Functions of Polycomb Proteins on Active Targets

Round 1

Reviewer 1 Report

This review describes the role of Polycomb complexes in gene transcription. The topic is timely and interesting for a wide audience.  Overall, this review accomplishes its goal of presenting evidence/characterizing the presence of Polycomb proteins at active gene targets. However, many recent studies elucidating the mechanisms by which Polycomb proteins activate target gene expression were performed using cancer cells lines. To improve the scholarly merit of this review, those studies in cancer should be included and more emphasis should be placed on discussing the mechanisms of PcG functions on active gene targets, such as the title suggests. Here are some suggestions and feedback that will hopefully help with formulating a final manuscript.

1. Section 2.5 begins with a discussion of phase separation as a mechanism of Polycomb-mediated higher-order chromatin structures. However, CBX2 is the only mammalian PcG found to form phase separated condensates. Phase separation as a Polycomb-mediated epigenetic mechanism is still not yet well understood. Although it may deserve a brief mention, it should not be portrayed as a primary mechanism of how Polycomb mediates higher order chromatin organization.

2. Section 3.1, the authors identifed a trend that the ratio of repressed to active Polycomb targets in primitive cells is high, and that the ratio tilts towards active targets in differentiated cells. Interestingly, cancer cells often exhibit features of de-differentiation. Therefore, in the context of the cancer-based studies cited in this paper, how do cancer cells fit into this trend?

3. The statement “formal proof for Polycomb recruitment to active targets via these transcription factors is still pending” is not entirely correct. Please see PMID: 17502350 and PMID: 32548262

4. Table 2 is very informative, but it should be revised: I think that the Nature Genetics paper from Cavalli did not show that RING1B and PCGF4 bind to active genes in human ESCs. Also, Di Croce and colleagues showed that RYBP and RING1B are recruited to active genes in mouse ESCs (PMID: 23273917). Reference 114 did not show that PRC1/2 are recruited to active genes, instead they showed that some genes have PRC complexes in one repressed allele but not in both. Finally, not only RING1B, but also PCGF2, CBX4 and CBX8 are recruited to active genes in breast cancer cells (PMID: 32548262, PMID: 30139998, PMID: 27346354, PMID: 25822021). All these references and clarifications should be included.

5. PcGs such as EZH2, CBX8, and PCGF2, are known to behave non-canonically and/or associate with non-PRC complexes to promote gene expression, particularly in cancer cells. Discussing these mechanisms of gene activation would add great value to this review.

6. The interchangeable use of semicolons and commas throughout the review makes it at times difficult to understand the text.

Author Response

1. Section 2.5 ….

We agree that the actual experimental evidence for condensates acting as major players in Polycomb-dependent higher order chromatin structures still lacks wide support. We, therefore, reordered and rephrased the corresponding paragraphs in section 2.5.

2. Section 3.1

Yes, we even considered a specific section for cancer cells given the relatively substantial contribution to the current knowledge on Polycomb supporting gene activity. A brief statement about our thinking on these activities in cancer cells compared to those in normal cell types has been added in the first paragraph of section 3.1.

3. The statement …

Indeed, reference PMID: 17502350, available at the time we were editing the final drafts of manuscript. The statement has been deleted and substituted by another describing the complex nature of the interactions involved.

4. Table 2

We have amended where necessary:

• Cavalli’s group paper re-analyzes data for ESCs for RING1B, but not, as noted by the reviewer, for PCGF4.
• Targets in ESCs occupied by RYBP and RING1B are enriched in transcriptionally active loci (PMID: 23273917) and the suggested reference of Morey et al, 2013 has been added.
• Brookes et al (originally ref. 114, now ref. 125) show, for six transcriptionally active genes, that the presence of RING1B occurs allelically, so that, precisely, binding would correlate with repression, no activation. In this regard, our intention including this piece of evidence in the table, is to indicate a functional relationship with active loci rather than to imply that occupancy equals activation. We distinguish between these two possibilities in section 3.3.
• Active PCGF2 targets in breast cancer cells (PMID:32548262, PMID:30139998) are now included. We have not included other references related to active targets because our intention is to focus on activities that could involve more than one Polycomb subunit. Thus, CBX8 acting together with WDR5 (PMID:27346354) can be considered similar to the interaction with components of the transcription elongation machinery (PMID: 23623499, PMID: 25921540) where activities appear CBX8-intrinsic, with little relation to the Polycomb set.
• PCGF2 role in transcriptional activation (PMID:25822021) inhibiting SUMOylation of DNA-binding proteins or by activating its deSUMOylation is, therefore, indirect and that is why we feel apart from activities associated to other subunits in the same table.

5. PcG such as…

The reviewer is right about these activities, and the point is partially mentioned also in the previous comment by the reviewer. However, the inclusion of all instances of Polycomb involvement in transcriptional activation exceeds the scope of our review. We have based our quest on Polycomb functions on active loci on cases that involve subunits known (or suspected) to act together with other Polycomb components rather than in non-PRC complexes. We thought it would be better focused and have added a sentence for clarification (second paragraph section 3).

6. The interchangeable use of semicolons…

Done.

Reviewer 2 Report

General evaluation:

This is a review carefully written about the topic of PcG proteins' participation
in transcriptionally active events. I consider that the text is comprehensive and
provides an elegant approach to multiple facets of this very interesting biological
problem. In sum, I found that the topic is of maximum interest, the writing style
of the article is correct and it is easy to read. However, I have several issues
that can be addressed by the authors to improve even more the final version of
the manuscript.

Major issues:

* 1. Introduction

1- It is surprising that, on a review on Polycomb, the manuscript does not contain a single ocurrence of
the words "epigenetics", "epigenome" or "epigenomics". In particular, the Introduction (line 23-30)
should distinguish between gene regulation governed at at genome level (DNA sequence, enhancers/promoters)
and at epigenetic level (histone marks, DNA methylation). The concept of histone tail modifications should be
introduced here also, together with the histone marks that will be used along the paper (e.g. H3K4me3 for
active promoters, H3K27me3 for repressed promoters, H3K27ac for active enhancers - which is shown in Figure2
but explained after in line 338, H3K4me1 for enhancers, etc.).

2- Genome topology effect over gene regulation is being intensively studied related to this problem. Although
the Abstract mentions this topic (line 10, "sometimes high order chromatin structures"), I missed the authors
include a paragraph and several key references in this field in the Introduction. Concepts such as TADs, LADs,
or A/B compartments could be enumerated here as they are related to chromatin accessibility of TFs to
promoters/enhancers. In fact, in line 222 (and later), the authors will talk about "long range promoter-promoter
and promoter-enhancer contacts" that contribute "to the overall 3D structure of the genome" so that this information
is necessary in the Introduction.

* 2. The Polycomb system

3- (line 82) The role of PRC1 and/or PRC2 can be affected by the action of ncRNAs. The authors can consider to
mention this line of research here (e.g. PMID: 23431328 Noncoding RNA and Polycomb recruitment / PMID: 29772549 Long
non-coding RNA and Polycomb: an intricate partnership in cancer biology / PMID: 30608221 Towards understanding
of PRC2 binding to RNA)

* 3. Polycomb and transcriptionally active loci

4- (line 259) I would say that one key aspect is the almost invariable presence of RING1B (alone) as the main PcG actor
in these positive transcriptional scenarios. Is it correct to add this fact to the list of items below? About this,
complementary to reference 128, the authors could introduce the novel work of the same lab (PMID: 32548262 (2020),
Estrogen induces dynamic ERα and RING1B recruitment to control gene and enhancer activities in luminal breast cancer),
which goes into a similar direction.

Minor issues:

* 1. Introduction

1- (line 35) I would be more specific and substitute "developmentally relevant genes" for "homeotic genes"
or even better "mutations in these genes produced homeotic transformations". There is a rich body of
literature reviewing these findings that can be helpful (e.g. see Box 1 in Blanco et al. The Bivalent
Genome: Characterization, Structure, and Regulation in Trends in Genetics 2020).

2- About Trithorax proteins (TrxG), I wonder whether it could be interesting to discuss what is their role
in the same active regulatory regions/scenarios that are now associated now to PcG (if known) or at least
to speculate about their co-ocurrence (or not) and provide further insight to future readers of this review.

3- (line 51) it is necessary to close the "(" in the previous line with another ")"

* 2. The Polycomb system

4- (line 70) in the sentence "terminal or histone H3", you should substitute "or" for "of"

5- (line 106) in "CBX/ and CBX8 in mammals", "CBX/" should be substituted for "CBX7"?

6- RYBP has been associated to positive transcriptional regulation in ESCs before (reference 32 and PMID: 23273917 - RYBP and
Cbx7 define specific biological functions of polycomb complexes in mouse embryonic stem cells). The authors
could mention this fact in line 111.

7- (line 132) A PRC2.1 subunit called EPOP has been recently discovered (acting opposite to PRC2.2, Jarid2).
EPOP could appear in Table 1 at least (PMID: 27863226 and PMID: 28061330, 2016).

8- (line 156) H3K27me1 (mono) is in fact associated to positive transcriptional regulation too (works in D. Pasini's lab).
I wonder whether the authors can correlate this with the activating role of PcG proteins that is explained next.

* 3. Polycomb and transcriptionally active loci

9- About reference 128, the authors employed a variant of ChIP-seq called ChIP-exo that seems to be more
precise in detecting the location of RING1B in its binding site. I wonder this could be commented when
mentioning discrepancies between references 123 and 128.

10- (line 374) The ubiquitination of H2B (differently from H2A) is associated to positive regulation of gene
expression. Is it known any association between H2B and RING1B acting as positive regulator in the literature?

Bibliography

11- Some references need a bit of cleaning (authors, additional text to be removed, etc.)

Author Response

*1. Introduction

1 - It is surprising…

We are sorry that a distinction between gene regulation at genome level and at epigenetic level is not clear. Possibly is because it reflects our idea that regulation at both levels are often functionally linked despite of being mechanistically different. Regarding the lack of the term epigenome and/or epigenetics, it was intentional to avoid the confrontation between the view that considers epigenetic everything related to histone modifications, etc and the view that the term should only be used for inherited events. To sort this out, we have added text (first paragraph, section 1) that introduces the term epigenetics, histone tail modifications and the marks that will appear later in the manuscript.

2- Genome topology…

Indeed, a great deal of work has been carried out to describe and understand genome compartmentalization and its possible function in gene control. We, however, intended to restrict our discourse on topological matters strictly related to Polycomb activities. Moreover, to our knowledge, the role(s) of these structures, for example TADs, on gene regulation is still under evaluation as transcriptional states, active or repressed, per se, appear as good predictors of compartmentalization. Nevertheless, as it is clear that readers may miss a reference to these topological structures a paragraph to add context has been added (first paragraph, section 1)

*2. The Polycomb system

3- (line 82)…

It is true that the role of ncRNAs on Polycomb function has been, surely is, an active line of research, mostly as a means of targeting to repressed targets. However, the reason why we did not mention ncRNAs regarding Polycomb activities is because, in our thinking, the recruiting functions are rather controversial, with models such as HOTAIR or XIST recruiting to Hox clusters of the inactive X chromosome, respectively, being re-evaluated. The finding that Polycomb-RNA interactions are of promiscuous nature or that can inhibit catalytic functions, made it, in our thinking, unnecessary to the general idea of the review. A sentence regarding this has been added in the second paragraph of section 2.4

*3. Polycomb and transcriptionally active loci

4- (line 259) …

Yes. We also realize the invariable presence of RING1B. We did not dare to state it like this in the light of the not so large body of evidence. However, after the suggestion we will add it to the list indicating the possibility of evaluation after new evidence.

About new work from Morey’s lab it has been added.

Minor issues:

* 1. Introduction

1- (line 35) I would be …

Phrases have been amended as requested. About including additional references reviewing the Polycomb field we included one (for historical reasons, not a review, reporting the first description of a Polycomb gene) together with two high-quality reviews. Note that, as stated in the introduction our manuscript does not intend to cover thoroughly the field but to focus on the less known, novel functions on active targets. Surely, additional references would not harm but, in our opinion, the ones quoted cover most of the potential interest of readers on the general aspects of the Polycomb field.

2- About Trithorax proteins (TrxG)…

The TrxG products are best known as antagonistic of repressive Polycomb functions. We mention them in the last paragraph of section 1. However in the absence of studies that assess their contributions to Polycomb functions on active targets we would prefer not to speculate.

Minor points 3, 4, 5 have been addressed as suggested.

6- RYBP has been associated to…

Rather than making a specific mention where suggested (line 111), we have included in table 2 that RYBP binds active targets in ESCs.

7- (line 132) A PRC2.1 subunit called EPOP…

The reviewer is right, but as stated in the text, our intention was not to produce a thorough collection of Polycomb subunits and complexes but to list only those that are mentioned in the text, in an attempt to assist setting them in context. EPOP is not mentioned and therefore was not included in the table.

8- (line 156) H3K27me1 (mono) is in fact…

As far as we understand work from Pasini’s lab, that H3K27me1 associates to transcriptional activation is not as much as an active function as the consequence of a non-targeted methylation. On the other hand, it is considered that H3K4me1 marks enhancers regardless of activity state (repressed, poised and active). Therefore, we do not find it likely to be related to activating roles of Polycomb proteins.

* 3. Polycomb and transcriptionally active loci

9- About reference 128,…

We think that ChIP-exo is not used in the work quoted as reference 128, and therefore the presumed discrepancy with results reported in reference 123 should not lie on the ChIP method (perhaps, more likely, in the computational strategies).

10- (line 374) The ubiquitination of H2B…

This ubiquitination is related to transcriptional elongation but as far as we can tell, it has no link with Polycomb activity. Obviously, it could be a target of Polycomb activity but we do not know about any reports supporting such as function yet.

* Bibliography

11- Some references need…

Done

Reviewer 3 Report

• Line 70: “N-terminal or histone H3” should be “N-terminal of histone H3”;
• Line 72: “…is catalyzed by heterodimeric RING E3 ligases present in PRC1 complexes [19,20].”. Also it is not clear why ref 20 (Keeping cell identity in Arabidopsis… ) is cited here since the review is focused on the vertebrates and drosophila. It could be pointed that (for consideration): It was shown in Drosophila that Sce is more active in PRC1-related dRAF complex (Lagarou A., Mohd-Sarip A., Moshkin Y.M., Chalkley G.E., Bezstarosti K., Demmers J.A., Verrijzer C.P. // Genes Dev. 2008. V. 22. № 20. P. 2799-2810.).
• Table 1: The fly homolog of L3mbtL2 is dSfmbt (not l(3)mbt). In column «protein motif» for L3MBTL2 “MBT domains” should be indicated.
• Line 107: “Polycomb (Pc)c in Drosophila” should be “Polycomb (Pc) in Drosophila”.
• Line 185: “through DNA subunits” should be “through DNA-binding subunits” ?
• Line 192: “depends on the concurrent activity of a varied set of DNA binding proteins [84]”. This multifaceted problem was discussed in excellent review from Kassis’s lab (Kassis and Brown, 2013. PMID: 23419717) and more recently in Epigenomes (Erokhin M., Georgiev P., Chetverina D. // Epigenomes. 2018. V. 2. № 1. P. 1.). These papers also should be cited here.
• Line 209: “2.5 High structure chromatin structures mediated by Polycomb” should be “2.5 High order chromatin structures mediated by Polycomb” ?
•  
• Lines 244-246: Similarly, in Drosophila tissues, subunits from PRC2 and PhoRC complexes associate to Ultrabithorax (Ubx) - another gene in the BX-C complex- both in wing and halter imaginal discs  where Ubx is active and inactive, respectively.

Here authors should also add the reference: “Langlais et al  2012 PMID: 23139817).

It should also be indicated that in transgenic Drosophila system was reported that neither a robust activator placed near PRE nor even active transcription through the PRE fails to dislodge the PcG proteins (Erokhin et al., 2015, PMID: 26504232; Kassis and Muller, 2015, PMID: 26567151).

Author Response

We have addressed all typos and reference amendments requested.

The only exception is that we rather do not add the suggested mention of a higher E3 ligase activity in Drosophila dRAF complex. It is true that, as in mammals, it appears that E3 ligases in variant PRC1 complexes are more active than those of canonical complexes. However, we feel it is a matter of (important) detail but dispensable, in our opinion, once that we stated in the text that our intention is not to review the Polycomb field as a whole but mostly what concerns to transcriptionally activate targets.

It should also be indicated that in transgenic Drosophila…

We have made reference to these observations to state that the presence of Polycomb proteins on chromatin is compatible with transcription (first paragraph section 3).