Defining the Pseudomonas aeruginosa SOS Response and Its Role in the Global Response to the Antibiotic Ciprofloxacin

Bacterial strains and growth. P. aeruginosa PAO1 was obtained from G. Sundin. Unless specified, solid medium was Lennox LB (28) plus 1.6% agar (LBA); liquid medium was Miller LB (28) (LB). For selection, antibiotics were used for E. coli and P. aeruginosa PAO1, respectively, as follows: streptomycin (Sm), 30 μg/ml and 250 μg/ml; gentamicin (Gm), 15 μg/ml and 50 μg/ml. Ciprofloxacin was obtained from MP Biomedicals (Aurora, Ohio) and used at the concentrations indicated below. All bacteria were grown aerobically at 37°C.

Strain construction.Primer sequences were designed based on the P. aeruginosa genome database (http://v2.pseudomonas.com ) (35, 39). A lexA allelic exchange cassette was assembled containing ∼800 bp of homology surrounding lexA, the lexA open reading frame, and the Gm^r marker from vector pBBR1MCS-5 (22) using assembly PCR and the primers listed in Table S1 at the website http://www.scripps.edu/chem/romesberg/ . The resulting cassette was cloned into vector pKNG101 (20) to create pRTC0021; the S125A mutation was then introduced using primers PA_lexA_S125A_QCF and PA_lexA_S125A_QCR and the QuikChange site-directed mutagenesis kit (Stratagene) to create vector pRTC0022.

pRTC0022 was transformed into E. coli strain SY-17 and introduced into P. aeruginosa by conjugative transfer with selection on M9 plus 0.2% citrate and Gm to select clones that integrated the allelic exchange cassette into the chromosome by either a single- or double-crossover event. Replica plating onto M9 plus 0.2% citrate containing either Gm or Sm identified clones containing the allelic exchange cassette and lacking the vector sequences due to a double-crossover event. Colonies were verified as Gm^r and Sm^s, and the mutation was confirmed by sequencing.

Confirmation that the LexA(S125A) mutant is not cleaved in response to ciprofloxacin.For each strain, five clones were grown in LB for 18 h. Cultures were diluted 1:500 and grown to mid-log phase (optical density at 60 nm [OD₆₀₀], ∼0.4 to 0.5), and then ciprofloxacin was added to a final concentration of 1 μg/ml. At 0, 30, and 120 min following ciprofloxacin addition, cell aliquots were removed and stored at −20°C. During the experiment, the OD₆₀₀ and viable CFU per ml were monitored for each of the cultures (see Fig. 1A and B, below). This protocol is identical to that used to prepare samples for the transcriptional studies (below). Whole-cell lysates were prepared by sonication in phosphate-buffered saline, and the soluble fraction was collected and normalized for total protein concentration (Bio-Rad protein assay). Samples were separated on a 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel and transferred to a 0.2-μm nitrocellulose membrane. Immunostaining was performed with a rabbit polyclonal antiserum to LexA (1:8,000; 2 h; kindly provided by J. Little) and horseradish peroxidase-linked anti-rabbit antibody (1:20,000; 1 h; Upstate Biotechnology), followed by detection with ECL Plus (GE Biosciences).

• Open in new tab
• Download powerpoint

FIG. 1.

Population kinetics of P. aeruginosa strains during microarray and whole-cell lysate sample preparation monitored by OD (A) and CFU/ml (B). Filled and open triangles represent PAO1 and the LexA(S125A) mutant, respectively. Time zero is defined as the point immediately following ciprofloxacin addition. As described in the text, samples for total RNA extraction and whole-cell lysates were collected at 0, 30, and 120 min post-ciprofloxacin addition. (C) Analysis of total full-length LexA measured by Western blotting with anti-LexA antibody.

Growth rate measurement.For each strain, three cultures were inoculated into 2 ml tryptic soy broth (Difco) and grown for 16 h with shaking, followed by continued growth after a 100-fold dilution in tryptic soy broth. At the indicated time points, the OD₆₀₀ was measured and the number of viable CFU per ml was determined by dilution plating.

Sensitivity to UV light and MMS.Three independent cultures of each strain were grown overnight in LB. Appropriate dilutions were plated onto LBA, and UVC irradiations were performed using a G8T5 germicidal tube (Ushio America, Cypress, CA). UV fluences were determined using a UVX radiometer with a UVX-25 sensor (UV Products). After irradiation, plates were protected from light and incubated for 2 days before colonies were counted. To determine the methyl methanesulfonate (MMS; Aldrich) sensitivity, three independent cultures were grown overnight in LB. Appropriate dilutions were plated onto LBA containing MMS at the indicated concentrations and incubated for 2 days before colonies were counted.

MIC determination.For each strain, three independent cultures were grown for 25 h in LB containing no antibiotic. From each culture, ∼10⁵ CFU were used to inoculate LB containing increasing concentrations of ciprofloxacin in 96-well plates. Inoculations were done in duplicate to yield a total of six data points per strain. After 18 h of incubation, growth was measured by reading the OD₆₅₀ in a V _max Kinetic microplate reader (Molecular Devices, California). The MIC was defined as the lowest concentration of ciprofloxacin that prevented any detectable growth.

Transcriptional analysis. P. aeruginosa genome arrays containing 25-mer probe sets for over 5,500 open reading frames from PAO1, 199 probe sets corresponding to 100 intergenic regions, and 117 probe sets from other P. aeruginosa strains were obtained from Affymetrix (Santa Clara, CA). A complete description and annotation for this P. aeruginosa genome array is available at http://www.affymetrix.com .

Sample preparation and data analysis.For each strain, five clones were inoculated in LB and grown for 18 h. Cultures were diluted 1:500 and grown to mid-log phase (OD₆₀₀, ∼0.4 to 0.5), at which point ciprofloxacin was added to a final concentration of 1 μg/ml. At 0, 30, and 120 min following ciprofloxacin addition, appropriate volumes from each of the five cultures per strain were pooled and added to 2 volumes of RNAprotect reagent (QIAGEN); cell pellets were stored at 4°C until RNA extraction. Total RNA was extracted using the RNeasy Mini kit (QIAGEN) at the end of the sample collection period. This procedure was repeated three independent times to generate three samples each just prior to and 120 min post-ciprofloxacin addition. Details of data analysis and reverse transcription-PCR validation have been provided along with our supplementary data sets via the internet (http://www.scripps.edu/chem/romesberg/ ).

Microarray accession numbers.Microarray data have been deposited at the National Center for Biotechnology Information's Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/ ) under the accession number GSE5443.

General characterization of the P. aeruginosa SOS response.We constructed a lexA(S125A) mutant of P. aeruginosa PAO1, where the catalytic serine of LexA has been replaced with alanine. We monitored the levels of full-length LexA in response to added ciprofloxacin (1 μg/ml ciprofloxacin) and showed that while the wild-type protein underwent cleavage, the mutant protein remained intact (Fig. 1).

No significant growth attenuation was observed with the lexA(S125A) strain relative to its isogenic parental strain (log phase doubling times of 45.0 ± 3.5 and 45.5 ± 4.5 min were observed for the LexA mutant and PAO1 strains, respectively [means ± standard deviations]), as was previously observed with analogous mutants of E. coli (15). Relative to the wild-type strain, we found that the lexA(S125A) mutant was hypersensitive to UV irradiation (Fig. 2A) but not to MMS (Fig. 2B) or ciprofloxacin [the ciprofloxacin MIC for both wild-type PAO1 and the lexA(S125A) strain was 0.125 μg/ml]. This suggests that the P. aeruginosa SOS response is important for repairing DNA damage associated with UV irradiation but not with MMS or ciprofloxacin.

• Open in new tab
• Download powerpoint

FIG. 2.

Killing kinetics of P. aeruginosa strains, as shown by survival following UV (A) or MMS (B) treatment. PAO1 and the LexA(S125A) mutant are represented by filled and open triangles, respectively.

Characterization of the P. aeruginosa transcriptional response to ciprofloxacin.Using DNA microarrays, we transcriptionally profiled mid-log-phase PAO1 at 30 and 120 min after exposure to suprainhibitory concentrations of ciprofloxacin (1 μg/ml; 8× MIC) (Fig. 1A and B). Only the magnitude of the observed changes was different, and so we focused the analysis on the 120-min data set. We observed that the levels of 196 transcripts increase at least twofold and the levels of 408 transcripts decrease at least twofold relative to levels immediately before the addition of the drug (Table 1; see also our supporting information at http://www.scripps.edu/chem/romesberg/ ).

Sixty-four of the genes that are up-regulated in response to ciprofloxacin are in regulons that are likely controlled by LexA-like repressors (Table 2). For example, the autoregulated LexA-like Ser-Lys dyad repressor PtrR (25) is up-regulated by 5-fold, and seven genes thought to be directly or indirectly under its control are up-regulated by up to 100-fold. One of these genes, ptrB, acts to repress the type III secretion system (40). Thus, via induced cleavage of PtrR, exposure to ciprofloxacin results in a down-regulation of the type III secretion system. In addition, 35 nearby cryptic prophage genes spanning from PA0614 to PA0648 are strongly up-regulated. Another putative phage repressor that shares similarities with LexA and is highly homologous to PtrR is PA0906. PA0906 is divergently transcribed from a putative operon that spans PA0907 to PA0911. PA0906 and the five genes of this operon are up-regulated 20- to 80-fold after exposure to the drug. Overall, the data suggest that at least one-third of the positive transcriptional response to ciprofloxacin is controlled by LexA-like repressors.

The number of genes that are down-regulated in response to ciprofloxacin is more than twice the number that are up-regulated. The down-regulated response appears to involve virtually every facet of cellular metabolism, including general metabolism, cell wall/capsule biosynthesis, DNA replication/repair, cell division, motility, and quorum sensing (Table 1; see also our supporting information at http://www.scripps.edu/chem/romesberg/ ). Changes observed were similar for all genes in a given operon, supporting the physiological significance of their regulation.

We observed significant and consistent changes in the operons encoding the subunits of ATP synthase (PA5553 to PA5561), which all decrease between 4- and 13-fold, and the subunits of NADH dehydrogenase complex I (PA2637 to PA2649), which all decrease between 2- and 5-fold. Similar decreases were observed in the response to acute H₂O₂ damage (30). In addition, while nrdA and nrdB, which encode the ribonucleotide reductase complex, are both up-regulated in response to ciprofloxacin four- to eightfold, two genes in a separate operon (PA5496 and PA5497) that are predicted to encode an alternate ribonucleotide reductase are down-regulated three- to fourfold. In addition, genes encoding many other proteins involved in metabolism were down-regulated after exposure to the drug (Table 1; see also our supporting information at the website http://www.scripps.edu/chem/romesberg/ ).

In addition to the decreased transcription of genes involved in general metabolism, decreases are also observed with genes involved in DNA metabolism. An operon containing genes that encode components of the replication machinery, including dnaA, dnaN, recF, and gyrB, is down-regulated ∼4-fold in response to ciprofloxacin. In addition, genes encoding DNA polymerase I, the HolB subunit of DNA polymerase III, and the DNA binding protein HU are all down-regulated two- to fivefold. The recG and ruvABC genes, all encoding proteins thought to be important for repairing ciprofloxacin-induced damage (7, 11), are down-regulated, albeit less than twofold. In contrast, recA, recX, and recN are up-regulated 7- to 17-fold. Interestingly, the three genes encoding damage-inducible DNA polymerases, PA0923, PA0670, and PA0669, are up-regulated in response to ciprofloxacin. PA0923 encodes a dinB-like Y-family polymerase and is up-regulated fourfold in response to ciprofloxacin. PA0670 and PA0669 encode two polymerases recently shown to be involved in damage-induced mutagenesis in Caulobacter crescentus (16) and are up-regulated two- and sixfold, respectively. The overall pattern of expression in the DNA replication genes suggests a shift from the canonical DNA replication enzymes to the inducible polymerases in response to ciprofloxacin.

Nearly all of the major cell division and lipopolysaccharide genes are significantly down-regulated. Of particular note are the changes observed in the wbp region, which encodes the B-band lipopolysaccharide O antigen and spans from PA3141 to PA3160. Transcription of these genes decreased by two- to sixfold after exposure to ciprofloxacin. Another interesting trend is the down-regulation of 41 genes that encode proteins involved in motility. We also observe a two- to fourfold increase in transcription of two major efflux proteins (MexC and MexR) and a four- to fivefold decrease in transcription of three major membrane pore proteins (OprD, OprG, and OprI). These changes in mobility and permeability are consistent with the general trend toward reduced metabolic activity in response to antibiotic exposure.

Contribution of the P. aeruginosa SOS genes to the ciprofloxacin-induced transcriptional response.We next characterized the transcriptional response to ciprofloxacin in a lexA(S125A) PAO1 mutant under the same conditions as those used to characterize the wild-type strain. As expected, lexA, recA, recX, and recN are induced by ciprofloxacin in a LexA cleavage-dependent manner (Table 3). In addition, PA3413 and PA1045 are regulated by LexA. PA3413 is a probable homolog of E. coli yebG, which is LexA regulated in E. coli (29), but its biological function is not known. PA1045 appears to encode a DinG helicase (37), which is related to the mammalian XPD family of helicases, and it may play a role in transcription-coupled repair.

The induction of several hypothetical genes was found to depend on LexA cleavage: PA2288, PA3414, PA1044, PA0069, and PA0922. PA3414 is predicted to encode a protein of unknown function, but its location next to yebG (see above) suggests that these genes may be coordinatively transcribed. PA2288 encodes a hypothetical protein of no known function. However, it has recently been shown that mutation of wspF, which disrupts a signal cascade involved in biofilm formation, causes a mild induction of lexA and recA (each 1.6-fold) and a 4.0- and 2.8-fold increase in PA3414 and PA2288 expression, respectively, supporting the association of these genes with the SOS response (19). PA1044 encodes a hypothetical protein with no known function that is divergently transcribed from PA1045 (see above). PA0922 and PA0069 are predicted to encode a hypothetical protein and a photolyase-like protein, respectively.

In E. coli there are three nonessential polymerases, each of which is LexA regulated: Pol II (encoded by polB), Pol IV (encoded by dinB), and Pol V (encoded by umuC and umuD). The P. aeruginosa genome encodes three nonessential polymerases, PA0923, PA0669, and PA0670. As mentioned above, transcription of PA0923, which encodes a polymerase that is highly homologous to E. coli dinB, is induced by ciprofloxacin; however, its induction is not LexA regulated. This agrees with recent findings in other organisms (6, 36) and suggests that a LexA-regulated dinB polymerase may be more the exception than the rule. In contrast, PA0669 and PA0670, which appear to be encoded in the same operon, are induced by ciprofloxacin in a LexA cleavage-dependent manner. PA0669 is predicted to encode an alternate alpha-subunit, and PA0670 is predicted to encode a Y-family polymerase. This operon resembles one that was recently found to play a role in damage-induced mutagenesis in the α-Proteobacteria Caulobacter crescentus (16), although a PAO1 mutant lacking this operon shows no signs of increased sensitivity to UV, MMS, or ciprofloxacin-mediated damage (R. T. Cirz and F. E. Romesberg, unpublished results). While we were able to detect sufficient levels of PA0670 in the microarray studies, the level of PA0669 was too low to observe a rigorous statistical difference between PAO1 and the LexA mutant directly (supporting information can be found at http://www.scripps.edu/chem/romesberg/ ). However, using real-time PCR we were able to detect PA0669 mRNA after ciprofloxacin treatment, but not in the LexA mutant, confirming that this gene is LexA regulated (see Table S2 in our supporting information at http://www.scripps.edu/chem/romesberg/ ).

Analysis of the P. aeruginosa LexA box and other potential SOS genes.By identifying the SOS regulon empirically, we were also able to identify a consensus binding sequence for LexA, CTG-TATAA-ATATA-CAG (bold residues are 100% conserved) (Table 3). The consensus is essentially the same as that in E. coli with the exception of position eight, where it is most frequently a dA in P. aeruginosa and a dT in E. coli.

We searched the Pseudomonas genome (35, 39) for other potential LexA binding sites using the sequence CTGN₂TN₇CAG with up to four mismatches in the central 10-bp region. In addition to the sites that regulate the 15 genes identified in our microarray studies, eight potential LexA binding sites were identified (see Table S4 in our supporting information at http://www.scripps.edu/chem/romesberg/ ). Four are positioned between 128 and 154 bp from a gene, and four are intragenic. The microarray data suggest that either these sites do not bind LexA in vivo or that they do not effectively regulate expression. Thus, the data suggest that the 15 genes identified experimentally represent the entire LexA regulon.