⚗️ biochemistry

Supercoiling twists Cas9 off-target discrimination when nicking and cleaving

This study reveals that negative DNA supercoiling dramatically accelerates Cas9 off-target cleavage and alters its cutting sites, demonstrating that DNA topology critically influences Cas9 specificity and necessitates topological considerations for safer gene editing.

Original authors: Jaskovikaite, I., Offerhaus, H. S., Vinogradovas, M., Barkauskaite, U., Depken, M., Jones, S. K.

Published 2026-04-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jaskovikaite, I., Offerhaus, H. S., Vinogradovas, M., Barkauskaite, U., Depken, M., Jones, S. K.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: A Scissor That Gets Confused by Twisted Paper

Imagine Cas9 (the gene-editing tool) as a highly skilled scissor-wielding robot. Its job is to find a specific sentence in a massive library of books (your DNA) and cut it out so you can rewrite it. To do this, the robot uses a "search card" (the guide RNA) to find the exact page and line it needs.

Usually, this robot is very precise. But sometimes, it gets confused. It might find a sentence that looks almost right and cut the wrong page. This is called an off-target effect, and it's dangerous because it could break the wrong gene.

This new study discovered that the shape of the DNA plays a huge role in how confused the robot gets. Specifically, it looks at supercoiling—which is like twisting a rubber band or a piece of paper.

The Analogy: The Twisted Rubber Band

Think of DNA not as a straight string, but as a rubber band.

Relaxed DNA: The rubber band is lying flat on the table. It's easy to read, but the robot has to work hard to find the right spot.
Negatively Supercoiled DNA: The rubber band is twisted tightly in the opposite direction of its natural curl. This creates tension, making the rubber band want to unwind or pop open slightly.

The researchers found that when the DNA is twisted (supercoiled), it acts like a "pressure cooker" for the robot's mistakes.

Key Findings Explained

1. The "Twist" Makes Mistakes Explode

When the DNA is relaxed, the robot is picky. If the sentence on the page has a typo (a mismatch), the robot usually says, "Nope, that's not the right page," and moves on.

But when the DNA is twisted (negatively supercoiled), the tension forces the DNA to open up slightly. This makes it much easier for the robot to grab onto the wrong page.

The Result: The robot might cut a "wrong" page 1,000 times faster than it would on flat DNA. It's like the twisted rubber band is shouting, "Cut me here!" even if the text is slightly wrong.

2. The Robot Gets "Stuttery" (Changing Where It Cuts)

Sometimes, the twist doesn't just make the robot cut faster; it makes it cut in the wrong place.

The Analogy: Imagine you are trying to cut a piece of paper at a specific line. If the paper is crumpled or twisted, your scissors might slip and cut two inches to the left or right.
The Science: The study found that on twisted DNA, the robot sometimes shifted its cut by two letters (nucleotides). This is critical because cutting in the wrong spot can change the outcome of the gene edit entirely.

3. The "One-Handed" Scissors (Nicking vs. Cutting)

Cas9 has two blades (domains): one cuts the top strand, and one cuts the bottom strand. Usually, it uses both to make a clean double-cut.

The Discovery: The researchers found that on certain twisted, "typo-ridden" DNA, the robot gets confused enough that it only uses one blade. It makes a single nick (a small scratch) instead of a full cut.
The Twist: This happens because the twist in the DNA messes with the robot's internal sensors. One blade works fine, but the other gets jammed. This is actually good news! Scientists can use this "glitch" to create safer gene editors that only nick DNA (which is easier for the cell to fix without errors) rather than chopping it in half.

4. The Crystal Ball (The Computer Model)

The team built a computer simulation (called CRISPRzip) to predict this behavior.

The Analogy: Think of it like a weather forecast for gene editing. Instead of just guessing if it will rain, the model calculates how the "wind" (DNA twisting) will push the "clouds" (the robot) toward a mistake.
The Result: The model successfully predicted that twisted DNA makes the robot faster and more prone to errors. This helps scientists design better guides that won't get confused by the natural twists in our cells.

Why Does This Matter?

In our bodies, DNA is constantly being twisted and untwisted by processes like reading genes (transcription) and copying them (replication). It is rarely flat and relaxed.

The Problem: Current gene-editing tools are often tested on flat DNA in a lab dish. This paper warns us that in the real world (inside a living cell), the DNA is twisted, which means the robot might make way more mistakes than we thought.
The Solution: By understanding how the "twist" affects the robot, scientists can:
1. Predict where the robot might make mistakes in a human body.
2. Design better tools that ignore the twist or work safely with it.
3. Create new types of "nicking" tools that are safer for editing human genes.

The Bottom Line

This paper tells us that DNA topology (how twisted it is) is a secret variable that changes how gene editing works. If we want to edit genes safely in humans, we can't just look at the letters (A, C, T, G); we have to respect the shape and tension of the DNA strand itself. The "twist" can turn a precise surgeon into a clumsy one, or a clumsy one into a specialized tool, depending on how we use it.

1. Problem Statement

CRISPR-Cas9 is a powerful genome-editing tool, but its clinical and research utility is limited by off-target effects, where the nuclease cleaves DNA sequences that partially match the guide RNA (gRNA). While previous studies have established that DNA topology (specifically negative supercoiling) influences Cas9 activity, a comprehensive, high-resolution understanding of how supercoiling affects:

Cleavage kinetics across thousands of diverse off-target sequences.
The precise location of cleavage sites.
The differential activity of Cas9's two nuclease domains (HNH and RuvC) in the presence of mismatches and topological stress.

...has been missing. Current in silico prediction models often fail to identify bona fide off-targets because they lack variables accounting for DNA topology, which fluctuates dynamically in cells due to transcription and replication.

2. Methodology

The authors adapted the NucleaSeq method to analyze Cas9 activity on negatively supercoiled (nSC) plasmid libraries rather than just relaxed linear DNA.

Library Construction: They generated two plasmid libraries (Library 1 and 2) containing thousands of target sequences, including:
- Perfectly matched on-targets.
- Off-targets with single and double mismatches, insertions, and deletions relative to the gRNA.
- Alternative PAM sequences and non-target controls.
Supercoiling: Plasmids were purified from E. coli to achieve a negative superhelical density ( $\sigma$ ) of approximately -0.033.
Kinetic Assay: The plasmid libraries were exposed to Cas9 ribonucleoprotein (RNP) complexes. Samples were quenched at multiple time points (0 to 1000 minutes).
Sequencing & Analysis:
- Cleaved DNA fragments were separated from intact plasmids via restriction digestion (SapI) and size selection.
- Time-specific barcodes were added to samples for Next-Generation Sequencing (NGS).
- Read counts were normalized and fitted to exponential decay functions to determine effective cleavage rates ( $k_{clv}$ ) and cleavage positions with nucleotide resolution.
Domain-Specific Analysis: To dissect domain activity, the authors used:
- Wild-type (WT) Cas9.
- HNH-active/RuvC-inactive nickase (D10A).
- RuvC-active/HNH-inactive nickase (H840A).
- Fluorescently labeled linear DNA substrates to visualize single-strand nicking vs. double-strand cleavage.
Biophysical Modeling: The authors utilized and refined the CRISPRzip model, a stochastic simulation of R-loop formation. They integrated DNA torque parameters to simulate how negative supercoiling lowers the energy barrier for R-loop extension, allowing the model to predict cleavage kinetics based on sequence and topology.

3. Key Contributions & Results

A. Supercoiling Dramatically Accelerates Off-Target Cleavage

Magnitude: Negative supercoiling can accelerate off-target cleavage rates by up to 1,000-fold compared to relaxed DNA.
Sequence Dependence:
- Seed Region Mismatches: Off-targets with mismatches in the PAM-proximal "seed" region (positions 1–10) showed massive rate increases (up to 100-fold) under negative supercoiling.
- PAM-Distal Mismatches: These were already cleaved rapidly on relaxed DNA; supercoiling made them indistinguishable from on-target rates.
- Insertions vs. Deletions: Insertion mispairs showed the most dramatic topology dependence (some >10,000-fold increase). Conversely, deletions in the first 4 PAM-proximal positions were cleaved slower when supercoiled, likely due to misalignment exacerbated by torque.

B. Topology-Dependent Shifts in Cleavage Sites

Supercoiling alters the precise nucleotide position where Cas9 cuts.
Insertions: Off-targets with PAM-proximal insertions showed cleavage site shifts of up to two nucleotides further from the PAM on the target strand when supercoiled.
Deletions: Deletions often caused a shift of one nucleotide closer to or further from the PAM, independent of topology.
Significance: Shifts in cleavage position directly alter the resulting indel outcomes (e.g., frameshifts vs. in-frame edits), making prediction of editing outcomes difficult without considering topology.

C. Differential Inhibition of HNH and RuvC Domains

The study revealed that specific mispairs combined with supercoiling can convert WT Cas9 from a double-strand nuclease into a nickase (cutting only one strand).
Case Study (3A4): An off-target with an insertion at position 3 (3A4) showed topology-dependent nicking.
- On relaxed DNA, both HNH and RuvC domains were active (though weak).
- On supercoiled DNA, the HNH domain activity was suppressed, while RuvC remained active.
This suggests that local structural distortions caused by the combination of sequence mispairing and DNA torque can selectively disable one nuclease domain while leaving the other functional.

D. Validation via CRISPRzip Modeling

The biophysical model successfully reproduced the experimental cleavage kinetics across the library.
The model confirmed that negative torque effectively "tilts" the free-energy landscape of R-loop formation, stabilizing intermediates that would otherwise be unstable on relaxed DNA.
The model predicts that positive supercoiling (found in transcription bubbles) would have the opposite effect, potentially inhibiting cleavage.

4. Significance and Implications

Improved Off-Target Prediction: Current computational tools for predicting Cas9 off-targets often fail because they ignore DNA topology. This study provides the kinetic data necessary to integrate topological variables into prediction algorithms, leading to safer guide RNA design.
Mechanistic Insight: The work demonstrates that Cas9 does not just "read" sequence; it actively "senses" DNA topology. The interplay between sequence mismatches and DNA torque dictates whether Cas9 acts as a nuclease, a nickase, or remains inactive.
Therapeutic Applications:
- Supercoiling Sensors: Cas9 could be engineered to act as a sensor for DNA torque (e.g., detecting transcriptional activity or replication stress).
- Nickase Engineering: The discovery that specific mispairs can suppress HNH activity suggests a method to create "mutation-free" nickases without relying solely on protein engineering (D10A/H840A mutations), potentially useful for base and prime editing.
Biological Context: Since DNA supercoiling is a dynamic feature of the cellular environment (varying by cell cycle, gene expression, and chromatin state), these findings explain why Cas9 editing efficiency and specificity vary significantly between different genomic loci and cell types.

Conclusion

Jaskovikaitė et al. provide a genome-agnostic, high-resolution map of how DNA supercoiling modulates Cas9 specificity. By combining high-throughput NucleaSeq experiments with biophysical modeling, they reveal that negative supercoiling can turn weak off-targets into potent cleavage sites, shift cleavage positions, and even alter the enzymatic mechanism of Cas9 from double-strand cutting to single-strand nicking. These findings are critical for developing more predictable and safer gene-editing therapies.