Fig. 4

Chromatin reorganization between cell types is modulated by stochastic clustering between epigenetic domains. a Two-color dSTORM image of active (H3K4me3, blue) and repressive (H3K27me3, red) chromatin marks in a representative S2 cell. Images of early and late embryos are displayed in Supplementary Fig. 4a and panel c. Scale bar = 1 µm. b Quantification of co-occurrence (CA > 0.5) between active and repressive chromatin using aCBC³². Violin plots of CAs for H3K4me3 and H3K27me3 are shown in the upper panel and lower panels, respectively. The black line represents the median of the distribution. c Representative zoomed images of two-color dSTORM for the three cell types investigated. Black arrows indicate the localization of small active chromatin domains in the periphery of large repressive domains. The lower panel displays active and repressive marks of Chip-Seq enrichment profiles for late embryo. Scale bar = 200 nm. d,e dSTORM-rendered images of Alexa-647-labeled d H3K27me3 and e H3K4me3. Images show density maps computed from the area of the polygons obtained from the Voronoï diagram with scale defined on top. Scale bar = 1 µm. Zoomed regions display detected compartments (highlighted with different colors). Scale bar = 200 nm. Additional images for all cell types and chromatin marks are displayed in Supplementary Fig. 5a, b. f,g Population-based distribution of epigenetic domain sizes as obtained from dSTORM and predicted from ChiP-seq data for H3K27me3 f and H3K4me3 g. PDF is probability density function. Single-cell distributions of physical sizes and Chip-Seq data are shown in Supplementary Figs. 5c, d and 6b, respectively. N = 60, from two to three biological replicates in microscopy imaging. h Percentage of clustering for active and inactive chromatin marks for each cell type. Error bars = SD. One-sample t test p-values: *p < 0.01; **p < 0.001. i Box plots of the distributions of normalized Hi-C counts between chromatin domains of H3K27me3 or H3K4me3 in embryos and S2 cells. The results were independent of matrix resolution (10, 20, and 50 kb). Boxes contain 50% of the data (0.67σ), and red lines mark the median values. Outliers (>3.3σ away from the mean values) are shown as black dots. p-values were calculated using the Welch t test