Maintenance of spatial gene expression by Polycomb-mediated repression after formation of a vertebrate body plan

Julien Rougot, Naomi D Chrispijn, Marco Aben, Dei M Elurbe, Karolina M Andralojc, Patrick J Murphy, Pascal WTC Jansen, Michiel Vermeulen, Bradley R Cairns, Leonie M Kamminga

Preprint posted on November 12, 2018

Though ezh2-mutant zebrafish form grossly normal body plans, key developmental factors are dysregulated early on, which dooms the embryo to later failure of organogenesis and lethality.

Selected by Yen-Chung Chen

Background and context

Polycomb repressive complex 2 (PRC2) mediates methylation of lysine 27 on histone 3 (H3K27me3). H3K27me3 is a chromatin mark associated with gene repression, and the disruption of PRC2 results in grave disruption of development. In flies, body segments are organized wrongly when H3K27me3 is disrupted, and mouse embryos die before gastrulation, presumably as a result of failed differentiation. These severe consequences have made H3K27me3 one of the best studied histone modifications in development, and several critical genes governing segmentation or fate specification have been shown to be under the control of PRC2 via H3K27me3 [Reviewed in 1 and 2].

Because mammalian embryos fail to gastrulate without PRC2, the function of PRC2 in development is mostly studied by differentiating stem cells in vitro. Although the in vivo functions of PRC2 are well characterized in flies, the mode of action of Drosophila PRC2 is starkly different from vertebrate PRC2. In a previous study from the Kamminga lab, they used a germ cell transplantation technique in zebrafish to generate a PRC2 loss-of-function model by mutating ezh2, a core component of PRC2. They found out that even without ezh2, the fish remains grossly normal through segmentation until organogenesis [3]. The power of this model enables Rougot et al. to investigate the consequences of the absence of H3K27me3 in gene regulation in vivo in the current preprint.

Key findings

Ezh2 mutant embryos
Zebra fish provides a novel and versatile model to study the consequences of the loss of H3K27me3 (Figures are adapted from the manuscript under a CC-BY-NC-ND 4.0 International license).
(A) ezh2-mutant embryo develops largely normal up to 48 hours post-fertilization (from Figure 1A)
(B) ezh2-mutants are steadily deprived of ezh2 during early embryogenesis (from Figure 1B).
(C) The landscape of histone marks and the binding of writers are perturbed (from Figure 1G).
(D) Spatial expression of histone mark-regulated genes is altered in the absence of ezh2 (from Figure 6A).


Rougot et al. showed that in maternal-zygotic ezh2mutant zebrafish, the level of H3K27me3 decreases to a level barely detectable with ChIP-seq. Interestingly, besides the absence of H3K27me3, rnf2, the core component of PRC1, failed to find its normal targets in ezh2 mutants, though the subunits of PRC1 are present in ezh2mutants. It has been unclear whether the action of PRC1 is dependent on the disposition of H3K27me3, and the current study provides a new line of evidence that in the early stages of zebrafish development, PRC1 requires functional PRC2 to work properly.

In the absence of H3K27me3, H3K4me3, a histone modification associated with transcriptionally active state, is enriched in otherwise H3K27me3-occupied gene loci. Consistent with the important role H3K27me3 plays in developmental gene regulation, genes regulating developmental processes gain H3K4me3 and are upregulated in the absence of H3K27me3. The authors further performed mass spectrometry to check if these epigenetic and transcriptomic changes resulted in altered proteome. Indeed, the proteomic study shows an up-regulation of both a subset of genes occupied by H3K27me3 in the wildtype and the genes gaining H3K4me3 in the ezh2 mutant.

To further examine segment-specific effects of H3K27me3 disruption, the authors first performed RT-qPCR for different body segments to evaluate the expression pattern of hox genes, which are expressed with a clear segmental pattern and govern the head-to-tail body plane. In ezh2 mutants, thoracic hox genes (e.g. hoxa9a and hoxa9b) are up-regulated in the rostral part of body. More caudal hox genes (e.g. hoxa11b and hoxa13b) are also up-regulated but to a lesser extent. These caudal hoxgenes might be repressed by other mechanisms in addition to H3K27me3. Then, the authors performed in situhybridization of several transcription factors with known spatial expression patterns, including the tbx family, isl1, and gsc. Ectopic expression of every transcription factor examined in mutants. The most extreme change is observed in gsc. The normal expression pattern for gsc is confined in telencephalon and diencephalon, branchial arches, and otic vesicle, while in ezh2 mutants, the expression becomes ubiquitous despite being weak.

In short, Rougot et al. used ezh2 mutant zebrafish to show that the early developmental process is robust against the absence of H3K27me3 and proceeds in a grossly normal pattern without H3K27me3. Nonetheless, dysregulation of expression of developmental regulators is detected at the RNA and protein level at 24 hpf, and the spatial patterns of these developmental regulators are perturbed to various extent without H3K27me3.

Why I like this preprint

Histone modifications have long been proposed as an important factor in development to ensure precise coordination of differentiation processes. Indeed, mechanistic studies corroborate the roles of histone modifications in gene regulation and identify many processes they are involved in. Nevertheless, many studies on histone modification-mediated regulation in mammals are performed in vitro, and phenotypic and mechanistic discrepancies between in vivo and in vitromodels are not uncommon. Besides, although the well-defined and scalable nature of cell models enables detailed characterization of the mechanism and a great variety of manipulations, the lack of higher order organization and interactions between different cell types in cell models poses a great challenge for researchers hoping to assess the crosstalk between different regulatory mechanisms. For example, it is common that a striking mutant phenotype in cell models is less severe and presumably compensated in vivobecause environmental cues might provide information that compensates the dysfunction. Thus, it is an invaluable resource to have a vertebrate model of PRC2 loss of function, and this study shows the degree of erosion of spatial expression pattern differs a lot among the genes examined, which might suggest H3K27me3 contributes differently to the repression of H3K27me3-marked genes.

Open questions

  1. H3K27me3 is reported to safeguard cell identities. Different cell identities are often controlled by antagonizing developmental cues and undergo transcriptomic changes in opposite direction. Would it be possible to examine each germ layer individually to see if the cells gain more plasticity or dual identity in the absence of H3K27me3? This could also rule out the possibility that the transcriptomic changes in different germ layers of ezh2mutant cancel out each other when an embryo is profiled as a whole.
  2. It is still unclear how PRC2 finds its target loci to tri-methylate. Would it be feasible to re-introduce ezh2 later in development in an ezh2mutant zebrafish to see if H3K27me3 at certain loci could be established properly to see whether the guiding signal that leads PRC2 to its target loci is transient or not?


  1. Kassis, J. A., Kennison, J. A. & Tamkun, J. W. Polycomb and Trithorax Group Genes in Drosophila. Genetics. 206,1699–1725 (2017).
  2. Aloia, L., Di Stefano, B. & Di Croce, L. Polycomb complexes in stem cells and embryonic development. Development 140, 2525 LP-2534 (2013).
  3. San, B. et al.Normal formation of a vertebrate body plan and loss of tissue maintenance in the absence of ezh2. Rep. 6,24658 (2016).

Tags: epigenetics, h3k27me3, prc2, zebra fish

Posted on: 7th January 2019 , updated on: 18th January 2019

Read preprint (No Ratings Yet)

  • Author's response to the question raised in this highlight

    Leonie M Kamminga shared

    1. This is an interesting thought. We performed our analysis on whole embryo lysates at 24hpf, a stage when many different cell types are already present. By using germ layer specific transgenes or markers it would be possible to study the individual germ layers. However, in case the identity of cells changed because of loss of Ezh2, these cells might have lost the germ layer specific identity and will therefore not be picked up. In addition, and the same goes for single cell techniques, in case cell identity has changed, it will be difficult to reconstruct the embryo based on expression profiles. A good alternative, in this case, would be an approach using tomo-seq. This technique ensures the spatial information of gene expression will be maintained and changes between wildtype and mutant embryos will be better and easier to compare.
    2. The way PRC2 finds its target loci is still under debate. A recent study shows Mtf2 is involved in recruitment of PRC2. The authors of the paper describing this work, propose that recruitment of PRC2 is dependent on the helical shape of the DNA. There are different methods to reintroduce Ezh2 later during development in MZezh2mutants to get insight into recruitment of PRC2. One of the approaches would be to cross a germ line mutant female with a heterozygous male, which results in having only zygotic expression of ezh2. However, with this method it is difficult to control when ezh2starts to express. Another method, which allows control of timing of ezh2expression involves a transgene with an inducible promoter, for instance a heat shock promoter.

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here