TAD boundaries and gene activity are uncoupled

Using high-throughput single-cell imaging, this study demonstrates that Topologically Associating Domain (TAD) boundary architecture and gene activity are largely uncoupled, as TAD boundary interactions are infrequent and independent of transcriptional status, while disrupting TAD boundaries does not alter gene expression.

The Gist

The Big Question: Do "Room Dividers" Control the Party?

Imagine your DNA isn't just a long string of letters, but a massive, bustling city. Inside this city, the streets are folded and packed into distinct neighborhoods called TADs (Topologically Associating Domains).

Think of a TAD like a specific apartment building or a gated community.

• The Walls (Boundaries): The edges of these neighborhoods are marked by "walls" made of special proteins (like CTCF). These walls are supposed to keep the residents inside the neighborhood interacting with each other and keep them from hanging out with the neighbors next door.
• The Residents (Genes): Inside the neighborhood are the "houses" (genes). Some houses are active (lights on, people talking), and some are inactive (lights off, quiet).

The Old Theory: Scientists used to think these "walls" were the strict bouncers of the city. They believed that if the walls were close together, the party inside (gene activity) would happen. If the walls were far apart, the party would stop. Essentially, the idea was: Structure dictates function. If you build the room right, the party happens.

The New Discovery: The Walls and the Party are Unrelated

This new paper, led by researchers at the NIH, asked a simple question: Do the walls actually control the party?

To find out, they didn't just look at a map of the city (which averages out millions of cells). Instead, they used a high-tech "super-microscope" to look at individual cells and even individual copies of genes (alleles) in real-time. They watched the distance between the walls and checked if the lights were on inside the houses.

Here is what they found, broken down into three simple experiments:

1. The "Active vs. Inactive" Test

The Analogy: Imagine checking 10,000 different houses in a neighborhood.

• The Expectation: You'd think the houses with the lights on (active genes) would have their walls pulled tight together, while the dark houses (inactive genes) would have walls far apart.
• The Reality: It didn't matter. The distance between the walls was exactly the same whether the gene was screaming "Party!" or sleeping soundly. The walls didn't care if the gene was working or not.

2. The "Turn the Music On/Off" Test

The Analogy: Imagine you have a giant switch to turn the music on (stimulate the gene) or off (inhibit the gene) for the whole neighborhood.

• The Expectation: If you turn the music on, the walls should snap together to create a cozy dance floor. If you turn it off, the walls should drift apart.
• The Reality: You could blast the music or silence the room completely, and the walls didn't budge. They stayed exactly where they were. The structure of the neighborhood didn't change just because the gene activity changed.

3. The "Knock Down the Walls" Test

The Analogy: This is the big one. The researchers used a special tool to dissolve the "walls" (the CTCF proteins) entirely.

• The Expectation: If the walls are the bouncers, removing them should cause chaos. The party should stop, or the wrong people should start talking to each other.
• The Reality: Surprisingly, the genes kept working just fine! Even without the walls, the genes continued to produce their proteins. The "party" kept going even though the "gated community" was gone.

(Note: They did find that removing a different protein called RAD21 did change things, but that's like removing the foundation of the building, not just the walls. The specific "boundary walls" (CTCF) weren't the ones holding the gene activity together.)

The Takeaway: The City is More Flexible Than We Thought

The Conclusion:
The paper concludes that TAD boundaries and gene activity are "uncoupled."

Think of it like this:

• Old View: The walls of a room determine what happens inside. If the walls are close, you can dance. If they are far, you can't.
• New View: The walls are more like a fence around a park. The fence exists to keep the park distinct from the street, but the fence doesn't decide if people are playing soccer or having a picnic inside. The people (genes) can play or stop playing regardless of how close the fence is.

Why does this matter?
For a long time, scientists thought that if we could fix the "folding" of DNA (the walls), we could cure diseases caused by genes going haywire. This paper suggests it's more complicated. The "folding" might be a side effect or a loose guideline, but it isn't the strict boss of the gene.

The genome is a dynamic, messy, and flexible place. The genes have their own internal logic for turning on and off, and they don't need the "walls" to hold their hands to do it. The structure of the genome and the activity of the genes are two different things that happen to live in the same neighborhood, but they aren't running the same show.

Technical Summary

1. Problem Statement

Topologically Associating Domains (TADs) are fundamental structural units of the genome, defined by boundaries enriched with the architectural protein CTCF and formed via the loop extrusion mechanism driven by cohesin. The prevailing hypothesis suggests that TADs function as regulatory insulators, facilitating enhancer-promoter interactions within a domain while preventing cross-talk between domains. However, a critical gap exists in understanding the causal relationship between TAD architecture and gene expression at the single-cell and single-allele level.

Population-based methods (e.g., Hi-C) average out the high heterogeneity and dynamics of chromatin structures, masking the fact that TAD boundaries are transient and that gene expression is stochastic (bursting). It remains unclear whether:

1. TAD boundary proximity correlates with the transcriptional state of genes within the TAD.
2. Acute changes in transcription (activation or inhibition) alter TAD boundary structure.
3. Disruption of TAD boundaries (via loss of architectural proteins) directly alters gene expression.

2. Methodology

The authors developed a high-throughput imaging pipeline to simultaneously visualize TAD boundaries and nascent transcription in single cells.

• High-Throughput DNA/RNA FISH (HiFISH):
  • Technique: Combined DNA-FISH (to detect TAD boundaries) and RNA-FISH (to detect nascent RNA transcripts) in a single hybridization step within 384-well plates.
  • Probes: Large BAC probes (~165 kb) targeting the 5' and 3' boundaries of specific TADs (EGFR and MYC) and control non-TAD regions. Stellaris RNA probes targeted intronic regions of nascent transcripts.
  • Cell Models: Human bronchial epithelial cells (HBEC), human foreskin fibroblasts (HFF), and HCT116 cells.
• Image Analysis (HiTIPS):
  • Custom software (HiTIPS) was used for automated segmentation and measurement of center-to-center distances between TAD boundary signals in 2D and 3D.
  • Interaction Definition: Alleles with boundary distances < 250 nm were classified as "paired" or interacting.
  • Activity Status: Alleles were classified as "active" if a nascent RNA signal was detected within 1 µm of the DNA signal.
• Experimental Perturbations:
  • Transcription Inhibition: Treatment with DRB (CDK9/7 inhibitor) to block RNA Polymerase II elongation.
  • Transcription Activation: Treatment with Dexamethasone (Dex) to induce glucocorticoid receptor target genes (ERRFI1, FKBP5, VARS2).
  • Protein Depletion: Auxin-inducible degron (AID) systems used to rapidly deplete RAD21 (cohesin component) and CTCF (boundary protein) in HCT116 cells.

3. Key Contributions

• Methodological Advance: Established a robust, high-throughput single-allele imaging pipeline capable of correlating structural proximity with transcriptional status in thousands of alleles per condition, overcoming the limitations of population-averaged data.
• Decoupling of Structure and Function: Provided direct evidence that TAD boundary proximity is largely independent of the transcriptional activity of genes within the TAD.
• Differential Roles of Architectural Proteins: Demonstrated that while cohesin (RAD21) depletion affects both structure and expression, the loss of the boundary-defining protein CTCF disrupts TAD architecture without significantly altering gene expression.

4. Key Results

A. TAD Boundaries Pair Infrequently and Transiently

• TAD boundaries (5' and 3') interact more frequently than non-TAD control regions (approx. 2–3 fold higher), but pairing is still a low-frequency event (occurring in only ~20–30% of alleles at any given time).
• This confirms that TADs are dynamic, probabilistic structures rather than static, stable domains.

B. No Correlation Between Boundary Proximity and Gene Activity

• Active vs. Inactive Alleles: In HBEC, HFF, and HCT116 cells, the median distance between TAD boundaries was identical for transcriptionally active and inactive alleles of EGFR and MYC.
• Statistical Significance: No significant difference in boundary distances was found regardless of whether the gene was bursting (active) or silent (inactive).

C. Transcription Dynamics Do Not Drive TAD Structure

• Global Inhibition: Acute treatment with DRB reduced nascent RNA levels by >90% but did not alter TAD boundary distances for EGFR or MYC.
• Gene-Specific Activation: Treatment with Dexamethasone robustly induced ERRFI1, FKBP5, and VARS2 (2–7 fold increase in RNA), yet TAD boundary distances remained unchanged compared to controls.

D. Disruption of Boundaries Does Not Necessarily Alter Expression

• RAD21 Depletion: Loss of the cohesin component RAD21 significantly increased TAD boundary distances (disrupting the loop) and reduced gene expression of EGFR and MYC. The authors suggest this is likely due to local effects on enhancer-promoter interactions rather than global TAD loss.
• CTCF Depletion: Loss of CTCF significantly increased MYC TAD boundary distances (disrupting the boundary) but had no significant effect on the expression levels of MYC or EGFR.
• Conclusion: The physical integrity of the TAD boundary (defined by CTCF) is not strictly required for the transcriptional regulation of genes within the domain.

5. Significance and Implications

• Redefining TAD Function: The study challenges the dogma that TADs are strict regulatory compartments essential for gene expression. Instead, it supports a model where TADs act as modulatory or probabilistic features that may limit inter-TAD interactions but are not the primary drivers of intra-TAD gene activation.
• Scale of Regulation: Transcriptional regulation appears to occur at finer scales (e.g., local enhancer-promoter loops or sub-TAD structures) rather than being governed by the global configuration of the TAD.
• Dynamic Nature: The uncoupling of structure and activity highlights that the genome is highly dynamic; genes can be active or inactive regardless of whether their TAD boundaries are currently "paired."
• Disease Context: While structural variations disrupting boundaries can cause disease (e.g., via enhancer hijacking), the study suggests that the mere presence or absence of a TAD boundary is not the sole determinant of gene expression levels, implying a degree of robustness in gene regulatory networks.

In summary, the authors conclude that TAD boundary architecture and gene activity are largely uncoupled, suggesting that the structural organization of the genome and its transcriptional output operate as partially independent layers of regulation.