In brief: Scientists have converted human Argonaute 2, a natural RNA-handling protein, into a programmable editor. In cultured cells, the resulting ALTER system changed selected cytosines to uracils and produced fewer predicted immune-triggering peptides than editors built around foreign proteins.
Many programmable RNA tools borrow their targeting machinery from bacteria. That strategy has produced powerful editors, but it can complicate delivery and raise questions about immune recognition. A new study starts from a protein human cells already use.
Published in Nature Biotechnology on 20 October 2026, the work repurposes human Argonaute 2, or AGO2, as an RNA-targeting platform. The researchers fused engineered AGO2 to a deaminase, an enzyme that can convert cytosine to uracil in RNA. They call the system AGO-led targeted editing of RNA, or ALTER.
Why edit RNA instead of DNA
DNA editing can make a permanent genomic change. RNA editing acts on temporary messenger molecules copied from DNA, so its effects fade as those molecules are replaced. That reversibility may be useful when a lasting genome change is unnecessary or too risky, although repeated treatment and reliable delivery may then be required.
A cytosine-to-uracil change can alter the instructions carried by a messenger RNA. In principle, that could correct certain disease-causing messages or change how a protein is produced. In practice, an editor must reach the right tissue, modify enough target RNA, avoid unintended sites and do so without harmful immune or cellular effects.
ALTER changes the guide–target geometry
AGO2 normally uses short guide RNAs to find matching transcripts and can cut them. The team designed guides that remodel the target RNA into a loop, exposing the chosen cytosine to the attached deaminase while reducing AGO2’s native cutting activity. The loop size, target position and orientation can be tuned.
In HEK293T cells—a widely used human kidney-derived laboratory cell line—ALTER achieved an average of 1.9 times more on-target C-to-U editing than the Cas13-based CURE and RESCUE editors across the messenger-RNA targets tested. Transcriptome-wide sequencing found similar or lower levels of off-target editing in those comparisons.
Human-derived does not automatically mean invisible
Using human AGO2 could reduce one translational obstacle, but the finished editor is still engineered. It includes altered amino acids, a fused deaminase and junctions the immune system has not necessarily encountered. The researchers therefore used a machine-learning model to predict peptides that might be presented to immune cells.
The model predicted fewer potentially immunogenic peptides for ALTER than for the Cas13 editors, and the team applied an AI-assisted “deimmunization” workflow to redesign candidate regions. These are computational predictions, not clinical immune-safety results. Human immune responses depend on delivery method, dose, tissue and genetic variation, and must ultimately be tested experimentally.
The study is a platform demonstration
The paper’s data come from cultured cells, not animals or patients. It does not show that ALTER can be packaged, delivered to a diseased organ or maintained at a therapeutic concentration. It also compares systems under specific experimental conditions; a 1.9-fold average advantage is not a universal performance ranking for every target.
The authors have released analysis code through the ALTER-code repository, supporting scrutiny of the RNA-sequencing work. Further studies will need to reproduce the result, compare delivery formats, measure longer-term effects and test disease-relevant models.
Researchers will also need target-by-target design rules. RNA folds into structures that can hide or expose a base, and expression levels vary between tissues. A guide that performs well against a fluorescent reporter may behave differently against a scarce disease transcript in a primary cell.
The important advance is conceptual: a central human RNA-interference protein can be expanded beyond its natural role and used as a modular programmable scaffold. ALTER is not an RNA medicine yet. It is a new candidate chassis from which future RNA tools might be built.
Leave a Reply