Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Efficient and iterative retron-mediated in vivo recombineering in Escherichia coli

Efficient and iterative retron-mediated in vivo recombineering in Escherichia coli Submitted: 11 March 2022; Received (in revised form): 22 April 2022; Accepted: 2 May 2022 © The Author(s) 2022. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com for making mutations conferring easily selectable phenotypes, 1. Introduction this rate of recombination is not robust enough to obviate the need Genome editing technologies have proven to be invaluable molec- for extensive screening when making nonselectable edits (19). ular tools, enabling rapid advances in functional genomics, Farzadfard et al. (20) improve upon their original SCRIBE design metabolic engineering, therapeutics development, bioremedia- to achieve highly efficient ssDNA recombineering via transient tion and more (1–4). Traditional methods for genome editing knockdown of host exonuclease expression through CRISPR inter- use allelic exchange via the RecA-dependent process of homol- ference (CRISPRi), reporting nearly 100% recombination efficiency ogous recombination (HR). In HR, circular or linear DNA sub- for one of their chosen targets. Additionally, the authors demon- strates, encoding the desired modification flanked by targeted strate that SCRIBE enables the incorporation of multiple muta- regions of homology, are introduced into recombination-proficient tions at distinct loci (18, 20). However, with no easy method (ΔrecBCΔsbcBC or ΔrecD) cells (5, 6). Endogenous recombination for plasmid curing, doing so requires that each editing plasmid proteins facilitate crossover between the editing template and the contain a unique selection marker for plasmid maintenance, a target locus, incorporating the desired mutations into the host strategy limited by the number of orthogonal selection markers chromosome. While this process is useful for targeted mutage- and compatible plasmid origins available. nesis, it is sometimes inefficient and always laborious, requiring In this work, we sought to enable iterative genome editing numerous cloning steps to incorporate sizable regions of homol- by moving the functional components to easily curable plas- ogy (0.5–5 kb) and extensive screening to identify recombinant mids while maintaining the SCRIBE system’s high recombination cells (7). efficiency. We hypothesized that combining an improved SCRIBE The development of in vivo recombination-mediated engineer- system with Cas9 counterselection against wild-type cells would ing (recombineering), which relies on phage-derived proteins to provide an easy-to-use genome editing platform that precluded facilitate DNA integration, has greatly improved the speed and the need to electroporate ssDNA for recombineering. In addition, efficiency with which targeted mutations can be made. For exam- the system was designed so that the editing plasmid can be retar- ple, the λ-phage-derived Red system in Escherichia coli yields more geted through a single round of PCR amplification and cloning than a 100-fold increase in recombinant cells over traditional HR to reduce construction times and enable the rapid generation of through a process independent of RecA (8, 9). Recombineering multilocus mutants. reduces editing template design constraints by enabling efficient recombination with homology arms of <50 bp on double-stranded 2. Materials and methods DNA (10). In addition, short single-stranded DNA (ssDNA) oligonu- 2.1 Strains, plasmids and culture conditions cleotides (oligos) can be used as suitable recombineering sub- strates. In fact, ssDNA oligos have produced recombinant cells Bacterial strains used in this work are listed in Supplementary with just one of the three λ-Red proteins, the ssDNA-annealing Table S1. Plasmid construction methods and primers are described protein Beta, further simplifying the requirements for success- in Supplementary Table S2 (along with Addgene #s for avail- ful recombineering (9). Furthermore, the Clustered Regularly able plasmids) and the Primers_GeneFrags.fa file, respectively. The Interspaced Short Palindromic Repeats (CRISPR)–Cas9 bacterial pFF745 plasmid was a gift from Timothy Lu (Addgene #61450). immune system has been adapted as a programmable counter- Cloning steps for retargeting the editing plasmid are illustrated in selection tool to selectively target and kill unedited cells after Supplementary Figure S2. Briefly, primers with overhangs encod- recombineering. This technique enhances the apparent efficiency ing the new target sequence were used to PCR amplify the vector by increasing the mutant to wild-type ratio of the population, using Q5 High-Fidelity 2x Master Mix (New England Biolabs, NEB), allowing highly efficient scarless genome editing (11–14). Addi- followed by DpnI digestion to remove template DNA, agarose gel tional efforts for the improvement of λ-Red editing efficiency electrophoresis to verify amplicon size and column-based DNA have used engineered mutator strains, where the overexpres- cleanup. DNA fragments were then assembled using the 2x NEB- sion of HR-involved enzymes, the deletion of host exonucle- uilder HiFi DNA Assembly Master Mix (NEB) with the following ases and/or disabling the mismatch repair system significantly modifications: 1 μl Master Mix+ 0.5μl each DNA fragment, incu- improve recombination efficiency but can unintentionally lead to bated at 50 C for 15 min. Chemically competent E. coli NEB5α the accumulation of off-target mutations (15–17). cells were used for cloning. Antibiotics were used at the following Farzadfard and Lu (18) described a novel approach for recom- concentrations unless otherwise noted: spectinomycin (50μg/ml), bineering, termed Synthetic Cellular Recorders Integrating Biolog- kanamycin (50μg/ml), chloramphenicol (40μg/ml), rifampicin ical Events (SCRIBE), where the ssDNA is generated in vivo and (20μg/ml), tetracycline (20μg/ml) and carbenicillin (100μg/ml). then incorporated into the chromosome. The in vivo ssDNA is pro- The tetA gene was PCR amplified from pUC18-mini-Tn7T-Gm- duced by the Ec86 retron of E. coli BL21, composed of the msr TetAR using primers 2490 and 2491 and integrated into E. coli and msd RNA elements and the reverse transcriptase (RT). Upon MG1655 using λ-Red recombination with pKD46 (21), creating the transcription of the msr–msd sequence, inverted repeats flanking MG1655 yhiS::tetAR strain. the msr–msd RNA form a secondary structure specifically recog- 2.2 Recombineering assays and recombination nized by the RT and the msdRNA is reverse transcribed to ssDNA. efficiency determination Simultaneous expression of bet on the SCRIBE plasmid allows for efficient recombination of the newly synthesized ssDNA into Since the number of recombinant cells should increase with the lagging strand of the targeted locus during DNA replication time, all experiments were performed in freshly transformed cells. (Figure 1A). By placing control of this system under an inducible Approximately 100 ng of editing plasmid was transformed into promoter, the cells record their exposure to the inducer as the frac- competent E. coli cells via either heat shock at 42 C or electropo- tion of cells with the msd encoded mutation. By simply changing ration in a 0.1-cm gap cuvette with a single 1.8-kV pulse. Cells the msd sequence, SCRIBE can be easily retargeted to modify any were recovered in 1 ml SOB at 30 C for 2 h and then outgrown desired genomic site. The authors report a maximum efficiency for overnight at 30 C in 5 ml Luria Bertani (LB) broth+ appropriate –4 SCRIBE of 10 recombinants per generation (18). While sufficient antibiotic(s) to select transformants. The total number of viable cells was determined by spotting 10μl 10-fold serially diluted (22) were inserted upstream of mutL to allow for inducible overnight cultures on LB plates+ appropriate antibiotic(s). For the expression by the addition of cumate, yielding pCas9CyMutL. rpoB and tetR assays, 10μL of 10-fold serially diluted overnight cul- Escherichia coli MG1655 cells harboring pCas9CyMutL were then tures were spotted on LB plates with rifampicin or tetracycline to transformed with pTJV1Sc-rpoB1 or pTJV1Sc-tetA and recovered determine the number of recombinant cells. For the ackA assays, in 1 ml SOB at 30 C for 2 h, after which cells were transferred 1 ml of cells from overnight cultures were pelleted, washed and to 5 ml LB+ chloramphenicol+ spectinomycin with or without resuspended with sterile phosphate buffered saline after which cumate (100μM) and outgrown overnight at 30 C. Cultures were washed cells were spotted on M9 chloroacetate plates (M9 min- diluted 1:10 into LB+ chloramphenicol+ spectinomycin+ cumate imal media+ 10 mM sodium chloroacetate+ 2% glycerol+ 0.1% with or without aTc (0.1μg/ml) and grown overnight at 30 C. SOB) to determine the number of ackA mutant cells. Recom- Recombination efficiency and efficiency of Cas9 counterselec- bination efficiency was determined by dividing the number of tion were determined the following day, and the results are recombinant cells by the total number of viable transformants. reported as the mean with standard error for three independent Efficiencies reported are the mean and standard error for three replicates. independent replicates. For the rpoB experiments, spontaneous rifampicin resistance was assessed for the wild-type strain as 2.6 Off-target mutation frequency analysis described in Section 2.6 below and the spontaneous mutation Escherichia coli MG1655 was separately transformed with pTV1β- rate was subtracted from all samples plated on rifampicin. All rpoB and pCas9CyMutL and plated on LB+ spectinomycin and statistical analyses were performed using the GraphPad Prism 5 LB+ chloramphenicol, respectively, for the selection of transfor- program. mants. Single colonies from each transformation were used to inoculate 5 ml LB broth+ appropriate antibiotic in triplicate and 2.3 Promoter optimization grown overnight at 30 C, with pCas9CyMutL grown with or with- out cumate (100μM). Cultures were then diluted 1:10 into the The P promoter of the Ec86 retron cassette in pTLlSc-rpoB lacO same media and grown for an additional overnight at 30 C, after was replaced with P , and P was inserted upstream of J23101 VanCC which 10μL of 10-fold serial dilutions of each were spotted onto bet as described in Supplementary Table S2. Expression of the LB only and LB+ rifampicin to assess the spontaneous mutation Ec86 retron cassette and beta recombinase by P and P , J23101 VanCC frequency. respectively, was further optimized by PCR amplifying pTJVSc- For genome sequencing, a single colony from each transforma- rpoB with primers 2326 & 2358 and 2325 & 2359, followed by tion and wild-type MG1655 were grown in LB broth+ appropriate assembly of the two resulting amplicons using the 2x NEBuilder antibiotic overnight at 30 C, with pCas9CyMutL grown with and HiFi DNA Assembly Master Mix (NEB). The assembled pTJVSc- without cumate (100μM) for induction of mutL expression. Cul- rpoB promoter library was transformed into E. coli NEB5α cells, tures were then diluted 1:100, grown overnight and repeated and recombination efficiency was determined. In addition, the for a total of two setbacks. Genomic DNA was extracted after remainder of the pTJVSc-rpoB culture was diluted 1:100 into 5 ml the first setback for wild-type and after the second setback for LB broth+ rifampicin and grown overnight again at 30 C. The fol- the other samples using a cetyltrimethyl ammonium bromide lowing morning, the plasmid library was extracted, retransformed (CTAB)/phenol-chloroform extraction protocol (23). into E. coli NEB5α cells, and the process was repeated three more Samples were prepared for whole-genome sequencing using times. After the fourth round of selection, the pTJVSc-rpoB plas- the NEBNext Ultra II FS DNA Library Prep Kit for Illumina (NEB). mid was extracted from four randomly chosen transformants and Sequencing was performed using the NovaSeq6000 sequencing the P and P promoters were sequenced with primers 746 J23101 VanCC platform (Novogene Co. Sacramento, CA). Sequencing data were and 2252, respectively, to assess the convergence of the population quality filtered and adapters were trimmed using the Trim Galore on a single optimal promoter sequence, yielding pTJV1Sc-rpoB as script (24). Mutations were identified using the Breseq mutational indicated in Supplementary Table S2. analysis pipeline (25) set to polymorphism mode with the default parameters and a minimum coverage cutoff of 20× reads. The E. 2.4 Cas9 counterselection coli MG1655 reference genome (National Center for Biotechnology For pCas9CR4 for CRISPR/Cas9 directed counterselection against Information (NCBI) accession: NC_000913) was used as the refer- unedited wild-type cells, transformants with editing plasmid ence sequence. The wild-type MG1655 parent strain was used as were outgrown overnight. The following morning cultures were a control to assess differences between the reference genome and diluted 1:10 into LB+ appropriate antibiotics with or without our lab strain and sequence variations identified were subtracted anhydrotetracycline (aTc; 0.1μg/ml) to induce Cas9 expression from those found in our experimental samples. and then grown overnight at 30 C. Serial dilutions were spotted onto LB+ rifampicin for the rpoB assays, LB+ tetracycline for the 3. Results tetA reversion assays, or M9 chloroacetate for the ackA assays, then incubated overnight at 37 C. The efficiency of Cas9 counters- 3.1 Improvement of recombineering with election was determined by dividing the number of colonies from reverse-transcribed ssDNA the aTc-induced cultures that grew with selection by the total To enable an efficient and iterative system for genome editing number of viable cells plated on a nonselective plate. Results are with ssDNA reverse-transcribed in vivo, we moved the SCRIBE sys- reported as the mean with standard error for three independent tem’s functional components to the pKDsgRNA plasmid that has replicates. the temperature-sensitive variant of the pSC101 origin to allow for easy curing with growth at 37–42 C (21, 26, 27). This plas- 2.5 Efficiency improvement with negative mid, pTLlSc, possesses the msr-msd coding sequence and Ec86 RT mutator alleles for the synthesis of ssDNA, the bet gene to facilitate incorpora- The mutL gene was cloned into pCas9CR4 as described in tion of the ssDNA and the S. pyogenes single-guide RNA (sgRNA) Supplementary Table S2. The cymR repressor and P promoter to enable Cas9 targeting. The judicious design of this plasmid cymRC Figure 1. Optimization of the origin of replication and promoter elements enables an efficient and curable system for in vivo recombineering . (A) In vivo retron-generated ssDNA serves as an editing substrate for beta recombination at the lagging strand of the replication fork. (B) Rifampicin resistance conferred by the P564L mutation of the rpoB gene was used to measure the recombination efficiency. The pSC101 origin of replication allows for curing of the editing plasmid by growth at 42 C after desired mutation(s) are made. (C) Modifications made to promoters and origin of replication for enhanced recombination efficiency and curability. (D) Recombination efficiency of each editing plasmid reported as the average of three independent replicates. Error bars represent SEM, and statistical significance relative to pFF745–rpoB is denoted by asterisks (ns = not significant; **P < 0.01; ***P < 0.001; one-way analysis of variance and Tukey’s test of log-transformed values). enables easy one-step cloning to retarget both the msd and sgRNA was slightly higher than the original construct (Supplementary (Supplementary Figure S2). To assess the recombination efficiency Figure S1B), and sequencing of the promoter from randomly cho- using an easily selectable mutation, the msr-msd was retargeted to sen transformants revealed a convergence of the pool toward a introduce three consecutive nucleotide substitutions and create single promoter sequence that had a recombination efficiency the P564L point mutation of rpoB that confers rifampicin resis- of nearly 10-fold greater than the parent plasmid (Supplemen- tance (Figure 1B). Here, we define recombination efficiency as the tary Figure S1C–D). Moreover, the efficiency of the improved con- number of cells obtained on selective media divided by the total struct, pTJV1Sc-rpoB, was similar to the original SCRIBE plasmid, number of viable cells. First, we compared the efficiency of pFF745, pFF745–rpoB (Figure 1D). described by Farzadfard et al., which possesses a high-copy pUC 3.2 CRISPR/Cas9 counterselection against origin targeted to rpoB, with our temperature-sensitive plasmid wild-type cells pTLlSc-rpoB. As expected, recombination efficiency was about 100-fold lower with the temperature-sensitive plasmid, presum- We next investigated whether our improved in vivo recombi- ably because of decreased expression of the retron elements and neering system could be combined with CRISPR/Cas9 counter- bet compared to the high-copy pUC origin on pFF745 (Figure 1D). selection against wild-type cells for a highly efficient genome We sought to improve this efficiency by increasing the transcrip- editing system. The pCas9CR4 plasmid has cas9 under the con- tion of the msr-msd, RT and bet through promoter replacements. trol of an inducible P promoter that was engineered to enable tet The strong constitutive promoter P (28) replaced the P co-maintenance with a genome targeting sgRNA until induc- J23101 lacO promoter driving expression of the msr-msd RNA and RT. The tion with anhydrotetracycline (aTc) (27). We first transformed promoter of bet was changed to P (22) because it is both pCas9CR4 into the E. coli strains NEB5α, BL21-DE3 and MG1655, VanCC strong and nonhomologous to P . Replacing these promoters and subsequently transformed the pTJV1Sc-rpoB editing vector, J23101 individually did not increase recombination efficiency, but when which possessed an sgRNA targeted to wild-type rpoB. After recov- combined, efficiency increased nearly 10-fold (Figure 1D ). ery, the cells were transferred to LB broth with spectinomycin To further increase recombination efficiency, we used a selec- and chloramphenicol to select for both plasmids and grown tion strategy where a library of variant plasmids was created by overnight. The following morning, the cultures were then passed cloning degeneracies into both the P and P promoters. into media with or without aTc and grown an additional overnight J23101 VanCC We hypothesized that variants with increased efficiency would for induction of Cas9 expression and killing of unedited cells. overtake the population after repeated rounds of outgrowth, selec- Cells were then spotted onto plates with and without rifampicin tion, plasmid extraction and re-transformation (Supplementary to directly select for the rpoB mutation and assess total viable Figure S1A). After four passages, the efficiency of the evolved pool cells, respectively (Figure 2A). The recombination efficiency was Figure 2. Targeted counterselection against unedited cells by inducible cas9 expression. (A) After Beta-recombination, cells are sub-cultured 1:10 into WT media with aTc to induce cas9 expression from pCasCR4. Constitutive expression of an sgRNA from pTJV1Sc-rpoB targets rpoB alleles for DNA DSB by Cas9 cleavage, thereby eliminating nonmutated cells. (B) Escherichia coli NEB5α, BL21 and MG1655 cells were transformed with pTJV1Sc-rpoB and the RpoB frequencies were determined by plating on LB+ rifampicin. (C) Escherichia coli MG1655::tetAR was transformed with pTJV1Sc-rpoB, P564L pTJV1Sc-tetA and pTJV1Sc-ackA. Frequencies of the RpoB , TetA and AckA mutations were determined by plating cells on LB+ rifampicin, P564L *70Y E54* tetracycline and chloroacetate, respectively. (D) Sequential mutation of distinct loci was performed by transforming E. coli MG1655 with pTJV1Sc-rpoB, determining recombination efficiency by plating on LB + rifampicin, curing pTJV1Sc-rpoB by growing cells at 42 C, then transforming once more with pTJV1Sc-ackA. Double mutant frequency was assessed by plating on LB+ rifampicin+ sodium chloroacetate. Averages are based on three independent replicates. Error bars represent SEM. Statistical significance is denoted by asterisks (** = P-value < 0.01; ***= P-value <0.001; two-tailed Student’s t-test of log-transformed values). –5 –3 between 1 × 10 and 3 × 10 without induction of Cas9 expression and are thus resistant to the toxic acetate analog chloroacetate. for all three strains. However, when cas9 was induced, the average A recombination efficiency of 5.6% without Cas9 counterselec- efficiency was ∼5% for BL21 and 60–80% for NEB5α and MG1655, tion and over 100% with induction of Cas9 was obtained, as indicating a substantial reduction in wild-type cells (Figure 2B). slightly more colonies were observed on the selection plate than To demonstrate that our system could mutagenize different on the nonselective plate on average (Figure 2C). These experi- targets, we examined the recombination efficiency of tetA and ments show that targeted counterselection of unedited cells with ackA. We inserted the tetA encoded tetracycline efflux pump that Cas9 can successfully enrich for recombinants across distinct loci. possessed a premature stop codon of about 200 bp from the The pCas9CR4 plasmid used in these experiments was start codon into MG1655 using traditional λ-Red recombineering designed for tight repression of cas9 and an ssrA degradation tag (21). Our experiments then mutated the stop codon back to a decreases Cas9 stability, enabling co-maintenance of both genome sense codon, creating the TetA mutation to enable the full- targeting sgRNA template and cas9 on two plasmids. Although *70Y length translation of tetA and confer tetracycline resistance to the induction of cas9 is required for cell death by double-strand break cells. We obtained tetracycline-resistant colonies with efficiencies (DSB), we questioned whether some transient level of Cas9 expres- –4 –5 of 5.4 × 10 and 1.8 × 10 with and without Cas9 counterse- sion also enriches for recombinants, thus we performed exper- lection, respectively, much lower than those observed for rpoB iments to compare the number of mutants obtained with the (Figure 2C). Next, the acetate kinase gene, ackA, was mutated SCRIBE system in the presence and absence of pCas9CR4. More with three nucleotide point mutations to create a premature stop recombinant cells were produced for all three targets with cells codon (AckA ) so that the cells are unable to metabolize acetate harboring pCas9CR4 than cells without, even without induction E54* of cas9 expression (Figure 2C). In fact, we were unable to obtain repair of mismatches within 3 bp (17, 31). The introduction of tetracycline-resistant cells when using the pTJV1Sc-tetA plasmid several adjacent mismatches also prevents repair by MMR (15). alone. These results suggest that even low levels of cas9 expres- However, these strategies present design constraints that can- sion in the uninduced state can increase recombination efficiency not always be followed. An alternate strategy is the recombinant possibly by slowing the growth of wild-type cells that must repair expression of negative mutS or mutL variants that are dominant DSB or because this low-level DSB stimulates recombination. over wild-type enzymes and prevent efficient repair (32). Accord- The proposed mechanism of ssDNA recombineering asserts ingly, we cloned mutL with the dominant negative E32K muta- that allelic replacement occurs at the replication fork, where tion into the pCas9CR4 plasmid under the control of a cumate the supplied ssDNA replaces an Okazaki fragment on the lag- inducible promoter (22) to maintain a two-plasmid system with ging strand. This mechanism results in a lagging strand bias, independent control of cas9 and mutL (Figure 3A). To test this sys- where recombination efficiencies for oligos targeting the lagging tem, we introduced a single bp mismatch of A:C in rpoB with strand are higher than those targeting the leading strand. To and without induction of the negative mutL. In cells harboring assess whether in vivo produced ssDNA proceeds through the pCas9CyMutL with mutL in the off-state, the average frequency –6 same mechanism, we targeted the leading strand of rpoB for edit- of rifampicin-resistant mutants was 4.9 × 10 without Cas9- –4 ing (Supplementary Figure S3A). Surprisingly, this resulted in a induction and 4.5 × 10 with Cas9 counterselection (Figure 3B). similar efficiency as targeting the lagging strand (Supplementary In contrast, the induction of mutL enabled rpoB mutations with Figure S3B). We wondered whether this was a function of the in an average recombination efficiency of 0.2% and 18% without vivo generation of ssDNA or if the same would be true of the rpoB and with Cas9 counterselection (Figure 3B). We also re-examined target using traditional λ-Red recombineering. We constructed a the tetA reversion assays, which introduced a single bp T:T mis- control plasmid in which the msr-msd and RT coding sequences match that is less efficiently repaired than the A:C mismatch used were removed (pTV1β-rpoB), resulting in constitutive expression for rpoB. Induction of the mutator allele increased the recombi- of Beta only and sgRNA targeting rpoB. We then performed a tra- nation efficiency by about 100-fold both with and without Cas9 ditional recombineering experiment by transforming exogenous counterselection, achieving a maximum efficiency of 0.3% on aver- ssDNA into cells with this plasmid and pCas9CR4 and observed age (Figure 3C). When generating a single A:G mismatch in the that the lagging oligo was much more efficient, consistent with ackA gene, mutL induction improved recombination by 1000- and previous observations (7, 9, 29) and the proposed mechanism of 100-fold with and without Cas9 counterselection, respectively, ssDNA recombineering (Supplementary Figure S3C). To evaluate resulting in an average maximum efficiency of 56% ( Figure 3D). whether the same phenomenon is true of other targets using These results further highlight the versatility of our system in SCRIBE, we targeted the leading strand of tetA with the in vivo sys- generating precise mutations across a diverse range of loci. tem (Supplementary Figure S3A). In this case, over 1000-fold fewer To further illustrate the efficiency enhancement these tools recombinant cells were obtained than when targeting the lagging offer over previous methods, we performed the same experiments strand (Supplementary Figure S3D). While we suspect that recom- using the no-SCAR method described previously (12). Single- bineering with in vivo ssDNA occurs at the lagging strand of the stranded editing oligos targeting the lagging strand were used to replication fork, our data suggest that the constant availability introduce the same single bp mismatches as before in cells harbor- of ssDNA may be sufficient to overcome the lagging strand bias ing pCas9CyMutL and pKDsgRNA, a plasmid encoding the λ-Red in some cases or that an alternative mechanism may influence proteins Exo, Beta and Gam under the control of the arabinose- recombination rates at some locations. inducible P promoter, as well as an sgRNA targeting the wild BAD type gene sequences. When targeting rpoB, induction of mutL before transformation, during recovery, or both achieved sim- ilarly high recombination rates as the retron system, yielding 3.3 Iterative mutagenesis of two targets average efficiencies >2% with direct selection on rifampicin and Above all, the high efficiency of mutagenesis found at our target 100% with Cas9 counterselection, a 20-fold increase in recom- sites confirms a robust system for genome editing, precluding the binants than without mutL (Supplementary Figure S5A). Regard- need to screen large numbers of colonies to identify mutants. To less of whether mutL was induced, Cas9 counterselection effec- demonstrate that our system could iteratively construct genome tively removed unedited cells with 95% of cells plated on aTc modifications, cells that were mutated at the rpoB locus were being rifampicin-resistant even without mutL expression. On the grown at 42 C to cure the cells of pTJV1Sc-rpoB. Subsequent trans- contrary, when targeting ackA, mutL induction improved the formation with pTJV1Sc-ackA enabled mutation of the ackA gene recombination efficiency less than 10-fold, yielding a maximum and produced cells resistant to both rifampicin and chloroacetate –4 efficiency of 4.8 × 10 when plated on chloroacetate (Supplemen- (Figure 2D). Amplification and Sanger sequencing of the rpoB and tary Figure S5B). However, sequencing the ackA gene of several ackA loci showed that both intended mutations were made, con- putative mutant colonies on the chloroacetate selection media firming that iterative use of the system could edit multiple loci revealed none had acquired the intended mutation. Furthermore, independently (Supplementary Figure S4). Cas9 counterselection did not substantially enrich for mutants, with <5% of the colonies on the aTc-induced plate able to grow with chloroacetate when patched (Supplementary Figure S5B), 3.4 Co-Expression of dominant-negative mutL indicating most of the colonies on the counterselection plate Recombineering with ssDNA results in transient production of escaped Cas9 killing. These results are consistent with previous heteroduplex DNA that the methyl-directed mismatch repair studies examining the rate of Cas9 escape in bacteria, which −3 −4 (MMR) system can repair, thus decreasing the observed rate of report frequencies ranging from ∼10 to 10 in multiple species mutagenesis. Several strategies are known to decrease the rate of (11, 27, 33, 34). Overall, the combined mutL and retron-based sys- MMR and enable more efficient mutagenesis (15, 17, 30–32). For tems result in consistently high mutagenesis rates for single bp example, single bp mismatches are repaired with different effi- edits and can outperform traditional oligo recombineering meth- ciencies, while C:C mismatches evade repair entirely and prevent ods for certain target loci. Figure 3. Expression of dominant-negative MutL allele enhances recombination efficiency for single nucleotide point mutations. (A) The MutL gene E32K was cloned into the pCas9CR4 plasmid under the control of the P promoter for inducible expression. When cumate is added, MutL is CymRC E32K expressed and competes with the native MutL for binding to MutS during mismatch repair. The E32K mutation prevents MutH binding, inhibiting removal and repair of the mismatched base. Cells harboring pCas9CyMutL were transformed with (B) pTJV1Sc-rpoB1, (C) pTJV1Sc-tetA or (D) pTJV1Sc-ackA1 encoding single point mutations for generating the RpoB , TetA and AckA mutations, respectively, and grown in liquid P564L *70Y E54* culture with or without cumate. The unmodified pCas9CR4 in both wild-type E. coli and a mutS mutant were used as controls. Recombination efficiency with and without Cas9 counterselection was assessed by plating on LB + rifampicin, LB+ tetracycline or M9+ chloroacetate. Data reported are the average of three independent replicates. Error bars represent SEM and statistical significance relative to pCas9CR4 is denoted by asterisks (ns= not significant; * = P-value < 0.05; **= P-value < 0.01; ***= P-value <0.001; one-way analysis of variance and Tukey’s test of log-transformed values). Hindrance of MMR increases the frequency of background However, induction of the mutator mutL resulted in an average –5 genome mutations (35, 36); however, the inducibility of our sys- mutation frequency of about 1.3 × 10 , an ∼20-fold increase over tem minimizes the amount of time in which MMR is inhibited. the wild-type (Supplementary Figure S6A). Similarly, uncontrolled expression of the λ-Red genes has also While these results indicate that the unexpressed mutL and been shown to increase the rate of spontaneous mutations (37). constitutive bet did not increase the rate of point mutations, As such, constitutive expression of bet in our system could con- we wanted to assess further whether other genomic mutations ceivably produce unintended recombination events in the cell. and rearrangements that would not manifest as rifampicin resis- To assess the background rate of mutagenesis in cells with our tance were present. Accordingly, we performed whole-genome SCRIBE system, we measured spontaneous rifampicin resistance sequencing on the strains grown for two overnights to iden- (38, 39). Cells were grown overnight, diluted 1:10, grown overnight tify polymorphisms within the entire population. A complete and then spotted on rifampicin to assess the mutation frequency. list of the mutations identified is in Supplemental File 2. Our The wild-type cells spontaneously acquired rifampicin resistance analysis revealed 41 total mutations in the population constitu- –7 –11 at a frequency of 6 × 10 on average. Cells constitutively express- tively expressing bet, constituting a mutation rate of 5.9 × 10 ing bet and cells with the uninduced pCas9CyMutL did not dif- mutations per cell per generation (Supplementary Figure S6B), fer significantly from the wild-type (Supplementary Figure S6A). with one new IS5 mobile element insertion into the rclA gene. Figure 4. Alternative recombinases can enhance efficiency for some targets. (A) The bet CDS on pTJV1Sc was replaced with genes encoding the CspRecT and EcRecT recombinases. Recombination efficiency for each of the recombinase variants with and without induction of cas9 expression was assessed for (B) the RpoB mutation by plating cells on LB+ rifampicin, (C) the AckA mutation by plating cells on LB+ chloroacetate and (D) the P564L E54* TetA mutation by plating cells on LB+ tetracycline. The pCas9CyMutL plasmid was used for the tetA experiments. Data reported are the average of *70Y three independent replicates. Error bars represent SEM and statistical significance is denoted by asterisks (ns = not significant; *** = P-value <0.001; one-way analysis of variance and Tukey’s test of log-transformed values). The uninduced pCas9CyMutL cells displayed a slightly higher binding protein promoted the highest number of recombinants –10 mutation rate of 2.4 × 10 ; however, both rates were similar to the between the different targets. For the rpoB mutation, EcRecT was –10 wild-type control, which exhibited a mutation rate of 1.0 × 10 . more efficient than both Beta and CspRecT, with 92% efficiency In contrast, 478 mutations were found in the population with when using Cas9 counterselection (Figure 4B). On the contrary, the –9 mutL induced for a mutation rate of 1.59 × 10 mutations per ackA mutation was more efficient with Beta, where 80% of cells cell per generation (Supplementary Figure S6B), as well as a dele- were mutants compared to 4.5% for CspRecT and no ackA mutants tion of the ∼1200 bp insH21 mobile element. Even still, this low with EcRecT (Figure 4C). Given the improvement in efficiency frequency of background mutations indicates selecting a mutant seen for tetA when expressing MutL , we used pCas9CyMutL E32K with unintended mutations would be a rare occurrence. Over- in combination with the alternative recombinases. We obtained all, these results demonstrate that maintenance of plasmids with over 100% efficiency for the tetA reversion with CspRecT, both the mutL mutator allele or constitutive expression of bet do not with and without Cas9 counterselection. Although slightly more increase the frequency of background mutations to a level that colonies were observed on the tetracycline plate than on the non- would cause concern from a genetic engineering standpoint. selective plate for two of the replicates (Figure 4D), the CspRecT clearly outperformed Beta and EcRecT where relatively low levels of mutants were found. Thus, in circumstances where a muta- tion is difficult to obtain, it is advisable to try these alternative 3.5 Alternative recombinases enhance efficiency recombinases, while the mechanism causing these discrepancies for some targets warrants further investigation. Recently, Wannier et al. identified recT from a Collinsella stercoris phage (CspRecT) with improved recombination efficiency com- pared to Beta for oligo mediated recombineering in E. coli (40). 4. Discussion To examine the effect of alternative recombinases in our sys- tem, we replaced bet with CspRecT, as well as EcRecT from In moving the SCRIBE machinery onto the temperature-sensitive the E. coli BL21-DE3 chromosome (Figure 4A). We then compared pSC101 origin of replication and combining it with Cas9 coun- the efficiency of these plasmids across three different targets. terselection, we successfully constructed an in vivo system for Surprisingly, our results showed inconsistency for which DNA- ssDNA recombineering that is highly efficient and iterative. Targeted counterselection using CRISPR/Cas9 enabled efficiencies distant loci within the same cell through repeated rounds of near 100% after a single night of outgrowth post-transformation, library recombineering. reducing the number of colonies needing to be screened for mutant identification. Moreover, both the ssDNA tem- Supplementary data plate and sgRNA sequence can be retargeted for modification Supplementary data are available at SYNBIO online. of any genomic locus with an appropriate PAM in a single cloning step using sequence overlap cloning methods. Numer- Data availability ous loci can be mutated without the need for orthogonal selection markers or extensive screening procedures, increas- Raw sequence reads for the off-target mutation frequency exper- ing the speed and ease of generating desirable mutants. In iments are available under the BioProject ID PRJNA813865 in the addition, the pCas9CR4 and pCas9CyMutL plasmids can be NCBI databases. cured using pKDsgRNA-p15 (Addgene #62656) as previously described (12), yielding plasmid-free mutants for downstream Funding applications. This work was supported by the Department of Microbiology and Recombineering is an important tool in the biological sci- Cell Science, Institute of Food and Agricultural Sciences, Univer- ences, and continued enhancement of recombination efficiency sity of Florida. across diverse target loci will decrease strain construction times. Recently, Lopez et al. showed that increasing the abundance of Acknowledgments ssDNA produced by retrons significantly improves recombina- tion efficiencies (41), consistent with the enhanced efficiencies We would like to thank the 2019 University of Florida International we observed when increasing the msr–msd/RT expression and bet Genetically Engineered Machine team for inspiring the initial work through promoter modifications. A complementary strategy is to on this project and L. Trujillo Rodriguez and L.A. Schuster for reduce the rate of ssDNA degradation by host nucleases, as the their support and advice during experimentation and manuscript High-efficiency SCRIBE (HiSCRIBE) system by Farzadfard et al. (20) preparation. describes. This system improves upon their original SCRIBE design by coupling a strong ribosome-binding site for bet on the SCRIBE Author contributions plasmid with CRISPRi-enabled transcriptional interference of the C.R.R. conceived this study. A.J.E. performed the experiments and endogenous exonucleases. Recombination efficiencies of nearly data analysis under the supervision of C.R.R. A.J.E. and C.R.R. 100% for galK and ∼25% for a kanamycin resistance gene tar- prepared the manuscript. get were obtained with these improvements. However, employing the nuclease-deficient dCas9 for CRISPRi prevents the use of the Conflict of interest statement. No potential conflict of interest was nuclease-active Cas9 for counterselection. Although the authors reported by the authors. show that CRISPR/Cas9 counterselection can select for recombi- nant cells at high efficiencies in an exonuclease knockout strain, this strategy prevents easy system portability since a mutant References strain must be used. 1. Baba,T., Ara,T., Hasegawa,M., Takai,Y., Okumura,Y., Baba,M., Lim et al. (42) also describe a CRISPR/retron-based editing sys- Datsenko,K.A., Tomita,M., Wanner,B.L. and Mori,H. (2006) Con- tem enabling trackable editing and functional screening across struction of Escherichia coli K-12 in-frame, single-gene knockout multiple loci in a population. However, this system requires the mutants: the Keio collection. Mol. Syst. Biol., 2, 1–11. sequential transformation and maintenance of three plasmids. 2. Choi,K.R., Jang,W.D., Yang,D., Cho,J.S., Park,D. and Lee,S.Y. (2019) Moreover, the construction of their retron/sgRNA plasmid requires Systems metabolic engineering strategies: integrating systems multiple enzymatic digestions for cloning, contrary to our two- and synthetic biology with metabolic engineering. Trends Biotech- plasmid system, which simultaneously retargets the retron and nol., 37, 817–837. sgRNA in a single two-fragment cloning step. Most importantly, 3. Jaiswal,S., Singh,D.K. and Shukla,P. (2019) Gene editing and sys- both the HiSCRIBE and the CRISPR/retron system of Lim et al. tems biology tools for pesticide bioremediation: a review. Front. (20, 42) would require multiple orthogonal selection markers to Microbiol., 10, 87. mutate multiple loci within a single cell, as no easy method of 4. Chen,Y., Batra,H., Dong,J., Chen,C., Rao,V.B. and Tao,P. (2019) plasmid curing is included in either design, preventing iterative Genetic engineering of bacteriophages against infectious dis- mutagenesis. Although the recombination efficiencies achieved eases. Front. Microbiol., 10, 954. with our system in the absence of counterselection are less 5. Jasin,M., Schimmel,P., Winans,S.C., Elledge,S.J., Krueger,J.H. and than those reported for HiSCRIBE, Cas9-induced cell death of Walker,G.C. (1984) Deletion of an essential gene in Escherichia unedited cells in the population enabled apparent efficiencies on coli by site-specific recombination with linear DNA fragments. J. par with HiSCRIBE. The ease with which our system can be retar- Bacteriol., 159, 783–786. geted and used iteratively to incorporate multiple edits combined 6. Russell,C.B., Thaler,D.S. and Dahlquist,F.W. (1989) Chromosomal with its high efficiency in mutant selection provides a quick and transformation of Escherichia coli recD strains with linearized plas- straightforward method for constructing mutants relevant to both mids. J. Bacteriol., 171, 2609–2613. research and biotechnological applications. Methods enabling 7. Court,D.L., Sawitzke,J.A. and Thomason,L.C. (2002) Genetic engi- multiplexed editing of numerous genomic loci within populations neering using homologous recombination. Annu. Rev. Genet., 36, and continuous evolution of targeted loci, using retron-mediated 361–388. recombineering, have been recently described (20, 42–44) and rely 8. Muyrers,J.P.P., Zhang,Y., Testa,G. and Stewart,A.F. (1999) on efficient editing for generating a library of variants. Our sys- Rapid modification of bacterial artificial chromosomes by tem would be well suited for constructing such libraries, with ET-recombination. Nucleic Acids Res., 27, 1555–1557. the added advantage of being curable, thus enabling editing of 9. Ellis,H.M., Yu,D., DiTizio,T. and Court,D.L. (2001) High efficiency 28. Anderson,J.C. (2006) http://parts.igem.org/Promoters/Catalog/ mutagenesis, repair, and engineering of chromosomal DNA using Anderson (10 January 2021, date last accessed). single-stranded oligonucleotides. Proc. Natl. Acad. Sci. U. S. A., 98, 29. Zhang,Y., Muyrers,J.P.P.P., Rientjes,J. and Stewart,F. (2003) 6742–6746. Phage annealing proteins promote oligonucleotide-directed 10. Zhang,Y., Buchholz,F., Muyrers,J.P.P. and Stewart,A.F. (1998) A new mutagenesis in Escherichia coli and mouse ES cells. BMC Mol. logic for DNA engineering using recombination in Escherichia coli. Biol., 4, 1. ̈ ́ Nat. Genet., 20, 123–128. 30. Nyerges,A., Csorgo,B., Nagy,I., Latinovics,D., Szamecz,B., Posfai,G. 11. Jiang,W., Bikard,D., Cox,D., Zhang,F. and Marraffini,L.A. (2013) and Pal,C. (2014) Conditional DNA repair mutants enable highly RNA-guided editing of bacterial genomes using CRISPR–Cas sys- precise genome engineering. Nucleic Acids Res., 42, e62. tems. Nat. Biotechnol., 31, 233–239. 31. Kramer,B., Kramer,W. and Fritz,H.J. (1984) Different base/base 12. Reisch,C.R. and Prather,K.L.J. (2017) Scarless Cas9 assisted recom- mismatches are corrected with different efficiencies by the bineering (no-SCAR) in Escherichia coli, an easy-to-use system for methyl-directed DNA mismatch-repair system of E. coli. Cell, 38, genome editing. Curr. Protoc. Mol. Biol., 117, 31.8.1–31.8.20. 879–887. ̈ ́ ́ ́ 13. Pyne,M.E., Moo-Young,M., Chung,D.A. and Chou,C.P. (2015) Cou- 32. Nyerges,A., Csorgo,B., Nagy,I., Balint,B., Bihari,P., Lazar,V., pling the CRISPR/Cas9 system with lambda red recombineering Apjok,G., Umenhoffer,K., Bogos,B., Posfai,G. et al. (2016) A highly enables simplified chromosomal gene replacement in Escherichia precise and portable genome engineering method allows compar- coli. Appl. Environ. Microbiol., 81, 5103–5114. ison of mutational effects across bacterial species. Proc. Natl. Acad. 14. Jiang,Y., Chen,B., Duan,C., Sun,B., Yang,J. and Yang,S. (2015) Multi- Sci. U. S. A., 113, 2502–2507. gene editing in the Escherichia coli genome via the CRISPR–Cas9 33. Oh,J.H. and Van Pijkeren,J.P. (2014) CRISPR–Cas9-assisted recom- system. Appl. Environ. Microbiol., 81, 2506–2514. bineering in Lactobacillus reuteri. Nucleic Acids Res., 42, e131. 15. Sawitzke,J.A., Costantino,N., Li,X.T., Thomason,L.C., 34. Liu,J., Wang,Y., Lu,Y., Zheng,P., Sun,J. and Ma,Y. (2017) Develop- Bubunenko,M., Court,C. and Court,D.L. (2011) Probing cellu- ment of a CRISPR/Cas9 genome editing toolbox for Corynebac- lar processes with oligo-mediated recombination and using the terium glutamicum. Microb. Cell Fact., 16, 205. knowledge gained to optimize recombineering. J. Mol. Biol., 407, 35. Schaaper,R.M. and Dunn,R.L. (1987) Spectra of spontaneous 45–59. mutations in Escherichia coli strains defective in mismatch correc- 16. Mosberg,J.A., Gregg,C.J., Lajoie,M.J., Wang,H.H. and Church,G.M. tion: the nature of in vivo DNA replication errors. Proc. Natl. Acad. (2012) Improving lambda red genome engineering in Escherichia Sci. U. S. A., 84, 6220–6224. coli via rational removal of endogenous nucleases. PLoS One, 7, 36. Isaacs,F.J., Carr,P.A., Wang,H.H., Lajoie,M.J., Sterling,B., Kraal,L., e44638. Tolonen,A.C., Gianoulis,T.A., Goodman,D.B., Reppas,N.B. et al. 17. Costantino,N. and Court,D.L. (2003) Enhanced levels of λ Red- (2011) Precise manipulation of chromosomes in vivo enables mediated recombinants in mismatch repair mutants. Proc. Natl. genome-wide codon replacement. Science, 333, 348–353. Acad. Sci. U. S. A., 100, 15748–15753. 37. Murphy,K.C. and Campellone,K.G. (2003) Lambda Red-mediated 18. Farzadfard,F. and Lu,T.K. (2014) Genomically encoded analog recombinogenic engineering of enterohemorrhagic and memory with precise in vivo DNA writing in living cell popula- enteropathogenic E. coli. BMC Mol. Biol., 4, 11. tions. Science, 346, 1256272. 38. Garibyan,L., Huang,T., Kim,M., Wolff,E., Nguyen,A., Nguyen,T., 19. Wang,H.H., Isaacs,F.J., Carr,P.A., Sun,Z.Z., Xu,G., Forest,C.R. and Diep,A., Hu,K., Iverson,A., Yang,H. et al. (2003) Use of the rpoB Church,G.M. (2009) Programming cells by multiplex genome engi- gene to determine the specificity of base substitution muta- neering and accelerated evolution. Nature, 460, 894–898. tions on the Escherichia coli chromosome. DNA Repair (Amst), 2, 20. Farzadfard,F., Gharaei,N., Citorik,R.J. and Lu,T.K. (2021) Efficient 593–608. retroelement-mediated DNA writing in bacteria. Cell Syst., 12, 39. Lee,H., Popodi,E., Tang,H. and Foster,P.L. (2012) Rate and molec- 860–872.e5. ular spectrum of spontaneous mutations in the bacterium 21. Datsenko,K.A. and Wanner,B.L. (2000) One-step inactivation of Escherichia coli as determined by whole-genome sequencing. Proc. chromosomal genes in Escherichia coli K-12 using PCR products. Natl. Acad. Sci. U. S. A., 109, E2774–E2783. Proc. Natl. Acad. Sci. U. S. A., 97, 6640–6645. 40. Wannier,T.M., Nyerges,A., Kuchwara,H.M., Czikkely,M., Balogh,D., 22. Meyer,A.J., Segall-Shapiro,T.H., Glassey,E., Zhang,J. and Voigt,C.A. Filsinger,G.T., Borders,N.C., Gregg,C.J., Lajoie,M.J., Rios,X. et al. (2019) Escherichia coli “Marionette” strains with 12 highly optimized (2020) Improved bacterial recombineering by parallelized protein small-molecule sensors. Nat. Chem. Biol., 15, 196–204. discovery. PNAS, 117, 13689–13698. 23. Feil,W.S., Feil,H. and Copeland,A. (2012) Bacterial genomic DNA 41. Lopez,S.C., Crawford,K.D., Lear,S.K., Bhattarai-Kline,S. and Ship- isolation using CTAB. https://www.jgi.doe.gov (10 January 2021, man,S.L. (2022) Precise genome editing across kingdoms of life date last accessed). using retron-derived DNA. Nat. Chem. Biol., 18, 199–206. 24. Krueger,F. (2021) TrimGalore: a wrapper around Cutadapt and 42. Lim,H., Jun,S., Park,M., Lim,J., Jeong,J., Lee,J.H. and Bang,D. (2020) FastQC to consistently apply adapter and quality trimming to Multiplex generation, tracking, and functional screening of sub- FastQ files, with extra functionality for RRBS data. GitHub. stitution mutants using a crispr/retron system. ACS Synth. Biol., 9, 25. Deatherage,D.E. and Barrick,J.E. (2014) Identification of mutations 1003–1009. in laboratory-evolved microbes from next-generation sequencing 43. Simon,A.J., Morrow,B.R. and Ellington,A.D. (2018) Retroelement- data using breseq. Methods Mol. Biol., 1151, 165–188. based genome editing and evolution. ACS Synth. Biol., 7, 26. Hashimoto-Gotoh,T. and Sekiguchi,M. (1977) Mutations of tem- 2600–2611. perature sensitivity in R plasmid pSC101. J. Bacteriol., 131, 405. 44. Schubert,M.G., Goodman,D.B., Wannier,T.M., Kaur,D., Lu,T.K., 27. Reisch,C.R. and Prather,K.L.J.J. (2015) The no-SCAR (Scarless Shipman,S.L. and Church,G.M. (2021) High throughput functional Cas9 Assisted Recombineering) system for genome editing in variant screens via in-vivo production of single-stranded DNA. Escherichia coli. Sci. Rep., 5, 15096. PNAS, 118, e2018181118. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Synthetic Biology Oxford University Press

Efficient and iterative retron-mediated in vivo recombineering in Escherichia coli

Synthetic Biology , Volume 7 (1): 1 – May 3, 2022

Loading next page...
 
/lp/oxford-university-press/efficient-and-iterative-retron-mediated-in-vivo-recombineering-in-e-Rn5ZgTv5YU
Publisher
Oxford University Press
Copyright
© The Author(s) 2022. Published by Oxford University Press.
eISSN
2397-7000
DOI
10.1093/synbio/ysac007
Publisher site
See Article on Publisher Site

Abstract

Submitted: 11 March 2022; Received (in revised form): 22 April 2022; Accepted: 2 May 2022 © The Author(s) 2022. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com for making mutations conferring easily selectable phenotypes, 1. Introduction this rate of recombination is not robust enough to obviate the need Genome editing technologies have proven to be invaluable molec- for extensive screening when making nonselectable edits (19). ular tools, enabling rapid advances in functional genomics, Farzadfard et al. (20) improve upon their original SCRIBE design metabolic engineering, therapeutics development, bioremedia- to achieve highly efficient ssDNA recombineering via transient tion and more (1–4). Traditional methods for genome editing knockdown of host exonuclease expression through CRISPR inter- use allelic exchange via the RecA-dependent process of homol- ference (CRISPRi), reporting nearly 100% recombination efficiency ogous recombination (HR). In HR, circular or linear DNA sub- for one of their chosen targets. Additionally, the authors demon- strates, encoding the desired modification flanked by targeted strate that SCRIBE enables the incorporation of multiple muta- regions of homology, are introduced into recombination-proficient tions at distinct loci (18, 20). However, with no easy method (ΔrecBCΔsbcBC or ΔrecD) cells (5, 6). Endogenous recombination for plasmid curing, doing so requires that each editing plasmid proteins facilitate crossover between the editing template and the contain a unique selection marker for plasmid maintenance, a target locus, incorporating the desired mutations into the host strategy limited by the number of orthogonal selection markers chromosome. While this process is useful for targeted mutage- and compatible plasmid origins available. nesis, it is sometimes inefficient and always laborious, requiring In this work, we sought to enable iterative genome editing numerous cloning steps to incorporate sizable regions of homol- by moving the functional components to easily curable plas- ogy (0.5–5 kb) and extensive screening to identify recombinant mids while maintaining the SCRIBE system’s high recombination cells (7). efficiency. We hypothesized that combining an improved SCRIBE The development of in vivo recombination-mediated engineer- system with Cas9 counterselection against wild-type cells would ing (recombineering), which relies on phage-derived proteins to provide an easy-to-use genome editing platform that precluded facilitate DNA integration, has greatly improved the speed and the need to electroporate ssDNA for recombineering. In addition, efficiency with which targeted mutations can be made. For exam- the system was designed so that the editing plasmid can be retar- ple, the λ-phage-derived Red system in Escherichia coli yields more geted through a single round of PCR amplification and cloning than a 100-fold increase in recombinant cells over traditional HR to reduce construction times and enable the rapid generation of through a process independent of RecA (8, 9). Recombineering multilocus mutants. reduces editing template design constraints by enabling efficient recombination with homology arms of <50 bp on double-stranded 2. Materials and methods DNA (10). In addition, short single-stranded DNA (ssDNA) oligonu- 2.1 Strains, plasmids and culture conditions cleotides (oligos) can be used as suitable recombineering sub- strates. In fact, ssDNA oligos have produced recombinant cells Bacterial strains used in this work are listed in Supplementary with just one of the three λ-Red proteins, the ssDNA-annealing Table S1. Plasmid construction methods and primers are described protein Beta, further simplifying the requirements for success- in Supplementary Table S2 (along with Addgene #s for avail- ful recombineering (9). Furthermore, the Clustered Regularly able plasmids) and the Primers_GeneFrags.fa file, respectively. The Interspaced Short Palindromic Repeats (CRISPR)–Cas9 bacterial pFF745 plasmid was a gift from Timothy Lu (Addgene #61450). immune system has been adapted as a programmable counter- Cloning steps for retargeting the editing plasmid are illustrated in selection tool to selectively target and kill unedited cells after Supplementary Figure S2. Briefly, primers with overhangs encod- recombineering. This technique enhances the apparent efficiency ing the new target sequence were used to PCR amplify the vector by increasing the mutant to wild-type ratio of the population, using Q5 High-Fidelity 2x Master Mix (New England Biolabs, NEB), allowing highly efficient scarless genome editing (11–14). Addi- followed by DpnI digestion to remove template DNA, agarose gel tional efforts for the improvement of λ-Red editing efficiency electrophoresis to verify amplicon size and column-based DNA have used engineered mutator strains, where the overexpres- cleanup. DNA fragments were then assembled using the 2x NEB- sion of HR-involved enzymes, the deletion of host exonucle- uilder HiFi DNA Assembly Master Mix (NEB) with the following ases and/or disabling the mismatch repair system significantly modifications: 1 μl Master Mix+ 0.5μl each DNA fragment, incu- improve recombination efficiency but can unintentionally lead to bated at 50 C for 15 min. Chemically competent E. coli NEB5α the accumulation of off-target mutations (15–17). cells were used for cloning. Antibiotics were used at the following Farzadfard and Lu (18) described a novel approach for recom- concentrations unless otherwise noted: spectinomycin (50μg/ml), bineering, termed Synthetic Cellular Recorders Integrating Biolog- kanamycin (50μg/ml), chloramphenicol (40μg/ml), rifampicin ical Events (SCRIBE), where the ssDNA is generated in vivo and (20μg/ml), tetracycline (20μg/ml) and carbenicillin (100μg/ml). then incorporated into the chromosome. The in vivo ssDNA is pro- The tetA gene was PCR amplified from pUC18-mini-Tn7T-Gm- duced by the Ec86 retron of E. coli BL21, composed of the msr TetAR using primers 2490 and 2491 and integrated into E. coli and msd RNA elements and the reverse transcriptase (RT). Upon MG1655 using λ-Red recombination with pKD46 (21), creating the transcription of the msr–msd sequence, inverted repeats flanking MG1655 yhiS::tetAR strain. the msr–msd RNA form a secondary structure specifically recog- 2.2 Recombineering assays and recombination nized by the RT and the msdRNA is reverse transcribed to ssDNA. efficiency determination Simultaneous expression of bet on the SCRIBE plasmid allows for efficient recombination of the newly synthesized ssDNA into Since the number of recombinant cells should increase with the lagging strand of the targeted locus during DNA replication time, all experiments were performed in freshly transformed cells. (Figure 1A). By placing control of this system under an inducible Approximately 100 ng of editing plasmid was transformed into promoter, the cells record their exposure to the inducer as the frac- competent E. coli cells via either heat shock at 42 C or electropo- tion of cells with the msd encoded mutation. By simply changing ration in a 0.1-cm gap cuvette with a single 1.8-kV pulse. Cells the msd sequence, SCRIBE can be easily retargeted to modify any were recovered in 1 ml SOB at 30 C for 2 h and then outgrown desired genomic site. The authors report a maximum efficiency for overnight at 30 C in 5 ml Luria Bertani (LB) broth+ appropriate –4 SCRIBE of 10 recombinants per generation (18). While sufficient antibiotic(s) to select transformants. The total number of viable cells was determined by spotting 10μl 10-fold serially diluted (22) were inserted upstream of mutL to allow for inducible overnight cultures on LB plates+ appropriate antibiotic(s). For the expression by the addition of cumate, yielding pCas9CyMutL. rpoB and tetR assays, 10μL of 10-fold serially diluted overnight cul- Escherichia coli MG1655 cells harboring pCas9CyMutL were then tures were spotted on LB plates with rifampicin or tetracycline to transformed with pTJV1Sc-rpoB1 or pTJV1Sc-tetA and recovered determine the number of recombinant cells. For the ackA assays, in 1 ml SOB at 30 C for 2 h, after which cells were transferred 1 ml of cells from overnight cultures were pelleted, washed and to 5 ml LB+ chloramphenicol+ spectinomycin with or without resuspended with sterile phosphate buffered saline after which cumate (100μM) and outgrown overnight at 30 C. Cultures were washed cells were spotted on M9 chloroacetate plates (M9 min- diluted 1:10 into LB+ chloramphenicol+ spectinomycin+ cumate imal media+ 10 mM sodium chloroacetate+ 2% glycerol+ 0.1% with or without aTc (0.1μg/ml) and grown overnight at 30 C. SOB) to determine the number of ackA mutant cells. Recom- Recombination efficiency and efficiency of Cas9 counterselec- bination efficiency was determined by dividing the number of tion were determined the following day, and the results are recombinant cells by the total number of viable transformants. reported as the mean with standard error for three independent Efficiencies reported are the mean and standard error for three replicates. independent replicates. For the rpoB experiments, spontaneous rifampicin resistance was assessed for the wild-type strain as 2.6 Off-target mutation frequency analysis described in Section 2.6 below and the spontaneous mutation Escherichia coli MG1655 was separately transformed with pTV1β- rate was subtracted from all samples plated on rifampicin. All rpoB and pCas9CyMutL and plated on LB+ spectinomycin and statistical analyses were performed using the GraphPad Prism 5 LB+ chloramphenicol, respectively, for the selection of transfor- program. mants. Single colonies from each transformation were used to inoculate 5 ml LB broth+ appropriate antibiotic in triplicate and 2.3 Promoter optimization grown overnight at 30 C, with pCas9CyMutL grown with or with- out cumate (100μM). Cultures were then diluted 1:10 into the The P promoter of the Ec86 retron cassette in pTLlSc-rpoB lacO same media and grown for an additional overnight at 30 C, after was replaced with P , and P was inserted upstream of J23101 VanCC which 10μL of 10-fold serial dilutions of each were spotted onto bet as described in Supplementary Table S2. Expression of the LB only and LB+ rifampicin to assess the spontaneous mutation Ec86 retron cassette and beta recombinase by P and P , J23101 VanCC frequency. respectively, was further optimized by PCR amplifying pTJVSc- For genome sequencing, a single colony from each transforma- rpoB with primers 2326 & 2358 and 2325 & 2359, followed by tion and wild-type MG1655 were grown in LB broth+ appropriate assembly of the two resulting amplicons using the 2x NEBuilder antibiotic overnight at 30 C, with pCas9CyMutL grown with and HiFi DNA Assembly Master Mix (NEB). The assembled pTJVSc- without cumate (100μM) for induction of mutL expression. Cul- rpoB promoter library was transformed into E. coli NEB5α cells, tures were then diluted 1:100, grown overnight and repeated and recombination efficiency was determined. In addition, the for a total of two setbacks. Genomic DNA was extracted after remainder of the pTJVSc-rpoB culture was diluted 1:100 into 5 ml the first setback for wild-type and after the second setback for LB broth+ rifampicin and grown overnight again at 30 C. The fol- the other samples using a cetyltrimethyl ammonium bromide lowing morning, the plasmid library was extracted, retransformed (CTAB)/phenol-chloroform extraction protocol (23). into E. coli NEB5α cells, and the process was repeated three more Samples were prepared for whole-genome sequencing using times. After the fourth round of selection, the pTJVSc-rpoB plas- the NEBNext Ultra II FS DNA Library Prep Kit for Illumina (NEB). mid was extracted from four randomly chosen transformants and Sequencing was performed using the NovaSeq6000 sequencing the P and P promoters were sequenced with primers 746 J23101 VanCC platform (Novogene Co. Sacramento, CA). Sequencing data were and 2252, respectively, to assess the convergence of the population quality filtered and adapters were trimmed using the Trim Galore on a single optimal promoter sequence, yielding pTJV1Sc-rpoB as script (24). Mutations were identified using the Breseq mutational indicated in Supplementary Table S2. analysis pipeline (25) set to polymorphism mode with the default parameters and a minimum coverage cutoff of 20× reads. The E. 2.4 Cas9 counterselection coli MG1655 reference genome (National Center for Biotechnology For pCas9CR4 for CRISPR/Cas9 directed counterselection against Information (NCBI) accession: NC_000913) was used as the refer- unedited wild-type cells, transformants with editing plasmid ence sequence. The wild-type MG1655 parent strain was used as were outgrown overnight. The following morning cultures were a control to assess differences between the reference genome and diluted 1:10 into LB+ appropriate antibiotics with or without our lab strain and sequence variations identified were subtracted anhydrotetracycline (aTc; 0.1μg/ml) to induce Cas9 expression from those found in our experimental samples. and then grown overnight at 30 C. Serial dilutions were spotted onto LB+ rifampicin for the rpoB assays, LB+ tetracycline for the 3. Results tetA reversion assays, or M9 chloroacetate for the ackA assays, then incubated overnight at 37 C. The efficiency of Cas9 counters- 3.1 Improvement of recombineering with election was determined by dividing the number of colonies from reverse-transcribed ssDNA the aTc-induced cultures that grew with selection by the total To enable an efficient and iterative system for genome editing number of viable cells plated on a nonselective plate. Results are with ssDNA reverse-transcribed in vivo, we moved the SCRIBE sys- reported as the mean with standard error for three independent tem’s functional components to the pKDsgRNA plasmid that has replicates. the temperature-sensitive variant of the pSC101 origin to allow for easy curing with growth at 37–42 C (21, 26, 27). This plas- 2.5 Efficiency improvement with negative mid, pTLlSc, possesses the msr-msd coding sequence and Ec86 RT mutator alleles for the synthesis of ssDNA, the bet gene to facilitate incorpora- The mutL gene was cloned into pCas9CR4 as described in tion of the ssDNA and the S. pyogenes single-guide RNA (sgRNA) Supplementary Table S2. The cymR repressor and P promoter to enable Cas9 targeting. The judicious design of this plasmid cymRC Figure 1. Optimization of the origin of replication and promoter elements enables an efficient and curable system for in vivo recombineering . (A) In vivo retron-generated ssDNA serves as an editing substrate for beta recombination at the lagging strand of the replication fork. (B) Rifampicin resistance conferred by the P564L mutation of the rpoB gene was used to measure the recombination efficiency. The pSC101 origin of replication allows for curing of the editing plasmid by growth at 42 C after desired mutation(s) are made. (C) Modifications made to promoters and origin of replication for enhanced recombination efficiency and curability. (D) Recombination efficiency of each editing plasmid reported as the average of three independent replicates. Error bars represent SEM, and statistical significance relative to pFF745–rpoB is denoted by asterisks (ns = not significant; **P < 0.01; ***P < 0.001; one-way analysis of variance and Tukey’s test of log-transformed values). enables easy one-step cloning to retarget both the msd and sgRNA was slightly higher than the original construct (Supplementary (Supplementary Figure S2). To assess the recombination efficiency Figure S1B), and sequencing of the promoter from randomly cho- using an easily selectable mutation, the msr-msd was retargeted to sen transformants revealed a convergence of the pool toward a introduce three consecutive nucleotide substitutions and create single promoter sequence that had a recombination efficiency the P564L point mutation of rpoB that confers rifampicin resis- of nearly 10-fold greater than the parent plasmid (Supplemen- tance (Figure 1B). Here, we define recombination efficiency as the tary Figure S1C–D). Moreover, the efficiency of the improved con- number of cells obtained on selective media divided by the total struct, pTJV1Sc-rpoB, was similar to the original SCRIBE plasmid, number of viable cells. First, we compared the efficiency of pFF745, pFF745–rpoB (Figure 1D). described by Farzadfard et al., which possesses a high-copy pUC 3.2 CRISPR/Cas9 counterselection against origin targeted to rpoB, with our temperature-sensitive plasmid wild-type cells pTLlSc-rpoB. As expected, recombination efficiency was about 100-fold lower with the temperature-sensitive plasmid, presum- We next investigated whether our improved in vivo recombi- ably because of decreased expression of the retron elements and neering system could be combined with CRISPR/Cas9 counter- bet compared to the high-copy pUC origin on pFF745 (Figure 1D). selection against wild-type cells for a highly efficient genome We sought to improve this efficiency by increasing the transcrip- editing system. The pCas9CR4 plasmid has cas9 under the con- tion of the msr-msd, RT and bet through promoter replacements. trol of an inducible P promoter that was engineered to enable tet The strong constitutive promoter P (28) replaced the P co-maintenance with a genome targeting sgRNA until induc- J23101 lacO promoter driving expression of the msr-msd RNA and RT. The tion with anhydrotetracycline (aTc) (27). We first transformed promoter of bet was changed to P (22) because it is both pCas9CR4 into the E. coli strains NEB5α, BL21-DE3 and MG1655, VanCC strong and nonhomologous to P . Replacing these promoters and subsequently transformed the pTJV1Sc-rpoB editing vector, J23101 individually did not increase recombination efficiency, but when which possessed an sgRNA targeted to wild-type rpoB. After recov- combined, efficiency increased nearly 10-fold (Figure 1D ). ery, the cells were transferred to LB broth with spectinomycin To further increase recombination efficiency, we used a selec- and chloramphenicol to select for both plasmids and grown tion strategy where a library of variant plasmids was created by overnight. The following morning, the cultures were then passed cloning degeneracies into both the P and P promoters. into media with or without aTc and grown an additional overnight J23101 VanCC We hypothesized that variants with increased efficiency would for induction of Cas9 expression and killing of unedited cells. overtake the population after repeated rounds of outgrowth, selec- Cells were then spotted onto plates with and without rifampicin tion, plasmid extraction and re-transformation (Supplementary to directly select for the rpoB mutation and assess total viable Figure S1A). After four passages, the efficiency of the evolved pool cells, respectively (Figure 2A). The recombination efficiency was Figure 2. Targeted counterselection against unedited cells by inducible cas9 expression. (A) After Beta-recombination, cells are sub-cultured 1:10 into WT media with aTc to induce cas9 expression from pCasCR4. Constitutive expression of an sgRNA from pTJV1Sc-rpoB targets rpoB alleles for DNA DSB by Cas9 cleavage, thereby eliminating nonmutated cells. (B) Escherichia coli NEB5α, BL21 and MG1655 cells were transformed with pTJV1Sc-rpoB and the RpoB frequencies were determined by plating on LB+ rifampicin. (C) Escherichia coli MG1655::tetAR was transformed with pTJV1Sc-rpoB, P564L pTJV1Sc-tetA and pTJV1Sc-ackA. Frequencies of the RpoB , TetA and AckA mutations were determined by plating cells on LB+ rifampicin, P564L *70Y E54* tetracycline and chloroacetate, respectively. (D) Sequential mutation of distinct loci was performed by transforming E. coli MG1655 with pTJV1Sc-rpoB, determining recombination efficiency by plating on LB + rifampicin, curing pTJV1Sc-rpoB by growing cells at 42 C, then transforming once more with pTJV1Sc-ackA. Double mutant frequency was assessed by plating on LB+ rifampicin+ sodium chloroacetate. Averages are based on three independent replicates. Error bars represent SEM. Statistical significance is denoted by asterisks (** = P-value < 0.01; ***= P-value <0.001; two-tailed Student’s t-test of log-transformed values). –5 –3 between 1 × 10 and 3 × 10 without induction of Cas9 expression and are thus resistant to the toxic acetate analog chloroacetate. for all three strains. However, when cas9 was induced, the average A recombination efficiency of 5.6% without Cas9 counterselec- efficiency was ∼5% for BL21 and 60–80% for NEB5α and MG1655, tion and over 100% with induction of Cas9 was obtained, as indicating a substantial reduction in wild-type cells (Figure 2B). slightly more colonies were observed on the selection plate than To demonstrate that our system could mutagenize different on the nonselective plate on average (Figure 2C). These experi- targets, we examined the recombination efficiency of tetA and ments show that targeted counterselection of unedited cells with ackA. We inserted the tetA encoded tetracycline efflux pump that Cas9 can successfully enrich for recombinants across distinct loci. possessed a premature stop codon of about 200 bp from the The pCas9CR4 plasmid used in these experiments was start codon into MG1655 using traditional λ-Red recombineering designed for tight repression of cas9 and an ssrA degradation tag (21). Our experiments then mutated the stop codon back to a decreases Cas9 stability, enabling co-maintenance of both genome sense codon, creating the TetA mutation to enable the full- targeting sgRNA template and cas9 on two plasmids. Although *70Y length translation of tetA and confer tetracycline resistance to the induction of cas9 is required for cell death by double-strand break cells. We obtained tetracycline-resistant colonies with efficiencies (DSB), we questioned whether some transient level of Cas9 expres- –4 –5 of 5.4 × 10 and 1.8 × 10 with and without Cas9 counterse- sion also enriches for recombinants, thus we performed exper- lection, respectively, much lower than those observed for rpoB iments to compare the number of mutants obtained with the (Figure 2C). Next, the acetate kinase gene, ackA, was mutated SCRIBE system in the presence and absence of pCas9CR4. More with three nucleotide point mutations to create a premature stop recombinant cells were produced for all three targets with cells codon (AckA ) so that the cells are unable to metabolize acetate harboring pCas9CR4 than cells without, even without induction E54* of cas9 expression (Figure 2C). In fact, we were unable to obtain repair of mismatches within 3 bp (17, 31). The introduction of tetracycline-resistant cells when using the pTJV1Sc-tetA plasmid several adjacent mismatches also prevents repair by MMR (15). alone. These results suggest that even low levels of cas9 expres- However, these strategies present design constraints that can- sion in the uninduced state can increase recombination efficiency not always be followed. An alternate strategy is the recombinant possibly by slowing the growth of wild-type cells that must repair expression of negative mutS or mutL variants that are dominant DSB or because this low-level DSB stimulates recombination. over wild-type enzymes and prevent efficient repair (32). Accord- The proposed mechanism of ssDNA recombineering asserts ingly, we cloned mutL with the dominant negative E32K muta- that allelic replacement occurs at the replication fork, where tion into the pCas9CR4 plasmid under the control of a cumate the supplied ssDNA replaces an Okazaki fragment on the lag- inducible promoter (22) to maintain a two-plasmid system with ging strand. This mechanism results in a lagging strand bias, independent control of cas9 and mutL (Figure 3A). To test this sys- where recombination efficiencies for oligos targeting the lagging tem, we introduced a single bp mismatch of A:C in rpoB with strand are higher than those targeting the leading strand. To and without induction of the negative mutL. In cells harboring assess whether in vivo produced ssDNA proceeds through the pCas9CyMutL with mutL in the off-state, the average frequency –6 same mechanism, we targeted the leading strand of rpoB for edit- of rifampicin-resistant mutants was 4.9 × 10 without Cas9- –4 ing (Supplementary Figure S3A). Surprisingly, this resulted in a induction and 4.5 × 10 with Cas9 counterselection (Figure 3B). similar efficiency as targeting the lagging strand (Supplementary In contrast, the induction of mutL enabled rpoB mutations with Figure S3B). We wondered whether this was a function of the in an average recombination efficiency of 0.2% and 18% without vivo generation of ssDNA or if the same would be true of the rpoB and with Cas9 counterselection (Figure 3B). We also re-examined target using traditional λ-Red recombineering. We constructed a the tetA reversion assays, which introduced a single bp T:T mis- control plasmid in which the msr-msd and RT coding sequences match that is less efficiently repaired than the A:C mismatch used were removed (pTV1β-rpoB), resulting in constitutive expression for rpoB. Induction of the mutator allele increased the recombi- of Beta only and sgRNA targeting rpoB. We then performed a tra- nation efficiency by about 100-fold both with and without Cas9 ditional recombineering experiment by transforming exogenous counterselection, achieving a maximum efficiency of 0.3% on aver- ssDNA into cells with this plasmid and pCas9CR4 and observed age (Figure 3C). When generating a single A:G mismatch in the that the lagging oligo was much more efficient, consistent with ackA gene, mutL induction improved recombination by 1000- and previous observations (7, 9, 29) and the proposed mechanism of 100-fold with and without Cas9 counterselection, respectively, ssDNA recombineering (Supplementary Figure S3C). To evaluate resulting in an average maximum efficiency of 56% ( Figure 3D). whether the same phenomenon is true of other targets using These results further highlight the versatility of our system in SCRIBE, we targeted the leading strand of tetA with the in vivo sys- generating precise mutations across a diverse range of loci. tem (Supplementary Figure S3A). In this case, over 1000-fold fewer To further illustrate the efficiency enhancement these tools recombinant cells were obtained than when targeting the lagging offer over previous methods, we performed the same experiments strand (Supplementary Figure S3D). While we suspect that recom- using the no-SCAR method described previously (12). Single- bineering with in vivo ssDNA occurs at the lagging strand of the stranded editing oligos targeting the lagging strand were used to replication fork, our data suggest that the constant availability introduce the same single bp mismatches as before in cells harbor- of ssDNA may be sufficient to overcome the lagging strand bias ing pCas9CyMutL and pKDsgRNA, a plasmid encoding the λ-Red in some cases or that an alternative mechanism may influence proteins Exo, Beta and Gam under the control of the arabinose- recombination rates at some locations. inducible P promoter, as well as an sgRNA targeting the wild BAD type gene sequences. When targeting rpoB, induction of mutL before transformation, during recovery, or both achieved sim- ilarly high recombination rates as the retron system, yielding 3.3 Iterative mutagenesis of two targets average efficiencies >2% with direct selection on rifampicin and Above all, the high efficiency of mutagenesis found at our target 100% with Cas9 counterselection, a 20-fold increase in recom- sites confirms a robust system for genome editing, precluding the binants than without mutL (Supplementary Figure S5A). Regard- need to screen large numbers of colonies to identify mutants. To less of whether mutL was induced, Cas9 counterselection effec- demonstrate that our system could iteratively construct genome tively removed unedited cells with 95% of cells plated on aTc modifications, cells that were mutated at the rpoB locus were being rifampicin-resistant even without mutL expression. On the grown at 42 C to cure the cells of pTJV1Sc-rpoB. Subsequent trans- contrary, when targeting ackA, mutL induction improved the formation with pTJV1Sc-ackA enabled mutation of the ackA gene recombination efficiency less than 10-fold, yielding a maximum and produced cells resistant to both rifampicin and chloroacetate –4 efficiency of 4.8 × 10 when plated on chloroacetate (Supplemen- (Figure 2D). Amplification and Sanger sequencing of the rpoB and tary Figure S5B). However, sequencing the ackA gene of several ackA loci showed that both intended mutations were made, con- putative mutant colonies on the chloroacetate selection media firming that iterative use of the system could edit multiple loci revealed none had acquired the intended mutation. Furthermore, independently (Supplementary Figure S4). Cas9 counterselection did not substantially enrich for mutants, with <5% of the colonies on the aTc-induced plate able to grow with chloroacetate when patched (Supplementary Figure S5B), 3.4 Co-Expression of dominant-negative mutL indicating most of the colonies on the counterselection plate Recombineering with ssDNA results in transient production of escaped Cas9 killing. These results are consistent with previous heteroduplex DNA that the methyl-directed mismatch repair studies examining the rate of Cas9 escape in bacteria, which −3 −4 (MMR) system can repair, thus decreasing the observed rate of report frequencies ranging from ∼10 to 10 in multiple species mutagenesis. Several strategies are known to decrease the rate of (11, 27, 33, 34). Overall, the combined mutL and retron-based sys- MMR and enable more efficient mutagenesis (15, 17, 30–32). For tems result in consistently high mutagenesis rates for single bp example, single bp mismatches are repaired with different effi- edits and can outperform traditional oligo recombineering meth- ciencies, while C:C mismatches evade repair entirely and prevent ods for certain target loci. Figure 3. Expression of dominant-negative MutL allele enhances recombination efficiency for single nucleotide point mutations. (A) The MutL gene E32K was cloned into the pCas9CR4 plasmid under the control of the P promoter for inducible expression. When cumate is added, MutL is CymRC E32K expressed and competes with the native MutL for binding to MutS during mismatch repair. The E32K mutation prevents MutH binding, inhibiting removal and repair of the mismatched base. Cells harboring pCas9CyMutL were transformed with (B) pTJV1Sc-rpoB1, (C) pTJV1Sc-tetA or (D) pTJV1Sc-ackA1 encoding single point mutations for generating the RpoB , TetA and AckA mutations, respectively, and grown in liquid P564L *70Y E54* culture with or without cumate. The unmodified pCas9CR4 in both wild-type E. coli and a mutS mutant were used as controls. Recombination efficiency with and without Cas9 counterselection was assessed by plating on LB + rifampicin, LB+ tetracycline or M9+ chloroacetate. Data reported are the average of three independent replicates. Error bars represent SEM and statistical significance relative to pCas9CR4 is denoted by asterisks (ns= not significant; * = P-value < 0.05; **= P-value < 0.01; ***= P-value <0.001; one-way analysis of variance and Tukey’s test of log-transformed values). Hindrance of MMR increases the frequency of background However, induction of the mutator mutL resulted in an average –5 genome mutations (35, 36); however, the inducibility of our sys- mutation frequency of about 1.3 × 10 , an ∼20-fold increase over tem minimizes the amount of time in which MMR is inhibited. the wild-type (Supplementary Figure S6A). Similarly, uncontrolled expression of the λ-Red genes has also While these results indicate that the unexpressed mutL and been shown to increase the rate of spontaneous mutations (37). constitutive bet did not increase the rate of point mutations, As such, constitutive expression of bet in our system could con- we wanted to assess further whether other genomic mutations ceivably produce unintended recombination events in the cell. and rearrangements that would not manifest as rifampicin resis- To assess the background rate of mutagenesis in cells with our tance were present. Accordingly, we performed whole-genome SCRIBE system, we measured spontaneous rifampicin resistance sequencing on the strains grown for two overnights to iden- (38, 39). Cells were grown overnight, diluted 1:10, grown overnight tify polymorphisms within the entire population. A complete and then spotted on rifampicin to assess the mutation frequency. list of the mutations identified is in Supplemental File 2. Our The wild-type cells spontaneously acquired rifampicin resistance analysis revealed 41 total mutations in the population constitu- –7 –11 at a frequency of 6 × 10 on average. Cells constitutively express- tively expressing bet, constituting a mutation rate of 5.9 × 10 ing bet and cells with the uninduced pCas9CyMutL did not dif- mutations per cell per generation (Supplementary Figure S6B), fer significantly from the wild-type (Supplementary Figure S6A). with one new IS5 mobile element insertion into the rclA gene. Figure 4. Alternative recombinases can enhance efficiency for some targets. (A) The bet CDS on pTJV1Sc was replaced with genes encoding the CspRecT and EcRecT recombinases. Recombination efficiency for each of the recombinase variants with and without induction of cas9 expression was assessed for (B) the RpoB mutation by plating cells on LB+ rifampicin, (C) the AckA mutation by plating cells on LB+ chloroacetate and (D) the P564L E54* TetA mutation by plating cells on LB+ tetracycline. The pCas9CyMutL plasmid was used for the tetA experiments. Data reported are the average of *70Y three independent replicates. Error bars represent SEM and statistical significance is denoted by asterisks (ns = not significant; *** = P-value <0.001; one-way analysis of variance and Tukey’s test of log-transformed values). The uninduced pCas9CyMutL cells displayed a slightly higher binding protein promoted the highest number of recombinants –10 mutation rate of 2.4 × 10 ; however, both rates were similar to the between the different targets. For the rpoB mutation, EcRecT was –10 wild-type control, which exhibited a mutation rate of 1.0 × 10 . more efficient than both Beta and CspRecT, with 92% efficiency In contrast, 478 mutations were found in the population with when using Cas9 counterselection (Figure 4B). On the contrary, the –9 mutL induced for a mutation rate of 1.59 × 10 mutations per ackA mutation was more efficient with Beta, where 80% of cells cell per generation (Supplementary Figure S6B), as well as a dele- were mutants compared to 4.5% for CspRecT and no ackA mutants tion of the ∼1200 bp insH21 mobile element. Even still, this low with EcRecT (Figure 4C). Given the improvement in efficiency frequency of background mutations indicates selecting a mutant seen for tetA when expressing MutL , we used pCas9CyMutL E32K with unintended mutations would be a rare occurrence. Over- in combination with the alternative recombinases. We obtained all, these results demonstrate that maintenance of plasmids with over 100% efficiency for the tetA reversion with CspRecT, both the mutL mutator allele or constitutive expression of bet do not with and without Cas9 counterselection. Although slightly more increase the frequency of background mutations to a level that colonies were observed on the tetracycline plate than on the non- would cause concern from a genetic engineering standpoint. selective plate for two of the replicates (Figure 4D), the CspRecT clearly outperformed Beta and EcRecT where relatively low levels of mutants were found. Thus, in circumstances where a muta- tion is difficult to obtain, it is advisable to try these alternative 3.5 Alternative recombinases enhance efficiency recombinases, while the mechanism causing these discrepancies for some targets warrants further investigation. Recently, Wannier et al. identified recT from a Collinsella stercoris phage (CspRecT) with improved recombination efficiency com- pared to Beta for oligo mediated recombineering in E. coli (40). 4. Discussion To examine the effect of alternative recombinases in our sys- tem, we replaced bet with CspRecT, as well as EcRecT from In moving the SCRIBE machinery onto the temperature-sensitive the E. coli BL21-DE3 chromosome (Figure 4A). We then compared pSC101 origin of replication and combining it with Cas9 coun- the efficiency of these plasmids across three different targets. terselection, we successfully constructed an in vivo system for Surprisingly, our results showed inconsistency for which DNA- ssDNA recombineering that is highly efficient and iterative. Targeted counterselection using CRISPR/Cas9 enabled efficiencies distant loci within the same cell through repeated rounds of near 100% after a single night of outgrowth post-transformation, library recombineering. reducing the number of colonies needing to be screened for mutant identification. Moreover, both the ssDNA tem- Supplementary data plate and sgRNA sequence can be retargeted for modification Supplementary data are available at SYNBIO online. of any genomic locus with an appropriate PAM in a single cloning step using sequence overlap cloning methods. Numer- Data availability ous loci can be mutated without the need for orthogonal selection markers or extensive screening procedures, increas- Raw sequence reads for the off-target mutation frequency exper- ing the speed and ease of generating desirable mutants. In iments are available under the BioProject ID PRJNA813865 in the addition, the pCas9CR4 and pCas9CyMutL plasmids can be NCBI databases. cured using pKDsgRNA-p15 (Addgene #62656) as previously described (12), yielding plasmid-free mutants for downstream Funding applications. This work was supported by the Department of Microbiology and Recombineering is an important tool in the biological sci- Cell Science, Institute of Food and Agricultural Sciences, Univer- ences, and continued enhancement of recombination efficiency sity of Florida. across diverse target loci will decrease strain construction times. Recently, Lopez et al. showed that increasing the abundance of Acknowledgments ssDNA produced by retrons significantly improves recombina- tion efficiencies (41), consistent with the enhanced efficiencies We would like to thank the 2019 University of Florida International we observed when increasing the msr–msd/RT expression and bet Genetically Engineered Machine team for inspiring the initial work through promoter modifications. A complementary strategy is to on this project and L. Trujillo Rodriguez and L.A. Schuster for reduce the rate of ssDNA degradation by host nucleases, as the their support and advice during experimentation and manuscript High-efficiency SCRIBE (HiSCRIBE) system by Farzadfard et al. (20) preparation. describes. This system improves upon their original SCRIBE design by coupling a strong ribosome-binding site for bet on the SCRIBE Author contributions plasmid with CRISPRi-enabled transcriptional interference of the C.R.R. conceived this study. A.J.E. performed the experiments and endogenous exonucleases. Recombination efficiencies of nearly data analysis under the supervision of C.R.R. A.J.E. and C.R.R. 100% for galK and ∼25% for a kanamycin resistance gene tar- prepared the manuscript. get were obtained with these improvements. However, employing the nuclease-deficient dCas9 for CRISPRi prevents the use of the Conflict of interest statement. No potential conflict of interest was nuclease-active Cas9 for counterselection. Although the authors reported by the authors. show that CRISPR/Cas9 counterselection can select for recombi- nant cells at high efficiencies in an exonuclease knockout strain, this strategy prevents easy system portability since a mutant References strain must be used. 1. Baba,T., Ara,T., Hasegawa,M., Takai,Y., Okumura,Y., Baba,M., Lim et al. (42) also describe a CRISPR/retron-based editing sys- Datsenko,K.A., Tomita,M., Wanner,B.L. and Mori,H. (2006) Con- tem enabling trackable editing and functional screening across struction of Escherichia coli K-12 in-frame, single-gene knockout multiple loci in a population. However, this system requires the mutants: the Keio collection. Mol. Syst. Biol., 2, 1–11. sequential transformation and maintenance of three plasmids. 2. Choi,K.R., Jang,W.D., Yang,D., Cho,J.S., Park,D. and Lee,S.Y. (2019) Moreover, the construction of their retron/sgRNA plasmid requires Systems metabolic engineering strategies: integrating systems multiple enzymatic digestions for cloning, contrary to our two- and synthetic biology with metabolic engineering. Trends Biotech- plasmid system, which simultaneously retargets the retron and nol., 37, 817–837. sgRNA in a single two-fragment cloning step. Most importantly, 3. Jaiswal,S., Singh,D.K. and Shukla,P. (2019) Gene editing and sys- both the HiSCRIBE and the CRISPR/retron system of Lim et al. tems biology tools for pesticide bioremediation: a review. Front. (20, 42) would require multiple orthogonal selection markers to Microbiol., 10, 87. mutate multiple loci within a single cell, as no easy method of 4. Chen,Y., Batra,H., Dong,J., Chen,C., Rao,V.B. and Tao,P. (2019) plasmid curing is included in either design, preventing iterative Genetic engineering of bacteriophages against infectious dis- mutagenesis. Although the recombination efficiencies achieved eases. Front. Microbiol., 10, 954. with our system in the absence of counterselection are less 5. Jasin,M., Schimmel,P., Winans,S.C., Elledge,S.J., Krueger,J.H. and than those reported for HiSCRIBE, Cas9-induced cell death of Walker,G.C. (1984) Deletion of an essential gene in Escherichia unedited cells in the population enabled apparent efficiencies on coli by site-specific recombination with linear DNA fragments. J. par with HiSCRIBE. The ease with which our system can be retar- Bacteriol., 159, 783–786. geted and used iteratively to incorporate multiple edits combined 6. Russell,C.B., Thaler,D.S. and Dahlquist,F.W. (1989) Chromosomal with its high efficiency in mutant selection provides a quick and transformation of Escherichia coli recD strains with linearized plas- straightforward method for constructing mutants relevant to both mids. J. Bacteriol., 171, 2609–2613. research and biotechnological applications. Methods enabling 7. Court,D.L., Sawitzke,J.A. and Thomason,L.C. (2002) Genetic engi- multiplexed editing of numerous genomic loci within populations neering using homologous recombination. Annu. Rev. Genet., 36, and continuous evolution of targeted loci, using retron-mediated 361–388. recombineering, have been recently described (20, 42–44) and rely 8. Muyrers,J.P.P., Zhang,Y., Testa,G. and Stewart,A.F. (1999) on efficient editing for generating a library of variants. Our sys- Rapid modification of bacterial artificial chromosomes by tem would be well suited for constructing such libraries, with ET-recombination. Nucleic Acids Res., 27, 1555–1557. the added advantage of being curable, thus enabling editing of 9. Ellis,H.M., Yu,D., DiTizio,T. and Court,D.L. (2001) High efficiency 28. Anderson,J.C. (2006) http://parts.igem.org/Promoters/Catalog/ mutagenesis, repair, and engineering of chromosomal DNA using Anderson (10 January 2021, date last accessed). single-stranded oligonucleotides. Proc. Natl. Acad. Sci. U. S. A., 98, 29. Zhang,Y., Muyrers,J.P.P.P., Rientjes,J. and Stewart,F. (2003) 6742–6746. Phage annealing proteins promote oligonucleotide-directed 10. Zhang,Y., Buchholz,F., Muyrers,J.P.P. and Stewart,A.F. (1998) A new mutagenesis in Escherichia coli and mouse ES cells. BMC Mol. logic for DNA engineering using recombination in Escherichia coli. Biol., 4, 1. ̈ ́ Nat. Genet., 20, 123–128. 30. Nyerges,A., Csorgo,B., Nagy,I., Latinovics,D., Szamecz,B., Posfai,G. 11. Jiang,W., Bikard,D., Cox,D., Zhang,F. and Marraffini,L.A. (2013) and Pal,C. (2014) Conditional DNA repair mutants enable highly RNA-guided editing of bacterial genomes using CRISPR–Cas sys- precise genome engineering. Nucleic Acids Res., 42, e62. tems. Nat. Biotechnol., 31, 233–239. 31. Kramer,B., Kramer,W. and Fritz,H.J. (1984) Different base/base 12. Reisch,C.R. and Prather,K.L.J. (2017) Scarless Cas9 assisted recom- mismatches are corrected with different efficiencies by the bineering (no-SCAR) in Escherichia coli, an easy-to-use system for methyl-directed DNA mismatch-repair system of E. coli. Cell, 38, genome editing. Curr. Protoc. Mol. Biol., 117, 31.8.1–31.8.20. 879–887. ̈ ́ ́ ́ 13. Pyne,M.E., Moo-Young,M., Chung,D.A. and Chou,C.P. (2015) Cou- 32. Nyerges,A., Csorgo,B., Nagy,I., Balint,B., Bihari,P., Lazar,V., pling the CRISPR/Cas9 system with lambda red recombineering Apjok,G., Umenhoffer,K., Bogos,B., Posfai,G. et al. (2016) A highly enables simplified chromosomal gene replacement in Escherichia precise and portable genome engineering method allows compar- coli. Appl. Environ. Microbiol., 81, 5103–5114. ison of mutational effects across bacterial species. Proc. Natl. Acad. 14. Jiang,Y., Chen,B., Duan,C., Sun,B., Yang,J. and Yang,S. (2015) Multi- Sci. U. S. A., 113, 2502–2507. gene editing in the Escherichia coli genome via the CRISPR–Cas9 33. Oh,J.H. and Van Pijkeren,J.P. (2014) CRISPR–Cas9-assisted recom- system. Appl. Environ. Microbiol., 81, 2506–2514. bineering in Lactobacillus reuteri. Nucleic Acids Res., 42, e131. 15. Sawitzke,J.A., Costantino,N., Li,X.T., Thomason,L.C., 34. Liu,J., Wang,Y., Lu,Y., Zheng,P., Sun,J. and Ma,Y. (2017) Develop- Bubunenko,M., Court,C. and Court,D.L. (2011) Probing cellu- ment of a CRISPR/Cas9 genome editing toolbox for Corynebac- lar processes with oligo-mediated recombination and using the terium glutamicum. Microb. Cell Fact., 16, 205. knowledge gained to optimize recombineering. J. Mol. Biol., 407, 35. Schaaper,R.M. and Dunn,R.L. (1987) Spectra of spontaneous 45–59. mutations in Escherichia coli strains defective in mismatch correc- 16. Mosberg,J.A., Gregg,C.J., Lajoie,M.J., Wang,H.H. and Church,G.M. tion: the nature of in vivo DNA replication errors. Proc. Natl. Acad. (2012) Improving lambda red genome engineering in Escherichia Sci. U. S. A., 84, 6220–6224. coli via rational removal of endogenous nucleases. PLoS One, 7, 36. Isaacs,F.J., Carr,P.A., Wang,H.H., Lajoie,M.J., Sterling,B., Kraal,L., e44638. Tolonen,A.C., Gianoulis,T.A., Goodman,D.B., Reppas,N.B. et al. 17. Costantino,N. and Court,D.L. (2003) Enhanced levels of λ Red- (2011) Precise manipulation of chromosomes in vivo enables mediated recombinants in mismatch repair mutants. Proc. Natl. genome-wide codon replacement. Science, 333, 348–353. Acad. Sci. U. S. A., 100, 15748–15753. 37. Murphy,K.C. and Campellone,K.G. (2003) Lambda Red-mediated 18. Farzadfard,F. and Lu,T.K. (2014) Genomically encoded analog recombinogenic engineering of enterohemorrhagic and memory with precise in vivo DNA writing in living cell popula- enteropathogenic E. coli. BMC Mol. Biol., 4, 11. tions. Science, 346, 1256272. 38. Garibyan,L., Huang,T., Kim,M., Wolff,E., Nguyen,A., Nguyen,T., 19. Wang,H.H., Isaacs,F.J., Carr,P.A., Sun,Z.Z., Xu,G., Forest,C.R. and Diep,A., Hu,K., Iverson,A., Yang,H. et al. (2003) Use of the rpoB Church,G.M. (2009) Programming cells by multiplex genome engi- gene to determine the specificity of base substitution muta- neering and accelerated evolution. Nature, 460, 894–898. tions on the Escherichia coli chromosome. DNA Repair (Amst), 2, 20. Farzadfard,F., Gharaei,N., Citorik,R.J. and Lu,T.K. (2021) Efficient 593–608. retroelement-mediated DNA writing in bacteria. Cell Syst., 12, 39. Lee,H., Popodi,E., Tang,H. and Foster,P.L. (2012) Rate and molec- 860–872.e5. ular spectrum of spontaneous mutations in the bacterium 21. Datsenko,K.A. and Wanner,B.L. (2000) One-step inactivation of Escherichia coli as determined by whole-genome sequencing. Proc. chromosomal genes in Escherichia coli K-12 using PCR products. Natl. Acad. Sci. U. S. A., 109, E2774–E2783. Proc. Natl. Acad. Sci. U. S. A., 97, 6640–6645. 40. Wannier,T.M., Nyerges,A., Kuchwara,H.M., Czikkely,M., Balogh,D., 22. Meyer,A.J., Segall-Shapiro,T.H., Glassey,E., Zhang,J. and Voigt,C.A. Filsinger,G.T., Borders,N.C., Gregg,C.J., Lajoie,M.J., Rios,X. et al. (2019) Escherichia coli “Marionette” strains with 12 highly optimized (2020) Improved bacterial recombineering by parallelized protein small-molecule sensors. Nat. Chem. Biol., 15, 196–204. discovery. PNAS, 117, 13689–13698. 23. Feil,W.S., Feil,H. and Copeland,A. (2012) Bacterial genomic DNA 41. Lopez,S.C., Crawford,K.D., Lear,S.K., Bhattarai-Kline,S. and Ship- isolation using CTAB. https://www.jgi.doe.gov (10 January 2021, man,S.L. (2022) Precise genome editing across kingdoms of life date last accessed). using retron-derived DNA. Nat. Chem. Biol., 18, 199–206. 24. Krueger,F. (2021) TrimGalore: a wrapper around Cutadapt and 42. Lim,H., Jun,S., Park,M., Lim,J., Jeong,J., Lee,J.H. and Bang,D. (2020) FastQC to consistently apply adapter and quality trimming to Multiplex generation, tracking, and functional screening of sub- FastQ files, with extra functionality for RRBS data. GitHub. stitution mutants using a crispr/retron system. ACS Synth. Biol., 9, 25. Deatherage,D.E. and Barrick,J.E. (2014) Identification of mutations 1003–1009. in laboratory-evolved microbes from next-generation sequencing 43. Simon,A.J., Morrow,B.R. and Ellington,A.D. (2018) Retroelement- data using breseq. Methods Mol. Biol., 1151, 165–188. based genome editing and evolution. ACS Synth. Biol., 7, 26. Hashimoto-Gotoh,T. and Sekiguchi,M. (1977) Mutations of tem- 2600–2611. perature sensitivity in R plasmid pSC101. J. Bacteriol., 131, 405. 44. Schubert,M.G., Goodman,D.B., Wannier,T.M., Kaur,D., Lu,T.K., 27. Reisch,C.R. and Prather,K.L.J.J. (2015) The no-SCAR (Scarless Shipman,S.L. and Church,G.M. (2021) High throughput functional Cas9 Assisted Recombineering) system for genome editing in variant screens via in-vivo production of single-stranded DNA. Escherichia coli. Sci. Rep., 5, 15096. PNAS, 118, e2018181118.

Journal

Synthetic BiologyOxford University Press

Published: May 3, 2022

Keywords: genome editing; retron; recombineering; Cas9; MutL E32K

References