Fig. 2

Micro-H3K27me3 domains are altered upon insulator protein depletion. a Distribution plot showing the density of micro-H3K27me3 domains (y-axis) depending on the sizes of the domains (x-axis), as detected by scoring H3K27me3 levels by scanning the genome with a resolution of 40 bp bin size using NormR (see “Methods” [61];). Micro-domains’ sizes distribute into mers of nucleosome arrays (see dotted lines every 200 bp) thus reinforcing the relevance of the detected signal unlike for swap controls (see “Methods”). The red dotted lines highlight the interval of micro-domain sizes. b Plot showing the averaged log ratio of H3K27me3 levels of micro-domains in Beaf-KD as compared to WT control cells, depending on domain sizes (x-axis). c Average H3K27me3 levels in micro-domains in Beaf-KD as compared to WT control cells. Note that Beaf-KD leads to a significant reduction in H3K27me3 levels as compared to WT cells (***; Wilcoxon pairwise test, p value <1e−4; see Additional file 2: Fig. S2D). d Distribution plot quantifying the normalized densities of H3K27me3 levels from normalized ChIP-seq reads in bins corresponding to micro-domains compared to randomized control bins of the same distribution sizes selected out of euchromatin or heterochromatin domains (see “Methods”). The dashed lines highlight the mean values of H3K27me3 densities for micro-domains controls as indicated. e Distribution of insulator sites (bins) bound by GAF/dCTCF that harbor decreasing H3K27me3 levels. Sites were plotted relative to heterochromatin borders (x-axis) with a Beaf32 site (blue bars) or not (brown bars). Note that the influence of Beaf32 borders is detected even for GAF/dCTCF sites localizing over long distances (> 10 kb) from such borders. The error bars represent the variability of signal from independent replicates with all bins in the indicated intervals (see “Methods” for details). f Bar plot showing the relative distribution of Beaf32 sites (black), H3K27me3 micro-domains (blue) and sites with increasing H3K27me3 spreading (yellow) within euchromatin domains. All sites were distributed as a function of the positioning with respect to borders (0 and 100 are the two sides of euchromatin domains, and 50% the middle of such domains)