Engineering ADAR1p150- and dCasRx-based systems for programmable RNA base editing in plants

Programmable RNA base editing enables precise, transcript-level modification without altering the underlying genomic DNA, offering a reversible approach to manipulating gene function. However, its application to endogenous plant transcripts remains largely unexplored. Here, we systematically evaluated two programmable RNA base editing strategies based on ADAR1p150 and dCasRx for targeted transcript engineering in plant cells. Using an A-to-I editing-responsive GUS-LUC reporter, we quantitatively characterized engineered ADAR1p150 variants and cognate recruiting RNAs (arRNAs), with the optimal combination yielding a 12.4-fold enhancement in reporter editing activity. However, this optimized ADAR1p150-arRNA system failed to produce detectable editing events across seven endogenous Arabidopsis transcripts with varied expression levels in Arabidopsis protoplasts, nor in stable Arabidopsis and rice lines. We therefore screened a panel of dCasRx fusion constructs harboring candidate A-to-I and C-to-U deaminase domains and identified TadA8e-dCasRx (8e-RX) as a functional A-to-I RNA editor for plant transcripts, while three cytidine deaminase fusions (APOBEC3A, PpPPR65^DYW, and AtMEF1^DYW) exhibited no measurable C-to-U editing activity. The 8e-RX editor achieved robust A-to-I editing efficiencies ranging from 24% to 37% at three endogenous Arabidopsis transcript sites. Further protein engineering failed to substantially improve editing performance, whereas a loop-forming guide RNA (gRNA) architecture successfully repositioned the editing window and enabled targeted modification at the AtALS locus, which corresponds to the herbicide-resistance-associated K256E amino acid substitution. Notably, stable transgenic plants exhibited inconsistent accumulation of the 8e-RX editor and lacked reproducible target editing. Collectively, this study establishes the feasibility of dCasRx-mediated programmable RNA base editing in plant cells and lays a foundation for further optimization of editing efficiency and stable in planta editing performance.