Geminiviruses are a family of plant viruses with circular single‐stranded DNA genomes. They have been deconstructed by researchers for multiple biotechnological applications, including protein expression, gene silencing and genome editing, in plants (Lozano‐Durán, 2016). Under the deconstructed virus strategy, the coat protein and movement protein genes were removed from the geminiviruses, while the sequences required for replication were retained. The replicons replicate after delivery to plant cells and increase the copy number of carried DNA; this leads to high levels of gene expression (Lozano‐Durán, 2016). Although a few geminiviral replicon‐based vectors have been used in gene targeting (Baltes et al., 2014; Cermak et al., 2015; Wang et al., 2017), the list of DNA replicon‐based vectors is still limited in plants, especially for food crops. In this study, we developed sweet potato leaf curl virus (SPLCV) replicon‐based expression vectors. We tested the efficiency of these vectors in CRISPR‐Cas‐mediated RNA targeting and gene editing by using Nicotiana benthamiana as model plant.

SPLCV is a monopartite geminivirus belonging to the genus Begomoviruses (Bi and Zhang, 2012). The coding region of SPLCV‐JS (accession number: KF040468.1) replication‐associated proteins (Rep:1900 bp, encoding four proteins: AC1, AC2, AC3 and AC4) and intergenic region (IR:284 bp) were synthesized and cloned into the binary vector pCambia0390 together with GFP expression cassette (U4:GFP) in an IR‐GFP‐Rep‐IR origination to produce the reporter vector SPLCV‐GFP (Figure 1a). A regular T‐DNA vector (T‐GFP) harbouring the same expression cassette was used as control. N. benthamiana leaves infiltrated with Agrobacterium tumefaciens containing SPLCV‐GFP significantly showed stronger GFP fluorescence than T‐GFP (Figure 1b). We confirmed the circularization between the two IRs by PCR with a prime pair facing opposite directions in the SPLCV‐GFP‐infiltrated leaves (Figure 1c). The GFP transcripts in SPLCV‐GFP‐infiltrated leaves were 22.5 times higher than those in T‐GFP (Figure 1d).

After several days of infiltration, evident necrosis was found in most SPLCV‐GFP‐infiltrated leaves. To decrease cell lethality, three mutational SPLCV vectors were constructed: (i) AC4^m: a T‐to‐A mutation was introduced in the coding region of Rep, thereby resulting in a premature stop of translation of AC4 (AC4:26T→A(Leu9TAA); AC1:183T→A(Leu61Leu)); (ii) AC2^m: a premature termination codon mutation was introduced in the coding region of AC2, and this mutation changed one amino acid of AC1 (AC2:31A→T(Lys11TAG); AC1:1034A→T(Glu345Val)); and (iii) AC2^m/AC4^m double mutant. All constructs were sequenced to confirm the correct mutation sites. The GFP expression cassette was cloned into the three mutational vectors, and the expression efficiency was compared with that of SPLCV‐GFP (WT). No visible decrease in GFP fluorescence was observed in N. benthamiana leaves infiltrated with A. tumefaciens containing these mutational constructs at days 3 and 9 (Figure 1e). The GFP transcripts exhibited no significant difference between the WT and AC2^m/AC4^m vectors (Figure 1f). These mutational constructs did not induce necrosis at day 12 (Figure 1g). We then tested the RNA‐silencing suppressor (RSS) activity of these basic viral vectors through a GFP silencing experiment in N. benthamiana (16C) leaves. Figure 1h shows that GFP silencing was only observed in empty and AC2^m/AC4^m vectors. This result implied that the AC2^m/AC4^m construct lost RSS activity. Insertion of up to 9 kb noncoding DNA sequence in AC2^m/AC4^m still resulted in strong GFP expression (Figure 1i). This result suggested that the cargo capacity of SPLCV was sufficient to deliver CRISPR nucleases.

We then tested whether the SPLCV vector can enhance LwaCas13a‐mediated RNA targeting activity in plants (Abudayyeh et al., 2017). The AC2^m/AC4^m construct was used for this series of experiments. We artificially synthesized the DNA sequence of plant codon‐optimized LwaCas13a and cloned into AC2^m/AC4^m and regular vectors with 3×Flag fusion on the C terminus and a dual‐flanking nuclear localization sequence (NLS) under the expression of the U4 promotor. Moreover, we designed two gRNAs against the mGFP5 transcript and recombined the targeting vectors (SPLCV‐based vector: SV; T‐DNA vector: TV; Figure 1j). The same vectors containing a non‐targeting (NT) spacer were used as control. After 4 days of infiltration, a higher accumulation of LwaCas13a and gRNA transcripts and the LwaCas13a protein was found in SV‐infiltrated leaves than in TV‐infiltrated ones (Figure 1k and l). In comparison with the left side of leaves infiltrated with mGFP5/NT constructs, a visible decline in GFP fluorescence was observed in the corresponding right side infiltrated with SVs at day 4. This phenomenon was not apparent in TVs (Figure 1m). The SVs resulted in significantly higher levels of knockdown of mGFP5 (67% for gRNA1, 70% for gRNA2) than TVs (25% for gRNA1, 32% for gRNA2; Figure 1n). The SVs‐infiltrated samples accumulated less mGFP5 protein than TVs (Figure 1o). In addition, we obtained the similar results when targeting an endogenous transcript NbPDS1 (Figure 1p). These results clearly showed that SPLCV‐based RNA‐targeting vectors were more efficient than regular ones.

We further tested whether SPLCV vector can enhance SpCas9‐ and LbCas12a‐mediated gene editing in plants. The AC4^m‐derived constructs could be efficient for gene editing in plants due to the loss of cell lethality but maintained RSS activity (Figure 1h; Mao et al., 2018). Thus, the AC4^m vector was used in this series of experiments. We cloned plant codon‐optimized SpCas9 (with NLS fusion on the C terminus) and LbCas12a (with 3×HA and NLS fusion on the C terminus) into AC4^m and regular vectors under the expression of the U4 promotor. We designed several crRNAs against NbPDS1 and mGFP5 in wild type and 16C N. benthamiana, respectively, and recombined the SVs and TVs (Figure 1q). The same vectors containing NT crRNA were used as control. After 6 days of infiltration, a higher accumulation of nucleases, crRNA transcripts and the nuclease protein was found in SV‐infiltrated leaves than in TV‐infiltrated ones (Figures 1r and s). The average relative amplification values of the target sites obtained using the qPCR‐based method were approximately 1.01 and 0.90 in NT‐infiltrated and TV‐infiltrated samples, respectively (Figures 1t and u; Peng et al., 2018). Such a notion suggests that the mutation frequency for TVs was approximately 10%, and this result was similar to a previous report (Bernabé‐Orts et al., 2019). However, the average relative amplification values of five target sites were approximately 0.65 in SV‐infiltrated samples (Figures 1t and u). Thus, the mutation frequencies for SVs were 35% and much higher than those of TVs.

In summary, our work expands the existing list of geminiviral replicon‐based vectors in plants. The SPLCV‐based vectors offer an opportunity to enhance the efficiency of CRISPR‐Cas‐mediated gene knockdown, knockout and knockin in plants, especially for its host sweet potato; gene editing through de novo induction of meristems may also be conducted in this important hexaploid crop in the near future (Maher et al., 2020).

双生病毒是一类具有环状单链 DNA 基因组的植物病毒。研究人员已将它们解构用于多种生物技术应用，包括植物中的蛋白质表达、基因沉默和基因组编辑（Lozano-Durán，2016）。在解构病毒策略下，双生病毒的外壳蛋白和运动蛋白基因被去除，而复制所需的序列被保留。复制子在递送至植物细胞后进行复制并增加所携带DNA的拷贝数；这导致高水平的基因表达（Lozano-Durán，2016）。尽管一些基于双生病毒复制子的载体已用于基因打靶（Baltes等，2014；Cermak等，2015；Wang等，2017），但基于DNA复制子的载体在植物中的列表仍然有限，特别是对于粮食作物。在这项研究中，我们开发了基于复制子的甘薯曲叶病毒（SPLCV）表达载体。我们以本塞姆氏烟草为模型植物，测试了这些载体在 CRISPR-Cas 介导的 RNA 靶向和基因编辑中的效率。

SPLCV 是一种单联双生病毒，属于Begomoviruses属（Bi 和Zhang，2012）。合成并克隆了SPLCV-JS的编码区（登录号：KF040468.1）复制相关蛋白（Rep：1900 bp，编码四种蛋白：AC1、AC2、AC3和AC4）和基因间区（IR：284 bp）与 IR-GFP-Rep-IR 起始中的 GFP 表达盒 (U4:GFP) 一起进入二元载体 pCambia0390，以产生报告载体 SPLCV-GFP（图 1a）。使用含有相同表达盒的常规 T-DNA 载体 (T-GFP) 作为对照。用含有 SPLCV-GFP 的根癌农杆菌浸润的本塞姆氏烟草叶子显着显示出比 T-GFP 更强的 GFP 荧光（图 1b）。我们通过 PCR 确认了两个 IR 之间的环化，其中一对质数对在 SPLCV-GFP 渗透的叶子中面向相反的方向（图 1c）。SPLCV-GFP 渗透叶子中的 GFP 转录本比 T-GFP 中的转录本高 22.5 倍（图 1d）。

渗透几天后，大多数 SPLCV-GFP 渗透的叶子发现明显的坏死。为了降低细胞致死率，构建了三个突变 SPLCV 载体：（i）AC4 ^m：在 Rep 编码区引入 T 至 A 突变，从而导致 AC4 翻译过早停止（AC4：26T→A） (Leu9TAA);AC1:183T→A(Leu61Leu)); (ii)AC2m ^：在AC2的编码区引入提前终止密码子突变，该突变改变了AC1的一个氨基酸(AC2：31A→T(Lys11TAG)；AC1：1034A→T(Glu345Val))；(iii)AC2m ^/ AC4m^双突变体。对所有构建体进行测序以确认正确的突变位点。将GFP表达盒克隆到三个突变载体中，并与SPLCV-GFP（WT）的表达效率进行比较。在第 3 天和第 9 天，用含有这些突变结构的农杆菌浸润的本塞姆氏烟草叶子中没有观察到 GFP 荧光明显减少（图 1e）。GFP 转录本在 WT 和 AC2 ^m /AC4 ^m载体之间没有显着差异（图 1f）。这些突变构建体在第 12 天时并未诱导坏死（图 1g）。然后，我们通过本塞姆氏烟草(16C) 叶子中的 GFP 沉默实验测试了这些基本病毒载体的 RNA 沉默抑制子 (RSS) 活性。图 1h 显示仅在空载体和 AC2 ^m /AC4 ^m载体中观察到 GFP 沉默。该结果暗示 AC2 ^m /AC4 ^m构建体失去了 RSS 活性。^{在 AC2 m} /AC4 ^m中插入长达 9 kb 的非编码 DNA 序列仍然会导致强烈的 GFP 表达（图 1i）。这一结果表明 SPLCV 的负载能力足以递送 CRISPR 核酸酶。

然后我们测试了 SPLCV 载体是否可以增强植物中 LwaCas13a 介导的 RNA 靶向活性（Abudayyeh等，2017）。AC2 ^m /AC4 ^m构建体用于这一系列实验。我们人工合成了植物密码子优化的 LwaCas13a 的 DNA 序列，并克隆到 AC2 ^m /AC4 ^m和常规载体中，C 端有 3×Flag 融合，在 U4 启动子的表达下有双侧翼核定位序列 (NLS) 。此外，我们针对mGFP5转录本设计了两种 gRNA，并重组了靶向载体（基于 SPLCV 的载体：SV；T-DNA 载体：TV；图 1j）。使用包含非靶向（NT）间隔区的相同载体作为对照。渗透 4 天后，在 SV 渗透的叶子中发现 LwaCas13a 和 gRNA 转录本以及 LwaCas13a 蛋白的积累量高于 TV 渗透的叶子（图 1k 和 l）。与用 mGFP5/NT 构建体浸润的叶子左侧相比，在第 4 天用 SV 浸润的相应右侧观察到 GFP 荧光明显下降。这种现象在 TV 中并不明显（图 1m）。SV 导致mGFP5敲低水平显着高于 TV（gRNA1 为 25%，gRNA2 为 32%；图 1n）（gRNA1 为 67%，gRNA2 为 70%）。SVs 渗透样本积累的 mGFP5 蛋白比 TVs 少（图 1o）。此外，当靶向内源转录本NbPDS1时，我们获得了类似的结果（图 1p）。这些结果清楚地表明，基于 SPLCV 的 RNA 靶向载体比常规载体更有效。

我们进一步测试了 SPLCV 载体是否可以增强 SpCas9 和 LbCas12a 介导的植物基因编辑。由于细胞致死性的丧失，AC4 ^{m衍生的构建体可以有效地进行植物基因编辑，但仍保持 RSS 活性（图 1h；Mao}等人，2018）。因此，AC4 ^m载体被用于这一系列的实验中。我们将植物密码子优化的 SpCas9（C 端有 NLS 融合）和 LbCas12a（C 端有 3×HA 和 NLS 融合）克隆到 AC4 ^m和常规载体中，在 U4 启动子的表达下。我们分别在野生型和16C本塞姆氏烟草中设计了几种针对NbPDS1和mGFP5的crRNA ，并重组了SV和TV（图1q）。含有 NT crRNA 的相同载体用作对照。渗透 6 天后，在 SV 渗透的叶子中发现核酸酶、crRNA 转录本和核酸酶蛋白的积累量高于 TV 渗透的叶子（图 1r 和 s）。在 NT 渗透和 TV 渗透样本中，使用基于 qPCR 的方法获得的目标位点的平均相对扩增值分别约为 1.01 和 0.90（图 1t 和 u；Peng 等， 2018 ）。这样的概念表明电视的突变频率约为 10%，这一结果与之前的报告类似（Bernabé-Orts等，2019）。然而，在 SV 渗透样本中，五个目标位点的平均相对扩增值约为 0.65（图 1t 和 u）。因此，SV 的突变频率为 35%，远高于 TV。

总之，我们的工作扩展了植物中现有的基于双生病毒复制子的载体列表。基于 SPLCV 的载体为提高植物中 CRISPR-Cas 介导的基因敲低、敲除和敲入的效率提供了机会，特别是对其宿主甘薯；在不久的将来，通过分生组织从头诱导的基因编辑也可能在这种重要的六倍体作物中进行（Maher等人，2020）。