Design and Construction of Plasmids for Genome Editing in Monocots

The CRISPR/Cas9 system has been successfully used for genome editing in a variety of monocots, including rice (Oryza sativa), wheat (Triticum sp.), barley (Hordeum vulgare), and maize (Zea mays) (Feng et al. 2018; Gasparis et al. 2018; Hu et al. 2018; Kis et al. 2019; Okamoto et al. 2017; Wang et al. 2017; Zhang et al. 2016). To perform CRISPR/Cas9-mediated genome editing, the Cas9 endonuclease is guided by a single guide RNA (sgRNA) to recognize the complementary sequence and create double-strand breaks (DSBs), thereby generating a short deletion or insertion. Genome-edited plants can be generated either by stable or transient transformation. For stable transformations, the Agrobacterium-mediated transformation method is typically used to deliver transfer DNA (T-DNA) into the plant cell, where it is then inserted into the plant genome (Mikami et al. 2015a; Nandy et al. 2019). For transient transformations, particle bombardment and polyethylene glycol (PEG)-mediated methods are used to deliver plasmids or ribonucleoproteins (RNPs) into the plant cells (Woo et al. 2015; Svitashev et al. 2016; Zhang et al. 2016; Lin et al. 2018). Plasmids used for the stable genome editing of plants require a selection cassette, known as a sgRNA cassette, and a clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) cassette in the T-DNA region, while the selection cassette is not necessary for transient transformations (Fig. 1).

  1. 1.

    Selection cassette: In monocots, several genes have served as useful selection markers for the efficient selection of transgenic plants, such as neomycin-phosphotransferase (NPTII), bar, mutated acetolactate synthase (ALS), plant phosphomannose isomerase, and hygromycin phosphotransferase (Hpt) (Miki and McHugh 2004; Hu et al. 2016). Among these genes, Hpt is the most widely used selection marker which confers tolerance to the herbicide hygromycin because several crops have a natural tolerance to kanamycin. Plant phosphomannose isomerase can also be used as a selectable marker for rice transformation (Hu et al. 2016). The expression of Hpt is usually driven by a strong constitutive promoter, such as maize Ubiquitin 1 (ZmUbi1), rice ACTIN 1, or Cauliflower mosaic virus (CaMV) 35S, for the ubiquitous expression of the antibiotic-tolerance gene (Mikami et al. 2015a).

  2. 2.

    sgRNA cassette: sgRNA is a programmable 20-nucleotide (nt) sequence that recognizes the target DNA sequence and an invariant scaffold sequence (Ran et al. 2013), then directs the Cas nuclease to cleave the target sequence. Two factors are important for the function of sgRNA; promoter activity and the specificity of the sgRNA. A mixed dual promoter system is generally used in CRISPR/Cas9 system. In this system, Cas9 is directed by the RNA Polymerase II (Pol II) promoter while sgRNA expression is regulated by a Pol III promoter such as U6 or U3. To increase the transcription of the sgRNA, several monocot-specific U3 or U6 promoters have been cloned and used to direct the expression of the sgRNA, such as those from rice (Ma et al. 2015), maize (Qi et al. 2018), and wheat (**ng et al. 2014). In rice, the sgRNA driven by the OsU6 promoter produces more transcripts than when driven by the OsU3 promoter (Mikami et al. 2015a); however, several promoters, including OsU3, OsU6a, OsU6b, and OsU6c, have been used to direct the expression of sgRNA, and all of them could effectively direct genome editing with mutation rates of 81.4–90.0% (Ma et al. 2015; Shan et al. 2013; **e and Yang 2013; Zhou et al. 2014). In addition to promoters, the specificity of the sgRNA for its target DNA sequence is another factor that affects the efficiency of genome editing. Several resources available on the web can be used to design highly specific sgRNAs for use with the CRISPR/Cas9 system (Table 1; modified from Zhang lab, https://zlab.bio/guide-design-resources). Nevertheless, even if these 20-nt sequences perfectly match the target gene, some sgRNAs do not work well. The online tools listed in Table 1 recommend targets with a low risk of an off-target match, but not all predicted target sequences may result in an efficient mutation. In addition, targeted DNA sequences with GC contents higher than 50% have higher genome-editing efficiencies (88.5–89.6%) than those with GC contents lower than 50% (77.2% efficiency) (Ma et al. 2015). Successive Ts in 20-nt target sequence is not good when sgRNA expression is driven by the U3 or U6 promoters (Wu et al. 2014).

  3. 3.

    Cas9 cassette: Aspects of the Cas9 cassette that affect the mutation rate during genome editing include the expression level and codon usage of Cas9 (Ma et al. 2015; **e et al. 2015). Several strategies have been conducted to improve the expression of Cas9, including the use of a strong constitutive promoter, the addition of a translational enhancer, and the addition of nuclear localization signals. Constitutive promoters have been used to direct the expression of Cas9, including the ZmUbi1 and 35S. The Cas9 nucleases from different bacteria may have variations in the protospacer adjacent motif (PAM) sequence that they require for cleavage; therefore, many Cas9 homologs with different PAM requirements have been isolated from different bacteria. These Cas proteins are introduced in ‘Alternatives to SpCas9’ section. Streptococcus pyogenes Cas9 (SpCas9) is the most common Cas used in CRISPR/Cas9-mediated genome editing. It has been codon-optimized for maize (**ng et al. 2014) and rice (Miao et al. 2013) to improve its expression levels in these monocots.

Fig. 1
figure 1

Schematic of the T-DNA region in a binary vector for genome editing in monocots. PPol III: Polymerase III promoter. PPol II: Polymerase II promoter. PUbi1: maize Ubiquitin 1 promoter and the first exon. Hpt: Hygromycin phosphotransferase. (Modified from Howells et al. 2018)

Full size image
Table 1 Web-based tools for sgRNA design
Full size table

Strategies for Multiplex Genome Editing in Monocots

One advantage of the CRISPR/Cas9 system over other crop-breeding strategies is its flexibility for multiplex genome editing (Wang et al. 2017; Wang et al. 2018). The editing of multiple functional genes allows for the rapid improvement of multiple agronomic traits at one time, while editing the cis-acting elements of a promoter affects transcriptional regulation. The deletion of larger fragments between two sgRNA-targeted sites on the same chromosome following the generation of multiple DSBs has been reported in many species. The CRISPR/Cas9 system has been used to delete DNA fragments ranging from dozens of bases to greater than 1 Mb (Mali et al. 2013; Shan et al. 2013). In addition, targeted deletions of 10 bp to over 200 kb between two target sites have been reported in rice (Mikami et al. 2016; Zhou et al. 2014). The elimination of the Tos17 retrotransposon using CRISPR/Cas9 was reported in rice, providing a rapid breeding route for making reverting the agronomically important genes that have been inactivated by the insertion of transposable elements (Saika et al. 2019).

Multiplex genome editing can be achieved by the simultaneous delivery and expression of multiple sgRNAs; however, since most CRISPR/Cas9 components are transferred into plants via Agrobacterium-mediated transformations, an efficient plasmid construction strategy is required. Traditionally, multiple sgRNA expression cassettes (including a Pol III promoter, a sgRNA, and a terminator) can be stacked into one T-DNA (Fig. 2a); however, this may increase the cloning difficulties due to the limited restriction sites available, and the fact that large T-DNAs may decrease the transformation efficiency. Alternative strategies have therefore been developed to facilitate multiplexed genome editing in plants. These alternative strategies are based on the expression of multiple sgRNAs as a single transcript, after which multiple functional sgRNAs are generated following the processing of the transcripts by exogenous ribozymes (Gao and Zhao 2014), Csy-type ribonuclease 4 (Csy4) (Cermak et al. 2017), or the plant endogenous transfer RNA (tRNA)-processing system (**e et al. 2015). The endogenous tRNA-processing system exists in almost all organisms and has been successfully used to perform multiplex genome editing in rice (**e et al. 2015). In the tRNA-processing system, the Pol III promoter is used to direct the expression of a single synthetic gene containing multiple tRNA-sgRNAs or a polycistronic tRNA-sgRNA (PTG) gene. The PTG gene can be generated using the Golden Gate assembly method (Lowder et al. 2015) (Fig. 2b). After transcribing the PTG, the endogenous RNases specific to tRNA would recognize the tRNA components and cleave the individual sgRNAs from the PTG transcript. The resulting sgRNAs would then guide Cas9 to multiple target sites for genome editing. Many toolboxes are currently available in the public database Addgene (Cermak et al. 2017; Castel et al. 2019; Hahn et al. 2019), from where these vectors can be purchased by academic researchers.

Fig. 2
figure 2

Diagram of a plasmid construct used for multiplex genome editing. a Stacking of multiple sgRNA expression cassettes in one T-DNA. P Pol III: Polymerase III promoter. b Cloning of four tRNA-gRNAs into the CRISPR/Cas9 binary vector using a single Golden Gate ligation. (Modified from Kurata et al. 2018).

Type V Cas Systems

The CRISPR RNA (crRNA)-guided SSN activity of the type V Cas system, which includes Cas12a and Cas12b, also generates DSBs. Several Cas12a (formerly Cpf1) orthologs, such as F. novicida Cas12a (FnCas12a), Acidaminococcus sp. Cas12a (AsCas12a), and Lachnospiraceae sp. Cas12a (LbCas12a), have been engineered as a class of genome-editing tools distinct from the Cas9 system (Zetsche et al. 2015). The Cas12a proteins induce staggered DSBs at sites distal to a 5′ T-rich PAM, generating relatively longer deletions via the non-homologous end joining repair pathway in various plant species, including rice (Endo et al. 2016a; Begemann et al. 2017; Hu et al. 2017; Kim et al. 2017c; Tang et al. 2017; Xu et al. 2017). Similar to the SpCas9 systems, transgene-free Cas12a-mediated mutants can be generated by simply segregating the T-DNA fragment in the transition from the T0 to T1 generations (Xu et al. 2017). In addition, mixing RNP with Cas12a and crRNA has allowed transgene-free genome editing in soybean and tobacco (Kim et al. 2017b). Interestingly, the Cas12a proteins have different sensitivities to temperature in plants (Malzahn et al. 2019); therefore, the optimization of the incubation temperature used in crop culture systems may facilitate a higher editing efficiency when using Cas12a proteins.

Cas12a proteins not only induce DSBs, but also process their own pre-crRNA for maturation. This RNase activity provides a simple pattern by which single or multiplex crRNA(s) can be expressed from the same transcription unit of Cas12a, facilitating the highly efficient genome editing of plants (Wang et al. 2017; Ding et al. CRISPR-Edited Protoplast Regeneration

In addition to the validation of transformation efficiency, mutated protoplasts have the advantage of being able to regenerate into entire mutant plants. Protoplasts isolated from meristematic tissues or totipotent cells were first used for plant regeneration in the early 1970s (Takebe et al. 1971), and just a few years later, researchers used protoplasts as materials for plant transformation (Marton et al. 1979). The progeny of N. tabacum regenerated from transformed protoplasts displayed the mutant phenotype, indicating that the transgenic tumor markers (octopine and nopaline) were inherited through meiosis (Wullems et al. 1981a; Wullems et al. 1981b). Monocot protoplast regeneration (Abdullah et al. 1986; Fujimura et al. 1985; Rhodes et al. 1988a) and transformation (Rhodes et al. 1988b; Shimamoto et al. 1989; Toriyama et al. 1988) protocols have also been established.

Cas proteins and sgRNAs are sufficient for CRISPR/Cas genome editing and are no longer required once the genes have been edited. Transient expression or the direct delivery of sgRNAs and Cas proteins into the cells is sufficient for editing; therefore, the DNA encoding the Cas proteins and sgRNAs does not need to be integrated into the genome for their continued expression, making the plants regenerated from these edited cells transgene-free. In 2015, Prof. **-Soo Kim’s group published a milestone article using protoplasts (Woo et al. 2015), in which RNP was used as the genome-editing reagent to edit lettuce (Lactuca sativa) protoplasts, which were subsequently regenerated into transgene-free plants.

There are several advantages to generating transgene-free edited crops using protoplasts:

  1. 1.

    Protoplast transformation can be applied to edit hybrid and long-juvenile-phase crops, which are typically propagated using vegetative methods. Traditionally, the use of Agrobacterium tumefaciens-mediated or other stable transformation techniques means the transgene (selection markers, sgRNA, and Cas9) must be integrated into the genome. In inbred crops, such as rice, this transgene can be removed through crossing; however, this causes the segregation of the desired traits in the offspring. This separation issue also occurs in systems with a low editing efficiency, in which the transformants must be crossed in order to obtain the homozygous genotype. By contrast, homozygous edited crops can be achieved from heterozygous edited protoplasts using a second transfection (Hsu et al. 2019). Several methods are available for achieving this goal, including the delivery of RNP into the callus using a biolistic approach (Liang et al. 2017, 2019) or the transient expression of Cas9 and sgRNA using an Agrobacterium-mediated transfection (Chen et al. 2018); however, the gene-editing/mutation efficiencies of these methods are low (less than 10%). The protoplast transformation strategy had a higher efficiency; for example, over 50% of tobacco protoplasts were mutagenized using this technique, which was similar to the numbers transformed during the efficiency evaluation without antibiotic selection.

  2. 2.

    Protoplast transformation can deliver more than one reagent. The current Agrobacterium vectors used for stable transformations are multiplex sgRNAs. The genes encoding Cas proteins are large, and no multiplex Cas protein vectors are currently available. Three different Cas protein vectors (SaCas9, FnCas12a, and nCas9-Target-AID) can be delivered into a single protoplast to edit three different target sites without interference (Hsu et al. 2019). The introduction of these three Cas protein vectors requires three subsequent Agrobacterium-mediated transformations, the crossing of individual mutants, or the co-transformation of Agrobacterium harboring independent Cas9s and different selection markers.. In addition to sgRNA and Cas proteins, donor DNA fragments are also required for knock-in genome editing, and must be delivered using Agrobacterium (Endo et al. 2016b; Miki et al. 2018;, Wolter and Puchta 2019). Using these protocols, only a few copies of donor DNA can be used in genome editing, which may be one of the reasons for the low efficiency of knock-in genome editing. Although multiple T-DNAs can be delivered into plant nuclei to increase the editing efficiency, this also increases the difficulty of removing the T-DNAs to produce marker-free plants. In protoplasts, however, the donor DNA can be delivered in microgram quantities, which may help increase the knock-in efficiency. Not only can the sgRNAs, Cas proteins, and donor DNAs be more easily delivered into protoplasts, but the application time and amounts of genome-editing reagents can be controlled to improve the editing efficiency.

  3. 3.

    Using RNPs, the issues of promoter and codon modification in different species can be solved. The original Cas proteins were obtained from microorganisms; therefore, their codons must be modified for vector construction in plant species, and their expression must be driven by plant promoters. These issues have been investigated in Arabidopsis, resulting in the identification of a T-DNA architecture causing homozygous mutations in the first generation after transformation (Castel et al. 2019); however, the transcription and translation of these genes can still be problematic in different target crops. The resulting amount of Cas proteins in the cell can cause a low editing efficiency; however, in protoplasts, RNPs can be used to solve this problem and provide a higher delivery efficiency for transfection.

  4. 4.

    With the exception of Arabidopsis, most Agrobacterium-mediated transformation protocols were performed using a tissue culture platform. In dicots, the regenerated plants were derived using organogenesis, meaning they were derived from multiple cells. Many edited transformants are therefore chimeric (Kaya et al2016). To validate their edited sequences, the transformant DNAs were sequenced. Diploids contain two alleles in each cell; therefore, if the transformants contained more than two alleles they were determined to be chimeric (Kaya et al. 2016). If the edited alleles are not present in the reproductive organs, the alleles cannot be passed on to the progeny. In contrast, protoplasts are single cells that are edited before the first cell division is completed. The regenerates are then derived from a single edited protoplast, meaning all cells have same genomic background, which enables the edited alleles to be transmitted to the next generation. In our previous studies, non-chimeric regenerates were derived from protoplasts edited using the Cas proteins Cas9, Cas12a, and Target-AID, and the genotypes were inherited in a Mendelian manner (Hsu et al. 2019). This phenomenon was also reported in lettuce (Woo et al. 2015).

RNPs (Andersson et al. 2018; Woo et al. 2015) and plasmids (Andersson et al. 2017; Zong et al. 2018) have been delivered into lettuce and potato (Solanum tuberosum) protoplasts, which were subsequently regenerated into transgene-free genome-edited plants. In our lab, we established N. tabacum (Hsu et al. 2019; Lin et al. 2018), rapid cycle brassica (Brassica oleracea), wild tomato (Solanum peruvianum), and N. benthamiana (Lin et al. in preparation) protoplast regenerations, and used these systems to establish RNP and plasmid DNA gene-editing platforms (Hsu et al. in preparation). Some issues are yet to be resolved in protoplast regeneration during genome editing, however:

  1. 1.

    The regeneration protocol is difficult to establish. Only a few protocols for genome editing and protoplast regeneration have been developed, all of which were achieved in dicots (Andersson et al. 2017, 2018; Hsu et al. 2019; ** et al. 2019; Lin et al. 2018; Tuncel et al. 2019; Woo et al. 2015). Many protoplast regeneration protocols are available in other species, including rice (Shimamoto et al. 1989; Toriyama et al. 1988) and other important Poaceae species (Rhodes et al. 1988a, 1988b). Recent discoveries have elucidated the mechanisms by which plants can regenerate, which could be applied to improve protoplast regeneration in the future (Lowe et al. 2016). In rice, it is possible to solve the lack of an efficient protoplast regeneration protocol using an alternative method; Cas9-sgRNA RNPs could be directly delivered into rice zygotes, which can then be cultured into mature plants (Toda et al. 2019). A total of 14–64% of the transgene-free plants obtained using this method were found to contain the target mutations.

  2. 2.

    Unexpected mutations can occur during protoplast regeneration. When 15 protoplast potato regenerants were sequenced, they were found to include a variety of mutations, including insertions/deletions, chromosome rearrangements, and aneuploidy (Fossi et al. 2019). Indeed, somaclonal mutations occur in all tissue culture strategies, including micropropagation and somatic embryogenesis. In our experience using bamboo, somaclonal mutations occurred after a long-term subculture (Lin and Chang 1998; Lin et al. 2006, 2007; Liu et al. 2019); rather, they simply result in more than one edited transformant for each transformation. In addition, mutation breeding is a strategy used in traditional breeding. We can select the transformants with good traits and use these edited lines as parental lines during crop production. In our opinion, although mutations can occur during regeneration, the protoplast regeneration techniques currently available are useful tools for transgene-free genome editing.

Screening for Marker-Free CRISPR/Cas9-Induced Mutants

Genome editing is widely used produce new genetic variants in plants. Several approaches for creating genome-edited crops have been developed, including the CRISPR/Cas system, which can be used with various tissue types including protoplasts (as described in the previous section), callus, leaf discs, and germline cells. Agrobacterium-mediated, PEG-mediated, particle bombardment, and virus infection transformation methods are commonly used for the delivery of the CRISPR/Cas system into plants, resulting in stable or transient expression patterns.

The transient expression of the CRISPR/Cas system can deliver DNA-based sgRNA and Cas9 RNP sequences or the proteins themselves (Chen et al. 2018; Zhang et al. 2016). In a DNA-based CRISPR/Cas9 system, recombinant DNA can be transferred to the plant using Agrobacterium- or PEG-mediated transformations or particle bombardment, eliminating the need for herbicide or antibiotic selection steps. This method allows the expression of the CRISPR/Cas9 plasmid without requiring its integration into the plant genome. This approach can reduce regeneration time via tissue culture while producing mutation frequencies similar to the stable expression of the CRISPR/Cas9 vector at the target site. More than 86% of wheat T0 mutants generated using this technique were transgene free (Zhang et al. 2016). In the Transgene-free CRISPR/Cas9 system, RNPs are assembled in vitro and directly delivered into the protoplasts using a PEG fusion approach. The RNP complex directly targets the recognized sequences and induces DSBs (Park and Choe 2019; Woo et al. 2014), high-resolution melting analysis (HRM) (Thomas et al. 2014), PCR- and labeling-based method (Biswas et al. 2019), and annealing at critical temperature PCR (ACT-PCR) (Hua et al. 2017).

Table 2 Advantages and disadvantages of different methods for CRISPR/Cas9-mediated mutant screening. (Adapted from Bao et al. 2019)
Full size table

The CAPS, indCAPS, and T7E1 cleavage assays for identifying gene-edited mutants are based on an enzymatic approach. Wild-type and mutant sequences are amplified using PCR then subjected to enzyme digestion. A typical CAPS assay can be used if a RE cutting site is present at the CRISPR target site, which is disrupted once the sequences are mutated; however, RE sites are not always present within target regions. A similar dCAPS assay has been developed for genome regions not possessing different RE sites between the wild type and the mutant. This dCAPS assay introduces or disrupts a RE site near the mutation site by amplifying several mismatched nucleotides, which greatly increases the flexibility for selecting the sgRNA target sites. T7 endonuclease 1 can be useful for digesting the mismatched heteroduplexes formed between the wild-type and mutated strands; however, this method lacks the sensitivity to distinguish between homozygous mutants and the wild type.

Other PCR-based analyses include PAGE, HRM, ACT-PCR, and PCR−/labeling-based assays. To distinguish genome-edited mutations from wild-type target sequences, target region amplicons are migrated on a native polyacrylamide gel. Homoduplex DNA migrates faster than heteroduplex DNA. To distinguish homozygous mutants from the wild type, unknown amplicons can be pre-mixed with wild-type amplicons before the assay. The migration pattern of the pre-mixed homozygous mutant and wild type will be similar to that of the heterozygous mutant. HRM is a fluorescence-based technique for determining the differences in the melting temperatures of heteroduplex and homoduplex DNA fragments. This technique can detect differences as small as 0.1 °C; however, its sensitivity is influenced by the amplicon length and variation of the mutated sequence. ACT-PCR assay can also distinguish homozygous mutant sequences at the target site using optimal annealing temperature and specific primers; however, the assay requires the design of specific primers and is time consuming and labor intensive. A simple PCR- and amplicon labeling-based method was recently used to identify the CRISPR/Cas9-generated mutants in rice (Biswas et al. 2019). This approach requires two pairs of primers or a FAM-labeled allele-specific primer. The sensitivity, precision, and reliability of the FAM-labeled method allows for the detection of indels with a high sensitivity (down to ±1 bp).

Although all these PCR-based analyses enable the effective, accurate, and economical screening of CRISPR/Cas9-generated mutants, the identification of sequence changes resulting from SDN1-generated mutations requires the Sanger sequencing of amplicons generated from the target region. In addition, whole-genome sequencing is a powerful tool for identifying not only on-target and off-target mutations, but also transgene-free plants produced by genome editing. The only drawback of whole-genome sequencing is its cost and time requirement.