An open-source AI gene editor grants rice crops immune defense without patent fees

An AI-generated gene editor has proven just as powerful as traditional CRISPR in rice, offering plant breeders a high-precision toolkit free from restrictive patent claims.

The Giant Problem

CRISPR-Cas9 has revolutionised agricultural research, enabling precise genetic tweaks across dozens of crop species. Yet despite hundreds of successful lab trials, very few genome-edited crops ever make it to commercial farm fields. A major bottleneck is intellectual property. The foundational patents covering standard Streptococcus pyogenes Cas9 (SpCas9) are tightly held by a small group of institutions. For agricultural biotech companies and public breeding programs, acquiring the rights to commercialise a Cas9-edited crop variety can be prohibitively expensive. Furthermore, bacterial nucleases are not naturally optimised for plant biology and often require extensive engineering to function reliably without introducing unintended genetic errors.

The Science

To bypass these legal and biological hurdles, a research team led by Bing Yang at the University of Missouri turned to OpenCRISPR-1, a synthetic gene editor generated by a protein language model. Because it was designed by an AI exploring sequence possibilities unconstrained by natural evolution, OpenCRISPR-1 differs from canonical SpCas9 by 403 amino acids. Yet it retains the same molecular blueprint needed to slice double-stranded DNA. To adapt this synthetic nuclease for plants, the researchers built an editing package tailored for monocot crops. They added plant-specific nuclear localisation signals to guide the enzyme into the cell nucleus and paired it with an AI-designed, open-source guide RNA scaffold called OpsgRNA. In tests targeting the OsSWEET gene family—which dangerous bacterial pathogens hijack to steal sugars from rice plants—the AI tool achieved editing efficiencies matching SpCas9, reaching up to 100% mutation rates in plant calli and regenerated seedlings. Deep sequencing confirmed that the AI nuclease cuts DNA at the exact same position as standard Cas9, generating clean, predictable loss-of-function edits. The researchers also engineered a prime editor variant named OpenPE6c by fusing a modified version of the AI nuclease to a reverse transcriptase. In rice cells, OpenPE6c installed precise, single-letter sequence rewrites while producing significantly fewer unintended genetic errors than standard prime editors.

How They Did It

The team assembled binary vectors driving the AI-designed nuclease under a maize ubiquitin promoter and delivered them into immature rice embryos via Agrobacterium-mediated transformation. They validated the edits across single, dual, and triple gene targets using restriction fragment length polymorphism assays and deep-amplicon sequencing on an Illumina MiSeq platform.

Why You Should Care

By combining an AI-designed nuclease with a synthetic guide RNA, the researchers created a high-performance genome editing platform that operates entirely outside traditional CRISPR patents. When inoculated with three distinct strains of Xanthomonas oryzae, triple-knockout rice plants produced by the system demonstrated broad-spectrum resistance to bacterial blight, showing virtually no disease lesions. Subsequent plant generations stably inherited the edited genes, and the team successfully recovered transgene-free, disease-resistant offspring. This open-access system provides a clear blueprint for public institutions and global breeders to develop resilient, high-yield commercial crops without facing severe licensing barriers.