One Dose to Correct a Typo at the RNA Level: New Rett Syndrome Therapy Takes a Preclinical Step Forward

The challenge in treating Rett syndrome is not only replacing the missing protein, but also precisely controlling how much is restored in neurons throughout the brain. Shape Therapeutics and the Rett Syndrome Research Trust (RSRT) announced a collaboration to advance an AI-designed RNA-editing therapy intended to require only a single administration, with the aim of restoring MeCP2 protein from the disease-causing message itself.

Rett syndrome is primarily caused by mutations in the MECP2 gene on the X chromosome, occurs mostly in females, and can lead to developmental regression, motor and communication impairments, seizures, and breathing abnormalities. Shape is currently targeting the R168X nonsense mutation. This mutation inserts a premature stop signal into the instructions for making the protein, preventing cells from producing full-length MeCP2.

At the core of the therapy is an engineered guide RNA that directs the cell’s existing ADAR enzyme to a specific site on MECP2 messenger RNA, converting adenosine into inosine, which the cell reads as guanosine, thereby rewriting the erroneous stop codon. This operation does not directly alter DNA. The AI described by Shape is used to identify guide RNAs with better editing efficiency and specificity from a large number of possible sequences, rather than to directly determine by algorithm whether a patient is suitable for treatment.

A preprint study released by Shape in 2026 showed that after the therapy was administered intravenously to mice carrying the R168X mutation, editing of the target RNA reached approximately 70%, MeCP2 protein was markedly restored across multiple brain regions, and Rett-like manifestations and survival also improved. These results are still limited to an animal model of a single mutation, the paper has not yet completed peer review, and the same efficacy in humans cannot be directly inferred.

The concept of a “single administration” relies on a viral capsid developed by Shape to deliver a genetic payload capable of continuously producing guide RNA into the central nervous system. The correction to the RNA molecule itself is not permanent, but if the vector can remain in cells over the long term, it may continuously recruit ADAR. This differs from strategies requiring regular intrathecal injections of small RNAs. RSRT said the collaboration also allows it to compare viral delivery with other delivery routes. The organization has currently committed a total of $10.4 million to its Rett RNA-editing research portfolio, but the amount allocated to this individual collaboration and the development timeline have not been disclosed.

The biggest hurdles remain delivery and dosing. Whether viral vectors can safely and evenly reach the human brain, whether they will trigger immune responses, and whether the guide RNA might inadvertently edit other transcripts all require more comprehensive toxicology studies and long-term follow-up. Too little MECP2 expression causes disease, while too much may likewise damage the nervous system. Therefore, “how much to restore” and “which cells are corrected” will be important questions before human trials can begin.

In addition, R168X accounts for only a subset of patients with Rett syndrome. Different MECP2 mutations may each require separately designed and validated guide RNAs. This collaboration moves a technology that has already demonstrated proof of concept in mice toward more systematic preclinical development, but there are currently no human trial data, and no regulatory filing or date for the first clinical trial has been announced.