Corneal buttons were collected aseptically from patients at the eye unit of a large hospital in the south of England. After removing the corneal epithelium, the stromal layer was dissected and digested with collagenase Type-1 (1 mg/mL; Life Technologies, Warrington, UK) for 3 h at 37°C. Digested stroma was cultured in Corneal Culture Medium (CCM) and hCF were characterized as described previously (Wong et al., 2011). Human monocytic cell line THP-1 was cultured in RPMI 1640 containing glutamine (Lonza, Biologics, Tewkesbury, UK), 1% (v/v) Penicillin-Streptomycin and 10% (v/v) decomplemented Fetal Calf Serum (dFCS; Life Technologies, Warrington, UK). Cells were cultured in a humidified incubator at 37°C and 5% (v/v) CO₂.

Patients provided written informed consent to use surplus corneal tissue specimen for research via the NHS Blood Transplant Eye Retrieval Service based at Southampton General Hospital (Southampton, UK). Protocols were approved by the NRES Committee South Central-Berkshire 06/Q1602/56.

Pseudomonas aeruginosa strain PAO1 (Holloway1C Stanier131) was obtained from the National Collection of Industrial, Food and Marine Bacteria, UK. P. aeruginosa strain PA14 and PA14 ΔflgK and PA14 ΔpopB mutants were provided by George O'Toole (Dartmouth Medical School, Dartmouth, USA). PA14 ΔflgK mutant is unable to synthesize a functional flagellum and the PA14 ΔpopB mutant lacks the PopB protein, an essential protein without which the TTSS cannot deliver exotoxins within eukaryotic cells (Berg, 2003; Hauser, 2009). Green Fluorescence Protein (GFP) was inserted into PAO1, PA14 and mutant strains by transformation of the PAO1/GFP plasmid provided by Alice Prince (Columbia University, New York, USA) (Parker and Prince, 2013). To complement PA14 ΔflgK and PA14 ΔpopB mutation, the flgK and popB coding region sequences from the PA14 genome were amplified using the primers listed in Supplementary Table 1 and cloned into the HindIII and BamHI sites of the shuttle cloning vector pUCP19 (Schweizer, 1991). pUCP19-flgK and pUCP19-popB were transformed into Escherichia coli DH5α and into PA14 ΔflgK and PA14 ΔpopB strains (Irani and Rowe, 1997). Carbenicillin (Sigma-Aldrich, Poole, UK) at 100 and 300 μg/mL was added to the medium for plasmid maintenance for E. coli and P. aeruginosa, respectively. Bacterial strains were grown in Luria-Bertani broth and agar (Oxoid, Basingstoke, UK) at 37°C.

hCFs were grown to 30–50% of confluence in a 24-well plate and transfected with four TLR5 siRNA (FlexiTube GeneSolution, Qiagen, Manchester, UK) using INTERFERin® (Polyplus transfection, Illkirch-Graffenstaden, France). Mock RNA was used as a transfection control. A total of 1 nM of TLR5-siRNA per well was mixed with 100 μl of dFCS-free CCM and 2 μl of INTERFERin®. After 10 min at room temperature (RT), the transfection mix was added to the CCM (500 μl) already present in the wells and incubated for 4 h. Transfection solution was then removed and 1 mL of fresh CCM was added per monolayer and cells were incubated for 48 h. Mock transfection with an irrelevant siRNA was done to ensure that TLR5 expression remained unaltered. TLR5 silencing was checked by measuring TLR5 expression in PA14-stimulated hCFs by Real Time quantitative Polymerase Chain Reaction (RT-qPCR) and western-blot.

The fliC (flagellin) coding region of PAO1 is 100% identical to that in PA14, and it was amplified using the primers listed in Supplementary Table 1 and cloned into the NdeI and XhoI restriction sites of the T7 promoter vector pET22b(+) (Merck Millipore, Hertfordshire, UK). After checking pET22b-fliC by PCR with T7 universal oligonucleotides (Supplementary Table 1) and sequencing, the plasmid was transformed into E. coli BL21 using ampicillin (Sigma-Aldrich, Poole, UK) at 50 μg/mL to maintain the plasmid. BL21 pET22b-fliC culture was grown to an OD λ₆₀₀ nm = 0.5 and induced with isopropyl-β-D-1-thiogalactopyranoside (IPTG; 1 mM) for 4 h, harvested and lysed. Cell supernatant and pellet from the lysed culture were analyzed by SDS-PAGE, which demonstrated presence of the recombinant (r)FliC in the supernatant. rFliC was purified from the soluble fraction of harvested cultures using nickel-nitrilotriacetic acid (Ni-NTA) metal-affinity chromatography (Qiagen, Manchester, UK) under native conditions, as described previously (Hung et al., 2015). Bound rFliC was eluted with 50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole buffer (pH 8.0) and elution fractions analyzed by SDS-PAGE. Pooled fractions were dialyzed against PBS, pH7.4 for 2 days. Lipopolysaccharide (LPS) content was quantified with the Pierce LAL Chromogenic Endotoxin Quantitation Kit (Thermo Fisher Scientific, Basingstoke, UK) following the manufacturer's instructions. SDS-PAGE of purified rFliC confirmed the purity and molecular mass (~49 KDa) of the protein (Supplementary Figure 1).

hCFs were seeded into 24-well culture plates (Greiner bio-one, Stonehouse, UK, ~10⁵ cells/well) and infected over time with P. aeruginosa at a Multiplicity Of Infection (MOI) of 10 in dFCS-free DMEM medium (Sigma-Aldrich, Poole, UK). Stimulatory conditions included 100 ng/mL of P. aeruginosa LPS (Sigma-Aldrich, Poole, UK) and 5 μg/mL of purified P. aeruginosa rFliC protein for 6 h and when required, streptolysin-O (SLO), was used for the intracellular delivery of the molecule as described previously (Walev et al., 2001). Bacterial association to hCFs was quantified over time as described previously (Hardy et al., 2000). Briefly, monolayers were washed gently 4 times with phosphate buffered saline (PBS), pH7.4 and lysed with 250 μl of PBS containing saponin [1% (w/v); Sigma-Aldrich, Poole, UK] and dFCS (1% w/v) for 15 min. Determination of colony forming units was done by serial dilution and viable counting on agar plates.

Inhibition experiments were done using 50 μM of Myd88 peptide inhibitor (MyD88i, Bio-techne, Abington, UK) and 1 μM of TAK-242 (Merck Millipore), respectively (Loiarro et al., 2005; Takashima et al., 2009; Matsunaga et al., 2011). hCFs were treated with MyD88i and TAK-242 for 24 and 12 h, respectively, prior to hCFs stimulation with P. aeruginosa strains, rFliC or LPS for 6 h. Uninfected hCFs monolayers in inhibited and non-inhibited conditions were used as controls.

In this study, total RNA was extracted from infected- and stimulated- hCF cells during different experimental conditions including hCF infection with the PAO1, PA14 wild-type and ΔflgK, ΔpopB mutant strains for 1, 3 or 6 h, and after stimulation with rFliC (5 μg/mL) and P. aeruginosa LPS (100 ng/mL) for 6 h. Total RNA was extracted from siTLR5-hCFs monolayers after 6 h of similar stimulatory and infection conditions. RNA extraction was done using the RNeasy Mini Kit (Qiagen, Manchester, UK). Control conditions were uninfected and un-stimulated hCFs or siTLR5-hCFs. RNA samples were treated with DNAse-50 (PrimerDesign, Chandler's Ford, UK) and total concentration was quantified using a NanoDrop spectrophotometer (ND-1000, NanoDrop, Thermo Fisher Scientific, Basingstoke, UK). cDNA was synthetized from 1 μg of RNA using the RT-nanoscript system (PrimerDesign, Chandler's Ford, UK) and the reverse oligonucleotides listed in Supplementary Table 1. RT-qPCRs were done with 1 μl of each cDNA in a Rotorgene-Q 5plex HRM (Qiagen, Manchester, UK) using the Power SYBR green PCR Master Mix (Life Technologies, Warrington, UK) and oligonucleotides listed in Supplementary Table 1. β-actin was used as an endogenous control and melting curves were done to ensure specific gene amplification. The 2^−ΔΔCT method was used for data analyses (Livak and Schmittgen, 2001).

Intracellular protein analysis of TLR2, TLR4, TLR5, NLRC4, caspase-1, caspase-4, and the cytokines IL-18 and IL-1β was done after 6 h in similar infection and stimulatory conditions, including the experiments with siTLR5-hCFs cells, as for RT-qPCR experiments. hCFs infected with complemented PA14 ΔflgK pUCP19-flgK and PA14 ΔpopB pUCP19-popB strains were included for protein analyses. Intracellular mature IL-18 and IL-1β were also measured in hCFs treated with TAK-242, MyD88i and TAK-242/MyD88i and infected with PA14 and mutant strains. Cells were washed in PBS and total protein extract was obtained by using Cytobuster™ protean extraction reagent (Merck Millipore, Nottingham, UK). Total protein (5 μg) was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electro-transferred onto polyvinylidene difluoride (PVDF) membranes (GE Healthcare, Amersham, UK). Membranes were blocked for 1 h in PBS containing 0.05% (v/v) Tween20 (PBST) and 5% (v/v) dry milk powder. Antibodies used were rabbit monoclonal anti-TLR2, -TLR4, -TLR5; rabbit polyclonal anti-NLRP3, -Caspase-4, -IL-18, and -IL-1β (Abcam, Bristol, UK); rabbit polyclonal anti-NLRC4 (Bio-techne Ltd, Abingdon, UK). All antibodies were used at the manufacturers' recommended dilutions and left overnight at 4°C. Membranes were washed three times in PBST and incubated for 1 h with goat anti-Rabbit IgG (H&L)-HRP pre-adsorbed secondary antibody (Abcam, Bristol, UK) according to the manufacturer's recommendations. Rabbit anti-β-actin polyclonal antibody (Abcam, Bristol, UK) was used as an endogenous and loading control. Immuno-detection was carried out in a Versadoc 4000 and data analyzed with Quantity One 4.6.9 software (BioRad, Langford, UK).

Extracellular IL-18 and IL-1β cytokines were quantified with the Meso Scale Discovery (MSD) electro-chemiluminescence assays (Meso Scale Diagnostics, Gaithersburg, USA.) and by western blot after protein precipitation with trichloroacetic acid (TCA). Supernatants from triplicate hCF monolayers infected with wild-type PAO1, PA14, mutant PA14 ΔflgK, and PA14 ΔpopB strains (MOI = 10) were collected after 6 h of infection.

For visual comparison of bacterial association with hCFs, hCF monolayers, grown on borosilicate glass cover slips, were infected for 9 h with GFP-expressing wild-type PAO1, PA14 and mutant PA14 ΔflgK and PA14 ΔpopB strains (MOI = 10). After infection, cells were washed three times with warmed DMEM medium and live-stained with Wheat Germ Agglutinin (1/1,000) for 10 min at RT. Monolayers were washed with PBS and fixed with paraformaldehyde (4% v/v) for 15 min at RT. After three more PBS washes, cells were permeabilized with PBS containing saponin 0.1% (v/v) and 10% (v/v) dFCS for 2 h at 4°C and counterstained with PBS containing DAPI (1/1,000) for 15 min at RT. hCFs monolayers were washed three times in PBS and examined with a Leica SP5 LSCM confocal microscope (Leitz).

Data were analyzed by one-way ANOVA with Dunnett's multiple comparison test. P < 0.05 denoted significance.

In preliminary experiments, hCFs were infected over time with different MOI (1-100) of P. aeruginosa. These experiments demonstrated that a MOI of 10 was optimal for infecting hCFs in order to study both logarithmic association of P. aeruginosa to hCFs and maximal gene and protein expression in intact cell monolayers (data not shown).

Adhesion patterns of P. aeruginosa wild-type bacteria and mutant strains for flagellum (ΔflgK) and TTSS (ΔpopB) to hCFs were first examined as an initial step required for cellular PRR stimulation. No significant differences (P > 0.05) in the association dynamics of wild-type PAO1 and PA14 strains and the PA14 ΔpopB mutant with hCF monolayers were observed (Figure 1A). By contrast, the PA14 ΔflgK mutant showed a significant ~1 log reduction in bacterial association with hCFs compared to the wild-type and ΔpopB strains at every time point examined (P < 0.05). The complemented PA14 ΔflgK pUCP19-flgK strain showed a similar association pattern (P > 0.05) as the wild-type PA14 strain by up to 6 h, but by 9 h there was an ~1 log divergence in adherence of the complemented strain compared with the parent strain (Figure 1A). There was no significant difference in adherence between the PA14 wild-type strain, the ΔpopB mutant and the complemented PA14 ΔpopB pUCP19-popB strain over the 9 h time period (P > 0.05; Figure 1A). Representative confocal microscopy images confirmed visually that the numbers of PA14 ΔflgK bacteria associated to hCFs monolayers were reduced compared to PA14 wild-type strain and ΔpopB mutant (Figure 1B). Bacterial growth rates were similar for all the strains (Supplementary Figure 2), except for the PA14 ΔflgK mutant, which was marginally slower between 0 and 3 h compared with the wild-type PA14 strain, although the growth rates of both were similar thereafter (P > 0.05). In addition, cytotoxicity induced by P. aeruginosa infection (MOI = 10) as determined by release of lactate dehydrogenase enzyme was not significant by 9 h (<5% observed enzyme release, data not shown). Death of the cell monolayers occurred between 9 and 24 h of infection, regardless of phenotype (data not shown).

Induction of inflammasome-associated signaling molecules in hCFs during the early stages of infection by P. aeruginosa and the effects caused by the ΔflgK and ΔpopB mutations were analyzed. Expression of the tlr2, tlr4, tlr5, nlrp3 and nlrc4 receptor genes, il-18 and il-1β cytokine genes and caspase-1 and caspase-4 genes was quantified by RT-qPCR after 1, 3 and 6 h of bacterial infection (Figure 2A). Each point in the graph represents the fold-change in expression caused by the wild-type and mutant bacteria compared against uninfected cells, using the 2^−ΔΔCT method. No significant changes in expression were observed for tlr2, nlrp3 or caspase-1 (data not shown). The highest levels of gene induction were observed 6 h after infection with wild-type PAO1 and PA14 strains (Figure 2A), with significant increases at 3 h observed for il-18 and il-1β only with PA14 (P < 0.05). A one-way ANOVA with Dunnett's test was used to compare the fold-changes for these genes induced by the PA14 wild-type compared with the PAO1 wild-type, and the fold-changes induced by PA14 ΔflgK and ΔpopB mutants compared with their parental PA14 strain. At 3 h, the levels of il-18 and il-1β cytokine genes induced by PA14 wild-type infection of hCF were ~3-fold higher than the levels induced by PAO1 wild-type (P < 0.05; Figure 2A). At 6 h, PA14 induced statistically significant ~4-fold higher levels of tlr4 (P < 0.05) compared with PAO1, whereas for the other genes, the observed fold-increases induced by both strains were similar (P > 0.05). hCF infection with the PA14 ΔflgK and ΔpopB mutants, showed that at 3 h, there were significantly lower (~13-fold) levels of il-18 (P < 0.001) and il-1β (P < 0.01) cytokine gene induction, compared with the PA14 wild-type strain (Figure 2A). By 6 h, infection with the PA14 ΔflgK mutant showed significantly decreased levels (~ 5–10-fold) of tlr4, tlr5, nlrc4, casp-4, il-18, and il-1β cytokine gene expression compared with PA14 wild-type (P < 0.05; Figure 2A). By contrast, infection with the PA14 ΔpopB mutant showed significant reductions (~ 4-fold) in expression of tlr4, tlr5, nlrc4, and il-18 genes (P < 0.05) compared to PA14 wild-type, but not significantly different for casp-4 and il-1β genes (P > 0.05).

Protein levels detected by western blot were consistent with the levels of gene expression quantified by RT-qPCR. After 6 h of hCFs infection with wild-type PAO1 and PA14, significant levels of TLR4, TLR5, NLRC4, native form of caspase-4 and IL-1β and IL-18 mature-formed proteins were detected by western blot with quantification of protein densitometry ratios relative to endogenous β-actin (Figure 2B, Supplementary Table 2). Significantly lower levels of production of these molecules were induced by the PA14 ΔflgK and ΔpopB mutant strains (P < 0.05) and the percentage reductions in inflammatory molecule protein densitometry values are shown in Supplementary Table 2. In comparison with wild-type bacteria, infection with the complemented PA14 ΔflgK pUCP19-flgK and PA14 ΔpopB pUCP19-popB strains restored to varying degrees the protein levels of TLR4, TLR5, NLRC4, caspase-4 and mature IL-1β and IL-18 (Figure 2B, Supplementary Table 2). Thus, complementation of ΔflgK restored ~50, 85, 60, 40, 32, and 61% of TLR4, TLR5, NLRC4, Caspase-4, IL-18, and IL-1β protein levels, respectively (Supplementary Table 2). Complementation of ΔpopB restored ~65, 60, 62, 57 37, and 73% of TLR4, TLR5, NLRC4, Caspase-4, IL-18, and IL-1β protein levels, respectively (Supplementary Table 2).

Protein levels of the caspase-4 pro-enzyme, and IL-18 and IL-1β pro-forms were also analyzed by western blot. Caspase-4 pro-enzyme and pro-IL-18 protein expression was found to be constitutive, whereas pro-IL-1β was induced by P. aeruginosa wild-type infection, reduced by infection with the ΔpopB mutant and very weakly induced by infection with the PA14 ΔflgK mutant (Figure 2C).

The inflammatory cytokine profile of hCFs infected with wild-type P. aeruginosa is characterized predominantly by the production of IL-1β and IL-18 with the expression of other cytokines and chemokines not significant (manuscript submitted elsewhere). The early release of mature IL-1β and IL-18 was quantified by MSD-ELISA from the supernatants of hCFs infected with wild-type PAO1 and PA14 strains and isogenic PA14 ΔflgK and ΔpopB mutants after 6 h. No significant differences in IL-1β were found between PAO1 and PA14 (P > 0.05). Both P. aeruginosa wild-type strains induced ~30 pg/mL of IL-1β secretion in hCFs after 6 h of infection (Figure 3A). By contrast, a significant reduction in IL-1β secretion to levels of ~5 pg/mL was observed from hCFs infected with the PA14 ΔflgK mutant strain. These lower levels of extracellular IL-1β detected in the supernatants of hCFs infected with the PA14 ΔflgK are in agreement with the low intracellular IL-1β levels shown in Figure 2B under the same infection conditions. There were no significant differences (P > 0.05) in the levels of IL-1β secretion induced by the PA14 ΔpopB mutant (~20 pg/mL) compared to wild-type PA14 after 6 h (Figure 3A). Complementation of each mutation restored the levels of IL-1β release similar to those induced by the PA14 wild-type strain.

Measurement of IL-18 in infected hCFs supernatants revealed levels of extracellular IL-18 of ≤0.25 pg/mL for all infection conditions, which was similar to uninfected controls (P > 0.05; Figure 3B). Western blotting of IL-18 protein precipitated from infected hCF-supernatants collected after 6 h infection did not show the presence of mature IL-18 (Figure 3C), which was in agreement with the MSD-ELISA data.

Pathway inhibition was used to target key intracellular signaling molecules stimulated in hCFs by P. aeruginosa flagellum and TTSS that potentially could contribute to IL-18 and IL-1β production. Since our results showed similar cytokine production in hCFs caused by PAO1 and PA14 infection, only infection with PA14 was used for the subsequent inhibition experiments. Confluent monolayers were treated with MyD88i, TAK-242, and MyD88i/TAK-242 inhibitors and then infected with PA14 wild-type and PA14 ΔflgK and ΔpopB mutant strains for 6 h. MyD88i blocks the signal through the MyD88 molecule and TAK-242 interferes with the association between the TIRAP and TRAM molecules to TLRs and inhibits the consequent signaling triggered by the receptor. Although TAK-242 has been reported to have selectivity for the LPS receptor TLR4, inhibitory effect has also been seen with other TLRs including TLR5 (Loiarro et al., 2005; Takashima et al., 2009; Matsunaga et al., 2011). Intracellular IL-18 and IL-1β protein expression was examined by western blot (Figure 4A) and extracellular release of these cytokines was quantified by MSD-ELISA (Figure 4B). Treatment of hCF monolayers with MyD88i, TAK-242 alone and MyD88i/TAK-242 together, did not induce IL-18 or IL-1β production (Supplementary Figure 3).

Infection with PA14 wild-type induced intracellular IL-18 protein production and protein densitometry values (from n = 3 independent western blots) revealed no significant differences in intracellular IL-18 protein expression induced by addition of any of the inhibitors (Figure 4A and Supplementary Table 3). Infection with the PA14 ΔpopB mutant also induced mature IL-18 intracellular expression, but addition of TAK-242 (alone or in combination with MyD88i) appeared to significantly decrease its intracellular production (P < 0.01; Figure 4A and Supplementary Table 3), whereas MyD88i alone had no significant effect. PA14 ΔflgK did not induce intracellular IL-18 production, and secretory IL-18 was absent in all treated hCFs, as previously observed (Figures 3B,C).

Infection with wild-type PA14 induced intracellular IL-1β protein production, which was not inhibited by MyD88i or TAK-242 (Figure 4A and Supplementary Table 3). In cells infected with PA14 wild-type, levels of extracellular IL-1β were apparently stimulated by the addition of MyD88i or the MyD88i/TAK-242 combination, but no statistically significant differences were found compared to untreated cells (P > 0.05, Figure 4B). PA14 ΔflgK mutant induced low levels of intracellular and extracellular IL-1β in agreement with the results in Figures 2B, 3B, and were null with the addition of TAK-242, alone or in combination with MyD88i (Figures 4A,B and Supplementary Table 3B). Addition of TAK-242 also appeared to diminish intracellular and extracellular IL-1β production in hCFs infected with the PA14 ΔpopB mutant (Figures 4A,B and Supplementary Table 3).

Our results showed that infection with the PA14 ΔflgK mutant led to reduced IL-18 and IL-1β expression by hCFs. In order to demonstrate that the flagellum may be involved in inducing cytokine expression by these cells, we next tested the hypothesis that silencing the flagellum receptor TLR5 in hCFs would lead to a reduction of cytokine production regardless of bacterial adhesion. For these experiments, the flagellum filament protein FliC of P. aeruginosa was used as a control to demonstrate TLR5 stimulation. Recombinant (r) FliC was cloned and over-expressed in E. coli using the pET expression system (Supplementary Figure 1). Wild-type hCFs and hCFs that were TLR5-silenced using siRNA-TLR5 were stimulated with the PA14 wild-type, ΔflgK mutant strain and with 5 μg/mL of rFliC, and IL-18 and IL-1β gene expression was measured by RT-qPCR. Non-stimulated cells were used as a control.

Control experiments demonstrated that hCFs transfected with siRNA-TLR5 and infected with PA14 for 6 h significantly diminished TLR5 gene (P < 0.0001; Supplementary Figure 4A) and protein (Supplementary Figure 4B) expression compared to control Mock transfection. Expression of il-18 gene was reduced ~2 fold in hCF-siTLR5 compared to hCFs when stimulated with PA14 and rFliC for 6 h (Figure 5A). Infection with PA14 ΔflgK mutant did not induce IL-18 RNA expression and TLR5-silencing did not affect IL-18 expression, which is in agreement with the results shown in Figure 4A. Similarly, decreased levels of IL-1β RNA expression in hCFs-siTLR5 were found following stimulation with PA14 and rFliC for 6 h (Figure 5B). Otherwise, IL-1β RNA expression was slightly increased in hCFs-siTLR5 infected with the PA14 ΔflgK mutant compared to no infection (Figure 5B).

Secreted cytokines (IL-18 and IL-1β) were quantified from each supernatant after 6 h of hCF and hCF-siTLR5 incubation with PA14, ΔflgK mutant and rFliC (Figure 5C). SLO-treatment was used to deliver rFliC intracellularly. As expected, no extracellular IL-18 was detected. Levels of secreted IL-1β protein were significantly reduced in supernatants of hCFs-siTLR5 compared to those detected in hCFs, both after stimulation with PA14 (P < 0.05 and ~30% of reduction). Most significantly, hCFs-siTLR5 treated with rFliC showed a reduction of IL-1β secretion of >95% (P < 0.0001) compared to wild-type cells (Figure 5C), which was similar to the data obtained by RT-qPCR (Figure 5B). No significant extracellular IL-1β secretion was induced by SLO treatment alone of hCFs (~0.7 pg/mL, data not shown).

To examine if LPS was responsible, possibly, for the IL-1β induction observed in hCFs-siTLR5 infected with the PA14 ΔflgK mutant strain (Figure 5B), IL-1β gene expression was measured in hCFs-siTLR5 treated with the TAK-242 inhibitor. Under these inhibitory conditions, signaling through TLR5 (flagellin receptor) and TLR4 (LPS receptor) should be blocked. Repression of IL-1β expression was detected in both hCFs and hCFs-siTLR5 infected with PA14 ΔflgK mutant (P < 0.05, Figure 6A) treated with TAK-242. These transcriptional results were confirmed by detecting IL-1β protein expression under the same inhibitory conditions: addition of TAK-242 blocked the little IL-1β production induced in hCFs infected with the PA14 ΔflgK mutant (Figure 6B). In order to examine inhibitory effects on TLR4 sensing of LPS, IL-1β gene expression was also measured by RT-qPCR in hCFs that were stimulated with 100 ng/mL of pure Pseudomonas LPS, with and without TAK-242, Myd88i and TAK-242/Myd88i treatments. As shown in Figure 6C, addition of either inhibitor decreased IL-1β expression in hCFs treated with pure Pseudomonas LPS, compared to untreated cells. Similar experiments were done measuring IL-18 expression, but no increases in IL-18 expression were detected in LPS-stimulated hCFs (data not shown). No extracellular IL-1β protein was detected after hCF stimulation with P. aeruginosa LPS alone or in the presence of any inhibitor (data not shown).

Thus far, our data suggest that the P. aeruginosa flagellin (FliC) appears to plays an important role in inducing IL-18 and IL-1β expression through TLR5 stimulation. We tested the hypothesis that chemical inhibition of TLR5-mediated signaling pathways reduces cytokine gene expression induced by FliC. hCF monolayers were pre-treated with MyD88i, TAK-242 and both inhibitors combined, then stimulated with rFliC for 6 h and cytokine gene expression quantified by RT-qPCR (Figure 7). Controls were hCF monolayers without chemical pre-treatment and no stimulation with rFliC.

rFliC stimulation of hCF increased IL-18 gene expression ~10 fold compared to unstimulated cells and addition of MyD88i, TAK-242 and MyD88i/TAK-242 significantly reduced gene expression (P < 0.0001, Figure 7A). Similarly, rFliC stimulation increased IL-1β expression by ~20 fold, compared to unstimulated monolayers (Figure 7B). Inhibition with MyD88i alone reduced IL-1β gene expression by ~40% (P < 0.05), whereas treatment with TAK-242 and MyD88i/TAK-242 combined decreased IL-1β gene expression by >75% (P < 0.0001) compared to untreated cells (Figure 7B). Secreted IL-1β was quantified from supernatants of hCF stimulated with rFliC alone or in presence of the inhibitors. SLO-treatment was used to deliver the purified protein intracellularly. As shown in Figure 7C, MyD88i treatment reduced release of IL-1β by ~20%. By contrast, addition of TAK-242, alone or in combination with MyD88i, significantly reduced extracellular IL-1β release by >95% (P < 0.0001) compared to non-inhibited rFliC-stimulated hCFs.

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.