Preformed Chromatin Topology Assists Transcriptional Robustness of Shh during Limb Development

Christina Paliou, Philine Guckelberger, Robert Schöpflin, Verena Heinrich, Andrea Esposito, Andrea Maria Maria Chiariello, Simona Bianco, Carlo Annunziatella, Johannes Helmuth, Stefan Haas, Ivana Jerković, Norbert Brieske, Lars Wittler, Bernd Timmermann, Mario Nicodemi, Martin Vingron, Stefan Mundlos, Guillaume Andrey

Preprint posted on January 23, 2019

Article now published in Proceedings of the National Academy of Sciences at

If chromatin loops are preformed and ubiquitous, can they play a role in the regulation of a process as dynamic as developmental gene expression?

Selected by Rafael Galupa



The advent of chromosome conformation capture technologies has prompted discoveries regarding how chromosomes are organised and how the chromatin fibre folds, such as the realization that chromosomes are organised in topologically associating domains (TADs) – chromatin regions at the sub-megabase scale that physically interact with each other more frequently than with sequences outside the respective TADs. This topological organisation has been proposed to play a role in the regulation of gene activity, namely in helping genomic elements such as enhancers to convey regulatory information to their target genes, which are sometimes located far away from each other on the “linear” scale of the genome. This is the case for example for the Shh locus and its limb-specific enhancer, ZRS, which are located 1Mb from each other. Shh is crucial for limb development and its expression in the mouse limb is known to be solely dependent on ZRS – knock-outs of either ZRS or Shh lead to limb malformation. As expected for most enhancer-promoter pairs, Shh and ZRS are found within the same TAD. While the looping between enhancers and promoters has been described as dynamic for certain loci, for many others those interactions are preformed. As the authors show in this study, the contacts between Shh and ZRS are present even in tissues where ZRS is not important to activate Shh expression, or where Shh is not expressed. So how important can these preformed contacts be for Shh expression in the limb?


Key findings

  • The authors abrogated the contacts between ZRS and Shh by generating homozygous deletions of binding sites for CTCF flanking ZRS; CTCF binding has been shown to mediate topological loops in mammals. Limbs of these mutant mice showed a 50% reduction in Shh Given that Shh expression in the limb is solely dependent on ZRS, these findings mean that ZRS is still able to activate Shh expression in the absence of looping. This also means that the contacts between ZRS and Shh are therefore not essential for Shh expression in the limb. Of note, the 50% reduction in Shh expression levels led to no phenotype in limb development; this is consistent with findings in heterozygous ZRS-knockouts, which also show 50% reduction of Shh expression and no phenotypical consequences.


  • The authors generated the same deletion in the context of a hypomorph allele of ZRS, in which Shh expression is already reduced 75%, with phenotypical consequences. Limbs of the double mutant mice showed no Shh expression left and a more severe phenotype, implying that the remaining expression of Shh in the context of the hypomorph ZRS allele depends on the contacts formed by ZRS and Shh. As the authors suggest, these contacts might therefore provide “robustness in gene regulation that can buffer variations in enhancer activity by maintaining Shh mRNA in high excess”.



Open questions & What I like about this preprint

  • (1) The most pressing question that stems out from this paper is: how do ZRS and Shh communicate if they do not depend on the chromatin loops between them? Other studies have started suggesting that enhancers and promoters might not need to be close in space for cis-regulation. It would be interesting nevertheless to determine whether in the mutant limbs ZRS and Shh could still be within the same range of physical proximity as in wildtype limbs. Might the TAD structure itself be important to mediate their proximity, or do they communicate in a manner that is completely independent of spatial proximity?


  • (2) I like the concept that the topological organisation of chromosomes provides robustness to the regulation of gene expression, as this implies a non-essential but important regulatory layer, which could be more amenable to evolutionary tinkering. The authors point out in their discussion that the expression of other loci (or of Shh in other tissues) often relies on multiple enhancers that can provide robustness to the system, while Shh expression in the limb relies only on ZRS – chromatin topology might thus be more important in this case. It would be interesting to investigate whether indeed expression of loci relying on fewer regulatory elements would be more sensitive to their topological organization – or whether increasing the number of regulatory elements for a given locus would make it more resistant to changes in its topological organisation.


  • (3) The authors also made some interesting observations regarding the binding of CTCF and cohesin in the different mutants they observed – when deleting some CTCF sites, they observed ectopic binding on previously unused sites, but this was not always accompanied by a cohesin signal as would be expected. The relationship between CTCF and cohesin binding certainly deserves further exploration.


  • (4) Finally, I think this study also raises interesting questions about dosage of gene expression. I spent my PhD studying regulatory loci of X-chromosome inactivation, an essential developmental process that serves to equalise X-linked gene expression between XX and XY individuals. Failure in X-inactivation – and thus a double dosage of X-linked genes – leads to early female lethality. Expression levels of Shh in the developing limb are clearly in excess, as a 50% reduction does not result in phenotypical consequences – how much is thus enough to ensure proper development? And how is this related to the morphogen gradient formed by Shh and to the interpretation of that gradient? These questions are surely beyond the scope of the preprint by Paliou et al, but they certainly lie at the core of our interest in gene regulation during development.


Tags: chromatin loops, enhancers, gene regulation, mouse limb, sonic hedgehog, tads

Posted on: 8th April 2019 , updated on: 16th April 2019

Read preprint (1 votes)

  • Author's response

    Guillaume Andrey and Christina Paliou shared

    (1) In normal conditions, Shh and the ZRS are involved in a preformed chromatin loop unlike a majority of characterized developmental enhancers and promoters. When the loop is abrogated by the deletion of CTCF binding sites, the interaction decreases to a level similar to an average intra-TAD interaction. To us, this correspond to a more transient and thus less stable interaction occurring within the frame of a TAD. Thus, the enhancer and the promoter still find one another, but to a lower frequency. However, the resolution of HiC to precisely quantify the Shh-ZRS contact in the CTCF mutants might be insufficient. Accordingly, we would argue that further research and alternative molecular tools are needed to better characterize the remaining interaction and the factors that might mediate it.

    (2) So far, an overwhelming number of studied developmental loci (for example Hoxd, Sox9 or Ihh) seem to have adopted a strategy involving increased number of regulatory regions rather than preformed enhancer-promoter interactions. The evolution of this latter strategy might come from the exceptionally high specificity of activity of the ZRS limb enhancer and its “mutability”. Indeed, many mutations, that slightly alters the enhancer sequence have been shown to result in Shh gain-of-function and severe digit malformations. Thereby, duplicating such a mutation-prone region to increase robustness might be less favorable than reaching a similar effect via the establishment of a preformed chromatin interaction.

    (4) It is interesting that one needs to hit a threshold of 25-30% of remaining Shh expression to observe clear skeletal phenotypes. In general, the production of a morphogen like SHH is regulated at many levels, for example at the secretion level or by post-translational modifications. Thus, it appears that a good transcriptional strategy is to enable a large reservoir of transcripts, which effects are further controlled during later regulatory steps at the protein level. In this view, the transcription could occur in “excess”. More generally, the loss of single copy of many developmental genes result in marginal effects in mice whereas the same haploinsufficiency might already be problematic in human patients, suggesting that mice development might be more resilient to partial loss-of-function.

    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