Neither Pseudomonas aeruginosa nor flagellin affected cytosolic Ca2+ concentration ([Ca]i) in airway epithelial cell lines JME and Calu-3, but bacteria or flagellin activated NF-κB, IL-8 promoter, and IL-8 secretion. ATP (purinergic agonist) and thapsigargin (blocks Ca2+ pump, releases endoplasmic reticulum Ca2+, and triggers Ca2+ entry through plasma membrane channels) both increased [Ca]i but hardly stimulated NF-κB and IL-8. ATP and thapsigargin elicited larger, synergistic activations of NF-κB and IL-8 secretion when combined with flagellin. BAPTA-AM (to buffer [Ca]i) or Ca2+-free solution reduced increases in [Ca]i due to ATP or thapsigargin and also reduced NF-κB activation and IL-8 secretion triggered by flagellin, ATP, thapsigargin, ATP + flagellin, and thapsigargin + flagellin. IL-8 promoter analysis showed that AP-1 and CCAAT/enhancer-binding protein (C/EBP)β/nuclear factor for IL-6 (NF-IL6) sites were important for IL-8 expression, and the NF-κB-binding site was critical for activation by all agonists and for activation by [Ca]i. Thus increased [Ca]i was not required for P. aeruginosa- or flagellin-activated NF-κB and IL-8 expression and secretion, and increased [Ca]i was only weakly stimulatory during activation by ATP or thapsigargin. However, ATP- or thapsigargin-induced increases in [Ca]i synergized with flagellin or P. aeruginosa, and buffering or reducing [Ca]i reduced these responses. Thus [Ca]i plays an important regulatory role in P. aeruginosa- or flagellin-activated innate immune responses in airway epithelia. Dose-dependent responses indicated that flagellin-ATP synergism occurred most prominently at ATP concentrations ([ATP]) > 10 μM and [flagellin] >10−8 g/ml and during steady increases rather than oscillations in [Ca]i.
- innate immune response
- 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid
- cystic fibrosis
during bacterial infections of the lungs the airway epithelial cells play a central role in activating the innate immune response, i.e., release of proinflammatory cytokines and subsequent recruitment of neutrophils to fight the infection. These inflammatory reactions are particularly pronounced in the disease cystic fibrosis (CF), in which there is usually a large accumulation of Pseudomonas aeruginosa in the mucus and airway surface liquid. It appears that the initial steps in activating the epithelial cells involve P. aeruginosa releasing monomeric flagellin that binds to a receptor in the airway epithelial cells [Toll-like receptor (TLR)-2 and -5 and asialoGM1 have all been proposed] followed by activation of second messengers, NF-κB migration to the nucleus, and upregulation of multiple cytokine and other proinflammatory genes (1, 7, 17, 28, 35, 39).
In addition to the TLR-related signaling pathway, a number of experiments have implicated Ca2+ in the activation of NF-κB and inflammatory signaling responses to P. aeruginosa. Addition of intact P. aeruginosa, flagellin, or anti-asialoGM1 antibodies to some epithelial cell lines (HAEo−, 16 HBE, HM3, and NCIH292) elicited increases in cytosolic Ca2+ concentration ([Ca]i) and subsequent activation of Src, Ras, ERK1/2, and NF-κB, resulting in increased expression of MUC2 and IL-8 (1, 22, 23, 28). Increased ERK-NF-κB signaling in response to P. aeruginosa or flagellin was mimicked by thapsigargin, the Ca2+-ATPase/sarco(endo)plasmic reticulum Ca2+-ATPase pump blocker that increases [Ca]i in cells, and blocked by the cellular Ca2+ buffer BAPTA-AM (28). Based partly on the fact that flagellin triggered release of ATP and apyrase (hydrolyzes ATP) blocked the activating effects of flagellin, it has been proposed (22, 23) that flagellin interactions with asialoGM1 induced the release of ATP, which activated purinergic receptors (P2Y2) and downstream Ca2+ signaling that was critical for the TLR-mediated response. Recent experiments on primary airway epithelia have shown that [Ca]i-elevating agonists like bradykinin and ATP also increased cytokine expression and secretion (30, 31).
However, there is also evidence indicating that elevations of [Ca]i are not involved in activating the innate immune responses triggered by P. aeruginosa. Whereas flagellin is critical for P. aeruginosa activation of inflammatory signaling in airway epithelial cells, likely through effects on TLRs (34, 39), TLRs are not known to trigger Ca2+ signaling (3). In addition, recent studies have shown for NCIH292 cells that ATP-induced increases in [Ca]i alone may be insufficient to activate innate immune response signaling (22). Furthermore, very recent experiments have shown that apical flagellin reduces Na+ absorption by murine airway epithelial cells without affecting [Ca]i (20)
The main goal of this study was to test the role of [Ca]i in controlling activation of innate immune responses to P. aeruginosa flagellin, which is a necessary and sufficient stimulus for activating an innate immune response in primary airway epithelial cells (35, 39). We also tested effects of the purinergic agonist ATP, which raises [Ca]i in airway epithelia (26), and the Ca2+-ATPase blocker thapsigargin, which leads to loss of endoplasmic reticulum (ER) Ca2+ and activation of plasma membrane Ca2+ channels (5), to test for possible additive or synergistic effects with flagellin that might be mediated through changes in [Ca]i. We measured [Ca]i, NF-κB activation, and IL-8 secretion in both the human nasal cell line JME/CF15 (expresses ΔF508CFTR) (16) and the lung gland, serouslike cell Calu-3 (expresses wild-type CFTR) (21, 33) during treatment with flagellin, ATP, and thapsigargin and then during conditions designed to buffer or lower [Ca]i (BAPTA, the cellular Ca2+ chelator, and Ca2+-free solutions). The comparative roles of NF-κB vs. AP-1 or CCAAT/enhancer-binding protein (C/EBPβ)/unclear factor for IL-6 (NF-IL6)-related signaling in controlling IL-8 secretion during these treatments were established with a luciferase reporter driven by full-length IL-8 promoter or constructs in which NF-κB, AP-1, or NF-IL6 sites were mutated.
MATERIALS AND METHODS
Unless otherwise specified, reagents and chemicals were obtained from Sigma (St. Louis, MO). Thapsigargin was dissolved in DMSO at 0.1–1.0 mM and then dissolved into solutions at 0.1–1.0 μM; these concentrations yielded similar effects on cellular functions. Stock solutions of ATP (100 mM) were diluted into the medium at concentrations indicated below.
CF airway JME cells (also called CF15; see Ref. 16), a continuous SV40 large T antigen-transformed human nasal epithelial cell line homozygous for ΔF508 CFTR, were cultured in Dulbecco's modified Eagle's medium-F-12 medium supplemented with 10% FBS, 2 mM l-glutamine, 1% penicillin-streptomycin, 10 ng/ml EGF, 1 μM hydrocortisone, 5 μg/ml insulin, 5 μg/ml transferrin, 30 nM triiodothyronine, 180 μM adenine, and 5.5 μM epinephrine. Calu-3 cells, a human gland epithelial cell line homozygous for wild-type CFTR (21, 33), were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 2 mM l-glutamine, and 1% penicillin-streptomycin. For most experiments, cells were passaged at 1:5–1:10 dilution, and the remaining cell suspension was seeded directly onto 25-mm diameter cover glasses or onto a 24-well or 12-well tissue culture plate (BD Falcon, Bedford, MA). In some experiments, cells were passaged onto either 1.0- or 4.2-cm2 Transwell membranes (0.4-μm pore size; BD Falcon) and then grown until cells formed confluent monolayers. Control experiments were performed in Ussing chambers to ensure that filter-grown cells attained confluence and were polarized. JME cells on filters had transepithelial resistances >200 Ω·cm2 and exhibited amiloride-sensitive, serosa-positive currents of 10–25 μA/cm2, consistent with Na+ absorption through epithelial Na+ channels. Forskolin treatment increased this apparent Na+ absorption but did not increase anion currents, consistent with their CF genotype. Calu-3 cells had transepithelial resistance >500 Ω·cm2 and responded to apical ATP (100 μM) with brief increases in currents of 15–30 μA/cm2 that were blocked by the apical CFTR blocker CFTRinh-172, consistent with purinergic stimulation of apical CFTR. Basolateral ATP elicited larger currents, as expected from experiments on primary cells (30, 31). Thus both JME and Calu-3 cells exhibited polarized responses consistent with previous studies.
P. aeruginosa and flagellin.
P. aeruginosa strains PAK and PAO1 were grown overnight in Luria-Bertani culture medium. Before experiments, bacteria were washed three times with PBS, resuspended in medium minus antibiotics and supplements at a concentration of 108 cfu/ml (OD600 = 0.1), and then diluted into the medium for experimentation as required. P. aeruginosa flagellin (10−3 g/ml in solution containing 10 mM phosphate buffer, pH 7.4, 140 mM NaCl and 3 mM KCl) (Inotek, Beverly, MA) was stored at −20°C and diluted from the stock into the incubation medium at stated concentrations. This solution was vortexed vigorously and heated to 37°C before being added to the solutions to ensure dispersal as monomers. As described by Inotek, recombinant flagellin is expressed with tags in Escherichia coli and purified to >95% homogeneity by SDS-PAGE. Previous experiments showed that LPS contamination of this preparation is small and cannot account for effects of flagellin to activate NF-κB and IL-8 secretion (35).
NF-κB-luciferase adenovirus and NF-κB activation assays.
A recombinant adenoviral vector expressing a luciferase reporter gene driven by NF-κB transcriptional activation (adv-NF-κBluc) was used for studies to determine effects of either P. aeruginosa or flagellin, ATP, and/or thapsigargin as described previously (14, 35). This vector contained the luciferase gene driven by four tandem copies of the NF-κB consensus sequence (32). Recombinant adenoviral stocks were generated as previously described (32) and were stored in 10 mM Tris with 20% glycerol at −80°C. The virus was added to JME or Calu-3 cells at 100 multiplicity of infection (MOI) and returned to the incubator for 48 h. Control experiments with a β-galactosidase-expressing adenovirus showed that this infection protocol generated expression in 75–100% of the cells (not shown). Cells were then washed three times to remove viruses and exposed to the various agonists for 4 h. Cells were then washed and processed with the luciferase assay system with Reporter Lysis Buffer (Promega, Madison, WI) to measure NF-κB-mediated transcriptional induction according to the manufacturer's protocol. Measurements of luciferase activity (relative light units) were performed in triplicate for each sample and normalized to the protein concentration (Bradford assay). These averages were then expressed relative to the average control value in the epithelial cells, which was set equal to 1.0.
Enzyme-linked immunosorbent assay of IL-8 secretion.
Samples were collected either from plastic wells in which epithelial cells were grown or from the basal chamber of cells grown on filters. Samples from control or treated cells were collected, cleared of any P. aeruginosa or cellular debris by centrifugation (5 min, 1,000 g), stored at −20°C until use, and then thawed, diluted 1:100 or 1:200 in 100 μl of Assay Diluent (BD Pharmingen, San Diego, CA), run in triplicate per the manufacturer's protocol (OptEIA Human IL8 Set, BD Pharmingen), and read at 450 nm with an ELX808 Ultra Microplate Reader (Bio-Tek Instruments, Winooski, VT). Averages of the triplicates are reported.
IL-8 promoter mutation analysis.
As described elsewhere (4), the 127-bp upstream region of the IL-8 transcriptional start site contains the key and well-characterized binding sites for the transcription factors AP-1, C/EBPβ/NF-IL6, and NF-κB. This region was amplified from HeLa cell genomic DNA isolated with TRIzol (Invitrogen) according to the manufacturer's instructions. Transcription factor binding sites in the IL-8 promoter were chosen to be analyzed based on their previously described location (2, 14, 25, 27). The 127-bp IL-8 promoter region was amplified through PCR from HeLa cell DNA with the following primer sequences: “IL-8 +7”: 5′-GCT ACT AGC TAG CAT GGA GTG CTC CGG TG-3′ and “IL-8 −127”: 5′-CGC GAG CTC GAT GAC TCA GGT TTG CC-3′. The NheI restriction endonuclease recognition sequence, 5′-GCT ACT AGC TAG C-3′, was incorporated into the 5′ end of the IL-8 +7 primer, and the SacI restriction endonuclease recognition sequence, 5′-CGC GAG CTC-3′, was incorporated into the 5′ end of the IL-8 −127 primer to allow the amplified product to be digested with the appropriate enzymes and ligated (T4 DNA ligase, Promega) into the pGL3 Basic plasmid vector (Promega). The promoter sequence was inserted upstream of the luciferase reporter gene in the vector pGL3 Basic and was verified to contain the correct insert by DNA sequencing.
To make the sequence changes in the AP-1 transcription factor binding site, the 127-bp region was amplified with the primer 5′-GAG CTC GAT GgC Ttg GGT TTG CCC TGA GGG GAT-3′ (lowercase letters indicate location of base changes) instead of the IL-8 −127 primer and inserted into the pGL3 Basic vector as described above. For NF-κB and C/EBPβ/NF-IL6 binding site mutation the Stratagene Quikchange site-directed mutagenesis kit was used. Both NF-κB and NF-IL6 binding site mutations were made according to Wu et al. (38). Site-directed mutagenesis was performed in two rounds to make the site changes with the following primers: NF-κB 1st round: F 5′-GGG CCA TCA GTT GCA AAT CGT taA ATT TCC TCT GAC ATA ATG-3′, R 5′-CAT TAT GTC AGA GGA AAT Tta ACG ATT TGC AAC TGA TGG CCC-3′; NF-κB 2nd round: F 5′-GGG CCA TCA GTT GCA AAT CGT TAA cTT TCC TCT GAC ATA ATG-3′, R 5′-CAT TAT GTC AGA GGA AAg TTA ACG ATT TGC AAC TGA TGG CCC-3′; NF-IL6/C/EBPβ 1st round: F 5′-GAG GGG ATG GGC CAT CAG cTa CAA ATC GTG GAA TTT CCT CT-3′, R 5′- AGA GGA AAT TCC ACG ATT TGt AgC TGA TGG CCC ATC CCC TC-3′; NF-IL6/C/EBPβ 2nd round, F 5′-GAG GGG ATG GGC CAT CAG CTA CgA gTC GTG GAA TTT CCT CT-3′, R 5′-AGA GGA AAT TCC ACG AcT cGT AGC TGA TGG CCC ATC CCC TC-3′. All transcription factor binding site changes were verified by DNA sequencing. Activation of the full-length IL-8 promoter and of the mutated versions (AP-1 mut, NF-IL6 mut, and NF-κB mut—see Table 1) was then assessed by transfecting JME cells with the IL-8 promoter-firefly luciferase plasmid along with a second plasmid expressing Renilla luciferase.
JME cells grown to 80–100% confluence were passaged and seeded onto the 24-well plate at a 1:3 dilution so that cells reached 50–60% confluence after overnight incubation. On the following day cells on the plate were cotransfected with the firefly and Renilla luciferase vectors, using Effectene Transfection Reagent (Qiagen) according to the manufacturer's protocol. Briefly, cells were incubated in the mixture of plasmid, transfection reagent, and medium for 12–18 h, and then the mixture was removed and replaced by regular medium. Cells were grown for a further 48 h before the experiment. Cells were treated with flagellin (10−7 g/ml), flagellin + ATP (100 μM), or flagellin + thapsigargin (1 μM) for 4 h, followed by washing and processing with the luciferase assay system. Firefly and Renilla luciferase expression were determined by the Dual-Luciferase Reporter Assay System (Promega, WI) according to the manufacturer's instructions. Relative luciferase activity was calculated for each sample by normalizing firefly luciferase readings with the Renilla luciferase readings.
Measurement of [Ca]i.
Cells grown on cover glasses or on filters were incubated with the original growth medium in sealed tissue culture plastic ware containing 1–10 μM fura-2 AM for 40–60 min at room temperature and then washed three times with Ringer solution to remove the extra dye. One to three micromolar fura-2 AM was used to load nonconfluent cells, whereas 5–10 μM fura-2 AM was used to load confluent cells, including those grown on filters. Similar loading and fura-2 responses were obtained with Ringer fura-2-loaded cells on the cover glasses or on filters mounted onto a chamber on the stage of the imaging microscope and maintained at room temperature or at 37°C. There was no significant difference in responses to agonists at the two temperatures.
Treatments with agonists were made by diluting stock solutions 1,000× into Ringer solution at the stated concentrations. Fluorescence ratio imaging measurements of [Ca]i were performed with methods that were reported previously (15, 24). Briefly, a Nikon Diaphot inverted microscope was used with a 40× Neofluar [1.4 numerical aperture (NA)] or a long-working-distance water immersion 40× (0.75 NA) lens. A charge-coupled device camera collected emission (>510 nm) images during alternate excitation at 350 ± 5 and 380 ± 5 nm with a filter wheel (Lambda-10, Sutter Instruments, Novato, CA). Axon Imaging Workbench 4.0 (Axon Instruments, Foster City, CA) controlled both filters and collection of data. Calibration of fura-2 signals was performed by calculation according to the equation presented by Grynkiewicz et al. (11): [Ca]i = Kd (Fmax/Fmin)[(R − Rmin)/(Rmax − R)], where Rmin is the ratio of fluorescence intensities at 350 and 380 nm obtained at zero [Ca]i, Rmax is the ratio at saturating [Ca]i, R is the measured ratio, Kd is the dissociation constant for fura-2, and Fmin and Fmax are the fluorescence intensities at 380 nm minus and plus calcium, respectively. The apparent Kd for fura-2 used in all calibrations was 224 nM (11). In situ calibration of fura-2 was carried out by treating the cells with ionomycin (10 μM) and then perfusing the cells sequentially with a Ca2+-free external solution to determine Rmin and then with a solution containing 2 mM Ca2+ to determine Rmax. All images were corrected for background (region without cells).
For measurements of NF-κB activation, IL-8 secretion, and IL-8 promoter activity, epithelial cells were incubated in Eagle's minimum essential medium (MEM, Mediatech) or in a Ca2+-free MEM supplemented with 2 mM l-glutamine, 1 mM Mg2SO4, and 0.5 mM Na2HPO4. Cells were washed twice with MEM or Ca2+-free MEM before experiments started, and treatment was performed by diluting stock solutions of reagents into corresponding media. In experiments to measure [Ca]i, epithelial cells were incubated in solutions containing (mM) 145 NaCl, 1.2 MgSO4, 2 CaCl2, 2.4 K2HPO4, 0.6 KH2PO4, 10 HEPES, and 10 glucose (pH 7.4) or in growth medium in which phenolphthalein had been removed. Ca2+-free Ringer solution was composed of the same solution without added Ca2+. Measurements of [Ca]i in response to flagellin, ATP, and thapsigargin yielded similar results whether Ringer or MEM was used.
Unpaired or paired t-tests were used to compare groups and effects, depending on the experiments (StatView, Abacus Concepts, Berkeley, CA). P < 0.05 was considered significant. Data are presented as averages ± SD or including values from all individual experiments; n refers to the number of experiments.
Flagellin, P. aeruginosa, ATP, and thapsigargin on [Ca]i and NF-κB activation.
Previous studies testing the role of [Ca]i in P. aeruginosa or flagellin activation of innate immune response signaling have tested the effects of bacteria, flagellin, ATP, or thapsigargin alone, but not in combination. We were interested to correlate changes in [Ca]i and in NF-κB activation with multiple stimulation regimes, including use of a range of concentrations to test the potential role of [Ca]i in controlling innate immune responses in response to these stimulants. Measurements of [Ca]i, NF-κB activation, and IL-8 secretion were performed on JME cells grown on cover glasses or in cell culture wells or filters. For Calu-3 cells, [Ca]i was measured on cells grown on cover glasses, and NF-κB and IL-8 responses were measured in cells grown in cell culture wells or on filters. [Ca]i measurements included 20–30 cells per microscope field.
Typical [Ca]i records are shown for JME cells in Fig. 1, A–F and H, and for Calu-3 cells in Fig. 2, A and B. Baseline [Ca]i in the absence of any agonists was 50–75 nM in JME cells and somewhat higher in Calu-3 cells. JME cells responded to 1–10 μM ATP with rapid increases in [Ca]i (∼150–200 nM) that were followed by smaller oscillations in [Ca]i that increased in frequency with increasing [ATP] (Fig. 1A). In contrast, a high dose (100 μM) of ATP caused rapid, large (400–600 nM) increases in [Ca]i that did not oscillate in both JME (Fig. 1B) and Calu-3 (Fig. 2A) cells. In JME cells the initial, rapid peak in [Ca]i in response to 100 μM ATP was followed by a second, smaller and slower increase in [Ca]i (Fig. 1B) that was not present in Calu-3 cells (Fig. 2A). Further addition of thapsigargin (1 μM) to either JME or Calu-3 cells caused further increases in [Ca]i that often reached micromolar concentrations. Addition of thapsigargin alone similarly caused large increases in [Ca]i in JME cells (Fig. 1C).
In contrast to these vigorous [Ca]i responses to ATP and thapsigargin, flagellin (10−7 g/ml, a dose that yielded robust increases in NF-κB activation; Fig. 1, G and I) routinely had no effect on [Ca]i in JME cells during short-term (Fig. 1D) or long-term (Fig. 1E) incubations and also for JME cells grown on filters (Fig. 1H).
Similar experiments were performed on Calu-3 cells, although in this case the experiments were restricted to cells on cover glasses since filter-grown cells did not load enough fura-2 to allow reliable [Ca]i measurements. Results were similar to those obtained from JME cells—10−7 g/ml flagellin did not alter [Ca]i, while ATP and thapsigargin had typical effects to raise [Ca]i.
Previous experiments had shown that the noncytotoxic P. aeruginosa strain PAO1 had no effect on [Ca]i in Calu-3 cells (15). We tested the effects of another noncytotoxic P. aeruginosa strain, PAK, on [Ca]i in JME cells. There was no effect of 106-108 cfu/ml PAK on [Ca]i in two experiments (total of >50 cells examined), although in one experiment, 2 of ∼30 cells in the field exhibited very small increases in [Ca]i (20–100 nM)—these 2 responding cells are shown in Fig. 1E, with responses of several nonresponding cells shown for comparison. All these cells responded typically to ATP and thapsigargin (not shown).
Flagellin, ATP, and thapsigargin were also tested for their effects on NF-κB activation. ATP and thapsigargin both elicited small, but significant increases in NF-κB activity (measured from luciferase) in JME cells grown both in tissue culture wells (Fig. 1G) and on filters (Fig. 1I). Flagellin elicited much larger activation of NF-κB than ATP or thapsigargin (Fig. 1, G and I). Furthermore, responses to flagellin + ATP and flagellin + thapsigargin were larger than responses to flagellin, and there were larger than additive, i.e., synergistic, effects of ATP and thapsigargin to increase NF-κB activation in the presence of flagellin. Comparison of Fig. 1, G and I, shows that the magnitudes of the responses were only slightly different when flagellin, ATP, and thapsigargin were added to the apical surface of JME cells grown on filters (Fig. 1I) compared with additions of the stimulants to both sides of cells grown in culture wells (Fig. 1G). This result is consistent with the idea that agonist-triggered signaling was qualitatively similar when triggered from the apical or apical and basolateral surfaces of the cells. Previous experiments have shown that primary cultures of airway epithelial cells are similarly responsive to flagellin addition to either the apical or basolateral surface of the cells (35). Given the similar responses of JME cells to agonists added to the apical (filters) vs. apical and basolateral sides (cells in culture wells), we performed most of the rest of the experiments with cells grown in culture wells.
Similar experiments to assay NF-κB activity were performed on Calu-3 cells grown on filters to ensure polarized responses. As shown in Fig. 2C, neither thapsigargin nor ATP activated NF-κB or IL-8 secretion. Apical flagellin elicited much larger activation of NF-κB than apical ATP or thapsigargin (Fig. 2C). There were similar larger effects of flagellin + thapsigargin vs. flagellin on NF-κB activation in Calu-3 cells, but there were no significant differences in responses to flagellin vs. flagellin + ATP. Thus both ATP and thapsigargin activated NF-κB in JME cells, and these stimulatory effects became synergistic in the presence of flagellin. In Calu-3 cells, thapsigargin and ATP were stimulatory only in combination with flagellin, again indicating a synergistic effect of the [Ca]i-raising agonists on flagellin-triggered NF-κB response.
Concentration-dependent effects of flagellin, P. aeruginosa, and ATP on NF-κB.
Previous experiments investigating the effects of ATP, P. aeruginosa, and flagellin on airway epithelial cells have generally tested only single concentrations of the agonists (see, e.g., Refs. 1, 22, 23, 28). The ability of ATP to augment flagellin responses was investigated here in dose-response studies. Adv-NF-κB-luc-infected JME cells were treated with 10−10-10−5 g/ml flagellin or with flagellin + 100 μM ATP. This range of flagellin concentrations was chosen to compare to the dose dependence of flagellin activation of cloned mouse TLR-5 (34). Flagellin alone elicited dose-dependent increases in NF-κB activity beginning at 10−10 g/ml, with steadily increasing activation at higher concentrations and near saturation at 10−5 g/ml (Fig. 3). This concentration dependence of the response to P. aeruginosa flagellin was very similar to that exhibited by mouse TLR-5-expressing human embryonic kidney (HEK) cells (34). Data in Fig. 3 also showed, similar to experiments in Fig. 1, G and I, that ATP alone (i.e., with [flagellin] = 0) caused a small activation of NF-κB, but this stimulatory effect was larger in the presence of flagellin, particularly at high [flagellin], an effect that was synergistic over the entire range of [flagellin].
Similar experiments were performed to test the concentration dependence of ATP's effects on NF-κB activation, alone and in the presence of either an intermediate concentration of flagellin (10−7 g/ml; Fig. 4A) or P. aeruginosa strain PAK (106 cfu/ml; Fig. 4B). We also tested 0.1 μM thapsigargin. Although 1, 5, 10, and 100 μM ATP and 0.1 μM thapsigargin all triggered increases in [Ca]i in JME cells (Fig. 1, A–C), the threshold for ATP to activate NF-κB was 10 μM, with larger NF-κB activation occurring with 100 μM ATP and 1 μM thapsigargin. The stimulatory effects of ATP and thapsigargin on NF-κB activation were synergistic in the presence of either flagellin (Fig. 4A) or PAK (Fig. 4B).
Effects of BAPTA-AM and Ca2+-free solution on [Ca]i, NF-κB, and IL-8 triggered by flagellin, P. aeruginosa, ATP, and thapsigargin.
Although previous studies have tested effects of BAPTA or Ca2+-free solutions on innate immune response signaling (1, 22, 23, 28), we wanted also to measure [Ca]i under these conditions to make a quantitative assessment of the role of [Ca]i in the activation of innate immune responses. We therefore measured [Ca]i, NF-κB activation, and IL-8 expression and secretion in cells that had been treated with BAPTA-AM or Ca2+-free solutions. BAPTA-AM prevented [Ca]i responses to ATP and slowed and blunted responses to thapsigargin (Fig. 5A vs. Fig. 1B). Similar results were obtained from Calu-3 cells (Fig. 5B vs. Fig. 2A). The fact that BAPTA abolished [Ca]i responses to ATP but only blunted and slowed responses to thapsigargin is predicted based on their actions: ATP elicited a large transient increase in [Ca]i (likely resulting from partial release of Ca2+ from ER) followed by a smaller, sustained increase in [Ca]i, likely due to opening of store-operated Ca2+ channels (SOCs) in the plasma membrane (5), while thapsigargin caused larger and more persistent increases in [Ca]i, likely resulting from thapsigargin completely releasing Ca2+ from the ER and persistent opening of the SOCs, which permits continued Ca2+ entry into the cells from the essentially infinite volume of the extracellular fluid to overwhelm BAPTA's buffering of cytosolic Ca2+.
Effects of nominally Ca2+-free solutions on [Ca]i were also investigated in JME cells. Brief treatment (30 s) had no effect on basal [Ca]i or on initial response to ATP, but the secondary increase in [Ca]i was abolished (compare Fig. 1B and Fig. 5C). These results were consistent with the idea that the initial rise in [Ca]i resulted from the effect of ATP to release Ca2+ from the ER store, while the second, delayed rise in [Ca]i resulted from Ca2+ entry into the cells from outside. Ca2+-free solution also reduced the [Ca]i response to thapsigargin, as predicted from the effect to activate store-operated channels in the plasma membrane. Longer treatment (25 min) with Ca2+-free solution had even more pronounced inhibitory effects on [Ca]i responses of JME cells to ATP and thapsigargin (Fig. 5D).
Effects of 1-h treatment with BAPTA-AM on NF-κB activation by ATP, thapsigargin, and flagellin are shown in Fig. 6A for JME cells and in Fig. 6B for Calu-3 cells. Control responses to ATP, thapsigargin, flagellin, flagellin + ATP, and flagellin + thapsigargin were similar to those presented above (Figs. 1F and 2C). For JME cells (Fig. 6A), BAPTA reduced NF-κB activation in response to ATP and thapsigargin to roughly control levels. BAPTA also reduced NF-κB activation elicited by flagellin, even though flagellin did not increase [Ca]i. BAPTA also reduced NF-κB activation by flagellin + ATP to roughly the level elicited by flagellin treatment, while BAPTA effects on flagellin + thapsigargin activation were less pronounced, perhaps resulting from BAPTA's incomplete capability to buffer [Ca]i in the presence of thapsigargin (see Fig. 5A). In contrast to the potent effects of 25 μM BAPTA-AM to inhibit NF-κB activation, 25 μM BCECF-AM did not alter the NF-κB responses of JME cells to any of the agonists (not shown). This result showed that the inhibitory effects of BAPTA-AM were not due to toxic effects of hydrolysis products formed during the entry and cleavage of the membrane-permeant probes into the cells (24). Thus effects of BAPTA-AM likely result from the Ca2+-buffering properties of BAPTA. BAPTA similarly reduced NF-κB activation in response to ATP, thapsigargin, flagellin (even though flagellin did not increase [Ca]i in these cells either), flagellin + ATP, and flagellin + thapsigargin in Calu-3 cells (Fig. 6B). Inhibitory effects of BAPTA were somewhat more potently expressed in Calu-3 cells than in JME cells (compare Fig. 6, A and B).
Ca2+-free solutions were also used to test the role of [Ca]i in activating NF-κB. Cells had their normal medium removed by three washes with a Ca2+-free medium, followed by incubation in the Ca2+-free medium and treatment with ATP, thapsigargin, and/or flagellin. This condition approximated the condition shown in Fig. 5D, where it can be seen that [Ca]i responses to ATP and thapsigargin were both reduced substantially. Results from paired experiments are shown in Fig. 7. Although there were variable responses in each condition, Ca2+-free solutions reduced responses to ATP and thapsigargin. Ca2+-free solution reduced NF-κB activation induced by flagellin in four of five experiments, by flagellin + ATP in four of five experiments, and by flagellin + thapsigargin in five of five experiments. It was also noted that the inhibitory effects of Ca2+-free solution were absent in experiments in which stimulation by the agonists was relatively low, consistent with the idea that flagellin elicited one level of activation and the [Ca]i agonists were augmenting this response.
Effects of flagellin, ATP, and thapsigargin on IL-8 secretion and IL-8 promoter (control and mutated AP-1, NF-IL6, and NF-κB sites).
IL-8 expression and secretion are controlled by signaling leading to activation of NF-κB as well as by signaling leading to activation of the other transcription factors, AP-1 and NF-IL6/C/EBPβ (3, 13, 25). The role of Ca2+ in controlling IL-8 secretion was tested by measuring IL-8 secretion into the medium during treatment with ATP, thapsigargin, and/or flagellin under control conditions and in BAPTA-treated cells (Fig. 8A, JME cells; Fig. 8B, Calu-3 cells). Cells were infected with adenovirus expressing NF-κB-luciferase for 48 h, followed by treatment of one set of cells with 25 μM BAPTA-AM for 1 h and then washing and replacement with normal medium. Paired cells were not treated with BAPTA. Data in Fig. 8 show for the control cells that ATP and thapsigargin both increased IL-8 secretion, but these effects were small compared with the stimulation triggered by flagellin, and responses to flagellin + ATP or to flagellin + thapsigargin were larger than to flagellin alone. BAPTA reduced IL-8 secretion in response to all the agonists (Fig. 8).
Measurements of IL-8 promoter activity were also used to test the stimulatory effects of ATP, thapsigargin, and flagellin on IL-8 gene expression. Mutation of specific regions of the promoter were then used to test the relative roles of the AP-1 vs. NF-IL6/C/EBPβ vs. NF-κB sites in controlling IL-8 gene expression by the agonists. As summarized in Fig. 9 and Table 2, flagellin increased IL-8 promoter activity by ∼10-fold, and this was further increased in the presence of flagellin + ATP and flagellin + thapsigargin. Responses of the cells transfected with mutated IL-8 promoter constructs showed reductions in activity when any of the sites were mutated, and inhibitory effects occurred in an order indicating that the most important regulator of the IL-8 promoter was the NF-κB-binding site, followed by the NF-IL6/C/EBPβ and AP-1 sites. Thus there was significant stimulation by flagellin compared with no-treatment control for the full-length IL-8 promoter and also for each of the mutated versions.
These data also showed that, compared with flagellin treatment, ATP + flagellin and thapsigargin + flagellin increased IL-8 promoter activity of the AP-1 and NF-IL6 mutants but not the NF-κB mutant. These results showed that multiple signaling pathways may mediate the IL-8 responses to flagellin and the [Ca]i agonists, but that activation of the NF-κB binding site is most important both for activation by flagellin and also for the stimulatory effects of flagellin + ATP and flagellin + thapsigargin vs. flagellin.
Both JME and Calu-3 cells responded to monomeric flagellin by activating NF-κB and IL-8 secretion. JME cells also responded to flagellin by activating the full-length IL-8 promoter. Previous work has shown that flagellin expression by P. aeruginosa is required to activate innate immune responses in a number of airway epithelia, including JME cells (14, 39). The concentration dependence of flagellin's stimulation of NF-κB in JME cells was very similar to that observed previously for activation of mouse TLR-5-transfected HEK293 cells (34), consistent with the idea that TLR-5 is mediating these responses of airway epithelial cells to monomeric flagellin (and also to P. aeruginosa, see Refs. 14, 35, 39).
Flagellin-induced activations of NF-κB and IL-8 occurred in the absence of any detectable increase in [Ca]i in either JME or Calu-3 cells, even over a 1-h incubation. Similarly, P. aeruginosa strain PAK activated NF-κB in JME cells without increasing [Ca]i, except for a very small response in 2 of >60 cells examined. These results were consistent with previous experiments showing that P. aeruginosa strain PAO1 had no effect on [Ca]i in Calu-3 cells (15), although PAO1 potently activated NF-κB and IL-8 secretion in these cells (J. Tseng, Z. Fu, and T. E. Machen, unpublished observations). Recent experiments have similarly shown for both CFDE (CF) and 16 HBE14o− (non-CF) cells that apical application of flagellin had no effect on [Ca]i (20). Results showing that flagellin and P. aeruginosa had no effect on [Ca]i are consistent with previous (9) and more recent (20) experiments showing that P. aeruginosa did not trigger increases in transepithelial or patch-clamp currents of primary airway epithelial cells, as would be expected if the bacteria were increasing [Ca]i and activating Ca2+-activated Cl− channels and transepithelial Cl− secretion (see, e.g., Ref. 33). Thus it appears that global changes in [Ca]i were not required for apical P. aeruginosa or flagellin to activate innate immune responses by airway epithelial cells. These results contrast with previous experiments showing that flagellin or P. aeruginosa caused brief increases in [Ca]i in HM3, 1HAEo−, 16 HBE, and primary nasal epithelial cells (1, 23, 28). Recent experiments on NCIH292 cells showed that flagellin triggered responses in a minority (∼30%) of the cells, although neither the magnitudes nor the time courses of the responses in this 30% of the cells were quantitated (22).
The reason for the discrepancy between these previous experiments and ours is not apparent but may have arisen from technical differences or in differences in cell culture methods and/or in cell differentiation. One possibility is that there may be differences in flagellin-triggered ATP release by the airway epithelial cells, with HM3, 1HAEo−, 16 HBE, and NCIH cells releasing ATP (and then triggering [Ca]i) in response to flagellin-activation of TLR-5 (22) while JME, Calu-3, and primary cells do not release ATP in response to flagellin and therefore do not increase [Ca]i. Consistent with this idea, apyrase, which cleaves ATP and inhibits flagellin-triggered inflammatory signaling in NCIH cells (22), had no effect on flagellin-triggered NF-κB activation in JME cells (Z. Fu and T. E. Machen, unpublished observations). A related possibility is that cellular differentiation may affect flagellin- or P. aeruginosa-triggered [Ca]i responses differently in JME and Calu-3 cells vs. HM3, 1HAEo−, 16 HBE, and NCIH cells. Differentiation-dependent effects on [Ca]i signaling have been noted previously in airway epithelia: 1 μM ATP induces [Ca]i oscillations in subconfluent bovine tracheal epithelial cells but not in confluent monolayers (18). Whatever the explanation for the different responses, it is clear that for JME and Calu-3 cells grown for 3–7 days in culture, flagellin triggered robust increases in NF-κB activity and IL-8 gene activity without affecting [Ca]i. This result indicated that increases in [Ca]i were not necessary for flagellin or P. aeruginosa to activate an innate immune response in differentiated airway epithelia.
Although increases in [Ca]i were not required to obtain flagellin-activation of NF-κB, other results indicated that increases in [Ca]i were sufficient to activate inflammatory signaling. Thus ATP and thapsigargin each increased [Ca]i and activated NF-κB and IL-8 secretion, and these changes were reduced by either BAPTA-AM or Ca2+-free solutions. In JME cells low [ATP] (1–10 μM) triggered [Ca]i oscillations similar to those reported previously (18) but little or no activation of NF-κB, while high [ATP] (100 μM) triggered both a rapid increase and a sustained elevation in [Ca]i as well as a larger activation of NF-κB, i.e., compared with the low [ATP]. These results may indicate that in epithelial cells sustained increases in [Ca]i are more potent than oscillatory [Ca]i in activating NF-κB, consistent with findings in lymphocytes (7).
In contrast to the relatively weak activation of innate immune responses in response to ATP- or thapsigargin-triggered increases in [Ca]i, there were larger effects of the [Ca]i-raising agonists when they were added in combination with P. aeruginosa or flagellin. This was seen clearly in the dose-dependent effects of flagellin on JME cells (Fig. 3), which showed small stimulations of NF-κB by ATP or flagellin and larger, synergistic stimulations when the cells were exposed to flagellin + ATP. These results were consistent with recent experiments showing that flagellin triggering of ERK and NF-κB signaling was enhanced by ATP released from the cells into the bathing medium (23). The synergism noted in the present experiments appeared to become most pronounced at the higher concentrations of both ATP and flagellin, indicating that quite large amounts of ATP will be required to be released from the epithelial cells (23) to attain the concentrations required to elicit the synergism noted here.
Similar synergistic effects of thapsigargin + flagellin (vs. flagellin alone) on NF-κB activation and IL-8 secretion were observed in both JME and Calu-3 cells, and these effects were reduced by BAPTA. Ca2+-free solution also reduced activation by these agonists in JME cells. These data indicated that increases in [Ca]i act synergistically with flagellin's normal signaling pathways (3) to activate innate immune responses, although the variable magnitudes of the inhibitory effects of BAPTA and Ca2+-free solutions in JME and Calu-3 cells on NF-κB vs. IL-8 secretion indicate that Ca2+ may have multiple and complex roles in the signaling pathways. This idea is consistent with recent work by McNamara et al. (22), who propose that flagellin-triggered release of ATP leads to increases in [Ca]i that synergize in TLR-5-activated signaling leading to NF-κB activation. We found that in general there were larger synergistic effects of thapsigargin + flagellin than with ATP + flagellin, although, as shown in Fig. 3, ATP and thapsigargin sometimes elicited similar synergistic effects in the presence of flagellin. These results may indicate that in general thapsigargin induced larger increases in NF-κB and IL-8 activation than ATP due to larger [Ca]i responses elicited by thapsigargin than ATP, but there may also be a critical stimulatory level of [Ca]i that is reached with ATP in some experiments, so further increase in [Ca]i with thapsigargin yielded no further activation of NF-κB. Synergistic effects among [Ca]i and NF-κB activation and IL-8 secretion may also explain why flagellin activation of NF-κB was reduced by BAPTA and Ca2+-free solutions even though flagellin did not increase [Ca]i. For example, flagellin-triggered NF-κB signaling may require specific levels of [Ca]i, and Ca2+-free solutions and, especially, BAPTA may reduce [Ca]i enough to inhibit these signaling pathways, although this does not explain the larger inhibitory effect of BAPTA vs. Ca2+-free treatment on flagellin stimulation. This also does not explain the fact that BAPTA inhibited activation in response to flagellin + thapsigargin even though BAPTA only slowed and slightly depressed the [Ca]i response to thapsigargin. A possibility to explain these apparent contradictions is that Ca2+-free treatment reduces the stimulatory effect of Ca2+, while BAPTA buffers both Ca2+ and other divalent cations that may be required for flagellin stimulation. In this regard, we note that BAPTA binds Zn2+, which could be involved as a cofactor in many signaling pathways, at least 10 times more avidly than it binds Ca2+ (11). Further work will be required to test this hypothesis.
The signaling following flagellin binding to TLR-5 and the activation of IL-8 expression and secretion appears to involve multiple pathways (22, 28, 39). Data obtained with the IL-8 promoter and mutants indicated that pathways leading to activation of AP-1, NF-IL6, and NF-κB sites were all important for responses to either flagellin or flagellin + [Ca]i agonists. These results were consistent with recent experiments showing that the p38 MAPK pathway (22, 40), which likely activates both the AP-1 and NF-IL6 sites, as well as the TLR-MyD88-TRAF-IκB pathway, which activates NF-κB (3), are both triggered by flagellin and P. aeruginosa activation of TLR-5. Mutating the NF-κB site had the most profound inhibitory effect on IL-8 promoter activity in response to flagellin, flagellin + ATP, or flagellin + thapsigargin. Mutating the NF-IL6 and AP-1 sites was less effective in preventing activation by flagellin. These data were consistent with previous work showing that the NF-κB site was the most important in activation of inflammatory signaling in a number of different cells during treatment with other bacteria—Helicobacter pylori (3) and Shigella flexneri (27)—or TNF-α + IL-1β (13).
In addition, the wild-type IL-8 promoter and also the AP-1 mutant and NF-IL6 mutant constructs exhibited larger stimulation by flagellin + ATP and flagellin + thapsigargin than stimulation by flagellin alone (Fig. 9). In contrast, the NF-κB mutant exhibited a small and similar activation by flagellin, flagellin + ATP and flagellin + thapsigargin. These results, combined with the other data showing activation of NF-κB by ATP and thapsigargin in the presence of flagellin (Figs. 3, 4, and 6), indicated that Ca2+-mediated synergism of flagellin stimulation of IL-8 expression and secretion was mediated through effects of [Ca]i on NF-κB activation, while [Ca]i effects on AP-1 and NF-IL6 sites may be less pronounced.
Flagellin-activated NF-κB and IL-8 responses were larger in JME cells than in Calu-3 cells (Fig. 1F vs. Fig. 2C and Fig. 8, A vs. B), consistent with the idea that CF cells exhibit hyperinflammatory signaling compared with non-CF cells (6, 8, 19). However, because the cells are not genetically matched, further experiments will be required to test this possibility. The present data do, however, offer potential insights into this question of the role of CFTR in inflammatory signaling. CF cells have hyperpolarized basolateral membrane potentials compared with non-CF cells (29, 36, 37), and this could have an impact on [Ca]i signaling and, as shown here, activation of innate immune responses. Because [Ca]i-raising agonists augment innate immune responses induced by flagellin, it might be expected that the innate immune responses to ATP + flagellin will be larger in CF cells than in non-CF cells, resulting from larger Ca2+ influx into these cells due to hyperpolarized basolateral membrane potential. According to this model, hyperinflammatory signaling in CF would occur only during the presence of both flagellin (or other NF-κB-activating agonists) and [Ca]i-raising agonists.
In conclusion, we propose that P. aeruginosa-triggered innate immune responses by airway epithelia, including secretion of IL-8, are initiated by release of flagellin monomers by the bacteria into the airway surface liquid. Flagellin activates airway epithelia by triggering TLR-5 signaling leading to activation of AP-1, NF-IL6/C/EBPβ, and NF-κB sites on the IL-8 promoter. The NF-κB site is critical for controlling IL-8 expression, while the AP-1 and NF-IL6 sites likely enhance responses induced by NF-κB binding (13). Increases in [Ca]i were not required for activation of NF-κB and IL-8 expression and secretion by the epithelial cells during treatment with P. aeruginosa or flagellin, and increases in [Ca]i served only a weak stimulatory role during activation by ATP or thapsigargin. However, the specific level of [Ca]i appeared to be an important regulator of flagellin-triggered inflammatory signaling. Thus ATP- or thapsigargin-induced increases in [Ca]i appeared to play an important, often synergistic, stimulatory role during simultaneous treatments with P. aeruginosa or flagellin. These responses were reduced by buffering (BAPTA) or reducing (Ca2+-free treatment) [Ca]i. A similar regulatory role for [Ca]i in flagellin-triggered inflammatory signaling has recently been proposed by McNamara et al. (22), although this apparently requires that flagellin-TLR-5 signaling releases ATP from cells (thereby activating [Ca]i). Our data further indicate that steady increases in [Ca]i synergize better than oscillations in [Ca]i during ATP or thapsigargin + flagellin activation of NF-κB and IL-8 secretion. Finally, it is predicted that the synergistic effects of ATP and other [Ca]i-raising agonists to augment activation by flagellin will be larger in CF cells than in non-CF cells, potentially contributing to hyperinflammation in CF airways.
This research was supported by grants from the National Institutes of Health (1RO1-DK-51799, T32-AI-007620, and 1R01-HL-071730), CF Foundation, California Tobacco-Related Disease Research program, and Cystic Fibrosis Research Inc.
We thank Kevin Hybiske, William Reenstra, and Christian Schwarzer for discussions and suggestions for improving the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society