Bridge recombinases close the gap in human genome editing

A highlight of Science https://doi.org/10.1126/science.adz1884 (2026)

NB: This is not a preprint, but rather a short highlight I wrote for a (sadly unsuccessful) editor job application. I am, however, pleased enough with it to share it here!

Current gene editing strategies support targeted repair of small aberrations, but scar-free editing at larger scales remains out of reach. Writing in Science and publishing back-to-back with Perry et al., Pelea et al. establish the Insertion Sequence 110 (IS110) RNA-guided bridge recombinase ISCro4 as a promising candidate for programmable human genome editing.

Bridge recombinases are newly described bacterial editing tools that facilitate mobile genetic element integration. The recombinase is in complex with two linked bridge RNA (brRNA) loops, target-binding (TBL) and donor-binding (DBL), containing a 14 bp recognition sequence. This recognition sequence enables payload delivery and integration via site-specific recombination, without the need for host repair machinery. Here, the authors assayed multiple IS110 recombinases in a human embryonic kidney cell line, and ISCro4 emerged as the lead candidate with the highest gene deletion efficiency in a plasmid reporter. Engineering of the brRNA to split the TBL and DBL, natively joined by a short linker, doubled ISCro4 recombinase activity. ISCro4 was then validated against diverse editing functions beyond deletion, including plasmid reporter inversion, insertion, and with multiple targets.

The authors followed these data with the killer experiment: genomic targeting of the recombinase to disease-relevant loci. Recombination efficiencies were expectedly lower than for plasmid reporters, sitting at 0.5%, 3.5%, and 6% for genomic deletion, inversion, and insertion, but nonetheless these data provide proof-of-principle that ISCro4 can perform programmable genome editing in human cells. Further engineering is required to address the identified off-targeting of ISCro4, due to a combination of mismatch tolerance in the T/DBLs and the limited size of the 14 bp recognition sequence.

This work and the accompanying article in Science are the first demonstration of successful human genome editing using bridge recombinases. The theoretical payload of these bridge recombinases exceeds that of existing CRISPR-based technologies, and may prove more effective at correcting whole gene deletions, inversions, and repeat expansions. Improvement of the recognition sequence specificity will be an important area of future investigation.