|Home | About | Journals | Submit | Contact Us | Français|
Bacterial persister cells constitute a small portion of a culture which is tolerant to killing by lethal doses of bactericidal antibiotics. These phenotypic variants are formed in numerous bacterial species, including those with clinical relevance like the opportunistic pathogen Pseudomonas aeruginosa. Although persisters are believed to contribute to difficulties in the treatment of many infectious diseases, the underlying mechanisms affecting persister formation are not well understood. Here we show that even though P. aeruginosa cultures have a significantly smaller fraction of multidrug-tolerant persister cells than cultures of Escherichia coli or Staphylococcus aureus, they can increase persister numbers in response to quorum-sensing-related signaling molecules. The phenazine pyocyanin (and the closely related molecule paraquat) and the acyl-homoserine lactone 3-OC12-HSL significantly increased the persister numbers in logarithmic P. aeruginosa PAO1 or PA14 cultures but not in E. coli or S. aureus cultures.
Over the last 50 years, Pseudomonas aeruginosa has emerged as a major cause of nosocomial infections in immunocompromised patients, including those relying on mechanical ventilation and suffering from neutropenia or severe burns, and is perhaps most well known as the agent primarily responsible for the decline in lung function leading to death in cystic fibrosis (CF) patients (8, 14, 16, 23, 38). P. aeruginosa possesses a multiplicity of virulence factors that are elicited upon access to susceptible individuals, including various toxins, secretion systems, siderophores, surface appendages, endotoxin (lipopolysaccharide [LPS]), alginate, and phenazines. These can generate acute toxicity/injury, leading P. aeruginosa to be labeled as the “hyena” of the bacterial world (15). Treatment of infections by P. aeruginosa is hindered by its high level of intrinsic resistance to antibiotics due primarily to a combination of the impermeable outer membrane and a number of broad-spectrum efflux pumps (50, 51). P. aeruginosa is also thought to enter into a biofilm mode of growth in CF lung infections (36, 48, 63), contributing both to pathogenicity/colonization and resistance to therapeutic intervention.
In P. aeruginosa, global regulation, mediated by at least 3 quorum-sensing (QS) systems, controls population behaviors and synthesis of the majority of these pathogenicity factors (59, 66). This bacterium possesses two N-acyl-homoserine lactone (HSL)-mediated quorum-sensing systems, las and rhl (26, 43, 46, 47), and a Pseudomonas quinolone signal (PQS) system mediated by 2-heptyl-3-hydroxy-4-quinolone (49). In the HSL-mediated systems, the HSL synthases LasI and RhlI are responsible for the synthesis of the autoinducers N-(3-oxododecanoyl)-l-HSL (3-OC12-HSL) and N-butyryl-l-HSL (C4-HSL), respectively. Expressions of the lasI and rhl genes are regulated by the transcriptional activators LasR and RhlR in response to their cognate HSL signal molecules. Among the numerous cellular and secreted virulence factors in P. aeruginosa whose biosynthesis is regulated by quorum sensing is the phenazine pyocyanin (PYO). PYO is one of the predominant phenazines secreted by P. aeruginosa in stationary phase (11) and has been shown to act as both an important virulence factor (27, 28) and a signaling molecule (11). Phenazines are redox-active secondary metabolites whose antibiotic effects on competing organisms have been of research interest for more than 50 years (28). More recently, the effect of PYO on P. aeruginosa itself has been explored (11, 52), showing that PYO acts as the final signal in the quorum-sensing system cascade and regulates at least 22 genes in P. aeruginosa (11). Phenazine synthesis is mediated by the gene products of two homologous operons, phzA1B1C1D1E1F1G1 (phz1) and phzA2B2C2D2E2F2G2 (phz2) (37); however, the expression of phz1 accounts for the majority of phenazine production (7, 64). Expression of genes for phenazine biosynthesis is regulated by several mechanisms in P. aeruginosa, including quorum-sensing systems. Whereas both the PQS and rhl systems positively regulate phz1 expression (12, 26), the orphan LuxR-type quorum-sensing regulator QscR negatively regulates the expression of phz1 and phz2 (29).
In recent years, the resistance/tolerance of biofilms to antibiotic treatment has been associated with the existence of a subpopulation of bacteria called persister cells (60, 61). Persisters were first described in 1944 by Joseph Bigger, who noted that penicillin could not completely sterilize a culture of Staphylococcus (3). This observation was generally ignored until Harris Moyed again examined persisters in the 1980s (40, 41, 55). Persisters constitute a small subpopulation of apparently nongrowing multidrug-tolerant cells present in all bacterial populations studied to date (31). In contrast to resistance caused by genetic mutations, tolerance to antibiotics by persisters occurs as phenotypic variance of the wild type (WT): reinoculation (growth) of isolated persister cells regenerates the original sensitive population (3, 21, 67), indicating that persistence is an epigenetic phenomenon. Given the potential clinical importance of multidrug-tolerant persister cells, it is surprising how little is known about them, in the cases of some of the more intractable pathogens confronting us today, and whether one or more regulatory circuits mediate switching from a normal growth mode to a persistent cell. The phenomenon of persister formation so far has been characterized mainly with the organism Escherichia coli (reviewed in reference 32), whereas less is understood regarding persister cells of P. aeruginosa. One hypothesis about the underlying mechanism for persister formation is that persister cells are phenotypic variants which are not regulated per se but result from stochastic fluctuations of toxic proteins (32). However, several genes, including genes encoding the global stationary phase-related regulators SpoT, RelA, DskA, and RpoS (42, 62), that appear to be involved in persistence in P. aeruginosa have been discovered (9, 42, 62). We undertook to investigate the formation of persister cells in P. aeruginosa cultures in a comprehensive analysis under various conditions related to cidal and static antibiotic challenge, quorum-sensing-mediated signaling, and environmental stress conditions. Here, we demonstrate that even though P. aeruginosa produces less persisters than does E. coli or S. aureus, it can increase the persister cell fraction up to 90-fold in response to 3-OC12-HSL or PYO but not to a number of tested antibiotics or general stress conditions like hyperosmotic or heat stress.
Bacterial strains and plasmids that were used or constructed in this study are summarized in Table Table11.
Bacteria were cultivated aerobically (220 rpm) at 37°C, unless otherwise indicated. LB broth (Difco) or LB agar (Difco, Remel) was used for routine culturing. Cells were typically grown by diluting overnight cultures 1:1,000 in 15 ml LB using 125-ml Erlenmeyer flasks with vented caps (Corning Inc.). Overnight cultures were prepared by inoculating 5 to 7 single colonies from the corresponding LB agar plates into 15 ml LB broth and incubating for 16 to 18 h at 37°C with shaking at 220 rpm.
Restriction digestion, ligation, transformation, and agarose gel electrophoresis were done according to standard protocols (54). Plasmids were prepared from E. coli using a QIAprep Spin miniprep kit from Qiagen according to the manufacturer's protocol. Electroporation of plasmids into P. aeruginosa cells was performed as previously described (6). Genomic DNA of P. aeruginosa was extracted using the Puregene DNA purification kit (Gentra Systems) according to the manufacturer's instructions. DNA fragments used for cloning were extracted from agarose gels using a QIAquick gel extraction kit from Qiagen. PCR products were purified using a QIAquick PCR purification kit (Qiagen) and, when cloned, sequenced to confirm the correct sequences (Agencourt Bioscience Corp., Beverly, MA). Oligonucleotide synthesis was carried out by Agencourt.
Unmarked deletions of the two redundant operons for phenazine biosynthesis, phz1 and phz2, were constructed in P. aeruginosa PA14 with a procedure based on that recently described by Dietrich et al. (11), by using the pEX18Tc suicide vector (19) for the sac-based selection method described by H. Schweizer (56). Briefly, a single deletion was first made for phzA2-phzG2, which was then used for generating the Δphz1 Δphz2 double knockout (the PA14 Δphz mutant). For the deletion of phzA2-phzG2, the flanking regions of approximately 1 kb were amplified from the chromosome of P. aeruginosa PA14 using the primer pairs phzA2-1000-s/as (upstream of phzA2; Table Table2)2) and phzG2+1000-s/as (downstream of phzG2; Table Table2).2). The resulting DNA fragments contained 21-bp overlaps and were joined by overlap extension PCR. The PCR product contained the phzA2-phzG2 deletion and was first cloned into pCR2.1TOPO (Invitrogen) and subsequently subcloned into pEX18Tc using the HindIII and BamHI restriction sites. The resulting plasmid, pEX18Tc-Δphz2, was mobilized into P. aeruginosa from E. coli S17-1 (4). Briefly, P. aeruginosa PA14 was inoculated in 10 ml LB and incubated overnight at 42°C without shaking, while the donor E. coli/pEX18Tc-Δphz2 was cultivated overnight at 37°C with shaking at 220 rpm. For matings, 300 μl of each culture were mixed, the cells collected, and the pellets spread onto the center of an LB agar plate. After 18 h of growth at 37°C, the cells were plated on Pseudomonas isolation agar (PIA; Difco) containing 100 μg/ml tetracycline to select for merodiploids. After regrowth on PIA/tetracycline, single colonies were plated on LB agar containing 5% (wt/vol) sucrose for selecting cointegrants that are sucrose resistant due to the loss of the sacB gene. These were then screened for the appropriate deletion using colony PCR. For the double deletion, the procedure was repeated starting from the amplification with the primer pairs phzA1-1000-s/as (upstream of phzA1; Table Table2)2) and phzG1+1000-s/as (downstream of phzG1; Table Table2)2) and using the single-deletion PA14 Δphz2 mutant strain.
For MIC determination for different antibiotics, two independent P. aeruginosa cultures were grown to early logarithmic phase and treated with or without 2 mM PYO prior to MIC testing according to our persister assay conditions (see below). For the non-PYO samples, the same volume of dimethyl sulfoxide (DMSO) was added. After cultivation in the presence of PYO/DMSO for 1 h 15 min (37°C, 220 rpm), the cultures were diluted 1:1 in LB prior to addition of 1.5 μl bacterial suspensions to 100 μl LB containing various concentrations of carbenicillin, piperacillin, imipenem, meropenem, ciprofloxacin, ofloxacin, tobramycin, or kanamycin. The LB used for both dilution steps contained an amount of PYO/DMSO identical to that in the cultures. The MICs of PYO, PCA, 3-OC12-HSL, or paraquat were determined accordingly, without the additional PYO/DMSO treatment. The MICs were determined after ≥24 h of incubation at 37°C.
Persisters were determined, following a modified procedure previously described (32), by exposure of logarithmic or stationary cultures to antibiotics using concentrations exceeding five times the MIC for each antibiotic. Persister numbers were determined by plating the antibiotic-treated cultures on LB agar plates and subsequent counting of CFU representing the cell numbers which survived antibiotic exposure. Logarithmic cultures of P. aeruginosa were challenged with different antibiotics after 3 to 3.5 h of cultivation, whereas logarithmic cultures of E. coli or S. aureus were treated after 3 h or 2.5 h of incubation at 37°C with shaking at 220 rpm, respectively. The difference in the times of incubation allows these bacterial species to be compared at similar growth stages. Stationary cultures were generally challenged after 18 h of cultivation. To kill all nonpersister cells, treatment with antibiotics was performed for 3.5 h. Following the challenge, the samples were washed once in ice-cold PBS or LB and diluted in the same buffer or medium before 10 μl of each sample was plated on LB agar plates for determination of the number of CFU of surviving persisters. If ciprofloxacin or ofloxacin was used for persister isolation, samples were washed twice in ice-cold PBS or LB prior to dilution and plating, to avoid the carryover effects of these antibiotics. In an initial screen for P. aeruginosa persisters, various antibiotics over different concentration ranges were used: tobramycin, ≤800 μg/ml; gentamicin, ≤400 μg/ml; carbenicillin, ≤5 mg/ml; piperacillin, ≤1 mg/ml; imipenem, ≤200 μg/ml; ciprofloxacin, ≤50 μg/ml; and ofloxacin, ≤ 150 μg/ml. Persister cells were reported to not grow (or grow very slowly) in the presence of the killing agent (reviewed in reference 32). To identify antibiotic concentrations ensuring that only drug-tolerant persister cells survived, we performed killing curves for each antibiotic. Increasing the antibiotic concentration resulted in a rapid killing of the majority of the cultures followed by a plateau of surviving drug-tolerant persister cells (Fig. (Fig.1).1). We chose concentrations at which the killing curves were well within this plateau, in which further increase of the concentration does not result in further killing. The compounds were then used as listed: 0.5 to 5 mg/ml carbenicillin, 25 to 75 μg/ml imipenem, and 10 to 25 μg/ml ofloxacin or ciprofloxacin. Antibiotics were purchased from Sigma-Aldrich, except ciprofloxacin and imipenem, which were purchased from United States Pharmacopeia (USP).
To test the effect of different agents on P. aeruginosa persister formation, we cultivated P. aeruginosa to early logarithmic phase (2 h 20 min to 2 h 25 min) and subsequently added ≤2 mM PYO (Cayman Chemical) or phenazine-1-carboxylic acid (PCA; isolated by the Novartis Natural Products Unit from a Streptomyces strain), ≤1 mM paraquat (Sigma-Aldrich), 0.5 mM 3-OC12-HSL (MP Biomedicals), ≤2.5% H2O2, ≤0.5 μg/ml ofloxacin, ≤0.12 μg/ml ciprofloxacin, or ≤32 μg/ml chloramphenicol to the samples. Besides pyocyanin, which was diluted 1:40 into the cultures, all agents were prepared as 100× stock solutions. For the mock-treated control samples, the same amount of the solvent was added (DMSO for PYO, PCA, or 3-OC12-HSL controls, 0.7% ethanol for the chloramphenicol controls, and H2O for all other compounds). After further cultivation to mid-logarithmic growth phase (an additional 1 h 10 min to 1 h 15 min), the persister fraction of the cultures was determined as described above. To ensure that the different bacterial strains were treated at similar growth stages, cultures of E. coli or S. aureus were cultivated for 1 h 20 min and subsequently treated for 1 h 10 min to 1 h 15 min.
To investigate the effect of culture supernatants of different strains on persister formation, the P. aeruginosa PAO1 and PA14 strains, the PA14 Δphz mutant, E. coli K-12, and S. aureus RN4220 were grown for 16 to 18 h at 37°C, 220 rpm, before the cells were removed by centrifugation at 37°C. The resulting supernatants were supplemented with 25 g/liter powdered LB (Difco) before sterile filtration (0.2-μm-pore diameter; Millipore) to restore depleted nutrients. As the control, the same concentration of LB was dissolved in H2O and sterile filtered. These supernatants, tested to ensure that no contaminating live cells were present and prewarmed at 37°C, were used to reculture bacterial cells collected at early logarithmic phase for 1.5 h before they were challenged with 1 mg/ml carbenicillin or piperacillin for 3.5 h.
The influence of hyperosmotic or heat stress conditions on P. aeruginosa persister formation was tested as follows. To test for hyperosmotic stress, the P. aeruginosa cells were collected at 37°C and resuspended in the same culture volume of prewarmed (37°C) LB broth or LB broth supplemented with 0.2 M or 0.5 M NaCl. Heat stress conditions were applied by splitting the cultures for cultivation at 37 and 45°C. After 1 h 20 min of stress exposure, the persister fraction of the cultures was determined as described above.
In order to characterize persister cells of P. aeruginosa, it was first determined if treatment with antibiotics of various classes left a fraction of surviving cells and how large this proportion of persisters in cultures of P. aeruginosa was. Logarithmic or stationary P. aeruginosa PAO1 or PA14 cultures were exposed to several antibiotic classes over different concentration ranges including aminoglycosides (tobramycin and gentamicin), penicillins (carbenicillin and piperacillin), carbapenems (imipenem), or quinolones (ciprofloxacin and ofloxacin) for 3.5 h before determining the fractions of surviving persister cells by plating for CFU counting. Treatment with increasing concentrations of the different antibiotics resulted in the expected biphasic killing curves, showing a rapid killing of the majority of cells using up to the MIC of each antibiotic, followed by a plateau of persister cells surviving antibiotic concentrations of several times the MICs (data not shown). Persisters were previously reported to constitute a remarkably high fraction of stationary (>1%) as well as logarithmic (~0.1%) E. coli or S. aureus (≤1%) cultures (reviewed in reference 32). In comparison, the fraction of persisters that we initially observed with P. aeruginosa cultures was significantly lower. We therefore performed a direct comparison of the persister fractions in mid-logarithmic and stationary cultures of P. aeruginosa PAO1 and PA14, E. coli K-12, and S. aureus RN4220 by treatment with up to 5 mg/ml carbenicillin or up to 25 μg/ml ciprofloxacin for 3.5 h. In all four instances, rapid killing of the majority of cells in logarithmic cultures occurred at carbenicillin concentrations of <0.5 mg/ml, followed by plateaus of cells surviving up to 5 mg/ml of this cell wall inhibitor (Fig. (Fig.1).1). The concentration range we chose for carbenicillin was therefore suitable to compare the fractions of persisters in these different bacterial species. Surprisingly, even though the P. aeruginosa cultures had higher cell numbers in the untreated control samples, they had significantly lower persister fractions than did E. coli or S. aureus (Fig. (Fig.1).1). In logarithmic-phase cultures, the portion of P. aeruginosa survivor cells (~0.04%) was almost 10-fold lower than that of E. coli (~0.3%) and at least 60-fold lower than that of S. aureus (~2.5%). Comparison of persisters in logarithmic P. aeruginosa and E. coli cultures isolated by ciprofloxacin resulted in even lower levels in P. aeruginosa cultures of ~0.0005% compared to the ~0.15% found in identically treated E. coli cultures (data not shown). One possible explanation for the higher persister numbers in the carbenicillin-treated Pseudomonas cultures is that we observed the cultures still undergoing growth during carbenicillin treatment prior to lysis (data not shown), whereas ciprofloxacin appeared to more rapidly halt the growth of Pseudomonas cultures.
We found similar species-specific differences in persister numbers in stationary cultures. In stationary E. coli cultures, up to 7% of the cells were persisters tolerant to ciprofloxacin treatment. In comparison, only 0.02 to 0.1% of drug-tolerant persister cells survived ciprofloxacin treatment in stationary P. aeruginosa cultures (Fig. (Fig.1).1). To ensure that the higher persister numbers in E. coli K-12 and S. aureus RN4220 cultures were not a phenomenon specific to these laboratory-adapted strains, we confirmed these results using clinical isolates (data not shown). Taken together, these findings somewhat unexpectedly show that P. aeruginosa forms persister fractions in logarithmic and stationary cultures which are significantly lower than those of the other tested bacteria.
As P. aeruginosa has a comparatively small fraction of persisters, we speculated that certain environmental conditions or chemical treatments might increase the persister level. P. aeruginosa PAO1 and PA14 persister formation was investigated under hyperosmotic (≤0.5 M NaCl), heat (45°C), and oxidative stress (≤2.5% H2O2) conditions. We also tested exposure to bacteriostatic or cidal antibiotics, using concentration ranges from well below to up to the MICs (≤0.5 μg/ml ofloxacin, ≤0.12 μg/ml ciprofloxacin, and ≤32 μg/ml chloramphenicol), as well as ≤2 mM of the phenazine PYO (MIC of >2 mM). While general stress or the listed antibiotics showed no significant impact on persister formation (data not shown), addition of PYO, a predominant phenazine secreted by P. aeruginosa at stationary phase, to early logarithmic cultures of P. aeruginosa PAO1 or PA14 substantially increased the proportion of cells surviving carbenicillin treatment applied in mid-log phase (Fig. (Fig.2).2). This effect was dose dependent, reaching an almost 90-fold increase over untreated cells when 2 mM PYO was used (Fig. (Fig.2A).2A). To confirm that these survivor cells indeed represented persisters, we measured the effect of 2 mM PYO over different carbenicillin concentrations. The carbenicillin killing curve of P. aeruginosa cultures grown with PYO was similar to that without PYO, showing a rapid killing of the bulk of the population at concentrations around the MIC, followed by a plateau, in which no further killing occurred when the concentration was further increased, at levels extending well above MIC values. This plateau showed the survival of persister cells at elevated numbers of CFU (Fig. (Fig.2B),2B), supporting the notion that PYO positively affects the formation of persisters in P. aeruginosa cultures. This elevation of persister numbers in Pseudomonas cultures grown in the presence of PYO was further confirmed by selecting with ciprofloxacin (see Fig. S1 in the supplemental material). Ciprofloxacin is known to efficiently kill even nongrowing Pseudomonas cells (5), indicating that the increased fractions of surviving cells are indeed multidrug-tolerant persisters and not a consequence of any slight growth reduction possibly mediated by PYO. To rule out a general effect of PYO on drug susceptibility, the impact of PYO on P. aeruginosa susceptibility to various antibiotics was also tested. Treatment of logarithmic P. aeruginosa cultures with 2 mM of the phenazine had only marginal effects on the sensitivity to 8 standard antibiotics (6 out of 8 antibiotics had either unchanged or twofold increased MICs, while the MICs of the quinolone antibiotics ciprofloxacin and ofloxacin were increased fourfold when cultures were grown in the presence of PYO [see Table S1 in the supplemental material]). These slight decreases in susceptibility are unlikely to account for the increased number of cells surviving challenges with antibiotics at well over the MICs (up to 8-fold the MIC of carbenicillin or up to 120-fold the MIC of ciprofloxacin), as used here.
The impact of PYO on persisters at stationary phase was also investigated. Although directly incubating stationary cells with PYO had no effect on persister formation (data not shown), addition of 2 mM PYO to early-logarithmic cultures resulted in P. aeruginosa stationary cultures with an ~20-fold higher proportion of cells surviving ciprofloxacin killing (Fig. (Fig.3).3). It should be noted that PYO treatment also reduced the growth of the P. aeruginosa culture, resulting in a lower density than that of nontreated cultures (Fig. (Fig.3A3A).
To determine if the increase of persister cells in logarithmic cultures in response to PYO is a general phenomenon or is specific to certain bacteria, such as P. aeruginosa, we investigated the effect of PYO addition on early-logarithmic cultures of the strains E. coli K-12 and S. aureus RN4220. Addition of PYO to early-logarithmic cultures of E. coli and S. aureus resulted in efficient killing of these bacterial species (Fig. (Fig.4),4), consistent with the fact that phenazines are a class of antibiotics (1, 2, 18, 22). Nonetheless, the total numbers of persisters that tolerated ampicillin killing in the PYO-treated samples did not increase. The killing of E. coli or S. aureus by PYO reached the plateau at 0.5 mM and higher, at which the small fractions of cells that survived the PYO treatment were also tolerant to ampicillin. This suggests that an increase in persister numbers in response to PYO is not a general phenomenon and also that other bacterial persisters are in fact tolerant to killing by PYO.
To examine whether the positive effect on persister formation in P. aeruginosa cultures is a common feature of phenazines and related molecules or more specific to PYO, we tested the effect of two structurally related compounds, paraquat and PCA, on logarithmic P. aeruginosa cultures.
It was recently demonstrated that paraquat and PYO induce partially overlapping transcriptional changes in P. aeruginosa, including the induction of genes encoding the RND efflux pump MexGHI-OpmD, as well as PA2274, a putative flavin-dependent monooxygenase, and that this induction is mediated by the transcriptional regulator SoxR (gene product of PA2273) (11, 24, 45). The addition of ≥0.2 mM paraquat (MIC of 1 mM) increased persisters in P. aeruginosa PAO1 and PA14 cultures significantly compared to untreated cultures (Fig. (Fig.5A).5A). As in the case of PYO, cultivation of logarithmic cultures of E. coli K-12 in the presence of paraquat did not affect the persister fraction (data not shown), showing that the paraquat-mediated increase may be specific for P. aeruginosa. A P. aeruginosa deletion mutant lacking the soxR gene (the PAO1 ΔsoxR mutant) produced numbers of persisters in all growth stages similar to those of the parent strain and also exhibited the paraquat-mediated increase in logarithmic culture persisters, suggesting that SoxR does not play a role in the formation of persisters in PAO1.
PCA was shown to be a second predominant phenazine excreted at similar levels by stationary PA14 cultures (~30 μM when grown in minimal medium) (12). Interestingly, in contrast to PYO and paraquat, addition of up to 2 mM of the phenazine molecule PCA (MIC of >2 mM) did not have an effect on persister fractions in PAO1 or PA14 cultures (data not shown). Hence, elevated persister numbers are not mediated by all phenazines but seem to be more specific to PYO (or paraquat). PCA, like paraquat, is structurally related to PYO and is also thought to have oxidative activity (52). The finding that PYO, but not PCA, leads to increased persister numbers in P. aeruginosa cultures indicates that the underlying mechanism is more specific than exposure to oxidative stress, which is in accordance with our observation that up to 2.5% H2O2 did not enhance persister formation (data not shown). Further, despite the close structural similarity, there seems to be a differential recognition of PYO versus PCA.
The observation that the phenazine PYO can increase the fraction of persisters in logarithmic cultures suggested a direct role for phenazines in regulating persister numbers. To test whether the loss of PYO production would alter the persister cell fraction, a Δphz1 Δphz2 double-deletion mutant, which was previously reported to exhibit a complete lack of phenazine production (11), was constructed in PA14 (the PA14 Δphz mutant). The persister levels in logarithmic, early-stationary (18 h), and late-stationary (24 to 72 h) cultures of the PA14 strain and PA14 Δphz mutant were then examined. Interestingly, the deletion strain lacking phenazine production did not differ in its fraction of persister cells in all the cultures examined (data not shown). These findings suggest that even though the phenazine PYO is able to modulate persister formation in P. aeruginosa, an additional and possibly compensatory signaling system(s) capable of controlling the persister phenotype is likely also present in this organism.
Since PYO acts as quorum-sensing signaling molecule in P. aeruginosa, we decided to further examine the effect of another QS signal, the acyl-homoserine lactone 3-OC12-HSL, on persister formation. This signal molecule is produced by LasI, which sits atop the currently proposed hierarchy of the QS circuit (10). The effect of increasing concentrations of 3-OC12-HSL (MIC of >1 mM) on logarithmic cultures of PAO1 and PA14 was determined. Similar to the effect seen with PYO, 3-OC12-HSL significantly increased persister numbers in strain PAO1 (Fig. (Fig.5B).5B). However, and in contrast to PYO, 3-OC12-HSL did not affect persister numbers in PA14 (data not shown). Hence, the signaling effects of 3-OC12-HSL on early-logarithmic cultures seem to differ between the two P. aeruginosa strains.
Since the phenazine PYO and 3-OC12-HSL, each a QS signal, increased persister numbers in logarithmic cultures in PAO1 when added exogenously, we then asked if these molecules potentiate each other to promote the persister formation. This was tested by coincubating the PAO1 cultures with 0.2 mM PYO, which is closer to the reported physiological level (11) and alone does not affect persister numbers (Fig. (Fig.2A),2A), and various amounts of 3-OC12-HSL (Fig. (Fig.6).6). It is clear that in the presence of 0.2 mM PYO, even low concentrations of 3-OC12-HSL (≤40 μM) resulted in increases of persister cells in the treated PAO1 cultures (Fig. (Fig.66).
The PYO- and 3-OC12-HSL concentrations used in the above-described experiments to trigger an increase in persisters are higher than physiological concentrations typically accumulated in stationary P. aeruginosa cultures; in the supernatant of PA14, a strain that secrets ~10-fold more PYO than PAO1, about 60 μM PYO was detected after growth in LB (11), whereas 3-OC12-HSL was shown to be excreted at ~5 μM by PAO1 (47). However, it is clear that some signaling molecules are able to potentiate each other to enhance the persister phenotype (Fig. (Fig.6).6). To address if other unidentified signaling molecule(s) in P. aeruginosa cultures would allow PYO to exert its persister-promoting activity at the physiological concentrations, spent media prepared from both wild-type PA14 and PA14 Δphz mutant stationary cultures, which were supplemented with fresh LB, were used to cultivate early-logarithmic P. aeruginosa cultures prior to antibiotic challenge. As shown in Fig. Fig.7,7, the spent media from the stationary cultures of PA14 WT induced a significant increase of the persister fraction in the PA14 cultures. In contrast, the spent growth medium derived from the phenazine-deficient mutant did not increase the persister fraction (Fig. (Fig.7).7). This suggests that physiological levels of the phenazine PYO excreted by stationary PA14 cultures may be sufficient to induce elevated persister formation in P. aeruginosa in the context of the entire secretome. Persisters in E. coli or S. aureus logarithmic cultures were not affected by the above-described spent medium prepared from PA14 (or by the spent medium prepared from stationary cultures of their own species; data not shown), consistent with the results obtained with the synthetic PYO.
Bacterial persisters comprise a small subpopulation of a culture that is tolerant to killing with lethal doses of bactericidal antibiotics (3) and are believed to contribute to the difficulties in effectively treating many infectious diseases (30, 32, 33). Although the persister theory was proposed in the 1940s, there is still an incomplete understanding of the underlying mechanisms that affect the persistence of bacterial cells to antibiotic killing. A current theory proposed for persister formation in E. coli postulates that stochastic fluctuation in the production of toxic proteins known to block cell metabolism mediates entrance of cells into a dormant state enabling the persister cell to escape corruption of cellular functions by antibiotics (32). However, in both E. coli and P. aeruginosa a number of the genes that were identified to affect the persister phenotype encode global regulators, such as DnaK, DnaJ, DksA, histone-like protein, integration host factor (17), and PhoU (34) in E. coli or SpoT, RelA, DksA (62), and RpoS (42) in P. aeruginosa, suggesting that environmental signaling plays an important role in modulating persistence.
Here we demonstrate that even though cultures of the ubiquitous environmental organism and therapeutically challenging opportunistic pathogen P. aeruginosa have significantly less persisters than cultures of E. coli or S. aureus under standard laboratory conditions, this phenotype can be affected by QS-linked molecules. The phenazine pyocyanin and the acyl-homoserine lactone 3-OC12-HSL, which are normally secreted by this bacterium at later stages of growth, as well as paraquat, significantly increased the persister numbers in logarithmic P. aeruginosa cultures. During the time we were preparing the manuscript, Kayama et al. (20) reported that overexpression of LasR in P. aeruginosa, which should lead to elevated production of 3-OC12-HSL, resulted in increased tolerance to ofloxacin. This observation is consistent with our results obtained from 3-OC12-HSL treatment. In contrast, overexpression of RhlR in the same study did not show any effect on ofloxacin tolerance.
Interestingly, testing the PA14 Δphz mutant showed that a complete lack of phenazine biosynthesis did not result in altered persister levels in logarithmic or stationary cultures. There is, however, a clear and specific response of log-phase cells to these molecules, which in various environmental or clinical niches would likely be present via production by neighboring stationary-phase cells. Our data showing that spent P. aeruginosa culture medium from wild-type in which QS molecules as a whole are present in physiological levels, but not phenazine-defective cultures, increased persister numbers support this notion and also suggest that additional molecules may be secreted that could contribute alone or together with PYO to the regulation of persister numbers in log-phase cultures. Consistent with this, PYO and 3-OC12-HSL, when added alone, required high concentrations to exert persister-promoting activities but worked synergistically at lower concentrations to increase persister fractions in log-phase cultures of strain PAO1.
Also, many phenomena involved in persister formation have been reported to show high redundancy, with prominent examples given by the toxins RelE and MazF in E. coli (32); whereas plasmid-mediated overexpression of RelE or MazF in E. coli was shown to strongly increase tolerance to antibiotics, deletion of the relE or mazF gene was reported to have no effection persister production phenotype.
It should be noted that under the conditions tested 3-OC12-HSL affected persister levels in PAO1 but not in PA14. This may be related to the fact that PAO1 produces much lower levels of PYO than PA14 (11). So in this case, it can be speculated that additional acyl-HSL would have a more profound synergistic effect with a low concentration of PYO and/or other factors affecting the persister phenotype in PAO1. Indeed, the results obtained with the spent media also suggest that additional signaling molecule(s) are likely involved in modulating persister formation in P. aeruginosa, as the relatively lower concentration of PYO that accumulates in PA14 stationary culture medium is sufficient to enhance P. aeruginosa persisters.
The observed persistence enhancement is a specific response to PYO and 3-OC12-HSL treatments and is not associated with general environmental stress, growth arresting or cidality exerted by the static or cidal antibiotics tested here, or oxidative stress. In addition, the ability of these signaling molecules to increase persistence to antibiotic killing appears to be bacteria specific. Clearly, when PYO alone, PYO in PA14 spent medium, or the spent medium of stationary E. coli or S. aureus cultures were used to treat logarithmic E. coli or S. aureus cells, no significant change to the persistence of these bacteria could be observed. It is intriguing to speculate that phenazines (alone or in synergy with other QS molecules) may not only inhibit other bacteria but also enhance persister formation to potentially protect the phenazine-producing P. aeruginosa population from complete killing by antibiotics produced by competing organisms.
Interestingly, although we did not observe a difference in the levels of persister production in cultures of the PA14 WT and the phz mutant, the spent media of the two strains differed in their effects. Since in our studies PYO increased persister numbers when added to logarithmic but not to stationary cultures, it can be speculated that once they reach later stages of growth, the cultures are unaffected. The effects of QS molecules on persister formation would then become important when growing cells are exposed to exogenous QS molecules, for example, in a setting like the CF lung, where it was recently shown that the majority of the P. aeruginosa cells analyzed were actively growing (68) and could therefore benefit from enhancing effects on persister formation by PYO or HSL molecules in their environment.
In this context it is intriguing to note that acyl-HSLs can be readily detected in the sputa of CF patients (39, 58) and that the activities of the las and rhl systems were shown to be enhanced in the presence of sputum extracts from CF patients (13). Furthermore, PYO in the sputa of Pseudomonas-infected CF patients has been detected in concentrations up to 100 μM (65). Future experiments will be required to delineate the signaling pathway that leads to the increased persister formation in response to PYO or acyl-HSL treatments and to better understand the clinical relevance of this phenomenon.
We thank Keith Poole for providing the PAO1 ΔsoxR strain. We thank E. Peter Greenberg for very helpful discussions and methodical input. We thank Steven J. Projan and Neil S. Ryder for careful reading of the manuscript and their suggestions for improvement. We thank Esther Schmitt for helpful advice on naturally produced phenazine analogues.
Published ahead of print on 22 January 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.