The impact of CHD4 depletion on chromatin accessibility.

A. Western blots of nuclear soluble (Nucleoplasm) and chromatin fractions across CHD4 depletion probed with antibodies directed against indicated proteins. Times in hours of Auxin addition are indicated across the top. Lamin B1 and Histone H3 act as loading controls. B. Cell cycle analysis across CHD4 depletion time course. Hours post Auxin addition are indicated at the bottom, “CTRL” indicates DMSO control. Data represent an average of three replicates. Asterisks indicate significant differences from CTRL using a mixed-eKects model with Dunnett’s multiple comparisons correction. *p < 0.05; **p < 0.01. C. Volcano plots of differentially accessible ATAC-seq peaks between 0 and 1 hour of Auxin addition (left) or 0 and 4 hours of Auxin addition (right). Magenta spots indicate statistically significant differences (FDR > 0.05). Numbers of significant peaks decreased or increased are indicated on the plots. D. Heatmaps of ATAC-seq signal for all regions displaying increased accessibility (N=52041) or decreased accessibility (N=8384) at any time across the CHD4 depletion time course are displayed for each time point. E. Heatmaps of Cut&Run data for indicated histone modifications at sites increasing or decreasing in accessibility (as in panel D) (H3K27Ac and H3K4Me1 from this study; H3K27Me3 and H3K4Me3 taken from (Lando et al., 2024)) at indicated times after CHD4 depletion. F. Percentages of sites increasing in accessibility (top, blue) or decreasing in accessibility upon CHD4 depletion (bottom, red) which localise to indicated genomic features. Active enhancers are defined as having H3K4Me1 and K3K27Ac but not K3K4Me3, and inactive enhancers as having H3K4Me1 but not H3K4Me3 or H3K27Ac. G. Heatmaps of CHD4 and MBD3 Cut&Run data at upDARs and downDARs in 2iL conditions. H. Heatmaps of ATAC-seq signal at active enhancers (N=4707) across the CHD4 depletion time course.

CHD4 acutely regulates gene expression.

A and B. Heatmaps from nascent RNA-seq (A) or bulk RNA-seq (B) of genes showing significant (Padj <0.05) changes in expression at any point during the CHD4 depletion time course. C. Volcano plots showing significant gene expression changes at indicated time points in nascent RNAseq (top) and bulk RNAseq (bottom). Genes increasing upon CHD4 depletion are shown in red, and those decreasing are shown in blue. The number of significantly misexpressed genes at each time point is indicated in the figure. D. Violin plots showing the average Log2FoldChange of significant upregulated (red) and downregulated (blue) genes during the CHD4 depletion timecourse. E. Gene Ontology (GO) enrichment analysis of genes increased or decreased after four hours or 24 hours of CHD4 depletion. The top 10 biological processes are shown for each category, based on smallest adjusted p-value.

Chromatin opening upon CHD4 depletion.

A. Frequency density distribution of the distance of increasing Differentially Accessible Regions (“upDARs;” red lines) and decreasing regions (“downDARs;” blue lines) to the TSS of genes showing increased (solid lines) or decreased (dotted lines) expression within four hours of CHD4 depletion. B. For genes either increasing or decreasing at each time point, we have plotted the average number of DARs for which genes of that category were found to be the closest gene. C. Heatmaps of Cut&Tag signal for H3K27Ac, and H3K4Me1 at sites increasing in accessibility at indicated times of CHD4 depletion. D. Heatmaps of NANOG and SOX2 Cut&Run signal at increasing accessibility sites at indicated times of CHD4 depletion. E. Pairwise comparisons of called peaks of binding for NANOG (top) and SOX2 (bottom) between undepleted cells (0h) and 30 minutes, 1 hour or 4 hours of CHD4 depletion. Significantly changed (FDR>0.05) binding sites are shown in blue when log2FC>0 and red when log2FC<0. F. IGV screenshot of the upstream region of the mouse Eomes locus displaying ATAC-seq, Cut&Run and Cut&Tag data as indicated at left. Boxed regions labelled 1-5 are upDARs, while the box labelled P corresponds to the Eomes promoter.

CHD4 and SALL4 both restrict chromatin accessibility.

A. Volcano plots of differentially accessible ATAC-seq peaks when comparing 1 hour, 4 hours or 24 hours of SALL4 depletion with those seen in undepleted cells (0). Magenta spots indicate statistically significant differences (FDR > 0.05). Numbers of significant peaks decreased or increased are indicated on the plots. B. Heatmaps of ATAC-seq signal for all regions displaying increased accessibility (N=26510) across the SALL4 depletion time course are displayed for each time point. C. Heatmap of SALL4 Cut&Run signal in undepleted ES cells (taken from (Ru et al., 2022)) at all regions displaying increased accessibility across the SALL4 depletion time course (left, N=26510) or at active enhancers (right, N=4339). D. Overlap of sites showing increased accessibility upon SALL4 depletion with those increasing upon CHD4 depletion (upDARs). The % A/T base composition of the different categories of sites is indicated. E, F. Heatmaps of ATAC-seq signal at sites increasing upon either SALL4 or CHD4 depletion (SALL4+CHD4), sites increasing upon CHD4 depletion but not upon SALL4 depletion (CHD4 Only), or sites increasing upon SALL4 depletion but not upon CHD4 depletion (SALL4 Only) plotted at indicated time points of CHD4 depletion (E) and SALL4 depletion (F). G, H. Cut&Run signal for SALL4 (G) or for CHD4 and MBD3 (H) in undepleted ES cells at the three different classes of sites.

NuRD regulates NANOG and SOX2 binding to active sites.

A. Heatmaps of NANOG and SOX2 Cut&Run signal at indicated times of CHD4 depletion across sites decreasing in accessibility. B. Heatmaps of H3K27Ac and K3H4Me1 Cut&Tag signal at indicated times of CHD4 depletion across decreasing accessibility sites. C. Heatmaps of Cut&Run signal for NANOG and SOX2 across active enhancers at indicated times of CHD4 depletion. D. IGV screenshot of the enhancer cluster downstream of the Klf4 gene displaying ATAC-seq, Cut&Run and Cut&Tag data as indicated at the left. Boxed regions are labelled with the distance in Kb from the annotated Klf4 transcription start site. E. Fluorescence survival curves of chromatin-bound NANOG-HALO molecules in 2iL (blue line) or after 1 hour of CHD4 depletion (purple line). The grey dotted line represents the fluorescence survival curve for molecules in a fixed-cell control imaged under identical conditions. F. Apparent dissociation rates (ko+) of chromatin-bound NANOG molecules calculated through fitting a single exponential decay model to the survival curves in panel E. Error bars represent 95% confidence intervals for each fit applied to data taken from three independent experiments. The horizontal dashed line represents the upper 95% confidence limit for a fixed-cell control. ** indicates that 99% confidence intervals do not overlap, i.e. p < 0.01. G.H. As in F but for SOX2 (G) and KLF4 (H). The fixed cell control was not imaged in the SOX2 experiments in panel F. I. Western blots of nuclear soluble (Nucleoplasm) and chromatin fractions across MBD3 depletion probed with antibodies directed against the indicated proteins. Times in hours of Auxin addition are indicated across the top. Lamin B1 and Histone H3 act as loading controls. J, K. Apparent dissociation rates (Ko+) for NANOG-HALO (I) and KLF4-HALO (J) before and after 60 minutes of MBD3 depletion. For calculation of Ko+ the trajectories were pooled from 4 replicates of each time point obtained over 2 days.

CHD4 controls accessibility differently at different classes of sites.

A,C, E, G Tn5 integration frequency was determined from ATAC-seq data and plotted across indicated sites across the CHD4 depletion time course. The number of sites is shown in parentheses. B, D, F, H Vplots across sites indicated above (corresponding to the Tn5 integration plots) at indicated times of CHD4 depletion. See Supplementary Figure 2 for a Vplot schematic. I. MNase-seq data collected 0 (red), 30 minutes (dark red) or 24 hours (black) after of NuRD reformation in Mbd3-/- ES cells are plotted across indicated sites. Top graphs show results from “Hard” MNase treatment while bottom graphs show “Soft” MNase treatment (see text). The y-axis shows normalised read coverages, while the X-axis shows distance in base pairs from the centre of the feature. Curves show mean and standard error from 3 biological replicates.

Model of CHD4 function.

A. Highly accessible sites. In undepleted cells these sites are extensively bound by CHD4/NuRD, where it acts to promote the oK rate of transcription factors to promote accessibility. Tn5 is able to access the central NFR but also can integrate into the flanking nucleosomal DNA. After CHD4 depletion, the on rate for TFs does not change, but the oK rate is now much reduced, resulting in increased TF binding. The sites become less accessible to Tn5, such that although it can still access the hypersensitive site within the NFR, there are fewer integrations extending outwards. These regulatory regions cannot quickly respond to receipt of external signals. B. Model of CHD4 function at inaccessible, silent enhancers. In undepleted conditions, there is low CHD4 enrichment at these sites. Here, CHD4 acts to prevent binding of transcription factors and maintain low accessibility, such that Tn5 cannot frequently access the DNA. After CHD4 depletion, the locus becomes more accessible, and transcription factors can stably bind. This leads to spurious activation of distal promoters and an increase in transcriptional noise.

gRNA sequences used for gene targeting

Antibodies used in this study with dilutions used for western blot.

Cell cycle arrest and cell death following CHD4 depletion.

Phase contrast images and example flow cytometry plots of CHD4-mAID ES cells in 2iL conditions after indicated times of Auxin addition. Cell cycle stage and relative percentage of cells are indicated on the flow cytometry plots.

Schematic diagram of Vplots.

Adapted from (Henikoff et al., 2011). The Vplot is derived by plotting the midpoint of all recovered fragments (horizontal lines) onto the graph, with midpoint position (white circle) on the x-axis and fragment length on the y-axis. Red lines correspond to reads obtained from fragments with both ends (i.e. Tn5 integration sites) in the NFR, while blue lines represent reads spanning a nucleosome. The vertical dotted line indicates the position of the central hyperaccessible site. The inferred chromatin structure of the locus is shown below.