Chemical herbicides, which are commonly used to kill weeds in the field, have been extensively applied and transformed the way of weed management in modern agriculture due to its efficiency, ease of use and relatively low cost. In last decades, a large number of transgenic herbicide‐tolerant crops (i.e. corn, soybean, cotton, rice) have been developed and commercialized, reshaping the global seed market (Schutte et al., 2017). However, the use of chemical herbicides (i.e. ALS inhibitors, EPSPS inhibitors, ACCase inhibitors) has also drastically increased, resulting in the occurrence of weed species resistant to these herbicides. Thus, developing novel herbicide‐resistant crops to diversify weed management is of great value in slowing the evolution of weed resistance to herbicides and to maintain sustainable crop production in the future. Thanks to cutting‐edge CRISPR‐mediated base editing technologies which have emerged lately (Ren et al., 2019; Wang et al., 2020; Yan et al., 2018); important genetic variations related to any herbicide resistance can be manipulated directly and rapidly by the relevant base editors. Tubulin genes have been reported to endow resistance to trifluralin and other dinitroaniline herbicides in a number of crops (Chu et al., 2018; Lyons‐Abbott et al., 2010). Compared to other major herbicides, the frequency of dinitroaniline resistance in weeds is quite low (Heap, 2014). This is likely because mutations in tubulin genes could affect microtubule polymerization, an important process in cell division and elongation, and consequently lead to plant death. Therefore, tubulin genes are promising target genes to develop herbicide‐resistant germplasms for future crop breeding. In this study, we successfully generated a novel artificial rice germplasm with trifluralin and pendimethalin resistance without fitness penalty by precisely editing the endogenous OsTubA2 gene within the rice genome. Theoretically, this important trait can also be rapidly introduced into other major crops using CRIPSR‐mediated adenine base editors.

It has been previously shown that a Met‐268‐Thr mutation in the α‐tubulin gene EiTUA1 correlates with dinitroaniline resistance in a number of goosegrass (Eleusine indica) biotypes (Yamamoto et al., 1998). Proteins belonging to the α‐tubulin and β‐tubulin family are highly conserved, showing up to 88% amino acid similarity (Rao et al., 2016). Therefore, we hypothesized that introduction of this point mutation into the rice genome might transform common rice into a herbicide‐tolerant variety. To this end, precise base editing of the α‐tubulin homologue gene OsTubA2 (LOC_Os11g14220) was carried out using the previously described rice adenine base editor rBE14 (Yan et al., 2018). A sgRNA corresponding to a NGG PAM was designed to target the complementary genomic DNA strand of OsTubA2 at T1981 site (Figure 1a). The oligos were synthesized, constructed and shuttled into rBE14 binary vector through Gateway recombination as described previously (Yan et al., 2018). The rBE14/sgRNA system was then introduced into rice cultivar Kitaake (Oryza sativa spp. japonica) through Agrobacterium‐mediated transformation.

Rice plants were regenerated from the independent calli and genotyped directly by Sanger sequencing of the target region. Out of 63 independent rice lines obtained, 8 lines (12.7% efficiency) were identified with A > G conversion occurring at T1981, replacing the Met268 residue with threonine residue (Figure 1b). All eight mutant lines were heterozygous and had no random indels detected. The potential off‐targets of OsTubA2‐targeting rBE14/sgRNA in the rice genome were predicted. No nucleotide changes at the potential off‐target sites were detected in the eight positive transgenic lines (Figure 1c).

Next, T1 progenies of each OsTubA2‐edited T0 lines were genotyped to determine the lines in which the T‐DNA was segregated out. PCR amplification with specific primers corresponding to the Cas9, sgRNA and Hyg transgenes was conducted. As shown in Figure 1d ande, some T1 individuals lacked the transgenes due to genetic segregation. Furthermore, Sanger sequencing revealed that some plants (i.e. #2‐5, #2‐10 etc.) were homozygous. Given that germination of wild‐type Kitaake seeds is sensitive to dinitroaniline herbicide treatment (Figure 1f), a phenotypic and genotypic analysis was further conducted. Rice seeds of T1 population of heterozygous T0 lines #2 and #5 were germinated in cylinders containing 6.6 mg/L pendimethalin which is sufficient to inhibit the hypocotyl and root growth of wild‐type rice. All homozygous seeds carrying the M268T mutation showed resistance to pendimethalin treatment, whereas germination of wild‐type seed was arrested. On the other hand, heterozygous plants exhibited different resistance phenotypes according to the growth of roots and hypocotyls under the conditions tested, it likely results from the dosage effect of M268T mutation (Figure 1g andh). We also analysed the resistance to 4.0 mg/L trifluralin (another type of dinitroaniline herbicides) which can inhibit the root and hypocotyl growth of wild‐type rice. When trifluralin was applied to T1 progenies, similar phenotypes were observed (Figure 1h). Thus, the M268T mutation of OsTubA2 truly confers resistance to both trifluralin and pendimethalin herbicides in rice and is stably inherited in the subsequent generations.

All the OsTubA2‐edited plants lacking the T‐DNA transgenes were grown under natural light conditions in the greenhouse. The plants were morphologically indistinguishable from the wild‐type plants (Figure 1i). Furthermore, the thousand‐grain weight and germination rate of T2 seeds were investigated, no significant difference was observed between M268T and wild‐type plants (Figure 1j). Together, our results indicate that introduction of Met‐268‐Thr mutation in OsTubA2 by base editing does not cause growth penalty in rice.

Genetic diversity of OsTubA2 gene in 4,726 re‐sequenced rice accessions was investigated using RiceVarMap v2.0 (Zhao et al., 2015). The data indicate that M268 in OsTubA2 has not been targeted by natural or human selection during rice domestication (Figure 1k). Thus, M268T plants are a novel artificial germplasm with a great potential for future rice improvement. A number of α‐tubulin genes from other major economic crops, including wheat, maize, barley, sorghum, oilseed rape, cotton and soybean, were further analysed. Sequence alignments revealed that the sequence of M268 locus is highly conserved among different species (Figure 1l), opening the possibility that M268T‐mediated dinitroaniline resistance can be rapidly introduced into other crops through precise base editing.

In summary, our study shows that the M268T mutation in the endogenous OsTubA2 gene, generated by adenine base editor, endows dinitroaniline herbicide resistance in rice without inducing fitness cost. Any other rice cultivars or cash crops can be enhanced with the herbicide resistance trait using this strategy in the future.

由于化学除草剂的效率，易用性和相对较低的成本，在现代农业中通常用于杀死杂草的化学除草剂已被广泛应用并改变了杂草的管理方式。在过去的几十年中，已经开发了许多转基因耐除草剂作物（即玉米，大豆，棉花，水稻）并实现了商品化，重塑了全球种子市场（Schutte等人，2017年））。但是，化学除草剂（例如ALS抑制剂，EPSPS抑制剂，ACCase抑制剂）的使用也已大大增加，导致出现了对这些除草剂具有抗性的杂草物种。因此，开发新型的抗除草剂作物以使杂草多样化管理在减缓杂草对除草剂的抗性演变以及维持未来作物可持续生产方面具有重要价值。由于最近出现了尖端的CRISPR介导的碱基编辑技术（Ren等人，2019 ; Wang等人，2020 ; Yan等人，2018）; 有关的除草剂抗性相关的重要遗传变异可以由相关的基本编辑直接而迅速地处理。据报道，在许多农作物中，微管蛋白基因赋予了对氟乐灵和其他二硝基苯胺除草剂的抗性（Chu等人，2018 ; Lyons‐Abbott等人，2010）。与其他主要除草剂相比，杂草中对二硝基苯胺的抗药性相当低（Heap，2014年）。这可能是因为微管蛋白基因的突变会影响微管聚合，这是细胞分裂和伸长的重要过程，并因此导致植物死亡。因此，微管蛋白基因是有希望的靶基因，可开发出抗除草剂的种质，以供将来的作物育种。在这项研究中，我们通过精确编辑水稻基因组中的内源OsTubA2基因，成功地产生了具有氟乐灵和二甲戊乐灵抗性的新型人工水稻种质，而没有适应性损失。从理论上讲，也可以使用CRIPSR介导的腺嘌呤碱基编辑器将这一重要性状迅速引入其他主要农作物。

先前已经表明，在一个的Met-268-THR突变α微管蛋白基因EiTUA1在许多与二硝基苯胺抗性相关因素牛筋草（牛筋）生物型（山本等人。，1998）。属于α-微管蛋白和β-微管蛋白家族的蛋白质高度保守，显示出高达88％的氨基酸相似性（Rao等，2016）。因此，我们假设将此点突变引入水稻基因组可能会将普通水稻转化为耐除草剂品种。为此，对α-微管蛋白同源基因OsTubA2（LOC_Os11g14220）使用先前描述的水稻腺嘌呤碱基编辑器rBE14（Yan等人，2018）进行。对应于NGG PAM的sgRNA被设计为在T1981位点靶向OsTubA2的互补基因组DNA链（图1a）。如先前所述（Yan等人，2018），通过Gateway重组来合成，构建和寡核苷酸穿梭到rBE14二元载体中。然后将rBE14 /因组系统导入水稻品种Kitaake（稻属粳稻）通过农杆菌介导的转化。

水稻植株由独立的愈伤组织再生并通过靶区的Sanger测序直接进行基因分型。在获得的63个独立水稻系中，鉴定出8个系（效率12.7％），在T1981发生A> G转化，用苏氨酸残基代替Met268残基（图1b）。所有八个突变体系都是杂合的，没有检测到随机插入/缺失。潜在脱靶OsTubA2-在水稻基因组靶向rBE14 /因组进行了预测。在八个阳性转基因品系中未检测到潜在的脱靶位点的核苷酸变化（图1c）。

接下来，对每个经过OsTubA2编辑的T0系的T1后代进行基因分型，以确定分离出T-DNA的系。用对应于Cas9，sgRNA和Hyg转基因的特异性引物进行PCR扩增。如图1d ande所示，由于遗传隔离，一些T1个体缺乏转基因。此外，桑格（Sanger）测序揭示了一些植物（例如2-5、2-10等。）是纯合子。鉴于野生型Kitaake种子的发芽对二硝基苯胺除草剂处理敏感（图1f），因此需要进一步进行表型和基因型分析。杂合性T0系＃2和＃5的T1群体的水稻种子在装有6.6 mg / L二甲戊乐灵的圆筒中发芽，这足以抑制野生型水稻的下胚轴和根系生长。所有带有M268T突变的纯合种子均显示对二甲戊乐灵的抗性，而野生型种子的发芽被阻止。另一方面，在测试条件下，杂合植物根据根和下胚轴的生长表现出不同的抗性表型，这很可能是由于M268T突变的剂量效应所致（图1g和h）。我们还分析了对4的阻力。0 mg / L三氟拉林（另一种二硝基苯胺除草剂）可以抑制野生型水稻的根和下胚轴生长。当三氟拉林应用于T1后代时，观察到相似的表型（图1h）。因此，M268T的突变OsTubA2确实赋予水稻对三氟拉林和二甲戊乐灵除草剂的抗性，并在后代中稳定遗传。

所有缺少T-DNA转基因的经过OsTubA2编辑的植物都在温室的自然光照条件下生长。这些植物在形态上与野生型植物没有区别（图1i）。此外，研究了T2种子的千粒重和发芽率，M268T和野生型植物之间没有发现显着差异（图1j）。总之，我们的结果表明，通过碱基编辑在OsTubA2中引入Met-268-Thr突变不会导致水稻生长障碍。

使用RiceVarMap v2.0研究了4,726个重测序水稻种质中OsTubA2基因的遗传多样性（Zhao等，2015）。数据表明，在水稻驯化过程中，OsTubA2中的M268未被自然或人为选择作为目标（图1k）。因此，M268T植物是一种新型的人工种质，具有未来水稻改良的巨大潜力。多种α微管蛋白进一步分析了其他主要经济作物的基因，包括小麦，玉米，大麦，高粱，油菜，棉花和大豆。序列比对显示，M268基因座的序列在不同物种之间高度保守（图1l），这为通过精确碱基编辑将M268T介导的二硝基苯胺抗性迅速引入其他作物的可能性提供了可能性。

总而言之，我们的研究表明，由腺嘌呤碱基编辑产生的内源性OsTubA2基因中的M268T突变赋予水稻对二硝基苯胺除草剂的抗性，而不会降低适应性成本。将来，使用该策略可以提高除草剂抗性的任何其他水稻品种或经济作物。