Packing the guide inside the nuclease: TnpB's single-transcript editing trick

In Short. ISDra2 TnpB can keep its guide RNA inside its own coding sequence, process that mRNA into a working reRNA, and edit plant genomes from one transcript. That compactness is what lets a Tobacco rattle virus cassette carry both the editor and a visual silencing screen in Nicotiana benthamiana.

---

Gene editors need two parts that usually travel separately: a nuclease protein and a short guide RNA that tells it where to cut. In plants, those parts are often delivered by viruses with tight cargo limits. Split the nuclease and the guide into two expression units and the size advantage of a compact nuclease shrinks before the particle is even built.

TnpB proteins from IS200/IS605-family transposons are RNA-guided nucleases and evolutionary relatives of Class 2 type V CRISPR effectors such as Cas12. ISDra2 TnpB is among the best studied of them. In bacteria, its guide (called reRNA) already overlaps the 3′ end of the tnpB gene. Wang and colleagues show that this native overlapping layout can drive detectable genome editing in plants when kept as a single transcript, without exogenous ribozymes or a separate guide promoter (bioRxiv DOI 10.64898/2026.09.09.750255).

One transcript, two jobs

Their single-transcript unit (STU) places a rice codon-optimized ISDra2 TnpB coding sequence, leaves the reRNA-overlapping stretch unoptimized, then adds only a 7 bp conserved motif (GGTTCAA) and a 16-20 bp spacer downstream. A dual-transcript unit (DTU) expresses TnpB and reRNA from separate promoters for comparison. In stable transgenic rice, STU editing efficiencies matched DTU at four of eight tested loci, lagged at two (OsYSA and OsPDS-T2), and exceeded DTU at two (OsWaxy and OsPi21). Nanopore full-length transcript sequencing recovered 5′-truncated TnpB RNAs consistent with processed reRNA-like molecules in planta. The same STU logic worked for a second ortholog, ISAam1 TnpB, in soybean hairy roots, again with STU efficiencies comparable to DTU.

The platform claim is architectural. The nuclease mRNA is also the guide precursor. Separating those jobs is optional, not required, once the native overlap is respected.

Why viral packaging cares

Tobacco rattle virus (TRV) is a workhorse for virus-induced genome editing (VIGE) in N. benthamiana. Cargo space is limited. The authors loaded STU into TRV2 and targeted NbChlH, whose knockout turns tissue chlorotic. Editing frequencies reached as high as 86.3% in chlorotic systemic leaves for an early STU construct. Optimized STU variants (full 224 bp overlapping reRNA region, or a 116 bp truncated core) improved systemic editing further relative to a DTU that joined TnpB and guide with a hammerhead ribozyme on one viral transcript.

Heritability followed the same pattern. Seeds from chlorotic capsules yielded about 10-40% chlorotic progeny for the first STU design, with many green seedlings still showing roughly 50% editing. Optimized STU constructs pushed chlorotic seedling fractions up to 100% in fully chlorotic capsules. The ribozyme-linked DTU produced about 30-40% chlorotic progeny. RNA FISH sixty days after infiltration helps explain the gap. In STU-infected plants, both viral coat-protein and TnpB signals persisted in developing leaf primordia. In DTU-infected plants, coat-protein signal remained while TnpB signal was gone. Compact cargo that stays on the virus during systemic spread is the packaging lesson. Visible editing early does not guarantee the editor still rides along weeks later.

A second module that fits because the first is small

Silent edits are hard to find among uneven viral infection. The authors therefore packed an NbPDS silencing fragment onto the same TRV-STU backbone. Transient albino tissue marks capsules worth screening for heritable edits at a second gene (NbChlH in their demo), while PDS silencing itself rarely inherits. Adding that module generally lowered editing efficiency, and a DTU-plus-silencing version performed worst, with weak or undetectable systemic albino phenotypes. STU versions still supported clear dual VIGE-VIGS behavior. Compactness bought room for a screen, not magic immunity to cargo trade-offs.

Scope that stays plant-shaped

This preprint is about rice, soybean hairy roots, and N. benthamiana under TRV. It does not establish a human therapy path, a clinical packaging standard, or a Cas-brand replacement story. TnpB’s appeal for constrained delivery has been discussed broadly in the literature the authors cite. The new evidence here is that the native overlapping reRNA architecture can be the operating system for plant editing and for a single viral cassette that edits and marks in one pass. Anyone comparing “smaller editors” should ask whether the guide still needs its own cassette. That question is measured in cargo diagrams and systemic retention, not in slogans.

Wang et al. A native single-transcript TnpB architecture enables efficient virus-induced genome editing and visual screening of heritable progeny harboring phenotypically silent edits. bioRxiv (posted 2026-09-10). Preprint, not peer-reviewed.