High-Frequency Flp Recombinase-Mediated Inversions of the oriC-Containing Region of the Pseudomonas aeruginosa Genome

Abstract

The genomes of the two clonally derived Pseudomonas aeruginosa prototypic strains PAO1 and DSM-1707 differ by the presence of a 2.19-Mb inversion including oriC. Integration of two Flp recombinase target sites near the rrn operons containing the inversion endpoints in PAO1 led to Flp-catalyzed inversion of the intervening 1.59-Mb fragment, including oriC, at high frequencies (83%), favoring the chromosome configuration found in DSM-1707. The results indicate that the oriC-containing region of the P. aeruginosa chromosome can readily undergo and tolerate large inversions.

Pseudomonas aeruginosa is an opportunistic pathogen that can be found in diverse habitats. It causes a variety of acute infections and is also responsible for chronic life-threatening lung infections of cystic fibrosis (CF) patients (3, 9, 18). CF isolates are characterized by certain phenotypes, including rough lipopolysaccharide structure, mucoid phenotype, and loss of motility (3, 9, 17). Comparative genome mapping of Pseudomonas aeruginosa PAO with P. aeruginosa C, which belongs to a major clone found in CF patient infections and aquatic habitats, also revealed variations at the genomic level (13). CF isolates contained large chromosomal inversions, and the exclusive detection of inversions in isolates from the lungs of patients with CF, which represent atypical habitats for this bacterium, was cited as supporting the theory that features of this particular ecological niche may select, cause, or tolerate the observed genomic changes (10). This is what might have occurred between Escherichia coli and several closely related Salmonella species, including Salmonella enterica serovar Typhimurium (6). Since large chromosomal inversions could be constructed and stably maintained under laboratory conditions in E. coli (6), Salmonella serovar Typhimurium (8), and Bacillus subtilis (1), bacteria evidently have the inherent ability to tolerate and even select for gross chromosomal rearrangements. During construction of a Δ(mexAB-oprM) Δ(mexCD-oprJ) chromosomal double mutant in the PAO1 background by using a Flp recombinase-based method (5), we observed that the intervening 1.59-Mb region containing oriC (Fig. 1A) underwent inversions at high frequencies and decided to further investigate this phenomenon.

FIG. 1.

Genomic maps of P. aeruginosa PAO236 (A) and its inversion derivative PAO238 (B). Position 1 is defined as the first nucleotide of oriC. The locations of FRT sites in the Δ(mexAB-oprM)::FRT and Δ(mexCD-oprJ)::FRT mutants and their orientations are indicated by solid circles. The four rrn operons (16) and their chromosome coordinates are indicated by open circles. Since PAO236 is derived from PAO1, their map coordinates are identical except that PAO236 contains a FRT sequence inserted at the mexAB-oprM locus and a FRT-Gm^r-FRT cassette at the mexCD-oprJ locus.

Deletion of the mexCD-oprJ operon from a Δ(mexAB-oprM) strain.

Strain PAO200 [Δ(mexAB-oprM)::FRT] was previously described (14). The Δ(mexCD-oprJ)::Gm^r-FRT strains PAO236 and PAO237 were derived from PAO200 in several steps. First, the mexC-mexD-oprJ genes were deleted from pKMJ002 (2) by digestion with ClaI, followed by religation to form pPS1088. One of the delimiting ClaI sites is located 156 bp upstream of the mexC operon start at the ATG codon of nfxB, and the second ClaI site is located 209 bp downstream of the oprJ termination codon. Second, the deleted 6,138-bp DNA segment was replaced by the gentamycin resistance (Gm^r)-Flp recombinase target (FRT) cassette from pPS856 (5), followed by return of the deletion into the PAO200 chromosome by a previously described method (15). This procedure placed the Gm^r cassette and its flanking FRT sites into the chromosome at 5.15 Mb in the orientation shown in Fig. 1A, i.e., opposite the FRT site previously integrated into the chromosome at the mexAB-oprM locus at 0.47 Mb. Flp-catalyzed deletions or inversions were obtained after conjugal transfer of pFLP2 from E. coli mobilizer strain SM10 (5). Maintenance of pFLP2 was selected by plating the exconjugants on VBMM (5) with 500 μg of carbenicillin (Cb) per ml, and this plasmid was cured by plating cells on VBMM with 5% sucrose. The Flp-catalyzed deletion of the Gm^r-FRT cassette from PAO236, followed by inversion of the mexAB-oprM and mexCD-oprJ intervening chromosomal DNA segment, yielded strains PAO238 and PAO239. Similarly, two isolates containing the desired Δ(mexAB-oprM) Δ(mexCD-oprJ) mutations without the inversions were obtained and designated PAO277 and PAO278.

Evidence for Flp-mediated inversion of a large chromosomal DNA fragment.

To verify the deletion in PAO238, we isolated chromosomal DNA and performed genomic Southern analysis (5) with BamHI-digested chromosomal DNA, utilizing the insert of pPS1088 as the probe. This analysis did not yield the expected banding pattern, i.e., a single 3.9-kb BamHI fragment (see Fig. 2B and 2D, lane PAO277), but rather two BamHI fragments of 2.8 and 14.9 kb, respectively (Fig. 2D, lane PAO238). The latter pattern could only be explained by the fact that we had obtained a strain which had undergone the desired deletion event removing the Gm^r cassette (Fig. 2B), followed by an inversion between the FRT sites which would provide regions of homology with pPS1088 in two separate regions of the chromosome and on two distinct BamHI fragments (Fig. 2C). This was verified by PCR analysis with primers homologous to regions flanking the FRT insertion sites in the mexAB-oprM and mexCD-oprJ regions, respectively. Template DNA from strain PAO238 was obtained by a boiling preparation procedure, and PCR was performed as previously described (5). The results are shown in Fig. 3A. Using PAO238 DNA, PCR fragments were only obtained when the reactions were primed either with primer pair AB_up (5′-GTGAGCAAGCAGCAGTACGC) and CD_down (5′-AAGCGCTACGCGAGCTGATC) (392 bp) (Fig. 3A, lane 4) or AB_down (5′-GCCGAAGAGATCGAGTTCCC) and CD_up (5′-ACGGTCTCTCCGTGGTCCTC) (350 bp) (lane 5). This result supports the notion that strain PAO238 contained an inversion between the chromosomal FRT sites located in the mexAB-oprM and mexCD-oprJ regions of the chromosome, and the sizes of the PCR fragments were consistent with the ones expected from the inversion event.

FIG. 2.

Chromosomal maps of strains constructed in this study and PCR analysis. Maps of strain PAO236 [Δ(mexAB-oprM)::FRT Δ(mexCD-oprJ)::Gm^r-FRT] (A), its Gm^s derivative PAO277 (B), and a Gm^s derivative (PAO238) that has undergone a chromosomal inversion between the indicated FRT sites (C) are shown. DNA sequences and their transcriptional orientations from the mexAB-oprM and mexCD-oprJ operons are indicated in white and black boxes, respectively. (D) Genomic Southern analyses of strains isolated in this study. One-microgram samples of chromosomal DNAs isolated from the indicated strains were digested with BamHI and separated by electrophoresis on a 1% agarose gel in Tris-acetate-EDTA buffer (11). The separated fragments were transferred to Immobilon-P membranes (Millipore, Bedford, Mass.) and probed with the biotinylated insert from pPS1088. The probe was biotinylated, and the fragments were detected with the NEBlot Phototype labeling and Photostar detection kits from New England Biolabs (Beverly, Mass.), respectively. Lane M contained biotinylated λHindIII standards from New England Biolabs, and their sizes in kilobases are indicated on the left. Lanes: PAO1, wild-type; PAO236 [Δ(mexAB-oprM)::FRT Δ(mexCD-oprJ)::Gm^r-FRT]; PAO277 and PAO278 [Δ(mexAB-oprM)::FRT Δ(mexCD-oprJ)::FRT]; PAO238 [Δ(mexAB-oprJ)::FRT Δ(mexCD-oprM)::FRT], containing the 1.59-Mb chromosomal inversion; PAO238-F [Δ(mexAB-oprJ)::FRT Δ(mexCD-oprM)::FRT], a strain retaining the inversion after reintroduction of pFLP2 into PAO238; PAO281 and PAO282 [Δ(mexAB-oprM)::FRT Δ(mexCD-oprJ)::FRT], two strains that reverted back to the chromosomal configuration found in strains PAO277 and PAO278 after reintroduction of pFLP2 into PAO238.

FIG. 3.

(A) PCR analysis of strains PAO238, PAO277, and PAO281 utilizing the primer pairs indicated in the highlighted box. Aliquots of the PCRs were analyzed on 1.5% agarose gels in Tris-acetate-EDTA buffer (11) and stained with ethidium bromide. (B) Sequence analysis of PCR fragments from the mexAB-oprJ and mexCD-oprJ regions of the PAO chromosome. The PCR fragments from the reaction mixtures obtained from PAO238 (panel A, lanes 4 and 5) were purified from an agarose gel using a Geneclean kit (BIO 101, Vista, Calif.), and their sequences were determined by automated DNA sequencing at the University of Colorado at Boulder sequencing facility employing the same primers used in the PCRs. Top three lines, partial sequence of the PCR fragment obtained with primers AB_down and CD_up. Inverted arrows delimit chromosomal sequences and FRT sequences, respectively. Restriction enzyme cleavage sites found in the FRT site are indicated in lowercase letters, as is the initiation codon of nfxB (top). Bottom three lines, partial sequence of the PCR fragment obtained with primers AB_up and CD_down.

To further verify the inversion, the nucleotide sequences of the PCR fragments were determined. The sequences obtained (Fig. 3B) showed that the two PCR fragments contained hybrid sequences formed by the inversion event, i.e., mexC′ separated from ′oprM by FRT and mexA′ separated from ′oprJ by FRT. Finally, chromosomal macrorestriction patterns were examined after digestion of diverse chromosomal DNAs with PacI and separation of the restriction fragments by pulsed-field gel electrophoresis (PFGE) (10, 13) (Fig. 4). Digestion with PacI revealed the presence of the inversion in strains PAO238 and PAO239 versus their respective progenitor strains, PAO236 and PAO237. The 2,335- and 1,282-kb PacI fragments seen in PAO236 and PAO237 were changed in size to ∼2,200- and ∼1,500-kb PacI fragments in PAO238 and PAO239. For comparison, we also examined the PacI patterns of the chromosomes of PAO1, the wild-type strain used for genome sequencing, and those of DSM-1707 (13), another commonly used prototype strain that is clonally derived from the same PAO precursor strain as PAO1. Interestingly, the patterns observed in our inversion strains were similar to the ones seen in DSM-1707, whereas the patterns of PAO236 and PAO237 were most similar to those of PAO1. From genome sequence analysis, it is known that strains PAO1 and DSM-1707 differ by the presence of a 2.19-Mb inversion, including oriC, between the rrnA and rrnB operons located at 0.72 and 4.79 Mb, respectively (16) (Fig. 1A). This inversion changes the sizes of the involved PacI fragments from 2,335 and 1,282 kb in PAO1 to 2,160 and 1,454 kb in DSM-1707. The latter pattern is very similar to that observed in our inversion strains, since the inversions happened between loci that are not too distant from one another (Fig. 1A).

FIG. 4.

PFGE analysis of PacI digests of PAO genomic DNAs. Samples were prepared and digested with PacI, and fragments were separated by PFGE as previously described (10, 13). The strains analyzed were PAO236 and PAO237 [Δ(mexAB-oprM)::FRT Δ(mexCD-oprJ)::Gm^r-FRT], PAO238 and PAO239 [Δ(mexAB-oprJ)::FRT Δ(mexCD-oprM)::FRT], PAO1 (a prototrophic strain used for determination of the genomic sequence), and DSM-1707 (a prototrophic strain, clonally derived from the same PAO isolate as PAO1). White asterisks mark fragments that differ in individual strains. The sizes of PacI fragments from strain PAO1 and its derivatives are indicated on the left.

Frequency of Flp-mediated inversions.

Since the inversion occurred in three isolates obtained in two separate gene replacement experiments, inversions seemed to happen at very high frequencies, despite the absence of apparent selective pressure. To assess the frequency of inversion, we reintroduced pFLP2 into PAO236 and processed 20 individual exconjugants. All 20 exconjugants became Gm susceptible (Gm^s), indicating excision of the Gm^r cassette. The pFLP2 plasmid was cured from the same 20 Gm^s isolates, chromosomal DNA templates were obtained by the boiling method, and PCR analysis was performed with the primer pair AB_up and CD_down to determine which isolates contained the previously observed inversion. PCR fragments were observed in 16 of 20 isolates (data not shown), suggesting that 80% of the isolates contained the 1.59-Mb inversion of the chromosomal region located between mexAB-oprM and mexCD-oprJ. We suspected that the remaining four isolates did not contain the inversion and therefore should have the chromosome organization depicted in Fig. 2B. This was verified by performing PCR analyses with primer pairs AB_up and AB_down and CD_up and CD_down. Both primer pairs yielded PCR fragments of the expected sizes with DNA templates from all four isolates tested, and representative results obtained with one of these isolates are shown in Fig. 3A, panel PAO277. Whereas primer pairs AB_up and AB_down and CD_up and CD_down yielded the expected PCR fragments (274 and 470 bp) (Fig. 3A, lanes 1 and 2), all other primer pairs tested did not yield any PCR fragments. This result was verified by genomic Southern analysis of the same and a second isolate using pPS1088 insert DNA as the probe (Fig. 2D, lanes PAO277 and PAO278). The probe hybridized to a single 3.9-kb BamHI fragment, consistent with the chromosomal organization depicted in Fig. 2B.

Flp recombinase can revert the inversion in absence of selective pressure.

Since 19 of 23 isolates tested to this point contained an inversion, we entertained the idea that this may have been due to some fortuitous selective pressure exerted by the experimental conditions employed in the Flp recombinase-mediated step. To examine this possibility, we decided to perform the opposite experiment, i.e., reversion of the inverted segment back to the configuration found in the progenitor strain, PAO236. To do this, we introduced pFLP2 by conjugation into PAO238 and selected 20 individual exconjugants for further experimentation. The plasmid was cured from these isolates by plating on VBMM with 5% sucrose, and after single colony purification, chromosomal DNA templates were prepared by the boiling procedure. The presence of the original inversion was assessed by PCR analysis utilizing the primer pair AB_up and CD_down as described above. As evidenced by the presence of a PCR product, 16 of 20 isolates examined retained the original inversions, and 4 isolates yielded no PCR fragments (data not shown), indicating that they had potentially reverted back to the chromosomal configuration illustrated in Fig. 2B. When the PCR reactions were performed with primer pairs AB_up and AB_down and CD_up and CD_down, these primer pairs yielded fragments of 274 and 470 bp, respectively, in all four isolates tested (data not shown). Representative results obtained with one of these isolates are shown in Fig. 3A (strain PAO281). All other primer pairs tested did not produce any PCR fragments. This result was verified by genomic Southern analysis of the same and a second isolate using pPS1088 insert DNA as the probe (Fig. 2D, lanes PAO281 and PAO282). The probe hybridized to a single 3.9-kb BamHI fragment in both strains, consistent with the notion that they had reverted to the chromosomal organization depicted in Fig. 2B. Analysis of a nonrevertant, PAO238-F (Fig. 2D), using the same probe yielded the original pattern (2.8 and 14.9 kb) observed in PAO238.

Conclusions.

The results presented in this study suggest that the oriC-containing region of the P. aeruginosa chromosome can undergo inversions at high frequencies. Although our inversions were catalyzed by Flp recombinase after insertion of FRT sites into the genome, similar large inversions are found in at least two prototrophic P. aeruginosa isolates, i.e., strains PAO1 and DSM-1707, that are both clonally derived from the same original PAO isolate. In these strains, the inversions were probably RecA mediated and occurred between the rrnA and rrnB operons (16). Since the genome of laboratory strain PAO1 seems quite stable, RecA-mediated inversions seemingly do not happen at high frequencies, although no studies have ever been performed to address this issue. Even though our starting strains were derived from PAO1 and by PFGE resembled the sequenced P. aeruginosa wild-type strain, the frequencies of Flp-catalyzed inversions (>80%) favored the DSM-1707 versus the PAO1 chromosome arrangement. This finding was somewhat surprising, since inversions do not increase the amount of repetitive DNA in the chromosome, and thus the rate of reversal due to homologous recombination between the repeats would be expected to be equal the rate of initial occurrence, especially in the absence of any obvious selective pressure (6). It has been speculated that chromosome rearrangements may impact growth rate (4, 6). This notion is supported by the existence of an E. coli laboratory strain with a remarkable imbalanced inversion between two rrn operons with respect to oriC (4). This strain suffers from a severe growth defect, with strong selection for compensatory rearrangements restoring the natural gene order. In our case, however, there were no obvious growth rate differences when the inversion mutant PAO238, its parental strain PAO236 (a PAO1 derivative), and the revertant PAO280 were grown on VBMM, the medium on which the inversions were originally selected (data not shown). This finding is perhaps not surprising, since aside from the above-described example most inversions isolated in E. coli and Salmonella serovar Typhimurium have no significant effects on growth rates in vitro (6, 12). However, such chromosome rearrangements might affect bacterial infection, fitness, and growth in other niches, and it is therefore tempting to speculate that the differences observed in the PAO1 and DSM-1707 chromosomes might reflect different handling during propagation. It has recently been noted that P. aeruginosa population structure and genome evolution seem to be quite different when compared to the Enterobacteriaciae (7).

In an earlier publication describing the Flp-FRT procedure (5), we pointed out that a possible drawback to such an approach might be that recombination between FRT sites placed in the same chromosome could lead to a deletion or inversion of a large chromosomal segment, depending on the orientation of the FRT sites and the distance between them (19). Although we dismissed such large chromosomal rearrangements as unlikely, we now know that they can indeed happen and over large distances. To minimize such problems, it is therefore advisable when integrating a cassette into a second gene in the same genome to place the second cassette in the same orientation as the first, since large deletion events are more unlikely than inversions, especially when mutants are selected on minimal medium. With the available PAO1 and other bacterial genome sequences, such experiments will be possible and FRT cassettes will remain valuable tools for micro- and macromanipulations of entire bacterial chromosomes (19). Our observations also underscore the importance of verifying chromosomal mutations by PCR and/or Southern analysis.