Menu

Close

Orphan CpG islands boost the regulatory activity of poised enhancers and dictate the responsiveness of their target genes

Tomás Pachano, Víctor Sánchez-Gaya, María Mariner-Faulí, Thais Ealo, Helena G. Asenjo, Patricia Respuela, Sara Cruz-Molina, Wilfred F. J. van Ijcken, David Landeira, Álvaro Rada-Iglesias

Preprint posted on August 05, 2020 https://www.biorxiv.org/content/10.1101/2020.08.05.237768v1

No gene (regulation) is an island: the presence of CpG islands on enhancers and promoters defines the ‘language’ of their communication.

Selected by Jesus Victorino

Categories: genetics

*If you liked this preLights, please click on the thumb-up icon at the end of the page. Any comment, suggestion or question related to either scientific discussion or format will be more than welcome and very much appreciated. You can write directly at the bottom of this page or contact me by email or Twitter.

 

A couple of weeks ago, I saw this very interesting preprint on my regular bioRxiv alert on developmental biology, genetics and genomics. What might sound a bit depressive though, is that despite being August, this preprint has been the closest I have been to anything related to an island this year. While I obviously did not get a sun tan, I enjoyed a lot learning about enhancer-promoter compatibility —which somehow paid off.

Here’s my highlight on the most relevant messages of this manuscript from Álvaro Rada-Iglesias’ Lab

Summary

Enhancers are essential for the precise control of gene expression during vertebrate development. Poised enhancers are a subset of enhancers that reside in CpG rich regions and show enhancer marks (P300 & H3K4me1) while also featuring repressive histone marks (H3K27me3) [1]. Whereas a number of biochemical or functional assays can be employed to identify enhancers, their target genes are harder to predict and require experimental validation such as (epi)genetic editing. Enhancer-promoter communication is believed to happen within topological associated domains (TADs) which organize chromatin in functional compartments. However, in many instances enhancers do not regulate all genes within a TAD and their specific behaviours suggest that other layers of genetic information shape this sophisticated communication.

In this preprint by Tomás Pachano and colleagues, they found that CpG islands at poised enhancers interact with CpG-rich promoters (usually related to developmental genes), determining enhancer-promoter compatibility. The authors studied poised enhancers in mouse embryonic stem cells (mESC) that are later activated in anterior neural precursor and discovered that CpG islands are an essential component that favoured binding by polycomb group complexes. Despite the presence of polycomb proteins, which are traditionally associated with gene repression, CpG islands boosted enhancer activity. Therefore, polycomb binding seemed to contribute to putting enhancers and promoters in close proximity, rather than preventing their expression in mESC. Altogether, this work strengthens the entity of CpG islands as a functional genetic element and provides mechanistic insights towards a better understanding of enhancer responsiveness.

 

Figure 1. Graphical summary of the proposed model on enhancer-promoter compatibility. As shown in the left panel, enhancer-promoter communication is delimited by TAD boundaries and the transcriptional control of gene expression relies on the presence of CpG islands in both the poised enhancer and the promoter. This new layer of genetic information can potentially help explaining the molecular mechanisms underlying some cases of congenital diseases (right panel).

Key findings:

  • Poised enhancers do not function as silencers in mESCs: deletion of a Sox1 poised enhancers (or the CpG islands within it) downregulated Sox1 after differentiation rather than upregulating it in mESCs.
  • CpG islands promote polycomb-mediated long-range interactions between poised enhancers and promoters: genetic insertions of a poised enhancer together with a CpG island in the locus of Gata6, generated binding of polycomb group of protein and new chromatin contacts between promoter and enhancer.
  • Long-range responsiveness to poised enhancers is exclusive to CpG rich promoters: CpG-poor promoters do not show enhancer responsiveness over the long range while they do respond to enhancer regulation when artificially placed near the promoter region.

Why I chose this paper

Since I started my PhD almost five years ago, I have been obsessed with regulatory elements and how genetic variation affects gene regulation. Thanks to the large amount of ChIPseq and HiC data available one can take a quick look at a genomic locus and have a rough idea of the epigenetic status of a gene of interest. However, despite selecting candidate enhancers within the same TAD, we do not fully understand how promoters communicate with enhancers and other regulatory elements. Therefore, we are still far from predicting gene regulation without experimental validation and, even more, the effect of genetic variation which is identified in uncharacterized regions in most cases.

While there is literature about differential gene regulation between developmental vs housekeeping genes [2], I was very enthusiastic to read about differential enhancer responsiveness based on the features of promoter and enhancer sequences. We have only written a few pages of the epigenetic ‘dictionary’ and more studies are needed in this direction to find new general rules hidden in the genome. The ability of CpG islands located within or near enhancers to facilitate its interaction with target genes is a major advance that will help us interpret the mechanisms underlying gene regulation and raises the question of whether this would be the case for other types of less studied elements such as silencers. Last but not least, this preprint presents CpG islands as new regulatory players that can be susceptible to disease-causing mutations.

Questions

  • Gene regulation can be achieved from very distal elements located over 1 megabase from their target gene and CpG islands seem to play an important role in long-range interaction between poised enhancers and promoters. Have the authors checked whether there are CpG islands in such distal elements that are not specifically poised in mESC? Do the authors think this can be a general mechanism of connecting distal chromatin regions?
  • This work suggests that polycomb function at poised enhancers is not to prevent them from activating gene expression in mESC. What do the authors think might keep enhancers unable to boost transcription? Might this information still be encoded in the pool of binding sites for transcription factors?
  • In this preprint, deletion of poised enhancers did not increase gene expression in mESC. However, in a paper published earlier this year by Ngan and colleagues [3], they identified silencers in mESC associated with polycomb repressive complex 2 (PRC2) and showed that these elements transition to enhancers during differentiation. According to their data, PRC2-bound regions contain features of poised enhancers, including CpG islands, that keep genes repressed during pluripotency. How do the authors explain such diverse behaviours among very similar regulatory regions?

References

  1. Rada-Iglesias A et al. 2011. A unique chromatin signature uncovers early developmental enhancers in humans. Nature, 470, 279–283.
  2. Zabidi MA et al. 2015. Enhancer–core-promoter specificity separates developmental and housekeeping gene regulation. Nature 518, 556–559.
  3. Ngan CY et al. 2020. Chromatin interaction analyses elucidate the roles of PRC2-bound silencers in mouse development. Nature Genetics. doi:10.1038/s41588-020-0581-x 

Tags: cpg island, crispr, gene expression, histone marks, mouse stem cells, ocgis

Posted on: 7th September 2020

doi: https://doi.org/10.1242/prelights.24401

Read preprint (2 votes)




  • Author's response

    Tomás and Álvaro shared

    • Gene regulation can be achieved from very distal elements located over 1 megabase from their target gene and CpG islands seem to play an important role in long-range interaction between poised enhancers and promoters. Have the authors checked whether there are CpG islands in such distal elements that are not specifically poised in mESC? Do the authors think this can be a general mechanism of connecting distal chromatin regions?
      Most enhancers, including some classical ones like the ZRS enhancer controlling Shh expression in the limb, do not have CpG islands nearby. Therefore, we do not think that orphan CpG islands represent the only mechanism to achieve long-range communication between genes and enhancers. Instead, there might be several alternative mechanisms through which different enhancers can efficiently communicate with their target genes. Therefore, we favour the idea of enhancers being a heterogeneous group of regulatory elements with diverse genetic, epigenetic and topological features that determine the mechanism by which they control gene expression.On the other hand, there are many orphan CpG islands that are not associated to ESC poised enhancers. Although these orphan CpG islands could play other functions, such as alternative gene promoters (10.1371/journal.pgen.1001134), it is certainly possible that some of them are part of enhancers that could become poised and subsequently activated in other developmental contexts that we have not investigated yet. Therefore, the importance of CpG islands for enhancer-gene communication might be quite widespread and general across different cell types and tissues.
    • This work suggests that polycomb function at poised enhancers is not to prevent them from activating gene expression in mESC. What do the authors think might keep enhancers unable to boost transcription? Might this information still be encoded in the pool of binding sites for transcription factors?That is exactly what we think is happening. Poised enhancers have a modular   composition consisting of a cluster of transcription factor binding sites (TFBS) and a nearby CpG island. Our data clearly shows that the TFBS confer the cis-activation capacity to the poised enhancers, while the CpG islands can boost the regulatory activity of the poised enhancers but have no cis-activation capacity on their own. Therefore, the poised enhancers might not activate their target genes in mESC simply because their cognate transcription factors are still not expressed in that cellular state, and not because of polycomb repression. This can be seen in Fig. 3, where we show that the PE Sox1(+35)TFBS insert at the Gata6-TAD is not bound by polycomb in mESC yet it does not get activated (i.e. gain H3K27ac) until the ESC are differentiated into anterior neural progenitors. A non-mutually exclusive possibility is that the TFBS modules within poised enhancers contain binding sites for repressors expressed in ESC that prevent the activation of the enhancers. These repressors might become silenced upon ESC differentiation and allow other transcription factors to activate the poised enhancers. 
    • In this preprint, deletion of poised enhancers did not increase gene expression in mESC. However, in a paper published earlier this year by Ngan and colleagues [3], they identified silencers in mESC associated with polycomb repressive complex 2 (PRC2) and showed that these elements transition to enhancers during differentiation. According to their data, PRC2-bound regions contain features of poised enhancers, including CpG islands, that keep genes repressed during pluripotency. How do the authors explain such diverse behaviours among very similar regulatory regions?Although these Polycomb-associated silencers indeed resemble poised enhancers, our experiments based on either deletions of CpG islands associated with endogenous poised enhancers (Fig. 1) or insertions of CpG islands together with TFBS sequences at exogenous loci (Fig 3) do not support a repressive function for CpG islands, and thus for Polycomb complexes, within poised enhancers.   Moreover, in our previous work (1016/j.stem.2017.02.004), we deleted five different poised enhancers (including the TFBS alone or together with their nearby CpG islands) and did not observe any significant gene expression changes in ESC. Similarly, the complete loss of PRC2 in mESC did not result in the activation of either poised enhancers or their target genes (10.1016/j.stem.2017.02.004), which is in agreement with PRC2 not being required for gene silencing in mESC (10.1016/j.molcel.2014.06.005). Nevertheless, the overall number of distal elements deleted by us (5 loci) and Ngan et al. (4 loci) is still quite limited, so it is certainly possible that there are different types of poised enhancers, with some of them having a dual function as both silencers and enhancers depending on the cellular context (10.1016/j.molcel.2019.10.004). Future work should aim at characterizing the genetic and epigenetic differences between these groups of distal elements, which might explain their diverse behaviours. On the other hand, the silencer effects reported by Ngan et al. are in general quite modest, with the target genes increasing their expression by less than 2-fold on average upon deletion of the investigated distal elements. This further suggests that PRC2 is a weak repressor in ESC, although it can lead to stronger gene silencing upon differentiation (10.1016/j.molcel.2014.06.005). Therefore, it might be interesting to investigate whether poised enhancers might activate their target genes in certain lineages (e.g. neuroectoderm) while contributing to their silencing in others (e.g. endoderm).

    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
    Close