Breeders have contributed to altering animal and plant genomes for years through traditional hybridization and selection to produce new genetic combinations for improved commodity development. This process, however, can take years (Orton, 2020). To overcome this problem, engineered nucleases such as zinc finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs) were discovered to effectively carry out specific genetic changes in plant genomes (Carroll, 2011; Nemudryi et al., 2014). Despite their ability to carry out site-specific gene editing, engineering through ZFNs and TALENs is time-, effort-, and cost-consuming. Clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated nucleases such as Cas9 have emerged recently as breakthrough tools for genome editing studies (Doudna and Charpentier, 2014).

First reported in 2011, these molecular scissors are now widely used in labs to make precise cuts in genomic DNA (Zhu et al., 2020). CRISPR was first discovered in the genome of Escherichia coli in 1987, where the interspaced sequences in the CRISPR are derived from past genome invasions (Ishino et al., 1987). There are three main participants in the CRISPR-Cas9 machinery: CRISPR-RNA (crRNA), trans-activating crRNA (tracrRNA), and a Cas nuclease protein. The transcription of the CRISPR loci, along with the tracrRNA(s) and Cas nuclease protein, acts as an adaptive bacterial defense against viral attacks. As its name states, the CRISPR locus has multiple palindromic repeats with spacer DNA between the repeats (Jiang and Doudna, 2017). To facilitate the cloning procedure and improve genome editing efficiency, scientists have engineered dual tracrRNA:crRNA as a single guide RNA (sgRNA) where a 20-nucleotide single-target matching sequence makes up the 5’ end and a secondary structure at the 3’end helps for Cas9 binding (Doudna and Charpentier, 2014). The double-stranded DNA break generated by CRISPR-Cas9 triggers the DNA repair mechanism. DNA repair by NHEJ can result in errors in the target gene, often resulting in deletions or insertions of bases, which can cause a coding frameshift, rendering the gene inactive (Jeggo, 1998). This technique is widely used today by laboratories in studying gene knockout phenotypes in a wide range of organisms. Plants are routinely edited for genes of interest in order to study knockout phenotypes without substantial off-target mutations, which can be associated with random mutagenesis approaches such as TILLING (Kurowska et al., 2011; Zaidi et al., 2020).

Despite the ability of CRISPR-Cas9 to efficiently induce desirable mutations, as long as the target sequence is present, the existing CRISPR-Cas9 complex can still induce additional mutations. This includes the possibility of additional insertions or deletions that could potentially revert the initial lesion back to a (near)-wild-type state. Thus, it is important to remove the CRISPR- Cas9 complex-containing T-DNA from the plants in order to ensure stable inheritance of the desired induced mutation (Gao et al., 2016). In addition, the prolonged presence of the CRISPR- Cas9 complex can cause potential off-target mutagenic activity, resulting in undesirable or unknown genomic mutations. The presence of multiple orthologs of a gene due to genome duplication can also increase the risk of off-target gene editing by the continued presence of the CRISPR- Cas9 complex (Zhang et al., 2018; Hahn and Nekrasov, 2019).

Removing the CRISPR-Cas9 complex from plant genomes can be achieved through several methods (Gao et al., 2016; He and Zhao, 2019; Kang et al., 2023; Yang et al., 2023). However, some of these techniques face challenges regarding their acceptability across different plant types. For example, the use of visual selectable markers such as the fluorescent protein DsRed to identify Cas9-transgene-free individuals may be difficult due to autofluorescence or in plants harboring a previously introduced fluorescent marker. In addition, the color or translucency of the seed coat may block the visualization of a fluorescent marker. A recently developed approach is to use grafting where the CRISPR- Cas9 complex can be transported from a transgenic rootstock to wild-type scions (Yang et al., 2023). Although this technique can result in transgene-free gene-edited seeds, it can only be reliably used in dicotyledonous plants.

Here, we report a novel but simple strategy to obtain Cas9-free mutants with a stable inheritance in Arabidopsis thaliana (Arabidopsis). In this approach, we used a Cas9+gRNA in a vector with a phosphinothricin acetyltransferase gene (blpR), conferring resistance to Basta^® (glufosinate ammonium). Our approach to removing the CRISPR-Cas9 machinery from successfully gene-edited Arabidopsis plants is based on rescuing non-transgenic Arabidopsis seedlings from Basta-selection media. As a proof of concept, we targeted sites in the At-Hook containing nuclear Localized 26 (AHL26) gene using a CRISPR-Cas9 complex. AHL26 is part of the AHL gene family, which features a conserved AT-hook motif at their N-terminal end and a plant and prokaryote conserved (PPC) domain, also known as the domain of unknown function #296 (DUF296), at their C-terminal end. Our lab routinely works with this family of genes to study seedling development and flowering time (Street et al., 2008; Zhao et al., 2013, 2014; Favero et al., 2016; Tayengwa et al., 2020; Jacques et al., 2022). After obtaining CRISPR-generated AHL26 mutants, we grew the segregating generation of Cas9 mutants on Basta selection and rescued and sequenced the non-transgenic offspring lacking the Cas9 construct. Once identified, we screened Cas9-free mutants for two more generations to demonstrate the heritability of the mutation. Using this strategy, we were able to isolate stable Cas9-free mutants in Arabidopsis for the AHL26 gene. We also demonstrate how the principles of this technique can be applied to field crops such as canola and wheat.

In this study, we report a technique to obtain Cas9-transgene-free Arabidopsis mutants and its potential application in other crops such as wheat and canola. This method is based on the ability of some plants to recover from Basta selection. Basta herbicide has an active ingredient, glufosinate-ammonium (Hoerlein, 1994; Takano and Dayan, 2020) with a mode of action that inhibits the glutamine synthetase enzyme. Glutamine synthetase plays a role in nitrogen metabolism in plants, and the inactivation of this enzyme results in the disruption of multiple metabolic pathways, including photosynthesis (Krogmann et al., 1959; Wendler et al., 1990; Wild and Wendler, 1993; Dan Hess, 2000; Dayan et al., 2019). The inhibition of glutamine synthetase results in the accumulation of toxic amounts of ammonia in the chloroplast (Bernard and Habash, 2009; Takano and Dayan, 2020). Plants exposed to this herbicide stop growing, resulting in death approximately two weeks later. Our lab routinely uses the phosphinothricin acetyltransferase gene (blpR) as a positive selection marker in plasmids for selecting transgenic plants which have been exposed to Basta. The blpR gene in transgenic plants confers resistance to glufosinate-ammonium by acetylating it to create an inactive conjugate, which allows plants to grow on growth media with the Basta selection agent (Block et al., 1987).

During our studies with transgenics plants expressing the blpR gene, we observed the ability of the non-transgenic plants to recover their growth once the plants were moved to media without the Basta selection agent. Since Basta doesn’t directly disrupt the photosynthesis system, the absence of the herbicide allowed some plants to regain their normal growth. Non-transgenic Arabidopsis seedlings growing on Basta media display stunted growth with discolored leaves. We observed 100 percent growth recovery once the affected seedlings were moved to normal growth media ( Figure 1D ). This method is consistent with findings in a field where injury to plants due to herbicide application can be corrected by removing the herbicide contaminated soil with clean soil (Mathers, 2006). The recovered seedlings were then able to grow normally in soil under greenhouse conditions. In contrast, Arabidopsis seedlings grown on kanamycin or hygromycin selection media could not recover their growth after being transferred to non-selection growth media ( Figure 1D ). Basta, being an herbicide, differs from kanamycin and hygromycin in the mode of action as both antibiotics interfere with protein synthesis in the chloroplast, causing permanent damage to plants (Davies et al., 1965; Weisblum and Davies, 1968; Davis, 1987; Hoerr et al., 2016). It is important to clarify that the purpose of our study is not to compare the effectiveness of different selection agents against each other. Instead, we aimed to determine which selection agents allow seedlings to be successfully rescued after selection. By testing Basta, kanamycin, and hygromycin, we sought to identify whether our seedling rescue method could be applied across different selection agents. Our findings show that this recovery phenomenon is specific to Basta, suggesting that the method is most effective when using Basta as the selection marker.

Non-transgenic siblings are plants from the same genetic background as transgenic plants but do not contain transgene of interest. These plants can be obtained by growing the progeny of a heterozygous transgenic plant on selection media to identify sensitive non-transgenic siblings. These siblings can serve as a control group, allowing for more accurate comparisons by minimizing genetic variation unrelated to the transgene. The recovery of Basta-sensitive seedlings from Basta-selection media provides an opportunity to easily identify non-transgenic siblings in plant transformation studies. The isolation of non-transgenic siblings not only allows to select Cas9-transgene-free mutants in Arabidopsis, as reported in this study, it also offers researchers several advantages in experimental design and result interpretation. For example, comparing transgenic plants to closely related non-transgenic siblings is particularly important when studying traits that may be influenced by minor genetic variations caused by the transformation process. Another more labor-intensive approach for identifying non-transgenic siblings in plant transformation studies involves self-pollination of heterozygous transgenic plants or backcrossing to the wild type, which is then followed by PCR assays or transgene selection analysis in paired sibling populations.

While other methods isolating Cas9-free mutants have been described, such as fluorescent-based isolation or grafting, these systems may not be compatible with all plant species or varieties (Gao et al., 2016; Yang et al., 2023). For example, although DsRed fluorescence selection offers a simple alternative for selecting Cas9-free plants, its application is limited to seeds with translucent seed coats, such as Arabidopsis and camelina. Plant seeds with thick seed coats, such as canola and wheat, limit the detection of DsRed fluorescent signals. In addition, grafting presents a practical challenge with the need for special equipment along with skilled personnel. It also has a potential limitation in monocots due to the lack of a cambium layer of cells. The ease of identifying and isolating non-transgenic plants using our method from the progeny of CRISPR-Ca9 generated mutants offers a simple and reproducible alternative for isolating Cas9-free mutants.

Although our method offers a simple alternative to isolate Cas9-free Arabidopsis plants, the method only allows the isolation of T-DNAs that still contain a selection marker gene. It does not track unknown T-DNA pieces inserted into the plant genome, the isolation of which is necessary to make true non-transgenic plants. Thus, if true non-transgenic plants are the goal, a whole genome sequence will be necessary to determine the presence of any such foreign DNA elements in any of the methods used for transgenes selection such as DsRed. The efficiency of mutation is highly dependent on the promoter driving the CRISPR gene. In our study, the mutation efficiency of the CRISPR system used was relatively low. However, recent advancements in CRISPR systems, such as those described by (Tsutsui and Higashiyama, 2017), have achieved mutation efficiencies exceeding 90%. With these more efficient CRISPR systems, our method could be significantly more effective, increasing the likelihood that non-transgenic segregating plants from mutant lines would harbor the desired mutation.

In addition to Arabidopsis, we investigated the recovery of field crops like camelina, canola, and wheat from Basta selection media. Based on our defined recovery criteria, camelina failed to recover after being moved from selection to non-selection media. The lack of recovery in camelina from Basta media suggests that our method may be limited in effectiveness for crops other than Arabidopsis, canola, and wheat. Fortunately, DsRed system offers a reliable transgene selection method for camelina (Sharma Koirala and Neff, 2020) with proven efficacy in isolating Cas9-free mutants within this species (Gao et al., 2016). The response of non-transgenic canola to Basta selection involved growth slowdown followed by leaf yellowing, with varying recovery rates upon transplantation to normal growth media. Although canola seedlings had a lower recovery percentage in soil than Arabidopsis, more seedlings exhibited green hypocotyls, suggesting potential for tissue culture approaches. Conversely, wheat had a higher recovery percentage in both media and soil, most likely attributed to its tightly packed crown structure and higher regenerative ability compared to canola (Huang and Taylor, 1993; Shang et al., 2021). These findings underscore the versatility of our Basta recovery methods across different crops, each presenting unique challenges and advantages. For our camelina, canola, and wheat experiments, we did not include a Basta control due to the unavailability of transgenic lines for these plants. As a result, further testing is required to fully assess the effectiveness of the Basta recovery method in eliminating trans-mutagenic CRISPR-Cas9. However, we anticipate that the recovery observed in this study could be applicable to segregating transgenic plants as well.