Efficient CRISPR/Cas9‐based plant genomic fragment deletions by microhomology‐mediated end joining

Clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein (Cas) systems produce double strand breaks (DSBs) in targeted sites and generally cause insertion/deletions of one or several bases by non-homologous end joining (NHEJ)-mediated DNA repair. However, many edited cell lines with small insertions/deletions produce abnormal transcripts or proteins causing unexpected effects that complicate functional analysis (Tuladhar et al., 2019).

Clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein (Cas) systems produce double-strand breaks (DSBs) in targeted sites and generally cause insertion/ deletions of one or several bases by non-homologous end joining (NHEJ)-mediated DNA repair. However, many edited cell lines with small insertions/deletions produce abnormal transcripts or proteins causing unexpected effects that complicate functional analysis (Tuladhar et al., 2019). Therefore, the ability to delete larger genomic fragments would have advantage for making true null alleles of coding and non-coding genes. The NHEJ pathway also produces fragment deletion with low efficiency (Pathak et al., 2019). Microhomology-mediated end joining (MMEJ) is another DNA repair mechanism based on recombination between microhomologous sequences (MHSs) of 4-25 bp located near DSBs (Decottignies, 2007). Studies in model organisms including plants showed that MMEJ usually results in deletions of sequences between pairs of MHSs near DSBs generated by genome editing (Pathak et al., 2019;Shen et al., 2017). The MMEJ approach has been explored for deletion or insertion of genomic fragments in animals and microorganisms (Bae et al., 2014;Nakade et al., 2014;Zhang et al., 2016), but this type of application as well as systematic investigation of the frequency of MMEJ events in plant genome editing is lacking.
In addition, generation of fragment deletions could remove antibiotic or herbicide selectable marker genes from transgenic lines for minimizing public and regulatory concerns. Several strategies for obtaining marker-free transgenic plants have been reported (Perez and Angenon, 2013), but these methods require the use of a marker-free system for plant transformation. Therefore, a high-efficiency method for removing target DNA fragments in plants is needed.
Here, we analysed the efficiency of CRISPR/Cas9-targeted deletion of genomic fragments mediated by MMEJ in rice (Oryza sativa L.) and demonstrated this approach as an effective tool for deleting marker genes from transgenic plants. Using the CRISPR/ Cas9 multiplex genome editing system for plants (Ma et al., 2015), our laboratory produced hundreds of rice mutant lines. A proportion of these lines were edited at two or more sites within the target genes. Although we did not consider MHSs for MMEJ in our initial experimental design, we selected 48 edited populations (each representing a group of T 0 plants generated with the same CRISPR/Cas9 construct) that had fragment deletions between the target sites, and determined whether some deletion events were mediated by MMEJ (two MHS sites recombined as one in the deletion junctions) or by NHEJ (no MHSs in the deletion junctions). Of these edited populations, 24 had MMEJ-mediated fragment deletions ranging from 28 to 20 667 bp with an average size of 1653 bp. The average fragment deletion efficiency was 18.9% ( Figure 1a). Moreover, 43 edited populations had NHEJ-mediated deletions ranging from 25 to 21 000 bp with an average size of 948 bp. Thus, if the design of editing targets does not consider facilitating MMEJ, the frequencies of fragment deletions induced by MMEJ and NHEJ are similar, but the sizes of MMEJ-mediated deletions are somewhat larger.
We next attempted to induce MMEJ to delete a marker gene from an existing transgenic line. We previously generated a transgenic line carrying a homozygous, single-copy T-DNA insertion of a transgene carrying MONOSACCHARIDE TRANSFER-ASE 3 (MST3). The MST3 transgene was in a pCAMBIA1300based binary construct with the hygromycin-resistance (HPT) gene cassette ( Figure 1b). To delete this HPT cassette using genome editing, we first determined the sequences flanking the T-DNA insertion by the modified high-efficiency thermal asymmetric interlaced PCR (mhiTAIL-PCR) . We then used our online microhomology-finder tool (http://skl.scau.edu.cn/repf inder/) to analyse the sequences flanking the HPT cassette in the T-DNA and the inserted genome to identify MHSs. This analysis identified four groups of MHSs (MHS1-4) of 5-7 bp in the flanking sequences (Figure 1b,c). To produce DSBs as close as possible to the MHSs for efficient MMEJ-mediated fragment deletion, we used the web-based tool CRISPR-GE (http://skl.sca u.edu.cn/) to design three target sites (T1-T3) that were located near any pair of the MHSs, including two target sites on the genomic flanking sequence and one target site on the T-DNA flanking sequence (Figure 1b,c).
To further determine whether these deletions were mediated by MMEJ or NHEJ, we examined the deletion-junction sequences. To this end, all the PCR-amplified fragments with reduced sizes and full length (Figure 1d) were recovered and sequenced. In the three hoM plants, the HPT cassette was precisely excised with deletionjunction events between the pair of MHS1, MHS2 or MHS3 (Figure 1e,f). Of the 11 heM plants, 8 had deletion-junction events between pair of MHS1, MHS 2 or MHS 4 (Figure 1f). The remaining 3 heM plants (#3, #5 and #9) had fragment deletions generated by NHEJ, with junction sites between the targeted nonhomologous DSB ends. In these heM plants, sequencing of the fulllength fragments showed the presence of the HPT cassette but with various base insertion/deletion mutations at the three targeted sites. In total, 3/21 (14.3%) of the edited plants had a homozygous marker gene deletion by MMEJ, and 11/21 (52.4%, including 8 plants by MMEJ and 3 plants by NHEJ) underwent mono-allelic (heterozygous) gene deletion (Figure 1f). Our results demonstrate that by designing suitable target sites near MHSs (see below), the efficiency of MMEJ-mediated fragment deletion is much higher than that of NHEJ-mediated deletions. In addition, we obtained marker-free T 1 plants derived from the hoM #1, #8 and #15 plants, which had no HPT and the T-DNA (containing the CRISPR/Cas9 components and Bar) removed by genetic segregation.
Although studies have reported genomic fragment deletions by CRISPR/Cas systems, the efficiency is not high. The MMEJ pathway is active during cell cycle, which provides more chances for DSB repair than the other DNA repair pathways (Shen et al., 2017). Our results suggest that three factors may be related to the efficiency of genomic deletion using MMEJ-assisted CRISPR/ Cas editing. First, the MHS1 and MHS4 sites, which are closer to the targeted sites (Figure 1c), had higher efficiency (in total 8/21) than that of the other two MHSs (in total 3/21, Figure 1f). It seems that the closer the MHSs are to the DSB positions, the higher the fragment deletion efficiency will be. Second, the length of MHSs may be less important than their positions relative to the DSBs; for example, MHS3 has the longest size (7 bp) but it produced only one heM plant. Therefore, a MHS size of 4-5 bp (or even 3 bp, such as MHS4(a)) is sufficient for efficient MMEJmediated DNA repair. Third, production of multiple DSBs using multiplex sgRNAs may increase the frequency of genomic fragment deletions.
In conclusion, the high-efficiency genomic fragment deletion strategy using MMEJ-assisted CRISPR/Cas9 editing is flexible in target design for many genomic regions to remove targeted genomic fragments. Moreover, this approach can facilitate functional genomics research and genetic improvement of organisms.