Acute Activation of Genes Through Transcriptional Condensates Impact Non-target Genes in a Chromatin Domain
Curation statements for this article:-
Curated by eLife
eLife Assessment
The authors use single molecule imaging and in vivo loop-capture genomic approaches to investigate estrogen mediated enhancer-target gene activation in human cancer cells. These potentially important results suggest that ER-alpha can, in a temporal delay, activate a non-target gene TFF3, which is in proximity to the main target gene TFF1, even though the estrogen responsive enhancer does not loop with the TFF3 promoter. To explain these results, the authors invoke a transcriptional condensate model. The reviewers were split on the strength and interpretation of the evidence presented, which is considered incomplete at this stage. We encourage a revision which buttresses the findings with additional control experiments and careful consideration of alternative explanations and mathematical models. Further, the depth of the discussion on existing literature could be improved. This work will be of interest to those studying transcriptional gene regulation and hormone-aggravated cancers.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Transcription activation of genes by estrogens is driven by enhancers, which are often located within the same Topologically Associating Domain (TAD) as non-targeted promoters. We investigated how acute enhancer-driven activation affects neighbouring non-target genes within the same TAD. Using single-molecule RNA FISH (smFISH), we tracked the transcription of TFF1 (enhancer-targeted) and TFF3 (non-targeted) during estrogen stimulation. We observed mutually exclusive expression patterns: TFF1 expression peaked at 1 hour, while TFF3 reached its peak at 3 hours, after TFF1 ’s activation had diminished. Chromatin looping data indicated that the enhancer loops with TFF1 but not TFF3 , suggesting that TFF3 upregulation is not due to direct enhancer-promoter interactions. CRISPR deletion of the TFF1 enhancer and 1,6-hexanediol (HD) exposure revealed that the TFF1 enhancer:promoter undergo Liquid-Liquid Phase Separation (LLPS), which sequesters the transcriptional machinery and inhibits TFF3 expression. As estrogen signalling wanes or LLPS is disrupted, TFF1 expression declines while TFF3 expression increases. Our findings reveal that enhancer-driven activation can indirectly influence neighbouring genes, highlighting a dynamic shift in gene expression as signalling progresses.
Article activity feed
-
-
-
eLife Assessment
The authors use single molecule imaging and in vivo loop-capture genomic approaches to investigate estrogen mediated enhancer-target gene activation in human cancer cells. These potentially important results suggest that ER-alpha can, in a temporal delay, activate a non-target gene TFF3, which is in proximity to the main target gene TFF1, even though the estrogen responsive enhancer does not loop with the TFF3 promoter. To explain these results, the authors invoke a transcriptional condensate model. The reviewers were split on the strength and interpretation of the evidence presented, which is considered incomplete at this stage. We encourage a revision which buttresses the findings with additional control experiments and careful consideration of alternative explanations and mathematical models. Further, the depth of …
eLife Assessment
The authors use single molecule imaging and in vivo loop-capture genomic approaches to investigate estrogen mediated enhancer-target gene activation in human cancer cells. These potentially important results suggest that ER-alpha can, in a temporal delay, activate a non-target gene TFF3, which is in proximity to the main target gene TFF1, even though the estrogen responsive enhancer does not loop with the TFF3 promoter. To explain these results, the authors invoke a transcriptional condensate model. The reviewers were split on the strength and interpretation of the evidence presented, which is considered incomplete at this stage. We encourage a revision which buttresses the findings with additional control experiments and careful consideration of alternative explanations and mathematical models. Further, the depth of the discussion on existing literature could be improved. This work will be of interest to those studying transcriptional gene regulation and hormone-aggravated cancers.
-
Reviewer #1 (Public review):
Summary:
The manuscript by Bohra et al. describes the indirect effects of ligand-dependent gene activation on neighboring non-target genes. The authors utilized single-molecule RNA-FISH (targeting both mature and intronic regions), 4C-seq, and enhancer deletions to demonstrate that the non-enhancer-targeted gene TFF3, located in the same TAD as the target gene TFF1, alters its expression when TFF1 expression declines at the end of the estrogen signaling peak. Since the enhancer does not loop with TFF3, the authors conclude that mechanisms other than estrogen receptor or enhancer-driven induction are responsible for TFF3 expression. Moreover, ERα intensity correlations show that both high and low levels of ERα are unfavorable for TFF1 expression. The ERa level correlations are further supported by …
Reviewer #1 (Public review):
Summary:
The manuscript by Bohra et al. describes the indirect effects of ligand-dependent gene activation on neighboring non-target genes. The authors utilized single-molecule RNA-FISH (targeting both mature and intronic regions), 4C-seq, and enhancer deletions to demonstrate that the non-enhancer-targeted gene TFF3, located in the same TAD as the target gene TFF1, alters its expression when TFF1 expression declines at the end of the estrogen signaling peak. Since the enhancer does not loop with TFF3, the authors conclude that mechanisms other than estrogen receptor or enhancer-driven induction are responsible for TFF3 expression. Moreover, ERα intensity correlations show that both high and low levels of ERα are unfavorable for TFF1 expression. The ERa level correlations are further supported by overexpression of GFP-ERa. The authors conclude that transcriptional machinery used by TFF1 for its acute activation can negatively impact the TFF3 at peak of signaling but once, the condensate dissolves, TFF3 benefits from it for its low expression.
Strengths:
The findings are indeed intriguing. The authors have maintained appropriate experimental controls, and their conclusions are well-supported by the data.
Weaknesses:
There are some major and minor concerns that related to approach, data presentation and discussion. But I think they can be fixed with more efforts.
-
Reviewer #2 (Public review):
Summary:
In this manuscript by Bohra et al., the authors use the well-established estrogen response in MCF7 cells to interrogate the role of genome architecture, enhancers, and estrogen receptor concentration in transcriptional regulation. They propose there is competition between the genes TFF1 and TFF3 which is mediated by transcriptional condensates. This reviewer does not find these claims persuasive as presented. Moreover, the results are not placed in the context of current knowledge.
Strengths:
High level of ERalpha expression seems to diminish the transcriptional response. Thus, the results in Fig. 4 have potential insight into ER-mediated transcription. Yet, this observation is not pursued in great depth however, for example with mutagenesis of ERalpha. However, this phenomenon - which falls under …
Reviewer #2 (Public review):
Summary:
In this manuscript by Bohra et al., the authors use the well-established estrogen response in MCF7 cells to interrogate the role of genome architecture, enhancers, and estrogen receptor concentration in transcriptional regulation. They propose there is competition between the genes TFF1 and TFF3 which is mediated by transcriptional condensates. This reviewer does not find these claims persuasive as presented. Moreover, the results are not placed in the context of current knowledge.
Strengths:
High level of ERalpha expression seems to diminish the transcriptional response. Thus, the results in Fig. 4 have potential insight into ER-mediated transcription. Yet, this observation is not pursued in great depth however, for example with mutagenesis of ERalpha. However, this phenomenon - which falls under the general description of non monotonic dose response - is treated at great depth in the literature (i.e. PMID: 22419778). For example, the result the authors describe in Fig. 4 has been reported and in fact mathematically modeled in PMID 23134774. One possible avenue for improving this paper would be to dig into this result at the single-cell level using deletion mutants of ERalpha or by perturbing co-activators.
Weaknesses:
There are concerns with the smRNA FISH experiments. It is highly unusual to see so much intronic signal away from the site of transcription (Fig. 2) (PMID: 27932455, 30554876) which suggests to me the authors are carrying out incorrect thresholding or have a substantial amount of labeling background. The Cote paper cited in the manuscript is likewise inconsistent with their findings and is cited in a misleading manner: they see splicing within a very small region away from the site of transcription.
One substantial way to improve the manuscript is to take a careful look at previous single cell analysis of the estrogen response, which in some cases has been done on the exact same genes (PMID: 29476006, 35081348, 30554876, 31930333). In some of these cases, the authors reach different conclusions than those presented in the present manuscript. Likewise, there have been more than a few studies which characterized these enhancers (the first one I know of is: PMID 18728018). Also, Oh et al. 2021 (cited in the manuscript) did show an interaction between TFF1e and TFF3, which seems to contradict the conclusion from Fig. 3. In summary, the results of this paper are not in dialog with the field, which is a major shortcoming.
In the opinion of this reviewer, there are few - if any - experiments to interrogate the existence of LLPS for diffraction limited spots such as those associated with transcription. This difficulty is a general problem with the field and not specific to the present manuscript. For example, transient binding will also appear as a dynamic 'spot' in the nucleus, independently of any higher order interactions. As for Fig. 5, I don't think treating cells with 1,6 hexanediol is any longer considered a credible experiment. For example, there are profound effects on chromatin independent of changes in LLPS (PMID: 33536240).
Summary:
In conclusion, I suggest that the authors look at alternative explanations and analyses -- many of which are experimentally and mathematically rigorous and pre-date the condensate model -- to explain their data.
-
