Epigenetic reprogramming by TET enzymes impacts co-transcriptional R-loops
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
The work provides new insight into the potential role of 5hmc DNA in specific transcriptional processes. This implies that 5hmC containing DNA has specific epigenetic features beyond being a simple intermediate in interconversion between repressive 5mC and active C DNA. This work has the merit to focus the attention on a rare DNA modification, helping defining its functions, starting from in vitro evidence and extending these findings in cellular context. There are some weaknesses in the presentation of the data, the controls and the statistical analyses.
(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. The reviewers remained anonymous to the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
DNA oxidation by ten-eleven translocation (TET) family enzymes is essential for epigenetic reprogramming. The conversion of 5-methylcytosine (5mC) into 5-hydroxymethylcytosine (5hmC) initiates developmental and cell-type-specific transcriptional programs through mechanisms that include changes in the chromatin structure. Here, we show that the presence of 5hmC in the transcribed gene promotes the annealing of the nascent RNA to the template DNA strand, leading to the formation of an R-loop. Depletion of TET enzymes reduced global R-loops in the absence of gene expression changes, whereas CRISPR-mediated tethering of TET to an active gene promoted the formation of R-loops. The genome-wide distribution of 5hmC and R-loops shows a positive correlation in mouse and human stem cells and overlap in half of all active genes. Moreover, R-loop resolution leads to differential expression of a subset of genes that are involved in crucial events during stem cell proliferation. Altogether, our data reveal that epigenetic reprogramming via TET activity promotes co-transcriptional R-loop formation, disclosing new mechanisms of gene expression regulation.
Article activity feed
-
-
Evaluation Summary:
The work provides new insight into the potential role of 5hmc DNA in specific transcriptional processes. This implies that 5hmC containing DNA has specific epigenetic features beyond being a simple intermediate in interconversion between repressive 5mC and active C DNA. This work has the merit to focus the attention on a rare DNA modification, helping defining its functions, starting from in vitro evidence and extending these findings in cellular context. There are some weaknesses in the presentation of the data, the controls and the statistical analyses.
(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. The reviewers remained anonymous to the authors.)
-
Reviewer #1 (Public Review):
This study shows that the 5hmC DNA modification facilitates the formation of DNA-RNA hybrids during transcription. It is a hypothesis-driven study based on the known fact that 5hmC weakens the interaction between DNA and H2A.2B dimers and reduces the energy needed to separate the DNA strands. The study shows a correlation between 5hmC and R loops in mES cells and human HEK293 cells. This is shown first by dot blot, DRIP-qPCR at specific loci subjected to in vitro transcription in cells depleted of the TET enzymes Tet1 and Tet3 that convert 5mC into 5hmC, and later by bioinformatic analyses of genome-wide data of DNA-RNA hybrids and 5hmC of public databases, concluding that around 50% of active genes show an overlap between genes showing R loops and 5hmC. Thus there is a good correlation between R loop- and …
Reviewer #1 (Public Review):
This study shows that the 5hmC DNA modification facilitates the formation of DNA-RNA hybrids during transcription. It is a hypothesis-driven study based on the known fact that 5hmC weakens the interaction between DNA and H2A.2B dimers and reduces the energy needed to separate the DNA strands. The study shows a correlation between 5hmC and R loops in mES cells and human HEK293 cells. This is shown first by dot blot, DRIP-qPCR at specific loci subjected to in vitro transcription in cells depleted of the TET enzymes Tet1 and Tet3 that convert 5mC into 5hmC, and later by bioinformatic analyses of genome-wide data of DNA-RNA hybrids and 5hmC of public databases, concluding that around 50% of active genes show an overlap between genes showing R loops and 5hmC. Thus there is a good correlation between R loop- and 5hmC-containing genes, but there are also regions with 5hmC with no R loops and the vice versa. Expression of TET fused to an inactive Cas9 increases R loops formation at the APOE locus. Although its physiological meaning is unclear, the study is nice and of interest but additional experiments are required to validate the model.
-
Reviewer #2 (Public Review):
These data describe a new potential role for 5hmC modified DNA in enforcing R-loop structures especially over the 3' ends of protein coding genes in both mESCs and human cell lines (HEK293). While these data are at present largely correlative, they certainly make an interesting connection between 5hmC DNA modification and a key feature of transcribed genes (R-loops) that are known to be associated with DNA damage in numerous pathological cellular conditions. Various additional controls are needed to fully justify the claims made. Especially the reliance on the S9.6 mab to detect R-loops and their level changes based on reducing or targeting TET enzymes (that metabolise mC to 5hmC) needs tight controls. The potential higher affinity of S9.6 for 5hmC vs unmodified DNA as a possible cause of increased signal …
Reviewer #2 (Public Review):
These data describe a new potential role for 5hmC modified DNA in enforcing R-loop structures especially over the 3' ends of protein coding genes in both mESCs and human cell lines (HEK293). While these data are at present largely correlative, they certainly make an interesting connection between 5hmC DNA modification and a key feature of transcribed genes (R-loops) that are known to be associated with DNA damage in numerous pathological cellular conditions. Various additional controls are needed to fully justify the claims made. Especially the reliance on the S9.6 mab to detect R-loops and their level changes based on reducing or targeting TET enzymes (that metabolise mC to 5hmC) needs tight controls. The potential higher affinity of S9.6 for 5hmC vs unmodified DNA as a possible cause of increased signal needs to be investigated. Also, the redundancy of TET enzymes needs further investigation, rather than separately depleting either Tet1 and Tet3 (but not Tet2).
-
Reviewer #3 (Public Review):
Sabino et al. investigate the role of a rare DNA modification, 5-hydroxymethylcytosine (5hmC), in the formation of R-loops at transcribed loci. In an in vitro setting they demonstrate that 5hmC favors co-transcriptional R-loops formation. They further extend their findings in a cellular model, where they modulate the expression or the activity of TET enzymes, responsible for the conversion of 5mC into 5hmC. They observe that depletion of TET leads to a reduced amount of R-loops in cells, while targeting TET to a specific locus increases the formation of R-loops. Then, they take advantage of published datasets to show that 5hmC presence correlates with R-loops in active genes (and validate this observation performing PLA analysis), and with H2AX. Finally, they demonstrate that overexpression of RNaseH in …
Reviewer #3 (Public Review):
Sabino et al. investigate the role of a rare DNA modification, 5-hydroxymethylcytosine (5hmC), in the formation of R-loops at transcribed loci. In an in vitro setting they demonstrate that 5hmC favors co-transcriptional R-loops formation. They further extend their findings in a cellular model, where they modulate the expression or the activity of TET enzymes, responsible for the conversion of 5mC into 5hmC. They observe that depletion of TET leads to a reduced amount of R-loops in cells, while targeting TET to a specific locus increases the formation of R-loops. Then, they take advantage of published datasets to show that 5hmC presence correlates with R-loops in active genes (and validate this observation performing PLA analysis), and with H2AX. Finally, they demonstrate that overexpression of RNaseH in mouse embryonic stem cells leads to differential expression of genes that contain 5hmC modifications and are involved in diapause establishment. Nevertheless, RNaseH overexpression does not affect cell cycle progression of mouse embryonic stem cells in the time points investigated.
This work provides new information on the interplay between epigenetic modifications, R-loops formation and transcription. In the discussion, the authors propose some interesting points to pursue in future research on this topic. Nevertheless, there are some important aspects of experimental setting, data presentation and statistical analysis that need improving.
The main point of the manuscript, namely the fact that 5hmC favors co-transcriptional R-loops formation, is supported by data that in many cases lack statistical significance (Fig 1E, 2C). If this significance is not reached, the whole manuscript loses impact. The experiments where error bars and statistical significance are shown (Fig 1D, 2F) would be more convincing if the authors showed that 5hmC does not affect transcription levels in those particular settings.The authors provide a quantification of dot-blots based on the normalization of the signal of interest (5hmC, 5mC, S9.6, Fig 1B, 2B, 2C) on dsDNA. From source data, it seems that these signals are quantified in two different membranes: this is not correct, should be carried out on the same membrane.
The proximity ligation assay performed in figure 4 to confirm the previous results is not convincing. Better resolution and magnification are needed to better gauge the signal. Single plane confocal images would be clearer. Moreover, the PLA signal seems to be mainly extranuclear. It is worth noticing that the use of S9.6 antibody for imaging techniques proved to be problematic for its non-specific binding to dsRNA (see Hartono et al., The EMBO Journal, 2021). Additional control conditions are necessary.
-
