The pore-forming toxin Panton–Valentine leukocidin (PVL) is carried by community-acquired methicillin-resistant Staphylococcus aureus and associated with necrotizing pneumonia together with poor prognosis of infected patients. Although the cell-death–inducing properties of PVL have previously been examined, the pulmonary immune response to PVL is largely unknown. Using an unbiased transcriptional profiling approach, we show that PVL induces only 29 genes in mouse alveolar macrophages, which are associated with TLR signaling. Further studies indicate that PVL directly binds to TLR2 and induces immune responses via NF-κB in a TLR2, CD14, MyD88, IL-1R–associated kinase 1, and TNFR-associated factor 6-dependent manner. PVL-mediated inflammation is independent of pore formation but strongly depends on the LukS subunit and is suppressed in CD14/TLR2−/− cells. In vivo PVL or LukS induced a robust inflammatory response in lungs, which was diminished in CD14/TLR2−/− mice. These results highlight the proinflammatory properties of PVL and identify CD14/TLR2 as an essential receptor complex for PVL-induced lung inflammation.
Panton–Valentine leukocidin (PVL) is a β-barrel pore-forming toxin secreted from Staphylococcus aureus (1), which is the leading cause of bacterial infections in developed countries (2). As of lately, PVL gained enormous attention because epidemiological and clinical studies linked the presence of PVL to serious skin infections and life-threatening necrotizing pneumonia (3, 4). Although S. aureus strains carrying the gene for PVL used to be rare (5), the recent emergence and rapid global spread of community-acquired methicillin-resistant S. aureus (CA-MRSA) clones not only alerted us of the imminent threat of drug-resistant bacteria outside of hospitals but also raised awareness of PVL (6). In sharp contrast to healthcare-associated–MRSA and methicillin-sensitive S. aureus, the majority of CA-MRSA strains carry genes encoding PVL (6, 7). Moreover, the emergence of CA-MRSA increased the overall rate of staphylococcal infections, which range from skin to bloodstream infections and fatal necrotizing pneumonia in previously healthy people (8).
PVL consists of two subunits, LukS-PV and LukF-PV, which, in an equimolar ratio, shape the octamer structure that is essential for pore formation on host cells (9, 10). Previous studies show that human and rabbit neutrophils are highly sensitive to the pore-forming properties of PVL and rapidly undergo cell death (11, 12). It is generally accepted that myeloid cells are the prime target of PVL and that low concentrations of the toxin cause apoptosis, whereas higher amounts induce lysis of neutrophils (13). This way, PVL is considered a virulence factor that mediates an important immune escape mechanism and thus contributes to CA-MRSA pathogenesis.
Within the respiratory tract, alveolar macrophages (AMs) are considered to represent the first line of defense and express a plethora of pattern recognition receptors, including TLRs, which recognize pathogen-associated molecular patterns (14). Considering the constant exposure to inhaled bacteria, the fast and accurate recognition of pathogen-associated molecular patterns is an extremely important defense mechanism. TLR activation results in downstream signaling pathways such as activation of MAPK and the transcription factor NF-κB. These pathways then modulate inflammatory gene expression, which is crucial in shaping the innate immune response within the respiratory tract (15).
Although the death-inducing role of PVL in neutrophils has been previously examined (13), PVL’s role within the lung is somewhat controversial, and the specific responsiveness of AMs to this toxin is unknown. We therefore explored the detailed function of PVL on AMs and, using a microarray profiling approach, now show that PVL induces a highly specific inflammatory transcriptional response in AMs. Further biochemical and genetic studies indicate that this inflammatory response is independent from PVL’s pore-forming ability and is mediated via NF-κB, through a CD14-TLR2–dependent mechanism.
Materials and Methods
Protein production and purification
To generate PVL, we amplified the sequence from S. aureus V8 (ATCC 49775) using the following primers, close to the coding region of LukF-PV and LukS-PV, as described earlier (13): LukF-FW (5′-caccGCTCAACATAT CACACCTGTAAg-3′), LukF-RV (5′-TTAGCTCATAGGATTTTTTTCCTTAGATTg-3′), LukS-FW (5′-caccGAATCTAAAGCTGATAACAATATTGAGAATATTg-3′), and LukS-RV (5′-TCAATTATGTCCT TTCACTTTAATTTCA TGAg-3′). PCR products were digested with XhoI and NcoI and ligated into pETM11 (16). Respective LukF-pETM11 and LukS-pETM11 constructs were transformed into competent Escherichia coli DH5-α TM cells (Invitrogen), and clones were confirmed by sequencing. BL21 (DE3) pLys-competent cells (Invitrogen) were used for expression of pETM11 plasmids for 6 h following induction with 0.05 mM isopropyl β-d-thiogalactoside (Promega). Cells were lysed using Emulsiflex-C3 (Avestin Europe); His-tagged proteins were isolated using Ni-NTA resin (Qiagen) and desalted using ZEBA columns (Thermo Scientific). Finally, proteins were subjected to LPS removal using DetoxiGel columns (Thermo Scientific) until a final LPS concentration of <0.02 EU/ml was ensured (Charles River Analytics). Homogeneity of the final products (LukF and LukS) was checked by silver stains, which showed single protein bands (Supplemental Fig. 1A). LukS and LukF were aliquoted and stored at −20°C until use, when both subunits were mixed at equimolar ratios immediately before added to cells or mice, respectively.
Polymorphonuclear cell isolation
Human polymorphonuclears (PMNs) were isolated from peripheral blood of healthy volunteers using gradient centrifugation (Histopaque), according to the manufacturer’s instruction (Sigma-Aldrich). Mouse PMNs were isolated from bone marrow of wild-type (WT) mice by flushing femurs and tibias with HBSS (Sigma-Aldrich) supplemented with 0.5% FCS. Cell suspensions were passed through a 70-μm filter, washed, and separated using a 62.5 and 81% Percoll gradient (Sigma-Aldrich). After centrifugation, PMNs were found in the 81–62.5% interface. Cell viability was assessed using trypan blue exclusion and was >90%.
Cell culture and treatments
MH-S cells (American Type Culture Collection) were cultured in RPMI 1640 containing 50 mM 2-ME, 1% penicillin, streptomycin, and 10% FCS at 37°C. For generation of MH-S cells stably expressing dominant-negative TLR2, 2 × 106/ml cells were transfected with 1 μg purified pZERO dTIR-TLR2 plasmid (Eubio), according to the suppliers’ instructions (Amaxa), and were selected using 10 μg/ml puromycin and then grown in the presence of 5 μg/ml puromycin. Human embryonic kidney (HEK)293 cells stably transfected with CD14, TLR2+/−CD14, TLR4+/−CD14, or mock-transfected cells were provided by Dr. D.T. Golenbock (University of Massachusetts Medical School, Worcester, MA) and maintained in DMEM (Life Technologies) supplemented with 10% FCS and 1% penicillin-streptomycin in the presence of 50 μg/ml G418 or puromycin, respectively (17). Supernatant from HEK293 MD2 cells was filtered and added (1:2) to HEK293 CD14+/−TLR4 cells. For purity tests, LukS, LukF, PVL (280 nM), or Pam3CSK4 (10 μg/ml) were pretreated with 1 μg/μl proteinase K at 37°C for 30 min, followed by heat inactivation of the enzyme at 65°C; alkalizing the compounds was carried out with 0.2 M NaOH at room temperature for 2 h, followed by neutralization with HCl. Primary AMs were obtained from C57BL/6, TLR2−/−, CD14−/−, and CD14/TLR2 double-knockout (DKO) mice by bronchoalveolar lavage and cultured in RPMI 1640. Bone marrow-derived macrophages (BMDMs) were retrieved from the tibia and the femur of mice and differentiated in RPMI 1640 supplemented with 10% L929-conditioned medium for 7 d as described previously (18). Bronchoscopically retrieved bronchoalveolar lavage fluid (BALF) from intubated patients without underlying lung disease was used to isolate primary human AMs. This procedure was approved by the local Ethical Review Board of the Medical University of Vienna. Lipoteichoic acid (LTA) was provided by S. von Aulock (University of Konstanz, Konstanz, Germany), and Pam3CSK4 was purchased from EMC Microcollection.
Cell death and apoptosis assays
Cell death was measured using the CytoTox 96 Nonradioactive Cytotoxicity Assay (Promega), according to the manufacturer’s instruction. For this purpose, 105/ml cells were incubated with indicated amounts of PVL, and lactate dehydrogenase (LDH) release was determined at indicated times at OD 450 nm. Apoptotic laddering was assayed using a commercially available kit (Roche Diagnostics). A total of 2 × 106/ml MH-S cells were treated with 280 nM PVL or 1 μM staurosporine (19) for indicated time points, after which, total DNA was isolated as suggested by the manufacturer (Roche Diagnostics). Two micrograms of fragmented DNA was run on a 1% agarose gel and visualized by ethidium bromide staining.
Gene expression profiling
Isolated total RNA was purified using the RNeasy kit per manufacturer’s instructions (Qiagen). Total RNA (200 ng) was then used for GeneChip analysis. Preparation of terminal-labeled cDNA, hybridization to genome-wide murine GeneLevel 1.0 ST GeneChips (Affymetrix), and scanning of the arrays were carried out according to manufacturer’s protocols. Affymetrix microarray cell intensity files were combined, and the expression was normalized with the robust multi-array average algorithm (20) to obtain an expression matrix. Identification of regulated genes, comparing control and 1-h conditions, was performed with significance analysis of microarrays as described previously (21); a false-discovery rate of 1% was imposed. All data are deposited in the ArrayExpress database at www.ebi.ac.uk/arrayexpress, and the accession number is E-MEXP-2513. Functional clustering of annotated genes was performed using DAVID bioinformatical tools (22, 23).
Total lung RNA was isolated and RT-PCR performed as described previously (24). Mouse gene-specific primer sequences used are shown in Supplemental Table I.
Western blotting was conducted as described previously (24). Abs specific for phospho-IκB kinase (IKK)α/β (S177/181), phospho-IκBα (S32/36), IκB-α (R&D Systems), phospho-p38 (T180/Y182), p38, and phospho-SAPK/JNK (T183/Y185) (R&D Systems) were used at dilutions of 1:1000; β-actin (Sigma-Aldrich) was used as a loading control at a dilution 1:500. All primary Abs were obtained from Cell Signaling Technology, unless otherwise indicated.
Nuclear extracts were isolated from MH-S or BMDM cells of WT and TLR2−/− mice as previously described (25) and mixed with fluorescently labeled oligonucleotides containing the NF-κB consensus binding site (5′-AGTTGAGGGGACTTTCCCAGGC-3′ and 3′-TCAACTCCCCTGAAAGGGTCCG-5′; underlined nucleotides represent the binding site). Mixtures were run on a 5% TGE native gel and visualized using the LI-COR Odyssey Imaging System.
Cytokine and chemokine measurements
Patch clamp measurements
Currents were recorded with the patch clamp technique in whole-cell and inside-out mode (27). MH-S cells were plated on glass coverslips 12–24 h before experiments were performed. Electrodes pulled of borosilicate capillaries had a tip resistance of 5–6 MΩ for whole-cell recording and 8–10 MΩ for single-channel recording. After patch formation, an equilibrium period of 5 min followed, which was succeeded by control recordings at holding potentials ranging from −80 to +80 mV with 20-mV increments. During the control period, no electrical activity could be observed. After adding single components (LukF-PV or LukS-PV), or both components (PVL) to the bathing solution (140–280 nM), first channel openings were detected after ∼1 min. Electrophysiological measurements were carried out with an Axopatch-1D patch clamp amplifier (Axon Instruments).
Gene reporter assays
HEK 293 cells were seeded at 1.5 × 105cells/ml and transiently transfected with 3.6 μg NF-κB, AP-1, CRE reporter vectors, or empty backbone (Panomics), respectively, and 0.36 μg pRenilla luciferase gene vector (Promega). In the second set of experiments, HEK293 cells were transfected with 1 μg msc-pDENY negative control or dominant-negative MyD88, Toll–IL-1R domain-containing adaptor protein (TIRAP), IL-1R–associated kinase 1 (IRAK1), or TNFR-associated factor 6 (TRAF6) pDENY plasmids (InvivoGen), respectively, together with 2.5 μg NF-κB reporter vector and 0.25 μg pRenilla luciferase gene vector. Twenty-four hours postexpression, cells were treated with PVL for 16 h and lysed, and luciferase activity was measured by luminometry, according to the manufacturer’s instructions.
TLR2 binding assays
PVL, LukF, LukS, Pam3CSK4, or 3-deoxy-d-manno-oct-2-ulosonic acid (KDO)–lipid A was immobilized on high-binding 96-well plates by overnight incubation at 4°C. After washing steps, plates were blocked using Superblock (Thermo Scientific) for 1 h, followed by incubation with TLR2–Fc or triggering receptor expressed on myeloid cells (TREM1)–Fc fusion proteins (2 μg/ml) (R&D Systems), respectively, at room temperature for 2 h. Specific binding was quantified using anti-Fc IgG–biotin (Sigma-Aldrich), followed by streptavidin–HRP and 3,3′,5,5′-tetramethylbenzidine solution, and OD was measured on a spectrophotometer (OD 450 nm).
In vivo experimental procedures
Age- and sex-matched, pathogen-free 7- to 9-wk-old C57BL/6, TLR2−/−, CD14−/−, and CD14/TLR2 DKO mice were used in all experiments. The Animal Care and Use Committee of the Medical University of Vienna approved all experiments. TLR2−/− mice were provided by Dr. S. Akira (University of Osaka, Osaka, Japan) (28), CD14−/− mice were obtained from The Jackson Laboratory, and CD14/TLR2 DKO were generated at the Medical University of Vienna animal facility. Lung inflammation was induced as described previously (29, 30). Briefly, mice were anesthetized by inhalation of isoflurane (Baxter), and 1 μg/g PVL was administered intranasally. At indicated times, mice were sacrificed, and cell counts were enumerated from BALF. Cytokine levels were determined in BALF and lung homogenates.
Lungs for histology were harvested 24 h after induction of lung inflammation, fixed in 10% formalin, and embedded in paraffin. Four-micrometer sections were stained with H&E and analyzed by a pathologist who was blinded for groups. To score lung inflammation and damage, the entire lung surface was analyzed with respect to the following parameters: interstitial inflammation, edema, endothelitis, bronchitis, pleuritis, and thrombi formation. Each parameter was graded on a scale of 0–4, with 0, absent; 1, mild; 2, moderate; 3, moderately severe; and 4, severe. The “inflammation score” was expressed as the sum of the scores for each parameter, the maximum being 24.
Following Ag retrieval and blocking, lung sections were incubated with rat anti-mouse Ly-6G FITC (BD Biosciences) or isotype control Ab (Cemfret Analytics) at room temperature. After rinsing, rabbit anti-FITC Ab (Zymed) was added, followed by incubation with powervision poly HRP–anti-rabbit (ImmunoLogic), and visualization with 3,3-diaminobenzidin-tetra-hydrochloride (Vector Laboratories). Nuclear counterstaining was done with Mayer’s hemalaun solution.
Differences between groups were analyzed using unpaired t test or one-way ANOVA, followed by Bonferroni post hoc analyses, where appropriate, using GraphPad Prism software. Time course experiments were calculated using two-way ANOVA. Values are expressed as mean ± SD. A p value <0.05 was considered statistically significant (***p < 0.001, **p < 0.01, and *p < 0.05).
PVL activates transcription of a small subset of NF-κB–regulated genes in AMs, associated with TLR signaling
To understand the impact of PVL on lung-specific immune cells, we generated recombinant PVL and conducted a microarray profiling of mouse AMs. Considering primary cells are rather heterogeneous and to avoid potential donor effects, we used MH-S cells, which are a routinely used the immortalized mouse AM cell line, for our microarray studies. Before conducting this, we determined that PVL is functional and capable of inducing cell death in both human and mouse neutrophils (Fig. 1A) at a concentration similar to described previously (13). PVL also caused cell death in MH-S cells, albeit much later than in neutrophils (starting at 10 h), as determined by an LDH assay (Fig. 1B) and DNA laddering, which is a hallmark of apoptosis (Fig. 1C). In addition, PVL-mediated cell death occurred in a dose-dependent manner (Fig. 1D), and it relies on both the native confirmation of PVL as well as the presence of both subunits (Fig. 1E). Furthermore, primary AMs behaved similarly to MH-S cells, because PVL-induced cell death of primary murine and human AMs occurred only after prolonged incubation (Fig. 1F). These data show that PVL was biologically active and gave us the confidence to treat MH-S cells with PVL for 1 h (i.e., at a time when no cell death was observed) and to conduct microarrays.
Genome-wide analysis surprisingly showed that PVL induced the expression of only 29 genes (Fig. 2A, 2B, Supplemental Table II), suggesting that a 1-h treatment of AMs with PVL produces a highly specific response, where only a limited set of transcription factors may be induced. Importantly, gene ontology analysis suggested that TLR as well as MAPK signaling pathways were activated with the greatest likelihood (Supplemental Table III). We noticed that out of our 29 genes, many of them were previously identified as being regulated by NF-κB or to negatively regulate the NF-κB pathway such as MIP-2, NF-κB-INS, TNF-α, ZPF36, and IκB-α (25, 31), suggesting that PVL had the ability of inducing NF-κB in AMs. Finally, to verify whether the microarray results were accurate, we performed RT-PCR and could show that all tested genes were indeed bona fide transcriptional targets of PVL (Fig. 2C). Therefore, microarray profiling suggested that PVL had the ability of inducing NF-κB as well as MAPK pathways in MH-S cells, possibly via a TLR.
The canonical activation of NF-κB is typified by IκB-α phosphorylation at Ser32 and Ser36 and subsequent ubiquitin-induced degradation of IκB-α by the 26S proteasome and translocation of NF-κB into the nucleus (15). To verify whether PVL is activating NF-κB in a canonical way, we treated MH-S cells with PVL for 15, 30, or 60 min and could show that PVL resulted in phosphorylation of IκB-α at 15 min, at a time when it was degraded and was synthesized 60 min posttreatment (Fig. 3A). Furthermore, PVL treatment also resulted in phosphorylation of p38 MAPK as well as JNK at 30 and 60 min (Fig. 3A). Degradation of IκB-α 15 min post-PVL treatment resulted in increased nuclear NF-κB, showing that NF-κB was being activated in a canonical manner in response to PVL (Fig. 3B). Supershift analysis verified that the NF-κB consisted of transcriptionally competent RelA/p50 heterodimers (data not shown). As microarray profiling had uncovered that PVL had the ability to induce proinflammatory cytokines such as TNF-α (Fig. 2C), we next addressed whether p38 and/or NF-κB signaling were contributing to transcription of proinflammatory genes in MH-S cells in response to PVL using inhibitors of these pathways. SB203580 acts as a potent inhibitor of ATP binding and inhibits phosphorylation of p38α, p38β, and p38β2. MG132 is a proteasome inhibitor that prevents IκB degradation, and BMS-345541 is a highly selective IKKβ inhibitor that prevents IκBα phosphorylation and thereby blocks NF-κB–dependent transcription. As shown in Fig. 3C and 3D, TNF-α synthesis at both the mRNA and protein level was significantly reduced in MH-S cells, which had been pretreated with both of the NF-κB inhibitors but not with the p38 inhibitor. Collectively, our data strongly suggest that PVL has the ability to induce proinflammatory cytokine synthesis in MH-S cells, and although PVL induces MAPK pathways in this cell type, it is the NF-κB pathway that is required for synthesis of proinflammatory mediators such as TNF-α.
The ability of PVL to induce inflammatory gene expression is independent of pore formation
One of the first effects of pore-forming toxins is the permeabilization of the plasma membrane to ions, leading to changes in cytoplasmic ion composition, which have been previously shown to modulate inflammatory gene expression (32). To determine whether pore formation is a prerequisite for inflammatory cytokine synthesis following PVL treatment, we stimulated MH-S cells with different doses of single subunits of PVL (LukS or LukF) or an equimolar combination of both subunits (PVL) and performed whole-cell patch clamp. No pore formation was observed following treatment of MH-S cells with single subunits of toxin, but multiple ion channels were opened following a short treatment with both subunits (Fig. 4A). These data are in line with previous observations, showing that both subunits of PVL in an equimolar ratio are required to perform a pore (10). Significantly, although single subunits were incapable of forming a pore in MH-S cells, LukS was capable of inducing TNF-α gene expression (Fig. 4B). Furthermore, LukS, but not LukF, was able to induce an inflammatory response by primary AMs (Fig. 4C). These data indicate that inflammatory gene expression relies on cellular pathways that are independent of pore formation.
TLR2 plays an important role in PVL-mediated cytokine secretion, and this response is enhanced in the presence of CD14
Bacterial cell wall components from S. aureus have very potent proinflammatory activities in vitro and in vivo (30, 33) and are being sensed by different host receptors including TLRs. Recently, it has been observed that bacterial pore-forming toxins can also be recognized by TLRs (34–36). Because PVL treatment induced a specific set of genes associated with TLR signaling (Supplemental Table III), we wondered whether TLRs are important for inflammatory cytokine synthesis in response to PVL. For this purpose, we used stably transfected HEK293 cells overexpressing TLR2, TLR4, or mock-transfected cells with/without CD14. TLR4/CD14 cells exhibited absolutely no response to PVL or the TLR2 ligand LTA, whereas TLR2 cells responded to LTA as described previously (Fig. 5A) (30). Importantly, we observed that in response to PVL treatment, HEK cells stably transfected with TLR2 secreted significantly higher amounts of IL-8, with synthesis being enhanced in the presence of CD14 (Fig. 5A). These overexpression data suggest that PVL has the ability to induce proinflammatory cytokine synthesis via TLR2 and that this response is enhanced by CD14, a previously described coreceptor for TLR2 (37).
Because LPS induces TLR4 signaling and is a potent activator of NF-κB (31), we wanted to rule out LPS contamination as a possible cause of NF-κB activation and inflammation. Although HEK cells overexpressing TLR4/CD14 exhibited no response to PVL (Fig. 5A), IL-8 was secreted following LPS treatment, and this was significantly reduced by polymyxin B (Supplemental Fig. 1B). Stimulating TLR2/CD14 HEK cells with PVL in the presence of polymyxin B showed that IL-8 synthesis was not affected by polymyxin B, conclusively ruling out LPS contamination as a cause of inflammation mediated by PVL (Supplemental Fig. 1C). Furthermore, we discovered that stimulating TLR2/CD14 HEK cells with single subunits (LukF and LukS) or PVL induced IL-8 secretion in the presence of LukS or PVL but not in response to LukF (Fig. 5B). Because we observed earlier that, although both subunits of PVL were generated in an identical way, only LukS was able to activate macrophages via TLR2, whereas LukF did not induce substantial inflammation (Fig. 4C), we concluded that our recombinant proteins were not contaminated with TLR2 ligands. To further conclusively demonstrate that the proinflammatory effects of LukS and PVL depended on the proteins and not contaminating lipoproteins, we stimulated MH-S cells with LukS, PVL, and the lipoprotein Pam3CSK4 in their native forms or after treatment with proteinase K or NaOH, which is known to hydrolyze the N-terminal acyl chains of bacterial lipoproteins (38). As shown in Supplemental Fig. 1D, proteinase K abolished the TNF-inducing capacity of LukS and PVL, whereas alkalizing agents did not reduce TNF-α levels in supernatants.
We next sought out to determine which transcription factors are important in PVL-induced inflammation and transiently transfected TLR2/CD14 HEK cells with NF-κB, AP-1, and CRE reporter plasmids and could show that only NF-κB was activated by PVL (Fig. 5C) and that this activation was enhanced in the presence of CD14 (Fig. 5D). Again, LukS was the responsible PVL subunit that activated NF-κB in CD14/TLR2 HEK cells (Supplemental Fig. 2). Activation of NF-κB through TLR2 engages MyD88 and TIRAP, enrollment of IRAK4 and IRAK1, followed by formation of complexes of TRAF6 with TGF-β–activated kinase 1 and TGF-β–activated protein kinase 1-binding protein 1, leading to activation of the IKK complex (39). To identify cellular signaling molecules needed for NF-κB activation upon PVL challenge, we transfected TLR2/CD14 HEK cells with dominant-negative plasmids for MyD88, TIRAP, IRAK1, and TRAF6. Cells treated with LTA in the presence of these plasmids exhibited significantly less NF-κB activation (Fig. 5E). Importantly, all dominant-negative plasmids (other than the vector control) also inhibited PVL-mediated NF-κB activation (Fig. 5E), suggesting that PVL signals to NF-κB via a TLR2-MyD88-IRAK1 and TRAF6 pathway in a manner analogous to LTA.
PVL binds to the extracellular domain of TLR2
Because we observed that LukS and PVL had the ability to induce IL-8 secretion in HEK 293 cells overexpressing TLR2, we next hypothesized that LukS or PVL could directly interact with TLR2. For this purpose, we developed a modified ELISA where we coated plates with increasing doses of PVL, LukF, LukS, the TLR2 ligand Pam3CSK4 (40), or the TLR4 ligand KDO–lipid A (41) and subsequently used either a chimeric molecule containing the extracellular domain of TLR2 or TREM1, fused to human Fc (IgG) to capture ligands as described in 1Materials and Methods. TLR2 interacted with Pam3CSK4 in a dose-dependent manner, as evidenced by increased signal intensity above background, but not with KDO–lipid A (Fig. 6). Importantly, PVL and LukS also interacted with TLR2 in a dose-dependent manner comparable to wells coated with Pam3CSK4. Not surprisingly, LukF, which was unable to induce an inflammatory response in macrophages (Fig. 4B, 4C) or CD14/TLR2 HEK cells (Fig. 5B), did not bind to TLR2. Binding of PVL and LukS to TLR2 was specific, because PVL or LukS did not bind to TREM1, a receptor that plays an important role in innate immunity, but whose ligand is unknown (42). These biochemical data suggest that the ability of LukS and PVL to induce inflammation in HEK cells overexpressing TLR2 (Fig. 5) is a result of its ability to directly bind to the extracellular domain of TLR2 and to induce downstream signal transduction events.
Lack of PVL-induced activation in CD14/TLR2-deficient macrophages
Dimerization of the TIR domain of TLRs is essential for TLR-mediated inflammation (39). To verify our overexpression data, we stably transfected MH-S cells with a dominant-negative mutant of TLR2, where the TIR domain is deleted (MHS TLR2-ΔTIR), producing MH-S cells, which are still capable of recognizing TLR2 ligands but are not capable of signaling (43). MHS TLR2-ΔTIR cells treated with PVL exhibited significantly lower levels of TNF-α than control cells in response to both LTA and PVL, showing that an intact TIR domain of TLR2 was essential for PVL-mediated inflammation (Fig. 7A). Because we had shown earlier that PVL has the ability to activate NF-κB signaling and that NF-κB was crucial for PVL-mediated inflammation (Figs. 3, 5), we hypothesized that NF-κB activation may be defective in TLR2−/− cells following PVL treatment. Indeed, although WT BMDM exhibited a prolonged phosphorylation of IKKβ in its activation loop at Ser177/181, beginning 15 min post-PVL treatment and becoming stronger at 60 min, TLR2−/− BMDMs exhibited only a transient phosphorylation in response to PVL 15 min posttreatment (Fig. 7B). Activation of IKKβ precedes phosphorylation of IκB-α at conserved serine residues (Ser32/36), which primes IκB-α for degradation via the 26S proteasome (15). We could show that the reduced IKKβ activation observed in response to PVL in TLR2−/− cells translated to defective IκB-α phosphorylation and degradation (Fig. 7B). This resulted in lower NF-κB translocation to the nucleus (Fig. 7C). Decreases in NF-κB activation in TLR2−/− BMDMs were specific for PVL because IKKβ, IκB-α phosphorylation, degradation, and NF-κB translocation to the nucleus were equivalent in TLR2−/− BMDMs following LPS or KDO–lipid A treatment (Fig. 7B, 7C, Supplemental Fig. 3) (data not shown).
Interestingly, although there was a significant decrease in NF-κB activation and cytokine synthesis in the absence of functional TLR2, the inflammatory response was not completely abolished. Arguing that additional receptors might be involved in these PVL-mediated responses, we reasoned that overexpression studies with HEK cells suggested a role for CD14 and therefore studied the importance of CD14 alone and in conjunction with TLR2. Although WT AMs responded robustly to PVL, TLR2−/− and CD14−/− AMs secreted lower levels of MIP-2. Importantly, the inflammatory response to PVL was completely abolished in the absence of both CD14 and TLR2 (Fig. 7D). In line with our observation of LukS being the proinflammatory component of PVL, LukS-induced MIP-2 secretion showed the same CD14/TLR2 dependency as PVL. These results illustrate that CD14 and TLR2 play important roles in PVL-induced cytokine synthesis by AMs and strongly suggest that these receptors could be important in PVL-induced lung inflammation in vivo.
CD14 and TLR2 are essential for PVL-mediated lung inflammation in vivo
As we discovered CD14 and TLR2 to be essential for the induction of an inflammatory response in AMs in vitro, we sought to gain insight into the nature of PVL-induced lung inflammation in vivo. Intranasal administration of PVL or LukS resulted in an enhanced neutrophil influx to the alveolar space and increased pulmonary levels of IL-1β, TNF-α, MIP-2, and KC 6 h posttreatment (Fig. 8A) (data not shown). These data not only illustrate the in vivo effects of PVL but also confirm that LukS is the active subunit. We next hypothesized that PVL-mediated lung inflammation would be reduced in TLR2−/− mice and indeed observed reduced PMN influx as well as lower TNF-α, MIP-2, KC, and IL-1β levels in TLR2−/− mice as compared with WT animals 6 h after intranasal inoculation (Supplemental Fig. 4).
Considering the in vitro importance of TLR2 and CD14, we next investigated the contribution of these receptors to lung inflammation in vivo. Our findings confirm that TLR2 and CD14 partially contribute to PVL- or LukS-induced lung inflammation and that no inflammatory response can be initiated in the absence of both receptors (Fig. 8B). Taken together, these data show that CD14 and TLR2 play indispensable and previously unappreciated roles in PVL-induced lung inflammation in vivo.
In this study, using microarray profiling and biochemical studies, we could show that PVL and LukS bind to the extracellular domain of TLR2 and induce inflammation in a TLR2- and CD14-dependent manner. PVL-induced gene expression could be blocked with inhibitors of the NF-κB pathway, suggesting NF-κB activation is central to PVL’s ability to induce inflammation. Consistent with this, we could show that PVL signals to NF-κB via a TLR2/MyD88/TIRAP/TRAF6 axis. Finally, experiments in TLR2, CD14, and TLR2/CD14-deficient mice and cells ratify our overexpression experiments and show that PVL induces inflammation and NF-κB activation via TLR2 and CD14 in vivo.
The early transcriptional response activated by PVL in AMs is surprisingly specific as only 29 genes were significantly expressed following a 1-h treatment. This is in contrast to LPS treatment of macrophages, which activates the expression of hundreds of genes (25, 44) and induces activation of a plethora of transcription factors such as NF-κB, CREB, EGR-1, and members of the AP-1 family (45). Inflammatory gene transcription is a combinatorial process, and NF-κB family transcription factors can frequently associate with other transcription factors, such as AP-1, to influence gene expression following a particular stimulus such as LPS or TNF-α (46). Because PVL only activates NF-κB but not other transcription factors such as AP-1, the specificity of the early gene signature activated by PVL could simply be explained by its ability to solely induce NF-κB activation. This is highlighted by the observation that although PVL-induced transcription of TNF-α in AMs is almost completely abolished by a specific IKKβ inhibitor, its protein levels, although significantly reduced, are not completely abolished, possibly because of positive feedback by PVL-induced cytokines. In line with this hypothesis, the number of genes influenced by PVL after 8 h of treatment goes up from 29 to 136 (data not shown).
We show that the ability of PVL to induce inflammation is uncoupled from its pore-forming properties and that single subunits of PVL are not able to induce a pore, even at higher concentrations. This situation is different from other toxins such as α-hemolysin of E. coli, which have been shown to induce inflammation in a dual dose-dependent manner, where low concentrations of α-hemolysin induce calcium oscillations and inflammation but high concentrations induce cell death (47). For our microarray studies, we used an identical dose of PVL that has the ability to cause cell death in MH-S cells and primary AMs and could show that inflammation and NF-κB activation precede cell death. In this respect, it is of interest that NF-κB is generally believed to be an antiapoptotic transcription factor (48). Many cancer cell lines produce high levels of inflammatory cytokines, which results in constitutive NF-κB activation and resistance to chemo- or radiotherapy (49, 50). It is tempting to speculate that the activation of NF-κB observed in response to PVL is a strategy of AMs to combat the death-inducing capability of PVL. In line with this hypothesis, we find that expression of antiapoptotic genes regulated by NF-κB increased in AMs at later time points (data not shown). Cell death following PVL seems to be more complicated and dependent on cell type. Human and mouse neutrophils are more susceptible to PVL than AMs, which could be enlightened by differences in the induction of antiapoptotic genes, as well as the activation of cell repair mechanisms (51).
Computational analysis of our microarray experiments suggested that TLR signaling was crucial for PVL’s ability to induce transcription in AMs. The idea that TLRs could play an important role in bacterial toxin recognition is not uncommon as other pore-forming toxins have been shown to mediate inflammation via TLRs, particularly TLR2 and 4. Peritoneal macrophages from TLR4 null mice exhibited blunted TNF-α secretion accompanied by decreased NF-κB activation following pneumolysin challenge compared with WT mice, suggesting that pneumolysin activates inflammation in a TLR4-dependent manner (52). Stimulation of WT peritoneal macrophages with Haemophilus influenzae porin induced an increase in cytokine secretion compared with TLR2−/− and MyD88−/− macrophages (35). Porin from Neisseria meningitidis (PorB) has been shown to induce activation of dendritic cells in a TLR2- and MyD88-dependent manner (53). In addition, it was shown that PorB binds TLR2 directly and that the presence of TLR1 enhanced this binding, whereas TLR6 did not seem to play a role (36). In this paper, we show that PVL induced inflammatory cytokine synthesis via TLR2 and CD14. Although the absence of either TLR2 or CD14 reduced the inflammatory response to PVL, CD14/TLR2 double-deficient cells and mice were incapable of responding to PVL at all. However, our study is in contrast to data showing that LukF from S. aureus is able to induce inflammation in a TLR4-dependent manner in bone marrow-derived dendritic cells (54). We could not confirm a role for TLR4 in the inflammatory response to LukF, at least in our HEK TLR4-CD14–MD2 overexpressing system (data not shown). Quite the opposite, we discovered that LukS is the active component of the toxin, because LukS stimulation of macrophages resulted in an inflammatory response, whereas LukF induced no such effect. Furthermore, LukS bound to TLR2 and required CD14 and TLR2 for signaling in vitro as well as in vivo. Overexpression of TLR2 but not CD14 was sufficient for LukS to induce an inflammatory response, indicating that TLR2 is required for signaling, whereas CD14 might act as a coreceptor.
The recent emergence of CA-MRSA and rapid expansion of highly virulent strains carrying PVL such as USA300 (3) dramatically enhanced the number of studies performed on PVL. Patients with PVL-positive S. aureus in their lungs develop necrotizing pneumonia and have exceedingly high mortality rates, indicating that PVL might be an important virulence factor (4). However, the precise role of PVL as a virulence factor of CA-MRSA has only been recently investigated and resulted in controversial observations. Whereas some studies used WT and isogenic ΔPVL S. aureus strains and failed to show any harmful effects of PVL (55, 56), others that used clinical isolates and isogenic strains carrying PVL, as well as purified PVL, clearly demonstrated the tissue-damaging properties of PVL and were able to mimic the necrotizing pneumonia observed in humans (57). A recent study performed vaccination experiments using LukS and/or LukF before infection with USA300 and disclosed improved survival in vaccinated mice, suggesting that PVL significantly contributes to S. aureus pathogenesis (58). Arguing that mouse models might not be ideal to study the in vivo role of PVL, Olsen et al. (59) infected nonhuman primates with USA300 and isogenic PVL deletion-mutant strains and ultimately did not identify a major role for PVL in aggravating pneumonia in vivo. Nevertheless, no previous report studied the precise innate immune mechanisms associated with PVL. Of great interest, while this work was in progress, Yoong and Pier (60) investigated the impact of PVL on bacterial replication in vivo and discovered that PVL-positive MRSA strains replicated less efficiently when compared with isogenic PVL-negative bacteria. Using blocking Abs, the authors could illustrate that PVL induced an inflammatory response and activated neutrophils, which ultimately counteracted bacterial multiplication. The authors argue that, at first, PVL induces a protective immune response and that neutralizing Abs that are generated upon first encounter of this bacterium might enhance the bacterial spread during subsequent infections. Hence, to our knowledge, this work is the first to predominantly shed light on the proinflammatory role of PVL. Significantly, we hereby show the molecular mechanisms underlying PVL’s proinflammatory properties within the lung in vivo and disclose that both TLR2 and CD14 are required for this response.
We thank Tiina Berg for sequencing LukS and LukF and Thomas R. Burkard for submitting the microarray data to the repository.
This work was supported in part by the Austrian Society of Antimicrobial Chemotherapy.
The online version of this article contains supplemental material.
Abbreviations used in this article:
bronchoalveolar lavage fluid
bone marrow-derived macrophage
human embryonic kidney
IL-1R–associated kinase 1
methicillin-resistant Staphylococcus aureus
Toll–IL-1R domain-containing adaptor protein
TNFR-associated factor 6
triggering receptor expressed on myeloid cell 1
The authors have no financial conflicts of interest.