Base Editing‐Mediated Targeted Evolution of PopulusCesA8 Boosts Cellulose Production

Improving wood properties is crucial for sustainable forestry (Zhu and Li 2024). Base editing (BE) results in nucleotide changes that could improve traits of importance to crop and tree improvement (Mishra et al. 2020; Wei et al. 2021; Yang et al. 2024). Although BEs have been developed in trees (Li et al. 2021; Yao et al. 2023), they are rarely applied in the targeted evolution of tree genes. Here, we report pABE8e, an adenine base editor with a hyperactive TadA8e (V106W) variant, which functions efficiently in Populus. We applied pABE8e to the targeted mutagenesis of Populus cellulose synthase A8 (PdCesA8), a key enzyme in wood cellulose synthesis. This approach identified a conserved S550G substitution in PdCesA8 that significantly increases crystalline cellulose content without impacting plant growth. This work demonstrates efficient base editing in Populus and reveals a promising target for enhancing tree biomass characteristics. We constructed two base editing vectors, pABE8e and pABEmax, using the deaminase TadA8e V106W and TadA-TadA7.10 developed by the David R. Liu lab (Figure 1a; Figure S1) (Koblan et al. 2018; Richter et al. 2020). We evaluated their A-to-G editing efficiency by targeting eight distinct sites in key cellulose biosynthesis genes, including PdCesA8a/b, PdKOR1a/b and PdCSI1a/b (Figure 1b) (Zhu and Li 2024). Pooled transformation yielded 91 T0 transgenic lines for pABE8e and 119 for pABEmax. Molecular analysis revealed that pABE8e consistently showed superior editing efficiency, with frequencies ranging from 5% (PdKOR1a) to 85.8% (PdCesA8b). In contrast, pABEmax generally exhibited lower editing efficiencies, from 3.7% (PdKOR1a) to 25.6% (PdCesA8b). For example, at the PdCesA8b target site, pABE8e's efficiency (85.8%) was over threefold higher than that of pABEmax (25.6%) (Figure 1b). We explored whether the sgRNA promoter influenced these outcomes; as previous work suggested, the AtU3 promoter was superior to AtU6 (Li et al. 2021). By swapping promoters (AtU6::PdCesA8a to AtU3::PdCesA8a and AtU3::PdCesA8b to AtU6::PdCesA8b), we found that editing efficiency was primarily dictated by the sgRNA target sequence itself, not the choice of promoter (Figure 1b). pABE8e showed high activity across protospacer nucleotides 4–9 (from the 5′ end), peaking at position 5 (A5) with relaxed PAM sequences (e.g., GGC, AGG, AGT, AGA) (Figure 1c,d). In addition, pABE8e could also edit multiple adenines simultaneously within this window (Figure S2). pABE8e lines showed frequent homozygous A-to-G substitutions at A5 for the efficient PdCesA8b target (40.9% with AtU6, 42.9% with AtU3) (Figure 1b). This was significantly higher than pABEmax, which yielded only 4.7% homozygous substitutions for the same target (Figure 1b,c). These results confirm pABE8e as an efficient, PAM-relaxed ABE, ideal for generating homozygous substitutions in one generation for directed protein evolution in Populus. Plant CesAs have three unique regions: the N-terminal domain (NTD), the plant-conserved region (PCR) and the class-specific region (CSR). These regions are likely essential for the assembly and function of the cellulose synthase complex. To generate targeted mutagenesis of CesA8, we designed 30 sgRNAs targeting 20 sites across these regions in both PdCesA8a and PdCesA8b, which includes 10 sgRNAs that simultaneously target both genes (Figure 1e; Figures S3 and S4). Following Agrobacterium-mediated transformation, genotyping of T0 lines revealed variable editing efficiencies, with eight sgRNAs showing high efficiency (T1, T2, T5, T6, T7, T13, T20, T30), 12 showing low efficiency and the remaining 10 showing no editing (Table S1). Our data suggest that the lower GC content and thymine enrichment at positions 18–20 of the sgRNA target sequence may impair editing success (Table S1; Figure S5), which is supported by a previous report (Hiranniramol et al. 2020). The editing process successfully generated novel CesA8 alleles with missense mutations, including S95G and S102G in the NTD, T393A and S393P in the PCR, S550G and T595A in the CSR (Figure 1e,f). Phenotyping of homozygous T0 mutants revealed distinct effects. For example, mutations T393A and S393P in the PCR domain significantly reduced plant height and cellulose content, leading to thin-walled fibres and deformed vessels (Figure 1g–k). In contrast, the S550G mutation in the CSR domain maintained normal growth while significantly increasing crystalline cellulose content and fibre wall thickness, highlighting S550 as a crucial target for manipulating cellulose biosynthesis (Figure 1g–k). The S550G alteration might regulate CesA complex assembly or activity, enhancing crystalline cellulose production. To investigate the evolutionary role of S550, we examined PdCesA8 orthologues in Ochroma pyramidale (balsa), Khaya senegalensis (African mahogany) and Tectona grandis (teak), species selected for their diverse wood properties and available genome sequences (Sahu et al. 2023). Phylogenetic analysis showed S550 to be serine/threonine (S/T) in balsa, but alanine (A) in teak and asparagine (N) in African mahogany (Figure 1l). This divergence is significant because teak and African mahogany have thicker secondary cell walls and higher crystalline cellulose content than balsa (Figure 1m,n). The correlation strongly suggested that S550 has been a target of natural selection, shaping wood characteristics. In summary, we developed pABE8e, an efficient base editor that enables direct T0 mutagenesis in Populus. Through pABE8e-mediated targeted mutagenesis, we identified that the S550G substitution in PdCesA8b enhanced cellulose production without hindering growth. The evolutionary conservation and functional significance of this site across tree species indicate that S550G is a promising target for precision breeding to improve wood quality. This study demonstrated that base editing-mediated targeted evolution is a powerful strategy for rapidly enhancing complex traits in trees. J.G. designed and supervised the study. P.Z., M.Z., Y.Z. and X.J. performed experiments and analysed data. J.G. wrote the paper. We thank Dr. Yaoguang Liu for providing the pYLCRISPR/Cas9 vector, Dr. Hui Zhang for providing the rABE8e and ABEmax vector and Dr. Sunil Kumar Sahu for providing the three genome sequences. This work was supported by the Science and Technology Innovation 2030-Major Project (2023ZD04057), the National Natural Science Foundation of China (32022055) and the Zhejiang A&F University Starting Funding (2021FR026). The data that supports the findings of this study are available in the Supporting Information of this article. Figures S1–S5 . Tables S1–S2 . Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.