Stationary Phase-Specific Virulence Factor Overproduction by a lasR Mutant of Pseudomonas aeruginosa

Secreted virulence factors of the human pathogen Pseudomonas aeruginosa are often under quorum sensing control. Cells lacking the quorum-sensing regulator LasR show reduced virulence factor production under typical laboratory conditions and are hypo-virulent in short-term animal infection models, yet lasR mutants are frequently associated with long-term infection in cystic fibrosis patients. Here, I show that in stationary-phase or slow-growth conditions, lasR cells continuously and strongly produce the important virulence factor pyocyanin while wild-type cells do not. Pyocyanin overproduction by lasR cells is permitted by loss of repression by RsaL, a LasR-dependent negative regulator. lasR cells also contribute pyocyanin in mixed cultures, even under “cheating” conditions where they depend on their wild-type neighbors for nutrients. Finally, some clinical P. aeruginosa isolates with lasR mutations can overproduce pyocyanin in the laboratory. These results imply that slow-growing clinical populations of lasR cells in chronic infections may contribute to virulence by producing pyocyanin under conditions where lasR + cells do not.


Introduction
Pseudomonas aeruginosa is a common opportunistic bacterial pathogen that causes human infections in a variety of clinical situations [1] but is especially important in cystic fibrosis lung infections [2]. Many of the numerous virulence factors produced by P. aeruginosa are under the control of quorum sensing, which uses diffusible autoinducer molecules as a way to monitor cell density [3]. Specific genes are thus activated when bacterial cell population density, and hence autoinducer concentration, exceeds a threshold. P. aeruginosa has at least three quorum-sensing systems, with distinct autoinducer molecules and partially overlapping regulons, that are hierarchically arranged. The Las system is the first to become activated, and it in turn stimulates additional systems known as Rhl and PQS, which additionally regulate each other ( Fig. 1) [3]. Finally, pyocyanin, a phenazine small molecule and virulence factor, acts as a terminal signaling factor in the quorum-sensing cascade [4]. Consistent with this hierarchy, inactivation of LasR, the regulatory protein of the Las quorumsensing system, has been reported to severely attenuate quorum sensing, the production of quorum-regulated factors, and virulence in typical laboratory culture and in short-term animal models [5,6,7,8,9,10,11].
Decreased quorum sensing can permit lasR mutants to become social cheaters that gain a growth advantage by utilizing quorumregulated ''public goods'' (including virulence factors) produced by nearby wild-type cells rather than producing their own [12,13]. Cheating was thus proposed as one reason [13] why lasR mutants are detected in highly chronic human infections such as those occurring in cystic fibrosis patients [14]. Consistent with this idea, lasR mutant cells outcompeted co-infected wild-type cells and lowered overall virulence in a murine burn-infection model, consistent with their being non-producing cheaters [11]. However, recent work has revealed that cells lacking LasR function can in fact accomplish quorum sensing by employing the Rhl and PQS systems [15,16]. Without activation by LasR, the quorum response is substantially delayed, but it appears to resemble the wild-type response in terms of gene expression and virulence factor synthesis [15]. Such LasR-independent virulence factor production is another potential explanation for why lasR mutants may not reduce overall virulence in long-term cystic fibrosis infections. In accord with this idea, the presence of lasR mutant cells has been correlated with disease progression and declining lung function in cystic fibrosis patients [17].
Pyocyanin is one of the most important quorum-regulated virulence factors of P. aeruginosa. It has numerous toxic effects on host tissues at such infection sites as the respiratory epithelium, where its toxicity is thought to be related to the generation of reactive oxygen species when pyocyanin is oxidized [18]. Pyocyanin is under the control of the Rhl and PQS systems and can accordingly be produced even in the absence of LasR after a delay [15]. As with the presence of lasR mutants, high levels of sputum pyocyanin have been associated with advanced infection in cystic fibrosis patients [19]. Pyocyanin also serves as an antibiotic thanks to its redox activity, can act as a terminal electron acceptor for P. aeruginosa, and is a terminal signaling molecule in the quorum-sensing cascade [4]. It is therefore useful for monitoring quorum-sensing activity in P. aeruginosa, especially given its bright blue color when oxidized.
Most previous laboratory studies of P. aeruginosa quorum sensing have observed bacteria exponentially growing in shaking culture. Under such conditions, wild-type quorum-sensing behaviors begin during late exponential phase and continue into stationary phase, while LasR-independent behaviors do not begin to appear until the onset of stationary phase, between 8 and 24 h [15]. However, most bacterial cells in nature are not growing in optimal laboratory-like conditions. Instead, pathogens often form biofilms during infection [20]. The physiology of slowly growing bacteria resembles that of bacteria growing in biofilms [21], and P. aeruginosa expresses the stationary-phase sigma factor RpoS both in clinical sputum samples [22] and in continuously fed biofilms in vitro [23]. Indeed, one reason for the treatment resistance of cells growing in biofilms is their relatively slow growth [24]. Therefore, I reasoned that slow-growing or stationary-phase cells maintained in longer-term culture might manifest phenotypes that reflect their behavior in a more physiologically relevant state. Here, I report that wild-type and lasR cells exhibit clearly distinct yet complementary stationary-phase phenotypes. Moreover, wild-type/lasR mixtures can collaborate to enact behaviors inaccessible to the individual strains.

Routine bacterial culture
Pseudomonas aeruginosa and Escherichia coli strains were routinely cultured on LB Lennox solid (1.5% agar) and liquid media at 37uC. Culture stocks were stored in 25% glycerol at -80uC, and fresh plates were grown for each experiment. The following antibiotics were used for selection/maintenance for P. aeruginosa; the maintenance concentration was used for E. coli culture: gentamycin (75/15 mg/ml) and tetracycline (75/25 mg/ml). Irgasan (25 mg/ml) was used as an E. coli-specific selective agent. P. aeruginosa strains are listed in Table 1. E. coli strains are listed in Table S1 in File S1, plasmids are listed in Table S2 in File S1, and primer sequences are listed in Table S3 in File S1.
M9 medium was based on a salt solution of 12.8 g/L NaHPO 4 ?7H 2 O, 3 g/L KH 2 PO 4 , 0.5 g/L NaCl, 1 g/L NH 4 Cl. A 5X salts stock was prepared and autoclaved. To make the final medium, the 5X stock was mixed with 2 mM MgSO 4 and 0.1 mM CaCl 2 from sterile 1M stocks, the appropriate carbon sources (sodium caseinate was added from a 10% autoclaved stock), and was adjusted to 1X with sterile H 2 O.
SCFM medium was made as described by Palmer et al. [25] and was prepared and used freshly, as it displayed a short shelf life.

Specialized culture conditions
Static cultures of P. aeruginosa were grown in 4-ml volumes in 12well microtiter plates, in 2-ml volumes in 24-well plates, or in 200ml volumes in 96-well plates. A 1% volume of stationary-phase LB starter culture (OD 600 ,2-3), adjusted to OD 600 = 1.0, was used for inoculation (hence, the starting OD 600 was equivalent to 0.01). Pure autoinducer molecules (3OC12-HSL, C4-HSL, and PQS; Sigma) were added from 100 mM stocks in DMSO, and equivalent volumes of DMSO were used for controls.

Reporter assays
Luminescent lux reporter strains were grown in static 200-mL LB cultures in 96-well microtiter plates. The cultures were inoculated at an initial OD 600 of 0.01 (from starter cultures diluted to OD 600 = 1.0) and grown at 25uC and 40% relative humidity. At daily time points, the luminescence values of the plate were read with a BioTek Synergy 2 plate reader. Data were collected using BioTek Gen5 software. At the same time points, static cultures with isogenic backgrounds, bearing an empty reporter and prepared identically, were taken, vortexed to disperse cells (effective dispersion was verified by phase-contrast microscopy) and serially diluted to obtain colony-forming unit (CFU) counts. The resulting CFU counts were then used to normalize the luminescence values of their respective reporter strains.

Pyocyanin extraction and quantification
Pyocyanin was extracted from the supernatants of liquid cultures by adding an equal volume of chloroform (CHCl 3 ) and vigorously vortexing. The lower, pyocyanin-containing organic layer was then taken and vortexed with an equal volume of 0.2 M HCl. The pink pyocyanin-containing aqueous layer was taken, and its absorbance at 520 nm (A 520 ) was read in a BioTek Synergy 2 plate reader.
For some experiments, pyocyanin was quantified directly in culture supernatants by reading the absorbance at 691 nm as previously described [19]. Sterile medium was used as a blank.

Cheating experiments
For cheating experiments, PA14 cells or mixtures of PA14 or PA14 phz cells with lasR cells were grown in liquid M9 medium (6 mL) containing 1% casein at 25uC in a tube roller. Pyocyanin was quantified as described above. The fraction of lasR cells within a mixture was determined using a lasR strain chromosomally marked with gentamycin resistance (MTC637). Cultures were serially diluted in 1X M9 salts and plated on LB or LB containing 5 mg/ml gentamycin to obtain CFU counts.

Statistical analysis
Comparisons between samples were analyzed using unpaired equal-variance two-tailed Student's t-tests. The threshold for significance was set as p,0.01.

Pyocyanin is overproduced by lasR cells in extended stationary-phase culture
To observe the behavior of stationary-phase cells over a time period of days rather than hours, as in traditional laboratory studies, I examined static liquid LB cultures of PA14 and a lasR mutant derivative growing at 25uC. Under these conditions, cells grow to a high density that then very gradually falls over the course of several days ( Fig. 1B) but do not exhibit the ''death phase'' that typically precedes long-term adaptation to stationary phase [26]. In shaking culture, wild-type cells produce noticeable pyocyanin beginning in late exponential phase [4], while lasR cells begin to produce it by 24 h of culture [15]. After 3-4 days in static culture, I unexpectedly observed strong and continuing production of pyocyanin by stationary-phase lasR cells that turned the cultures dark blue, while wild-type cells produced virtually no visible pyocyanin at any time during the experiment (Fig. 1C). This effect was strongest in LB at 25uC, but the same trend appeared in static cultures of minimal M63 medium and in a nutritional mimic of cystic fibrosis sputum [25] at both 25uC and 37uC (Fig. S1 in File S1). Therefore, the wild type and lasR mutant display distinct stationary-phase phenotypes in that lasR cells continually produce pyocyanin while wild-type cells barely produce any pyocyanin. The phenotype of the lasR mutant was not due to additional mutations accumulated during the experiment, as cells from 6day-old blue cultures displayed the same time course of pyocyanin production when inoculated into liquid LB, grown overnight at 37uC, and re-inoculated into static LB (Fig. S2 in File S1).

Stationary-phase wild-type and lasR cells express distinct quorum-regulated virulence genes
Because stationary-phase wild-type and lasR cells displayed distinct phenotypes with respect to pyocyanin production, I analyzed the expression of additional quorum-regulated genes with roles in virulence factor production. Two distinct expression patterns were apparent. The first, typified most strongly by lasB but also seen for rhlA (encoding the LasB elastase and a rhamnolipid biosynthesis protein, respectively), showed strong early expression in the wild-type but only weak expression in lasR cells (Fig. 2). The second, seen most strongly for phzA1 but also for hcnA (involved in phenazine and hydrogen cyanide synthesis, respectively), showed delayed but stronger expression by lasR mutant cells but weaker expression by the wild type (Fig. 2). These results revealed that wild-type cells were successfully performing quorum sensing, as they very strongly expressed lasB and also expressed rhlA. However, phzA1 was notable for being largely turned off in the wild-type. The lasR mutant displayed the opposite phenotype, most strongly expressing genes that were weakly expressed by the wild type. Among the sampled quorum-regulated virulence genes, the wild-type and lasR strains thus showed distinct but complementary expression profiles, and the lasR profile was characterized by strong phzA1 expression and pyocyanin production.

LasR-independent expression requires the Rhl and PQS quorum-sensing systems
Previously reported LasR-independent quorum sensing in shaking culture required the Rhl quorum sensing system [15], in accord with its position in the quorum-sensing network (Fig. 1A). I thus tested whether the Rhl and PQS systems were also required for quorum expression in stationary-phase lasR cells. Indeed, additional deletion of rhlR, encoding the RhlR regulator, in a lasR background abolished the expression of all tested genes (Fig. 2). Similarly, pyocyanin production did not occur in lasR rhlI or lasR pqsA double mutants, which are unable to produce the Rhl autoinducer N-butyryl-L-homoserine lactone (C4-HSL) or 2heptyl-4-quinolone (HHQ) and 2-heptyl-3-hydroxy-4-quinolone (PQS), respectively (Fig. 3A). Each of these double mutants could be complemented for pyocyanin production by exogenous addition of the appropriate autoinducer (C4-HSL or PQS), with stronger induction at 100 mM than at 10 mM (Fig. 3A). Consistent with these results, a triple lasR rhlI pqsA mutant required the addition of both autoinducers to restore pyocyanin production (Fig. 3A). Moreover, exogenous addition of PQS alone or in combination with C4-HSL to the lasR mutant accelerated pyocyanin production, while C4-HSL alone did not (Fig. 3B). This result is consistent with the idea that cellular RhlR levels are a limiting factor for LasR-independent pyocyanin production, as PQS signaling can stimulate rhlR transcription [27] and addition of constitutively expressed plasmid-borne rhlR greatly accelerated and increased pyocyanin production in a lasR mutant in shaking culture [15].
A lasR pqsH double mutant, which is unable to convert HHQ to PQS [28,29], was able to produce pyocyanin (Fig. 3C), suggesting that HHQ is itself a signaling molecule that can functionally substitute for PQS to induce pyocyanin production under stationary-phase conditions. This result contrasts with a previous report [29], but the difference may be due to the different strain background, culture media and conditions used in this work.
It has been suggested that LasR-independent quorum sensing and pyocyanin production may occur through the PhoB-mediated phosphate starvation pathway [27] or use the newly discovered signaling molecule IQS, whose synthesis requires the AmbB protein [30]. To test whether pyocyanin production by stationaryphase lasR cells required either of these proteins in addition to Rhl and PQS quorum signaling, I constructed lasR phoB and lasR ambB double mutants and assayed them for pyocyanin production in static culture. Each of the double mutants produced pyocyanin indistinguishably from the lasR mutant (Fig. S3 in File S1), showing that neither of these pathways is required for LasR-independent overproduction of pyocyanin in stationary-phase culture.

Repression by RsaL explains the different quorum profiles of wild-type and lasR cells
The weak expression by wild-type cells of genes that were strongly expressed by the lasR mutant suggested that they might be under negative regulation. Notably, phzA1 and hcnA, which displayed the strongest LasR-independent expression and the weakest expression by the wild type, are direct targets of negative regulation by RsaL, a repressor whose primary role is to provide negative homeostatic feedback to Las quorum sensing [31]. Meanwhile, lasB and rhlA, which are not under RsaL repression [31], were strongly expressed in the wild type. Because expression of rsaL is under LasR control [32], RsaL was an excellent candidate for a negative repressor that would be present in the wild type but absent in a lasR mutant. Indeed, stationary-phase rsaL expression in static culture was very strong in wild-type cells but quite weak in lasR cells (Fig. 4A). Presumably, the RsaL protein produced during the initial peak of expression in wild-type cells continues to stably bind its target DNA sequences [33], such as phzA1, in subsequent days, ensuring their continued repression.
If RsaL were responsible for repressing genes such as phzA1 in otherwise quorum-active wild-type cells in stationary phase, inactivation of rsaL in a wild-type background would relieve this repression. Consistent with this hypothesis, an rsaL mutant in static culture displayed copious pyocyanin production that began significantly earlier than in a lasR mutant (Fig. 4B), suggesting that RsaL normally blocks pyocyanin production by the wild type. Deletion of rsaL also disrupts Las homeostasis, resulting in overabundance (25 mM, compared to 2 mM in the wild type) of the Las autoinducer N-3-oxo-dodecanoyl-L-homoserine lactone (3OC12-HSL) [31]. It was thus possible that high concentrations of 3OC12-HSL abetted the early production of pyocyanin. To correct for any such effect, I constructed an rsaL lasI double mutant unable to produce 3OC12-HSL and exogenously added a low (100 nM) concentration of 3OC12-HSL at the time of inoculation. The double mutant displayed 3OC12-HSL-dependent early pyocyanin production that was even stronger than that of the rsaL mutant (Fig. 4B), confirming that stationary-phase wildtype cells are capable of pyocyanin production but that it is repressed by the presence of the RsaL repressor. Therefore, expression of a specific set of quorum-regulated genes (including phzA1 and hcnA) in lasR cells is caused by LasR-independent Rhl and PQS quorum-sensing activity in combination with deactivation of RsaL-mediated repression.
lasR cells contribute pyocyanin in mixed culture even under conditions that permit cheating A lasR mutant is a well-known example of a ''cheater''. Typical cheating experiments use defined medium containing casein as the sole carbon source [12,13,34]. Because casein utilization requires quorum-regulated extracellular proteases such LasB, whose production in early phases of growth is induced by the Las system [35], a lasR mutant fails to grow on casein medium (Fig. S4 in File S1). When a wild-type strain is grown together with a lasR mutant, the lasR mutant benefits from the casein proteolysis performed by wild-type-derived LasB without the associated costs of producing quorum-regulated factors and thereby gains an advantage [12,13]. In light of the distinct quorum-sensing profiles of stationary-phase wild-type and lasR cells, I hypothesized that lasR cells might be able to contribute quorum-regulated factors such as pyocyanin even while ''cheating'' with respect to nutrition. To test this hypothesis, I cultivated wild-type and lasR cells alone and in a 1:4 mutant-to-wild-type mixture for several days in shaking liquid M9 medium with 1% casein, a typical cheating medium. As expected, the lasR mutant alone did not grow in this medium (Fig. S4 in File S1), while the wild-type grew (Fig. S4 in File S1) and produced some pyocyanin (Fig. 5A), indicating quorum sensing. The mixture of the two strains, however, produced much more pyocyanin than the wild-type alone (Fig. 5A), suggesting that the lasR mutant was contributing to pyocyanin production. To test this idea, I grew 1:4 lasR-to-phz mixtures in which only the lasR mutant could contribute pyocyanin. Such mixtures produced only slightly less pyocyanin than mixtures with the wild-type and substantially more pyocyanin than the wild-type alone (Fig. 5A), confirming that the lasR mutant contributed the majority of pyocyanin in lasR Cells Overproduce Pyocyanin mixtures. In such mixtures, the relative lasR population increased from its initial 20% level to 30-40% after 7 days of culture (Fig. 5B), suggestive of a slight advantage gained by nutritional cheating. These results demonstrated that the lasR mutant cooperated with respect to pyocyanin production even under conditions (casein medium) that forced it to cheat with respect to LasB. Notably, pyocyanin production by the lasR cells was not detected until the third day of culture (Fig. 5A), explaining why this lasR phenotype is not seen in cheating assays when cultures are diluted every 1-2 days [12,13].

Clinical lasR isolates can overproduce pyocyanin
The overproduction of pyocyanin by stationary-phase lasR cells in monoculture and in mixtures with wild-type cells raised the interesting possibility that lasR cells may overproduce pyocyanin in clinical infections, thereby increasing virulence. A relationship between the presence of lasR cells and high pyocyanin is at least suggested by separate studies associating high sputum pyocyanin [19] and lasR cell presence [17] with P. aeruginosa disease progression in cystic fibrosis patients. Assessing the relative contributions of lasR + and lasR cells to pyocyanin or virulence factor production in actual chronic human infections is very difficult, due to the spatial and genetic complexity of human lung infections. However, as a simple test of principle, I subjected a small set of clinical cystic fibrosis isolates that were wild type or mutant for lasR (Fig. 6A) to a static culture assay and looked for pyocyanin production. The lasR + strains exhibited minimal pyocyanin production, whereas lasR mutant strains showed widely varied production ranging from minimal to very strong (Fig. 6B). This variability may reflect differences in the severity of the different lasR mutant alleles (Fig. 6A) and other mutations in the genetic backgrounds of these strains that modulate pyocyanin production. Nonetheless, these results demonstrate that some clinical lasR mutant strains have the ability to overproduce pyocyanin under stationary-phase conditions where lasR + isolates cannot.

Discussion
This work shows that the quorum response by lasR mutants in slow-growth or stationary-phase conditions is distinct from the wild-type response and is characterized by strong expression of virulence factor genes that are repressed in wild-type cells by RsaL. For example, the pattern of low pyocyanin production by wildtype and high production by lasR cells in static stationary-phase culture is a reversal of the pattern seen for cells growing exponentially in shaking culture, showing that ''typical'' lab conditions uncover only part of the full range of cell behaviors. Experiments conducted in shaking culture for 24 hours showed that lasR cells could manifest a quorum response [15], but did not reveal the distinctions between the wild-type and lasR stationaryphase phenotypes that develop after longer-term culture under slow-growth conditions. Stationary-phase phenotypes are highly relevant for bacterial physiology in natural settings, including within infective biofilms [20,21,22]. The relative metabolic inactivity of some biofilmembedded cells is one mechanism of resistance against killing by host defenses or by antibiotics [24]. Moreover, host environments like the cystic fibrosis lung contain hypoxic niches that slow bacterial growth [36,37]. It is therefore likely that many P. aeruginosa cells in long-term infections are enacting stationary-phase behaviors. In such conditions, the presence of lasR mutants within the P. aeruginosa population may permit the expression of important virulence factors such as pyocyanin that are repressed by RsaL in lasR + cells, thus expanding the range of phenotypes available to the total population. In this way, niches containing lasR cells could make a key contribution to virulence.
If repression by RsaL prevents lasR + cells from producing important virulence factors, why are mutations in rsaL not commonly isolated in clinical samples from chronic infections? One likely reason is because of the homeostatic function of RsaL in the normal quorum response. Cells lacking RsaL function display constitutive overproduction of quorum-regulated factors [38], perhaps making an rsaL cell population less competitive than wild-type cells under faster-growth conditions in the same way that wild-type cells can be cheated on by lasR cells. In contrast, a lasR mutant can be competitive under fast-growth conditions (as illustrated by its well-known cheating behavior) before overproducing a more narrowly defined set of quorum-regulated factors specifically during stationary phase. This fine tuning is made possible by a combination of three features of the quorum-sensing regulatory circuit: first, RsaL is under LasR control and thus is not produced in a lasR mutant [32]; second, RsaL has many other targets in addition to its homeostatic regulation of lasI [31]; and third, the Rhl and PQS systems, which are normally activated by LasR, can also self-activate in a lasR mutant [15,16].
The distinct contributions of lasR + and lasR cells in a mixture allows them to collaborate to produce otherwise inaccessible phenotypes. This is seen most clearly in casein medium, where the lasR + cells secrete LasB to break down casein and feed the lasR cells, and the lasR cells in turn produce high levels of pyocyanin. It is conceivable that such a division of labor, where lasR cells overproduce pyocyanin and other virulence factors, may have a role in host infection. In this scenario, slow-growing or stationaryphase lasR cells within an infecting population might continually produce pyocyanin under conditions where lasR + cells do not. Overproduction of pyocyanin by some clinical lasR isolates under stationary-phase laboratory conditions suggests that they may do likewise in an infection setting, in accord with the findings that lasR strains [17] and high sputum pyocyanin [19] are both correlated with disease progression in cystic fibrosis patients. One corollary of this idea is that treatment strategies based on strong pharmacological inhibition of LasR (thus making cells functionally null for LasR) may in fact increase pyocyanin production by lasR + cells in stationary phase.

Supporting Information
File S1 This file contains Tables S1-S3 and Figure S1-S4. Table S1, Escherichia coli strains used in this study. Table S2, Plasmids used in this study. Table S3, Primer sequences used in this study. Figure S1, Pyocyanin production in static culture. Figure S2, Pyocyanin production by lasR cells after overnight regrowth. Figure S3, LasR-independent pyocyanin production does not depend on AmbB or PhoB. Figure S4, Growth of wildtype PA14 or lasR cells in M9 medium with added casamino acids or casein. (PDF)