Introduction

Bacterial quorum sensing (QS) regulates multiple cooperative behaviors like production of extracellular proteases, siderophores, exopolysaccharides, antibiotics and biosurfactants (Miller and Bassler, 2001; Fuqua and Greenberg, 2002; Williams et al., 2007). The opportunistic pathogen Pseudomonas aeruginosa provides a particularly well-understood example of QS-regulated virulence gene expression (Juhas et al., 2005; Schuster and Greenberg, 2006). P. aeruginosa infects immunocompromised individuals such as those suffering from cystic fibrosis and burns. It possesses two complete QS systems, las and rhl, that each produce and respond to a distinct diffusible acyl-homoserine lactone (acyl-HSL) signal molecule. The las system consists of the signal synthase LasI that produces 3-oxo-dodecanoyl-HSL (3OC12-HSL), and the cognate transcriptional regulator LasR. The rhl system consists of the signal synthase RhlI that produces butanoyl-HSL (C4-HSL), and the cognate transcriptional regulator RhlR. LasR and RhlR, in complex with their respective acyl-HSL signals, control the expression of over 300 genes (Hentzer et al., 2003; Schuster et al., 2003; Wagner et al., 2003). The las system controls activation of the rhl system (Latifi et al., 1996; Pesci et al., 1997), but this regulatory hierarchy is nutritionally conditional (Medina et al., 2003; Dekimpe and Deziel, 2009).

The production of QS-controlled ‘public goods’ is subject to exploitation by social cheaters that avoid the cost of cooperation but reap the benefits (Keller and Surette, 2006; West et al., 2007a). This has been demonstrated both in vitro (Diggle et al., 2007; Sandoz et al., 2007) and in vivo (Kohler et al., 2009; Rumbaugh et al., 2009). Thus, explaining the stability of cooperative behaviors has been a major challenge to evolutionary biology. A common explanation is kin selection, according to which cooperative behaviors are favored if they increase the reproductive potential of related individuals (Hamilton, 1964). In microbes, kin selection is thought to be achieved primarily through limited dispersal, which keeps related individuals together (West et al., 2007a). Pleiotropic constraints, repetitive population fragmentation and prudent metabolic regulation of public goods may also contribute to the stability of cooperation (Cremer et al., 2012; Foster et al., 2004; Kummerli and Brown, 2010; West et al., 2007b; Xavier and Foster, 2007).

It is generally assumed that QS further optimizes cooperative behaviors by restricting the expression of public goods to when they are beneficial, such as at high population density or limited diffusion. A recent study experimentally confirmed that QS may indeed provide a density-dependent fitness benefit (Darch et al., 2012). Other modeling studies and experimental work with a synthetic QS pathway generally indicated that QS-controlled public goods production is a trade off between the cost of constitutive or very early exoenzyme activation and the loss of attainable benefit from no or very late activation (Czaran and Hoekstra, 2009; Pai and You, 2009; Pai et al., 2012).

In many bacteria, the basic QS circuitry is subject to regulation by other pathways that have the potential to affect the quorum threshold (Mellbye and Schuster, 2011). In P. aeruginosa, negative regulation by two proteins, QscR and QteE, is intimately linked to the basic QS circuitry. QscR is an orphan LuxR-type protein (Chugani et al., 2001) that responds to the LasI-generated signal 3OC12-HSL and other long-chain acyl-HSLs (Lee et al., 2006). It primarily represses the expression of las and rhl-dependent genes, including lasI and rhlI, possibly by forming inactive heterodimers with LasR and RhlR (Chugani et al., 2001; Ledgham et al., 2003; Lequette et al., 2006). QscR also functions as a transcriptional activator of at least two genes (Lee et al., 2006).

QteE is a unique regulator without close homologs in other species (Siehnel et al., 2010; Liang et al., 2011). Its transcription appears to be activated by LasR (Gilbert et al., 2009), and it is thought to function analogously to the QS antiactivator TraM in Agrobacterium tumefaciens (Fuqua et al., 1995; Piper and Farrand, 2000). QteE represses the expression of several las and rhl-dependent target genes by independently reducing LasR and RhlR protein stability, probably through direct protein–protein interaction (Siehnel et al., 2010). Both QscR and QteE therefore have the capacity to modulate the quorum threshold and maximum gene expression levels through analogous mechanisms, namely their interaction with LasR and RhlR. Some of the outstanding questions concern the evolutionary purpose of these regulators: Why is it that in some bacterial species the basic QS circuitry of diffusible signal and cognate receptor is not sufficient to control target gene expression, and how does the presumed mechanism help optimize QS-controlled gene expression?

Here, we investigate the functional significance of the precise timing and level of QS-controlled public goods production under defined growth conditions, including those that favor QS. We compared QS gene expression kinetics, associated metabolic burden and fitness consequences of high, normal and no cooperation by qscR/qteE, wild-type and lasR strains, respectively, in individual and mixed culture.

Materials and methods

Bacterial strains and plasmids

Bacterial strains used in this study are listed in Table 1. Mutant strains of P. aeruginosa PAO1 were obtained from a nonredundant transposon library at the University of Washington (Jacobs et al., 2003). The insertions are in individual genes rather than in operons eliminating the possibility of polar effects, they contain a suitable antibiotic resistance marker to distinguish strains in coculture, eliminating the need for an additional genetic manipulation, and they are out-of-frame such that translational lacZ fusions are not generated. A defined qteE deletion mutant was obtained from Richard Siehnel, University of Washington (Siehnel et al., 2010).

Table 1 Bacterial strains and plasmids
Full size table

Luria broth (LB) medium (Lennox) buffered with 3-(N-morpholino)propanesulfonic acid, pH 7.0, was used for routine culturing unless specified. Antibiotics tetracycline (50 μg ml−1) and carbenicillin (200 μg ml−1) for P. aeruginosa, and ampicillin (100 μg ml−1) for Escherichia coli were used where appropriate.

Strain construction

To distinguish strains in some of the cocultures, the wild-type and the lasR mutant were tagged with a carbenicillin resistance gene using site-specific integration (Becher and Schweizer, 2000). A 1200-bp ampicillin resistance gene with its promoter was PCR-amplified from pUCP18 (Schweizer, 1991) using primers pRG12-F (5′-N6AAGCTTTTCTTAGACGTCAGGTGGCAC-3′) and pRG12-R (5′-N6CTCCAGTGTTTGCAAGCAGCAGATTACG-3′) and digested with HindIII and XhoI (restriction sites underlined). This fragment was ligated into the equally digested miniCTX3a vector (Hoang et al., 2000). The resulting construct, pRG12, was confirmed by sequencing. Plasmid pRG12 was introduced into the wild-type and the lasR mutant by electroporation and selected for integration into the P. aeruginosa genome as described (Hoang et al., 2000).

To construct the lasB'-gfp transcriptional fusion, a 240-bp lasB promoter region was PCR-amplified from the PAO1 genome using the primers pRG13-F (5′-N6AAGCTTGGCCTACAAGCTCGACGTCA-3′) and pRG13-R (5′-N6GAACTTGTTTTCGACGGTGCTTTCGT-3′). The PCR product was digested with HindIII and EcoRI (restriction sites underlined) and ligated with an equally digested promoter probe vector, pProbeAT (Miller et al., 2000). The resulting construct pRG13 was confirmed by sequencing.

Growth assays

Cultures of wild-type, qscR, qteE and lasR strains were grown in M9 minimal medium with 1% (w/v) sodium caseinate, 0.5% (w/v) casamino acids (CAA) or 0.5% (w/v) monosodium glutamate as the sole carbon (C-) sources (Sandoz et al., 2007; Wilder et al., 2011). Caseinate medium represents a condition that requires cooperation, whereas CAA and glutamate media represent conditions that do not require cooperation. Experimental cultures were started from overnight (18 h) LB cultures that had been inoculated with a freshly grown single colony of the respective strain. The starting optical density (OD600) of the experimental cultures was 0.05 (based on a 1-cm pathlength), corresponding to ∼108 colony forming units (CFU) per ml. Cells from the overnight cultures used for inoculation were not washed, as washing resulted in greatly delayed growth in caseinate medium and in poor reproducibility (Sandoz et al., 2007; Wilder et al., 2011). All cultures were grown at 37 °C with shaking. The growth of individual strains in caseinate medium was assessed in glass culture tubes containing 4 ml of medium. At 0, 12, 24, 36 and 48 h, an aliquot was removed and appropriate dilutions were plated onto LB plates to determine CFU per ml. This scheme is different from our previous studies (Sandoz et al., 2007; Wilder et al., 2011) and was chosen because the PAO1 strain (University of Washington) used here grows more slowly on caseinate medium than the PAO1 strain (Mike Vasil, University of Colorado) used previously.

To measure the growth of individual strains in CAA and glutamate media, 200-μl cultures were grown in transparent 96-well plates (Greiner bio-one, Monroe, NC, USA; Cat. No. 655185) in a multifunction plate reader (Tecan Infinite M200, San Jose, CA, USA). OD600 was measured every 15 min for up to 12 h. The reported OD600 values are directly from these microtiter plate cultures and were not converted to a 1-cm pathlength. Growth rates were quantified during exponential phase between 120 and 210 min in CAA medium and between 240 and 360 min in glutamate medium. To confirm that the presence of an antibiotic tag did not affect growth, cultures of the tagged wild-type and lasR mutant strains were compared with their respective untagged parents in a similar manner.

To verify that the presence of a transposon in the negative regulator mutants did not affect their phenotype, we compared individual growth and gene expression of the qteE transposon mutant with that of the defined qteE deletion mutant (Supplementary Figure). To confirm that the enhanced growth of the negative regulator mutants in caseinate medium was not caused by higher amounts of exoproducts present in the inoculum source (LB overnight cultures), we performed the following experiment. LB overnight cultures of the qteE mutant and the wild-type were centrifuged to separate the supernatant from the cells. The cells were washed and resuspended in either their own or in the other strain’s filter-sterilized supernatant. Specifically, qteE mutant cells were resuspended in qteE mutant or wild-type supernatant, and wild-type cells were resuspended in qteE mutant or wild-type supernatant. Individual caseinate cultures were inoculated with these cell suspensions and growth was assessed as described above.

Cocultures were initiated from LB overnight cultures such that the frequency of the rare strain was 1% based on OD600. The rare strain carried an antibiotic resistance marker. At the indicated times, culture aliquots were removed, appropriately diluted and plated to determine CFU per ml. Strains were distinguished by plating onto LB plates with and without the appropriate antibiotic. Percent enrichment was calculated by dividing the final ratio by the initial ratio.

Relative fitness was also calculated as the ratio of Malthusian growth parameters (w), which is essentially the ratio of the number of doublings by two competitor populations (Lenski et al., 1991). Here, w=ln(X1/X0)/ln(Y1/Y0), where X0 and X1 are the initial and final CFU per ml of the lasR mutant, and Y0 and Y1 are the initial and final CFU per ml of the wild-type or qteE/qscR mutants, respectively.

Statistical significance was determined by either a two-sample t-test (two-tailed, equal variance) for pairwise comparison of experimental data or by a one-sample t-test (two-tailed) for comparison of experimental data to a fixed value (that is, no enrichment in coculture).

Measurement of lasB'-gfp expression

lasB promoter activity was quantified by fluorescence spectroscopy in the wild-type, lasR, qscR and qteE strains containing plasmid pRG13. Experimental cultures were started from overnight LB cultures that had been inoculated with a freshly grown single colony. The cultures were washed once in M9 salts and inoculated into fresh LB at an OD600 of 0.002 (based on a 1-cm pathlength). The cultures were grown to an OD600 of 0.2, washed again and finally diluted to an OD600 of 0.01 in CAA and glutamate medium in 96-well plates (Greiner bio-one, Cat. No. 655090). This washing and culturing scheme was done as previously reported to reduce preinduction of lasB'-gfp in the qteE and qscR strains (Siehnel et al., 2010). Absorbance at 600 nm and GFP fluorescence at excitation and emission wavelengths of 480 and 535 nm, respectively, were measured using a multifunction plate reader (Tecan Infinite M200). The GFP protein encoded by pProbeAT is GFPmut1, a fast-folding and very stable variant (Cormack et al., 1996). Thus, fluorescence measurements at any given time represent cumulative expression levels. Background fluorescence of a strain without the reporter fusion, normalized to its OD, was subtracted from the fluorescence signal of the strain carrying the reporter fusion, also normalized to its OD, at the same time point. To determine the rate of lasB expression, we differentiated the fluorescent signal over time as described (Zaslaver et al., 2006). The change in GFP/OD within a 15-min timespan, corresponding to two successive measurements, was calculated (dGFP/dt/OD). The data were smoothed by averaging three consecutive measurements.

Results

Effect of QS dysregulation on public goods expression and growth

We first assessed the fitness characteristics of wild-type, lasR, qscR and qteE strains by comparing their culture growth in a minimal medium with either CAA or glutamate as the sole C-source. Both C-sources do not favor QS; they can be taken up and metabolized without prior processing by QS-dependent extracellular enzymes. This allowed us to investigate QS target gene expression and associated metabolic burden independent from a potential growth benefit afforded by QS. P. aeruginosa grows faster in CAA than in glutamate medium, as growth in the latter necessitates de novo synthesis of amino acids. Amino-acid starvation and concomitant reduction in growth rate prematurely activate QS gene expression via the stringent response (van Delden et al., 2001). Thus, growth with the two different C-sources permitted modulation of the quorum threshold.

In individual culture, the growth rates of the qscR and qteE mutants were very similar to those of the wild-type and the lasR mutant in CAA medium, but substantially slower in glutamate medium (Figures 1a and b and Table 2). Relative fitness of strains in coculture reflected their growth in individual culture (Figure 1c). The lasR mutant and the wild-type strain, at 1% initial frequency, invaded qteE and qscR mutant populations in glutamate but not in CAA medium.

Figure 1
figure 1

P. aeruginosa fitness under conditions that do not favor QS. Wild-type, lasR, qscR and qteE strains were grown in CAA and glutamate medium. (a) Culture density (OD600) of individual strains grown in CAA medium. (b) Culture density (OD600) of individual strains grown in glutamate medium. (c) Coculturing of different strain combinations in CAA and glutamate medium. Two-strain cultures were initiated with 1% initial frequency of the rare strain as indicated and grown for 12 h. The error bars indicate s.d. of the mean of three independent biological replicates. The asterisk indicates statistically significant difference from the initial inoculum (dashed horizontal line; P<0.05). The asterisk in parentheses indicates borderline significance (P=0.052). For the lasR cocultures, the dagger indicates statistically significant difference from the wild-type (P< 0.05). s.d., standard deviation.

Full size image
Table 2 Growth rates in CAA and glutamate media
Full size table

The observed growth characteristics of each of the strains correlated with individual differences in the timing and magnitude of QS-controlled public goods expression. We chose LasB elastase, a las and rhl-responsive extracellular virulence factor (Pearson et al., 1997) and major casein protease in P. aeruginosa (Galloway, 1991), as a relevant public good. We quantified lasB promoter activity in the wild-type, lasR, qscR and qteE strains using a gfp transcriptional reporter fusion (Figure 2). For reasons that are unclear, the growth of the qscR and qteE mutants in glutamate medium was less distinct from that of the wild-type and lasR mutant when the strains harbored the lasB'-gfp reporter plasmid (Figure 2b vs Figure 1b). We measured both the cumulative levels as well as the rates of lasB expression (see Materials and Methods for technical details). Interestingly, we obtained two distinct rate peaks (Figures 2e and f), the potential significance of which is discussed below. The qscR and qteE mutants showed higher lasB expression as compared with the wild-type and the lasR mutant in both CAA and glutamate media (Figures 2c–f). However, in CAA, the induction was observed at the transition from exponential to stationary phase (onset ∼300 min, rate peak 350 min; Figures 2a, c and e), whereas in glutamate medium it was observed earlier in growth at lower cell density (onset ∼200 min, rate peak 270 min; Figures 2b, d and f). Induction upon entry into stationary phase had little impact on bacterial growth rate, whereas induction during active growth did. The transient decrease in growth rate in glutamate medium around 200 min correlated with the sharp increase in the rate of lasB expression. The timing of lasB induction was slightly advanced in the qteE and qscR mutants compared with the wild-type (Figures 2c–f insets). Both mutants also exhibited a low basal level of lasB expression (Figures 2c and d insets). However, in contrast with the high expression rate post induction, the low expression rate early in growth had a negligible impact on fitness.

Figure 2
figure 2

P. aeruginosa lasB expression in individual cultures. Wild-type, lasR, qteE and qscR strains containing the lasB'-gfp reporter plasmid pRG13 were grown in CAA medium (left panels a, c, e) and in glutamate medium (right panels b, d, f). (a, b) Culture density (OD600). (c, d) Cumulative lasB expression (GFP/OD). (e, f) Rate of lasB expression (dGFP/dt/OD). The insets are a magnification of the gene expression early in growth. The error bars indicate s.d. of the mean of three independent biological replicates.

Full size image

Effect of QS dysregulation on bacterial fitness under conditions that favor QS

Next, we assessed P. aeruginosa growth in defined medium with caseinate as the sole C-source. Caseinate utilization requires QS-dependent extracellular proteases, including LasB (Iglewski and van Delden, 1998; Sandoz et al., 2007). We found that both qscR and qteE mutants grew substantially faster than the wild-type in caseinate medium (Figure 3a). This is to be expected if growth is limited by exoenzyme concentration. The qscR and qteE mutants highly express LasB and other exoenzymes, and therefore degrade caseinate faster than the wild-type. We verified that the observed differences in growth were not caused by the presumably different amounts of exoenzymes present in the culture inoculum (Figure 3b). The lasR mutant grew to a density about 30-fold lower than that of the other three strains (Figure 3a).

Figure 3
figure 3

P. aeruginosa fitness under conditions that favor QS. Wild-type, lasR, qteE and qscR strains were grown in caseinate medium. (a) Density (CFU per ml) of individual cultures. (b) Density (CFU per ml) of qteE and wild-type cultures when inoculated with their own supernatant and with the supernatant from the respective other seed culture. Supernatant swapping had no statistically significant effect on growth (P<0.05). (c) Coculturing of different strain combinations. Two-strain cultures were initiated with 1% initial frequency of the rare strain as indicated and grown for the time specified. The error bars indicate s.d. of the mean of three independent biological replicates. The asterisk indicates statistically significant difference from the initial inoculum (dashed horizontal line; P<0.05). For the lasR cocultures, the dagger indicates statistically significant difference from the wild-type (P<0.05).

Full size image

To investigate how a QS-deficient cheater would benefit from varied levels of public goods production, we cocultured the lasR mutant, at 1% initial frequency, with the wild-type, the qscR or the qteE mutant. Enrichment was quantified after 24 h of competition, because in an individual culture, the qscR and qteE mutants reached stationary phase at this time (Figure 3a). This allowed us to restrict competition to the period of active growth, and keep the duration of competition identical for all cocultures. We found that enrichment of the lasR cheater was significantly higher in qscR or qteE mutant coculture than in the wild-type coculture (Figure 3c). However, this measure of relative fitness based on initial and final frequencies does not consider the density dependence of cooperation (Ross-Gillespie et al., 2009). During the 24-h timespan, the qteE and qscR mutant populations reached a higher population density and underwent more rounds of replication, in principle affording cheaters increased opportunity to enrich. To address this point, we calculated the relative fitness of the lasR mutant based on the ratio of the number of doublings (Malthusian parameter) and found no difference in wild-type coculture compare with qteE or qscR coculture (wlasR vs WT=1.67±0.09; wlasR vs qteE=1.69±0.09; wlasR vs qscR=1.64±0.11; P>0.7 for both pairwise comparisons).

To investigate whether the wild-type with a normally functioning QS system can invade qscR and qteE mutant populations, we cocultured the wild-type at 1% initial frequency with either the qscR or the qteE mutant for 24 h. The wild-type enriched significantly (P<0.05; Figure 3c). Conversely, the qscR or qteE mutant, when rare, did not enrich in wild-type coculture (Figure 3c).

Discussion

We have shown that negative regulation of QS by QteE and QscR has a major impact on absolute and relative fitness in the opportunistic bacterial pathogen P. aeruginosa. Inactivation of each of the regulators had very similar effects on QS-controlled public goods expression and QS-dependent growth, suggesting that it is indeed the negative regulation of QS common to both that causes the observed behaviors rather than potential less well-characterized pleiotropic effects. Almost identical phenotypes of both mutants point to a similar—albeit not completely redundant—mechanism of action, presumably heterodimerization with LasR and RhlR (Ledgham et al., 2003; Siehnel et al., 2010).

First, we demonstrated that both regulators mainly affect the magnitude and to a lesser extent the timing (threshold) of expression of a representative public goods gene, lasB. The effect on the quorum threshold appears less pronounced than that reported previously (Chugani et al., 2001; Siehnel et al., 2010). Both previous studies used a complex rather than a defined growth medium, but they used plasmid-borne reporter constructs similar to ours. Siehnel et al. (2010) used the same qteE transposon insertion mutant and Chugani et al. (2001) used a qscR insertion mutant that, although differently constructed, was at least in the same PAO1 strain background. However, Siehnel et al. (2010) also conceded that their measurement of the cumulative expression levels of lasB and other genes could not discern the contribution of early induction to maximum expression in stationary phase. This is where our rate measurements can provide further insight. It is clear from our data that there is a major induction of lasB expression in the qteE and qscR mutants that largely overlaps with the induction kinetics seen in the wild-type and that is largely responsible for the cumulative expression levels seen in stationary phase.

It is possible that a more prominent role of QteE and QscR in controlling the QS threshold may be uncovered by simultaneous inactivation of both (and perhaps other) regulators. Nevertheless, we did observe a low basal level of lasB expression in qscR and qteE mutants at very low cell densities. This could represent truly constitutive expression, presumably caused by self-induction or ‘short-circuiting’ of the QS pathway in individual cells (Pai and You, 2009), although we cannot exclude the possibility that this is residual GFP from previous induction. We attempted to minimize this effect by diluting and pre-growing P. aeruginosa strains at low density.

Next, we showed that the magnitude and timing of gene expression correlates with absolute and relative fitness. Early induction of QS in glutamate medium, but not late induction in CAA medium, significantly slowed the growth of the qteE and qscR mutants, allowing invasion by lasR mutant and wild-type strains. The congruence between growth and gene expression kinetics implies that the response of the chosen gene, lasB, is representative for the QS regulon as a whole, even though the responses of other individual genes might differ (Chugani et al., 2001; Siehnel et al., 2010). QteE and QscR dampened the expression of QS target genes after transition to the ‘on’ state, providing homeostatic regulation. A similar mechanism balances 3OC12-HSL production in P. aeruginosa: RsaL, another QS repressor, is activated by LasR and directly represses lasI transcription (Rampioni et al., 2006; Rampioni et al., 2007). Thus, the initial positive feedback mechanism in QS, autoinduction, is tightly coupled with and immediately followed by multiple negative feedback mechanisms that confine the metabolic cost of public goods production. The latter regulatory principle is analogous to the well-established endproduct inhibition of biosynthetic pathways.

Interestingly, the expression rate graphs revealed two distinct peaks (Figures 2e and f), possibly signifying the relative contribution of las vs rhl-dependent induction. The second peak could represent rhl-dependent expression as it coincides with the lasR-independent lasB activation of the lasR mutant in glutamate medium. Additional mutant experiments would be required to substantiate these findings.

Under growth conditions that strictly favor QS, qteE and qscR mutants overproducing public goods have an absolute fitness advantage in individual culture. This is consistent with a previous study reporting enhanced virulence of a qscR mutant in an insect infection model (Chugani et al., 2001). However, the relative fitness of qteE and qscR mutants in mixed culture is low as they are not only invaded by lasR-deficient cheaters but even by the QS-proficient wild-type. On the basis of growth rate ratios, the relative fitness of the lasR mutant is actually no higher in qteE and qscR mutant coculture than it is in wild-type coculture (the lasR mutant does grow faster in qteE and qscR mutant coculture, but so do the qteE and qscR mutants themselves). This appears puzzling initially as the increased metabolic burden from public goods overproduction by the qteE and qscR mutants should afford the lasR mutant a larger growth advantage. We surmise that the maximum growth rate of the lasR mutant in qteE and qscR coculture is limited by the rate of uptake or intracellular processing of the digested protein substrate rather than by the rate of extracellular proteolysis.

More broadly, our assessment of the costs and benefits of negative regulation permits inferences about the predominant adaptive environment of P. aeruginosa. Conditions that strictly require QS, including certain infections, do not appear to be the dominant growth environment for clonal, cooperating populations of P. aeruginosa. Otherwise, negative regulation would be lost. Maintenance of negative regulation would require substantial residence in environments that favor QS less or not at all, or would require significant mixing of producing and nonproducing strains. It is evident from our experiments (Figures 1 and 2) and numerous others that P. aeruginosa is unable to restrict QS-controlled public goods expression, through appropriate co-regulation, to those nutritional environments that require QS. For example, the constraints of transmembrane signaling may simply prevent the direct sensing of a generic extracellular protein substrate for the purpose of exoprotease expression. The tuning of public goods expression by negative regulation can therefore be considered a bet-hedging strategy that optimizes cost and benefits for life in alternating environments that do and do not favor QS. This scenario is of particular relevance to opportunistic pathogens such as P. aeruginosa that are thought to maintain their ‘virulence’ factors and associated control mechanisms primarily due to advantages in nonparasitic contexts (Brown et al., 2012). Here, tuning could be important because QS-controlled products may sometimes—but not always—confer a fitness advantage by nonspecifically inhibiting competing microbes (Hibbing et al., 2010). As a bet-hedging strategy, tuning may complement cell-to-cell variability of QS gene expression in isogenic populations (Anetzberger et al., 2009; Garmyn et al., 2011; Perez and Hagen, 2010). Such phenotypic heterogeneity can be caused by random fluctuations in the levels of regulatory proteins, and may also serve to enhance population fitness in dynamic and unpredictable environments (Veening et al., 2008).

QscR may integrate additional information in natural multispecies environments; it may attenuate QS gene expression by eavesdropping on acyl-HSL signals produced by other species (Lee et al., 2006). Taken together, ‘normal’ QS with functional QscR and QteE appears to be the strategy best suited to balance absolute and relative fitness. The tuning of QS-controlled gene expression through feedback inhibition can thereby help stabilize public goods cooperation in bacteria.