Human Argonaute RNA base editor cuts predicted immune triggers while nearly doubling cytoplasmic editing
A Nature Biotechnology study published October 20 engineers human Argonaute-2 to build an RNA base editor with 18 predicted MHC-II peptides versus 370–389 for bacterial Cas13 editors.
Programmable RNA base editors built from bacterial Cas13 proteins can correct disease-linked transcripts without permanently altering genomic DNA, but their foreign bacterial scaffolds present hundreds of non-human peptides that risk triggering adaptive immune responses.
In a peer-reviewed Nature Biotechnology study published October 20 (doi:10.1038/s41587-026-03359-y), Stanford researchers led by Robert F. Lusi and Steven M. Banik repurposed human Argonaute-2 (AGO2) fused to human APOBEC3A (Y132D) to create the ALTER platform. [1][2]
By engineering AGO2 (ΔPAZ, Q633R) in ALTER-4 and pairing it with loop-inducing guide RNAs (14L8 and 12L6), the team remodeled target RNA structure to achieve selective editing while suppressing unwanted cleavage. In an EGFP reporter assay, ALTER-4 achieved 46.6% to 61.6% C-to-U editing versus 23.9% for RESCUE-S and 3.6% for CURE-N. Across four endogenous mRNAs (PPIB, TYMS, ACTB, SMARCA4) in HEK293T cells, ALTER-4 achieved 1.9-fold higher average editing with 4-fold to 25-fold lower off-targets. In MARIA deep-learning MHC-II screening, ALTER-4 displayed just 18 predicted immunogenic peptides, compared with 389 for CURE-N and 370 for RESCUE-S. [1]
Important biological boundaries apply. Because human AGO2 localizes primarily to the cytoplasm, ALTER edited the nuclear transcript MALAT1 at only 2.52% (9.95% with an added nuclear localization signal) versus 42.90% for CURE-N, and induced 374 to 443 differentially expressed genes. All experiments were conducted in cultured HEK293T cells, and the reduction in MHC-II epitopes is a neural-network prediction awaiting in vivo immunological validation. [1][3]
Take it further
Inspect the ALTER analysis and MARIA screening code For RNA-editing and protein-design researchers. Contains scripts used to score junction peptides and transcriptome-wide editing.
Sources & details 3 sources
Discuss in your voice app
Take the reporting, visual explanations and original sources into a conversation. Copy the brief, paste it into your preferred assistant, then start its voice mode.
Paste the brief into your voice app to discuss this reporting.Reporting sources
- Nature Biotechnology · Engineered human Argonaute 2 for targeted C-to-U RNA base editing
Peer-reviewed study by Lusi, Moncayo von Hase, Model, and Banik, published October 20, 2026 (doi:10.1038/s41587-026-03359-y). Reports ALTER-4 architecture, loop-inducing gRNAs, 1.9× endogenous editing, and 18 vs 370–389 MARIA peptides.
- GitHub · rflusi/ALTER-code analysis pipeline
Authors’ public repository for RNA-seq off-target quantification and MARIA immunogenicity screening scripts.
- NCBI BioProject · PRJNA1522030 sequencing deposit
Primary raw sequencing accession for transcriptome-wide C-to-U off-target profiling.
Images & figures
Replacing bacterial Cas13 with human AGO2 and MARIA screening cut predicted MHC-II epitopes from 370–389 to 18.
Euangelion · original data figure · Source · CC BY 4.0 · Download figure