Hyperosmotic stress induces rewiring of 3D chromatin interactions

Key results

1. 01

   ShownOne hour of 200 mM sorbitol removes most pre-existing loops in HEK293T eGFP-YAP1 cells while a few hundred mostly de novo loops appear; aggregate analysis shows loss and gain in wild-type HEK293T and HCT116 too, with weaker gain in reanalysed NaCl-treated T47D Hi-C.

   Hi-C (8.18 billion reads, 4 replicates per condition): 14,714 lost vs 361 gained loops (DESeq2, padj < 0.1). 97.2% of gained loops absent in untreated cells; 49% of gained anchors (354/722) reuse untreated anchor positions. Aggregate peak analysis (APA) on the same loop coordinates showed loss and gain in all three sorbitol-treated lines; signal was weaker in T47D with 110 mM NaCl (0.51 billion reads). · Hyperosmotic stress triggers widespread rewiring…; Figure legends

2. 02

   ShownGained loops are transient: detectable within 10 minutes, strongest at 1 hour and weakened by 24 hours. They are more punctate than size- and contact-matched pre-existing loops, enclose weaker domains with weaker stripes, and are longer than lost loops. Pre-existing loops start recovering from 3 hours.

   8-point Hi-C time course (0 h to 24 h) in wild-type HEK293T kept in 200 mM sorbitol, scored by APA; at 24 h gained-loop signal was lower but still above baseline. Aggregates of 241 gained loops of at least 150 kb (10-kb resolution) were compared with pre-existing loops matched on size and contact frequency (nullranges matchRanges). · Sorbitol-induced loops are transient and…; Figure legends

3. 03

   ShownStress-induced looping needs cohesin but not CTCF or YAP1: acute RAD21 degradation abolished gained-loop enrichment, whereas CTCF degradation or a dominant-negative YAP1 lacking its activation domain had no clear effect, even though CTCF remains bound at most gained anchors.

   Hi-C APA in HCT116 auxin-inducible degron lines (1 µM 5-Ph-IAA; near-complete loss of nuclear GFP) and low-depth eGFP-YAP1ΔTAD HEK293T. CUT&Tag after 1 h sorbitol: 13,742 CTCF and 204 RAD21 peaks decreased vs 421 and 42 increased; YAP1 12,553 decreased vs 21 increased. Over 70% of gained anchors were CTCF-bound before treatment and mostly stayed bound; retained CTCF peaks carried stronger CTCF motifs. · CTCF and cohesin are retained at…; Sorbitol-induced loop formation require cohesin…; Figure legends

4. 04

   ShownGained-loop anchors favour active promoters and enhancer-promoter contacts, and their promoters are enriched for GC-rich SP/KLF motifs; broad overlap with ChIP-seq peaks of most factors suggests the anchors mark highly active regions rather than SP/KLF-specific sites.

   Promoter overlap: 52.1% of gained anchors vs 12.6% of lost and 22% of static anchors, mostly promoters active in baseline RNA-seq. Enhancer-promoter loops: 53.7% of gained vs 16.0% of lost and 20.4% of static. HOMER against an expression-matched promoter background found SP and KLF motifs; ENCODE HEK293 ChIP-seq showed gained anchors overlap sites of nearly all tested transcription factors more often. · Sorbitol-induced loop anchors are enriched at…; Motif Enrichment Analysis; Figure legends

5. 05

   SuggestedGenes whose promoters anchor gained loops are induced hours after the loops form, largely through new transcription, and RAD21 loss modestly dampens this; most stress-responsive genes, however, lie outside rewired anchors and causality is not established.

   RNA-seq time course: differentially expressed genes (|log2FC| > 2, padj < 0.05) rose from a few hundred at 1 h to over 900 by 24 h. Exon-intron split analysis showed larger nascent-RNA gains at gained-anchor genes. 27.2% of responsive genes sit at gained or lost anchors. RAD21-depleted HCT116: gained-anchor genes lower (p < 0.001, n = 405), pre-existing-anchor genes not (p = 0.423, n = 3,026). · Sorbitol-induced loop formation precedes…; Discussion; Figure legends