P. aeruginosa PAO1 (PAO1 Lausanne sub-line), a PAO1 ΔpqsA isogenic mutant, (see below) and the pqsA-reporter strain PAO1 pqsA::lux (Fletcher et al., 2007) were used in this study. Bacteria were grown on LB agar or in LB broth at 37°C, 200 rpm. The HBEC line Calu-3 was obtained from the American Type Culture Collection (ATCC no. HTB-55) and used between passages 19 and 35. Undifferentiated Calu-3 cells were maintained in minimum essential medium α GlutaMAX^TM I (MEM, Life Technologies, UK) supplemented with 10% fetal bovine serum (Sigma-Aldrich, UK). Calu-3 cells were differentiated at the air-liquid interface (Calu-3-ALI) as previously described (Zhu et al., 2010; Stewart et al., 2012). Briefly, Calu-3 cells (10⁵) were seeded on 0.4 μm pore size transwell inserts (Corning Life Sciences, NE) and fed with 500 μl of culture medium in the lower chamber every other day for 21 days. Calu-3-ALI cultures exhibited epithelial integrity and polarity featured with the expression of tight junction protein zonula occludin 1 (ZO-1), cilia, and mucus production at the apical surface (Figure S1).

An in-frame deletion of pqsA constructed in PAO1 that combined a 410-bp upstream fragment of the pqsA gene with a 423-bp downstream fragment was generated by overlap extension polymerase chain reaction (PCR) using P. aeruginosa PAO1 chromosomal DNA as template. The upstream 410-bp fragment was amplified using the forward primer pqsAf1 which carries a XbaI restriction site 5′-ATATCTAGACGCCTCGAACTGTGAGATTT-3′, and the reverse primer pqsAr1 containing the first 12 nucleotides of the pqsA gene and an overhanging end containing the last 15 nucleotides of pqsA gene (underlined) 5′-TCATGCCCGTTCCAATGTGGACATGACAGAACG-3′; the downstream 423-bp fragment was amplified using forward primer pqsAf2 containing the last 15 nucleotides of pqsA gene and an overhanging end containing the first 12 nucleotides of pqsA with its 3 upstream nucleotides (underlined) 5′-GTCATGTCCACATTGGAACGGGCATGTTGATTCAGG-3′ and the reverse primer pqsAr2 which carries a HindIII restriction site 5′-TATAAGCTTACTCGCTGTCCACTTCCAAT-3′. To perform the overlap extension PCR, a second PCR was performed using the 410 and 423-bp fragments as templates and the primers pqsAf1 and pqsAr2. The final PCR product was cloned using the XbaI and HindIII restriction sites into the suicide vector, pME3087, resulting in plasmid pYCL1. The ΔpqsA deletion was generated by allelic exchange using pYCL1 as described (41). Bacterial strains used to generate the ΔpqsA mutant are listed in Table S1. The ΔpqsA deletion was verified by nucleotide sequencing and phenotypic analysis (see Figures S2A–D).

Overnight bacterial cultures in LB broth were normalized to OD₆₀₀ 0.1 in fresh LB broth and cultured for 4 h. Bacterial cells were washed twice with PBS containing Ca²⁺ and Mg²⁺ (PBS-Ca²⁺/Mg²⁺) and suspended in PBS-Ca²⁺/Mg²⁺ at a concentration of 3.5 × 10⁸ CFU/ml for multiplicity of infection (MOI) 50 or 3.5 × 10⁶ CFU/ml for MOI 0.5. Bacterial suspensions (100 μl) were added to the apical surface of Calu-3-ALI cultures and incubated at 37°C, 5% CO₂ for different time periods as stated in the text. Controls were Calu-3-ALI cultures treated with 100 μl of PBS-Ca²⁺/Mg²⁺ (untreated) and cell culture medium inoculated with bacteria (P. aeruginosa only). In some instances, Calu-3-ALI cultures were treated with 2-heptyl-3-hydroxy-4(1H)-quinolone (PQS, synthesized in house (Diggle et al., 2003) dissolved in dimethyl-sulfoxide (DMSO) to a final concentration of 20 and 40 μM 1 h prior to infection. Control cultures were treated with equivalent amount of DMSO.

Extraction and quantification of AQs was based on a previously described method (Ortori et al., 2011). Briefly, culture medium (400 μl) was added to the transwell inserts, and the medium in the transwell insert and lower chambers was collected from cultures. Cell-free supernatants were prepared by filtration through a 0.2-μm polytetrafluoroethylene (PTFE) syringe filter. From each supernatant, a 200-μl sample was spiked with 10 μl of an internal standard solution (10 μM uniformly deuterated PQS (d4-PQS in MeOH)), and extracted three times with 0.5 ml of acidified ethyl acetate (0.1% (v/v) AcOH in EtOAc) (Sigma-Aldrich, UK). Combined organic extracts from each sample were dried under vacuum (Jouan RC10 Speedvac, Thermo Scientific) and re-dissolved in 50 μl of MeOH prior to liquid chromatography mass spectrometry (LC-MS/MS) analysis. Analysis was conducted using reversed phase chromatography (Phenomenex Gemini C18 column (3.0 μm, 50 × 3.0 mm) installed in a Shimadzu Series 10AD VP LC system), in tandem with an Applied Biosystems QTRAP 4000 hybrid triple quadrupole-linear ion trap mass spectrometer. Calibration samples were prepared by spiking 200 μl of sterile media with 50 μl of methanolic calibration mix containing HHQ, PQS, and HQNO at concentrations from 1 to 100 nM, and extracting with ethyl acetate.

Differentiated Calu-3-ALI cultures were infected with WT and PAO1 pqsA::lux reporter strain in the presence and absence of exogenous PQS as described above. The wells were visualized using a Hamamatsu luminometer and Wasabi software. Luminescence was quantified using Image J.

For quantification of cell-associated bacteria, Calu-3-ALI cultures were lysed with 250 μl of 0.1% Triton X-100 (Sigma-Aldrich, UK) and homogenized by repeated pipetting. Serial dilutions were prepared in PBS-Ca²⁺/Mg²⁺ and 100 μl of each dilution was plated on LB agar plates. Colony-forming unit (CFU) was calculated by multiplying the number of colonies by the corresponding dilution factor. For quantification of bacteria in the lower chamber, the basal culture media were collected, subjected to serial dilutions, and CFU calculated as described above.

Reverse phase protein microarray analysis (Spurrier et al., 2008) was performed to evaluate the presence of β-actin, nucleoporin 98-kDa (NUP98), human cilia-associated tubulin IV, human mucin MUC5AC, human zonula occludin-1 (ZO-1) protein, and human E-cadherin by their correspondent antibodies on lysates of Calu-3-ALI cultures post infection (Life Technologies, USA; Sigma-Aldrich, UK; Cell Signalling technology, UK). Calu-3-ALI cultures with or without WT or ΔpqsA infection at 3 and 6 hpi were lysed with 70 μl of RIPA buffer (Thermo Scientific Pierce, UK) supplemented with protease and phosphatase inhibitors (Thermo Scientific Pierce, UK). Data were analyzed using RPPanalyzer software written in R. Evaluation of cytotoxicity was based on the loss of fluorescence signal quantified as arbitrary fluorescence units (AFUs).

Cytotoxicity in undifferentiated Calu-3 cultures was assessed using the lactate dehydrogenase (LDH) colorimetric assay according to the protocol provided by the manufacturer (Roche Applied Sciences, UK). Calu-3 cells were seeded in 24-well tissue culture plates (Corning Life Sciences, Netherlands) at a density of 100,000 cells per well and cultured for 3 days. On the day of infection, culture medium was replaced with minimum essential medium α GlutaMAX I without fetal bovine serum. Cells were infected with WT PAO1 or ΔpqsA in PBS-Ca²⁺/Mg²⁺ at MOI 50 for 3 and 6 h. Cells incubated with PBS-Ca²⁺/Mg²⁺ were used as negative controls. Cell-free supernatants were collected and centrifuged to remove cellular debris and bacteria. Supernatants from uninfected cells were used as negative controls and those collected from Calu-3 cells lysed with 2% of Triton X-100 represented the 100% cell death positive control. The percentage of cell death was calculated by comparison with the optical density (OD) readings from 100% cell death controls.

Calu-3-ALI cultures were lysed as above and processed for Western blot analysis to detect actin using mouse monoclonal anti-β-actin (1 μg/ml, Sigma-Aldrich, UK) followed by HRP-conjugated goat anti-mouse secondary antibody (1:10,000 dilution, Sigma-Aldrich, UK). Antibody binding was detected using the enhanced chemiluminescence (ECL) detection reagent (GE Healthcare, UK).

RNAs were extracted from Calu-3-ALI cultures using the Absolutely RNA Miniprep Kit (Stratagene, UK) according to manufacturer’s instructions. The quantity and quality of RNA were determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, UK). First cDNA strand was synthesized from 3 μg of total RNA using random hexamers and Moloney Murine Leukemia Virus (M-MLV) Reverse Transcriptase (Life Technologies, UK) according to manufacturer’s instructions (Life Technologies, UK). Reverse transcript negative controls were also performed.

Quantitative real-time PCR (qPCR) was performed to determine transcription of genes encoding granulocyte-macrophage colony-stimulating factor (GM-CSF), TNF-α, IL-6, IL-17C, and IL-8, according to the MIQE guidelines (Bustin et al., 2009). Primers are shown in Table S2. Primers were tested for efficiency via construction of a standard curve performed with templates in 4-log serial dilutions and efficiencies between 99 and 110% were considered appropriate. 2x SYBR Green I Master Mix with 2.5 mM MgCl₂ and ROX as reference dye were used (Stratagene, UK). Each sample was tested in triplicate and the thermal cycling setting was as follows: 95°C for 3 min, 40 cycles of 95°C for 10 s, and 72°C for 30 s. The relative gene expression changes between samples were calculated using the ΔΔCq method using hypoxanthine-guanine phosphoribosyltransferase (HPRT) as reference. Analysis was carried out with the program MxPro PCR software Mx3005P (Stratagene, UK). Total mRNA was prepared at specific times.

For reverse transcriptase PCR (RT-PCR), total RNA was extracted from Calu-3 cell culture infected with P. aeruginosa PAO1 wild type or ∆pqsA using a Qiagen RNeasy extraction kit (Qiagen). cDNA synthesis was performed from 1 μg of total purified RNA by using random hexamer primers and GoScript reverse transcriptase (Promega). A total of 50 ng of resulting cDNA was PCR-amplified using Expand High Fidelity PCR System (ROCHE) and primers pqsERT For and pqsERT Rev (for pqsE), pqsART For and pqsART Rev (for pqsA), oprLRT For and oprLRT Rev (for oprL), mexGRT For and mexGRT Rev (for mexG), and lecART For and lecART Rev (for lecA) (Table S3). After 5 min of denaturation at 95°C, the following reaction cycle was used for 35 cycles: 95°C for 30 s, 55°C for 30 s, and 72°C for 1 min. The PCR products were analyzed on a 2% (w/v) agarose gel and stained with Tracklt Cyan/Orange (Invitrogen).

The levels of GM-CSF in the basal culture media were quantified using a Capture ELISA following the manufacturers’ instructions (R&D, UK). The detection limit for this assay ranges between 10 and 1000 pg/ml. Simultaneous quantification of the levels of IL-6 and TNF-α in the basal media was conducted using the FlowCytomix Simplex kit bead-based immunoassay (eBioscience, UK) according to the manufacturers’ instructions. Samples were analyzed by flow cytometry (Beckman Coulter FC500) and the concentrations were analyzed using the FlowCytomix Pro 2.4 software (eBioscience, UK). The detection limit for these assays ranges between 20 and 20,000 pg/ml. In later experiments, a multiplex system (R&D systems) was employed to quantify levels of TNF-α, IL-6, IL-17C, IL-1β, IL-1α, and IL-33. Samples were analyzed using a Bio-Rad Bio-Plex 200 system.

Statistical analyses were performed in GraphPad Prism. Significance was assessed using unpaired t-tests when comparing two groups or one-way analysis of variance (ANOVA) with Tukey’s post hoc test when comparing more than two groups.

To determine whether AQs are produced during infection of differentiated HBECs, differentiated Calu-3 cells in air-liquid interface cultures (Calu-3-ALI; Figure S1) were infected with PAO1 or the isogenic AQ-negative ΔpqsA mutant (Figure S2) at MOI 50 and AQs in the infected cultures were quantified by LC-MS/MS as described in Materials and Methods. The results show that HHQ, PQS, and HQNO are all detectable in Calu-3-ALI cultures infected by PAO1 at concentrations of 2.5 ± μM, 2 ± μM, and 2.5 ± μM, respectively, in the upper chamber, and 600 ± nM, 350 ± nM and 200 ± nM in the lower chamber. In contrast, no AQs were detected in uninfected cultures or cultures infected with the ΔpqsA mutant (Figure 1). These results demonstrate production of AQs during PAO1 infection of Calu-3-ALI cultures in a pqsA-dependent manner.

To investigate the contribution of AQs to P. aeruginosa growth and cellular damage during infection, Calu-3-ALI cultures were infected with PAO1 or the ΔpqsA mutant at MOI 50 and cell-associated bacteria, and bacteria in the lower chamber were quantified after 6 hpi. Growth of ΔpqsA was similar to that of PAO1 (Figure 2A, cell-associated) and consistent with their growth profiles in wells containing medium only (Figure 2A, PA only) or when cultured in LB broth (Figure S2A). The bacterial counts in the lower chamber at 6 hpi represented a minor proportion of the total bacterial population and, in agreement with the data obtained from the quantification of cell-associated bacteria, no significant differences were observed between PAO1 and ΔpqsA-infected cultures (4.7 ± 7.2 × 10⁴ (SD) and 5.8 ± 9.2 × 10⁴ (SD), respectively). These results indicate that the AQ-QS system is not required for promoting bacterial survival or growth during infection of Calu-3-ALI cultures at high MOI.

To investigate whether AQs contribute to P. aeruginosa virulence by altering its ability to cause cell damage and, in turn, promote breakdown of the epithelial barrier, the degradation of Calu-3 cell actin after infection with either PAO1 or ΔpqsA was quantified using a microarray-based assay and Western blot analysis. This methodology was chosen because of the variable infection patterns observed during infection ( Figure S3). Initial confocal analysis failed to provide a consistent account of the infection process with areas within the same infection displaying differential levels of cell damage and bacterial growth (Figure S3). PAO1 infection caused significant loss of the actin fluorescence signal at both 3 and 6 hpi compared with untreated cells and no differences were observed between PAO1 and ΔpqsA-infected cultures (Figure 2B). A reduction in the actin-specific signal occurred alongside reduction in the levels of ZO-1, E-cadherin, HSP90, mucin MUC5AC, and type IV β-tubulin in Calu-3-ALI cultures (Figure S4). Levels of the nuclear protein NUP98 were similarly maintained at 3 and 6 hpi after infection with both ΔpqsA and PAO1 (Figure 2B). The comparable loss of cellular proteins during PAO1 and ΔpqsA mutant infections of Calu-3-ALI cultures was in line with the levels of lactate dehydrogenase produced by undifferentiated Calu-3 cells after infection with these strains (Figure 2C). These results indicate that AQs do not contribute to P. aeruginosa-induced cytotoxicity in Calu-3-ALI cultures under these experimental conditions.

The production of AQs by P. aeruginosa in laboratory culture flasks is maximal during the stationary phase of growth, suggesting that AQ-dependent QS may play a role only at a late stage post-infection (Diggle et al., 2003). We postulated that the cell damage caused by PAO1 infection at high MOI provided a narrow window of opportunity to assess the contribution of ΑQs to lung epithelium colonization by P. aeruginosa. Hence, the infection assay was modified to model chronic infection and Calu-3-ALI cultures were infected at MOI 0.5 with both PAO1 and ΔpqsA for 6, 9, 12, and 24 h (Figure S5). PAO1 and ΔpqsA displayed similar growth when infections were performed at low MOI. Bacterial growth was detected both in the cell-associated fraction and in the lower chamber. Together, these data show that AQs do not influence PAO1 growth during high and low MOI infection of Calu-3-ALI cultures.

To examine the impact of the AQ system on the induction of innate immune responses in airway epithelial cells in response to P. aeruginosa infection, the transcriptional expression and production of pro-inflammatory mediators by Calu-3-ALI cultures after infection with PAO1 or ΔpqsA were determined by quantitative real-time PCR and immuno-assays, respectively. Comparable upregulation of GM-CSF, TNF-α, IL-6, IL-17C, and IL-8 transcripts by PAO1 and ΔpqsA was found at 2 hpi (Figure 3A). In agreement with these observations, GM-CSF, TNF-α, and IL-6 were detected at similar levels in supernatants of PAO1 and ΔpqsA infected cultures (Figure 3B). Hence, lack of AQ synthesis does not affect production of pro-inflammatory cytokines by Calu-3-ALI cultures during P. aeruginosa PAO1 infection.

P. aeruginosa-infected CF lung is exposed to AQs that could precondition epithelial cells to P. aeruginosa infection (Collier et al., 2002; Barr et al., 2015). Also, the addition of PQS to bacterial supernatants from a pqsA mutant has been shown to inhibit TNF-α and IL-6 production by mouse macrophages (Kim et al., 2010a,b) and purified PQS suppressed the production of HIP-1α in human epithelial cells with a cystic fibrosis transmembrane conductance regulator (CFTR) mutation (IB3-1) (Legendre et al., 2012). Therefore, it was important to determine whether exogenous PQS could alter activation of HBEC in response to infection with P. aeruginosa.

In the first instance, exogenous PQS was tested for its ability to influence gene expression in P. aeruginosa during infection of Calu-3-ALI cultures. For this, pqsA promoter activity in the presence and absence of exogenous PQS (20 and 40 μM) was investigated using a pqsA promoter reporter strain. pqsA promoter activity was detected 4 h post-infection (Figure 4) and was significantly enhanced by addition of exogenous PQS (40 μM), suggesting that exogenous PQS can regulate gene expression in P. aeruginosa PAO1 during infection of Calu-3-ALI cultures.

Next, Calu-3-ALI cultures were pre-treated with 40 μM PQS and then infected with PAO1 or ΔpqsA. This concentration was chosen because it significantly increased the activity of the pqsA promoter. Supernatants were collected at 3 hpi and tested for the presence of IL-17C, IL-6, TNF-α, IL-1β, IL-1α, and IL-33. Levels of IL-1β, IL-1α, and IL-33 were below the detection limit of the assay (data not shown). TNF-α, IL-6, and IL-17C were readily detected in supernatants from infected cultures (Figure 5A). In the absence of exogenous PQS, there was a tendency for increased production of IL-6 and IL-17C in the ΔpqsA-infected cultures compared to PAO1 with this trend becoming significant in the case of IL-6. Exogenous PQS reduced TNF-α and IL-6 production in WT and ΔpqsA-infected cultures and IL-17C production in the ΔpqsA-infected cultures. The growth of WT and ΔpqsA was similar in the presence and absence of exogenous PQS (Figure 5B). Exogenous PQS did not affect infection-mediated cytotoxicity (Figure S6A). These findings further corroborate the immunosuppressive role of PQS during P. aeruginosa infection (Hooi et al., 2004; Skindersoe et al., 2009; Kim et al., 2010b) albeit at doses probably not achieved in this PAO1-Calu-3-ALI infection model.

Following the analysis of PAO1 growth, cytotoxicity, and immunostimulatory ability in the presence and absence of endogenous and exogenous PQS, it was important to determine the expression of selected AQ-regulated virulence factors during PAO1 infection of Calu-3-ALI cultures under these conditions. Since PqsE has a key role in controlling a distinct virulome by an as yet unknown mechanism (Rampioni et al., 2016) and its expression is upregulated by AQ activation of the pqsABCDE operon, transcription of pqsE and virulence factors specifically controlled by PqsE was investigated during infection with PAO1 and ΔpqsA in the presence and absence of exogenous PQS. Samples were collected at 3 hpi and transcription of pqsE alongside that of pqsA and those of mexG and lecA, which are specifically controlled by PqsE (Rampioni et al., 2016), was examined by RT-PCR. The mexG gene codes for a multidrug efflux pump component and lecA for a cytotoxic galactophilic lectin known to inhibit growth, ciliary beating frequency, and the morphology of human respiratory cells (Bajolet-Laudinat et al., 1994; Aendekerk et al., 2005; Chemani et al., 2009). In agreement with the comparable growth of PAO1 and ΔpqsA in the presence and absence of exogenous PQS (Figure 5B), expression of the housekeeping gene oprL was consistent during infection with both strains in the presence and absence of PQS (Figure 6).

Transcription of pqsA was observed in the PAO1-infected samples while levels increased on provision of exogenous PQS (Figure 6). As expected, the pqsA transcript, because of the nature of the gene deletion, was not detected in ΔpqsA-infected cultures in the absence or presence of PQS (Figure 6). Interestingly, the pqsE and mexG transcripts were clearly expressed in the absence of PQS not only in the PAO1 but also in the ΔpqsA-infected cultures. In both instances, exogenous PQS increased expression of both target genes although levels were consistently lower in the case of ΔpqsA (Figure 6 and Figure S6B). Limited lecA expression could be detected in PAO1 and ΔpqsA infections. As above, lecA transcription was lower in the ΔpqsA infection and was upregulated by exogenous PQS in both WT and ΔpqsA infections (Figure 6). These results support AQ-independent transcription of pqsE and PqsE-controlled virulence factors during PAO1 infection of Calu-3-ALI cultures which can be upregulated by endogenous AQ production and exogenous PQS.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.