Inserting molecular marker sequences in living plant cells for protein labeling is a great challenge in functional genomic research. We established a simple and easy-to-use tag insertion method using an ePE2 system and Genome editing with Reverse transcription templates (RTTs) partially Aligned to each other but Nonhomologous to target sequences within Duo pegRNA (GRAND) strategy (Li et al., 2023; Wang et al., 2022). Because the insertion efficiency of longer tags, such as 66-bp 3×FALG, remains insufficient (Li et al., 2023), sequential optimizations are urgently needed.

Recently, a series of mammalian PE6 prime editors have been developed for improving efficiency via phage-assisted evolution and rational design of different reverse transcriptases (RTs) (Doman et al., 2023). Among them, PE6c and PE6d exhibited favourable activities at most application scenarios (Doman et al., 2023). To construct corresponding plant tools, the RNaseH-truncated evolved M-MLV RT of ePE2 was replaced with the Schizosaccharomyces pombe Tf1 retrotransposon RT variant to form ePE6c or was introduced the T128N/N200C/V223Y mutations to generate ePE6d (Figure 1a; Supplemental Materials and Methods). To test ePE6s in plants, three epegRNAs were designed for installing small mutations, including a T insertion, a G-to-A substitution and a TGTG insertion, in the rice Pid3, Pik-h and TB1 genes, respectively. After Agrobacterium-mediated stable transformation, editing efficiencies were determined in calli using amplicon next-generation sequencing (NGS). On average, 7.08% and 19.82% of the reads were precisely edited by ePE6c and ePE6d, respectively, showing that both editors are active in rice. In HEK293T cells, the efficiencies of PE6c and PE6d were similar to those of PEmaxΔRNaseH for the installation of point mutations (Doman et al., 2023). However, side-by-side comparisons showed that the efficiencies of ePE6c were 1.83- to 10.12-fold lower than those of ePE2 (P < 0.05, Figure 1b). In contrast, a significant decrease in precise edits by ePE6d was not observed throughout the targets (Figure 1b), while the ratios of the pegRNA scaffold-derived byproducts of ePE6d were 11.62- to 580.98-fold greater than those of ePE2 at the three sites (P < 0.05, Figures S1 and S2). Recent advances indicated that pegRNA scaffold-derived byproducts could be alleviated by modifying the stem structure of pegRNA (Shuto et al., 2024). In this case, we presumed that ePE6d would be as compatible as ePE2 for small edits in plants after further epegRNA optimization.

Given that most RT mutations in PE6c and PE6d have evolved for the installation of longer edits (Doman et al., 2023), epegRNAs were designed to insert a 27-bp sequence of the HA tag into the 3′ end of the OsERF141 (LOC_Os02g42585), OsHLH109 (LOC_Os01g64780) and OsTubA1 (LOC_Os03g51600) genes (Figure 1c). Amplicon-NGS showed that the average HA insertion efficiency of ePE6c was 6.87-fold lower than that of ePE2 (Figure 1d). Together with the above-described lower efficiencies of ePE6c for point mutations (Figure 1b), our results suggested that Tf1 RT may be less effective in plants under the current PE architecture. On the other hand, ePE6d generated precise insertions with efficiencies of 15.56%, 14.10% and 32.15% at the ERF141-T, HLH109-T and TubA1-T sites, respectively, with values that were 8.36-, 2.72- and 2.65-fold greater than those of ePE2 (P < 0.05, Figure 1d). The main types of ePE6d byproducts were incomplete insertions followed by short replicates of flanking genome sequences (Figure S3). Along with the enhancement of precise editing activity, the byproduct efficiencies of ePE6d remained at the same level as those of ePE2 at HLH109-T and TubA1-T and were slightly increased at ERF141-T. The insertion of a 30-bp c-MYC tag was further examined at the TubA1-T site. Precise c-MYC edits were obtained with insignificant different efficiencies by ePE2 and ePE6d (P > 0.05, Figure S4), suggesting that ePE6d may be less adaptable for improving the editing of c-MYC than HA at the same genomic target. It has been reported that the lower structural stability of the pegRNA sequence disrupts PE6d activity (Doman et al., 2023). Intriguingly, NUPACK (Fornace et al., 2022) prediction demonstrated that the secondary structure of the pegRNA 3′ extension for c-MYC insertion was more disordered than that for HA (Figure S5), providing a potential clue regarding the behaviour of ePE6d in c-MYC tag editing. In addition, the tagging vectors were retransformed to assess the activity in the transgenic plants. Precise insertions were induced by ePE6d in 20.83% to 70.83% of the T₀ lines, which is superior to 4.17% to 18.75% of ePE2 (Figure 1e,f; Table S1). Consistent with the calli results, byproducts occurred with a comparable frequency in 10.94% of ePE2 and 11.46% of ePE6d transgenic plants. Collectively, our data showed that ePE6d outperforms ePE2 for tag insertions in plants.

Next, duo epegRNAs with a 10-bp RTT overlap were designed to conduct GRAND editing to insert a relatively longer 78-bp calmodulin-binding peptide (CBP) tag and 90-bp 3×c-MYC tag (Figure 1g). NGS of the callus samples revealed that the CBP insertion of ePE6d in ERF141-T, HLH109-T and TubA1-T increased by 2.26-, 5.65- and 7.56-fold, respectively, compared with that of ePE2 (P < 0.05, Figure 1h). For 3×c-MYC, ePE6d had a 4.47-fold improvement in overall efficiency, achieving a maximum editing ratio of 33.84% at ERF141-T. Moreover, ePE6d offered higher edit:byproduct ratios in five out of the six editing than ePE2 (P < 0.05, Figure S6), suggesting a favourable profile for protein tagging. Further validation of the transgenic plants indicated that the average percentage of tagged lines increased from 7.64% of ePE2 to 50% of ePE6d for CBP and from 13.89% of ePE2 to 56.94% of ePE6d for 3×c-MYC (Figures 1i,j and S7; Table S2). Precise editing occurred maximally in 81.25% of the T₀ lines for CBP at TubA1-T and in 66.67% of the T₀ lines for 3×c-MYC at ERF141-T, indicating the outstanding tagging activity of ePE6d. Our previous work showed that ePE2 tagging of 3×FLAG was largely disrupted by incomplete insertions (Li et al., 2023). Although partial sequences remained the main unintended edits of ePE6d (Figure S8), no significant increase in byproducts was observed in the examined cases except for the CBP insertion at ERF141-T. To further demonstrate the compatibility of ePE6d, three additional tags were tested at the TubA1-T sites. The precise insertion of an 81-bp 3×HA tag and 135-bp 3×AVI tag was obtained in 45.83% and 18.75% of the transgenic lines, respectively (Figure S7; Table S2). However, we did not observe the insertion of the intact 171-bp GB1 tag in the plants, suggesting that the ePE6d knock-in ability might sharply reduce with size increasing of tags.

Overall, we showed the robust ability of ePE6d to install small edits and to insert approximately 100 bp of epitope tags in rice genome, among which some in situ protein labeling was validated using western blotting (Figure S9). We believe that ePE6d is an efficient, versatile and reliable tool for plant gene tagging as well as multiple types of genetic manipulations.

在活植物细胞中插入分子标记序列进行蛋白质标记是功能基因组研究中的巨大挑战。我们使用 ePE2 系统和 Genome 编辑建立了一种简单易用的标签插入方法，其中 Reverse 转录模板 （RTT） 部分 A彼此相连，但 N与 Duo pegRNA （GRAND） 策略中的目标序列同源（Li等人 ，2023 年;Wang et al.， 2022）。由于较长标签（如 66-bp 3×FALG）的插入效率仍然不足 （Li et al.， 2023），因此迫切需要顺序优化。

最近，已经开发了一系列哺乳动物 PE6 引物编辑器，用于通过噬菌体辅助进化和不同逆转录酶 （RT） 的合理设计来提高效率 （Doman等 人，2023 年）。其中，PE6c 和 PE6d 在大多数应用场景中表现出有利的活动 （Doman et al.， 2023）。为了构建相应的植物工具，将 ePE2 的 RNaseH 截短进化的 M-MLV RT 替换为粟酒裂殖酵母 Tf1 反转录转座子 RT 变体以形成 ePE6c，或引入 T128N/N200C/V223Y 突变以产生 ePE6d（图 1a;补充材料和方法）。为了在植物中测试 ePE6s，设计了三种 epegRNA，分别在水稻 Pid3、Pik-h 和 TB1 基因中植入小突变，包括 T 插入、G-to-A 替换和 TGTG 插入。农杆菌介导的稳定转化后，使用扩增子下一代测序 （NGS） 测定愈伤组织中的编辑效率。平均而言，7.08% 和 19.82% 的读数分别被 ePE6c 和 ePE6d 精确编辑，表明这两个编辑者在水稻中都很活跃。在 HEK293T 细胞中，PE6c 和 PE6d 的效率与 PEmaxΔRNaseH 在安装点突变方面的效率相似（Doman等人 ，2023 年）。然而，并排比较表明，ePE6c 的效率比 ePE2 低 1.83 至 10.12 倍（P < 0.05，图 1b）。相比之下，在整个靶标中未观察到 ePE6d 精确编辑的显着降低（图 1b），而 ePE6d 的 pegRNA 支架衍生副产物的比率为 11.62 至 580。在三个位点比 ePE2 高 98 倍（P < 0.05，图 S1 和 S2）。最近的进展表明，可以通过修饰 pegRNA 的茎结构来缓解 pegRNA 支架衍生的副产物（Shuto等人 ，2024 年）。在这种情况下，我们假设在进一步优化 epegRNA 后，ePE6d 与 ePE2 一样兼容植物中的小编辑。

鉴于 PE6c 和 PE6d 中的大多数 RT 突变已经进化为安装更长的编辑（Doman等 人，2023 年），epegRNA 被设计为将 HA 标签的 27 bp 序列插入 OsERF141 （LOC_Os02g42585）、OsHLH109 （LOC_Os01g64780） 和 OsTubA1 （LOC_Os03g51600） 基因（图 1c）。扩增子-NGS 显示，ePE6c 的平均 HA 插入效率比 ePE2 低 6.87 倍（图 1d）。结合上述 ePE6c 对点突变的较低效率（图 1b），我们的结果表明，在当前 PE 结构下，Tf1 RT 在植物中可能不太有效。另一方面，ePE6d 在 ERF141-T、HLH109-T 和 TubA1-T 位点产生精确插入，效率分别为 15.56%、14.10% 和 32.15%，值比 ePE2 高 8.36 倍、2.72 倍和 2.65 倍（P < 0.05，图 1d）。ePE6d 副产物的主要类型是不完全插入，然后是侧翼基因组序列的短重复（图 S3）。随着精确编辑活性的增强，ePE6d 的副产物效率与 ePE2 在 HLH109-T 和 TubA1-T 中的副产物效率保持在同一水平，在 ERF141-T 时略有提高。在 TubA1-T 位点进一步检查 30 bp c-MYC 标签的插入。ePE2 和 ePE6d 获得精确的 c-MYC 编辑，效率差异不显著 （P > 0.05，图 S4），表明 ePE6d 在改善 c-MYC 编辑方面的适应性可能低于 HA 在相同基因组靶标上。据报道，pegRNA 序列的较低结构稳定性会破坏 PE6d 活性（Doman等人 ，2023 年）。 有趣的是，NUPACK（Fornace等 人，2022 年）预测表明，用于 c-MYC 插入的 pegRNA 3' 延伸的二级结构比 HA 的二级结构更无序（图 S5），为 ePE6d 在 c-MYC 标签编辑中的行为提供了潜在线索。此外，重新转化标记载体以评估转基因植物中的活性。ePE6d 在 20.83% 至 70.83% 的 T₀ 细胞系中诱导精确插入，优于 ePE2 的 4.17% 至 18.75%（图 1e，f;表 S1）。与愈伤组织结果一致，副产物在 10.94% 的 ePE2 和 11.46% 的 ePE6d 转基因植株中以相当的频率出现。总的来说，我们的数据表明，ePE6d 在植物中插入标签的性能优于 ePE2。

接下来，设计具有 10 bp RTT 重叠的二重奏 epegRNAs 进行 GRAND 编辑，以插入相对较长的 78 bp 钙调蛋白结合肽 （CBP） 标签和 90 bp 3×c-MYC 标签（图 1g）。愈伤组织样品的 NGS 显示，与 ePE2 相比，ePE6d 在 ERF141-T、HLH109-T 和 TubA1-T 中的 CBP 插入分别增加了 2.26 倍、5.65 倍和 7.56 倍（P < 0.05，图 1h）。对于 3×c-MYC，ePE6d 的整体效率提高了 4.47 倍，在 ERF141-T 下实现了 33.84% 的最大编辑率。此外，ePE6d 在 6 次编辑中有 5 次编辑中提供了比 ePE2 更高的编辑：副产物比 （P < 0.05，图 S6），表明蛋白质标记具有良好的特征。对转基因植物的进一步验证表明，对于 CBP，标记品系的平均百分比从 ePE2 的 7.64% 增加到 ePE6d 的 50%，对于 3×c-MYC，标记品系的平均百分比从 ePE2 的 13.89% 增加到 ePE6d 的 56.94%（图 1i、j 和 S7;表 S2）。在 TubA1-T 处，81.25% 的 CBP T₀ 行和 ERF141-T 处 3×c-MYC 的 T₀ 行中，精确编辑发生率最高，表明 ePE6d 的标记活性突出。我们之前的工作表明，3×FLAG 的 ePE2 标记在很大程度上被不完全插入所破坏（Li等 人，2023 年）。尽管部分序列仍然是 ePE6d 的主要意外编辑（图 S8），但除了在 ERF141-T 处插入 CBP 外，在检查的病例中没有观察到副产物的显着增加。为了进一步证明 ePE6d 的兼容性，在 TubA1-T 位点测试了三个额外的标签。在 45.83% 和 18.75% 的转基因品系中分别获得了 81 bp 3×HA 标签和 135 bp 3×AVI 标签的精确插入（图 S7;表 S2）。 然而，我们没有观察到完整的 171 bp GB1 标签在植物中的插入，这表明 ePE6d 敲入能力可能会随着标签尺寸的增加而急剧降低。

总体而言，我们展示了 ePE6d 在水稻基因组中安装小编辑和插入约 100 bp 表位标签的强大能力，其中一些原位蛋白质标记使用蛋白质印迹进行了验证（图 S9）。我们相信 ePE6d 是一种高效、通用且可靠的植物基因标记以及多种类型的遗传操作工具。