This study is led by Professor Jian Tian (Institute of Animal Science, Chinese Academy of Agricultural Sciences, Beijing, China). The authors developed a TnpB generation model based on the protein large language model MP-TRANS. Using this model, they directly generated 100,000 multi-site mutant sequences of ISDra2 TnpB. Based on strategies of minimal recovery rate and optimal energy, the top five sequences were selected for subsequent experimental validation.
The results showed that in HEK293T cells, the wild-type ISDra2 TnpB and the mutants TnpB-TA, TnpB-TC, and TnpB-TD exhibited high editing efficiencies at the AGBL-1, AGBL-2, ROSA26-1, ROSA26-3, and AAVS-2 loci. Notably, at the AGBL-1, AGBL-2, ROSA26-1, and ROSA26-3 loci, the editing efficiency of the TnpB-TD mutant was 1.2-, 1.2-, 1.46-, and 1.5-fold higher than that of the wild-type ISDra2 TnpB, respectively. At the same time, both the wild-type ISDra2 TnpB and the TnpB-TD mutant maintained low off-target rates at these target sites. Further analysis of editing events with efficiencies above 0.1% revealed that the same nuclease produced different types of edits at different targets, and that the wild-type ISDra2 TnpB and TnpB-TD also showed distinct editing profiles at the same target. Overall, the TnpB-TD mutant generated more diverse editing outcomes than the wild-type and demonstrated stronger long-fragment (>10 bp) editing capabilities at all four target sites.
To investigate the molecular mechanism underlying the enhanced editing activity of the TnpB-TD mutant, the authors performed a 500 ns molecular dynamics simulation analysis. The results indicated that TnpB-TD exhibited improved overall stability, occupied a more confined conformational space, and displayed a broader energy basin compared to the wild-type ISDra2 TnpB. In addition, structural alignment and Coulomb potential analysis revealed that TnpB-TD formed a more stable interaction with ωRNA.
See the article:
Engineering high-efficiency mutants of the transposase ISDra2 TnpB via protein-generative models and energy-optimized design
https://www.sciencedirect.com/science/article/pii/S266217382600041X
aBIOTECH
16-Feb-2026