Variant PCGF1-PRC1 links PRC2 recruitment with differentiation-associated transcriptional inactivation at target genes

Polycomb repressive complexes-1 and -2 (PRC1 and 2) silence developmental genes in a spatiotemporal manner during embryogenesis. How Polycomb group (PcG) proteins orchestrate down-regulation of target genes upon differentiation, however, remains elusive. Here, by differentiating embryonic stem cells into embryoid bodies, we reveal a crucial role for the PCGF1-containing variant PRC1 complex (PCGF1-PRC1) to mediate differentiation-associated down-regulation of a group of genes. Upon differentiation cues, transcription is down-regulated at these genes, in association with PCGF1-PRC1-mediated deposition of histone H2AK119 mono-ubiquitination (H2AK119ub1) and PRC2 recruitment. In the absence of PCGF1-PRC1, both H2AK119ub1 deposition and PRC2 recruitment are disrupted, leading to aberrant expression of target genes. PCGF1-PRC1 is, therefore, required for initiation and consolidation of PcG-mediated gene repression during differentiation.

Minor points: 1. In the introduction, the authors should update to also mention recent work that established that PRC2.1 and PRC2.2 are both required for deposition of H3K27me3 (Helin and Bracken labs, 2019) and that variant PRC1 mediated H2AK119ub1, promotes primarily PRC2.2 mediated deposition of H3K27me3 (Klose and Pasini labs, 2020). 2. For Figure 1a and 1b, the authors should state that what type of ESCs were used for the ESC-EB differentiation. 3. The authors should show gene ontology enrichment analysis of the three groups of genes in Figure 1b. 4. In the results section, the authors should clarify whether their statement "The same trend was also seen, albeit in a weaker level, in Group 2 and 3 genes (Extended Fig. 2a)". relates to Figure 2a and not Extended Figure 2a. 5. For Figure 4a and 4b, the authors could perform a western blot analysis of Pcgf1 to confirm it was fully depleted upon Dox induction.
Reviewer #2 (Remarks to the Author): In this manuscript, Koseki and colleagues interrogated how PRC1 complexes regulate gene expression during embryonic stem cell (ESC) differentiation into embryoid bodies (EBs). By using gene expression profiles, ChIP-seq experiments, and KO cells each of the six PCGF orthologs, they found that PCGF1 was required for initial downregulation of a sub-set genes in two days-old EBs. Also, they suggest that PCGF1 was required for efficient RING1B, PRC2 and canonical PRC1 complexes during differentiation in this order of events. Albeit I believe that two day-old EBs are quite heterogeneous and therefore I would have chosen another differentiation model (i.e ESCs to EpiSCs, 2i +LIF to -LIF), their conclusions are well supported by the results. I do have the following specific comments which I hope will improve the impact of the study: 1. How many genes were deregulated in the EBs compared to the ESC? Also, I think it would be informative to include a 24h time point in both the RNA-seq experiments in WT cells. Moreover, given the fact that the 8h time point seems to be quite important for the interpretation of the results, occupancy of PCGF1, RING1B, PCGF2 and PRC2 should be assessed by ChIP-qPCR in differentiation time course (0, 8h, 24 and 48h). Analysis of KLF4, TBX3, and PDGFA genes should be enough. These experiments will confirm the recruitment events proposed in Figures 3 and 4. 2. Group 1 genes were affected in PCGF6 KO cells, thus authors should determine if PCGF6 binds to this group of genes in ESCs and EBs. Could it be possible to generate PCGF1/6 dKO cells? These experiments will elucidate whether both PCGF orthologs functionally cooperate in regulating Group 1 genes.
3. I was surprised that PCGF1 was not bound to Groups 1 to 3 in ESCs (Fig. 2a) and more specially to Group 2. These groups are based on RING1B occupancy in both ESC and EBs, and we know that PCGF1 is expressed and co-localizes with RING1B in ESCs. In line with this, I was also surprised with the low number of RING1B targets the authors found compared to published RING1B ChIP-seq in ESCs. The authors should clarify this. Moreover, H2AK119ub1 results from Figure 3a suggest that either PCGF1 was recruited to Group 1 genes in ESCs or that a PRC1 complex with PCGF6 deposited the mark.
4. If RING1B was not recruited to KLF4, TBX3, and PDGFA genes in ESCs but H2AK119ub1 was present, RING1A must be the E3-ligase involved in depositing H2AK119ub1 at these sites. Can the authors check this hypothesis by doing RING1A ChIP-qPCR? 5. The key control in Figure 3C should be to compare by RNA-seq the expression of Group 1 genes in control and RING1B mutant ESCs.
6. To functionally demonstrate that PCGF1 and PRC2 are required for full downregulation of Group 1 genes during ESC differentiation, authors should deplete EED (by shRNA) in PCGF1 KO cells.
Minor points: 1. PcG genes are regulated during the cell cycle. The authors should check the cell cycle profile of the cells during the differentiation time course presented in Figure S1C. If possible, PCGF orthologs expression by RT-qPCR should also be evaluated. 2. The authors should acknowledge the published work on PCGF2/MEL18 and PCGF4/BMI1 during ESC differentiation towards the mesoderm and ectoderm lineages, respectively. 3. "The same trend was also seen, albeit in a weaker level, in Group 2 and 3 genes (Extended Fig. 2a)." should read (Fig. 2a). 4. Are SOX4 and GRHL2 genes PCGF1 targets in ESCs? 5. I don't agree that KDM2B binding was unchanged in In Fig. S4, I think it was not recruited.
Reviewer #3 (Remarks to the Author): In the manuscript entitled "Variant PCGF1-PRC1 links PRC2 recruitment with differentiation-associated transcriptional inactivation at target genes" the authors elucidate the contribution of distinct Polycomb activities in the downregulating the expression of target genes upon differentiation of mESC into EBs. The authors provide evidences that the PCGF1-containing PRC1 complex is crucial to facilitate differentiationassociated down-regulation of a specific group of genes upon transition from mESC to EBs. They suggest a model in which the PCGF1-vPRC1 complex is able to initiate PcG-repressive domain formation by recruiting RING1B to deposit H2AK119ub1 via CpG island recognition by KDM2B at target genes. This allows subsequent recruitment of PRC2, deposition of H3K27me3 followed by PCGF2-cPRC1 recruitment enforcing the PcG-mediated gene silencing.
Overall the data presented are of good quality and largely clear. The way some expression-location analyses are presented is sometime difficult to follow and revising their presentation may help to follow their experiments. Some data are also redundant and repeated in the figures ( Fig.1d and Fig.3c), this should be avoided. The presented data and proposed model is rather confirmatory with limited novelty. However, the major claim of this manuscript is the central involvement of PCGF1 in the de novo establishment of PcGrepressive domains at developmental genes. This is in some part convincing, however, the manuscript also fell short in fully demonstrating the exclusive role of PCGF1 in mounting repression at targets during differentiation.
The use of EBs remains rather dishomogeneous as cell differentiation is not fully restricted to a single fate and the results could be affected by a diverse abundance in cell types. I do understand that the time at which EBs are taken is rather short but these results will be strengthen by also differentiating ESC into more homogenous fates using alternative protocols (see also following comment).
The authors undermine the effects observed with the PCGF6 KO, however, it remains unclear why the same group of genes fail to be repressed in the absence also of PCGF6 (Fig.1d). Is PCGF6 involved in PCGF1 mediated silencing? If so, how does this reconcile with the central recruitment of H3K27me3 and canonical PRC1 shown later in the figures? The interplay between non-canonical PRC1 complexes must be better investigated in order to determine whether PCGF1 plays an apical role in de novo PcG repression of simply acts redundantly with other PRC1 complexes as previously suggested by the authors in ESC (Fursova et al. Mol Cell 2019). The authors cannot exclude that lack of repression is a secondary effect of differentiation failure of PCGF1 KOs in EBs (or PCGF6 KOs). De novo establishment of target repression with other stimuli is therefore required to reinforce their observations. The authors also fail to determine how PCGF1 activity is regulated at physiological levels. Using ectopic expression of PCGF1 they show a nice recruitment at Group 1 genes in EBs (Fig.2a). However the same trend is also evident for Group 3 genes which undergo transcriptional activation in normal EBs (Fig.2b). This suggest that the observed PCGF1 recruitment is rather the consequence of overexpression conditions. Indeed, it also remains unclear why H2Aub deposition is not modulated in the presence of a clear recruitment of PRC1 activity at these genes. This deposition is not fully dependent on PCGF1 activity ( Fig.3a) suggesting that multiple complexes are already active at these sites (Fig.3a). The only activity which is clearly modulated is the recruitment of PRC2 and canonical PRC1. This likely explains the strong recruitment of RING1B as was previously shown by others to be dependent on non-canonical PRC1 forms (Tavares et al. Cell 2012). Using catalytically dead RING mutants, the authors further show that deposition of H2Aub is required for PRC2 recruitment at de novo EBs targets (group 1). However, lack of modulation in H2Aub deposition does not fit with a central role of H2Aub in building de novo repressive PcG modules. The How is it possible that H2Aub is required to repress these genes in EBs but is not differentially deposited?
The figure 4 is fine, however, it does not add much to the story and shows a poor modulation of PRC2 recruitment. The modulation of H3K27me3 and H2Aub at artificial CGI is also not convincing and it remains unclear while these modifications have not been tested with quantitative PCRs like the rest.
We would like to submit the revised version of our manuscript N-COMMS-20-04383 to Nature Communications. We greatly appreciate the comments and suggestions made by the reviewers. To address the reviewers' comments, we have newly added (i) Figure 4 Point-by-point replies to reviewers:

Reviewer #1:
Polycomb repressive complexes (PRC) are essential for maintaining cellular identity during development and can be divided to two major classes, PRC1 and PRC2. This manuscript explores the roles and molecular mechanisms of the Pcgf1-PRC1 complex during mouse embryonic stem cell (ESC) differentiation to embryoid bodies (EBs). The authors generated comprehensive genomic data, including RNA-seq, ChIP-seq and CUT&Tag-seq, in series of conditional knockout ESCs of various Pcgf proteins, both before and after induction to differentiate to EBs. Their main finding is that the Pcgf1-PRc1 complex is recruited to Polycomb target genes and then this leads to deposition of H2AK119ub1 and subsequent deposition of PRC2. The experiments are very well executed and the work is clearly explained. This will be of broad interest to those in the chromatin and stem cell fields who are interested in the regulation of differentiation by Polycomb proteins. My comments below are minor suggestions to further help and clarify some aspects. We thank the reviewer for pointing out these very relevant papers, which are now included in the introduction (page 2, lines 20-24) and discussion (page 10, lines 24-29). Figure 1a and 1b, the authors should state that what type of ESCs were used for the ESC-EB differentiation.

For
We have added the information on origin and culture conditions of ESCs in the result (page 3, lines 17-23), and materials and methods (page 18, line 24-page 19, line 6). Figure 1b.

The authors should show gene ontology enrichment analysis of the three groups of genes in
We appreciate the reviewer's suggestion. We have added the GO data in Extended Figure 1. Please also see the results section (page 3, lines 30-32).

4.
In the results section, the authors should clarify whether their statement "The same trend was also seen, albeit in a weaker level, in Group 2 and 3 genes (Extended Fig. 2a)". relates to Figure 2a and not Extended Figure 2a.
We have corrected the text accordingly. Figure 4a and 4b, the authors could perform a western blot analysis of Pcgf1 to confirm it was fully depleted upon Dox induction.

For
The cell line used in our paper is the same one reported in Almeida et al. (2017) and . As both of those papers show western blot results, we did not perform this experiment again. We would also like to point out that in our experiment, Dox is used to activate TRE-mediated expression of EGFP, but not modulate PCGF1 expression.

Reviewer #2:
In this manuscript, Koseki and colleagues interrogated how PRC1 complexes regulate gene expression during embryonic stem cell (ESC) differentiation into embryoid bodies (EBs). By using gene expression profiles, ChIP-seq experiments, and KO cells each of the six PCGF orthologs, they found that PCGF1 was required for initial downregulation of a sub-set genes in two days-old EBs. Also, they suggest that PCGF1 was required for efficient RING1B, PRC2 and canonical PRC1 complexes during differentiation in this order of events.  (0, 8h, 24 and 48h). Analysis of KLF4, TBX3, and PDGFA genes should be enough. These experiments will confirm the recruitment events proposed in Figures 3 and 4.
Upon ESC-to-EB differentiation, we found 765 upregulated genes (log2FC > 2) and 625 downregulated genes (log2FC < -2). (page4, lines 2-4). We appreciate the suggestion to investigate bindings of RING1B, PCGF2 and SUZ12 during ESC-to-EB transition. We performed time course ChIP-qPCR analysis and We appreciate the reviewer's comment. Unfortunately, ChIP-grade antibody for PCGF6 is not available, and therefore we could not perform this experiment.
Instead, we investigated RING1B distribution in Pcgf6-KO in ESCs and EBs, which we believe is the major issue here. We did not find noticeable changes in RING1B binding in Pcgf6-KO ESCs or EBs and, therefore, concluded that PCGF1 played a major role to bring RING1B to the Group 1 genes during ESCto-EB differentiation. These results are shown in Extended Figures 2d and 2e, and described in the results section (page 4, line 31-page 5, line 7). (Fig. 2a) and more specially to Group 2. These groups are based on RING1B occupancy in both ESC and EBs, and we know that PCGF1 is expressed and co-localizes with RING1B in ESCs. In line with this, I was also surprised with the low number of RING1B targets the authors found compared to published RING1B ChIP-seq in ESCs. The authors should clarify this. Moreover, H2AK119ub1 results from Figure 3a suggest that either PCGF1 was recruited to Group 1 genes in ESCs or that a PRC1 complex with PCGF6 deposited the mark.

I was surprised that PCGF1 was not bound to Groups 1 to 3 in ESCs
Upon the comments from the reviewer, we re-examined PCGF1 distribution by using anti-PCGF1 antibody raised by , and replaced TY1tagged PCGF1 (expressed in Pcgf1-KO ESCs) ChIP-seq with endogenous PCGF1 ChIP-seq. We successfully detected binding of endogenous PCGF1 to target genes in both ESCs and EBs (not only Group 2 and 3 genes, but also For RING1B, we identified 1615 and 1706 target genes in ESCs and EBs, respectively. In our previous study (Endoh et al., 2017, eLife), we identified 2959 genes bound by RING1B. Although the number of genes found in this study is smaller than our previous study, such differences are mainly caused by threshold setting. To support this notion, we observed most of the genes identified in this study were also included in our previous study (given the list is quite long, the comparison data is not shown in the manuscript).

If RING1B was not recruited to KLF4, TBX3, and PDGFA genes in ESCs but
H2AK119ub1 was present, RING1A must be the E3-ligase involved in depositing H2AK119ub1 at these sites. Can the authors check this hypothesis by doing

RING1A ChIP-qPCR?
We thank the reviewer for this comment. The short answer is that there is no ChIP-grade antibody for RING1A as far as we know. Instead, our results indicate that newly recruited RING1B to Group 1 genes upon ESC-to-EB differentiation (Figures 2b, 2c) mainly represents RING1B incorporated in canonical PRC1 rather than variant PCGF1-PRC1, as described in page 6, line 5-page 7, line 3.
We indeed found substantial binding of PCGF1 and RING1B in Group 1 genes in ESCs (Figures 1c, 1d, and Figures 2b, 2c); which mediates H2AK119ub1, as confirmed by deletion of Pcgf1 (Figure 3a), and disruption of the catalytic activity of Ring1A/B (Figure 3d). Therefore, modest but clear binding of PCGF1-PRC1 is observed at Group 1 genes, which we believe contributes to H2AK119ub1 deposition (page 6, line 5-page 8, line 18). Figure 3C should be to compare by RNA-seq the expression of Group 1 genes in control and RING1B mutant ESCs.

The key control in
We appreciate the reviewer's comment. We have now modified Figure 3c

To functionally demonstrate that PCGF1 and PRC2 are required for full downregulation of Group 1 genes during ESC differentiation, authors should deplete EED (by shRNA) in PCGF1 KO cells.
We thank the reviewer for this comment. We agree that it is important to elucidate how PCGF1 and PRC2 synergistic downregulate target genes. In this study, we propose that PCGF1-dependent H2AK119ub1 recruits PRC2 (Figure 5d We have investigated proliferation of Pcgf1-KO during ESC-to-EB differentiation and observed moderate decrease of the proliferation rate. We have also examined the expression of Pcgf factors during during ESC-to-EB differentiation as shown in Extended Figure 2a (page 4, lines19-21).

The authors should acknowledge the published work on PCGF2/MEL18 and PCGF4/BMI1 during ESC differentiation towards the mesoderm and ectoderm lineages, respectively.
We thank the reviewer for pointing out this issue. We have included a paper by Morey et al (Cell stem cell, 2015) in the discussion section (page 12, lines 7-8). Fig. 2a)." should read (Fig. 2a).

3."The same trend was also seen, albeit in a weaker level, in Group 2 and 3 genes (Extended
Thank you for this comment. We have corrected the text accordingly.

Are SOX4 and GRHL2 genes PCGF1 targets in ESCs?
Yes, this is shown in Figure 2c. Fig. S4, I think it was not recruited.

I don't agree that KDM2B binding was unchanged in In
Thank you for this comment. We have revised the text accordingly (Page 10, lines 4-5).

Reviewer #3 (Remarks to the Author):
In the manuscript entitled "Variant PCGF1-PRC1 links PRC2 recruitment with differentiation-associated transcriptional inactivation at target genes" the authors elucidate the contribution of distinct Polycomb activities in the downregulating the expression of target genes upon differentiation of mESC into EBs. The authors provide evidences that the PCGF1-containing PRC1 complex is crucial to facilitate differentiation-associated down-regulation of a specific group of genes upon transition from mESC to EBs. They suggest a model in which the PCGF1-vPRC1 complex is able to initiate PcG-repressive domain formation by recruiting RING1B to deposit H2AK119ub1 via CpG island recognition by KDM2B at target genes. This allows subsequent recruitment of PRC2, deposition of H3K27me3 followed by PCGF2-cPRC1 recruitment enforcing the PcG-mediated gene silencing.
１．Overall the data presented are of good quality and largely clear. The way some expression-location analyses are presented is sometime difficult to follow and revising their presentation may help to follow their experiments. Some data are also redundant and repeated in the figures (Fig.1d and Fig.3c), this should be avoided. The presented data and proposed model is rather confirmatory with limited novelty. However, the major claim of this manuscript is the central involvement of PCGF1 in the de novo establishment of PcG-repressive domains at developmental genes. This is in some part convincing, however, the manuscript also fell short in fully demonstrating the exclusive role of PCGF1 in mounting repression at targets during differentiation.
We regret the confusion in Figures 1d and 3c. Figure 3c showed the results for ２．The use of EBs remains rather dishomogeneous as cell differentiation is not fully restricted to a single fate and the results could be affected by a diverse abundance in cell types. I do understand that the time at which EBs are taken is rather short but these results will be strengthen by also differentiating ESC into more homogenous fates using alternative protocols (see also following comment).
We appreciate the reviewer's concern. A similar comment was also made by reviewer 2. We maintained the ESCs in LIF and 3i, on feeder MEFs, and induced differentiation to EBs by depleting LIF, 3i, and feeder MEFs. In this condition, we observed that ESCs differentiate into EBs in a synchronous manner. The detailed description of the culture conditions is now added to the text (page 3, lines 17-23: page 18, line 24-page 19, line 6). In addition, we confirmed PCGF1dependent accumulation of RING1B to Klf4 and Tbx3, by differentiating ESCs into epiblast like cells (epiLCs) (Extended Fig. 2g, 2h) (page 5, lines 11-16).
３．The authors undermine the effects observed with the PCGF6 KO, however, it remains unclear why the same group of genes fail to be repressed in the absence also of PCGF6 (Fig.1d) (Fursova et al. Mol Cell 2019).
We appreciate this comment. A similar concern was also raised by reviewer 2 (Comment #2). We therefore investigated if RING1B distribution was altered in Pcgf6-KO in ESCs and EBs. We did not find changes in RING1B binding in Pcgf6-KO ESCs or EBs and, therefore, concluded that PCGF1 played a major role to bring RING1B to the Group 1 genes during ESC-to-EB differentiation. These results are shown in Extended Fig. 2d and 2e  The idea that H2AK119ub1 silences gene expression, is still controversial to some extent, and needs further validation (see discussion, page 10, lines 16-33).
However, we did note a role for H2AK119ub1 to recruit PRC2, and canonical PRC1, upon transcriptional down-regulation. This role is independent from differentiation of ESCs, and instead is dependent on prior transcriptional downregulation (see Figures 4 and 5).
５ ． The authors also fail to determine how PCGF1 activity is regulated at physiological levels. Using ectopic expression of PCGF1 they show a nice recruitment at Group 1 genes in EBs (Fig.2a). However the same trend is also evident for Group 3 genes which undergo transcriptional activation in normal EBs (Fig.2b). This suggest that the observed PCGF1 recruitment is rather the consequence of overexpression conditions. Indeed, it also remains unclear why H2Aub deposition is not modulated in the presence of a clear recruitment of PRC1 activity at these genes. This deposition is not fully dependent on PCGF1 activity (Fig.3a) suggesting that multiple complexes are already active at these sites (Fig.3a) (Blackledge et al. Mol Cell 2020) as well as with the marginal role of PRC2 and canonical PRC1 complexes in transcriptional repression (Riising et al. Mol Cell 2014, Tamburri et al. Mol Cell 2020and Fursova et al. Mol Cell 2020.
How is it possible that H2Aub is required to repress these genes in EBs but is not differentially deposited?
To address this comment, we have re-examined PCGF1 distribution by using an anti-PCGF1 antibody raised by , and replaced Ty1-PCGF1 ChIP-seq data with endogenous PCGF1 ChIP-seq. We successfully detected binding of endogenous PCGF1 to target loci including not only Group 2 and 3 genes, but also Group 1 genes in ESCs (Fig. 2b, 2c). PCGF1, therefore, binds target CGIs in a constitutive manner; and mediates H2AK119ub1 at Group 1 genes, as well as Group 2 and 3 genes, in ESCs (Fig. 3a, Extended Fig. 3a).
Importantly, we also found modest but clear binding of RING1B in Group 1 genes in ESCs (Fig. 1c, d, and Fig. 2b, c). This binding likely mediates H2AK119ub1, as represented by a lack of H2AK119ub1 depositions in RING1A/B catalytic mutant ESCs (Fig. 3d). We, therefore, conclude that a fraction of RING1B that forms complexes with PCGF1, and constitutively binds to the Group 1 genes, irrespective of the transcriptional status. This RING1B/PCGF1 fraction mediates H2AK119ub1 at the Group 1 genes in ESCs. In contrast, increase of RING1B binding at the Group 1 genes during ESC-to-EB transition represents RING1B incorporated in canonical PRC1, which is recruited to target genes via recognition of PRC2-mediated H3K27me3 (3a, 3b). We therefore suggest that PCGF1-PRC1 mediates H2AK119ub1 at not only transcriptionally silenced genes, but also active genes (such as Group 1 genes). We further reveal that at transcriptionally active genes, H2AK119ub1-dependent recruitment of PRC2 and canonical PRC1 is inhibited by transcriptional activity per se (Figures 4, 5d). We speculate that the H2AK119ub1-dependent repressive pathway counteracts transcriptional activity.
６．The figure 4 is fine, however, it does not add much to the story and shows a poor modulation of PRC2 recruitment. The modulation of H3K27me3 and H2Aub at artificial CGI is also not convincing and it remains unclear while these modifications have not been tested with quantitative PCRs like the rest.
In Figure 4 of the revised version, we reveal the impact of prior transcriptional inactivation by Triptolide on H2AK119ub1-dependent recruitment of PRC2 and canonical PRC1 (Figure 4, Extended Figure 4). In Figure 5 (Figure 4 in the previous version), we addressed the same issue using a different approach to avoid potential off-target effects by Triptolide. These results were further validated by ChIP-qPCR (Figure 5c, Extended Figures 5b, 5c).
data shows very nicely that in the control cell line, TetR-KDM2B recruits PCGF1 (and RING1B) to the TetO array. However, following PCGF1 depletion there is a loss of PCGF1 ChIP-qPCR signal (and a failure to recruit RING1B). Taken these together, we concluded the specificity of the antibody was sufficiently validated.
In the present study, as the specificity of this antibody is sufficiently validated, we did not perform additional ChIP-seq analysis by using PCGF1-KO ES cells, which should provide the best control. I, however, think the quality of our data in this present study could be assessed by comparing them with those of Fursova paper at the genes of interests (in this case, Klf4, Tbx3, Pdgfa, Sox4, Grhl2 and Gapdh)(Appendix Figure 2F GSE119618). In Fursova paper, PCGF1 signals, which were completely depleted upon tamoxifen-induced deletion of PCGF1/3/5, associated with CpG islands associated with these genes (Appendix Figure 2F GSE119618). In our data shown in Appendix Figure  2F, PCGF1 signals accumulated at CpG islands in ES cells and EB and overlapped with those identified by Fursova et al. (Appendix Figure 2F This study). I, therefore, think specificity of the antibody in our ChIP-seq experiments is sufficiently qualified. Accordingly, we have added a sentence to state the origin of this antibody in page 6, lines 13-14 in the revised manuscript.