Dynamic spreading of chromatin-mediated gene silencing and reactivation between neighboring genes in single cells
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This study describes a novel approach to investigate how the transcriptional repressors KRAB and HDAC4 repress gene expression, how repression spreads over differing genomic distances, and what the role of insulator elements is in blocking the spread of repression and in reactivation of repressed genes. The results of this study allow modeling of the coordinated repression or activation of closely linked genes and should be of wide interest to researchers interested in chromatin and gene expression.
(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.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Synthetic Biology (eLife)
Abstract
In mammalian cells genes that are in close proximity can be transcriptionally coupled: silencing or activating one gene can affect its neighbors. Understanding these dynamics is important for natural processes, such as heterochromatin spreading during development and aging, and when designing synthetic gene regulation circuits. Here, we systematically dissect this process in single cells by recruiting and releasing repressive chromatin regulators at dual-gene synthetic reporters, and measuring how fast gene silencing and reactivation spread as a function of intergenic distance and configuration of insulator elements. We find that silencing by KRAB, associated with histone methylation, spreads between two genes within hours, with a time delay that increases with distance. This fast KRAB-mediated spreading is not blocked by the classical cHS4 insulators. Silencing by histone deacetylase HDAC4 of the upstream gene can also facilitate background silencing of the downstream gene by PRC2, but with a days-long delay that does not change with distance. This slower silencing can sometimes be stopped by insulators. Gene reactivation of neighboring genes is also coupled, with strong promoters and insulators determining the order of reactivation. Our data can be described by a model of multi-gene regulation that builds upon previous knowledge of heterochromatin spreading, where both gene silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment.
Article activity feed
-
-
Author Response:
Reviewer #2 (Public Review):
This study explores how the transcriptional status of a gene can affect its neighbours. For this, the authors insert a dual reporter system location in two cell lines, separating or not the two reporters by a 5kb-spacer with or without an insulator, and subjecting one of them to repression either with a KRAB module or the histone deacetylase HDAC4, analysing by time-lapse microscopy and FACS how repressing one reporter impacts expression from the second. The approach is elegant, the work well conducted and its results nicely presented, with interesting differences observed between the impacts of the two repressors. However, a major concern regards the generalities drawn by the authors from a very limited set of observations (2 cell lines, 2 genomic loci, 2 chromatin modifiers, 2 …
Author Response:
Reviewer #2 (Public Review):
This study explores how the transcriptional status of a gene can affect its neighbours. For this, the authors insert a dual reporter system location in two cell lines, separating or not the two reporters by a 5kb-spacer with or without an insulator, and subjecting one of them to repression either with a KRAB module or the histone deacetylase HDAC4, analysing by time-lapse microscopy and FACS how repressing one reporter impacts expression from the second. The approach is elegant, the work well conducted and its results nicely presented, with interesting differences observed between the impacts of the two repressors. However, a major concern regards the generalities drawn by the authors from a very limited set of observations (2 cell lines, 2 genomic loci, 2 chromatin modifiers, 2 promoters, a single distance of 5Kb as spacer, one insulator). Short of scaling up their analyses very significantly they should tone down their conclusions and refrain from statements such as "we propose a new model of multi-gene regulation, where both gene silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment" or (lines 66-68) "we used these findings to form a kinetic model that leverages information from changes in histone modifications to understand the dynamics of gene expression during both silencing and reactivation. Our results show that targeted transcriptional silencing affects neighboring genes". By comparison, a number or previous studies including by Amabile et al. 2015 and Groner et al. 2010 (both referenced but not much commented on), although centered exclusively on the impact of KRAB and devoid of the nice time-lapse analyses presented here, were much broader and already made many of the points stated in the present work, including for the latter study the demonstration that KRAB-mediated silencing can spread overall several tens of kb with an average maximal efficiency for promoters located within 15kb or less, results in loss of histone H3 acetylation, drop in RNA PolII and gain of H3K9me3 at these promoters through long-range spreading of this mark and of HP1beta from the initial KRAB-binding site and is dependent on the ability of KAP1 to recruit HP1. The present paper would considerably gain from placing its results in the context of the sum of these previous studies, discussing specifically what it truly adds to their conclusions.
Thank you for the useful suggestions; we have modified the text of the paper to put the work into context, describe better what was known already (that KRAB-mediated silencing and associated modifications can spread across tens of kb), and emphasize the new findings (the dependence of silencing times on distance, the inability of cHS4 insulators to stop spreading at these short distances, and the fact that reactivation can also spread between genes in a distance-dependent manner). We also made it more explicit when statements are claims based on our data (especially throughout the results section), versus hypotheses that we put forward in the discussion that require further testing (please see more details in the following section).
Changes to the text include:
-We added a more thorough description of what was known of KRAB spreading in the introduction (including summarizing results from Amabile et al. 2015, Groner et al. 2010, and Meylan et al. 2011).
-We edited the model section and the abstract to tone down the generality of our conclusions and acknowledge previous work:
• Original: “We propose a new model of multi-gene regulation, where both gene silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment, where both gene silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment.”
• New: “Our data can be described by a model of multi-gene regulation that builds upon previous knowledge of heterochromatin spreading, where both silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment.”
-We provided more general background information in the introduction:
• Edited sentence: “Chromatin-mediated gene regulation is crucial in development, aging, and disease, with classical examples of Xchromosome inactivation and the spatial-temporal control of Hox genes (Soshnikova and Duboule 2009; Payer 2017).
-We better explain why a synthetic biology approach is useful and important:
• Original: “it is also important in synthetic biology, where precise control of gene expression is necessary. This Site-specific recruitment of CRs is a common method of regulating gene expression in synthetic systems”
• New: “It is also important in synthetic biology, where precise control of gene expression is necessary for probing gene regulatory networks with CRISPRi type screens, for better understanding mechanisms of epigenetic regulation such as cellular reprogramming, and for therapeutic applications such as gene therapy (Lienert et al. 2014; Keung et al. 2015; Thakore et al. 2016). Due to the limit in length of DNA constructs that can be successfully delivered and integrated into cells (Lukashev and Zamyatnin 2016; Liu et al. 2017), multiple genes are often placed close together, such as an antibiotic resistance selective marker next to a gene of interest. In these synthetic systems, a common method of regulating gene expression is through site-specific recruitment of CRs. CRs modulate gene expression with varying kinetics and can establish long-term epigenetic memory through positive feedback mechanisms, which enable spreading of epigenetic effects beyond the target locus, and lead to undesirable changes in gene expression, which can be implicated in aging and cancer (Sedivy et al. 2008; Wang et al. 2015)."
-We added more background on the discovery of this spreading phenomenon:
• Added sentence: “This phenomenon of spreading of epigenetic effects was discovered in Drosophila and was originally coined as position effect variegation(Muller 1930; Wang et al. 2014). The mechanism of action has since been elucidated to involve readers and writers of histone modifications forming a feedback loop that causes modifications to spread (Elgin and Reuter 2013), and has also been shown to occur in mammalian cells in vivo (Groner et al. 2012).”
-We delineated our specific questions clearer in the introduction: “We wanted to know how silencing by these two rapid-acting chromatin regulators with and without positive feedback mechanisms, KRAB and HDAC4, respectively, affect gene expression of neighboring genes when separated by either distance or the cHS4 insulator, as well as how the dynamics of reactivation are affected after removal of the chromatin regulator.”
-We delineated our specific findings clearer in the introduction:
• Original: “We used these findings to form a kinetic model that leverages information from changes in histone modifications to understand the dynamics of gene expression during both silencing and reactivation.”
• New: “Reactivation also spreads between the two genes with a delay that is distance-dependent, and is affected by promoters and insulators. We can summarize our findings with a simple kinetic model that describes the dynamics of silencing and reactivation as a competition between: (1) the silencing rates associated with each repressive CR and (2) the activation rates associated with strong promoters and insulators, where both of these rates decrease with genomic distance.”
-We clarified that the loss of acetylation and gain in H3K9me3 were expected from KRAB recruitment:
• “To see if the changes in gene expression were accompanied by the expected changes in chromatin modifications, we used CUT&RUN to map activating and repressive histone modifications.”
• “KRAB recruitment is known to result in loss of acetylation and gain of H3K9me3 (Ayyanathan et al. 2003; Groner et al. 2010; Amabile et al. 2016; Feng et al. 2020), while HDAC4 is associated with loss of acetylation(Wang et al. 1999; Wang et al. 2014).”
• Add sentence: “This loss of acetylation and gain of H3K9me3 and its spreading across a large domain including over neighboring genes is in line with what has been previously shown (Groner et al. 2010; Amabile et al. 2016), confirming that our system works as expected.”
Reviewer #3 (Public Review):
Heterochromatin can spread to neighbouring regions via feedforward "reader-writer" mechanisms and repress its target genes through physical compaction. However, little is known about the dynamics of heterochromatin spreading and the kinetics of target gene repression. In this manuscript, Dr. Bintu and colleagues use a synthetic approach by recruiting tetR-KRAB or tetR-HDAC in doxycycline inducible manner to upstream regions of two constitutive promoter driving reporters in tandem or spaced by 5kb to allow real-time measurement of heterochromatin spreading based on reporter gene expression upon dox induction/withdrawn. The authors also established the system in CHO cells and in K562 cells, for comparison. The authors revealed that KRAB-mediated H3K9me3 spreading is fast, and a function of spatial distance, leading to distance-related delay in gene silencing. In contrast, HDAC-mediated deacetylation is slow, and is subject to potential stochastic interactions between neighbouring nucleosomes, and therefore displays less delays in silencing. Furthermore, in contrast to common belief, widely adopted insulator has little effects on heterochromatin spreading in both KRAB and HDAC-mediated silencing. Finally, mathematic modelling reveals potential roles of histone acetylation and the acetylation reader-writer feedforward loop in fighting against HDAC-mediated spreading of gene repression.
A few limitations of the general conclusions are that, perhaps not unexpectedly, differences were seen in the behavior of the reporters at the integration sites in one cell line versus the other. This, of course, is not a fault of the authors and rather reflects the rigor of their approach. For example, no insulator configurations inhibited spreading of silencing upon HDAC4 recruitment in CHO cells, but insulators did attenuate HDAC4-mediated silencing in K562 cells. There were also differences in background expression of the constructs in the two cells. These issues raise challenges in general conclusions from the study, and underscore the particularities of genome function in different contexts.
Although the synthetic approaches adopted in this study can help tease out individual functions of chromatin regulators, previous studies using dCas9-fused with KRAB domains to inoculate heterochromatin domains in megabases of natural sites in the human genome indicated that spreading of H3K9me3 heterochromatin domains does not necessarily lead to gene silencing, and gene repression is more associated with the loss active histone marks, such as H3K27ac and H3K4me3. Therefore, it is possible that the model of kinetics of gene silencing in this current synthetic system may be valid in short distance (~5kb), but may over-estimate the roles of H3K9me3/HDAC spreading on gene expression in much larger scale.
Thank you for your review and comments. We agree that at larger length scales of megabases and at endogenous genes the kinetics of silencing may be different from our proposed model. Our model serves as a set of hypotheses for understanding dynamics of genes on the kilobase length scale in synthetics systems. We have clarified the introduction of our model as follows:
• Original: “We used our gene dynamics and chromatin modifications data to develop a generalizable kinetic model that captures distance-dependent silencing associated with a tethered chromatin regulator and distills the roles of promoters and insulators as elements associated with high reactivation rates.”
• New: “We developed a kinetic model that summarizes our observations of gene dynamics and chromatin modifications as a competition between the distance-dependent silencing rates associated with each tethered chromatin regulator and the reactivation rates driven by our promoters and insulators.”
We have also added findings from previous studies about the histone modifications primarily driving the silencing of KRAB-mediated repression.
• “The strength of recruitment also affects silencing and spreading as we saw from low dox recruitment (Figure 1 - figure supplement 3)... In previous studies, targeting dCas9-KRAB to hundreds of repeated sgRNA sites forms a large domain of H3K9me3 heterochromatin on the order of megabases in a few days, but does not result in widespread gene silencing, rather the silencing of genes is controlled by the loss of H3K27Ac and H3K4me3 (Feng et al. 2020).”
-
Evaluation Summary:
This study describes a novel approach to investigate how the transcriptional repressors KRAB and HDAC4 repress gene expression, how repression spreads over differing genomic distances, and what the role of insulator elements is in blocking the spread of repression and in reactivation of repressed genes. The results of this study allow modeling of the coordinated repression or activation of closely linked genes and should be of wide interest to researchers interested in chromatin and gene expression.
(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):
In this study, Lensch et al., describe a novel approach to study the mechanisms of gene repression at the single cell level using a reporter construct expressing fluorescent proteins whose expression can be repressed by inducible expression of the KRAB or HDAC4 repressor proteins. The fluorescent reporter genes were either adjacent, separated by a 5 kb spacer or flanked by one or two cSH4 insulator elements. The reporter constructs were integrated into fixed genomic sites in two different cell lines and expression of the fluorescent reporter proteins monitored at the single cell level by imaging and flow cytometry following Doxycycline-inducible expression of the KRAB or HDAC4 repressor proteins. This approach revealed that repression KRAB or HDAC4 followed different kinetics, different mechanisms and with …
Reviewer #1 (Public Review):
In this study, Lensch et al., describe a novel approach to study the mechanisms of gene repression at the single cell level using a reporter construct expressing fluorescent proteins whose expression can be repressed by inducible expression of the KRAB or HDAC4 repressor proteins. The fluorescent reporter genes were either adjacent, separated by a 5 kb spacer or flanked by one or two cSH4 insulator elements. The reporter constructs were integrated into fixed genomic sites in two different cell lines and expression of the fluorescent reporter proteins monitored at the single cell level by imaging and flow cytometry following Doxycycline-inducible expression of the KRAB or HDAC4 repressor proteins. This approach revealed that repression KRAB or HDAC4 followed different kinetics, different mechanisms and with different distance dependency. Interestingly, the authors show that the cSH4 insulator did not always block the spreading of repression, but can play a role in coordinating the reactivation of gene expression. Despite the fact that the system used is artificial and simplified, the approach highlights how such approaches can be used to make mathematical modeling of the coordinated repression or activation of closely linked genes.
-
Reviewer #2 (Public Review):
This study explores how the transcriptional status of a gene can affect its neighbours. For this, the authors insert a dual reporter system location in two cell lines, separating or not the two reporters by a 5kb-spacer with or without an insulator, and subjecting one of them to repression either with a KRAB module or the histone deacetylase HDAC4, analysing by time-lapse microscopy and FACS how repressing one reporter impacts expression from the second. The approach is elegant, the work well conducted and its results nicely presented, with interesting differences observed between the impacts of the two repressors. However, a major concern regards the generalities drawn by the authors from a very limited set of observations (2 cell lines, 2 genomic loci, 2 chromatin modifiers, 2 promoters, a single distance …
Reviewer #2 (Public Review):
This study explores how the transcriptional status of a gene can affect its neighbours. For this, the authors insert a dual reporter system location in two cell lines, separating or not the two reporters by a 5kb-spacer with or without an insulator, and subjecting one of them to repression either with a KRAB module or the histone deacetylase HDAC4, analysing by time-lapse microscopy and FACS how repressing one reporter impacts expression from the second. The approach is elegant, the work well conducted and its results nicely presented, with interesting differences observed between the impacts of the two repressors. However, a major concern regards the generalities drawn by the authors from a very limited set of observations (2 cell lines, 2 genomic loci, 2 chromatin modifiers, 2 promoters, a single distance of 5Kb as spacer, one insulator). Short of scaling up their analyses very significantly they should tone down their conclusions and refrain from statements such as "we propose a new model of multi-gene regulation, where both gene silencing and gene reactivation can act at a distance, allowing for coordinated dynamics via chromatin regulator recruitment" or (lines 66-68) "we used these findings to form a kinetic model that leverages information from changes in histone modifications to understand the dynamics of gene expression during both silencing and reactivation. Our results show that targeted transcriptional silencing affects neighboring genes". By comparison, a number or previous studies including by Amabile et al. 2015 and Groner et al. 2010 (both referenced but not much commented on), although centered exclusively on the impact of KRAB and devoid of the nice time-lapse analyses presented here, were much broader and already made many of the points stated in the present work, including for the latter study the demonstration that KRAB-mediated silencing can spread overall several tens of kb with an average maximal efficiency for promoters located within 15kb or less, results in loss of histone H3 acetylation, drop in RNA PolII and gain of H3K9me3 at these promoters through long-range spreading of this mark and of HP1beta from the initial KRAB-binding site and is dependent on the ability of KAP1 to recruit HP1. The present paper would considerably gain from placing its results in the context of the sum of these previous studies, discussing specifically what it truly adds to their conclusions.
-
Reviewer #3 (Public Review):
Heterochromatin can spread to neighbouring regions via feedforward "reader-writer" mechanisms and repress its target genes through physical compaction. However, little is known about the dynamics of heterochromatin spreading and the kinetics of target gene repression. In this manuscript, Dr. Bintu and colleagues use a synthetic approach by recruiting tetR-KRAB or tetR-HDAC in doxycycline inducible manner to upstream regions of two constitutive promoter driving reporters in tandem or spaced by 5kb to allow real-time measurement of heterochromatin spreading based on reporter gene expression upon dox induction/withdrawn. The authors also established the system in CHO cells and in K562 cells, for comparison. The authors revealed that KRAB-mediated H3K9me3 spreading is fast, and a function of spatial distance, …
Reviewer #3 (Public Review):
Heterochromatin can spread to neighbouring regions via feedforward "reader-writer" mechanisms and repress its target genes through physical compaction. However, little is known about the dynamics of heterochromatin spreading and the kinetics of target gene repression. In this manuscript, Dr. Bintu and colleagues use a synthetic approach by recruiting tetR-KRAB or tetR-HDAC in doxycycline inducible manner to upstream regions of two constitutive promoter driving reporters in tandem or spaced by 5kb to allow real-time measurement of heterochromatin spreading based on reporter gene expression upon dox induction/withdrawn. The authors also established the system in CHO cells and in K562 cells, for comparison. The authors revealed that KRAB-mediated H3K9me3 spreading is fast, and a function of spatial distance, leading to distance-related delay in gene silencing. In contrast, HDAC-mediated deacetylation is slow, and is subject to potential stochastic interactions between neighbouring nucleosomes, and therefore displays less delays in silencing. Furthermore, in contrast to common belief, widely adopted insulator has little effects on heterochromatin spreading in both KRAB and HDAC-mediated silencing. Finally, mathematic modelling reveals potential roles of histone acetylation and the acetylation reader-writer feedforward loop in fighting against HDAC-mediated spreading of gene repression.
A few limitations of the general conclusions are that, perhaps not unexpectedly, differences were seen in the behavior of the reporters at the integration sites in one cell line versus the other. This, of course, is not a fault of the authors and rather reflects the rigor of their approach. For example, no insulator configurations inhibited spreading of silencing upon HDAC4 recruitment in CHO cells, but insulators did attenuate HDAC4-mediated silencing in K562 cells. There were also differences in background expression of the constructs in the two cells. These issues raise challenges in general conclusions from the study, and underscore the particularities of genome function in different contexts.
Although the synthetic approaches adopted in this study can help tease out individual functions of chromatin regulators, previous studies using dCas9-fused with KRAB domains to inoculate heterochromatin domains in megabases of natural sites in the human genome indicated that spreading of H3K9me3 heterochromatin domains does not necessarily lead to gene silencing, and gene repression is more associated with the loss active histone marks, such as H3K27ac and H3K4me3. Therefore, it is possible that the model of kinetics of gene silencing in this current synthetic system may be valid in short distance (~5kb), but may over-estimate the roles of H3K9me3/HDAC spreading on gene expression in much larger scale.
-
