Bystander activation across a TAD boundary supports a cohesin-dependent hub-model for enhancer function

Article activity feed

1. 

   Review Commons

   Jan 20, 2025

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   We thank the reviews for their thorough assessment of our manuscript and their constructive suggestions for further improvement. We are pleased that the reviewers recognise that “*this work represents an important and substantive contribution” *to the field of genome organization and gene transcription.

   Reviewer 1

   1. Does the CTCF degron substantially remove CTCF from the Mnx1/Shh TAD border? In prior AID-CTCF degron studies a considerable fraction of cohesin dependent TAD borders are retained upon CTCF removal. Moreover, CTCF sites at these retained borders still have clear ChIP-seq peaks - even though the protein is >95% depleted and scarcely detectable …

   Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Reply to the reviewers

   We thank the reviews for their thorough assessment of our manuscript and their constructive suggestions for further improvement. We are pleased that the reviewers recognise that “*this work represents an important and substantive contribution” *to the field of genome organization and gene transcription.

   Reviewer 1

   1. Does the CTCF degron substantially remove CTCF from the Mnx1/Shh TAD border? In prior AID-CTCF degron studies a considerable fraction of cohesin dependent TAD borders are retained upon CTCF removal. Moreover, CTCF sites at these retained borders still have clear ChIP-seq peaks - even though the protein is >95% depleted and scarcely detectable by western. Thus, while I suspect that the authors are correct that the shorter distance of the 35 kb border deletion contributes substantially to the increased crosstalk between the Mnx1 and Shh-enhancers, I suspect part of the reason for a lack of a similar effect in the CTCF degron is due to the known challenges in removing CTCF from this border. To argue that the border but not the CTCF is important, I think it would be helpful to show the CTCF signal is sufficiently lost in the degron by ChIP-seq and/or show that this TAD border has been lost by Hi-C. Alternatively, the authors could tone down this claim to something more conservative, as I did not find it to be presented as a key conclusion of the paper as a whole.

   We used the CTCF-AID mESC line published by Nora et al (2017). In our previous manuscript (Kane et al., 2022) we presented the published Hi-C and CTCF-ChIP-seq data from these cells at the Shh TAD (Fig 2c of Kane et al) – reproduced below for the reviewer’s benefit. This shows the loss of insulation at the Shh/Mnx1 TAD boundary when CTCF is degraded, and the loss of CTCF ChIP-seq signal at this boundary.

   1. In my opinion, the authors' description of existing data for the importance of TAD borders in enhancer promoter regulation is not described in a sufficiently balanced and complete manner, and overall impression given by the text is that CTCF marked borders have little serious evidence for a role in developmental enhancer specificity and are maybe a cancer thing. This is doubly unfortunate, as it undermines the impact of the authors work in expanding our view of what TAD borders are in a regulatory sense, as well as presents an unbalanced view of work in the field. This is of course easily corrected. In particular, I recommend the following revisions: It is " depletion of CTCF has only a small effect on transcription in cell culture (Nora et al., 2017; Hsieh et al., 2022)." It should be clarified that there is only a small *acute * effect on transcription (in the first 6-12 hours), which may tell us more about the timescale at which promoters sample, integrate and respond to changes in their enhancer environment than about the roles of CTCF particularly. Notably, this degradation is *lethal*, it results in massive changes in transcription after 4 days, and I suspect the authors agree that this lethal affect arises from CTCF's role in transcription regulation (if you remove some key cytoskeletal protein or metabolic enzyme the primary cause of cell death is not transcriptional, but almost all the evidence for CTCF's vital role in the cell is linked in one way or another to transcription).

   As suggested by the reviewer we have inserted the word “acute” into that sentence.

   The discussion of TAD border deletions is more one-sided than ideal. I appreciate the discussion is usually even more unbalanced when presenting the opposite view in the literature - many works only cite the examples where border deletion does lead to ectopic expression and phenotypes. The current text presented a subset of these border deletion data in such a way as to give me the impression the authors are deeply skeptical that CTCF plays a role as an insulator of E-P interactions in a developmental context (rather than just as a weird cancer thing). For example: Pennacchio's lab has analyzed a series of TAD border deletions with more examples of both lethal effects and effects with no apparent phenotype 3

   I appreciate that Bickmore and colleagues found quite phenotypically normal mice upon deletion of CTCF sites from Shh, but it might be balanced to still reference the work from Uishiki et al that indicate in humans the CTCF site does play a role in Shh - ZRS communication. As the authors are doubtless aware, Andrey and colleagues show a CTCF dependent enhancement of a sensitized ZRS enhancer. Zuin et al. in an elegant experiment in which an enhancer is mobilized to different distances away from its promoter using transposon induction, reported a complete lack of detection of enhancers mobilizing outside the TAD to activate gene expression. A balanced presentation of the data on CTCF role might include some discussion of the above. In light of these earlier works, the findings the authors report about border bypass are all the more surprising.

   We thank the reviewer for highlighting some of these studies, especially for drawing our attention to the interesting recent preprint from Chakraborty et al. (doi.org/10.1101/2024.08.03.606480), which we now discuss in the revised manuscript. * As suggested by the reviewer, we now also cite Ushiki et al., 2021 in the Introduction in the context of CTCF-associated phenotypes, rather than just in the Discussion as in the original submission. We already cited the work of Andrey and colleagues (Paliou et al). However, we chose not to cite the Pennacchio study, because the deletions used were large – all >10kb and some as large as 80kb. Therefore, we consider it highly likely that other regulatory sequences beyond CTCF sites themselves may have been deleted, complicating conclusions drawn about the function of the TAD boundaries per se*. We have also chosen to focus our discussion on studies of enhancers in their native genomic locus, and predominantly in vivo analyses, rather than ectopic enhancer integrations (such as Zuin et al) in cell lines.

   1. By contrast, direct evidence for cross TAD interactions at endogenous loci has not to my knowledge been shown as clearly as described in the current manuscript. Recent work from Rocha and colleagues showed evidence that some enhancers upstream of Sox2 can pass ectopically induced boundaries. While recent work has described examples of 'TAD border bypass' at endogenous loci (e.g. for Pitx1 8, Hoxa regulation 9), these reports really just expand the view of regulatory boundaries rather than provide evidence against it. They invoke a 3D stacking of boundaries that allows boundary proximal enhancers and promoters to stack with (and so bypass) an intervening TAD boundary. Notably, in this view enhancers and promoters that lie away from the border of their respective TADs are still separate, and indeed intervening genes between distal enhancers for Pitx1 and Hoxa appear to follow these rules.2 Mnx1 and the Shh enhancers by contrast do not appear to be an example of border stacking. Given that Sox2 at least is also a TAD border, and the position of the bypassing enhancers is not precisely known in the work from Rocha, it is possible that that case is also an example of boundary stacking, which appears less likely in the case of Mnx1 (which does not appear to be at CTCF marked border, at least in mESCs).

   We thank the reviewer for highlighting some of these studies. We had already discussed the study from Rocha and colleagues (Chakraborty et al., 2023) and we had discussed the boundary stacking paper from Hung et al, (2024). However, based on the reviewer’s comment we now include a specific discussion about TAD boundary stacking and boundary proximal enhancer bypass, noting that Mnx1 is not close to a TAD boundary. This will become even more relevant in our planned revised manuscript where we will investigate possible Mnx1 activation by Shh enhancers (SBE2/3) located even further away from the Shh/Mnx1 TAD boundary.

   Statistics: Some of the bar graphs quantifying the %-expressing cells do not obviously have associated n-values, as are some of the violin plots of the distances. I think all these bar graphs could also benefit from adding error bars (e.g. by bootstrapping from the sampled population). This will help the reader more easily appreciate how sampling error and sample size affect the variation seen in the plots.

   We will add the n-values to all graphs. Regarding error bars, we think that showing the data from the two biological replicate separately is a better way to show the data reproducibility to the reader, than using boostrapping to estimate error bars.

   Figures 2 and 3: I would have preferred the authors zoom in more on the FISH spots to help the reader appreciate the proximity. I do appreciate also seeing a field of more than 1 cell (to give some sense of the variability), but these images mostly have only 1 spot pair per panel, which is exceedingly small as they contain parts of more than 1 nucleus. There is also unnecessary white space in this figure that could have been used to show zoom in panels.

   The same applies to the image panels in Figure 3 as for figure 2 - there is considerable unused whitespace, the image panels capture mostly a single nucleus and its pattern of DAPI dense heterochromatin (which isn't particularly relevant to the narrative) while the fluorescent spots that are the focus of the narrative are quite small. It is nice to have an example of the cell to see that this isn't just random background (that there is just one spot per cell) - in that sense though it's equally helpful to show its not just 1 cell in the field that has the signal-to-noise (SNR) shown. For this figure and the panels in figure 2, I'd recommend showing a zoom out showing ~3 nuclei with transcription foci (at least in the regions where the % transcribing is >60% it should be fine to have adjacent nuclei transcribing, for those where it is 10%, 1 of 3 nuclei transcribing in the image selected would also help get the sense of the data). These zoom out images would also give a sense of the SNR in the image, and then a zoom in where the FISH spots are sizable would make it easier to see the neighboring transcripts. Extended Data Fig 3 does a better job showing the context of the limb and then zooming in to an image where the RNA spots are appreciable. It looks like the resolution of the zoom in is lower, such that zooming in further on the spots in this data may not enhance the image.

   In response to the reviewer’s comment, we will present zoomed-out and zoomed-in images as suggested.

   1. Figure 3 - DNA FISH It would be helpful to include a diagram indicated where the DNA FISH probes are located on the genome and their size in kb as an inset in the figure.

   We will indicate the locations of DNA-FISH probes in a revised version of Figure 1a. Probe sizes are listed in the supplementary tables. We have now made this clearer in the legend to Figure 3.

   Reviewer 2

   The authors claim that co-expression of Mnx1 and Shh in the foregut and lung buds is also driven by boundary crossing contacts with the MACS1 enhancer. However, the effect of the boundary deletion on the co-transcription of Shh and Mnx1 is only showed for the ZPA. In this sense I find potentially misleading the statement of the authors in the following paragraph: "In the ZPA, the foregut, and the lung buds, the majority of Mnx1 RNA-FISH signals are at alleles that show simultaneous signal for Shh nascent transcript from the same allele (closely apposed signals) (Fig. 2a, b and Extended Data Fig. 2a). In del 35 embryos, an even higher proportion of Mnx1 transcribing alleles also transcribe Shh (Fig. 2b,Extended Data Fig. 2a, Extended Data Table 3.). These data suggest that both the ZRS and MACS1 enhancers are able to simultaneously activate transcription at two gene loci on the same chromosome". In my opinion this phrasing implicitly extends the increase in Mnx1-Shh co-expressing nuclei observed in the ZPA of 35 del embryos to the expression of these two genes in the foregut and lung buds (driven by the MACS1 enhancer) while this effect has not been specifically addressed. In a previous work, the authors showed that boundary deletion does not impact Mnx1 expression in the foregut and lungs. It would be important to clarify whether more precise analysis in this study have led to different conclusions or, alternatively, appropriately discuss the results. Ideally the authors should analyse the effect of the 35 del allele in the foregut / lung buds or rephrase the statement about the sharing of the MACS1 enhancer.

   The reviewer is correct that in our previous publication (Williamson et al., 2019) we did not detect Mnx1 expression in the lungs of 35kb del embryos. However, we only examined this by in situ hybridisation so we probably lacked the sensitivity to detect weak Mnx1 expression. In response to the reviewer’s comments, we now propose to do RNA FISH for transcription at Mnx1 in other tissues of 35kb del embryos.

   The authors use the quantifications of nuclei co-expressing Mnx1 and Shh from the same allele as an indicator of simultaneous transcription of the two genes by the sharing of the enhancer as opposed to a model of alternate transcriptional bursts. However, I am concerned that the time scale at which looping and transcriptional bursts occur is at odds with the detection of nascent transcription in FISH experiments, thus not excluding that shifting of the enhancer from one promoter to the other could still result in detection of nascent RNA of the two genes in the same allele. In any case, following the argumentation of the authors, the fraction of nuclei expressing Mnx1 alone does not appear to be significantly different from those expressing Mnx1 and Shh, and the increase of Mnx1 expressing nuclei upon boundary deletion seem proportionally similar to the increase of Mnx1+/Shh+ nuclei. In my opinion, this makes it difficult to interpret the detection of Mnx1 alone or both Mnx1-Shh expression as a reflection of alternate looping and transcriptional burst from enhancer sharing. Determining whether the two promoters compete for the interaction with the enhancer or share it would require estimate whether in the 35 del homozygote embryos Shh expression is reduced compared to wts, as a result of the increased interaction of the ZRS with the enhancer. The authors claim that there are no differences in the % of cells expressing Shh upon boundary deletion but in my opinion measurement is not sufficient to estimate a change in transcriptional rate (frequency of bursting). Nascent mRNA level detection in single cells would allow to better asses competition or concomitant activation of the two gene. Not being an expert in the RAN FISH technique it is not clear to me whether fluorescence intensity could be used as an estimator of transcription. From the images of the authors, in some cases it seems that expression of Shh alone is higher than when both Shh and Mnx1 are transcribed from the same allele (Fig. 2a, left panel, Fig 2c left vs right panel. However, in other cases an opposite trend can be observed (Mnx1 intensity in Fig2a central vs right panel). Thus, a single nuclei PCR or RNAseq approach may be more suited for this assessment.

   We respectfully disagree with the reviewer. We argue that nascent RNA FISH, using probe pools that for the most part detect the introns of Shh and Mnx1, is a better measure of transcription bursting/frequency (on or off) than probe signal intensity and therefore is a measurement of transcription rate. Single nuclei PCR or RNAseq would not assay nascent transcription and would not distinguish between alleles.

   Minor comments:

   1. In the mESC model overexpressing the tZRS-VP64 construct, Shh and Mnx1 seem to be transcribed at similar rates compared to what observed in vivo (where only a minor fraction of Shh+ cells express Mnx1). Thus, despite the fact that TAD boundary deletion increases Mnx1, but not Shh, expression, the ZRS activity seems to more easily overcome the border in this context than in vivo. Could the authors comment on this interesting observation? May it relate to the insulation score of TAD boundaries in the mESCs compared to in vivo? Alternatively, could it reflect that combinatorial TF binding to an enhancer contribute to its directionality?

   *These are interesting speculations by the reviewer, but we would argue that it is hard to compare in vivo and in vitro experiments. For example, in the limb bud, the ZPA region where the ZRS is active cannot be distinguished morphologically from the surrounding mesenchymal cells, therefore it is likely that some nuclei that are just outside the ZPA may be included in the analysis. *

   Overall, figure organization and clarity could be improved. For example, enlargement of RNA fish images in Fig. 1 could be enlarged (to the same size than the broad view image) and RNA FISH signal could be highlighted with arrowheads. Panel distribution could also be optimized.

   We will try to clarify these figures – see also response to reviewer 1 (point 6).

   Reviewer 3

   There are a couple of claims and conclusions that are not fully supported by the data, and which I think could be resolved by rephrasing them and/or qualify them as preliminary or speculative. The authors often indicate co-expression as suggestive of co-regulation by a single enhancer, when in most cases this is not formally shown; such suggestion remains one among other possibilities. For instance, co-expression of Shh and Mnx1 in the developing bud is attributed to the ZRS enhancer, co-expression of Shh and Mnx1 in the foregut is attributed to MACS1 enhancer. Do the authors have any evidence that when deleting these enhancers, Mnx1 expression is abolished (or reduced) in the respective tissues?

   If not, I think the following sentences need revision, because causality is implied by the way it is written but it is not formally shown (and the data could suggest other options too):

   "However, we have previously identified that ZRS can also drive low level expression of Mnx1, located 150kb away in the adjacent TAD, in the developing limb bud (Williamson et al., 2019)." No genetic evidence is provided in Williamson et al. 2019

   i) It is true that in Williamson et al., we did not provide genetic evidence that ZRS is the enhancer responsible for Mnx1 expression in the limb bud ZPA. However, there is no other known enhancer in biology with activity specific to the ZPA, and when the ZRS is deleted the ZPA no longer functions as a signaling centre for the limb bud. As a compromise, we have rephrased the indicated text to “However, we have previously identified that ZRS also appears to be able to drive low level expression of Mnx1, located 150kb away in the adjacent TAD, in the ZPA of the developing limb bud”.

   "However, we also detect nascent transcription from Mnx1 in the Shh expressing portions of the developing ventral foregut and the lung bud of E10.5 embryos, an activity that is driven by the Shh MACS1 enhancer, located a further 100kb into the Shh TAD from ZRS (Sagai et al., 2017) and therefore able to induce transcription at Mnx1 across a TAD boundary from a distance of >260 kb (Fig. 1a)."

   *ii) We have modified the text to now read “However, we also detect nascent transcription from Mnx1 in the Shh expressing portions of the developing ventral foregut and the lung bud of E10.5 embryos, an activity that is likely to bedriven by the Shh *MACS1 enhancer, located a further 100kb into the Shh TAD from ZRS”.

   "These data suggest that both the ZRS and MACS1 enhancers are able to simultaneously activate transcription at two gene loci on the same chromosome."

   iii) We have modified this statement to now read that these enhancers “may be able to simultaneously activate transcription at two gene loci on the same chromosome”.

   "This is the first report of two endogenous mammalian genes transcribed simultaneously under the control of the same enhancer" (can the authors really claim this without genetic evidence, i.e., deleting the enhancer? Isn't that the golden standard in the field?).

   iv) We stand by this claim, because we have been able to provide evidence in support of our observations in tissues, by using synthetic enhancer activation in cell culture where we can be absolutely be sure what the enhancer responsible for activation is.

   "Therefore, the Shh ZRS enhancer can simultaneously activate transcription at two genes and across an intact, but porous, TAD boundary. See response (iv) above

   "This is a consequence of ZRS-driven activation, not Mnx1 transcription per se."

   v) We stand by this claim.

   The mathematical model, even if simple, is very poorly described. In the results section, it is not easy to understand what the model takes into account, etc; it would be important for non-experts to understand as well what is at stake. In the methods section, it does not seem to be properly described; it is only stated "The association between the transcription of Shh and Mnx1 regulated by the same enhancer was done by linear modelling with binomial link function." Would this be enough to recreate / reproduce the same model? I am not a mathematician, but I suspect more details would be needed.

   *We apologize if our approach was not clear. We used logistic regression not a mathematical model. We have now expanded the relevant Methods section to now read: *

   “

   “To test whether or not there is a tendency of coexpression between two loci on the same chromatid, only nuclei with exactly one signal of each locus are informative. For these nuclei, we scored how many had expression in cis and how many in trans. To assess whether there was chromatid-specific coexpression, we tested statistically whether there was an excess of nuclei showing expression in cis. We did this using logistic regression, a form of generalized linear regression model. More specifically, we tested, for each model, whether the model intercept was significantly different from zero by using the z-scaled test statistic returned by these models and converting it to a p-value.”

   The authors claim that an enhancer working exclusively on one gene at a time would lead to a preference in individual expression - is this really the case? Could the authors show the expected scenarios for [one enhancer - two common targets] versus [two enhancers - two independent targets] and how this compares to the data?

   Our statistical analysis is restricted to the scenario of one enhancer acting on two genes (either simultaneously, or alternately). We do not test a two enhancers two target genes scenario because it is not relevant to our experimental analyses using synthetic activation of a single enhancer (with tZRS-Vp64, Extended Data Table 4).

   1. The results obtained with the VP64 activation (activation of ZRS leads to increased expression of Mnx1) are used by the authors as another piece of evidence that ZRS controls Mnx1 - but could VP64 activation be inducing chromatin opening / enhanced accessibility and therefore increased expression across the TAD boundary? I am not sure the authors need to test this, but they should at least acknowledge other possibilities (in relation to point 1).

   *We have previously shown (Benabdallah et al., 2019) that tal-VP64 activators alter chromatin structure (H3K27ac) in the Shh TAD only locally at the site of binding and at the Shh gene, and that this does not spread more generally. We have clarified this in the revised text. We also note that the effect of both the 35kb deletion and cohesin degradation on Mnx1 activation from the tZRSVp64 activator would not be consistent with a model of general chromatin opening/accessibility. The same argument applies to the DNA-FISH experiment (Fig 3) showing Mnx1 activation in the limb bud (ZPA) occurs specifically in the context of a compact chromatin conformation. *

   "In the nuclei of pre-motor neurons, where Mnx1 expression is driven from its own proximal enhancers (Fig. 1a), Mnx-ZRS and Mnx1-Shh distances are not different between Mnx1 expressing and non-expressing alleles." The authors use this as an argument to claim that Mnx1 expression per se does not explain the distance differences observed in the limb bud - but can such comparisons of expression and distances between loci be made between different cell types? Is there enough evidence for this to be a valid assumption? If not, then the assumption should be explicitly presented.

   We believe that the reviewer is confused here. We are not suggesting that Mnx1 expression per se doesn’t explain the distance differences in the limb bud, rather that these distance differences in the limb bud associated with Mnx1 transcription do not occur in the pre-motor neurons where activation is not dependent on distal enhancers, particularly in the Shh TAD.

   1. In Fig. 3b the authors show that shorter distances between the loci (Mnx1, Shh, ZRS) were associated with simultaneous transcription at Mnx1 and Shh, implying throughout that this would be associated with common activation by ZRS; but the shorter distances between the three loci are also associated with Mnx1 transcription alone. How is this explained?

   *This is explained by the configuration of the Shh TAD and the general spatial proximity of Shh-ZRS in both expressing and non-expressing tissues due to the CTCF-mediated loop and that is apparent in Hi-C heat maps. *

   1. The text could be revised to look out for "expression levels" versus "expression frequency" - in several instances the authors mention expression "levels" when they are referring to % of cells expressing a given gene, which would thus be more appropriate called "expression frequency"?

   The reviewer makes an important point. In the revised manuscript we have removed all mention of “expression levels” and have replaced these with “ frequency”.

   Read the original source
2. 

   Review Commons

   Jan 20, 2025

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   Summary:

   Williamson et al. set out to investigate in further detail the previously described co-expression of Shh and Mnx1 in the same cells of the developing mouse embryo (Williamson et al. 2019). The authors suspect that this co-expression is caused by the activity of a single enhancer (ZRS), which is intriguing because Shh and Mnx1 lie in adjacent TADs, i.e., there is a TAD boundary in between them, which in principle would prevent ZRS (within the same TAD as Shh) from activating Mnx1 expression. Using RNA-FISH in mouse embryo sections, the authors first confirmed the co-expression of Shh and Mnx1 in the developing bud, which was …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #3

   Evidence, reproducibility and clarity

   Summary:

   Williamson et al. set out to investigate in further detail the previously described co-expression of Shh and Mnx1 in the same cells of the developing mouse embryo (Williamson et al. 2019). The authors suspect that this co-expression is caused by the activity of a single enhancer (ZRS), which is intriguing because Shh and Mnx1 lie in adjacent TADs, i.e., there is a TAD boundary in between them, which in principle would prevent ZRS (within the same TAD as Shh) from activating Mnx1 expression. Using RNA-FISH in mouse embryo sections, the authors first confirmed the co-expression of Shh and Mnx1 in the developing bud, which was not observed in other tissues where Shh and Mnx1 are normally expressed (e.g., neural tube) - further suggesting the involvement of the limb-specific ZRS enhancer. Using mouse embryos harbouring a 35kb-deletion encompassing the TAD boundary between Shh and Mnx1, the authors observed by RNA-FISH that Mnx1 was more frequently expressed in the mutant limb buds, in accordance to their previous results via RT-qPCR (Williamson et al. 2019). Based on RNA FISH and a mathematical model, the authors conclude that Shh and Mnx1 are frequently expressed from the same allele, also supporting the idea of common cis-regulation. Using a TALE-based system in mESC, the authors induce specific recruitment of VP64 to ZRS and observe increased expression for both Shh and Mnx1. This effect was enhanced for Mnx1 (but not for Shh) when the TAD boundary was missing (in the context of the 35kb-deletion). Using DNA-FISH following RNA-FISH on the same samples, the authors were then able to correlate transcriptional states with distances between the loci. This analysis led them to show that shorter distances between the loci (Mnx1, Shh, ZRS) were associated with simultaneous transcription at Mnx1 and Shh (but also with Mnx1 transcription alone). Finally, the authors combine their synthetic recruitment of VP64 to ZRS with protein-degron systems for CTCF and cohesin. These investigations showed that CTCF degradation did not impact the VP64-induced expression of Shh nor of Mnx1. In contrast, acute depletion of cohesin led to impaired VP64-induced expression of both Shh nor of Mnx1. Using DNA FISH, the authors showed that this was correlated with increased Mnx1-ZRS and Shh-ZRS distances. Overall, the authors conclude that their data support a model by which ZRS regulates Mnx1 across a TAD boundary and in a cohesin-dependent manner.

   Major comments:

   1. There are a couple of claims and conclusions that are not fully supported by the data, and which I think could be resolved by rephrasing them and/or qualify them as preliminary or speculative. The authors often indicate co-expression as suggestive of co-regulation by a single enhancer, when in most cases this is not formally shown; such suggestion remains one among other possibilities. For instance, co-expression of Shh and Mnx1 in the developing bud is attributed to the ZRS enhancer, co-expression of Shh and Mnx1 in the foregut is attributed to MACS1 enhancer. Do the authors have any evidence that when deleting these enhancers, Mnx1 expression is abolished (or reduced) in the respective tissues?

   If not, I think the following sentences need revision, because causality is implied by the way it is written but it is not formally shown (and the data could suggest other options too):

   "However, we have previously identified that ZRS can also drive low level expression of Mnx1, located 150kb away in the adjacent TAD, in the developing limb bud (Williamson et al., 2019)." No genetic evidence is provided in Williamson et al. 2019

   "However, we also detect nascent transcription from Mnx1 in the Shh expressing portions of the developing ventral foregut and the lung bud of E10.5 embryos, an activity that is driven by the Shh MACS1 enhancer, located a further 100kb into the Shh TAD from ZRS (Sagai et al., 2017) and therefore able to induce transcription at Mnx1 across a TAD boundary from a distance of >260 kb (Fig. 1a)."

   "These data suggest that both the ZRS and MACS1 enhancers are able to simultaneously activate transcription at two gene loci on the same chromosome."

   "This is the first report of two endogenous mammalian genes transcribed simultaneously under the control of the same enhancer" (can the authors really claim this without genetic evidence, i.e., deleting the enhancer? Isn't that the golden standard in the field?)

   "Therefore, the Shh ZRS enhancer can simultaneously activate transcription at two genes and across an intact, but porous, TAD boundary.

   "This is a consequence of ZRS-driven activation, not Mnx1 transcription per se."

   1. The mathematical model, even if simple, is very poorly described. In the results section, it is not easy to understand what the model takes into account, etc; it would be important for non-experts to understand as well what is at stake. In the methods section, it does not seem to be properly described; it is only stated "The association between the transcription of Shh and Mnx1 regulated by the same enhancer was done by linear modelling with binomial link function." Would this be enough to recreate / reproduce the same model? I am not a mathematician, but I suspect more details would be needed.
   2. The authors claim that an enhancer working exclusively on one gene at a time would lead to a preference in individual expression - is this really the case? Could the authors show the expected scenarios for [one enhancer - two common targets] versus [two enhancers - two independent targets] and how this compares to the data?
   3. The results obtained with the VP64 activation (activation of ZRS leads to increased expression of Mnx1) are used by the authors as another piece of evidence that ZRS controls Mnx1 - but could VP64 activation be inducing chromatin opening / enhanced accessibility and therefore increased expression across the TAD boundary? I am not sure the authors need to test this, but they should at least acknowledge other possibilities (in relation to point 1)
   4. "In the nuclei of pre-motor neurons, where Mnx1 expression is driven from its own proximal enhancers (Fig. 1a), Mnx-ZRS and Mnx1-Shh distances are not different between Mnx1 expressing and non-expressing alleles." The authors use this as an argument to claim that Mnx1 expression per se does not explain the distance differences observed in the limb bud - but can such comparisons of expression and distances between loci be made between different cell types? Is there enough evidence for this to be a valid assumption? If not, then the assumption should be explicitly presented. 5.1/ In Fig. 3b the authors show that shorter distances between the loci (Mnx1, Shh, ZRS) were associated with simultaneous transcription at Mnx1 and Shh, implying throughout that this would be associated with common activation by ZRS; but the shorter distances between the three loci are also associated with Mnx1 transcription alone. How is this explained?
   5. The text could be revised to look out for "expression levels" versus "expression frequency" - in several instances the authors mention expression "levels" when they are referring to % of cells expressing a given gene, which would thus be more appropriate called "expression frequency"?

   Otherwise, I find the experimental and analytical work very solid.

   Minor comments:

   1. The authors claim that their study "is the first report of two endogenous mammalian genes transcribed simultaneously under the control of the same enhancer" - despite the fact that this claim is not formally demonstrated given the absence of genetic data (as mentioned above), it is still an intriguing idea that to date no such case has been formally demonstrated in mammals. I think the authors could emphasise this more.
   2. Cis-regulation across TAD borders has been suggested/demonstrated before, for both enhancers (two examples cited by the authors) and other types of elements (e.g., silencers). The authors might want to consider mentioning/discussing other previous studies that have either also suggested that TAD boundaries are not "absolute barriers" and/or found cis-regulatory elements working across TAD boundaries, such as those from the Andrey/Mundlos lab (Kragesteen et al. 2020), Heard lab (Galupa et al. 2020 and 2022), Ren lab (Diao et al. 2017), Rinn lab (Groff et al. 2018), Spitz lab (Tsujimura et al. 2015); the last one is cited by the authors but not in the context of inter-TAD communication. Some of these have been recently reviewed in Szalay et al. 2024 (PMID: 39592879).
   3. The authors state that "Conventional enhancer-promoter looping models predict that transcriptional bursts from two genes under the control of the same enhancer should not be coincident." Could the authors provide references for this statement, or explain this further?

   Significance

   This study set out to understand whether co-expression of Shh and Mnx1 in the same cells of the developing mouse embryo was due to regulation by a common enhancer (ZRS), located across a TAD boundary from Mnx1, which would strengthen a growing body of studies showing that TAD boundaries are not as impermeable as once thought and postulated. Despite lacking formal genetic evidence to fully support their hypothesis, this study will nevertheless be an important addition to the field of transcriptional regulation and 3D chromosome structure and to its specialized audience. My field of expertise: transcriptional regulation and 3D chromosome structure.

   Read the original source
3. 

   Review Commons

   Jan 20, 2025

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   In this work, Williamson L et al address two intriguing questions in the field of gene regulation and chromatin organization: whether enhancer activity can overcome TAD boundaries and whether regulatory elements- shared between two genes interact with them concomitantly or competitively. For that, they use as a paradigm the well characterized regulatory landscape of the mouse Shh locus. This gene is located in a large TAD containing enhancers for different embryonic structures, including the ZPA of the developing limbs, foregut, lung buds and floor plate of the embryonic neural tube. Mnx1, a gene located in the adjacent TAD near the …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #2

   Evidence, reproducibility and clarity

   In this work, Williamson L et al address two intriguing questions in the field of gene regulation and chromatin organization: whether enhancer activity can overcome TAD boundaries and whether regulatory elements- shared between two genes interact with them concomitantly or competitively. For that, they use as a paradigm the well characterized regulatory landscape of the mouse Shh locus. This gene is located in a large TAD containing enhancers for different embryonic structures, including the ZPA of the developing limbs, foregut, lung buds and floor plate of the embryonic neural tube. Mnx1, a gene located in the adjacent TAD near the boundary of the Shh TAD, is mainly expressed in differentiating motor neurons of the spinal cord. However, the authors observed that Mnx1 transcription overlpas that of Shh in the ZPS, lung and foregut.

   Combining high resolution RNA and DNA FISH experiments they demonstrate that: (a) Shh enhancers located near (up to few hundreds kb) from the TAD border can overcome TAD insulation and drive Mnx1 transcription, while enhancers located near the Mnx1 and Shh loci or more internally in the Shh TAD specifically activate only their respective loci, (b) Deletion of the TAD boundary increases the fraction of cells coexpressing Shh and Mnx1 in the ZPA, (c) Coactivation of Shh and Mnx1 occurs in cis (i.e. in the same allele) suggesting concomitant activity of the shared ZRS enhancer on the two promoters (d) the co-activation correlates with the tightening of the distances between Shh, Mnx1 and the ZRS enhancer and is dependent on the ZRS activity as shown by the overexpression of tZRS-VP64 in mouse ES cells, (e) Cohesin, but not CTCF activity is required for the looping of the ZRS element with Shh and Mnx1. The experiments are well designed and the findings provide important insights improving our understanding that TAD boundaries constitute somewhat permeable rather than absolute barriers, reinforcing previous evidences in the fields, although the functional significance of this permeability is not addressed in this work. There are two main points which, in my opinion, require to be addressed by the authors to improve the overall quality and clarity of the work:

   Major comments:

   • The authors claim that co-expression of Mnx1 and Shh in the foregut and lung buds is also driven by boundary crossing contacts with the MACS1 enhancer. However, the effect of the boundary deletion on the co-transcription of Shh and Mnx1 is only showed for the ZPA. In this sense I find potentially misleading the statement of the authors in the following paragraph: "In the ZPA, the foregut, and the lung buds, the majority of Mnx1 RNA-FISH signals are at alleles that show simultaneous signal for Shh nascent transcript from the same allele (closely apposed signals) (Fig. 2a, b and Extended Data Fig. 2a). In del 35 embryos, an even higher proportion of Mnx1 transcribing alleles also transcribe Shh (Fig. 2b,Extended Data Fig. 2a, Extended Data Table 3.). These data suggest that both the ZRS and MACS1 enhancers are able to simultaneously activate transcription at two gene loci on the same chromosome". In my opinion this phrasing implicitly extends the increase in Mnx1-Shh co-expressing nuclei observed in the ZPA of 35 del embryos to the expression of these two genes in the foregut and lung buds (driven by the MACS1 enhancer) while this effect has not been specifically addressed. In a previous work, the authors showed that boundary deletion does not impact Mnx1 expression in the foregut and lungs. It would be important to clarify whether more precise analysis in this study have led to different conclusions or, alternatively, appropriately discuss the results. Ideally the authors should analyse the effect of the 35 del allele in the foregut / lung buds or rephrase the statement about the sharing of the MACS1 enhancer.
   • The authors use the quantifications of nuclei co-expressing Mnx1 and Shh from the same allele as an indicator of simultaneous transcription of the two genes by the sharing of the enhancer as opposed to a model of alternate transcriptional bursts. However, I am concerned that the time scale at which looping and transcriptional bursts occur is at odds with the detection of nascent transcription in FISH experiments, thus not excluding that shifting of the enhancer from one promoter to the other could still result in detection of nascent RNA of the two genes in the same allele. In any case, following the argumentation of the authors, the fraction of nuclei expressing Mnx1 alone does not appear to be significantly different from those expressing Mnx1 and Shh, and the increase of Mnx1 expressing nuclei upon boundary deletion seem proportionally similar to the increase of Mnx1+/Shh+ nuclei. In my opinion, this makes it difficult to interpret the detection of Mnx1 alone or both Mnx1-Shh expression as a reflection of alternate looping and transcriptional burst from enhancer sharing. Determining whether the two promoters compete for the interaction with the enhancer or share it would require estimate whether in the 35 del homozygote embryos Shh expression is reduced compared to wts, as a result of the increased interaction of the ZRS with the enhancer. The authors claim that there are no differences in the % of cells expressing Shh upon boundary deletion but in my opinion measurement is not sufficient to estimate a change in transcriptional rate (frequency of bursting). Nascent mRNA level detection in single cells would allow to better asses competition or concomitant activation of the two gene. Not being an expert in the RAN FISH technique it is not clear to me whether fluorescence intensity could be used as an estimator of transcription. From the images of the authors, in some cases it seems that expression of Shh alone is higher than when both Shh and Mnx1 are transcribed from the same allele (Fig. 2a, left panel, Fig 2c left vs right panel ). However, in other cases an opposite trend can be observed (Mnx1 intensity in Fig2a central vs right panel). Thus, a single nuclei PCR or RNAseq approach may be more suited for this assessment.

   Minor comments:

   • In the mESC model overexpressing the tZRS-VP64 construct, Shh and Mnx1 seem to be transcribed at similar rates compared to what observed in vivo (where only a minor fraction of Shh+ cells express Mnx1). Thus, despite the fact that TAD boundary deletion increases Mnx1, but not Shh, expression, the ZRS activity seems to more easily overcome the border in this context than in vivo. Could the authors comment on this interesting observation? May it relate to the insulation score of TAD boundaries in the mESCs compared to in vivo? Alternatively, could it reflect that combinatorial TF binding to an enhancer contribute to its directionality?
   • Overall, figure organization and clarity could be improved. For example, enlargement of RNA fish images in Fig. 1 could be enlarged (to the same size than the broad view image) and RNA FISH signal could be highlighted with arrowheads. Panel distribution could also be optimized.

   Significance

   Significance. The work presented by Williamson L et al provide interesting insights on how TAD borders contribute to the insulation of topologically domains and restricting enhancer interactions, showing that some enhancers are able to overcome TAD insulation and showing that enhancer looping and TAD border crossing rely on enhancer activity and cohesin loop extrusion. As mentioned above, these findings reinforce and extend previous reports (Chakraborty et al 2023, Kessler S et al 2023, Balasubramanian et al 2024, Tzu-Chiao H. et al 2024,). This work does not specifically address whether the fact that their tested enhancer (really focusing on the ZRS enhancer) can overcome the Shh TAD boundary is dependent on their intrinsic properties (e.g. TFBS composition) or whether it relates with their distance to the border. This would require more complex genetic rearrangements (for example bringing floor plate enhancers in proximity of the border, and in combination with the TAD boundary deletion) and would significantly increase the scientific relevance of the work, yet at the expense of significant amount of work that could not be addressed in a reviewing process. In summary, the research of Williamsons l et al constitute an overall well performed piece of work that integrates well within other pieces of evidence of the field of gene regulation and chromatin organization. Thus, without constituting a major conceptual breakthrough in the field, it constitutes a valuable contribution to our understanding of basic principles of genome organization and gene transcription.

   Read the original source
4. 

   Review Commons

   Jan 20, 2025

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   Summary and significance in the context of the field:

   In this work, the authors conduct a detailed investigation of the 'ectopic'/'bystander' activation of the gene Mnx1 by enhancers of Shh, located in the neighboring TAD. TAD borders have been shown in a number of works to contribute to the remarkable specificity of enhancer-promoter choice, and the current dogma in the field is to view them as perfect boundaries to enhancer-promoter interaction. Notably, this current dogma also highlights a conundrum in our understanding of gene regulation, as available 3D genome data from both sequencing and microscopy show that TAD borders are …

   Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

   Learn more at Review Commons

   ---

   Referee #1

   Evidence, reproducibility and clarity

   Summary and significance in the context of the field:

   In this work, the authors conduct a detailed investigation of the 'ectopic'/'bystander' activation of the gene Mnx1 by enhancers of Shh, located in the neighboring TAD. TAD borders have been shown in a number of works to contribute to the remarkable specificity of enhancer-promoter choice, and the current dogma in the field is to view them as perfect boundaries to enhancer-promoter interaction. Notably, this current dogma also highlights a conundrum in our understanding of gene regulation, as available 3D genome data from both sequencing and microscopy show that TAD borders are regions of abrupt decrease in 3D proximity, but far from perfect borders, with numerous cross-TAD interactions detected by Hi-C and its variants and by single-cell microscopy (albeit fewer than the local intra-TAD interactions).

   The authors show convincing data that Mnx1 indeed responds transcriptionally to several Shh-enhancers located over 100 kb distal and on the wrong side of the TAD boundary. The data come from developing mouse embryos, span several tissues, and include key controls for specificity of the method. This provides convincing data with which to challenge the currently widely accepted view of as TADs a significant boundary, complimenting the few examples that indicate that such regulation is possible in special cases (see further discussion in 2b below). I believe this work represents an important and substantive contribution to the field and should ultimately be published, after a few notable issues have been addressed.

   Major comments:

   Does the CTCF degron substantially remove CTCF from the Mnx1/Shh TAD border?
   In prior AID-CTCF degron studies 1,2, a considerable fraction of cohesin dependent TAD borders are retained upon CTCF removal. Moreover, CTCF sites at these retained borders still have clear ChIP-seq peaks - even though the protein is >95% depleted and scarcely detectable by western. Thus, while I suspect that the authors are correct that the shorter distance of the 35 kb border deletion contributes substantially to the increased crosstalk between the Mnx1 and Shh-enhancers, I suspect part of the reason for a lack of a similar effect in the CTCF degron is due to the known challenges in removing CTCF from this border. To argue that the border but not the CTCF is important, I think it would be helpful to show the CTCF signal is sufficiently lost in the degron by ChIP-seq and/or show that this TAD border has been lost by Hi-C. Alternatively, the authors could tone down this claim to something more conservative, as I did not find it to be presented as a key conclusion of the paper as a whole.

   Minor comments:

   I believe the manuscript could be strengthened by some textual revisions of the introduction: 2a) In particular, in my opinion, the authors' description of existing data for the importance of TAD borders in enhancer promoter regulation is not described in a sufficiently balanced and complete manner, and overall impression given by the text is that CTCF marked borders have little serious evidence for a role in developmental enhancer specificity and are maybe a cancer thing. This is doubly unfortunate, as it undermines the impact of the authors work in expanding our view of what TAD borders are in a regulatory sense, as well as presents an unbalanced view of work in the field. This is of course easily corrected. In particular I recommend the following revisions:

   It is " depletion of CTCF has only a small effect on transcription in cell culture (Nora et al., 2017; Hsieh et al., 2022)." It should be clarified that there is only a small *acute * effect on transcription (in the first 6-12 hours), which may tell us more about the timescale at which promoters sample, integrate and respond to changes in their enhancer environment than about the roles of CTCF particularly. Notably, this degradation is lethal, it results in massive changes in transcription after 4 days, and I suspect the authors agree that this lethal affect arises from CTCF's role in transcription regulation (if you remove some key cytoskeletal protein or metabolic enzyme the primary cause of cell death is not transcriptional, but almost all the evidence for CTCF's vital role in the cell is linked in one way or another to transcription). The discussion of TAD border deletions is more one-sided than ideal. I appreciate the discussion is usually even more unbalanced when presenting the opposite view in the literature - many works only cite the examples where border deletion does lead to ectopic expression and phenotypes. The current text presented a subset of these border deletion data in such a way as to give me the impression the authors are deeply skeptical that CTCF plays a role as an insulator of E-P interactions in a developmental context (rather than just as a weird cancer thing). For example:

   Pennacchio's lab has analyzed a series of TAD border deletions with more examples of both lethal effects and effects with no apparent phenotype 3

   Deletion of TAD borders upstream of the FGF3/4/15 locus in mouse is embryonic lethal (particularly the border Kim et al label TB1 and didn't delete in their cancer model). https://www.biorxiv.org/content/10.1101/2024.08.03.606480v1

   I appreciate that Bickmore and colleagues found quite phenotypically normal mice upon deletion of CTCF sites from Shh, but it might be balanced to still reference the work from Uishiki et al that indicate in humans the CTCF site does play a role in Shh - ZRS communication: 4

   As the authors are doubtless aware, Andrey and colleagues show a CTCF dependent enhancement of a sensitized ZRS enhancer. 5

   Zuin et al. in an elegant experiment in which an enhancer is mobilized to different distances away from its promoter using transposon induction, reported a complete lack of detection of enhancers mobilizing outside the TAD to activate gene expression 6.

   A balanced presentation of the data on CTCF role might include some discussion of the above. In light of these earlier works, the findings the authors report about border bypass are all the more surprising.

   2b) By contrast, direct evidence for cross TAD interactions at endogenous loci has not to my knowledge been shown as clearly as described in the current manuscript.

   Recent work from Rocha and colleagues 7 showed evidence that some enhancers upstream of Sox2 can pass ectopically induced boundaries. While recent work has described examples of 'TAD border bypass' at endogenous loci (e.g. for Pitx1 8, Hoxa regulation 9), these reports really just expand the view of regulatory boundaries rather than provide evidence against it. They invoke a 3D stacking of boundaries that allows boundary proximal enhancers and promoters to stack with (and so bypass) an intervening TAD boundary. Notably, in this view enhancers and promoters that lie away from the border of their respective TADs are still separate, and indeed intervening genes between distal enhancers for Pitx1 and Hoxa appear to follow these rules.2 Mnx1 and the Shh enhancers by contrast do not appear to be an example of border stacking. Given that Sox2 at least is also a TAD border, and the position of the bypassing enhancers is not precisely known in the work from Rocha, it is possible that that case is also an example of boundary stacking, which appears less likely in the case of Mnx1 (which does not appear to be at CTCF marked border, at least in mESCs).

   Statistics

   Some of the bar graphs quantifying the %-expressing cells do not obviously have associated n-values, as are some of the violin plots of the distances. I think all these bar graphs could also benefit from adding errorbars (e.g. by bootstrapping from the sampled population). This will help the reader more easily appreciate how sampling error and sample size affect the variation seen in the plots.

   Recommendations for improving the figures

   Figure 2

   I would have preferred the authors zoom in more on the FISH spots to help the reader appreciate the proximity. I do appreciate also seeing a field of more than 1 cell (to give some sense of the variability), but these images mostly have only 1 spot pair per panel, which is exceedingly small as they contain parts of more than 1 nucleus. There is also unnecessary white space in this figure that could have been used to show zoom in panels.

   Figure 3 -image panels

   The same applies to the image panels in this figure as for figure 2 - there is considerable unused whitespace, the image panels capture mostly a single nucleus and its pattern of DAPI dense heterochromatin (which isn't particularly relevant to the narrative) while the fluorescent spots that are the focus of the narrative are quite small. It is nice to have an example of the cell to see that this isn't just random background (that there is just one spot per cell) - in that sense though it's equally helpful to show its not just 1 cell in the field that has the signal-to-noise (SNR) shown.
   For this figure and the panels in figure 2, I'd recommend showing a zoom out showing ~3 nuclei with transcription foci (at least in the regions where the % transcribing is >60% it should be fine to have adjacent nuclei transcribing, for those where it is 10%, 1 of 3 nuclei transcribing in the image selected would also help get the sense of the data). These zoom out images would also give a sense of the SNR in the image, and then a zoom in where the FISH spots are sizable would make it easier to see the neighboring transcripts. Extended Data Fig 3 does a better job showing the context of the limb and then zooming in to an image where the RNA spots are appreciable. It looks like the resolution of the zoom in is lower, such that zooming in further on the spots in this data may not enhance the image.

   Figure 3 - DNA FISH

   It would be helpful to include a diagram indicated where the DNA FISH probes are located on the genome and their size in kb as an inset in the figure.

   References cited above

   1. Nora, E. P., Goloborodko, A., Valton, A.-L., Gibcus, J. H., Uebersohn, A., Abdennur, N., Dekker, J., Mirny, L. A. & Bruneau, B. G. Targeted Degradation of CTCF Decouples Local Insulation of Chromosome Domains from Genomic Compartmentalization. Cell 169, 930-944.e22 (2017).
   2. Kubo, N., Ishii, H., Gorkin, D., Meitinger, F., Xiong, X., Fang, R., Liu, T., Ye, Z., Li, B., Dixon, J., Desai, A., Zhao, H. & Ren, B. Preservation of Chromatin Organization after Acute Loss of CTCF in Mouse Embryonic Stem Cells. bioRxiv 118737 (2017).
   3. Rajderkar, S., Barozzi, I., Zhu, Y., Hu, R., Zhang, Y., Li, B., Alcaina Caro, A., Fukuda-Yuzawa, Y., Kelman, G., Akeza, A., Blow, M. J., Pham, Q., Harrington, A. N., Godoy, J., Meky, E. M., von Maydell, K., Hunter, R. D., Akiyama, J. A., Novak, C. S., Plajzer-Frick, I., Afzal, V., Tran, S., Lopez-Rios, J., Talkowski, M. E., Lloyd, K. C. K., Ren, B., Dickel, D. E., Visel, A. & Pennacchio, L. A. Topologically associating domain boundaries are required for normal genome function. Commun. Biol. 6, 435 (2023).
   4. Ushiki, A., Zhang, Y., Xiong, C., Zhao, J., Georgakopoulos-Soares, I., Kane, L., Jamieson, K., Bamshad, M. J., Nickerson, D. A., University of Washington Center for Mendelian Genomics, Shen, Y., Lettice, L. A., Silveira-Lucas, E. L., Petit, F. & Ahituv, N. Deletion of CTCF sites in the SHH locus alters enhancer-promoter interactions and leads to acheiropodia. Nat. Commun. 12, 2282 (2021).
   5. Paliou, C., Guckelberger, P., Schöpflin, R., Heinrich, V., Esposito, A., Chiariello, A. M., Bianco, S., Annunziatella, C., Helmuth, J., Haas, S., Jerković, I., Brieske, N., Wittler, L., Timmermann, B., Nicodemi, M., Vingron, M., Mundlos, S. & Andrey, G. Preformed chromatin topology assists transcriptional robustness of Shh during limb development. Proc. Natl. Acad. Sci. U. S. A. 116, 12390-12399 (2019).
   6. Zuin, J., Roth, G., Zhan, Y., Cramard, J., Redolfi, J., Piskadlo, E., Mach, P., Kryzhanovska, M., Tihanyi, G., Kohler, H., Eder, M., Leemans, C., van Steensel, B., Meister, P., Smallwood, S. & Giorgetti, L. Nonlinear control of transcription through enhancer-promoter interactions. Nature 604, 571-577 (2022).
   7. Chakraborty, S., Kopitchinski, N., Zuo, Z., Eraso, A., Awasthi, P., Chari, R., Mitra, A., Tobias, I. C., Moorthy, S. D., Dale, R. K., Mitchell, J. A., Petros, T. J. & Rocha, P. P. Enhancer-promoter interactions can bypass CTCF-mediated boundaries and contribute to phenotypic robustness. Nat. Genet. 55, 280-290 (2023).
   8. Hung, T.-C., Kingsley, D. M. & Boettiger, A. N. Boundary stacking interactions enable cross-TAD enhancer-promoter communication during limb development. Nat. Genet. 56, 306-314 (2024).
   9. Hafner, A., Park, M., Berger, S. E., Murphy, S. E., Nora, E. P. & Boettiger, A. N. Loop stacking organizes genome folding from TADs to chromosomes. Mol. Cell 83, 1377-1392.e6 (2023).

   Significance

   The authors show convincing data that Mnx1 indeed responds transcriptionally to several Shh-enhancers located over 100 kb distal and on the wrong side of the TAD boundary. The data come from developing mouse embryos, span several tissues, and include key controls for specificity of the method. This provides convincing data with which to challenge the currently widely accepted view of as TADs a significant boundary, complimenting the few examples that indicate that such regulation is possible in special cases (see further discussion in 2b below). I believe this work represents an important and substantive contribution to the field and should ultimately be published, after a few notable issues have been addressed.

   Audience: I believe this work will be of general interest to the eukaryotic transcription community, the 4D genome community, and the developmental biology community.

   My expertise: developmental biology, 4D genome biology, microscopy

   Read the original source
5. 

   Version published to 10.1101/2024.11.01.621524 on bioRxiv

   Nov 3, 2024