize the impact of temperature on Rhino localization, we varied culture temperature and directly assayed GFP-Rhino by confocal microscopy. To define the temperature required to disrupt Rhino localization, flies were raised at 25°C and shifted to elevated temperatures for 1 h. Ovaries were then dissected, and living egg chambers were imaged. Following a 1 h shift to 34°C, Rhino foci were retained, but a 1 h shift to 35°C led to complete loss of foci (Supplemental Fig. S2E). To determine if Rhino localization is sensitive to the 12°C change from 25°C to 37°C or absolute temperature, we raised GFP-Rhino flies at 18°C and shifted them to 30°C for 1 h. Rhino foci remained following this treatment (Supplemental Fig. S2F). Localization of Rhino to nuclear foci is not disrupted by temperature shift but is sensitive to temperatures that are encountered in the environment.

To characterize chromatin binding following recovery from heat shock, we used CUT&RUN (Skene and Henikoff 2017; Ahmad 2018) to determine the genome-wide distribution of Rhino. Consistent with earlier observations, Rhino is enriched at 42AB in control ovaries (Fig. 2E). One hour after 37°C heat shock, in contrast, Rhino was undetectable across 42AB (Fig. 2E). Surprisingly, Rhino binding remained undetectable 24 h after heat shock, when transcription had partially recovered and Rhino cytologically localized to 42AB (Fig. 2E). At 72 h of recovery post-heat shock, low-level Rhino binding to 42AB cluster was detected (Fig. 2E), while cluster transcript levels had returned to control levels (Supplemental Fig. S1D).

Rhino is an HP1 homolog and binds to H3K9me3 at piRNA clusters (Klattenhoff et al. 2009; Mohn et al. 2014; Zhang et al. 2014). To determine if loss of Rhino localization following heat shock is associated with changes in chromatin modification, we used CUT&RUN to assay the genome-wide distribution of H3K9me3. H3K9me3 enrichment at piRNA clusters at 1 and 24 h post-heat shock was comparable to no heat shock controls (Supplemental Fig. S2A,B). The repressive H3K27me3 mark is also enriched at clusters and unaltered by heat shock (Supplemental Fig. S2C,D). Heat shock thus does not alter epigenetic marks associated with clusters, yet cluster transcription and cytological localization of Rhino are restored upon recovery from heat shock. It is unclear why Rhino binding fails to recover when assayed by CUT&RUN, but functional and cytological measures indicate that the RDC reassembles and promotes transcription following heat shock.

Disassembly and reassembly of the piRNA precursor transcription and export machinery

The RDC components, Rhino, Del and Cuff, colocalize at piRNA clusters and promote noncanonical transcription from both strands, generating precursor transcripts (Klattenhoff et al. 2009; Pane et al. 2011; Mohn et al. 2014; Zhang et al. 2014; Andersen et al. 2017). To define the kinetics of RDC disassembly during heat shock, we shifted female flies expressing GFP-Cuff to 37°C for 15, 23, and 30 min, fixed the ovaries, immunolocalized Rhino, and quantified Cuff and Rhino signals in nuclear foci. As anticipated, Rhino and Cuff colocalize prior to heat shock (Fig. 3A), and the signal for both proteins is highly correlated (scatter plot, Fig. 3J). After a 15 min heat shock, in contrast, Rhino is preferentially lost from nuclear foci, reflected in a shift to the left in the scatter plot comparing Cuff to Rhino signal (Fig. 3J). Between 23 and 30 min, Rhino is almost undetectable, and Cuff signal has declined, but remains detectable at 23 min (Fig. 3B,C,J). We also find that 30 min heat shock displaces Deadlock from nuclear foci (Supplemental Fig. S3A). Heat shock thus temporally uncouples Cuff from Rhino localization.

FIGURE 3.

Disassembly and reassembly of the piRNA precursor transcription and export machinery. (A–C) Confocal images of a nurse cell nucleus in GFP-Cuff expressing flies labeled for GFP-Cuff (anti-GFP, red), Rhino (anti-Rhino, green), and merge for the following conditions: (A) no heat shock controls, (B) 23 min heat shock at 37°C, and (C) 30 min heat shock at 37°C. The nucleus is outlined in white, and levels have been adjusted to achieve the maximum dynamic range. Arrows point to a focus that is enlarged in the inset. Colocalization of GFP-Cuff and Rhino is detected in no heat shock controls and reduced upon 23 min of heat shock at 37°C. Heat shock at 37°C for 30 min showing the loss of Cuff and Rhino signal. Scale bar, 2 μm. (D–I) Confocal images of a nurse cell nucleus in GFP-Thoc5 expressing flies labeled for GFP-Thoc5 (anti-GFP, red), Rhino (anti-Rhino, green), and merge for the following conditions: no heat shock controls (D) and heat shock at 37°C for 1 h with 1, 2, 4, 6, and 8 h recovery (E–I, respectively). The nucleus is outlined in white, and levels have been adjusted to achieve the maximum dynamic range. Arrows point to a focus that is enlarged in the inset. No heat shock controls show colocalization of GFP-Thoc5 and Rhino signal. Heat shocked at 37°C for 1 h and allowed to recover for 1, 2 and 4 h showing loss of Rhino and GFP-Thoc5. Rhino foci were detected at 6 h recovery after heat shock, and both Rhino and GFP-Thoc5 foci colocalization were detected at 8 h recovery post-heat shock. Scale bar, 2 μm. (J) Scatterplot comparing the normalized signal of Rhino and GPF-Cuff foci in control with a 37°C heat shock for the indicated time. Minimally, four nurse cell nuclei were selected by DAPI staining for each condition and subsequently optically sectioned with 1 μm steps. Each point represents a focus of Rhino or GFP-Cuff 5 SD above the background with the signal normalized to the maximum fluorescence in control. (K) Scatterplot comparing the normalized signal of Rhino and GPF-Tho5 foci in control and after a 37°C heat shock for 1 h and allowed to recover for the indicated time. Each point represents focus of Rhino and/or GFP-Thoc5. The quantification method used was the same as above.

The RDC promotes production of piRNA cluster transcripts that are bound by the TREX complex and exported to the cytoplasm (Mohn et al. 2014; Zhang et al. 2014, 2018). The TREX colocalizes with Rhino and is displaced from nuclear foci in rhi mutants (Zhang et al. 2018). To determine the relationship between TREX and Rhi localization and recovery from heat shock, we incubated flies expressing Thoc5-GFP at 37°C for 1 h and assayed Rhino and Thoc5-GFP localization at 1, 2, 4, 6, and 8 h of recovery at room temperature. Figure 3D–I shows representative images before heat shock and during recovery. Quantification of Rhino and Thoc5 signal at nuclear foci is shown in the scatter plots in Figure 3K. In no heat shock control egg chambers, Rhino and Thoc5 colocalize to a subset of foci (Fig. 3D), but Rhino or Thoc5 individually dominates other foci. Following heat shock and 1 h of recovery, Rhino is lost from nuclear foci, but Thoc5-GFP persists at a subset of foci (Fig. 3E). Rhino foci are first detected in a subset of nuclear foci following 4 h of recovery, and the number and intensity of foci increase at 6 h (Fig. 3G,H). Cuff colocalizes with Rhino at the earliest time points where Rhino recovers (image and scatter plot Supplemental Fig. S3B,C), consistent with coassembly. In contrast, Thoc5-GFP is not detected at Rhino foci at 4 h (Fig. 3K) but is detected at a subset of Rhino foci at 6 h, with colocalization progressively increasing at later time points (Fig. 3H,I,K). Localization of UAP56, which associates with THO in the TREX complex, shows a similar recovery pattern (Supplemental Fig. S3D). These findings suggest that the RDC association with clusters drives transcription, leading to subsequent cotranscriptional localization of the TREX.

Heat shock control of cytoplasmic piRNA machinery

TREX-dependent precursor nuclear export is followed by ping-pong processing and phased piRNA biogenesis in perinuclear nuage and at the surface of mitochondria. Aub and Vasa localize to nuage and are required for ping-pong amplification (Brennecke et al. 2007; Malone et al. 2009; Xiol et al. 2014). To determine the impact of heat shock on the ping-pong machinery, we analyzed Vasa and Aub GFP fusion protein distribution in live egg chambers. Before heat shock, GFP-Vasa localizes to nuage, and GFP-Aub is concentrated in nuage and present at lower levels in the cytoplasm (Fig. 4A,C). After heat shock, GFP-Aub is reduced at nuage and associates with large cytoplasmic aggregates (Fig. 4C), and GFP-Vasa is reduced at nuage and localizes to nurse cell nuclei (Fig. 4A). Mutations in rhi, thoc7, or uap56 disrupt Vasa and Aub localization to nuage but do not lead to Vasa localization to the nucleus or Aub localization to cytoplasmic aggregates (Klattenhoff et al. 2009; Zhang et al. 2012, 2018). Accumulation of Vasa in the nucleus and Aub in cytoplasmic aggregates thus appears to be a direct consequence of heat shock. Reduced nuage localization, in contrast, could be secondary to heat shock disruption of the RDC and TREX, or a more direct response to heat shock.

FIGURE 4.

Acute heat shock causes accumulations of Vasa in the nucleus and Aub in mitochondria. (A) Confocal images of an egg chamber stained for the piRNA pathway protein Vasa (anti-Vasa) in w¹ no heat shock control and 1 h recovery after a 37°C heat shock for 1 h. w¹ flies that were heat shocked showed Vasa staining inside the nucleus. Scale bar, 10 μm. (B) Live confocal images of GFP-Vasa dissected egg chambers treated with 10 mM leptomycin for 2 h or DMSO control. Graphs to the right of images show line scans of GFP-Vasa signal from the single nurse cell nucleus shown in the inset. Scale bar, 10 μm. (C) Confocal images of an egg chamber expressing GFP-Aub (green) and RFP-Zucchini (red) from flies that were heat shocked for 1 h at 37°C with 1 h recovery and no heat shock controls. GFP-Aub is predominantly at perinuclear foci (nuage) and RFP-Zucchini localize in the cytoplasm for no heat shock controls. GFP-Aub localization to cytoplasmic structure is observed after heat shock, whereas RFP-Zucchini signal remains cytoplasmic. Scale bar, 10 μm.

Vasa binds to cluster transcripts and localizes to nuage granules that are biased toward regions of the nuclear envelope opposite clusters (Zhang et al. 2012), and the C-terminal HELICc domain of Vasa tethered to GFP localizes in the nucleus and in the cytoplasm (Wang et al. 2015a). We therefore speculated that Vasa transiently localizes to the nucleus, where it associates with cluster transcripts, and heat shock traps Vasa in the nucleus. Cluster transcripts are exported from the nucleus through a CRM1-dependent process (ElMaghraby et al. 2019; Kneuss et al. 2019; Mendel and Pillai 2019), and leptomycin B blocks CRM-1-dependent nuclear export (Findley et al. 2003; Prasad et al. 2007). Significantly, we find that GFP-Vasa localizes to nuage and nurse cell nuclei after 2 h of incubation in 10 µM leptomycin B (Fig. 4B). The line scans in Figure 4B show Vasa signal across nurse cells. In control egg chambers, the signal drops to near background levels in the nucleus, but nuclear and cytoplasmic signals are comparable following leptomycin B treatment (Fig. 4B). Vasa thus appears to cycle between nuage and the nucleus, with heat shock trapping a transient nuclear intermediate.

To further characterize the impact of heat shock on the piRNA biogenesis machinery, we assayed localization of Ago3, Maelstrom, Squash, Armi, and Zuc. Ago3 localizes to nuage and functions with Aub and Vasa in ping-pong amplification (Malone et al. 2009; Wang et al. 2015b; Webster et al. 2015). Maelstrom suppresses canonical cluster transcription and is concentrated in nuage but also found dispersed in the nucleus and cytoplasm (Chang et al. 2019). Squash localizes to nuage and appears to function downstream from piRNA biogenesis in transposon silencing (Pane et al. 2007; Malone et al. 2009). Armi localizes to nuage and mitochondria and functions with the mitochondrial nuclease Zuc in phased piRNA biogenesis (Ge et al. 2019). Using a combination of live imaging and immunofluorescence, we find that heat shock displaces Squash from nuage and leads to Maelstrom localization to cytoplasmic aggregates (Supplemental Fig. S4A,B). However, heat shock has little impact on Ago3, Zuc, or Armi localization (Fig. 4C; Supplemental Fig. S4C,D). Heat shock thus disrupts the piRNA cluster transcription complex, the associated nuclear export machinery, and nuage localization of critical ping-pong amplification and silencing factors. In contrast, heat shock has relatively little impact on the localization of the phased piRNA biogenesis machinery.

Genetic control of cluster chromatin assembly

piRNAs mapping to clusters are maternally deposited in the oocyte (Brennecke et al. 2008), and transient depletion of Piwi during embryogenesis has been reported to compromise piRNA cluster heterochromatin assembly in adults (Akkouche et al. 2017). In contrast, zygotic piwi mutations, which deplete this nuclear PIWI clade Argonaute through later stages of development, do not block cluster propagation or piRNA production in adults (Akkouche et al. 2017). These findings suggest that maternal piRNAs bound to Piwi direct RDC localization to heterochromatin in the embryo and that RDC localization is then maintained by a Piwi-independent epigenetic mechanism. To determine the role of Piwi in cluster chromatin recovery following heat shock, we incubated piwi^2/NLS mutant females at 37°C for 1 h and assayed Rhino localization following a shift back to room temperature. The piwi^NLS allele produces a protein lacking a nuclear localization signal (NLS) and therefore cannot direct Rhino to piRNA cluster chromatin (Klenov et al. 2011), and piwi² is a null allele. However, Rhino foci were present at 2 h post-heat shock and appeared to have fully recovered by 4 h post-heat shock in piwi^2/NLS mutant ovaries (Fig. 5A). In w¹ controls, in contrast, Rhino foci are first detected at 4 h postheat shock (Fig. 5B). To quantify recovery, we determined the number of Rhino foci with signal 5 SD above average in confocal z-section images of multiple nuclei at each time point. In w¹, computationally defined Rhino foci were first observed at 4 h post-heat shock, and the number of foci had not fully recovered at 8 h. In piwi^2/NLS mutants, numerous Rhino foci are detectable at 1 h post-heat shock, and localization has fully recovered by 4 h. (Fig. 5D). The piwi mutations result in a depletion of Piwi in the nucleus, and thus, the accelerated recovery of Rhino localization implies that the presence of Piwi slows Rhino localization.

FIGURE 5.

piwi^2/NLS mutants accelerate Rhino foci recovery, and thoc7^d/Df mutants delay recovery. (A,B) Confocal images of egg chambers in piwi^2/NLS(A) and w¹ (B) flies before heat shock, and 1, 2, and 4 h after a 1 h 37°C heat shock stained for Rhino. Nurse cell nuclei were identified using DAPI staining and outlined in white. Rhino foci are seen 4 h after heat shock in w¹, while piwi^2/NLS show Rhino foci 2 h after heat shock and numerous foci at 4 h post-heat shock. Scale bar, 10 μm. (C) Confocal images of egg chambers stained for Rhino in thoc7^d/Df mutants before heat shock, and 1, 8, and 24 h after a 1 h 37°C heat shock. Nurse cell nuclei were identified using DAPI staining and outlined in white. Eight hours after heat shock, many nuclei show no Rhino foci. Even 24 h after heat shock, some nuclei show few Rhino foci. Scale bar, 10 μm. (D) Quantification of the number of Rhino foci in w¹, piwi^2/NLS, and thoc7^d/Df mutants after a 37°C heat shock for 1 h with the indicated recovery time and no heat shock controls. Minimally, six nurse cells were imaged for each condition. Individual nurse cell nucleus was identified with DAPI staining and optical sections with 1 μm step size. Rhino foci were then identified in each z-section as 5 SD above background and counted. Each dot represents a single z-section. P-values compared to the no shock control are recorded above each bar, and not significant is marked as (n.s.).

piRNAs bound to Piwi function with Panoramix in piRNA-guided transcriptional silencing of transposons (Sienski et al. 2015; Yu et al. 2015). To determine if the enhanced recovery of Rhino localization in piwi mutants is linked to transposon activation, we assayed Rhino recovery from heat shock in panx^M4/dDf mutant ovaries. panx^M4 allele contains a 4 bp deletion resulting in a frameshift mutation in panx gene, and panx^dDf is a large deletion that eliminates the entire gene (Yu et al. 2015). In this genetic background, Rhino recovery was comparable to w¹ controls (Supplemental Fig. S5A). Piwi thus delays reassembly of cluster chromatin following heat shock, through a process independent of transposon silencing.

The TREX complex localizes to clusters, binds cluster transcripts, and has been implicated in suppressing ectopic Rhino localization outside of piRNA clusters (Zhang et al. 2018). To determine the impact of TREX on cluster chromatin assembly, we assayed recovery of Rhino foci after heat shock in thoc7^d/Df and uap56^sz/28 mutants. The thoc7 gene encodes a component of the heteropentameric THO complex, which interacts with UAP56 to form the TREX complex. However, THO and UAP56 also have TREX-independent functions (Rehwinkel et al. 2004; Hur and Chung 2016; Hur et al. 2016). In thoc7^d/Df mutants, which disrupt THO complex formation, some Rhino foci are present 1 h after heat shock; however, the number of Rhino foci in each z-section did not begin to increase until 8 h of recovery, and the number of Rhino foci were significantly lower than in no heat shock controls at 24 h after heat shock (Fig. 5C). The thoc7^d/Df mutation thus delays Rhino recovery. In contrast, recovery of Rhino foci is comparable to w¹ in uap56^sz/28 mutants (Supplemental Fig. S5B). uap56^sz/28 allelic combination is a hypomorph and known to decrease piRNA production and significantly reduce UAP56 interaction with the THO complex (Zhang et al. 2012, 2018). These findings suggest that the THO complex has a TREX-independent role in promoting Rhino localization.

Maternal piRNAs are not required to license piRNA cluster chromatin

Rapid recovery of Rhino localization in piwi mutants implies that Piwi and piRNAs are not required for cluster chromatin assembly in adult ovaries. We therefore sought to reevaluate the role of piRNAs in cluster chromatin assembly during early embryogenesis. Gebert et al. (2021) found that deletion of the dominant germline piRNA cluster at 42AB does not disrupt transposon silencing or female fertility, despite removing ∼30% of all piRNAs. Females homozygous for this deletion thus produce viable embryos that lack maternal piRNAs uniquely mapping to 42AB. To directly determine the role of maternal piRNAs in cluster chromatin assembly and piRNA biogenesis, we generated embryos lacking maternal piRNAs mapping to 42AB and carrying a wild-type copy of the cluster by crossing homozygous 42AB Df females to wild-type OreR males. The reciprocal cross generated genetically identical embryos with maternally deposited 42AB piRNAs (Fig. 6A). We assayed Rhino localization by CUT&RUN and piRNA expression by small RNA-seq in the adults that developed from these embryos.

FIGURE 6.

Maternal 42AB piRNAs are not required for licensing Rhino binding to and piRNA production from 42AB cluster. (A) Experimental crosses: OreR females were crossed to 42AB Df males to produce F1 +/42AB Df heterozygotes that inherit maternal 42AB piRNAs. The reciprocal cross with 42AB Df females crossed to OreR males generates a genetically identical F1 population; however, they do not inherit maternal 42AB piRNAs. (B) CUT&RUN signal tracks at 42AB piRNA cluster for IgG control and anti-Rhino in F1 females with and without maternal piRNA. Graphs show similar enrichment of Rhino at 42AB piRNA cluster in both F1 populations. (C) Uniquely mapping small RNA signal tracks against 42AB piRNA cluster for OreR, 42AB Df, and the two F1s with and without maternal 42AB piRNAs. No detectable differences between the two F1 populations were observed in three replicates. (D) Example scatter plot comparing expression (rpkm log₁₀) of mapping unique cluster piRNAs between two F1s shows no global changes. Dotted line is twofold change. 42AB piRNA cluster is highlighted in red. (E) Example scatter plot comparing expression (rpkm log₁₀) of all transposon mapping piRNAs between two F1s also shows no global changes. Dotted line is twofold change.

Figure 6B shows genome browser profiles of Rhino binding across 42AB in the F1 progeny from the genetic crosses above. Remarkably, Rhino enrichment at the 42AB cluster is nearly identical in ovaries from both sets of adults. Additionally, levels of piRNAs produced from the 42AB cluster also appear to be identical in both F1 progeny (Fig. 6C). Quantification of cluster and transposon mapping piRNAs confirms these qualitative observations (Fig. 6D,E). To control for genetic background effects, we performed a parallel cross using w¹ as the control strain. Again, piRNA production and Rhino binding were identical in both sets of F1 progeny, independent of the maternal deposition of 42AB piRNAs (Supplemental Fig. S6A–E). We speculated that maternally deposited piRNAs shared by 42AB and other clusters could license cluster chromatin. The αγ genomic region shares transposable elements with 42AB, and the 42AB Df does not eliminate these multimapping piRNAs. However, Rhino does not bind to this region, with and without complementary maternal piRNAs (Supplemental Fig. S6F), strongly suggesting that maternal piRNAs are not required to license cluster chromatin assembly.

DISCUSSION

Environmental stress leads to transposon activation in evolutionarily diverse systems and may generate genetic diversity that facilitates adaptive evolution (McClintock 1984; Grandbastien et al. 1997; Van Meter et al. 2014; Fouche et al. 2020). To drive evolution, mutations must be transmitted through the germline, and the piRNA pathway silences transposons during germline development. We therefore analyzed the impact of acute heat shock on the organization and function of the Drosophila piRNA pathway.

Heat shock control of cluster chromatin

In D. melanogaster, the HP1 homolog Rhino binds to clusters and anchors the RDC complex, which drives piRNA precursor transcription and TREX-dependent nuclear export (Zhang et al. 2014; Hur and Chung 2016; Andersen et al. 2017). We show that heat shock leads to the rapid loss of RDC localization to nuclear foci, followed by loss of the TREX complex. These cytological changes are associated with a decrease in cluster transcript accumulation from both strands of germline clusters, consistent with Rhino function in driving cluster transcription. Expression from clusters progressively increases during recovery from heat shock, with the concurrent restoration of Rhino and Cuff, then followed by TREX localization. While wild-type Rhino binding to clusters, assayed by CUT&RUN, is not re-established after transient heat shock (Fig. 2), we also find that transient heat shock has no impact on fertility (Supplemental Fig. S7). These findings suggest that functional RDC complexes assemble during recovery from heat shock (Figs. 1–3).

Rhino is known to bind to H3K9me3 histone marks (Klattenhoff et al. 2009; Mohn et al. 2014); however, recent studies suggest that zinc finger protein Kipferl and H3K27me3 marks also contribute to Rhino localization (Baumgartner et al. 2022; Akkouche et al. 2025). Although we do not observe changes in H3K9me3 or H3K27me3 levels at piRNA clusters after heat shock (Supplemental Fig. S2A–D), whether heat shock affects Kipferl localization remains unknown. Examining how heat shock influences Kipferl dynamics and how its recovery kinetics compare with Rhino could reveal additional mechanisms regulating protein assembly at piRNA clusters. Additionally, Rhino localizes to district chromosomal locations like piRNA clusters or transposons. Although we see global disruption of Rhino location after heat shock, it is possible that Rhino recovery to these different genomic locations follows different kinetics and warrants further investigation.

Heat shock control of the cytoplasmic piRNA biogenesis machinery

Following nuclear export, piRNA precursors undergo ping-pong and phased piRNA processing. The PIWI proteins Aub and Ago3, with the conserved DEAD box protein Vasa, are required for ping-pong amplification, and all three proteins localize to perinuclear nuage. Phased biogenesis requires Ago3, the conserved helicase Armi, and the nuclease Zuc. Ago3 localizes to nuage, Armi localizes to nuage and mitochondria, and Zuc associates with the mitochondrial membrane. Upon heat shock, Vasa rapidly localizes to the nucleus and is reduced at nuage, and Aub is displaced from nuage and localizes to granular material that lies adjacent to Zuc (Fig. 4). Significantly, CRM-1 is required for nuclear export of piRNA precursors, Vasa binds to these transcripts, and we show that Vasa localizes to the nucleus in ovaries treated with the inhibitor leptomycin-B (ElMaghraby et al. 2019; Kneuss et al. 2019; Mendel and Pillai 2019). We therefore propose that Vasa cycles between the nucleus and nuage and may directly facilitate nuclear export to piRNA precursors. The functional significance of Aub localization to structures near Zuc is unclear, as aub mutations do not disrupt phased biogenesis. However, Vasa is required for Aub localization to nuage, and reduced Aub localization to nuage on heat shock could be secondary to reduced Vasa localization.

Two-phase response to heat shock?

RDC mutations disrupt cluster transcription and lead to a profound reduction in cluster mapping piRNAs, with broad increases in transposon expression (Mohn et al. 2014). In contrast, loss of the RDC on heat shock reduces cluster transcript levels but does not reduce piRNAs mapping to clusters or transposons, and expression of most transposon families is suppressed. (Fig. 1) Heat shock drives a reduction in gene transcription (Jamrich et al. 1977; Lindquist 1986; Teves and Henikoff 2011), and we speculate that heat shock also suppresses transposon transcription. In contrast, piRNAs are stable following heat shock, and we propose that they guide posttranscriptional transcript cleavage, suppressing transposon transcript accumulation. The existing pool of piRNAs may therefore act as a buffer to allow time for the nuclear machinery to reassemble following acute heat shock. In contrast, prolonged heat stress has been reported to activate transposons and lead to transposition-induced mutations (Fanti et al. 2017; Cappucci et al. 2019). Prolonged disruption of the RDC with longer heat shock treatments may therefore allow time for turnover of stable piRNAs, leading to transposon activation. The result is a two-phase germline response to heat shock that maintains genome integrity in response to transient stress, but allows transposon mobilization and potential generation of genetic diversity on long-term stress, allowing for adaptive evolution under conditions that compromise viability.

Heat shock as a tool

The ability to reversibly disrupt cluster chromatin using heat shock provides a novel tool to probe cluster chromatin assembly. The sequential recovery of Rhino followed by TREX after heat shock is consistent with the proposed model of RDC function in promoting cluster transcription, producing precursors which are then bound by TREX and exported from the nucleus. How the RDC is directed to specific loci, in contrast, is not fully understood. piRNAs mapping to clusters and Piwi protein are maternally deposited in the oocyte, and transient depletion of Piwi during embryogenesis compromises, but does not block, Rhino localization and piRNA production in adults. In contrast, zygotic piwi mutations, which deplete the protein later in development, do not block Rhino localization or piRNA expression from clusters. Together, these observations suggest that maternal piRNAs bound to Piwi promote Rhino localization and cluster chromatin assembly during embryogenesis, and that these heterochromatic domains are then propagated to the adult stage by a Piwi-independent epigenetic mechanism. However, we find that piwi mutations accelerate recovery of Rhino localization after heat shock. Conversely, thoc7 mutations disrupt the TREX complex and delay recovery (Fig. 5). We speculate that Piwi may compete with Thoc, which promotes Rhino recovery; alternatively, Piwi may directly compete with Rhino binding at piRNA cluster chromatin. Regardless, cluster assembly in the adult ovary appears independent of Piwi and may be facilitated by TREX.

These findings led us to reevaluate the role of maternal piRNAs in cluster heterochromatin assembly. A deletion that removes the major germline cluster at 42AB is homozygously viable and fertile, and embryos derived from homozygous females thus lack maternally deposited piRNA mapping specifically to this cluster. Using this deletion, we show that Rhino binding to 42AB and piRNA production from this cluster do not require matching maternal piRNAs. Rhino also fails to localize to genome regions complementary to maternal “multimapping” piRNAs, suggesting that piRNAs are not sufficient to trigger cluster chromatin assembly.

Thoc7 protein is a component of the THO complex which appears to cotranscriptionally bind to cluster transcripts. Mutations in thoc7 delay reassembly of cluster chromatin following heat shock. We therefore speculate that Rhino-independent canonical transcription of clusters, perhaps initiated at transposon promoters, generates transcripts that are recognized by a complex that includes THO, which initiates cluster chromatin assembly. What additional factors function with THO and how these transcripts are differentiated from gene transcripts remain to be determined.

MATERIALS AND METHODS

Fly strains and husbandry

Fly strains used are in Table 1. All flies were maintained at 25°C on cornmeal medium. One to two day old females were collected and fed on yeast for 2 days. Heat shock was conducted on 2–4 day old flies with yeast for 1 h in a water bath kept at 37°C or at the stated temperature and duration. Flies were allowed to recover for 1 h after heat shock at room temperature unless otherwise stated. Ovaries were collected from aged-matched 2–4 day old females for a no heat shock control.

TABLE 1.

Resources

Reagent or resource	Source	Identifier
Antibodies
Guinea pig anti-Rhino—(Immuno-staining 1:500, CUT&RUN 1:50)	Klattenhoff et al. 2009	N/A
Mouse antinuclear pore complex, MAb414 clone—(immuno-staining 1:1000)	Covance MAb414 clone	MMS-120R
Rabbit anti-Deadlock (immuno-staining 1:1000)	Julius Brennecke laboratory	RRID: AB_2568875
Rat IgM anti-Vasa—(Immuno-staining 1:250)	Developmental Studies Hybridoma bank	ID: AB_760351
Rabbit anti-Ago3—(immuno-staining 1:3000)	MAb Technologies, Inc.	Product # Ago3-3
Mouse anti-Squash	Developmental Studies Hybridoma bank	SQUASH 1F3-1B10 AB_10571462
Rabbit anti-Armi-CT (immuno-staining 1:500)	Cook et al. 2004	N/A
Rabbit anti-H3K9me3 (CUT&RUN 1:100)	Abcam	ab8898
Rabbit anti-H3K27me3 (CUT&RUN 1:100)	Cell Signaling Technology	#9733
GFP Booster_ATTO488 (immuno-staining 1:200)	ChromoTek	Cat# gba488, RRID: AB_2631434
RFP-Booster ATTO594 (immuno-staining 1:200)	ChromoTek	Cat# rba594
Wheat germ agglutinin (WGA) Alexa Fluor 647	Thermo Fisher Scientific	Cat# W32466
Antidigoxigenin POD	Roche	Cat# 11207733910
Chemicals, peptides, and recombinant proteins
SuperScript III	Thermo Fisher Scientific	Cat# 18080-085
RNase OUT	Thermo Fisher Scientific	Cat# 10777-019
TURBO DNase	Thermo Fisher Scientific	Cat# AM2238
RNase H	Thermo Fisher Scientific	Cat# 18021-071
T4 RNA Ligase	Thermo Fisher Scientific	Cat# AM2141
dNTP set (100 mM)	Thermo Fisher Scientific	Cat# 10297018
dUTP solution (100 mM)	Thermo Fisher Scientific	Cat# R0133
dNTP mix	NEB	Cat# N0447L
DNA polymerase I	NEB	Cat# M0209S
T4 DNA polymerase	NEB	Cat# M0203L
Klenow DNA polymerase	NEB	Cat# M0210S
T4 PNK	NEB	Cat# M0201L
Klenow 3′ to 5′ exo	NEB	Cat# M0212L
UDG	NEB	Cat# M0280S
Phusion polymerase	NEB	Cat# M0530S
T4 RNA ligase 2, truncated	NEB	Cat# M0242L
50% PEG8000	NEB	Cat# B1004S
T4 DNA ligase	Enzymatics Inc.	Cat# L6030-HC-L
Hybridase Thermostable RNase H	Biosearch Technologies	Cat# H39500
16% Formaldehyde	Ted Pella Inc.	Cat# 18505
Formamide	Millipore Sigma	Cat# F7503
Digitonin	Millipore Sigma	Cat# 300410-250MG
EDTA-free Protease Inhibitor Cocktail (Roche)	Millipore Sigma	Cat# 11873580001
CUT&RUN pAG-MNase and Spike-In DNA	Cell Signaling	Cat# 40366S
RNase A	Thermo Fisher Scientific	Cat# EN0531
Protease K
Grace's Insect Medium, unsupplemented	Thermo Fisher Scientific	Cat# 11595030
Schneider's Drosophila Medium	Thermo Fisher Scientific	Cat# 21720024
Insulin solution from bovine pancreas	Sigma-Aldrich	Cat# I0516-5ML
Critical commercial assays
mirVana miRNA isolation kit	Thermo Fisher Scientific	Cat# AM1560
RNA Clean & Concentrator-5	Zymo Research	Cat# R1015
Agencourt AMPure XP	Beckman Coulter	Cat# A63880
TSA Cyanine 3 System	Akoya Biosciences	Cat# NEL704A001KT
BioMagPlus Concanavalin A	Bangs laboratories	BP531
Deposited data
High-throughput sequencing	This study	GEO: GSE234723
Small RNA-seq for w1 rep1 for cluster definition	Zhang et al. 2018	SRA: SRR7408119
Small RNA-seq for w1 rep2 for cluster definition	Zhang et al. 2021	SRA: SRR10541189
Experimental models: organisms/strains
D. melanogaster/w1	William Theurkauf lab	N/A
D. melanogaster/thoc5 promoter > Thoc5-GFP	Moon et al. 2011	N/A
D. melanogaster: rhiP > GFP-mel-Rhi	Parhad et al. 2017	N/A
D. melanogaster: rhiP > GFP-mel-Cuff	Parhad et al. 2020	N/A
D. melanogaster/uap56 promoter > UAP56Venus	Zhang et al. 2012	N/A
D. melanogaster/P{vasT:GFP}	Styhler et al. 1998	N/A
D. melanogaster: nanos > Gal4	William Theurkauf lab	N/A
D. melanogaster/P[UAS-GFP-aub]	Harris and Macdonald 2001	Bloomington Drosophila Stock Center—Stock# 42219
D. melanogaster/P{Ubi-zuc-RFP}	Vagin et al. 2013	N/A
D. melanogaster/P{GFP-mael.wt}	Sienski et al. 2012	N/A
D. melanogaster/armi 72.1	Cook et al. 2004	N/A
D. melanogaster/piwi 02	Lin and Spradling 1997	Bloomington Drosophila Stock Center—Stock #43319
D. melanogaster/panoramix M4	Yu et al. 2015	N/A
D. melanogaster/Df(2R)BSC821 (panxdDf)	Yu et al. 2015	Bloomington Drosophila Stock Center—Stock# 27582
D. melanogaster/armi 1	Cook et al. 2004	Bloomington Drosophila Stock Center—Stock# 8513
D. melanogaster/armi 72.1	Cook et al. 2004	Bloomington Drosophila Stock Center—Stock# 8544
D. melanogaster/thoc7d (thoc7d05792)	Zhang et al. 2018	Harvard Exelixis Stock Collection—Stock# d05792
D. melanogaster/thoc7Df (DF(3L)BSC128)	Zhang et al. 2018	Bloomington Drosophila Stock Center—Stock# 9293
D. melanogaster/uap56 28	Zhang et al. 2012	N/A
D. melanogaster/uap56 sz15	Zhang et al. 2012	N/A
D. melanogaster/42ABDf1.1 (Df(2R)42ABΔ)	Gebert et al. 2021	N/A
D. melanogaster/Oregon-R	William Theurkauf lab	N/A
Oligonucleotides
Random primers	Thermo Fisher Scientific	Cat# 48190011
Software and algorithms
RStudio		https://www.rstudio.com/
ImageJ		https://imagej.nih.gov/ij/
Inkscape		https://inkscape.org
Gimp		https://www.gimp.org
Leica Application Suite X (LAS X)	Leica Microsystems	https://www.leica-microsystems.com/products/microscope-software/p/leica-las-x-ls/
BEDTools	Langmead et al. 2009
UCSC Genome Browser		https://genome.ucsc.edu/cgi-bin/hgGateway
piSet		https://github.com/tianxiongbb/piSet

Female fertility assays

Assay for measuring female fertility was previously described in Parhad et al. (2020) and modified/conducted in the following manner. Five to six females that were 2–4 days old were subjected to heat shock at 37°C for 1 h or no heat shock treatments. Following treatment, flies were transferred and maintained on grape juice agar plates for 1 day (0–24 h collection) before transferring to a new grape juice agar plate for a second day (24–48 h collection). After removing flies, the total number of eggs were counted. Eggs were incubated at 25°C for 24 h before the number of hatched eggs were counted. The fertility bar graphs show the mean and standard deviation for three biological replicates.

Immunofluorescence

Staining of fixed ovaries was performed as previously described with buffer A protocol in Theurkauf (1994) and Zhang et al. (2012). Briefly, 2–4 day old female ovaries were dissected in Robb's media, fixed in 4% formaldehyde for 10 min, washed, incubated in primary overnight, washed, incubated in fluorophore conjugated secondary, and DAPI for DNA staining overnight if required. If nuclear envelop staining was required, we incubated tissue for 30 min at room temperature with wheat germ agglutinin with a final concentration of 5 ng/μL, washed and mounted on the slide. Primary antibody concentrations used were as follows: guinea pig anti-Rhino 1:500, rat IgM anti-Vasa 1:500, rabbit anti-Ago3 1:3000, mouse anti-Squash 1:30, and GFP and RFP booster at 1:200. See Table 1 for antibody details.

Live ovary dissection was carried out using media described in Prasad et al. (2007) and was mounted directly on a coverslip in halocarbon oil for live imaging. Treatment with leptomycin B was carried out in the media mix in Prasad et al. (2007) at 25°C for the indicated time in a rocking incubator.

Immuno-DNA FISH

Immunofluorescence staining was performed in the same manner as above up until completion of secondary conjugation overnight. After secondary incubation, samples were washed in BAT buffer for 10 min and then fixed in buffer A fixative containing 4% formaldehyde for 10 min as described in the immunofluorescence protocol. Samples were rinsed in 2×SSCT (2× SSC, 0.1% Tween-20) three times to exchange buffering system and subsequently washed at room temperature for 10 min with 2×SSCT + 20% formamide, then 2×SSCT + 40% formamide, and lastly 2×SSCT + 50% formamide to gradually increase formamide concentration. Fresh 2×SSCT + 50% formamide was added to the sample and incubated at 37°C water bath for 2 h to allow for equilibration. For each sample, solution was removed and replaced with 1x Hyb solution (1× SSC, 50% formamide, and 10% [m/v] dextran sulfate) with 0.3 μg probe. The probe to 42AB was against a unique region directly adjacent to 42AB itself as 42AB is repetitive and may bind elsewhere in the genome. Samples were then denatured at 95°C for 5 min, and allowed to hybridize overnight at 37°C. Samples were then washed for 20 min with prewarmed 2×SSCT + 50% formamide at 37°C three times, followed by 10 min washes at room temperature with solutions in the following order: 2×SSCT + 40% formamide, 2×SSCT + 20% formamide, 2×SSCT wash three times, and PBS+0.2% Tween-20 (PBST) wash two times. Anti-DIG POD conjugation (Roche) and tyramide signal amplification (Akoya Biosciences) were performed according to the manufacturer's instructions. DAPI staining was performed simultaneously with Cy3 addition when required, followed by three washes with PBS-0.05% Triton X for 5 min before mounting on slides.

Image acquisition and automated image analysis

All images were taken using a Leica TCS SP8 confocal microscope, and line scans for Figure 4B were produced using Leica Application Suite X. Around 10 females’ ovaries were dissected and pooled for each staining condition and/or treatment. Visual inspection of ovaries was performed, and representative images were chosen for each figure unless otherwise stated. Experimental conditions being compared to controls were always taken under the same imaging conditions, and these conditions were set to maximize the dynamic range in WT control/no heat shock conditions as much as possible while minimizing saturation. In instances where maximum dynamic range was not achieved during image acquisition step, images were adjusted to have the maximum dynamic range for the WT/no heat shock controls and applied to all images in the same experiment. This method allows for direct comparison of localization patterns as well as relative signal intensity. For quantification purposes, minimally, four to six nuclei were imaged for each staining/treatment condition, and 10–25 optical z-sections were acquired for each nucleus. These data sets were used in the automated image quantification pipelines developed to provide an unbiased overview.

To quantify the presence of Rhino adjacent to 42AB foci, we developed an automated and unbiased image analysis pipeline using ImageJ. z-sections of nurse cell nuclei stained for Rhino and 42AB-DNA were taken using confocal microscopy. Signal was defined as 5 SD above the mean signal for each marker. The number of 42AB foci were counted for each z-section and designated as regions of interest using analyze particles. To measure the number of 42AB adjacent to Rhino, we looked for overlap between Rhino signal in each region of interest. Rhino and 42AB signals are usually just shy of colocalization. To ensure that we capture adjacent localization of these two markers, we expanded the Rhino signal equally in all directions and measured overlap with each exaptation. A range of zero to six pixel expansions were tested, and all expansions showed the same trend. Expansion 3 was chosen for the main figure.

For Thoc7-Rhi and Cuff-Rhi colocalization analysis, z-stacks were taken of nurse cell nuclei stained for two different nuclear markers. Images were analyzed in ImageJ, where signal was defined the same way as above. Signals for the two markers were merged and used to pick particles to define region of interest. For every focus identified, the signal intensity for each marker was measured.

To count Rhino foci, z-stacks were acquired of nurse cell nucleus stained for Rhino, and signal was defined the same way as above. Rhino foci for each z-section were defined using analyze particle and subsequently counted.

Small RNA-seq

Total RNA-seq was performed as in Zhang et al. (2021). Briefly mirVana kit (Ambion) was used to extract total RNA from 2–4 day old female ovaries. Approximately 18–30 nt small RNAs were isolated using polyacrylamide gel purification. Sequencing libraries were constructed as follows: 2S rRNA depletion, 3′ adaptor ligation, gel purification, 5′ adaptor ligation, gel purification, reverse transcription, and PCR amplification. Libraries were sequenced using Illumina NextSeq platform, and sequencing analysis mapping to dm6 was performed using piSet_srnaseq pipeline on GitHub. Two to three replicates were performed for each condition.

Long RNA-seq

Total RNA sequencing was performed as described in Zhang et al. (2021). Total RNA was extracted from 2–4 day old female ovaries using the mirVana kit (Ambion). rRNA was depleted using antisense rRNA oligo hybridization with subsequent RNase H digestion. Libraries were made as follows: RNA fragmentation, reverse transcription, dUTP incorporation, end repair, size selection with Ampure XP beads (∼100–500 nt), A-tailing, adaptor ligation, UDG treatment, and PCR amplification. Sequencing was performed using the Illumina NextSeq, and sequencing analysis mapping to dm6 was performed using piSet_rnaseq pipeline on GitHub. Two to three replicates were performed for each condition.

CUT&RUN

CUT&RUN protocol was adapted from Ahmad (2018) (Skene and Henikoff 2017; Ahmad 2018). A brief description and modification are as follows. Ovaries were dissected from 2–4 day old females in Robb's medium (unless otherwise stated), bound to ConA beads, incubated overnight with the indicated primary antibody or anti-IgG control, and subsequently incubated with pAMA. DNA was digested for 30 min to 1 h before reaction was stopped. DNA recovered from supernatant fraction was used to prepare sequencing libraries by end repair, A-tailing, adaptor ligation, and then PCR amplification. Libraries were sequenced using Illumina NextSeq. Sequencing analysis mapping to dm6 was performed using piSet_chipseq pipeline on GitHub. Two to three replicates were performed for each condition. Primary antibodies used were guinea pig anti-rhino (1:50), rabbit anti-H3K9me3 (1:100), rabbit anti-H3K27me (1:100), guinea pig IgG (1:100), and rabbit IgG (1:100). See Table 1 for antibody details.

Bioinformatics analysis

The bioinformatic analysis was previously described in Zhang et al. (2021). The Drosophila reference genome (dm6), rRNA sequences, gene annotations, and hairpin sequences were obtained from FlyBase (version 6.13). Transposon consensus sequences were obtained from Repbase (Bao et al. 2015).

piRNA cluster annotation

piRNA clusters were defined using a similar method as Yu et al. (2019). w¹ control small RNA-seq data from Zhang et al. (2021) was used to annotate piRNA clusters on dm6 (see Table 1). Data mapping to rRNA, miRNA, tRNA, snRNA, and snoRNA were removed, and the remaining 24–32 nt small RNA reads that mapped to the dm6 genome were considered piRNAs. These piRNAs were then assigned to 20 kb sliding windows (with a 1 kb step). Windows containing more than 100 piRNAs per million uniquely mapped piRNAs were considered potential piRNA clusters. Twenty kilobase genomic windows containing <200 distinct reads or species were also filtered out to remove false positives resulting from unannotated miRNA, rRNA, tRNA, snRNA, and snoRNA that mostly produce reads of the same sequence. Additionally, we calculated the first-nucleotide content for each 20 kb window and discarded those with 1U/10A percentage <50%. The remaining contiguous 20 kb windows were deemed putative piRNA clusters. Lastly, we performed manual curation for putative piRNA clusters using piRNA profile. piRNA clusters were classified as uni-strand and dual-strand piRNA clusters based on the direction of the piRNAs produced.

RNA-seq analysis

piSet_rnaseq pipeline from GitHub was used to analyze long RNA sequence reads. Raw reads were mapped to rRNA sequence using Bowtie2 (version 2.2.5) with default settings (Langmead and Salzberg 2012). Remaining reads were mapped to the Drosophila genome (dm6) and transposon consensus sequences using STAR (version 2.5.2b) and HISAT2 with default parameters (Dobin et al. 2013; Kim et al. 2015). The transcript abundance was counted for each gene, transposon, and cluster (reads per kilobase per million mapped reads [RPKM]) using BEDTools (version 2.27.1) (Quinlan and Hall 2010) and normalized to the total number of genome mapping reads excluding rRNA mapping reads. DESeq2 was used to evaluate differential gene expression (Love et al. 2014).

Small RNA-seq analysis

piSet_srnaseq was used to analyze small RNA sequencing data. Cutadapt (Version 1.15) (Martin 2011) was used to remove 3′ end adaptor. The raw small RNA-seq reads were first mapped to rRNA, miRNA hairpin, snoRNA, snRNA, and tRNA sequence allowing for no mismatches using Bowtie (version 1.1.0) (Langmead et al. 2009). The remaining reads were then mapped to the Drosophila genome (dm6) and transposon consensus sequences. Small RNA abundance was normalized to uniquely mapping reads. For ping-pong analysis, the 5′ to 5′ overlaps between piRNAs mapping to opposite genomic strands were calculated, and Z-score for 10 nt overlap was calculated by using 1–9 and 11–30 nt overlaps as background (Li et al. 2009).

CUT&RUN-seq analysis

CUT&RUN sequencing analysis was performed using piSet_chipseq pipeline on GitHub. Briefly, Bowtie2 (version 2.2.5) with default settings was used to map sequencing reads to the Drosophila genome (dm6) and transposon consensus sequence (Langmead and Salzberg 2012). Signal mapping to transposons was normalized to the total number of genome mapping CUT&RUN reads.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.