Sodium salicylate interferes with quorum-sensing-regulated virulence in chronic wound isolates of Pseudomonas aeruginosa in simulated wound fluid

Introduction. An important factor for delayed healing of chronic wounds is the presence of bacteria. Quorum sensing (QS), a cell density-dependent signalling system, controls the production of many virulence factors and biofilm formation in Pseudomonas aeruginosa . Aim. Inhibition by sodium salicylate (NaSa) of QS-regulated virulence expression was evaluated in QS-characterized clinical wound isolates of P. aeruginosa, cultured in serum-containing medium. Methodology. Fourteen clinical P. aeruginosa strains from chronic wounds were evaluated for the production of QS signals and virulence factors. Inhibition of QS by NaSa in P. aeruginosa clinical strains, wild-type PAO1 and QS reporter strains was evaluated using in vitro assays for the production of biofilm, pyocyanin, siderophores, alkaline protease, elastase and stapholytic protease. Results. Six clinical strains secreted several QS-associated virulence factors and signal molecules and two were negative for all factors. Sub-inhibitory concentrations of NaSa downregulated the expression of the QS-related genes lasB, rhlA and pqsA and reduced the secretion of several virulence factors in PAO1 and clinical strains cultured in serum. Compared to serum-free media, the presence of serum increased the expression of QS genes and production of siderophores and pyocyanin but decreased biofilm formation. Conclusions. Pseudomonas aeruginosa from chronic wound infections showed different virulence properties. While very few strains showed no QS activity, approximately half were highly virulent and produced QS signals, suggesting that the targeting of QS is a viable and relevant strategy for infection control. NaSa showed activity as a QS-inhibitor by lowering the virulence phenotypes and QS signals at both transcriptional and extracellular levels.


InTRoduCTIon
Chronic or hard-to-heal wounds have a significant impact on quality of life and are associated with a large health-economic burden (2-4 % of health care budgets) [1][2][3]. Between 1 and 1.5 % of the population in the industrialized world will have a hard-to-heal wound during their lifetime, and infection is an important complication. One reason for wound chronicity is the high prevalence of biofilms (in 80 % of chronic wounds) [4], which results in a drastic increase in bacterial tolerance towards antibiotics [5,6]. Poor treatment outcome together with the rapid spread of antibiotic resistance, which is estimated to cause more yearly deaths than cancer by 2050 [7], stresses the need for alternative antimicrobial treatment strategies [3,8].
Two of the most frequently isolated bacterial species in chronic wounds are Staphylococcus aureus and Pseudomonas aeruginosa, with respective prevalences ranging between 32-94 % and 17-52 % [9,10]. Particularly, P. aeruginosa has been shown to destroy host tissues and is associated with stalled and deteriorating wounds [10][11][12]. The virulence of the infecting bacteria is believed to play an important role in wound chronicity [13]. Today, it is well established that quorum sensing (QS), a cell and signal density-dependent communication system used by a wide range of microorganisms, regulates the synthesis of many virulence factors (for a recent review see Lee and Zhang [14]). Pseudomonas aeruginosa is globally regulated by at least three QS systems: two N-acyl-homoserine lactone (HSL)-mediated quorumsensing systems, las and rhl, and the Pseudomonas quinolone signal (pqs) system (2-heptyl-3-hydroxy-4-quinolone, PQS). The HSL synthases LasI and RhlI mediate the synthesis of the autoinducers N-(3-oxo-dodecanoyl)-L-homoserine lactone (3-OC 12 -HSL) and N-butanoyl-L-homoserine lactone (C 4 -HSL), respectively.
The toxic effects of P. aeruginosa are largely mediated by QS-regulated virulence factors, including pyocyanin, elastase and alkaline protease. Pyocyanin is both toxic [15] and involved in biofilm architecture [16,17]; elastase and alkaline protease degrade host tissues and are involved in iron acquisition from host proteins [18] and host immune evasion [19,20]. Other virulence-associated factors are rhamnolipids, siderophores and stapholytic proteases. The siderophore pyoverdine plays an essential role in P. aeruginosa virulence by capturing iron and regulating the production of exotoxin A and an endoprotease [21]. Rhamnolipids are surfactants with haemolytic and necrotic activities [22], which influence the structure of biofilms [23], and like stapholytic protease [24], are toxic towards a variety of micro-organisms conferring upon P. aeruginosa a competitive advantage in the colonization of wound-tissue niches. QS has been found to influence biofilm formation in vitro [25,26]. However, factors other than QS also play a role in biofilm formation since some QS-deficient clinical strains of P. aeruginosa were found to form biofilms in vivo [27] and in vitro with larger surface coverages than the fully QS-competent PAO1 [28]. All of these virulence factors, which are QS-dependent, are major contributors to the ability of this bacterium to cause disease.
Although there is evidence that P. aeruginosa clinical isolates from chronic wounds can produce QS signal molecules [13], there is less literature on the characterization of virulence factors. Since QS signals affect virulence-factor production and biofilm development [14], a wound-treatment concept based on QS inhibition would represent an attractive strategy for the prevention and/or treatment of wound infections. Salicylic acid, a metabolite of aspirin, has previously been evaluated as a quorum-sensing inhibitor (QSI) in vitro [29,30]; however, its role as a potential anti-virulence compound to treat human infectious diseases remains to our best knowledge limited to keratitis caused by P. aeruginosa [31]. The poor solubility of salicylic acid led us to investigate the potential QS inhibitory effect of the sodium salt of salicylic acid, sodium salicylate (NaSa), which is several hundred-fold more soluble in water than salicylic acid [32,33].
Screening and evaluation of QSIs are often carried out in simplified in vitro test systems based on the cultivation of bacteria in standard media, e.g. minimal medium or lysogeny broth (LB) [34][35][36]. Although of great importance in evaluating the QS pathways and potential QSIs, these standard media are far from in vivo wound conditions, which could result in the risk of limited compound efficacy when QSIs are tested in vivo and thus also clinically. A method that represents being one step closer to the evaluation of QSIs in a wound-relevant milieu is to supplement the test medium with serum to better simulate the characteristics of the wound exudate. The total protein content of chronic wound fluid has been determined to be approximately 50 % of the protein content found in serum [37,38]. Serum and its abundant serum protein albumin are known to interact with and potentially decrease the bioavailability of a wide variety of compounds [39,40]. Hence, including 50 % serum in the test medium when evaluating QSIs or when studying the phenotypic behaviour of wound pathogens in general increases the relevance of in vitro chronic wound models. Taken together, to evaluate the efficacy of potential QSI compounds in the treatment of chronic wounds, it is important to gain more knowledge on the virulence of wound clinical isolates and to use relevant media that resemble the wound bed environment.
The aim of this study was to first characterize 14 clinical P. aeruginosa strains isolated from chronic ulcers in terms of their ability to produce QS signal molecules and a range of virulence factors, and then to evaluate the effect of NaSa on the expression of QS genes and production of virulence factors in selected clinical and reference strains grown in serum-containing simulated wound fluid.

Characterization of clinical strains isolated from chronic wounds
To add to the current understanding of QS in chronic ulcers and to test the NaSa treatment effect, 14 clinical P. aeruginosa strains isolated from chronic wounds were included in the study. All clinical strains were confirmed to belong to the P. aeruginosa species using the API kit (bioMérieux SA, Marcy-l'Etoile, France) based on 20 miniature biochemical tests to determine an analytical profile index (API) with scores ≥99.5 % (data not shown). All the strains, including the reference strain PAO1, were then characterized in terms of their ability to produce a number of QS-modulated virulence factors (Table 1), QS signals and biofilm (Table 2). Four clinical strains and PAO1 were positive for all seven tested virulence factors, and two additional strains were positive for five virulence factors. In contrast, two strains were completely negative for all virulence factors evaluated. Of the 14 clinical strains, pyocyanin and siderophores were found in eight strains, stapholytic protease and rhamnolipid in seven strains, alkaline protease in six strains, and elastase and swarming ability were detected in five strains.
Regarding QS signal production, six out of 14 strains were positive for all three QS signals, nine strains and PAO1 were positive for ≥1 QS signal, while in five strains, none of the signals were detected (limits of detection: 0.39 nM, 0.39 nM and 19.5 nM for C 4 -HSL, 3-OC 12 -HSL and PQS, respectively) ( Table 2). Concerning their abilities to form biofilm on polystyrene pegs, no clear connection was observed between signal/virulence-factor production and biofilm formation. Three strains (strains 4, 5 and 6) resulted in the highest OD values (similar biofilm biomass as PAO1); the same strains also produced all three QS signals and ≥5 virulence factors. Among the strains with the lowest biofilm biomass, strains 11 and 12 produced the three QS signals, and strain 13 produced two QS signals. In contrast, strains 2 and 7 were moderate biofilm producers with low production of virulence factors.

Effect of naSa on QS modulation in PAo1
First, the QS modulatory effect of NaSa was evaluated using PAO1-based QS-reporter strains cultured in a commonly used minimal medium [34,41]. Fluorescent signals from green fluorescent protein (GFP) were acquired and normalized by OD to account for any NaSa-dependent differences in growth rate (Fig. S1, available in the online version of this article).
The highest concentration of NaSa tested was below the MIC for all strains (Table S1, available in the online version of this article). Treatment with up to 10 mM NaSa resulted in an overall downregulation of the las, rhl and pqs QS systems in a concentration-dependent manner, as measured by the three reporter strains (Fig. 1a). NaSa treatment was more effective against the las (lasB::gfp) and pqs (pqsA::gfp) systems than the rhl (rhlA::gfp) system. The highest dose of NaSa (10 mM) inhibited the expression levels of lasB by 12-fold, of pqsA by 11-fold and of rhlA by 2-fold. NaSa concentrations of 0.02 mM (for lasB), 0.08 mM (for pqsA) and 5 mM (for rhlA) resulted in a statistically significant reduction in QS expression compared to the control.
Next, to investigate the effect of NaSa on QS in serumcontaining media, a series of experiments were performed using the same reporter strains. NaSa significantly reduced the gene expression of the three QS reporter strains, both in the absence and presence of up to 50 % serum, in a concentrationdependent manner (Fig. 1b-d). In 50 % serum, 10 mM NaSa reduced the gene expression of lasB, rhlA and pqsA genes by four-, two-and three-fold, respectively ( Fig. 1b-d). In the control group (non-NaSa treated), the addition of 50 % serum resulted in an upregulation in gene expression by 1.85-fold for lasB, 1.2-fold for rhlA and 1.2-fold for pqsA compared to minimal media (no serum) ( Fig. 1b-d).
To investigate whether NaSa treatment also influences the synthesis of the actual virulence factors controlled by QS in serum-containing media, wild-type P. aeruginosa PAO1 was cultured in minimal AB medium supplemented with increasing concentrations of NaSa and serum using a Table 1. Characterization of the ability of P. aeruginosa clinical isolates from chronic wounds to produce virulence factors. All assays were agar-plate-based except for rhamnolipid production, which was evaluated in a glycerol-based liquid medium. Serum-supplemented agar was used for characterization of siderophore production checkerboard approach. NaSa did not affect the growth rate of PAO1 when grown in 50 % serum (Fig. 2a). In minimal media, NaSa reduced the growth rate (Fig. S2, available in the online version of this article), although after 24 h incubation, there was no difference in c.f.u. counts (approx. 2×10 9 c.f.u. ml −1 ) between the bacterial cultures taken from the four extreme conditions (NaSa/serum: +/-, -/+, +/+, -/-) represented by the four corners of the checkerboard assay (Fig. 2b).   NaSa treatment of PAO1 in medium supplemented with 50 % serum resulted in a linear decrease in the production of pyocyanin (five-fold decrease for 10 mM NaSa, R 2 =0.82, P<0.05) and siderophores (2.7-fold decrease for 10 mM NaSa, R 2 =0.99, P<0.001) compared to the control ( Fig. 2b-d).
The lowest concentration of NaSa tested (1.25 mM) in 50 % serum was enough to significantly reduce pyocyanin production compared to the control, while 5 mM NaSa was required to significantly reduce siderophore production.
In control wells (not NaSa treated), increasing concentrations of serum resulted in a linear increase of siderophore (R 2 =0.99, P<0.01) and pyocyanin (R 2 =0.93, P<0.01) production ( In the majority of the strains, 11 out of 14, serum-supplemented culture medium resulted in a significant decrease in biofilm formation on polystyrene surfaces after 24 h (Fig. S4c), with an average reduction of 59 % in biofilm biomass.

Effect of naSa on selected virulence factors in strains isolated from wounds
To investigate the effect of NaSa treatment on clinical isolates, a subset of three highly virulent strains (1, 4 and 12), positive for all investigated QS signals and virulence factors, were treated with 0, 2 or 10 mM NaSa in 50% serum-supplemented medium. These concentrations were chosen based on pyocyanin inhibition in pilot experiments (data not shown). While 2 mM NaSa significantly reduced pyocyanin production in Error bars represent ±sd. * Indicates a statistically significant difference compared to the untreated 0 mM NaSa group (in the 0 and 50% serum groups, respectively) with a P-value <0.05 based on one-way ANOVA Dunnett's post hoc test.

dISCuSSIon
Although some studies include clinical strains, the vast majority of research on QS in P. aeruginosa related to clinical issues is carried out using laboratory strains and various reporter strains [42][43][44][45][46][47]. As suggested by Soukarieh et al. [48], two important drawbacks for the further clinical development of QSIs are uncertainty in regard to anti-virulence profiles and efficacy towards P. aeruginosa clinical isolates, as well as the lack of standardization of methods used to assess the compounds. A recent study by Cornforth et al. [49] further highlights the need of proper evaluation and evidence-based selection of infection models. The authors propose a quantitative framework for model evaluation based on comparing transcriptome data acquired from a target system, such as clinical samples, resulting in an accuracy score [49].
The literature on P. aeruginosa QS from a clinical perspective has largely focused on cystic fibrosis [44,47,50,51], whereas studies regarding the QS of clinical P. aeruginosa isolates from chronic wounds are limited. To increase the current knowledge, this study included 14 clinical strains of P. aeruginosa isolated from patients suffering from infected chronic ulcers and evaluated their ability to produce QS signals and selected virulence factors. PAO1 was included as a virulent reference strain. Although the strain collection was relatively small, this study demonstrates that six out of 14 strains showed a highly virulent phenotype, being positive for at least five out of seven investigated virulence factors. These results are in line with a previous study on chronic wounds where 56 % of the isolates produced two of three investigated virulence factors [46]. In addition, eight of the 14 strains produced at least two QS signals. This high prevalence of signal production was accompanied by increased virulence, as five of the six highly virulent strains were positive for all three QS signals, suggesting that an anti-QS strategy could be a viable treatment option for P. aeruginosa strains that cause wound infections.
According to the methodology used in this study, some of the clinical strains do not appear to produce any QS signals but still produce virulence factors (strains 8 and 14), while others produced two QS signals but only ≤1 virulence factor (strains 9 and 13). However, the apparent absence of signals in some strains does not necessarily mean that the strains are signal negative. Signal detection could be influenced by methodological limitations regarding signal stability, sampling time, and the limit of detection of the analytical technique. In fact, the lactone ring of AHL signals undergoes lactonolysis in a pH-dependent manner [52], resulting in signal inactivation. Pyocyanin production for PAO1 was evaluated after 48 h due to low production after 24 h. Biofilm formation was evaluated in 50 % serum in saline. Data are expressed as % activity or production relative to untreated PAO1 (100 % correspond to increase in fluorescence intensity per min of 9333 for elastase, decrease in OD per min of 0.0012 for stapholysis and to OD readings of 0.53, 0.42, 0.73 and 2.0 for pyocyanin, alkaline protease, siderophores and biofilm, respectively). Error bars represent ±sd, N=3. * Indicates a statistically significant difference compared to the untreated 0 mM NaSa group with a P-value <0.05 using one-way ANOVA Dunnett's post hoc test.
Furthermore, the presence of serum could inactivate as well as degrade signals [53,54], making the addition of serum relevant to evaluate the potential effect of compounds. Moreover, sampling time also matters [55] and should thus be taken into account. In pilot experiments, different timepoints were evaluated to choose the time-point with the highest detection value for some of the factors studied (data not shown). Furthermore, the observation that some of the strains were positive for QS signals but negative for virulencefactor production could be due to mutations in QS receptors.
In particular, the LasR receptor is prone to mutation [56], which has been demonstrated in clinical isolates from cystic fibrosis [47] and intubated patients [45]. Hence, non-virulent strains could during establishment of an infection have been virulent, but at some point lost this phenotype. Similarly, strains negative in signal and virulence-factor production in vitro could potentially be virulent clinically in close proximity of QS signal-producing strains, since QS signals can be shared between strains [57]. Considering all the above aspects, the in vivo pathogenicity of P. aeruginosa is multifactorial and dependent on the infectious dose, virulence characteristics of the strains, the ability to produce, sense and respond to QS signals, as well as the host immune response. In this study, any potential relationship between the strains' virulence and the wound healing outcome could not be studied, since clinical outcome data on the patients were not available.
As further confirmed in this study, P. aeruginosa in chronic wounds often expresses QS signals and virulence factors [13,46], which to a large extent are QS-regulated [14]. Inhibition of QS has been proposed as an alternative treatment strategy for chronic infections [58] and has generated much research interest in the last decades, especially in light of the increasing global threat of antibiotic resistance. Treatment with a QSI could result in reduced expression of virulence factors, aiding the host immune system to clear the infecting pathogen. NaSa is the sodium salt of salicylic acid, a metabolite of the anti-inflammatory and analgesic drug aspirin (acetylsalicylic acid; ASA). This study reports that NaSa reduces QS gene expression in a physiologically relevant milieu, resulting in attenuated secretion of several virulence factors, which could be beneficial clinically. NaSa partially inhibited all three QS systems las, rhl and pqs at the gene expression level in a concentration-dependent manner, using reporter strains for lasB, rhlA and pqsA grown in medium with and without 50 % serum. Furthermore, a concentrationdependent effect could be demonstrated for the production of virulence factors, as NaSa significantly reduced the secretion of pyocyanin, siderophores, alkaline protease and elastase. A few previous in vitro studies have demonstrated the effect of ASA and salicylic acid on QS and/or the production of virulence factors in both Gram-negative and Gram-positive bacteria, e.g. those concerning P. aeruginosa [29,30,59,60], S. aureus [61][62][63] and S. epidermidis [64]. Interestingly, a clinical study showed that orally administered aspirin reduced the healing time of wounds associated with chronic venous insufficiency, although the mechanism of action was not elucidated [65]. Since NaSa is closely related to ASA and salicylic acid, and because ASA is readily deacetylated in vivo [66], it is likely that NaSa could have similar effects. The in vitro studies performed on P. aeruginosa [29,30,59,60] are all based on laboratory reference strains and two corneal isolates cultured in minimal medium or LB, highlighting the need and relevance of proving treatment efficacy with multiple clinical isolates and serum-supplemented culture conditions to closer resemble the wound environment. Mimicking the conditions where the treatment will be applied, e.g. the wound environment, potentially increases the chance of identifying active compounds with greater potential to be effective in vivo and, ultimately, in a clinical setting.
To further evaluate the potential of NaSa as a QSI, its effect was also assessed in three highly virulent clinical strains grown in serum-supplemented media. NaSa decreased the production of four of the five virulence factors tested, indicating that the substance is also effective on clinical P. aeruginosa strains. NaSa did not significantly affect the stapholytic activity, except for strain 1. Further, NaSa did not display any effect on the amount of biofilm biomass formed by any of the clinical wound isolates (only three shown here) or the PAO1 strain, when biofilms were grown on the surface of the Calgary biofilm device under dynamic conditions (Fig. 3f).
A drawback of the widely used microtitre plate assay for biofilm quantification using crystal violet is that bacterial aggregates formed in the bulk solution are not considered [67], and it could still be possible that NaSa may affect biofilm aggregation in suspension. It is well established that biofilm is formed in chronic wounds [4,68] and that aggregated bacteria are abundant in the bulk solution of bacterial cultures [69,70]. Further studies in our lab will evaluate the effect of NaSa in a three-dimensional biofilm model [67], using confocal laser scanning microscopy and labelling of biofilm matrix components, to evaluate how biofilm composition and architecture may be influenced by NaSa.
Previously, it was shown that ASA decreased biofilm formation on polystyrene in serum-free medium, although a high concentration of 34 mM was required [59]. Similarly, medium with 0.11 mM ASA was shown to reduce the total biofilm biomass produced in a biofilm flow-chamber system using minimal media supplemented with glucose [30]. However, since this is based on subjecting the bacterial cells to a constant flow of ASA-containing medium, it is difficult to determine the total amount of ASA that the cells were exposed to.
Including serum in the test medium drastically reduced P. aeruginosa biofilm formation on polystyrene pegs for the majority of clinical isolates after 24 h under dynamic conditions. Serum has previously been shown to reduce early biofilm formation by P. aeruginosa on plastic surfaces, possibly by increasing twitching motility [71,72] or because of the low levels of free iron caused by iron-binding proteins such as lactoferrin [73][74][75].
Furthermore, in control samples without NaSa, supplementing the medium with increasing concentrations of serum resulted in increased QS activation, most notably in the lasB::gfp and pqsA::gfp reporter strains. Serum also increased pyocyanin and siderophore production in P. aeruginosa clinical strains and PAO1. These results agree with a previous study in which serum was shown to increase the expression of the lasI, rhlA and pqsA genes in P. aeruginosa, accompanied by increased virulence-factor production in late stage cultures (16 h) [76]. It was proposed that in the late phase, serum enhances vfr expression, resulting in increased levels of the transcription modulator Vfr, which activates the lasR gene. This leads to increased 3-OC 12 -HSL production via enhanced lasI transcription. Additionally, LasR activates rhlR/rhlI transcription, increasing C 4 -HSL levels, and LasR enhances mvfR expression, inducing the transcription of the pqsA-E operon leading to higher levels of PQS. Since Vfr is a positive regulator of lasR-and rhlR transcription [77,78], it could also explain the increased levels of siderophores and pyocyanin in the presence of serum, although vfr expression was not evaluated in the present study. In contrast, albumin, a major component of serum, has been shown to decrease las-and rhl-associated QS, but not that associated with pqs, by binding QS signal molecules [53], indicating that serum components other than albumin also play a role in las and rhl. The serum-dependent increased production of siderophores could also be due to changes in iron concentrations. Studies have shown that low levels of free iron, as in serum due to iron-binding proteins such as transferrin [79,80], induce siderophore production. It has also been shown that serum increases the virulence of the problematic wound pathogen S. aureus, further highlighting the relevance of including serum when studying wound infection in vitro [81].
In the present study, a trend was observed in which the inclusion of serum in the test medium decreased the potency of NaSa treatment compared to that achieved in serum-free medium, although only the las system was significantly reduced with 10 mM NaSa treatment. It was previously shown that serum components such as albumin interact with a wide variety of pharmaceuticals [40], e.g. aspirin [82,83]. Importantly, the present results show that NaSa was still effective in medium supplemented with 50 % serum. This indicates that NaSa could potentially be effective as a QSI in the wound environment where serum is a major component.
In conclusion, this study shows that NaSa significantly decreases the gene expression of the las, rhl and pqs QS systems and attenuates the production of pyocyanin, siderophores, alkaline protease and elastase in P. aeruginosa wound strains and PAO1 cultured in simulated wound fluid. Serum alone increased QS gene expression of the las system and production of pyocyanin and siderophores, while biofilm formation was reduced, stressing the importance of including serum in vitro when appropriate.
The characterization performed on the clinical isolates from chronic wounds demonstrated that approximately half of the strains were highly virulent and expressed several QS signals, highlighting the relevance of using QSIs such as NaSa in the clinical setting as an alternative to conventional antibiotics.
The observed QS inhibitory effect of NaSa will be further evaluated in vivo.  The AB medium for QS signal-and QS interference reporter strains was supplemented with 10 µg ml −1 gentamicin and 100 µg ml −1 ampicillin or 60 µg ml −1 gentamicin, respectively. All serum used was of foetal calf origin (non-heat inactivated, HyClone, GE Healthcare Life Sciences, USA). The strains used are listed in Table 3.

Characterization of P. aeruginosa clinical strains from chronic wounds (i) Virulence characterization
Pyocyanin and siderophore production Overnight cultures of the clinical strains and PAO1 on blood agar were diluted in saline [0.9 % (w/v) NaCl] until OD 546 =0.1 was reached, which corresponds to 10 8 c.f.u. ml −1 , and then 100 µl was added to Kings agar A in 24-well plates and 5 µl was added to Pseudomonas F-agar in 9 cm diameter petri dishes for detection of pyocyanin and siderophore (pyoverdine and pyochelin) production [84], respectively. Plates were incubated for 24 h (siderophores) and 48 h (pyocyanin) at 35 °C in a humidified incubator. Visual detection of green colouration in the agar indicated that a strain was positive for pyocyanin production. Fluorescent colonies under UV light were considered positive for siderophore production [85]. At least two of three independent experiments needed to be positive for a strain to be considered a pyocyanin or siderophore producer.

Alkaline protease and elastase
The presence of alkaline protease and elastase was detected using modified agar-based assays [86,87]. Overnight cultures of the clinical isolates and PAO1 on blood agar were diluted to OD 546 =0.1 in saline, and 5 µl were plated on agar plates containing 1.5 % (w/v) tryptic soy agar (TSA, Merck, USA) in 100 mM Tris adjusted to pH=10 and supplemented with 2 % (w/v) skimmed milk powder (Scharlau, Spain) for alkaline protease determination or on plates containing 1.5 % (w/v) TSA and 0.3 % (w/v) insoluble elastin from bovine neck ligaments for elastase determination. The plates were incubated at 35 °C for up to 48 h in a humidified incubator. Strains showing visual signs of degraded milk proteins (24 h) or elastin (48 h) around the colonies were considered positive for alkaline protease and elastase, respectively. Strains with at least two of three positive independent trial results were considered alkaline protease or elastase producers.

Swarming
Overnight cultures of the clinical isolates and PAO1 on blood agar were diluted to OD 546 =0.1 in saline, and 2 µl spots were plated on agar containing 1.5 % (w/v) TSA and 2 % (w/v) skimmed milk powder. Swarming behaviour was observed during a period of 48 h of incubation in a humidified incubator at 35 °C. The experiment was repeated three times.

Rhamnolipids
Rhamnolipid production was evaluated in a glycerol-based medium [88]. Overnight cultures of the clinical isolates and PAO1 on blood agar were diluted to OD 546 =0.1 in saline. A total of 100 µl of the OD suspension was added to 3 ml medium containing 42 g l −1 glycerol, 1.4 g l −1 NaNO 3 , 7 g l −1 K 2 HPO 4 , 3 g l −1 KH 2 PO 4 and 0.2 g l −1 MgSO 4 ×7 H 2 O. Rhamnolipids were quantified based on the orcinol assay [89]. In brief, after 6 days of culture at 35 °C and 100 r.p.m., the supernatants were collected by centrifugation at 14 000 g for 10 min. A volume of 300 µl of supernatant was added to 1.7 ml diethyl ether (VWR, USA), vortexed and centrifuged using a bench centrifuge (iFuge, Neuation, India) at 2000 g for 1 min. The diethyl ether fraction was collected and evaporated overnight prior to the addition of 300 µl dH 2 O. After vortexing, 100 µl of the rhamnolipid-water solution was added to 900 µl 0.19 % orcinol (w/w) in 53 % (w/w) H 2 SO 4 . The mixture was incubated at 80 °C for 30 min before cooling down at RT. Of this mixture, 200 µl was added to the wells of a 96-well plate (Nunc, Denmark), and the OD was read at 421 nm using a plate reader (FluostarOmega, BMG LABTECH, Germany). Test medium without bacteria, using the same sample preparation, was used as a blank. A strain was considered a rhamnolipid producer if the average OD from three independent experiments differed from that of the blank with statistical significance.

Stapholytic activity
LasA protease activity was measured by determining the ability of P. aeruginosa culture supernatants to lyse boiled Staphylococcus aureus cells [24]. Staphylococcus aureus ATCC 25923 was grown in 5 ml TSB at 35 °C and 100 r.p.m. in a shaking incubator (KS4000i, IKA, Germany). The cells were harvested by centrifugation at 4000 g for 10 min and added to 30 ml of 10 mM Na 2 PO 4 (dH 2 O) containing 1.5 % (w/v) TSA. The mixture was boiled for 15 min and added to two sterile petri dishes (9 cm ø). After solidification, 5 µl of bacterial suspension of OD 546 =0.1 in saline of each clinical isolate and PAO1 was spotted on each agar plate. The plates were incubated upside down at 35 °C in a humidified incubator for 24 h. Clear zones in the agar around the colonies in at least two out of three independent experiments indicated that a strain was positive for stapholytic activity.
Overnight cultures of the clinical isolates on blood agar were diluted to OD 546 =0. The area under the curve (AUC) data of baseline-and blank-adjusted fluorescence intensity (FI) (OD normalized) from three independent experiments, with at least two technical replicates, were used to characterize the clinical strains in terms of signal production. As a cut-off value for signal production characterization, strains with AUC data within the standard curve and above that of the signal deficient mutant were considered positive for QS signal production.

Effect of naSa on PAo1 QS activation and virulencefactor production in minimal media with/without serum supplementation (i) MIC of naSa
To ensure that the NaSa concentrations did not inhibit growth, the MIC of the clinical isolates, PAO1 and the QS reporter strains was determined in 50 % (v/v) serum in saline. Colonies from overnight cultures on blood agar plates were diluted to OD 546 =0.1 and further diluted 1 : 1000 in 50 % (w/v) serum to approximately 10 5 c.f.u ml −1 . In a flat-bottomed 96-well plate (Nunc, Denmark), 150 µl inoculum was mixed 1 : 1 with different concentrations of NaSa in 50 % (w/v) serum in saline. Final concentrations of NaSa ranged between 2 and 250 mM using two-fold dilution steps. The plate was incubated for 24 h at 35 °C under static conditions. Next, the OD 600 was measured, and concentrations of NaSa resulting in OD 600 <0.01 were used to define the MIC. At least two of the three trials needed to match for the results to be considered valid.

(ii) Effect of naSa on QS activation in minimal media
To evaluate the effect of NaSa on QS activation, reporter strains were used as described previously [41] with modifications. In brief, overnight cultures of the reporter strains lasB::gfp(ASV), rhlA::gfp(ASV) and pqsA::gfp(ASV) grown in TSB supplemented with 60 µg ml −1 gentamicin were washed twice in AB medium by 5 min centrifugation at 4000 g, diluted to OD 546 =0.1 and further diluted 1 : 10 in AB medium. NaSa was diluted to 10 mM in the reporter-strain-containing AB medium and further two-fold diluted in same medium to 20 µM. Growth curves of the reporter strains were obtained to ensure that the chosen serum and NaSa levels did not significantly inhibit bacterial growth (Fig. S1, available in the online version of this article).
The same checkerboard approach was used with P. aeruginosa PAO1, using the same medium but without antibiotics. After 72 h of incubation at 35 °C in a humidified incubator, cultures were centrifuged at 14 000 g for 10 min, and 50 µl supernatants were transferred to a clear 384-well plate (Nunc, Denmark). The OD was measured at 405 and 690 nm for indirect quantification of siderophores and pyocyanin production, respectively [93,94]. For samples resulting in OD readings above the limit of detection of the plate reader, the well volumes were reduced and subsequent OD readings were pathwaycorrected. Average data from three independent experiments are presented.

Effect of naSa on clinical P. aeruginosa strains from chronic wounds
Overnight cultures of strains 1, 4, 12 and PAO1 on blood agar plates were diluted to OD 546 =0.1 in LB supplemented with 50 % (w/v) serum and NaSa (0, 2 or 10 mM final concentrations) and cultured in a 24-well microtiter plate (Nunc, Denmark) at 35 °C and 100 r.p.m. for 24 h and 48 h. Cell-free LB with 50 % (w/v) serum served as a blank sample. After 24 h and 48 h, 500 µl of bacterial suspension from each well was transferred to an Eppendorf tube and centrifuged at 14 000 g for 10 min. Next, the production of the following virulence factors was evaluated.

(i) Pyocyanin and siderophore production
A volume of 300 µl of the supernatant was transferred to a 96-well microtitre plate (Nunc, Denmark). The presence of pyocyanin and siderophores was evaluated after 48 h of growth using a plate reader by determining their blankcorrected ODs at 405 and 690 nm, respectively. Average data from three independent experiments were calculated.

(ii) Alkaline protease activity
Alkaline protease activity was evaluated in an azocasein-based assay [95] after 24 h of growth. Altogether, 100 µl of supernatant was added to 400 µl 1 % (w/v) azocasein solution in 100 mM TRIS (pH 10) and incubated at 35 °C and 100 r.p.m. for 30 min before 500 µl 10 % (w/v) trichloroacetic acid was added to stop the degradation of azocasein. The tubes were centrifuged at 14 000 g for 10 min before the supernatant was mixed with 1 M NaOH in a 1 : 1 ratio. Of the mixture, 100 µl was further diluted 1 : 1 in 1 M NaOH and transferred to a 384well microtitre plate, and OD was recorded at 420 nm using a plate reader. Data were blank-subtracted against cell-free LB supplemented with 50 % (v/v) serum. Average data from three independent experiments were calculated.

(v) Biofilm formation
Biofilm formation was evaluated as described previously using strains 1, 4, 12 and PAO1 cultured in saline supplemented with 50 % (v/v) serum and 0, 1 or 10 mM NaSa. The data are presented as blank-corrected average OD values from three independent experiments, each with three technical replicates.

Effect of serum on selected virulence factors in clinical P. aeruginosa strains from chronic wounds
The effect of serum on the production of siderophores and pyocyanin was investigated in LB with and without the addition of 50 % (v/v) serum after 24 and 48 h cultures, respectively. Overnight cultures of strain 1-14 and PAO1 in TSB were washed twice by centrifugation at 4000 g for 10 min and resuspended in saline to OD 546 =0.1. In total, 5 µl of these bacterial suspensions were added to 195 µl LB or LB supplemented with 50 % (v/v) serum in 96-well microtitre plates. After incubation at 35 °C in a humidified incubator, 50 µl of bacterial supernatants of each sample were collected by centrifugation at 4000 g for 10 min and transferred to a 384-well plate (Nunc Denmark). Siderophore and pyocyanin production was evaluated by absorbance measurements at 405 nm and 690 nm, respectively. The experiment was repeated three times. Biofilm formation was studied as described before, but using LB with or without 50 % (v/v) serum as media; N=6.

Statistics
One-way ANOVA followed by two-sided Dunnett's post hoc tests was performed to evaluate significant differences between the control (no treatment) and each of the different concentrations of NaSa and between the control (no serum) and each of the different concentrations of serum in medium with regard to the expression of lasB::gfp, rhlA::gfp and pqsA::gfp genes, production of siderophores, pyocyanin, elastase, alkaline protease, stapholytic protease and biofilm. Two-sided Student's t-tests were performed to calculate differences in pyocyanin, siderophores and biofilm production between the control (no serum) and the serum-containing medium as well as for the production of rhamnolipids between the strains and the blank. Statistical analyses were performed using SPSS Statistics 21 (IBM Corporation, USA) with a significance level set at P<0.05.

Funding information
This study was financially supported by the Swedish