Mapping the Transcriptional Machinery of the IL-8 Gene in Human Bronchial Epithelial Cells

Cystic fibrosis (CF) is a severe inherited disease caused by mutations of a gene encoding a chloride channel termed CF transmembrane conductance regulator (CFTR) (1). Although most of the CF patients are affected by multiple organ pathologies, lung disease is the major cause of morbidity and mortality in CF. In the airway tract, CFTR mutations lead to defective chloride secretion, excessive sodium reabsorption, and reduction of airway surface liquid volume and of mucociliary clearance, which favor, by still not completely defined mechanisms, the chronic bacterial colonization with Pseudomonas aeruginosa (2, 3). A hallmark of CF lung pathology is a hyperinflammatory condition, characterized by predominant infiltrates of polymorphonuclear neutrophils in bronchial lumina and increased expression of proinflammatory cytokines and chemokines, in particular the CXC chemokine IL-8 (4–8). The excessive inflammation has been related with the molecular defect of CF (for review, see Ref. 9). Moreover, it has been proposed that excessive proteolytic activity of massive polymorphonuclear neutrophil infiltrates within the bronchial lumina of CF patients cleaves the chemokine receptor CXCR1 expressed on the membrane of neutrophils and inactivates their bacterial killing capacity (10). Recently, it has been established that excessive IL-8 and IL-6 release from lung epithelial cells may also play a role in bacterial proliferation and adhesion, enhancing the amount of sialyl Lewis X carbohydrate and its derivatives localized in respiratory mucins of CF patients (11). Thus, hyperexpression of the chemokine IL-8, which mainly originates from the surface of epithelial cells lining the conductive airways, is presently considered a critical pharmacological target to reduce the excessive inflammation in CF lungs. However, a thorough understanding of the transcription machinery of the IL-8 gene in human bronchial epithelial cells is still uncomplete.

The interaction of P. aeruginosa with pattern recognition receptors (PRRs) expressed on the airway epithelial cells is a potent inducer of expression of IL-8. Among the PRRs involved in P. aeruginosa recognition, it has been reported that TLR5, in cooperation with TLR2 and/or asialo-GM1 receptor (ASGM1R), binds bacterial flagellum to transduce a MyD88-dependent signaling cascade activating protein kinases and a calcium signaling pathway, with the latter being mediated by the release of ATP into the extracellular milieu as a so-called “danger” signal (12–15). Even if it is known that TLR4 can bind the LPS of bacterial walls, generating a MyD88-dependent proinflammatory signaling, this pathway seems to be reduced in CF, although this remains controversial (16, 17). Activation of TLR2/4/5 and ASGM1R induces IL-8 expression mainly through the phosphorylation of three MAPKs, namely MAPK p38, ERK1/2, and JNK1/2/3. ERK1/2 and MAPK p38 can transduce signals to other substrates, such as the 90-kDa ribosomal S6 kinase (RSK) family. Among RSK components, RSK1 is the most represented form and its phosphorylation may occur downstream of both ERK and 3-phosphoinositide–dependent protein kinase-1 pathways, leading to activation of C/EBPβ (also known as NF-IL6), as documented in human hepatocytes, and/or to activation of CREB, as reported in human airway epithelial cells (18). Another member of the RSK family, the mitogen- and stress-activated kinase-2 (MSK2), has been reported to be a substrate of p38α, which is able to activate both CREB and AP-1 (19). The p38 effectors include also the small heat shock protein 27 (HSP27), which is involved in inflammatory processes, together with MK2 kinase, with both activating NF-κB (20). Additionally, HSP27 has been reported to be recruited in IL-1β– and TNF-α–dependent IL-8 gene transcription in HeLa cells (21).

These phosphorylations eventually activate transcription factors (TFs), first of all NF-κB (22–28). Lung inflammation in CF has been related to an increased NF-κB signaling causing the induction of IL-8 gene expression (22, 23, 29). CF airway epithelial cells exposed to proinflammatory stimuli, such as TNF-α and IL-1β, regulate IL-8 gene transcription through the activation of NF-κB, NF-IL6, AP-1, and CHOP (22, 23, 29, 30). Moreover, CREB has been proposed to be a TF involved in IL-8 gene transcription machinery in the U937 monocytic cell lines exposed to Helicobacter pilori VacA toxin (28), and it has been shown that CREB collaborates with NF-κB in CXC chemokine expression in human respiratory carcinoma cells (31). Additionally, it has been recently observed that CREB and NF-κB synergistically induce IL-6 but not IL-8 gene expression in astrocytes stimulated with TNF-α (32). Therefore, the complete understanding of all the different TFs specifically involved in P. aeruginosa-dependent IL-8 gene transcription in CF respiratory epithelial cells has not yet been reached.

To address this issue we applied in this study the TF “decoy” strategy (33, 34), based on oligodeoxynucleotides (ODNs) mimicking the consensus sequences for the major TFs identified within the proximal promoter region of the IL-8 gene in order to obtain a complete mapping of the TFs involved. In this paper, we report that the induction of IL-8 gene transcription by P. aeruginosa in human bronchial epithelial IB3-1 cells involves a cooperative activation of the MAPKs ERK and p38, as well as their substrate kinases RSK, MSK, and HSP27, and of the TFs NF-κB, NF-IL6, AP-1, CHOP, and CREB.

IB3-1 cells (LGC Promochem) are human bronchial epithelial cells immortalized with adeno12/SV40, derived from a CF patient with a mutant F508del/W1282X genotype (35). Cells were grown in LHC-8 basal medium (Biofluids, Rockville, MO) supplemented with 5% FBS. CuFi-1 cells, a gift from A. Klingelhutz, P. Karp, and J. Zabner (University of Iowa, Iowa City, IA), were derived from human bronchial epithelium from a patient with CF (CFTR DF-508/DF-508 mutant genotype) and had been transformed by a reverse transcriptase component of telomerase, hTERT, and human papillomavirus type 16, E6, and E7 genes. This cell line was grown on human placental collagen type IV (Sigma-Aldrich, St. Louis, MO)-coated flasks or plates in bronchial epithelial growth medium (Cambrex Bio Science, Walkersville, MD). Calu-3 cells were obtained from a human lung adenocarcinoma and derived from serous cells of proximal bronchial airways. Calu-3 cells were cultured in culture flasks or plates in DMEM containing 4.5 g/l glucose, supplemented with 10% FBS in a humidified atmosphere of 5% CO₂. All culture flasks and plates were coated with a solution containing 35 μg/ml bovine collagen (BD Biosciences, Franklin Lakes, NJ), 1 μg/ml BSA (Sigma-Aldrich), and 1 μg/ml human fibronectin (BD Biosciences). P. aeruginosa, PAO1 laboratory strain, was provided by A. Prince (Columbia University, New York, NY). Bacteria were grown in trypticase soy broth or agar (Difco, Detroit, MI).

Human rIL-1β and rTNF-α were purchased from Sigma-Aldrich. Pharmaceutical inhibitors AG1288 and GSK3 inhibitor II were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), KRIBB3, parthenolide, and SB203580 were from Sigma-Aldrich, and SL0101 and U1026 were from Tocris Bioscience (Ellisville, MO).

Cells were seeded in 6-cm Petri dishes at a density of 2.5 × 10⁶ cells to obtain 1 × 10⁷ cells for each array. Cells were starved in LHC-8 basal medium serum-free before stimulation with PAO1 (100 CFU/cell), human rIL-1β (10 ng/ml), or human rTNF-α (50 ng/ml) for 30 min. According to the manufacturer’s protocol, 200 μg cell lysate was incubated with each human phospho-MAPK array (R&D Systems, Minneapolis, MN). Arrays were exposed to chemoluminescent reagent, and nitrocellulose membranes were then exposed to x-ray films. Phospho-MAPK array spot signals developed on x-ray films were quantitated by scanning the film on a high-resolution transmission-mode scanner and analyzing the array image file using the image analysis software Digimizer (MedCalc Software, Mariakerke, Belgium).

Nuclear extracts were obtained from IB3-1 cells as previously described (36). IB3-1 cells were exposed to P. aeruginosa PAO1 strain (100 CFU/cell), IL-1β (10 ng/ml), or TNF-α (50 ng/ml) for 4 h, washed twice with iced PBS, and detached by trypsinization. Nuclear proteins were separated by hypotonic lysis followed by high-salt extraction treatment of nuclei. Protein concentration was determined using the Bradford method. Nuclear extracts were brought to a concentration of 0.5 μg/ml to perform EMSA.

EMSA experiments were performed as previously described (34). Double-stranded synthetic ODNs, designed on the putative consensus sequences of TFs NF-κB, NF-IL6, AP-1, CHOP, and CREB localized within proximal promoter region of the IL-8 gene (see Table I), were used. ODNs were labeled with [γ-³²P]ATP using 10 U T4 polinucleotide kinase (MBI Fermentas, St. Leon-Rot, Germany) in 500 mM Tris-HCl (pH 7.6), 100 mM MgCl₂, 50 mM DTT, and 1 mM EDTA in the presence of 50 mCi [γ-³²P]ATP in a volume of 20 μl for 45 min at 37°C. Complementary ODNs (150 ng) were added in 150 mM NaCl buffer for annealing reaction, performed at 100°C for 5 min before being left diminish to room temperature overnight. Nuclear extracts (2.5 ng) from IB3-1 cells were used and poly(dI:dC) (1 mg/reaction) was also added to abolish nonspecific binding. After 30 min binding at room temperature, the samples were run at constant voltage (200 V) under low ionic strength conditions (0.25× TBE buffer: 22 mM Tris-borate, 0.4 mM EDTA) on 6% polyacrilamide gels. Gels were dried and finally subjected to standard autoradiographic procedures.

A chromatin immunoprecipitation (ChIP) assay was performed by using a ChIP assay kit (Upstate Biotechnology, Lake Placid, NY) according to the manufacturer’s instructions. Briefly, a total of 5 × 10⁶ IB3 cells (from two 6-cm-diameter Petri dishes) were treated for 10 min at room temperature with 1% formaldehyde culture medium. Cells were washed in PBS, and then glycine was added to a final concentration of 0.125 M. The cells were then suspended in 0.5 ml lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-Cl [pH 8.1]) plus protease inhibitors (1 μg/ml pepstatin A, 1 μg/ml leupeptin, 1 μg/ml aprotinin, 1 mM PMSF) and the chromatin was subjected to sonication (using a Sonics Vibra-Cell VC130 sonicator with a 2-mm probe). Fifteen 15-s sonication pulses at 30% amplitude were required to shear chromatin to 200- to 1000-bp fragments. Aliquots (0.2 ml) of chromatin were diluted to 2 ml in ChIP dilution buffer containing protease inhibitors and then cleared with 75 μl salmon sperm DNA/protein A-agarose 50% gel slurry (Upstate Biotechnology) for 1 h at 4°C before incubation on a rocking platform with either 10 μg NF-κB p65-specific antiserum (sc-372X; Santa Cruz Biotechnology) or normal rabbit serum (Upstate Biotechnology). Diluted chromatin (20 μl) was saved and stored for later PCR analysis as 1% of the input extract. Incubations occurred overnight at 4°C and continued an additional 1 h after the addition of 60 μl protein A-agarose slurry. Thereafter the agarose pellets were washed consecutively with low-salt, high-salt, and LiCl buffers. DNA/protein complexes were recovered from the pellets with elution buffer (0.1 M NaHCO₃ with 1% SDS), and cross-links were reversed by incubating overnight at 65°C with 0.2 M NaCl. The samples were treated with RNase A and proteinase K, extracted with phenol/chloroform, and ethanol-precipitated. The pelleted DNAs were washed with 70% ethanol and dissolved in 40 μl Tris-EDTA. Aliquots (2 μl) were used for each real-time PCR reaction to quantitate immunoprecipitated promoter fragments.

For quantitative real-time PCR reactions, a pair of primers that amplify a 301-bp region on the IL-8 promoter, containing the NF-κB binding site, was designed (forward, 5′-TCA CCA AAT TGT GGA GCT TCA GTA T-3′; reverse, 5′-GGC TCT TGT CCT AGA AGC TTG TGT-3′). PCR reactions were also performed using negative control primers that amplify a 255-bp genomic region ∼5 kb upstream of the IL-8 promoter, lacking NF-κB binding sites (forward, 5′-TCC CTA AGT CAC TTT CTT CAA GTT GC-3′; reverse, 5′-CGT GCA TTT AAT TGT GTC TTG TGG-3′). Each real-time PCR reaction was performed in 25 μl final volume, using 2 μl template DNA (from ChIPs), 10 pmol primers, and 1× iQ SYBR Green Supermix (Bio-Rad) for a total of 45 cycles (96°C for 15 s, 66°C for 30 s, 72°C for 20 s) using an iCycler IQ (Bio-Rad). The relative proportions of immunoprecipitated promoter fragments were determined based on the threshold cycle (Ct) value for each PCR reaction. Real-time PCR data analyses were obtained using the comparative Ct method: a ΔCt value was calculated for each sample by subtracting the Ct value for the sample amplified with IL-8 promoter primers from the Ct value obtained for the same sample amplified with negative control primers. For each kind of immunoprecipitation (IgG or NF-κB p65 antiserum), a ΔΔCt value was then calculated by subtracting the ΔCt value for the untreated cells sample from the ΔCt value for the treated cell samples. Fold differences were then determined by the 2^−ΔΔCt method. Each sample was quantitated in duplicate for at least three separate experiments. Mean ± SD values were determined.

Human bronchial epithelial cells (IB3-1, CuFi-1, or Calu-3) were seeded in 24-well plates at a density of 30,000 cells/cm² and transfected with ODNs using cationic liposome vector Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. Lipofectamine 2000 (4 μl) was diluted in 1 ml serum-free LHC-8 basal medium (Biofluids) and double-stranded decoy or scrambled ODNs (2 μg) were added and incubated for 10 min to generate liposome/DNA complexes as previously described (34). Liposome/DNA complexes were then added to IB3-1 cells and incubated for 6 h. After this time of incubation cells were washed twice with serum-free culture medium and left at 37°C and 5% CO₂ for 20–24 h before proinflammatory challenge with P. aeruginosa (100 CFU/ml), IL-1β (10 ng/ml), or TNF-α (50 ng/ml) for a further 4 h.

Total RNA from IB3-1 cells was purified using a High Pure RNA Isolation Kit (Roche, Mannheim, Germany), and 2.5 μg RNA was reverse-transcribed to cDNA using the High Capacity cDNA Archive Kit and random primers (Applied Biosystems, Foster City, CA) in a final reaction volume of 100 μl. cDNA (5 μl) was mixed with Fast SYBR Green Master Mix (Applied Biosystems) and IL-8 primers (15 nM forward sequence, 5′-GACCACACTGCGCCAACA-3′; 15 nM reverse sequence, 5′-GCTCTCTTCCATCAGAAAGTTACATAATTT-3′). GAPDH gene expression was determined as the normalizer gene (2.5 nM primer forward sequence, 5′-GTGGAGTCCACTGGCGTCTT-3′; 15 nM reverse sequence, 5′-GCAAATGAGCCCAGCCTTC-3′). Primer sets were purchased from Sigma-Genosys (The Woodlands, TX). Real-Time PCR was performed in duplicate for both target and normalizer genes using the ABI Prism 7900 HT Fast real-time PCR system (Applied Biosystems) as follows: 50°C for 2 min, 95°C for 20 s, and 40 two-step cycles of 95°C for 1 s and 60°C for 20 s. Results were collected with SDS 2.3 software (Applied Biosystems), and relative quantification was performed using the Ct method. Data were analyzed with RQ Manager software 1.2 (Applied Biosystems). Changes in mRNA expression levels were calculated after normalization to calibrator gene. The ratios obtained after normalization are expressed as fold change over untreated samples.

It has been already defined that surface structures and soluble products from P. aeruginosa interact with TLR2/4/5, activating directly a MyD88-dependent cascade and leading to the expression of IL-8 and other proinflammatory genes in human bronchial epithelial cells (14, 15). P. aeruginosa is known to activate phosphorylation of MAPKs p38 and ERK1/2, which has been related to IL-8 gene expression (37). MAPK p38 was shown to phosphorylate HSP27, the latter leading to activation of IKK and NF-κB, in human colonic epithelial cells upon induction by dextran sodium sulfate (38). Moreover, ERK1/2 phosphorylates RSK and MSK (39). RSK has been shown to be activated also by phosphorylation of AKT and it has been implicated in the activation of other transcription factors relevant to IL-8 gene expression, such as CREB, AP-1, and NF-IL6 (39). Another relevant kinase is GSK3, which has been shown to induce IL-8 expression in neuroblastoma cells (40). As a possible feedback mechanism, GSK3 activation can be inhibited through phosphorylation elicited by activated AKT (41). Thus, we started investigating the potential implication of MAPK JNK, RSK, MSK, HSP27, GSK3, and AKT, besides that known of MAPK ERK1/2 and p38, in the P. aeruginosa-dependent induction of IL-8 in the bronchial epithelial IB3-1 cell line. Cells were exposed to the P. aeruginosa laboratory strain PAO1 for 30 min, and cell lysates were extracted to detect the phosphorylation of these different kinase substrates with a human phospho-MAPK array.

As far as the MAPK p38 pathway is concerned, a clear phosphorylation of p38 isoforms α and γ and of the downstream substrate HSP27 can be observed (Fig. 1A, 1B). No striking phosphorylation of the MAPK ERK1/2 was found. However, a clear phosphorylation of RSK1 and MSK2, which are substrates of activated ERK1/2, was evident (Fig. 1A, 1B). With regard to the ERK1/2-independent kinases known to phosphorylate RSK1/2, we did not observe further phosphorylation of AKT-1/2/3 by P. aeruginosa, whereas phosphorylation of GSK3 was increased (Fig. 1A, 1B). No significant basal or P. aeruginosa-stimulated phosphorylation of the MAPK JNK family was observed (Fig. 1A, 1B). To test whether the kinases activated by P. aeruginosa are implicated in the activation on IL-8 gene transcription, we preincubated the cells with the pharmacological inhibitors of MAPK p38 (SB203580), HSP27 (KRIBB3), ERK1/2 (AG1288 and U1026), RSK (SL0101), GSK3 (GSK3-inh), IKK (parthenolide), and, as a negative control, of JNK (SP600125) before the 4-h infection with P. aeruginosa. As shown in Fig. 1C, the inhibitors of MAPK p38 and that of its substrate HSP27 sharply reduced IL-8 mRNA content, confirming and extending the role of this signaling pathway. Similarly, inhibition of ERK1/2 resulted in a significant inhibition of P. aeruginosa-dependent IL-8 gene transcription by testing both AG1288 and U1026 inhibitors, possibly partly dependent on its downstream kinase RSK (Fig. 1C). A partial reduction was also evident upon GSK3 inhibition, which is in agreement with the data reported in other cell models indicating GSK3 as an IL-8 inducer (40). Finally, IKK inhibition by parthenolide, which in turn inhibits NF-κB activation, showed a strong reduction of IL-8 mRNA content (Fig. 1C). In summary, the results obtained in IB3-1 cells suggest that P. aeruginosa-dependent transmembrane signaling involved in IL-8 gene transcription appears associated with MAPK p38, ERK, and GSK3 pathways. Interestingly, MAPK p38 downstream IKK effector HSP27 and the ERK substrate RSK1 seem implicated in this model of inflammation. As previously reported (29), IKK was closely related to IL-8 gene expression in CF respiratory epithelial cells. The pilot experiments in IB3-1 cells were extended to the CuFi-1 and Calu-3 human epithelial bronchial cell lines, expressing either F508del-mutated CFTR or wild-type CFTR protein, by analyzing the effect of the same inhibitors on the expression of IL-8 gene both at the transcript and protein levels. As shown in Fig 1D–G, a consensus is evident for the involvement of p38, HSP27, ERK, RSK, and IKK in all the three cell lines. The modest, albeit statistically significant, involvement of GSK3 in IB3-1 cells is not consistently confirmed in the other cell lines, whereas SP600125, which inhibits JNK, shows an unexpected reduction of IL-8 mRNA expression and protein release in CuFi-1 and Calu-3 cells.

Because P. aeruginosa activates in parallel both TLR/MyD88-dependent and cytosolic calcium-dependent pathways by promoting the release of nucleotides in the extracellular milieu, which bind to purinergic P2Y receptors (12, 13, 37), to further elucidate the role of the MyD88-dependent cascade in IL-8 gene expression, we stimulated bronchial IB3-1 cells with the proinflammatory cytokine IL-1β, which activates MyD88 through its IL-1 receptor. Through MyD88 adaptor molecule and Toll/IL-1R (TIR) domain, IL-1R was shown to activate downstream kinases such as p38 and ERK in HeLa cells, finally inducing IL-8 gene expression (42). Additionally, IL-1β has been proposed to activate a sequential ERK/RSK1 cascade both in human nasal epithelial cells and in rat vascular smooth muscle cells (43, 44). Interestingly, HSP27 has been reported to regulate IL-1–dependent IKK activation in HeLa cells (45). To verify and extend these findings in our experimental model, we tested the effect of IL-1β–dependent phosphorylation with a human phospho-MAPK assay. As shown in Fig. 2A and 2B, IL-1β strongly induced phosphorylation of MAPK p38 isoforms α and γ. A modest induction of the isoform δ and no phosphorylation of the isoform β were observed (Fig. 2A, 2B). Additionally, an increased phosphorylation of HSP27 was detected upon exposure of bronchial cells to IL-1β stimulation (Fig. 2A, 2B). Furthermore, enhancement of phosphorylation of both ERK2 and its substrates RSK1 and MSK2 was observed. No changes of phosphorylation levels of JNK and AKT were detected, whereas a reduction of phosphorylation of GSK3β was shown in respect to basal levels (Fig. 2A, 2B). Similar to observations obtained with P. aeruginosa-dependent stimulation, inhibition of MAPK p38 and HSP27 by SB203580 and KRIBB3, respectively, caused a significant reduction of IL-1β–dependent IL-8 gene transcription (Fig. 2C). Moreover, pharmacological inhibition of ERK1/2, with both AG1288 and U1026 and RSK and JNK, showed a sharp inhibition of IL-8 mRNA content (Fig. 2C). On the contrary, GSK3 inhibitor did not reduce IL-8 gene transcription elicited by IL-1β (Fig. 2C). Finally, parthenolide strongly inhibited IL-8 gene transcription (Fig. 2C). To study a completely MyD88-independent signaling pathway of regulation of IL-8 gene transcription, we exposed IB3-1 cells to TNF-α, which has been reported to transduce p38 and JNK signaling through TNFR-associated factor protein activation (27, 46). Additionally, HSP27 has been reported to be activated after TNF-α stimulation through TAK1 kinase (21). As shown in Fig. 3A and 3B, TNF-α induced phosphorylation of MAPK p38 isoform δ but not α, β, and γ. Additionally, a 2-fold induction of HSP27 was found (Fig. 3A, 3B). In agreement with previously reported results (27, 46), we observed phosphorylation of JNK1 and JNK2 phosphokinases (Fig. 3A, 3B). Interestingly, TNF-α inhibited to different extents the basal phosphorylation of several substrates, such as AKT-1 and AKT-3, ERK 1/2, MSK2, and GSK3β. As shown in Fig. 3C, pharmacological inhibition of MAPK p38 and HSP27 led to strong reduction of TNF-induced IL-8 gene transcription, confirming the role of the p38 pathway and extending the comprehension of HSP27 activation. Also, inhibition of JNK kinases led to a significant decrease of IL-8 mRNA expression, whereas no effect was observed using both ERK1/2 inhibitors AG1288 and U1026 and RSK inhibitor SL0101 (Fig. 3C). Furthermore, the GSK3 inhibitor partially reduced the transcription of IL-8 gene (Fig. 3C). Finally, IKK inhibition by parthenolide showed a potent decrease of IL-8 mRNA expression mediated by TNF-α (Fig. 3C). Taken together, the consensus obtained with the results from the three different cell lines, expressing either the wild-type or mutated CFTR protein, so far suggests that the MyD88/TIR-dependent pathways elicited by P. aeruginosa and IL-1β occur through the MAPK p38 isoforms α, δ, and γ, the p38 substrate HSP27, and the MAPK ERK2 and its substrate RSK. Additionally, the MyD88-independent TNF-α–induced pathway does not activate ERK1/2, but it does activate MAPK JNK and only the isoform δ of MAPK p38.

Despite the fact that NF-κB and AP-1 are widely established as TFs implicated in the expression of IL-8 gene in different cell models (27, 34), other TFs have been recently reported as possible regulators of IL-8 gene transcription, such as the TF CHOP in T-lymphocytes and in human bronchial IB3-1 cells (30, 47), the TF CREB in U937 monocytic cells and in A549 cells (28, 48), and the TF NF-IL6 in human conjunctival and bronchial cells induced by P. aeruginosa (48). To build a comprehensive picture of the different TFs intervening in human bronchial epithelial cells challenged with P. aeruginosa, we first made an in silico analysis of the proximal region of the IL-8 gene promoter (from start site up to −180 bp) with the TF search software TESS (http://www.cbil.upenn.edu/cgi-bin/tess/tess) to obtain a prediction of consensus sequences for TFs. Fig. 4 summarizes the major TF regulatory elements within the proximal region of IL-8 gene promoter indicated by in silico analysis, namely the consensus sequences for NF-κB (−80/−72 bp), NF-IL6 (−93/−84 bp), AP-1 (−126/−120 bp), and CREB (−171/−164 bp). Because a CHOP consensus sequence has been described at –137/−130 bp from the start site of the IL-8 promoter region (47), this has been included in our scheme, although it is not evidenced by TESS. We previously observed that TF decoy ODNs directed against the TF NF-κB strongly and selectively interfere with the transcription of the IL-8 gene in human bronchial IB3-1, CuFi-1, Calu-3, and BEAS-2B cell lines (34). Starting from our previous experience and this in silico analysis, decoy ODNs homologous to these sequences have been synthesized with the addition of short 5′ and 3′ flanking regions, as reported in Table I. In addition to the sequences identified for known TFs, the sequences localized between −119/−94 and −163/−138 bp were termed intermediate sequences A and B (ISA and ISB) and tested to evaluate their effects on IL-8 gene transcription. To check whether the decoy ODN molecules synthesized are able to interfere with the biological activity of TFs NF-κB, NF-IL6, AP-1, CHOP, and CREB, nuclear extracts of stimulated cells were produced and preincubated with each TF decoy ODN and run by EMSAs. As a source of TFs, IB3-1 cells were stimulated with IL-1β, which induces abundant amounts of TFs in the unfractionated nuclear preparation. The results shown in Fig. 5 demonstrate that the decoy ODNs completely suppress the molecular interactions of these NFs with their specific target sequences. Additionally, similar inhibitory activity of the ODN decoys have been obtained, as expected, using nuclear extracts prepared from untreated IB3-1 cells, as well as cells stimulated with P. aeruginosa or TNF-α, in which >95% inhibition of DNA/protein interactions was obtained. Finally, the inhibitory effects were considered specific, since inhibition of DNA/protein interactions were obtained only with each specific ODN decoy molecule, with the others being nonactive or exhibiting a significantly lower activity. This is shown in Fig. 6, which reports the results of a competitive EMSA experiment in which 100 ng cold competitor decoy NF-κB, NF-IL6, AP-1, CHOP, and CREB oligonucleotides were incubated for 20 min with nuclear extracts before addition of the [³²P]-labeled EMSA probes. Fig. 6B clearly shows that NF-IL6, AP-1, CHOP, and CREB double-stranded oligonucleotides do not compete with the NF-κB probe for binding to nuclear extracts. As expected, full suppression of the interactions between the [³²P]-labeled NF-κB probe and nuclear extracts was observed when a cold NF-κB double-stranded ODN competitor was employed. Similarly, no cross-competition was observed using [³²P]-labeled CREB (Fig. 6C), NF-IL6 (Fig. 6D), AP-1 (Fig. 6E), and CHOP (Fig. 6F), with the exception of the CREB oligonucleotide, which to some extent competes with AP-1 (but with lower efficiency, as expected; see Fig. 6E). The competitive EMSA analysis reported in Fig. 6 does support the concept that the decoy approach leads to specific inhibition of TF/DNA interaction; however, these results do not formally demonstrate that the decoy molecules interfere with the transcription factor activity on the IL-8 promoter in intact cells. This possibility was assessed using the NF-κB model system with the objective of determining whether the decoy approach leads to reduction of the recruitment of NF-κB to the IL-8 gene promoter; toward this end, ChIP experiments were performed. IB3-1 cells were treated with PAO1 or PAO1 plus NF-κB decoy ODN, nuclear/DNA interactions were stabilized with formaldehyde, and the shared chromatin was immunoprecipitated with Abs against NF-κB. In parallel, the same procedure was performed with untreated IB3-1 cells as control. The DNA from immunoprecipitates was isolated and amplified by PCR with primers specific for the IL-8 promoter, as shown in Fig. 7A and 7B. Quantitative real-time PCR profiles for the amplification of the IL-8 promoter demonstrate that treatment of IB3-1 cells with PAO1 induces a fast and sharp recruitment of NF-κB to the IL-8 gene promoter, as shown in Fig. 7C and 7D. This effect is inhibited in the presence of NF-κB decoy ODN, indicating that the TF decoy approach is able to interfere with the recruitment and binding of TFs to the IL-8 gene promoter.

IB3-1 cells were transfected with IL-8 NF-κB, IL-8 NF-IL6, IL-8 AP-1, IL-8 CHOP, IL-8 CREB decoy ODNs, ISA and ISB ODNs, or scrambled ODN, each complexed with cationic liposomes for 20–24 h before exposure to P. aeruginosa, IL-1β, or TNF-α for further 4 h. As shown in Fig. 8, P. aeruginosa-dependent IL-8 transcription in bronchial epithelial cells was significantly inhibited using decoy ODNs against the TFs NF-κB, AP-1, CHOP, and CREB, whereas NF-IL6 decoy ODN showed no inhibitory activity in IB3-1 cells. Instead, IL-8 gene transcription upon exposure to IL-1β was diminished after preincubation of decoy ODNs interfering with TFs NF-κB, NF-IL6, AP-1, CHOP, and CREB. Third, TNF-α–mediated IL-8 gene transcription was reduced only by preincubation of decoy ODNs against NF-κB, NF-IL6, and AP-1. These results overall suggest a differential interference of TF decoy ODNs depending on the proinflammatory stimulus in that the TNF-α pathways seem independent of the participation of the TFs CHOP and CREB. Focusing again on P. aeruginosa, we previously demonstrated the strong inhibitory effect of TF NF-κB decoy ODNs on the expression of IL-8 gene in CuFi-1 and Calu-3 cells. In this study, we extended the analysis on the effect of the TF decoy ODNs against NF-IL6, AP-1, CHOP, and CREB. As shown in Fig. 9, all of these TF decoy ODNs significantly inhibited IL-8 gene transcription and protein secretion in CuFi-1 and Calu-3 cells. Finally, no effect on IL-8 gene transcription was observed by preincubating ISA and ISB ODNs before stimulation with P. aeruginosa, IL-1β, or TNF-α. The lack of inhibition of IL-8 transcription always observed with ISA and ISB ODNs, based on the sequence localized from −119 to −94 and from −163 to −138 bases from the transcription start site of the IL-8 gene promoter, suggests the absence of putative regulatory elements, as anticipated with in silico TESS analyses, in these intervening sequences. Collectively, the results presented in this study suggest that the induction of IL-8 gene transcription by P. aeruginosa in our human bronchial epithelial cell models is associated with the activation of MAPK ERK and its substrate kinase RSK, of the MAPK p38 and its substrate HSP27, and of the TFs NF-κB, NF-IL6, AP-1, CHOP, and CREB.

A thorough understanding of the molecular pathways regulating the chronic lung inflammation in CF is relevant to design novel approaches to reduce the progressive damage of the pulmonary tissue. Lung damage is thought to be mediated by an excessive recruitment of neutrophils, lead by overexpression of the chemokine IL-8, which is now considered a key therapeutic target. It is already known that P. aeruginosa, upon interaction with PRRs expressed in respiratory epithelial cells, induces a proinflammatory response involving the participation of the MAPKs p38 and ERK, as well as of the nuclear transcription factors NF-κB, NF-IL6, and AP-1, via MyD88-dependent cascades and intracellular calcium (13, 24). In this paper, we basically confirm this scenario also in our experimental model, and we further propose that RSK, GSK3, HSP27, and the transcription factors CREB and CHOP are implicated in the process.

To investigate the cascade of kinases activated by P. aeruginosa, we performed a phosphokinase assay. We confirm that P. aeruginosa induces phosphorylation of the MAPK p38 pathway and, specifically, of the p38 isoforms α, δ, and γ and of their downstream effector HSP27. Its role in IL-8 gene expression is confirmed by the pharmacological inhibition of both p38 and HSP27 with SB203580 and KRIBB3, respectively. As HSP27 was reported to participate always in IL-8 mRNA expression, our data may suggest a p38δ-dependent activation of HSP27, also because p38δ is the only isoform activated by all three proinflammatory challenges. We did not observe a direct P. aeruginosa-dependent phosphorylation of ERK1/2, as reported by other investigators (13). However, we found a clear increase of phosphorylation of RSK1 and MSK2, which are substrates of ERK1/2, and, in parallel, we found an inhibitory effect of P. aeruginosa-dependent IL-8 gene transcription with the inhibitors AG1288 and U1026 (ERK1/2) and SL0101 (RSK). A possible explanation of the lack of direct phosphorylation of ERK1/2 could be due to a very quick dephosphorylation of ERK within the 30 min incubation performed in the assay. With everything considered, we are in favor of a role of the ERK pathway in P. aeruginosa-dependent expression of the IL-8 gene. The phosphokinase assay showed also the increased phosphorylation of GSK3, and that its pharmacological inhibition significantly but modestly decreased P. aeruginosa-dependent transcription of IL-8 in IB3-1 cells. GSK3 has been described to phosphorylate >40 different substrates and to modulate the activation of 18 transcription factors, including NF-κB, AP-1, CREB, and C/EBP (49), which could be relevant in IL-8 gene expression. However, the role of GSK3 in our experimental model is very controversial, as its pharmacological inhibition reduces only slightly IL-8 gene transcription in IB3-1 cells and does not change it significantly in CuFi-1 and Calu-3 cells. Thus, the complete mechanisms of activation and the pattern of transcription factors modulated by GSK3 in our model deserve further specific investigations.

The results obtained in this study with three proinflammatory challenges could provide suggestions related to MyD88-dependent cascades involved in the regulation of transcription of the IL-8 gene. Both P. aeruginosa and the proinflammatory cytokine IL-1β are known to transduce intracellular signals through a MyD88/TIR-dependent cascade, whereas TNF-α utilizes alternative pathways (46). As expected from MyD88 activation, both P. aeruginosa and IL-1β activate MAPK p38 and ERK, whereas TNF-α involves p38 and JNK, but not the ERK pathway. Additionally, in this study we observed the involvement of RSK1 in the MyD88 pathway and of HSP27 in the common p38-related pathway.

A role for CHOP in the activation of transcription of the IL-8 gene induced by IL-1β and PGE₂ in the same IB3-1 bronchial epithelial cells has recently been proposed (30). In this study, we confirm the involvement of CHOP after exposure of IB3-1 cells to both IL-1β and P. aeruginosa by a TF decoy approach. The MAPK p38/HSP27 pathway has been related to CHOP activation in melanoma cells (50). Consistent with this hypothesis, we also observed phosphorylation of MAPK p38α and p38γ and a very relevant phosphorylation of HSP27 after testing proinflammatory stimuli in which CHOP seems involved, namely P. aeruginosa and IL-1β.

CREB has been previously related to IL-8 gene transcription in U937 monocytic cells exposed to VacA toxin (28), but never in bronchial epithelial cells challenged with P. aeruginosa and IL-1β, as shown in this study. As RSK has been reported as a potential activator of CREB (37), and we showed that RSK1 was phosphorylated after exposure to P. aeruginosa and IL-1β, our transcription factor decoy approach seems to confirm the relationship between the activations of RSK and CREB, at least in our experimental model. In parallel, the lack of inhibition of CREB by TF decoy ODNs in IB3-1 cells stimulated with TNF-α is consistent with the suppression of CREB activity by TNF-α, previously reported in murine Leydig tumor cells (51). In any case, we were not able to reach full suppression of IL-8 gene transcription using single TF decoys, suggesting that the single transcription factors in this study can be judged as required, but not sufficient, to sustain full activated transcription of the IL-8 gene.

In conclusion, the present study is intended to increase knowledge on different activatory pathways leading to transcription of the IL-8 gene in the specific context of bronchial epithelial cells. The working hypothesis, derived from the studies of others (13–32, 37–51) and from our results (12, 52, 53), is summarized in the model depicted in Fig. 10. From a practical point of view, our model confirms the redundancy of the phosphoproteins and TFs involved, a model in which no single step can completely inhibit IL-8 gene transcription. Additionally, this suggests several possible regulatory targets to modulate the excessive expression of this chemokine, besides the well-known TF NF-κB, for the treatment of the lung pathology of patients affected by CF.

We are grateful to Alice Prince for donating the P. aeruginosa strain PAO1, to Stephan J. Reshkin for critical revision of the manuscript and helpful discussions, to Maria Cristina Dechecchi and Elena Nicolis for helpful discussions, and to Federica Quiri for excellent technical assistance.

The authors have no financial conflicts of interest.