Highly efficient C-to-T and A-to-G base editing in a Populus hybrid.

CRISPR‐Cas‐based genome editing technologies hold great potential for genetic research and bioengineering in trees. CRISPR‐Cas9 was previously demonstrated to be a highly efficient genome editing system in poplar (Zhou et al., 2015). However, these applications only used the nuclease activity of Cas9 for targeted mutagenesis. By contrast, dead Cas9 (dCas9)‐derived base editors can install specific base changes in a plant genome. Cytosine base editors (CBEs) can introduce C‐to‐T and occasionally C‐to‐G base changes (Komor et al., 2016; Nishida et al., 2016) and adenine base editors (ABEs) can generate A‐to‐G base changes (Gaudelli et al., 2017). CBEs can be applied for constructing premature stop codons, while both CBEs and ABEs can be used for making amino acid codon changes or altering RNA splicing sites or cis‐regulatory elements in tailored applications. Despite their promising potential, base editing tools have not been fully established in trees.

To develop efficient base editors in trees, we worked on a P opulus hybrid (Populus tremula × P. alba hybrid clone INRA 717‐1B4). We first tried to establish an efficient C‐to‐T base editing system by comparing two promising CBEs, PmCDA1‐BE3 and A3A/Y130F‐BE3, which were in a base editor 3 (BE3) configuration by fusion of a cytidine deaminase and a uracil DNA glycosylase inhibitor (UGI) to the Cas9D10A nickase (Figure 1a). A PmCDA1 base editor in the BE3 configuration was previously demonstrated in rice (Tang et al., 2018). Recently, high‐efficiency base editing in rice, wheat and potato was demonstrated with A3A‐BE3 (Zong et al., 2018). Both PmCDA1‐BE3 and A3A‐BE3 showed much higher editing efficiencies than the widely used rAPOBEC1‐BE3 in plants (Tang et al., 2018; Zong et al., 2018). Further, introduction of the Y130F mutation to A3A‐BE3 appeared to be even more potent in making C‐to‐T conversions in human cells, especially at methylated target sites (Wang et al., 2018). Therefore, A3A/Y130F‐BE3 is a promising CBE for testing in plants.

Figure 1

C‐to‐T and A‐to‐G base editing in poplar. (a) Illustration of the C‐to‐T and A‐to‐G base editing systems. (b) Location and DNA sequence of sgRNAs and primers for Sanger sequencing, designed for both alleles of 4CL1 and PII genes for Populus tremula × P. alba. (c) Summary of C‐to‐T and A‐to‐G base editing efficiencies in poplar T0 lines for both 4CL1 and PII genes. (d) Base editing windows of base editors A3A/Y130F‐BE3 (blue line with circle) and PmCDA1‐BE3 (red line with square) at four sgRNA target sites. The number of cytosine (C) indicates its position in the protospacer. (e) Examples of T0 transgenic poplar with monoallelic and biallelic mutation by introducing a premature stop codon to 4CL1 via C‐to‐T conversion. Red arrows indicate introduction of stop codon by C‐to‐T conversion. Blue arrows and black arrows indicate biallelic and monoallelic editing bases, respectively. (f) Phenotype of the wild‐type (WT) and 4CL1 mutants, with green stem in WT and 4CL1 monoallelic mutants, and brown stem in 4CL1 biallelic mutants. (g) Examples of T0 transgenic poplar with A‐to‐G base editing by ABEmax_V1. Blue arrows and black arrows indicate biallelic and monoallelic editing bases, respectively. (h) Examples of T0 transgenic poplar with A‐to‐G base editing by ABEmax_V2. Blue arrows and black arrows indicate biallelic and monoallelic editing bases, respectively. (i) Base editing windows at the 4CL‐sgRNA6 target site by base editors ABEmax_V1 (green line with circle) and ABEmax_V2 (purple line with square). The number of adenine (A) indicates its position in the protospacer.

To compare PmCDA1‐BE3 and A3A/Y130F‐BE3, we targeted four sites in poplar with two sgRNAs editing 4‐coumarate: CoA ligase 1 (4CL1) (4CL1‐sgRNA1 and 4CL1‐sgRNA2) and the other two sgRNA editing PII (PII‐sgRNA1 and PII‐sgRNA2), which encodes an evolutionarily conserved protein involved in coordinating carbon and nitrogen assimilation (Figure 1b). All four sgRNAs were designed to target both P. alba and P. tremula genomes and were intended to introduce premature stop codons. To effectively assess the CBEs, we constructed multiplexed T‐DNA constructs where the two sgRNAs for editing the same gene were expressed under the AtU6 promoter and AtU3 promoter, respectively, and the Cas9D10A nickase‐UGI fusion gene was expressed under the AtUbi10 promoter (Figure 1a). The four T‐DNA vectors were constructed using our previously established multiplexed CRISPR assembly system (Lowder et al., 2015) and transformed into poplar petiole sections by Agrobacterium‐mediated transformation (Leple et al., 1992). Between 19 and 22 T0 lines were randomly selected for each construct and then genotyped by Sanger sequencing. At 4CL1 A3A/Y130F‐BE3 generated 50.0% and 95.5% editing frequencies with 4CL1‐sgRNA1 and 4CL1‐sgRNA2, respectively (Figure 1c). With PmCDA1‐BE3, 26.3% and 78.9% editing frequencies were obtained for these two target sites, lower than those by A3A/Y130F‐BE3 (Figure 1c). At both target sites, biallelic base editing lines were obtained with either A3A/Y130F‐BE3 or PmCDA1‐BE3 (Figure 1c). At PII, A3A/Y130F‐BE3 and PmCDA1‐BE3 generated 19.0% and 0% editing frequencies, respectively, with PII‐sgRNA1 driven by the AtU6 promoter (Figure 1c). By contrast, PII‐sgRNA2, driven by the AtU3 promoter, was of high efficiency as it resulted in 81.0% base editing frequency with A3A/Y130F‐BE3 and 100% base editing frequency with PmCDA1‐BE3 (Figure 1c). Altogether, we found both CBEs were highly efficient in generating targeted C‐to‐T base changes in poplar. Remarkably, we only identified insertion and deletion (indel) byproduct mutations at the 4CL1‐sgRNA1 and PII‐sgRNA2 target site (Figure 1c), suggesting high editing purity of both CBEs.

By combing the data from all four target sites in the T0 lines, it appeared that A3A/Y130F‐BE3 can edit a broader window from C5 to C18 on the protospacer (Figure 1d). By contrast, PmCDA1‐BE3’s editing window shifted slightly to the 5’ end of the protospacer, from C2 to C13 (Figure 1d). These observations were consistent with previous reports of editing windows for A3A/Y130F‐BE3 in human cells (Wang et al., 2018) and for PmCDA1‐BE3 in rice (Tang et al., 2018). We further conducted phenotypic investigation of four T0 lines edited by A3A/Y130F‐BE3, with two being monoallelic mutants and the other two being biallelic mutants (Figure 1e). Since the biallelic mutants carried premature stop codons in both alleles of P. alba and P. tremula, we expected to see loss‐of‐function phenotype. Indeed, both biallelic mutants, not the monoallelic mutants, showed brown stems (Figure 1f), which suggested reduced lignin accrual and altered monolignol composition as previously reported for the loss of function of 4CL1 in poplar (Zhou et al., 2015).

Since its first demonstration in human cells (Gaudelli et al., 2017), A‐to‐G base editor ABE7.10 was further improved to ABEmax based on expression optimization and ancestral reconstruction (Koblan et al., 2018). To develop an effective ABE in poplar, we compared two ABEmax systems. These systems were based on human codon‐optimized ecTadAwt‐ecTadA* adenine deaminase, and maize or human codon‐optimized Cas9D10A nickase, in ABEmax_V1 and ABEmax_V2, respectively (Figure 1a). Two targets sites at 4CL1, 4CL1‐sgRNA5 and 4CL1‐sgRNA6, were chosen for multiplexed A‐to‐G base editing (Figure 1a and b). Nineteen and 22 T0 lines were randomly selected for ABEmax_V1 and ABEmax‐V2, respectively. Neither ABEs generated edited events with 4CL1‐sgRNA5 which was driven by the AtU6 promoter (Figure 1c). However, with 4CL1‐sgRNA6 driven by the AtU3 promoter, ABEmax_V1 and ABEmax_V2 generated 84.2% and 95.5% editing frequencies (Figure 1c), respectively, and many of the events are biallelic editing (Figure 1g and h). Notably, no indel byproducts were found among all T0 lines (Figure 1c), suggesting high‐purity A‐to‐G conversions. Among seven adenines in the protospacer, A7 and A9 were both edited at high frequencies (Figure 1i). The data were in line with the general editing window from A4 to A10 for ABE systems (Gaudelli et al., 2017; Koblan et al., 2018).

In this study, we compared two CBEs and two ABEs in poplar for precise base editing. Based on the editing data from six independent multiplexed base editing constructs, the AtU3 promoter consistently yields much higher base editing frequencies (78.9%–100%) than the AtU6 promoter (0%–50.0%) (Figure 1a and c). Hence, with the AtU3 promoter for sgRNA expression, the CBEs and ABEs that we tested here should yield high‐efficiency base editing and will have promising applications for precise genome editing in poplar and other trees.

Conflicts of interests

The authors declare no competing financial interests.

Author contributions

Y.Q. and G.C. designed the study. Y.Q., G.C. and E.E. supervised the study. G.L. and S.S. performed the experiments and analysed the data. Y.Q., G.L, S.S., E. E. and G.C. wrote the paper.