Abstract
Precision gene editing has been recently achieved by homology-directed gene targeting (HGT) and prime editing (PE), but implementation remains challenging in plants. Here, we report a novel tool for precision genome editing in plants employing microhomology-mediated end joining (MMEJ). The MMEJ-mediated precise gene replacement produced much higher targeted editing efficiencies than the cNHEJ, up to 8.89 %, 4.47 %, and 8.98 % in tomato, lettuce, and cabbage, respectively.
cNHEJ and MMEJ mechanisms were used for CRISPR/Cas9-based genome editing at high efficiency and specificity1-3. However, a system using cNHEJ and MMEJ for precise gene replacement has not been developed in plants. Here, we proposed and evaluated a novel MMEJ-based system for precise gene replacement (Supplementary Fig. 1 and 2a,b)4. A predefined genomic site with a protospacer adjacent motif (PAM) is determined first, thus enabling the prediction of the DSB formation site. The sequence between the two predicted cut sites is employed as a donor template so that its DNA sequence is essentially modified allowing it to carry targeted base changes to avoid recurrent cuts after gene replacement (Supplementary Fig. 2a,b). In the case of MMEJ-mediated editing, the predicted flanking DSB ends would then be used to choose microhomologies (MH1, and MH2, Supplementary Fig. 1b).
We employed PEG-mediated tomato protoplast transfection to deliver the SpCas9 proteins, gRNAs, and donors for replacing six base pairs of the exon 5 of SlHPAT3 (Fig. 1a,b; Supplementary Table 1 and Supplementary file 1). Targeted deep-sequencing data revealed editing efficiency as high as 2.46 % and 0.09% (Supplementary Fig. 2c and Supplementary Table 3) with the MJ.HPAT3-1 and cNJ.HPAT3-1 donor. Significantly, the edited products of the cNHEJ donor did not include base changes located at its two ends (Supplementary Tables 2 and 3). Further analysis of the edited products revealed multiple repaired products by the MJ.HPAT3-1 donor with various frequencies (Fig. 1b). The precisely edited allele that contains all the intended base modification only accounted for 9.72 % of total edited reads (Supplementary Table 3). However, when we observed all the edited products containing the targeted B and C base changes, and excluded the others containing one-sided base changes (A1 and A2 or D1 and D2) that were designed to prohibit the recurrent cleavages of the SpCas9, we obtained a total of 805 reads corresponding to 33.01% of the total edited sequences (Fig. 1b; Supplementary Table 3). Our data also confirmed the simultaneous cutting activities of the gR1.HPAT3 and gR2.HPAT3 since most of the indel alleles were revealed with the traces of both the gRNA cleavages at the 3rd base upstream of PAM sequences (Fig 1a and Supplementary Figs. 4,5).
We next sought to investigate the implications of different donor doses on the frequency of MMEJ-mediated DNA replacement. The MMEJ-mediated editing frequencies with all base changes were increased with higher amounts of donor DNA, from 10.42 % (50 pmol) to 19.84 % (300 pmol), and the one-sided repair frequencies were reduced accordingly (Supplementary Fig. 6 and Supplementary Table 4). Moreover, when the MJ.HPAT3-1 donor dose was increased, the portions of targeted products (containing B and C base changes) were not different between 50 and 100 pmol but increased at higher doses, from 31.38 % (50 pmol) to 47.81 % (300 pmol) (Supplementary Table 4-5). In the case of the cNJ.HPAT3-1 donor, the editing efficiency did not vary much among the donor doses and was much lower compared to that of the MJ.HPAT3-1 (Fig. 1c). Again, most of its products contained only B and C changes (Supplementary Fig. 6).
Effective microhomology lengths (8-20 bases) increased MMEJ-mediated gene insertion in mammalian cells1, 5, 6. When the microhomology was shorter than 20 bp, the total editing efficiency was significantly reduced from 4.22 ± 0.47 % (MJ.HPAT3-1) to 2.33 ± 0.31 % (MJ.HPAT3-2) and 2.65 ± 0.58 % (MJ.HPAT3-3) (Fig. 1d and Supplementary Table 6). More importantly, the all-base-change precise editing efficiency was significantly higher for 20-bp microhomology (0.28 ± 0.05 %) compared to that of the 10-bp (0.12 ± 0.03 %) and 5-bp (0.04 ± 0.01 %) microhomology lengths (Supplementary Table 6). Furthermore, when we consider all the reads containing the targeted base changes (B and C), the targeted editing efficiency was 1.16 ± 0.09; 0.66 ± 0.12; and 0.55 ± 0.06 % for the 20-bp; 10-bp; and 5-bp microhomologies, respectively. Interestingly, the activities of the gRNAs were not significantly different among all the treatments (Supplementary Table 6), indicating that the MMEJ-mediated editing efficiency obtained from the experiments depended on microhomology length5-7.
We next tested if NU7441, a small chemical that was shown to significantly enhance the MMEJ-mediated DSB repair products in mammalian cells8, can facilitate MMEJ repair. When the NU7441 concentration was increased from 0 to 1 µM, the editing efficiency was dramatically elevated in all the donors (Fig. 1e,f). The case of the MJ.HPAT3-2 donor was remarkable in that the precise editing efficiency was enhanced 23.9 folds, from 0.14 to 3.34 %, with 1µM of NU7441, similar to the highest MMEJ-mediated efficiency of MJ.HPAT3-1 (3.80%) under the same conditions (Fig. 1f and Supplementary Table 7). Moreover, the ratio of the repaired products containing only B and C base changes was dramatically increased with higher NU7441 concentrations, reaching up to 5.24 % at 1 µM of NU7441 in the case of the MJ.HPAT3-2 (Fig. 1e, f). To further check whether NU7441 negatively impacts the suspected cNHEJ-mediated DSB repair in the case of cNJ.HPAT3-1 donor, we conducted editing experiments using a cNJ.HPAT3-1 donor with the addition of 0.5, 1.0, and 2.0 µM of NU7441. Surprisingly, the editing efficiency gradually increased with the increment of NU7441 concentration, and most of the repaired products contained only B and C base changes. The editing efficiency reached 1.58 % when 2.0 µM of NU7441 was added (Fig. 1f), which is 1.88-fold higher than the treatment without NU7441.
The plant regeneration system for tomato protoplast was shown to be of very low efficiency and time-consuming. To overcome this challenge, we attempted to deliver the cNHEJ and MMEJ-mediated gene editing into a tomato by the Agrobacterium-mediated method9. SlHPAT3 and SlHKT1;2 loci were selected for editing via the cNHEJ and MMEJ approaches (Supplementary Fig. 7a,b and Supplementary file 1). Unexpectedly, the cNHEJ and MMEJ-mediated editing were extremely low at both the loci (Supplementary Table 8), possibly due to the low donor availability, the requirement of four simultaneous cleavages, and the nucleolytic damages to the unprotected ends of donors. The editing reads were mostly with one-side editing (Supplementary Fig. 8a) or one-sided insertion of the MMEJ donor (Supplementary Fig. 8b). However, subsequent screening of regenerated plants carried precisely edited alleles, albeit their editing rate was relatively low (up to 4% for the cNHEJ event cNJ1 and 3% for the MMEJ event #MJ5) (Supplementary Fig. 9).
To extend MMEJ-mediated precision gene editing to other plant species, we conducted MMEJ-mediated precise gene replacement in lettuce and cabbage using the RNP transfection method. The THERMO-TOLERANCE 1 (TT1)10, ORANGE (Or)11, 12, and ACETOLACTATE SYNTHASE 1 (ALS1)13 genes were selected as targets for both lettuce and cabbage were identified by NCBI Blastp (Supplementary Fig. 10 and Supplementary Tables 9,10). We designed and employed two gRNAs for cutting genomic loci and MMEJ donors containing 20-base microhomologies at two ends (Fig. 2a,b). In lettuce, the highest MMEJ-mediated precise editing efficiency for all the intended base changes was 1.81 ± 0.75 % for the LsALS1 locus, and the lowest efficiency was zero for the LsTT1 locus. The LsOr locus showed only 0.13 ± 0.10 % for the exchange of all the intended bases (Fig. 2c and Supplementary Table 11). Considering all the repaired products that contain the targeted base changes for the desirable a.a., the efficiency reached 4.47 %, 0.78%, and 0.42% for LsALS1, LsTT1, and LsOr locus, respectively (Fig 2c and Supplementary Table 11). In cabbage, the highest precise editing efficiency (7.27 ± 4.46 %) was obtained at the BoTT1 locus and lower at the BoOr gene (1.68 ± 0.24 %). The ALS1 locus for cabbage resulted in the least precise editing efficiency, at only 0.73 ± 0.43 %, nearly half of the editing efficiency obtained with the LsALS1 (Fig. 2d and Supplementary Table 12). The data from lettuce and cabbage indicate that MMEJ-mediated precision editing could be successfully extended to other plant species.
Taken together, we successfully engineered the error-prone MMEJ-mediated DSB repair mechanism for precision gene replacement in plants. However, it requires further optimization of efficiency, especially that of plant regeneration from the edited cells. This report offers another precision gene-editing tool that may help to advance crop breeding in the future.
Methods
Targeted genes and donor DNA preparation for RNP works
A tomato homolog of Arabidopsis hydroxyproline O-arabinosyltransferase 3 (SlHPAT3, accession no. Solyc07g021170.1) was chosen as the first target thanks to its highly active guide RNA (gRNA) pair14. We employed PEG-mediated tomato protoplast transfection experiments using RNPs with SpCas9 protein and two sgRNAs, gR1.HPAT3 and gR2.HPAT3 (Fig. 1a) for cutting the genomic sites, combining that with a chemically modified cNHEJ (cNJ.HPAT-1) donor or an MMEJ donor (MJ.HPAT3-1) (Supplementary Table 1 and Supplementary file 1) for the replacement of six base pairs of the exon 5 of SlHPAT3 (Fig. 1a). The 5’ end modified cNJ.HPAT3-1 was prepared by PCRs using 5’ modified oligos (5’ phosphorylated, phosphorothioate bond addition to the phosphodiester linkage between the first and the second nucleotides) (Supplementary Table 1) synthesized by Bioneer (Korea) without a template. The MMEJ donors were prepared by PCRs using oligos (Supplementary Table 1) synthesized by Bioneer (Korea) without templates. A high-fidelity DNA Taq polymerase was used for the PCR amplifications. The PCR products were cleaned using a BIOFACT PCR cleanup kit (BIOFACT, Korea). The donor concentrations were assessed by Nanodrop2000 spectrophotometer (Thermofisher, USA) and directly used for transfection stored at -20°C for further uses.
Construction of plasmid for Agrobacterium-mediated transformation in tomato
For stable transformation and assessment of the cNHEJ and MMEJ-mediated precision gene replacement in tomatoes, we designed and cloned the gRNA expression cassettes (Supplementary file 1) using the Golden-gate cloning system as described previously9, 15. Two gRNA expression cassettes (gR1.HPAT3 and gR2.HPAT3 for SlHPAT3; gR1.HKT1;2 and gR2.HKT1;2 for SlHKT1;2) were used to generate two DSBs at each targeted site. The cNHEJ donors (cNJ.HPAT3-1 for SlHPAT3; cNJ.HKT1;2 for SlHKT1;2) and MMEJ donor (MJ.HPAT3-1 for SlHPAT3; MJ.HKT1;2 for SlHKT1;2) were designed to be flanked by two gRNAs (gDR1.HPAT3 and gDR2.HPAT3 for cNJ.HPAT3-1; gDR3.HPAT3-1 and gDR4.HPAT3 for MJ.HPAT3; gDR1.HKT1;2 and gDR2.HKT1;2 for cNJ.HKT1;2; gDR3.HKT1;2 and gDR4.HKT1;2 for MJ.HKT1;2) cutting sites (Supplementary file 1). The binary plasmids were constructed to test the cNHEJ and MMEJ approaches using a conventional T-DNA and a geminiviral replicon system9. The NptII selection marker expression cassette (pNOS-NptII-tOCS) is driven by the NOS promoter and terminated by the OCS terminator (Addgene # 51144). An intron-containing plant codon-optimized SpCas9 driven by a CaMV 35S promoter and CaMV 35S terminator (p35S-pcoCas9I-t35S) was used (Supplementary file 1).
Isolation of protoplasts
Tomato (Solanum lycopersicum cv. Micro-Tom), lettuce (Lactuca sativa L. cv. Cheongchima), and cabbage (Brassica. oleracea) seeds were sterilized with 70% ethanol for 3 min, 1% hypochlorite solution for 15 min, and washed five times with distilled water. The sterilized seeds were inoculated in a medium containing 1/2 Murashige and Skoog salts, 0.4 mg/L thiamine HCl, 100 mg/L Myo-inositol, 30 g/L sucrose, and 8 g/L gelrite, pH 5.7. The seedlings were grown in a growth chamber under a 16 h light/8 h dark photoperiod (100–130 μmol/m2 s) at 25°C for tomato and 20°C for lettuce, and 23°C for cabbage.
For protoplast isolation of tomato and cabbage, the cotyledons of 4-day-old tomato seedlings and the cotyledons of 7-day-old cabbage seedlings were immersed in cell and protoplast washing solution (CPW) containing 0.5% cellulase (Novozymes, Basgsvaerd, Denmark), 0.5% pectinase (Novozymes), 1% viscozyme (Novozymes), 3 mM MES (pH 5.8) and 9% mannitol. After 15 min of vacuum infiltration, the suspension was incubated for 2-4 hr on a rotary shaker at 50 rpm at 25°C. The suspension was filtered through an eight-layer gauze and centrifuged for 5 min at 100 g. Protoplasts were separated on a 21% sucrose density gradient and then collected at the interface of W5 solution (2 mM MES pH 5.8, 154 mM NaCl, 125 mM CaCl2, 5 mM KCl). The harvested protoplasts were washed three times with W5 solution and then resuspended in MMG solution (4 mM MES pH 5.7, 0.4 M mannitol, 15 mM MgCl2). The concentration of protoplasts was determined using a hemocytometer.
For the lettuce protoplast isolation, the cotyledons of 7 d-old seedlings were digested with 10 mL of enzyme solution (1% [w/v] Viscozyme (Novozyme), 0.5% Celluclast (Novozyme), and 0.5% Pectinex (Novozyme), 3 mM MES (2-[N-Morpholino] ethanesulfonic acid), pH 5.7 and 9% mannitol in CPW salts with shaking at 40 rpm for 4–6 h at 25 °C in the dark. The protoplast mixture was then filtered through a 40 µm nylon cell strainer (Falcon) and collected by centrifugation at 800 rpm for 5 min in a 14 mL round tube (SPL). The collected protoplasts were re-suspended in W5 solution (2 mM MES [pH 5.7], 154 mM NaCl, 125 mM CaCl2, and 5 mM KCl) and further centrifuged at 800 rpm for 5 min. Finally, the protoplasts were re-suspended in W5 solution and counted under a microscope using a hemocytometer.
Protoplasts were adjusted to a density of 1 × 106/mL in MMG solution before transfection. The transfected protoplasts were cultured in protoplast culture medium (MS medium containing 0.4 mg/L thiamine HCl, 100 mg/L myo-inositol, 30 g/L sucrose, 0.2 mg/L 2,4-dichlorophenoxyacetic acid [2,4-D], and 0.3 mg/L 6-benzylaminopurine [BAP], pH 5.7) in the dark f at 25°C for 4 weeks.
PEG-mediated RNP and donor transfections
SpCas9 protein was purchased from ToolGen, Inc. (South Korea), and guide RNAs were synthesized by GeneArt Precision gRNA Synthesis Kit (Invitrogen) according to the manufacturer’s protocol. PEG-mediated RNP and donor transfections were performed in the previous study16.
For ribonucleoprotein (RNP) and donor DNA transfections with tomato and cabbage protoplasts, 2 × 105 protoplasts were transfected with the purified SpCas9 protein (20 μg) premixed with in vitro-transcribed sgRNA 1 (10 μg), sgRNA 2 (10 μg), and donor templates (300 pmol) in PBS buffer followed by incubating for 10 min at 25°C. The RNP complexes were mixed with protoplasts and then supplemented with an equal volume of 40% PEG transfection solution (40% PEG 4000, 0.2 M mannitol, and 0.1 M CaCl2). This suspension was mixed gently and then incubated at room temperature for 10 min. An equal volume of W5 solution was added for washing, followed by centrifugation at 100 g for 5 min. The supernatant was discarded, and the protoplasts were incubated with 1 ml of W5 solution in the dark at 25°C for 48 h. Afterwards, the cells were collected for gDNA isolation and subsequent targeted deep sequencing analysis.
For lettuce protoplast transfection, SpCas9 protein and sgRNAs were premixed in 1× NEB buffer 3 for at least 10 minutes at room temperature and 2×105 protoplast cells were transfected with SpCas9 protein (20 μg) premixed with sgRNAs (10 μg each) and donor DNA (300 pmol). A mixture of 2 × 105 protoplast cells was re-suspended in 200 μl MMG solution and then was slowly mixed with RNP complex, donor, and 350 μl of PEG solution (40% [w/v] PEG 4000, 0.2 M mannitol, and 0.1 M CaCl2). After incubation for 10 min, the transfected protoplast cells were gently re-suspended in 650 μL W5 solution. After additional incubation for 10 min, 650 μL W5 solution was added slowly again and was mixed well by inverting the tube. Protoplasts were pelleted by centrifugation at 556 rpm for 5 min and washed gently in 1 ml W5 solution. Protoplasts were pelleted by centrifugation at 556 rpm for 5 min and re-suspended gently in 1 ml WI solution (4 mM MES [pH 5.7], 0.5 M mannitol, and 20 mM KCl). Finally, the protoplasts were transferred into a 60 × 15 mm petri dish (Falcon), cultured under dark conditions at 25°C for 48 h and then analyzed for genome editing efficiency.
Agrobacterium-mediated tomato transformation
The Agrobacterium-mediated tomato transformation was conducted using a protocol published by Vu and coworkers 9 with or without 1 µM NU7441 treatment for 5 days post-washing. Ten-day post-transformation, thirty cotyledon fragments were collected per transformation plate to isolate genomic DNAs and subsequent miniseq analysis. Regenerated plants were selected in media containing 100 mg/L kanamycin and transferred to soil pots before analyzing for the editing performance. Genomic DNAs were extracted from the plants and analyzed by PCR amplification of the targeted sequences and by Sanger sequencing.
Targeted deep sequencing
Genomic DNAs were isolated from the protoplasts using the CTAB method. We used the miniseq sequencing service (MiniSeqTM System, Illumina, USA) to obtain targeted deep sequencing of the edited genomic sites. Miniseq samples were prepared in three PCRs according to the manufacturer’s guidelines, with genomic DNAs as the first PCR template. The first and second PCRs used primer listed in Supplementary Table 1, whereas the third PCRs were conducted with the manufacturer’s primers to assign a sample ID. A high-fidelity DNA Taq polymerase (Phusion, NEB, USA) was used for the PCRs. The miniseq raw data FASTQ files were analyzed using the Cas-Analyzer tool17. The indel analyzing window was set at 5 bases with a comparison range that covered both the read ends. A similar analysis was conducted for the targeted base changes of lettuce and cabbage genes.
Statistical analysis
The editing data, statistical analysis, and plots were further processed by the MS Excel and Graphpad Prism 9.0 programs and are explained in detail in the legends of figures and tables wherever applicable.
Author contributions
Conceptualization, T.V. V and J.Y.K.; Methodology, T.V. V and J.Y.K.; Conducted experiments, T.V. V., G.H.L, S.H.C, J.Y.M., S.W.K., J.C.J, N.T.N, S.D., M.T.T., Y.W.S, J.K, Y.J.S; Data analysis, T.V. V. and J.Y.K.; Writing – Original Draft, T.V. V.; Writing – Review & Editing, T.V. V. and J.Y.K.; Funding Acquisition, T.V.V., and J.Y.K.; and Supervision, T.V. V and J.Y.K.
Competing interests
The authors have applied for Korean patents (10-2021-0089814) and PCT patents (PCT-KR2021-008727) based on the results reported in this paper.
Supplementary item
Supplementary figure legends
Supplementary Fig. 1. HR-independent strategies for precision gene/allele replacement using CRISPR/SpCas9
Supplementary Fig. 2. HR-independent precision editing approaches using CRISPR/SpCas9 Supplementary Fig. 3. Sanger sequencing data revealed precise replacement of DNA by MMEJ donor
Supplementary Fig. 4. Indel alleles revealed from transfection of SpCas9, gR1.HPAT1 and gR2.HPAT3 with cNJ.HPAT3-1 donor
Supplementary Fig. 5. Indel alleles revealed from transfection of SpCas9, gR1.HPAT1 and gR2.HPAT3 with MJ.HPAT3-1 donor
Supplementary Fig. 6. The frequency of the MMEJ-mediated repaired products at different donor doses
Supplementary Fig. 7. Agrobacterium-mediated system for cNHEJ and MMEJ-mediated gene editing in tomato
Supplementary Fig. 8. Representative repaired products obtained by the pMJ1 and pMJ2 vector.
Supplementary Fig. 9. cNHEJ and MMEJ-mediated precision editing events revealed from Agrobacterium-mediated transformation
Supplementary Fig. 10. Identification of TT1, Or, and ALS1 genes in lettuce and cabbage for MMEJ-mediated gene targeting
Supplementary Tables
Supplementary Table 1. gRNA, donor, and primer sequences employed in the study
Supplementary Table 2. cNHEJ- and MMEJ-mediated editing rates at SlHPAT3 locus revealed by Sanger sequencing
Supplementary Table 3. cNHEJ- and MMEJ-mediated editing rates of various repaired products at SlHPAT3 locus
Supplementary Table 4. Editing frequency of various repaired products revealed from treatments of different donor types and doses
Supplementary Table 5. MMEJ-mediated gene editing using various MJ.HPAT3-1 donor amounts (replicate)
Supplementary Table 6. MMEJ-mediated editing efficiency revealed from DSBs repair using donors with different microhomology lengths.
Supplementary Table 7. The frequency of repaired products revealed from the treatment of NU7441
Supplementary Table 8. cNHEJ and MMEJ mediated gene editing in tomato using the Agrobacterium-mediated delivery of the editing tools
Supplementary Table 9. Sequence alignment for selection of targeted genes in lettuce Supplementary Table 10. Sequence alignment for selection of targeted genes in cabbage
Supplementary Table 11. Data revealed from the targeted deep-sequencing analysis of MMEJ-mediated editing in lettuce
Supplementary Table 12. Data revealed from the targeted deep-sequencing analysis of MMEJ-mediated editing in cabbage
Supplementary files
Supplementary file 1: Sequences used in the study
Acknowledgments
This work was supported by the National Research Foundation of Korea (Grant NRF 2020R1I1A1A01072130, 2020M3A9I4038352, 2020R1A6A1A03044344, 2021R1A5A8029490, 2022R1A2C3010331) and the Program for New Plant Breeding Techniques (NBT, Grant PJ01478401), Rural Development Administration (RDA), Korea.