This week we publish in @NatureBiotech "High-precision cytosine base editors by evolving nucleic-acid-recognition hotspots in deaminase". In this work, we show that both nucleotide and context specificity of deaminases can be reprogrammed by hotspot-focused directed evolution 1/n

A heroic amount of work went into this study. I'm incredibly grateful for the brilliant and driven scientists in my lab who made it possible, especially graduate student Yuan Wu, who led this effort and will be job hunting soon! 11/n.

All plasmids have been deposited to Addgene: If you are interested in trying these editors, we welcome independent testing and feedback! Forgot to post the paper link: 10/n.

Admittedly, not all NCN editors performed at the same level of specificity. We observed cases where highly active context-specific deaminases, while showing ideal precision in vitro, conferred reduced specificity in cells. 9/n.

The added selectivity of our context-specific deaminases helps pinpoint editing within a 4–10 nt window, providing a unique solution to the longstanding challenge of bystander editing. Parallel efforts on adenine base editing are underway - stay tuned! 8/n.

Finally, we showcased the utility of context-specific base editing by installing two cancer driver mutations, KRAS^G12D and TP53^R248Q, each demanding selective editing of one C in consecutive Cs, which is a persistent challenge for canonical editors. 7/n

To benchmark editor performance, we prepared a paired sgRNA-target library covering 10,099 ClinVar-documented T-to-C mutations. Our context-specific editors achieved bystander-free editing at many of these sites, unlike canonical editors that always edit across wide windows. 6/n

Encouraged by successful isolation of an ACG-specific deaminase, we designed additional selection sites and conducted hotspot-focused directed evolution. This effort yielded 15 more NCN-specific variants, completing a set of 16 deaminases that together cover all NCN contexts. 5/n

In this process, we found many TadACs acquired a preference for AC substrates, likely shaped by the ACG selection site. We suspected, and confirmed, that context specificity at −1/+1 can be reprogrammed by evolution, leading to TadAC4.1 with clear −1 A, +1 G preference. 4/n

Through site-saturation mutagenesis of loop 7, we accessed multipotent TadA variants that deaminate both A and C (TadAC1s). DNA shuffling produced TadAC2s, which predominantly target C. Further evolution at loop 3 yielded TadAC3s that fully reject A. 3/n

We developed 16 TadA-derived NCN-specific deaminases that together cover every possible −1 and +1 context for a target cytosine. These variants support editor customization to prioritize target C-to-T conversion while minimizing bystander and off-target editing. 2/n.

RT @NatureBiotech: High-precision cytosine base editors by evolving nucleic-acid-recognition hotspots in deaminase .

RT @NatureBiotech: CRISPR editors gain precision via evolved hotspots #NBTintheNews via @CrisprMedicine

If you are interested in the method but lack the biochemical expertise to purify the enzyme—just reach out! We are here to help! 8/n.

Lastly, we are committed to making eTAM-seq accessible to the research community. We have shared enzyme aliquots with many U.S. labs and have helped others prepare eTAM-seq libraries when there is shared research interest. 7/n.

Also, a shout-out to the brilliant recent developments in chemical deamination-based m⁶A detection methods, by Chuan he's lab and Chengqi Yi's lab: We are excited to see the field moving forward on multiple fronts! 6/n.

Using eTAM-seq-v2, we mapped the m⁶A landscape in six human cell lines and seven embryonic mouse tissues. Beyond global patterns, eTAM-seq-v2 reveals that most neighboring m⁶A sites are independently deposited. 5/n.

In terms of sensitivity, eTAM-seq-v2 delivers robust performance with 10 ng of total RNA (~500 cells), and we believe we have not reached the detection limit. 4/n.

Like other enzyme-assisted sequencing approaches, eTAM-seq-v2 preserves RNA integrity. With just 60 million uniquely mapped reads, we can effectively survey >50% of A sites (and hence m⁶A) in all expressed genes. 3/n.

In addition to replacing the enzyme, we streamlined the workflow by integrating enzymatic treatment into a commercial library prep kit that uses template switching instead of ligation. This standardizes the protocol and cuts processing time from 3 days to just 8 hours. 2/n.