The CRISPR/Cas system has rapidly become the preferred tool for genome engineering in various organisms due to high efficiency, specificity, simplicity and versatility. Currently, CRISPR/Cas‐mediated base editing, a novel genome editing strategy that enables irreversible nucleotide changes at target loci without double‐stranded DNA cleavage or any donor template, has been widely adopted for generating gain‐of‐function germplasms in functional genomics research and crop genetic improvement (Hua et al. , 2019; Ren et al. , 2018; Yan et al. , 2018). However, the recognition of a specific protospacer adjacent motif (PAM) for Cas protein restricts the targeting range of these tools, especially base editors, given that it requires a functional PAM for Cas protein interaction to localize the target base in the editing window within the protospacer for nucleotide deamination (Ren et al. , 2017; Ren et al. , 2018; Yan et al. , 2018). The PAM specificity is therefore the key limitation of the application of the CRISPR system in genome editing. Since the commonly used Streptococcus pyogenes Cas9 (SpCas9) recognizes canonical NGG PAM (Hu et al. , 2018), many efforts have been directed towards the identification of new Cas proteins for different PAM specificity in years (Li et al. , 2019; Qin et al. , 2019). Intriguingly, a promising candidate, an orthologous Cas9 protein from Streptococcus canis (ScCas9), which shares 89.2% sequence similarity with SpCas9, has been identified and characterized. ScCas9 recognizes minimal NNG PAM sequences, and it is capable of efficient genome editing in human cells (Chatterjee et al. , 2018). However, its PAM specificity has not been verified in other systems and its application in plants has not been previously reported. Here we show that ScCas9 achieves efficient target gene mutagenesis at NAG sites in comparison with NGG, NTG and NCG sites. Moreover, we also show that ScCas9 can be used in multiplex genome editing and base editing in rice plants.

To test PAM preference of ScCas9 in rice, we first assessed its nuclease activity, side‐by‐side with SpCas9, towards twelve endogenous genomic loci with NGG, NAG, NCG and NTG PAMs (Figure 1a). ScCas9 gene was codon‐optimized, fused with a nuclear localization signal at both termini and expressed in stable transgenic rice plants. Subsequently, individual lines were genotyped by Sanger sequencing as previously reported (Ren et al. , 2019). For NGG PAM sites, the editing efficiency of ScCas9 was comparable to that of SpCas9 at the OsCPK6 target site, averaging 97.92% indels dominated with mono‐allelic mutations compared to 94.12% indels dominated with bi‐allelic mutations for SpCas9. On the other two NGG sites in OsMPK9 and OsMPK17 , ScCas9 showed no activity, whereas SpCas9 did well. For NAG PAM sites, ScCas9 resulted in 91.18% efficiency on OsMPK16 , 94.74% on OsCPK7 , while SpCas9 yielded 7.69% and 92.31%, respectively. On three NTG and NCG PAM sites tested, indel frequencies were more variable, ranging from 0% to 46.67% (all plants were mono‐allelic mutants). Interestingly, the editing efficiency of ScCas9 was genomic locus‐dependent, since 7 out of 12 target sites tested were resistant to ScCas9. Taken together, these results indicate that ScCas9 nuclease recognizes NNG PAM on a locus‐dependent manner in targeted plant genome editing, and it is more suitable for editing NAG target sites in rice.

To further validate the capacity and efficacy of ScCas9 towards the NAG PAM, we tested ScCas9 in multiplex genome editing in transgenic rice plants. One sgRNA targeting both OsMPK14 and OsMPK15 at the conserved genomic region was transferred into rice. Genotyping data showed that ScCas9 achieved comparable activity to SpCas9 at both target sites (Figure 1b). Alternatively, two different sgRNAs targeting OsCPK9 and OsCPK10 simultaneously were used as well. ScCas9 showed notably improved genome editing, averaging 94.12% editing of OsCPK9 and 89.36% editing of OsCPK10 . By contrast, SpCas9 achieved 73.33% and 12.12% editing under the same condition tested, respectively (Figure 1b). In view of all data, we conclude that ScCas9 outperforms SpCas9 on target sites with NAG PAMs in rice.

In our previous studies, we developed a series of cytidine and adenine base editors for targeted base editing using the nickase version of SpCas9 and its variants (Ren et al. , 2019; Ren et al. , 2017). Thus, we speculated that ScCas9 could broaden the targeting scope of base editors considering the preference of NAG PAM. Therefore, we constructed the hAID*Δ‐ScCas9n‐UGI‐NLS chimeric gene, cytidine base editor named rBE25 (Figure 1c), and tested its activity towards an NAG PAM at the OsBZR1 site in transgenic rice plants (Figure 1d). Of 46 independent lines confirmed by Sanger sequencing, 17 heterozygous lines (36.96% efficiency) were identified with nucleotide changes in the target region and 2 lines carried indel mutations (Figure 1e and f). Meanwhile, the chimeric gene TadA‐TadA7.10‐ScCas9n‐NLS , adenine base editor named rBE26 (Figure 1g), was constructed and used to target the endogenous OsGS1 gene for generating potential herbicide‐resistant rice germplasm (Figure 1h). As a result, 19 of 40 independent lines (47.5% efficiency) were identified with a single A to G conversion at the desired site. All the mutated lines were heterozygous, and no indels were identified (Figure 1i and j). Taken together, our data indicate that ScCas9 is compatible with nucleotide deaminases and might serve as a useful RNA‐guided DNA‐targeting platform for other modification enzymes for genome engineering.

In this study, we have extensively investigated the nuclease activity of ScCas9 on different NNG PAM sequences in rice plants. We found that the cleavage activity of ScCas9, different to the report from human cells, is lower at NGG sites and more robust at NAG sites as compared to SpCas9. Furthermore, ScCas9 is less active at NTG and NCG sites. Interestingly, the performance of ScCas9 nuclease is locus‐dependent. It has previously been reported that SpCas9 is sensitive to chromatin state, DNA and/or histone modifications at the target region (Kallimasioti‐Pazi et al. , 2018). Therefore, we presume that ScCas9 might be more sensitive than SpCas9 to these factors. Further experiments with more target sites are required to address this question. Nevertheless, our data show that ScCas9 is a new genome editing player regarding NAG PAM, achieving considerable editing efficiency in multiplex genome editing, and in cytidine as well as adenosine base editing. In conclusion, ScCas9 nuclease and its derived editing tools expand the CRISPR toolbox for targeted genome editing in plants.

CRISPR / Cas系统具有高效，特异性，简单和多功能性，已迅速成为各种生物体中基因组工程的首选工具。目前，CRISPR / Cas介导的碱基编辑是一种新颖的基因组编辑策略，可在靶基因座处实现不可逆的核苷酸变化，而无需双链DNA切割或任何供体模板，已被广泛用于功能基因组学研究中的功能获得种质和作物遗传改良（Hua等，2019 ; Ren等，2018 ; Yan等，2018）。但是，对Cas蛋白的特定原间隔子相邻基序（PAM）的识别限制了这些工具（尤其是碱基编辑器）的靶向范围，因为它需要与Cas蛋白相互作用的功能性PAM才能将目标碱基定位在用于核苷酸脱氨的原间隔子（Ren等人，2017 ; Ren等人，2018 ; Yan等人，2018）。因此，PAM特异性是CRISPR系统在基因组编辑中应用的关键限制。由于常用的化脓性链球菌Cas9（SpCas9）识别典型的NGG PAM（Hu et al。，2018），多年来人们一直致力于鉴定具有不同PAM特异性的新Cas蛋白（Li等人，2019 ; Qin等人，2019）。有趣的是，已经鉴定并鉴定了一种有前途的候选基因，即犬链球菌的直系同源Cas9蛋白（ScCas9），该蛋白与SpCas9具有89.2％的序列相似性。ScCas9识别最小的NNG PAM序列，并且能够在人类细胞中进行有效的基因组编辑（Chatterjee et al。，2018）。但是，其PAM特异性尚未在其他系统中得到验证，并且以前尚未在植物中报道其应用。在这里，我们显示与NGG，NTG和NCG站点相比，ScCas9在NAG站点实现了有效的靶基因诱变。此外，我们还表明，ScCas9可用于水稻植物的多重基因组编辑和碱基编辑。

为了测试水稻中ScCas9的PAM偏好，我们首先与SpCas9一起评估了其对十二个内源基因组位点（NGG，NAG，NCG和NTG PAM）的核酸酶活性（图1a）。ScCas9基因经过密码子优化，在两个末端均与核定位信号融合，并在稳定的转基因水稻植株中表达。随后，如先前报道的那样，通过Sanger测序对单个品系进行基因分型（Ren等，2019）。对于NGG PAM位点，ScCas9在OsCPK6目标位点的编辑效率与SpCas9相当，平均97.92％的插入缺失为单等位基因突变，而94.12％的插入缺失为SpCas9双等位基因。在OsMPK9的其他两个NGG站点上和OsMPK17，ScCas9没有活性，而SpCas9表现良好。对于NAG PAM网站，ScCas9导致91.18％的效率上OsMPK16，在94.74％OsCPK7，而SpCas9分别取得了7.69％和92.31％。在测试的三个NTG和NCG PAM位点上，插入缺失频率变化更大，范围从0％到46.67％（所有植物均为单等位基因突变体）。有趣的是，ScCas9的编辑效率取决于基因组基因座，因为测试的12个靶位中有7个对ScCas9具有抗性。综上所述，这些结果表明，ScCas9核酸酶在目标植物基因组编辑中以基因依赖的方式识别NNG PAM，它更适合于编辑水稻中的NAG目标位点。

为了进一步验证ScCas9对NAG PAM的能力和功效，我们在转基因水稻植物的多重基因组编辑中测试了ScCas9。将一种在保守基因组区域靶向OsMPK14和OsMPK15的sgRNA转移到水稻中。基因分型数据显示，ScCas9在两个靶位点均实现了与SpCas9相当的活性（图1b）。可替代地，也同时使用了靶向OsCPK9和OsCPK10的两种不同的sgRNA 。ScCas9表现显着改善基因组编辑，平均94.12％的编辑OsCPK9和89.36％的编辑OsCPK10。相比之下，在同一测试条件下，SpCas9的编辑率分别为73.33％和12.12％（图1b）。根据所有数据，我们得出结论，在水稻中使用NAG PAM时，ScCas9的表现优于SpCas9。

在我们以前的研究中，我们开发了一系列胞嘧啶和腺嘌呤碱基编辑器，用于使用SpCas9及其变体的切口酶版本进行靶向碱基编辑（Ren等人，2019年; Ren等人，2017年）。因此，我们推测，考虑到NAG PAM的偏爱，ScCas9可以扩大基础编辑者的定位范围。因此，我们构建了hAID *Δ-ScCas9n-UGI-NLS嵌合基因，胞苷碱基编辑器命名为rBE25（图1c），并测试了其对OsBZR1对NAG PAM的活性。转基因水稻植株中的位点（图1d）。通过Sanger测序确认的46个独立品系中，鉴定出17个杂合品系（效率为36.96％），其靶区域的核苷酸发生了变化，其中2个品系带有indel突变（图1e和f）。同时，构建了嵌合基因TadA-TadA7.10-ScCas9n-NLS，腺嘌呤碱基编辑器，命名为rBE26（图1g），并用于靶向内源性OsGS1基因产生潜在的抗除草剂水稻种质（图1h）。结果，鉴定出40条独立系中的19条（效率为47.5％），并在所需位点进行了一次A到G的转化。所有突变株系均为杂合子，未鉴定出插入缺失（图1i和j）。综上所述，我们的数据表明ScCas9与核苷酸脱氨酶兼容，并且可以作为有用的RNA引导的DNA靶向平台，用于其他基因组工程修饰酶。

在这项研究中，我们已广泛研究了ScCas9对水稻植物中不同NNG PAM序列的核酸酶活性。我们发现，与CasCas9相比，ScCas9的切割活性与人细胞报告的不同，在NGG部位更低，在NAG部位更强劲。此外，ScCas9在NTG和NCG站点上的活性较低。有趣的是，ScCas9核酸酶的性能取决于基因座。以前有报道说SpCas9对目标区域的染色质状态，DNA和/或组蛋白修饰敏感（Kallimasioti-Pazi等人，2018）。因此，我们认为ScCas9可能比SpCas9对这些因素更敏感。为了解决这个问题，需要对更多目标位点进行进一步的实验。尽管如此，我们的数据显示，ScCas9是涉及NAG PAM的新型基因组编辑器，在多重基因组编辑，胞苷以及腺苷碱基编辑中实现了可观的编辑效率。总之，ScCas9核酸酶及其衍生的编辑工具扩展了CRISPR工具箱，可用于植物中的靶向基因组编辑。