AI-designed OpenCRISPR-1 enables efficient targeted mutagenesis and prime editing in rice

Recent advances in generative artificial intelligence (AI) have enabled the de novo design of genome-editing nucleases. For example, OpenCRISPR-1 offers an open-source alternative to naturally evolved CRISPR systems and expands the "freedom to operate" (FTO). Here, we report the development and systematic validation of a monocot-optimized OpenCRISPR-1-based genome-editing ecosystem in rice (Oryza sativa). By targeting the OsSWEET susceptibility (S) gene family, we demonstrate that OpenCRISPR-1 supports robust multiplexed editing in both rice calli and stable T0 plants, with mutation frequencies reaching 100% in some samples. Deep sequencing revealed that the OpenCRISPR-1 mutational landscape mirrors that of Streptococcus pyogenes Cas9 (SpCas9), facilitating the development of predictable loss-of-function alleles that confer broad-spectrum resistance to bacterial blight. To develop a fully open-source platform, we integrated an AI-designed Open sgRNA scaffold (OpsgRNA), which maintained high editing efficacy across multiple target loci, into our editing system. Furthermore, we expanded the toolkit by engineering OpenPE6c, an OpenCRISPR-1-based prime editing system. OpenPE6c exhibited precise editing rates in rice protoplasts comparable to that of canonical SpCas9-PE6c while significantly reducing imprecise byproducts, suggesting that the AI-designed nuclease has enhanced fidelity. Our results establish OpenCRISPR-1 as a versatile, high-performance, public-access platform for advanced plant genome engineering, offering a transparent framework for the global democratization of precision crop breeding.