Abstract
The three-dimensional organization of eukaryotic genomes into compartments, topologically associating domains, and loops is mediated by architectural proteins whose organizational principles vary across species. In Drosophila , insulator proteins including Su(Hw) and the histone variant γH2Av form liquid-liquid phase separation (LLPS) condensates, yet how this phase separation capacity relates to genome compartmentalization has remained unclear. Here we use hyperosmotic stress to simultaneously displace architectural proteins from chromatin in both Drosophila and human cells, enabling a comparative dissection of genome organization principles across species. We find that although human CTCF shares predicted LLPS properties with Drosophila insulator proteins, it does not form condensates upon osmotic stress, while Drosophila insulator proteins do. Hi-C analysis reveals that osmotic stress causes loss of compartments, TAD boundary strength, and loops in both organisms, but genome recovery after stress is rapid and near-complete in human cells while remaining substantially incomplete in Drosophila after one hour. Analysis of this recovery asymmetry reveals a fundamental difference in compartment organization: whereas human A and B compartments engage in robust homotypic long-range interactions, Drosophila B compartments rarely participate in long-range B-to-B contacts, indicating that the Drosophila genome does not replicate canonical A/B compartment organization. Instead, Drosophila genome architecture is dominated by A-to-A interactions, and A compartments are specifically enriched in γH2Av and Su(Hw), with moderate enrichment of cohesin subunits. Furthermore, loops in Drosophila are mechanistically independent from compartments and TADs, recovering before compartment structure is restored, and are anchored by Su(Hw) and cohesin rather than by dCTCF. Together, these findings suggest that the LLPS properties of γH2Av and Su(Hw) underlie A compartment formation in Drosophila through a mechanism distinct from the heterochromatin-driven B compartment interactions that predominate in vertebrates, revealing fundamentally different organizational principles between fly and human genomes.