Rapid and accurate construction of allelic exchange vectors using Gibson assembly
The construction of the allelic exchange vector for gene hasS was performed as shown in Fig. 1. First, the 500 bp upstream and downstream DNA sequences flanking the locus of interest were amplified (Fig. 2a). Gibson assembly was used to fuse and clone these fragments into the pEX18Tc allelic exchange vector as described in the Materials and Methods. Other high fidelity DNA polymerase such as Phusion DNA Polymerase can also be used in the PCR and Gibson assembly. The resulting deletion construct was transformed into E. coli DH5α. Colonies containing the deletion construct were confirmed by colony PCR using the universal pEX18 primer pair flanking the multiple cloning sites (Fig. 2b). Positive colonies containing both the upstream and downstream DNA fragments showed a 1 kbp insert and were further selected for DNA preparation and sequencing. The accuracy of the Gibson assembly was extremely high with more than 95% of the colonies tested during method development (n = 28) containing the mutant deletion allele with the correct sequence.
The pEX18 series of suicide vectors have been widely used for site-directed gene manipulation in P. aeruginosa and related bacteria [7]. The construction of a gene knockout vector usually involves several steps of subcloning. To accelerate the procedure, Wolfgang et al. [15] and Choi et al. [11] used overlap PCR and the Gateway cloning technology (Invitrogen) to generate gene deletion constructs. However, the Gateway technology procedure still requires several extra cloning steps: first, the upstream and downstream fragments are amplified by flanking PCR; secondly, the two fragments are fused together by overlap PCR; and finally the resulting deletion allele is purified by electrophoresis and cloned into a Gateway allelic exchange vector. Furthermore, the pEX18 suicide vectors must be modified to be compatible with the Gateway cloning system by adding the att flanking sites. Gateway proprietary reagents such as Clonase are used to recombine the deletion allele from the Gateway vector into the modified Gateway compatible pEX18 vector. Utilizing the method described herein the complete construct can be generated in two steps: first, the upstream and downstream DNA fragments are amplified by PCR; then the DNA fragments are ligated directly into a linearized pEX18 plasmid by single step Gibson assembly to generate the allelic exchange vector. Neither proprietary reagents nor modifications to the suicide vector are required.
Quick deletion construct transfer by electroporation and merodiploid selection
The transfer of the allelic exchange plasmid into P. aeruginosa is traditionally performed via conjugation. Specifically, the deletion construct is transformed into a conjugative E. coli strain, which is subsequently mated with recipient P. aeruginosa cells. After mating, transformed P. aeruginosa cells are isolated from the conjugation mixture by plating on media containing both triclosan and the antibiotic compatible with the plasmid resistance marker. Repetitive passage of the deletion construct under harsh double antibiotic selection places additional stress on the cells and may promote undesirable spontaneous mutations. Alternatively, we pursued a more direct approach via electroporation that avoids the multiple steps that may cause spontaneous mutations while shortening the protocol. Although transformation of P. aeruginosa by electroporation methods has previously been considered relatively inefficient, we have optimized the electroporation protocol described herein to transformation efficiencies similar to those obtained by conjugation.
To inhibit the DNA restriction-modification system of P. aeruginosa, the cells were grown at 42–43 °C overnight without shaking as described by Rolfe et al. [10]. To optimize the transformation efficiency, a panel of electroporation solutions was tested. Significant cell lysis was observed when water, 10–15% glycerol, HEPES buffer, 300–500 mM sucrose, 300–500 mM sorbitol or various combinations thereof were used. MgSO4 at 1 mM was found to effectively maintain cell structure integrity and not interfere with electroporation. Therefore, cells were harvested and washed in 1 mM MgSO4 at either 4 °C or 23 °C.
The optimum electric parameters were investigated by testing the transformation efficiency of a P. aeruginosa replicative plasmid pBSP11TcR at both 4 °C and 23 °C using 1 mm gap and 2 mm gap cuvettes at different voltages. The efficiencies of electroporation under these conditions are shown in Fig. 3.
Using these conditions the deletion construct was transferred into P. aeruginosa cells and integrated into the PAO1 chromosome by homologous recombination as shown in Fig. 4. Merodiploids in which the deletion construct is integrated into the chromosome usually appeared on the tetracycline selection plates within 40 h. In some cases, positive colonies appeared up to 60 h after electroporation. Beyond 72 h, spontaneous tetracycline resistant colonies began to appear. Suicide pEX18 vector constructs with alternate antibiotic resistance markers such as carbenicillin and gentamicin were also tested. However, colonies with spontaneous resistance against these antibiotics appeared as early as 20 h, which made selection of merodiploids more difficult. Therefore, tetracycline resistance appears to provide the most optimal selection marker for merodiploids by this method.
Tetracycline resistant colonies were analyzed by colony PCR to confirm the generation of merodiploids (Fig. 5). Using the universal pEX18 vector forward primer (primer 1) and a primer specific to a genomic sequence after the 500 bp downstream sequence (primer 4), colonies in which the plasmid integrated via the upstream homologous sequence yielded a ~ 2 kbp PCR product (Fig. 5, lane 3). Meanwhile, colonies in which the plasmid integrated via the downstream homologous sequence yielded a ~ 1 kbp band (Fig. 5, lane 2). The approximate crossovers for either of these crossover events are illustrated in Fig. 4. If the PCR was performed with the universal pEX18 vector reverse primer (primer 2) and a primer specific to a genomic sequence before the 500 bp upstream (primer 3), reversed patterns would be expected. Generally, 5–10 tetracycline resistant colonies were obtained, all of which were confirmed to be merodiploids by PCR.
Improved sucrose-sacB counter-selection of knockout mutants
Following confirmation by colony PCR, merodiploids were isolated by streaking selected colonies onto a fresh tetracycline selection plate to avoid carryover contamination. Isolated colonies were then streaked onto TYS10 plates containing 300 mM sucrose to select for colonies that have lost the integrated plasmid backbone containing sacB. The sacB gene was originally isolated from Bacillus subtilis and encodes levansucrase, which catalyzes the conversion of sucrose to levans, a molecule that is toxic to P. aeruginosa. The sucrose intolerance of merodiploids expressing sacB is caused by accumulation of levans in the periplasmic space, which is a slow process and does not cause immediate lethality. Direct streaking of tetracycline resistant colonies onto TYS10 plates and the alternative streaking from an intermediate culture in LB were directly compared and no difference was observed. However, considering the possibility that a wild type revertant may gain a growth advantage in an intermediate culture and outcompete the deletion mutant, streaking an isolated tetracycline resistant colony directly onto a TYS10 plate minimizes undesirable population bias arising prior to counter-selection. The TYS10 medium contains no salt and thus increases the sensitivity of cells to the accumulation of levans [16]. The sacB counter-selection was performed at room temperature to reduce sucrose hydrolysis and slow down the growth of P. aeruginosa to allow more effective selection of merodiploids having undergone the second homologous recombination event.
Sucrose-resistant colonies were analyzed by colony PCR. As illustrated in Fig. 4, two populations formed: one included the desired deletion mutants while the other included wild type revertants. The theoretical ratio between two populations is 1:1 and in the case of hasS, the ratio of ΔhasS vs wild type of 14 randomly selected colonies was 8:6, as judged by the 2 kb versus 1 kb insert, respectively (Fig. 6). However, the ratio may be biased towards one population if for example the gene to be deleted is beneficial for bacterial growth. The deletion of hasS was confirmed by DNA sequencing (Additional file 3).