PMC

(A) Profiles of sequence motifs across aligned LAD borders. X-axis depicts the position relative to the nearest LAD border; positive coordinates inside LADs and negative coordinates outside LADs. Colored lines show the median frequency within a running window of 10 kb for sequence motifs, grey lines for DamID identified peaks. (B) Profiles of CP190 peaks across aligned LAD borders; all Cp190 peaks (1^st panel), CP190 peaks without SU(HW) binding, defined as peaks with an average log₂(SU(HW)binding ratio) <1.0 (2^nd panel), CP190 peaks with SU(HW) binding, defined as CP190 peaks with an average log₂(SU(HW)binding ratio) >1.5 (3^rd panel).

Genome - NL interaction maps after knockdown and overexpression of SU(HW). (A) Western blot analysis of SU(HW) expression levels after knockdown (ctrl: control RNAi; kd1 and kd2: SU(HW) RNAi with two independent dsRNA fragments) and after overexpression (ctrl: control vector oe: overexression by transfection of SU(HW) under an Act5C promoter). 1^st lane in each panel: transfected with Dam-LAM, 2^nd lane with Dam-only. (B–C) Median NL interaction (log₂ Dam-LAM/Dam ratio) across all aligned LAD borders (824 borders, 1^st panel); border regions with SU(HW) present (220 borders, 2^nd panel, red triangle represents SU(HW) at the borders), borders without SUH(HW) present (604 borders, 3^rd panel). (B) Knockdown of SU(HW) (blue line) and control knockdown (grey line). (C) Overexpression of SU(HW) (orange line), and corresponding control (grey line).

(A–B) Median NL interaction (log₂ Dam-LAM/Dam ratio) across LADs without SU(HW) peaks (190 borders, 1^st panel), LADs with at least one SU(HW) peak (634 borders, 2^nd panel, red triangle represents the presence of one or more SU(HW) peaks at any position inside LADs), the 25% of LADs with the highest SU(HW) peak density (206 borders, 3^rd panel, red triangles represent high density of SU(HW) peaks). (A) After knockdown of SU(HW) (blue line), after control knockdown (grey line). (B) After overexpression of SU(HW) (orange line), after control overexpression (grey line). (C) Ave changes in NL interaction levels per LAD, for LADs without SU(HW) (grey), LADs with at least one SU(HW) peak (light blue or orange), the 25% of LADs with the highest SU(HW) peak density (dark blue or orange) after knockdown (blue, 1^st panel) and overexpression of SU(HW) (orange, 2^nd panel). * Wilcoxon test; p<10⁻³ (D) Median changes in NL interaction across aligned SU(HW) peaks (red triangle) inside LADs (bright lines), in border regions (pale lines) and outside LADs (grey lines) after knockdown of SU(HW) (blue, 1^st panel) and after overexpression of SU(HW) (orange, 2^nd panel). (E) Cis-spreading of SU(HW) DamID signals from aligned SU(HW) binding peaks (red triangle), inside LADs (red line) and outside LADs (grey line).

(A) Binding maps of insulator proteins along a 1 Mb region on chromosome 2L. Y-axes depict the Dam-insulator over Dam-only methylation ratio (high values are truncated at 10). Grey rectangles represent LADs. (B) Theoretical profiles of features that are respectively enriched at LAD borders, enriched inside LADs or depleted from LADs (upper panels). Profiles of insulator and EcR binding peaks across aligned LAD borders. 824 borders, left and mirrored right borders combined (lower panels). The region around each border from which data was taken ranges from the center of the inter-LAD region to the center of the LAD; this ensures that each data point is used only once. X-axis depicts the position relative to the nearest LAD border; positive coordinates inside LADs and negative coordinates outside LADs. Y-axes depict the median number of binding peaks within a running window of 10 kb. Y-axes are scaled to frequencies within the plotted window, depending on the genome-wide frequency of the feature. (C) Percentage of insulator binding peaks within LADs (top) and percentage of inter-LAD peaks within a 10 kb region just outside LADs (bottom). Black horizontal lines represent the percentage expected by chance. **significantly enriched or *depleted compared to random permutation simulations; p<10⁻³. (D) Model of the chromatin organization in LADs, with SU(HW) binding mainly at LAD borders and inside LADs; and DWG, BEAF-32, GAF and CTCF preferentially located in inter-LAD regions.

(A) Genome – NL interaction maps in Drosophila Kc cells and human Lung Fibroblasts along a 4 Mb region at respectively chromosome 2L and chromosome 18. Human data are from . Y-axes depict the log₂ transformed Dam-LAM over Dam-only methylation ratio, smoothed by a running median of respectively 15 and 5 probes. Rectangles below each map represent calculated LAD positions for Drosophila (red) and human (blue). Grey rectangles at the bottom represent genes at the + and - strand. (B) Distribution of LAD sizes in Drosophila (red) and human cells (blue). Dashed lines mark the median LAD sizes. (C) Histogram of the number of genes per LAD. (D–F) Profiles across aligned LAD borders (824 borders, left and mirrored right borders combined). Running window median (red line) and a random subset of 2001 single genes (black dots in E and F). The region around each border from which data was taken ranges from the center of the inter-LAD region to the center of the LAD; this ensures that each data point is used only once. X-axis depicts the position relative to the nearest LAD border; positive coordinates inside, negative coordinates outside LADs. (D) Median gene coverage. (E) mRNA levels in A-values, (log₂(Cy5)+log₂(Cy3))/2 (F) Median RpII18 occupancy on entire genes as determined by DamID [data from30].