The mutagenesis was carried out by electroporating a polymerase chain reaction (PCR)-product that contains an antibiotic cassette flanked by sequences homologous to the targeted DNA into a strain expressing the lambda Red operon. The presence of the specific PCR product promotes the deletion of the chromosomally located targeted region via homologous recombination between the genomic region and the flanking sequences of the PCR product. In order to use this system in P. aeruginosa, we first cloned the lambda Red operon into the pUCP18 vector. We also cloned the sacB gene into this vector to enable a rapid cure of the plasmid following mutagenesis. The generated vector is called pUCP18-RedS (Genbank EU073163) (Additional file 1). Initially, the pqsC gene was selected for knockout because the mutant phenotype can be differentiated from the wild-type strain by its inability to produce the Pseudomonas aeruginosa blue pigmented phenazine, pyocyanin, lack of which results in a yellowish culture (unpublished data). We used a 3-step PCR procedure to generate a fragment containing the kanamycin resistance cassette flanked by regions surrounding the pqsC gene (Fig 1A). In the first step, P. aeruginosa genomic DNA was used to amplify the upstream and downstream regions of the target sequence with primers F_uppqsC/R_uppqsC-kan and F_downpqsC-kan/R_downpqsC (all primer sequences are shown in Additional file 2). In each pair, one primer contains at its 5' end a 20 bp-region homologous to the kanamycin cassette. Simultaneously, the pUC4K plasmid was used to amplify the kanamycin resistance gene. In the second step, the three fragments and primers F_uppqsC/R_downpqsC were mixed and a product containing the upstream, kan and downstream regions was obtained. The third step is performed with primers F_uppqsC/R_downpqsC to obtain a large quantity of PCR product.

We first assessed whether flanking regions of ~600-nt were sufficient to enable homologous recombination between the target gene and the PCR product, as this is generally sufficient when a typical suicide vector is used. Primers F_uppqsC/R_uppqsC-kan and F_downpqsC-kan/R_downpqsC were designed to amplify regions 651-nt upstream and 610-nt downstream of pqsC, and to leave intact the 5' and 3' ends of the gene in the resulting recombinant mutant (Fig. 1A). The 2.2 kb PCR fragment obtained was electroporated into PA14/pUCP18-RedS previously induced by arabinose for the expression of the lambda Red proteins. Aliquots of 2 or 4 μg of electroporated DNA allowed the selection of 10–20 Km^R recombinants; that colony yield was approximately doubled when 8 μg of DNA was used. All Km^R colonies obtained were analyzed by assessing their ability to produce pyocyanin. Approximately 80% of the colonies potentially carried the insertion, as revealed by their inability to produce pyocyanin (Fig. 2). The correct integration of the kanamycin cassette was then confirmed by PCR in five independent colonies. Figure 3A shows results obtained for one representative ΔpqsC recombinant. A PCR product of 868 bp was obtained using primers F_upout-pqsC/R_inkan (Additional file 2) and the mutant genomic DNA as template, while no PCR product was obtained when the wild type genomic DNA was used. In addition when the wild type and ΔpqsC genomic DNA were used as template, primers F_upout-pqsC/R_downout-pqsC (Additional file 2) amplified a product of 2757 bp and 2550 bp (720 bp of pqsC was replaced by 927 bp of the kanamycin cassette) respectively (Fig 3A). The correct and unique insertion of the kanamycin resistance cassette was verified by Southern. HindIII-digested genomic DNA of three independent colonies was hybridized with the 5' end of the kanamycin cassette, and only one band of 8016 bp was obtained confirming that the cassette was inserted into pqsC (Fig 3B). Finally, to ensure inclusion of proper borders in the inserted kanamycin cassette, 700-bp regions upstream and downstream of the cassette were sequenced. The sequences indicated proper insertion of the cassette and that no mutations were incorporated during the 3-step PCR procedure as the flanking sequences were identical to the original.

Using 3-step PCR, we also successfully obtained two supplementary mutants: ΔlasR and ΔkynBU. The ΔlasR mutant was generated using 490-nt upstream and 445-nt downstream flanking homologous regions resulting in the replacement of 93 bp of lasR by the kanamycin cassette. As expected the lasR mutant generated was defected in pyocyanin production (Fig 2). Five independent colonies were analyzed and the correct insertion was confirmed in all five colonies using primers F_upout-lasR/R_inkan and F_upout-lasR/R_downout-lasR (Additional file 2 and Fig. 4A). The ΔkynBU mutant was generated in both wild type and trpE^-phnAB^- backgrounds using 471-nt upstream and 419-nt downstream flanking homologous regions. In these mutants 1041 bp region of kynBU was replaced by the 927 bp in length kanamycin cassette. The trpE^-phnAB^-ΔkynBU mutants were functionally validated also using pyocyanin as readout. In fact, P. aeruginosa is unable to produce pyocyanin if the bacterium is impaired in its capacity to synthesize 4-hydroxyl-quinolines (HAQs). Recently it was reported that two distinct pathways supply anthranilate as the precursor of HAQs 10. Anthranilate can be either derived from the conversion of chorismate to anthranilate by an anthranilate synthase: TrpEG and PhnAB; or produced by the degradation of tryptophan through the kynurenine pathway: KynBU. While kynBU and trpE^-phnAB^- mutants were still able to produce pyocyanin in rich medium (Fig 2) due to the fact that the anthranilate or kynurenine pathways supplied anthranilate respectively, no trace of pyocyanin was present in any trpEphnAB^-ΔkynBU recombinants tested (Fig 2). Also, DNA amplification was performed in five independent ΔkynBU and trpEphnAB^-ΔkynBU colonies using primer pairs F_upout-kynBU/R_inkan and F_upout-kynBU/R_downout-kynBU (Additional file 2) and confirmed the insertion of the kanamycin cassette in all five ΔkynBU and trpEphnAB^-ΔkynBU recombinants (Fig 4A). Note that PCR products of similar size were obtained with primers F_upout-kynBU/R_downout-kynBU when wild type or mutant genomic DNA was used as templates, as 1024 bp of kynBU were replaced by 927 bp of kanamycin cassette. Altogether these results demonstrate that the lambda Red technique is an efficient system for generating mutants in P. aeruginosa.

The mutagenesis process could be simplified by using a PCR fragment with a shorter region of homology to the target gene. This would enable the product to be obtained in a 1-step PCR using long primers that are homologous to the kanamycin resistance cassette at their 3' ends and homologous to the target gene at their 5' ends. Thus in order to reduce the length of homologous extension necessary to select for P. aeruginosa recombinants, the following four sets of primers were designed to amplify the pqsC::kan locus: F-pqsC₆₀₀/R-pqsC₆₀₀, F-pqsC₃₀₀/R-pqsC₃₀₀, F-pqsC₁₅₀/R-pqsC₁₅₀ and F-pqsC₁₀₀/R-pqsC₁₀₀ (Additional file 2). The PCR products generated contained the antibiotic cassette flanked with 600, 300, 150 or 100-nt homologous regions to the pqsC-surrounding region; when 8 μg of each product was electroporated, 35, 50, 60 and 40 Km^R colonies were obtained, respectively, and out of those 30, 46, 50 and 34 positive recombinants were identified based on their inability to produce pyocyanin, which corresponds to a 83–92% positive rate. Five recombinants from each transformation event were analyzed by PCR and the expected products were obtained systematically using primers F_upout-pqsC/R_inkan and F_upout-pqsC/R_downout-pqsC (Additional file 2). These results show that in the case of pqsC mutagenesis the number of positive recombinants was not dependent on the length of the homologous region used and that a 100-nt homologous sequence was sufficient to allow recombination.

To further test the feasibility of 1-step PCR for this application, we designed and purchased primers (Sigma Aldrich) that contain at their 5' extremity 100-nt homology to the flanking regions of kynBU (F-kynBU₁₀₀-kan/R-kynBU₁₀₀-kan) or lasR (F-lasR₁₀₀-kan/R-lasR₁₀₀-kan) and at their 3' extremity 22–24 nt homology to the kanamycin resistance cassette (Additional file 2). The kanamycin resistance cassette was amplified using pUC4K as the template and a mixture of two sets of primers: F₁₀₀-kan/R₁₀₀-kan and F_up/R_down (Fig 1B). The resultant PCR product was used for electroporation. ΔkynBU recombinants, but not ΔlasR mutants, were obtained using 8 μg of PCR product and confirmed by PCR as described above. Our inability to obtain the ΔlasR mutant with this protocol indicates that the 1-step PCR method cannot be considered highly reliable and that 100 nt is likely close to the minimal length of homology needed.

The lambda Red methodology was combined with 1-step PCR to transfer a knockout mutation from the original strain in which it was generated to other strains. We amplified the lasR::kan locus using the DNA of the previously obtained mutant as the template and the primer pair F-lasR/R-lasR. The amplification generated a PCR product containing 490-nt and 445-nt upstream and downstream flanking target regions. Eight-microgram aliquots of PCR product were electroporated into three PA14 isogenic mutants (ΔmvfR, ΔpqsE and ΔpqsA), and all three strains successfully acquired the mutated locus, which was demonstrated by DNA amplification using primer pairs F_upout-lasR/R_inkan and F_upout-lasR/R_downout-lasR.

A comprehensive non-redundant PA14 transposition mutant library was recently constructed by random insertion of the MAR2xT7 transposon which confers gentamycin-resistance 11. We tested whether the lambda Red methodology would allow the transfer of loci where the MAR2xT7 transposon has been randomly inserted. The gentamycin cassette that had been inserted into pqsC was amplified using primers F-pqsC₆₀₀/R-pqsC₆₀₀ to generate a product with 379-nt upstream and 471-nt downstream pqsC-surrounding regions. Similar to the deletion experiments above, 2, 4 or 8 μg samples of DNA were electroporated into PA14/pUCP18-RedS and yielded 14, 14 and 50 Gm^R recombinants, respectively. Five recombinants were tested for the inability to produce pyocyanin in cultures and the presence of the MAR2xT7 transposon was checked by PCR amplification using primers F-pqsC₆₀₀/R-pqsC₆₀₀, respectively (data not shown). These results demonstrate that the lambda Red technique can be used in conjunction with a mutant library to achieve rapid transfer of mutated genes into various PA14 isogenic mutants and thereby enable the creation of multi-locus knockout mutants.

Lambda Red-based methodology has the potential of enabling large regions of chromosomal DNA, including large gene clusters and operons and pathogenicity islands, to be deleted. We examined whether the lambda Red system could be also used to delete the entire P. aeruginosa HSI-II locus 12, an approximately 24 kb genomic region encoding a putative type VI secretion system located between PA14_43110 and PA14_42880 genes. The PCR procedure followed to obtain the PCR product necessary to delete the HSI-II locus is presented in Fig 1C. First, genomic DNA of mutants of the PA14 non-redundant transposon library: PA14_43100::MAR2xT7 and PA14_42880::MAR2xT7 were used as template to generate the PCR fragment containing the gentamycin cassette flanked by the HSI-II locus-borders. Primers F_up-PA14_43100/R_up-MAR2xT7 were designed to amplify MAR2xT7 and the 890 nt upstream sequence when inserted into PA14_43100; while F_down-MAR2xT7/R_down-PA14_42880 were used to amplify MAR2xT7 and the 955 nt downstream sequence when inserted into PA14_42880 (Additional file 2). In a second PCR-step, the resulting upstream and downstream fragments were mixed with the primers F_up-PA14_43100 and R_down-PA14_42880 and a PCR product containing MAR2xT7 flanked by the borders of the HSI-II locus obtained. PCR product aliquots of 2, 4 or 8 μg were electroporated into PA14/pUCP18-RedS, and 256, 280 and 300 recombinants were selected on gentamycin, respectively. All mutants tested were producing pyocyanin (Fig 2). DNA amplification using primers F_up-PA14_43100/R_up-MAR2xT7 and F_up-PA14_43100/R_down-PA14_42880 confirmed deletion of the target region in 5 independent colonies; Fig 4B shows the PCR products obtained for one representative recombinant. While no DNA amplification with the primers F_up-PA14_43100/R_down-PA14_42880 flanking the HSI-II locus was obtained when wild type genomic DNA was used as a template, a fragment of 2844 bp was amplified when mutant genomic DNA was used (Fig 4B).

Pseudomonas and E. coli strains were grown in Luria Bertani medium (LB) at 37°C with aeration and when necessary pseudomonas cultures were supplemented with kanamycin (400 μg/ml), gentamycin (15 μg/ml) or carbenicillin (300 μg/ml). Pseudomonas strains used in this study include ΔpqsA 16, ΔpqsE 17, ΔmvfR 18, PA14_42880::MAR2xT7 and PA14_43100::MAR2xT7 11. The trpE-phnAB- double mutant was constructed in this study: 1293 bp of phnA and 300 bp of phnB were deleted by allelic exchange in the auxotroph trpE::Mar2xT7 mutant background. The vector pKOBEG-sacB 4 contains the Red operon expressed under the control of the arabinose inducible pBAD promoter and the sacB gene that is necessary to cure the plasmid 24. Since pKOBEG-sacB was not capable of replication in P. aeruginosa, the Red operon-araC fragment obtained after digestion with KpnI and HindIII was cloned into the multi clonal site of the E. coli – P. aeruginosa shuttle vector pUCP18 (Genbank U07164) 19 to create the plasmid pUCP18-Red. In addition, the NdeI sacB fragment from pKOBEG-sacB was cloned into pUCP18-Red, previously digested with NdeI, to create pUCP18-RedS (Genbank EU073163) (Additional file 1). P. aeruginosa recombinants were easily cured from pUCP18-RedS by streaking the mutant strains on NaCl-free LB agar plates supplemented with 10% sucrose.

Approximately 5 μg of DNA from P. aeruginosa and 1 μg of pUC4K DNA was digested by HindIII. The 5' end of the kanamycin cassette was amplified using primers 5'-gccacgttgtgtctcaaaat-3' and 5'-gcatttctttccagacttgttc-3', and the PCR product was labeled using the ECL kit (Amersham) and used for hybridization following manufacturer's instructions.

All PCRs were performed using High Fidelity Taq polymerase (Roche), 1 μM of each primer, 100 μM of dNTPs. DMSO 8% was added to the reaction when P. aeruginosa genomic DNA was used as template. Most of the primers were generated by the Massachusetts General Hospital DNA oligo core facility. The long primers were obtained from Sigma Aldrich.

A kanamycin resistance cassette with long flanking regions homologous to the target gene was generated by a previously described 3-step PCR protocol 420 with minor modifications. P. aeruginosa genomic DNA was first used as a template to amplify the regions flanking the target gene with the primers F_up/R_up-kan and F_down-kan/R_down (Additional file 2 and Fig. 1A). R_up-kan and F_down-kan contain, respectively, at their 5' extremity 19 and 22-nt regions that are homologous to the 5' and the 3' terminals of the kanamycin resistance gene. The plasmid pUC4K 21 was used as a template to amplify the kanamycin cassette with the primers F-kan/R-kan (Additional file 2). In a 2^nd PCR, 10 ng, 25 ng or 50 ng of each of the three fragments were mixed with the primers F_up/R_down. A third PCR was finally performed using again the F_up/R_down primers and the product of the second PCR as the template in order to increase the yield of the sought band (Fig. 1A). Five hundred microliters (10 reactions) of the 3^rd PCR were ethanol precipitated, dissolved in 10 μl of water and dialyzed for 1 h to eliminate salt that can interfere with electroporation.

The 1-step kynBU or lasR PCR products were generated using pUC4K (Genbank X06404) as template with a mixture of long and short primers (Fig 1B). The long primers contained at their 5' extremity 100-nt homology to the flanking regions of kynBU (F-kynBU₁₀₀-kan/R-kunBU₁₀₀-kan) or lasR (F-lasR₁₀₀-kan/R-lasR₁₀₀-kan) and at their 3' extremity 22–24 nt homology to the kanamycin resistance cassette. The short primers F-kynBU/R-kunBU and F-lasR/R-lasR were homologous to the 5' extremities of the kynBU or the lasR long primers, respectively.

The DNA concentration was measured by spectrophotometer at 260 nm and the DNA molecular weight was visually confirmed on agarose gel.

P. aeruginosa PA14/pUCP18-RedS was grown in LB with carbenicillin to an optical density at 600 nm (OD₆₀₀) of 0.4 at 37°C. Then cells were induced for 1.5 h by treatment with 0.2% L-arabinose; the bacteria were rendered electrocompetent by four washings with 10% sucrose at room temperature and were condensed to a concentration of about 10¹¹ /ml. Electroporation was carried out using 100 μl of bacteria solution and no more than 10 μl of DNA. The electroporated cells were diluted in 2 ml LB and incubated for 2 h at 37°C. Transformants were selected by plating the electroporated cells on antibiotic-imbued plates and further verified by PCR for the correct insertion using two set of primers F_upout/R_inkan and F_upout/R_downout, with the latter pair hybridizing outside the DNA region used for the recombination event.