AI Meets Rett: RSRT Funds $1M Collaboration with Profluent to Advance Gene-editing with Artificial Intelligence

I’m excited to share with our community a promising development that has been in the works for quite some time — RSRT has made a $1 million award to fund a partnership with Profluent Bio to apply artificial intelligence (AI) to gene editing for Rett syndrome. The collaboration was further solidified at our recent Rett Syndrome Genetic Medicines Summit. It’s a prime example of why that meeting of scientists, biopharma leaders, clinicians and regulatory experts is so important. Among the most exciting discussions at the Summit was a visionary roadmap for accelerated development of base editing therapeutics, presented by gene editing pioneer, Fyodor Urnov. 

A faster path to therapy

With the high-profile success of the Baby KJ treatment and based on recent interactions between regulators and leaders in the gene editing field including himself and fellow summit speaker David Liu, Fyodor outlined a novel platform-based strategy that could dramatically reduce timelines and costs.

Central to the strategy is the concept of a single adenine base editor (ABE), which acts as the engine that makes the correction, while the RNA guide serves as the GPS, directing the editor to the specific mutation site. The time and cost saving approach would require conducting IND-enabling experiments for one single mutation while all the other mutations, which only differ in their RNA guides, would only require off-target testing in cells (not animal models).

The RSRT Editing Consortium has demonstrated the feasibility of base editing restoring MECP2 expression, extending lifespan, and reducing Rett-like symptoms in mice. Importantly, RSRT has also established humanized mouse models and repositories of patient cells to support this new strategy.

 

Challenges to overcome

To bring this vision to the clinic safely and efficiently, improvements are needed for the ABE:

• Compact design:  While RSRT is invested in non-viral delivery, adeno-associated virus is currently the best approach to deliver therapies throughout the brain, however, ABEs are generally large and typically cannot fit into a single AAV. Theoretically two AAVs can be used but this increases the complexity, dose, price and more. Thus, we need an ABE that can fit into a single AAV vector.
• Broad targeting:  A key component of CRISPR-Cas systems, including ABEs, is a small DNA sequence called the PAM, which Cas proteins must recognize to target specific sites in the genome. Each guide RNA must not only bring the Cas protein to the site of the mutation but be directly next to a PAM sequence, which can restrict the number of mutations an ABE can correct.  The currently known enzymes which can fit in a single AAV have a PAM requirement that is limiting. Identifying a compact ABE with an expanded PAM will allow us to take advantage of the platform-based strategy outlined by Fyodor and use a single ABE to target more mutations. 
• Regulated control: The CRISPR-based editor uses a bacterial enzyme. Ideally, we need the enzyme to edit the mutation and then become inactive so it doesn’t accidentally cause edits beyond the intended one.  Also, there is concern that having a bacterial enzyme in the brain for long periods of time could be detrimental. So, we need an editor that could be turned on and off. The systems needed to accomplish this take up space in the AAV so being compact is really important.

 

Enter Profluent: AI meets gene editing

To tackle these challenges, RSRT is partnering with Profluent, a leader in AI-assisted protein engineering. Profluent’s generative AI platform creates novel synthetic CRISPR-based editors that have never existed in nature.

Profluent’s technology analyzes millions of biological sequences to design synthetic base editors with properties optimized for safety, precision, and delivery. With $1 million in funding from RSRT, they will work to develop an ABE that can be tightly regulated, packaged efficiently and applied across a wide range of Rett mutations.

“Much like how ChatGPT learns to generate language, our models learn from sequences of amino acids and nucleic acids. We can now explore possibilities in gene editing that evolution has never produced.”

- Hilary Eaton | Chief Business Officer, Profluent

 

The implications are enormous. Traditional CRISPR methods, derived from natural bacterial defense systems, have revolutionized research and offered new treatments for hereditary diseases like sickle cell anemia. However, AI-generated editors could expand these capabilities, producing tools that are faster, more precise and tailored. “The hope is to eventually create gene editors that are nimble and powerful in ways nature never intended,” noted Peter Cameron, Profluent’s Senior Vice President of Gene Editing and Translation.

“From RSRT’s perspective it’s important we access cutting edge technologies that will move the needle on cures for Rett,” says Bob Deans, RSRT’s Chief Technology Officer. “When those technologies are being advanced by people who care about patients and are motivated and inspired by them, it’s a win-win for all. Profluent fits this description!”