Fig 2: Chromatin organization and consequences upon H1 depletion on genome structure. Chromatin organization and H1 variants distribution (upper panel): Hi-C data allows determination of B (inactive) and A (active) compartments. B compartment is characterized by closed chromatin, long TADs with a high H1.2/H1X ratio and a great overlap with low GC Giemsa bands, while the opposite occurs for A compartment. H1 variants were differentially distributed along the genome and two profiles could be distinguished in T47D breast cancer cells: H1.2, H1.5 and H1.0 co-localized at low GC regions whereas H1.4 and H1X occupied high GC regions. Upon multiple H1 depletion (H1 KD), H1.2 and H1.4 were strongly depleted while H1.0 became up-regulated but without changing its distribution. Remaining H1.4 redistributed to low GC regions, whereas H1.2, H1.5 and especially H1X were redistributed to higher GC regions. Consequences of H1 KD in chromatin structure (middle panel): Upon H1 KD, chromatin accessibility increased and H3K9me3 signal decreased, especially at A compartment. Intra-TAD interactions increased both at B and A compartments whereas inter-TAD interactions were reduced at A compartment. TAD-border strength increased, together with some TAD borders being lost or shifted. Upon H1 KD, shifts between and within A/B compartments occurred, being more frequent compaction shifts at A compartment (including A-to-B shifts) and decompaction at B compartment (incl. B-to-A). Consequences of H1 KD in genome structure are related to gene expression deregulation (bottom panel): TADs presenting a coordinated response to H1 KD were enriched compared to the expected frequency. Up-regulated genes accumulated within TADs with poor basal expression and low gene density. Gene-dense TADs contained both up- and down-regulated genes simultaneously. TADs with only down-regulated genes showed intermediate features.

Fig 3: Genomic distribution of histone H1 variants upon H1 knock-down in breast cancer cells. (A) Immunoblot analysis of H1 depletion in H1 KD cells. Chromatin extracts (1 or 5µg of protein) from T47D multiH1 KD cells cultured in the presence or not of Doxycycline for 6 days were run in SDS/PAGE and immunoblotted with the indicated antibodies against H1 variants or histone H3 as loading control. ImageJ Immunoblot quantification of multiple experiments is indicated as mean (ratio + Dox/untreated) and SD. Number of biological replicates used for quantification were: n = 6 (H1.2, H1.4 and H1.0), n = 4 (H1X), n = 2 (H1.3, H1.5). (B) ChIP-qPCR of H1 variants in multiH1 KD cells. Chromatin from untreated or Dox-treated H1 KD cells was used for ChIP with antibodies against H1 variants and unrelated IgG as a control. Resulting DNA was amplified by qPCR with oligos for distal promoter (3kb upstream TSS) and TSS regions of genes CDK2 and NANOG. ChIP amplification is shown relative to input DNA amplification. A representative experiment quantified in triplicate is shown. Statistical differences between Untreated (–Dox) and + Dox immunoprecipitated DNA for each H1 variant are supported by paired-t-test. (***) P < 0.001; (ns/non-significant) P > 0.05. (C) Heat map and cluster analysis of the input-subtracted ChIP-seq abundance of H1 variants within Gpos and Gneg bands from untreated T47D cells. The color of heatmap grids represents the relative input-corrected coverage of the H1 variant indicated at the X-axis within each G band, while the Y-axis shows to which group the band belongs. Two main clusters of H1 variant distribution are formed, one with abundant H1.X and H1.4 at high GC bands, the other with H1.2, H1.5 and H1.0 enriched at low GC bands. Two replicates are shown (r1, r2). (D) Scatter plots of the indicated H1 variant pairs input-subtracted ChIP-seq abundance within 100-kb bins of the human genome. The GC content at each bin is color-coded. Pearson's correlation coefficient is shown (P-value < 0.001). (E) Heat map and cluster analysis of the input-subtracted ChIP-seq abundance of H1 variants from WT or H1 KD T47D cells (-/+Dox) within 10 chromatin states (ChromHMM segmentation). The profile of H3K9me3 is included. For each heatmap grid, the color represents the input-corrected coverage of the H1 variant identified by the X-axis within each region, while the Y-axis shows the ChromHMM group the region belongs to. (F) Scatter plots of H1 variants input-subtracted ChIP-seq abundance within 100-kb bins of the human genome in multiH1 KD cells treated or not with Doxycycline. The GC content at each bin is color-coded. (C, E) Heatmaps were performed by using the R package ‘pheatmap’. The ‘euclidean’ distance measure and the ‘complete’ cluster method were used in clustering rows and columns.

Fig 4: TAD boundaries changes upon H1 KD. (A) Hi-C interaction maps of 6.25 Mb region in chromosome 3 at 50-kb resolution. Left panel is a heat map of Hi-C maps normalized by reads coverage in Log2 scale with TADs overlayed by black lines. Top triangle of the map corresponds to Hi-C in WT and lower triangle to H1 KD. Green arrow points to the de-novo detected TAD border in H1 KD. Right panel, differential Hi-C map showing the enrichment of internal interaction in the two separated TADs around the new detected border. (B) Box plot showing the H1.2 and H1.4 input-subtracted ChIP-seq signal in WT cells within TADs containing the TAD borders divided in conserved, shifted <100 kb and non-conserved according to their behavior upon H1 KD. (C) TAD border dynamics. Box plot of normalized border strength distribution for TAD borders in WT and H1 KD cells, divided in conserved, shifted <100 kb and non-conserved borders. (***) P < 0.001; (**) P < 0.01; (*) P < 0.05 (Mann–Whitney test).