Deciphering Plant Chromatin Regulation via CRISPR/dCas9-Based Epigenome Engineering

Round 1

Reviewer 1 Report

The manuscript by Dubois and Roudier is a nicely written, comprehensive, and very interesting review on tools to manipulate the epigenome in plants that will certainly interest researchers working on chromatin, transcription, and epigenome editing. I just have few comments that the authors should take into consideration:

General comments:

- I am missing information regarding off-target activities of these systems. This is a critical point for the use of these tools and therefore I think an effort should be made to compile what is currently known. Only information about targeted methylation using Suntag is provided.

- A summary table would be very useful for the reader, indicating the different constructs reported in this manuscript, their use, efficiency, off-target activities, references, etc…

Other comments:

Figure 1: reorganize panels b and c and corresponding figure legend.

Figure 1e: Figure legend says EBF1, please correct.

The section describing the Cascade approach could be explained a bit better. What is Cascade vs Cas9? Is it necessary to express all Cascade proteins? Is a gRNA needed….? It is not clear from the cartoon since the colors and shapes used for the gRNAs are different between Cas9 and Cascade.

Line 167: Triple SRDX (3x-SRDX) efficiently… and I would use the “3x-SRDX” expression thereafter and in the figure.

Line 177: Can you provide a reference for this sentence, which states: CpfI “occupies much larger porting of the targeted DNA region in comparison to dCas9”.

Line 189: “N-terminus (300 amino acids)”. Please add (N300TPL) right after this expression.

Line 222-223: This sentence could be more accurate as it is a bit misleading in its current form. Reduction of the ectopic hypermethylation was achieved by removing an NLS and therefore decreasing the accumulation of the NtDRM module.

Line 223-225: The Jacobsen lab has just published another paper where they report genome-wide hypermethylation using MQ1 (aka SssI) fused to an artificial ZF. Since this review is mostly about plants, I think this reference should be included together with reference 37 to refer to ectopic hypermethylation using SssI.

Figure 3: panel a shows target and off-target methylation using the Suntag-DRM approach. I think this is misleading somehow and should be modified since, as the authors mention in the manuscript, an optimized version of this construct does not cause genome-wide hypermethylation. Thus, I think the off-target cartoon should be removed or, if authors want to point to this important problem, depict both versions of the construct and point to tool optimization as an important requirement when using this sort of tools. Also, the figure legend should be modified accordingly. Moreover, a cartoon with the dCas9-MQ1 construct could be shown pointing to the different methylation context where DRM and MQ1 are more active (CHH and CG respectively).

Figure 3 figure legend: Replace Tet1 for TET1

Section 2.2.3. I would personally move sentences 276-281 describing successful targeted acetylation in mammals to the beginning of the section and then move to the experiments done in plants where this approach is not so well established.

Sentence 287. This is not correct. Reference 41 tested MS2-KYP over the FT locus with no effect on K9me2. Please correct.  

Sentence 299-301. Please provide a reference to support this sentence.

Sentence 333-337. I would tone down this sentence. This paper only shows that UBE2a cannot induce transcription but does not show actual UBE2a-mediated H2B ubiquitination. Also, the interdependency between H2Bub and K4me3/K79me is not demonstrated. This, however, is a very nice example for Section 3.1.1 since they combine the action of two histone modifiers to enhance gene activation. Perhaps one option would be to combine sections 3.1.1 and 3.1.2

Line 337: Please provide a reference that supports the absence of K79me pathway in Arabidopsis. Otherwise rephrase.

Line 364: Reference 50 does not study targeted K27me3. Please correct. Moreover, there is a well-known anticorrelation between DNA methylation and K27me3 accumulation. Please add this discussion to this section, and an appropriate reference, if authors want to make a point about the negative role of DNA methylation on targeted K27me3. The Jacobsen lab just published a paper where they show that SssI-mediated targeted CG-methylation triggers a reduction in the accumulation of K27me3 as well as H2A.Z, which supports the already known anticorrelation between K27me3 and H2A.Z accumulation in plants (Liu et al. 2021). This data would be a nice addition to this section.

Section 3.2.2. Authors described that DNA methylation affects transcription through its impact on TF binding. There are alternative ways by which DNA methylation can modify transcription. The Jacobsen lab has published two nice studies where SUVH and MBP methyl-readers recruit complexes involved in transcriptional activation and repression respectively (Harris et al. 2019, Ichino et al. 2021). Might be worth commenting briefly on these mechanisms as different ways to regulate gene expression through targeted DNA methylation.

Section 4.1 Besides CpG methylation or PRC2 targeted regions, authors should comment and refer to studies that show, both in vitro and in vivo, that nucleosomes are a direct physical barrier for Cas9 target binding.

Line 476: Why mention only CRISPRi? As far as I know, no CRISPR-mediated epigenome targeting has been done in a cell or organ-specific manner.

Author Response

- The manuscript by Dubois and Roudier is a nicely written, comprehensive, and very interesting review on tools to manipulate the epigenome in plants that will certainly interest researchers working on chromatin, transcription, and epigenome editing. I just have few comments that the authors should take into consideration:

We’d like to thank you for your careful reading and accurate and interesting comments and suggestions.

- I am missing information regarding off-target activities of these systems. This is a critical point for the use of these tools and therefore I think an effort should be made to compile what is currently known. Only information about targeted methylation using Suntag is provided.

Although of paramount importance, systematic analysis of off-target activities in the different systems remains scarce. Among all references herein, off-target activity was almost exclusively described in the work involving targeted DNA methylation or demethylation, with two other studies involving CRISPRa (Li et al., 2017 and Selma et al., 2019). We have added this information when describing these two CRISPRa tools (lines 143-146), and a sentence at the end of section 2 to underline this point (lines 312-316). Off-target activity description (or absence therein) is also mentioned in Supplementary Table 1.

- A summary table would be very useful for the reader, indicating the different constructs reported in this manuscript, their use, efficiency, off-target activities, references, etc…

A Summary Table was generated based on the different systems reported in this review (Supplementary Table 1).

Other comments:

Figure 1: reorganize panels b and c and corresponding figure legend.

Panels b and c were re-ordered in the correct way.

Figure 1e: Figure legend says EBF1, please correct.

Figure 1e legend was corrected.

The section describing the Cascade approach could be explained a bit better.

What is Cascade vs Cas9? Is it necessary to express all Cascade proteins? Is a gRNA needed….? It is not clear from the cartoon since the colors and shapes used for the gRNAs are different between Cas9 and Cascade.

Figure 1e: the cascade crRNA was redesigned according to design and colors that are more similar to the Cas9 sgRNAs in other panels.

A paragraph describing the Cascade system was added to the manuscript (lines 156-166).

Line 167: Triple SRDX (3x-SRDX) efficiently… and I would use the “3x-SRDX” expression thereafter and in the figure.

These corrections were done.

Line 177: Can you provide a reference for this sentence, which states: CpfI “occupies much larger portion of the targeted DNA region in comparison to dCas9”.

Further investigation of this specific point indicate that Cpf1 proteins vary in size and are sometimes smaller than Cas9 proteins. The fact that Cpf1 can be larger than Cas9 is not a general principle and we therefore removed the corresponding sentence.

Line 189: “N-terminus (300 amino acids)”. Please add (N300TPL) right after this expression.

We have modified the text and Figure 2.

Line 222-223: This sentence could be more accurate as it is a bit misleading in its current form. Reduction of the ectopic hypermethylation was achieved by removing an NLS and therefore decreasing the accumulation of the NtDRM module.

This sentence was modified.

Line 223-225: The Jacobsen lab has just published another paper where they report genome-wide hypermethylation using MQ1 (aka SssI) fused to an artificial ZF. Since this review is mostly about plants, I think this reference should be included together with reference 37 to refer to ectopic hypermethylation using SssI.

The additional reference was added.

Figure 3: panel a shows target and off-target methylation using the Suntag-DRM approach. I think this is misleading somehow and should be modified since, as the authors mention in the manuscript, an optimized version of this construct does not cause genome-wide hypermethylation. Thus, I think the off-target cartoon should be removed or, if authors want to point to this important problem, depict both versions of the construct and point to tool optimization as an important requirement when using this sort of tools. Also, the figure legend should be modified accordingly. Moreover, a cartoon with the dCas9-MQ1 construct could be shown pointing to the different methylation context where DRM and MQ1 are more active (CHH and CG respectively).

We thank reviewer 1 for this suggestion: an additional cartoon was added to Figure 3 and the legend modified accordingly.

Figure 3 figure legend: Replace Tet1 for TET1

This modification was done.

Section 2.2.3. I would personally move sentences 276-281 describing successful targeted acetylation in mammals to the beginning of the section and then move to the experiments done in plants where this approach is not so well established.

This paragraph was moved at the beginning of section 2.2.3.

Sentence 287. This is not correct. Reference 41 tested MS2-KYP over the FT locus with no effect on K9me2. Please correct.  

A sentence describing the attempt to target KRYPTONITE was added.

Sentence 299-301. Please provide a reference to support this sentence.

Reference was added (line 331) on the new version of the manuscript, (reference 44).

Sentence 333-337. I would tone down this sentence. This paper only shows that UBE2a cannot induce transcription but does not show actual UBE2a-mediated H2B ubiquitination. Also, the interdependency between H2Bub and K4me3/K79me is not demonstrated. This, however, is a very nice example for Section 3.1.1 since they combine the action of two histone modifiers to enhance gene activation. Perhaps one option would be to combine sections 3.1.1 and 3.1.2

Both paragraphs were brought together in a single part.

Line 337: Please provide a reference that supports the absence of K79me pathway in Arabidopsis. Otherwise rephrase.

A reference was added (reference 53).

Line 364: Reference 50 does not study targeted K27me3. Please correct. Moreover, there is a well-known anticorrelation between DNA methylation and K27me3 accumulation. Please add this discussion to this section, and an appropriate reference, if authors want to make a point about the negative role of DNA methylation on targeted K27me3. The Jacobsen lab just published a paper where they show that SssI-mediated targeted CG-methylation triggers a reduction in the accumulation of K27me3 as well as H2A.Z, which supports the already known anticorrelation between K27me3 and H2A.Z accumulation in plants (Liu et al. 2021). This data would be a nice addition to this section.

An additional paragraph was added describing these recent results.

Section 3.2.2. Authors described that DNA methylation affects transcription through its impact on TF binding. There are alternative ways by which DNA methylation can modify transcription. The Jacobsen lab has published two nice studies where SUVH and MBP methyl-readers recruit complexes involved in transcriptional activation and repression respectively (Harris et al. 2019, Ichino et al. 2021). Might be worth commenting briefly on these mechanisms as different ways to regulate gene expression through targeted DNA methylation.

An additional paragraph was added describing these recent results.

Section 4.1 Besides CpG methylation or PRC2 targeted regions, authors should comment and refer to studies that show, both in vitro and in vivo, that nucleosomes are a direct physical barrier for Cas9 target binding.

One additional sentence, including two references, was added.

Line 476: Why mention only CRISPRi? As far as I know, no CRISPR-mediated epigenome targeting has been done in a cell or organ-specific manner.

The sentence was modified accordingly.

Reviewer 2 Report

In this review paper, Dubois and Roudier present recent progress of CRISPR/dCas9 technology in plants. It seems likely that this manuscript is well-written and provides a guideline for the plant science community to consider using the CRISPR/dCas9 techniques.

Given that these techniques are mostly developed in animal systems, the majority of the systems were not applied in plants.  Thus, it will be better if the authors can provide a table summarizing the researches in plant systems.

In addition, I recommend writing a conclusion or future direction section at the last of the manuscript.

One more comment is on the title of this manuscript. Since the first part of the review is on transcriptional engineering, which may not be related to epigenome editing, the tile should be corrected.

Author Response

Responses to reviewer 2:

In this review paper, Dubois and Roudier present recent progress of CRISPR/dCas9 technology in plants. It seems likely that this manuscript is well-written and provides a guideline for the plant science community to consider using the CRISPR/dCas9 techniques.

Given that these techniques are mostly developed in animal systems, the majority of the systems were not applied in plants.  Thus, it will be better if the authors can provide a table summarizing the researches in plant systems.

A summary table is now provided.

In addition, I recommend writing a conclusion or future direction section at the last of the manuscript.

A conclusion was added.

One more comment is on the title of this manuscript. Since the first part of the review is on transcriptional engineering, which may not be related to epigenome editing, the tile should be corrected.

As mentioned in the introduction, besides directly recruiting chromatin modifiers, CRISPR-dCas9 epigenome alterations can be achieved via the targeting of a transacting factor to a gene regulatory region which can destabilize or trigger a shift in the local chromatin state. In addition, many limitations of the CRISPR approach have been pinpointed in transcriptional engineering systems and also apply to epigenome editing methods.

We have modified the title by changing “editing” to “engineering” in order to better include this indirect aspect.

Reviewer 3 Report

In this manuscript, authors summarized and discussed the recent advances in plant epigenome editing via CRISPR/dCas9 system. This manuscript also highlights the limitations of epigenome editing approaches and discussed the potential strategies to overcome such limitations. The manuscript is timely, well structured and written. I recommend this manuscript for publication. However, it would be appropriate to add the conclusion section.

Author Response

In this manuscript, authors summarized and discussed the recent advances in plant epigenome editing via CRISPR/dCas9 system. This manuscript also highlights the limitations of epigenome editing approaches and discussed the potential strategies to overcome such limitations. The manuscript is timely, well structured and written. I recommend this manuscript for publication. However, it would be appropriate to add the conclusion section.

A conclusion was added.

Round 2

Reviewer 1 Report

Thanks to the authors for the corrections and congrats again on the nice manuscript! Please find my responses in yellow.

Line 223-225: The Jacobsen lab has just published another paper where they report genome-wide hypermethylation using MQ1 (aka SssI) fused to an artificial ZF. Since this review is mostly about plants, I think this reference should be included together with reference 37 to refer to ectopic hypermethylation using SssI.

The additional reference was added.

This reference has not been included here (or I can´t fint it). The reference I meant is Liu et al, 2021 Nat Comm

Figure 3: panel a shows target and off-target methylation using the Suntag-DRM approach. I think this is misleading somehow and should be modified since, as the authors mention in the manuscript, an optimized version of this construct does not cause genome-wide hypermethylation. Thus, I think the off-target cartoon should be removed or, if authors want to point to this important problem, depict both versions of the construct and point to tool optimization as an important requirement when using this sort of tools. Also, the figure legend should be modified accordingly. Moreover, a cartoon with the dCas9-MQ1 construct could be shown pointing to the different methylation context where DRM and MQ1 are more active (CHH and CG respectively).

We thank reviewer 1 for this suggestion: an additional cartoon was added to Figure 3 and the legend modified accordingly.

Is there a reason to show in the new cartoon the direct fusión dCas9-MQ1 rather tan Suntag-MQ1, which is probably a more powerful tool?

• Sentence in lines 453-455 is confusing, please rewrite.
• Line 565: Please correct “CRIPSR”
• Line 568: Please correct “CAs9”
• Supp table: Suntag DRM2 was used with 3 simultaneous gRNAs. Please correct.
• Supp table: MQ1 was used with 3 simultaneous gRNAs. Please correct.
• Supp table: In the methylation section, lane 60 says “targeted CHH methylation” while referring to MQ1. Correct
• Supp table: Please make sure the rest of the info is accurate.

Author Response

Responses to reviewer 1 (round 2) are highlighted in blue.

Comments and Suggestions for Authors

Thanks to the authors for the corrections and congrats again on the nice manuscript! Please find my responses in yellow.

Thank you for your very nice review, it largely improved our manuscript.

Line 223-225: The Jacobsen lab has just published another paper where they report genome-wide hypermethylation using MQ1 (aka SssI) fused to an artificial ZF. Since this review is mostly about plants, I think this reference should be included together with reference 37 to refer to ectopic hypermethylation using SssI.

The additional reference was added.

This reference has not been included here (or I can´t fint it). The reference I meant is Liu et al, 2021 Nat Comm

The additional reference was added with a bit of rephrasing (lines 229-231).

Figure 3: panel a shows target and off-target methylation using the Suntag-DRM approach. I think this is misleading somehow and should be modified since, as the authors mention in the manuscript, an optimized version of this construct does not cause genome-wide hypermethylation. Thus, I think the off-target cartoon should be removed or, if authors want to point to this important problem, depict both versions of the construct and point to tool optimization as an important requirement when using this sort of tools. Also, the figure legend should be modified accordingly. Moreover, a cartoon with the dCas9-MQ1 construct could be shown pointing to the different methylation context where DRM and MQ1 are more active (CHH and CG respectively).

We thank reviewer 1 for this suggestion: an additional cartoon was added to Figure 3 and the legend modified accordingly.

Is there a reason to show in the new cartoon the direct fusión dCas9-MQ1 rather tan Suntag-MQ1, which is probably a more powerful tool?

We agree we oversimplified the figure. For Figure 3 panel b we changed the cartoon to SunTag MQ1. Figure legend was also modified.

• Sentence in lines 453-455 is confusing, please rewrite.

This sentence was re-written (lines 424-428 (somehow there’s a 29 lines shift) )

• Line 565: Please correct “CRIPSR”
• Line 568: Please correct “CAs9”

• We corrected these spelling mistakes (lines 536 and 539)

• Supp table: Suntag DRM2 was used with 3 simultaneous gRNAs. Please correct.
• Supp table: MQ1 was used with 3 simultaneous gRNAs. Please correct.
• Supp table: In the methylation section, lane 60 says “targeted CHH methylation” while referring to MQ1. Correct

All correction were done and appear in red in the table.

• Supp table: Please make sure the rest of the info is accurate.

We reconsidered each reference in the table and found a few more errors on the number of sgRNAs used for some of the CRISPRa proof of concept papers. They have now been corrected (in red).

Reviewer 2 Report

In this revised manuscript, the authors well-addressed my previous comments. I feel that the current manuscript is suitable to be published in Epigenomes.

Author Response

No more modifications asked.