Establishment of developmental gene silencing by ordered polycomb complex recruitment in early zebrafish embryos
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This manuscript is of broad interest to developmental biologists and those studying transcriptional/epigenetic regulation of cell-specific and housekeeping gene programs. The work demonstrates that Polycomb complexes coordinate the regulation of distinct groups of genes during early embryogenesis, which offers interesting insights into how very early embryos differentially control housekeeping versus specific developmental gene promoters/enhancers. The data are of high quality, and the conclusions are insightful yet measured.
(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 #3 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Vertebrate embryos achieve developmental competency during zygotic genome activation (ZGA) by establishing chromatin states that silence yet poise developmental genes for subsequent lineage-specific activation. Here, we reveal the order of chromatin states in establishing developmental gene poising in preZGA zebrafish embryos. Poising is established at promoters and enhancers that initially contain open/permissive chromatin with ‘Placeholder’ nucleosomes (bearing H2A.Z, H3K4me1, and H3K27ac), and DNA hypomethylation. Silencing is initiated by the recruitment of polycomb repressive complex 1 (PRC1), and H2Aub1 deposition by catalytic Rnf2 during preZGA and ZGA stages. During postZGA, H2Aub1 enables Aebp2-containing PRC2 recruitment and H3K27me3 deposition. Notably, preventing H2Aub1 (via Rnf2 inhibition) eliminates recruitment of Aebp2-PRC2 and H3K27me3, and elicits transcriptional upregulation of certain developmental genes during ZGA. However, upregulation is independent of H3K27me3 – establishing H2Aub1 as the critical silencing modification at ZGA. Taken together, we reveal the logic and mechanism for establishing poised/silent developmental genes in early vertebrate embryos.
Article activity feed
-
-
Author Response:
Reviewer #1 (Public Review):
Hickey et al. studied chromatin landscape changes in early Zebrafish embryos at three distinct stages: preZGA, ZGA and postZGA. Using ChIP-seq on these time-course samples, they examined developmental genes at their regulatory elements, including promoters and enhancers, that carry nucleosomes enriched with histone variant H2A.Z, as well as post-translational modifications H3K4me1 and H3K27ac, but with low DNA methylation, in early-stage embryos prior to turning on zygotic gene expression. During embryogenesis, this group of elements recruit a Polycomb Repressive Complex 1 (PRC1) component Rnf2 to "write" the ubiquitinated H2A or H2A.Z. The mono-ubH2A/Z then recruits a PRC2 component Aebp2 to further "write" the H3K27me3 repressive mark to silent these developmentally regulated genes in …
Author Response:
Reviewer #1 (Public Review):
Hickey et al. studied chromatin landscape changes in early Zebrafish embryos at three distinct stages: preZGA, ZGA and postZGA. Using ChIP-seq on these time-course samples, they examined developmental genes at their regulatory elements, including promoters and enhancers, that carry nucleosomes enriched with histone variant H2A.Z, as well as post-translational modifications H3K4me1 and H3K27ac, but with low DNA methylation, in early-stage embryos prior to turning on zygotic gene expression. During embryogenesis, this group of elements recruit a Polycomb Repressive Complex 1 (PRC1) component Rnf2 to "write" the ubiquitinated H2A or H2A.Z. The mono-ubH2A/Z then recruits a PRC2 component Aebp2 to further "write" the H3K27me3 repressive mark to silent these developmentally regulated genes in later stage embryos. Using a small molecule to inhibit Rnf2 abolishes H3K27me3 and leads to ectopic gene expression.
Most of the data for the first half of this manuscript are presented in a clear and logic manner. The conclusions based on these correlation assays are quite obvious and well supported (except a few minor points raised below for clarifications, #2-#3). The major concern is for the second half of the manuscript where a drug is used to draw causal relationships (see point #1 below).
- Using small molecule could have secondary effects. It also seems that the drug-induced defects cannot be reversed after being washed away. Furthermore, this drug treatment eliminates almost all H3K27me3 genome-wide, regardless of their occupancy status with mono-ubH2A/Z, making it difficult to make the causal connection between the prerequisite mono-ubH2A/Z occupancy and the subsequent de novo H3K27me3. I think it is important for the authors to address this point more directly as this is the main conclusion of this work. Could the authors perform genetic analyses to confirm the specificity of the phenotypes?
- Page 8, line 160-163: "Curiously, enhancer cluster 5 (Figure 2A) was unique - displaying high H3K4me1, very high H3K27ac, and open chromatin (via ATAC-seq analysis; Figure 2 - figure supplement C, D) - but bore DNA methylation - an unusual combination given the typical strong correlation between high H3K4me1 and DNA hypomethylation." I suspect that the authors are talking about the chromatin state at pre-ZGA stage as this is the only stage DNA methylation pattern was included, but it is hard to tell that this cluster displays high H3K4me1 at all.
We now see the confusion, and are happy to clarify this. We were intending to refer to to the histone marking at postZGA, and the DNAme at postZGA (for cluster #5) – as postZGA is the time when H3K4me1 is high, H3K27ac is very high, and DNAme remains high. The reviewer is right that we do not show the DNAme pattern at post ZGA, only preZGA. However, the DNAme pattern stays almost constant between preZGA (2.5 hpf) and postZGA (4.3 hpf) – a result we published previously in Potok et al., 2010 (note: the maternal genome shows DNA reprogramming prior to 2.5hr, and is then constant through ZGA). We did not include DNAme at every stage simply to save space in Panel A, which was getting crowded. However, to avoid the reader misunderstanding our point, we have taken care to make this clear in the revised manuscript. We thank the reviewer for raising this point.
- Page 10, line 206-207: "PRT4165 treatment also conferred limited new/ectopic Aebp2 peaks (Figure 4C, clusters 4, 6, 7,8)", it seems that clusters 4, 6, 7, 8 together are not "limited" compared to clusters 1, 3, and 5, and could be even more abundant.
Thank you for this comment - we agree with the reviewer and have clarified this in the text and Figure 4. In the initial version, the section where we mention ‘limited’ additional sites was intend to refer to promoters, and although as only a modest fraction of the ectopic sites are at promoters, but we did not provide that context in the text. Indeed, if one looks at all sites in the genome, there are a large number of ectopic sites after PRT4165 treatment. This is shown clearly in the revised Figure 4 (which shows all genomic sites) and we have clarified this in the text.
We were curious whether there is any feature that helps us understand what might unify the ectopic binding, and therefore underlie the mechanism(s). First, we tested whether binding sites for particular transcription factors might be enriched; however, we did not find a class of binding sites that represented more than 3% of the total sites. We note that others have reported some affinity of mammalian Aebp2 for DNA and some limited sequence specificity (Kim et al., NAR 2009), and in the absence of a high-affinity H2AUb target, that shadow DNA binding function may become more apparent. Furthermore, we did not observe chromatin marks that showed a highly significant degree of overlaps. Thus, although intriguing, there does not appear to yet be a logic to the ectopic binding observed.
- In the context of studying the chromatin state of developmental genes in early vertebrate embryos, there are two recent publications in mouse embryos which also investigated the crosstalk between mono-ubH2A and H3K27me3 at the ZGA transition in mouse (https://doi.org/10.1038/s41588-021-00821-2 and https://doi.org/10.1038/s41588-021-00820-3). It would be informative to add some discussion for comparisons between these two vertebrate organisms.
Reviewer #2 (Public Review):
One model for polycomb domain establishment suggests that PRC2 adds H3K27me3 first, and then recruits PRC1 for silencing. The key evidence for this model is the H3K27me3-binding module CBX proteins in canonical PRC1 complexes. This model has been revised by recent studies, and it is now well recognized that the polycomb domains can be de novo established in a different order. In other scenarios, including X inactivation, a non-canonical PRC1 complex that lacks CBX proteins catalyzes ubH2A first, and PRC2 complex is subsequently recruited through recognizing ubH2A modification by its Jarid2 and Aebp2 subunits.
In this manuscript, Hickey and co-workers analyzed the temporal change of various epigenetic marks around ZGA stages during zebrafish early embryo development. Based on their experimental data and bioinformatic analysis, they suggest that polycomb establishment in zebrafish embryo is following the 'non-canonical' order, in which H3K27me3 establishment is dependent on ubH2A pre-deposition and the following recruitment of Aebp2-PRC2 complex. Moreover, they suggest that polycomb-silenced developmental genes are solely repressed by ubH2A, independent of H3K27me3. Overall, the functional analysis (RNF2 inhibitor experiments) conducted in the current study highlights the critical function of PRC1 and ubH2A in silencing developmental genes during early embryo development. Moreover, this study provides clues that could reconcile with the earlier observations that H3K27me3 seems largely dispensable for silencing developmental genes in zebrafish early embryo (e.g. PMID: 31488564).
The main concern is two similar studies have just been published in Nature Genetics using mouse early embryos, and the observation of this manuscript largely agree with the two mouse studies, rendering the novelty of this study.
In addition, certain conclusions in the manuscript requires further experimental support:
- While the authors claim that H3K27me3 is established after ZGA, it is quite surprising to me that they did NOT analyzed the H3K27me3 pattern before ZGA. While IF staining suggests a minimal level of H3K27me3 before ZGA (Fig1 S2B), previous ChIP-seq analysis demonstrate that H3K27me3 are present (e.g. PMID: 22137762).
Briefly, in our own work, we do not detect H3K27me3 by IF prior to ZGA, and we could not detect H3K27me3 peaks by ChIP during preZGA (also mentioned as ‘data not shown’ in Murphy et al., 2018).
- While the RNF2 inhibitor experiment clearly demonstrates that PRC1 is required for the deposition of both ubH2A and H3K27me3, that does not necessarily mean that PRC1-mediated ubH2A deposition precedes H3K27me3. The establishment and maintenance of polycomb domain usually requires the crosstalk and reinforcement between polycomb complexes. Therefore, the deficiency in either PRC1 or PRC2 complex may lead to the decreased level of both marks. To clarify a hierarchical order of the polycomb domain establishment, a phenotypic analysis of PRC2 deficiency is also necessary.
Here, we emphasize that prior to performing the inhibitor experiment, we addressed the temporal order of addition in Figure 1 and in Figure 1 – figure supplement 1. H2Aub1 is added extensively to thousands of developmental genes during preZGA, well before H3K27me3 is detected. We interpret this as evidence that H2Aub1 temporally precedes H3K27me3 during embryonic development. We will also mention (described in the Discussion) that maternal zygotic loss of Ezh2, which eliminates all H3K27me3 in the genome at all embryo stages does not result in the activation of developmental genes.
- Parental difference. As shown in Fig.1B, ubH2A level varies greatly in sperm and egg, which suggests that the reprogramming process of ubH2A (and perhaps H3K27me3) distribution could be significantly different for the two parental alleles. It would be interesting to analyze the ubH2A and H3K27me3 distribution in germ cells before fertilization.
We appreciate the reviewer’s comment and agree that this would be an interesting line of inquiry. However, this would require genomics analyses from reciprocal crosses of highly polymorphic fish strains. This would involve very considerable additional work. Therefore, we will consider this in our future studies.
- The role of Aebp2 subunit. Given the well-characterized function of Aebp2 in recognizing ubH2A, an involvement of Aebp2-PRC2 complex in establishing H3K27me3 on PRC1 pre-deposited regions is not unexpected. Indeed, Aebp2 co-localized well with ubH2A marked regions (Fig.3). However, an issue not clarified in the manuscript is whether Aebp2 is the sole subunit for the recruitment of PRC2 to ubH2A marked regions. Paralleled analysis of the changes for Aebp2 and H3K27me3 upon RNF2 inhibitor treatment is necessary, and Aebp2-dependent and -independent regions should be separately classified for analysis.
- Role of PRC1 on the temporal regulation of gene expression during early development. The authors only analyzed the RNA-seq results for RNF2i treated embryos post ZGA. Therefore, it is currently not clear if the role of PRC1 in transcriptional repression is restricted to post-ZGA stages. RNA-seq analysis of RNF2i treated embryos on those stages are also warranted.
-
Evaluation Summary:
This manuscript is of broad interest to developmental biologists and those studying transcriptional/epigenetic regulation of cell-specific and housekeeping gene programs. The work demonstrates that Polycomb complexes coordinate the regulation of distinct groups of genes during early embryogenesis, which offers interesting insights into how very early embryos differentially control housekeeping versus specific developmental gene promoters/enhancers. The data are of high quality, and the conclusions are insightful yet measured.
(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 #3 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
Hickey et al. studied chromatin landscape changes in early Zebrafish embryos at three distinct stages: preZGA, ZGA and postZGA. Using ChIP-seq on these time-course samples, they examined developmental genes at their regulatory elements, including promoters and enhancers, that carry nucleosomes enriched with histone variant H2A.Z, as well as post-translational modifications H3K4me1 and H3K27ac, but with low DNA methylation, in early-stage embryos prior to turning on zygotic gene expression. During embryogenesis, this group of elements recruit a Polycomb Repressive Complex 1 (PRC1) component Rnf2 to "write" the ubiquitinated H2A or H2A.Z. The mono-ubH2A/Z then recruits a PRC2 component Aebp2 to further "write" the H3K27me3 repressive mark to silent these developmentally regulated genes in later stage embryos. …
Reviewer #1 (Public Review):
Hickey et al. studied chromatin landscape changes in early Zebrafish embryos at three distinct stages: preZGA, ZGA and postZGA. Using ChIP-seq on these time-course samples, they examined developmental genes at their regulatory elements, including promoters and enhancers, that carry nucleosomes enriched with histone variant H2A.Z, as well as post-translational modifications H3K4me1 and H3K27ac, but with low DNA methylation, in early-stage embryos prior to turning on zygotic gene expression. During embryogenesis, this group of elements recruit a Polycomb Repressive Complex 1 (PRC1) component Rnf2 to "write" the ubiquitinated H2A or H2A.Z. The mono-ubH2A/Z then recruits a PRC2 component Aebp2 to further "write" the H3K27me3 repressive mark to silent these developmentally regulated genes in later stage embryos. Using a small molecule to inhibit Rnf2 abolishes H3K27me3 and leads to ectopic gene expression.
Most of the data for the first half of this manuscript are presented in a clear and logic manner. The conclusions based on these correlation assays are quite obvious and well supported (except a few minor points raised below for clarifications, #2-#3). The major concern is for the second half of the manuscript where a drug is used to draw causal relationships (see point #1 below).
1. Using small molecule could have secondary effects. It also seems that the drug-induced defects cannot be reversed after being washed away. Furthermore, this drug treatment eliminates almost all H3K27me3 genome-wide, regardless of their occupancy status with mono-ubH2A/Z, making it difficult to make the causal connection between the prerequisite mono-ubH2A/Z occupancy and the subsequent de novo H3K27me3. I think it is important for the authors to address this point more directly as this is the main conclusion of this work. Could the authors perform genetic analyses to confirm the specificity of the phenotypes?
2. Page 8, line 160-163: "Curiously, enhancer cluster 5 (Figure 2A) was unique - displaying high H3K4me1, very high H3K27ac, and open chromatin (via ATAC-seq analysis; Figure 2 - figure supplement C, D) - but bore DNA methylation - an unusual combination given the typical strong correlation between high H3K4me1 and DNA hypomethylation." I suspect that the authors are talking about the chromatin state at pre-ZGA stage as this is the only stage DNA methylation pattern was included, but it is hard to tell that this cluster displays high H3K4me1 at all.
3. Page 10, line 206-207: "PRT4165 treatment also conferred limited new/ectopic
Aebp2 peaks (Figure 4C, clusters 4, 6, 7,8)", it seems that clusters 4, 6, 7, 8 together are not "limited" compared to clusters 1, 3, and 5, and could be even more abundant.4. In the context of studying the chromatin state of developmental genes in early vertebrate embryos, there are two recent publications in mouse embryos which also investigated the crosstalk between mono-ubH2A and H3K27me3 at the ZGA transition in mouse (https://doi.org/10.1038/s41588-021-00821-2 and https://doi.org/10.1038/s41588-021-00820-3). It would be informative to add some discussion for comparisons between these two vertebrate organisms.
-
Reviewer #2 (Public Review):
One model for polycomb domain establishment suggests that PRC2 adds H3K27me3 first, and then recruits PRC1 for silencing. The key evidence for this model is the H3K27me3-binding module CBX proteins in canonical PRC1 complexes. This model has been revised by recent studies, and it is now well recognized that the polycomb domains can be de novo established in a different order. In other scenarios, including X inactivation, a non-canonical PRC1 complex that lacks CBX proteins catalyzes ubH2A first, and PRC2 complex is subsequently recruited through recognizing ubH2A modification by its Jarid2 and Aebp2 subunits.
In this manuscript, Hickey and co-workers analyzed the temporal change of various epigenetic marks around ZGA stages during zebrafish early embryo development. Based on their experimental data and …
Reviewer #2 (Public Review):
One model for polycomb domain establishment suggests that PRC2 adds H3K27me3 first, and then recruits PRC1 for silencing. The key evidence for this model is the H3K27me3-binding module CBX proteins in canonical PRC1 complexes. This model has been revised by recent studies, and it is now well recognized that the polycomb domains can be de novo established in a different order. In other scenarios, including X inactivation, a non-canonical PRC1 complex that lacks CBX proteins catalyzes ubH2A first, and PRC2 complex is subsequently recruited through recognizing ubH2A modification by its Jarid2 and Aebp2 subunits.
In this manuscript, Hickey and co-workers analyzed the temporal change of various epigenetic marks around ZGA stages during zebrafish early embryo development. Based on their experimental data and bioinformatic analysis, they suggest that polycomb establishment in zebrafish embryo is following the 'non-canonical' order, in which H3K27me3 establishment is dependent on ubH2A pre-deposition and the following recruitment of Aebp2-PRC2 complex. Moreover, they suggest that polycomb-silenced developmental genes are solely repressed by ubH2A, independent of H3K27me3. Overall, the functional analysis (RNF2 inhibitor experiments) conducted in the current study highlights the critical function of PRC1 and ubH2A in silencing developmental genes during early embryo development. Moreover, this study provides clues that could reconcile with the earlier observations that H3K27me3 seems largely dispensable for silencing developmental genes in zebrafish early embryo (e.g. PMID: 31488564).
The main concern is two similar studies have just been published in Nature Genetics using mouse early embryos, and the observation of this manuscript largely agree with the two mouse studies, rendering the novelty of this study.
In addition, certain conclusions in the manuscript requires further experimental support:
1. While the authors claim that H3K27me3 is established after ZGA, it is quite surprising to me that they did NOT analyzed the H3K27me3 pattern before ZGA. While IF staining suggests a minimal level of H3K27me3 before ZGA (Fig1 S2B), previous ChIP-seq analysis demonstrate that H3K27me3 are present (e.g. PMID: 22137762).
2. While the RNF2 inhibitor experiment clearly demonstrates that PRC1 is required for the deposition of both ubH2A and H3K27me3, that does not necessarily mean that PRC1-mediated ubH2A deposition precedes H3K27me3. The establishment and maintenance of polycomb domain usually requires the crosstalk and reinforcement between polycomb complexes. Therefore, the deficiency in either PRC1 or PRC2 complex may lead to the decreased level of both marks. To clarify a hierarchical order of the polycomb domain establishment, a phenotypic analysis of PRC2 deficiency is also necessary.
3. Parental difference. As shown in Fig.1B, ubH2A level varies greatly in sperm and egg, which suggests that the reprogramming process of ubH2A (and perhaps H3K27me3) distribution could be significantly different for the two parental alleles. It would be interesting to analyze the ubH2A and H3K27me3 distribution in germ cells before fertilization.
4. The role of Aebp2 subunit. Given the well-characterized function of Aebp2 in recognizing ubH2A, an involvement of Aebp2-PRC2 complex in establishing H3K27me3 on PRC1 pre-deposited regions is not unexpected. Indeed, Aebp2 co-localized well with ubH2A marked regions (Fig.3). However, an issue not clarified in the manuscript is whether Aebp2 is the sole subunit for the recruitment of PRC2 to ubH2A marked regions. Paralleled analysis of the changes for Aebp2 and H3K27me3 upon RNF2 inhibitor treatment is necessary, and Aebp2-dependent and -independent regions should be separately classified for analysis.
5. Role of PRC1 on the temporal regulation of gene expression during early development. The authors only analyzed the RNA-seq results for RNF2i treated embryos post ZGA. Therefore, it is currently not clear if the role of PRC1 in transcriptional repression is restricted to post-ZGA stages. RNA-seq analysis of RNF2i treated embryos on those stages are also warranted. -
Reviewer #3 (Public Review):
Hickey et al. continue their groups investigation of how vertebrate embryos establish the proper chromatin regulatory context around the time of zygotic genomic activation for both housekeeping genes, which are spatially and temporally widely transcribed, and for developmental genes, which must remain off but must remain available for later lineage-restricted expression. Employing an array of ChIP-Seq analyses of chromatin marks and specific chromatin modifiers (e.g. Rnf2) on pre ZGA, ZGA, and post ZGA zebrafish embryos, they find increased deposition of H2Aub1 by Rnf2-PRC1 at Placeholder nucleosomes (enriched at developmental genes) which, in their model, serves to recruit Aebp2-PRC2 to lay down H3K27me3 marks to repress these developmental genes until their appropriate time of expression later in …
Reviewer #3 (Public Review):
Hickey et al. continue their groups investigation of how vertebrate embryos establish the proper chromatin regulatory context around the time of zygotic genomic activation for both housekeeping genes, which are spatially and temporally widely transcribed, and for developmental genes, which must remain off but must remain available for later lineage-restricted expression. Employing an array of ChIP-Seq analyses of chromatin marks and specific chromatin modifiers (e.g. Rnf2) on pre ZGA, ZGA, and post ZGA zebrafish embryos, they find increased deposition of H2Aub1 by Rnf2-PRC1 at Placeholder nucleosomes (enriched at developmental genes) which, in their model, serves to recruit Aebp2-PRC2 to lay down H3K27me3 marks to repress these developmental genes until their appropriate time of expression later in development post ZGA. Overall, the conclusions are largely well-supported by the data, the proposed models provide a thoughtful interpretation and roadmap forward for future work. Overall, the study also opens a number of interesting questions including how tissue specific transcription factors fit into the overall process as well.
Strengths of the paper include the quality of the epigenetic profiling across multiple informative chromatin marks with chromatin accessibility and DNAme, with multiple highly concordant replicates for the profiling experiments. The resulting data is necessarily very dense and feature-rich, and the authors are largely successful in conveying the meaning of the many "metagene" displays and browser tracks and developing a coherent story of how these show the transcriptional regulatory events captured around ZGA. The high quality of this data will also be of great value for reanalysis across the field. The 'non-canonical' order of PRC component recruitment further broadens our understanding of the multiple mechanisms by which epigenetic regulators can function.
Perhaps the main weakness, and one which is not in an obvious way technically surmountable for the present study but is worth considering going forward, is the reliance on the single drug PRT4165 to block Rnf2 activity and prevent H2Aub1 deposition. An orthologous genetic tool producing similar results would strengthen this mechanistic insight, but seems challenging given the likely wide-ranging effects of loss-of-function in chromatin modifiers (as noted in the Discussion with death at 3 dpf with Rnf2 LOF) and narrow window of time around the ZGA over just a few hours making inducible (e.g. Cre/lox or CRISPR) approaches likely challenging. The use of MZ mutants for Rnf2 might allow for further understanding of the precise temporal requirements for Rnf2 activity by removing maternal contribution that might function even earlier.
-
