The Drosophila ZAD zinc finger protein Kipferl guides Rhino to piRNA clusters

  1. Lisa Baumgartner
  2. Dominik Handler
  3. Sebastian Wolfgang Platzer
  4. Changwei Yu
  5. Peter Duchek
  6. Julius Brennecke

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Interactions between transposons and the Drosophila host genome are governed by dedicated H3K9me3-enriched loci that are selected for producing anti-transposon piRNAs through binding by the HP1 variant Rhino in Drosophila. The authors identify Kipferl, a ZAD zinc-finger protein, as helping to guide Rhino to G-rich motifs found at piRNA-producing loci in the female germline. The work thus reveals the involvement of a factor binding specific DNA sequences in piRNA biogenesis. The findings are of broad interest to the fields of heterochromatin and transposon biology.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

RNA interference systems depend on the synthesis of small RNA precursors whose sequences define the target spectrum of these silencing pathways. The Drosophila Heterochromatin Protein 1 (HP1) variant Rhino permits transcription of PIWI-interacting RNA (piRNA) precursors within transposon-rich heterochromatic loci in germline cells. Current models propose that Rhino’s specific chromatin occupancy at piRNA source loci is determined by histone marks and maternally inherited piRNAs, but also imply the existence of other, undiscovered specificity cues. Here, we identify a member of the diverse family of zinc finger associated domain (ZAD)-C 2 H 2 zinc finger proteins, Kipferl, as critical Rhino cofactor in ovaries. By binding to guanosine-rich DNA motifs and interacting with the Rhino chromodomain, Kipferl recruits Rhino to specific loci and stabilizes it on chromatin. In kipferl mutant flies, Rhino is lost from most of its target chromatin loci and instead accumulates on pericentromeric Satellite arrays, resulting in decreased levels of transposon targeting piRNAs and impaired fertility. Our findings reveal that DNA sequence, in addition to the H3K9me3 mark, determines the identity of piRNA source loci and provide insight into how Rhino might be caught in the crossfire of genetic conflicts.

Article activity feed

  1. Version published to 10.7554/elife.80067 on eLife
  2. eLife

    Evaluation Summary:

    Interactions between transposons and the Drosophila host genome are governed by dedicated H3K9me3-enriched loci that are selected for producing anti-transposon piRNAs through binding by the HP1 variant Rhino in Drosophila. The authors identify Kipferl, a ZAD zinc-finger protein, as helping to guide Rhino to G-rich motifs found at piRNA-producing loci in the female germline. The work thus reveals the involvement of a factor binding specific DNA sequences in piRNA biogenesis. The findings are of broad interest to the fields of heterochromatin and transposon biology.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

  3. eLife

    Reviewer #1 (Public Review):

    In Drosophila germline, most piRNA loci use a non-canonical mechanism to transcribe piRNA precursors at the presence of H3K9me3, which depends on an HP1a paralog called Rhino/HP1d that specifically binds piRNA loci. How does Rhino find the right loci to bind? The current model in the field posits that maternally deposited piRNAs provide a specificity cue for Rhino. Now, Baumgartner et al. from Brennecke Group described a novel factor, the ZAD zinc-finger protein CG2678/Kipferl, that appears to provide another key specificity input to a subset of Rhino's chromatin binding, specifically in differentiated female germline (but not in males or stem/progenitor cell types in the female germline). Using genetics, genomics, genome editing, microscopy and biochemical approaches, Baumgartner et al. propose that Kipferl binds a G-rich DNA motif and, at the presence of local H3K9me3, recruits and/or stabilizes the binding of Rhino to these loci and then convert them from transcriptionally inert heterochromatin to piRNA-producing loci. Overall, the text is well written, the figure is clear, and the data is of high quality. With some additional experiments and text edits, this work represents a significant contribution to the field and should attract readers working on piRNA, transposon, satellite DNA, zinc-finger proteins, HP1 and heterochromatin.

    Specific concerns

    1. The genetic hierarchy between Kipferl and Rhino requires further clarification. Authors seem to propose a simple model where Kipferl acts genetically upstream of Rhino. This simple hierarchy is at odds with several observations. First, the center of Kipferl binding generally has less Kipferl binding without Rhino (Fig 5D). In some cases, Kipferl binding is completely gone without Rhino (Fig 7E middle, bottom). The text describes the loss of Kipferl spreading without Rhino but should also mention this reduction/loss in Kipferl binding. The effect of rhino-/- on Kipferl's chromatin binding should be shown along with wildtype level of Kipferl enrichment in Fig 5C for proper comparison. How should readers understand the effect of Rhino on Kipferl? What is the prominent Kipferl domain in rhino-/- in Fig 5B? Second, the broad binding of Kipferl is gone in rhino-/-, does it mean Kipferl requires Rhino to spread? Or, could Rhino (that is recruited by maternally deposited Piwi/piRNA) recruit Kipferl to neighboring sites, which look like a spreading phenomenon? Most importantly, the argument of Kipferl recruiting Rhino should be directly demonstrated by a sufficiency test in addition to the presented evidence of necessity. Could authors tether Kipferl in H3K9me3-decorated regions to see if Rhino is recruited and vice versa? Observations like 42AB in Fig 5E make one wonder if Rhino also recruits Kipferl, so their relationship is not simply Kipferl recruiting or acting upstream of Rhino, as described throughout this manuscript. Clarifying the relationship between Kipferl and Rhino is essential as it is a central claim made.

    2. DNA binding of Kipferl remains putative. Since the 4th zinc-finger is shown to impact Kipferl localization via interaction with Rhino, it remains formally possible that the first three zinc-fingers control Kipferl localization via protein-protein interaction rather than direct DNA binding. Unless direct biochemical evidence of Kipferl binding DNA is available, the DNA binding of Kipferl should be toned down and described as putative and requires further investigation in text.

    3. The relative contribution of maternally deposited piRNAs versus Kipferl in recruiting Rhino is unaddressed. Prior work from multiple groups including Mohn et al. 2014 Cell from the same group of this manuscript suggested a role of maternally deposited piRNAs in determining a subset of H3K9me3 domains as Rhino binding sites. Is Kipferl or maternally deposited piRNA a better predictor of Rhino binding? This manuscript proposes that Kipferl binds a simple G-rich motif and at the presence of H3K9me3 recruits Rhino binding. The readers are left wondering where maternally deposited piRNAs fit in the model of Rhino recruitment, which should be tested or discussed in text, as maternally deposited piRNA is seen as the key determinant of Rhino binding before this work.

  4. eLife

    Reviewer #2 (Public Review):

    This work seeks to understand how Rhino protein binds piRNA-producing loci in the ovarian genome to initiate the transcription of piRNA precursors in Drosophila. From yeast two hybrid screens, the authors find CG2678, named as Kipferl from this study, interacts with Rhino. ChIP-Seq experiments show that Kipferl and Rhi bind similar genomic regions--piRNA clusters. Mutating Kipferl leads to the loss of Rhino binding for many, but not all, piRNA clusters. Correspondingly, the piRNA production from fly ovaries is distorted, a few transposon families are activated, and animals drop their fertility. Mechanistically, the authors find that Kipferl binds with guanosine-rich DNA motifs and interacts with the chromodomain of Rhino. In summary, the authors propose that Rhino finds its binding sites in the genome via interacting with H3K9me3 and Kipferl.

    In piRNA field, one fundamental question is how the piRNA-producing loci--piRNA clusters--are determined in the genome. So far, the Drosophila ovary has been the leading system to answer this question. Although each species likely employs different mechanisms to drive piRNA production, a detailed study from Drosophila could provide a key piece of knowledge on understanding how piRNA system is evolved to provide the maximum protection to the host genome. Prior work has identified Rhino, an HP1 family protein, binds with piRNA clusters to initiate the transcription of piRNA precursors. The current model proposes that Rhino is recruited to piRNA clusters by H3K9me3. However, Rhino only binds with partial of H3K9me3 regions in the genome, suggesting additional mechanisms are required to recruit Rhino to piRNA clusters. This story identifies a key DNA-binding factor, Kipferl, as essential for recruiting Rhino to the most of, but not all, piRNA clusters. Although other mechanisms still likely exist to fully explain how Rhino is recruited, this solid story moves us one step further to get the full answer.

  5. eLife

    Reviewer #3 (Public Review):

    In this manuscript, Baumgartner et al investigated how cells control Rhino specific deposition on only a subset of the H3K9me3 chromatin domains to specify piRNA source loci. They identified a previously unknown protein, Kipferl, which by interacting with the chromodomain of Rhino guides and stabilizes its specific recruitment to selected piRNA source loci. Kipferl would be preferentially recruited to Guanine-rich DNA motifs. They show that in Kipferl mutant flies, Rhino nuclear subcellular localization and Rhino's chromatin occupancy changes dramatically. Then, they dissect all the domains of the Kipferl protein and show that the Rhino- and DNA-binding activities can be separated and that the 4th ZnF of Kipferl is required to interact with Rhino.

    It is a very elegant genetic work (CRISPR-edited, rescue, KD, overexpression fly lines). In addition, the authors used a combination of yeast two hybrid screen, ChIP, small-RNA-seq and imaging to dissect the function of this new protein. The data in this paper are compelling. Some conclusions might be more moderate. Even if the effect of Kipfler on 80F (Rhino binding, piRNA production) is very obvious, this study also clearly demonstrates that other protagonists are required for the specific binding of Rhino to other piRNA source loci (including 42AB and 38C).

    - Is Kipferl expressed early during oogenesis development? If Kipferl starts to be expressed only after the GSCs and cystoblast stage, Kipferl is probably not required to determine the specification of piRNA source loci identity but probably more for the maintenance of the specification. Could the authors discuss or comment on that?

    - To perform most of their ChIP-seq analysis, the authors have divided the genome into pericentromeric heterochromatin and euchromatin based on H3K9me3 ChIP-seq data performed on ovaries. With this classification the 42AB (2R:6,256,844-6,499,214) and the 38C (2L:20148259-20227581) piRNA clusters known to be heterochromatic fall in the euchromatic part of the genome. Was there a problem with the annotation?

    - Some regions exist in euchromatin that are strongly enriched in Rhino, in Kipferl and in H3K9me3 but are not producing piRNA. Does this type of region exist in heterochromatin?

    - Kipferl has been identified to interact with Rhino by a yeast two-hybrid screen (Figure 2). A co-IP which is the classical method for confirming the occurrence of this intracellular Rhino-Kipferl interaction should be provided.

    - Rhino is known to homodimerize and it has been reported that this homodimerization is important for its binding to H3K9me3 (Yu et al, Cell Res 2015). It is surprising not to find Rhino among the interactors that were picked up from the screen. Do the authors have any explanations or at least comments on these results?

    - In Kip mutants, the delocalization of Rhino to a very large structure at the nuclear periphery is a very clear phenotype (Figure 3). All the very elegant genetic controls are provided. This particular localization of Rhino is correlated with an increase in 1.688 Satellite expression and a colocalization of Rhino and the 1.688 RNAs in the nucleus. The authors propose that this increase is consistent with an elevated Rhino occupancy at 1.688 satellites. The authors should moderate their statements in the light of the results of ChIP experiments. Rhino is maintained on these loci in Kip mutants but an increase is not very clearly observed. Couldn't it be the RNA and not the DNA of this 1.688 region traps Rhino? The same in situ experiment should be performed after an RNAse treatment. The delocalization of Rhino is lost in the Kipferl, nxf3 double mutant flies. What is the chromosomal Rhino distribution in this context? Is the increase in nascent transcripts of 1.688 satellites lost?

    - The level of some Rhino dependent germline TE piRNAs is affected in Kipferl GLKD. Is there a direct correlation between TEs which lost piRNAs and those for which the level of transcripts increases (Diver, 3S18, Chimpo, HMS Beagle, flea, hobo) ?

    - Figure 5E, it seems that Kipferl binding is also dependent on Rhino. All the presented loci have much less binding of Kip in Rhino -/- (The scale for the 42AB locus should be the same between the Rhino -/- and the control MTD w-sh). In addition, the distribution of Rhino in the Kipferl-sh on the 42AB is maintained but seems to be different. Could the authors discuss these points?

    - It is not clear why the authors focus only on Kipferl binding sites in a Rhino mutant in the Figure 5D? Even if the authors mention in the text that "Kipferl binding sites in Rhino mutants ... often coincided with regions bound by Kipferl and Rhino in wildtype ovaries" it should be added the same analysis presented in figure 5D centered on Kipferl peaks detected in ChIP experiments in WT condition in the different genotypes.

    - There is a discrepancy between the results found Figure 3A and Supp figure 3B. In the Rhino mutant the level of Kipferl protein does not seem to be affected whereas in the Rhino GLKD, there is a strong decrease of Kipferl protein. The authors completely elude this point.

    - Comparing the figure 5E and the figure 6G presenting both the 80F piRNA cluster, depending of the scale and the control line that was chosen to illustrate the results we can draw different conclusions. In the figure 5E we can conclude that le level of Kipferl decreases on the 80F locus in Rhino (-/-) compared to the control MTD w-sh, whereas in the figure 6G we can conclude that the level of Kipferl is similar in the Rhino (-/-) compared to the control w1118.

    - gypsy8 or RT1b are enriched in GRGG motifs and are also the ones that among Rhino-independent Kipferl enrichment are the most Rhino enriched. Are these 2 elements present in the 80F cluster? Are these two elements derepressed upon Kipferl GLKD ? Where are these two elements in the figure presenting the change in TE transcript level upon Kipferl GLKD?

  6. eLife

    Author Response

    Reviewer 1

    In Drosophila germline, most piRNA loci use a non-canonical mechanism to transcribe piRNA precursors at the presence of H3K9me3, which depends on an HP1a paralog called Rhino/HP1d that specifically binds piRNA loci. How does Rhino find the right loci to bind? The current model in the field posits that maternally deposited piRNAs provide a specificity cue for Rhino. Now, Baumgartner et al. from Brennecke Group described a novel factor, the ZAD zinc-finger protein CG2678/Kipferl, that appears to provide another key specificity input to a subset of Rhino's chromatin binding, specifically in differentiated female germline (but not in males or stem/progenitor cell types in the female germline). Using genetics, genomics, genome editing, microscopy and biochemical approaches, Baumgartner et al. propose that Kipferl binds a G-rich DNA motif and, at the presence of local H3K9me3, recruits and/or stabilizes the binding of Rhino to these loci and then convert them from transcriptionally inert heterochromatin to piRNA-producing loci. Overall, the text is well written, the figure is clear, and the data is of high quality. With some additional experiments and text edits, this work represents a significant contribution to the field and should attract readers working on piRNA, transposon, satellite DNA, zinc-finger proteins, HP1 and heterochromatin.

    Specific concerns

    1. The genetic hierarchy between Kipferl and Rhino requires further clarification. Authors seem to propose a simple model where Kipferl acts genetically upstream of Rhino. This simple hierarchy is at odds with several observations. First, the center of Kipferl binding generally has less Kipferl binding without Rhino (Fig 5D). In some cases, Kipferl binding is completely gone without Rhino (Fig 7E middle, bottom). The text describes the loss of Kipferl spreading without Rhino but should also mention this reduction/loss in Kipferl binding. The effect of rhino-/- on Kipferl's chromatin binding should be shown along with wildtype level of Kipferl enrichment in Fig 5C for proper comparison. How should readers understand the effect of Rhino on Kipferl? What is the prominent Kipferl domain in rhino-/- in Fig 5B? Second, the broad binding of Kipferl is gone in rhino-/-, does it mean Kipferl requires Rhino to spread? Or, could Rhino (that is recruited by maternally deposited Piwi/piRNA) recruit Kipferl to neighboring sites, which look like a spreading phenomenon? Most importantly, the argument of Kipferl recruiting Rhino should be directly demonstrated by a sufficiency test in addition to the presented evidence of necessity. Could authors tether Kipferl in H3K9me3decorated regions to see if Rhino is recruited and vice versa? Observations like 42AB in Fig 5E make one wonder if Rhino also recruits Kipferl, so their relationship is not simply Kipferl recruiting or acting upstream of Rhino, as described throughout this manuscript. Clarifying the relationship between Kipferl and Rhino is essential as it is a central claim made.

    The relationship between Kipferl and Rhino is indeed complex and we agree that the linear hierarchy as stated in the first submission is too simplistic. We therefore added a clear statement that loss of Rhino impairs spreading as well as stability/strength of the Kipferlchromatin interaction (text relating to Figure 5, second paragraph, Figure 5C,D) . We furthermore edited relevant passages in the text to clarify the points raised by this reviewer. We added an analysis of Rhino/Kipferl domains and the binding of Rhino or Kipferl in kipferl mutants and rhino mutants, respectively (panel 5C). This strengthens the conclusion that Rhino and Kipferl are co-dependent at many sites. Together with the previous analysis that focuses on Kipferl peaks in rhino mutants (now panel 5E), we conclude that Kipferl does bind many Rhino domains by itself, albeit at considerably lower levels and in less broad peaks. We do not know what the prominent Kipferl accumulation observed in immuno fluorescence in rhino mutants corresponds to. The suggested tethering experiment is an interesting suggestion that we consider part of a follow-up study that also aims at understanding the exact molecular and structural basis of the Rhino-Kipferl interaction, ideally in complex with DNA.

    1. DNA binding of Kipferl remains putative. Since the 4th zinc-finger is shown to impact Kipferl localization via interaction with Rhino, it remains formally possible that the first three zinc-fingers control Kipferl localization via protein-protein interaction rather than direct DNA binding. Unless direct biochemical evidence of Kipferl binding DNA is available, the DNA binding of Kipferl should be toned down and described as putative and requires further investigation in text.

    We agree that definitive statements addressing this question require biochemical or structural insight into Kipferl-DNA interactions. We therefore made text changes to reflect this throughout the text relating to Figures 5 and 6.

    1. The relative contribution of maternally deposited piRNAs versus Kipferl in recruiting Rhino is unaddressed. Prior work from multiple groups including Mohn et al. 2014 Cell from the same group of this manuscript suggested a role of maternally deposited piRNAs in determining a subset of H3K9me3 domains as Rhino binding sites. Is Kipferl or maternally deposited piRNA a better predictor of Rhino binding? This manuscript proposes that Kipferl binds a simple G-rich motif and at the presence of H3K9me3 recruits Rhino binding. The readers are left wondering where maternally deposited piRNAs fit in the model of Rhino recruitment, which should be tested or discussed in text, as maternally deposited piRNA is seen as the key determinant of Rhino binding before this work.

    At this point, we cannot firmly separate the role of maternal piRNAs (which would act early in embryogenesis) from a guiding function of Kipferl, whose function during early embryogenesis is unclear (e.g. we see strongly reduced levels of Kipferl in germline stem cells). The current data in the field, together with the new Kipferl findings, indicate that Rhino requires H3K9me2/3 and an additional specificity factor/determinant for stable chromatin binding in the differentiating female germline. While maternal piRNAs might be essential to provide locus-specific H3K9-methylation, Kipferl has the capacity to alter the Rhino profile considerably at sites where H3K9me2/3 co-occurs with Kipferl recruitment sites to chromatin (presumably DNA motifs). Together the two pathways might act in parallel to explain why certain transposon insertions are bound by Rhino, while others are not. We aimed to clarify our view on this important topic in the revised Discussion section (second paragraph).

    Reviewer 3

    In this manuscript, Baumgartner et al investigated how cells control Rhino specific deposition on only a subset of the H3K9me3 chromatin domains to specify piRNA source loci. They identified a previously unknown protein, Kipferl, which by interacting with the chromodomain of Rhino guides and stabilizes its specific recruitment to selected piRNA source loci. Kipferl would be preferentially recruited to Guanine-rich DNA motifs. They show that in Kipferl mutant flies, Rhino nuclear subcellular localization and Rhino's chromatin occupancy changes dramatically. Then, they dissect all the domains of the Kipferl protein and show that the Rhino- and DNA-binding activities can be separated and that the 4th ZnF of Kipferl is required to interact with Rhino.

    It is a very elegant genetic work (CRISPR-edited, rescue, KD, overexpression fly lines). In addition, the authors used a combination of yeast two hybrid screen, ChIP, small-RNA-seq and imaging to dissect the function of this new protein. The data in this paper are compelling. Some conclusions might be more moderate. Even if the effect of Kipfler on 80F (Rhino binding, piRNA production) is very obvious, this study also clearly demonstrates that other protagonists are required for the specific binding of Rhino to other piRNA source loci (including 42AB and 38C).

    • Is Kipferl expressed early during oogenesis development? If Kipferl starts to be expressed only after the GSCs and cystoblast stage, Kipferl is probably not required to determine the specification of piRNA source loci identity but probably more for the maintenance of the specification. Could the authors discuss or comment on that?

    According to our image in Fig. 2D, Kipferl is very weakly expressed in GSCs and early cystoblasts. This is also supported by our unpublished observations on a cultured germline stem cell line (see above), where Kipferl is not detectable on chromatin by ChIP-seq. In these cells, Rhino has a remarkably different chromatin occupancy. Also in testes where Kipferl is not expressed, a different Rhino pattern was observed (Aravin laboratory) despite males inheriting the same complement of maternally deposited piRNAs. Together these data are consistent with a model where Kipferl acts as a specificity factor at its binding sites. We agree, however, that several Rhino domains exist where Kipferl does not show pronounced binding without Rhino. At these sites, Kipferl might act as a stabilizer or maintenance factor for Rhino, as it is nevertheless required for stable Rhino binding. In agreement with findings from testes (Aravin lab), we argue that Rhino’s chromatin occupancy in ovaries is not stable across developmental stages. And that it is respecified upon cystoblast differentiation, at least in part, by Kipferl. We also addressed this central point above (general comment #4 and 5).

    • To perform most of their ChIP-seq analysis, the authors have divided the genome into pericentromeric heterochromatin and euchromatin based on H3K9me3 ChIP-seq data performed on ovaries. With this classification the 42AB (2R:6,256,844-6,499,214) and the 38C (2L:20148259-20227581) piRNA clusters known to be heterochromatic fall in the euchromatic part of the genome. Was there a problem with the annotation?

    As stated in the text, clusters 42AB, 38C, and 80F were analyzed separately (as reference loci) and therefore were not included in either euchromatin or heterochromatin. The reviewer is correct that in the heatmaps, these clusters fall into the euchromatic compartment, as the classification into heterochromatin was not performed based on the presence or absence of the H3K9me3 chromatin mark at any given locus, but defined as inclusion in the continuous body of pericentromeric heterochromatin, which ends 400 kb upstream of 42AB and 2,000 kb downstream of 38C. We added a respective comment to the methods section (lines 945946).

    • Some regions exist in euchromatin that are strongly enriched in Rhino, in Kipferl and in H3K9me3 but are not producing piRNA. Does this type of region exist in heterochromatin?

    In euchromatin, roughly 80% of all Rhino-bound 1kb-tiles produce less than 10 piRNAs per kb per 1 million sequenced miRNAs. In heterochromatin, this is the case for only 10% of Rhino-bound tiles. This difference is likely caused by the high density of transposon fragments within heterochromatin, which allow the initiation of piRNA production from Rhino/Moonshiner-dependent transcripts through triggering.

    • Kipferl has been identified to interact with Rhino by a yeast two-hybrid screen (Figure 2). A co-IP which is the classical method for confirming the occurrence of this intracellular RhinoKipferl interaction should be provided.

    See our response to main comment #1.

    • Rhino is known to homodimerize and it has been reported that this homodimerization is important for its binding to H3K9me3 (Yu et al, Cell Res 2015). It is surprising not to find Rhino among the interactors that were picked up from the screen. Do the authors have any explanations or at least comments on these results?

    We can only speculate as to why Rhino was not identified in the Y2H screen. We are able to detect the homodimerization of Rhino in dedicated yeast two hybrid experiments in the lab, although the interaction was weak. One potential explanation is that dimerization of bait and prey is in competition with dimerization of bait and bait or prey and prey, reducing the efficiency of bait recruitment. For our yeast two hybrid experiments in the lab we use the Gal-4 system, while the screen was based on the more stringent LexA system, for which the homodimerization of Rhino might be too weak to be detected.

    • In Kip mutants, the delocalization of Rhino to a very large structure at the nuclear periphery is a very clear phenotype (Figure 3). All the very elegant genetic controls are provided. This particular localization of Rhino is correlated with an increase in 1.688 Satellite expression and a colocalization of Rhino and the 1.688 RNAs in the nucleus. The authors propose that this increase is consistent with an elevated Rhino occupancy at 1.688 satellites. The authors should moderate their statements in the light of the results of ChIP experiments. Rhino is maintained on these loci in Kip mutants but an increase is not very clearly observed. Couldn't it be the RNA and not the DNA of this 1.688 region traps Rhino? The same in situ experiment should be performed after an RNAse treatment. The delocalization of Rhino is lost in the Kipferl, nxf3 double mutant flies. What is the chromosomal Rhino distribution in this context? Is the increase in nascent transcripts of 1.688 satellites lost?

    The suggestion that RNA might trap Rhino at Satellite loci is a very interesting point. We performed the suggested RNAse treatment experiment in ovaries. This did lead to the disappearance of Rhino foci, however this is the case for both wildtype and kipferl depleted ovaries. While this indeed might point towards a role of RNA in stabilizing Rhino at chromatin, more in-depth experiments are required to clarify this.

    Regarding the Nxf3 point: we clarified this in the revised text (lines 293-296): In nxf3/kipf double mutants we still observe strongly increased RNA FISH signal for the Satellite transcripts of 1.688 and Rsp families, which colocalize with GFP-Rhino (Fig. S3H). We therefore assume that Rhino is still associated with the same chromosomal regions as it is in the Kipf mutant. Just the localization at the nuclear envelope is lost.

    • The level of some Rhino dependent germline TE piRNAs is affected in Kipferl GLKD. Is there a direct correlation between TEs which lost piRNAs and those for which the level of transcripts increases (Diver, 3S18, Chimpo, HMS Beagle, flea, hobo) ?

    We added a dedicated statement to this in the revised text: piRNAs antisense to the TEs that are upregulated at the RNA level are strongly reduced, but they are not the only TEs where piRNAs are decreased (lines 340-343).

    • Figure 5E, it seems that Kipferl binding is also dependent on Rhino. All the presented loci have much less binding of Kip in Rhino -/- (The scale for the 42AB locus should be the same between the Rhino -/- and the control MTD w-sh). In addition, the distribution of Rhino in the Kipferl-sh on the 42AB is maintained but seems to be different. Could the authors discuss these points?

    This point has been addressed above (main revision requests).

    • It is not clear why the authors focus only on Kipferl binding sites in a Rhino mutant in the Figure 5D? Even if the authors mention in the text that "Kipferl binding sites in Rhino mutants ... often coincided with regions bound by Kipferl and Rhino in wildtype ovaries" it should be added the same analysis presented in figure 5D centered on Kipferl peaks detected in ChIP experiments in WT condition in the different genotypes.

    We addressed this in the new revised Figure 5 and the corresponding text.

    • There is a discrepancy between the results found Figure 3A and Supp figure 3B. In the Rhino mutant the level of Kipferl protein does not seem to be affected whereas in the Rhino GLKD, there is a strong decrease of Kipferl protein. The authors completely elude this point.

    See our comment to reviewer 1 above.

    • Comparing the figure 5E and the figure 6G presenting both the 80F piRNA cluster, depending of the scale and the control line that was chosen to illustrate the results we can draw different conclusions. In the figure 5E we can conclude that le level of Kipferl decreases on the 80F locus in Rhino (-/-) compared to the control MTD w-sh, whereas in the figure 6G we can conclude that the level of Kipferl is similar in the Rhino (-/-) compared to the control w1118.

    We made a mistake with the axis label for the Kipferl ChIP in w1118 in Fig. 6F (former panel G), which goes up to 800 like for the Rhino ChIP. This has been fixed.

    • gypsy8 or RT1b are enriched in GRGG motifs and are also the ones that among Rhinoindependent Kipferl enrichment are the most Rhino enriched. Are these 2 elements present in the 80F cluster? Are these two elements derepressed upon Kipferl GLKD ? Where are these two elements in the figure presenting the change in TE transcript level upon Kipferl GLKD?

    Both TEs are indeed present in cluster 80F. However, Kipferl loss does not result in their derepression despite piRNA loss. Rt1b/a are also not significantly upregulated in the rhino KD, suggesting that only evolutionarily old copies exist that are not able to reactivate. Gypsy8 is slightly upregulated in a rhino KD, but not in kipf KD. This discrepancy might be due to the difference in developmental timing when the effect of Rhino or Kipferl depletion sets in. Neither element is known to react strongly to any perturbation of the piRNA pathway and they are mostly considered old and inactive.

  11. Version published to 10.1101/2022.05.09.491178v1 on bioRxiv