Essential role of Cp190 in physical and regulatory boundary formation

Abstract

Boundaries in animal genomes delimit contact domains with enhanced internal contact frequencies and have debated functions in limiting regulatory cross-talk between domains and guiding enhancers to target promoters. Most mammalian boundaries form by stalling of chromosomal loop-extruding cohesin by CTCF, but most Drosophila boundaries form CTCF independently. However, how CTCF-independent boundaries form and function remains largely unexplored. Here, we assess genome folding and developmental gene expression in fly embryos lacking the ubiquitous boundary-associated factor Cp190. We find that sequence-specific DNA binding proteins such as CTCF and Su(Hw) directly interact with and recruit Cp190 to form most promoter-distal boundaries. Cp190 is essential for early development and prevents regulatory cross-talk between specific gene loci that pattern the embryo. Cp190 was, in contrast, dispensable for long-range enhancer-promoter communication at tested loci. Cp190 is thus currently the major player in fly boundary formation and function, revealing that diverse mechanisms evolved to partition genomes into independent regulatory domains.

Fig. 1.. Cp190 is required to form nonpromoter boundaries in fly embryos.

(A) Percentages of indicated genotypes (with/without maternal/zygotic protein) that completed indicated developmental transitions in three biological replicates. Horizontal lines show means. (B) Top: Distribution of indicated datasets in 2-kb bins ±25 kb around all WT contact domain boundaries ranked by physical insulation defects in Cp190⁰. Lane 8 shows the presence of indicated DNA motifs in the central 2-kb bin. Summarized values (average physical insulation score or percentage of WT boundaries with boundary/ChIP peak/transcribed TSS present) across 2-kb bins are shown. Enrichments ±2 kb around the central boundary are indicated. Bottom: Same but for all WT Cp190 ChIP peaks ranked by ChIP occupancy. (C) Physical insulation scores or differences measured at all 1140 Cp190-occupied boundaries whose most boundary-proximal Cp190 peak is distal or proximal to a transcribed TSS in WT embryos (75). P values and W-statistics from two-sided Wilcoxon rank sum test with continuity correction are indicated. (D) Similar to (C) but at all 1140 Cp190-occupied boundaries whose most boundary-proximal Cp190 peak overlaps or does not overlap the indicated DNA motif. (E) Example locus (dm6 coordinates) Hi-C maps (2-kb resolution), eigenvector values (2-kb resolution, positive/negative for A/B compartments), physical insulation score (calculated with different window sizes in gray, average in black), and contact domain boundaries (red lines) from this (above) and published (below) Hi-C studies (8, 67) and Cp190 ChIP-seq (reads per million). Homeobox genes are blue. Differential (Cp190⁰ minus WT) values are shown below.

Fig. 2.. Cp190 is required for boundary formation at CTCF peaks.

(A) Distribution of indicated datasets in ±25-kb windows centered around all 1477 CTCF peaks identified in WT ranked by ChIP occupancy. (Lanes 1 to 4) Presence of contact domain boundaries called in each genotype by TopDom in 2-kb bins around the peak center. (Lanes 5 to 10) Physical insulation score differences measured in genotype X (top) minus genotype Y (bottom) by Hi-C. (Lanes 11 and 12) CTCF or (Lanes 13 and 14) Cp190 ChIP occupancy in indicated genotypes. (Lanes 15 and 16) Differential CTCF and Cp190 ChIP occupancy in genotype X (top) minus genotype Y (bottom). (Lane 17) CTCF motif presence in DNA. (Lane 18) Expressed TSSs in WT 4- to 6-hour-old embryos (75). Summarized values (average physical insulation score or percentage of WT CTCF peaks with boundary/ChIP peak/differentially bound region/CTCF motif/transcribed TSS present) across 2-kb bins are shown above with indicated enrichments ±2 kb around the central CTCF peak (highlighted in red on the x axis). (B) Example locus (dm6 coordinates) Hi-C maps presented as in Fig. 1E. Arrowheads point to a CTCF and Cp190 cobound peak located at a domain boundary in WT; empty arrowheads indicate their absence in the mutants. ChIP-seq scale is reads per million. Cartoons summarize CTCF and Cp190 binding status and boundary presence/absence. RPKM, reads per kilobase per million reads.

Fig. 3.. CTCF-occupied boundaries are differentially sensitive to CTCF or Cp190 loss.

(A) Same as Fig. 2A but for all 312 CTCF-occupied boundaries ranked by physical insulation defects in CTCF⁰ relative to WT. (B) Alluvial plot of the same data as in (A). Lost boundaries are absent in the mutant but present in WT. Weaker and intact boundaries are retained in the mutant but have a delta physical insulation score in the mutant minus WT ≥0.01 or <0.01, respectively. Cartoons show typical CTCF and Cp190 ChIP occupancy in each scenario. (C) Box plots of data shown in (A). Physical insulation score differences in indicated mutants minus WT at all 312 CTCF-occupied boundaries at which Cp190 recruitment to the nearest CTCF peak to the boundary is strictly CTCF dependent (lost Cp190 peak in CTCF⁰) or at least partially CTCF independent (residual Cp190 peak at the former CTCF peak in CTCF⁰). P values and W-statistics from two-sided Wilcoxon rank sum test with continuity correction are indicated. Box plot: Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile ranges; points, outliers; n = CTCF-occupied boundaries of indicated categories. (D) Same as (C) but for CTCF-occupied boundaries at which the nearest CTCF peak to the boundary is distal (further than ±200 bp) or proximal (within 200 bp) to an actively transcribed TSS (RPKM > 0) in WT embryos (75). (C and D) *** indicate high statistical significance (P ≤ 0.0001). The precise P values are shown in the figures.

Fig. 4.. Cp190 complexes exert similar enhancer-blocking activity in a quantitative reporter assay.

(A) Enrichments of indicated proteins (rows) in pull-downs with indicated GFP-tagged baits (columns, Su(Hw)[1-219], CTCF[1-293], Chro[613-926], and full-length Cp190) from the same batches of embryo nuclear extracts, analyzed by mass spectrometry (MS) and ranked by their specific enrichment in the Cp190 pull-down. Scale bar indicates log₂ fold change (FC) of average intensity based absolute quantification (iBAQ) values of biological duplicate pull-downs over negative control pull-downs with unrelated GFP-tagged bait. CTCF pull-downs were previously described (7). (B) Published ChIP-seq profiles in S2 or Kc cell lines of indicated insulator proteins ±1 kb around the cloned genomic fragments (345 to 888 bp long indicated by red boxes, loci separated by vertical lines). Scales show total counts. (C) Insulator strengths of cloned genomic fragments measured in S2 cells transiently transfected with reporters with indicated I (insulator) test fragments, expressed as percentage of gypsy insulator strength (set to 100%). Insulators block EGFP activation by the enhancer (E). A gypsy insulator (“G”) blocks EGFP activation by the enhancer from the left. Fragments were tested in biological duplicates (dots). Horizontal lines show average values obtained with gypsy or a neutral spacer (n = 8 biological replicates); dotted lines show SDs. Tested fragments were bound by CTCF (A and B), Su(Hw) (A′ to D′), or Chro+Pzg+BEAF-32 (A″ to M″). Single (pink dots) or two (red dots) pairs of non-overlapping BEAF motifs in B″ and F″ were mutated.

Fig. 5.. Cp190 prevents regulatory cross-talk between early patterning gene loci.

(A) Extended ftz locus (dm6 coordinates) Hi-C maps presented as in Fig. 1E. (B) NG Capture-C profiles (1-kb resolution) around Scr or ftz TSS viewpoints showing average normalized reads (in reads per million) of biological triplicates excluding bins ±2 kb around the viewpoint (gray). Differential Capture-C profiles in mutant versus WT are shown as log₂ fold change profiles obtained from diffHic. Yellow brackets mark the deleted boundary in SF1^KO. WT Cp190 ChIP peaks are highlighted in blue (two prominent Cp190 ChIP signals in WT overlap a blacklisted region and were not called peaks). Early stripe enhancers with schematized expression patterns are numbered 1 to 5 (references in Materials and Methods). Dotted arrow shows Scr regulation by a hypothetical distal enhancer (question mark). SF1^KO and SF2B^KO deletions are yellow. (C) RNA–fluorescence in situ hybridization (FISH) with cohybridized antisense probes against Scr (red) and ftz (green) mRNAs in 4′,6-diamidino-2-phenylindole (DAPI)–stained early gastrula embryos (anterior left; posterior right; scale bars, 100 μm; merged images on the right). In Cp190⁰, Scr is expressed in its WT stripe and in ectopic ftz stripes (filled arrowheads). In SF1^KO embryos, Scr is lost in its WT stripe (empty arrowhead) and only expressed in ftz stripes (filled arrowheads). In SF2B^KO embryos, Scr and ftz expression seem normal (embryo rotation reveals a normal ventral gap in Scr expression).

Fig. 6.. Cp190 is dispensable for ectopic Scr activation by a long-range enhancer.

(A) Similar to Fig. 5A but showing contact domains downstream of Scr. (B) NG Capture-C profiles presented as in Fig. 5B but around Dfd or Scr TSS viewpoints in indicated genotypes. Enhancer 7 drives schematized reporter gene expression in the hindgut and anal plate of older embryos in transgene assays (52) and is separated from the Scr promoter (30 kb away) by Cp190 ChIP peaks. (C) RNA-FISH with antisense probes (red) against Scr mRNA in late-stage (stage 16) DAPI-stained embryos (anterior left; posterior right; scale bars, 100 μm). Scr is normally expressed in labial and prothoracic segments and the anterior midgut and is additionally expressed in the hindgut and anal plate (left and right arrowheads) of Cp190⁰ mutants. (D) Summarized Scr misexpression phenotypes in Cp190⁰ early and late embryos. Effective (solid arrows) or blocked (dotted arrows) transcriptional activation of promoters by indicated enhancers is shown (hindgut and anal plate enhancer in blue; Scr enhancers in orange including a putative distal enhancer marked by a question mark; ftz enhancers in green). In Cp190⁰ embryos, Scr is activated by its endogenous enhancers and additionally by formerly insulated enhancers, resulting in cumulated expression patterns.

Fig. 7.. Cp190 supports enhancer-blocking but not long-range pairing by Homie.

(A) WT expression of Fujioka et al. (37) transgenes with divergently transcribed GFP and LacZ reporter genes, integrated 142 kb upstream of eve in the vicinity of local hebe enhancers (green). eve endogenous enhancers (pink) are respectively active in anal plate, cardiac mesoderm, and specific neurons. When Homie insulator is between GFP and LacZ, it pairs in a head-to-head orientation with endogenous Homie downstream of eve, leading to schematized GFP and LacZ expression patterns. Below, RNA-FISH with antisense probes against GFP (top) or LacZ (bottom) mRNAs in midstage (stage 13) DAPI-stained control embryos with Cp190 (anterior left; posterior right; scale bars, 100 μm). RNA-FISH signal was false-colored green or pink when it was respectively detected in a deep (showing hebe enhancer-driven neuronal expression) or surface (showing cardiac mesoderm and anal plate expression marked by arrowheads) confocal slice. Note that neuronal expression driven by hebe enhancers masks that driven by eve enhancers, and anal plate signal is visible in all confocal slices. (B) Expression of same transgenes in Cp190⁰. Homie still supports long-distance reporter gene activation by distal eve enhancers (arrowheads) but is a weaker enhancer-blocker (as seen both by LacZ activation by hebe enhancer and activation of LacZ/GFP in the Homie forward/reverse transgenes, respectively, by the eve anal plate enhancer, although Homie is still able to block the eve cardiac enhancer).