Menu

Close

CTCF Mediates Dosage and Sequence-context-dependent Transcriptional Insulation through Formation of Local Chromatin Domains

Hui Huang, Quan Zhu, Adam Jussila, Yuanyuan Han, Bogdan Bintu, Colin Kern, Mattia Conte, Yanxiao Zhang, Simona Bianco, Andrea Chiariello, Miao Yu, Rong Hu, Ivan Juric, Ming Hu, Mario Nicodemi, Xiaowei Zhuang, Bing Ren

Preprint posted on July 08, 2020 https://www.biorxiv.org/content/10.1101/2020.07.07.192526v1

Keeping enhancers and promoters apart with CTCF

Selected by Clarice Hong

Background

Insulators are cis-regulatory elements that are thought to act as boundary elements separating chromosomal domains and prevent improper enhancer-promoter communication. Specifically, enhancer blocking insulators inhibit enhancer-promoter communication when placed between an enhancer and promoter. Enhancer blocking insulators are largely defined by their activity in a plasmid assay, whereby the insulator is placed between an enhancer-promoter pair to prevent the enhancer from activating its target promoter. It is thus unclear whether they can block endogenous enhancer-promoter interactions in the genome.

The first discovered and best-known insulator is the 5’HS4 element in the chicken β-globin locus. It contains a CTCF binding site that is essential for its enhancer-blocking activity. CTCF binding has been implicated in many functions in the genome, including demarcating the boundaries of topologically associated domains (TADs), which are thought to segregate the genome into compartments to prevent aberrant enhancer-promoter interactions. However, acute depletion of CTCF has been found to disrupt boundary formation with minimal changes in gene expression, and CTCF deletion at TAD boundaries only sometimes leads to a functional gene expression change. Furthermore, many CTCF sites are not located at TAD boundaries. The capacity of any given CTCF site in the genome to act as an enhancer blocking insulator is thus unclear. In this preprint, the authors develop a system to test the requirements for CTCF-dependent enhancer-blocking insulators in the genome.

Key findings

The authors first developed a reporter system to test enhancer-blocking insulation in the genome. The Sox2 gene has been previously found to mostly be regulated by a long-range enhancer ~110kb downstream of the gene. Using a hybrid mouse embryonic stem cell line, they tagged one allele of Sox2 with eGFP and the other with mCherry. They then inserted a pair of asymmetric flippase recognition sites between the Sox2 gene and its downstream enhancers on the eGFP allele only, allowing for directional integration of enhancer-blocking insulators at a single location. By measuring the activity of eGFP with and without integration of insulators, the authors could determine the activity of each insulator.

The authors first asked if single CTCF binding sites (CBSs) were sufficient to insulate the Sox2 gene from its enhancer. They tested 11 CBSs (1-4kb long) from 2 TAD boundaries and other chromatin loop anchors. Surprisingly, most insulators were not functional or had modest effects. This suggests that single CBSs are not sufficient to act as enhancer blocking insulators. Thus, the authors hypothesized that multiple CBSs might be needed for insulator function. The authors then concatenated multiple CBSs in various combinations and tested them in the same reporter assay. Indeed, combining even 2 CBSs lead to higher levels of insulation, with the two CBSs generally acting in a synergistic manner. Concatenating up to 4 CBSs reduce expression by about 38% but does not completely block Sox2 expression. Thus, at least 2 CBSs appear to be required for insulator activity.

To understand the differences between boundary and non-boundary CBSs, the authors concatenated 139-bp genomic DNA sequences centered on the CTCF motif from 6 TAD boundaries or 6 non-boundary locations. While the insulator from TAD boundaries was effective as expected, the insulator from non-boundary locations was not functional despite having stronger CTCF binding as measured by ChIP-seq. Thus, CTCF requires appropriate flanking sequence context to act as an enhancer blocking insulator.

Finally, the authors looked at the impact of insulators on 3D genome structure. Using PLAC-seq and high-resolution DNA-FISH, the authors showed that there are reduced interactions between the Sox2 enhancer and promoter when insulators consisting of 2 or 4 CBSs are integrated into the genome. Using Hi-C and DNA-FISH, the authors also showed that there is an increased probability of TAD boundary formation at the insulator. Taken together, insulators at the Sox2 locus appear to function by separating the enhancer and promoter into 2 compartments, thereby reducing their interaction frequency and expression of the Sox2 gene.

What I liked

The assay that the authors developed to test enhancer-blocking activity in the genome is impressive. Enhancer blocking insulators have not generally been functionally tested between endogenous enhancer-promoter pairs in the genome. Clearly, the potency of insulators in the genome in this system is much less than previously observed, for example, the addition of human HS5 reduces reporter gene expression by 60% on plasmids but only about 11% in this system. This highlights the importance of using an integrated reporter system to test regulatory elements since plasmids do not form compartments. I also like how the authors directly asked whether CTCF sites at TAD or loop boundaries are sufficient to act as insulators. These sites have been thought to act as insulators simply because they are found at boundaries. Previous papers have addressed the question of whether CTCF is necessary by deleting CBSs, while this preprint showed that CBSs + its flanking sequences are sufficient. Finally, the results showing that there is increased boundary probability at the location where the insulator is located is very cool. Again, while it is not clear whether compartmentalization is necessary for proper expression, it is clearly sufficient.

Future directions and questions

  1. Given the pleiotropy of CTCF activities, it is important to understand how CTCF behaves differently depending on its context. While this preprint showed that CTCF insulating activity is clearly context-dependent, it would be interesting to know what makes boundary flanking sequences different from non-boundary ones. Are there enriched transcription motifs in boundary sequences vs non-boundary ones that might cooperate with CTCF?
  2. The boundary sequences also appear to be just as important as the CTCF core sequence. When concatenating CBSs, the authors conclude that the dosage of CTCF is crucial for effective insulator activity. However, can another interpretation be that concatenating CBSs simply adds more boundary sequences and therefore more transcription factors to cooperate with CTCF? Similarly, will boundaries that do not contain CTCF binding sites also function as insulators?
  3. Do CBSs at stronger boundaries (higher contact frequency by Hi-C) compared to sites at weaker boundaries act as stronger insulators in this system?
  4. How can we explain the fact that multiple CBSs seem to be necessary for insulation but there are boundaries that only have one CBS?
  5. The single cell imaging experiments showed reduced contact frequency between the Sox2 enhancer and promoter. I am curious about the interaction between the insulator and the enhancer or promoter, since one potential mechanism for how insulators work is by interacting directly with the enhancer and/promoter to prevent them from contacting each other. Is there any evidence for this mechanism?
  6. It is quite interesting that concatenating up to 4 CBSs only leads to a maximum of 38% reduction in expression. One explanation could be that the insulators used in this preprint are not very active. Alternatively, insulators could be very context specific, and placing it at a different locus or at a different distance from the enhancer or promoter could significantly change its activity.

 

Posted on: 25th July 2020

doi: https://doi.org/10.1242/prelights.23459

Read preprint (No Ratings Yet)




Author's response

Hui Huang shared

Hi Clarice,

I very much appreciate your picking our work to highlight in preLights. Thank you very much for the clear summary of our work and the excellent questions. In fact, we are actively working on these aspects to further our understanding of the molecular mechanism regarding how flanking sequences contribute to CTCF mediated insulation. I’m happy to provide my thoughts and speculations on these questions.

  1. Given the pleiotropy of CTCF activities, it is important to understand how CTCF behaves differently depending on its context. While this preprint showed that CTCF insulating activity is clearly context-dependent, it would be interesting to know what makes boundary flanking sequences different from non-boundary ones. Are there enriched transcription motifs in boundary sequences vs non-boundary ones that might cooperate with CTCF?

We are particularly interested in finding out the factors that assist CTCF mediated insulation. We did de novo motif search of the boundary CTCF sites tested. We also compared the motifs enriched in flanking sequences of boundary and non-boundary CTCF sites. We found slight enrichment of RNUX motifs as well as many others in flanking sequences of boundary CTCF sites. However, it is hard to judge which factor is actually involved in insulation based on motif analysis. We are currently taking different strategies to attack this question. 1. We applied genome-wide gene CRISPR knock out (GeCKO) screening to search regulators of insulators. 2. We tested a series of truncations of the boundary CTCF sequences to fine map essential motif in the flanking sequences.

  1. The boundary sequences also appear to be just as important as the CTCF core sequence. When concatenating CBSs, the authors conclude that the dosage of CTCF is crucial for effective insulator activity. However, can another interpretation be that concatenating CBSs simply adds more boundary sequences and therefore more transcription factors to cooperate with CTCF? Similarly, will boundaries that do not contain CTCF binding sites also function as insulators?

It is a great question. We believe that CTCF is essential for insulation. On top of that, other factors are involved. In the case of CBSs obtained from the Sox9-Kcnj2 boundary (~7.5kb), we also tested a mutant sequence where we deleted all 19bp-CTCF-motifs. This mutant boundary sequence showed no insulation effect when inserted between the Sox2 gene and its super-enhancer.

  1. Do CBSs at stronger boundaries (higher contact frequency by Hi-C) compared to sites at weaker boundaries act as stronger insulators in this system?

It is a very interesting question. In other words, the question asks whether Hi-C measured contact insulation positively correlated transcriptional insulation. We did not test weak boundary sequences in our system. I tend to believe that this correlation exists. We found that 4CBS provided stronger transcriptional insulation and contact insulation than 2CBS (Figure2b, Figure4 b-c).

  1. How can we explain the fact that multiple CBSs seem to be necessary for insulation but there are boundaries that only have one CBS?

Thank you for this question. Insulation effect of CBS is very likely also dependent on the strength of enhancer it blocks. The Sox2 super-enhancer we tested is a very potent enhancer in mouse ES cells. That’s probably why we only observed mild insulation effect by single CBSs. However, we want to highlight the quantitative relationship between insulation strength and the number of CBS. I believe CTCF can be less important in forming boundary at some genomic locations and some boundaries are CTCF-independent.

  1. The single cell imaging experiments showed reduced contact frequency between the Sox2 enhancer and promoter. I am curious about the interaction between the insulator and the enhancer or promoter, since one potential mechanism for how insulators work is by interacting directly with the enhancer and/promoter to prevent them from contacting each other. Is there any evidence for this mechanism?

Accumulating evidence supports that chromatin domains are formed through loop extrusion process. The cohesin complex can track along the chromatin and stalls at convergent CTCF sites, thus forming chromatin loops/TADs. In our case, the inserted CTCF sites are likely to interact with CTCF sites on Sox2 promoter and super-enhancer to form two smaller chromatin loops, therefore, spatially separate the Sox2 gene from its super-enhancer. Interestingly, our results suggest that other factors are needed to form sub-domains and exert insulation function in addition to CTCF. Because we observed strong CTCF binding to non-boundary CBSs, yet no insulation effect was observed. We are currently working on the chromatin tracing experiments in the insertion clone of non-boundary CTCF sites.

  1. It is quite interesting that concatenating up to 4 CBSs only leads to a maximum of 38% reduction in expression. One explanation could be that the insulators used in this preprint are not very active. Alternatively, insulators could be very context specific, and placing it at a different locus or at a different distance from the enhancer or promoter could significantly change its activity.

This question is related to question 4. The insulation effect(reduction in gene expression) is dependent on insulator activity as well as other factors such as the strength of the enhancer. I believe the same 4CBS sequence will have stronger insulation effect if inserted between a weaker enhancer-gene pair. I don’t know how the insulation effect will change along with the distance from the enhancer. I’d expect the difference to be mild.

Reference:

  1. Dixon, J.R. et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 485, 376-380 (2012).
  2. Despang, A. et al. Functional dissection of the Sox9-Kcnj2 locus identifies nonessential and instructive roles of TAD architecture. Nat Genet 51, 1263-1271 (2019).
  3. Davidson, I.F. et al. DNA loop extrusion by human cohesin. Science 366, 1338-1345 (2019).
  4. Fudenberg, G. et al. Formation of Chromosomal Domains by Loop Extrusion. Cell Rep 15, 2038-2049 (2016).
  5. Kim, Y., Shi, Z., Zhang, H., Finkelstein, I.J. & Yu, H. Human cohesin compacts DNA by loop extrusion. Science 366, 1345-1349 (2019).
  6. Sanborn, A.L. et al. Chromatin extrusion explains key features of loop and domain formation in wild-type and engineered genomes. Proc Natl Acad Sci U S A 112, E6456-6465 (2015).
  7. Hnisz, D., Day, D.S. & Young, R.A. Insulated Neighborhoods: Structural and Functional Units of Mammalian Gene Control. Cell 167, 1188-1200 (2016).

 

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