CHD3 regulates BMP signalling response during cranial neural crest cell specification

  1. Zoe H. Mitchell
  2. Joery den Hoed
  3. Willemijn Claassen
  4. Martina Demurtas
  5. Laura Deelen
  6. Philippe M. Campeau
  7. Karen Liu
  8. Simon E. Fisher
  9. Marco Trizzino

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

CHD3 is a component of the NuRD chromatin remodeling complex. Pathogenic CHD3 variants cause Snijders Blok-Campeau Syndrome, a neurodevelopmental disorder with variable features including developmental delays, intellectual disability, speech/language difficulties, and craniofacial anomalies. To unveil the role of CHD3 in craniofacial development, we differentiated CHD3 -KO induced pluripotent stem cells into cranial neural crest cells (CNCCs). CHD3 expression is low in wild-type iPSCs and neuroectoderm, but upregulated during CNCC specification, where it opens the chromatin at BMP-responsive enhancers, to allow binding of DLX5 and other factors. CHD3 loss leads to repression of BMP target genes and an imbalance between BMP and Wnt signalling, ultimately resulting in aberrant mesodermal fate. Consequently, CNCC specification fails, replaced by early-mesoderm identity, which can be partially rescued by titrating Wnt levels. Our findings highlight a novel role for CHD3 as a pivotal regulator of BMP signalling, essential for proper neural crest specification and craniofacial development.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    1. Description of the planned revisions

    Insert here a point-by-point reply that explains what revisions, additional experimentations and analyses are planned to address the points raised by the referees.

    The following revisions are in progress:

    - From Reviewer-1: The authors observe defects in CNCCs through genomic experiments. It would be really nice to perform simple wound healing/scratch assays and/or transwell assays to test if the CNCC migration phenotype is reduced in the CHD3 KO as well which would support the transcriptomic data.

    As recommended by the Reviewer, we are performing a transwell assays to investigate whether CHD3 loss leads to defects in cell migration. These experiments should be completed in the next two weeks.

    __- From Reviewer-2: __Since CHD3 shows a progressive upregulation in expression during CNCC differentiation (Fig. 2E), one hypothesis can be that it is not necessary involved in the activation of the CNCC programs but instead it is involved in maintaining these programs active - by keeping regulatory elements accessible. Thus, authors should check expression of CNCC markers, and EMT genes at the same time point than Fig. 2E in both WT and KO cells.

    As recommended by the reviewer we are differentiating the cells to perform RT-qPCR timecourse for CNCC and EMT markers. These experiments will be completed in the next two weeks.

    __- From Reviewer-2: __It has been shown that CNCC regulatory elements controlling differentiation genes are primed/accessible prior migration (PMID: 31792380; PMID: 33542111). Since the authors claim "CHD3 may have the role of priming the developing CNCCs to respond to BMP by opening the chromatin at the BMP responsive enhancers", it will be good to perform ATAC-seq are several time point during the differentiation process to assess the dynamic of chromatin reorganization to see when the switch to mesoderm fate occurs and how accessibility of BMP responsive element changes in WT and KO cells during CNCC differentiation to be able to demonstrate the KO fail to make BMP responsive element accessible or whether it is a defect in the maintenance of this accessibility.

    As recommended by the Reviewer, we are differentiating the cells to perform ATAC-seq timecourse. These experiments will be completed in the next two/three weeks.

    2. Description of the revisions that have already been incorporated in the transferred manuscript

    The following revisions have already been carried out:

    Reviewer1

    1. Figure 1 presents nice confirmation of the CHD3 KO cell lines being used. However, given that these cell lines were previously published, I suggest moving these data to the supplement. As suggested by the Reviewer, we moved most of Figure 1 to the supplement, merging the remaining Figure 1 with Figure 2.

    In the results section for Figure 1, the authors discuss the CHD3 heterozygotes, but I only see the KO cell line data presented. It would be especially nice to see the protein levels of Chd3 in the het.

    As suggested, we have now performed western blot and qPCR for CHD3 in the heterozygous line and added it to Supplementary Figure S1.

    The authors discuss which genes are up and downregulated in the Chd3 KO D18 RNA-seq, and show a clear heatmap in Figure 2A for WT cells. The same heatmap for candidate genes discussed in the results would be appreciated for Chd3 KO.

    As recommended by the Reviewer, we have added CHD3-KO RNA-seq to the heatmap in Fig. 2A.

    In general 2-3 replicates are presented. While the authors are showing heatmaps for selected locations for individual clones, which is appreciated (ex: Figure 4B and Fig 6), the QC for data quality is missing. For example, show spearmean correlation across the genome for datasets as a supplement.

    We performed spearman correlation of ATAC-seq and RNA-seq data, which confirmed the replicates are very highly correlated, and created new dedicated supplemental figures (Supplemental Figures S3, S4, S5, S6, S7).

    In the section discussing the results presented in Figure 4, the authors discuss the ATAC-seq peak number changes and overlap with gene expression changes. However, the overlap with gene expression changes is not shown. Making a simple Venn diagram would help readers.

    As suggested, we added a Venn diagram with ATAC-seq/RNA-seq overlap in Figure 3D.

    In addition, showing a heatmap for unchanged ATAC-seq peaks can help to demonstrate the increase/decrease.

    As recommended, we have added an heatmap for unchanged ATAC-seq regions as Supplementary Figure S7.

    In Figure 6, the authors present ChIPseq data for CHD3 in D14 and D18 samples, focusing on locations losing or gaining accessibility. What is enrichment at unchanged sites? Is CHD3 specifically enriched at changed locations? Then what about over genes with altered gene expression vs not changed? Is CHD3 only bound to distal elements? Performing an analysis of the peak distribution, perhaps with ChromHMM or other methods to look at promoter vs enhancer vs other locations. These types of analyses could really enrich the interpretation of direct CHD3 function.

    Unfortunately, there is no ChromHMM data for neural crest cells, nor for closely related cell types. Therefore, to address the Reviewer's suggestion, we have taken two approaches: 1) We have further broken down the distribution of the peaks, dividing them between intergenic, intronic, exonic and TSS. Moreover, we have leveraged publicly available H3K27ac ChIP-seq data generated (by our group) in iPSC-derived CNCCs to identify CHD3 peaks that are decorated by this histone modification which typically marks active enhancers. This analysis revealed that 91% of the peaks are either intergenic (50%) or intronic (41%) and that ~a third of the peaks are decorated with H3K27ac in human iPSC-derived CNCCs, suggesting that they are bona-fide active enhancers in this cell type.

    Related to the above, I am not sure if there is a phenotypic test for enhanced mesoderm. I suspect only IF/expression and morphology are possible, which the authors did. However, sorting the cells (with some defined markers) to ask how many are mesoderm-like vs CNCC in WT vs CHD3 KO would give some information outside of the bulk expression data.

    The manuscript already included IF experiments for mesodermal markers, which clearly show that nearly all the cells acquired the mesodermal fate. See for example Brachyury IF in Figure 2E.

    Minor points Reviewer-1:

    1. 1A seems to fit better with Figure 2. Done
    2. The authors say that the KO cell lines are not defective in pluripotency, but Figures 1G suggests a slight decrease in SSEA-1. Is this reproducibly observed? It is not statistically significant and not reproducibly observed.
    3. Would be nice to show number of up and downregulated genes in volcano plots for fast viewing of readers (ex: Fig 2B). We have modified the volcano plot as suggested.
    4. Is it fair to use violin plots when data points are only 2-3 replicates (as in Figures 2C, 3D). To address this, we have layered the actual datapoints on top of the violin plots.

    The labels in Fig 4A and 5E are very hard to read.We have changed color to improve readability.

    1. For browser tracks, the authors show very zoomed in examples (Fig 4C, and especially Fig 6C). showing a bit more of the area around these peaks would give readers a more clear appreciation of the data. Related to browser tracks, including more information just as including the gene expression changes (such as in Fig 6C) to enhance the interpretation of the impact of Chd3 binding, accessibility change and then, I presume, reduced Sox9 expression. Similar suggestion for Figure 4C, where I anticipate coordinate transcription changes of the associated genes. We have zoomed out the tracks, as suggested, and added expression data next to them.
    2. Do the authors observe any clone variability between the two CHD3 KO clones? There is variability I see in some of the heatmaps, but don't know if that it is because of clones or technical variation. We do not observe any significant variability between the clones.

    Reviewer-2

    1. What is the expression level of CHD3 in the heterozygote line? Does the remaining allele compensate for the loss which will explain the absence of phenotype?

    Ass suggested also by Reviewer-1, we have performed western blot for CHD3 in the heterozygous line and added it to Supplementary Figure S1. The bot shows that the remaining allele does not compensate. However it is likely that even a reduced amount of wild-type CHD3 is sufficient for proper CNCC specification.

    The authors should use the term "regulatory elements" instead of "enhancers" as they can act either as activator or repressors.

    As suggested, we have changed nomenclature from enhancers to cis-regulatory elements.

    On the same line, while the authors indicate "Motif analysis of the enhancers aberrantly active in CHD3-KO cells ", they haven't shown these are active. They should say they perform the analysis on regulatory elements aberrantly accessible in CHD3 KO. Done.

    See point 3 above.

    The rationale that led the authors to focus on genes typically expressed in the primitive streak and in the early pre-migratory mesoderm, and BMP responsive transcription factors could be better explained. Are they part of the most deregulated genes in the RNA-seq analysis?

    Not only mesodermal genes are among the most upregulated genes in the RNA-seq, but the motifs for the transcription factors encoded by these genes (e.g. TBR2, Brachyury, GATA, TBX3, TBX6) are among the most frequently represented in the aberrantly accessible cis-regulatory elements. The same applies to BMP responsive factor, but the other way around (they are downregulated and enriched in the aberrantly closed ATAC-seq regions).

    In the absence of CHD3, BMP response is not effective. While the authors nicely showed this is linked with changes in chromatin accessibility, it is necessary to check the expression levels of BMP receptors in CHD3 KO cells.

    We have checked the expression of these genes, and they were not differentially expressed. This is consistent with the downstream response being affected rather than ligand binding to the receptors.

    Aberrant early mesoderm signature of the CHD3-KO cells needs to be better shown. It is not obvious from the GO analysis in Fig. 2 and the authors then showed expression of some markers but it is unclear how they picked them up.

    See point 5: not only mesodermal genes are among the most upregulated genes in the RNA-seq, but the motifs for the transcription factors encoded by these genes (e.g. TBR2, Brachyury, GATA, TBX3, TBX6) are among the most frequently represented in the aberrantly accessible cis-regulatory elements. See for example expression levels of typical mesodermal genes below:

    EOMES - upregulated log2FC: 5.5

    TBXT - upregulated log2FC: 4.6

    MESP1 - upregulated log2FC: 4.7

    MIXL1 - upregulated log2FC: 5.4

    TBX6 - upregulated log2FC: 3.2

    MSGN1 - upregulated log2FC: 4.6

    HAND1 - upregulated log2FC: 5.5

    The authors claim CHD3 directly binds at BMP responsive enhancers, but in the figure, they show the data for all the region gaining or losing activity. It will be nice to add the information for the BMP responsive elements only.

    As recommended, we have added an heatmap for BMP responsive regions only, clearly showing that CHD3 binds them (Supplementary Figure S7).

    The authors need to support better that CHD3-KO express more Wnt signaling/activity.

    We have checked expression of many genes that are typically Wnt responsive during mesoderm specification (see also point 7). These include:

    EOMES - upregulated log2FC: 5.5

    TBXT - upregulated log2FC: 4.6

    MESP1 - upregulated log2FC: 4.7

    MIXL1 - upregulated log2FC: 5.4

    TBX6 - upregulated log2FC: 3.2

    MSGN1 - upregulated log2FC: 4.6

    HAND1 - upregulated log2FC: 5.5

    These data clearly support that the Wnt-mediated mesodermal program is markedly upregulated.

    Minor points Reviewer-2:

    1. In the discussion, the authors could indicate whether CHD3 mutants somehow phenocopies some of the craniofacial defects observed in DLX5 mutant patients. Done.
    2. It is not indicated were to find the data regarding expression epithelial and mesenchymal genes in the CHD3-KO cells. They are in the heatmap in Fig. 1C.
    3. Authors could add in the discussion what is known about how CHD3 function changes from opening or closing chromatin is very intriguing a could be discussed. To our knowledge, nothing is known on this. CHD3 is significantly understudied.

    OPTIONAL: While this is not necessary for the current study, it is very intriguing that other CHD family member do not compensate. How this tissue or DNA sequence activity is achieved could be discussed. What are CHD4 or CHD5 expressed during CNCC differentiation? Could they be used to rescue the CHD3 KO phenotype? While this may be difficult to test, it could perhaps be discussed.

    We have added a paragraph on this in the discussion.

    3. Description of analyses that authors prefer not to carry out* *

    From Reviewer 1: Given the changes in the CHD3-KO accessibility are mostly gene distal, are there existing Hi-C/microC/promoter CaptureC or other that can be used to ask if these are interacting with the predicted genes?

    We are not aware of this type of essays being performed genome-wide in human CNCCs. The only studies performed in human CNCCs are SOX9-centred. Looking at 3D chromatin conformation would also be out of the scope of the paper.

    From Reviewer-2:

    OPTIONAL: Does increasing BMP concentration early during CHD3 KO differentiation has a better effect at rescuing CNCC differentiation?

    Indicated by Reviewer as OPTIONAL. We do not think that adding BMP earlier on would make a significant difference in rescuing CNCC differentiation.

    From Reviewer-1: Are the results observed NuRD-based or CHD3 NuRD independent functions? Looking at other NuRD subunit binding or effects in differentiation would help to dig into this a bit more. I realize this is a bit of a big ask, so I am not asking for everything. Are there existing binding data in CNCCs for a NuRD subunit that could be examined for overlap in where these changes occur, for example? I want to be clear I am not asking the authors to do all the experiments for an alternative NuRD subunit.

    There are no existing data on NuRD binding in CNCCs. However, while the Reviewer is definitely not recommending generating new data in this regard, we still decided to make an attempt at performing ChIP-seq for the core NuRD subunit MBD3 in our CNCC. We will only make one attempt (multiple replicates), and if it does not work we will not pursue this any further as the Reviewer clearly stated that this is not necessary nor required and we do not want to delay the resubmission.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Summary

    In this manuscript, Mitchell et al. study the function of CHD3 - a member of the NuRD chromatin remodeling complex - during human cranial neural crest cells (CNCC) differentiation in vitro. To this end they use iPSC CHD3-KO lines. They first observed this deletion has no impact on pluripotency levels of mutant iPSC neither on their capacity to differentiate into the three germ layers derivatives. Yet, they showed these mutant cells are unable to form CNCC as they fail to induce EMT genes and undergo CNCC differentiation. Using ATACseq, the authors then showed CHD3 KO present a reorganization of the chromatin accessible landscape, biasing these cells from a CNCC to a mesoderm fate. They further determine that upon differentiation of CHD3 KO cells, BMP responsive regulatory elements are aberrantly closed, making the cells insensitive to the signaling, explaining how they then fail to generate CNCC. Using ChIP-seq, they confirmed a direct action of CHD3 in making these elements accessible as it normally binds to these chromatin regions to allow proper differentiation. In addition, they demonstrate these BMP responsive genes are bound by DLX5, a transcription factor essential for neural crest development. Finally, the authors showed that during CNCC differentiation, CHD3 KO cells experience an imbalance between BMP and WNT signaling, leading to these cells adopting a mesoderm instead of a CNCC identity. They finally, showed this can be partially rescued by reducing the amount of Wnt signaling - that decreases the mesoderm gene expression - however, it not sufficient to induce a neural crest identity.

    Major comments

    1. What is the expression level of CHD3 in the heterozygote line? Does the remaining allele compensate for the loss which will explain the absence of phenotype?
    2. Since CHD3 shows a progressive upregulation in expression during CNCC differentiation (Fig. 2E), one hypothesis can be that it is not necessary involved in the activation of the CNCC programs but instead it is involved in maintaining these programs active - by keeping regulatory elements accessible. Thus, authors should check expression of CNCC markers, and EMT genes at the same time point than Fig. 2E in both WT and KO cells.
    3. The authors should use the term "regulatory elements" instead of "enhancers" as they can act either as activator or repressors.
    4. On the same line, while the authors indicate "Motif analysis of the enhancers aberrantly active in CHD3-KO cells ", they haven't shown these are active. They should say they perform the analysis on regulatory elements aberrantly accessible in CHD3 KO.
    5. The rationale that led the authors to focus on genes typically expressed in the primitive streak and in the early pre-migratory mesoderm, and BMP responsive transcription factors could be better explained. Are they part of the most deregulated genes in the RNAseq analysis?
    6. In the absence of CHD3, BMP response is not effective. While the authors nicely showed this is linked with changes in chromatin accessibility, it is necessary to check the expression levels of BMP receptors in CHD3 KO cells.
    7. Aberrant early mesoderm signature of the CHD3-KO cells needs to be better shown. It is not obvious from the GO analysis in Fig. 2 and the authors then showed expression of some markers but it is unclear how they picked them up.
    8. It has been shown that CNCC regulatory elements controlling differentiation genes are primed/accessible prior migration (PMID: 31792380; PMID: 33542111). Since the authors claim "CHD3 may have the role of priming the developing CNCCs to respond to BMP by opening the chromatin at the BMP responsive enhancers", it will be good to perform ATACseq are several time point during the differentiation process to assess the dynamic of chromatin reorganization to see when the switch to mesoderm fate occurs and how accessibility of BMP responsive element changes in WT and KO cells during CNCC differentiation to be able to demonstrate the KO fail to make BMP responsive element accessible or whether it is a defect in the maintenance of this accessibility.
    9. The authors claim CHD3 directly binds at BMP responsive enhancers, but in the figure, they show the data for all the region gaining or losing activity. It will be nice to add the information for the BMP responsive elements only.
    10. Motifs enrichment analysis of regions gaining accessibility in CHD3 KO do not seems to be labeled as Wnt responsive elements. The authors need to support better that CHD3 KO express more Wnt signaling/activity.
    11. OPTIONAL: Does increasing BMP concentration early during CHD3 KO differentiation has a better effect at rescuing CNCC differentiation?
    12. OPTIONAL: While this is not necessary for the current study, it is very intriguing that other CHD family member do not compensate. How this tissue or DNA sequence activity is achieved could be discussed. What are CHD4 or CHD5 expressed during CNCC differentiation? Could they be used to rescue the CHD3 KO phenotype? While this may be difficult to test, it could perhaps be discussed.

    Minor comments

    1. In the discussion, the authors could indicate whether CHD3 mutants somehow phenocopies some of the craniofacial defects observed in DLX5 mutant patients.
    2. It is not indicated were to find the data regarding expression epithelial and mesenchymal genes in the CHD3-KO cells.
    3. Authors could add in the discussion what is known about how CHD3 function changes from opening or closing chromatin is very intriguing a could be discussed.

    Significance

    General assessment:

    The link between chromatin remodelers and craniofacial defects has been shown in several studies in the past, but it still remains unclear how mutation of a given factor leads to such tissue specific defects. This manuscript represents an interesting and detailed mechanistic study on the role of chromatin remodeler in cell fate decision, showing that reorganization of chromatin accessibility is essential to proper response to signaling pathway and cell differentiation.

    Advance:

    The authors manage to link how mutant-induced changes in chromatin accessibility biased the cells towards a mesoderm fate as they directly impact the capacity of the cells to respond to BMP signaling - these regions closing upon CHD3 loss. However, the question remains to know whether CHD3 acts as an initiating factor or instead in involved in maintaining these programs active. As noted by the authors, a clinical link (with patient-derived iPCS) would be of great interest but as it stands the story already provide a good mechanistic understanding on how CHD3 control CNCC differentiation.

    Audience:

    This manuscript will be of great interest for specialized audience, yet a broader public may find it interesting too.

    Reviewer field of expertise:

    Neural crest and craniofacial development, epigenetics, transcriptomics

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this manuscript, Mitchell et al examine the impact of CHD3 KO (or het) on iPSC differentiation to CNCCs to model how pathogenic CHD3 mutations promote a neurodevelopmental disorder. The authors perform genomic characterization of the KO and het mutants during this differentiation model, and identify loss of CHD3 results in BMP target gene repression and increased mesoderm fate. Finally, the authors attempt to rescue the observed defects by titrating Wnt levels and observe partial rescue. I think the paper is easy to follow, contains interesting data, and establishes a novel role for CHD3 in CNCC differentiation, which may have implications in the disorder highlighted. I have the following suggestions:

    1. Figure 1 presents nice confirmation of the CHD3 KO cell lines being used. However, given that these cell lines were previously published, I suggest moving these data to the supplement.
    2. In the results section for Figure 1, the authors discuss the CHD3 heterozygotes, but I only see the KO cell line data presented. It would be especially nice to see the protein levels of Chd3 in the het.
    3. The authors discuss which genes are up and downregulated in the Chd3 KO D18 RNAseq, and show a clear heatmap in Figure 2A for WT cells. The same heatmap for candidate genes discussed in the results would be appreciated for Chd3 KO. Only a subset of markers are shown in Fig 2C.
    4. In general 2-3 replicates are presented. While the authors are showing heatmaps for selected locations for individual clones, which is appreciated (ex: Figure 4B and Fig 6), the QC for data quality is missing. For example, show spearmean correlation across the genome for datasets as a supplement.
    5. In the section discussing the results presented in Figure 4, the authors discuss the ATAC-seq peak number changes and overlap with gene expression changes. However, the overlap with gene expression changes is not shown. Making a simple venn diagram would help readers. a. In addition, showing a heatmap for unchanged ATACseq peaks can help to demonstrate the increase/decrease.
    6. In Figure 6, the authors present ChIPseq data for CHD3 in D14 and D18 samples, focusing on locations losing or gaining accessibility. What is enrichment at unchanged sites? Is CHD3 specifically enriched at changed locations? Then what about over genes with altered gene expression vs not changed? Is CHD3 only bound to distal elements? Performing an analysis of the peak distribution, perhaps with ChromHMM or other methods to look at promoter vs enhancer vs other locations. These types of analyses could really enrich the interpretation of direct CHD3 function.
    7. Given the changes in the CHD3 KO accessibility are mostly gene distal, are there existing Hi-C/microC/promoter CaptureC or other that can be used to ask if these are interacting with the predicted genes?
    8. Are the results observed NuRD-based or CHD3 NuRD independent functions? Looking at other NuRD subunit binding or effects in differentiation would help to dig into this a bit more. I realize this is a bit of a big ask, so I am not asking for everything. Are there existing binding data in CNCCs for a NuRD subunit that could be examined for overlap in where these changes occur, for example? I want to be clear I am not asking the authors to do all the experiments for an alternative NuRD subunit.
    9. The authors observe defects in CNCCs through genomic experiments. It would be really nice to perform simple wound healing/scratch assays and/or transwell assays to test if the CNCC migration phenotype is reduced in the CHD3 KO as well which would support the transcriptomic data.
    10. Related to the above, I am not sure if there is a phenotypic test for enhanced mesoderm. I suspect only IF/expression and morphology are possible, which the authors did. However, sorting the cells (with some defined markers) to ask how many are mesoderm-like vs CNCC in WT vs CHD3 KO would give some information outside of the bulk expression data.
    11. I did not see a reviewer token for the GEO data, so I could not check the deposited datasets.

    Minor points

    1. 1A seems to fit better with Figure 2.
    2. The authors say that the KO cell lines are not defective in pluripotency, but Figures 1G suggests a slight decrease in SSEA-1. Is this reproducibly observed?
    3. Would be nice to show number of up and downregulated genes in volcano plots for fast viewing of readers (ex: Fig 2B).
    4. Is it fair to use violin plots when data points are only 2-3 replicates (as in Figures 2C, 3D)
    5. The labels in Fig 4A and 5E are very hard to read.
    6. For browser tracks, the authors show very zoomed in examples (Fig 4C, and especially Fig 6C). showing a bit more of the area around these peaks would give readers a more clear appreciation of the data.
    7. Related to browser tracks, including more information just as including the gene expression changes (such as in Fig 6C) to enhance the interpretation of the impact of Chd3 binding, accessibility change and then, I presume, reduced Sox9 expression. Similar suggestion for Figure 4C, where I anticipate coordinate transcription changes of the associated genes.
    8. Do the authors observe any clone variability between the two CHD3 KO clones? There is variability I see in some of the heatmaps, but don't know if that it is because of clones or technical variation.

    Referees cross-commenting

    I think that the other reviewer and I are inline with each other in terms of our reviews and thoughts on the manuscript, so I do not have anything to add.

    Significance

    The paper presented by Mitchell et al represents a new role for CHD3 in regulating CNCC differentiation and perhaps explains why CHD3 mutations exist in neurodevelopmental disorders such as Snijders Blok-Campeau Syndrome. Limitations are the reliance on genomic datasets and modeled differentiation, although this permits for more mechanistic studies.

    I believe the fields of neural development, stem cell, chromatin biology, and others will be interested in this manuscript.

  4. Version published to 10.1101/2025.02.14.638260 on bioRxiv