A 600-Amino-Acid Scissor

The trail begins in a bacterium isolated from places including tomato roots. A team from Japan’s National Institute of Advanced Industrial Science and Technology, or AIST, TOPPAN Holdings Inc. and Inplanta Innovations Inc. found a compact RNA-guided DNA-cutting protein in Agrobacterium salinitolerans and engineered it into a genome-editing candidate. Its name is AsaTnpB-L. It works as a complex of one nuclease protein—the molecular scissor—and one guide RNA that specifies where to cut.

The organizations report that an engineered version edited the AAVS1 locus in cultured human cells at roughly 60% efficiency and could also perform genome editing in plants. Its headline attribute is size: about 600 amino acids, compared with 1,368 for SpCas9, the nuclease in the most familiar CRISPR-Cas9 system. On the protein-length measure, it is less than half as large.

Searching unknown genes by shape

A conventional search often looks for amino-acid sequences resembling those of known DNA-cutting enzymes. The Japanese team took another route. It converted functionally uncharacterized genes in a plant-pathogenic bacterial genome database into predicted protein structures, using an AI-enabled structural-prediction program, and screened the resulting shapes for resemblance to TnpB, a known class of RNA-guided nuclease.

This is not a case of “AI inventing a genome editor.” Computation narrowed the field. Researchers then prepared guide RNAs, tested whether candidate proteins actually cut DNA and determined the short recognition sequence adjacent to the target. They named the newly found structural family TnpB-L—TnpB-like RNA-guided nucleases—and selected AsaTnpB-L for engineering.

TnpB proteins are encoded by transposable elements, stretches of DNA able to move within genomes. In 2021, researchers reported direct evidence that TnpB could function as a programmable RNA-guided DNA nuclease. TnpB is widely regarded as an evolutionary ancestor of CRISPR-Cas12 effectors. AsaTnpB-L is therefore not a shaved-down Cas9. It is a different natural lineage rediscovered and repurposed.

The scissor, its guide and a TAM signpost

AsaTnpB-L binds DNA complementary to its guide RNA, but a matching guide is not sufficient. A nearby short sequence called a target-adjacent motif, or TAM, acts as a recognition signpost. The team determined that the original enzyme recognizes 5′-TTCAT-3′.

That requirement is both guardrail and restriction. A longer, more demanding motif may support specificity, but it also confines editing to places where the motif exists. By substituting amino acids in the nuclease, the researchers altered its recognition to 5′-TBAT-3′, where B represents T, C or G. The change is intended to widen the set of addressable sites.

The team also introduced multiple amino-acid changes around the predicted interface between AsaTnpB-L and its guide RNA. It was this structure-guided version that produced the reported editing rate of approximately 60% at AAVS1. A percentage at one locus cannot establish performance at other genes, in other cell types or in whole organisms.

Why a smaller scissor matters for delivery

Genome editing has a delivery problem as much as a cutting problem. A protein and RNA—or genetic instructions that make them—must enter the right cells and often reach the nucleus. When the carrier has strict cargo limits, a smaller nuclease can leave room for regulatory sequences or additional functions.

Plants present another constraint. The team says many established editing enzymes show stronger activity at 37°C or above, while many plants grow at roughly 20°C to 30°C. Looking in bacteria occupying a plant environment was a rational way to seek molecules more comfortable at lower temperatures.

But the release does not demonstrate packaging in a specific viral vector, delivery inside an animal, or a full activity curve across temperatures. “Easier to deliver” and “suited to plants” are promising design propositions, not proof of therapeutic delivery to a target organ or superiority in the field. Smallness enlarges the engineering space; it does not solve delivery by itself.

What “homegrown” does—and does not—mean

The organizations call AsaTnpB-L a “purely domestic” genome-editing tool. The practical center of that claim is intellectual property. Many foundational editing systems originated abroad, so Japanese companies may face foreign licensing negotiations and rights clearance before commercialization. The team says AsaTnpB-L was independently designed after a patent search to avoid infringing existing third-party patents in Japan.

Applications have been filed domestically, and part of the work is registered as Japanese Patent No. 7821459, “TnpB-like RNA-guided nuclease complex.” The listed inventors span AIST, Inplanta Innovations and TOPPAN Holdings. The AIST release names Akiyoshi Nakamura, Shigeo Sugano, Hiroshi Yamamoto and Nobutaka Mitsuda among the researchers leading the work.

“Domestic” does not mean that the bacterium, every database, prediction method, reagent or instrument originated in Japan. Nor does a Japanese patent search guarantee freedom to operate in every overseas market. Patent scope, surrounding delivery technologies and licensing terms will require separate analysis as products move closer to market.

A Japanese toolbox built in stages

CRISPR-Cas9 spread after researchers turned a bacterial immune mechanism into a programmable editing method. Emmanuelle Charpentier and Jennifer Doudna received the 2020 Nobel Prize in Chemistry for developing the technique. The achievement did not create a universal scissor: recognition motifs, operating temperatures, delivery constraints and patents differ, so laboratories continue to expand the toolbox.

The three Japanese partners have advanced in steps. In 2023, they reported a sonication-assisted whisker RNP method, using microscopic needle-like crystals and ultrasound to introduce a Cas9 protein–guide RNA complex into rice cells. Later that year they announced AalCas9, obtained from a microorganism isolated from deep-sea sediment and able to recognize an unusual PAM. Peer-reviewed papers on the rice-delivery method and AalCas9 followed in 2023 and 2024.

The long road beyond 60%

The next task is to make the headline number travel. Researchers will need results across multiple targets and cell types, independent replication and controlled comparisons with Cas9, Cas12 and other TnpB platforms. Genome-wide off-target studies must examine unintended cuts; investigators must characterize the mutations produced after repair and test cellular toxicity, immune recognition and long-term stability.

Medical use would require delivery and safety studies in animals before any clinical path. Agricultural use would require data on regenerated whole plants, inheritance, traits and environmental consequences. The announcement says “plant applicability” was demonstrated, but does not identify the species, target, efficiency or regeneration outcome. Those omissions prevent meaningful comparison with established plant editors.

The partners plan to build a broadly accessible Japanese platform for drug development, crop breeding, biomanufacturing, environmental applications and materials. They are scheduled to show the technology at BioJapan 2026 in October and the BMB2026 meeting in December. No release date, reagent-access terms, price, collaboration model or overseas patent plan has been published.

AsaTnpB-L need not replace Cas9 to matter. Its value would be in opening carriers that were too cramped, temperatures at which other enzymes struggled, or DNA addresses other tools could not conveniently reach. Six hundred amino acids is not the answer. It is room to ask more useful questions.