Robust counterselection and advanced λRed recombineering enable markerless chromosomal integration of large heterologous constructs

Abstract

Despite advances in bacterial genome engineering, delivery of large synthetic constructs remains challenging in practice. In this study, we propose a straightforward and robust approach for the markerless integration of DNA fragments encoding whole metabolic pathways into the genome. This approach relies on the replacement of a counterselection marker with cargo DNA cassettes via λRed recombineering. We employed a counterselection strategy involving a genetic circuit based on the CI repressor of λ phage. Our design ensures elimination of most spontaneous mutants, and thus provides a counterselection stringency close to the maximum possible. We improved the efficiency of integrating long PCR-generated cassettes by exploiting the Ocr antirestriction function of T7 phage, which completely prevents degradation of unmethylated DNA by restriction endonucleases in wild-type bacteria. The employment of highly restrictive counterselection and ocr-assisted λRed recombineering allowed markerless integration of operon-sized cassettes into arbitrary genomic loci of four enterobacterial species with an efficiency of 50-100%. In the case of Escherichia coli, our strategy ensures simple combination of markerless mutations in a single strain via P1 transduction. Overall, the proposed approach can serve as a general tool for synthetic biology and metabolic engineering in a range of bacterial hosts.

Figure 1.

General outline of the cI-hok counterselection strategy and helper genetic constructs. (A) Illustration of the cI-hok counterselection principle. The cI-hok-neo cassette is delivered into the host genome by recombineering using the neo marker for positive selection and rhaSR-P_rhaB-driven λRed functions provided by the pRedCm helper plasmid. Upon chromosomal integration, cI confers tight repression of P_R-hok within the same dual selectable cassette and the P_L-cat unit residing within pRedCm, by which the cell remains viable and chloramphenicol-sensitive. (B) Three possible fates of cells carrying the cI-hok cassette and the pRedCm plasmid. As long as functional cI is present, the cell remains viable and chloramphenicol-sensitive (middle scenario). Spontaneous mutation of cI alone results in derepression of both P_L-cat and P_R-hok units, subsequent Hok-mediated membrane permeabilisation, and death of the mutant (upper scenario). ‘En bloc’ elimination of the cI-hok cassette because of a purposive recombination event causes derepression of P_L-cat and enables selection of chloramphenicol-resistant recombinants (bottom scenario). (C) A set of template constructs for amplifying cI-hok cassettes with neo (kanamycin), strA (streptomycin), or aacC4 (gentamycin) orthogonal positive selection markers. The two bottom constructs carry either the P_H207 (in truncated form) or P_A1O4 promoter in front of the T_L3S1P56 terminator. For each promoter, there are two alternative constructs in which the promoter is directed towards or away from the terminator.

Figure 2.

Evaluation of the efficiency of cI-hok counterselection. (A) Schematic representation of the construction of strains involved in the measurement of mutation rates to counterselection resistance. Each of the galETKM, lacZYA and araBAD operons was targeted in MG1655 with either the cI-hok or cat-sacB cassettes using the same homology arms for each locus. The cat-rpsL cassettes were inserted into the chromosome of B1733 (MG1655 rpsL^K43R). The lacZYA operon was additionally targeted using the cI-hok^L13ochre-neo cassette with inactive hok. (B) Mutation rate to counterselection resistance for cI-hok, sacB and rpsL markers. The cI-hok strains were transformed with pRedCm to enable counterselection. Spontaneous mutants were selected on LB with chloramphenicol (200 mg/l), LB with streptomycin (500 mg/l), and SuLB plates for cI-hok, rpsL and sacB strains, respectively. The values shown and expressed as mutations to counterselection resistance per generation were calculated from fluctuation assay data (the frequencies of the mutants are shown in Supplementary Figure S4) using the Ma-Sandri-Sarkar maximum likelihood method implemented in the FluCalc web tool as described in the section ‘Measurement of mutation rate to counterselection resistance’ of the Supplementary Materials and Methods. Error bars indicate 95% confidence intervals calculated using the same software. In experiments with B1826 [pRedCm], a portion of the cultures did not contain Cm^R mutants. Thus, the mutation rate was calculated using the p0 method and 95% confidence intervals were determined as described in (36). (C) Comparison of mutation rates of the cI-hok and cI-hok^L13ochre within the lac locus. All measurements and calculations were performed as described for Figure 2B.

Figure 3.

Stimulation of λRed recombineering by suppression of EcoKI endonuclease activity. (A) Schematic of the experiment for assessing the capability of antirestriction functions to improve recombineering efficiency. The ral, ardA, or ocr genes were cloned on pRedCm upstream of gam. The resulting plasmids were transformed into MG1655 and tested as helpers for replacement of the galETKM operon with the scrKYABR cassette, which includes three EcoKI recognition sites (shown as red rectangles). (B) Efficiency of replacing the galETKM operon with the scrKYABR cassette. MG1655 cells carrying pRedCm, pRedCmRal, pRedCmArdA, or pRedCmOcr were induced for expression of λRed and antirestriction functions and electroporated with 0.6 μg of the scrKYABR cassette. Scr⁺ recombinants were selected on M9 plus 0.5% sucrose plates and the viable cell titre was determined on non-selective LB plates. The values shown are the average of three biological replicates; error bars indicate SD. (C) Effect of antirestriction functions on the efficiency of plating (EOP) of λ and unmethylated λ.0 phages. A detailed description is presented under ‘Assay of λ phage plating efficiency’ in the Supplementary Materials and Methods. Both phages were plated on cells induced for expressing λRed and antirestriction functions as well as on cells of plasmidless MG1655 and TG1 strains. Normalised EOP for each strain was calculated by dividing the determined phage titre by the titre of the same phage on a lawn of TG1. The average and SD were then calculated using the normalised EOP. For TG1, normalised SD was calculated by dividing non-normalised SD by the non-normalised average. The values shown are the averages of three biological replicates; error bars indicate SD.

Figure 4.

Markerless integration of linear constructs via cI-hok counterselection. (A) Schematic of experiments for estimating the efficiency of markerless cloning. Upon electroporation, a linear cassette (shown in violet) replaces the cI-hok cassette within a specific locus (ara in the figure) via λRed recombination, thereby enabling selection of recombinant colonies for Cm^R due to derepression of the P_L-cat module within pRedCmOcr or the chromosomal ocr-γβexo-P_L-cat helper construct. Targeted genomic loci are indicated as open circles along the chromosome. The value in parentheses near the name of a cassette indicates its entire length. (B) Schematic of the construct used for estimating the efficiency of markerless integration. The total length of the cassettes and flanking homology arm are indicated. The cassettes were prepared as described under ‘Preparation of linear cassettes’ in the Supplementary Materials and Methods (also see Supplementary Figure S14). An asterisk in the designation ‘trpE*’ refers to the S40F amino acid substitution in the anthranilate synthase, which renders the enzyme insensitive to feedback inhibition by l-tryptophan. (C) The efficiency of markerless cassette integration across genomic loci. Briefly, induced cells were electroporated with 0.4–1 μg of a cassette, recovered overnight in LB and plated on LB plus chloramphenicol plates with supplements enabling discrimination of recombinants as described under ‘cI-hok counterselection and identification of recombinants’ in the Materials and Methods. Colonies exhibiting the desired phenotype (sucrose utilisation, violacein production, or luminescence) were recognised as recombinants. For each cassette, at least seven recombinant colonies were purified and used for verifying the construct presence within a targeted locus by a locus-specific PCR (Supplementary Figure S8). In experiments with the trp cassettes, 16 Cm^R colonies per biological replicate were tested using locus-specific PCR with no phenotypic screening. The presence of the pntAB-gdhA module in the sdaC and gltS loci was additionally verified by Sanger sequencing of the junctions between the construct and chromosome. The percentage of positive recombinants was calculated by dividing the titre of recombinants by the total number of Cm^R colonies. The exact numerator and denominator are listed in Supplementary Table S5. The bars indicate average efficiency of integration, and open diamonds indicate exact values for each biological replicate.

Figure 5.

cI-hok counterselection and ocr-assisted recombineering in non-conventional bacteria. (A) Schematic of experiments for estimating the efficiency of markerless cloning. Either the galETKM (galTKM in the case of Pantoea) or manXYZ operon was replaced with the cI-hok-neo cassette. Next, the cI-hok-neo cassette was replaced with a PCR-amplified construct carrying the wild-type operon. (B) The efficiency of markerless cassette integration in the gal and man loci of the Salmonella, Citrobacter, and Pantoea genomes. cI-hok counterselection and λRed recombineering were performed using pRedCmOcr, pRedCmOcr^SC101ts and pRedCmOcr^RSF helper plasmids for Citrobacter, Salmonella, and Pantoea, respectively. Both these techniques were implemented similarly to the experiments in E. coli with minor modifications, which are described in the section ‘Recombineering and cI-hok counterselection in Salmonella, Citrobacter, and Pantoea’ of the Supplementary Material and Methods. The recombinants were selected for Cm^R on tetrazolium agar supplemented with either 1% d-galactose or 1% mannose. Colonies exhibiting either the Gal⁺, or Man⁺ phenotype were recognised as recombinants. For each cassette, seven recombinant colonies were purified and used for verifying the construct presence within the targeted locus using locus-specific PCR (Supplementary Figure S8). The percentage of positive recombinants was calculated by dividing the titre of recombinants by the total number of Cm^R colonies. The exact numerator and denominator are listed in Supplementary Table S5. The bars indicate average efficiency of integration, and open diamonds indicate the exact values for each biological replicate.