Site‐directed mutagenesis facilitates the experimental validation of gene function and can speed up plant breeding by producing new genetic variability or by reproducing previously known gene variants in other than their original genetic backgrounds. However, its application is challenging in wheat owing to high genomic redundancy and highly genotype‐dependent DNA transfer methods (Koeppel et al., 2019). In wheat, large chromosomal regions are hardly amenable to meiotic recombination, which limits the potential for trait improvements. The era of transgenesis facilitated the generation of desired traits through the transfer of recombinant DNA into elite backgrounds. This technology, however, is limited by long and costly regulatory evaluation processes owing to publicly overrated method‐specific risks. As another option, the use of meiotically recombinant and genetically fixed doubled haploids proved very useful for accelerating crop improvement (Kalinowska et al., 2019). Viable methods of in planta haploid induction via uniparental genome elimination are available in species such as Arabidopsis through modification of CENTROMERIC HISTONE 3 (CENH3) (Ravi and Chan, 2010), in maize and rice via knockout of a sperm‐specific phospholipase gene (Kelliher et al., 2017; Yao et al., 2018), and in wheat through intergeneric crossing with maize (Laurie and Bennett, 1988). Haploid induction coupled with site‐directed mutagenesis has previously been reported in Arabidopsis, maize and wheat (Kelliher et al., 2019). However, in wheat, no unambiguous evidence has been provided yet, considering that a mutated target sequence was shown for just a single event. Furthermore, proof of heritability of site‐directed mutations is still lacking. The present study involves intergeneric pollination of wheat with cas9/guide RNA (gRNA)‐transgenic maize to facilitate site‐directed mutagenesis in any wheat germplasm of choice. For exemplification of this principle, new allelic variants were generated for the wheat genes BRASSINOSTEROID‐INSENSITIVE 1 (BRI1) and SEMI‐DWARF 1 (SD1) which are involved in the regulation of plant height.

The present approach relies on the expression of cas9 and wheat gene‐specific gRNA in maize sperm cells. Therefore, transgenic maize carrying a ubiquitously expressed GFP was analysed, and conspicuous fluorescence was found in sperm cells (Figure 1a). Two Cas9/gRNA target motifs for TaBRI1 and one for TaSD1 were selected. These proved to be conserved across all two (AABB) or three (AABBDD) homeologues of the target genes in durum and bread wheat, respectively. Corresponding gRNAs were cloned into generic vectors used to transform maize (Budhagatapalli et al., 2016). Two hundred maize T₀ plants carrying wheat target‐specific cas9/gRNA‐encoding T‐DNAs were prescreened by qRT‐PCR analysis. Per target motif, five maize transgenics with high cas9 and gRNA expression were selected for pollination of wheat (Figure 1b). Upon these intergeneric crosses, embryos were rescued in vitro. Regenerated wheat plants were then subjected to PCR‐based mutation analysis by Sanger sequencing. For BRI1 target motif 1, three, two and one mutants were obtained out of 83, 44 and 10 plants in genotypes BW, W5 and D6, respectively. Two plants out of 4 and 3 carried mutations for BRI1 target motif 2 in genotypes W5 and D7, respectively. In addition, seven mutants for the SD1 target motif 1 were obtained from 17, 5 and 8 plants in genotypes BW, K15 and S96, respectively (Figure 1c). Subcloning and Sanger sequencing of target motif‐derived amplicons of M₁ plants indicated that all bread wheat mutants for BRI1 and SD1 were invariably homozygous, whereas those in durum genotypes D6 and D7 were chimeric (Figure 1c). This may be due to differences in cas9 and gRNA expression, the time point of male genome activation and the activity of DNA repair.

In the present approach, mutations can be induced at various phases before and after the zygote undergoes mitosis (Figure 1d). Mutations induced during G₁ and early S phase are more likely to occur owing to Cas9 and gRNA molecules pre‐produced in the sperm rather than to zygotic de novo transgene expression. Resultant embryos are expectedly non‐chimeric with regard to the induced mutations (Figure 1d‐i). Alternatively, after chromatid duplication, Cas9 may trigger mutations in one chromatid or independently in either of the sister chromatids (Figure 1d‐ii). In this scenario, the daughter cell that has received a mutated wheat chromatid during the first embryonic mitosis itself undergoes S phase, by which the mutated allele becomes genetically fixed across the two sister chromatids, while the other daughter cell has received a non‐mutated or differently mutated chromatid and thus gives rise to a genetically distinct sector. Consequently, embryos formed via mutagenesis during G₂ phase are expectedly chimeric (Figure 1d‐ii). In the course of initial embryonic cell divisions upon wheat x maize crosses, maize chromosomes are eliminated due to asynchronous processing in terms of DNA replication, condensation and centromere formation (Laurie and Bennett, 1988).

In total, 15 independent target gene‐specific mutants were identified out of 174 wheat plants from which good‐quality Sanger sequences of target motifs had been retrieved. Mutants were obtained in six wheat backgrounds, including the three spring‐type bread wheats BW, W5 and K15, the winter‐type bread wheat S96, and the two durum wheats D6 and D7 (Figure 1c). Mutations were found in all three target motifs addressed (Figure 1c). None of the 15 mutants carried any transgene. Across the genotypes, the efficiency in mutant plant formation ranged from 3.6% to 50% (Figure 1c). The BRI1 and SD1 genes are known to play an important role in plant height. Therefore, loss‐of‐function mutants may entirely fail to develop. In addition, knockouts of BRI1 and SD1 (GA20ox) in Arabidopsis lead to male sterility, as they regulate key genes of anther and pollen development (Plackett et al., 2012; Ye et al., 2010). The haploid plants obtained in the present work were subjected to colchicine treatment. As a result, 7 out of 15 mutants were fertile (Figure 1c). In M₂, progenies of doubled‐haploid mutants SD1‐TM1‐DH04‐AABBdd of genotype K15 and SD1‐TM1‐DH07‐AABBdd of genotype BW proved to have invariably inherited the very same mutations detected in their M₁ progenitors (2‐bp deletion and 48‐bp insertion in the D subgenome, respectively) (Figure 1c). These primary mutants were thereby confirmed to be non‐chimeric and true breeding. The M₂ plants displayed a reduced plant height phenotype. At the anthesis stage, the height of tiller 1 exhibited an average reduction of 6 and 5 cm compared with the wild‐type in genotypes K15 and BW, respectively (Figure 1e). The weak phenotype of these mutants is likely due to the still functional SD1 homeologues of the A and B genomes which may largely compensate the loss of function of the sd1 alleles of the D genome.

In conclusion, the principle of haploid induction coupled with site‐directed mutagenesis was exemplified in wheat using the two target genes BRI1 and SD1 which control the agronomically important trait plant height. Major advances achieved in this work include reduced genotype dependence of site‐directed mutagenesis in wheat, the opportunity of creating a whole variety of mutations using just one cas9/gRNA‐transgenic (pollinator) plant as well as the production of T‐DNA‐free and frequently homozygous M₁ plants. There is still scope for increasing the efficiency of this approach, for example by stronger transgene expression at the relevant time point or by the development of improved protocols for in planta production of doubled haploids.

定点诱变有利于基因功能的实验验证，并且可以通过产生新的遗传变异或通过在原始遗传背景之外复制先前已知的基因变体来加速植物育种。然而，由于高基因组冗余和高度基因型依赖的 DNA 转移方法，其在小麦中的应用具有挑战性（Koeppel等人， 2019 ）。在小麦中，大的染色体区域很难进行减数分裂重组，这限制了性状改善的潜力。转基因时代通过将重组 DNA 转移到精英背景中，促进了所需性状的产生。然而，由于公开高估了特定方法的风险，该技术受到漫长且成本高昂的监管评估流程的限制。作为另一种选择，使用减数分裂重组和遗传固定的双单倍体被证明对于加速作物改良非常有用（Kalinowska等人， 2019 ）。通过单亲基因组消除进行植物内单倍体诱导的可行方法可在拟南芥等物种中通过修改着丝粒组蛋白 3 ( CENH3 ) (Ravi 和 Chan, 2010 )，在玉米和水稻中通过敲除精子特异性磷脂酶基因 (Kelliher)等， 2017 ；Yao等， 2018 ），以及通过与玉米属间杂交的小麦（Laurie 和 Bennett， 1988 ）。此前曾在拟南芥、玉米和小麦中报道过单倍体诱导与定点诱变（Kelliher et al ., 2019 ）。 然而，在小麦中，考虑到仅在单个事件中显示了突变的靶序列，因此尚未提供明确的证据。此外，仍然缺乏定点突变遗传性的证据。本研究涉及用cas9 /gRNA 转基因玉米对小麦进行属间授粉，以促进任何所选小麦种质的定点诱变。为了举例说明这一原理，我们为参与株高调节的小麦基因BRASSINOSTEROID-INSENSITIVE 1 ( BRI1 ) 和SEMI-DWARF 1 ( SD1 ) 生成了新的等位基因变体。

目前的方法依赖于玉米精子细胞中cas9和小麦基因特异性 gRNA 的表达。因此，对携带普遍表达的GFP的转基因玉米进行了分析，并在精子细胞中发现了明显的荧光（图 1a）。选择了TaBRI1的两个 Cas9/gRNA 靶基序和TaSD1的一个。事实证明，这些基因分别在硬粒小麦和面包小麦中目标基因的所有两个 (AABB) 或三个 (AABBDD) 同源物中是保守的。相应的 gRNA 被克隆到用于转化玉米的通用载体中（Budhagatapalli等人， 2016 ）。通过 qRT-PCR 分析对 200 株携带小麦靶标特异性cas9 /gRNA 编码 T-DNA 的玉米 T ₀植物进行了预筛选。每个目标基序，选择五个具有高cas9和 gRNA 表达的玉米转基因用于小麦授粉（图 1b）。通过这些属间杂交，胚胎在体外被拯救。然后通过桑格测序对再生小麦植株进行基于 PCR 的突变分析。对于BRI1靶基序 1，分别从基因型 BW、W5 和 D6 的 83 株、44 株和 10 株植物中获得了 3 个、2 个和 1 个突变体。 4 株和 3 株植物中的 2 株分别携带基因型 W5 和 D7 中BRI1靶基序 2 的突变。此外， SD1目标基序 1 的 7 个突变体分别从基因型 BW、K15 和 S96 的 17、5 和 8 株植物中获得（图 1c）。 对 M ₁植物的目标基序衍生的扩增子进行亚克隆和桑格测序表明， BRI1和SD1的所有面包小麦突变体始终是纯合的，而硬粒小麦基因型 D6 和 D7 中的突变体是嵌合的（图 1c）。这可能是由于cas9和gRNA表达、男性基因组激活的时间点和DNA修复活性的差异所致。

在本方法中，可以在受精卵经历有丝分裂之前和之后的各个阶段诱导突变（图1d）。 G ₁期和早期 S 期诱导的突变更有可能发生是由于精子中预先产生的 Cas9 和 gRNA 分子，而不是合子从头转基因表达。就诱导突变而言，所得胚胎预计是非嵌合的（图 1d-i）。或者，在染色单体复制后，Cas9 可能会触发一个染色单体的突变，或独立触发任一姐妹染色单体的突变（图 1d-ii）。在这种情况下，在第一次胚胎有丝分裂期间接收到突变的小麦染色单体的子细胞本身会经历 S 期，在此期间，突变的等位基因在两个姐妹染色单体之间进行遗传固定，而另一个子细胞则接收到非突变或非突变的等位基因。不同突变的染色单体，从而产生遗传上不同的部分。因此，G ₂期通过诱变形成的胚胎预计是嵌合的（图 1d-ii）。在小麦与玉米杂交的初始胚胎细胞分裂过程中，由于 DNA 复制、浓缩和着丝粒形成方面的异步处理，玉米染色体被消除（Laurie 和 Bennett， 1988 ）。

总共，从 174 株小麦植株中鉴定出 15 个独立的靶基因特异性突变体，并从中检索了靶基序的优质桑格序列。在六种小麦背景中获得突变体，包括三种春型面包小麦 BW、W5 和 K15，冬型面包小麦 S96，以及两种硬粒小麦 D6 和 D7（图 1c）。在所有三个目标基序中都发现了突变（图 1c）。 15 个突变体均未携带任何转基因。在各个基因型中，突变植物形成的效率范围为 3.6% 至 50%（图 1c）。已知BRI1和SD1基因在植物高度中发挥重要作用。因此，功能丧失突变体可能完全无法发育。此外，拟南芥中BRI1和SD1 （ GA20ox ）的敲除会导致雄性不育，因为它们调节花药和花粉发育的关键基因（Plackett等， 2012 ；Ye等， 2010 ）。对本工作中获得的单倍体植物进行秋水仙碱处理。结果，15 个突变体中有 7 个是可育的（图 1c）。在 M ₂中，基因型 K15 的双单倍体突变体 SD1-TM1-DH04-AABBdd 和基因型 BW 的 SD1-TM1-DH07-AABBdd 的后代被证明总是继承了在其 M ₁祖细胞中检测到的完全相同的突变（2-bp分别在 D 亚基因组中缺失和 48 bp 插入）（图 1c）。这些初级突变体由此被证实是非嵌合的和真正的育种。 M ₂植物表现出植物高度降低的表型。 在花期，K15和BW基因型的1号分蘖高度与野生型相比分别平均降低6厘米和5厘米（图1e）。这些突变体的弱表型可能是由于 A 和 B 基因组的SD1同源物仍然有功能，这可能在很大程度上补偿了 D 基因组的sd1等位基因的功能丧失。

总之，利用控制农艺重要性状株高的两个靶基因BRI1和SD1 ，在小麦中例证了单倍体诱导与定点诱变相结合的原理。这项工作取得的主要进展包括减少小麦定点诱变的基因型依赖性、仅使用一种cas9 /gRNA 转基因（传粉媒介）植物产生多种突变的机会以及不含 T-DNA 的生产和经常纯合的M ₁植物。提高这种方法的效率仍有空间，例如通过在相关时间点更强的转基因表达或通过开发用于在植物中生产双单倍体的改进方案。