Synthetic analysis of chromatin tracing and live-cell imaging indicates pervasive spatial coupling between genes
Listed in
- Evaluated articles (eLife)
- Structural Biology and Molecular Biophysics (eLife)
Abstract
The role of the spatial organization of chromosomes in directing transcription remains an outstanding question in gene regulation. Here, we analyze two recent single-cell imaging methodologies applied across hundreds of genes to systematically analyze the contribution of chromosome conformation to transcriptional regulation. Those methodologies are: 1) single-cell chromatin tracing with super-resolution imaging in fixed cells; 2) high throughput labeling and imaging of nascent RNA in living cells. Specifically, we determine the contribution of physical distance to the coordination of transcriptional bursts. We find that individual genes adopt a constrained conformation and reposition toward the centroid of the surrounding chromatin upon activation. Leveraging the variability in distance inherent in single-cell imaging, we show that physical distance – but not genomic distance – between genes on individual chromosomes is the major factor driving co-bursting. By combining this analysis with live-cell imaging, we arrive at a corrected transcriptional correlation of ϕ ≈ 0.3 for genes separated by < 400 nm. We propose that this surprisingly large correlation represents a physical property of human chromosomes and establishes a benchmark for future experimental studies.
Article activity feed
-
Author Response
Reviewer #1 (Public Review):
The authors use their expertise in live-cell imaging and mathematical modeling to further explore the relationship between chromatin structure, gene positioning and transcriptional coregulation. One of the strengths of the manuscript arises from the authors analysis of two publicly available datasets encompassing chromatin tracing and transcriptional activity. Using spatial analysis and modeling, the authors have impressively extended the findings of Su et. al, Cell 2020, who generated the analyzed dataset. A number of important concepts were explored including 1.) do genes re-position upon activation and 2.) can spatial proximity be correlated with transcriptional co-regulation. In general the authors conclusions are supported by their findings and should provide a blueprint for analysis …
Author Response
Reviewer #1 (Public Review):
The authors use their expertise in live-cell imaging and mathematical modeling to further explore the relationship between chromatin structure, gene positioning and transcriptional coregulation. One of the strengths of the manuscript arises from the authors analysis of two publicly available datasets encompassing chromatin tracing and transcriptional activity. Using spatial analysis and modeling, the authors have impressively extended the findings of Su et. al, Cell 2020, who generated the analyzed dataset. A number of important concepts were explored including 1.) do genes re-position upon activation and 2.) can spatial proximity be correlated with transcriptional co-regulation. In general the authors conclusions are supported by their findings and should provide a blueprint for analysis of additional related big imaging datasets in the future.
However there are a number of weaknesses including lack of statistical analysis or incomplete description (e.g. bootstrapping parameters, statistical methods, number of genes/cells/measurements, etc.) on some figures that make it difficult to interpret the significance of the trends. In addition, the modeling using live-cell studies is generalized based on a behavior (e.g. diffusion) of a single gene. The manuscript is densely written in a way that may be inaccessible for non-specialists. A final schematic model that summarizes biological findings would help alleviate this weakness.
We are glad that the reviewer considers the observed phenomenon important and that our overall findings are consistent with our results. We implemented changes in response to each of the above requests:
we added additional explanation of test statistics;
we analyzed diffusion of additional genes;
we tried to simplify the text;
we added a final schematic.
Reviewer #2 (Public Review):
In their manuscript, Bohrer and Larson reanalyse previously published imaging datasets in order to tackle a long-standing question in modern genome biology: does the physical proximity of transcribed genes correlate with their co-expression?
The authors start off by reanalysing fixed-cell data, in which they find that active genes (i.e., any gene with RNA FISH signal) often repositions towards the centroid of the imaged chromatin environment one transcriptionally active. The analysis is straightforward, but the notion of "closer to the centroid" remains a bit vague to me, and is not well defined as regards its functional significance. There is no doubt of the clear trend in the analysed data -- but the interpretation could be strengthened.
We tried to clarify this part of the text and also added a schematic illustration to the discussion to help clarify this important point (Fig. 5).
Then, using the same dataset, the question on physical gene proximity is addressed. This is not only an important and timely question, but also one which the authors address very nicely. They deduce that when a pair of loci are brought within sufficiently low physical 3D proximity (unrelated to their genomic distance) they are more likely than expected to be co-expressed. In cis, this distance can be defined to approx. <2.5 Mb of genomic separation. They also looked in trans, via a complex transfer of knowledge from live-cell imaging to the fixed-cell dataset, to show that genes brought within approx. 400 nm from one another display quite a high coexpression correlation. Despite the parsimonious nature of the model and assumptions that the authors use for this (testing more complex parameters might prove beneficial here), their postulations can quite adequately explain observations published by others that were previously left largely without interpretation.
In my opinion, the main strength of this manuscript lies with the initial analysis of the fixed-cell data and the clear trends therein. The latter part, which nicely identifies caveats in available data and analyses and which makes a solid effort to combine live-cell with fixed-cell data, leaves more scenarios to be tested. Nevertheless, based on the outcome of this analysis (mostly found in Fig. 4), the value of ~400 nm as a physical proximity cutoff for co-expression is reasonable (based on previous knowledge) and does provide a solid first step in the direction of deciphering the rules that allow coordinated gene expression in mammalian cells.
We agree that the modelling section is more of a first step and that future work will need to be done to investigate further. In the revision, we make this point explicit within the main text (See below).
Overall, this is a conceptual advance of merit that can re-shape ways of approaching the stillopen issue of gene co-bursting in light of novel (mostly imaging) technologies.
We appreciate the comment.
-
eLife Assessment:
The authors use their expertise in live-cell imaging and mathematical modeling to explore the relationship between chromatin structure, gene positioning and transcriptional co-regulation, using two publicly available datasets encompassing chromatin tracing and transcriptional activity. The resulting analysis reveals a weak association between transcription and proximity, but needs more statistical validation to strengthen the validity of the conclusions. With some clarifications and revisions, several findings, such as coupling of spatiotemporal positioning with activity, in-depth analysis of existing imaging/ChIP-seq datasets, could make this work impactful to both specialists and non-specialists.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive …
eLife Assessment:
The authors use their expertise in live-cell imaging and mathematical modeling to explore the relationship between chromatin structure, gene positioning and transcriptional co-regulation, using two publicly available datasets encompassing chromatin tracing and transcriptional activity. The resulting analysis reveals a weak association between transcription and proximity, but needs more statistical validation to strengthen the validity of the conclusions. With some clarifications and revisions, several findings, such as coupling of spatiotemporal positioning with activity, in-depth analysis of existing imaging/ChIP-seq datasets, could make this work impactful to both specialists and non-specialists.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)
-
Reviewer #1 (Public Review):
The authors use their expertise in live-cell imaging and mathematical modeling to further explore the relationship between chromatin structure, gene positioning and transcriptional co-regulation. One of the strengths of the manuscript arises from the authors analysis of two publicly available datasets encompassing chromatin tracing and transcriptional activity. Using spatial analysis and modeling, the authors have impressively extended the findings of Su et. al, Cell 2020, who generated the analyzed dataset. A number of important concepts were explored including 1.) do genes re-position upon activation and 2.) can spatial proximity be correlated with transcriptional co-regulation. In general the authors conclusions are supported by their findings and should provide a blueprint for analysis of additional …
Reviewer #1 (Public Review):
The authors use their expertise in live-cell imaging and mathematical modeling to further explore the relationship between chromatin structure, gene positioning and transcriptional co-regulation. One of the strengths of the manuscript arises from the authors analysis of two publicly available datasets encompassing chromatin tracing and transcriptional activity. Using spatial analysis and modeling, the authors have impressively extended the findings of Su et. al, Cell 2020, who generated the analyzed dataset. A number of important concepts were explored including 1.) do genes re-position upon activation and 2.) can spatial proximity be correlated with transcriptional co-regulation. In general the authors conclusions are supported by their findings and should provide a blueprint for analysis of additional related big imaging datasets in the future.
However there are a number of weaknesses including lack of statistical analysis or incomplete description (e.g. bootstrapping parameters, statistical methods, number of genes/cells/measurements, etc.) on some figures that make it difficult to interpret the significance of the trends. In addition, the modeling using live-cell studies is generalized based on a behavior (e.g. diffusion) of a single gene. The manuscript is densely written in a way that may be inaccessible for non-specialists. A final schematic model that summarizes biological findings would help alleviate this weakness.
-
Reviewer #2 (Public Review):
In their manuscript, Bohrer and Larson reanalyse previously published imaging datasets in order to tackle a long-standing question in modern genome biology: does the physical proximity of transcribed genes correlate with their co-expression?
The authors start off by reanalysing fixed-cell data, in which they find that active genes (i.e., any gene with RNA FISH signal) often repositions towards the centroid of the imaged chromatin environment one transcriptionally active. The analysis is straightforward, but the notion of "closer to the centroid" remains a bit vague to me, and is not well defined as regards its functional significance. There is no doubt of the clear trend in the analysed data -- but the interpretation could be strengthened.
Then, using the same dataset, the question on physical gene proximity …
Reviewer #2 (Public Review):
In their manuscript, Bohrer and Larson reanalyse previously published imaging datasets in order to tackle a long-standing question in modern genome biology: does the physical proximity of transcribed genes correlate with their co-expression?
The authors start off by reanalysing fixed-cell data, in which they find that active genes (i.e., any gene with RNA FISH signal) often repositions towards the centroid of the imaged chromatin environment one transcriptionally active. The analysis is straightforward, but the notion of "closer to the centroid" remains a bit vague to me, and is not well defined as regards its functional significance. There is no doubt of the clear trend in the analysed data -- but the interpretation could be strengthened.
Then, using the same dataset, the question on physical gene proximity is addressed. This is not only an important and timely question, but also one which the authors address very nicely. They deduce that when a pair of loci are brought within sufficiently low physical 3D proximity (unrelated to their genomic distance) they are more likely than expected to be co-expressed. In cis, this distance can be defined to approx. <2.5 Mb of genomic separation. They also looked in trans, via a complex transfer of knowledge from live-cell imaging to the fixed-cell dataset, to show that genes brought within approx. 400 nm from one another display quite a high co-expression correlation. Despite the parsimonious nature of the model and assumptions that the authors use for this (testing more complex parameters might prove beneficial here), their postulations can quite adequately explain observations published by others that were previously left largely without interpretation.
In my opinion, the main strength of this manuscript lies with the initial analysis of the fixed-cell data and the clear trends therein. The latter part, which nicely identifies caveats in available data and analyses and which makes a solid effort to combine live-cell with fixed-cell data, leaves more scenarios to be tested. Nevertheless, based on the outcome of this analysis (mostly found in Fig. 4), the value of ~400 nm as a physical proximity cutoff for co-expression is reasonable (based on previous knowledge) and does provide a solid first step in the direction of deciphering the rules that allow coordinated gene expression in mammalian cells.
Overall, this is a conceptual advance of merit that can re-shape ways of approaching the still-open issue of gene co-bursting in light of novel (mostly imaging) technologies.
-
