The stability of DRs-involved paired-gRNA plasmids pDG-A-X series
To study the stability of DRs-involved paired-gRNA plasmids, pDG-A-X series for co-expression of two gRNAs were employed. A functional gRNA contains a 20-bp sequence for targeting and a 82-bp scaffold that binds Cas9 protein . Each gRNA was transcribed by a 35 bp constitutive promoters J23119 (Figure 1A). Plasmid rearrangement was detected by PCR primers F1/R1 after plasmid construction and re-transformation processes.
pDG-A-100K for 100-kb genomic deletion was constructed by using E. coli DH10B strain as host. However, the deletion rate of 73.33% was observed after pDG-A-100K plasmid construction process. Similarly, the deletion rate was around 65% after re-transformation process of the correct pDG-A-100K plasmid (Figure 1B and 4B). PCR results indicated that the deletion occurred between the paired-gRNA regions of these mutant plasmids. DNA sequencing results demonstrated one of two gRNAs with its promoter was eliminated. Furthermore, the double restriction enzyme digestion analyses by using NdeI and CaiI showed the deletion only occurred between the paired-gRNA regions, rather than other parts of plasmids (Figure 1C).
To enhance the stability of pDG-A-100K, the effects of experimental conditions including DNA transformation methods and culture mediums were then assessed during re-transformation process in DH10B strain. Compared with transformation by heat shock, electrotransformation led to a 5.6-fold decrease in the deletion rate for cells cultured in LB medium and a 3.5-fold decrease for cells cultured in TB medium (Figure 1D). The nutrient supplies for plasmid propagation also influenced its stability. Replacing the LB medium with the nutrient-rich TB medium reduced the deletion rate by half when DNA was chemically transformed into cells, while no further decrease in the deletion rate was achieved when plasmids were transformed electrically (Figure 1D). Therefore, pDG-A-100K appeared to be more stable when introduced into cells by electroporation and propagated in rich medium, but neither could eliminate the events of plasmid rearrangement.
The patterns of plasmid rearrangement
Various recombination derivations were discovered during DRs-mediated recombination events of pDG-A-X series (Figure 2). Based on our observations, two main types of deletion were summarized: the deletion of the first gRNA expression cassette along with its promoter (MUT-1), and the deletion of the second gRNA expression cassette together with its promoter (MUT-2). Moreover, point mutations and insertions occurred when the candidate plasmids were sequenced for further analysis. For example, point mutations in the –10 regions or –35 regions of promoter J23119 (MUT-3/4) appeared frequently, which could affect the transcription process of gRNAs. A 12-bp repeated insertion at the end of the gRNA scaffold was also detected (MUT-5), which could influence the normal structure of gRNA. Taken together, pDG-A-X series produced the deletion of one of paired gRNA expression cassettes randomly and other spontaneous mutation between the paired-gRNA regions, which made it difficult to maintain plasmid stability.
The RecA dependency of paired-gRNA plasmids recombination
To test whether the recombination of pDG-A-X series relied on the RecA enzyme, the correct pDG-A-100K was re-transformed into various E. coli strains with the genotypes of recA1, ΔrecA1398, Δ(sr1-recA) or recA+ (Figure 1E). In MG1655 recA+ strain expressing functional RecA protein, the deletion rate was up to 91.67%. In various recA mutant strains, DRs-induced recombination still occurred with the frequencies of 63.33-95%. No distinct difference of deletion rates was found in XL10-Gold recA1 strains, DH10B recA1 strain and Mach1T1 ΔrecA1398 strain, while the deletion rate even was increased up to 95% in DB3.1 Δ(sr1-recA) strain. All results indicated that RecA-independent recombination played a great role on the deletion of pDG-A-X series in E. coli.
The replication slippage model for RecA-independent recombination [19, 20] was applied to explain the phenomenon, since the main recombination products of pDG-A-X series were plasmid deletion form. The pDG-A-X series have ColE1 origin which produces high plasmid copy numbers and determines unidirectional replication . During the plasmid replication process in E. coli, pDG-A-X series generated the Types I or Type II slipped misalignment of the Okazaki fragment, which formed a loop within the lagging strand template to facilitate the formation of deletion (Figure 3). When the second promoter J23119 was employed as mispaired position, the deletion of the first gRNA expression region occurred, leading to the formation of pDG-A-X-M1. When the repeated gRNA scaffold mediated the plasmid recombination, the second gRNA region was deleted to form pDG-A-X-M2.
Effects of promoters and origins on pDG-A-X series stability
The effects of different plasmid architecture on plasmid stability were then evaluated. As shown in Figure 1, there are two pairs of DRs in pDG-A-X. One is the repeated 35-bp promoter J23119 while the other one is the repeated 82-bp gRNA scaffold. To reduce the number of DRs, pDG-P-X was designed by replacing the second promoter J23119 with an alternative 49-bp PR promoter (Figure 4A). After the assembly products of pDG-P-100K were introduced into DH10B strain, the deletion rate of pDG-P-X was up to 81.67% when verified by primers F1/R1 (Figure 4B). These deletion derivatives of pDG-P-100K didn’t contain promoter PR region when verified by primers F2/R1. DNA sequencing demonstrated that pDG-P-100K generated spontaneous deletion of the second gRNA region to form pDG-A-100K-M2. Although it was difficult to obtain correct pDG-P-X series plasmids by Gibson Assembly method, these plasmids could be more stably maintained during the re-transformation process once the correct plasmid was obtained firstly (Figure 4B).
Since the copy number may have an impact on plasmid stability, pDG-S-X series were designed by replacing the ColE1 origin with pSC101 origin  which replicates at a relatively low copy number (<8 copies/cell) (Figure 4A). However, pDG-S-100K still had a high deletion rate of 65% and 53.33% during plasmid construction and re-transformation processes (Figure 4B). These results demonstrated that just changing the promoter or the origin of DRs-involved paired-gRNA plasmids pDG-A-X series didn’t eliminate the events of plasmid rearrangement.
Design of RPGPs cloning strategy
In attempt to avoid DRs-induced plasmid rearrangement genetically, a reversed paired-gRNA plasmids (RPGPs) cloning strategy was developed for pDG-R-X series (Figure 5A). Compared with pDG-A-X, the plasmid architecture of pDG-R-X was versatilely modified through changing the promoter of the second gRNA, the origin of replication, and the direction of gRNA cassettes. Two gRNA cassettes were placed in opposite directions with one expressed by J23119 promoter and another by PR promoter, thus turning the two 82-bp gRNA scaffolds into inverted repeats (IRs). Moreover, two different promoters ensured the 20-bp sequences specific for two targeted loci in order, when the 20-bp sequences and the 20-bp overlap sequences for assembly were embedded in primers as a part of insert. Since the overlap sequences were repeated but reversed, the insert could be assembled in two directions, leading to the formation of pDG-R1-X or pDG-R2-X (Figure 5A). As we expected, pDG-R-100K didn’t generate any plasmid rearrangement events during plasmid construction process, when verified by PCR reactions (F3/R3 and F4/R2) and DNA sequencing.
To further examine PRGPs stability, the correct pDG-R1-100K plasmid was retransformed into DH10B strain and verified by PCR reaction. All of 50 colonies produced a 449-bp and a 602-bp band when amplified by primer pair F3/R3 and F4/R2, respectively. Representative colony PCR results are shown in Figure 5B. Nine of corresponding plasmids were digested by Eco32I and PstI and produced two bands with correct sizes of 3973 bp and 1592 bp (Figure 5C). The following DNA sequencing also confirmed that pDG-R1-100K maintained the intact paired gRNA expression cassettes without any mutations.
Large genomic deletion mediated by RPGPs
To test the practicability of RPGPs, RPGPs-associated CRISPR/Cas9 system was used for large genomic deletion in E. coli MG1655 strain. Since the double-strand breaks (DSBs) in E. coli can be repaired through its native end-joining mechanism , two plasmids were required for editing: p-PBAD-Cas9 plasmid contained the p15A replication origin, a kan gene, and Cas9 protein under control of the arabinose-inducible araBAD promoter (PBAD); RPGPs pDG-R-X series contained pSC101 replication origin, a bla gene and paired-gRNA expression cassettes (Figure 6A). Cas9 used here was an evolved SpCas9 variant xCas9-3.7 , which could reduce the survival rate of WT cells and increase the positive rate of large genomic editing.
pDG-R1-100K plasmid was applied to coexpress two gRNAs for the deletion of a 100-kb nonessential fragment from the E. coli chromosome (1,449,590-1,549,496) (Figure 6B). The targeting sequences of 100-kb fragment were summarized in Table S2, Supporting Information. Since the native end-joining mediated DNA repair resulted in genomic deletion of stochastic length around two targeted loci, we designed three pairs of primer F5/R5, F6/R6, and F7/R7 to check positive mutants among forty randomly selected colonies, and representative PCR results are shown in Figure 6C. A proximate 83.33% editing efficiency was achieved in this test, while negative colonies (16.67%) in the experimental group were also obtained. Further investigation showed that these colonies were not the wild type, but contained sequence deletion of stochastic length in the two target sites. The results indicated that RPGPs-associated CRISPR/Cas9 system was successfully used for large genome editing in E. coli.