Recombinant α-toxin

Recombinant Hla (rHla) from S. aureus was expressed and purified as described previously [26]. The purity was evaluated with a Coomassie-stained SDS-gel and hemolysis activity was tested on blood agar plates. For controls, a mock purification from Escherichia coli containing vector DNA only was carried out.

Cell culture and SILAC

The two immortalized human airway epithelial cell lines 16HBE14o- and S9 are frequently used as model cells for studying cellular functions of human airways [27–30]. S9 cells were originally derived from a cystic fibrosis patient, subsequently corrected by introduction of the gene encoding wild-type cystic fibrosis transmembrane conductance regulator (CFTR) through adenoviral transfer. 16HBE14o- cells were derived from the bronchial epithelium of a transplant patient and express wild-type CFTR. The 16HBE14o- cells are highly polarized and all known characteristics indicate that these cells represent one type of serous cells at the surface of airway epithelia, whereas the less polarized S9 cells may represent an airway cell type that is derived from the lower layer of the bronchial epithelium [27, 30, 31].

16HBE14o- and S9 cells were cultivated at 37°C with 5% CO₂ in a humidified atmosphere in RPMI-1640 (with L-glutamine and sodium bicarbonate, without arginine, leucine, lysine; Sigma-Aldrich) supplemented with 382 μM L-leucine (Sigma-Aldrich), 218 μM L-lysine, 144/72 μM L-arginine (16HBE14o-/S9 cells), 10% dialyzed fetal bovine serum (FBS, Sigma-Aldrich) and penicillin 100 U/ml, streptomycin 100 μg/ml (Biochrom AG). For 16HBE14o- cells, growth medium was additionally supplemented with 2 mM L-glutamine (Biochrom AG). For stable isotope labelling with amino acids in cell culture (SILAC), cells were grown for at least 6 cell divisions in medium supplemented with the heavy isotope labelled forms of L-lysine and L-arginine (¹³C₆¹⁵N₂-L-lysine and ¹³C₆¹⁵N₄-L-arginine, Silantes). Medium was changed every three days. Cells were sub-cultivated routinely twice a week using EDTA (0.05% (w/v) in phosphate buffered saline (PBS), Sigma-Aldrich) and trypsin / EDTA (0.05% (m/v) / 0.02% (m/v) in PBS, Biochrom AG). Cells were detached by gently tapping and suspended in growth medium. An aliquot of the cell suspension obtained was mixed with trypan blue solution (Sigma-Aldrich) for checking cell viability and counted using a Buerker haemocytometer. All cultures were tested for absence of mycoplasma contamination by PCR on a regular basis.

For proteomic and transcriptomic experiments cells were seeded in 150 mm cell culture dishes at a density of 5.5 x 10⁶ cells. Confluent cell layers were treated with 2,000 ng/ml rHla or an equivalent volume of mock control for 2 hours. SILAC-experiments were carried out in duplicate with switched labels. Transcriptomic experiments were performed in biological duplicates.

Cell viability and survival assays

Cells were seeded in 24-well plates at a density of 0.2 x 10⁶ cells. After reaching confluency, cells were treated with 2,000 ng/ml rHla or mock control. After 2, 6 or 24 hours cell viability was determined by cell counting as described above. Experiments were carried out at least in triplicate with three parallels each. For analysis of Hla cytotoxicity cells were seeded in 96-well plates at a density of 0.5 x 10⁴ cells (for both, 16HBE14o- and S9 cells) for early time points, or 1 x 10⁵ and 1.5 x10⁵ cells for long term analysis for 16HBE14o- and S9 cells, respectively. Cells were treated with 2,000 ng/mL of rHla for the indicated time points. The general metabolic condition of cells was determined by resazurin staining (Sigma-Aldrich) and absorbance was measured at 570 nm and 600 nm using a multiwell plate reader (Synergy Mx, BioTek). Experiments were carried out in triplicate with 6 parallels each.

Protein kinase inhibitors

Cells were seeded in 24-well plates at a density of 0.2 x 10⁶ cells and treated with 10 μM tyrphostin AG1478 (Sigma-Aldrich), PD98059 (Cell Signaling Technology) or vehicle 1 h prior to addition of rHla. Cell viability was determined 6 h post rHla-treatment by cell counting. Experiments were carried out in triplicates with three parallels.

Preparation of samples for proteome and phosphoproteome profilings

Cells were harvested by directly scraping in lysis buffer (8 M urea, 10 mM sodium fluoride, 2.5 mM sodium pyrophosphate, 1 mM β-glycerol phosphate, 1 mM sodium orthovanadate, 1 mM tris (2-carboxyethyl) phosphine (TCEP), 1 mM EDTA, and 20 mM HEPES, pH 8.0) and immediately frozen. Lysed cells were thawed, sonicated and free protein thiols were alkylated by addition of iodoacetamide to a final concentration of 5 mM followed by incubation for 20 min at room temperature. Lysates were clarified by centrifugation (8,700 x g, 15 min) and protein content was quantified by the Bradford method (Bio-Rad). Equal amounts of differentially labelled proteins were mixed for further sample processing. For protein expression profiling, 20 μg protein mix was fractionated into 10 gel slices by one-dimensional SDS-PAGE. Peptides were obtained by in-gel digestion using trypsin (Promega). For enrichment of phosphopeptides, 5 mg (for TiO₂-based enrichment) or 20 mg (for immuno-based enrichment) of protein mixtures were diluted eightfold with 20 mM HEPES, pH 8.0 and digested with trypsin using a protein-enzyme ratio of 20:1. Obtained peptides were purified using C18 Sep-Pak cartridges (Waters) and freeze-dried. Peptides destined for TiO₂-based enrichment were dissolved in buffer A (5 mM KH₂PO₄, 30% (v/v) acetonitrile, pH 2.7) and separated in 15 fractions by strong ion exchange chromatography (SCX) with a ResourceS SCX column (1 ml column volume, GE) connected to an Äkta Avant chromatography system (GE) using a binary linear gradient with buffer B (5 mM KH2PO4, 350 mM KCl, 30% (v/v) acetonitrile, pH 2.7) at a flow rate of 1 ml/min. Fractions were lyophilized, dissolved in 14 ml TiO₂-binding buffer (73% (v/v) acetonitrile, 10% (v/v) lactic acid, 2% (v/v) TFA) and 100 μl from a TiO₂-stock solution (30 mg / ml Titansphere TiO₂ bulk material (GL sciences) in 100% acetonitrile) was added, followed by incubation for 20 min at room temperature. Samples were centrifuged, beads were washed four times with 80% (v/v) acetonitrile, 2% (v/v) TFA and phosphopeptides were sequentially eluted with 5% (v/v) NH4OH and 30% (v/v) acetonitrile. Samples for enrichment of tyrosine-phosphorylated peptides by immunoprecipitation were dissolved in IP buffer (50 mM MOPS, 10 mM Na2HPO4, 130 mM NaCl, 0.5% (v/v) NP40, pH 7.5) and incubated overnight at 4°C with P-Tyr-100 beads prepared by coupling of 300 μg of a monoclonal mouse anti-phosphotyrosine antibody (P-Tyr-100, Cell Signaling Technology) to rec-Protein G-Sepharose 4B beads (Life Technologies) in a 1:4 ratio. Beads were washed three times with IP buffer and twice with water. Tyrosine-phosphorylated peptides were eluted twice with 50 μl 0.1% (v/v) TFA and once with 15% (v/v) acetonitrile, 0.1% (v/v) TFA. Eluates were vacuum-dried and purified with C18 Stage-Tips (Thermo Scientific). Peptides were again vacuum-dried and stored at -20°C.

Mass spectrometry and spectra analysis

LC-MS/MS analyses were performed using an EASY-nLCII nanoflow HPLC system coupled directly to an LTQ Orbitrap Velos hybrid mass spectrometer (Thermo Fisher Scientific). Peptide samples were dissolved in 20 μl 5% (v/v) acetonitrile, 0.1% (v/v) acetic acid and loaded onto a 20 cm-long self-packed C18 (Aeris Peptide 3.6 μm, pore size 100 Å; Phenomenex) analytical column. Gradual elution of peptides was achieved by running a binary linear gradient from 1% (v/v) acetonitrile/0.1% (v/v) acetic acid to 75% (v/v) acetonitrile/0.1% (v/v) acetic acid over a period of 46 minutes (TiO₂ samples) or 80 minutes (pTyr-IP samples) with a flow rate of 300 nl/min. The MS was operated in data-dependent mode, automatically switching between full survey scan (m/z 300–1700) with a resolution of 30,000 followed by fragmentation analyses of the top 10 precursors with a charge state greater than one. MS/MS analyses were either with collision induced dissociation (CID) or higher-energy collisional dissociation (HCD). Fragmentation by CID was performed in the linear ion trap with an AGC target value of 5 x 10³ ions and normalized collision energy of 35%. Spectra for HCD fragmentation were acquired in the Orbitrap mass analyzer with a target value of 5 x 10⁴ ions and 40% normalized collision energy. Precursors were dynamically excluded for repeated fragmentation for 30 s and 20 s for the CID and HCD method, respectively.

Obtained raw data files were processed using MaxQuant (vers. 1.2.2.5) with the integrated Andromeda search engine [32, 33] against the reviewed human proteome deposited in UniProtKB (Swiss-Prot database vers. 2011_11; released date: 2011-11-16; 20,251 sequences). Trypsin was specified as enzyme and a total number of 2 missed cleavages were allowed. Arg10 and Lys8 were set as heavy labels. Carbamidomethylation of cysteine residues was set as fixed modification. Oxidation of methionine, amino-terminal acetylation and phosphorylation of serine, threonine and tyrosine were selected as variable modifications. The ‘requantify’ and ‘match between runs’ options were enabled. For peptide, protein or site-specific identifications a false discovery rate cut-off of 0.01 was applied. Provided SILAC ratios are based on average values. Only SILAC ratios calculated from at least two independent events were taken into account. For protein expression analysis, only SILAC ratios from protein groups with at least two independent peptide identifications, including at least one unique peptide, were considered.

Immunoblotting

Primary antibodies used in this study were anti-ADAM10 (abcam #ab1997), PE conjugated anti-ADAM10 (Biolegend #352704), anti-v-Src (Calbiochem #OP07), anti-Src family pY416 (CST #2101), anti-MAPK14 (CST #9212, anti-MAPK14 pT180/pY182 (CST #9211), anti-FAK (BD #610088), anti-FAK pY397 (Biosource #44-625G), anti-FAK pY576 (Santa Cruz #sc-16563), anti-FAK pS910 (Biosource #44-596G), anti-MAPK3/1 (CST #9102), anti-MAPK3/1 pT202/pY204 (CST #9106), anti-PAK2 (CST #2608), anti-PAK1/2 pS144/pS141 (CST #2606), anti-BCAR1 (Santa Cruz #sc-860), anti-BCAR1 pY249 (CST #4014), anti-Paxillin (Santa Cruz #sc-5574), anti-Paxillin pY118 (CST #2541), anti-Shc pY239/pY240 (CST # 2434), anti-Gab1 pY627 (CST #3231), PE-conjugated anti-E-cadherin (BioLegend #324106), anti-EGFR (Biolegend #352904), anti-Vinculin (Life Technologies #700062).

Protein extracts were separated by one-dimensional SDS gel electrophoresis using 4–15% TGX gradient gels (Biorad) and subsequently transferred on a PVDF membrane (Merck-Millipore). Membranes were blocked for 1.5 h in 5% (w/v) dried milk in PBS-0.05% (v/v) Tween-20 and probed with various phospho-site specific primary antibodies and their corresponding pan antibodies overnight at 4°C. Detection was performed using either IRDye680RD or IRDye800CW secondary antibodies (LI-COR) and the Odyssey infrared imaging system (LI-COR).

Microscopy

For fluorescence microscopic analysis 1 x 10⁴ cells were seeded onto coverslips. After rHla or mock treatment for 2 h cells were washed with DPBS and fixed with 4% (w/v) paraformaldehyde (Sigma) for 10 minutes at room temperature, permeabilized in 0.2% (v/v) Triton X-100 (Sigma) for 5 min on ice and blocked with 1% (w/v) BSA in DPBS for 10 minutes. Cells were stained for 30 min with phycoerythrin-conjugated anti-E-cadherin in DPBS containing 1% (w/v) BSA, washed twice with BSA-DPBS and sealed with fluoromount (Sigma) or stained for 20 min with an anti-Vinculin antibody (Life Technologies) in BSA-DPBS followed by probing with an AlexaFluor488- conjugated secondary antibody (Life Technologies), rhodamine phalloidin (Life Technologies) and Hoechst 33342 (Sigma) staining for 45 min. Coverslips were sealed with fluoromount and fluorescence microscopy was performed using a Zeiss Axio Observer.

Gene expression analysis

Total RNA was isolated using the TRIzol reagent (Invitrogen). RNA was purified using the RNA Clean-Up and Concentration Micro Kit (Norgen) and concentrations were measured using a ND-1000 spectrophotometer (Thermo Fisher Scientific Inc). RNA integrity was validated by means of the lab-on-chip capillary electrophoresis technology (Bioanalyzer 2100, Agilent Technologies). Only RNA samples with an RNA integrity number (RIN)>9.5 [34], 260/280 nm≥1.8, 260/230 nm≥1.9 were used for microarray analyses. For each sample, 200 ng of total RNA was reverse transcribed into cDNA, amplified, and in vitro transcribed to cRNA. Sense-strand cDNA was generated from 10 μg of purified cRNA using random primers, followed by fragmentation and labelling using 5.5 μg of purified sense-strand DNA. Biotinylated sense-strand DNA was then hybridized onto the Affymetrix GeneChip Human Gene 1.0 ST arrays for 16 h. Arrays were washed and stained using the Fluidics Station 450. Scanning was performed by GeneChip Scanner 3000 7G (Affymetrix); raw CEL files were generated using the GCOS software. Quality assessment of all hybridizations was carried out by inspecting scan images and by reviewing external and endogenous controls using the Expression Console software (Affymetrix).

Data analysis was carried out using Rosetta Resolver system for gene expression data analysis (Rosetta Bio software). In brief, the raw signals of the gene-specific probes were summarized using the Robust Multi-array Average algorithm and data transformation for array comparability was achieved by performing quantile normalization. Genes exhibiting significantly different expression on the RNA level were identified using the following cut-off criteria: one-way analysis of variance with subsequent Benjamini and Hochberg false discovery rate multiple-testing correction on pair-wise comparisons (ANOVA, p≤0.05), signal correction statistics (Ratio Builder, p≤0.05) and fold-change≥1.5-fold. Probe-set transformation into genes was performed by using the Rosetta Resolver transformation tool based on the Entrez Genes/Unigenes search engine (NCBI).

Functional classification, downstream effects and upstream regulator analysis

For protein and phosphoprotein data, functional annotation enrichment was based on KEGG pathways (DAVID Bioinformatics Resources 6.7; [35]. Kinase-substrate relationships and protein-protein interactions were downloaded from the PhosphoSitePlus (www.phosphosite.org; [36]) and STRING database [37] and visualized with Cytoscape version 2.8.3 [38] and Adobe Illustrator CS6.

Functional trends caused by the change of gene expression were assessed using IPA Down Stream Effects Analysis (Ingenuity Systems). In brief, protein IDs or transcript-specific probe sets corresponding to differentially expressed genes were tabulated and IPA core analysis was performed. Top hits from the Molecular and Cellular Functions category were then extracted based on Fisher’s exact test (p-value≤0.01, Benjamini-Hochberg multiple testing corrected) as an estimation of the probability of the association between groups of genes and cellular functions due to random chance. IPA regulation z-score greater than 2 (increased activation) or lower -2 (decreased activation) was used to predict whether observed cellular functions are activated or deactivated after rHla-treatment. Identification of upstream regulators was achieved through the correlation with differentially expressed genes by the use of IPA Upstream regulator analysis (Ingenuity Systems). Upstream regulators of the molecule type growth factor, kinase, and transmembrane receptor were assumed as valid effectors of gene expression after rHla-treatment if the corresponding p-value obtained by Fisher’s exact test was equal to or less than 0.01. Activation z-score algorithm was used to allow for prediction whether an upstream regulator is activated (z≥2) or inactivated (z≤-2) based on the direction of expressional change of the associated genes.

Flow cytometry

Flow cytometry was used to determine the surface expression of ADAM10, EGFR and E-cadherin on 16HBE14o- and S9- cells. For flow cytometric analysis, cells were trypsinized and 1 x 10⁶ cells were stained with PE-conjugated anti-ADAM10 (BioLegend), anti-E-cadherin (BioLegend), anti-EGFR (BioLegend) or appropriate isotype control (IgG1k, eBioscience) antibodies in 100 μl medium for 30 min at 4°C in the dark. Cells were washed twice with FACS buffer (DPBS, 1% (v/v) FBS, 3.8 mM sodium azide) and resuspended therein. Stained cells were analyzed on an Attune Acoustic Focusing Cytometer (Life Technologies). Ten thousand events were gated and analyzed with Attune software V2.1.0 or FlowJo V10.07 (Tree Star).

ADAM10 knockdown

For ADAM10 knockdown in 16HBE14o- and S9 cells, siRNAs HS_ADAM10_4 (target sequence ATG GTG TTG CTG AGA CTG TTA), HS_ADAM10_5 (target sequence TTG GTG GGC AGT ATT ACT TAT) or negative control (Qiagen) were used according to the manufacturer`s protocol. Cells were seeded in 96-well plates at a density of 0.5 x 10⁴ cells or in 60 mm plates at a density of 0.5 x 10⁶ cells. 24 h after seeding, cells were transfected with siRNAs using Lipofectamine for 24 h. Medium was exchanged and cells were left undisturbed for additional 48 h.

Staphylococcal alpha-toxin decreases viability in 16HBE14o- but not in S9 cells

In previous studies, different rHla concentrations were tested on 16HBE14o- and S9 epithelial cells. The dose-response relationship for rHla-mediated cyto-/chemokine release thereby showed highest values for 2,000 ng/ml rHla [16]. Furthermore, retardation of cell growth for 16HBE14o- cells was unaffected by up to 200 ng/ml rHla as determined by impedance measurements and most pronounced when cells were treated with 2,000 ng/ml [25]. In addition, microscopic inspection revealed that this concentration resulted in the liberation of cells from the layer and irreversible formation of paracellular gaps for 16HBE14o-, A549 and primary human epithelial cells isolated from nasal polyps in the long term but this effect was only moderate and transient in S9 cells [25]. For the characterization of rHla-mediated effects on the metabolome [18], transcriptome and (phospo-)proteome, we adopted the concentration of 2,000 ng/ml for our omic studies by carefully testing cell survival of confluent cell layers of S9 and 16HBE14o- human bronchial epithelial cells under our experimental conditions.

As shown in Fig. 1A, proportions of viable 16HBE14o- cells were 80%, 30% and 5% of the respective controls after incubation for 2, 6 and 24 h, respectively. In contrast, only decreases of less than 30% in viable cells were observed for S9 cells within the same periods. We also determined rHla-mediated effects on the general metabolic condition of the cells using a Resazurin-based assay. Corresponding to the cell type-specific changes in cell numbers, a rHla-induced drop in fitness was observed from 90% at 10 min to 50% at 24 h for 16HBE14o- cells, whereas S9 cells were much less affected by rHla (less than 9%) over the same period of time.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Cell survival and general metabolic fitness of the human bronchial epithelial cells 16HBE14o- and S9 after treatment with 2,000 ng/ml rHla.

A. Cell counts of 16HBE14o- and S9 cells 2 h, 6 h and 24 h post rHla-treatment (black bars) compared to mock treated cells (white bars). Data represent mean ± SEM (n = 3). B. General metabolic condition of rHla-treated 16HBE14o- (diamonds) and S9 (triangles) cells as indicated by a resazurin-based assay. Measurements after rHla-treatment are normalized to the initial reading before rHla addition. Data represent mean ± SEM (n = 3).

https://doi.org/10.1371/journal.pone.0122089.g001

To test if surviving S9 cells recover from rHla-treatment, sub-confluent cell cultures were assayed with the Resazurin assay over a period of 72 h (Fig. 1B). As expected, values for 16HBE14o- cells declined. However, S9 cells showed an increase in Resazurin conversion rates for subsequent sampling points indicating that they are capable of overcoming Hla-mediated cytotoxicity.

In summary, in agreement with previous studies 2,000 ng/ml rHla is cytotoxic to 16HBE14o- but not to S9 cells under the experimental conditions used. As the aim of our work was to study early Hla-mediated cellular responses associated with cellular signaling and gene expression, we decided on a time point of two hours after rHla addition for further system-wide quantitative analyses. At this time point, rHla-mediated cellular signaling processes were expected to be fully activated, but cytotoxic effects of rHla were not yet eminent.

Alpha-toxin induces substantial changes in the phosphoproteomes of 16HBE14o- and S9 cells

To elucidate alterations in phosphorylation-dependent intracellular signaling pathways associated with the action of alpha-toxin, we utilized metabolic protein labelling (SILAC, [39, 40]), phosphopeptide enrichment techniques and high-accuracy mass spectrometric (MS) characterization combining high-low and high-high strategies [41–44]. Using this workflow, we were able to quantify 5,432 modification-specific peptide sequences (based on at least two independent SILAC ratio counts) corresponding to 1,827 proteins (Fig. 2A, S1 Table). For a total of 4,523 phosphorylations (4,036 phosphoserine, 321 phosphothreonine and 167 phosphotyrosines), the modification site within the amino acid sequence of the protein was thereby mapped with high confidence (Class-I hits: Localization Probability > 75%, Score Difference > 5; S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Comparative overview about the phosphoproteomic data of 16HBE14o- and S9 cells treated with rHla for two hours.

A. Histogram depicting the distribution of SILAC phosphopeptide ratios of Hla- vs. mock-treated cells. B. Euler diagram illustrating the set of proteins assigned with significant alterations in their phosphorylation status (filled circle) in relation to all phosphoproteins. C. Venn diagram showing the relation of altered phosphorylations common to both and unique to either cell types.

https://doi.org/10.1371/journal.pone.0122089.g002

Considering the third quartile (Q_0.75) of all unsigned log₂-transformed SILAC phosphopeptide ratios from both cell lines as cut-off (1.58-fold on linear scale), 43% and 35% of the derived phosphoproteomes are affected in 16HBE14o- and S9 cells following rHla-treatment, respectively (Fig. 2B). From the 1106 proteins that are derived from overlapping phosphopeptide quantitation of both models, the half showed alterations in either system and most of them (approximately 80%) were commonly up- or down-regulated (Fig. 2C). Interestingly, most of the cell type-specific altered proteins were less phosphorylated in 16HBE14o- cells and more phosphorylated in S9 cells.

Alpha-toxin affects major signaling pathways of cell-cell and cell-matrix contacts as well as the actin cytoskeleton

To identify general and cell type-specific effects on the phosphoproteome of S9 and 16HBE14o- cells by Hla, we assessed the data sets for overrepresented biological pathways and processes. Strikingly, proteomic enrichment analysis with the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway resulted in identical hits for major pathways for both cell lines (Fig. 3A). Top pathways that were significantly enriched in regulated phosphoproteins in both cell types include the spliceosome, actin cytoskeleton signaling, focal adhesion, as well as ErbB/EGFR signaling and cell-cell contact signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Functional annotation analysis of proteins with altered phosphorylation in rHla-treated 16HBE14o- and S9 cells.

A. KEGG pathways with statistically noticeable number of proteins with altered phosphorylation following rHla-treatment for 2 h. The dotted vertical line indicates the cut-off for significant enrichment (Benjamini-Hochberg corrected p-value ≤0.01). Numbers next to the horizontal bars provide the number of assigned proteins. B. Matrix for cross-comparison of recurrent proteins with altered phosphorylation in both cell types between KEGG pathways from panel A. Numbers greater than the half of the maximum possible are highlighted. C. Ranking of phosphoproteins with altered phosphorylation that appear in at least three enriched KEGG pathways. D. Network visualization of proteins from the enriched KEGG pathways that have been identified with altered phosphorylation in both cell lines. Grey lines indicate experimentally validated protein-protein-interactions and red arrows kinase-substrate relationship taken from STRING and PhosphoSitePlus, respectively. rHla-induced changes are indicated in ruby (increased p-sites); cyan (decreased) or purple (both) for kinases (diamond shape) and substrates (circles) for 16HBE14o- cells (left half of shape) and S9 cells (right).

https://doi.org/10.1371/journal.pone.0122089.g003

The comparison of the individual regulated proteins across all affected pathways revealed a unique set of altered phosphoproteins associated with the spliceosome and a high degree of overlap of the remaining phosphoproteins across the other enriched pathways (Fig. 3B). In order to highlight key players, we ranked the proteins that were affected by rHla-treatment at their level of phosphorylation according to their frequency of occurrence in different pathways. In total, 15 proteins were found to be present in at least three different KEGG pathways (Fig. 3C). Remarkably, among them are 10 protein kinases—MAPK1 (ERK2), MAPK3 (ERK1), MAP2K2 (MEK2), SRC, FAK, EGFR, MAPK14 (p38 alpha), MAPK12 (p38 gamma), PAK2 and PAK4.

For selected kinases and down-stream targets, we performed Western blot analyses with cell extracts harvested 0.5, 2 and 6 hours after addition of rHla or mock treatment. Phosphosite-specific and corresponding pan antibodies for SRC, FAK1, MAPK1/3, BCAR1 and PAK2 were thereby utilized to validate the identified up- or down-regulation based on MS-SILAC ratios (S1 Fig.). As expected, the Western blot results of these key signal transduction proteins correlated well with those detected after 2 h of rHla challenge based on the MS-based phospho-profiling. The corresponding total protein levels remained largely unchanged over the monitored 6 h time period indicating direct modulation of the phosphorylation events by up-stream regulators.

Within the significantly enriched KEGG pathways, 65 proteins were phospho-regulated by rHla-mediated action in both cell lines. These proteins were subjected to interaction analyses based on known protein-protein and kinase-substrate relations. In Fig. 3D, the result is visualized in a two-circuit network separating spliceosome-associated proteins (left circle; 10 proteins) from the remaining (38 proteins). Both sub-networks show a high degree of intrinsic interconnectedness based on protein-protein interactions. Strikingly, the non-spliceosome sub-network contains 12 protein kinases, most of them with multiple and partly overlapping target proteins in the same network. The results of the pathway and network analyses suggested that Hla had a major impact on proteins that have key signaling functions in cell contact and cell anchorage processes. We therefore also used fluorescence microscopy to monitor actin, vinculin and E-cadherin as surrogate markers for morphological changes of these cell features. After 2 h of rHla-treatment, we observed a loss of actin stress fibers which was more pronounced in 16HBE14o- cells compared with S9 cells (S2B Fig.), redistribution of actin to the cellular periphery and a loss of dot-like distribution of vinculin in the cell periphery (S2B Fig.) as well as a partial loss in E-cadherin from the cell membranes (S2C Fig.) indicating down-regulation of cell-cell- or cell-matrix contacts, respectively.

Alpha-toxin induces divergent phosphorylation patterns in key signaling proteins in S9 and 16HBE14o- cells

From the signaling pathway analyses and the phenotypical observations, we hypothesized that variations in the rHla-induced activity of individual signaling nodes between both cell models might exist that are associated with the different sensitivity towards Hla. We therefore especially screened for proteins with differential phosphorylation patterns in both models.

We ranked the proteins according to the largest difference between oppositely altered phosphopeptides. Thereby, 14 phosphoproteins could be highlighted and the mapped phosphorylation sites were assigned. Most of them, i.e. 11 proteins, showed decreased phosphorylation in 16HBE14o- cells and increased phosphorylation in S9 cells (GAB1, EGFR, EPHA2, NU153, MAPK3, MAPK1, CTNB1, TOM1, CHD1, VIGLN, LASP1) and only 3 proteins (IF4G3, PSIP1, ODPA) were found with a reciprocal phosphorylation behavior. Notably, several of the proteins with differing phosphorylation including the receptor tyrosine kinase EGFR and the cytoplasmic serine/threonine kinases MAPK3 and MAPK1 are associated with signaling pathways regulating cell contacts and the actin cytoskeleton (Fig. 3D).

The kinases EGFR, MAPK3 and MAPK1 were found to be differentially regulated at sites directly involved in the induction of down-stream kinase activity. For pT202/pY204 in MAPK3/MAPK1, increased phosphorylation in S9 and decreased phosphorylation in 16HBE14o- cells were confirmed by Western blotting (S1 Fig.). Since we failed to detect signals with phosphosite-specific antibodies in EGFR, we used the known down-stream target sites Y239 in SHC and Y627 in GAB1 as surrogate markers for validation of potentially altered EGFR activity. Following rHla-treatment, SHC-Y239 and GAB1-Y627 were found to be markedly increased in phosphorylation in S9 cells, whereas at least phosphorylation of SHC-Y239 was strongly decreased in 16HBE14o- cells (S1 Fig.) indicating increased EGFR activity in S9 and decreased activity in 16HBE14o- cells as predicted by the MS results (S1 Table).

Finally, we extended the evaluation to all 71 protein kinases detected with altered phosphorylation levels following exposure of cells to rHla. Focusing on mapped sites that are known to directly modulate kinase activity, we detected up-regulation of the mitogen-activated protein kinase 14 (p38 alpha), the focal adhesion kinase FAK and Src-family kinases common to S9 and 16HBE14o- cells (S1 Table and S1 Fig.). In contrast, the receptor tyrosine kinases MET and EphA2 were found to be strongly de-phosphorylated at activation sites indicative for reduced kinase activity following rHla exposure (S1 Table).

Alpha-toxin causes distinctive expression changes in signaling modules relevant for cell viability and apoptotic cell death

Phosphorylation-mediated signaling more often than not involves modulation of gene expression. Therefore, we analyzed 16HBE14o- and S9 cells at the level of protein and mRNA expression under control conditions and 2 h after rHla-treatment. Examining our proteome data, we identified 3,579 different proteins based on at least two individual peptide identifications with at least one peptide unique per protein (false discovery rate ≤0.01). Within this set of proteins, 3,168 fulfill our criteria for quantification. Of note, protein kinases (e.g. EGFR, MAPK3, MAPK1, YES, FAK) and phosphatases (e.g. PTN23, ILKAP, DUS23, PGAM5, PP4C) as well as transcriptional and translational regulators (e.g. ATRX, CTCF, SMAD2, PURA/B) were included in this set indicating a sufficient depth of the analysis to cover even the expression of regulatory proteins, which are in general at the bottom of protein abundance. The application of the phosphoproteome derived cut-off values of more than 1.58-fold and less than -1.58-fold revealed moderate up- and down-regulation in protein expression following 2 h of rHla challenge. In summary, we detected 29 or 26 proteins, respectively, that appeared to be significantly increased or decreased in 16HBE14o- cells (S2 Table). For S9 cells, 33 proteins were observed at higher expression levels upon rHla-treatment of cells while 37 proteins were found in lower amounts. Glycosyltransferase glycogenin GYG-1, transmembrane 4 superfamily member TSN11, syndecan SYND4 and centromere protein F showed overlapping regulation at their protein levels in both cell lines. Interestingly, for both S9 and 16HBE14o- cells, functional category analysis of the proteins with altered abundances revealed an association with cellular growth and proliferation as well as cellular movement by trend.

Hence, we asked if the observed moderate changes in the proteome following rHla-treatment are preceded or accompanied by alterations at the transcriptional level. For this purpose, global transcriptome analyses of rHla-treated S9 and 16HBE14o- cells were performed. Two hours after the addition of rHla, we monitored 205 (S9) and 915 (16HBE14o-) gene-specific probe sets pointing to their corresponding genes with differential expression (Fig. 4A, S3 Table). Probe-set transformation into genes based on Entrez Genes/Unigenes allowed the identification of 44 and 44 genes in S9 cells, and 266 and 318 genes in 16HBE14o- cells that appeared up-regulated and down-regulated under influence of rHla, respectively. From these, 19 genes are transcribed at higher levels in both S9 and 16HBE14o- cells at comparable rates. For example, mRNAs of NR4A members 1 and 2, FOS, IL6, DUSP1, EGR1 and JUN were observed in this group. 25 genes show decreased transcription levels in both cell lines, including GPR21, CCL2, PMEPA1 and SERPINE1. The comparison of the transcriptome and proteome data revealed a number of direct correlations between mRNA and protein expression changes, e.g. down-regulation of plasminogen activator inhibitor PAI1/SERPINE1 (proteome: -1.9, transcriptome: -2.1) and SDC4 (-2.4; -1.3), and up-regulation of Golgi phosphoprotein 3-like protein GLP3L (1.6; 2.5), observed for 16HBE14o- cells and histone H1.2 (HIST1H1C, -1.9; -1.5) in S9 cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. DNA-microarray-based transcription profiles of 16HBE14o- and S9 cells under influence of rHla.

A. Volcano plots of mRNA expression differences under control conditions and after treatment with rHla in 16HBE14o- and S9 as a function of statistical significance (ANOVA post hoc p≤0.05) and intensity ratio at the level of probe-sets. Probe sets indicating significant higher expression after rHla-treatment are indicated in ruby, probe sets indicating higher expression under control conditions are in cyan, and probe sets with no significant expression differences are in gray. Numbers indicate the number of gene-specific mRNAs increased (ruby) and decreased (cyan) after assigning probe-sets to genes. B. Expression level changes for selected immediate early response genes after rHla treatment of 16HBE14o- (open circles) and S9 (filled circles) cells (***p<0.001; **p<0.01; *p<0.05). C. Top five cellular functions affected after rHla-treatment in the category molecular functions for 16HBE14o- (black bars) and S9 (gray bars) as predicted by IPA Downstream Effects Analysis based on differentially expressed genes. D. Functional trends after rHla-treatment in all categories as predicted by IPA Downstream Effects Analysis based on direction of change of differentially expressed genes (16HBE14o- upper panel, S9 lower panel). The word cloud depicts the frequency of terms by font size and predicted increase in functional activity is shown in ruby and decrease in activity is shown in cyan.

https://doi.org/10.1371/journal.pone.0122089.g004

We further examined genes with altered mRNA levels with regard to cellular function and activities. The observed up-regulation of NR4A1, FOS, EGR1 and JUN specific mRNAs indicates the induction of an immediate early response mediated by rHla in S9 and 16HBE14o- cells (Fig. 4B). We performed a downstream effects analysis using the semi-automated pathway analysis software IPA in order to highlight biological trends in 16HBE14o- and S9 cells exposed to rHla. Considering the top five categories in the group molecular and cellular functions we identified cellular growth and proliferation, cellular development, and cell death as enriched in differentially expressed genes in 16HBE14o- and S9 cells similarly (Fig. 4C and S4 Table). Gene expression and cell cycle functional categories ranked in the top five for 16HBE14o- cells, whereas cellular movement and lipid metabolism ranked in the top five for S9 cells. Although affected categories in 16HBE14o- cells show overlap with S9 cells, and vice versa, we detected a marked difference between both types of cells when looking at the associated predicted increase or decrease of activity (Fig. 4D and S5 Table). The functional terms associated with cell proliferation (proliferation, tumorigenesis) appear as decreased in 16HBE14o- cells, whereas proliferation is predicted to be enhanced in S9 cells after rHla-treatment. Likewise, for 16HBE14o- cells cell death is predicted to be increased and cell viability is decreased, but in S9 cells cell viability is increased and apoptosis is decreased as predicted by the change of gene expression.

Differential effects of EGFR and MAPK on transcriptional activation in rHla-treated 16HBE14o- cells or S9 cells

We hypothesized that rHla-mediated changes observed at the RNA level are preceded by activation changes in regulators upstream of gene expression that may match with observations of our phosphoproteome analysis. The correlation analysis of regulated genes to upstream regulator activity was limited to the molecule types ‘growth factor’, ‘kinase’ and ‘transmembrane receptor’ with the aim to specifically extract key factors of phosphorylation-dependent signaling. Since numerous upstream regulators with significant p-values (p≤0.05) were identified, we focused on the top 20 upstream regulators with highest p-values (S6 Table). Strikingly, EGFR and its agonist EGF were predicted as upstream regulators in rHla-treated S9 cells. As per activation z-score, EGF is significantly activated (p = 1.13E-18, z = 3.323) and EGFR shows activation by trend (p = 2.88e-16, z = 1.858). For 16HBE14o- cells in contrast, EGF (p = 1.86E-19, z = -1.033) and EGFR (p = 7.54E-18, z = -1.492) appeared inactivated after rHla-treatment by trend. We observed a similar association between differentially expressed genes and the upstream regulators MAPK3/1. While no change of activity was indicated in S9 cells (p = 5.19E-14, z = 0.836), a clear inhibition was inferred from data for 16HBE14o- cells (p = 3.52E-14, z = -2.98).

Alpha-toxin-mediated EGFR activation in S9 cells is associated with its resistant phenotype

To clarify if the elevated EGFR and MAPK activity in S9 cells is associated with its resistant phenotype, we inhibited EGFR and MAPK activity by co-treatment of cells with the EGFR-selective inhibitor tyrphostin AG1478 and the MAPK-activating kinase MAP2K1/2 (MEK1/2) with PD98059, respectively.

Whereas S9 cell counts decreased by approximately 10% 6 h after addition of rHla in the control experiment, we monitored a significant decrease in survival of about 50% when cells were co-treated with rHla and tyrphostin AG1478 (Fig. 5A). This level of reduction in cell survival is comparable to 16HBE14o- cells treated with rHla for the same period of time. Interestingly, no significant effect on cell counts was observed when S9 cells were treated with the MAP2K1/2-selective inhibitor PD98059.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. rHla-induced EGFR activation mediates resistance of S9 cells towards Hla.

A. Cell counts of rHla-treated S9 cells co-treated with or without the EGFR-selective inhibitor tyrphostin AG1478. Graphs represent means ± SEM (n = 3). B. Quantification of EGFR surface expression in 16HBE14o- (red), A549 (blue) and S9 (orange) epithelial cells as analyzed by flow cytometry. The corresponding isotype control is indicated by a dashed line. MFI: mean fluorescence intensity.

https://doi.org/10.1371/journal.pone.0122089.g005

To further exclude an involvement of cell-type specific protein amounts of EGFR in EGFR and MAPK activity we quantified the cell surface-exposed EGFR fraction of 16HBE14o- and S9 cells as well as in a human alveolar cell line A549 by flow cytometry [45, 46]. The fraction of positively stained cells was in any case almost 100%, suggesting that all three cell lines are consistently equipped with surface exposed EGFR (Fig. 5B). EGFR-amounts were highest on 16HBE14o- cells followed by significantly reduced amounts on S9 and A549 cells at levels of 43% and 30%, respectively. The results therefore do not indicate a relation between EGFR content and EGFR activity.

ADAM10 expression levels are different in three lines of airway epithelial cells and correlate with their rHla-sensitivities

Primarily based on experiments with human A549 alveolar cells, ADAM10 has recently been identified as a proteinaceous membrane-receptor for Hla which is involved in pore formation and mediation of toxic effects [9]. We therefore compared the level of ADAM10 in 16HBE14o-, S9 cells and A549 cells. Western blot analysis revealed a higher expression of ADAM10 in whole cell lysates of 16HBE14o- cells compared to S9 cells (Fig. 6A). Similar results were obtained for surface-exposed ADAM10 as analyzed by flow cytometry. S9 cells possess only 25% and A549 cells 80% of the surface-accessible ADAM10 content compared to 16HBE14o- cells (Fig. 6B, right panel). To characterize the role of ADAM10 in rHla-mediated responses in S9 and 16HBE14o- cells in more detail, we manipulated the level of ADAM10 by siRNA-mediated knockdown and studied the influence on cell viability and activation of specific kinases during rHla-treatment (S3 and S4 Figs.). Treatment with siRNAs resulted in efficient ADAM10 reduction in either cell line (S3A Fig.). Upon reduction of ADAM10 protein levels we observed a marked reversal of the deleterious effect of 6 h rHla-treatment on general metabolic condition of S9 cells (S3B Fig.).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Involvement of surface proteins in Hla mediated cytotoxicity.

A. Western blot analysis of ADAM10 expression in 16HBE14o- and S9 cells. Signals corresponding to mature and processed ADAM10 are indicated. B. Quantification of surface E-cadherin (left panel) and ADAM10 (right panel) in 16HBE14o- (red), A549 (blue) and S9 (orange) cells as analyzed by flow cytometry. Corresponding isotype controls are indicated as dashed shapes. C. Flow cytometry analysis of surface expression of E-cadherin (left panels) and ADAM10 (right panels) in 16HBE14o- (upper panels) and S9 (lower panels) cells after 2 h under control conditions (blue) or following 2 h treatment with rHla (red). The corresponding isotype control is indicated by a dashed line. MFI: mean fluorescence intensity.

https://doi.org/10.1371/journal.pone.0122089.g006

Recently, activation of ADAM10 with subsequent cleavage of its substrate E-cadherin and concomitant dissolution of adherence junctions has been highlighted as a major mechanism of Hla-induced cytotoxicity [22, 47]. In line with this, we detected drastically higher levels of surface exposed E-cadherin in the highly Hla-sensitive 16HBE14o- cells compared to S9 cells exhibiting 7-fold decreased membrane located E-cadherin content (Fig. 6B, left panel). In addition, the numbers of positive cells suggest that a sub-population of roughly 20% did not carry E-cadherin at the cell surface whereas 16HBE14o- cells were homogenously decorated with E-cadherin. When challenged with rHla for 2 h, the amount of cell surface located E-cadherin dropped by approximately 50% in 16HBE14o- but not in S9 cells whereas the E-cadherin content was unaffected (Fig. 6C, left panel). Interestingly, a similar decline was observed for ADAM10 rHla-treated 16HBE14o-, but not in S9 cells (Fig. 6C, right panel).

If ADAM10 would be responsible for meditating rHla-triggered intracellular signaling, we would expect that suppression of ADAM10 expression results in inertness of intracellular signaling pathways to rHla-treatment of cells. To exemplify this, we suppressed ADAM10 expression using siRNAs and measured rHla-mediated changes in the phosphorylation levels of Y141 in PAK2 or Y576 in FAK, respectively. As indicated by the data reported in S4 Fig., we did not observe negative effects of suppression of ADAM10 expression on rHla-mediated changes in phosphorylation levels in PAK2 or FAK (S1 Fig.).