Editing of the plastid genome helps understand the molecular functions of plastid genes and engineer desired traits in crops (Maliga, 2022). The DddA-derived cytosine base editors (DdCBEs) enable C-to-T editing in mitochondrial and plastid genomes (Kang et al., 2021; Li et al., 2021; Mok et al., 2020; Nakazato et al., 2021). Recently, Cho et al. (2022) developed TALE-linked deaminases (TALED) that can catalyse A-to-G base conversions in human mitochondria.

Harnessing the discovery of DddA_tox (Cho et al., 2022), we generated new monomeric TALE-linked CBEs for plastid editing by exploring two cytidine deaminases: a human APOBEC3A variant (hA3A-Y130F) with broad editing windows (Ren et al., 2021) and an improved cytidine deaminase based on TadA (Lam et al., 2023), generating mTCBE and mTCBE-T respectively. Also, we explored a TadA-derived deaminase that can simultaneously deaminate cytosines and adenines (Lam et al., 2023) towards engineering a dual base editor, named mTCABE-T. None of these deaminases have been previously investigated for organellar genome editing in either plants or humans.

We first assembled the left or right TALE arrays targeted to three rice plastid genes encoding core components of photosystem II (OsPsbA), photosystem I (OsPsaA), and RNA component of the 30S ribosome subunit (Os16SrRNA). The three monomeric plastid base editors, along with DdCBE and Split-TALED controls, were constructed for expression in rice (Figure 1a). We evaluated base editing efficiency in regenerated rice calli through targeted amplicon deep sequencing. Impressively, mTCBE induced high-efficiency C-to-T conversions, with average editing frequencies of 42.3%, 21.6%, and 19.4% at OsPsbA, OsPsaA, and Os16SrRNA respectively (Figure 1b–d). DdCBE catalysed C-to-T conversions with average editing efficiencies of 7.8%, 33.5% and 34.2% at these target sites (Figure 1b–d). By contrast, mTCBE-T was less efficient than mTCBE, displaying C-to-T editing efficiencies of only 5.8%, 3.3%, and 0.8% in these targets (Figure 1b–d). Overall, mTCBE appears to be a robust base editor alongside DdCBE (Figure 1e).

Remarkably, mTCABE-T exhibited C-to-T substitutions with average efficiencies of 33.2%, 21.3%, and 5.0% as well as A-to-G substitutions with average efficiencies of 17.5%, 7.2%, and 7.8%. The Split-TALED induced C-to-T editing with average efficiencies of 2.7%, 15.0%, and 8.9% as well as A-to-G editing with average efficiencies of 12.9%, 4.8%, and 3.3% in the same targets (Figure 1b–d), consistent with Mok et al. (2022) report, which noted that UGI-free split can induce C-to-T and A-to-G edits. mTCABE-T is a better dual base editor with higher C-to-T and A-to-G editing efficiencies than Split-TALED (Figure 1e,f). Further sequence analysis confirmed frequent simultaneous C-to-T and A-to-G editing by six mTCABE-T editors at the three target loci (Figure 1g, Figure S1).

We further examined plastid base editing profiles by different editors. As expected, DdCBE favoured editing cytosines of a 5′-TC context within spacer regions (Figure 1h, Figure S2). Notably, mTCBE often generated distinct editing patterns and efficiently converted cytosines in a 5′-TCC, AC, and GC context at the OsPsbA target (Figure 1h). Similarly, at OsPsaA, mTCBE converted cytosines in TC and TCC context (Figure S2). At Os16SrRNA, four cytosines were converted to T (Figure S2). Although editing frequently happened in the spacer regions, both mTCBE and mTCABE-T also showed base editing at positions upstream or downstream of the spacer region, resulting in only partly overlapping editing profiles by the left and right TALEs (Figure 1h,i, Figure S2). Our data suggest that these monomeric base editors have broader editing ranges than DdCBEs.

We next investigated these monomeric TALE-linked base editors (mTBEs) in T₀ transgenic plants. At OsPsbA, targeted deep sequencing revealed that C-to-T base editing occurred at multiple positions, with editing frequencies reaching up to 81% and 86% based on 16 and 26 T₀ plants by mTCBE and mTCABE-T respectively (Figure 1j). Significantly, such high-frequency targeted mutagenesis by mTCBE and mTCABE-T translated to a loss-of-function albino phenotype (Figure 1k). Hence, we genetically confirmed the role of OsPsbA in photosynthesis in rice. Consistent with the calli data, mTCBE-T exhibited much lower efficiency, and none of the 29 T₀ seedlings plants showed an albino phenotype (Figure 1k). Our data suggest that high-efficiency editing of plastid genome copies at multiple editable positions could effectively knockout gene function (Figure 1l, Figure S3). Sanger sequencing showed consistent mutation types in different leaves of each edited plant, including homoplasmic edits, which suggests high probability of germ-line transmittable inheritance (Figure S4). At Os16SrRNA, C-to-T editing efficiency of up to 80% was also observed in T₀ plants with mTCBE and mTCABE-T (Figure S5). Together, our data demonstrates mTCBE and mTCABE-T are potent plastid base editors in transgenic rice plants.

We analysed off-target activity of mTCBE and mTCABE-T at the OsPsbA site. Off-target mutations were induced at two of the four top candidate sites, only in few edited lines (Figure S6). Thus, sequences of TALE binding sites and expression of the editors can both affect the off-target editing outcomes.

In summary, we demonstrated two novel mTBEs, mTCBE, and mTCABE-T, for efficient plastid genome editing in rice. These mTBEs offer several advantages over DdCBEs. First, use of single TALE protein streamlines the vector construction process. Second, these mTBEs have broad editing windows. Third, mTCABE-T is an efficient dual base editor. Since mTCABE-T edits both strands of DNA, it can simultaneously induce four different types of base conversions on a single strand such as a coding sequence: C-to-T, A-to-G, T-to-C, and G-to-A. Thus, mTCABE-T in theory could potentially edit every base within its editing window. Hence, mTCABE-T would be a powerful high-density-targeted mutagenesis tool for protein evolution. Although we only demonstrated mTCABE-T in plastids, it is anticipated that mTCABE-T, when equipped with mitochondrial localization signals, could enable base editing in mitochondria in plants and beyond.

质体基因组的编辑有助于了解质体基因的分子功能并设计作物所需的性状（Maliga，  2022）。 DddA 衍生的胞嘧啶碱基编辑器 (DdCBE) 能够在线粒体和质体基因组中进行 C 到 T 编辑（Kang等人，  2021；Li等人，  2021；Mok等人，  2020；Nakazato等人，  2021 ） ）。最近，Cho等人。 ( 2022 ) 开发了 TALE 相关脱氨酶 (TALED)，可以催化人类线粒体中的 A 到 G 碱基转化。

_{利用 DddA tox}的发现（Cho等人，  2022），我们通过探索两种胞苷脱氨酶生成了用于质体编辑的新单体 TALE 连接 CBE：具有广泛编辑窗口的人 APOBEC3A 变体（hA3A-Y130F）（Ren等人，2022）。 ，  2021）和基于TadA的改进的胞苷脱氨酶（Lam等，  2023），分别生成mTCBE和mTCBE-T。此外，我们还探索了一种源自 TadA 的脱氨酶，它可以同时使胞嘧啶和腺嘌呤脱氨（Lam等人，  2023），以设计双碱基编辑器，命名为 mTCABE-T。这些脱氨酶之前都没有被研究过用于植物或人类的细胞器基因组编辑。

我们首先组装了针对三个水稻质体基因的左侧或右侧 TALE 阵列，这些基因编码光系统 II ( OsPsbA )、光系统 I ( OsPsaA ) 的核心组件和 30S 核糖体亚基的 RNA 组件 ( Os16SrRNA )。构建了三个单体质体碱基编辑器以及 DdCBE 和 Split-TALED 对照，用于在水稻中表达（图 1a）。我们通过靶向扩增子深度测序评估了再生水稻愈伤组织的碱基编辑效率。令人印象深刻的是，mTCBE 诱导了高效的 C-T 转换，OsPsbA、OsPsaA和Os16SrRNA的平均编辑频率分别为 42.3%、21.6% 和 19.4%（图 1b-d）。 DdCBE 催化 C 到 T 的转化，在这些目标位点的平均编辑效率分别为 7.8%、33.5% 和 34.2%（图 1b-d）。相比之下，mTCBE-T 的效率低于 mTCBE，在这些目标中显示 C-to-T 编辑效率仅为 5.8%、3.3% 和 0.8%（图 1b-d）。总体而言，mTCBE 似乎是与 DdCBE 一样强大的碱基编辑器（图 1e）。

值得注意的是，mTCABE-T 的 C 至 T 替代平均效率分别为 33.2%、21.3% 和 5.0%，A 至 G 替代平均效率分别为 17.5%、7.2% 和 7.8%。 Split-TALED 在相同目标中诱导 C-to-T 编辑的平均效率分别为 2.7%、15.0% 和 8.9%，以及 A-to-G 编辑的平均效率为 12.9%、4.8% 和 3.3% （图 1b-d），与 Mok等人的观点一致。 ( 2022 ) 报告指出，无 UGI 的拆分可以诱导 C 到 T 和 A 到 G 编辑。 mTCABE-T 是一种更好的双碱基编辑器，比 Split-TALED 具有更高的 C 到 T 和 A 到 G 编辑效率（图 1e、f）。进一步的序列分析证实了六个 mTCABE-T 编辑器在三个目标位点频繁同时进行 C 到 T 和 A 到 G 编辑（图 1g，图 S1）。

我们进一步检查了不同编辑器的质体碱基编辑配置文件。正如预期的那样，DdCBE 有利于编辑间隔区域内 5'-TC 上下文的胞嘧啶（图 1h，图 S2）。值得注意的是，mTCBE 通常会在OsPsbA靶点的 5'-TCC、AC 和 GC 环境中生成不同的编辑模式并有效转换胞嘧啶（图 1h）。同样，在OsPsaA中，mTCBE 转换了 TC 和 TCC 中的胞嘧啶（图 S2）。在Os16SrRNA中，四个胞嘧啶转化为 T（图 S2）。尽管编辑频繁发生在间隔区，但 mTCBE 和 mTCABE-T 也在间隔区上游或下游位置显示碱基编辑，导致左右 TALE 的编辑配置文件仅部分重叠（图 1h,i，图 S2） ）。我们的数据表明这些单体碱基编辑器比 DdCBE 具有更广泛的编辑范围。

_{接下来我们研究了 T 0}转基因植物中的这些单体 TALE 连接碱基编辑器 (mTBE) 。在OsPsbA ，靶向深度测序显示，C-to-T碱基编辑发生在多个位置，基于mTCBE和mTCABE-T的16和26个T ₀植物，编辑频率分别高达81%和86% （图1j） 。值得注意的是，mTCBE 和 mTCABE-T 的这种高频定向诱变转化为功能丧失的白化表型（图 1k）。因此，我们从遗传学角度证实了OsPsbA在水稻光合作用中的作用。与愈伤组织数据一致，mTCBE-T表现出低得多的效率，并且29株T ₀幼苗植物中没有一株表现出白化表型（图1k）。我们的数据表明，在多个可编辑位置对质体基因组拷贝进行高效编辑可以有效地敲除基因功能（图1l，图S3）。桑格测序显示每个编辑植物的不同叶子中存在一致的突变类型，包括同质编辑，这表明种系可传递遗传的可能性很高（图S4）。在Os16SrRNA中，在使用 mTCBE 和 mTCABE-T 的T ₀植物中也观察到高达 80% 的 C 到 T 编辑效率（图 S5）。总之，我们的数据表明 mTCBE 和 mTCABE-T 是转基因水稻植物中有效的质体碱基编辑器。

我们分析了 mTCBE 和 mTCABE-T 在OsPsbA位点的脱靶活性。脱靶突变在四个顶级候选位点中的两个被诱导，仅在少数编辑行中（图 S6）。因此，TALE 结合位点的序列和编辑器的表达都会影响脱靶编辑结果。

总之，我们展示了两种新型 mTBE：mTCBE 和 mTCABE-T，可用于水稻中有效的质体基因组编辑。与 DdCBE 相比，这些 mTBE 具有多项优势。首先，使用单一 TALE 蛋白简化了载体构建过程。其次，这些 mTBE 具有广泛的编辑窗口。第三，mTCABE-T是一种高效的双碱基编辑器。由于 mTCABE-T 编辑 DNA 的两条链，因此它可以同时在单链上诱导四种不同类型的碱基转换，例如编码序列：C 到 T、A 到 G、T 到 C 和 G -到-A。因此，mTCABE-T 理论上可以在其编辑窗口内编辑每个碱基。因此，mTCABE-T 将成为蛋白质进化的强大高密度靶向诱变工具。尽管我们仅在质体中证明了 mTCABE-T，但预计当配备线粒体定位信号时，mTCABE-T 可以实现植物及其他线粒体中的碱基编辑。