New system for cloning a multiplex tRNA-gRNA construct
To express multiple gRNAs in plants, we first developed a simple, fast and PCR-free cloning method (Fig. 1). This cloning platform consists of a pre-cloned vector, pGRNAs, which carries a single unit of a tRNA-gRNA scaffold, and an acceptor vector, which harbors the SpCas9-coding sequence and a selection marker (Fig. 1c). The first step is to add a target-binding sequence (19 ~ 20-nt) into the pGRNA by a PCR-free method (Fig. 1b). Two BsaI restriction enzyme (Type IIS) recognition sites were inserted between the tRNA sequence and the gRNA scaffold in pGRNA vector (Fig. 1): this insertion allowed us to easily ligate a short double-stranded DNA (23 ~ 24-nt) containing the target sequence into the pGRNA. To prepare the short double-stranded DNA, we ordered two complementary single-stranded oligos, which are matched respectively to the 3’ and 5’ overhangs made by the BsaI cut of pGRNA. We prepared each tRNA-gRNA unit within three days (no PCR step needed).
The pGRNA vector has two AarI restriction enzyme-binding sites on the outside of the tRNA-gRNA unit, and the AarI treatment produces a tRNA-gRNA unit with 4-nt overhang sequence. Each pGRNA is designed to produce specific overhang sequences that connect the tRNA-gRNA unit in the order of vector number (pGRNA1, pGRNA2, pGRNA3, pGRNA4, and pGRNA5e): tRNA-gRNA units could be sequentially ligated into a plant binary vector (acceptor vector, pECO100, pECO200, and pECO300) by using the Golden Gate assembly method. Thus, the plant binary vector with the desired multiplex gRNA combination, which we call pGG, could be easily and quickly (within a week) produced. The pGG vector was numbered according to the number of tRNA-gRNA units. For example, pGG-3 is the binary vector with three consecutive tRNA-gRNA units.
Validation of the editing efficiency of pGG-1 and pGG-2 vectors in protoplasts
A part of precursor tRNAGly sequences has been used to produce multiplex gRNAs from a single polycistronic transcript driven by U6/U3 promoters. This tRNA sequence has been reported to increase the expression of gRNA in rice protoplasts, which in turn improves genome editing efficiency (19). To determine whether the tRNA could also increase genome editing efficiency in a dicot plant, we edited twelve genes with a total of 28 gRNAs with and without the tRNA sequence in the protoplasts of wild tobacco, N. attenuata (Fig. 2b) (26,27). Single-stranded gRNA was expressed under the control of either AtU6 or AtU6-tRNA (Fig. 2a). Results show that the tRNA does not increase the editing efficiency of SpCas9-gRNA complexes in N. attenuata (P = 0.56) (Fig. 2b).
We then examined whether two gRNAs targeting the proximal site of one gene increase the genome editing efficiency. We chose six target genes of N. attenuata -- NaEAH1, NaNEC5b, NaNEC3a, NaAOC, NaMYC2, and NaNEC1c -- and then designed two adjacent gRNAs to target each one (Fig. 3a). The distance between two gRNAs varied from 37-nt to 85-nt. The pGG vectors containing one tRNA-gRNA (pGG-1) and two tRNA-gRNA (pGG-2) units were transformed into the protoplasts, and their editing efficiency and mutation patterns were determined by targeted deep sequencing (Fig. 3a). When two gRNAs were expressed, rather than one, the overall editing efficiency was increased at each target site: 3.0% (one gRNA) to 15% (two gRNAs) for NaEAH1-gRNA12 (g12), 4.5% to 17% for NaEAH1-g14, 3.6% to 8% for NaNEC5b-g20, 3.6% to 8% for NaNEC5b-g21, 3.0% to 6.4% for NaNEC3a-g4, 6.4% to 8.1% for NaNEC3a-g5, 7.1% to 16.7% for NaAOC-g2, 6.4% to 17.2% for NaAOC-g4, 5.0% to 9.3% for NaMYC2-g2, 4.5% to 8.2% for NaMYC2-g3, and 6.4% to 8.8% for NaNEC1c-g1, 4.4% to 8.3% for NaNEC1c-g2 (Fig. 3b).
We found that two proximal cleavages by SpCas9-gRNA induced large deletions between two cleavage sites (Fig. 3b). Although total editing frequencies of six pGG-2 constructs varied, the relative ratio of large deletions to total mutations was similar: the mean frequency of the large deletions was ~ 85% for NaEAH1-g12-g14, ~ 97% for NaNEC5b-g20-g21, ~ 76% for NaNEC3a-g4-g5, ~ 90% for NaAOC-g2-g4, ~ 87% for NaMYC2-g2-g3, and ~ 90% for NaNEC1c-g1-g2 (Fig. 3c). The large deletion occurred by rejoining the blunt end of two cleaved sites at three nucleotides upstream of the protospacer adjacent motif (PAM) sequence without any insertion or deletion of nucleotides: the mean frequencies of the large deletion were ~ 60% for NaEAH1-g12-g14, ~ 38% for NaNEC5b-g20-g21, ~ 84% for NaNEC3a-g4-g5, ~ 95% for NaAOC-g2-g4, ~ 28% for NaMYC2-g2-g3, and ~ 63% for NaNEC1c-g1-g2 (Fig. 3d and Additional file 1). The next abundant mutation patterns revealed by the large deletions with one nucleotide insertion or deletion at each cleaved site. For instance, either the C or A nucleotide was added at the NaEAH1-g14-cleaved site; A was added at the NaNEC5b-g21-cleaved site; three different nucleotides -- A, T, or C -- were added at the NaNEC3a-g5-cleaved site; A was removed at the NaAOC-g4-cleaved site and NaMYC2-g4-cleaved site; and T was added at the NaNEC1c-g1-cleaved site (Fig. 3d and Additional file 1).
Genome editing with three (pGG-3) and four gRNAs (pGG-4) in protoplasts and in planta
Furthermore, we examined the editing efficiency of pGG-3 constructs in protoplasts. In Fig. 3b, we examined the efficiency with which two guide RNAs edit the NaNEC1c gene. The third gRNA, NaNEC1c-g3 was designed to cleave the double-stranded DNA at 64-nt apart from the NaNEC1c-g2 cleavage site (Fig. 4b). We then examined the mutation patterns induced by simultaneously expressing three gRNAs binding on the proximal target sites. Total mutation frequency of NaNEC1c-g1-g2-g3-transformed protoplasts was 26% including small indels (4% for NaNEC1c-g1, 2% for NaNEC1c-g2, and 2% for NaNEC1c-g3) and large deletions (17% for NaNEC1c-g1 and -g3, 2% for NaNEC1c-g1 and -g2, 1% for NaNEC1c-g2 and -g3) (Fig. 4a).
We next tested whether the pGG system could effectively edit target genes in planta and induce the similar mutation patterns observed in the protoplasts. The pGG-3 vector carrying NaNEC1c-g1-g2-g3 was delivered into N. attenuata hypocotyl explants using Agrobacterium-mediated transformation (28) and whole plants were regenerated on the selection media. Among 24 T0 transformants, 21 lines were edited at least one binding site of three gRNAs (87.5%, Fig. 4b). As shown in the protoplasts, the editing frequency at the NaNEC1c-g2-binding site was lower than the editing frequency at the NaNEC1c-g1 and -g3-binding sites (Figs. 4a and b). Some T0 lines (T0-8, -9, -10) had large deletions at the target site: the major mutation pattern of the large deletion occurred when the blunt ends of two cleaved sites were rejoined at three nucleotides upstream of the PAM sequence of NaNEC1c-g1 and four nucleotides upstream of the PAM sequence of NaNEC1c-g3 (Fig. 4b and Additional file 2). However, unlike the results with the protoplasts, the results with several T0 transformants (T0-1, 2, 3, 4, 5, 6, 7, 12) had small indel mutations (Fig. 4b). Major small indel patterns in transformed plants exhibited an A or T insertion at the three nucleotides upstream of the PAM sequence of NaNEC1c-g3-binding site (Additional files 3 and 4).
We also confirmed that the pGG-4 vector carrying four gRNAs can successfully edit two genes in plants: g12 and g14 for targeting NaEAH1, and g1 and g2 for targeting NaNEC1c (Fig. 4c). Genomic DNA was extracted from the transformed calli grown in the selection media and the mutation frequency of each callus was measured. Out of 16 calli, at least one gene was edited from 15 (94%). Furthermore, the four-gRNA expression with SpCas9 successfully generated mutations both on NaEAH1 and NaNEC1c (more than 50% mutation frequency) in the calli 1, 2, 4, and 5. The mutation patterns of NaEAH1 in the protoplasts and the calli were quite different: the dominant mutation pattern in the calli (Fig. 4c) was insertion mutations, whereas the dominant mutation pattern in the protoplast was the large deletions (Fig. 3b). On the other hand, NaNEC1c-g1-g2 induced large deletions in both protoplasts (Fig. 3b) and the calli (Fig. 4c).