Fig. 4

TAD insulation does not account for micro-domains. a Genomic view of Mio locus aligning the ChIP-seq data of H3K27me3 upon Beaf-KD and control cells identifying large heterochromatin domains by hidden Markov model (HMM) along with micro-domains (red rectangle), the ChIP-seq of Beaf32, and GAF insulator proteins and the 2D map of long-range interactions as obtained from sub-kb-resolution Hi-C data from S2 cells [21]. Beaf32 and GAF sites are represented by orange and blue triangles, respectively. Note that the Mio locus is representative of the enrichment of Beaf32 at TADs/compartments borders [26, 62]. b Scheme representing the genomic context of euchromatic micro-domains with respect to nearby Beaf32 sites near repressive heterochromatin TADs (see panel c), as detected by high-resolution Hi-C. c Enrichment test of micro-domains in euchromatin as a function of the presence or not of a particular arrangement of Beaf32 sites at borders of the heterochromatin domain. Colors represent odds ratio and asterisks the corresponding p values (by Fisher’s exact test), as calculated relatively to domains without any Beaf32 site (see accolades). Note that presence of Beaf32 sites on the left of the heterochromatin domain belong to loci enriched in micro-domains found on the opposite side of such domains (as illustrated by the Mio locus). d Scattered plot representing the scores obtained from genome-wide aggregated Hi-C assessing long-range interactions [9] with data from S2 cells [20], between the indicated binding sites of Beaf32, GAF, dCTCF, and their co-factors CP190 or cohesin. X-axis: aggregation of global long-range interactions estimating the strength of TADs (as normalized Hi-C reads in LRIs-2; see Fig. 5a) depending on the binding at their borders of the indicated insulator proteins versus control sites. Y-axis: estimate of strength of looping (as normalized Hi-C reads in LRIs-3; see Fig. 5a) between the indicated binding sites or control sites (gray dots). The vertical and horizontal lines represent the threshold of the top-5% of Hi-C interactions (of the total Hi-C bins). Note that similar results were obtained using various sources of Hi-C data (see Additional file 4: Fig. S4; see “Methods”). e Gene set enrichment analysis (GSEA) testing the influence of TAD strength on formation of H3K27me3 micro-domains. TAD strength was estimated by computing Hi-C data in Beaf32-depleted and control cells (see “Methods”). Differential TAD strength was measured as net variations between Hi-C data in Beaf32-depleted cells compared to control cells. GSEA was performed for all TADs (left) or among the TADs bordered by a Beaf32 site. Of note, this requires using the high-resolution 2000 TADs to reflect the genomic context of the test with repressive H3K27me3 TADs, as shown (Fig. 4b) (see “Methods”) [15, 20]. The normalized differential LRI scores were estimated for all loci defined by a couple of bins corresponding to one Beaf32 site interacting with any distant gene (> 5 kb) harboring a H3K27me3 micro-domain or not. Genes were also classified depending on additional parameters of aggregated Hi-C (see Fig. 5)