Reducing competition between msd and genomic DNA improves retron editing efficiency

Abstract

Retrons, found in bacteria and used for defense against phages, generate a unique molecule known as multicopy single-stranded DNA (msDNA). This msDNA mimics Okazaki fragments during DNA replication, making it a promising tool for targeted gene editing in prokaryotes. However, existing retron systems often exhibit suboptimal editing efficiency. Here, we identify the msd gene in Escherichia coli, which encodes the noncoding RNA template for msDNA synthesis and carries the homologous sequence of the target gene to be edited, as a critical bottleneck. Sequence homology causes the msDNA to bind to the msd gene, thereby reducing its efficiency in editing the target gene. To address this issue, we engineer a retron system that tailors msDNA to the leading strand of the plasmid containing the msd gene. This strategy minimizes msd gene editing and reduces competition with target genes, significantly increasing msDNA availability. Our optimized system achieves very high retron editing efficiency, enhancing performance and expanding the potential for in vivo techniques that rely on homologous DNA synthesis.

Keywords: Escherichia coli; DNA Replication; Gene Editing; Retron.

Figure 1. Effect of plasmid replicons on genomic lacZ editing efficiency by the retron system.

(A) Schematic diagram of plasmids with different replicons. The detailed sequences of the plasmids are provided in Appendix Figs. S2–S4. (B) The efficiency of lacZ gene editing was assessed at various time points in ΔmutSrecJsbcB strains harboring p15AlacZ, pSClacZ, or pUClacZ plasmids. The data are displayed as a scatter plot, with each data point representing an individual biological replicate (n = 3). (C) Plasmid and msDNA copy numbers were quantified in the ΔmutSrecJsbcB strain harboring p15AlacZ, pSClacZ, or pUClacZ plasmids at the time points indicated by arrows in panel B (10, 18, and 14 h, respectively). The data were presented as a bar graph, with each bar representing the mean, and error bars indicating the standard deviation. Each data point represents an individual biological replicate (n = 4). Statistically significant differences were determined using an unpaired Student’s t-test. The p values for the msDNA copy number comparisons between ΔmutSrecJsbcB strains harboring pSClacZ and p15AlacZ, and pSClacZ and pUClacZ were 0.000093 and <0.000001, respectively. For the plasmid copy number comparisons between pSClacZ and p15AlacZ, and pSClacZ and pUClacZ, the p values were 0.009991 and <0.000001, respectively. *P < 0.01, ****P < 0.0001. Source data are available online for this figure.

Figure 2. Additional lacZ homology on the plasmid reduces genomic lacZ editing efficiency.

(A) A schematic presentation of the p15AlacZ plasmid. The ΔmutSrecJsbcB strain carrying the p15AlacZ plasmid can produce msDNA-PSlacZ, which has the potential to integrate into two targets: the genomic lacZ gene and the PSlacZ coding sequence in msd on the plasmid. (B) A schematic representation of the p15AlacZ-lacZ plasmid, containing an additional PSlacZ coding sequence. The ΔmutSrecJsbcB strain harboring the p15AlacZ-lacZ plasmid can produce msDNA-PSlacZ, which has the potential to integrate into three target sites: the genomic lacZ gene, the PSlacZ coding sequence in msd, and the additional PSlacZ coding sequence on the plasmid. The detailed sequence of the plasmid is provided in Appendix Fig. S5. (C) Editing efficiency of the lacZ gene at different time points in the ΔmutSrecJsbcB strain carrying p15AlacZ or p15AlacZ-lacZ plasmids. The data were displayed as a scatter plot, with each data point representing an individual biological replicate (n = 3). (D) Sequencing results of the lacZ sequence from the target region on plasmid. Red letters represent mutated bases. Source data are available online for this figure.

Figure 3. msDNA-mediated editing of msd sequences on plasmids.

(A) A schematic diagram illustrates four plasmids: pBR-PkanX, p15A-PKanY, pBR-ØkanX, and p15A-ØKanY. Each plasmid contains a 92 bp fragment within its msd region homologous to the kanamycin resistance gene (green line). Notably, these homologous sequences share a high degree of similarity (90 bp), differing by only two specific base pairs: “T-T” in pBR-PkanX and “G-A” in pBR-PkanY. Green and purple stars indicate mutation sites. Plasmids pBR-PkanX and p15A-PKanY possess the J23119 promoter, enabling retron operon transcription and subsequent msDNA expression. In contrast, the J23119 promoter is absent in pBR-ØkanX and p15A-ØKanY, preventing retron operon transcription and msDNA production. The detailed sequences of the plasmids are provided in Appendix Figs. S6–S9. (B) The ΔmutSrecJsbcB strain harboring two plasmids, pBR-PkanX and p15A-PKanY (strain 1), was constructed. The msDNA produced by each plasmid can edit the homologous sequence within the msd gene of the other plasmid. Sequencing results of the targeted msd gene region are provided. (C) Schematic diagrams illustrate strains 2, 3, and 4, each carrying different plasmid combinations, as shown in the figure. Sequencing data of the targeted msd gene regions from these strains reveal mutations introduced through editing, denoted by red letters.

Figure 4. Genomic lacZ gene editing in ΔmutSrecJsbcB strains harboring pUClacZ1, pUClacZ2, or pUClacZ3 plasmids.

(A) In E. coli strain carrying the pUClacZ plasmid, msDNA-PSlacZ shares the same sequence as the lagging strand in both the plasmid and genomic DNA replication forks. This enables msDNA integration into both plasmid msd and genomic lacZ. (B) Schematic diagram of the pUClacZ1 plasmid. Left: The overall structure of the pUClacZ1 plasmid. Note that the msd has been relocated in pUClacZ1. msDNA-PSlacZ integrates into the plasmid replication fork as the lagging strand due to its sequence complementarity with the red line sequence. This lagging-strand integration enables efficient msd editing. The detailed sequence of the plasmid is provided in Appendix Fig. S10. (C) Editing efficiency and growth in a colony-forming unit (CFU) of ΔmutSrecJsbcB-pUClacZ1 strain at different time points. The data were displayed as a scatter plot, with each data point representing an individual biological replicate (n = 3). (D) Schematic diagram of the pUClacZ2 plasmid. Left: The overall structure of the pUClacZ2 plasmid. Note that the msd has been relocated in pUClacZ2. The blue and orange dashed arrows represent the regions where the replication forks move on the left and right sides of the plasmid, respectively. The T-shaped symbol indicates the replication termination (Ter) site. msDNA-PSlacZ integrates into the plasmid replication fork as the leading strand. This leading-strand integration hinders efficient msd editing. The detailed sequence of the plasmid is provided in Appendix Fig. S11. (E) Editing efficiency and growth in CFU of ΔmutSrecJsbcB-pUClacZ2 strain at different time points. The data were displayed as a scatter plot, with each data point representing an individual biological replicate (n = 3). (F) Schematic diagram of the pUClacZ3 plasmid. Left: The overall structure of the pUClacZ3 plasmid. Note that the replication origin (ori) has been flipped horizontally, which in turn swaps the plasmid’s left and right replication forks. This reorientation shifts the replication terminus (ter) to the opposite end of the plasmid and positions msDNA-PSlacZ into the leading strand of the plasmid replication fork. This leading-strand integration hinders efficient msd editing. The detailed sequence of the plasmid is provided in Appendix Fig. S12. (G) Editing efficiency and growth in CFU of ΔmutSrecJsbcB-pUClacZ3 strain at different time points. The data were displayed as a scatter plot, with each data point representing an individual biological replicate (n = 3). Source data are available online for this figure.

Figure EV1. Schematic representation of the retron operon and msDNA-mediated genomic lacZ editing process.

(A) Top panel: The retron operon is represented by double-stranded DNA labeled as the coding strand and template strand, respectively. A strong constitutive promoter, J23119 (black arrow), drives transcription of the operon, which encodes msDNA (msr/msd), reverse transcriptase (ret), and CspRecT (recT). The coding regions of msr and msd are depicted as blue and gray rectangles, respectively. A segment within the msd coding region (gray rectangle), which shares significant sequence similarity with the genomic lacZ gene, is denoted as the PSlacZ-anti coding sequence (PS stands for partial sequence). The blue line aligns with the lacZ coding strand, while the red line represents its reverse complement. The mutation sequence is indicated by a yellow star. Transcription and reverse transcription produce msDNA, a hybrid of msr RNA (light blue) and cDNA (black). The msDNA incorporates lacZ homologous sequences, designated PSlacZ-anti (red). E. coli genomic DNA and plasmid are shown as double-stranded structures. The genomic ori and ter are marked in blue and yellow, respectively. The lacZ gene and its transcription direction are indicated by a purple arrow. During E. coli genomic DNA replication, msDNA (red lines) integrates into the single-stranded DNA region at the replication fork through base pairing. Notably, msDNA-PSlacZ-anti has the same sequence as the leading strand. The detailed sequence of the plasmid is provided in Appendix Fig. S1. Bottom panel: The retron operon expresses msDNA-PSlacZ, which incorporates lacZ homologous sequences designated PSlacZ (blue). Importantly, PSlacZ is the reverse complement of PSlacZ-anti. PSlacZ shares the same sequence as the lagging strand. (B) Effect of strain and editing template on genomic lacZ gene editing efficiency at 24 h of incubation. We compared the editing efficiency in WT (no mutation control), ΔmutS, ΔmutSrecJ, and ΔmutSrecJsbcB strains expressing msDNA-PSlacZ-anti or msDNA-PSlacZ. The data are presented as a bar graph, with each bar representing the mean, and error bars indicating the standard deviation. Each data point represents an individual biological replicate (n = 4). Statistically significant differences were determined using an unpaired Student’s t-test. The p value for the comparison of the editing efficiency between msDNA-PSlacZ-anti and msDNA-PSlacZ was 0.002714 in strain MG1655. In the ΔmutS, ΔmutSrecJ, and ΔmutSrecJsbcB strains, the p value for this comparison was <0.000001. **P < 0.005, ****P < 0.0001. (C) Sequencing results of the genomic lacZ sequence of the target region. Mutations change the codons for tryptophan and glutamate at positions 17 and 18 of the LacZ protein from TGG and GAA to stop codons (TGA and TAA). Red letters represent mutated bases. Source data are available online for this figure.

Figure EV2. Genomic ung and betI gene editing in ΔmutSrecJsbcB strains harboring pUCung or pUCbetI plasmids, respectively.

(A) A 92 bp homologous sequence of the ung gene (PSung coding sequence) was inserted into the msd region. The msDNA-PSung, aligning with the genomic DNA lagging strand, mediates ung gene editing, resulting in double base mutations. Notably, the msDNA-PSung aligns with the pUCung plasmid’s leading strand, preventing msd editing. The detailed sequence of the plasmid is provided in Appendix Fig. S13. (B) A 92 bp homologous sequence of the betI gene (PSbetI coding sequence) was inserted into the msd region. The msDNA-PSbetI, aligning with the genomic DNA lagging strand, mediates betI gene editing, resulting in one base deletion. Notably, the msDNA-PSbetI aligns with the pUCbetI plasmid’s leading strand, preventing msd editing. The detailed sequence of the plasmid is provided in Appendix Fig. S14. (C) Sequencing results of the targeted region in the genomic ung and betI genes. Four colonies were randomly selected from each plate, and these colonies were pooled separately for each strain (one pool for ΔmutSrecJsbcB-pUCung and one pool for ΔmutSrecJsbcB-pUCbetI) to increase the number of cells analyzed for mutations. The regions surrounding the targeted sequences in the genomic ung and betI genes were separately amplified by PCR for each pool, followed by DNA sequencing. The DNA sequencing results reflect the average editing efficiency within each pool. If editing is highly efficient (close to 100%), we expect to see a dominant peak corresponding to the mutated sequence, with a very small or undetectable peak for the wild-type sequence. Conversely, lower editing efficiency will result in a mixture of peaks, representing both mutated and wild-type sequences. Mutated bases are indicated in red.