Quorum sensing (QS) in a bacterial population is activated when extracellular concentration of QS signal reaches a threshold, but how this threshold is determined remains largely unknown. In this study, we report the identification and characterization of a novel anti-activator encoded by qslA in Pseudomonas aeruginosa. The null mutation of qslA elevated AHL-dependent QS and PQS signalling, increased the expression of QS-dependent genes, and enhanced the virulence factor production and pathogenicity. We further present evidence that modulation of QS by QslA is due to protein–protein interaction with LasR, which prevents LasR from binding to its target promoter. QslA also influences the threshold concentration of QS signal needed for QS activation; in the absence of qslA, QS is activated by nine times less N-3-oxo-dodecanoyl-homoserine lactone (3-oxo-C12-HSL) than that in wild type. The findings from this study depict a new mechanism that governs the QS threshold in P. aeruginosa.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The model of quorum sensing (QS) was formulated based on work done in marine bacterium Vibrio fischeri. It was discovered that bioluminescence in V. fischeri, which is activated at a critical bacterial population density (quorum), could also be activated at low population density when cell-free supernatant from high cell density cultures were added (Nealson et al., 1970). The diffusible QS signal molecule in the supernatant that is responsible for bioluminescence activation was later identified as N-3-oxohexanoyl-homoserine lactone (Eberhard et al., 1981). QS was subsequently reported in other bacterial pathogens and its generic role in cell density-dependent gene regulation was hence established (Fuqua et al., 1994; 1996; Lazdunski et al., 2004). The acyl-homoserine lactone (AHL)-dependent QS systems are widespread in Gram-negative bacteria. AHL signals are commonly synthesized by LuxI-type synthases and released extracellularly. Diffusion of QS signals back into bacterial cells occurs along with bacterial proliferation. Upon reaching a threshold concentration, the QS signal binds to cognate LuxR-type regulator to form a functional complex, which then activates the expression of QS-dependent genes including the gene encoding AHL synthase to generate a positive feedback.
Pseudomonas aeruginosa is a ubiquitous pathogen responsible for nosocomial infections as well as chronic infections in cystic fibrosis patients. Virulence during acute P. aeruginosa infection is mediated by many QS-regulated factors such as pyocyanin, elastase and exotoxin (Smith and Iglewski, 2003; Bjarnsholt and Givskov, 2007; Lesic et al., 2009), hence disruption of QS leads to decreased pathogenicity in P. aeruginosa (Rumbaugh et al., 2000). The AHL-mediated QS in P. aeruginosa is dependent on the QS signals 3-oxo-C12-HSL (N-3-oxo-dodecanoyl-homoserine lactone) and C4-HSL (N-butyryl-L-homoserine lactone) and their corresponding receptors LasR and RhlR. The Las and Rhl QS systems are organized in a hierarchy during exponential growth phase in rich medium, whereby 3-oxo-C12-HSL produced earlier activates LasR, which subsequently induces the expression of rhlI and rhlR (Pesci et al., 1997; de Kievit et al., 2002; Gilbert et al., 2009) (Fig. S1). The control of Rhl QS system by LasR is, however, dependent on growth and environmental conditions (Duan and Surette, 2007) and during stationary phase growth or growth in phosphate-limited medium, the Rhl QS is induced independent of the Las QS system (Medina et al., 2003; Dekimpe and Deziel, 2009). The P. aeruginosa quinolone signal 3,4-dihydroxy-2-heptylquinoline (PQS) signalling system is integrated with the AHL QS signalling networks. PQS and its precursor 4-hydroxy-2-heptylquinoline (HHQ) are 4-hydroxy-2-alkylquinoline (HAQ) molecules that act as PqsR (MvfR) ligands (Xiao et al., 2006). HAQ is synthesized by the enzymes encoded by pqsABCD and pqsH, which are transcriptionally regulated by LasR/3-oxo-C12-HSL complex (Gallagher et al., 2002; Gilbert et al., 2009) (Fig. S1). The Rhl QS system is in turn upregulated by HAQ-dependent PqsR activity and by PqsE (Farrow et al., 2008; Hazan et al., 2010), and it is also suppressed by the PQS signalling system depending on growth conditions (McGrath et al., 2004). As a result of the regulatory link between QS and PQS signalling systems, these two systems are sometimes co-regulated by the same factors (Venturi, 2006; Williams and Camara, 2009).
A threshold concentration of QS signal is crucial for QS activation. The rate at which QS signal accumulates extracellularly is governed by factors such as pH changes, degradation enzymes and mass transfer processes (Boyer and Wisniewski-Dye, 2009). Intrinsic factors, especially anti-activators, also play an important role in setting the signal threshold concentration. Anti-activators that inhibit QS activation by interaction with LuxR-type regulators have been identified in Agrobacterium tumefaciens. In the absence of conjugal opines, the functionality of QS signal receptor TraR, which is a LuxR-type regulator in A. tumefaciens, is inhibited by anti-activators, such as TraM, TraM2 and TrlR, through protein–protein interaction (Piper and Farrand, 2000; Chai et al., 2001; Wang et al., 2006). Among them, TrlR is a truncated version of TraR without the C-terminal DNA-binding domain and it dimerizes with TraR to form an inactive protein complex (Chai et al., 2001). TraM and TraM2, on the other hand, are Agrobacterium species-specific 102 amino acids (aa) proteins with no homology to the LuxR-type regulators (Hwang et al., 1995; Wang et al., 2006). In P. aeruginosa, at least two QS putative anti-activators have been identified, but their speculated molecular interaction with QS signal receptor remains to be determined experimentally. QscR is homologous to LasR, and was postulated to inhibit LasR and RhlR by protein–protein interaction (Ledgham et al., 2003) to delay the induction of QS signal production and moderate bacterial virulence (Chugani et al., 2001). However, QscR also has a functional DNA-binding domain and regulates the expression of a distinct set of genes when bound to 3-oxo-C12-HSL (Lee et al., 2006; Lequette et al., 2006). A more recent report has unveiled another possible anti-activator in P. aeruginosa, QteE, which is a 190 aa protein involved in controlling QS activation threshold (Liang et al., 2010; Siehnel et al., 2010). In the absence of QteE, LasR stability is increased but its transcription and translation are not affected (Siehnel et al., 2010).
Besides the putative anti-activators, QS-dependent expression of repressor RsaL is also responsible for the control of QS homeostasis. When optimal QS response is reached, a homeostatic mechanism mediated by RsaL is triggered to limit the 3-oxo-C12-HSL biosynthesis (Rampioni et al., 2007a). Deletion of rsaL results in unchecked production of 3-oxo-C12-HSL (Rampioni et al., 2006). RsaL inhibits the transcription of lasI by directly binding to the lasI promoter (Rampioni et al., 2007a,b).
In our attempt to discover novel negative regulators of QS in P. aeruginosa, we were curious whether a TraM-type anti-activator, which modulates QS by interaction with QS signal receptor, is also conserved in P. aeruginosa. Considering that anti-activators commonly share little sequence homology, we used random transposon mutagenesis approach in a specifically designed reporter strain to screen for potential regulatory proteins that governs QS threshold. Screening of transposon mutant library led to the identification of a novel regulatory protein designated as QslA. The role of QslA and its regulatory mechanisms have been characterized using genetic and biochemical approaches. The results show that QslA modulates LasR activity through protein–protein interaction and plays a crucial role in defining the QS activation threshold in P. aeruginosa.
Identification of QslA as a suppressor of rhlR expression
To identify potential regulatory proteins that might interact with LasR, we selected the promoter of rhlR, which is directly activated by LasR, for generating a transcriptional fusion with the coding region of lacZ. The resultant reporter gene plasmid PrhlR-lacZ was introduced into the parental strain PAO1 together with the plasmid pUCP-lasRHTH carrying the DNA-binding domain of LasR (designated as LasR-HTH) to generate QR1 as the parental strain for transposon mutagenesis. LasR-HTH, which enables activation of PrhlR-lacZ transcriptional expression in the absence of QS signal 3-oxo-C12-HSL (Anderson et al., 1999; Kiratisin et al., 2002), was used in this study so that the QS regulatory genes lasR and lasI, as well as their spontaneous mutants would not be selected in the genetic screen. From about 15 000 transposon insertion mutants, 20 were found to have altered PrhlR-lacZ activities (Table S1). The mutants QRM92 and QRM130, which showed substantially higher PrhlR-lacZ activities than QR1 (Fig. 1A), were chosen for further characterization in this study.
DNA sequencing analysis showed that the transposons were, respectively, inserted in the region between PA1243 and PA1244 and between PA1244 and aprX in QRM130 and QRM92 (Fig. 1B). PA1243 and PA1244 encode putative proteins of 858 aa and 113 aa, respectively, whereas aprX encodes a 414 aa protein substrate of the type I secretion system (Duong et al., 2001). Reverse transcription polymerase chain reaction (RT-PCR) analysis showed that only PA1244 and aprX transcript levels were consistently reduced or increased in both mutants (Fig. 1C), suggesting that they might be related to the change in expression pattern of rhlR in mutants QRM92 and QRM130. Further analysis showed that in trans expression of PA1244 in QR1 reduced PrhlR-lacZ activity to a basal level, but expression of PA1243 or aprX did not affect the reporter activity (Fig. 1A). These data suggest that PA1244 could be involved in negative regulation of QS. For the convenience of discussion, PA1244 was referred to as qslA hereafter according to its role in encoding a QS LasR-specific anti-activator as described below.
To verify that QslA also affects rhlR expression in wild-type strain, the qslA in-frame deletion mutant, ΔqslA, was generated using PAO1 as the parental strain and in trans expression of qslA in pDSK vector was introduced to ΔqslA to generate complemented strain ΔqslA(qslA). As expected, PrhlR-lacZ activity was increased in ΔqslA but greatly reduced in the complemented strain ΔqslA(qslA) (Fig. 1D).
Translational start site and in silico analysis of QslA
QslA was annotated as a putative protein with 113 aa in the Pseudomonas Genome Database (http://www.pseudomonas.com), but our in silico analysis predicted a translational start site at 27 bp downstream of the original annotated start site and the open reading frame would generate a 104 aa protein (FGENESB, Softberry). To address this discrepancy, the FLAG peptide coding sequence was fused to chromosomal qslA at the C-terminal to generate the strain QF1, and the fusion gene PrhlI-lacZ was introduced in wild-type PAO1, QF1 and the mutant ΔqslA. Genetic analysis showed that both PAO1 and QF1 had a similar level of lacZ expression, which was substantially lower than in the genetic background of deletion mutant ΔqslA, indicating that the resultant QslA-FLAG is functional (Fig. S2A). The purified QslA-FLAG fusion protein was analysed by Edman degradation and the results showed that the first eight amino acids of the QslA-flag fusion protein was matched to the 10th–17th amino acid of the originally annotated QslA protein (Fig. S2B). QslA protein therefore consists of 104 aa (Table S2A), with a predicted molecular weight of 11.8 kDa.
QslA homologues are found in certain Pseudomonas species, including P. aeruginosa, Pseudomonas mendocina and Pseudomonas fluorescens (Table S2B). These homologues share over 30% identity at amino acid level but none of them have been characterized previously. Domain analysis and motif search did not reveal any useful clues. Analysis of QslA showed that it has a highly helical secondary structure (61%) and the helical regions are mainly found in two regions (amino acid 51–70 and 80–96) as predicted by the PredictProtein (Rost et al., 2004) and the DNASTAR Lasergene Protean software. These putative structural features were also present in the predicted protein structure of QslA (Fig. S3A). It was also noted that QslA contains a relatively high proportion of hydrophobic amino acid residues (40/104), but hydrophobicity plot did not identify an obvious hydrophobic region (Fig. S1C).
The function of QslA was not evident from analysing its protein sequence, and this was expected for TraM-like anti-activator proteins that do not have identifiable motifs or domains (Hwang et al., 1995). Interestingly, sequence alignment of QslA with known and putative QS anti-activators QscR, QteE and TraM, revealed little sequence similarity among these proteins (Table S3C, Fig. S4A). QslA also did not resemble the predicted structures of QscR and QteE (Fig. S3C and D). In contrast, QslA predicted protein structure shared conserved alpha helix structural features with TraM and TraM2 (Fig. S2B), whereby two long alpha helices were present in these three proteins (Chen et al., 2004; 2006; Vannini et al., 2004). In addition, TraM and QslA were similar in protein size containing 102 aa and 104 aa respectively. However, superimposition of QslA predicted structure on TraM protein structure (Vannini et al., 2004) resulted in only partial overlap (Fig. S4B) and not all of the crucial residues essential for TraM interaction with TraR (Hwang et al., 1999; Swiderska et al., 2001) are present in the overlapping region. The similarities in protein size and predicted protein structure between QslA and TraM led us to postulate that QslA may function as a QS anti-activator, and its effect on QS was further investigated.
Effect of QslA on QS gene expression and signal production
Given that QslA affects Rhl QS system, which is linked to the Las QS and PQS signalling systems (Latifi et al., 1996; Pesci et al., 1999; McKnight et al., 2000), the impact of QslA on these three signalling systems was investigated. The expression of QS and PQS system signal synthase genes and their corresponding signal production were determined at different cell densities in wild-type PAO1, deletion mutant ΔqslA and its complemented strain ΔqslA(qslA). Interestingly, the results revealed two patterns of QslA influence. Similar to the effect on rhlR expression (Fig. 1D), the expression of pqsA and pqsR (Figs 2A and S5A) as well as the production of HAQ detected by the reporter strain, which are primarily PQS and HHQ (Fig. 2D), were significantly elevated in the absence of QslA. The expression of rhlI and production of C4-HSL were also increased in the deletion mutant ΔqslA at the early stage of bacterial growth (Fig. 2B and E). In addition, the rhlR and lasR expression levels were significantly higher in ΔqslA compared with wild type (Fig. S5B and C). Surprisingly, however, deletion of qslA resulted in decreased lasI expression and reduced level of 3-oxo-C12-HSL (Fig. 2C and F). Nevertheless, without exception, in trans expression of qslA in ΔqslA abolished or significantly reduced the transcriptional expression of three signal synthase genes (Fig. 2A, B and C), and production of three QS signals (Fig. 2D, E and F). The above results have established the regulatory role of QslA on three signalling systems, i.e. PQS, Las and Rhl, but the extent of influence varies depending on individual systems and the protein level of QslA.
RsaL-dependent negative feedback regulation inhibits lasI expression in ΔqslA
The unusual effect of qslA deletion on lasI expression and 3-oxo-C12-HSL production led us to investigate whether there is additional factor(s) influencing lasI expression. Considering that the expression of lasI is also regulated by RsaL, which is a lasI promoter specific repressor (Rampioni et al., 2007a), we tested whether mutation of qslA might cause a change in rsaL expression. The results showed that the rsaL transcriptional expression was significantly elevated in ΔqslA compared with its parental strain PAO1 (Fig. 3A). Thus, it is plausible that the increased level of RsaL in ΔqslA might be the cause of the decreased lasI expression in ΔqslA. To test this possibility, the single deletion mutant ΔrsaL and double deletion mutant ΔrsaLΔqslA were generated. As expected, deletion of rsaL from PAO1 substantially increased the lasI expression level (Figs 3B and 2C). In the absence of rsaL, null mutation of qslA caused an increase of about 25–40% in the transcriptional expression level of lasI than the single deletion mutant ΔqslA (Fig. 3B). As a control, the transcriptional expression level of rhlI was affected by mutation of qslA but not affected by deletion of rsaL (Fig. S6). Taken together, the above findings suggest that the increased level of RsaL in ΔqslA plays a role in negative feedback inhibition of lasI expression. However, the likelihood that lasI expression is also inhibited by other mechanism(s) in the absence of QslA cannot be ruled out at this stage, as the expression of lasI is not necessarily dependent on LasR (Duan and Surette, 2007).
QslA affects virulence factor production and bacterial virulence on Caenorhabditis elegans
The AHL-dependent QS and PQS signalling systems regulate the expression of genes encoding a range of virulence factors such as elastase, protease and pyocyanin (Latifi et al., 1995; Pesci et al., 1999). Because QslA alters QS and PQS signalling, we reasoned that the production of these virulence factors would be similarly affected. The results showed that deletion of qslA resulted in about 20–200% increase in elastase production compared with the wild-type strain PAO1 at three growth stages (Fig. 4A), and there was also a corresponding increase in expression of lasB that encodes for elastase (Fig. S7A). Protease produced in ΔqslA was about threefold and 50% more than that in wild type at OD600 = 0.5 and OD600 = 3.0, respectively, but at OD600 = 5.0, no substantial difference was detected in wild type and the mutant ΔqslA (Fig. 4B). Compared with elastase and protease, deletion of qslA caused the most substantial changes in pyocyanin production, whereby level in ΔqslA was increased to about 150–280% of the wild-type level at three growth stages from OD600 = 3.0 to 5.5 (Fig. 4C). Furthermore, expression of rhlA that is responsible for QS-dependent rhamnolipid production was also significantly induced in ΔqslA compared with wild type (Fig. S7B). Consistent with its impact on QS (Fig. 2), in trans expression of qslA in the deletion mutant ΔqslA drastically reduced the production of three virulence factors to almost undetectable levels (Fig. 4A, B and C).
Caenorhabditis elegans killing assay was also conducted to examine the effect of QslA on the pathogenicity of P. aeruginosa. The results showed that the mortality of C. elegans was significantly increased when infected with ΔqslA compared with wild type (Fig. 4D). In contrast, complementing the deletion mutant ΔqslA with a wild-type qslA significantly attenuated the pathogenesis of C. elegans (Fig. 4D).
Because the expression of QS and PQS signalling genes was enhanced in ΔqslA (Fig. 2), we speculated that the negative regulatory role exerted by QslA on QS and PQS signalling was due to QslA inhibition of LasR, as LasR controls Rhl QS and PQS signalling under the experimental conditions used in this study (Gilbert et al., 2009). Given that QslA does not contain a DNA-binding domain and that lasR expression is auto-regulated (Pesci et al., 1997), we reasoned that QslA might control lasR transcription through its post-transcriptional effect on LasR. To explore this possibility, lasR was fused with the constitutive promoter Plac in vector pUCP19 and expressed in deletion mutant ΔlasR that contains either the qslA expression construct (pDSK-qslA) or the empty vector pDSK. LasR activity was gauged by its activation of PlasI-lacZ activity. The results showed that QslA inhibited LasR activity when lasR was expressed constitutively (Fig. 5A), demonstrating that QslA inhibits LasR post-transcriptionally. QslA was also able to inhibit the LasR-HTH activity (Fig. 5A), as previously shown for QR1 (Fig. 1A).
QslA inhibits LasR by protein–protein interaction
To determine whether QslA post-transcriptionally modulates LasR activity through protein–protein interaction, QslA-FLAG fusion protein was expressed in strain PAO1 and the deletion mutant ΔlasR. As a control, untagged QslA protein was also expressed in PAO1 using the same vector as the QslA-FLAG fusion gene. Immunoprecipitates eluted from anti-FLAG affinity gel were immunoblotted using anti-LasR and anti-FLAG antibodies separately and the results showed that QslA-FLAG co-immunoprecipitated with LasR (Fig. 5B, top and middle panels lane 2). As a specificity control for anti-FLAG affinity chromatography, anti-LasR antibody did not detect any signal from the immunoprecipitates of untagged QslA (Fig. 5B, top panel lane 1), although LasR was present in the cell lysate (Fig. 5B, bottom panel lane 1). The specificity of anti-LasR antibody against LasR was confirmed by the absence of band in the immunoblot of the cell lysates from ΔlasR(qslA-FLAG) (Fig. 5B, bottom panel lane 3). The above results have demonstrated that QslA formed a heterologous protein complex with LasR in vivo.
To test whether DNA-binding ability of LasR is disrupted by QslA and to further verify the interaction between LasR and QslA, electrophoretic mobility shift assay (EMSA) was performed using purified LasR and QslA proteins (Fig. S8A). Previous study showed that LasR binds to the intergenic promoter region between lasI and rsaL (Schuster et al., 2004; Gilbert et al., 2009), so the biotinylated DNA probe of this region was generated. EMSA analysis confirmed that LasR formed DNA–protein complexes with the DNA probe (Fig. 5C, lane 3). But when QslA was added to the reaction mixture, the DNA-binding activity of LasR was disrupted in a dose-dependent manner (Fig. 5C, lanes 4–8). As a control, QslA alone could not form a complex with the DNA probe (Fig. 5C, lane 1). Interestingly, heat-treatment of QslA reduced but did not abolish the DNA-binding ability of LasR (Fig. 5C, lane 9), suggesting that either QslA is relatively heat-stable or denatured QslA remains partially active in interaction with LasR. The specificity of QslA interaction with LasR was also demonstrated using equimolar amounts of GST, which did not disrupt DNA-binding ability of LasR (Fig. S8B). To determine whether QslA could disrupt the preformed LasR-DNA complex, LasR was pre-incubated with the DNA probe before addition of QslA. As a control, QslA was pre-incubated with LasR for the same period of time before addition of the DNA probe under the same experimental conditions. The results showed that the formation of LasR-DNA complex was disrupted in a QslA concentration-dependent manner regardless of the order in which QslA was added (Fig. 5D). We further showed that QslA interaction with LasR was not affected by addition of up to 125 µM of 3-oxo-C12-HSL (Fig. S8C).
In addition, LasR/QslA interaction was analysed by sedimentation velocity. The results showed that LasR and QslA alone, respectively, gave sedimentation coefficient of 0.8 S and 2.1 S (Fig. S9A and B), while incubation of LasR with QslA gave rise to an additional peak at 5.8 S (Fig. S9C). Size-exclusion chromatography showed that QslA (11.8 kDa) was approximately 66 kDa and exists as a hexamer (Fig. S9D). LasR was previously reported to exist as a dimer (Schuster et al., 2004), hence based on these results and together with the sedimentation coefficients of LasR, QslA and the protein complex, it was postulated that three sets of LasR dimers interact with QslA hexamer at a 1:1 stoichiometry ratio. Taken together, the above results have established that QslA is a LasR anti-activator that modulates QS and PQS signalling by interacting with LasR.
3-oxo-C12-HSL sensitivity is modulated by QslA
Despite reduced 3-oxo-C12-HSL level, the qslA deletion mutant showed increased expression of QS and PQS signalling genes (Fig. 2), and enhanced QS-dependent virulence (Fig. 4). This led us to question whether the bacterial population could become more sensitive to 3-oxo-C12-HSL in the absence of QslA. To test this intriguing possibility, single deletion mutant ΔlasI and double mutant ΔlasIΔqslA were generated and the amount of exogenous 3-oxo-C12-HSL necessary for induction of QS-dependent elastase and protease production in these mutants were determined. Results showed that in the absence of qslA, less than 200 nM 3-oxo-C12-HSL was sufficient to trigger the production of elastase and protease to wild-type levels (Fig. 6A and B), which was about nine times less than the 3-oxo-C12-HSL concentration needed in ΔlasI (Fig. 6A and B).
In this study, we present evidence that a hypothetical protein PA1244, designated as QslA, is a key negative regulatory factor in the QS and PQS signalling pathway of P. aeruginosa. Null mutation of qslA resulted in enhanced expression of QS and PQS genes (Figs 1 and 2), promoted virulence factor production (Fig. 4A, B and C), and increased bacterial virulence against the animal model C. elegans (Fig. 4D). Consistent with the above findings, in trans expression of qslA in ΔqslA almost abolished the transcriptional expression of pqsA, rhlI and lasI (Fig. 2) as well as production of QS-regulated virulence factors, and drastically attenuated the pathogenicity of P. aeruginosa on C. elegans (Fig. 4). In addition, we showed that QslA inhibited QS and virulence factor production through counteracting the QS signal receptor LasR. Co-immunoprecipitation analysis indicated that QslA formed a protein–protein complex with LasR under in vivo conditions (Fig. 5B). The finding was further verified by EMSA study, which showed that QslA prevented LasR from binding to its target promoter (Fig. 5C and D). The work from this study has established that QslA is a novel anti-activator that acts by modulating QS through molecular interaction with the QS signal receptor LasR in P. aeruginosa.
The impact of QslA influence on QS gene expression in wild type is gene-dependent. The expression of rhlI was slightly upregulated in the deletion mutant ΔqslA compared with wild type (Fig. 1B), whereas QS and PQS signalling genes such as rhlR (Fig. S5C), pqsA (Fig. 1A) and pqsR (Fig. S5B) were induced in ΔqslA compared with wild type. Similarly, deletion of qslA caused about fourfold increase in transcriptional expression of the QS-activated genes lasB and rhlA (Fig. S7). It is possible that the expression level of the LasR-dependent genes is related to the LasR affinity to corresponding promoters (Schuster et al., 2004). The promoters with relatively low affinity for LasR may not be ‘saturated’ under normal conditions, and thus could direct a substantially higher increase in gene expression than those ‘saturated’ promoters when more LasR molecules become available. In addition, null mutation of QslA may affect the expression level of QS negative regulators, which could change the expression pattern of a set of QS genes by influencing the subtle balance of positive and negative regulation. For example, deletion of qslA resulted in increased level of the lacI promoter specific inhibitor RsaL (Fig. 3A), and decreased expression of lasI (Fig. 2C), whereas same deletion in the background of RsaL null mutant led to increased expression of lasI compared with the wild type (Fig. 3B). These variations reflect the complexity of QS modulation. Much remains to be done for fully understanding the sophisticated regulatory mechanisms associated with the QS signalling networks of P. aeruginosa (Fig. S1).
Identification of QslA adds a new member to the list of negative regulatory factors in the QS signalling pathway of P. aeruginosa, which includes RsaL, QscR and QteE (Ledgham et al., 2003; Rampioni et al., 2007a; Siehnel et al., 2010). The available data, albeit limited, seem to suggest different mechanisms of regulation among these regulatory factors. RsaL acts by directly binding to the lasI promoter, thus represses the lasI transcription and limits 3-oxo-C12-HSL biosynthesis (Rampioni et al., 2007a,b). The regulatory mechanism of QscR and QteE on QS also differs from that of QslA a loss of QscR or QteE resulted in advanced activation of QS gene expression (Chugani et al., 2001; Siehnel et al., 2010), which was not that clear and obvious in ΔqslA (Fig. 2C, E and F). This is not due to growth-dependent expression of qslA as RT-PCR analysis revealed its constitutive expression pattern throughout bacterial growth (Fig. S10). In addition, both QscR and QteE were shown to inhibit RhlR activity (Ledgham et al., 2003; Siehnel et al., 2010). In contrast, QslA did not display any significant inhibitory effect on the RhlR-dependent transcriptional activation of the rhlI promoter when co-expressed in Escherichia coli (data not shown). Besides, different from QteE that acts by reducing LasR protein stability (Siehnel et al., 2010), overexpression of QslA did not influence the protein level of LasR in vivo (Fig. S11). Furthermore, QscR is a homologue of LasR (29% protein identity, 47% protein similarity) with a DNA-binding domain and uses 3-oxo-C12-HSL as a cofactor to regulate a set of LasR/RhlR-independent genes (Lee et al., 2006; Lequette et al., 2006), whereas QslA shares little similarity with LasR and does not carry a DNA-binding domain.
Our data showed that QslA defines the QS threshold by governing the sensitivity of bacterial cells to QS signals. Null mutant of QslA in P. aeruginosa substantially reduced the amount of 3-oxo-C12-HSL signal molecules necessary for QS activation (Fig. 6). The increased 3-oxo-C12-HSL sensitivity (Fig. 6), however, did not seem to have obvious effect on advancing QS activation under the experimental conditions used in this study (Fig. 2C, E and F), even though the tested 3-oxo-C12-HSL concentrations were within the detected range of wild type (Pearson et al., 1995). This may not be surprising because expression of most QS-regulated genes in P. aeruginosa is not advanced by addition of exogenous QS signals (Whiteley et al., 1999; Schuster et al., 2003). The non-quorum nature of these QS-regulated genes was suggested to be due to complex promoter architecture such that expression of QS-regulated genes requires other growth-dependent factors as well as removal of inhibitors in medium (Yarwood et al., 2005; Schuster and Greenberg, 2006; 2007). Furthermore, studies have found that the timing of QS activation is dependent on multiple factors such as MvaT, GacA, RsmA, QscR and QteE (Reimmann et al., 1997; Chugani et al., 2001; Pessi et al., 2001; Diggle et al., 2002; Siehnel et al., 2010). Why would QslA-deficient mutant of P. aeruginosa become more sensitive to QS signals? This was probably due to the role of QslA in raising the ‘threshold hurdle’ for QS activation to a high level of 3-oxo-C12-HSL concentration (Fig. 7). According to this model, QslA serves as a ‘hurdle’ that inhibits and sequesters LasR/3-oxo-C12-HSL complex at 3-oxo-C12-HSL concentration below threshold level, and QS is only activated when more LasR/3-oxo-C12-HSL complex are formed when 3-oxo-C12-HSL concentration reaches above threshold level. However, in the absence or reduced level of QslA, the LasR/3-oxo-C12-HSL complex are freely available, and thus QS can be activated at a low 3-oxo-C12-HSL concentration. This model varies from that proposed for QscR whereby QS threshold is set by formation of inactive LasR/QscR (or RhlR/QscR) heterodimers that are dissociated by increasing level of 3-oxo-C12-HSL (or C4-HSL) (Ledgham et al., 2003). Because QslA and QscR affect LasR DNA-binding and dimerization, respectively, it is probable that both proteins are involved in setting QS threshold by different mechanisms, where 3-oxo-C12-HSL concentration modulates QscR inhibition of LasR (and RhlR) while QslA controls the intrinsic QS threshold, which is not affected by increasing 3-oxo-C12-HSL concentrations (Fig. 7). Thus, QslA and QscR might perform different modulatory roles in setting QS threshold.
Identification of QslA provides further insight into the sophisticated QS regulatory mechanisms in P. aeruginosa. The pathogen may recruit QslA to prevent premature QS activation at low bacterial population density by raising the QS threshold, and this anti-activation mechanism could serve an important role in ensuring that virulence factors are only produced at optimal bacterial quorum when they can overwhelm host defence responses. Moreover, the presence of QslA in P. aeruginosa and its functional analogue TraM in A. tumefaciens suggests that modulation of QS by anti-activator(s) could be a fairly conserved mechanism in Gram-negative bacteria.
Bacterial strains, plasmids, media and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were routinely maintained at 37°C in Luria–Bertani (LB) medium. For analysis, overnight starter cultures were diluted to OD600 = 0.03 in LB and cultured under the same temperature with shaking at 250 r.p.m. Basic minimal medium (BM medium) supplemented with 0.2% mannitol (Zhang et al., 2002) was used in transposon mutangenesis. Antibiotics at the following concentrations were added when necessary: gentamicin, 30 µg ml−1; kanamycin, 500 µg ml−1; tetracycline, 100 µg ml−1 for P. aeruginosa; and kanamycin, 100 µg ml−1; tetracycline, 10 µg ml−1; carbenicillin 100 µg ml−1 for E. coli. E. coli strain OP50 were cultured on nematode growth medium (NGM) agar plates containing 0.35% Bacto-Peptone, 2.9% NaCl, 1 mM CaCl2, 5 µg ml−1 cholesterol, 1 mM MgSO4, 25 mM KPO4 (pH = 6), and 1.7% Bacto-Agar as the food source for C. elegans wild-type strain Bristol N2.
Table 1. Bacterial strains and plasmids used in this study.
pDSK519 containing PA1243 under the control of Plac
pDSK519 containing aprX under the control of Plac
pDSK519 containing qslA under the control of Plac
pDSK519 containing qslA fused with FLAG peptide under the control of Plac
Protein expression vector, AmpR
pET14b containing lasR
GST fusion protein expression vector, AmpR
pGEX6P1 containing qslA
Transposon mutagenesis and identification of transposon insertion site
A mariner-based transposon carried by vector pBT20 was used for mutagenesis of P. aeruginosa strain QR1, which is a derivative of strain PAO1 containing plasmid constructs pUCP-lasRHTH and prhlR-lacZ (Table 1). E. coli S17-1 containing pBT20 was used for conjugal mating with recipient P. aeruginosa at 37°C for 6 h. BM medium agar plates supplemented with gentamicin and X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) (Biosynth) were used to select for transconjugants. Gentamicin-resistant colonies that were more or less blue than the parental strain QR1 were selected. Thermal asymmetric interlaced PCR (TAIL-PCR) was carried out using transposon specific primers and arbitrary degenerate primers to identify the transposon insertion sites as described previously (Kulasekara et al., 2005). The PCR product was then sequenced and analysed using NCBI BLAST server.
β-galactosidase activity assays
The lacZ transcriptional fusion reporter strains were constructed by amplifying the promoter regions of PlasI, PlasR, PpqsA, PpqsR, PrhlI, PrhlR, PrsaL, PlasB and PrhlA at −174 to +63 (PlasI), −345 to +72 (PlasR), −454 to +128 (PpqsA), −756 to +159 (PpqsR), −238 to +60 (PrhlI), −401 to +66 (PrhlR), −500 to +157 (PrsaL), −332 to +138 (PlasB) and −365 to +132 (PrhlA) relative to their translational start sites and ligating these promoter fragments to pME2-lacZ vector respectively. Unless otherwise stated, the cells of P. aeruginosa lacZ transcriptional fusion reporter strains were collected OD600 = 1.5 and assayed for β-galactosidase activity, each with triplicates. The results were analysed with unpaired t-test.
RNA extraction and RT-PCR analysis
Bacterial cells were collected after 5 h growth or at specific time points as indicated. Total RNA samples were purified using RNeasy miniprep kit (QIAGEN). Genomic DNA was removed by on-column treatment with DNase (QIAGEN) and recombinant DNaseI (Roche). RT-PCR was carried out using OneStep RT-PCR kit (QIAGEN) and band intensity was determined by ImageJ (http://rsbweb.nih.gov/ij/). For qslA RT-PCR in QR1, QRM92 and QRM130, reverse transcription was carried out using RT-PA1244 primer listed in Table S3.
DNA manipulation and deletion mutagenesis
The qslA in-frame deletion mutant in PAO1 was generated using pK18mobSac ligated to approximately 450 bp upstream and downstream flanking regions of qslA. The construct was introduced into PAO1 by conjugal mating using E. coli S17-1, and recombination of the plasmid with the genomic DNA resulted in an internal deletion from 1–288 bp of the qslA coding region. The mutant was confirmed by PCR analysis. For complementation, the coding region of qslA was amplified from 27 bp upstream of the translational start site to 108 bp downstream of the stop codon. The PCR product was cloned into expression vector pDSK519 and confirmed by DNA sequencing. In complementation analysis, plasmid vector pDSK519 was introduced into wild-type strain and corresponding mutants when necessary as controls.
FLAG recombinant protein purification and sequencing
FLAG peptide and glycine spacer was fused in-frame to QslA protein at the C-terminal by PCR, and the PCR product was ligated to pDSK519. The qslA-FLAG fusion plasmid was transformed into wild-type strain PAO1. Chromosomal-encoded qslA-FLAG fusion gene was constructed by ligating the qslA-FLAG PCR product to pK18mobSac and carrying out recombination as described (Kulasekara et al., 2005). Expression of the FLAG-fusion protein was confirmed by Western blot analysis. Plasmid- and chromosomal-encoded FLAG-fusion protein were purified from overnight cultures using ANTI-FLAG M2 Affinity Gel (Sigma) according to manufacturer's protocol. The purified FLAG-fusion protein was resolved by 15% SDS-PAGE, transferred onto PVDF membrane for protein sequencing by Edman degradation analysis.
Assessment of supernatant AHL and HAQ levels
Reporter strains DL1, DR1 and PP1 were, respectively, used to evaluate 3-oxo-C12-HSL, C4-HSL and HAQ levels, according to previously reported methods whereby AHL and HAQ levels were assessed using corresponding reporter strains (Pearson et al., 1995; 1997; Fletcher et al., 2007). In brief, cell-free supernatants from PAO1, ΔqslA and ΔqslA(qslA) were added separately to 1:20 diluted overnight cultures of corresponding reporter strains and grown at 37°C with shaking for 2.5 h. β-galactosidase activity was then measured. 3-oxo-C12-HSL and C4-HSL levels were determined by using standard curves while HAQ levels were shown as a percentage of the maximum wild-type level.
Analysis of virulence factor production
Elastase activity was determined by elastin-Congo red assay as previously described with minor modifications (Bjorn et al., 1979). Briefly, 500 µl of culture supernatant was added to tubes containing 1 ml of 5 mg ml−1 of elastin-Congo red (Sigma) in ECR buffer (0.1 M Tris-Cl pH 7.2, 1 mM CaCl2). Tubes were shaken at 37°C for 2 h. The unreacted elastin Congo-red was pelleted down and the supernatant was measured at OD495. Elastase activity units were determined using the equation: 1 unit of elastase activity = (OD495/OD600) × 100. Proteolytic activity assay was carried out according to the method previously described (Denkin and Nelson, 2004). Briefly, 100 µl of culture supernatants was incubated with 100 µl of 5 mg ml−1 azocasein dissolved in protease buffer (50 mM Tris-Cl pH 8.0, 0.04% NaN3) for 40 min at 37°C. Reaction was stopped by addition of 10% trichloroacetic acid to a final concentration of 6.7% and unreacted azocasein was spun down. Supernatants were added to 700 µl of 525 mM NaOH and OD442 was measured. Protease activity units were determined using the equation: 1 unit of protease activity = (OD442/OD600) × 100. Pyocyanin levels were measured in supernatants of P. aeruginosa strains as previously described (Dong et al., 2008). Pyocyanin levels were determined according to a previously published method (Essar et al., 1990).
C. elegans killing assays
Mortality rates of C. elegans when incubated with bacteria were carried out as previously described with slight modifications (Tan et al., 1999). Bacterial strains to be tested were grown overnight in LB and 300 µl was plated on 6-well 40 mm diameter NGM agar plates. The plates were incubated overnight at 37°C and then at room temperature overnight, and 15–20 adult worms were seeded on each plate. The plates were incubated at room temperature for 3 days and the alive and dead worms were scored every 24 h or otherwise indicated. A worm was considered dead when it did not respond to plate tapping. The experiment was carried out in triplicates.
Co-immunoprecipitation and Western blot analysis
Bacteria were grown to OD600 = 3.0 and cells were resuspended in lysis buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 1% Tween 20, 1 µM 3-oxo-C12-HSL). Cell lysates were obtained by sonication and incubated with ANTI-FLAG M2 Affinity Gel (Sigma) for 2 h at 4°C. Eluted proteins were resolved by 15% SDS-PAGE and transferred onto a PVDF membrane. Immunoblotting was performed using mouse monoclonal ANTI-FLAG M2 antibody (Sigma) and the LasR rabbit polyclonal antibody respectively.
Expression and purification of LasR and QslA
The lasR coding region was amplified and ligated into pET14b (Novagen) using NcoI and XhoI restriction sites to produce pET14b-lasR. Strain BL21 star (Invitrogen) transformed with pET14b-lasR was grown in 1 l of LB medium containing 2 µM 3-oxo-C12-HSL to OD600 = 0.5 before being induced with 500 µM of isopropyl-β-d-thiogalactopyranoside overnight at 18°C. Cell pellet was resuspended in 25 ml of LasR purification buffer (LRPB) (25 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM DTT, 1 mM EDTA, 10% glycerol, 0.05% Tween 20, 200 nM 3-oxo-C12-HSL) (Schuster et al., 2004) and lysed by sonication. The cell debris was removed by spinning at 26 000 g for 1 h and the cell lysates were applied to a 5 ml Hi-Trap Heparin HP affinity column (GE). Bound LasR proteins were eluted using a 100 ml linear gradient of 150–1000 mM of NaCl. Fractions containing LasR were pooled and applied to a 30 ml Mono Q column (GE) and the unbound flow-through fractions were collected and further purified using the Hi-Trap Heparin HP affinity column to obtain the fraction containing > 90% pure LasR protein.
For purification of QslA, the PCR product of qslA coding region was digested using BamHI and EcoRI and ligated into pGEX6P1 to generate the construct pGEX6P1-qslA. Strain BL21 Star carrying pGEX6P1-qslA or pGEX6P1 for purification of GST-QslA fusion protein and GST protein, respectively, were grown in 2.5 l of LB medium to OD600 = 0.5 before being induced with 200 µM of isopropyl-β-d-thiogalactopyranoside overnight at 18°C. Cell pellet was resuspended in 25 mM Tris-HCl buffer (pH 7.8) containing 150 mM NaCl and lysed by sonication. GST-QslA fusion protein and GST protein were purified using Glutathione Sepharose column chromatography. GST tag was cleaved from QslA fusion protein using PreScission Protease (GE). The protein purity was judged by SDS/PAGE to be > 99% pure.
Biotinylated DNA probe was amplified using 5′-biotin labelled primer (Table S1) to generate a 350 bp DNA fragment. LasR, QslA and GST proteins, at indicated concentrations, were incubated in DNA-binding buffer [10 mM Tris-Cl pH 7.5, 50 mM KCl, 1 mM DTT, 2.5% glycerol, 5 mM MgCl2, 10 ng µl−1 poly (dI·dC), 0.05% NP-40, 5 µM 3-oxo-C12-HSL] with 1.8 fmol of DNA probe unless otherwise stated. Loading buffer was then added to the reaction mixture and electrophoresis was conducted in native 6% Tris-borate-EDTA polyacrylamide gel at 4°C. DNA probes were detected using LightShift Chemiluminescent EMSA Kit (Pierce) according to the manufacturer's protocol.
Analytical ultracentrifugation and size-exclusion chromatography
Sedimentation velocity experiments were conducted using Beckman ProteomeLab XL-A analytical ultracentrifuge using an An-60 Ti rotor at 42 000 r.p.m. and 4°C. LasR and QslA proteins were purified in 25 mM Tris-HCl buffer (pH 7.8) containing 200 mM NaCl. Quartz cells fitted with double-sector centerpieces were loaded with 18 µM LasR or 180 µM QslA or 9 µM LasR together with 180 µM QslA. Absorption measurements were made at 280 nm at 3 min intervals. The data were analysed using the c(S) method in SEDFIT program (Schuck, 2000).
Size-exclusion chromatography was carried out using Superdex 200 HR 10/30 column (GE) and the elution profiles were calibrated using Bio-Rad gel filtration standards. The column was ran using 25 mM Tris-HCl buffer (pH 7.8) containing 150 mM NaCl at a flow rate of 1 ml min−1.
Domain and motif scans were carried out using InterProScan (http://www.ebi.ac.uk/Tools/InterProScan/). Protein homologues were identified using NCBI BlastP. Hydrophobicity plot analysis, amino acid sequence percent identity and alignment of QslA and TraM were carried out using CLC Main WorkBench (CLC Bio, Denmark). Protein structures were predicted by Hidden Markov Model (HMM)-based method (http://compbio.soe.ucsc.edu/SAM_T08/T08-query.html) and viewed using PyMol.
The authors thank Dr Martin Schuster (Oregon State University) for kindly providing the anti-LasR antibody. We also thank Dr Portia Loh and Dr Sharon Hee for assistance in protein purification and analysis. This work was financially supported by the Biomedical Research Council, Agency of Science, Technology and Research (A*STAR), Singapore.