Distinct principles of genome compartmentalization in Drosophila and humans revealed by osmotic stress

This study reveals that while osmotic stress disrupts genome architecture in both *Drosophila* and humans, the two species employ fundamentally distinct organizational principles, with human genomes relying on robust homotypic A/B compartment interactions and rapid recovery, whereas *Drosophila* genomes are dominated by A-to-A interactions driven by LLPS-capable proteins like Su(Hw) and $\gamma$H2Av, exhibit incomplete recovery, and utilize loops anchored by Su(Hw) and cohesin rather than dCTCF.

The Gist

Imagine your genome (your DNA) isn't just a long, messy string of code. It's more like a massive, bustling city. To keep things organized, the city is divided into neighborhoods (compartments), specific zones with their own rules (domains), and tiny, direct phone lines connecting important buildings (loops).

This paper is a story about what happens when we throw a "stress test" at two different cities: one in a human and one in a fly (Drosophila). The stress test was simple: we made the environment very salty (hyperosmotic stress), which is like suddenly draining the water out of the city, forcing everything to shrink and squeeze together.

Here is what the scientists found, broken down into everyday analogies:

1. The "Shrink Ray" Effect

When they applied the salty stress, both the human and fly cells reacted the same way physically: their DNA "shrunk" or condensed. It was like the city buildings were pushed closer together to save space.

• The Human Reaction: Once the salt was washed away and normal conditions returned, the human city bounced back quickly. The buildings spread out, and the neighborhoods reformed almost perfectly within an hour.
• The Fly Reaction: The fly city also shrank, but when the salt was washed away, it didn't fully bounce back. After an hour, it was still squished and disorganized. It was as if the fly city got stuck in a "traffic jam" and couldn't get its rhythm back.

2. The "Architects" (Proteins)

Every city needs architects to build and maintain its structure. In our DNA, these architects are proteins like CTCF (in humans) and Su(Hw) (in flies).

• The Human Architect (CTCF): When the stress hit, the human architect stepped back from the job site but didn't change much. It didn't form any special structures.
• The Fly Architect (Su(Hw)): This one was dramatic. When the stress hit, the fly architects didn't just step back; they grabbed their tools and formed giant, sticky bubbles (condensates) in the middle of the cell. Think of them like a group of construction workers suddenly huddling together in a giant, gelatinous blob to wait out the storm. This "blob" behavior is called liquid-liquid phase separation.

3. The Neighborhoods (Compartments)

DNA is divided into "Active" neighborhoods (A) where genes are working, and "Quiet" neighborhoods (B) where genes are sleeping.

• In Humans: The "Quiet" neighborhoods (B) like to hang out with other "Quiet" neighborhoods, and the "Active" ones (A) hang out with other "Active" ones. It's a very balanced social life. When the stress hit, this social order broke, but once the salt was gone, the neighborhoods quickly found their friends again.
• In Flies: Here is the big surprise. The fly "Active" neighborhoods (A) still hung out with other "Active" ones. But the "Quiet" neighborhoods (B) didn't hang out with each other at all! They were loners. The fly city is organized mostly by the "Active" areas sticking together, while the "Quiet" areas are just kind of floating around without a strong social network. Because of this, when the stress hit, the fly city had a much harder time reorganizing because it lacked that strong "Quiet neighborhood" glue.

4. The Phone Lines (Loops)

Loops are like direct phone lines connecting two specific points on the DNA to make sure a gene gets turned on.

• The Discovery: In flies, these phone lines are built by a different team than the one building the neighborhoods. The "phone lines" rely heavily on the Su(Hw) architect and a helper called Cohesin.
• The Twist: Even though the fly city was a mess after the stress, the phone lines (loops) started to reconnect before the neighborhoods (compartments) were fixed. This proves that in flies, the phone lines and the neighborhoods are built by different crews and operate on different schedules.

The Big Takeaway

This study reveals that flies and humans build their genetic cities using different blueprints.

• Humans rely on a "loop extrusion" machine (like a train pulling a rope) to organize their DNA. When the machine stops (due to stress), it's easy to restart it, so the city recovers fast.
• Flies rely more on sticky, phase-separated blobs (like the Su(Hw) architects forming bubbles) to organize their DNA. When the stress breaks these bubbles, it takes much longer for the architects to reform the blobs and reorganize the city.

Why does this matter?
It shows that even though flies and humans look very different, their DNA is organized in fundamentally different ways. Understanding these differences helps us learn how cells handle stress, how they recover from damage, and why certain diseases might affect one species but not the other. It's like realizing that while both New York and Tokyo are big cities, one is built on a grid system (humans) and the other on a winding, organic system (flies)—and they react to earthquakes very differently!

Technical Summary

1. Problem Statement

The three-dimensional (3D) organization of eukaryotic genomes involves hierarchical structures: chromosomal territories, A/B compartments, Topologically Associating Domains (TADs), and loops. While architectural proteins like CTCF, cohesin, and insulator proteins mediate these structures, the underlying organizational principles vary significantly between species.

• The Gap: In mammals, genome folding is largely explained by the loop extrusion model (CTCF-cohesin mediated). In Drosophila, the role of loop extrusion is debated, and alternative mechanisms like Liquid-Liquid Phase Separation (LLPS) involving insulator proteins (e.g., Su(Hw)) and histone variants (γH2Av) have been proposed.
• The Question: How do the mechanisms governing genome compartmentalization differ between Drosophila and humans? Specifically, does the LLPS capacity of Drosophila architectural proteins translate to a distinct mode of genome organization compared to the CTCF-driven model in humans?

2. Methodology

The authors employed a comparative approach using hyperosmotic stress as a perturbation tool to displace architectural proteins from chromatin in both Drosophila (S2 cells) and human (A375 cells) models.

• Experimental Design:
  • Stress Induction: Cells were treated with 250 mM NaCl for 1 hour to induce hyperosmotic stress.
  • Recovery: Cells were returned to normal media for 1 hour to assess structural recovery.
  • Controls: Untreated cells served as controls.
• Techniques:
  • Immunofluorescence: Stained for CTCF (human) and dCTCF/Su(Hw) (Drosophila) and DAPI to visualize chromatin volume and protein localization (condensate formation).
  • Hi-C Sequencing: Performed at high resolution (250 kb for humans, 20 kb for flies) to map genome-wide chromatin contacts.
  • Computational Analysis:
    • Compartment Analysis: Principal Component Analysis (PCA) to define A (active) and B (inactive) compartments.
    • Saddle Plots: To quantify homotypic (A-A, B-B) vs. heterotypic (A-B) interaction strengths.
    • TAD/Loop Calling: Insulation scores for TAD boundaries and "Mustache" algorithm for loop detection.
    • ChIP-seq Integration: Mapped publicly available ChIP-seq data for architectural proteins (Su(Hw), γH2Av, cohesin, Polycomb, dCTCF) onto Hi-C features.

3. Key Results

A. Differential Response of Architectural Proteins

• Human (A375): Osmotic stress caused chromatin condensation and reduced CTCF occupancy, but CTCF did not form visible condensates. Upon recovery, chromatin volume and CTCF distribution returned to near-control levels within 1 hour.
• Drosophila (S2): Stress induced chromatin condensation and the formation of insulator body condensates (LLPS) containing Su(Hw) and cohesin. While chromatin volume partially recovered, the genome structure remained significantly altered after 1 hour compared to human cells.

B. Global Genome Reorganization

• Contact Shifts: Both species showed a loss of short-range contacts and a gain in long-range contacts during stress.
• Recovery Asymmetry: Human genomes recovered compartment and TAD structures rapidly and nearly completely. Drosophila genomes showed incomplete recovery; TAD boundary strength and compartment integrity remained significantly degraded after 1 hour.

C. Fundamental Differences in Compartment Organization

• Human: Exhibits canonical A/B compartmentalization with robust homotypic long-range interactions for both A-A and B-B compartments.
• Drosophila: Displays a non-canonical organization:
  • A-A interactions: Strong and dominant, similar to humans.
  • B-B interactions: Rarely participate in long-range contacts. The Drosophila B compartments lack the strong homotypic interactions seen in vertebrates.
  • Protein Association: Drosophila A compartments are specifically enriched in γH2Av and Su(Hw), whereas B compartments are depleted of these. This suggests A-compartment formation is driven by the LLPS properties of these specific proteins.

D. Decoupling of Loops, TADs, and Compartments in Drosophila

• Loop Recovery: Loops in Drosophila recovered faster than TAD boundaries or compartments.
• Loop Anchors: Unlike humans (CTCF-cohesin), Drosophila loops are anchored by Su(Hw) and cohesin (Nipped-B/Rad21), but not by dCTCF or Polycomb.
• Mechanistic Independence: The study demonstrates that loops can reform independently of TAD boundaries and compartments in flies, suggesting these structures are mechanistically distinct in Drosophila, unlike the tight coupling often observed in mammals.

4. Key Contributions

1. Revealed Species-Specific LLPS Roles: Demonstrated that while human CTCF has predicted LLPS properties, it does not form stress-induced condensates in vivo, whereas Drosophila insulators (Su(Hw)) and γH2Av do, driving a distinct organizational mechanism.
2. Redefining Drosophila Compartments: Challenged the canonical A/B model for Drosophila, showing that the genome is dominated by A-A interactions while B compartments fail to engage in long-range B-B contacts.
3. Identified Alternative Loop Mechanisms: Provided evidence that Drosophila loops are mediated by Su(Hw)-cohesin interactions rather than the CTCF-cohesin loop extrusion model dominant in mammals.
4. Stress as a Dissection Tool: Validated hyperosmotic stress as a powerful tool to decouple and compare the recovery kinetics and dependencies of different 3D genome features across species.

5. Significance

This study fundamentally alters the understanding of genome architecture evolution. It suggests that while the phenotypes of 3D genome organization (compartments, TADs, loops) appear conserved, the molecular mechanisms driving them are distinct:

• Vertebrates: Rely heavily on loop extrusion (CTCF-cohesin) for TADs and heterochromatin-driven B-B interactions for compartmentalization. Recovery is rapid due to the re-loading of cohesin.
• Drosophila: Relies on LLPS-driven mechanisms involving Su(Hw) and γH2Av for A-compartment formation and loop anchoring. The lack of a fast loop-extrusion recovery pathway and the unique B-compartment behavior explain the slower, incomplete structural recovery observed under stress.

These findings imply that the "canonical" mammalian model of genome folding cannot be universally applied to invertebrates and highlight the critical role of phase separation in shaping the Drosophila epigenome.