Chrysanthemum (Chrysanthemum morifolium Ramat.) is an economically important ornamental crop worldwide. This typical obligate short-day (SD) herbaceous perennial in the Asteraceae (Compositae) family is a useful model for studying the photoperiodic control of flowering. However, the complex, heterozygous chrysanthemum genome and the self-incompatibility of this species have hindered basic and applied research on its horticultural and physiological properties.

Isolating and characterizing mutants in specific genes is critical to dissecting gene function for both basic and applied research. In the first report of genome editing in chrysanthemum, six CmDMC1 (DISRUPTION OF MEIOTIC CONTROL 1) genes were simultaneously targeted by TALENs (Shinoyama et al., 2020). However, the design of sequence-specific TALENs is cumbersome and the needed recombinant plasmids are large. In this regard, CRISPR/Cas9-mediated genome editing holds advantages in vector design and assembly, especially when targeting multiple genes, and has been widely used in many organisms. Nevertheless, the high frequency of chimeric events and low editing efficiency has hindered the application of the CRISPR/Cas9 system in chrysanthemum (Kishi-Kaboshi et al., 2017). Only a single study has reported a successful knockout for a TCP transcription factor gene in chrysanthemum (C. morifolium) using a conventional CRISPR/Cas9-mediated system (Li et al., 2022).

Chrysanthemum indicum L. is often used as a model for cultivated chrysanthemum since it is a progenitor of cultivated chrysanthemum, and as a health food and anti-inflammatory herb in traditional Chinese medicine for over 2000 years. In this study, we chose a diploid C. indicum as material (2n = 2x = 18) (Figure S1) and first targeted its single-copy Phytoene desaturase (CiPDS) gene. We cloned four single-guide RNAs (sgRNAs) based on the CiPDS sequence (Figure 1a,b) into pDIRECT-22C, which can simultaneously express multiple sgRNAs using the Csy-type ribonuclease 4 (Csy4) enzyme (Čermák et al., 2017). We measured the editing efficiency of each sgRNA in C. indicum protoplasts by high-throughput sequencing (Data S1). The editing efficiencies of sgRNA1-4 were 8.22 ± 1.2%, 7.82 ± 0.3%, 4.02 ± 0.4%, and 9.14 ± 1.0%, respectively (Figure 1b). We chose sgRNA1 and sgRNA2 to knockout CiPDS because they had the highest editing efficiencies and target the first CiPDS exon (Figure 1a).

Since the selection of optimal promoters is important for high-efficiency genome editing, we tested five different promoters to drive Cas9 and the sgRNAs (Figure 1c): the cauliflower mosaic virus (CaMV) 35S, Cestrum yellow leaf curling virus (CmYLCV), parsley (Petroselinum crispum) Ubiquitin (PcUbi), Arabidopsis thaliana YAOZHE (YAO), and C. indicum Ubiquitin (CiUbi) promoters. We also codon-optimized Cas9 based on the codon usage of A. thaliana (AtCas9) and C. indicum (CiCas9). We tested the editing efficiencies of the resulting nine constructs in at least two independent experiments each (Table S1). In total, we transformed 7786 C. indicum leaf discs with the nine constructs via Agrobacterium tumefaciens-mediated transformation (Figure 1c). The construct harbouring 35S:AtCas9 and PcUbi:sgRNA showed the highest editing efficiency among all experiments and was designated pDIRECT-Ci-opti (Construct 5 in Figure 1c and Table S1).

We obtained eight albino plants by transformation using pDIRECT-Ci-opti, with a biallelic editing efficiency of 0.74% (Figure 1c). One albino plant showed a bushy phenotype (Figure 1d1), whereas most plants developed normal roots and leaves (Figure 1e). Sanger sequencing revealed that all mutations in the albino plants occurred solely at the sgRNA1 target site and consisted of deletions and insertions (Figure 1e and Figure S2).

Among the 14 albino plants generated (Figure 1c), four showed variegated phenotypes (28.6%), suggesting a high level of chimerism (Figure 1d2,d3). When CiPDS is used as a visual marker to validate genome editing, chimerism is easily scorable, but scoring is time-consuming and laborious for most target genes without a visual mutant phenotype. To further improve the screening efficiency and solve the chimerism issue, we generated transgenic C. indicum expressing eYGFPuv (35S:eYGFPuv), encoding a protein with bright fluorescence under ultraviolet (UV) light that is visible to the naked eye (Chin et al., 2018) and has been used as a visible reporter for gene expression and stable transformation in plants (Yuan et al., 2021). All organs of 35S:eYGFPuv C. indicum plants exhibited fluorescence (Figure 1f). We then targeted eYGFPuv with specific sgRNAs for exploitation as a visible marker to efficiently screen gene-edited plants while eliminating chimeric plants. We cloned both CiPDS sgRNA1 and sgRNA2 and eYGFPuv sgRNA1 and sgRNA2 into pDIRECT-Ci-opti (pDIRECT-Ci-opti-CiPDS-eYGFPuv, Figure 1g) and transformed this construct into 35S:eYGFPuv leaf discs. While leaf discs from 35S:eYGFPuv C. indicum exhibited bright fluorescence, those from the wild type (WT) did not exhibit fluorescence under UV light (Figure 1h1). Following transformation with pDIRECT-Ci-opti-CiPDS-eYGFPuv, some 35S:eYGFPuv-derived calli lacked eYGFPuv fluorescence, suggesting that they harboured eYGFPuv mutation(s) (Figure 1h2). In addition, some regenerated non-fluorescent plantlets were not albino (Figure 1h3), suggesting that these plantlets were eygfpuv biallelic mutants but not Cipds biallelic mutants. Other regenerated plantlets without fluorescence were albino (Figure 1h4), indicating that they were eygfpuv Cipds biallelic mutants. We also identified variegated plantlets based on eYGFPuv fluorescence (Figure 1h5).

Using traditional antibiotic selection, we obtained 50 hygromycin-resistant transgenic plantlets. Among them, we identified eight Cipds biallelic mutants, representing a biallelic mutant screening efficiency of 16.0% (Figure 1i). We also selected 11 plants without eYGFPuv fluorescence that were eygfpuv biallelic mutants. Among them, seven were albino, with a biallelic mutant screening efficiency for eygfpuv of 63.6% (Figure 1i). Therefore, using mutated eYGFPuv as a visible marker improved screening efficiency and largely eliminated chimeric plants.

The production of flowering in chrysanthemum usually requires the artificial regulation of day length. One major goal of chrysanthemum breeding is to create photoperiod-insensitive or early-flowering chrysanthemum cultivars. TERMINAL FLOWER 1 (TFL1) inhibits flowering in chrysanthemum. Overexpressing TFL1 delays flowering time in chrysanthemum (Gao et al., 2019; Higuchi and Hisamatsu, 2015). Here, we identified two TFL1 homologues in C. indicum: CiTFL1a and CiTFL1b. We generated three Citfl1a biallelic mutants and one Citfl1b homozygous mutant using our CRISPR/Cas9 platform. The three Citfl1a biallelic mutants contained small deletions and insertions in the first CiTFL1a exon. The Citfl1b mutant contained the same 1-bp insertion in the third CiTFL1b exon in both gene copies (Figure 1j and Figure S3). The edited plants showed different degrees of early flowering, with the Citfl1a mutants exhibiting the earliest flowering (Figure 1k). Under SD conditions, flower buds emerged at 41.3 ± 1.2 days in WT plants but at 16.7 ± 0.6 and 32.7 ± 1.2 days in Citfl1a and Citfl1b plants, respectively (Figure 1l).

The visible reporter-assisted CRISPR/Cas9 system developed in this study should facilitate research and breeding of chrysanthemum.

菊花（Chrysanthemum morifolium Ramat.）是世界范围内重要的经济观赏作物。这种典型的菊科（菊科）专性短日照（SD）多年生草本植物是研究开花的光周期控制的有用模型。然而，菊花复杂的杂合基因组和该物种的自交不亲和性阻碍了其园艺和生理特性的基础和应用研究。

分离和表征特定基因中的突变体对于剖析基础和应用研究的基因功能至关重要。在菊花基因组编辑的第一份报告中，TALENs 同时靶向6 个CmDMC1（减数分裂控制破坏）基因（Shinoyama等，  2020）。然而，序列特异性TALEN的设计很麻烦，并且所需的重组质粒很大。在这方面，CRISPR/Cas9介导的基因组编辑在载体设计和组装方面具有优势，特别是在靶向多个基因时，已广泛应用于许多生物体中。然而，高频率的嵌合事件和低编辑效率阻碍了CRISPR/Cas9系统在菊花中的应用(Kishi-Kaboshi et al .,  2017 )。只有一项研究报告使用传统 CRISPR/Cas9 介导的系统成功敲除菊花 ( C. morifolium ) 中的 TCP 转录因子基因（Li等人，  2022）。

野菊花（Chrysanthemum indicum L.）是栽培菊花的祖先，常被用作栽培菊花的模型，并且作为保健食品和抗炎中药已有2000多年的历史。在本研究中，我们选择二倍体C. indicum作为材料 (2 n  = 2 x  = 18)（图 S1），并首先靶向其单拷贝八氢番茄红素去饱和酶( CiPDS ) 基因。我们将基于CiPDS序列（图 1a、b）的 4 个单引导 RNA（sgRNA）克隆到 pDIRECT-22C 中，该 pDIRECT-22C 可以使用 Csy 型核糖核酸酶 4（Csy4）酶同时表达多个 sgRNA（Čermák 等， 2017  ））。我们通过高通量测序测量了C. indicum原生质体中每种 sgRNA 的编辑效率（数据 S1）。sgRNA1-4的编辑效率分别为8.22±1.2%、7.82±0.3%、4.02±0.4%和9.14±1.0%（图1b）。我们选择 sgRNA1 和 sgRNA2 来敲除CiPDS，因为它们具有最高的编辑效率并靶向第一个CiPDS外显子（图 1a）。

由于最佳启动子的选择对于高效基因组编辑非常重要，因此我们测试了五种不同的启动子来驱动Cas9和 sgRNA（图 1c）：花椰菜花叶病毒 (CaMV) 35S、Cestrum黄叶卷曲病毒 (CmYLCV)、欧芹( Petroselum Crisum )泛素( PcUbi )、拟南芥 YAOZHE ( YAO ) 和C. indicum 泛素( CiUbi ) 启动子。我们还根据拟南芥( AtCas9 ) 和印度C. indicum ( CiCas9 )的密码子使用对Cas9进行了密码子优化。我们在至少两次独立实验中测试了所得九个构建体的编辑效率（表S1）。总共，我们通过根癌农杆菌介导的转化用九种构建体转化了 7786 个C. indicum叶盘（图 1c）。含有35S:AtCas9和PcUbi:sgRNA 的构建体在所有实验中显示出最高的编辑效率，并被命名为 pDIRECT-Ci-opti（图 1c 和表 S1 中的构建体 5）。

我们通过 pDIRECT-Ci-opti 转化获得了八株白化植物，双等位基因编辑效率为 0.74%（图 1c）。一株白化植物表现出浓密的表型（图1d1），而大多数植物发育出正常的根和叶（图1e）。桑格测序显示，白化植物中的所有突变仅发生在 sgRNA1 靶位点，并且由缺失和插入组成（图 1e 和图 S2）。

在生成的 14 株白化植物中（图 1c），有 4 株表现出杂色表型（28.6%），表明嵌合水平较高（图 1d2、d3）。当使用CiPDS作为视觉标记来验证基因组编辑时，嵌合现象很容易进行评分，但对于大多数没有视觉突变表型的目标基因来说，评分既费时又费力。为了进一步提高筛选效率并解决嵌合问题，我们生成了表达eYGFPuv（35S：eYGFPuv ）的转基因C. indicum，其编码的蛋白质在紫外（UV）光下具有肉眼可见的明亮荧光（Chin等人，2015 ）。 ，  2018）并已被用作植物中基因表达和稳定转化的可见报告基因（Yuan等，  2021）。35S：eYGFPuv C. indicum植物的所有器官均表现出荧光（图 1f）。然后，我们用特定的 sgRNA 靶向eYGFPuv，将其用作可见标记，以有效筛选基因编辑植物，同时消除嵌合植物。我们将CiPDS sgRNA1 和 sgRNA2 以及eYGFPuv sgRNA1 和 sgRNA2 克隆到 pDIRECT-Ci-opti 中（pDIRECT-Ci-opti-CiPDS-eYGFPuv，图 1g），并将该构建体转化为35S：eYGFPuv叶盘。虽然来自35S : eYGFPuv C. indicum 的叶盘表现出明亮的荧光，但来自野生型 (WT) 的叶盘在紫外光下不表现出荧光（图 1h1）。用 pDIRECT-Ci-opti-CiPDS-eYGFPuv 转化后，一些35S : eYGFPuv衍生的愈伤组织缺乏 eYGFPuv 荧光，表明它们含有eYGFPuv突变（图 1h2）。此外，一些再生的非荧光小植株不是白化的（图1h3），表明这些小植株是eygfpuv双等位基因突变体，而不是Cipds双等位基因突变体。其他没有荧光的再生植株是白化的（图1h4），表明它们是eygfpuv Cipds双等位基因突变体。我们还根据 eYGFPuv 荧光鉴定了杂色小植株（图 1h5）。

采用传统的抗生素筛选方法，我们获得了50株抗潮霉素转基因植株。其中，我们鉴定了8​​个Cipds双等位突变体，双等位突变体筛选效率为16.0%（图1i）。我们还选择了 11 株没有 eYGFPuv 荧光的植物，它们是eygfpuv双等位基因突变体。其中，7名白化病患者，eygfpuv双等位基因突变体筛选效率为63.6%（图1i）。因此，使用突变的eYGFPuv作为可见标记提高了筛选效率并很大程度上消除了嵌合植物。

菊花的开花通常需要人工调节日长。菊花育种的主要目标之一是培育光周期不敏感或早花的菊花品种。TERMINAL FLOWER 1 (TFL1) 抑制菊花开花。过度表达TFL1会延迟菊花的开花时间（Gao等，  2019；Higuchi 和 Hisamatsu，  2015）。在这里，我们在C. indicum中鉴定了两个TFL1同源物：CiTFL1a和CiTFL1b。我们使用 CRISPR/Cas9 平台生成了三个Citfl1a双等位突变体和一个Citfl1b纯合突变体。三个Citfl1a双等位基因突变体在第一个CiTFL1a外显子中包含小的缺失和插入。Citfl1b突变体在两个基因拷贝的第三个CiTFL1b外显子中包含相同的 1-bp 插入（图 1j 和图 S3）。编辑后的植物表现出不同程度的早期开花，其中Citfl1a突变体表现出最早的开花（图 1k）。在 SD 条件下，WT 植物的花芽出现时间为 41.3 ± 1.2 天，而Citfl1a和Citfl1b植物的花芽出现时间分别为 16.7 ± 0.6 和 32.7 ± 1.2 天（图 1l）。

本研究开发的可见报告辅助CRISPR/Cas9系统应有助于菊花的研究和育种。